Fluorescent probe for the imaging of superoxide and peroxynitrite during drug-induced liver injury†

Drug-induced liver injury (DILI) is an important cause of potentially fatal liver disease. Herein, we report the development of a molecular probe (LW-OTf) for the detection and imaging of two biomarkers involved in DILI. Initially, primary reactive oxygen species (ROS) superoxide (O2˙−) selectively activates a near-infrared fluorescence (NIRF) output by generating fluorophore LW-OH. The C 
<svg xmlns="http://www.w3.org/2000/svg" version="1.0" width="13.200000pt" height="16.000000pt" viewBox="0 0 13.200000 16.000000" preserveAspectRatio="xMidYMid meet"><metadata>
Created by potrace 1.16, written by Peter Selinger 2001-2019
</metadata><g transform="translate(1.000000,15.000000) scale(0.017500,-0.017500)" fill="currentColor" stroke="none"><path d="M0 440 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z M0 280 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z"/></g></svg>
 C linker of this hemicyanine fluorophore is subsequently oxidized by reactive nitrogen species (RNS) peroxynitrite (ONOO−), resulting in cleavage to release xanthene derivative LW-XTD, detected using two-photon excitation fluorescence (TPEF). An alternative fluorescence pathway can occur through cleavage of LW-OTf by ONOO− to non-fluorescent LW-XTD-OTf, which can react further with the second analyte O2˙− to produce the same LW-XTD fluorescent species. By combining NIRF and TPEF, LW-OTf is capable of differential and simultaneous detection of ROS and RNS in DILI using two optically orthogonal channels. Probe LW-OTf could be used to detect O2˙− or O2˙− and ONOO− in lysosomes stimulated by 2-methoxyestradiol (2-ME) or 2-ME and SIN-1 respectively. In addition, we were able to monitor the chemoprotective effects of tert-butylhydroxyanisole (BHA) against acetaminophen (APAP) toxicity in living HL-7702 cells. More importantly, TPEF and NIRF imaging confirmed an increase in levels of both O2˙− and ONOO− in mouse livers during APAP-induced DILI (confirmed by hematoxylin and eosin (H&E) staining).


Introduction
Drug-induced liver injury (DILI) is a leading cause of acute liver failure in the USA and Europe, and has as such raised serious concerns for public health. 1-3 The potential to induce DILI is also one of the most common causes of compound attrition in drug development, oen leading to drug withdrawals, restrictions, and project termination. 3 Minimizing hepatotoxicity is therefore crucial, requiring effective techniques for preclinical screening of DILI. 4,5 Unfortunately, probes capable of imaging DILI in living animals remain limited, 6-8 making the development of methods for accurate diagnosis vital for improving the treatment of DILI. [9][10][11] A common cause of DILI is overdose of acetaminophen (APAP), causing oxidative and nitrosative stress through elevated levels of reactive oxygen species (ROS) and reactive nitrogen species (RNS). Metabolism of APAP in the liver proceeds via transformation into a toxic metabolite, N-acetyl-pbenzoquinone imine (NAPQI). 12 APAP hepatotoxicity is known to arise from the interference of NAPQI with complex I/II of the mitochondrial electron transport chain (ETC), resulting in the leakage of electrons from the ETC to oxygen which induces superoxide (O 2 c À ) formation. 13,14 O 2 c À is then converted into hydrogen peroxide (H 2 O 2 ) and oxygen (O 2 ) by manganese superoxide dismutase, leading to additional oxidative stress. In addition, O 2 c À can react with endogenous nitric oxide (NO) to generate RNS peroxynitrite (ONOO À ). 15 Considering these ROS and RNS are products of different pathways, and exhibit different biological effects, their simultaneous detection could improve our understanding of the in vivo mode of action in DILI. 8,[16][17][18] Fluorescence imaging is commonly used as a non-invasive method to image and measure these types of analytes with high temporal and spatial resolution suitable for diagnostic applications in living organisms. 19 Improvements to uorescent methods can be made by using near-infrared uorescence (NIRF, 650-900 nm), which benets from minimized auto-uorescence of endogenous biomolecules and reduced light scattering in tissues. [20][21][22] While a number of NIRF-based probes have been used for the detection of O 2 c À , [23][24][25][26][27][28] none have yet been used to evaluate changes in O 2 c À in DILI. Further improvements to biological uorescence imaging can be achieved with the use of two-photon excitation uorescence (TPEF), which brings unique benets such as increased spatial resolution and enhanced penetration depths. 29,30 Nevertheless, such probes capable of two-photon excitation suitable for investigating the role of ONOO À in DILI are still rare. 7 Our groups' research interests lie in developing new uorescent probes for the detection and imaging of ROS and RNS, with a recent focus on dual-response probes. [30][31][32][33][34] Carrying on this work, uorescent probe LW-OTf was designed and synthesized with the intention imaging of DILI. O 2 c À , a primary ROS, and ONOO À , a prominent RNS, were chosen as pertinent DILI-related biomarkers for this study. To the best of our knowledge, LW-OTf represents the rst reaction-based smallmolecule uorescent probe having NIRF and TPEF capabilities with two independent optical channels: NIRF for O 2 c À and TPEF for ONOO À (Scheme 1a and b).
The synthesis of LW-OTf was carried out over two steps (Scheme 1c and S3 †). Hemicyanine-based uorophore LW-OH was rst prepared by retro-Knoevenagel reaction 35 followed by addition of a O 2 c À -reactive triuoromethylsulfonyl unit (triyl, Tf) to afford LW-OTf 36,37 (for further discussion of the TfO À counteranion see the ESI †). In the presence of O 2 c À triyl deprotection occurs, which leads to an increased NIRF signal by generation of uorophore LW-OH. Subsequent reaction with ONOO À results in oxidative cleavage of the alkene linker of LW-OH to generate xanthene derivative LW-XTD, [38][39][40] capable of twophoton uorescence (see ESI † for further discussion of hemicyanine-xanthene uorescent turn-on mechanism and selectivity). An alternative uorescence activation pathway can also occur, in which LW-OTf is rst cleaved by ONOO À to produce non-uorescent LW-XTD-OTf, which can subsequently react with superoxide to produce the same nal xanthene derivative LW-XTD. Unfortunately, uorescence experiments could not detect this second pathway, as generation of ONOO À requires an aqueous medium, which led to rapid decomposition of KO 2 -derived O 2 c À in an assay setting, and so sequential addition of peroxyntrite then superoxide could not be carried out. 41 This dual-response molecular design allows LW-OTf to produce either NIRF or TPEF signals in response to O 2 c À and ONOO À , respectively (Scheme 1b).   and concentration of ONOO À (0-4.2 mM). l ex/em ¼ 360/461 nm. Note: O 2 c À was prepared by dissolving KO 2 in DMSO, and was then added to LW-OTf (in DMSO), followed by addition of ONOO À (in water). The mixture was diluted with PBS buffer (10 mM, pH 7.4) before each measurement. See ESI for detailed procedures. †

Results and discussion
With LW-OTf in hand, we rst evaluated its optical properties. As shown in Fig. S1, † LW-OTf (10 mM, pH 7.4) exhibited absorption maxima at 546 and 576 nm, whilst for LW-OH (formed in situ by addition of O 2 c À , 20 mM) those absorptions decreased, with a new maximum at 687 nm. Subsequent addition of ONOO À (17.5 mM) to the solution resulted in the emergence of a peak at 353 nm. These observations are in good agreement with the proposed mechanism for the sequential reaction of LW-OTf with O 2 c À followed by ONOO À (Scheme 1). The uorescence behavior ( Fig. 1 and 2) of this sensing system was then evaluated. Initially, negligible uorescence was observed, with incremental addition of O 2 c À (0-25 mM) causing a continuous increase in emission intensity at 710 nm using excitation at 675 nm ( Fig. 1a). Removal of the triyl unit in the presence of O 2 c À released uorophore LW-OH, causing a 15.6fold enhancement in uorescence emission intensity. An excellent linear relationship between the emission intensity at 710 nm and the concentration of O 2 c À over the 0-18 mM range was observed (linear equation: Fig. 1b), and the detection limit was calculated to be 46.5 nM. The uorescence behavior of LW-OTf in the presence of both O 2 c À and ONOO À was then evaluated. Excitation at 360 nm was selected for onephoton uorescence experiments, matching the observed maximum absorption for LW-XTD at 353 nm, as well as previous reports of this system. 38 LW-OTf initially exhibited a weak emission signal at 461 nm in the presence of O 2 c À , however upon subsequent addition of ONOO À the D-p-A-based oxidation product LW-XTD exhibited a strong uorescence emission at 461 nm, upon excitation at 360 nm (Fig. 2a). As the concentration of ONOO À was increased from 0 to 4.2 mM the uorescence intensity at 461 nm gradually increased as well (linear equation: y 2 ¼ 5657 + 5074 Â [ONOO À ] (mM), R 2 ¼ 0.994, y 2 is the intensity at 461 nm), with a detection limit for ONOO À of 38.2 nM (Fig. 2b). The use of two-photon microscopy, was rst described by Webb et al. in 1990, and has since been adopted for bioimaging applications. 42 Given that the excitation of a uorophore using two-photon uorescence is twice that of one-photon uorescence, an excitation wavelength of 720 nm was chosen for two-photon measurements. LW-OTf also exhibited two-photon uorescence for the detection of ONOO À in the presence of O 2 c À in vitro using an excitation of 720 nm (Fig. S2 †).
The optical selectivity of LW-OTf towards the selected ROS and RNS was then evaluated in vitro, conrming that its uorescence response was most sensitive to the presence of O 2 c À (Fig. 3a). As previously noted, C]C cleavage by ONOO À could also occur, however as this does not lead to a uorescent signal, no selectivity issues arose. Selectivity towards ONOO À was then  determined using a stepwise approach, rst incubating LW-OTf with superoxide, then adding either ONOO À or a range of other ROS. Fragmentation of LW-OH to LW-XTD resulting from RNSmediated oxidative cleavage of the C]C linker was monitored by measuring the increase in emission at 461 nm aer excitation at 360 nm. These experiments demonstrated that intermediate LW-OH was specically responsive to ONOO À over H 2 O 2 , NOc, cOH, 1 O 2 , and ClO À (Fig. 3b). pH titrations indicated that the uorescence intensity of LW-OTf was greatest at pH 7-8, matching the physiological pH at which this probe would operate in vivo (Fig. S3 and S5 †). Fluorescence intensities at 710 nm decreased signicantly at lower pH, likely due to phenolic protonation of LW-OH, resulting in reduced intramolecular charge transfer (ICT) (Fig. S3 †). [43][44][45] Decomposition of ONOO À at acidic pH is likely responsible for the decreased emission intensities at 461 nm (Fig. S5 †). 46 In order to better understand and ultimately conrm the suggested mode of action of probe LW-OTf, we assessed the time course of the reaction of LW-OTf with both O 2 c À and ONOO À (Fig. S4 and S6 †). Whilst the reaction of LW-OTf with O 2 c À was nished within 10 min (Fig. S4 †), the subsequent addition of ONOO À resulted in an instantaneous and signicant uorescence increase (Fig. S6 †). Both reaction proles are consistent with the known reactivity of both analytes, and offer promising prospects for future applications, since rapid detection is particularly important for real-time detection of O 2 c À and ONOO À in living systems. In addition, high-resolution LC-MS experiments were performed to conrm the proposed uorescence mechanisms (Fig. S7-S12 †). Addition of KO 2 (3 equiv. in DMSO) to a solution of LW-OTf (in MeOH) resulted in the representative cation of LW-OH ([M] + , m/z ¼ 412.2283), indicating the triyl-deprotection of LW-OTf by superoxide (Fig. S8 †). The mass spectra following the addition of ONOO À (1 equiv. in water) were found to be consistent with the mechanism proposed above, with detection of the mass ion for LW-XTD ([M + H] + , m/z ¼ 229.0860) as well as the indoline byproduct conrming oxidative cleavage of LW-OH by ONOO À (Fig. S9 and S10 †). 38,47 HRMS was also used to prove the alternate activation pathway, with direct addition of ONOO À (5 equiv. in water) to LW-OTf (in MeOH) generating a mass ion for LW-XTD-OTf ([M + H] + , m/z ¼ 361.0357, Fig. S12 †). 39,40 These promising in vitro results prompted us to explore the duplex imaging of LW-OTf in living cells and in mice. Using MTT assays, it was conrmed that LW-OTf was non-toxic to HL-7702 cells (Fig. S13 †). Pre-treatment of the cells with a superoxide scavenger Tiron (10 mM, 30 min), 48,49 followed by incubation with LW-OTf (2.4 mM) for a further 15 min resulted in only weak uorescence in the red channel (Fig. S14a †). Conversely, pre-treatment with Tiron followed by stimulation by varying amounts of 2-methoxyestradiol 50-52 (2-ME, 0, 0.5, 2.0, 3.0 mg mL À1 ), an O 2 c À promoter, resulted in signicant uorescence intensity enhancement in the red channel (Fig. S14a †). Only weak uorescence and no signicant change was observed in the blue channel, for which an excitation wavelength of 405 nm was used (closest available to 360 nm). These results clearly conrm the ability of LW-OTf to selectively detect superoxide in cells using NIRF.
As shown in Fig. S14b, † cells were rst pre-incubated with Tiron (10 mM), then exposed to 2-ME (3.0 mg mL À1 ), followed by staining with LW-OTf (2.4 mM), and nally incubated with SIN-1 (ONOO À donor). 53 A concentration-dependent change in uorescence emission intensity in the blue and red channel was observed. The addition of 3.0 mM SIN-1 led to a 4.25-fold enhancement of the average blue uorescence intensity and 2.54-fold decrease in the average red uorescence intensity (Fig. S14d †). Similar to the results shown in Fig. S16, LW-OTf demonstrated the ability to visualize ONOO À in lysosomes with a Pearson correlation coefficient of 0.90 (Fig. S17 †).
Since overdose of APAP leads to the overproduction of ROS and RNS, 54 APAP-induced DILI was chosen as a representative model for liver toxicity in which to evaluate the effectiveness of LW-OTf. Although this model is oen used for single detection of ONOO À , 7,55-57 simultaneous detection of RNS and ROS in DILI is still rare. 8 Treatment of HL-7702 cells with APAP and LW-OTf produced a marked increase in uorescence in both the red and blue channels (Fig. 4b and e), indicating upregulation of intracellular O 2 c À and ONOO À aer administration of APAP, and demonstrating the ability of our probe to detect concentration changes of O 2 c À by NIRF and ONOO À by TPEF in DILI. This was further conrmed using tert-butylhydroxyanisole (BHA), a ROS and RNS scavenger 58 which has been used to eliminate ROS and relieve APAP-induced liver injury. 59,60 Upon addition of BHA, the uorescence intensity for both the red and blue channels decreased ( Fig. 4c and f). Similarly, LW-OTf exhibited the expected one-photon uorescence changes in the blue and red channel from APAP-induced hepatotoxicity and remediation using BHA (Fig. S15 †).
Inspired by these cell imaging experiments, LW-OTf was used for in vivo imaging of O 2 c À and ONOO À in DILI. Towards that aim, C57 mice were treated with APAP either at analgesic low dosage (200 mg kg À1 , control group), or high dosages (400 mg kg À1 and 600 mg kg À1 ). Aer 6 h, 61,62 all three groups were given intraperitoneal injections of LW-OTf (200 mL, 48 mM), and imaging was performed aer a further 15 min. The NIRF imaging ability for O 2 c À in exposed livers was investigated rst. As shown in Fig. 5a (le mouse) and Fig. 5b, only weak uorescence at 710 nm was observed in the control group mice, implying only low concentrations of O 2 c À for low doses of APAP. In contrast, aer the administration of high doses of APAP, the mouse livers displayed signicant uorescence enhancements in a concentration-dependent manner, indicating DILI-induced overproduction of O 2 c À aer APAP treatment. Deep tissue penetration imaging of O 2 c À in depilated mice using probe LW-  (c-g) Spleen, lung, heart, kidney, liver tissue section of control group. (h-l) Spleen, lung, heart, kidney, liver tissue section of model group with APAP (600 mg kg À1 )-induced liver injury. (g) The structure of the liver lobules was clear, the hepatocytes were arranged neatly. No obvious degeneration and necrosis of hepatocytes were observed. There was no obvious congestion of hepatic sinusoid. No obvious inflammatory cell infiltration was seen. (l) Compared to the control group, there were obvious liver damages in the model group: significant congestion and hemorrhage in the sinusoids of the acinar III zone (black arrow); a large amount of hepatocyte steatosis with dense vacuoles in the cytoplasm (blue arrow); some hepatocytes with dissolved and disappeared nucleus were necrotic (green arrow); slight inflammatory infiltration and individual inflammatory foci in the liver lobules (yellow arrow). Scale bar ¼ 100 mm. The data are expressed as the mean AE SD. Five mice in each group.
OTf indicated a 1.54-fold (APAP 400 mg kg À1 ) and 2.46-fold (APAP 600 mg kg À1 ) increase in the emission at 710 nm ( Fig. 5c  and d). Furthermore, as displayed in Fig. 5e and f, the highest dosage group (APAP 600 mg kg À1 ) showed signicant uorescence enhancement over time, indicating continued DILIinduced ROS overproduction. To conrm this, N-acetylcysteine (NAC), a hepatoprotective agent, was injected into the DILI mice (APAP 600 mg kg À1 ), resulting in a signicant attenuation in the uorescent signal back to levels comparable to the control group mice.
Our attention then turned to two-photon in vivo bio-imaging of ONOO À using LW-OTf, using the same mouse models as discussed above. Following surgical treatment, the liver of each mouse was assessed using two-photon uorescence imaging with excitation at 720 nm. The livers of mice under nitrosative stress (Fig. 6c) exhibited distinct uorescence enhancements (approximately 3.68-fold) when compared to the control mice (Fig. 6a). These results validate the ability of LW-OTf to image both superoxide and peroxynitrite in vivo, using NIRF for the former, and TPEF for the latter.
Following on from these in vivo results, we wished to explore the distribution of DILI-induced ROS/RNS in mice. Again, DILI model group mice were injected intraperitoneally with a dose of APAP (600 mg kg À1 ), whilst a control group was given only physiological saline. Aer 6 h both groups were injected with LW-OTf (200 mL, 48 mM) and le for 15 min before being killed and dissected to isolate their major organs for ex vivo NIRF imaging ( Fig. 7a and b). As shown in Fig. 7a, when compared to the other organs (spleen, lung, heart, and kidney), a signicant uorescence signal was observed in the liver of both DILI and control group mice. In addition, LW-OTf exhibited a stronger uorescence signal in the livers of DILI mice than in those of healthy mice (Fig. 7b). Furthermore, hematoxylin and eosin (H&E) staining of the liver tissues and other major organ tissues (spleen, lung, heart, and kidney) was carried out to identify the histological changes during the APAP treatment. All tissue types from both the control and model groups were examined for tissue architecture, degeneration, necrosis, hemorrhage, and inammatory cell inltration, looking for splenic, pulmonary, cardiac, renal, and hepatic damage. No obvious differences or damage were observed upon comparison of the control and model tissue samples of the spleen, lung, heart and kidney, indicating clearly that no APAP-induced damage had occurred (Fig. 7c-f and h-k).
As displayed in Fig. 7g, the control group liver appeared healthy, with the structure of the liver lobules clear, neatly arranged hepatocytes, and no obvious degeneration, necrosis, inammatory cell inltration or congestion of hepatic sinusoids. The hepatocytes from the model group (Fig. 7l), on the other hand, exhibited clear signs of liver damage. Signicant congestions and hemorrhage was visible in the sinusoids of the acinar III zone (black arrow), and a large amount of hepatocyte steatosis with dense vacuoles in the cytoplasm (blue arrow) were observed. Some hepatocytes with dissolved and disappeared nuclei were necrotic (green arrow). Slight inammatory inltration and individual inammatory foci in the liver lobules could be seen (yellow arrow). Thus, these H&E staining results are in good agreement with the results of in vitro and in vivo uorescence imaging using O 2 c À and ONOO À as biomarkers.

Conclusions
In conclusion, we have established LW-OTf as the rst molecular uorescent probe for the in situ imaging of RNS and ROS associated with drug-induced liver injury in living cells and mice. LW-OTf was able to detect O 2 c À and ONOO À via nearinfrared uorescence and two-photon uorescence, respectively. We believe that the molecular design of LW-OTf can be generalized for dual imaging of other biomarkers (e.g. alkaline phosphatase) 63 and ONOO À in DILI by simply changing the protecting group on the NIRF signaling moiety. In principle, the detection of RNS and ROS in real time with uorescence imaging agents could signicantly help guide the understanding of ROS-and RNS-related diseases and potentially contribute to the development of new approaches for the treatment of DILI.

Conflicts of interest
There are no conicts to declare.