Sulfur mustard, sarin, soman, potassium cyanide, VX, Lewisite, and BZ are highly toxic chemical warfare agents and present an extreme hazard. All experimental procedures involving these substances must be conducted under strictly controlled safety conditions. Operators are required to wear full-face respirators and other necessary personal protective equipment.

Chemical warfare agents sulfur mustard, soman, sarin, potassium cyanide, VX, Lewisite, and BZ were obtained from the Institute of Chemical Defense (China). Unless otherwise specified, all reagents were of analytical grade and used directly without further purification. Nuclear Magnetic Resonance (NMR) spectra were acquired on a Bruker AV spectrometer (400/101 MHz) in DMSO-d₆ or CDCl₃. Chemical shifts are reported in parts per million (ppm) relative to tetramethylsilane (TMS) as the internal standard. High-resolution mass spectrometry (HRMS) data for TrioAlert and its reaction products with chemical warfare agents were obtained using an Agilent 6520 Q-TOF LC/MS spectrometer.

Compound 1 underwent a Wittig reaction with ethyl (triphenylphosphoranylidene)acetate in ethanol, yielding compound 2. The resulting coumarin derivative was then subjected to a Vilsmeier–Haack reaction using DMF and POCl₃, affording the key intermediate 3-formylcoumarin. Finally, condensation of 3-formylcoumarin with o-phenylenediamine in methanol, catalyzed by sodium bisulfite, successfully afforded NNP-1 in 66.3% yield. ¹H NMR (400 MHz, Chloroform-d) δ 11.28 (s, 1H), 8.93 (s, 1H), 7.78 (d, J = 6.1 Hz, 1H), 7.51 (d, J = 5.9 Hz, 1H), 7.46 (d, J = 8.9 Hz, 1H), 7.24–7.32 (m, 2H), 6.68 (dd, J = 8.9, 2.1 Hz, 1H), 6.56 (d, J = 1.9 Hz, 1H), 3.46 (q, J = 7.1 Hz, 4H), 1.26 (t, J = 7.1 Hz, 6H). ¹³C NMR (101 MHz, Chloroform-d) δ 162.10, 156.82, 151.90, 147.99, 143.41, 142.69, 133.55, 130.40, 122.77, 122.53, 118.74, 111.14, 110.08, 108.82, 108.03, 96.93, 45.08, 12.46. ESI-HRMS (m/z): [M + H]⁺ calculated for NNP-1 (C₂₀H₂₀N₃O₂, (M + H)⁺): 334.1550, found 334.1546.

Compound 6 was treated with 2-cyanomethylbenzimidazole in ethanol and pyridine under reflux overnight to undergo Knoevenagel condensation-cyclization, forming the coumarin core. The crude product was then directly refluxed in concentrated HCl for 3 h to accomplish hydrolysis and aromatization, successfully afforded NNP-2 in 72.5% yield. ¹H NMR (600 MHz, DMSO-d₆) δ 12.22 (s, 1H), 8.79 (s, 1H), 7.67–7.56 (m, 2H), 7.29 (s, 1H), 7.20–7.12 (m, 2H), 3.35–3.31 (m, 4H), 2.81 (t, J = 6.4 Hz, 2H), 2.75 (t, J = 6.2 Hz, 2H), 1.95–1.88 (m, 4H). ¹³C NMR (151 MHz, DMSO-d₆) δ 160.84, 151.85, 148.01, 147.61, 143.47, 135.08, 130.12, 126.98, 122.24, 122.12, 119.72, 118.22, 112.79, 108.32, 106.96, 105.48, 49.95, 49.43, 27.35, 21.13, 20.21, 20.19. ESI-HRMS (m/z): [M + H]⁺ calculated for NNP-2 (C₂₂H₂₀N₃O₂, (M + H)⁺): 358.1550, found 358.1565.

A mixture of compound 1 and (3-methoxypropyl) benzoxazol-2-ylacetate (10) in isopropanol was refluxed for 12 h to undergo Knoevenagel condensation-cyclization, affording the 3-(2-benzoxazolyl)coumarin core. Upon completion, the solvent was evaporated under reduced pressure, and the crude product was purified by silica gel column chromatography to yield the title compound as a yellow solid ONP-1 in 53% yield. ¹H NMR (400 MHz, Chloroform-d) δ 8.62 (s, 1H), 7.79–7.85 (m, 1H), 7.57–7.64 (m, 1H), 7.43 (d, J = 8.9 Hz, 1H), 7.32–7.38 (m, 2H), 6.66 (dd, J = 8.9, 2.4 Hz, 1H), 6.55 (d, J = 2.3 Hz, 1H), 3.47 (q, J = 7.1 Hz, 4H), 1.26 (t, J = 7.1 Hz, 6H). ¹³C NMR (101 MHz, Chloroform-d) δ 159.89, 158.09, 157.69, 152.47, 150.39, 145.25, 142.11, 130.56, 124.96, 124.50, 120.12, 110.45, 109.73, 108.20, 106.50, 96.98, 45.14, 12.46. ESI-HRMS (m/z): [M + H]⁺ calculated for ONP-1 (C₂₀H₁₉N₂O₃, (M + H)⁺): 335.1390, found 335.1403.

A mixture of compound 6 and (3-methoxypropyl) benzoxazol-2-ylacetate (10) in isopropanol was refluxed for 12 h to undergo Knoevenagel condensation-cyclization, affording the vital intermediate. Upon completion, the solvent was evaporated under reduced pressure, and the crude product was purified by silica gel column chromatography to yield the title compound as a yellow solid ONP-2 in 33% yield. ¹H NMR (400 MHz, Chloroform-d) δ 8.50 (s, 1H), 7.77–7.84 (m, 1H), 7.56–7.62 (m, 1H), 7.30–7.37 (m, 2H), 6.99 (s, 1H), 3.28–3.37 (m, 4H), 2.94 (t, J = 6.4 Hz, 2H), 2.78 (t, J = 6.2 Hz, 2H), 1.94–2.05 (m, 4H). ¹³C NMR (101 MHz, Chloroform-d) δ 160.39, 158.37, 152.73, 150.36, 148.08, 145.33, 142.22, 126.49, 124.68, 124.36, 119.95, 119.39, 110.40, 108.07, 106.04, 105.11, 50.24, 49.86, 27.47, 21.17, 20.21, 20.18. ESI-HRMS (m/z): [M + H]⁺ calculated for ONP-2 (C₂₂H₁₉N₂O₃, (M + H)⁺): 359.1390, found 359.1383.

A solution of compound 13 (10 g, 60.6 mmol) and dimethyl malonate (9.2 g, 69.7 mmol) was degassed by purging with N₂ for 15 min. The mixture was then heated to 180°C and stirred for 1 h. After cooling to room temperature, the crude product was purified by flash column chromatography to afford compound 14 as an orange solid in 48% yield.

Phosphorus oxychloride (POCl₃, 1.2 mL) was added dropwise to DMF (5 mL) at 0°C. The mixture was stirred at 50°C for 30 min. A solution of compound 14 (4 g, 21.2 mmol) in DMF (80 mL) was added dropwise, and stirring was continued at 50°C for 3 h. After stirring overnight at room temperature, the reaction mixture was carefully poured onto ice-water (200 mL). The resulting precipitate was collected by filtration, washed thoroughly with water (3 × 50 mL), and dried under vacuum. The solid was then stirred in a 1:1 mixture of ethyl acetate (EA) and methanol (MeOH, 100 mL) at 50°C for 30 min, filtered, and dried to yield compound 15 as an orange solid in 45% yield.

A mixture of compound 15 (0.244 g, 1.124 mmol), compound 16 (0.15 g, 0.936 mmol), and NaOH (0.019 g, 0.468 mmol) was stirred at room temperature for 12 h. The precipitate was collected by filtration, washed with methyl tert-butyl ether (MTBE, 3 × 20 mL), and dried. The crude solid was suspended in methanol (20 mL), stirred for 30 min, filtered, washed with MTBE (3 × 10 mL), and dried. Purification by preparative HPLC afforded TrioAlert as a red-brown solid in 8.9% yield. ¹H NMR (400 MHz, DMSO-d₆) δ 13.40 (s, 1H), 8.51 (s, 1H), 8.40 (d, J = 15.6 Hz, 1H), 7.86 (t, J = 7.2 Hz, 1H), 7.80 (s, 1H), 7.58 (d, J = 8.8 Hz, 2H), 7.37–7.41 (m, 1H), 7.33 (t, J = 7.6 Hz, 1H), 6.83 (dd, J = 2.4, 8.8 Hz, 1H), 6.63 (d, J = 2.4 Hz, 1H), 3.10 (s, 6H). ¹³C NMR (101 MHz, DMSO-d₆) δ 181.56, 160.15, 156.66, 154.79, 149.80, 147.47, 143.60, 140.22, 135.24, 131.13, 126.05, 123.53, 121.56, 113.93, 113.27, 110.73, 109.26, 97.23. ESI-HRMS (m/z): [M + H]⁺ calculated for TrioAlert (C₂₁H₁₈N₃O₃, [M + H]⁺): 360.1343, found 360.1343.

A mixture of compound 13 (1 g, 6.057 mmol), ethyl acetoacetate (0.95 g, 7.302 mmol), and piperidine (1 mL) in ethanol (13 mL) was stirred at 90°C for 12 h. After cooling to room temperature, the precipitate was collected by filtration, washed with cold ethanol (2 × 10 mL), and dried under vacuum to yield compound 19 as a yellow solid in 92% yield.

A mixture of compound 19 (0.25 g, 1.081 mmol), compound 20 (0.174 g, 1.189 mmol), and NaOH (0.022 g, 0.540 mmol) was stirred at room temperature for 10 h. The reaction mixture was filtered, and the collected solid was washed with ethanol (3 × 10 mL) to give an intermediate (0.05 g, 87% purity by HPLC). The filtrate was concentrated under reduced pressure. The residue was triturated with a 1:5 mixture of methanol and MTBE (30 mL), and the resulting solid was collected by filtration (0.14 g). Combined crude material was purified by reversed-phase column chromatography. Fractions containing the pure product were concentrated. Trituration of the residue with methanol (10 mL) afforded additional pure product. The combined fractions yielded TrioAlert-B as a brown solid in 23% yield. ¹H NMR (400 MHz, DMSO-d₆) δ 8.68 (s, 1H), 8.41 (d, J = 16 Hz, 1H), 7.68–7.76 (m, 3H), 7.57 (d, J = 15.6 Hz, 1H), 7.35–7.38 (m, 2H), 6.87 (d, J = 8 Hz, 1H), 6.66 (s, 1H), 3.14 (s, 6H). ¹³C NMR (101 MHz, DMSO-d₆) δ 185.15, 160.40, 158.58, 155.91, 149.69, 148.24, 132.88, 132.32, 127.55, 124.87, 115.66, 115.39, 111.20, 108.80, 96.95. ESI-HRMS (m/z): [M + H]⁺ calculated for TrioAlert-B (C₂₁H₁₈N₃O₃, [M + H]⁺): 360.1343, found 360.1351.

All fluorescence spectra were measured at room temperature using a Hitachi F-2710 fluorescence spectrophotometer. The excitation and emission slit widths were set to 5/10 nm and the photomultiplier tube voltage was 400/700 V for NNP-1, NNP-2, ONP-1, ONP-2, TrioAlert, and TrioAlert-B. These sensors were dissolved in acetonitrile to prepare a 1 mM stock solution, which was then diluted with PBS buffer to a working concentration of 20 µM. Sulfur mustard, sarin, and cyanide were added to the solution. The resulting solution (500 µL) was transferred to a quartz cuvette, and the fluorescence spectra were recorded by a fluorescence spectrophotometer. For sulfur mustard and sarin, the excitation wavelength was 480 nm, and the emission spectrum was scanned from 510 to 700 nm. For cyanide, the excitation wavelength was 380 nm, and the emission spectrum was scanned from 410 to 700 nm.

HaCaT cells were purchased from the Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. Cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with penicillin/streptomycin and 10% fetal bovine serum (FBS), and were incubated at 37°C in a humidified atmosphere containing 5% CO₂. For cellular experiments, HaCaT cells in the logarithmic growth phase were seeded into 96-well plates at a density of 10⁵ cells/mL. All operations were performed using aseptic techniques. After cell attachment, fresh medium containing TrioAlert (20 µM) was added, and the cells were incubated for 10 min. Cells were washed three times with PBS to remove excess TrioAlert. Then, different concentrations of sulfur mustard, cyanide or sarin were added to the cell culture medium and incubated for various durations. Fluorescence imaging was performed using an Agilent BioTek Cytation 5 cell imaging multimode reader equipped with a 20× objective lens. Images and data were analyzed using BioTek Gen5 software (Excitation wavelength Ex = 380/469 nm). All data are presented as the mean ± standard deviation (SD) (n = 5).

Six-week-old C57BL/6J mice were purchased from the Animal Center of Naval Medical University and acclimatized for one week prior to experiments under standard housing conditions at a temperature of 22 ± 2°C and 12 h light/dark cycle with free access to standard laboratory chow and purified water. For acute toxicity experiments, mice were randomly divided into two groups (n = 3 per group). The control group: intraperitoneal (i.p.) injection of acetonitrile-PBS; The TrioAlert group: i.p. injection of TrioAlert (100 mg/kg) formulated in acetonitrile-PBS. After 24 h, all mice were anesthetized by intraperitoneal injection of 1.25% 2,2,2-tribromoethanol (0.2 mL/10 g). Then mice were euthanized, and the tissues were harvested immediately for H&E or Nissl staining, including brain, lung, intestine, spleen, liver, and kidney.

For animal fluorescence imaging experiments, mice were anesthetized by intraperitoneal injection of 2,2,2-tribromoethanol prior to the experiment, followed by depilation of the abdominal area. Mice were randomly divided into 5 groups (n = 5 per group): (1) Control group: mice received a topical application of acetonitrile-PBS (5 µL) only. (2) TrioAlert group: mice received a topical application of TrioAlert (5 µL, 50 µM) only. (3) TrioAlert + sulfur mustard group: mice were pretreated with a topical application of sulfur mustard (5 µL, 100 µM), followed by a topical application of TrioAlert (5 µL, 50 µM). (4) TrioAlert + sarin group: mice were pretreated with a topical application of sarin (5 µL, 100 µM), followed by a topical application of TrioAlert (5 µL, 50 µM). (5) TrioAlert + cyanide group: mice were pretreated with a topical application of cyanide (5 µL, 200 µM), followed by a topical application of TrioAlert (5 µL, 50 µM). Fluorescence imaging was performed using a Tanon Prime In Vivo Imaging System. Images and data were analyzed using Tanon Prime In Vivo Imaging System software. All data are presented as the mean ± SD (n = 5).

As previously reported²⁵, filter paper was immersed in an acetonitrile solution of TrioAlert (5 mM), sonicated for 40 min, and then soaked for 1 h. The filter paper was removed and air-dried to obtain TrioAlert-loaded filter paper. This paper was cut into small strips (2 cm × 1 cm) for chemical warfare agent detection. Following the reaction with the CWA, the TrioAlert test strip was scanned using a smartphone RGB color analysis application. The RGB values (R, G, B) at the central point were extracted for analysis. These digital values were first normalized to a 0–1 range and then linearized to compensate for the non-linearity (gamma correction) inherent in standard RGB color spaces, yielding linear RGB values (R_linear, G_linear, B_linear). The conversion from linear RGB to CIE 1931 XYZ tristimulus values was performed using a standard transformation matrix [M], which is specific to the color space of the capturing device (e.g., sRGB):

[X, Y, Z]^T = [M] · [R_linear, G_linear, B_linear]^T (1)

Finally, the XYZ values were converted to the CIE 1931 xyY colorimetric coordinates using the following equations:

x = X / (X + Y + Z) (2a)

y = Y / (X + Y + Z) (2b)

Y = Y (2c)

The resulting xyY coordinates provided a robust basis for the quantitative analysis of the test strip's color change.

All animal experimental protocols in this study were reviewed and approved by the Institutional Animal Care and Use Committee of Naval Medical University (NMUMREC-2021-013), and the approval certificate is available upon request. Animal experiments were performed in accordance with the regulations and guidelines of the Animals Research Committee at Naval Medical University (SMMU, Licence No. 2011023). The animals received humane care throughout the procedures in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (publication no. 85 − 23, revised 1996).

To develop practical and reliable fluorescent sensors for blister agent sulfur mustard, G-type nerve agent sarin, and blood agent cyanide, the vital step is to construct a stable recognition unit specifically for sulfur mustard. Compared to the numerous recognition units available for nerve and blood agents, thione and thiol groups represent the only two commonly used recognition units for sulfur mustard.^{24, 25, 33} However, both of these recognition units are prone to oxidation in air, particularly under UV irradiation.³⁴ Although we previously developed several thione/thiol-based fluorescent sensors for sulfur mustard detection, subsequent practical application revealed that this susceptibility to oxidation significantly limits their utility.^35–37 Due to the weak electrophilicity and the absence of hydrogen-bond donor or acceptor groups in sulfur mustard, designing a suitable recognition unit for it remains challenging.¹ Consequently, we sought inspiration for designing a sulfur mustard recognition unit by examining how this agent binds to biomacromolecules in organisms. It is now widely accepted that blister agents can form covalent bonds with proteins and DNA.^{38, 39} The N7 position of guanine is the primary binding site for sulfur mustard on DNA.^{38, 39} Cysteine residues have long been considered the main binding sites for sulfur mustard on proteins.³⁹ However, recent studies indicate that histidine residues also exhibit strong binding affinity towards blister agents.^{40, 41} Intriguingly, the covalent binding of sulfur mustard to both guanine and histidine residues occurs specifically at their imidazole moieties. Thus, we propose that the imidazole moiety itself could serve as a potential recognition unit for sulfur mustard.

The G-type nerve agent sarin is a strong electrophile. Consequently, fluorescent sensors for its detection typically incorporate nucleophilic functional groups such as hydroxyl, amino, pyridyl, or piperazinyl.^{27, 28, 42} We reasoned that an imidazole group also can be employed as the recognition unit for sarin. Thus, two sensors with imidazole moieties NNP-1 and NNP-2 were designed and synthesized (Fig. 1A). We also synthesized two additional sensors, ONP-1 and ONP-2, to investigate the detection capabilities of a similar recognition unit oxazole toward sulfur mustard and sarin. As shown in Fig. 2A, the fluorescent emission of all four sensors significantly quenched upon contact with sarin. However, only NNP-1 and NNP-2, which contain the imidazole group, exhibited pronounced fluorescence changes after treatment with sulfur mustard. This result clearly demonstrates that the imidazole moiety uniquely functions as an effective recognition unit for both sulfur mustard and sarin.

Similar to nerve agents, various strategies have been reported for the detection of cyanide, which incorporate carbonyls, dicyanovinyls, and indolium/pyridinium.^{29, 30, 43} Inspired by these units, α,β-unsaturated ketone was used as a recognition unit for cyanide in this study. Taken together, we employed an imidazole group as the recognition unit for sulfur mustard and sarin, while α,β-unsaturated ketone served as the recognition unit for cyanide and the linker between imidazole moiety and the coumarin fluorophore. As shown in Fig. 1B, two sensors TrioAlert and TrioAlert-B were synthesized in a few steps. The synthesis details are provided in the Methods and Materials section.

Having synthesized the sensors, we first evaluated the spectral responses of TrioAlert toward sulfur mustard, sarin, and cyanide. As shown in Fig. 2B, upon addition of sulfur mustard (100 µM) or sarin (100 µM), significant quenching of the fluorescence emission at 560 nm was observed in acetonitrile-PBS (1:9). Concurrently, the color of the reaction solution changed from yellow to red (Fig. 2E). The fluorescence quenching efficiency of sarin was slightly higher than that of sulfur mustard. The fluorescence quenching of TrioAlert exhibited an excellent linear relationship with the concentrations of sulfur mustard and sarin (Fig. 2C). In contrast to sulfur mustard and sarin, the blood agent cyanide (200 µM) caused a blue shift in the fluorescence emission of TrioAlert from 560 nm to 442 nm (Fig. 2B). These spectral changes were accompanied by a solution color change from yellow to colorless (Fig. 2E). As the cyanide concentration increased, the fluorescence intensity at 442 nm progressively intensified. Cyanide (200 µM) increased the fluorescence intensity at 442 nm from 700 to about 5500. We also explored the reaction of TrioAlert-B with sulfur mustard, sarin, and cyanide. TrioAlert-B itself exhibits a stronger fluorescence emission at 576 nm than TrioAlert in acetonitrile-PBS (1:9). However, the addition of high concentrations (200 µM) of sulfur mustard, sarin, and cyanide all resulted in no significant fluorescence or color change in the TrioAlert-B solution (Fig. 2D and F). Taken together, TrioAlert can react with three high-threat chemical warfare agents, sulfur mustard, sarin, and cyanide and produced distinct color and fluorescence changes.

We further investigated the recognition mechanisms of the sensor TrioAlert toward sulfur mustard, sarin, and cyanide. Since high-resolution mass spectrometry (HRMS) analysis of the reaction products had to be conducted in a laboratory on a different floor, and considering the safety risks associated with transferring real chemical warfare agent samples, simulants CEES (2-chloroethyl ethyl sulfide, as a simulant for sulfur mustard) and DCP (diethyl chlorophosphate, as a simulant for sarin) were used for HRMS characterization. In fact, due to the general lack of access to real chemical warfare agents, or the absence of authorization to handle them in most research laboratories, CEES and DCP are widely employed to evaluate the detection performance and investigate the reaction mechanisms of sensors.^{31, 44, 45} Fluorescence experiments confirmed that the fluorescence spectra of TrioAlert after reaction with CEES and DCP showed no significant differences from those obtained with same concentrations of sulfur mustard and sarin (Fig. S2), supporting the applicability of the simulants. Given the safety concerns regarding sample transfer and the relatively well-understood reaction mechanisms between TrioAlert and sulfur mustard/sarin, we also adopted CEES and DCP for studying the sensor’s reaction pathways in this section.

For sulfur mustard and sarin, some literature reports have detailed their reactions with histidine in biological systems.^46–48 The N1 and N3 positions of histidine in TrioAlert act as nucleophiles, attacking the sulfonium ion intermediate formed by CEES. This reaction forms an adduct, TrioAlert-SM, which quenches the fluorescence of the sensor solution (Fig. 2B and S2A). DCP features an electrophilic phosphonyl group that is susceptible to nucleophilic attack by the nitrogen atom of the benzimidazole group in TrioAlert, leading to the formation of an adduct, TrioAlert-DCP. In cyanide detection, CN⁻ undergoes a nucleophilic addition reaction with the α,β-unsaturated ketone structure of TrioAlert, leading to cleavage of the carbon-carbon double bond and formation of a new C-CN bond.⁴⁹ This results in the disruption of the TrioAlert conjugated system, restricted intramolecular rotation, and weakened intramolecular charge transfer (ICT) process, ultimately causing a blue shift in the emission wavelength (Fig. 2B). The reaction products of TrioAlert with CEES, DCP, and CN⁻ were all confirmed by high-resolution mass spectrometry (see Supporting Information), validating the reaction pathway illustrated in Fig. 3A.

To gain deeper insight into the fluorescence changes of TrioAlert upon reaction with blister agents, nerve agents, and blood agents, we performed density functional theory (DFT) calculations for TrioAlert, TrioAlert-SM, TrioAlert-NA, and TrioAlert-CN. Figure 3B displays the optimized molecular structures at the PBE0/6-31G(d,p) level of theory, and Fig. 3C presents the computed electronic transition energies of TrioAlert and its reaction products, TrioAlert-SM, TrioAlert-NA, and TrioAlert-CN. In the TrioAlert molecule, the benzimidazole and coumarin moieties exhibit favorable coplanarity, forming an effective π-conjugated system. This allows electrons to flow freely from the benzimidazole donor to the coumarin acceptor via the unsaturated ketone bridge, establishing an intramolecular charge transfer (ICT) framework. Upon nucleophilic substitution with CEES or DCP, the D-π-A architecture of the molecule remains intact, and the conjugated planarity is preserved. After reaction with CEES and DCP, the HOMO/LUMO energy gap remained largely unchanged, and no significant redshift or blueshift was observed in the fluorescence spectra, which is consistent with the spectral response results (Fig. 2B and C). The introduced sulfur and phosphorus atoms, being relatively heavy elements, can enhance intersystem crossing from the excited singlet state to the triplet state. This process competes with the radiative transition back to the ground state via photon emission, thereby depleting excited-state energy and leading to fluorescence quenching. The reaction between TrioAlert and CN⁻ disruptes of the TrioAlert conjugated system, restricted intramolecular rotation, and weakened ICT process, ultimately causing a blue shift in the emission wavelength (Fig. 2B). This process is accompanied by an increase in the HOMO–LUMO energy gap, which is consistent with the calculated electronic transition energies (Fig. 3C).

Next, the response time and LOD of TrioAlert for sulfur mustard, sarin, and cyanide were measured. As shown in Fig. 4A and S1B, the fluorescence intensity of TrioAlert solution remained constant over time while all three CWAs elicited obvious fluorescence/color changes within 30 s in different solvents. Among these three CWAs, sarin exhibited the fastest response rate with TrioAlert, achieving a complete response within 10 s (Fig. 4A). The response time of sulfur mustard and cyanide with TrioAlert is 60 s and 90 s, respectively (Fig. 4A and S1B). Meanwhile, the real-time fluorescence change of the TrioAlert solution upon reagent addition was recorded. After addition of sulfur mustard (5 µL, 100 µM) or sarin (5 µL, 100 µM) to the sensor solution, the fluorescence intensity of TrioAlert rapidly decreased under the UV radiation, which reached its minimum within 10 s. Upon contact with cyanide (5 µL, 200 µM) the orange fluorescence gradually decreased within 10 s, while blue fluorescence simultaneously increased. The blue fluorescence intensity reached the maximum at 40 s (Fig. 4B).

As shown in Fig. 4C, the fluorescence emission intensity of TrioAlert at 560 nm/442 nm displayed excellent linear relationships with the concentrations of sarin, sulfur mustard, or cyanide. According to LOD = 3σ/κ, the LODs of TrioAlert were 0.28 µM for sarin, 0.32 µM for sulfur mustard, and 0.76 µM for cyanide, respectively. These results demonstrate that TrioAlert exhibits a rapid response rate and low LOD for sulfur mustard, sarin, and cyanide.

To evaluate the selectivity of TrioAlert for sulfur mustard, sarin, and cyanide, TrioAlert was treated with other CWAs and some other potential interferents. As shown in Fig. 4D, sulfur mustard and sarin can quench the fluorescence emission of TrioAlert at 560 nm. In contrast, no significant fluorescence quenching was observed upon treatment with various other CWAs (e.g., Lewisite, Soman, VX, BZ) or common alkylating agents and solvents, including chloroethane, ethylene oxide, formaldehyde, diethylamine, and dimethyl sulfate. For cyanide, TrioAlert was treated with other relevant anions such as SCN⁻, OCN⁻, S₂O₃²⁻, HPO₄²⁻, NO₃⁻, HSO₃⁻, or neutral substances such as glutathione (GSH) and Triethylamine. All these analytes failed to induce significant fluorescence enhancement at 442 nm, while cyanide treatment resulted in an obvious fluorescence increase of TrioAlert solution. These results demonstrate the high selectivity of TrioAlert for sulfur mustard, sarin, and cyanide over the tested interferents.

To enable simpler and faster detection of these three CWAs, TrioAlert was employed to construct test strips (Fig. 5A). The TrioAlert-loaded test strips exhibited orange fluorescence under UV radiation. The test strips were then exposed to different concentrations of sulfur mustard, sarin, and cyanide. As shown in Fig. 5B-D, the fluorescence/color changes of the test paper can be observed by the naked eye. Furthermore, the RGB values obtained using a smartphone application showed strong linear correlations with the analyte concentrations: the Red value with sulfur mustard, the Blue value with sarin, and the Green value with cyanide (Fig. 5E-J).

To mitigate the influence of ambient brightness on color detection of the test strips, seven points were uniformly selected along the edge and center of the strips for detection and analysis. The RGB values were converted to CIE 1931 xy chromaticity coordinates (Fig. 5K). Since brightness (Y) was excluded and only the chromaticity (x, y) was relied upon, the method exhibits certain robustness across different smartphones and light sources. The results demonstrated strong linear correlations between the xy values and the concentrations of sulfur mustard, sarin, and cyanide, as shown in Fig. 5L-N. Therefore, the combined system of TrioAlert-loaded test strips and a smartphone equipped with an RGB color application provides a practical platform for the quantitative detection of sulfur mustard, sarin, and cyanide.

These results inspired us to explore whether TrioAlert can be used to image CWAs in complex biological samples. First, the cytotoxicity of TrioAlert was tested with CCK8 assays. As shown in Fig. S3, the sensor had almost no effect on cell viability. Next, we used TrioAlert to image sulfur mustard in living HaCaT cells. After incubation with TrioAlert (20 µM), the cells exhibited strong fluorescence in green channel and almost no fluorescence in blue channel (Fig. 6A). Upon addition of sulfur mustard (100 µM) to TrioAlert-loaded HaCaT cells, the green fluorescence of TrioAlert-loaded HaCaT cells significantly quenched while the fluorescence in the blue channel remained unchanged. The sulfur mustard-induced reduction of the green fluorescence in cells was completed within 4 min (Fig. 6B-D). Furthermore, the cells treated with lower concentrations of sulfur mustard (25 and 50 µM) exhibited stronger residual green fluorescence after 4 min (Fig. 6E and F) compared to cells treated with a higher concentration (100 µM). To further confirm this, TrioAlert-loaded cells were pretreated with reactive skin decontamination lotion (RSDL, 200 µM), a decontamination lotion developed by Defence Research and Development Canada that chemically inactivates CWAs.⁵⁰ Potassium 2,3-butanemonooxime is the main active ingredient of RSDL that can neutralise sulfur mustard and nerve agents.⁵¹ Pretreatment with RSDL largely abolished the sulfur mustard-induced fluorescence quenching (Fig. 6G). Sarin-treated cells exhibited similar fluorescence changes to sulfur mustard, but its quenching effect on the fluorescence of TrioAlert was more pronounced than that of sulfur mustard (Fig. 7). Next, the abilities of TrioAlert to visualize the cyanide were investigated. Upon contact with cyanide (200 µM), TrioAlert-loaded cells exhibited an obvious fluorescence transition from green to blue fluorescence (Fig. 8D). The fluorescence transition was completed within 6 min (Fig. 8C). The TrioAlert-loaded cells were also treated with different concentrations of cyanide (50 and 100 µM) for 6 min. As shown in Fig. 8D-F, the blue fluorescence of sensor-loaded cells enhanced with increasing concentrations of cyanide (0 µM, 50 µM, 100 µM and 200 µM) while the fluorescence intensity gradually decreased in green channel. Collectively, these results demonstrate that TrioAlert serves as an effective sensor for imaging sulfur mustard, sarin, and cyanide in living HaCaT cells.

To further evaluate its potential in medical diagnostics following exposure to sulfur mustard, sarin, and cyanide, we first investigated the acute toxicity of TrioAlert. As shown in Fig. 9A and B, the mice treated with TrioAlert (100 mg/kg) showed no significant signs of acute toxicity in histopathological and hematological analyses. Subsequently, TrioAlert was used to image sulfur mustard, sarin, and cyanide in living mice. Compared to acetonitrile-PBS-treated mice, living mice showed strong green fluorescence and no fluorescence in the blue channel when treated with TrioAlert alone (Fig. 9D). One group was cutaneously exposed to sulfur mustard (5 µL, 100 µM) and then treated with TrioAlert (5 µL, 50 µM) after 10 min. The green fluorescence of these mice was significantly weaker than that of TrioAlert-only-treated mice. Mice pretreated with sarin (5 µL, 100 µM) also showed a similar fluorescence quenching effect (Fig. 9F). In contrast to sarin and sulfur mustard, mice pretreated with cyanide (5 µL, 200 µM) displayed obvious fluorescence increase in the blue channel and weaker green fluorescence (Fig. 9G). These results demonstrate the low acute toxicity of TrioAlert and its successful application in detecting sulfur mustard, sarin, and cyanide in live mice.

Although TrioAlert is the first fluorescent sensor capable of simultaneously detecting three high-threat CWAs, its detection spectrum remains limited. Certain high-threat CWAs, such as V-series nerve agents and the blister agent Lewisite, cannot be detected. While TrioAlert should theoretically be capable of detecting A-series nerve agents like Novichok, experimental confirmation was not conducted due to the unavailability of Novichok agent sources. Selectivity represents the other primary challenge for TrioAlert. Although the combined use of ONP-1 enables the distinction between blister agent and nerve agent, TrioAlert alone can only detect the presence of nerve agents and sulfur mustard; it cannot differentiate between these two classes of agents. More importantly, strong acids can also decrease the fluorescence of TrioAlert. Despite limitations in the detection spectrum and selectivity of sensor TrioAlert, its development represents a meaningful step forward. As highlighted in the OPCW’s Practical Guide for Medical Management of Chemical Warfare Casualties, all portable detection/identification devices, regardless of the technology used, sometimes yield false positives and false negatives due to their sensitivity and selectivity. From the perspective of applying fluorescent sensors to CWA detection, the primary task at this stage is to design and synthesize a diverse range of fluorescent sensors targeting different CWAs, particularly those with relatively lower reactivity such as sulfur mustard. Only by developing a sufficiently diverse library of sensors can we, in the future, leverage artificial intelligence to assemble these sensors into sensor arrays, which might realize full-spectrum monitoring of all CWAs and even rapid on-site identification of unknown toxic agents.