Copper-Catalyzed Covalent Dimerization of Near-Infrared Fluorescent Cyanine Dyes: Synergistic Enhancement of Photoacoustic Signals for Molecular Imaging of Tumors

Abstract

Photoacoustic (PA) imaging relies on the absorption of light by chromophores to generate acoustic waves used to delineate tissue structures and physiology. Here, we demonstrate that Cu(II) efficiently catalyzes the dimerization of diverse near-infrared (NIR) cyanine molecules, including a peptide conjugate. NMR spectroscopy revealed a C–C covalent bond along the heptamethine chains, creating stable molecules under conditions such as a wide range of solvents and pH mediums. Dimerization achieved >90% fluorescence quenching, enhanced photostability, and increased PA signals by a factor of about 4 at equimolar concentrations compared to the monomers. In vivo study with a mouse cancer model revealed that dimerization enhanced tumor retention and PA signal, allowing cancer detection at doses where the monomers are less effective. While the dye dimers highlighted peritumoral blood vessels, the PA signal for dimeric tumor-targeting dye-peptide conjugate, LS301, was diffuse throughout the entire tumor mass. A combination of the ease of synthesis, diversity of molecules that are amenable to Cu(II)-catalyzed dimerization, and the high acoustic wave amplification by these stable dimeric small molecules ushers a new strategy to develop clinically translatable PA molecular amplifiers for the emerging field of molecular photoacoustic imaging.

Photoacoustic (PA) imaging is a fast-growing hybrid technology that combines optical with ultrasound methods to provide multiscale imaging capability and a high contrast ratio, depending on the difference in optical absorption characteristics of the target tissue.^[1–6] A major limitation of optical imaging methods is the poor penetration of light into tissue, confining the technology to the interrogation of shallow tissue.^[7–8] PA imaging overcomes this limitation by stimulating chromophores in tissue with light to generate spatially resolved ultrasound signals. Because the scattering of ultrasound is less than that of light from tissues, PA imaging can provide higher spatial resolution in deep tissue than optical imaging methods.^[9] Several reports have harnessed the high-resolution and deep penetration capability of PA signal to image the brain,^[10] delineate the boundary of cutaneous melanoma,^[11–12] perform ocular vasculature imaging,^[13–14] and demonstrate multispectral whole-body imaging of small animals.^[15–17] Most studies relied on endogenous chromophores such as oxy- and deoxyhemoglobin, lipids, and melanin for contrast enhancement.^[18–19] Driven by the need to understand the molecular underpinnings of cancer via imaging platforms, exogenous contrast agents are needed to enhance PA sensitivity to delineate molecular events and complement the limited structural and physiologic data endogenous chromophores provide.

Contrast-enhancing agents are expected to possess high molar absorptivity, photostability, and high conversion efficiency of absorbed light to heat.^[20–21] Examples of these imaging agents include metallic nanoparticles,^[1,22] carbonated particles,^[23–24] quantum dots,^[25] and organic dyes.^[26–29] Although nanoparticles can generate exceptionally high PA signals, the growing need to translate molecular photoacoustic imaging (MPI) to human subjects has stimulated interest in repurposing FDA-approved organic dyes such as indocyanine green (ICG) for PA imaging. In addition to their known safety profiles, these compounds have sharp absorption peaks in the near-infrared (NIR) window (650 to 900 nm), which is suitable for PA imaging of deep tissue. Unfortunately, the low photostability, fractional fluorescent components, and relatively low absorption coefficient compared to nanoparticles culminate into poor PA contrast. One approach to enhance PA signal is to develop chromophore molecular systems that can cooperatively quench the dye fluorescence and induce molecular aggregation to enhance thermal expansion in finite regions of the tumor microenvironment.

Carbon carbon (C–C) dimerization of cyanine molecules is an attractive approach to achieve complete fluorescence quenching, improve the absorption coefficient, and increase the local chromophore concentration while maintaining the small size that favors extravasation and excretion from the body. A previous study using electrochemical methods^[30] demonstrated one-electron oxidation of dicarbocyanine molecules^[31] to afford visible light-absorbing C–C dimers, among multiple other products. Further, while irreversible dimerization occurred at the even-numbered methane positions, the presence of substituents on the polymethine chains enhanced the reversibility of the dimers back to monomers. Given the constraints of electrochemical methods for large-scale production, a versatile boron trifluoride-catalyzed dimerization of functionalized dicarbocyanine molecules was shown to generate stable meso-substituted dimers. However, the absorption spectra of the dimers were broad and in the visible wavelengths, with two distinct maxima and a similar absorption coefficient at each maximum as the monomer. These characteristics negate the anticipated advantage of dimeric chromophore molecular systems for PA signal enhancement. Several reports have demonstrated that Cu(I) and Cu(II) are suitable catalysts for a variety of reactions, including the dimerization of aromatic compounds, typically in the presence of oxidants.^[32–33] Leveraging this phenomenon, we discovered a new chemical strategy to enhance the PA signal of carbocyanine molecules and used the concept to develop cancer-targeting dimeric chromophore systems for MPI.

Predicated on the versatility of Cu(II) to catalyze oxidative cross-coupling reactions and the susceptibility of the polymethine chains one-electron oxidation, we heated NIR light-absorbing tricarbocyanine molecules, cypate,^[34–35] IR780, and HITC (Figure 1) at 70°C for 24 h in a mixture of DMSO or DMF and ammonium acetate buffer containing CuCl₂. HPLC purification and analysis showed a major peak that was distinct from the starting monomeric dyes, but LC–MS (ESI) exhibited similar mass as the monomeric compound (Supplementary Methods; Table S1 and Table S2). Subsequent HRMS analysis of the product (illustrated with cypate product) showed a molecular ion peak at 1247.5892 Da, corresponding to C₈₂H₇₉N₄O₈, which is double the monomeric formula for cypate (C₄₁H₄₁N₂O₄) with the loss of 2 protons (Figure S1–S3). The result suggests the formation of a dimeric molecular interaction. Although we observed low levels of the dimeric products in solutions, the process was catalyzed by Cu(II) in different organic solvents and over a wide pH range, except when cypate decomposed above pH 8 (Figure S4).

Figure 1.

Chemical structures of dicarbocyanines and tricarbocyanines. Cypate, IR780, and HITC are monomeric tricarbocyanines, while cypate-3 and HIDC are dicarbocyanines. LS301 is a tumor-avid peptide-cypate conjugate.

To determine the point of dimerization, we performed 2D NMR analysis. Proton integration of cypate spectra showed seven protons in the polymethine chain of cypate (Figure S5–S6). Overlapped protons H14–H20, H15–H19 and H16–H18 are reminiscent of a symmetrical polymethine chain structure (see Figure S5 for numbering). Further analysis using ¹H- total correlated spectroscopy (TOCSY) revealed continuous spin propagation from H14 to H20 in the vinyl bridge (Figure S6). H1–C13 heteronuclear multiple-quantum correlation (HMQC) spectrum exhibited correlated resonances, suggesting the presence of seven C–H carbons in the polymethine chain (Figure S7). In comparison, the polymethine chain in the dimeric molecule displayed an unsymmetrical structure, particularly with H20 showing an upfield shift of 1.2 ppm from H14 (Figure S8). Proton H17 also showed a 0.5 ppm downfield shift compared to cypate H17 resonances. Furthermore, TOCSY resonances in the dimer break into two sections of spin propagation, H19–H20 and H14–H17, with the absence of H18 (Figure S8). Quaternary carbon C18 was confirmed by the absence of H1–C13 correlation in HSQC (Figure S9). Moreover, multiple bond correlations between C18 and H16, H17, H19, and H20 in heteronuclear multiple bond correlation (HMBC) indicated that C18 still resided in the chain and the substitution of H18 may contribute to the formation of cypate dimer (see Figure S10 for a full spectrum of ¹³C assignments).

We next explored if the conjugation of peptides with cypate will alter the dimerization. Using a tumor-avid cypate-peptide conjugate, LS301,^[36] we demonstrate that dimer formation achieved with about 60% yields for all the tricarbocyanines, pointing to a versatile Cu-mediated C–C dimerization of these chromophores. The TOCSY spectra of LS301 monomer and dimer (Figures S11–S12) followed a similar trend as cypate and its dimer, which was further confirmed by LCMS analysis (Figures S13). However, attempts to apply the method to dicarbocyanines (HIDC and cypate-3; Figure 1) was less successful, resulting in a complex mixture of products containing less than 1% of the HIDC and cypate-3 dimers. Probably, stability of the one-electron oxidation was low under the reaction condition, suggesting that Cu(II) catalyzed dimerization favors the less sterically hindered tricarbocyanine molecules. Fortunately, the NIR spectral properties of the tricarbocyanines are best suited for in vivo MPI because the reduced number of endogenous chromophores in this region enhances the selective absorption of light by the exogenous chromophores to boost PA signals for MPI.

Two characteristics of tricarbocyanine dimers and H aggregates are the enhancement of the 680 nm shoulder peak in the absorption spectra and loss of fluorescence via static quenching process. However, the spectral properties of C–C dimeric tricarbocyanines are not known. We found that the dimers exhibited a small but consistent 3–5 nm hypsochromic shift of the peak absorption, accompanied by a significant increase in the shoulder peak (Figure 2A, Figure S14). In the same solution, more than 90% fluorescence quenching was observed in all cases (Figure 2B and Figure S14). In general, tricarbocyanine H aggregates will dissociate in organic solvents such as DMSO, methanol, and DMF or different pH mediums. Our data show that the dimers are stable under these conditions (Figures S4 and S14). The spectroscopy results corroborate with the NMR analysis, supporting strong intermolecular interactions that are not readily reversible under conditions that disrupt H aggregates.

Figure 2.

Optical properties of cypate and LS301 monomers and dimers determined by spectroscopy and PA measurements. (A) Absorption spectra show hypsochromic shift in the dimer compared to monomer. (B) Fluorescence spectra show ~98% loss of fluorescence after dimerization. Spectra were measured in methanol. (C) Photoacoustic spectra of LS301 dimer, cypate dimer, LS301 monomer, cypate monomer, ICG, and 10% DMSO. (D) PA signal amplitude versus concentration of ICG, LS301 monomer and LS301dimer.

Previous attempts to repurpose NIR fluorescent tricarbocyanine dyes for in vivo PA utilized high dye doses of dyes such as ICG,^[37] a level that is not consistent with receptor-targeted imaging due to rapid saturation of available molecular targets. An alternative approach is to load the chromophores inside nanoparticles, which quench the dye fluorescence and enhanced PA signals.^[28] For ease of clinical translation, simple molecules that can extravasate into target tissue are preferred. We postulated that the newly prepared dimers could boost PA signal by minimizing radiative decay processes and creating a rich NIR chromophore microenvironment to maximize localized thermal expansion in the vicinity of the compounds. Achieving this goal would require enhanced photostability at high laser power used for PA imaging. Using variable LED fluence intensities, we found that the dimers were consistently more photostable in aqueous buffers than the monomers at equivalent dye concentrations (Figure S15). The superior stability of the dimer, including under high fluence intensity, provides additional advantages for its use in PA imaging, which employs high laser power. To test if these dimers are useful for MPI, we measured the PA signal amplitude at multiple laser wavelengths (680–950 nm), which generated the PA spectra of cypate and its dimer (Figure 2C). ICG, which is has been reported as a PA contrast agent,^[37] was also evaluated under similar conditions as cypate monomer and dimer. The shape of the PA spectra was similar to that of the absorption spectrum of the cyanine molecules, with the dimer exhibiting maximum PA signal amplitude at 800 nm (absorption peak at 780 nm). Comparison of the PA signal amplitude of cypate and cypate dimer at an equimolar concentration (200 μM) shows about a 3.6-fold increase for the dimer at 800 nm. A similar trend was observed for LS301 and its dimer, demonstrating the consistency of the signal enhancement regardless of whether the dye is free or conjugated. We also found that the rate of PA signal change vs. concentration of LS301 dimer was significantly higher than that of the monomer (Figure 2D). In all cases, equimolar amounts of the ICG displayed similar PA signals as cypate monomer. The observed synergistic generation of acoustic waves due to dimerization can be attributed to several factors, including fluorescence quenching and high photostability for the dimeric vs. the monomeric chromophores.

With the goal to explore in vivo application of the new molecular probe, we determined if the synergistic PA signal can be used for MPI in tumor-bearing mice. Previous studies using fluorescence imaging have shown that cypate clears rapidly from circulation and accumulates in the liver.^[38] However, the increased molecular weight of the dimers could improve the uptake in large tumors by the enhanced permeability and retention (EPR) effect. To that effect, we implanted a subcutaneous human lung cancer model (A549) in the dorsal flanks of female Balb/c nude mice. MPI was initiated after the tumors reached about 100 mm³ (Figure 3). The mouse was fixed on an animal heating pad to expose the dorsal plane for PA imaging. Given that the maximum PA intensity for both the monomers and dimers was highest at 800 nm in solution (Figure 2C), we performed the in vivo study at this nm excitation wavelength. Whole-body PA maximum amplitude projection (MAP) images of the tumor-bearing mice before injection of the imaging agents showed blood vessels based on hemoglobin absorption (Figure 3A). Subsequent intravenous injection of cypate dimer (200 μM, 120 μL) and PA imaging showed the dynamic distribution of the agent throughout the body. In addition to accumulating in some organs such as the liver, cypate dimer highlights an intricate vascular network throughout the body. Visualization of the vasculature beyond 6 h post-injection demonstrates that dimerization extended the blood resident time compared to the monomer, which was previously shown by fluorescence imaging to clear from circulation within 30 min after injection.^[38] Focusing on the tumor region, PA images highlighted peritumoral vessels with no noticeable increase in the tumor compared with the control image.

Figure 3.

Representative whole-body photoacoustic imaging of A549 tumor-bearing mice using (A) cypate dimer and (B) LS301 dimer at the indicated time points. Scale bar=5 mm. The inset image represents the magnified image of the tumor region. Two subfigures presented on the right are longitudinal cross-section B-mode PA image of liver and tumor, scale bar=5 mm and 2.5 mm, respectively. Liver and tumors region were marked with dark red and orange dotted circles, respectively. (C) A whole-body depth-encoded image 6-hour posterior to LS301 dimer injection. Scale bar=5 mm. Quantification of PA amplitude changes in the liver (D) and tumor (E) region over time before and after injection of LS301 dimer (n=3), cypate dimer (n=3), and LS301 monomer (n=3). Error bar=standard error.

To quantify the signal increase in the liver and the tumor region, we delineated the region of interest (ROI) by polygonal selection from the MAP image (Figure 3D for liver and 3E for tumor). The upper 10 to 30% of the selected pixels was used in data analysis to obtain statistically significant results by minimizing noise level. Consistent with previous fluorescence imaging reports, the liver PA signal increased rapidly, attaining a maximum of 5.4 fold at 2 h post-injection before gradually decreasing over time. Similarly, the PA signal level in the tumor region concurrently increased to 50% of its original level at 4 h after injection, which was followed by a rapid decrease over time. These results point to a biodistribution profile where a large amount of the cypate dimer was cleared through the hepatobiliary pathway, but the persistence of PA contrast in the blood vessels after 6 h provides evidence that dimerization enhanced vascular retention of the compound. As a result, the dimers had sufficient time to accumulate in tumors, which is not detectable with equimolar concentrations of the monomer. The peritumoral PA signal, which decreased rapidly after 4 h indicates that the non-tumor targeted dimeric cypate was unable to penetrate the tumor mass. Further, the high vascular and background signals decreased the effective tumor contrast (Figure 3A).

While the EPR effect is a practical approach to deliver contrast agents and drugs to tumors, molecularly targeted agents provide additional information about specific tumor-associated biomarkers. Using the same tumor model and imaging conditions described above for cypate dimer, we investigated the use of cancer-targeted agents for MPI. First, we inject the tumor-bearing mice with LS301 monomer (200 μM, 120 μL). We found that the PA signal was too weak for imaging purposes, demonstrating that the combination of EPR effect and binding to phosphorylated annexin A2, the target of LS301, was not capable of elevating the localized agent concentration for effective MPI (Figure 3 D,E). Although higher doses of LS301 monomer could overcome this impediment, receptor-targeted imaging methods rely on the bioavailability of a finite number of receptors to accumulate in tumors. As such, increasing the amounts of injected doses is less likely to offset the constraints imposed by the limited receptor density on tumor cells. To determine the benefits of dye dimerization strategy for MPI contrast agents, we evaluated the contribution of LS301 dimer compared to the monomer and cypate dimer.

Using a similar in vivo procedure described above, we observed high PA signal in the spleen of some mice before LS301 injection (Figure 3B, for example), probably due to high levels of the old or damaged red blood cell in the organ. Whole-body PA MAP images of the tumor-bearing mice after administering LS301 dimer (200 μM, 120 μL) showed significant PA signal increase throughout the mouse body, which persisted over 12 h (Figure 3B). A whole-body depth-encoded image of LS301 dimer demonstrates the differential distribution of the dimer in different organs at various depths at 6 h post-injection (Figure 3C). A time-dependent enhancement of PA signal amplitude in the liver (Figure 3D) and the tumor region (Figure 3E) was observed post-LS301 dimer administration. Unlike cypate dimer, the PA signal in the liver region was moderate at early time points but gradually increased over time, attaining a maximum at 6 h post-injection. Compared to the monomer, PA signal for LS301 dimer increased by 150% at 4 h post-injection. Using equivalent dose where the PA contrast for LS301 monomer was insignificant, we found that the PA signals for LS301 dimer in the tumor increased rapidly, attaining a maximum (~50% increase) at 6 h post-injection. Interestingly, both LS301 and cypate dimers exhibited similar tumor uptake profiles, suggesting that EPR effect was responsible for the initial localization in the tumor. However, significant differences were observed in the tumor-associated PA images. While the cypate dimer exhibited predominantly peritumoral distribution, PA signal for LS301 dimer was diffuse throughout the entire tumor mass, with a focal intratumoral intense signal. In contrast to cypate dimer, which rapidly decreased after 6 h, the PA signal for LS301 dimer in the tumor was maintained through the duration of the imaging study. Although the tumor sizes in the top view images shown in Figure 3A,B appear to be different, measurements of the tumors from different dimensions show that they are similar (Figure S16, Table S3). Because tumor cells were inoculated on the right hind leg position in all the mice, the 3D growth of the tumors is at an incline, which is not fully captured by the top view images shown in Figure 3A,B. The ensemble of these data highlights the contributions of EPR and cancer-targeting molecules to the uptake and retention of molecular dimers to enhance MPI.

In conclusion, we demonstrated a new Cu(II)-catalyzed method to synthesize C–C dimers of tricarbocyanines in >60% yields and successfully extended the approach to a dye-peptide conjugate, LS301, which enhanced the duration of PA signal retention in tumors. NMR and HRMS analyses revealed that the dimerization occurred at the C18–C18 polymethine chains, creating a stable, non-reversible C–C bond. While dimerization preserved the absorption spectra of the monomers, the fluorescence quenching was >90%, suggestive of static quenching. Our study revealed that a combination of fluorescence quenching and elevated local chromophore density within a confined tissue microenvironment increased the probability of light absorption and non-radiative relaxation of the excited molecules, which synergistically enhanced the PA signals. Whereas the acoustic waves generated by equimolar amounts of the monomeric dyes or dye-peptide conjugate were too low for MPI, the dimers exhibited sufficiently high PA signals for whole-body mouse imaging over 12 h. Although the mass equivalent of the dye per mole of dimers is twice that of the monomer, dimerization increased the net molecular weight of the dimers, thereby enhancing EPR effect compared to the monomers. This C–C approach to linking two dye molecules at the meso-position of the heptamethine chain preserved the π-π conjugation of the individual dyes while efficiently quenching the fluorescence for improved PA signal enhancement. Perhaps, the dimerization strategy will be advantageous in studies where a limited number of receptors are available for binding targeted molecular probes, thereby leveraging each binding event to deliver the equivalent of two dye molecules. These results point to a new MPI paradigm for interrogating cellular and molecular processes in vivo using small molecule dimers instead of large nanoparticles.

Experimental Section

See Supporting Information for details of the experimental methods. All animal studies were performed in accordance with the approved protocol by the Institutional Animal Care and Use Committee, Pohang University of Science and Technology (POSTECH) in Pohang, Republic of Korea (approval #: POSTECH-2017-0115-C2-R3).