Introduction

Activity-based molecular optical probes that produce a signal in response to biomarkers are promising tools for noninvasive, real-time, and accurate bioimaging1,2,3,4,5,6,7. Chemiluminescence (CL) molecular imaging, which eliminates the use of external excitation light and maximally reduces autofluorescence and light scattering, has evolved as a state-of-the-art imaging modality8,9,10. In previous studies, advances in the bioimaging of CL illuminators were limited and focused only on the visible (400 − 650 nm) region, and it was difficult to employ these tools in mammals owing to their low photon penetration11,12,13,14,15,16,17,18. Thus, molecular engineering approaches for developing luminous scaffolds in the near-infrared region [NIR, which is regarded as the tissue penetration optical window, usually further divided into NIR-I (650–950 nm) and NIR-II (950–1700 nm)] beyond 650 nm are important for mammalian imaging to enable superior spatial and temporal resolutions19,20,21,22,23,24. For the bathochromic emission and amplification of CL intensity in the NIR-I or NIR-II window, the following two versatile strategies can be used (Fig. 1A). The first strategy involves the use of a paired CL donor and fluorophore acceptor by means of resonance energy transfer process (indirect luminescence mode) such as the CL resonance energy transfer (CRET)25,26,27. The second strategy includes the extended conjugation of the original CL molecule skeleton (direct luminescence mode) such as the π-conjugation scaffold expansion or heteroatom substitution in the donor/acceptor group28,29,30. The significant difference between these two modes is whether the emitted photons directly originate from the CL molecule skeleton or indirectly from the nearby excited fluorophore via CRET.

Fig. 1: General strategy for constructing NIR-II CL probes and illustration of duplex in vivo imaging of CL-P.
figure 1

A Scheme of CRET-based CL emission (indirect mode) and conjugate-based CL emission (direct mode). B Design of probes and their maximum emission wavelengths presented using arrows. C Activation mechanism of CL-P in response to O2•− and illustration of duplex imaging of CL-P in APAP-induced liver injury.

Full size image

In the work of predecessors, Schaap’s dioxetane, a metastable CL illuminator caged with an analyte-specific mask, was subjected to tailored deprotection (phenoxide formation) for spontaneous chemiexcitation via a chemically initiated electron exchange luminescence mechanism to yield chemiexcited benzoate and consequently emit light31,32,33. In recent years, many CRET-based Schaap-like chemiluminophores have been developed for producing wavelength redshifts to detect various reactive small molecules and enzyme. For example, Shabat et al. and Lippert et al. independently developed several unimolecular NIR-I chemiluminophores that linked a dioxetane-based CL donor with a fluorophore acceptor via an intramolecular CRET process to produce a redshift in the CL wavelength34,35. The present group recently developed an H2S-activated NIR-II CL unimolecular probe, in which the longer-wavelength emissive fluorophore in proximity to the NIR-I CL donor could be illuminated through intramolecular CRET to produce a redshift in the maximum CL emission from 700 to 950 nm36. Although progress from a unimolecular NIR-I CL emitter to a unimolecular NIR-II CL emitter has advanced the diversity of biological applications, imaging penetration depth, and imaging contrast, challenges remain in this CRET-based luminescence mode. First, the CL intensity of CRET-based luminophore essentially depends on the energy or light obtained from the excited visible chemiluminophores rather than on the highly emissive intermediates (e.g., the benzoate ester generates from Schaap’s dioxetane) and consequently dissipate the emission brightness. Second, high acquisition barriers, such as number of steps, low yield, or some combination of these factors, increase the workload of chemists, further hindering access to target probes34,36. Finally, energy conversion efficiency is a primary limiting factor for CRET-based emitting intensity. For example, to obtain sufficient conversion efficiency, the spectral overlap and spatial distance between the paired donor and acceptor require careful considerations. Unfortunately, the current research on these aspects, particularly on the impact of distance on CRET efficiency, is very rare27,37. Therefore, exploiting direct Schaap-like chemiluminophores with enhanced brightness in the NIR region might unlock previously untapped superiority compared to luminescence below 650 nm.

According to the response and luminescence mechanism of Schaap’s dioxetane, the direct CL emission wavelength and intensity of these CL probes essentially depend on the emissive nature of the chemiexcited benzoate ester generated in its chemiexcitation process38,39,40,41. Consequently, several excellent studies developed methods to extend the π-conjugation of the benzoate ester scaffold. For instance, Shabat et al.29 and Pu et al.30 performed the direct conjugation of the electron-withdrawing group adjacent to Schaap’s dioxetane core and the additional adjustment of the atomic radius of the terminal acceptor groups, respectively. In this manner, the increase of delocalized π electrons via lengthening of the conjugated scaffolds might provide the possibility in controlling intramolecular charge transfer and reduce the energy gap to produce a redshift of emission wavelength of Schaap’s dioxetane-based chemiluminophores. Thus, the development of direct activatable NIR CL unimolecular probe with enhanced intensity could be further insighted the specific and unsolved biomarkers in deep tissue. However, the short half-life of CL emitter hampers its capacity to monitor in vivo metabolism or to track course of disease during a persistent period. If the chemiexcitation product from a flash-type chemiluminescent emission could be excited by laser to achieve fluorescent imaging, this would greatly facilitate real-time tracking of the metabolism and degradation of these chemiluminescent probes under FL mode, as well as enable long-term tracking of the disease progression. From the perspective of complementation, crosstalk-free duplex imaging by integrating NIR CL and NIR fluorescence (FL) channels could not only achieve precise disease detection through point-of-care feedback but also enable durative response tracking.

In this study, a modular and practical synthesis route was used to construct three direct NIR-II CL unimolecular skeletons [dubbed as CL-X3 (X = O, S, and Se)] with their maxima emission at 980 nm (CL-O3), 1025 nm (CL-Se3), and 1060 nm (CL-S3), respectively, based on Schaap’s dioxetane architecture. These probes contained three moieties, including a fluoride (F-)-activated Schaap’s dioxetane core, hexamethine chain, and dicyanomethylene-4H-benzopyran (DCMO) or its variant units that function as a CL donor, a π-conjugation bridge, and an electron-withdrawing acceptor, respectively (Fig. 1B). The extension of the π-conjugation and heteroatom substitution of acceptor led to the desired direct NIR-II CL probe CL-S3, which provided unique advantages in SBR (43.5) and imaging depth (up to 1.2 cm). The relevant calculation, including the energy gap and planarity of the chemiexcited benzoate ester, detailed the redshift mechanism and subtle differences in the maximum absorption/emission wavelength among three NIR-II chemiluminophores. Furthermore, the real-time duplex (NIR-II CL/FL) signals of unimolecular emitter CL-P were used to visualize superoxide anion (O2•−) in acetaminophen (APAP)-induced hepatotoxicity, and the superiority of NIR-II CL was confirmed after its comparison with NIR-II FL (Fig. 1C). Considering the potential advantage of the activated dual-channel NIR-II CL/FL imaging unimolecular platform, this study highlighted its possibility for integration in a duplex biomedical imaging platform.

Results

Design and synthesis of NIR CL probes

To synthesize direct NIR CL unimolecular skeletons based on Schaap’s dioxetane, Horner–Wittig reactions were integrated twice to prepare a conjugated quatra-methine aldehyde at the ortho position of the phenol group (Fig. 2A, compound 9). Then, Knoevenagel condensation was utilized to couple quatra-methine–aldehyde with DCMO or its variants dicyanomethylene-4H-benzothiopyran (DCMS) or dicyanomethylene-4H-benzoselenopyran (DCMSe). Lastly, these precursors underwent [2 + 2] cycloaddition to produce the corresponding F--response direct NIR-emitting CL probes.

Fig. 2: Synthesis of intermediates and target probes.
figure 2

A Synthesis route of compounds 6 and 9. B Synthesis route of NIR-I CL probes (CL-O1 and CL-O2). C Synthesis route of NIR-II CL probes (CL-O3, CL-S3, and CL-Se3).

Full size image

The detailed procedure consisted of three parts, including the synthesis of compounds 6 and 9 (Fig. 2A), synthesis of NIR-I CL probes (CL-O1 and CL-O2, Fig. 2B), and synthesis of NIR-II CL probes (CL-O3, CL-S3, and CL-Se3, Fig. 2C). The synthesis procedure and characterization of corresponding electron-withdrawing DCMX (X = O, S or Se) moieties were shown in the supporting information (Scheme. S1 and Figs. S1–6). First, compound 1 (Scheme. S2 and Figs. S7–11) was facilely obtained at the gram scale according to a previously reported synthesis procedure42, which was used to prepare the aldehyde-based compound 2 (compound 2-OTBS, in which the phenolic hydroxyl is protected, was used as an aldehyde to prepare the next compound, CL-O1). After multistep reactions, including the Horner–Wittig reaction, reduction of the ester group, and oxidation of alcohol, compounds 6 and 9 were obtained with excellent yields. To ensure the implementation of the oxidation reaction of compound 5, the phenolic hydroxyl group of the compound 3 was pre-protected with tert-butyldimethylsilyl chloride (TBSCl). With aldehyde-derivatives (2-OTBS, 6, and 9) and acceptor DCMX (X = O, S, or Se) in hand, Knoevenagel condensation was utilized to couple each other to produce the precursor of extended π-conjugation dioxetane-based skeletons, termed as O1, O2, O3, S3, and Se3. Despite differences in the heteroatom of the benzopyran portion of DCMO, DCMS, and DCMSe, the uniform synthesis of the corresponding precursors (O3, S3, and Se3) did not require adjustment. Via the oxidation of singlet oxygen, these precursors underwent [2 + 2] cycloaddition to produce the corresponding F--response NIR-I and NIR-II CL probes, which were termed as CL-O1, CL-O2, CL-O3, CL-S3, and CL-Se3. The results suggest that these simple and versatile synthesis routes have obvious advantages for the later design and molecular engineering of NIR-II CL probes. All intermediates and desired probes were assessed via 1H NMR, 13C NMR, and HRMS (Scheme. S3 and Figs. S12–38). Although two, four, or six methylene units bridged the donor and acceptor that can potentially lead to several isomers arising from the Z/E configuration of “C = C” double bond, a unique probe peak emerged from a mixture of olefin isomers based on 1H NMR. Therefore, the experimental results and relevant literature43 suggest that Knoevenagel condensation is thermodynamically controlled. While the HRMS spectra of CL-O1 and CL-O2 are almost entirely the target m/z peaks, some deprotected Schaap’s dioxetane species were observed in the HRMS spectra of CL-O3, CL-S3, and CL-Se3. This indicates that fresh stock of CL-O3, CL-S3, and CL-Se3 were needed to ensure the accuracy of the subsequent experiments (Figs. S30, S34, and S36).

Characterization of CL probes

Because these new NIR CL skeletons were reconstructed based on Schaap’s dioxetane, a similar response and luminescence mechanism of these structures was proposed before proceeding the relevant in vitro experiments. As shown in Fig. 3A, upon treatment with tetra-n-butylammonium fluoride (TBAF, an F- donor), the TBS-tailored mask was uncaged to release the unstable phenolate (I), which decomposed to generate benzoate ester (II) in an excited state. As the chemiexcited state of the benzoate ester derivative relaxed to its ground state, photons were released in concert. The relevant characterization, 1H NMR and HRMS of FL-O3 (Scheme. S4 and Figs. S39, 40) and HRMS of other benzoate esters, are provided in Figs. S41–44, confirming the reliability of the above mechanism.

Fig. 3: Characterization of NIR CL in vitro.
figure 3

A Proposed response and luminescence mechanism. B Absorbance spectra of CL-O1, CL-O2, FL-O1, and FL-O2 in DCM. C CL spectra of CL-O1 and CL-O2 in the presence of TBAF in DCM. D FL spectra of FL-O1 and FL-O2 in the presence of TBAF in DCM (the excitation wavelength of FL-O1 is 520 nm and that of FL-O2 is 550 nm). E Absorbance spectra of CL-O3, CL-S3, CL-Se3, FL-O3, FL-S3, and FL-Se3 in DCM. F CL spectra of CL-O3, CL-S3, and CL-Se3 in the presence of TBAF in DCM. G FL spectra of FL-O3, FL-S3, and FL-Se3 in the presence of TBAF in DCM (the excitation wavelength of FL-X3 is 808 nm). H CL kinetic profiles of CL-O3, CL-S3, and CL-Se3 in the presence of TBAF. I CL enhancements of CL-O3, CL-S3, and CL-Se3 after treatment with different kinds of ions. J CL intensities of CL-X3 (X = O, S, or Se) as a function of [F–] (0–200 μM). Data are the mean ± SD; n = 3 independent experiments.

Full size image

The absorption spectra of CL-O1 and CL-O2 in Fig. 3B were acquired before treatment with TBAF. Owing to the extension of the polymethine length from di-methine to tetra-methine, the maximum absorption peak of CL-O2 was slightly redshifted than that of CL-O1. Then, upon reacting with TBAF, the new absorption spectra of both FL-O1 and FL-O2 significantly more redshifted than its corresponding CL probes [Note: FL-OX (X = 1, 2, 3) is the post-TBAF treated version of CL-OX (X = 1,2,3); FL-X3 (X = O, S, Se) is the post-TBAF treated version of CL-X3 (X = O, S, Se)]. Meanwhile, the maximum absorption peak of FL-O2 was significantly longer than that of FL-O1. As expected, when the bridge of the polymethine chain was lengthened, the maximum F--activated CL emission peak of CL-O2 reached up to 845 nm, indicating a bathochromic shift of approximately 85 nm, compared with that of CL-O1 (Fig. 3C). The corresponding FL emission spectra of benzoate ester FL-O1 and FL-O2 were acquired under laser radiation and compared with its CL emission spectra (CL-O1 and CL-O2). The results clearly showed that the maximum emission wavelength of FL-O2 (853 nm) was significantly longer than that of FL-O1 (762 nm) (Fig. 3D) and that the maximum emission peaks of FL-O1 and FL-O2 were well matched with their corresponding CL emission peaks (CL-O1 and CL-O2). These results suggest that the CL spectrum of the Schaap-like emitter largely matched with the photoluminescence spectrum of its corresponding benzoate ester. For the NIR-II chemiluminophores CL-O3, CL-S3, and CL-Se3, their maximum absorption peaks exhibited a higher bathochromic shift than those of CL-O1 and CL-O2, with a further redshift in absorption peaks when O was replaced with S or Se (Fig. 3E).

Different from the tendency of the absorption peak of NIR-I CL probes, the absorption wavelength of FL-S3 (761 nm) exhibited a higher redshift than that of FL-Se3 (735 nm). Likewise, the maximum CL emission peaks of CL-S3 (1060 nm) and CL-Se3 (1025 nm) were significantly redshifted relative to those of CL-O3 (980 nm), but CL-Se3 exhibited a clear blueshift compared with CL-S3 (Fig. 3F). Thus, after F--response, CL-S3 enabled the maximum absorption/emission wavelength of these three skeletons. Of note, the corresponding sequence of NIR-II FL wavelengths (λem, FL-S3 > λem, FL-Se3 > λem, FL-O3) also followed the tendency of their NIR-II CL wavelengths (λem, CL-S3 > λem, CL-Se3 > λem, CL-O3) (Fig. 3G). Thus, these results indicated that subtle improvements, such as extended π-conjugation and heteroatom substitution of the electron-withdrawing group, could preferentially redshift the CL emission wavelengths of chemiluminophores. In addition, CL-O3 has the strongest CL intensity, followed by that of CL-S3 and CL-Se3, which correlates with the fluorescence quantum yield of its corresponding benzoate esters (Fig. S45 and Table S1).

It should be noted that the FL signal of the unresponsive probe is very weak, while the post-response signal of FL-O3 is in a turn-on state. Based on the changes in the electron-donating and electron-withdrawing abilities of the functional groups during the response process (These substituent properties are described by Hammett σ constants using quantum-chemically derived parameter)44, a possible explanation is proposed. For example, the functional groups, TBSO, in the CL-O3, have weak electron-donating abilities (as illustrated in Table S2, TBSO: σm: -0.110, σp: -0.039, Hansch π: 2.967), while the phenolic group in FL-O3 have strong electron-donating abilities (Phenolic: σm: -0.735, σp: -1.570, Hansch π: -3.116). This change in the electron-donating ability might enhance the intramolecular charge transfer process, thereby achieving a strong ‘on’ state in the FL emission. The kinetic profiles of these NIR-II emitters are shown in Fig. 3H. The NIR-II CL half-lives of CL-X3 (274 s for CL-O3, 205 s for CL-S3, and 156 s for CL-Se3) were sufficient for real-time in vivo imaging. The selectivity of F- was investigated based on two channels: NIR-II CL and NIR-II FL. As shown in Fig. 3I and Fig. S46A, all NIR-II CL probes showed a specific response to F- and exhibited a negligible signal in the presence of other ions. The NIR-II CL enhancements (Fig. 3I, 78.1-, 31.5-, and 17.8-fold for CL-O3, CL-S3, and CL-Se3, respectively) of these chemiluminophores were higher than NIR-II FL enhancements (Fig. S46A, 15.1-,9.5-, and 4.5-fold for FL-O3, FL-S3, and FL-Se3, respectively) after F- activation. This preferential signal enhancement was mainly attributed to the lower background of NIR-II CL than that of NIR-II FL. Upon response to F-, these NIR-II CL probes also held an excellent linear correlation between the NIR-II CL intensities and F- concentration from 0 to 200 μΜ (Fig. 3J), and a similar correlation was observed in the NIR-II FL channel (Fig. S46B). The above data demonstrated that these dioxetane-based scaffolds with extended π-conjugation and heteroatom substitution could efficiently enable NIR emission after F- activation and lead to a remarkable prospect in duplex NIR-II CL/FL emission.

Theoretical calculation

To elucidate the experimental absorption/emission wavelengths, the density functional theory (DFT) and time-dependent density functional theory (TD-DFT) were calculated based on PBE0/aug-cc-pvtz. Considering that the charge separation in the molecular orbitals of the ground and excited state is not significant, we have attempted to use the PBE0 general function, which is commonly used for local excitation45,46. The solvation model adopted in this work is Solvation Model based on Density (SMD) in the Gaussian 09 program and the chosen solvent is dichloromethane, the same as in the experiment47. The optimized geometry and electron density distribution of the highest occupied molecular orbitals (HOMO) and lowest unoccupied molecular orbitals (LUMO) of these NIR CL probes are showed in Fig. S47. Firstly, the electron cloud density of HOMO and LUMO were mainly delocalized on molecule skeletons except for the adamantane fragment, and a small fraction is concentrated on the electron-withdrawing group of DCMX in their LUMO orbits. Secondly, owing to the increase in methine units and the heteroatom radius, the energy band gaps of these molecules decrease sequentially, and the delocalization of the π-electrons in these molecules is further enhance. Furthermore, the intramolecular charge transfer process is closely related to the charge separation on the molecular skeleton. An increasement of conjugation system will decreases the charge transfer potential, and in turn improves the likelihood of charge transfer process. Thus, the above calculation results were in accordance with the redshifted absorption peaks of CL skeletons obtained from experimental measurements.

The optimized ground state (S0) and excited state (S1) geometries as well as the electron density distribution of the corresponding post-product benzoate esters are presented in Fig. 4A,B, and Fig. S48, respectively. The results showed that π-electrons on the molecular skeleton were dispersed and the excitation energy of these compounds gradually decreased with the increase in methane units and heteroatom radius (except for FL-S3 and FL-Se3). Exchange of heteroatom, such as O, S, and Se, in terminal group is a common and practical molecular engineering strategy towards bathochromically shift the absorption/emission wavelength of luminophore48,49. Contrary to the predictable red-shift of wavelength in the order of O, S, and Se, the detectable results in this work show a slight difference for S and Se due to the combined influence of atomic radius and the planarity of the entire molecule. Our theoretical calculations provided a plausible explanation based on the basic set of PBE0/aug-cc-pvtz (Fig. 4B). For example, the detectable emission wavelength of FL-S3 (λexp, em = 1060 nm) was more redshifted than that of FL-Se3 (λexp, em = 1025 nm), and the calculative emission wavelength of FL-S3 (λcal, em = 1018.73 nm) also exhibited a redshift compared with that of FL-Se3 (λcal, em = 1010.27 nm). A similar trend was observed in their absorption wavelengths (FL-S3, λexp, abs = 761 nm, λcal, abs = 715.77 nm; FL-Se3, λexp, abs = 735 nm, λcal, abs = 713.44 nm).

Fig. 4: Theoretical calculations on the investigation of the NIR CL probe and its corresponding post-products.
figure 4

A Structure of the benzoate esters (post-products). B Calculated adiabatic energies of the excited state and vertical absorption energies of the ground state (all optimizations are based on the PBE0/aug-cc-pvtz level, SMD). C Optimized S0 geometries (considering S0 as an example) and the selected distance of FL-O3. D Specific distance of corresponding benzoate esters. Calculated dihedral angle of products in the ground state E and excited state F.

Full size image

To investigate subtle anomalies in the experimental results of FL-S3 and FL-Se3, the distortion of the molecular planarity was further analyzed and compared. First, as the radius of the terminal heteroatom increases, the planarity of the six-membered ring based on a dicyano-skeleton undergoes gradual deformation (Fig. 4C, D and S49). This deformation is quantified by an increasement in the distance between the two ortho-carbon (purple arrow) and an increasement in the distance from the heteroatom to the para-carbon (cyan arrow). For example, in the ground state, the distance of the purple arrow has increased from 2.36 Å (FL-O3) to 2.86 Å (FL-Se3) and the distance of the cyan arrow has increased from 2.84 Å (FL-O3) to 3.23 Å (FL-Se3). Likewise, the expansion of the six-membered ring occurs in the excited state. The expansion also results in a noticeable change in the dihedral angle of skeleton. As shown in Fig. 4E, F, the ground states and excited states of these benzoate esters possessed almost planar geometry along the polymethine skeleton but with small dihedral angles in the terminal heterocycle. The calculated dihedral angle for the ground state (S0) between the terminal groups and the polymethine skeleton of FL-Se3 (24o) was larger than that of FL-S3 (18°) owing to different atomic radius in their heterocyclic cores. The calculated dihedral angle for the excited state (S1) of FL-Se3 (31o) was also significantly larger than that of FL-S3 (18°). Thus, to some extent, the terminal group of FL-Se3 is more distorted than those of FL-S3, leading to a smaller contribution to the conjugation of skeleton50. It is noteworthy that both of dihedral angle and radius in this work might influence the final outcomes. The dihedral angle may have a greater prestige in FL-S3 and FL-Se3, dominating the blue shift of FL-Se3. Between FL-O3 and FL-S3, even though the skeleton of FL-O3 is close to planarity, the increase in radius might lead to a greater extension in π-conjugation, which shown a superiority and determined the presentation in wavelength. Therefore, from the calculation perspective, the above data detailed the rationality and distinctiveness of the design strategies of π-conjugation extension and heteroatomic substitution.

Comparing the penetration depth and resolution of NIR-II CL and FL imaging

To verify the potential superiority of NIR-II CL for bioimaging, the tissue penetration depths of the NIR-II CL signal were investigated and compared with those of the NIR-II FL signal (excitation under 808 nm). As described in Fig. S50, photon–tissue interactions under external excitation in the FL mode included interface reflection (yellow), scattering (green), absorption (black), and autofluorescence (cyan), and all interactions dissipated signal intensity and generated interference noise. By contrast, owing to the elimination of excitation light, CL mode exhibited reduced photon–tissue interactions and low background signal. Encouraged by the above results, a tissue penetration experiment was performed in a biological tissue mimic (1% Intralipid in Fig. S51) via the NIR-II CL and NIR-II FL channels51,52,53,54.

A series of NIR-II CL probes were activated with TBAF in a small dish and then immediately covered with various thicknesses of the mimic tissue, consequently enabling real-time dual-channel imaging (Fig. 5A). When no mimic tissue was present over the activated probes, CL-S3 apparently produced the brightest NIR-II CL intensity and highest SBR (43.5) among these three probes (Fig. 5B,G). When covered with the different thicknesses of the mimic tissue, the detectable signal was distinctly attenuated in both NIR-II CL and NIR-II FL channels (Fig. 5B,C). In particular, the detection threshold of CL-S3 reached up to 1.2 cm, whereas that of CL-O3 and CL-Se3 was barely detectable at 0.9 and 0.6 cm, respectively. At a depth of 0.6 cm, the SBR of CL-S3 was as high as 13.5, which was 1.7- and 6.8-fold higher than that of CL-O3 (7.7) and CL-Se3 (2.2), respectively.

Fig. 5: Penetration depth and resolution of NIR-II CL/FL imaging.
figure 5

A Schematic illustration of tissue penetration (1% Intralipid) in both NIR-II FL/CL channels. B NIR-II CL imaging with different thicknesses of 1% Intralipid. C NIR-II FL imaging with different thicknesses of 1% Intralipid. D–F Cross-sectional intensity profiles of CL-O3, CL-S3, and CL-Se3 (green-dashed) in the NIR-II CL images in B at the depths of 0, 0.3, 0.6, 0.9, and 1.2 cm. G SBR for NIR-II CL images as a function of tissue depth in D–F. H–J Cross-sectional intensity profiles of FL-O3, FL-S3, and FL-Se3 (green-dashed) in the NIR-II FL images in C at the depths of 0, 0.3, 0.6, 0.9, and 1.2 cm. K SBR for NIR-II FL images as a function of tissue depth in H-J. Data are the mean ± SD; n = 3 independent experiments.

Full size image

Thus, among the three probes, CL-S3 yielded the maximum penetration depth and best imaging result. Although CL-O3 had the highest CL brightness, light before 1000 nm was uncollected because of the utilization of 1000 nm long-pass camera lens, resulting in a shallower penetration depth than that of CL-S3 (Fig. 5B, D, and G). Meanwhile, the activated NIR-II FL imaging was acquired under an 808-nm laser and its acquiring signal decreased with the increasing depth of the mimic tissue (Fig. 5C). First, FL-S3 achieved the highest SBR (6.5) when there was no mimic tissue overlay, with a maximum detectable tissue depth of 0.6 cm. Owing to a more centralized collection window, the SBR of FL-S3 (3.3) was higher than that of FL-O3 (2.4) and FL-Se3 (1.4) while their depths were up to 0.3 cm (Fig. 5K). As a result, NIR-II CL imaging had a significant advantage in terms of penetration depth and imaging quality compared with NIR-II FL imaging. Because the overall yield of CL-O3 was significantly higher than that of CL-S3, the skeleton of CL-O3 was preferred choosing for the subsequent in vivo experiments.

In vivo multiple imaging of O2 •− in APAP-induced hepatotoxicity

APAP is a common analgesic and antipyretic, and overdosed APAP will cause drug-induced acute liver failure55,56,57. As shown in Fig. 6A, the toxicity of APAP is primarily induced by a metabolite, N-acetyl-p-benzoquinone imine (path a), which rapidly dissipates glutathione (path b) and binds to mitochondrial membrane proteins, causing oxidative/nitrifying stress and mitochondrial damage (path c). Thus, substantial reactive oxygen species (such as O2•−, H2O2, and HClO) and other toxic reactive molecules are generated, leading to liver injury. To verify the superiority of NIR-II CL for in vivo imaging and execute the complementary dual-channel imaging, the real-time and longitudinal duplex detection (NIR-II CL/FL) of O2•− in an APAP-induced hepatotoxicity model was performed using CL-P (Fig. 6A). The CL-P was successfully synthesized using intermediate O3 (Scheme S5), and comprehensively characterized by 1HNMR, 13CNMR, and HRMS (Fig. S52–55). Meanwhile, the relevant characterizations of CL-P were performed, included characteristic spectra, selectivity, stability, kinetic profile, and biocompatibility (Fig. S56–57). These data indicated that CL-P produced an activated duplex NIR-II CL/FL signal with high selectivity and sensitivity upon response to O2•−.

Fig. 6: Activated duplex NIR-II FL/CL in vivo imaging.
figure 6

A Illustration of APAP-induced liver injury and schema of the CL-P activation process. (a. activated by the cytochrome P450 enzyme in cells. b reduced by small molecules (glutathione) in organisms. c. binding to mitochondrial membrane proteins). B Representative continuous NIR-II CL and FL images of living mice at different APAP dosages before and after intravenous CL-P injection. NIR-II CL (C) and NIR-II FL (D) signal intensities in the liver region at various APAP dosages (n = 3 independent mice). E NIR-II CL imaging (3 min) and NIR-II FL imaging (32 min) at 320 mg/kg APAP dosage selected from B. F Cross-sectional NIR-II CL and FL intensity profiles in line with E. G Ex vivo NIR-II CL imaging (2 min and 32 min) and NIR-II FL imaging (2 min and 32 min) at 320 mg/kg APAP dosage. H Normalized luminous intensity of different organs from F. Yellow and pink represent NIR-II CL and blue and cyan represent NIR-II FL. Unpaired two-tailed Student’s t-test was applied for data analysis of (H, Liver group versus other organs groups **P < 0.01, ***P < 0.001). Data are the mean ± SD; n = 3 independent experiments. Figure 6A adapted with permission from ref. 57. Copyright 2021 American Chemical Society.

Full size image

Various doses of APAP were first intraperitoneally injected into live, followed by the intravenous injection of CL-P at 5 h post-treatment of APAP. When the mice were pretreated with APAP at a hepatotoxic dosage (80 mg/kg), the maximal NIR-II CL intensity was observed at 1 min, which was 2.7-fold higher than that in the control group (Fig. 6B, C). Furthermore, its NIR-II CL intensity had sharply diminished at 3 min and disappeared completely at 8 min. Next, when the dosage of APAP was enhanced up to 320 mg/kg, the intensity of the NIR-II CL increased significantly with a maximum of 12.6 SBR and was still faintly detectable at 24 min. To explore the advantages of NIR-II CL imaging over visible CL imaging, we constructed a control probe, CL-540, which emitted a maximum CL intensity at 540 nm (Scheme S6 and Fig. S58-61). Compared to CL-P imaging, probe CL-540 exhibited substantial enrichment in the lung via tail vein injection, resulting in a more pronounced signal in the lung compared to the liver site (Fig. S62). In addition, the intensity of the visible CL signal acquired a maximum of 5.6 SBR in liver site (Fig. S62D). Therefore, it could be concluded that the reported CL-P enabled higher resolution imaging in NIR-II CL window compared to the visible CL window (CL-540) due to lower absorption and scattering of photons in tissue.

In contrast to a previous work36, in which the NIR-II FL signal was always “on” and lacked a response process, the NIR-II FL signal of CL-P was nearly blocked until it responded to O2•−. Compared to most reported dual-channel (CL/FL) activated probes, CL-P is one of the few unimolecular probes that enable two channels (CL/FL) simultaneously activated in the NIR-II region (Table S3). The NIR-II FL signal of CL-P at each dosage gradually increased as the reaction progressed and then climbed to the top, which contrasted with the tendency of the NIR-II CL signal. Thus, an activated and incremental NIR-II FL signal was also observed in the liver after treatment with different APAP dosages (Fig. 6B, D). Upon treatment with 320 mg/kg APAP, the detectable NIR-II FL signal in the liver area became stronger and reached a maximum at 32 min, with a 3.1-fold increase compared with that in the control group. Likewise, when a nephroprotective antioxidant (N-acetyl-l-cysteine (NAC)), an O2•− scavenger, was administered before CL-P treatment, the duplex luminescence signals were weakened to a background level. More importantly, the intensity profiles in the liver site (red dashed line in Fig. 6E) showed that NIR-II CL imaging exhibited a better imaging resolution, with a suitable width of 17.6 mm and 2.2-fold SBR over NIR-II FL.

The ex vivo NIR-II CL and FL imaging of the main organs (e.g., the liver, lung, heart, kidneys, and spleen) resected from an APAP-treated mouse (320 mg/kg) was performed (Fig. S63). As shown in Fig. 6G, H, the detectable NIR-II CL signal in the liver region was gradually weakened from 2 to 32 min, whereas the NIR-II FL was gradually increased. The difference in the imaging result verified that the NIR-II CL signal appeared to fade as the reaction proceeded and chemical energy dissipated and that the post-response product was gradually enriched in the liver site to enhance NIR-II FL intensity. The H&E staining of the liver tissue slice dissected from APAP-treated mice showed an obvious inflammatory and injury characteristic compared with that of the control and NAG-pretreated groups (Fig. S64). Meanwhile, the levels of alanine aminotransferase and aspartate aminotransferase were markedly elevated after APAP treatment. These results indicate that the real-time duplex imaging of O2•− in APAP-induced liver injury could be performed, demonstrating the superiority of NIR-II CL over NIR-II FL for in vivo imaging.

Discussion

In summary, a new class of direct NIR-II CL unimolecular probes with a maximum emission wavelength of up to 1060 nm was developed based on Schaap’s dioxetane using a simple and feasible synthesis route. Via a conjugated polymethine chain, these activated NIR-II CL emitters were feasibly coupled with a primitive Schaap’s dioxetane CL donor along with three acceptors (DCMO and its derivatives DCMS and DCMSe). After response, these activated chemiluminophores enabled bright NIR-II CL emission and generated extended π-conjugation benzoate esters, whose spectra of photoluminescence emission were essentially consistent with CL spectra of their parent. Meanwhile, the desired NIR-II CL emission provided clear advantages, including a specific response, continuous luminescence, and deep penetration depth (1.2 cm), and high SBR (43.5). The correlation of the tendency of wavelength shifts and luminescence intensity with extended π-conjugation skeleton and acceptor heteroatom radius was detailed in terms of theoretical calculations. More importantly, a duplex activated NIR-II CL/FL signal was constructed to detect O2•− in APAP-induced liver injury, enabling real-time and continuous signal collection in this unimolecular imaging platform and demonstrating the higher sensitivity and resolution of NIR-II CL imaging than that of NIR-II FL imaging. Overall, the facile construction of a direct NIR-II CL emission unimolecular skeleton not only opened an avenue toward the diversity of the NIR-II CL sensing platform but also provided a possibility for its integration in multiplex biomedical imaging.

Moreover, research on Schapp’s dioxetane precursor-based afterglow imaging has been exceptionally notable in recent years. For example, ultrasonically excited afterglow luminescence58,59,60,61 and X-ray excited afterglow luminescence6. Owing to evident light scattering and absorption in tissue, it is obstructive to execute light-induced afterglow (photo-afterglow) to generate a signal in deep-tissue3. Unlike light irradiation, other irradiation sources, including mechanical and non-radiative energy (Ultrasound)62,63 or electromagnetic and radiation radiative energy (X-ray)64, can overcome the above problems. To achieve deep-tissue afterglow luminescence based on ultrasound-induced or X-ray-induced Schaap’s dioxetane precursor platform, the most critical factor is the generation of sufficient singlet oxygen (1O2), which then oxidizes precursor to generate Schaap’s dioxetane and thus completing the afterglow process6,58,61. Of course, the latest instance involve Schaap’s dioxetane-based afterglow imaging induced by direct ultrasonic trigged chemical bond cleavage, but here we focus solely on the afterglow luminescence based on the precursor of Schapp’s dioxetane develpoed in this work65. Although the imaging apparatus may be more complex under ultrasonic or X-ray excitation modes, and some associated side effects (mechanical or ionization damage) with these methods, their advantages are significant. One major highlight is the increased tissue penetration depth compared to optical excitation. Given that Schaap’s dioxetane usually exists in a metastable state, if a relatively stable precursor is injected into a living body and then oxidized into dioxetane accordingly, the utilization efficiency of these probe can be further improved, enhancing its signal intensity. Thus, theoretically, by integrating these diret NIR-II CL skeleton with applicable sonosensitizers, or modifying the developed skeleton to directly produce adequate singlet oxygen under ultrasound or X-ray stimulation, our platform could potentially realize corresponding ultrasound-induced or X-ray-induced afterglow luminescence in the NIR-II window.

Methods

Animal handling

All the animal experimental tests were conducted according to the guidelines of the Regional Ethics Committee for Animal Experiments and the Care Regulations approved by the Institutional Animal Care and Use Committee of Fuzhou University. BALB/c female nude mice at 5–6 weeks old were purchased from the Shanghai Slack Laboratory Animal Center. The purchased mice were randomly grouped to different cages in specific-pathogen free (SPF) and standard environment (a 12/12 h diurnal alternation, at 20–25 °C temperature, and 45–55% humidity).

In vitro NIR CL and FL imaging

The abs spectra and NIR CL spectra of CL-O1, CL-O2, CL-O3, CL-S3, and CL-Se3 (30 μM) were acquired in the presence of TBAF (100 µM) in pure DCM. The NIR FL spectra were also obtained under the same conditions except with laser irradiation. The NIR signal (CL or FL) of these probes was obtained using two detectors (the detection range of red PMT is 400–900 nm and that of NIR PMT is 950–1700 nm) respectively. For the determination of chemiluminescence kinetic profiles. CL-X3 solutions (100 µM of CL-X3, containing 50% DMSO and 50% saline) were added to a black 96-well plate. The NIR-II CL intensities were continuously acquired after the addition of TBAF (100 µM) and plotted as a function of time. For in vitro selectivity studies, a 100 µM aqueous solution of CL-X3 containing 50% DMSO and 50% saline was treated with different kinds of aqueous ion solutions (200 µm). The NIR-II FL or NIR-II CL enhancement of CL-X3 was measured under an NIR-II imaging setup under optimized conditions (for NIR-II CL signal collection, the exposure time was 1200 ms, and for NIR-II FL signal collection, the exposure time was 50 ms under an 808-nm laser).

Theoretical calculation

All density functional theory (DFT) and time dependent (TD)-DFT calculations were performed using Gaussian 16 software, unless otherwise specified. For corresponding atomic coordinates, see Supplementary Data 1. In order to provide a better description of the real status of the probe molecule in solution, we employed the final reacted phenol anion to probe the properties of its ground and excited states.(1) Geometry optimization. The ground state (S0) structures for all the molecules were optimized using density functional theory (DFT) at the PBE0/aug-cc-pvtz level with Solvation Model based on Density (SMD) and the dichloromethane (DCM) solvent in the Gaussian 09 program. Meanwhile, the frequency calculations were also performed for all the optimized ground structures to confirm that stable structures without vibrational imaginary frequencies with the same basis set as the optimized process. (2) Absorption process, i.e., vertical absorption, corresponds to the UV-visible absorption spectrum of the molecule. Based on the optimized ground state structure, the excitation energy from the ground state to the excited state is calculated with the basis set of PBE0/aug-cc-pvtz level in DCM solvation. (3) Emission process, i.e., the energy change that causes vertical emission during the transition of the molecule from the excited state back to the ground state, corresponds to the molecular fluorescence emission process. Firstly, the excited state geometry is optimized on the basis of the ground state structure to get its excited state minima energy. Secondly, based on the above excited state configuration, the energy change from the excited state to the ground state is obtained, which is the vertical emission energy. The basis set of PBE0/aug-cc-pvtz level and SMD model is the same as the absorption process.

Comparison of penetration depth and resolution of NIR-II CL and NIR-II FL imaging

To compare the penetration depth and SBR between NIR-II CL imaging and NIR-II FL imaging, a small dish was filled with CL-O3, CL-S3, and CL-Se3 (200 μM of CL-X3 in 100 μL solution containing 80% saline and 20% DMSO) and immediately covered with a cylindrical dish filled with different depths of 1% Intralipid. The depth was obtained by measuring the height of the surface in the cylindrical dish. The various thicknesses employed in the experiment were 0.3, 0.6, 0.9, and 1.2 cm. The NIR-II CL images were directly acquired upon reacting with TBAF (the exposure time was 1200 ms). After 10 min, the NIR-II FL images were acquired using an 808-nm laser (the exposure time was 50 ms). The depth-dependent signal intensity, feature width, and SBR were assessed by fitting cross-sectional intensity profiles with Gaussian function.

In vivo whole-body NIR-II duplex imaging of O2 •− in APAP-induced liver injury

Six-week-old male naked mice under fasting for 12 h were randomly classified into five groups (n = 3). Mice in the experimental group were intraperitoneally injected with different APAP doses (80, 160, and 320 mg/kg) for 5 h. The control group was administered saline (200 μL) while NAC-treated group was pretreated with NAC (200 mg/kg) for 3 h and then intraperitoneally injected with APAP for 160 mg/kg. Subsequently, the mice were anesthetized (isoflurane), executed using 300 μM CL-P in 400 μL solution (containing 75% saline, 20% DMSO, and 5% Tween-80) via an intravenous injection, and then randomly classified into experimental and control groups. Then, the real-time and continuous whole-body imaging was performed for 32 min, and data and images were obtained at 0-, 1-, 3-, 8-, 16-, and 32-min time points. Duplex NIR-II CL and NIR-II FL images were acquired under the following order: for NIR-II CL imaging, the exposure time was 1200 ms in an 808-nm laser, and for NIR-II FL imaging, the exposure time was 50 ms. For the ex vivo imaging of main organs, representative ex vivo duplex NIR-II CL and NIR-II FL images were acquired at 2- and 32-min time points after the resection of the organs from the test group.

Procedure for resecting the main organs and ex vivo imaging

(1) Animal Anesthesia. Use appropriate anesthetics (such as isoflurane or sodium pentobarbital) to induce deep anesthesia in the mouse. (2) Euthanizing the Mouse. Under deep anesthesia, euthanize the mouse humanely using cervical dislocation. Confirm that the mouse shows no signs of life before beginning the dissection. (3) Dissection and Organ Retrieval. Make a longitudinal incision in the abdomen, starting from the lower abdomen up to the thoracic cavity, using surgical scissors. Carefully separate tissues to avoid damaging target organs. Retrieve the heart, liver, spleen, lungs, and kidneys. Rinse the organs with saline to remove blood and other contaminants. Place each organ into pre-prepared containers and immediately label them for further in vivo imaging.

Software

For chemical structure: ChemDraw 20.0. For NMR analysis: TopSpin 3.6.3; MestReNova. For HRMS analysis: Thermo Xcalibur Qual Browser 2.2 SP1.48. For emission spectra analysis: Fluoracle from FS5 software (Edinburgh Instruments, England). For abs spectra analysis: Thermo Scientific™ GENESYS™ 30. For in vivo imaging analysis: ImageJ software (Fiji). For NIR-II chemiluminescence and fluorescence imaging were analyzed using the FLS980 software (Edinburgh Instruments, England).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.