Cyanine‐Flavonol Hybrids for Near‐Infrared Light‐Activated Delivery of Carbon Monoxide

Abstract Carbon monoxide (CO) is an endogenous signaling molecule that controls a number of physiological processes. To circumvent the inherent toxicity of CO, light‐activated CO‐releasing molecules (photoCORMs) have emerged as an alternative for its administration. However, their wider application requires photoactivation using biologically benign visible and near‐infrared (NIR) light. In this work, a strategy to access such photoCORMs by fusing two CO‐releasing flavonol moieties with a NIR‐absorbing cyanine dye is presented. These hybrids liberate two molecules of CO in high chemical yields upon activation with NIR light up to 820 nm and exhibit excellent uncaging cross‐sections, which surpass the state‐of‐the‐art by two orders of magnitude. Furthermore, the biocompatibility and applicability of the system in vitro and in vivo are demonstrated, and a mechanism of CO release is proposed. It is hoped that this strategy will stimulate the discovery of new classes of photoCORMs and accelerate the translation of CO‐based phototherapy into practice.


Materials and Methods
Reagents and solvents of the highest purity available were used as purchased, or they were purified/dried using standard methods when necessary. Synthetic procedures were performed under an ambient atmosphere unless stated otherwise.
Flash column chromatography was performed using silica gel (230−400 mesh). 1 H NMR spectra were recorded on 300 or 500 MHz spectrometers; 13 C NMR were obtained on 125 MHz or 75 MHz instruments in CDCl3, CD3OD, and d6-DMSO. 1 H chemical shifts are reported in ppm relative to tetramethylsilane ( = 0.00 ppm) using the residual solvent signal as an internal reference. 13 C chemical shifts are reported in ppm with CDCl3 ( = 77.67 ppm) and CD3OD ( = 49.30 ppm) as internal references. The deuterated solvents were kept under nitrogen atmosphere.
UV-vis spectra were obtained with matched 1.0 cm quartz cuvettes. Fluorescence was measured on an automated luminescence spectrometer in 1.0 cm quartz fluorescence cuvettes at 26 ± 1 °C. The corresponding optical filters were used to avoid the second harmonic excitation/emission bands induced by the grating. Sample concentrations with the absorbance below 0.1 at the excitation wavelength at the absorption maxima were used. Each sample was measured five times, and the spectra were averaged. Emission and excitation spectra were normalized and smoothed using standard protocols.
The exact masses of the synthesized compounds were obtained using a triple quadrupole electrospray ionization mass spectrometer in a positive or negative mode coupled with directinlet or liquid chromatographies.
General Procedure for Irradiation in UV Cuvettes. A solution of a compound in the given solvent (3 mL) in a matched 1.0 cm quartz PTFE screw-cap cuvette equipped with a stirring bar was stirred and irradiated with a light source of 32 LEDs (max = 770 or 820 nm at ~60 or 14 mW/cm -2 , respectively; Supplementary Fig. 67). The progress of the reactions was monitored at the given time intervals by UV−vis spectrometry using a diode-array spectrophotometer.
Determination of CO Yields. Stock solutions of 5a in CH2Cl2 or 5b in methanol or DMSO (c ~ 1×10 -4 M) were diluted with methanol or PBS (pH 7.4, 10 mM, I = 100 mM) to give a final concentration of c ~3-50 × 10 −6 M, so that the amount of a co-solvent does not exceed 2.5% (v/v). The solutions (100-1000 μL) in closed GC vials fitted with PTFE septa were irradiated with LEDs at 770 or 820 nm (~30 and ~7 mW cm -2 , respectively) for the given times or to complete conversion. The released CO was analyzed and quantified by a GC-headspace instrument (5Å molecular sieve packed column) equipped with a TIC/MS detector in a SIM mode, which was calibrated using the quantitative photoreaction of cyclopropenone photoCORM (50-600 μL, c ~ 1 × 10 −5 M, in methanol). 1 Determination of CO Release Quantum Yields. Solutions of 5a or 5b in methanol (1000 μL, c ~ 1 × 10 -6 M) in closed GC vials fitted with PTFE septa were irradiated by a xenon short-arc lamp through a monochromtator set to 791 and 793 nm, respectively. The samples were S3 irradiated through the bottom of the vial to minimise reflection of light. The absolute photon flux was measured by a calibrated Si-photodiode. The total molar amount of the released CO was analyzed by GC-headspace as described above and used to calculate the quantum yields of CO release.
Fluorescence Measurements. Fluorescence and excitation spectra were measured using a fluorescence spectrometer in a 1.0 cm quartz fluorescence cuvette at 23 ± 1 °C. The sample concentrations were adjusted to keep the absorbance below 0.2 at the corresponding excitation wavelength. Each sample was measured five times, and the spectra were averaged. Emission and excitation spectra were normalized and corrected by the photomultiplier sensitivity function using correction files supplied by the manufacturer.
Singlet Oxygen Production Quantum Yields. Solutions of 1,3-diphenylisobenzofuran (DPBF; c = 4 × 10 −5 ) and either one of 5a-b (c ~ 1 × 10 −6 M) or indocyanine green (ICG, c = 6 × 10 −6 M) as photosensitizers in methanol were prepared. The stirred solution (3.5 mL) in a quartz cell (1 cm) was irradiated using LEDs at 770 nm, and the UV−vis spectra were recorded periodically. The overall irradiation time was selected to reach <10% conversion of DPBF. The procedure was repeated 3 times. The decomposition of DPBF monitored at 411 nm was fitted with a pseudo-first-order rate law. The data were corrected for light absorbed by the samples, calculated as an integral of the overlap between the absorption and LED emission spectra. The singlet oxygen formation quantum yield ΦΔ was calculated using ICG as a reference (ΦΔ = 0.008 2,3 ) Quantum Yields of Decomposition. A solution of a commercially available IR-783 dye (c = 6.5 × 10 -6 M, 3.0 mL) in PBS (pH 7.4, 10 mM, I = 100 mM) was stirred and left to equilibrate for 2-3 min at 20 °C. Afterwards, irradiation using LEDs at 730 or 770 nm was initiated, and UV−vis spectra were recorded periodically. The total irradiation time was chosen to reach a <10% conversion and to obtain 10 experimental points. The procedure was repeated 3 times. The quantum yield of decomposition Φdec was calculated using the equation Eq. S1: where Φ REF is the decomposition quantum yield of the reference compound, Δn is the number of moles of the photodecomposed IR-783 dye calculated from the absorbance change at λmax, and I is the total amount of the light absorbed by a sample in the given time period, calculated according to the equation Eq. S2: where A(λ,t) is absorbance of the sample at the wavelength λ in time t, and Iλ em is the emission intensity of a LED source at the wavelength λ. Due to the lack of available chemical actinometers, photooxygenation of DPBF by ICG (ΦΔ = 0.008 2,3 ) was used to measure the photon flux.

S4
Reaction Rate with 1 O2. A solution containing 14 (c = 5 × 10 −5 M) or 4 (c = 4 × 10 −6 M) and rose bengal (RB, c = 1 × 10 −5 M) as a singlet oxygen sensitizer in glycine buffer (pH 9.8, 20 mM) solution (3.0 mL) in a quartz cell (1.0 cm) was stirred and irradiated with LEDs at 545 nm. The UV-vis absorption spectra were recorded periodically. The total irradiation time was selected to reach a <10% conversion of 14B (or 4) and to obtain 10 experimental points. The procedure was repeated 3 times. The bimolecular reaction rate constant for 14B (or 4) with 1 O2 (kr) was calculated from the equation Eq. S3: where Δn is the number of moles of the photodecomposed 14B or 4, calculated from the absorbance change at λmax, c is the initial concentration of 14B or 4, kd is the known rate constant of singlet oxygen quenching in methanol (kd = 9.7 × 10 4 s −1 ) 4 or PBS (kd = 2.5 × 10 5 s −1 ), and I is the total amount of the light absorbed by the sample in the given time period calculated according to Eq. S2. The bimolecular rate constant kr for 14B and 4 was found to be 5.8 × 10 8 and 1.7 × 10 7 M −1 s −1 , respectively, determined relative to that of 1B (kr = 4.7 × 10 8 M −1 s −1 ). 5 In Vitro Toxicity Determination. Human hepatoblastoma HepG2 cells were grown according to the standard protocol in a 96-well plate. After reaching 80% confluence, cells were treated with a solution of 5b or its photoproducts in the concentration range of 6−200 μM (with 2% DMSO) for 2, 6, or 24 h. The cell viability was determined by a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (MTT). The absorbance of a formazan-reduction product was measured at 545 nm with a standard ELISA reader. Compound 5b was found not to interfere with the reduction of MTT.
In Vivo Experiments. Male nude SKH1 mice were allowed water and standard granulated diet ad libitum. Mice were anesthetised and then received an intraperitoneal injection of saline with 5% DMSO (a control group) or a solution of 5b (50 μmol kg -1 of body weight) in saline (10 µL g -1 , with 5% DMSO). An experimental group of mice was irradiated with 780 nm LED (4 × 500 mW) focused to the abdominal area for 2 h. Subsequently, the animals were sacrificed. Blood from the superior vena cava of each animal was transferred to sodium heparin-coated test tubes containing aq. EDTA (2 μL, 0.5 M). The CO (as COHb) and total hemoglobin concentrations in the sample of blood (1 μL) were determined using GC and a Drabkin cyanmethemoglobin method described previously. 6 Selected organs of each animal were then harvested, washed, put into an ice-cold potassium phosphate buffer (pH 7.4, 100 mM) in a ratio of 1:4 (w/w) and homogenised by sonication. The homogenate (40 μL) was added to CO-free septum-sealed vials containing sulfosalicylic acid (5 μL, 60% v/w). After incubation on ice (30 min), the amount of the CO released into the vial headspace was determined by GC/RGA. All studies in this work met the criteria for the care and use of animals and were approved by the Animal Research Committee of the 1 st Faculty of Medicine, Charles University, Prague.

S5
Statistical Analysis. Normally distributed data are presented as the mean ± SD and analyzed by the Student t-test. The Mann-Whitney U test was used in skewed data expressed as a median ± interquartile range. Differences in P ˂ 0.05 were considered significant.
S64 Figure S63. The depiction of photoproducts resulting from the irradiation of 5a. The primary photoproducts related to the release of CO are depicted in red (15 and 16). The products of benzoate ester hydrolysis in the dark are depicted in black (17 and 18). The secondary photoproducts, which follow the general photooxygenation pathway of the cyanine scaffold, are depicted in blue.  Figure S71. Irradiation of 14 in glycine buffer (pH 9.8, 10 mM) at 465 nm followed by UV-vis spectroscopy at 30-s intervals (from blue to red).

Competition Between CO Release and Cyanine Photooxygenation
The rate of photooxygenation of cyanine 4 can be expressed as: where kr is the rate of decomposition, [4] is the concentration of 4, n(4) is the amount of cyanine 4 in moles, V is the sample volume, k∑ is the bimolecular reaction rate of 4 and singlet oxygen, and [ 1 O2] is the concentration of singlet oxygen.