Low molecular weight 4,4’-quinocyanines for in vivo NIR-II fluorescence imaging

Abstract

Fluorescence-guided surgery (FGS) is limited by the shallow penetration and high autofluorescence of currently available NIR-I fluorophores. To address these constraints, we developed a class of low molecular weight organic fluorophores, 4,4′-quinocyanines (QuCy), that extend emission into the NIR-II (950–1700 nm) region. Rational design through substitution of indolenine head groups with N-substituted quinolinium moieties enhanced π-conjugation, reduced HOMO–LUMO gaps, and induced bathochromic shifts of up to 225 nm, confirmed by density functional theory and spectroscopic analyses. Five derivatives (JAM317–319, JAP331, JAP334) were synthesized via a modular route and exhibited emission maxima of 976–1004 nm. Compared with NIR-I dyes, QuCy dyes demonstrated improved depth penetration and contrast due to reduced tissue autofluorescence. In vivo evaluation of JAM317 revealed efficient tumor cell uptake, low cytotoxicity, and high-resolution vascular imaging. These findings establish QuCy dyes as a promising platform for clinically translatable NIR-II contrast agents.

Introduction

Fluorescence-Guided Surgery (FGS) has emerged as a promising technique for cancer diagnosis and treatment, providing real-time visualization of tumor margins and metastatic lesions^1–6. However, fluorescent dyes that emit in the visible light spectrum suffer from poor tissue penetration, light scattering, and tissue autofluorescence, which limit their clinical utility^7–10. To address these challenges, researchers have focused on developing probes that fluoresce in the near-infrared (NIR-I) window (650–950 nm)^11–15. Near-infrared light penetrates deeper into biological tissues than visible light, due to reduced absorption and scattering by water and endogenous chromophores such as hemoglobin^11,16.

NIR-I fluorophores, while exhibiting improved performance over visible dyes, still encounter limitations in penetration depth and tissue contrast^7,11,17. To further enhance tissue penetration and reduce autofluorescence, probes emitting in the short-wave infrared (SWIR), also referred to as the NIR-II region, between 950-1,700 nm, have emerged as a compelling alternative^12,18–21. The SWIR window offers several advantages over NIR-I, including reduced scattering, minimal absorption by tissue chromophores, and negligible autofluorescence, resulting in significantly improved imaging depth and resolution^8,22–25.

Current FDA-approved dyes for optical surgery navigation include Indocyanine Green (ICG)²⁶, Cytalux (OTL38)²⁷, and Lumisight (LUM015)²⁸, which emit in the NIR-I window. ICG and Cytalux contain fluorophores that are part of a class of organic small molecules known as indocyanines. Lumisight contains an indocyanine fluorophore linked to a quencher by a cathepsin-cleavable linker. Indocyanines consist of two N-substituted dimethylindolinium head groups connected by an odd-numbered polymethine linker. Indocyanine dyes are referred to as Cy3, Cy5, or Cy7, based on the number of carbons in the polymethine linker, with emission maxima of approximately 570, 670, and 780 nm, respectively²⁹. The incorporation of an additional benzo moiety into the indolinium head results in a 35 nm bathochromic shift, leading to the development of benzodimethylindolinium dyes such as ICG (Cy7.5), which has an emission maximum at approximately 815 nm³⁰.

Various classes of molecules fluorescing in the NIR-II region have been explored, including quantum dots^31,32, carbon nanotubes^33,34, and small organic dyes³⁵. Quantum dots, composed of inorganic materials such as indium arsenide (InAs) or lead sulfide (PbS), are notable for their size-tunable emission properties, which allow for precise adjustment into the NIR-II region^20,32. Quantum dots offer distinct advantages in terms of their bright fluorescence, high quantum yields, and photostability. However, the heavy metal composition of quantum dots has raised significant concerns regarding their potential for toxicity, thereby limiting their clinical translation and application^36,37.

Carbon nanotubes represent another class of NIR-II fluorescent molecules. These high molecular weight nanoparticles have shown promising fluorescence properties and have been extensively studied for various biomedical applications^1,3,38. Despite their potential, carbon nanotubes have demonstrated carcinogenic effects in the lungs, and their excretion and overall safety profiles remain inadequately understood^2,39,40. This uncertainty regarding their long-term biocompatibility has consequently impeded their progression toward clinical use^41,42.

In contrast, small organic dyes present a more favorable option for NIR-II fluorescence. These dyes are characterized by simpler molecular structures, which typically result in lower toxicity profiles⁴³. The chemical versatility of small organic molecules allows for extensive customization, enabling the development of dyes tailored specifically for biological applications. Their enhanced biocompatibility, coupled with the ability to conjugate with various targeting moieties, short peptides, or antibodies, makes them particularly attractive for clinical applications⁴⁴.

In addition, small organic dyes often exhibit improved pharmacokinetic properties, such as increased renal excretion and reduced liver accumulation, further mitigating toxicity concerns^45,46. Their widespread use across various branches of medicine reinforces their safety profile, particularly when administered at imaging doses. Thus, while both quantum dots and carbon nanotubes offer unique advantages in NIR-II fluorescence, small organic dyes remain the most promising candidates for safe and effective clinical translation in the realm of biomedical imaging and diagnostics.

In this paper, we propose a novel family of 4,4′-quinocyanines (QuCy) for fluorescent imaging. By replacing the dimethylindolenium head groups with N-substituted 4-quinoliniums, these QuCy dyes induce a significant bathochromic shift, resulting in emission maxima exceeding 800 nm for QuCy5, and 1000 nm for QuCy7. This structural modification extends the electron-coupled region, increasing the wavelengths of excitation and emission, leading to distinct advantages such as lower molecular weight compared to traditional cyanines. This paper focuses on the synthesis and early optical characterization of these prototype QuCy dyes, highlighting their potential for advanced biomedical imaging.

Results

HOMO-LUMO Analysis

Density Functional Theory (DFT) calculations were performed on Cyanine (Cy) dyes (Cy5, Cy7L, Cy7C) and their quinolinium-substituted (QuCy) analogs (QuCy5, QuCy7L, QuCy7C), where L and C denote linear and cyclic polymethine linkers, respectively (Supplementary Figs. 1 and 2). Figure 1A shows the calculated orbital distributions. In Cy dyes, electron density was localized between the nitrogen atoms of the polymethine chain, indicating a confined electronic system with higher transition energies. In contrast, QuCy dyes exhibited extended electron delocalization into the quinolinium head groups, producing stronger push–pull effects and reducing the HOMO–LUMO gap (Fig. 1B).

Fig. 1. Computational design and DFT analysis of Cy and QuCy dyes.

A HOMO-LUMO structures of Cy dyes (Cy5, Cy7L, and Cy7C) and their quinolinium-substituted derivatives (QuCy5 (JAM320, JAM335), QuCy7L (JAM319, JAP334), and QuCy7C (JAM318, JAP331, JAM317), showing the increased electron delocalization that extends into the ring structures in QuCy dyes. B Calculated HOMO-LUMO energy gaps for Cy and QuCy dyes, illustrating the reduction in band gaps upon quinolinium substitution. C Summary of computationally derived band gaps, showing significant decreases in QuCy dyes compared to their Cy counterparts.

Band gap calculations supported these observations (Fig. 1C). Cy5 exhibited a band gap of 2.51 eV, which decreased to 2.06 eV in QuCy5 derivatives (QuCy5Me, QuCy5OH). Cy7L’s band gap (2.22 eV) was reduced to 1.86 eV in QuCy7L derivatives (JAM319, JAP334). Cy7C derivatives (JAM317, JAM318, JAP331) showed further reductions to 1.76–1.77 eV, with little variation across the series (Supplementary Fig. 2).

Synthesis of QuCy

QuCy dyes were synthesized through a modular approach combining Menshutkin, Knoevenagel, and Vilsmeier–Haack reactions (Fig. 2A and Supplementary Fig. 1). This strategy enabled the preparation of both linear and cyclic polymethine derivatives, allowing structural diversity to tune optical properties.

Fig. 2. Synthetic routes for linear and cyclic QuCy fluorophores and bio-conjugation.

A Synthetic scheme for QuCy dyes, involving Menshutkin alkylation, Knoevenagel condensation, and Vilsmeier-Haack reactions. Linear dyes (QuCy5 and QuCy7L derivatives) were synthesized via alkylation of the quinolinium head group, while cyclic QuCy7C derivatives were formed using Vilsmeier-Haack intermediates. B Representative targeting modalities for fluorescent dyes, including conjugation to small-molecule ligands, antibodies, nanobodies, proteins, and peptides to enable biomarker-specific imaging.

For the linear dyes (QuCy7L derivatives JAM319 and JAP334), alkylation of the quinoline via the Menshutkin reaction produced intermediates 3 and 4. This step allowed incorporation of functional groups to modify hydrophilicity or lipophilicity, thereby improving solubility or membrane permeability. Targeting moieties such as biotin, folate, or choline could be introduced at this stage. Subsequent Knoevenagel condensation with precursor 2 yielded the final products.

Cyclic chlorocyclohexylidene derivatives (QuCy7C: JAM317, JAM318, JAP331) were prepared through Vilsmeier–Haack formylation to generate the dialdehyde intermediate 1, followed by Knoevenagel condensation. These compounds contained a reactive chlorocyclohexene moiety that enhanced rigidity and photostability, while also providing a functional handle for substitution at the chlorine atom. The resulting cyclic dyes contain a reactive chlorocyclohexene moiety that serves as a chemically addressable site for post-synthetic functionalization. Substitution at the chlorine position enables conjugation to targeting modalities including small-molecule ligands, peptides, proteins, nanobodies, or antibodies, facilitating biomarker-specific imaging applications (Fig. 2B).

Photophysical properties of QuCy Dyes

The synthesized QuCy dyes had cationic molecular weights ranging from 377.51 Da (JAP334) to 512.07 Da (JAM318). These values are significantly smaller than most reported NIR-II dyes, such as IR-FEP-PEG (6500 Da), p-FE (7000 Da), IR-1061-PEG (6000 Da), LZ-1105 (1090 Da), and IR-BEMC6P (5000 Da)^47–49. They are also smaller than commonly used NIR-I dyes, including ICG (751.98 Da) and IRDye800 (999.15 Da). Spectroscopic measurements confirmed marked bathochromic shifts relative to Cy dyes. Absorption maxima moved from 743 nm (Cy7L, JAS239) and 749 nm (Cy7C)^29,50 to 943–944 nm in QuCy7L derivatives (JAM319, JAP334) and 969–971 nm in QuCy7C derivatives (JAM318, JAP331, JAM317) (Fig. 3A and Supplementary Fig. 3). Emission maxima similarly shifted from 767 nm (Cy7L) and 775 nm (Cy7C) to 976–977 nm (QuCy7L) and 1000–1004 nm (QuCy7C). Extinction coefficients (ε) ranged from 41,990 to 160,800 M⁻¹ cm⁻¹. JAM319 had the highest ε (160,800 ± 2,702 M⁻¹ cm⁻¹), whereas JAP334 had the lowest (41,990 ± 635 M⁻¹ cm⁻¹) (Fig. 3B and Supplementary Fig. 4). Beer–Lambert plots show strong linearity (R² > 0.99). Quantum yields (φ) measured in DMSO ranged from 3.2% to 8.6%. JAM318 emerged as the brightest dye (ε × φ = 6,385 ± 909 M⁻¹ cm⁻¹), while JAP334 exhibited the lowest brightness (1344 ± 420 M⁻¹ cm⁻¹). Absorption behavior and relative brightness of the QuCy dyes were evaluated in 10% FBS to mimic plasma conditions (Supplementary Table 1). Under these conditions, all QuCy dyes exhibited a slight blue shift in absorption relative to DMSO and displayed differential fluorescence output when normalized to their DMSO brightness. To contextualize performance against established NIR-II fluorophores, the photophysical properties of JAM317 were compared with benchmark dyes FD-1080 and Flav7 (Supplementary Table 2)^51,52

Fig. 3. Photophysical characterization of QuCy dyes.

A Photophysical Properties of QuCy Dyes Absorption and emission spectra of QuCy dyes in DMSO, illustrating their Stokes shift and showing significant bathochromic shifts compared to Cy dyes B Summary of photophysical properties of the different dyes, including molecular weights, absorption and emission maxima, extinction coefficients, and quantum yields in DMSO. ^aquantum yield measured relative to IR-1048 in ethanol (Φ = 0.001) as a reference standard.

Photostability analysis

The photostability of JAM317 was tested in EggPC and DSPC liposomal formulations across different solvents under continuous and intermittent laser excitation (Fig. 4) at 5X the imaging power density of 2.66 mW/cm². Dyes encapsulated in EggPC liposomes exhibited an average hydrodynamic diameter of 105.7 nm with a zeta potential of –15.13 ± 2.4 mV, whereas DSPC liposomal formulations measured 100.1 nm with a zeta potential of –17.20 ± 3.6 mV. Under continuous irradiation for 60 min, JAM317 retained 75–83% of its fluorescence in DSPC formulations and 86–88% in EggPC formulations. Retention was highest in PBS for DSPC (82.9%) and in FBS for EggPC (87.8%). When subjected to intermittent excitation (every 2 s), stability improved further, with all formulations retaining >92% of initial fluorescence. Notably, EggPC–JAM317 in FBS maintained complete stability, preserving almost 100% of fluorescence after one hour.

Fig. 4. Photostability of encapsulated JAM317 dye.

A Photostability of JAM317 in Liposomal Formulations Photostability of encapsulated JAM317 under continuous laser excitation at 940 nm. JAM317 formulated in EggPC and DSPC liposomes demonstrated minimal photobleaching across solvents (DI water, PBS, and FBS). B Photostability under intermittent excitation, where JAM317 formulations maintained fluorescence over the entire duration. C Percentage fluorescence change after 60 minutes under continuous or intermittent excitation, illustrating stability of JAM317 formulations in both conditions, with EggPC liposomes providing the highest protection against photobleaching.

Depth penetration and tissue autofluorescence

Autofluorescence was assessed in chicken tissue samples under both NIR-I and NIR-II excitation (Fig. 5A). Excitation at 760 and 808 nm (NIR-I) produced significantly higher autofluorescence compared to 890 and 940 nm (NIR-II), where signals were negligible. Autofluorescence intensity scaled with tissue area (Fig. 5B), being most pronounced at 760 nm (e.g., 3124 a.u. at 0.86 mm², 800 nm LP filter), while the lowest signal was observed at 940 nm LP1250 (6.68 a.u. at 0.86 mm²). Increased exposure time amplified autofluorescence at 760 and 808 nm, but had no effect at 890 or 940 nm (Fig. 5C).

Fig. 5. Suppressed tissue autofluorescence under NIR-II excitation and detection.

A Autofluorescence Reduction in the NIR-II Window Autofluorescence analysis of chicken tissue under white light and different excitation wavelengths: 760 nm and 808 nm (NIR-I) and 890 nm and 940 nm (NIR-II) with two different long-pass filters. The experimental conditions were 760 nm 800 LP (excited with a 760 nm laser and viewed through a long-pass 800 nm filter), 808 nm 850 LP, 890 nm 925 LP, 940 nm 1000 LP, and 940 nm 1250 LP. Autofluorescence is prominent at 760 nm, reduced at 808 nm, and negligible at 890 nm and beyond. B Quantification of autofluorescence intensities at different excitation wavelengths with a 1 s exposure time, demonstrating significantly higher autofluorescence in the NIR-I range compared to NIR-II. C Autofluorescence analysis as a function of exposure time, showing an increase in autofluorescence with increasing exposure for 760 nm and 808 nm excitation, while no appreciable autofluorescence was detected at 890 nm and 940 nm across all exposure times.

Depth penetration was evaluated using capillary phantoms covered with intralipid layers (0–6 mm). JAS239 (760 nm excitation) and IR800 (808 nm excitation) rapidly lost structural fidelity by 1 mm depth, with diffuse, halo-like signals. ICG (808 nm excitation) retained partial definition to 3 mm but degraded at greater depths. In contrast, QuCy dyes (940 nm excitation) maintained sharp fluorescence localization even at 6 mm, with minimal blurring (Fig. 6A, D).

Fig. 6. Tissue penetration of NIR-II dyes over NIR-I dyes.

Fluorescence penetration of NIR-I (ICG, JAS239, IRDye 800) and NIR-II (JAM317, JAM318, JAM319, JAP331, JAP334) dyes through A increasing depth of intralipid solution (0–6 mm) and B increasing chicken tissue thickness (0–6 mm). NIR-I dyes were imaged under standard NIR-I excitation, while NIR-II dyes were imaged using a 940 nm laser and under either a long-pass (LP) 1000 or 1250 nm filters C Normalized quantification of (B) fluorescence intensity through chicken tissue phantom for NIR-I and NIR-II dyes imaged with an LP 1000 nm filter. NIR-II dyes exhibit superior depth penetration compared to NIR-I dyes at all thicknesses, with higher fluorescence retention at increased depths. Fluorescence signals from NIR-I dyes rapidly attenuate even at 2 mm, whereas NIR-II dyes maintain 50% intensity up to 6 mm. D Schematic of the experimental laser excitation and detector collection setup used to quantify fluorescence depth penetration through tissue-mimicking phantoms.

Tissue imaging experiments confirmed these results: at 100 μM, NIR-I dyes (ICG, JAS239, IR800) lost 75–84% of fluorescence intensity by 4 mm and 78–83% by 6 mm, while NIR-II dyes retained ~50% of their initial signal at 6 mm (Fig. 6B, C and Supplementary Fig. 5).

Cellular uptake and intracellular fluorescence

To better understand the determinants of cellular fluorescence, the aqueous solubility and photophysical behavior of the QuCy dyes were evaluated under physiologically relevant conditions (Supplementary Table 3). Upon transfer from DMSO into aqueous media, all dyes precipitated in PBS. In contrast, in 5% Dextrose solution (D5W) and 10% FBS (plasma-mimicking conditions), JAM317, JAM318, and JAP331 remained soluble, whereas JAM319 and JAP334 were only sparingly soluble (Supplementary Table 3). Although the soluble dyes formed optically clear yellow solutions in D5W and 10% FBS, this coloration differed from the green appearance observed in DMSO and was accompanied by a near-complete loss of fluorescence.

To assess whether fluorescence quenching could be reversed, quenched dye solutions in D5W were titrated with increasing concentrations of FBS (Supplementary Fig. 8). For the cyclized QuCy7C dyes, fluorescence intensity increased upon serum addition, indicating partial recovery of emission. In contrast, the linear dyes showed minimal to no fluorescence recovery under the same conditions (Supplementary Fig. 8).

Cellular uptake behavior differed markedly among the dyes. In intact A549 cells, only the cyclic dyes-JAM317 exhibited the strongest intracellular fluorescence, followed by JAP331 and JAM318, while the linear dyes JAM319 and JAP334 displayed weak or negligible fluorescence under physiological conditions (Fig. 7A, B). JAM319 showed negligible fluorescence in intact cells and remained non-fluorescent following cell lysis in DMSO, indicating limited cellular internalization (Fig. 7C). In contrast, JAP334 did not exhibit detectable fluorescence in intact cells but produced a measurable fluorescence signal following DMSO lysis, confirming successful cellular uptake, but is, however, non-fluorescent under native physiological conditions (Fig. 7C).

Fig. 7. Cellular uptake and fluorescence of QuCy dyes in A549 cells.

A White-light (top) and NIR-II fluorescence images (bottom) of A549 cells treated with QuCy dyes, showing intracellular fluorescence under physiological conditions. The red arrow indicates the position of the cell pellet as seen under white light. JAM317 exhibited the highest fluorescence, followed by JAP331 and JAM318. B Quantification of intracellular fluorescence intensities from Fig. 6A, showing the relative cellular uptake of different QuCy dyes. C Quantitative analysis of fluorescence in lysed cells treated with QuCy dyes, demonstrating dose-dependent cellular uptake across all dyes. The comparison of fluorescence between intact and lysed cell conditions provides insight into the ability of dyes to fluoresce under native environments versus their cellular uptake efficiency.

Finally, cellular uptake studies were performed using EggPC-encapsulated dye formulations. In this configuration, all tested dyes produced detectable intracellular fluorescence in A549 cells, indicating that liposomal encapsulation enabled cellular entry and fluorescence under conditions where free dyes exhibited limited or no signal (Supplementary Fig. 9).

In vivo imaging and biodistribution

The in vivo performance of JAM317 was evaluated in naive nude mice following intravenous administration of EggPC-encapsulated dye (40 nmol) and compared directly with EggPC-encapsulated ICG as a control (Fig. 8). Vascular imaging was performed using 940 nm excitation with long-pass (LP) emission filters to probe NIR-I and NIR-II performance.

Fig. 8. In vivo vascular fluorescence imaging in mice following intravenous ICG or JAM317 administration.

In vivo fluorescence imaging of mouse vasculature following intravenous injection of ICG (A–C) or JAM317 (D–G). ICG imaged in the NIR-I window A, NIR-II with a 1000 nm long-pass filter B, and NIR-II with a 1250 nm long-pass filter C demonstrates progressively reduced signal and spatial resolution at longer wavelengths. JAM317 (40 nmol, eggPC liposomes) imaged with 1000 nm D and 1250 nm E long-pass filters shows enhanced vessel contrast and improved visualization of fine vascular structures at longer wavelengths. Contrast-enhanced images F, G were generated using ImageJ’s CLAHE plugin, with inverted images providing clear delineation of vascular networks. A supplemental video shows real-time biodistribution and vascular illumination following JAM317 injection.

Following ICG injection, fluorescence was detectable in the NIR-I window; however, distinct vascular structures were not clearly resolved (Fig. 8A). Imaging ICG in the NIR-II window using a 1000 nm LP filter enabled visualization of major vessels, though the signal appeared diffuse and poorly defined (Fig. 8B). Further imaging with a 1250 nm LP filter modestly improved vessel delineation but remained limited in spatial resolution (Fig. 8C).

In contrast, JAM317 imaging in the NIR-II window demonstrated enhanced vascular contrast and resolution. At 1000 nm LP, JAM317 enabled clear visualization of vascular networks (Fig. 8D and Supplementary Movie 1), while imaging at 1250 nm LP further improved spatial resolution, revealing finer vascular architecture (Fig. 8E and Supplementary Movie 2). Contrast-enhanced processing using ImageJ’s CLAHE plugin and inverted display further accentuated vessel boundaries and microvascular features (Fig. 8F, G).

Real-time fluorescence video recordings following JAM317 injection showed rapid systemic perfusion, with initial signal detected in the heart, followed by distribution to the lungs and peripheral vasculature (Supplementary Movies 1 and 2). Biodistribution and pharmacokinetics were quantified by ROI-based fluorescence analysis of major organs (Fig. 9A). Fluorescence was first observed in the heart and subsequently distributed to peripheral tissues (Fig. 9B). Quantitative analysis of the vascular fluorescence decay revealed a rapid circulation half-life of 87 s for JAM317 (Supplementary Fig. 10). Over time, increasing liver-associated signal was observed, consistent with hepatic accumulation.

Fig. 9. Biodistribution and clearance pathways of JAM317 in mice.

A NIR-II fluorescence image of a mouse injected tail vein i.v. with JAM317 (40 nmol) in eggPC liposomes, with regions of interest (ROIs) drawn around key organs, including the heart, liver, lungs, bladder, lymph nodes, and intestines. B Fluorescence intensity versus time for the ROIs, showing the initial distribution of JAM317 through the heart, followed by systemic perfusion and subsequent accumulation in the liver over time. The periodic beat pattern in the time course represents the breathing rate of the animal C NIR-II fluorescence image of the mouse 24 h post-injection, illustrating dye localization primarily in the intestines. D Ex vivo fluorescence images of harvested organs 2 h post-injection, confirming high fluorescence in the liver and moderate fluorescence in the lungs, spleen, and intestines, with minimal fluorescence in the bladder, indicating hepatic metabolism as the dominant clearance pathway.

At 24 h post injection, fluorescence was predominantly localized to the intestines (Fig. 9C), consistent with clearance through fecal excretion pathways. Ex vivo organ imaging performed 2 h post injection confirmed the liver as the primary site of dye retention, with moderate fluorescence observed in the lungs, spleen, and intestines, and minimal signal detected in the bladder (Fig. 9D). In vitro albumin binding studies revealed that select QuCy7C fluorophores, including JAM317, JAM318, and JAP331, exhibit high binding affinity to Human Serum Albumin (HSA), Bovine Serum Albumin (BSA), and Fetal Bovine Serum (FBS) (>95% bound; Supplementary Table 4).

Discussion

We hypothesized that quinolinium substitution would extend conjugation within the cyanine scaffold, reduce the HOMO–LUMO gap, and red-shift optical properties into the NIR-II region. DFT calculations supported this prediction, revealing consistent band gap reductions across both linear and cyclic dye scaffolds (Fig. 1). Guided by these results, we synthesized a library of QuCy dyes using a modular strategy that allowed systematic variation of polymethine structure and head-group substitution (Fig. 2 and Supplementary Fig. 1). This approach provided flexible access to both linear and cyclic derivatives, while also enabling late-stage diversification with targeting moieties.

The synthetic platform demonstrated considerable adaptability. The Menshutkin reaction facilitated the introduction of hydrophilic or lipophilic groups, as well as molecular targeting units, while the incorporation of a chlorocyclohexylidene group in QuCy7C derivatives rigidified the scaffold and provided a site for further functionalization. This modularity establishes QuCy dyes as synthetically accessible, tunable fluorophores with potential for broad imaging applications. The chlorocyclohexylidene moiety present in the QuCy7C derivatives is, for example, reactive toward thiol nucleophiles under aqueous conditions, enabling straightforward post-synthetic modification. This reactivity allows conjugation to thiol-containing targeting motifs, including cysteine residues in peptides and proteins, facilitating the generation of targeted QuCy probes through simple thiol-substitution chemistry (Fig. 2B).

Photophysical characterization confirmed that quinolinium substitution induced large bathochromic shifts relative to conventional Cy dyes. Absorption maxima were red-shifted by ~200 nm, and emission maxima extended beyond 1000 nm (Fig. 3A), firmly positioning QuCy dyes in the NIR-II window. Importantly, these properties were achieved with molecular weights of 377–512 Da, far smaller than most PEGylated or polymeric NIR-II probes, which often exceed 5 kDa. The compact size of QuCy dyes facilitates bioconjugation to make targeted probes. Within the series, JAM318 emerged as the brightest derivative, balancing a moderate extinction coefficient with the highest quantum yield (Fig. 3B).

Both EggPC and DSPC liposomal formulations yielded nanoscale particles (~100–106 nm) with moderately negative surface charges (–15 to –17 mV). Nanoparticles within this size range are known to exploit the enhanced permeability and retention (EPR) effect, facilitating extravasation through leaky tumor vasculature and subsequent retention in the tumor microenvironment. The negative zeta potential further promotes stability in circulation by minimizing nonspecific protein adsorption and reducing rapid clearance, thereby extending blood residence time. We chose liposomal encapsulation because the free QuCy dyes showed poor solubility and stability in aqueous systems, with precipitation observed in PBS and fluorescence quenching in other solvents. Incorporation into lipid bilayers provides a discrete microenvironment that prevents aggregation and preserves fluorescence efficiency. At the same time, the liposomal carrier could function as a biocompatible nanodelivery system, enhancing circulation time and supporting passive tumor accumulation through the EPR effect.

In addition to favorable optical properties, the QuCys dyes exhibited robust photostability under both continuous and intermittent excitation (Fig. 4). Retention of >90% fluorescence after prolonged irradiation. Stability was particularly high in serum-like conditions, underscoring QuCys suitability for in vivo use.

Evaluation in tissue phantoms and biological samples highlighted two fundamental optical advantages of NIR-II probes: reduced autofluorescence and improved penetration depth. Autofluorescence was markedly lower at >900 nm excitation compared to NIR-I (Fig. 5A–C), yielding higher contrast and signal-to-background ratios. Imaging through intralipid and chicken tissue confirmed that conventional NIR-I dyes such as JAS239, ICG, and IR800 rapidly lost signal beyond 2–3 mm, while QuCy dyes preserved ~50% fluorescence intensity with clear structural fidelity up to 6 mm (Fig. 6A–C). These properties address key limitations of NIR-I fluorophores in deep-tissue surgical applications.

The observed loss of fluorescence in D5W despite apparent solubility suggests that QuCy dyes form soluble aggregates in aqueous and serum-containing environments, leading to aggregation-induced quenching (Supplementary Table 3). The color change from green in DMSO to yellow in aqueous media is consistent with excitonic interactions arising from dye aggregation.

Cellular studies revealed additional, scaffold-dependent differences among QuCy derivatives. While most compounds entered A549 cells, only JAM317, JAP331, and JAM318 produced strong fluorescence in intact cells (Fig. 7A, B). In contrast, JAP334 exhibited clear cellular uptake but weak intracellular fluorescence, and JAM319 appeared largely cell-impermeable (Fig. 7A–C). These findings highlight the importance of employing complementary assays—intact-cell fluorescence imaging alongside lysed-cell quantification—to decouple cellular uptake efficiency from intracellular fluorescence performance (Fig. 7C).

We attribute these differences, at least in part, to the intrinsic self-quenching behavior of QuCy dyes in aqueous environments. In solution, these dyes readily aggregate, resulting in diminished fluorescence. For robust intracellular emission to occur, dyes must not only cross the cellular membrane but also engage with proteins or other biomolecular scaffolds that provide a stabilizing environment capable of suppressing aggregation-induced quenching. This requirement likely explains why certain derivatives, such as JAP334, were internalized yet remained weakly fluorescent in intact cells, as insufficient intracellular stabilization preserved the quenched state.

Consistent with this hypothesis, serum titration experiments demonstrated that aggregation is reversible for cyclized QuCy7C dye (JAM317, JAM318, JAP334), but not for linear analogs (Supplementary Fig. 8), underscoring the importance of molecular architecture in dictating protein-mediated deaggregation. Moreover, liposomal encapsulation within EggPC provided a discrete lipophilic microenvironment that prevented aggregation, enabling intracellular fluorescence for dyes that were otherwise quenched under physiological conditions (Supplementary Fig. 9).

Collectively, these results establish that effective intracellular imaging with QuCy dyes requires a balance of cellular uptake and aggregation resistance. Among the derivatives studied, JAM317, JAP331, and JAM318 fulfilled both criteria, producing strong intracellular fluorescence, with JAM317 emerging as the brightest compound in a cellular environment. These findings define key design principles for NIR-II fluorophores and highlight formulation and protein interactions as critical determinants of biological performance.

Intravenous administration of JAM317 into naïve mice established it as a lead QuCy derivative for in vivo NIR-II imaging. Under 940 nm excitation with long-pass emission detection, JAM317 enabled high-resolution visualization of the vascular network, with vessel-level detail most clearly resolved at 1250 nm LP filter (Fig. 8). Direct comparison with EggPC-encapsulated ICG highlighted key performance differences across imaging windows. While ICG produced detectable fluorescence in the NIR-I range, vascular features were poorly resolved, and although imaging in the NIR-II window modestly improved vessel visibility, spatial resolution remained limited even at 1250 nm LP filter. In contrast, JAM317 exhibited substantially enhanced vascular contrast and finer structural definition in the NIR-II window, underscoring the advantages of its longer-wavelength emission and favorable photophysical properties under physiological conditions.

Real-time video imaging further demonstrated the rapid systemic perfusion of JAM317, with fluorescence fluctuations corresponding to cardiac pulsation and respiration, highlighting its sensitivity to dynamic physiological processes. Biodistribution analysis revealed rapid hepatic uptake followed by clearance through the intestines, with minimal renal excretion (Fig. 9). This hepatobiliary clearance profile is consistent with strong serum protein interactions and lipophilic formulation and aligns well with applications requiring sustained vascular contrast rather than rapid urinary elimination.

Collectively, these findings confirm that JAM317 outperforms conventional NIR-I and NIR-II ICG imaging for high-resolution vascular visualization, particularly at longer wavelengths where reduced photon scattering enhances spatial fidelity. Taken together with its compact molecular size, robust photophysical performance, and favorable in vivo behavior, JAM317 exemplifies the strengths of the QuCy platform. More broadly, these results establish QuCy dyes as a new class of compact NIR-II fluorophores with strong translational potential for fluorescence-guided surgery, tumor vasculature mapping, and real-time physiological monitoring.

In summary, we report the rational design, synthesis, and characterization of a new class of low molecular weight 4,4′-quinocyanine (QuCy) fluorophores that extend π-conjugation through quinolinium substitution to achieve efficient NIR-II emission. Guided by DFT analysis, these dyes demonstrate significant bathochromic shifts relative to conventional cyanines, with absorption and emission maxima reaching 970–1004 nm, high molar absorptivity, and quantum yields up to 8.6%. Their low molecular weights, modular synthetic accessibility, and photostability in liposomal formulations underscore their translational potential compared to existing NIR-I dyes. Functional evaluation revealed that QuCy dyes overcome the inherent limitations of ICG and related fluorophores by providing reduced tissue autofluorescence, enhanced penetration through scattering media, and robust fluorescence contrast at depths up to 6 mm. Among them, JAM317 emerged as a lead candidate, exhibiting strong cellular uptake, low cytotoxicity, and high-resolution vascular and biodistribution imaging in vivo. Collectively, these results establish QuCy dyes as a versatile and clinically promising platform of organic NIR-II fluorophores for fluorescence-guided surgery and deep-tissue imaging, paving the way toward the next generation of optical contrast agents. Building on this foundation, we are now developing targeted QuCy derivatives that engage upregulated enzymes and receptors in tumors, with the goal of enabling real-time, molecularly specific imaging for precision-guided oncologic surgery. We are now developing targeted QuCy derivatives that engage upregulated enzymes and receptors in tumors, with the goal of enabling real-time, molecularly specific imaging for precision-guided oncologic surgery. Beyond surgical applications, we envision that QuCy dyes will serve as broadly useful research tools, functioning as fluorescent tags for in vivo tracking of biomolecules or enzyme activity. Owing to their NIR-II emission, these fluorophores offer substantially improved spatial resolution, penetration depth, and signal-to-background compared to current visible fluorescent dyes, enabling more detailed and physiologically relevant tracking of molecular behavior in complex biological environments.

Methods

Solvents were purchased from ThermoFisher Scientific. Dry solvents were purchased from ACROS Organic. All other chemicals were purchased from Millipore Sigma. UV-vis absorption spectra were obtained on a JASCOV700 spectrophotometer, and fluorescence spectra were acquired on a Horiba QuantaMaster Fluorescence Spectrometer. The fluorescence imaging of the mice was performed with a small animal in vivo imaging system, IR VIVO (PhotonEtc, Canada). NMR spectroscopy was performed using deuterated DMSO (ThermoFisher Scientific) using Bruker 500 Uni, NEO400, and NEO600 machines. LC-MS analysis was carried out on the Waters UPLC-SQD2 (Supplementary Fig. 6). Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectra were acquired using a Bruker Ultraflex mass spectrometer (Bruker Daltonics, Billerica, MA, USA) in positive ion mode. Accurate mass measurements were performed using Bruker scimaX (Bruker Daltonics, Billerica, MA, USA) (Supplementary Fig. 7). The spectral workup was performed using MNOVA and flexAnalysis software.

Computational methods

HOMO-LUMO calculations were performed using Spartan’24 software (v.1.3, Wavefunction, Inc.) to evaluate the electronic properties of Cy (Cy5, Cy7L, and Cy7C) and QuCy dyes (QuCy5, QuCy7L, and QuCy7C, where L and C refer to polymethine linkers that are linear or contain a central cyclohexenyl region, respectively). Geometry optimizations were conducted using density functional theory (DFT) with the B3LYP functional and 6–31 G(d) basis set. Following optimization, single-point energy calculations were carried out to determine HOMO and LUMO energy levels. Electron density distributions and band gaps were derived to compare the localized π-electron system of Cy dyes with the extended delocalization observed in QuCy dyes. The solvent effects were modeled using the polarizable continuum model (PCM) for water to align computational predictions with experimental conditions. Visualization of HOMO and LUMO isosurfaces was performed on the Spartan software.

Chemical synthesis

Synthesis of (E)-2-chloro-3-(hydroxymethylene)cyclohex-1-ene-1-carbaldehyde (1).

20.0 mL of anhydrous dimethyl formamide (DMF) and 20.0 mL of dichloromethane (DCM) were mixed on ice. 18.5 mL (198 mmol) of POCl₃ mixed in 20.0 mL of DCM were added to the DMF-DCM solution dropwise under argon over 3 h. 5.00 g (5.28 mL, 50.9 mmol) of cyclohexanone were added to the solution and stirred on ice for 30 min, resulting in a bright yellow solution. The reaction mixture was refluxed at 60 °C for 2 h before increasing the temperature to 65 °C and 70 °C for 30 min and 10 min, respectively. The solution was allowed to rest for 10 min before pouring the mixture onto ice to crystallize. The resulting precipitate was filtered to isolate a yellow powder and dried by lyophilization for 2 h (5.17 g, 59%). ¹H NMR (500 MHz, DMSO-d6, δ, ppm): 8.53 (broad s, 2H, CHO), 2.36 (t, J = 6.3 Hz, 4H, cyclo-C-CH₂-CH₂-CH₂-C), 1.57 (quintet, J = 6.2 Hz, 2H cyclo-C-CH₂-CH₂-CH₂-C).

Synthesis of N-((E)-(2-chloro-3-((E)-(phenylimino)methyl)cyclohex-2-en-1-ylidene)methyl)aniline hydrochloride

3.95 g (22.9 mmol) of the resulting aldehyde 1 was dissolved in 30.5 mL of DMF and 50.0 mL of ethanol. The solution was cooled to −20 °C in a methanol-liquid nitrogen bath. 17.0 mL of 12.1 M HCl was added slowly by pipette while stirring. The resulting solution was a viscous orange liquid. The reaction mixture was warmed to 0 °C in an ice water bath. 6.30 mL (69.1 mmol) of aniline was added by pipette with slow stirring for 30 min. The solution turned to a dark violet and was poured onto ice to precipitate. The metallic purple solid was isolated by vacuum filtration and rinsed with ice-cold H₂O before drying overnight by lyophilization (5.54 g, 68%). ¹H NMR (500 MHz, DMSO-d6, δ, ppm): 8.56 (s, 2H, CHN), 7.60 (d, J = 7.8 Hz, 4H, Ph ortho-CCH), 7.49–7.43 (m, 4H, Ph meta-CCHCH), 7.36-7.26 (m, 4H, NH and Ph para-CCHCHCH), 2.74 (t, J = 6.2 Hz, 4H, cyclo-C-CH₂-CH₂-CH₂-C), 1.87 (quintet, J = 6.0 Hz, 2H cyclo-C-CH₂-CH₂-CH₂-C).

Synthesis of 1-(2-hydroxyethyl)-4-methylquinolin-1-ium chloride (3).

17.0 mL (1.84 g, 129 mmol) of lepidine were mixed with 18.0 mL (2.16 g, 269 mmol) of 2-chloroethan-1-ol. The solution was heated and stirred at 130 °C in a 45 mL Parr autoclave with a magnetic stirring bar for 2 h. The resulting viscous, brown liquid was poured into hot toluene while stirring. A brownish precipitate was obtained and recrystallized twice by dissolving in 100 mL of ethanol and pouring into 700 mL of toluene, resulting in a white powder following lyophilization. (13.25 g, 46%). ¹H NMR (600 MHz, DMSO-d6, δ, ppm): 9.30 (d, J = 6.0 Hz, 1H, Ar NCH), 8.64 (d, J = 8.9 Hz, 1H, Ar CHCHCHCH), 8.55 (dd, J = 8.5, 1.4 Hz, 1H, Ar CHCHCHCH), 8.25 (ddd, J = 8.7, 6.9, 1.4 Hz 1H, Ar, NCCHCHCHCH), 8.10-8.04 (m, 2H, Ar, NCCHCHCHCH and Ar NCHCH), 5.31 (t, J = 5.6 Hz, 1H, HOCH₂), 5.12 (t, J = 5.0 Hz, 2H, NCH₂), 3.91 (tdd, J = 5.9, 4.1, 2.0 Hz, 2H, HOCH₂), 3.02 (d, J = 0.9 Hz,C-CH₃) ¹³C NMR (151 MHz, DMSO-d6, δ, ppm): 158.52, 149.18, 136.99, 134.81, 129.45, 128.92, 127.05, 122.21, 119.47, 59.10, 58.87, 19.67. MALDI-TOF, m/z: (M-Cl+H)⁺ 189.97, calculated for C₁₂H₁₅NO 189.26.

Synthesis of 1,4-dimethylquinolin-1-ium iodide (4).

A 500 mL round-bottom flask equipped with a magnetic stirring bar was loaded with lepidine (28.3 mL, 214.0 mmol) and methyl iodide (10 mL, 160.6 mmol). As the reaction mixture temperature increased, a cold-water bath was used to maintain the temperature at 40 °C. A brownish solid formed, and 150 mL of methanol was added to solubilize the reaction mixture. Additional methyl iodide (4 mL) was added dropwise (for a total amount of 14 mL, 224.9 mmol). The reaction was stirred for 30 min at room temperature, then 30 min at 50 °C. The solution was poured into hot (60 °C) toluene. The precipitate was filtered off and twice recrystallized from toluene-ethanol (7:3 v/v), yielding yellow needles (26.99 g, 42%). NMR (400 MHz, DMSO-d6, δ, ppm): 9.38 (d, J = 6.0 Hz, 1H, Ar NCH), 8.56–8.45 (m, 2H, Ar NCCHCHCHCH), 8.27 (ddd, J = 8.8, 6.9, 1.4 Hz, 1H, Ar, NCCHCHCHCH), 8.10-8.02 (m, 2H, Ar, NCCHCHCHCH and Ar NCHCH), 4.59 (s, 3H, NCH₃), 3.00 (s, 3H, CCH₃). ¹³C NMR (151 MHz, DMSO, δ, ppm) 158.08, 148.88, 137.59, 134.87, 129.59, 128.40, 126.73, 122.40, 119.48, 45.05, 19.61. MALDI-TOF, m/z: (M-I + H)⁺ 159.47, calculated for C₁₁H₁₃N 159.23.

Synthesis of 4-((E)-2((E)-2-chloro-3-(2-((Z)-1-ethylquinolin-4(1H)-ylidene)-ethylidene)cyclohex-1-en-1-yl)vinyl)-1-ethylquinolin-1-ium iodide (JAM317).

2.0 g (5.6 mmol) of 2 were mixed with 3.33 g (11.0 mmol) of 1-ethyl-4-methylquinolin-1-ium iodide 5 in 46 mL of anhydrous pyridine. 4.0 mL of anhydrous triethylamine was added slowly by syringe. The solution changed color gradually from a deep red to blue. The solution was swirled and left to rest overnight at r.t., then 24 h in a −20 °C freezer to facilitate precipitation. The precipitate was isolated by vacuum filtration and dried by lyophilization. Purification was performed by dissolving the crude product in 25.0 mL of DMSO before pouring into 1.0 L of ice-cold H₂O for precipitation. The mixture was left to settle for 1 h before isolating the precipitate by vacuum filtration and drying overnight by lyophilization. The product was a brown powder (1.37 g, 42%). ¹H NMR (600 MHz, DMSO-d6, δ, ppm): 8.51 (dd, J = 8.6, 1.4 Hz, 2H, Ar NCH), 8.15 (d, J = 7.2 Hz, 2H, Ar NCCHCHCHCH), 8.02 (d, J = 13.8 Hz, 2H, linker CCHCHCCCl), 7.93 (d, J = 8.7 Hz, 2H, Ar, NCCHCHCHCH), 7.85 (ddd, J = 8.5, 6.9, 1.3 Hz, 2H, Ar NCCHCHCHCH), 7.59 (tm, J = 7.6 Hz, 2H, Ar NCCHCHCHCH), 7.41 (d, J = 7.2 Hz, 2H, Ar NCHCH), 7.01 (d, 13.8 Hz, 2H, linker CCHCHCCCl), 4.48 (q, J = 7.1 Hz, 4H, NCH₂CH₃), 2.809 (t, J = 6.2 Hz, 4H, cyclo-C-CH₂-CH₂-CH₂-C), 1.86 (quintet, J = 6.2 Hz, 2H, cyclo-C-CH₂-CH₂-CH₂-C), 1.41 (t, J = 7.1 Hz, 6H NCH₂CH₃). ¹³C NMR (151 MHz, DMSO-d6, δ, ppm): 145.89, 141.61, 141.08, 137.73, 135.89, 132.59, 126.84, 126.13, 125.30, 125.20, 117.14, 109.76, 109.34, 48.80, 26.77, 20.97, 14.47. LC-MS (ESI) m/z calcd for C₃₂H₃₂ClN₂: 479.225 [M-I]⁺; found 479.491. HRMS (ESI) m/z calcd for C₃₂H₃₂ClN₂: 479.22485 [M-I]⁺; found: 479.22453. UV/Vis (DMSO, nm (mol⁻¹dm³cm⁻¹)): λ_max (ε)=971 (93,420). Fluorescence (DMSO, nm): λ_ex = 971 nm, λ_em = 1002 nm.

Synthesis of 4-((E)-2((E)-2-chloro-3-(2-((Z)-1-hydroxyethylquinolin-4(1H)-ylidene)-ethylidene)cyclohex-1-en-1-yl)vinyl)-1-hydroxyethylquinolin-1-ium chloride (JAM318).

0.762 g (2.12 mmol) of 2 was mixed with 0.948 g (4.24 mmol) of 3 in 18.0 mL of anhydrous pyridine. 1.50 mL of anhydrous triethylamine was added slowly by syringe. The solution changed color gradually from a deep red to blue. The solution was swirled and left to rest overnight at r.t., then 24 h in a −20 °C freezer overnight to facilitate precipitation. The precipitate was isolated by vacuum filtration and dried by lyophilization. Purification was performed by dissolving the crude product dissolved in DMSO (10 mL) before pouring into ice-cold H₂O (1:40, v/v) to precipitate. The mixture was left to settle for 1 h before isolating the precipitate by vacuum filtration and drying overnight by lyophilization. The product was a brown powder (0.41 g, 36%).¹H NMR (500 MHz, DMSO-d6, δ, ppm): 8.49 (dd, J = 8.7, 1.4 Hz, 2H, Ar NCH), 8.03 (d, J = 7.3 Hz, 2H, Ar NCCHCHCHCH), 8.02 (d, J = 13.5 Hz, 2H, CCHCHCCCl), 7.94 (d, J = 8.7 Hz, 2H, Ar, NCCHCHCHCH), 7.81 (ddd, J = 8.5, 6.8, 1.3 Hz, 2H, Ar, NCCHCHCHCH), 7.58 (dd, J = 8.4, 7.0 Hz, 2H, Ar NCCHCHCHCH), 7.39 (d, J = 7.7 Hz, 2H, NCHCH), 7.00.(d, 13.8 Hz, 2H, CCHCHCCCl), 5.12 (broad m, 2H, OH), 4.51 (t, J = 5.4 Hz, 4H, NCH₂), 3.78 (m, 4H, HOCH₂), 2.80 (t, J = 6.2 Hz, 4H, cyclo-C-CH₂-CH₂-CH₂-C), 1.86 (quintet, J = 6.2 Hz, 2H, cyclo-C-CH₂-CH₂-CH₂-C). ¹³C NMR (151 MHz, DMSO-d6, δ, ppm): 146.04, 142.38, 141.75, 137.99, 135.91, 132.33, 126.81, 126.07, 125.30, 125.20, 117.18, 109.66, 108.66, 58.94, 55.90, 26.76, 20.72. LC-MS (ESI) m/z calcd for C₃₂H₃₂ClN₂O₂⁺: 511.215 [M-Cl]⁺; found 511.444. HRMS (ESI) m/z calcd for C₃₂H₃₂ClN₂O₂: 511.21468 [M-Cl]⁺; found: 511.214682. UV/Vis (DMSO, nm (mol⁻¹dm³cm⁻¹)): λ_max (ε)=969 (74,250). Fluorescence (DMSO, nm): λ_ex = 969 nm, λ_em = 1004 nm.

Synthesis of 1-(2-hydroxyethyl)-4-((1E,3E,5E)-7-((Z)-1-(2-hydroxyethyl)-quinoline-4(1H)-ylidene)hepta-1,3,5-trien-1-yl)quinoline-1-ium chloride (JAM319).

1.27 g (4.46 mmol) of N-((1E,3E,5E)-5-(phenylimino)penta-1,3-dien-1-yl)aniline hydrochloride were mixed with 2.00 g (8.94 mmol) of (1) in 37.0 mL of anhydrous pyridine. 3.20 mL of anhydrous triethylamine was added slowly by syringe. The solution changed color gradually from a deep red to blue. The solution was swirled and left to rest overnight at r.t., then 24 h in a −20 °C freezer to facilitate precipitation. The precipitate was isolated by vacuum filtration and dried by lyophilization. Purification was performed by precipitation of product dissolved in DMSO (25 mL) before pouring into ice-cold H₂O (1:40, v:v). The mixture was left to settle for 1 h before isolating the precipitate by vacuum filtration and drying overnight by lyophilization. The product was a brown powder (0.54 g, 26%). ¹H NMR (500 MHz, DMSO-d6, δ, ppm): 8.34 (d, J = 8.5 Hz, 2H, Ar NCH), 7.95 (d, J = 7.2 Hz, 2H, Ar NCCHCHCHCH), 7.89 (d, J = 8.8 Hz, 2H, Ar, NCCHCHCHCH), 7.77 (m, 2H, Ar, NCCHCHCHCH), 7.73 (t, J = 13.0 Hz, 2H, linker CCHCHCHCHCHCHCHC), 7.54 (t, J = 7.6 Hz, 2H, Ar NCCHCHCHCH), 7.36 (d, J = 7.8 Hz, 2H, Ar NCHCH), 7.16 (t. J = 12.8 Hz. 1H, linker CCHCHCHCHCHCHCHC), 6.97 (d, J = 13.6 Hz. 2H, linker CCHCHCHCHCHCHCHC), 6.54 (t, J = 12.7 Hz. 2H, linker CCHCHCHCHCHCHCHC), 5.12 (broad s, 2H, OH), 4.45 (t, J = 5.2 Hz, 4H, NCH₂), 3.76 (m, 4H, HOCH₂). ¹³C NMR (101 MHz, DMSO-d6, δ) 145.57, 142.18, 141.53, 138.81, 132.13, 128.92, 125.79, 125.71, 124.72, 124.55, 117.58, 112.01, 108.09, 58.90, 55.50. LC-MS (ESI) m/z calcd for C₂₉H₂₉N₂O₂: 437.222 [M-Cl]⁺; found 437.468. HRMS (ESI) m/z calcd for C₂₉H₂₉N₂O₂: 437.22235 [M-Cl]⁺; found: 437.222355. UV/Vis (DMSO, nm (mol⁻¹dm³cm⁻¹)): λ_max (ε)=943 (160,800). Fluorescence (DMSO, nm): λ_ex = 943 nm, λ_em = 976 nm.

4-((E)-2-((E)-2-chloro-3-(2-((E)-1-methylquinolin-4(1H)-ylidene)ethylidene)cyclohex-1-en-1-yl)vinyl)-1-methylquinolin-1-ium iodide (JAP331).

1.07 g (2.97 mmol) of 2 were mixed with 1.51 g (5.31 mmol) of 4 in 22 mL of anhydrous pyridine. 1.9 mL of anhydrous triethylamine was added slowly by syringe. The solution changed color gradually from a deep red to blue. The solution was swirled and left to rest overnight at r.t., then 24 h in a −20 °C freezer to facilitate precipitation. The precipitate was isolated by vacuum filtration and dried by lyophilization. Purification was performed by precipitation of the crude product dissolved in 10.0 mL of DMSO before pouring into 400 mL of ice-cold H₂O. The mixture was left to settle for 1 h before isolating the precipitate by vacuum filtration and drying overnight by lyophilization. The product was a brown powder (0.36 g, 24%). ¹H NMR (400 MHz, DMSO, δ) 8.48 (d, J = 8.5 Hz, 2H, Ar NCH), 8.10 (d, J = 7.2 Hz, 2H, Ar NCCHCHCHCH), 8.00 (d, J = 13.9 Hz, 2H, linker CCHCHCCCl), 7.89 – 7.80 (m, 4H, Ar NCCHCHCHCH), 7.59 (t, J = 7.5 Hz, 2H, Ar NCCHCHCHCH), 7.39 (d, J = 7.2 Hz, 2H, Ar NCHCH), 6.99 (d, J = 14.0 Hz, 2H, linker CCHCHCCCl), 3.99 (s, 6H, NCH₃), 2.79 (t, J = 6.2 Hz, 4H, cyclo-C-CH₂-CH₂-CH₂-C), 1.90-1.83 (m, 2H, cyclo-C-CH₂-CH₂-CH₂-C). ¹³C NMR (151 MHz, DMSO-d6, δ, ppm) 145.64, 141.95, 138.79, 136.20, 135.67, 132.38, 129.46, 128.29, 126.16, 124.82, 117.24, 109.62, 108.78, 41.72, 26.73, 20.96. LC-MS (ESI) m/z calcd for C₃₀H₂₈ClN₂⁺: 451.194 [M-I]⁺; found 451.489. HRMS (ESI) m/z calcd for C₃₀H₂₈ClN₂: 451.19355 [M-I]⁺; found: 451.193553. UV/Vis (DMSO, nm (mol⁻¹dm³cm⁻¹)): λ_max (ε)=970 (54,240). Fluorescence (DMSO, nm): λ_ex = 970 nm, λ_em = 1000 nm.

Synthesis of 1-methyl-4-((1E,3E,5E)-7-((Z)-1-methylquinolin-4(1H)-ylidene)hepta-1,3,5-trien-1-yl)quinolin-1-ium iodide (JAP334).

0.63 g (2.23 mmol) of N-((1E,3E,5E)-5-(phenylimino)penta-1,3-dien-1-yl)aniline hydrochloride 6 were mixed with 1.09 g (3.52 mmol) of 4 in 21.0 mL of anhydrous pyridine. 1.40 mL of anhydrous triethylamine was added slowly by syringe. The solution changed color gradually from a deep red to blue. The solution was swirled and left to rest overnight at r.t., then 24 h in a −20 °C freezer to facilitate precipitation. The precipitate was isolated by vacuum filtration and dried by lyophilization. Purification was performed by precipitation of the product dissolved in DMSO (9 mL) before pouring into ice-cold H₂O (1:40, v/v). The mixture was left to settle for 1 h before isolating the precipitate by vacuum filtration and drying overnight by lyophilization. The product was a brown powder (0.19 g, 21%). ¹H NMR (400 MHz, DMSO-d6, δ, ppm): ¹H NMR (400 MHz, DMSO-d6, δ, ppm) 8.33 (dd, J = 8.7, 1.4 Hz, 2H, Ar NCH), 8.01 (d, J = 7.3 Hz, 2H, Ar NCCHCHCHCH), 7.84 – 7.68 (m, 6H, Ar, NCCHCHCHCH, linker CCHCHCHCHCHCHCHC), 7.56 (ddd, J = 8.1, 6.6, 1.4 Hz, 2H, Ar NCCHCHCHCH), 7.35 (d, J = 7.3 Hz, 2H, Ar NCHCH), 7.15 (t, J = 12.9 Hz, 1H, linker CCHCHCHCHCHCHCHC), 6.97 (d, J = 13.6 Hz, 2H, linker CCHCHCHCHCHCHCHC), 6.59 – 6.50 (m, 2H, linker CCHCHCHCHCHCHCHC), 3.94 (s, 6H, NCH₃). ¹³C NMR (101 MHz, DMSO-d6, δ, ppm) 145.37, 142.98, 141.22, 139.00, 132.31, 128.75, 125.93, 125.89, 124.49, 124.46, 117.16, 112.13, 108.37, 41.47. LC-MS (ESI) m/z calcd for C₂₇H₂₅N₂: 377.201 [M-I]⁺; found 377.434. HRMS (ESI) m/z calcd for C₂₇H₂₅N₂: 377.20123 [M-I]⁺; found: 377.201225. UV/Vis (DMSO, nm (mol⁻¹dm³cm⁻¹)): λ_max (ε)=944 (41,990). Fluorescence (DMSO, nm): λ_ex = 944 nm, λ_em = 977 nm.

Absorbance spectra of Quinocyanine dyes

The absorbance spectra for the dyes were obtained using a JASCO spectrophotometer measuring absorbances in 1 nm steps from 500 to 1,200 nm. All compounds were dissolved in DMSO at a concentration of 5 μM.

Emission spectra of Quinocyanine dyes

The emissions for the dyes were obtained using a Horiba QuantaMaster fluorometer with excitation at λ_max and emissions recorded in 1 nm steps from (λ_max + 10) to 1500 nm. For NIR-I dyes, a Photomultiplier Tube (PMT) detector was used that covers wavelengths 182–900 nm, while for NIR-II dyes an Indium Gallium Arsenide (InGaAs) detector was used that covers wavelengths 800–1700 nm.

Determination of molar absorption coefficients

Molar extinction coefficients (ε) were obtained by measuring absorbance across a range of known concentrations (1–6 μM) in DMSO. Absorbance spectra were acquired using a 10 mm path length quartz cuvette on a UV–vis spectrophotometer. For each concentration point, measurements were repeated in triplicate, and ε values were calculated from the slope of the linear fit of absorbance versus concentration according to the Beer–Lambert law:

$$A=\epsilon cl$$ 1

where A is absorbance, c is concentration, and l is the path length.

Quantum yield measurement

Relative fluorescence quantum yields (Φ) were determined in DMSO using a previously described comparative method with IR-1048 in ethanol (Φ = 0.001) as the reference standard⁵³. Briefly, absorbance values at the excitation wavelength (940 nm) were maintained below 0.1 to minimize inner filter effects. Emission spectra were recorded on a Horiba Fluorolog-3 spectrofluorometer using matched quartz cuvettes (10 mm path length), and integrated fluorescence intensities were obtained by correcting for instrument response.

Quantum yields were calculated according to the equation:

$${\varnothing }_S={\varnothing }_R\frac{I_S}{I_R}\frac{A_R}{A_S}{\left(\frac{{\eta }_S}{{\eta }_R}\right)}^2$$ 2

where I is the integrated fluorescence intensity, A is the absorbance at the excitation wavelength, and η is the refractive index of the solvent (1.4772 for DMSO, 1.361 for ethanol). The subscript S is used to denote the sample compound, while R refers to the reference compound. All measurements were performed in triplicate at room temperature. Samples and standards were prepared in matched solvent systems and measured under identical instrumental settings.

Depth penetration studies

JAM317, JAS239, and ICG were diluted in DMSO and placed in a 96-well black-well clear-bottom plate at concentrations from 0.1 μM to 100 μM. The plate was imaged using the PhotonEtc IR Vivo scanner using the NIR-II emission filter (long-pass 1000 nm) using the excitation laser closest to each dye’s maximum absorbance: 760 nm for JAS239, 808 nm for ICG, and 940 nm for JAM317. The fluorescence intensity of ICG with an 808 nm laser at 1.10 mW/mm² and 0.05 s exposure time was recorded, and the parameters were adjusted for other dyes such that they all had the same initial fluorescent intensity. Wells were then covered with thinly sliced raw chicken breast tissue from 1 mm to 6 mm thick. Imaging was then repeated for each tissue thickness.

Depth penetration was also evaluated using intralipid phantoms prepared from 1% (w/v) intralipid solutions in PBS to mimic the optical scattering properties of biological tissue. Capillary tubes (inner diameter ~0.5 mm) were filled with dye solutions (10 µM) and submerged under increasing layers of intralipid (0–6 mm). Imaging was performed using excitation lasers at 760 nm (JAS239), 808 nm (ICG, IR800), and 940 nm (QuCy dyes) with corresponding long-pass emission filters (LP800, LP850, LP1000, LP1250). Fluorescence intensity was recorded at each depth, and normalized signal intensities were quantified to assess depth-dependent attenuation. Comparative analysis of NIR-I and NIR-II dyes demonstrated the effect of excitation/emission wavelength on penetration and spatial resolution in scattering media.

Cell culture

A549 human non-small cell lung cancer (NSCLC) cells were maintained in F-12K Nutrient Mixture (Kaighn’s modification) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. Cells were incubated at 37 °C in a humidified atmosphere containing 5% CO₂. Cells were routinely passaged using 0.25% trypsin-EDTA upon reaching ~80% confluency, and only low-passage cells (<P15) were used for all experiments.

Cell uptake assay using cell lysate measurements

A549 lung cancer cells were seeded into 96-well plates at densities ranging from 0 to 2.0 × 10⁴ cells per well and incubated in F-12K medium at 37 °C with 5% CO₂ for 24 h to allow adherence and growth. The cells were treated with 200 µL of 10 μM of each QuCy dye in serum-free media and incubated under the same conditions for one hour. After treatment, the dye-containing media were aspirated, and the cells were washed twice with 100 μL of ice-cold DPBS to remove extracellular dye. A subsequent wash with 0.05% Tween-20 in DPBS was performed to eliminate residual dye, followed by a final wash with ice-cold DPBS. Following the washes, cells were lysed in 100 μL of DMSO to release intracellular content. NIR-II fluorescence was measured directly from the 96-well plate using the PhotonEtc IR Vivo scanner, with excitation at 940 nm and detection through a 1000 nm long-pass filter.

Cell uptake assay using cell pellet measurements

A549 cells were seeded into 6-well plates at a density of 3 × 10⁵ cells per well and incubated at 37 °C with 5% CO₂ for 48 h. The cells were treated with 200 µL of 10 μM of either each free QuCy dye or EggPC encapsulated dye in serum-free media and incubated under the same conditions for one hour. Following treatment, the media was aspirated, and the cells were washed twice with 100 μL of ice-cold DPBS to remove residual extracellular dye. A subsequent wash with 0.05% Tween-20 in DPBS was performed, followed by a final wash with ice-cold DPBS. The cells were trypsinized using 0.25% trypsin-EDTA, and 1 mL of media containing FBS was added to neutralize the trypsin. The cell suspension was transferred to Eppendorf tubes, centrifuged at 13,000 rpm for 5 min, and the supernatant was aspirated. The cell pellets were washed twice with DPBS to remove any remaining extracellular dye. For the final wash, the cell pellet was left intact at the bottom of the tube and imaged for NIR-II fluorescence using the PhotonEtc IR Vivo scanner, with excitation at 940 nm and detection through a 1000 nm long-pass filter.

Preparation of dye-encapsulated liposomes

Dye-encapsulated liposomes were prepared using the thin-film hydration method as previously described^13,54. Briefly, the phospholipid component (EggPC or DSPC) was dissolved in chloroform to create a homogeneous solution. The lipid solution was transferred to a round-bottom flask, and the chloroform was evaporated under reduced pressure using a rotary evaporator to form a thin lipid film on the inner surface of the flask. The flask was then placed under vacuum for 2 h to remove any residual solvent. The dried lipid film was hydrated with 5% Dextrose containing JAM317 at a concentration of 200 µM. The hydration process was carried out by gently agitating the flask at RT, resulting in multilamellar liposomes. The liposome suspension was then passed through a polycarbonate membrane using the Avanti Polar Lipids Mini-Extruder Kit (Avanti Polar Lipids, Inc.). Extrusion through 100 nm membranes was performed at least 20 times at RT to yield unilamellar liposomes with an average size of approximately 100 nm, as verified by dynamic light scattering (DLS) using a Malvern Panalytical Zetasizer.

Encapsulation efficiency

To determine the encapsulation efficiency of the dye, unencapsulated dye was separated from liposome-encapsulated dye using centrifugal filtration. A sample of the liposome suspension was loaded into a centrifugal filtration tube with a 10,000 molecular weight cutoff membrane (Amicon Ultra, Millipore). The sample was centrifuged at 4000 × g for 10 min at 4 °C. The filtrate containing unencapsulated dye passed through the membrane, while the liposomes with encapsulated dye remained in the retentate. The concentration of dye in both the retentate and the filtrate was quantified by UV-Vis spectrophotometry at 970 nm. Encapsulation efficiency was calculated using the following formula:

$$EncapsulationEfficiency\left(\%\right)=\frac{EncapsulatedDye}{\mathrm{Total}\mathrm{Dye}}\times 100$$ 3

where Encapsulated Dye represents the amount of dye retained in the filtration tube, and Total Dye represents the sum of dye in the filtrate (unencapsulated) and the retentate (encapsulated).

Photostability studies

The encapsulated dyes were prepared at a concentration of 10 µM in deionized water, phosphate-buffered saline (PBS), or fetal bovine serum (FBS). Solutions were loaded into glass capillary tubes and irradiated using a 940 nm laser at 5X imaging power density of 2.66 mW/cm² for 60 min. Samples were either continuously exposed or subjected to intermittent irradiation with 2 s intervals. Fluorescence intensity was measured at defined time points. All measurements were performed in triplicate, and the mean fluorescence intensity was normalized to the initial value (F₀) and plotted as F/F₀ over time.

Autofluorescence studies

Fresh, skinless chicken breast tissue was used to assess intrinsic autofluorescence. Uniform tissue slabs (~5 mm thick) of varying sizes were selected across the same tissue sample. Imaging was performed using five excitation wavelengths and emission filter combinations: 760 nm (LP800), 808 nm (LP850), 890 nm (LP925), and 940 nm with both LP1000 and LP1250 filters. Fluorescence intensity from each region was quantified to compare background signal as a function of area size and excitation wavelength.

In vivo mice studies

In vivo biodistribution and fluorescence imaging experiments were conducted in healthy adult athymic nu/nu nude mice (8–12 weeks old). All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Pennsylvania (protocol no. 804641) and were performed in accordance with institutional and national guidelines. Animals were housed in a specific pathogen–free facility under standard conditions maintained by the University of Pennsylvania Unit for Laboratory Animal Resources (ULAR), with controlled temperature and humidity, a 12 h light–dark cycle, and ad libitum access to food and water. Animals were acclimated to the housing environment prior to experimentation. For all imaging and injection procedures, mice were anesthetized with 1% isoflurane delivered by inhalation. Intravenous administration of fluorescent agents was performed via a tail vein catheter to ensure consistent delivery and minimize handling-related stress. Animals were continuously monitored throughout anesthesia to maintain physiological stability. Following completion of imaging studies, mice were euthanized by CO₂ inhalation with cervical dislocation as a secondary confirmatory method per IACUC guidelines.

All imaging was conducted using the IR VIVO preclinical imaging system (Photon Etc., Montreal, Canada), equipped with an Alizé 1.7 InGaAs camera for detection in the NIR-II region. The system was operated using PhySpec™ software, which controlled laser excitation, filter selection, and image acquisition parameters. Excitation was performed using fiber-coupled laser diodes at 760, 808, 890, and 940 nm, with corresponding long-pass filters (LP800, LP850, LP925, LP1000, and LP1250) selected via the integrated motorized filter wheel. Exposure time, gain, and laser power density were optimized and held constant across comparative datasets.

Images and real-time video recordings were acquired at defined time points post-injection and processed using PhySpec™ and ImageJ software. Background subtraction and contrast adjustments were applied uniformly across groups. Quantitative analysis of fluorescence intensity was performed by drawing regions of interest (ROIs) over key anatomical areas, with data reported as mean intensity or signal-to-background ratio (SBR), where applicable.

Vasculature imaging

JAM317 or ICG was encapsulated in egg phosphatidylcholine (eggPC) liposomes prepared by the thin-film hydration method. The final dye concentration in the liposomes was 200 µM. Each mouse was anesthetized with isoflurane (1.5–2%) and administered 100 µL (40 nmol) of the JAM317-encapsulated liposomes via tail vein catheter injection. JAM317 fluorescence imaging was performed during injection and immediately after using the 940 nm excitation laser at 0.53 mW/cm² laser density and an exposure time of 50 ms under either 1000 LP or 1250 LP optical filters. ICG fluorescence imaging was performed during injection and immediately after using the 808 nm excitation laser at 0.78 mW/cm² laser density and an exposure time of 50 ms under either 850 LP for NIR-I or 1000 LP and 1250 LP for NIR-II optical filters. To evaluate the vascular distribution, whole body fluorescence images were acquired to monitor systemic circulation and overall distribution of the dye.

Biodistribution

Fluorescence images were obtained at either 2 h or 24 h post injection to track the temporal distribution of the dye-encapsulated liposomes. Signal intensities from various anatomical regions were quantified and analyzed using the PhySpec software. At 2 h post-injection, a set of mice were sacrificed, and organs, including the liver, spleen, kidneys, heart, lungs, and brain, were harvested. Ex vivo fluorescence imaging was performed on these organs to quantify dye biodistribution.

Blood circulation half-life measurement

JAM317 was formulated in EggPC liposomes and administered to naïve mice via intravenous bolus injection. In vivo fluorescence imaging was performed live during injection with 940 nm excitation and long-pass emission filtering. Blood circulation half-life was determined by quantifying fluorescence signal from arterial blood flow.ROIs were manually drawn over major superficial arteries, and mean fluorescence intensity was extracted at successive time points post-injection. All image analysis was performed using PhySPec software, with fluorescence intensity used as the quantitative readout. Background-subtracted arterial fluorescence values were normalized to the initial post-injection signal to generate circulation decay curves. Data are reported as mean ± SD, and the apparent blood circulation half-life was obtained by fitting the normalized fluorescence decay to a monoexponential decay model.

Albumin binding assay

Albumin binding of QuCy fluorophores was evaluated using human serum albumin (HSA), bovine serum albumin (BSA), and fetal bovine serum (FBS). 10 µM QuCy dyes were incubated with each protein solution at 37 °C for 30 min to allow binding equilibration. Following incubation, samples were subjected to centrifugal ultrafiltration using 10 kDa MWCO centrifugal filters to separate unbound dye from albumin-bound fractions. The filtrate, containing unbound dye, was collected, lyophilized to remove solvent, and subsequently reconstituted in DMSO. Dye concentrations were determined by comparison to calibration curves prepared in DMSO. The fraction of dye bound to albumin was calculated by subtracting the amount of unbound dye from the total input, and results are reported as percent bound ± SD. Values labeled as n.d. indicate conditions for which binding could not be reliably determined.

Author contributions

R.K.I., M.C.H., E.J.D. and A.V.P. conceived the study. R.K.I. curated the dataset, performed the formal analysis, and, together with M.C.H., S.A.P., and G.C.D., conducted the investigations. Methodology was developed by R.K.I., E.J.D., and A.V.P. R.K.I. prepared the original draft of the manuscript, and R.K.I., E.J.D., and A.V.P. reviewed and edited the final version.

Data availability

All data supporting the findings of this study are available within the Article and its Supplementary Information.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s44303-026-00140-3.