Ultra-small yet highly bright water-soluble single-benzene luminophore for glioblastoma surgical navigation

Abstract

Creating ultra-small, water-soluble fluorophores with low molecular weight (<200 g mol⁻¹) is critical for bioimaging, as it enables high bioavailability, low toxicity, and penetration of physiological barriers—yet resolving the long-standing conflict between ultra-small size, bright emission, and robust photostability has remained a major bottleneck in this field. Here, we report a class of single-benzene skeleton luminophores that break this trade-off: they exhibit ultra-small molecular weights (the smallest so far), near-unity absolute quantum yields (up to 97% in aqueous solution), excellent water solubility, and exceptional photostability against photobleaching, temperature fluctuations (20–80 °C), pH variations (3–12), and ionic interference. The key structural innovations driving these properties are: (1) hydroxyl modifications that enhance water solubility without disrupting the conjugated system; (2) intramolecular F⋯H–O hydrogen bonds that rigidify the planar skeleton, suppressing vibrational relaxation and non-radiative decay. These dyes are synthesized via a gram-scale one-pot S_NAr reaction, avoiding tedious multi-step synthesis of conventional dyes. In preclinical studies, they demonstrate excellent biocompatibility (no toxicity in vitro up to 500 µM; no organ damage in vivo) and efficient blood–brain barrier penetration—critical for brain imaging. When applied to fluorescence-guided resection of orthotopic glioblastoma in mice, these dyes enable precise tumor margin delineation, reduce residual tumor burden by >90%, and extend median survival by 2.3-fold compared to saline controls (42 days vs. 18 days). This work not only provides a new generation of bioimaging probes for neurosurgical oncology but also establishes a general design strategy for ultra-small, high-performance fluorescent materials—laying a solid foundation for clinical translation in GBM surgery and beyond.

Ultra-small single-benzene fluorophores with low molecular weights, near-unity quantum yields, enable guided resection of orthotopic glioblastoma, margin delineation, marked tumor burden reduction, and prolonged survival versus saline controls.

Introduction

Organic small-molecule fluorescent dyes are indispensable tools for bioimaging, owing to their tunable optics, high brightness, and biocompatibility.^1–5 As to biological applications—especially in live-cell or in vivo imaging—water solubility is non-negotiable, as organic co-solvents often induce cytotoxicity or probe aggregation.^6–12 Conventional water-soluble dyes, like fluorescein, cyanines, and Alexa Fluors, mainly rely on hydrophilic modifications (e.g., sulfonic acid groups) to improve solubility, but this comes at a cost: molecular weights exceed 300 g mol⁻¹ (sodium fluorescein: 376 g mol⁻¹; Alexa Fluor 488: 643 g mol⁻¹), molecular volumes expand (>500 ų), and synthesis becomes multi-step and low-yield (Table 2).^13–19 The difficulty in developing small water-soluble dyes arises from the need to balance molecular weight, solubility, and optical performance, since aggregation and disruption of the conjugation system would compromise fluorescence efficiency and optical stability.^20–25 These limitations are particularly problematic for central nervous system (CNS) imaging, where probes must cross the blood–brain barrier (BBB)—a tightly regulated barrier that excludes most molecules >500 Da.^26–32 For glioblastoma (GBM)—the most aggressive primary brain tumor, with a 5-year survival rate <5%—intraoperative imaging probes that penetrate the BBB and distinguish tumor from normal brain tissue are urgently needed, yet conventional dyes (mainly sodium fluorescein) often fail to resolve infiltrative tumor margins, leading to incomplete resection and recurrence.^33–36

Table 2. Comparison with the commercial water-soluble dyes.

Dyes	Ex/Em	QY (H2O, rt)	M w/g mol−1	Molecular volume/Å3	Synthetic step	Total yield/%
2	372/440	0.83	198.1	188	1	63
3	405/480	0.97	196.1	190	1	20
DAPI	358/461	0.80	277.3	500	3–4	50–70
Hoechst 33 342	350/461	0.70	452.5	650	5–6	20–30
Thioflavin T	450/482	0.40	342.4	400	3	60–70
Acridine orange	488/525	0.30	265.4	300	3–4	40–60
FITC	490/525	0.90	389.4	700	2–3	50–70
Alexa Fluor 488	485/519	0.92	643.4	800	3–4	60–75
Sodium fluorescein	490/514	0.95	376.3	600	2–3	60–80
ATTO 488	501/523	0.80	643.0	700	4–6	10–30
Rhodamine B	540/625	0.30	479.0	850	2–3	50–65
Cyanine 3	550/570	0.20	766.0	900	4–5	40–60

Another unmet challenge is balancing photostability with small size, where ultra-small molecules typically exhibit poor structural rigidity, leading to rapid non-radiative decay and photobleaching—rendering them unsuitable for long-term imaging.^37–39 Single-benzene fluorophores have recently emerged as potential candidates for optoelectronics due to their minimal size, but their inherent poor water solubility and structural instability have precluded biological applications.^40–44 Thus, designing ultra-small, water-soluble, photostable fluorescent dyes that can penetrate the BBB and guide GBM surgery remains an unresolved goal in chemical biology and oncology.

Here, we address these challenges by developing a class of single-benzene luminophores with three transformative features: (1) ultra-small molecular weight (<200 g mol⁻¹) and volume (<200 ų), enabling BBB penetration; (2) near-unity quantum yields (up to 97%) and exceptional photostability, supporting high-contrast imaging; (3) gram-scale one-pot synthesis, facilitating preclinical translation (Scheme 1a). We demonstrate that hydroxyl modifications and intramolecular hydrogen bonds stabilize the conjugated system, while intermolecular hydrogen bonds with water enhance solubility. Comprehensive in vitro and in vivo studies confirm biocompatibility, and in orthotopic GBM mouse models, these dyes outperform clinical sodium fluorescein in guiding complete tumor resection and prolonging survival (Scheme 1b). This work redefines the design of bioimaging probes for CNS diseases and establishes a path toward clinical translation.

Scheme 1. (a) Structure and property of the single benzene fluorescent dyes, and the comparison with other water-soluble dyes. (b) The application of these dyes in intraoperative navigation of tumors.

Results and discussion

We employed a single-benzene ring as the core scaffold. To enhance the water solubility, the hydroxyl group was introduced to the skeleton. Dye 2 and 3 were one-pot synthesized from the same reactant TFDCN (1) through a typical S_NAr reaction in acetonitrile (Fig. 1a). Upon optimization, dye 2 was obtained using Et₃N at 60 °C, whereas 3 required 4-dimethylaminopyridine (DMAP) to activate the nucleophile and higher temperature at 100 °C. To be precise, both 2 and 3 were facilely available on the gram scale. Starting from 3.0 g of TFDCN, the reactions under the two conditions afforded 1.9 g of 2 and 0.6 g of 3, corresponding to isolated yields of 63% and 20%, respectively. In particular, the reaction is also spontaneously accessible in aqueous solution under ambient conditions, which demonstrated gradually bright blue emission (Fig. 1b). As shown in Fig. 1c, dye 2 displayed a maximum λ_em of 440 nm and λ_ex at around 370 nm. Increasing the number of substituted hydroxyl groups, the single-benzene dye 3 (λ_em = 480 nm) exhibited a 40 nm red-shift in maximum emission wavelength compared to dye 2. Besides, two distinct excitation bands were observed at 305 nm and 395 nm, corresponding to π → π* and n → π* electronic transitions respectively (Fig. 1d). With the introduction of hydroxyl group to the non-fluorescent 1, two novel dyes were produced through a facile approach, which performed excellent water solubility and bright emission. The reference compound, pentafluorophenol (4), does not exhibit any fluorescence (Fig. S1). In this context, the presence of strong electron-withdrawing groups, the cyano group, is also essential for the single benzene skeleton. Besides, it has been reported that incorporating fluorine atoms into the molecular structure can enhance their permeability through the BBB and promote the targeting of the central nervous system (CNS).^45,46 Potential in brain bioimaging is highly expected.

Fig. 1. Synthesis and spectral properties of dyes 2 and 3. (a) One-pot synthetic route of target single-benzene dyes 2 and 3 at gram-scale. (b) Photo of fluorescence changes of the TFDCN/Et₃N system with reaction time by the spontaneously reaction in aqueous solution under ambient conditions. (c and d) Two-dimensional fluorescence spectra of 2 and 3 measured in deionized water, c = 1 × 10⁻⁴ M. Insets are corresponding emissive images.

There is often a trade-off between reducing molecular weight for better fluorescence efficiency and increasing molecular weight for improved stability or solubility. To evaluate the fluorescence efficiency and photostability of the dyes with the ultra-small molecular weight, we conducted thorough tests from the following aspects, including solvent effect, the influence of temperature and pH, photobleaching performance and electrochemical stability. Dye 2 exhibited excellent solubility both in water and common organic solvents (Fig. S2 and S3). From the normalized absorption spectrum (Fig. 2a), similar peaks of dye 2 were observed in DCM and THF, with a small blue shift in H₂O and a red shift in DMSO. The trend is mostly consistent with the polarity of the solvents, while the blue shift in water is attributed to the hydrogen bonds between hydroxyl group of dye and water. The absorption at around 370 nm is due to the n → π* electronic transitions, so that with the lone pairs involved, intermolecular hydrogen bonds stabilize the ground state more than the excited state, indicating a larger energy gap, thus requiring higher excitation energy. The tendency of the emission is like that of the absorption spectra, in which close emission were recorded in water, DCM and THF, but obvious red-shifted emission was observed in DMSO (Fig. 2b). Under the same conditions, dye 3, exhibits similar changes in absorption and emission behavior in different solvents (Fig. S4 and S5). However, the increase of the number of hydroxyl groups decreased the solubility in some non-polar and polar aprotic solvents such as toluene, THF and dioxane (Fig. S6). Notably, a quantum yield (QY) up to 97% was achieved for 3 in PBS at pH = 7.38 (Fig. 2c), which is consistent with its performance in deionized water. Lifetimes of dyes were also recorded, with that of dye 2 remaining nearly constant across different solvents, approximately 14.0 ns (Fig. 2d). In contrast, dye 3 displayed lifetimes around 6.0 ns, with a noticeable decrease in THF (Fig. 2e). It is worth noting that the QYs of both dye 2 and dye 3 are very high in mentioned solvents. Compared with other organic solvents, the QYs in water were the highest, reaching 83% and 97% respectively (Fig. 2f), which indicates that the dyes might demonstrate extremely high potential for bioimaging without the need for organic solvent auxiliaries. Generally, the photophysical properties of dyes 2 and 3 for powder and in different solvents were summarize in the Table 1 and Fig. S7–S29, where the radiative and non-radiative transition rates were attained. The above results well clarified that the solvent has a relatively minimal effect on dye 2, while it significantly affects dye 3. Aggregation of dye 3 for powder and in some solvents especially in THF would enhance the intermolecular interactions to varying extents, inducing faster non-radiative transition processes, thereby leading to the shortened fluorescence lifetimes and sharp decreases in QYs. The intramolecular hydrogen bonds would reduce excited state energy and prolong radiation lifetime, restricting intramolecular vibrations or twisting to decrease non radiative transition rates. Meanwhile, the smaller structures mean fewer opportunities for non-radiative decay pathways and energy loss, tend to maintain higher fluorescence performance.

Fig. 2. Photophysical properties and photostability evaluations of dyes 2 and 3. (a and b) Normalized absorption and emission spectra of 2 in different solvents. DCM: dichloromethane; THF: tetrahydrofuran; DMSO: dimethyl sulfoxide. (c) QY measurement of 3 in PBS (pH = 7.38). (d and e) Lifetime of 2 and 3 in different solvents. (f) QY of 2 and 3 in different solvents. (g) The curve graph of the fluorescence intensity of dye 2 in H₂O at 440 nm varying with temperature, λ_ex = 370 nm. (h) Photostability evaluation test of dye 2 showing the ratio of the fluorescence intensity after illumination to the initial intensity (I/I₀) at 440 nm with the increase of ultraviolet irradiation time. The time interval is 10 minutes and irradiation wavelength is 365 nm. (i) Cycling stability plot of relative emission intensity (I/I₀) variation at 440 nm through 50 repeated measurements under continuous 370 nm excitation. (j) Chemical stability analysis of dye 2 upon the addition of several metal ions and anions. The molar ratio of the ions to dyes is 10 : 1. (k and l) Cyclic voltammetry curves revealing the electrochemical stability of dyes 2 and 3. c = 1 × 10⁻⁴ M.

Table 1. Photophysical properties of dye 2 and 3 for powder and in different solvents.

Cmpd	Solvent	ε/M−1 cm−1	Ex/Em/nm	τ/ns	QY	K r	K nr
2	PBS	38 070	372/440	14.3	0.83	5.8 × 107	1.1 × 107
	DCM	26 440	383/436	13.9	0.76	5.5 × 107	1.6 × 107
	DMSO	42 660	400/452	13.4	0.53	4.0 × 107	3.5 × 107
	THF	32 880	382/441	12.0	0.60	5.0 × 107	3.3 × 107
	MeOH	40 210	377/448	11.7	0.56	4.8 × 107	3.8 × 107
	Toluene	28 170	382/428	10.7	0.63	5.9 × 107	3.5 × 107
	n-Hexane	—	373/423	8.4	0.44	5.2 × 107	6.7 × 107
	Powder	—	400/450	2.9	0.02	6.9 × 106	3.4 × 108
3	PBS	88 590	405/480	6.0	0.97	1.6 × 108	6.7 × 106
	DCM	54 960	447/497	6.4	0.57	7.3 × 107	8.3 × 107
	DMSO	83 510	447/493	4.5	0.49	1.1 × 108	1.1 × 108
	THF	—	452/497	2.9	0.37	1.3 × 108	2.2 × 108
	MeOH	—	410/492	6.4	0.82	1.3 × 108	2.8 × 107
	Toluene	—	424/517	3.5	0.29	8.3 × 107	2.0 × 108
	n-Hexane	—
	Powder	—	450/520	0.16	0.001	6.3 × 106	6.2 × 109

In addition, the variable-temperature fluorescence spectra (Fig. S30 and S31) from 20 °C to 80 °C were carried out to investigate the effect of temperature on the dyes. Obviously, a reversible cycle in emission intensity was observed for both dye 2 (Fig. 2g) and dye 3 (Fig. S32) in water, demonstrating intensified molecular thermal motion with rising temperature but stable molecular structures within the temperature range of 20 °C to 80 °C. In contrast, dye 3 exhibits lower sensitivity to temperature variations than dye 2, indicating that its molecular architecture possesses higher stability, effectively resisting thermally induced conformational changes and non-radiative decay pathways. Within the pH range of 3–12, the fluorescence intensity of dye 2 is consistent (Fig. S33 and S34). The extent of these changes is similar in both extreme pH conditions. Unlike dye 2, dye 3 shows remarkable acid stability, with negligible fluorescence variation between pH = 1 and pH = 9. However, exposure to alkaline environments (pH > 9) triggers rapid fluorescence quenching, as a result of the strong deprotonation affecting the conjugated system. Given that photobleaching frequently compromises the utility of fluorescent dyes for prolonged imaging applications, we systematically characterized the photostability of dyes 2 and 3 (1.0 × 10⁻⁴ M in aqueous solution) under continuous illumination. Upon 1-hour irradiation at 365 nm, dye 2 and 3 retained 78% (Fig. 2h) and 92% (Fig. S35) of their initial emission intensity (I/I₀). Meanwhile, the systems of 2 and 3 were subjected to 50 consecutive excitation cycles using 365 nm and 395 nm light, during which I/I₀ were also monitored with values of 86% and 97%, respectively (Fig. 2i and S36–S38). Despite the presence of 10-fold concentrations of various metal ions and anions, the dyes maintained bright emission with intensity comparable to the blank group (Fig. 2j and S39–S43). Furthermore, cyclic voltammetry reveals distinct redox behaviors (Fig. 2k and l). Dye 2 exhibits a high oxidation potential at +1.24 V (vs. Ag/AgCl), indicating a deep highest occupied molecular orbital (HOMO) energy level that resists thermodynamic oxidation. In contrast, no detectable oxidation current was recognized for dye 3, possibly due to irreversible hydroxyl group oxidative decomposition or an oxidation potential beyond the solvent window. Both dyes display multiple reduction peaks, suggesting stepwise electron filling into localized orbitals, possibly stabilized by electron-withdrawing cyano groups and asymmetric conjugated system. The stability of a molecule is influenced by both its molecular weight and structural rigidity, where our strategy enhances rigidity by the formation of hydrogen bonds to prevent degradation induced by the small molecular size. Therefore, the above results provide evidence that though dye 2 and 3 have small molecular weights, they are still promising as a stable platform for bioimaging, owing to the acceptable intensity loss after prolonged irradiation, minimal degradation under cyclic excitation, fluorescence stability within a larger pH range and tolerance to ionic and redox interferences, corresponding to the robust π-conjugation system resistant to photooxidation.

As summarized in Table 2, we list a comparative analysis of dye 2 and 3 against ten commercially available water-soluble organic small-molecule dyes. We mainly focus on the QY, molecular volume and synthesis information of the dyes. Dyes exhibit near-unity QY values in aqueous solution, revealing superior light conversion efficiency, minimal energy loss as well as excellent optical behavior. Importantly, molecular weight analysis and van der Waals volumes (MT algorithm) confirmed that dyes 2 and 3 possess significantly lower molecular masses and volumes, both numerically less than 200 g mol⁻¹ and 200 ų, respectively, which were clearly distinguishable with other commercial dyes. The extremely smaller size of dye 2 and 3 contributes to enhanced solubility across a broad range of organic and aqueous solvents and superior cell penetration, thereby improving the efficacy of dyes as fluorescent labels for live-cell imaging. Additionally, both dye 2 and dye 3 were synthesized through a one-pot synthetic method of hydrophilic hydroxyl modification, with moderate total yield and simplified preparation process. Therefore, dye 2 and 3 possess low synthesis cost and unique optical performance, enabling scaled synthesis, comparable or even better than commercial dyes.

Based on the transient absorption (TA) spectroscopy, we investigated the kinetic information of excited states. As illustrated in Fig. 3a, dye 2 displayed two positive excited state absorption (ESA) peaks at around 370 nm and 540 nm, ground state bleaching at 320 nm as well as strong stimulated emission signals at 440 nm consistent with the case of steady-state behavior. Decay related kinetic curve of dye 2 showed two stages with different time scales, characterized by a rapid lifetime of 4.6 ps and a slower component of 14.0 ns (Fig. 3b). The former could be assigned to vibrational relaxation inside excited state molecules, and the latter was impacted by the process of excited state returning to ground state and emitting fluorescence through radiative transition. In comparison with dye 2, dye 3 undergoes two decay lifetimes of 1.5 ps and 7.2 ns respectively (Fig. 3c and S44). The faster radiative decay contributes to the higher fluorescence emission efficiency, while the ultrafast processes imply enhanced intramolecular charge transfer (ICT) and stronger solvent–molecule interactions accelerating solvation relaxation. According to the above results, the energy level structure could be described by the Jablonski diagram shown in Fig. 3d. We further conducted a series of theoretical calculations.^47–51 By visualizing the distribution of ESP, negative charge was primarily distributed in the cyano group, while the most positive point presented in the region of hydroxyl group, confirming a pronounced D–π–A conjugated system (Fig. 3e and S45). Subsequently, frontier orbital analysis further revealed that the donor (hydroxyl) and acceptor (cyano) moieties, as well as the benzene ring as the linked bridge, dominated the relevant molecular orbitals (Fig. 3f and S46). The dominant contribution transition modes of dye 2 and 3 were identified as transitions from the HOMO to LUMO, with coefficients of 90.7% and 92.9%, companied with excited energy of 4.36 eV and 4.24 eV, respectively. Besides, the overlap integral of the two orbitals were calculated as 0.72 and 0.73 for dye 2 and 3, indicating fast transition efficiency, further improving QY and shortening lifetime. VMD was used to display the superposition images of S₀ and S₁ structures from top and side views, and the RMSD values for measuring structural differences were obtained (Fig. 3g and S47). As expected, tiny structural differences were observed with an RMSD value of 0.051 Å for dye 2, which agreed with the minor non-radiative energy loss caused by structural relaxation. We further explore the contribution of different frequency vibration modes to the reorganization energy of S₀ to S₁ state and S₁ to S₀ state of dyes 2 (Fig. 3h and i) and dye 3 (Fig. S48), as well as the atomic displacement vector of the maximum contribution vibration mode in low (<500) and high frequency modes. The highest proportion of vibration modes is concentrated in the bending vibration of hydroxyl groups whether for S₀ or S₁ state, while other molecular vibration was suppressed. The rigid structure and lower degrees of freedom reduce energy loss during vibrational transitions. The sum of the reorganization energy of S₀ (λ(i)) and S₁ (λ(ii)) of dye 2 were calculated as 0.20 eV and 0.19 eV, respectively (Fig. 3j). In the X-ray crystal structure of dye 2 (Fig. 3k), an intramolecular F⋯H–O hydrogen bond was observed that further anchors the conformation. Despite the current lack of a high-quality crystal structure for dye 3, its close structural analogy to dye 2 suggests a significantly higher potential for forming hydrogen bonds. Therefore, the total reorganization energy of the entire system remains at a relatively low level regardless of the contribution of these modes to energy.

Fig. 3. (a) TA spectroscopy of dye 2 recorded at different delay times following photoexcitation at 320 nm. (b and c) Decay associated kinetic fits of dyes 2 and 3 at 370 nm. (d) Jablonski diagram showing the fluorescence mechanism of dye 2. (e) Distribution of electrostatic potential (ESP) on molecular surface accompanied with energy of maximum and minimum points and schematic diagram of D–π–A conjugated structure of dyes. (f) Visualized frontier molecular orbitals of dye 2. (g) The image displaying the superposition of S₀ and S₁ structures from the top and side views and calculated RMSD (root mean square displacement) measuring the difference between the structures. (h and i) Decomposition of recombination energy of S₀ and S₁ and vibration modes with significant contributions of dye 2. (j) Schematic diagram of recombination energy change of S₀ → S₁ transition and calculated energy for different states. Theory level: B3LYP/def2svp. (k) X-ray structure of dye 2.

As the smallest reported fluorescent dye, their performance in bioimaging is crucial for determining its future utility. We performed intracellular imaging of dyes 2 and 3 in GL261 cells using confocal microscopy. We detected dye 2 emission using the Coumarin 343 channel and dye 3 emission using the ATTO 425 channel. The results revealed distinct spatial localization patterns for both dyes (Fig. 4a). To determine their specific subcellular localization, we conducted co-staining experiments with organelle-specific dyes. Consistent with the stability of the single-benzene ring structure, neither dye exhibited significant co-localization with mitochondria or the endoplasmic reticulum after 4 hours of internalization. In contrast, both dyes showed high colocalization with lysosomes (Fig. 4b, S49 and S50). The uptake of both dyes 2 and 3 was significantly reduced following chlorpromazine pretreatment and low-temperature incubation, whereas filipin and colchicine showed no significant inhibitory effects. Endocytosis inhibition studies demonstrated that both dyes are predominantly internalized via an energy-dependent, clathrin-mediated endocytic pathway (Fig. S51 and S52). Notably, both dyes maintained strong fluorescence even under the demanding pH and enzymatic conditions within lysosomes, further demonstrating their superior photostability.

Fig. 4. Intracellular fluorescence detection, organelle colocalisation, and safety assessment. (a) Fluorescence microscopy images of dye 2 in GL261 cells after 4 h incubation (DAPI: blue; dye 2: green), channel: Coumarin 343. Colocalisation of dye 2 with organelle probes: (i) mitochondria; (ii) endoplasmic reticulum; (iii) lysosomes. (b) Fluorescence microscopy images of dye 3 in GL261 cells after 4 h incubation (DAPI: blue; dye 3: green), channel: ATTO 425. Colocalisation of dye 3 with organelle probes: (i) mitochondria; (ii) endoplasmic reticulum; (iii) lysosomes. (c) In vitro cytotoxicity of dye 2 at different concentrations in HEK293 cells (n = 4). (d) In vitro cytotoxicity of dye 3 at different concentrations in HEK293 cells (n = 4). (e) AST and ALT levels in mouse livers and BUN and CREA levels in kidneys after intravenous injection of dyes 2 and 3 (1 mg mL⁻¹, 200 µL; n = 4). (f and g) In vivo toxicity of dyes 2 and 3 in various mouse tissues, observed in hematoxylin and eosin (HE) sections: (f) whole sections; (g) magnified sections.

Effective bioimaging requires high biocompatibility. Therefore, we systematically evaluated the safety profiles of the dyes. For in vitro assessment, HEK293 cells were incubated with the dyes. Results showed that cell viability remained high even at very high concentrations (Fig. 4c and d), indicating excellent in vitro safety. Notably, despite the nuclear localization observed in Fig. 4a, no detectable reduction in cell viability was observed under the experimental conditions, suggesting that nuclear entry of the probes does not necessarily result in cytotoxicity. To assess in vivo toxicity, we intravenously injected healthy mice with equivalent concentrations of the dyes. Blood samples collected 24 hours post-injection showed no significant alterations in key hematological parameters (Fig. 4e). Furthermore, hematoxylin and eosin staining of major organs revealed no signs of tissue toxicity (Fig. 4f), confirming that both dyes possess a favorable in vivo safety profile without damaging normal tissues. This favorable safety profile, coupled with their outstanding optical properties, provides a strong foundation for their use in bioimaging. Moreover, their notably low molecular weight suggested potential for crossing physiological barriers in vivo. Previous reports indicate that probes with molecular weights below 500 Da can leverage size effects to enhance endothelial uptake and facilitate passage across the BBB.⁵² Therefore, we hypothesized that these dyes could be particularly useful for imaging central nervous system diseases, where BBB penetration is a critical challenge.

GBM, the most common malignant brain tumor, is primarily treated with microsurgical resection.⁵³ However, achieving complete resection is challenging due to the invasive nature and poorly defined margins of tumors. To address this, fluorescent dyes such as sodium fluorescein are used intraoperatively to enhance tumor visualization and maximize the extent of resection (Tables S1 and S2).^54–56 Although not tumor-specific, sodium fluorescein accumulates in glioma via BBB disruption to aid fluorescence-guided surgery. Similarly, dye 2 and 3 lack active targeting ligands or tumor-specific metabolic pathways. Therefore, it is proposed that the preferential accumulation of the probes is primarily associated with favorable physicochemical properties. In the glioma model, intrinsic BBB permeability, together with the pathological disruption of the BBB and altered tumor microenvironment, is likely to facilitate probe delivery and retention within tumor tissue.

Given their favorable luminescence properties, biocompatibility, and low molecular weight, and intrinsic BBB penetration capability demonstrated by ex vivo biodistribution studies in healthy mice (Fig. S53 and S54), we evaluated the efficacy of both dyes in guiding resection in orthotopic glioblastoma mouse models, using sodium fluorescein as a control (Fig. 5a). Both probes exhibit similar biodistribution profiles, with the highest fluorescence intensity observed in the liver. Importantly, detectable fluorescence signals were also observed in the brain despite the presence of an intact BBB, demonstrating that both probes possess intrinsic BBB penetration capability. No unexpected accumulation was observed in other major organs, supporting a favorable systemic distribution profile.

Fig. 5. Dye-assisted guidance for glioblastoma resection. (a) Schematic illustration of dye groups, tumor resection, and signal monitoring schedules in orthotopically implanted GBM-bearing mice (all dye concentrations: 1 mg mL⁻¹; 200 µL is administered intravenously per group prior to resection). (b) Representative fluorescence images of GBM-bearing mice after resection guided by different dyes. (c) Survival curves of mice in each group (n = 6). (d) Monitoring of tumor signals in GBM-bearing mice after dye-guided resection. (e) Changes in head tumor signals before and after resection in GBM-bearing mice (n = 6). (f) Representative HE-stained sections from mice in each group. Statistical analysis in (c) is performed using the log-rank test. *p < 0.05; ns, not significant.

Orthotopic glioblastoma models were established by implanting Luci-GL261 cells into mice.⁵⁷ On day 10, tumor resection was performed following an established clinical protocol.⁵⁸ Mice in each group received 10 mg kg⁻¹ of the respective dye 30 minutes prior to surgery, which is attributed to evaluate the imaging performance of different probes under identical mass doses. After scalp preparation and craniotomy over the tumor site, hemostatic sponges were used to control bleeding. Sodium fluorescein was visualized at 465/520 nm, while both dyes were detected at 430/500 nm (Fig. 5b). Resection was performed under an in vivo microscope using ophthalmic forceps and a mastoid spoon.^59,60 The resection was deemed complete when the tumor region was nearly invisible under microscopy and the fluorescence signal was largely absent (Fig. S55).

Postoperative monitoring of bioluminescence signals revealed that the dye 3 group exhibited the most effective resection guidance, as shown by a significant signal reduction in all mice (Fig. 5d and e). Consistently, this group also demonstrated the longest survival (Fig. 5c). Although sodium fluorescein reduced postoperative bioluminescence signals compared with the saline group, this did not translate into improved survival. Given that postoperative survival is influenced by multiple factors, including surgical variability, residual tumor burden, and postoperative recovery, the underlying reason for this observation remains to be further investigated. In contrast, both novel dyes enabled near-total resection without this detrimental effect, significantly prolonging survival compared to non-resected controls. HE staining analysis further confirmed that both dyes facilitated near-complete resection of glioblastoma lesions (Fig. 5f), highlighting their ability to maximize tumor removal while sparing normal tissue. In summary, these dyes show significant advantages for intraoperative navigation in GBM, improving resection rates and reducing residual tumor burden while preserving normal brain tissue. Future studies directly comparing these dyes with 5-aminolevulinic acid (5-ALA) will be valuable for further evaluating their relative performance and clinical translational potential. Incorporating fluorescent protein-expressing orthotopic glioma models together with histological evaluation of resection margins would enable a more quantitative assessment of surgical precision and preservation of adjacent normal brain tissue. Future studies exploring their integration with multimodal intraoperative imaging (e.g., magnetic resonance imaging, optical coherence tomography) are highly promising. This strategy could enable the development of intelligent, high-precision surgical platforms for personalized tumor resection.

Conclusions

We report a class of ultra-small, water-soluble single-benzene luminophores that resolve the long-standing conflict between molecular size, water solubility, and photostability. Their key innovations—hydroxyl modifications for solubility, intramolecular hydrogen bonds for rigidity, and gram-scale one-pot synthesis—make them uniquely suited for bioimaging and clinical translation. Preclinical studies confirm BBB penetration, biocompatibility, and superior performance in GBM surgical navigation, with a 2.3-fold survival extension in mouse models. Beyond GBM, these dyes could potentially enable imaging of other CNS diseases, such as Alzheimer's and Parkinson's, where BBB penetration is critical. The design strategy—using minimal skeletons with targeted hydrogen-bonding modifications—provides a blueprint for next-generation fluorescent probes. With their favorable safety profile and scalable synthesis, these single-benzene luminophores represent a major step toward translating fluorescent imaging into precision neurosurgery.

Ethical statement

All animal procedures were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of Fudan University and approved by the Animal Ethics Committee of Fudan University (approval number: 2023-02-YJ-ST-02).

Author contributions

Y. Liu and Y. Chen carried out the main experiments and data analysing. T. Sun and P. Xing proposed the assumption and wrote the paper.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Supplementary information (SI): materials, experimental details, and additional CD, ¹H NMR, ¹³C NMR, SEM and MS spectra. See DOI: https://doi.org/10.1039/d6sc03762c.