Fluorescence umpolung enables light-up sensing of N-acetyltransferases and nerve agents

Intramolecular charge transfer (ICT) is a fundamental mechanism that enables the development of numerous fluorophores and probes for bioimaging and sensing. However, the electron-withdrawing targets (EWTs)-induced fluorescence quenching is a long-standing and unsolved issue in ICT fluorophores, and significantly limits the widespread applicability. Here we report a simple and generalizable structural-modification for completely overturning the intramolecular rotation driving energy, and thus fully reversing the ICT fluorophores’ quenching mode into light-up mode. Specifically, the insertion of an indazole unit into ICT scaffold can fully amplify the intramolecular rotation in donor-indazole-π-acceptor fluorophores (fluorescence OFF), whereas efficiently suppressing the rotation in their EWT-substituted system (fluorescence ON). This molecular strategy is generalizable, yielding a palette of chromophores with fluorescence umpolung that spans visible and near-infrared range. This strategy expands the bio-analytical toolboxes and allows exploiting ICT fluorophores for light-up sensing of EWTs including N-acetyltransferases and nerve agents.


Introduction
Intramolecular charge transfer (ICT)-based uorophores 1,2 , featured by donor-π-acceptor (D-π-A) scaffold, are essential research tools in biosensing [3][4][5][6][7] and bioimaging [8][9][10][11][12] . Their exceptionally high sensitivity to the electron disturbance and large Stokes shifts make these chromophores promising platforms for the construction of numerous high-performance uorescent dyes and probes [13][14][15][16][17] . In general, the substitution and/or interaction of the donor receptor with EWTs (which are widespread and found in living organisms and natural environments) can intrinsically suppress the ICT pathway, thereby undesirably quenching the uorescence (Fig. 1a) [18][19][20] . This quenching mode of ICT uorophores inevitably generates misleading information and severely restricts their applicability for accurate sensing/labeling. Therefore, it is essential to formulate a molecular design strategy that overcomes the essential issue of the EWTinduced uorescence quenching, which will pave a new pathway for the light-up and highly accurate analysis toolbox.
Herein, we report a simple and generalizable molecular engineering strategy -insertion of indazole building block into ICT chromophore to regulate the intramolecular rotation driving energy -for completely overturning the EWT-induced quenching mode into the light-up mode (Fig. 1b). Starting from D-π-A chromophore like a typical laser dye DCM (dicyanomethylene-4H-pyran) 21,22 , we further expand this uorescence umpolung strategy to other D-π-A featured chromophores that span the visible and nearinfrared (NIR) range. As demonstrated, we show that the electron density perturbation of indazole can lead to a mutational effect on intramolecular rotation driving energy (ΔE RDE ), ipping between positive (enhance rotation) and negative (suppress rotation). Speci cally, the straightforward insertion of an indazole building block into the D-π-A motif can (i) substantially increase the intramolecular rotation in Dindazole-π-A uorophores (ΔE RDE > 0), resulting in a rotation-induced dark state, whereas (ii) e ciently suppress the rotation with the incorporation of EWTs (ΔE RDE < 0), along with a signi cant uorescence enhancement. This uorescence umpolung strategy expands the understanding of ICT mechanism, and has for the rst time allowed us to elaborately design ICT probes for light-up sensing of EWTs, like Nacetyltransferases and nerve agents.

Results
Engineering ICT chromophores for uorescence umpolung Aiming to expand the utility of ICT uorophores along with tailoring their emission properties for highdelity bioimaging in vivo, our group has been engaged in a long-term research project investigating the use of high-performance donor-π-acceptor (D-π-A) uorescent dyes, such as DCM (dicyanomethylene-4Hpyran) [23][24][25] , QM (quinoline-malononitrile) 26,27 , and so on. To further extend the emission wavelength and optimize the biocompatibility, we grafted an indazole unit between the π-bridge and donor in the D-π-A motif because of indazole's biological and pharmacological activities 28 (Fig. 2, Supplementary Fig. 1-3). In the designed D-indazole-π-A uorophore named DCM-IN-NH 2 , the amino unit is employed as the donor, furan as the π-bridge, and DCM as the acceptor. To study the effects of the electron density perturbation on emission, we then substituted the electron-donating amino group with an electron-withdrawing tertbutoxycarbonyl unit (Boc) to obtain an EWT-indazole-π-A uorophore named DCM-IN-Boc. Unexpectedly, we observed that the DCM-IN-Boc uorophore with an electron-withdrawing Boc unit showed unprecedented strong NIR emission, thereby motivating us to more carefully study this uorescence umpolung phenomenon.
As mentioned above, the EWT-induced uorescence quenching is ubiquitous in ICT uorophores, such as DCM-NH 2 ( uorescence ON) and its acylate product DCM-Ac ( uorescence OFF) (Fig. 2a). However, upon the insertion of an additional indazole group into the D-π-A uorophore, the resulting DCM-IN-NH 2 (Dindazole-π-A type dye) shows unexpected very weak uorescence. After further attachment of the amino group with an electron-withdrawing Boc group, the corresponding DCM-IN-Boc (EWT-indazole-π-A type dye) shows a bright uorescence (120-fold enhancement) at around 700 nm (Fig. 2b). In this case, the insertion of indazole into DCM uorophore can quench the uorescence in the D-indazole-π-A featured system, but lighting-up the uorescence in the EWT-indazole-π-A motif. Clearly, this EWT-induced light-up emission is completely opposite to the EWT-induced quenching mode of general ICT uorophores and warranted further investigation.
Starting from the initial DCM-IN-NH 2 and DCM-IN-Boc, we engineered a series of uorescence umpolung chromophores using the following guidelines ( Fig. 2b and Supplementary Fig. 3): (i) starting with D-π-A uorophore as the motif; (ii) inserting an indazole building block between the π-bridge and donor. This strategy unlocks a great opportunity to rapidly establish a library of indazole-based uorophores for studying the regularity of uorescence umpolung. Speci cally, we employed DCM, malononitrile, or aldehyde group as the accepter (A), thiophene or furan as the π-bridge (π), 2-methyl-indazole or 1-methylindazole as the indazole building block (IN), amino group as the donor (D), and Boc or acetyl group as the EWT ( Fig. 2b and Supplementary Fig. 3). To our delight, all resulting D-indazole-π-A chromophores displayed extremely weak emission intensity, whereas their corresponding Boc-substituted products (EWTindazole-π-A chromophores) exhibited strong uorescence (from 620 to 700 nm) ( Fig. 2b and 2c).
Compiling these results together, we have successfully developed a simple and generalizable molecular engineering strategy -insertion of indazole building block into ICT chromophore -to achieve the uorescence umpolung that spans the visible and NIR range.

Reversing Intramolecular Rotation Driving Energy Enables Fluorescence Umpolung
To get a deeper understanding of this uorescence umpolung, it's critical to obtain the molecular geometries of these indazole-based chromophores. which further validates the planar conformation of EWT-indazole-π-A dyes in solution-state. Taken together, these single crystals and 2D-NOESY NMR analysis strongly con rm the twisted/planar conformation of D-indazole-π-A/EWT-indazole-π-A. More importantly, all the information leads us to suspect that the dark state of D-indazole-π-A dyes is due to the C-C bond rotations between indazole and the π-bridge, resulting in pronounced non-radiative decay losses (Fig. 3g).
To verify our hypothesis, we investigated the effect of viscosity on the uorescence of these chromophores. According to a general rule, when rotation is restricted in high-viscosity environments, the non-radiative deactivation is minimized, which results in a uorescence enhancement [29][30][31][32][33][34] . Indeed, the emission of DCM-IN-NH 2 is partly restored in a highly viscous environment (Fig. 3h), which is typical for molecular rotors. In contrast, DCM-IN-Boc shows almost no uorescent change with viscosity variation (Fig. 3i). Collectively, the viscosity sensitivity of DCM-IN-NH 2 and the viscosity insensitivity of DCM-IN-Boc help elucidate the uorescence umpolung of indazole-based uorophores: (i) When modi ed with EDG (electron-donating group, such as amino group), the obtained uorophores show obvious C-C bond rotations between indazole and the π-bridge, which leads to a uorescence dark state. (ii) On the other hand, when modi ed with EWT (such as Boc group), the C-C bond rotation is inhibited and the uorophores show bright emissions. Therefore, this uorescence umpolung of indazole-based chromophores could be attributed to the effects of the electron density perturbation on intramolecular rotation (Fig. 3g).
Furthermore, quantum chemical calculations were conducted by using model molecules DM-IN-NH 2 and DM-IN-Boc (Fig. 4). Their results strongly corroborate the above experimental data, and provide more details in the photoexcitation and deactivation process. Upon photoexcitation, a uorophore could be excited from the ground state to the locally excited (LE) state, and then experience a transition from the LE state to the circa 90 o twisted excited state 35 Fig. 5). These results con rm that the electron density perturbation of indazole leads to a mutational effect on the ΔE RDE , ipping between positive (enhance rotation) and negative (suppress rotation). It is thus concluded that (i) the uorescence quenching of DM-IN-NH 2 (D-indazole-π-A type dye) is largely related to the intramolecular rotation in the excited state, and (ii) the uorescence lighting-up of DM-IN-Boc (EWT-indazole-π-A type dye) is attributed to no rotation in its' photophysical process (Fig. 4a).
Overturning the ICT probes' quenching mode into light-up mode for sensing EWTs The above experiments have demonstrated that indazole-based uorophores keep silent in the Dindazole-π-A molecules, but emit brightly when substituted with an electron-withdrawing group. We thus hypothesized that this characteristic could be used for light-up (OFF-ON) sensing of EWTs. Upon encountering the corresponding EWT analytes, the transformation from D-indazole-π-A to EWT-indazole-π-A could dramatically light-up the uorescent signal. In this regard, various OFF-ON uorescent probes (named Lighter EW Trackers) could be built up via using our speci c indazole-based chromophores.
Detection of EWTs, especially for arylamine N-acetyltransferases (NATs) and nerve agents, has recently received growing attention in the elds of biological and environmental science. NATs are phase II metabolism enzymes that transfer an acetyl group from acetyl-CoA to aromatic amines and arylhydroxylamines 36 . The detection of NAT2 activity is signi cant for disease diagnosis and personalized therapy in a clinical setting. Unfortunately, although traditional ICT uorophores (such as DCM-NH 2 ) possess the large Stokes shift, they inevitably exhibit a turn-OFF response toward NAT2 ( Fig. 5a-5c) owing to the EWT-induced uorescence quenching. In contrast, our designed indazole-based probes could nely address this limitation (Fig. 5d). As shown in Fig. 5e, upon reaction with NAT2 mimic, Lighter EW Tracker (DCM-IN-NH 2 ) shows a blue-shift (from 525 to 500 nm) in the absorption spectrum.
Simultaneously, a signi cant uorescence enhancement (around 56-fold) is observed at 700 nm with a large Stokes shift (200 nm, Fig. 5f). Notably, both the emission wavelength, peak shape ( Supplementary  Fig. 1c), and high-resolution mass spectrum (HRMS, Supplementary Fig. 6) are coincident with our synthesized DCM-IN-Ac, which further con rms the generation of acylation product upon reaction with the NAT2 mimic. Overall, we have successfully developed an OFF-ON uorescence probe for light-up sensing of NAT2 based on our uorescence umpolung strategy.
Encouraged by the application in biosensing, we further explored the environmental monitoring applications of Lighter EW Tracker. Nerve agents (such as sarin, soman, phosgene, and so on) have gained infamy as chemical-warfare agents (CWAs) used in wars in undeveloped countries [37][38][39][40] . It's essential to analyze highly toxic CWAs and related chemicals in a rapid and precise manner. As known, the high toxicity of CWAs is due to their strong capability of nucleophilic attack. Therefore, we reasoned that the active amino group of Lighter EW Tracker could be utilized as the recognition site for light-up sensing of nerve agents.
Then, we studied the spectral response of Lighter EW Tracker toward the nerve-agent mimic diethyl chlorophosphate (DCP). As shown in Fig. 5h, the absorption band at 515 nm becomes gradually decreased and is replaced by a new peak at 470 nm. Simultaneously, the NIR emission becomes enhanced to an intensity that is around 12 times larger than that of the original solution (Fig. 5i). Moreover, the plot of the I 650 nm against the concentrations of DCP ranging from 0-6 µM display a good linear relationship (R 2 = 0.999, Supplementary Fig. 7). Hence, this linear curve allows for the convenient quantitative detection or tracing of DCP over this concentration range. In the HRMS of Lighter EW Tracker with DCP, the peak of DCM-IN-DCP is found at m/z 568.1739 ( Supplementary Fig. 8), which strongly supports that the nucleophilic substitution causes the generation of emissive DCM-IN-DCP (EWT-indazoleπ-A type dye). Consequently, Lighter EW Tracker enables the quantitative and light-up detection of nerve agents. All these results show that the indazole platform can provide a generalizable method for light-up sensing EWTs including NAT2 and nerve agents.
Light-up tracking of endogenous NAT2 in living cells and tissue homogenates After investigating the OFF-ON response characteristics of Lighter EW Tracker (DCM-IN-NH 2 ) as a NAT2 probe in vitro, we further explored the potential of the probe for live-cell imaging of NAT2 activation (Fig. 6). Cytotoxicity of Lighter EW Tracker was rst evaluated by the widely used MTT assay. As shown in Fig. 6d, when incubated with 2, 4, 8, 16, or 32 µM Lighter EW Tracker for 24 h, the cell viabilities are close to 100%, indicating the remarkable biocompatibility of the probe. Then, uorescence confocal microscope was used to image HepG2 and HeLa cells after incubation with Lighter EW Tracker. As well known, in humans, NAT2 is expressed only in speci c cells (such as HepG2 cells) 36 . As expected, much stronger NIR uorescence is detected in HepG2 cells than that of HeLa cells (Fig. 6a, 6b, and 6e). Indeed, this distinctly different uorescent intensity (p < 0.001) in HepG2 and HeLa cells is consistent with the biodistribution of NAT2. Then, co-staining studies con rm that the probes are mainly localized in the cytoplasm, such as the cell mitochondria ( Supplementary Fig. 9). Furthermore, the NIR uorescence in cells with quercetin (inhibitor of NAT2) is much weaker (p < 0.001) than that of the cells without quercetin (Fig. 6b, 6c, and 6e). All these results strongly support that Lighter EW Tracker could be used to speci cally detect endogenous NAT2 in cells with a remarkable lighting-up signal.
More convincing evidence was from the comparison of the time-dependent changes of the uorescence signal of DCM-NH 2 (turn-OFF probe) and Lighter EW Tracker (turn-ON probe). As shown in Fig. 6f and 6 g, DCM-NH 2 shows a bright NIR uorescent signal after cellular uptake within 0.5 h. Owing to the turn-OFF response toward NAT2, the uorescence of DCM-NH 2 becomes weaker with time, which inevitably results in misleading information, as the reduction in uorescence intensities could also be attributed to photobleaching, diffusion, and so on. In contrast, Lighter EW Tracker shows a weak uorescent signal within 0.5 h, indicating that the probe is non-uorescent initially with low background interference. As the incubation time elapses, Lighter EW Tracker is activated by endogenous NAT2, and the uorescence intensity increases gradually which reaches a maximum at 1.5 h (Fig. 6h and 6i). These results clearly demonstrate that Lighter EW Tracker can be used for real-time and light-up monitoring of cellular NAT2.
The cell imaging results encouraged us to further detect NAT2 in different tissue homogenates including heart, liver, spleen, lung, and kidney. Previous ndings suggest that human NAT2 expression is the highest in the liver but is expressed at functional levels in other tissues 41 . As expected, Lighter EW Tracker exhibits the highest uorescent intensity in livers than other tissues ( Fig. 6j and 6 k). These imaging results further highlight the potential of Lighter EW Tracker for the detection of NAT2 activity in a clinical setting, which will be very useful for personalized therapy and disease diagnosis.

Discussion
By employing a simple and generalizable structural modi cation strategy -insertion of the additional indazole group into ICT uorophores, we have successfully resolved the long-standing challenge of EWTs-induced uorescence quenching in ICT uorophores. This breakthrough is enabled by overturning the intramolecular rotation driving energy (ΔE RDE ), and thus allows the complete reversing of the traditional ICT uorophores' quenching mode into the light-up mode. Starting with classic D-π-A chromophore like a typical laser dye DCM, We have expanded this molecular engineering strategy to other D-π-A featured chromophores that span the visible and NIR range, and the corresponding chromophores with uorescence umpolung demonstrate the generalizability of our platform.
With the single crystal, 2D-NOESY NMR experiments, and quantum chemical calculations, we con rmed that the electron density perturbation could modulate the ΔE RDE with uorescence umpolung. The insertion of an indazole building block into the D-π-A motif plays a vital role in both the process from the ground state to the LE state, and the process from the LE state to the circa 90 o twisted excited state, and thus leading to the reversal ΔE RDE between positive (enhance rotation) and negative (suppress rotation).
The D-indazole-π-A uorophores display obvious intramolecular rotations between indazole and π-bridge (ΔE RDE > 0), thereby accounting for the dark state. On the other hand, the EWT-indazole-π-A uorophores show non-rotation (ΔE RDE < 0), and exhibit 120-fold emission enhancement. This uorescence umpolung strategy has for the rst time allowed for the construction of various probes that make a breakthrough to light-up detect EWTs. Based on the unique uorescent probe named as Lighter EW Trackers, we showed the quantitative and light-up detection of nerve agents, and real-time monitoring of endogenous NAT2 in living cells and tissue homogenates with high delity. We anticipate that our strategy of uorescence umpolung would greatly expand the bio-analytical toolboxes for both basic life science research and clinical applications, and altogether push the limits of biological imaging.

Methods
Synthesis methods for all compounds, characterization data are provided in the Supplementary Information.
Materials and general methods. Unless specially stated, all solvents and chemicals were purchased from commercial suppliers in analytical grade and used without further puri cation. The 1 H and 13 C NMR spectra were recorded on a Bruker AM 400 spectrometer using TMS as an internal standard. The highresolution mass spectrometry data were obtained with a Waters LCT Premier XE spectrometer. The singlecrystal data were obtained with a Bruker D8 Venture X-Ray Diffractometer. UV-vis absorption spectra were collected on a Varian Cary 500 spectrophotometer, and uorescence spectrum measurements were performed on a Varian Cary Eclipse uorescence spectrophotometer. Confocal uorescence images were taken on a Leica TCS SP8 (63 × oil lens).
Theoretical calculation details. All calculations were carried out using Gaussian 16 A 42 . Density functional theory (DFT) and time-dependent DFT (TD-DFT) were employed to investigate the uorescence (quenching) mechanism of all compounds. All structural optimizations in the ground and excited states were performed using M06-2X functional 43