Ultra-Sensitive Hydrogen Peroxide Sensor Based on Peroxiredoxin and Fluorescence Resonance Energy Transfer

: In this paper, a ﬂuorescence resonance energy transfer (FRET)-based sensor for ultra-sensitive detection of H 2 O 2 was developed by utilizing the unique enzymatic properties of peroxiredoxin (Prx) to H 2 O 2 . Cyan and yellow ﬂuorescent protein (CFP and YFP) were fused to Prx and mutant thioredoxin (mTrx), respectively. In the presence of H 2 O 2 , Prx was oxidized into covalent homodimer through disulﬁde bonds, which were further reduced by mTrx to form a stable mixed disulﬁde bond intermediate between CFP-Prx and mTrx-YFP, inducing FRET. A linear quantiﬁcation range of 10–320 nM was obtained according to the applied protein concentrations and the detection limit (LOD) was determined to be as low as 4 nM. By the assistance of glucose oxidase to transform glucose into H 2 O 2 , the CFP-Prx / mTrx-YFP system (CPmTY) was further exploited for the detection of glucose in real sample with good performance, suggesting this CPmTY protein sensor is highly practical.


Introduction
Hydrogen peroxide (H 2 O 2 ) is a strong oxidant, which is widely used in bleaching agents [1,2] and disinfectant [3,4]. In vivo, it is a main component of reactive oxygen species (ROS), which has long been considered to be harmful to cells [5] and involved in the development of many diseases [6,7]. Nonetheless, it is also found to participate in many signaling pathways [8,9] and to help defend against microbe infection [10,11] and abiotic stress [12]. Moreover, it is a by-product of many enzymatic reactions [13,14], with glucose oxidase (GOX) as a typical example [15,16]. The unique and significant role of H 2 O 2 attracts great research interest in biosensor development because one can know the amount of specific enzyme substrate indirectly by measuring its quantity. Therefore, the detection of H 2 O 2 is of practical significance.
At this moment, quite a lot of methods exist to detect H 2 O 2 , and they can be classified into four categories: colorimetric [17], spectrophotometry [18], electrochemistry [19], and fluorescence [20]. Among these, fluorescence-based methods have shown advantages like high sensitivity, fast response, and ability to fulfill in situ measurement in organelles within the cell. Besides fluorescent small molecules [21] and various fluorescent nano materials [22], fluorescent protein (FP) is another frequently utilized constituent in H 2 O 2 biosensors.
Many proteins in cell recognize H 2 O 2 with high selectivity and sensitivity by the advantage of specific activity of biological enzymes towards their substrate. One can easily fuse these recognition

Protein Expression and Purification
Trx (NCBI ACCESSION: XM_652682.1) and Prx (NCBI ACCESSION: XM_676869.1) utilized in this study were both from Aspergillus nidulans. Cysteine to serine mutation (C39S) in Trx was incorporated by site-directed mutagenesis according to the instruction manual of QuikChange™ Site-Directed Mutagenesis Kit from Stratagene, and the PCR kit of Premix PrimeSTAR HS (Takara, Code No.: R040A) was used in the experiment. CFP (mTurquoise2) was cloned into pET-28a (+) plasmid and the intermediate plasmid P1 obtained. Then, Prx was cloned into P1 at the 3 end of CFP by restriction enzyme digestion and ligation with linker sequence between them and the final expression plasmid P2 (for chimeric protein CFP-Prx) was obtained. YFP (mNeonGreen) and mTrx (TrxC39S) were cloned into pET-28a (+) by similar method to get expression plasmid P4 (for chimeric protein mTRX-YFP). Expression plasmid (P2 or P4) was transformed in to competent E. coli BL21 (DE3) through heat shock at 42 • C for 90 s, and protein synthesis was induced with 0.1 mM isopropyl thiogalactoside (IPTG) at

SDS-PAGE
Protein was analyzed by nonreducing 12% SDS-PAGE and gels were made up in our own laboratory and stained by Coomassie brilliant blue. CFP-PRX and mTrx-YFP were reduced by adding DTT and passed through a desalting column pre-equilibrated with appropriate buffer. Subsequently, they were mixed together with equimolar ratio and treated with indicated concentration of H 2 O 2 at 30 • C for 5 min. At last excessive amount of N-Ethylmaleimide (NEM) was added after treatments to block remaining thiol groups prior to dilution in gel-loading buffer.

Fluorescence Measurement
Emission spectra were measure on an F-4600 fluorescence spectrophotometer (HITACHI, Tokyo, Japan) in a cuvette with excitation at 400 nm and emission from 460 to 560 nm. The FRET ratio (518 nm/476 nm) was then calculated from specific emission spectrum.

Selectivity Test
Oxidized glutathione, cysteine, tert-butyl hydroperoxide (t-BOOH), superoxide radicals and peroxynitrite anion (ONOO − ) were exploited to test the selectivity of this detection method versus H 2 O 2 . Superoxide radicals were generated by xanthine-xanthine oxidase system and ONOO − by SIN-1.

Glucose Detection
An amount of 2 µL GOX (1 U/mL) and 10 µL glucose solution with various concentrations were mixed together. Then, 988 µL of CFP-Prx and mTrx-YFP mixture of equal molar ratio was added to GOX-glucose reaction system and the FRET ratio was measured. In this approach, the relation between glucose concentration and FRET ratio was established.

Construction of the Proposed H 2 O 2 Probe
We employed CFP and YFP to construct the proposed H 2 O 2 probe, as a FRET donor and acceptor, respectively. As shown in Figure 1, CFP was linked to Prx (to form Prx-CFP) and YFP to mutant Trx (to form mTrx-YFP). Two intermolecular disulfide bonds formed in Prx homodimer in the presence of H 2 O 2 , and then mTrx reacted with the homodimer to generate mixed disulfide dimers, bringing the tethered CFP and YFP close enough to generate FRET signals, which in reverse could indicate H 2 O 2 concentration in the medium. We named this new probe CPmTY, the abbreviation of the essential mixed disulfide dimer CFP-Prx/mTrx-YFP.

SDS-PAGE
Protein was analyzed by nonreducing 12% SDS-PAGE and gels were made up in our own laboratory and stained by Coomassie brilliant blue. CFP-PRX and mTrx-YFP were reduced by adding DTT and passed through a desalting column pre-equilibrated with appropriate buffer. Subsequently, they were mixed together with equimolar ratio and treated with indicated concentration of H2O2 at 30 °C for 5 min. At last excessive amount of N-Ethylmaleimide (NEM) was added after treatments to block remaining thiol groups prior to dilution in gel-loading buffer.

Fluorescence Measurement
Emission spectra were measure on an F-4600 fluorescence spectrophotometer (HITACHI, Tokyo, Japan) in a cuvette with excitation at 400 nm and emission from 460 to 560 nm. The FRET ratio (518 nm/476 nm) was then calculated from specific emission spectrum.

Selectivity Test
Oxidized glutathione, cysteine, tert-butyl hydroperoxide (t-BOOH), superoxide radicals and peroxynitrite anion (ONOO − ) were exploited to test the selectivity of this detection method versus H2O2. Superoxide radicals were generated by xanthine-xanthine oxidase system and ONOO − by SIN-1.

Glucose Detection
An amount of 2 μL GOX (1 U/mL) and 10 μL glucose solution with various concentrations were mixed together. Then, 988 μL of CFP-Prx and mTrx-YFP mixture of equal molar ratio was added to GOX-glucose reaction system and the FRET ratio was measured. In this approach, the relation between glucose concentration and FRET ratio was established.

Construction of the Proposed H2O2 Probe
We employed CFP and YFP to construct the proposed H2O2 probe, as a FRET donor and acceptor, respectively. As shown in Figure 1, CFP was linked to Prx (to form Prx-CFP) and YFP to mutant Trx (to form mTrx-YFP). Two intermolecular disulfide bonds formed in Prx homodimer in the presence of H2O2, and then mTrx reacted with the homodimer to generate mixed disulfide dimers, bringing the tethered CFP and YFP close enough to generate FRET signals, which in reverse could indicate H2O2 concentration in the medium. We named this new probe CPmTY, the abbreviation of the essential mixed disulfide dimer CFP-Prx/mTrx-YFP.

FRET Signals Respond to H 2 O 2 Concentration as a Result of CFP-Prx/mTrx-YFP Conjugation through Disulfide Bond
CFP-Prx and mTrx-YFP were successfully expressed and purified in this work. The fluorescence properties of both FPs were not interfered in each chimeric protein. At the same time, CFP-Prx retained the ability to react with H 2 O 2 to form dimer as free Prx did. When mixing and adding H 2 O 2 into the mixture, CFP-Prx/mTrx-YFP heterodimer formed as verified by SDS-PAGE ( Figure 2a). If one disulfide bond in Prx dimer remained, an mTRX and a PRX-CFP dimer formed a covalent heterotrimer. Otherwise, this heterotrimer further resolved by another mTrx-YFP and decomposed into two CFP-Prx/mTrx-YFP heterodimers. Therefore, there were five kinds of entities in the mixture with the presence of H 2 O 2 corresponding to the five bands in each lane of SDS-PAGE image from top to bottom: (1) CFP-Prx/CFP-Prx/mTrx-YFP heterotrimer, (2) CFP-Prx/CFP-Prx homodimer, (3) CFP-Prx/mTrx-YFP heterodimer, (4) monomer CFP-Prx, and (5) monomer mTrx-YFP, where (2) and (3) were very close to each other. Moreover, as the H 2 O 2 amount increased, the monomers decreased in concentration while the trimer and dimer increased. FRET happened between CFP-Prx and mTrx-YFP in hetero dimer and trimer, as revealed in the fluorescence spectra ( Figure 2b). Each spectrum corresponds to one lane in the SDS-PAGE image and more hetero dimer and trimer lead to higher FRET signals (decrease of CFP fluorescence and increase of YFP's). In this way, the H 2 O 2 concentration can be reflected by the FRET signal intensity.

FRET Signals Respond to H2O2 Concentration as a Result of CFP-Prx/mTrx-YFP Conjugation through Disulfide Bond
CFP-Prx and mTrx-YFP were successfully expressed and purified in this work. The fluorescence properties of both FPs were not interfered in each chimeric protein. At the same time, CFP-Prx retained the ability to react with H2O2 to form dimer as free Prx did. When mixing and adding H2O2 into the mixture, CFP-Prx/mTrx-YFP heterodimer formed as verified by SDS-PAGE (Figure 2a). If one disulfide bond in Prx dimer remained, an mTRX and a PRX-CFP dimer formed a covalent heterotrimer. Otherwise, this heterotrimer further resolved by another mTrx-YFP and decomposed into two CFP-Prx/mTrx-YFP heterodimers. Therefore, there were five kinds of entities in the mixture with the presence of H2O2 corresponding to the five bands in each lane of SDS-PAGE image from top to bottom: (1) CFP-Prx/CFP-Prx/mTrx-YFP heterotrimer, (2) CFP-Prx/CFP-Prx homodimer, (3) CFP-Prx/mTrx-YFP heterodimer, (4) monomer CFP-Prx, and (5) monomer mTrx-YFP, where (2) and (3) were very close to each other. Moreover, as the H2O2 amount increased, the monomers decreased in concentration while the trimer and dimer increased. FRET happened between CFP-Prx and mTrx-YFP in hetero dimer and trimer, as revealed in the fluorescence spectra ( Figure 2b). Each spectrum corresponds to one lane in the SDS-PAGE image and more hetero dimer and trimer lead to higher FRET signals (decrease of CFP fluorescence and increase of YFP's). In this way, the H2O2 concentration can be reflected by the FRET signal intensity.  To further prove that intermediate disulfide bonds was the cause for the FRET signal, we add DTT, which can destroy the bond, after H 2 O 2 and FRET signal diminished gradually in six minutes (Figure 3a). The reaction between DTT and protein disulfide bond was rather slow compared with that between Prx and H 2 O 2 , which completed in seconds. Thus we can measure the FRET signal immediately after H 2 O 2 addition, which is a benefit of this detection method. We also constructed wild-type Trx fused with YFP (wtTrx-YFP), which cannot induce FRET with CFP-Prx in the presence of H 2 O 2 (Figure 3b), as the resolving cysteine in wtTrx destroys the intermediate disulfide bond.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 5 of 11 To further prove that intermediate disulfide bonds was the cause for the FRET signal, we add DTT, which can destroy the bond, after H2O2 and FRET signal diminished gradually in six minutes (Figure 3a). The reaction between DTT and protein disulfide bond was rather slow compared with that between Prx and H2O2, which completed in seconds. Thus we can measure the FRET signal immediately after H2O2 addition, which is a benefit of this detection method. We also constructed wild-type Trx fused with YFP (wtTrx-YFP), which cannot induce FRET with CFP-Prx in the presence of H2O2 (Figure 3b), as the resolving cysteine in wtTrx destroys the intermediate disulfide bond.

Ability of CPmTY to Detect H2O2 with Low limit and Wide Range by Optimal Protein Concentration
First, we conducted the detection experiment in 20 mM Tris-HCl (pH 8.0) with 150 mM NaCl, the same buffer as used for protein purification. In this condition, the FRET ratio was very small, even after adding an excess amount of H2O2 (much like Figure 3b). We then removed NaCl from the buffer and found the FRET ratio increased significantly in response to H2O2 (much like Figure 2b). Salt ions may bind on protein interfaces to destroy hydrophobic interactions between protein dimer and prevent proteins from getting close to each other by electrostatic repulsion. Thus, NaCl may decrease the FRET ratio by impeding CFP-Prx dimerization and separating CFP and YFP in the CFP-Prx/mTrx-YFP conjugate. Moreover, YFP used in this study is pH sensitive, and shows stronger fluorescence intensity in pH 6-8. Therefore, we finally chose to examine the detection performance of CPmTY in 1 mM phosphate buffer (pH 7.0). Based on the working principle, the detection range of this method depends on the concentration of the two chimeric proteins. Here, we set the molar ratio as 1:1. With 200 nM CFP-Prx and 200 nM mTrx-YFP, the FRET ratio of CPmTY increased linearly in the range from 0 to 80 nM of H2O2. At a concentration four times of the former (i.e., 800 nM CFP-Prx and 800 nM mTrx-YFP) the linear range expanded to 120-320 nM. Based on these results, it could be concluded that H2O2 detection range by CPmTY varies with CFP-Prx and mTrx-YFP protein concentrations. However, the protein concentrations should neither be too high nor too low. As the concentration increased, spontaneous FRET occurred between separate CFP-Prx and mTRX-YFP, diminishing detection sensitivity (i.e., the difference between largest and smallest FRET ratio). As shown in Figure 4, the slope of the left fitting curve (i.e., 0.01) is greater than four times of the right (i.e., 0.0015). If the protein concentrations continue to increase, slope value of the fitting curve would tend to be zero, meaning that spontaneous FRET between separate CFP-Prx and mTrx-YFP equals to that within the CFP-Prx/mTrx-YFP conjugate. On the other hand, too low protein concentrations also damage detection performance, as the fluorescence intensity is similar to the background

Ability of CPmTY to Detect H 2 O 2 with Low Limit and Wide Range by Optimal Protein Concentration
First, we conducted the detection experiment in 20 mM Tris-HCl (pH 8.0) with 150 mM NaCl, the same buffer as used for protein purification. In this condition, the FRET ratio was very small, even after adding an excess amount of H 2 O 2 (much like Figure 3b). We then removed NaCl from the buffer and found the FRET ratio increased significantly in response to H 2 O 2 (much like Figure 2b). Salt ions may bind on protein interfaces to destroy hydrophobic interactions between protein dimer and prevent proteins from getting close to each other by electrostatic repulsion. Thus, NaCl may decrease the FRET ratio by impeding CFP-Prx dimerization and separating CFP and YFP in the CFP-Prx/mTrx-YFP conjugate. Moreover, YFP used in this study is pH sensitive, and shows stronger fluorescence intensity in pH 6-8. Therefore, we finally chose to examine the detection performance of CPmTY in 1 mM phosphate buffer (pH 7.0). Based on the working principle, the detection range of this method depends on the concentration of the two chimeric proteins. Here, we set the molar ratio as 1:1. With 200 nM CFP-Prx and 200 nM mTrx-YFP, the FRET ratio of CPmTY increased linearly in the range from 0 to 80 nM of H 2 O 2 . At a concentration four times of the former (i.e., 800 nM CFP-Prx and 800 nM mTrx-YFP) the linear range expanded to 120-320 nM. Based on these results, it could be concluded that H 2 O 2 detection range by CPmTY varies with CFP-Prx and mTrx-YFP protein concentrations. However, the protein concentrations should neither be too high nor too low. As the concentration increased, spontaneous FRET occurred between separate CFP-Prx and mTRX-YFP, diminishing detection sensitivity (i.e., the difference between largest and smallest FRET ratio). As shown in Figure 4, the slope of the left fitting curve (i.e., 0.01) is greater than four times of the right (i.e., 0.0015). If the protein concentrations continue to increase, slope value of the fitting curve would tend to be zero, meaning that spontaneous FRET between separate CFP-Prx and mTrx-YFP equals to that within the CFP-Prx/mTrx-YFP conjugate. On the other hand, too low protein concentrations also damage detection performance, as the fluorescence intensity is similar to the background interference, and at the same time both reaction possibilities between reduced CFP-Prx and H 2 O 2 and between oxidized CFP-Prx dimer and mTrx-YFP decrease drastically.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 6 of 11 interference, and at the same time both reaction possibilities between reduced CFP-Prx and H2O2 and between oxidized CFP-Prx dimer and mTrx-YFP decrease drastically. The optimal detection limit of CPmTY was determined to be 4 nM (calculated as 3SB/m, SB is the standard deviation of 20 blank responses and m the slope of calibration curve), which was much lower than most of the present H2O2 detection methods (as shown in Table 1). In the general detection test, we preferred to set the protein concentrations to a relatively low level to increase detection sensitivity. Hence, when applying a low detection range, the dilution of samples with high H2O2 concentration can eliminate the interference of other compositions such as salts in the sample.  The optimal detection limit of CPmTY was determined to be 4 nM (calculated as 3S B /m, S B is the standard deviation of 20 blank responses and m the slope of calibration curve), which was much lower than most of the present H 2 O 2 detection methods (as shown in Table 1). In the general detection test, we preferred to set the protein concentrations to a relatively low level to increase detection sensitivity. Hence, when applying a low detection range, the dilution of samples with high H 2 O 2 concentration can eliminate the interference of other compositions such as salts in the sample.

Detection Selectivity
The detection of H 2 O 2 in cells or other biological samples is frequently interfered with other oxidants, so the performance of CPmTY was tested in the presence of potential interfering species. The results were shown in Figure 5. CPmTY did not react with most of them (superoxide anion, cysteine, oxidized glutathione, and peroxynitrite) and showed excellent selectivity to H 2 O 2 . The only exception is T-BOOTH, which is synthetic and does not exist in real samples. This selectivity may be ascribed to the high specificity of Prx to H 2 O 2 .

Detection Selectivity
The detection of H2O2 in cells or other biological samples is frequently interfered with other oxidants, so the performance of CPmTY was tested in the presence of potential interfering species. The results were shown in Figure 5. CPmTY did not react with most of them (superoxide anion, cysteine, oxidized glutathione, and peroxynitrite) and showed excellent selectivity to H2O2. The only exception is T-BOOTH, which is synthetic and does not exist in real samples. This selectivity may be ascribed to the high specificity of Prx to H2O2.

Detection of Glucose
Detection of glucose involves glucose oxidation by glucose oxidase to generate H2O2 and the ensuing H2O2 measurement. First, the quantitative relationship between glucose concentration and the FRET ratio was calibrated. The glucose solution was diluted to a serial of concentrations (0, 2, 4, 6, 8, 10, 12 μM), and reacted with GOX. Subsequently, CFP-Prx and mTrx-YFP mixture, both 150 nM, was added to the glucose oxidation system and fluorescence spectra were measured. The glucose oxidation was conducted in a small volume in order to obtain relatively higher enzyme and substrate concentration for fast and complete reaction. Figure 6 showed the corresponding fluorescence spectrum from each measurement, labeled by the glucose concentration. The FRET ratio (518 nm/476 nm) was calculated and plotted versus glucose concentration as a calibration curve (inset in Figure  6). Finally, the calibration equation of R = 0.0039C + 0.822 (R 2 = 0.9809) was obtained for the glucose analysis. Herein, R is the FRET ratio (518 nm/476 nm), R 2 is correlation coefficient and C is the concentration of glucose (nM).

Detection of Glucose
Detection of glucose involves glucose oxidation by glucose oxidase to generate H 2 O 2 and the ensuing H 2 O 2 measurement. First, the quantitative relationship between glucose concentration and the FRET ratio was calibrated. The glucose solution was diluted to a serial of concentrations (0, 2, 4, 6, 8, 10, 12 µM), and reacted with GOX. Subsequently, CFP-Prx and mTrx-YFP mixture, both 150 nM, was added to the glucose oxidation system and fluorescence spectra were measured. The glucose oxidation was conducted in a small volume in order to obtain relatively higher enzyme and substrate concentration for fast and complete reaction. Figure 6 showed the corresponding fluorescence spectrum from each measurement, labeled by the glucose concentration. The FRET ratio (518 nm/476 nm) was calculated and plotted versus glucose concentration as a calibration curve (inset in Figure 6). Finally, the calibration equation of R = 0.0039C + 0.822 (R 2 = 0.9809) was obtained for the glucose analysis. Herein, R is the FRET ratio (518 nm/476 nm), R 2 is correlation coefficient and C is the concentration of glucose (nM).
To verify the applicability of CPmTY for glucose detection, Dulbecco's modified eagle medium (DMEM, Corning Cat. No. 10-013) was tested with labeled glucose concentration of 4500 mg/L, that is, 25 mM. The result was 24.58 mM, indicating a recovery of 98.32 ± 0.47% and confirming the applicability of CPmTY.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 11 To verify the applicability of CPmTY for glucose detection, Dulbecco's modified eagle medium (DMEM, Corning Cat. No. 10-013) was tested with labeled glucose concentration of 4500 mg/L, that is, 25 mM. The result was 24.58 mM, indicating a recovery of 98.32 ± 0.47% and confirming the applicability of CPmTY.

Conclusions
In this work, we developed a mild, sensitive, and fast detection method for H 2 O 2 by the distinctive reaction among H 2 O 2 , Prx, and mTrx, which can bring tethered CFP and YFP together to induce FRET. The detection limit was determined to be as low as 4 nM. This sensitive method shows satisfactory selectivity toward H 2 O 2 over other interfering oxidants. Moreover, this method can be applied to detect glucose content in DMEM with the aid of GOX and the result is very close to the labeled data. As there are numerous H 2 O 2 reacting proteins, such as Prx and H 2 O 2 transforming reactions, like GOX in nature, this study provides a novel idea to develop biosensors for H 2 O 2 and substances that can be transformed to H 2 O 2 by suitable enzymes.