Effects of Substituents on Photophysical and CO-Photoreleasing Properties of 2,6-Substituted meso -Carboxy BODIPY Derivatives †

: Carbon monoxide (CO) is an endogenously produced signaling molecule involved in the control of a vast array of physiological processes. One of the strategies to administer therapeutic amounts of CO is the precise spatial and temporal control over its release from photoactivatable CO-releasing molecules (photoCORMs). Here we present the synthesis and photophysical and photochemical properties of a small library of meso -carboxy BODIPY derivatives bearing different substituents at positions 2 and 6. We show that the nature of substituents has a major impact on both their photophysics and the efﬁciency of CO photorelease. CO was found to be efﬁciently released from π -extended 2,6-arylethynyl BODIPY derivatives possessing absorption spectra shifted to a more biologically desirable wavelength range. Selected photoCORMs were subjected to in vitro experiments that did not reveal any serious toxic effects, suggesting their potential for further biological research.

It has been found that, despite its overt toxicity when present in high concentrations, carbon monoxide (CO) as an endogenously produced signaling molecule can have beneficial effects in various physiological processes and cellular functions, including apoptosis, proliferation, and inflammation at sub-micromolar concentrations (≈0.2 µM) [19][20][21].
ficial effects in various physiological processes and cellular functions, including apoptosis, proliferation, and inflammation at sub-micromolar concentrations (≈0.2 μM) [19][20][21]. When considering its therapeutic applications, sufficiently low concentrations of CO must be administered in a controlled manner to avoid its toxic effects. One CO delivery strategy is based on metal carbonyl complexes that release a weakly bound CO by simple hydrolytic ligand exchange upon dissolution in aqueous media [22,23]. Another approach is the use of photochemically activatable CO-releasing molecules (photoCORMs), which allow precise spatial and temporal control over its release in tissues [24][25][26][27][28]. Visible/NIR light has a reduced tissue penetration due to high optical scattering and strong absorbance by endogenous chromophores, such as hemoglobin or melanin [29,30]; therefore, photo-CORMs activatable in this wavelength range are highly desirable. Most of the reported visible/NIR light-absorbing photoCORMs are metal carbonyl complexes, but several transition-metal-free (organic) photoCORMs, such as xanthene-9-carboxylic acid [31] or flavonol derivatives [32][33][34][35], have also been designed and studied. Some of us have introduced meso-carboxy BODIPY-based photoCORM derivatives 1a and b that can release CO upon irradiation at wavelengths of up to ≈750 nm (Scheme 1) [36]. The density functional theory (DFT) calculations suggested that a strained α-lactone intermediate, formed upon irradiation via a triplet biradical, is responsible for the subsequent CO liberation. Scheme 1. Carbon monoxide photorelease from meso-carboxy BODIPY derivatives.
In this work, we prepared several meso-carboxy BODIPY derivatives of 2, compounds 3-10 ( Figure 1), as potential new BODIPY-based photoCORMs, and studied their photophysical and photochemical properties. We evaluated the effects of various electron-donating and electron-withdrawing substituents at positions 2 and 6 on their absorption and emission properties, and the efficiencies of CO release and singlet oxygen production upon irradiation. The selected derivatives with the highest yields of CO were subjected to cell culture experiments to determine possible cytotoxic effects to assess their potential for future therapeutic use. Scheme 1. Carbon monoxide photorelease from meso-carboxy BODIPY derivatives.
In this work, we prepared several meso-carboxy BODIPY derivatives of 2, compounds 3-10 ( Figure 1), as potential new BODIPY-based photoCORMs, and studied their photophysical and photochemical properties. We evaluated the effects of various electrondonating and electron-withdrawing substituents at positions 2 and 6 on their absorption and emission properties, and the efficiencies of CO release and singlet oxygen production upon irradiation. The selected derivatives with the highest yields of CO were subjected to cell culture experiments to determine possible cytotoxic effects to assess their potential for future therapeutic use.

Results and Discussion
Synthesis: Methyl and benzyl esters of carboxylic acid derivative 2, compounds 11 and 12, were prepared from 2,4-dimethylpyrrole and the corresponding chlorooxalate in 34 and 31% chemical yields (Scheme 2), respectively, using a modified procedure previously described [37]. Compound 12 was treated with POCl3 in DMF to give aldehyde 13 as a synthetic precursor, which was subsequently converted to carboxylic acid 14 in the presence of NaClO2 and NH2SO3H in 55% yield (Scheme 3). Compound 13 was used for the preparation of oxime 15 by the reaction with hydroxylamine hydrochloride and sodium hydroxide in ethanol (52% yield), and then the resulting 15 was treated with oxalyl chloride in acetonitrile to give 2-cyano derivative 16 in 78% yield (Scheme 3).

Results and Discussion
Synthesis: Methyl and benzyl esters of carboxylic acid derivative 2, compounds 11 and 12, were prepared from 2,4-dimethylpyrrole and the corresponding chlorooxalate in 34 and 31% chemical yields (Scheme 2), respectively, using a modified procedure previously described [37]. Compound 12 was treated with POCl 3 in DMF to give aldehyde 13 as a synthetic precursor, which was subsequently converted to carboxylic acid 14 in the presence of NaClO 2 and NH 2 SO 3 H in 55% yield (Scheme 3). Compound 13 was used for the preparation of oxime 15 by the reaction with hydroxylamine hydrochloride and sodium hydroxide in ethanol (52% yield), and then the resulting 15 was treated with oxalyl chloride in acetonitrile to give 2-cyano derivative 16 in 78% yield (Scheme 3).

Results and Discussion
Synthesis: Methyl and benzyl esters of carboxylic acid derivative 2, compounds 11 and 12, were prepared from 2,4-dimethylpyrrole and the corresponding chlorooxalate in 34 and 31% chemical yields (Scheme 2), respectively, using a modified procedure previously described [37]. Compound 12 was treated with POCl3 in DMF to give aldehyde 13 as a synthetic precursor, which was subsequently converted to carboxylic acid 14 in the presence of NaClO2 and NH2SO3H in 55% yield (Scheme 3). Compound 13 was used for the preparation of oxime 15 by the reaction with hydroxylamine hydrochloride and sodium hydroxide in ethanol (52% yield), and then the resulting 15 was treated with oxalyl chloride in acetonitrile to give 2-cyano derivative 16 in 78% yield (Scheme 3). Scheme 4. Synthesis of 2,6-diphenyl (18) and 2,6-diethynyl (19)(20)(21) esters.
The target meso-carboxy BODIPY derivatives 2 and 4-10 were obtained by the deprotection of the corresponding methyl esters using lithium iodide or by catalytic hydrogenation from the benzyl esters in good isolated yields (65-92%; Scheme 5). 2,6-Dibromo analog 3 was prepared by the reaction of 2 with NBS in 49% yield (Scheme 5). The target meso-carboxy BODIPY derivatives 2 and 4-10 were obtained by the deprotection of the corresponding methyl esters using lithium iodide or by catalytic hydrogenation from the benzyl esters in good isolated yields (65-92%; Scheme 5). 2,6-Dibromo analog 3 was prepared by the reaction of 2 with NBS in 49% yield (Scheme 5).  and em max  of 6, bearing an electron-withdrawing group at position 2, are slightly hypsochromically shifted. The molar absorption coefficients (ε) were found to be in the range of 3.2-5.8 × 10 4 M -1 cm -1 , which is common for BODIPY chromophores [11]. The type of the solvent (methanol , Table 1, and PBS, Table 2) had only a marginal effect on the abs max  , em max  , and ε values. The solvent and properties of the substituents at positions 2 and 6 had a substantial effect on the fluorescence quantum yields (Φf). Except for compound 2, all Φf values were found to be higher in methanol; indeed, increasing the solvent polarity leads to a low fluorescence efficiency of BODIPY derivatives [12]. Besides, the Φf was found to be relatively small for bromo-and iodo-derivatives 3 and 4 due to the presence of heavy atoms (i.e., an efficient ISC). An efficient nonradiative decay must also be responsible for the small values of Φf in the case of 2-cyano (6) and 2phenylethynyl (9 and 10) derivatives. This is consistent with the low Φf found for 2-phenylethynyl [38,41] or 2,6-arylethynyl [40] BODIPY derivatives. Analogous meso-alkenyl substituted BODIPYs have also been reported to be practically non-fluorescent [39,42]. It has been explained by the effect of large stabilization upon excitation along with the bending of the fused BODIPY core, and the accessible S1/S0 conical intersection point [16]; the small geometrical evolution prompts the nonradiative relaxation to the ground state [43]. We also cannot exclude fluorescence quenching by internal charge transfer (ICT) in derivatives 9 and 10 [42,44,45]. Different amounts of DMSO were added to PBS to dissolve the BODIPY derivatives because, except for 10 (the p-phenyl polyethylene glycol substituents were introduced in 10 to improve its solubility in aqueous solutions), they are only par- Both λ abs max and λ em max of 6, bearing an electron-withdrawing group at position 2, are slightly hypsochromically shifted. The molar absorption coefficients (ε) were found to be in the range of 3.2-5.8 × 10 4 M -1 cm -1 , which is common for BODIPY chromophores [11]. The type of the solvent (methanol, Table 1, and PBS, Table 2) had only a marginal effect on the λ abs max , λ em max , and ε values. The solvent and properties of the substituents at positions 2 and 6 had a substantial effect on the fluorescence quantum yields (Φ f ). Except for compound 2, all Φ f values were found to be higher in methanol; indeed, increasing the solvent polarity leads to a low fluorescence efficiency of BODIPY derivatives [12]. Besides, the Φ f was found to be relatively small for bromo-and iodo-derivatives 3 and 4 due to the presence of heavy atoms (i.e., an efficient ISC). An efficient nonradiative decay must also be responsible for the small values of Φ f in the case of 2-cyano (6) and 2-phenylethynyl (9 and 10) derivatives. This is consistent with the low Φ f found for 2-phenylethynyl [38,41] or 2,6-arylethynyl [40] BODIPY derivatives. Analogous meso-alkenyl substituted BODIPYs have also been reported to be practically non-fluorescent [39,42]. It has been explained by the effect of large stabilization upon excitation along with the bending of the fused BODIPY core, and the accessible S 1 /S 0 conical intersection point [16]; the small geometrical evolution prompts the nonradiative relaxation to the ground state [43]. We also cannot exclude fluorescence quenching by internal charge transfer (ICT) in derivatives 9 and 10 [42,44,45]. Different amounts of DMSO were added to PBS to dissolve the BODIPY derivatives because, except for 10 (the p-phenyl polyethylene glycol substituents were introduced in 10 to improve its solubility in aqueous solutions), they are only partially soluble in aqueous solutions. The pK a of non-substituted meso-carboxy BODIPY derivatives has previously been determined to be 4.7; the absorption band of the conjugate acid is bathochromically shifted by ≈100 nm [36]. The pH titration of 9 in a PBS/methanol (95:5, v/v) solution showed that the signal of the conjugate acid appears at ≈680 nm and a pH below 5 ( Figure S99). BODIPY derivatives were present only as conjugate base forms in solutions under the experimental conditions used in this work.   Time-resolved spectroscopy: We performed nanosecond transient absorption (TA) spectroscopy of selected BODIPY derivatives in both aerated and degassed PBS/DMSO mixtures (c ∼ 10 −5 M; λ exc = 532 nm) to identify long-lived intermediates. Derivatives 3 and 4 gave strong transient signals with λ max at ≈436 nm (Figures S102-S111). A prominent ground state bleach with λ max = 506 and 530 nm (Figures S102, S103, S107 and S108) for 3 and 4, respectively, was also observed. The kinetic traces were fitted to a first-order rate law (the oxygen concentration in an aerated solution at 20 • C (∼2.7 × 10 −4 M [48]) was almost 30 times higher than that of a BODIPY derivative) to provide the lifetimes of 320 ns and 25 µs for 3 and 256 ns and 11.40 µs for 4 in aerated and degassed solutions, respectively (Table S1), which is in good agreement with the lifetimes reported for analogous BODIPY systems [36]. We assigned these signals to the triplet states [49]. Derivatives 2 and 7 did not show any apparent signals in the range typical for the triplet-excited BODIPYs; only fluorescence and ground-state bleach signals were detected (Figures S100, S101, S112 and S113); thus, triplet state concentrations were below the detection limit of our TA spectroscopy setup.
Intersystem crossing efficiency and singlet oxygen production: As the photochemical release of CO from meso-carboxy BODIPYs was reported to occur from the triplet-excited state [36], we evaluated the ISC efficiency (Φ ISC ) by quantitative analysis of the transient optical density [50,51] using nanosecond TA spectroscopy. Triplet state transient signals of 3 and 4 in methanol were sufficiently strong to obtain the Φ ISC values of 0.66 and 0.83, respectively. Besides, we evaluated quantum yields of the singlet oxygen production (Φ ∆ ) for selected derivatives in methanol using 1,3-diphenylisobenzofuran (DPBF) as a 1 O 2 trap (Table 1). The high Φ ∆ values of 0.59 and 0.72 for 3 and 4 ( Figures S95 and S96), respectively, bearing heavy halogen atoms directly attached to the BODIPY core, match those of analogous BODIPY derivatives with halogen or chalcogen atoms in various core positions [8,11,14,15]. These values are only slightly lower than those of Φ ISC , thus Φ ∆ can be used advantageously to estimate the lower limit of Φ ISC . As anticipated, a small Φ ∆ was found for compounds 2, 5, 6, 8, and 9 ( Figures S97 and S98).
Photorelease of CO: We evaluated the quantum yields of photochemical degradation (Φ r ) of selected derivatives in a PBS/DMSO mixture ( Table 2). In all cases, the efficiencies were greater in the absence of oxygen because of quenching of the reactive triplet state. Besides, the Φ r value for 3 (58 × 10 −4 ) was larger than that of 8 by two orders of magnitude (0.7 × 10 −4 ), indicating that the triplet state is considerably more photochemically active, as previously discussed [36]. Because the efficiency of photodegradation of all BODIPY derivatives in methanol was too low (irradiation of 8 in methanol under the same conditions led to a ≈5% conversion in 12 h, for example), we have not investigated their photochemistry in this solvent in detail.
Compounds 2-10 in PBS/DMSO mixtures were found to produce different amounts of CO upon irradiation at the corresponding λ abs max ( Table 2; Figure 2 and Figures S77-S94). The chemical yields obtained upon exhaustive photolysis were rather small for BODIPYs 2-7 (<15%) but considerably higher for phenethynyl derivatives 8-10 (up to 45% upon complete conversion in a degassed solvent). As expected (see above), the yields dropped approximately by half in aerated solutions. However, compounds 3 and 4 with an enhanced ISC, which could promote a more efficient CO release from the triplet state [36], did not liberate CO in high yields. Besides, lower chemical yields of CO found in degassed rather than aerated solutions for compounds 4 and 6 (Table 2) suggest that at least one of the competing degradation pathways is independent of the presence of oxygen.  The introduction of an electron-withdrawing group (6) or an additional carboxy group (5) at position 2 did not improve the CO yields. According to the proposed mechanism of the CO release from meso-carboxy BODIPYs 1a, b (Scheme 1), electron-withdrawing groups should enhance electron transfer from the carboxylate (with pKa of ≈4.7 [36]; the acids are fully dissociated at pH = 7.4) to the triplet-excited BODIPY core. However, Therefore, we analyzed the photoproducts formed upon irradiation of 10 in a PBS/DMSO mixture by high-resolution mass spectra (HRMS). Based on the proposed photoproduct structures (Supplementary Materials), we suggest that the CO production competes with bleaching of the starting material with the generated singlet oxygen [36] (for example, by oxidative cleavage of the triple bond or ring opening of the BODIPY core, reported before in [52]), or the compound undergoes a photoinduced attack of water as a nucleophile at position 3 [53] or an exchange of fluorine(s) by the OH group(s). Figure 2 shows that a photoproduct formed upon irradiation of 10, characterized by a slight hypsochromic shift of the main absorption band, is further consumed upon continuing irradiation. Because the absorption maximum of this intermediate is similar to that of the starting material, the compound must still retain a BODIPY or BODIPY-like chromophore (Supplementary Materials).
The introduction of an electron-withdrawing group (6) or an additional carboxy group (5) at position 2 did not improve the CO yields. According to the proposed mechanism of the CO release from meso-carboxy BODIPYs 1a, b (Scheme 1), electron-withdrawing groups should enhance electron transfer from the carboxylate (with pK a of ≈4.7 [36]; the acids are fully dissociated at pH = 7.4) to the triplet-excited BODIPY core. However, we could not verify this hypothesis because the competing dye degradation in both the presence and absence of oxygen was more efficient than the CO release. A low CO yield in 2 was also unexpected because analogous BODIPY derivatives with the unsubstituted 1,7-positions (1, Scheme 1) were reported to liberate CO in up to 87% yields in degassed solutions [36]. The methyl substituents at positions 2 and 6, which were introduced to improve the stability in the dark (Figures S74-S76) and simplify the synthesis of meso-carboxy BODIPYs, thus must play a detrimental role in the CO photorelease. This behavior may be related to an out-ofplane geometry of the BODIPY core induced by the steric hindrance of a meso-substituent and the 1,7-dimethyl groups that are responsible for a more efficient nonradiative decay via "butterfly motion" discussed for analogous substituted BODIPY derivatives [24,54].
Fortunately, π-extended derivatives 8-10 not only provided improved chemical yields of the CO formation but their absorption bands were bathochromically shifted toward the biologically desired longer wavelengths. Compounds 8 and water-soluble 10 were therefore selected for the assessment of the cellular cytotoxicity before their use in further biological studies.
Cellular toxicity: To determine the cellular toxicity of compounds 8 and 10 and their corresponding photoproducts, an MTT test was used to assess cell viability on the human hepatoblastoma HepG2 cell line. HepG2 cells were incubated with selected photo-CORMs or their respective photoproducts for 2, 6, and 24 h. Compound 8 displayed no cytotoxicity within the concentration range of 12.5-100 µmol L -1 ( Figure S116), whereas compound 10 showed no effects on cell viability up to the concentration of 50 µmol L -1 ( Figure S117). Their photoproducts were found to be non-cytotoxic in the concentration range of 12.5-200 µmol L -1 (Figures S117 and 118). These cytotoxicity data are comparable to those observed for meso-carboxy BODIPY photoCORMs 1a, b (Scheme 1) [36].

Experimental Section
Materials: Reagents and solvents of the highest purity available were used as purchased, or they were purified/dried using the standard methods when necessary.
Methods: The lowest possible intensity of incident light was used in the spectroscopic identification of the samples to prevent their photodegradation. All measurements were accomplished using fresh solutions prepared in the dark. 1 H and 13 C NMR spectra were obtained in CDCl 3 , CD 2 Cl 2 , DMSO-d 5 , or CD 3 OD on 75, 125, 300, and 500 MHz spectrometers (Bruker AVANCE III (300 MHz) and AVANCE III HD (500 MHz) spectrometers). 1 H chemical shifts are reported in ppm relative to the tetramethylsilane signal (TMS, δ = 0.00 ppm) using the residual solvent signal as an internal reference. 13 C NMR chemical shifts are reported in ppm relative to the solvent signal as an internal standard. High-resolution mass spectra (HRMS) were recorded on an Agilent 6224 Accurate-Mass TOF LC-MS instrument using ESI or APCl techniques. Absorption spectra and the molar absorption coefficients were obtained on a UV-vis spectrometer with matched 1.0 or 0.1 cm quartz cells. Molar absorption coefficients were determined from the absorption spectra; the average values were obtained from three independent measurements with solutions of different concentrations. No dependence of the molar absorption coefficient on the sample concentration was observed in the range from 1 × 10 −4 to 1 × 10 −6 M. All glassware was oven-dried before use. Purification procedures were performed using silica gel (Merck 60; 230-400 mesh) columns or by recrystallization.
Determination of CO yields: A solution of compounds 2-10 (100-10 mM) in a PBS (I = 0.1 M, pH ≈ 7.4)/DMSO solution was irradiated with LEDs emitting at the corresponding wavelengths (λ max = 490, 525, or 545 nm) in closed GC vials fitted with PTFE septa to complete the conversion of the starting material. The released CO into the vial headspace was determined by a GC-Agilent 5973 Mass Selective Detector headspace technique, which was calibrated using the photoreaction of cyclopropenone photoCORM [55] (50−500 µL, c ≈ 5 × 10 -4 M, in methanol).
Decomposition quantum yields: The quantum yields of the decomposition of BODIPY derivatives in both aerated and degassed (purged with argon for 20 min) PBS/DMSO solutions were determined at λ irr = 525 nm (LEDs) using the BODIPY derivative 22 [36] as an actinometer dissolved in PBS (I = 0.1 M, pH = 7.4) according to the published procedure [36]. All quantum yield measurements were repeated five times with independently prepared samples.
obtained in CDCl3, CD2Cl2, DMSO-d5, or CD3OD on 75, 125, 30 eters (Bruker AVANCE III (300 MHz) and AVANCE III HD (50 chemical shifts are reported in ppm relative to the tetramethyls ppm) using the residual solvent signal as an internal reference are reported in ppm relative to the solvent signal as an internal mass spectra (HRMS) were recorded on an Agilent 6224 Accu strument using ESI or APCl techniques. Absorption spectra an efficients were obtained on a UV-vis spectrometer with matche Molar absorption coefficients were determined from the absor values were obtained from three independent measurements concentrations. No dependence of the molar absorption coeffic tration was observed in the range from 1 × 10 −4 to 1 × 10 −6 M. All before use. Purification procedures were performed using sil mesh) columns or by recrystallization.
Determination of CO yields: A solution of compounds 2-1 = 0.1 M, pH ≈ 7.4)/DMSO solution was irradiated with LEDs em wavelengths (λmax = 490, 525, or 545 nm) in closed GC vials fitte plete the conversion of the starting material. The released CO in determined by a GC-Agilent 5973 Mass Selective Detector he was calibrated using the photoreaction of cyclopropenone phot ≈5 × 10 -4 M, in methanol).
Decomposition quantum yields: The quantum yields of th IPY derivatives in both aerated and degassed (purged with arg solutions were determined at λirr = 525 nm (LEDs) using the BO an actinometer dissolved in PBS (I = 0.1 M, pH = 7.4) according t [36]. All quantum yield measurements were repeated five time pared samples.
Fluorescence quantum yields: Fluorescence quantum yiel Edinburg Instrument FLS920 fluorimeter as the absolute va Singlet oxygen production quantum yields: Quantum yields for the singlet oxygen production, sensitized by BODIPYs derivatives in methanol at 525 (compounds 2, 3 and 4), 490 (compounds 5 and 6), and 545 nm (compounds 8 and 9), were determined by monitoring the photooxidation of 1,3-diphenylisobenzofuran (DPBF) using rose bengal (RB) [46] as a reference sensitizer. For derivatives 5 and 6, compound 3 was used as a reference. A solution of DPBF (c = 5 × 10 −5 ) and either BODIPY (c = 1 × 10 −5 M) or RB (c =5 × 10 −6 M) sensitizers in methanol was prepared. The stirred solution (3 mL) in a 1 cm quartz cell was irradiated using LEDs at a selected wavelength, and the UV−vis spectra were recorded periodically. The irradiation time was chosen to reach approximately 10% conversion of DPBF. The procedure was repeated five times.
Fluorescence quantum yields: Fluorescence quantum yields were determined on an Edinburg Instrument FLS920 fluorimeter as the absolute values using an integrating sphere. The quantum yields were measured three times and were averaged for each sample. The solution concentrations were kept low (A < 0.1). The fluorescence quantum yields were determined in methanol or a PBS (pH = 7.4, 10 mM, I = 0.1 M)/DMSO mixture.
Transient spectroscopy: The nanosecond laser flash spectroscopy setup was generally operated in a right-angle arrangement of the pump and probe beams. Laser pulses of ≤170 ps or 700 ps duration at 532 nm (20-240 mJ) were obtained from an Nd:YAG laser. The laser beam was dispersed onto a 40 mm long and 10 mm wide modified fluorescence cuvette held in a laying arrangement. An overpulsed Xe arc lamp was used as a source of the probe light. Kinetic traces were recorded using a photomultiplier. Transient absorption spectra were obtained using an ICCD camera equipped with a spectrograph. The samples were degassed by three freeze-pump-thaw cycles under reduced pressure (0.01 Torr). Absorption spectra of the sample solutions were measured regularly between laser flashes to test for possible photodegradation of the solution components using a diode-array spectrophotometer [56].
Intersystem crossing quantum yield: The ISC efficiency (Φ ISC ) was evaluated by a quantitative analysis of the transient optical density (see Supplementary Materials for details) [50,51] using nanosecond TA spectroscopy for selected derivatives in degassed methanol solutions (three freeze−pump−thaw cycles) at three different concentrations (3.0, 6.0, and 10.0 × 10 −6 M).
Cellular toxicity experiments: Human hepatobastoma HepG2 cell line (ATCC, Manassas, VA, USA) was used to test the cytotoxicity. The cells were grown in supplemented MEM media in 96-well plates according to the manufacturer´s instructions. The cells were kept at 37 • C and 5% CO 2 atmosphere during the experiment. The stock solution was prepared by dissolving a BODIPY derivative in DMSO; the final solutions were prepared by dissolution of stock solutions with Minimal Essential Medium (MEM). The final concentration of DMSO in media did not exceed 1%. The corresponding photoproducts were prepared by exhaustive irradiation of the solutions at λ irr = 505 nm for 24 h. Viability was determined using an MTT test (colorimetric assay based on the reduction of a yellow tetrazolium salt (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) as described before [57]. All experiments were performed at least in triplicates.
Synthesis of meso-carboxy BODIPYs: general procedure A. A methyl ester of mesocarboxy BODIPYs (1 equiv.) was dissolved in dry ethyl acetate and LiI (10 equiv.) was added. The reaction mixture was heated to reflux for 16 h under a nitrogen atmosphere. TLC was used to monitor the reaction. When the reaction was finished, the mixture was cooled to room temperature, and a small amount of HCl (0.2 mL) was added to quench the reaction. Water was added, and the mixture was extracted with ethyl acetate (3 × 10 mL). The combined organic layers were washed with water (20 mL), dried over anhydrous sodium sulfate, filtered, and concentrated to dryness under reduced pressure. The compounds were purified by flash chromatography on silica gel.

Conclusions
We report the synthesis and photophysical and photochemical properties of a series of 2,6-substituted meso-carboxy BODIPY derivatives designed as photoactivatable CO-releasing molecules (photoCORMs). The results provide valuable insights into the structural and electronic factors that affect their photoreactivity. We show that the methyl substituents at positions 1 and 7, originally introduced to improve the chemical stability of BODIPY derivatives, play an unfavorable role in the release of CO. Neither the enhancement of intersystem crossing by heavy-atom substituents nor the decrease in the electron density thanks to electron-withdrawing substituents improved the CO yields. However, CO was efficiently released from π-extended 2,6-arylethynyl BODIPY derivatives with absorption spectra shifted toward a more biologically desirable wavelength range. Subsequently, in vitro cytotoxicity experiments with the most potent meso-carboxy BODIPY photoCORMs and their photoproducts did not reveal any major toxic effects, which justifies them for further biological studies.