Dual-biomarker-triggered fluorescence probes for differentiating cancer cells and revealing synergistic antioxidant effects under oxidative stress

Dual-biomarker-triggered fluorescent probes were developed for simultaneous detection of the two biomarkers H2S and hNQO1 in cancer cells.


Introduction
Cancer, one of the most life-threating diseases, is characterized as uncontrolled growth and division of normal cells beyond their natural boundaries. The mortality of cancer remains high, which is mainly due to metastasis of primary cancer tumors. 1 The early stages of cancer development carry the maximum potential for therapeutic interventions, and the survival rate of certain cancers can be signicantly improved with early diagnosis and treatment. 2 Cancer biomarkers are endogenous molecules that are differentially expressed in cancer cells relative to their normal counterparts. Altered levels of such biomarkers can be measured to establish a correlation with the disease process and are useful for cancer diagnosis and therapy. 3 Furthermore, the simultaneous detection of multiple biomarkers can signicantly increase diagnostic accuracy. 4 Recent research has demonstrated that hydrogen sulde (H 2 S) and human NAD(P)H:quinine oxidoreductase 1 (hNQO1, EC 1.6.99.2) are potential biomarkers in certain cancer biology, which suggests that uorescent probes that detect these two species simultaneously would be of signicant utility.
As the third endogenous gasotransmitter, H 2 S is enzymatically generated from cystathionine g-lyase (CSE), cystathionineb-synthase (CBS) and 3-mercaptopyruvate sulfurtransferase (3-MPST)/cysteine aminotransferase (CAT). 5 H 2 S plays important roles in various biological and pathological progress, 6 and misregulation of endogenous H 2 S is associated with numerous diseases. 7 Specially, low levels of endogenous H 2 S appear to exhibit pro-cancer effects, whereas higher concentrations of H 2 S can lead to mitochondrial inhibition and cell death. 8 We note that some cancer cells, such as ovarian and colorectal cancer cell lines, exhibit increased H 2 S production. 9 This increased H 2 S may be useful for cell growth and proliferation due to H 2 Sinduced angiogenesis. 9c hNQO1 is a FAD-dependent avoprotein that catalyzes the obligatory 2-electron reduction of quinones to hydroquinones and provides versatile cytoprotection with multiple functions. 10 Levels of this reductase are elevated in a number of cancer types, including non-small cell lung cancer, colon cancer, liver cancer and breast cancer, 11 when compared to the surrounding normal tissue, making it an important cancer biomarker as well as an activator for some anticancer drugs. 12 In addition to their roles as potential cancer biomarkers, both H 2 S and hNQO1 are also vital participants in cellular redox homeostasis. H 2 S is recognized as a potential antioxidant, 13 can reduce disulde bonds, and can react with various reactive oxygen and nitrogen species. For example, Chang et. al. reported that vascular endothelial growth factor (VEGF)triggered H 2 S production is dependent on NADPH oxidasederived H 2 O 2 . 14a More recently, we as well as other groups found that endogenous H 2 S can be generated upon simulation of H 2 O 2 through the glutathionylation and subsequent increased activity of CBS in HEK 293 cells. 14b,c In addition, hNQO1 can reduce ubiquinone and vitamin E quinone to their active antioxidant forms and can also reduce superoxide to protect cells during oxidative stress. 15 Furthermore, hNQO1 can be an intracellular source of NAD + , which can fuel the activity of sirtuins to inhibit mitochondrial reactive oxygen production. 16 Despite the importance of H 2 S and hNQO1 in these systems, the response of these two biomarkers to oxidative stress remains largely unknown. To this end, our goal was to rationally design uorescent probes for simultaneous detection of H 2 S and hNQO1 to provide new chemical tools for investigating their possible crosstalk in redox homeostasis.
Recent research has demonstrated that uorescence-based methods are highly suitable and sensitive for in situ and realtime visualization of biomolecules. 17 Numerous uorescent probes have been developed for the detection of hNQO1 or H 2 S in living systems. 18 Until now, however, none of these probes allows for the simultaneous detection of the chemical (H 2 S) and enzymatic (hNQO1) biomarkers via a single probe. To achieve this goal, we utilized a dual-reactive and dualquenching strategy, which we reasoned would improve the sensitivity and selectivity of the system. 19 Dual-activation probes have recently gained attention due to their ability to ne-tune responses by requiring the presence of two specic analytes. For example, Chang et. al. reported the dual-analyte detection of H 2 O 2 and caspase 8 activity during acute inammation in living mice. 20a Similar strategies have also been used for the successful dual-analyte detection of small molecules. 20b-d Herein, we report the rational design and preparation of H 2 S and hNQO1 dual-responsive uorescent probes 1 and 2, which were successfully utilized to differentiate cancer cells and reveal the synergistic antioxidant effects in response to the oxidative stress.

Results and discussion
Rational design of the dual-biomarker-triggered uorescence probes To enable the simultaneous detection of H 2 S and hNQO1, we installed two chemoselective trigger groups that respond to H 2 S and hNQO1, respectively, into one uorophore. Such dualactivity probes are superior to traditional single-analyte detection probes because they provide specic advantages, including: (1) avoiding inhomogeneous intracellular distribution from different probes; (2) providing an enhanced off-on response due to the dual-quenching effects; and (3) enable a simple method to investigate the cooperative relationship of the analytes.
To enable access to such dual-responsive probes, we made use of the trimethyl-lock containing quinone propionic acid (Q 3 PA) moiety reported by McCarley's group 18a as the triggering group for hNQO1. For the H 2 S detection motif, we utilized the thiolysis of NBD (7-nitro-1,2,3-benzoxadiazole) amines, 21 which has been utilized by our group as well as others for development of excellent H 2 S probes. Additionally, this H 2 S sensing motif has been used for different biological applications including tumor bioimaging in mice. 9c Therefore, we combined the Q 3 PA and NBD amine moieties onto coumarin and naphthalimide uorophores to access dual-responsive systems. The Q 3 PA moiety can switch off the uorescence of the uorophore by the photoinduced electron transfer (PET) effect, while the NBD part can quench the uorescence through the uorescence resonance energy transfer (FRET) effect. We expected that the uorescence of the coumarin and naphthalimide uorophores would be quenched efficiently from this dual-quenching strategy, and that only dual activation of both the Q 3 PA and NBD motifs would result in uorescence turn-on (Scheme 1).

Synthesis and optical properties of the probes
As outlined in Fig. 1A, the synthesis of probe 1 started from a formylation reaction to generate 3, which was treated with dimethyl malonate to form the coumarin derivative 4. Then, single-reactive probe 6 was synthesized from coupling 4-nitro-7piperazinobenzofurazan (NBD-PZ) and the hydrolysis product 5. Aer N-boc deprotection and further coupling with Q 3 PA, probe 1 was obtained with relative good overall yield. Probe 2 was Scheme 1 Schematic illustration of the design for a dual-biomarkertriggered fluorescent probe, which should only be activated by the synergistic chemical reaction with H 2 S and enzymatic reaction with hNQO1.
prepared from a simple four-step synthesis from commercial available reagents (Fig. 1B). 4-Bromo-1,8-naphthalic anhydride was reuxed with N-boc-ethylenediamine to produce 8, aer which the piperazinyl group was introduced through a nucleophilic substitution to form 9. Further reaction with NBD-Cl afforded 10, which was then deprotected and coupled with the Q 3 PA motif to provide probe 2 in good yield. All compounds were characterized by 1 H and 13 C{ 1 H} NMR spectroscopy as well as high-resolution mass spectrometry (HRMS) (see ESI †).
With the probes in hand, we examined the optical response of 1 toward H 2 S and hNQO1 in phosphate buffered saline (PBS, 50 mM, pH 7.4 containing 0.007% BSA, 100 mM NADH). As shown in Fig. S1, † 1 displayed two absorption maxima around 405 nm and 500 nm due to the coumarin and NBD amine moieties, respectively. Aer reaction with both H 2 S and hNQO1, new peaks appeared at 395 and 520 nm, which corresponded to the production of coumarin uorophore and NBD-SH, respectively. 19b Notably, 1 remained water-solubile at concentrations over 25 mM (Fig. S2 †). Prior to activation, 1 was essentially non-uorescent (F 1 ¼ 0.15%) due to the PET-FRET dual-quenching effect. Aer treatment with both hNQO1 (1 mg mL À1 ) and H 2 S (200 mM) for 2 h, a large increase in emission (220-fold) appeared at 465 nm ( Fig. 2A). When 1 was treated by H 2 S alone for 2 h, only a 34-fold uorescence enhancement was observed (Fig. 2B), which was far lower than the response from hNQO1 and H 2 S together. When 1 was treated with hNQO1 alone for 2 h, the emission enhancement was negligible (2-fold) (Fig. 2C), implying a more efficient quenching from the NBD moiety in 1. Stability investigations showed that 1 was stable in PBS buffer in the absence of analytes (Fig. 2D). Taken together, probe 1 can be used to detect H 2 S and hNQO1 in tandem, whereas treatment with only one of the analytes resulted in a signicantly smaller response.
To achieve a more efficient single-and dual-quenching effect, we next assessed the uorescence response of 2 toward H 2 S and/or hNQO1. Emission spectra were also recorded in PBS buffer in the presence of NADH. As shown in Fig. 3A, 2 (F 2 ¼ 0.041%) was essentially non-uorescent due to the dual-   quenching effect, but a strong emission at 535 nm was observed when hNQO1 and H 2 S were added simultaneously. Aer 2 h, the uorescence increase at 535 nm was over 400-fold. Consistent with our design, treatment of 2 with hNQO1 or H 2 S alone resulted in only a negligible uorescence enhancement (3-or 7fold, Fig. 3B-D and S3 †). When compared with probe 1, we found that probe 2 not only resulted in a larger uorescence turn-on for combined H 2 S/hNQO1 activation, but also exhibited a lower single-analyte response. Because of these positive properties, we utilized probe 2 for subsequent bioimaging investigations.
Encouraged by the primary uorescence data, we further validated the chemistry associated with the sensing mechanism by using HRMS and UV-vis analysis. We rst conrmed the products of both the single-and dual-analyte reactions of 2 with H 2 S and/or hNQO1 with HRMS ( Fig. 4 and S4  respectively. We did not observe the cross reaction sideproducts (e.g. hNQO1-triggered 13 or H 2 S-triggered 12) in the MS spectra. We next performed UV-vis experiments to further probe the reaction mechanism. As shown in Fig. S5A, † the absorption spectrum of 2 displayed two maximum absorbance peaks near 350 and 500 nm. Aer treatment with H 2 S and hNQO1, both of these peaks disappeared and were replaced by peaks at 400 and 520 nm, which corresponded to the uorophore and NBD-SH, respectively. When H 2 S alone was added, new peaks at 400 and 520 nm were also observed ( Fig. S5B †). Furthermore, there was an obvious overlap between the absorbance prole of NBD-PZ and the emission prole of 11, indicating an intramolecular FRET effect in probe 2 (Fig. S5C †). When 2 was treated by hNQO1 alone, the absorbance peak at 500 nm increased (Fig. S5D †), implying that the PET process was abolished because the PET effect should result in small changes in absorbance spectra. 22 In addition, probe 2 maintained water solubility of over 20 mM under the experimental conditions (Fig. S6 †).
To gain more detailed information about the sensitivity of the dual-responsive probe, we incubated 2 with different levels of hNQO1 and H 2 S for 2 h, aer which the emission proles were measured. Probe 2 was rst treated with different concentrations of H 2 S (0-200 mM) in the presence of hNQO1 (1 mg mL À1 ). As shown in Fig. 5A and B, the emission at 535 nm was linearly related to the concentrations of H 2 S from 0 to 75 mM. When added to 1 mg mL À1 hNQO1, a 10 mM H 2 S solution resulted in a 46-fold uorescence response. Similarly, we treated 2 with various levels of hNQO1 (0.2-1 mg mL À1 ) in the presence of a constant H 2 S concentration (50 mM), and observed a uorescence enhancement of 180-fold (Fig. 5C).
One major requirement for a uorescent probe is that it must exhibit a response toward the targeted analytes but not for other competing species. In order to conrm that the turn-on response of 2 was selectively caused by the dual activation of hNQO1 and H 2 S, probe 2 was incubated with different reactive sulfur species (SO 3 2À and S 2 O 3 2À ), biothiols (Cys, Hcy and GSH) and reactive oxygen species (H 2 O 2 and HClO) in the presence of hNQO1 or H 2 S. As shown in Fig. 5D, only the co-incubation of hNQO1 and biothiols could trigger a very slight uorescence response (<10-fold, lanes 10-12), which was signicantly lower than the response triggered by hNQO1 and H 2 S (>400-fold, lane  15). No uorescence increase was observed when H 2 O 2 or HClO was added (lanes 6-7 and 13-14). Furthermore, treatment of 2 with dicoumarol, an hNQO1 inhibitor, resulted in a slower reaction rate than the inhibitor-free controls, conrming the requirement of hNQO1 for probe activation (Fig. S7 †).

Differentiation of cancer cells using the probe 2
We rst evaluated the cytotoxicity of 2 in HT29 cells (human colorectal epithelial cancer cells) by using the methyl thiazolyl tetrazolium (MTT) assay. The results showed that aer 2 h of cellular internalization of 33 mM probe, more than 90% of the cells remained viable (Fig. S8 †), implying a low cytotoxicity of 2. The cytotoxicity of 2 was further studied in HEK293A cells (human embryonic kidney cells) by monitoring of adherent cell proliferation through the xCELLigence RTCA system (Fig. S9 †). Compound 2 did not show signicant cytotoxicity from 0-15 mM aer 24 h incubation, and therefore 10 mM of 2 was used for bioimaging experiments. To investigate whether 2 could be employed to distinguish different types of cancer cells, several cell types were chosen as model biological systems. Given the elevated levels of both H 2 S and hNQO1 in some colorectal cancer cells, HT29 and HCT116 cells (human colorectal epithelial cancer cell lines) as well as FHC cells (human normal colorectal epithelial cell line) were initially selected. 9c Then HepG2 cells (human liver cancer cells) with a high level of endogenous H 2 S and HeLa cells (human cervical cancer cells) with a low level of endogenous H 2 S were also introduced. 8 We assumed that only the 2-stained cells with relatively high endogenous levels of both H 2 S and hNQO1 would display signicant uorescence. Aligned with this expectation, the confocal uorescence images showed clearly differentiable responses from the selected cells (Fig. 6A). The uorescence intensity in HT29 and HepG2 cells was much stronger than that in other cell lines. The relative uorescence increases in HT29 and HepG2 cells were about 5.3 and 3.7 fold higher than that of other cells (Fig. 6B). The signicantly different uorescence observed in cancerous versus non-cancerous cells is consistent with the probe design and suggests that the probe is differentially activated in cancerous versus non-cancerous cells.
In control experiments for single biomarker detection, two single-analyte probes NIR-H 2 S (for H 2 S detection) 9c and NIR-hNQO1 (for hNQO1 detection) 23 developed by us were separately incubated with these cells (Fig. S10 †). As shown in Fig. S11, † when cells were treated with NIR-H 2 S, the HT29, HepG2 and HCT116 cells displayed a uorescence response, implying the existence of endogenous H 2 S in the cells. When cells were incubated with NIR-hNQO1, the observed uorescence from the HT29 and HepG2 cells was stronger than that from the other three cell lines (Fig. S12 †). The results indicated the relatively high endogenous levels of both H 2 S and hNQO1 in HT29 and HepG2 cells, which is consistent with the bioimaging results of probe 2.
To further conrm the dual-activation of 2 in cancer cells, we added aminooxyacetic acid (AOAA, 200 mM), which is an inhibitor of enzymatic H 2 S synthesis, and dicoumarol (100 mM), which is an hNQO1 inhibitor. For the inhibitor-treated groups, HT29 cells were pretreated with the inhibitor for 30 min, then incubated with 2 (10 mM) for 1 h, washed and imaged (Fig. S13 †). HT29 cells showed strong uorescence aer incubation with 2 alone for 1 h. In contrast, pretreatment of one or two inhibitors led to a signicant decrease in uorescence, and the observed uorescence intensity was about a half of that in the group without inhibitors (Fig. 6C). These results clearly demonstrated the dual H 2 S and hNQO1 requirement for 2.
Investigation of the crosstalk between H 2 S and hNQO1 under oxidative stress H 2 O 2 , a common ROS, was introduced as a stimulus to investigate the potential crosstalk between H 2 S and hNQO1 in cellular redox homeostasis. HeLa cells were selected as the model biological systems due to the relative low levels of the both endogenous biomarkers. The cells were stained by 2, washed and imaged. As displayed in Fig. 7, 2-stained HeLa cells exhibited very weak uorescence. However, a signicant uorescence response was observed when cells were co-incubated with 2 and H 2 O 2 (50, 100 or 200 mM) for 1 h. To further understand the results, the inhibitors AOAA and dicoumarol were also used for control experiments (Fig. S14 †). The H 2 O 2stimulated cells displayed a signicant uorescence decrease Relative fluorescence intensity of images from inhibitor-pretreated HT29 cells. N ¼ 3 fields of cells, error bars are means AE sd. *P < 0.05; **P < 0.01. For (B), the black * was relative to HT29 group, and the red * was relative to HepG2 group. when pretreated with one or both inhibitors. The relative emission (Fig. 8A) showed that the stimulation by H 2 O 2 could trigger about 3.9-fold uorescence enhancement, which was much higher than the inhibitor-pretreated control groups (about 1.8-fold). In addition, aer co-incubation with H 2 O 2 and 2, AOAA-pretreated cells were further treated with Na 2 S (150 mM) for 30 min, and a small increase in uorescence was observed (1.5 fold) when compared with the AOAA-pretreated control group. These data suggest that endogenous H 2 S and hNQO1 could be spontaneously generated in living cells when cells were suffering from acute oxidative stress caused by exogenous H 2 O 2 .
Based on current knowledge, hNQO1 is regulated by the Keap1 (Kelch-like ECH-associated protein 1)/Nrf2 (nuclear factor-erythroid 2-related factor 2)/ARE (antioxidant response elements) pathway. 10 Nrf2 protein levels can rapidly increase in response to ROS, triggering the expression of hNQO1 to inhibit the formation of free radicals. 15a,24 Meanwhile, elevated Nrf2 can increase the expression of glutathione reductase (GSR), which can reduce GSSG to GSH. 24d,e Such GSH can be involved in the Sglutathionylation of CBS under H 2 O 2 to produce CBS -SG , which would enable more efficient biosynthesis of endogenous H 2 S. 14b,c Thus, we propose that the synergistic antioxidant effect of H 2 S and hNQO1 for handling oxidative stress in living cells is possibly regulated by Nrf2, which can trigger the expression of hNQO1 directly and improve endogenous H 2 S levels indirectly through controlling GSH (Fig. 8B). Taken together, these results support a synergistic antioxidant effect under cellular oxidative stress.

Conclusions
In summary, dual-biomarker-triggered uorescent probes were developed for the simultaneous detection of two potential cancer biomarkers. Probe 1 could detect the two biomarkers with a slight uorescence response toward one biomarker (34fold turn-on) and a signicantly enhanced uorescence by dual activation (220-fold turn-on). By contrast, the uorescence of probe 2 was signicantly enhanced and showed a greater response for the dual-activation from H 2 S and hNQO1 (>400fold turn-on). Moreover, probe 2 exhibited high sensitivity, excellent selectivity and good biocompatibility, which enabled us to differentiate activation levels in HT29 and HepG2 cells from FHC, HCT116 and HeLa cells due to the notably different endogenous levels of H 2 S and hNQO1 in the cell lines. Importantly, using the probe 2, we revealed a synergistic antioxidant effect between H 2 S and hNQO1 in living cells in response to the oxidative stress. These results clearly demonstrate the strengths of this dual reporter system, including the signicant off-on response, ability to distinguish cancer cells with both cancer biomarkers, and ability to investigate the crosstalk of analytes.  We also note, however, potential limitations of this system. For example, the developed tools only provide information on the relative levels of the biomarkers in different cell lines rather than precise quantication measurements. In addition, the development of probes with longer wavelength emissions would be needed to translate these systems into more complex systems, such as animal studies. Based on these needs, we are currently working to develop related dual-responsive probes with emission in the near-infrared region for in vivo applications. Overall, our work has demonstrated the research potential of dual-responsive uorescent probes in cancer biology and intracellular redox homeostasis.

Conflicts of interest
The authors declare no competing nancial interests.