J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 953-961 Photophysical Properties of Merocyanine 540 Derivatives Andrew C. Benniston and Anthony Harriman" Center for Fast Kinetics Research, The University of Texas at Austin, Austin, Texas 78712,USA Kirpal S. Gulliya Baylor Research Institute, Baylor Medical Center, 3812 Elm St., Dallas, Texas 75226,USA Photophysical properties have been measured for five merocyanine dyes having (i) different alkyl substituents on the thiobarbiturate subunit [i.e. ethyl (butyl Merocyanine UO), or hexyl], (ii) the sulfonate group replaced with a methyl group, or (iii) the benzoxazole residue replaced with 1,2-naphthoxazole, and for one of the correspond- ing oxonol derivatives. Neither the nature of the alkyl group nor the absence of the sulfonate group exert any significant effect on the photophysical properties of the dyes in ethanol solution at 22 "C.However, extending the size of the benzoxazole residue increases fluorescence and triplet yields and decreases the yield for photoiso-merization. A detailed two-dimensional NMR analysis of one of the dyes has shown that the ground state exists exclusively in an all-trans conformation while FTlR indicates hydrogen bonding between the thiobarbiturate subunit and the polymethine bridge. Using this structural information, the energetics and mechanics of the isomerization processes are discussed. It is further shown that these structural modifications affect the efficacy with which the dyes kill leukemic cells under illumination due to pronounced changes in lipophilicity.Merocyanine 540 shows considerable potential as a photo- sensitizer for selective eradication of leukemia and related viral contaminants from blood products'.' and as a radiation sensitizer for treatment of solid turn our^.^ Consequently, several research groups have reported on the photochemical properties of this dye in fluid solution and in micro-heterogeneous Fluorescence is moderately intense but, in common with most cyanine dyes, intersystem crossing to the triplet manifold is extremely inefficient, and the domi- nant photoprocess involves isomerization of the first excited singlet state to form a long-lived (unstable) isomer. The photophysical properties, and in particular the rate of isom-erization, of such merocyanine dyes have been found to depend on both viscosity and polarity of the so1vent.",l2 Isomerization is assumed to occur via trans-cis intercon-version and, as such, might be expected to involve large-scale torsional motion since the terminal subunits are quite b~1ky.l~Also, the ionized sulfonic acid residue, which pro- vides modest solubility in water and which anchors the dye close to an aqueous surface when dissolved in a lipid mem- brane,I4 might be involved in specific interactions with protic solvents. On close scrutiny, however, it appears that little is known about the intimate details of the isomerization process and, in fact, the conformation of the ground-state molecule remains unresolved.This is unfortunate because the compound might be converted into a much more potent photosensitizer if isomerization could be inhibited.With this in mind, we sought to explore deeper into the mechanism of the isomer- ization process by the study of derivatives of Merocyanine 540 having different alkyl chains attached to the thiobarbitu- rate subunit, having an extended benzoxazole residue, and having the water-solubilizing sulfonic acid residue removed. It is shown that such structural modifications have little effect on the rates of isomerization and it is concluded that the isomerization process involves only a modest structural per- turbation. Furthermore, the sulfonic acid group plays no obvious role in controlling the dynamics of isomerization in homogeneous solution, but replacing the benzoxazole moiety with a second thiobarbiturate subunit, forming an oxonol, results in greatly enhanced rates of isomerization. Since the ultimate objective of this work is to design improved photo- sensitizers for removal of the leukemia virus from blood and bone marrow, some attention has been given to their in situ concentration and photochemical behaviour.Experimental Merocyanine 540 (2) was obtained from Eastman-Kodak and purified by column chromatography on silica using acetone- methanol (98:2) as eluent. For the purified material in ethanol solution, the absorption maximum (Amax) was at 560 nm and the molar absorption coefficient at the peak maximum (cmax) was 167000 dm3 mol-' cm-'.The general synthetic route used to prepare merocyanine dyes has been reported previously' ',16 and was used for the preparation of the new compounds studied here. Interestingly, in the prep- aration of 4, a minor component was separated by column chromatography and identified as 5 on the basis of 'H NMR, 13C NMR and mass spectrometry; a sample of 5 was subse- quently synthesized by direct self-condenation of the thio- barbiturate residue. Starting materials were purchased from Aldrich Chemicals or Eastman-Kodak and were used as received. Water was double-distilled and deionized with a Millipore purification system. Ethanol (Aaper, Absolute grade) was used as received. All other solvents were spectro- scopic grade, used as received, or were redistilled under vacuum. Analytical data for the new compounds are provid- ed here; all compounds were purified by extensive chroma- tography on silica and it should be noted that the reported yields were not optimised.Compound 1 Yield: 0.3 g (33%). 'H NMR (C2H6]-dimethyl sulfoxide): S = 1.14-1.19 (6 H, m); 2.11 (2 H, m);2.58 (2 H, m); 4.40- 4.43 (6 H, m); 6.49-6.54 (1 H, d, J = 13.7 Hz); 7.49 (2 H, m); 7.74-7.84 (4 H,m); 7.98 (1 H, m). FAB MS (nitrobenzyl alcohol matrix): 491 (M+). UV-VIS (C,H,OH): A,,Jnm = 560 (&,,Jdm3 mol-' cm-' = 168000). FTIR (KBr disc): 1630; 1670 cm- ' (CO stretch). Compound 3 Yield 1.4 g (74%). 'H NMR (CDCl,): 6 = 0.86 (6 H, t, J = 7.2 Hz); 1.13-1.19 (9 H, t, J = 7.2 Hz); 1.28 (12 H, s); 1.60 (4 H, br); 2.11 (2H, m); 2.51-2.57 (2 H, t, J = 6.7 Hz); 3.04-3.13 (6 H, t, J = 7.2 Hz); 4.33-4.47 (4 H, m); 4.56 (2 H, m); 6.49-6.54 (1 H, d, J = 13.5 Hz); 7.17 (1 H, d, J = 8 Hz); 7.46-7.52 (2 H, q, J = 6.4 Hz); 7.73-7.84 (3 H, m); 7.91-7.99 (1 H, m).FAB MS (nitrobenzyl alcohol matrix): 602 [M-(C,H,),NH+]. UV-VIS (C,H,OH): A,,Jnm = 560 (&,Jdm3 mol-' cm-' = 178000). FTIR (KBr disc): 1630; 1670 cm-' (CO stretch). Compounds 4 and 5 Purification by chromatography on silica of the crude mixture, eluting first with ethyl acetate-hexane (1:1) gave 4 and 5 as impure products due to incomplete separation. Further purification of 4 on a second column, eluting with ethyl acetate-hexane (3 : 7), gave a deep blue solid, which appeared as a single spot on TLC (R, = 0.38).Yield: 300 mg (19%). 'H NMR (CDCl,): 6 = 0.88-0.91 (6 H, dt, J = 7.3 Hz, J' = 2 Hz); 0.93-0.96 (3 H, t, J = 7.3 Hz); 1.31-1.41 (6 H,m); 1.62-1.68 (4 H, m); 1.72-1.78 (2 H, m); 3.86-3.89 (2 H, t, J = 7.3 Hz); 4.39-4.42 (4 H, t, J = 7.8 Hz); 5.54-5.64 (1 H, d, J = 12.5 Hz); 7.09-7.11 (1 H, dd, J = 7.7 Hz. J' = 1 Hz); 7.23-7.26 (1 H, dt, J = 7.8 Hz, J' = 1.2 Hz); 7.28-7.31 (1 H, dt, J = 7.7 Hz, J' = 1 Hz); 7.39-7.41 (1 H, d, J = 8 Hz); 7.77- 7.82 (1 H, t, J = 13 Hz); 7.84-7.89 (1 H, t, J = 12.9 Hz); 7.95- 7.98 (1 H, d, J = 13 Hz). FAB MS (nitrobenzyl alcohol matrix): 482 (M+). UV-VIS (C,H,OH): A,,Jnm = 560 (&,,Jdrn3 mol-' cm-'= 165000). FTIR (KBr disc): 1630; 1670 cm-' (CO stretch). Similarly, purification of 5 on a second column, eluting with acetone-methanol (9 :l), gave a red solid, which appeared as a single spot on TLC (R, = 0.10).Yield: 30 mg (2%). 'H NMR (CDCl,): 6 = 0.9 (12 H, m); 1.25-1.31 (8 H, m); 1.61 (8H,m); 4.32 (8H, m); 7.89 (2 H, d, J = 13 Hz); 8.30 ( 1 H, t, J = 13 Hz). 13C NMR (CD,CN): J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 14.01; 20.92; 21.09; 30.10; 47.82; 48.26; 103.37; 120.30; 160.72; 161.27; 162.87; 179.72. FAB MS (nitrobenzyl alcohol matrix): 547 (M'). UV-VIS (C,H,OH): A,,,/nm = 540 (&,,Jdm3 mol-' cm-' = 142000). FTIR (KBr disc): 1630; 1670 cm-' (CO stretch). A further sample of 5 was synthe- sized by direct condensation of an equimolar mixture of NYN'-dibutyl-2-thiobarbituricacid and 5-(3-methoxypropen- 1,3-diene)-N,N'-dibutyl-2-thiobarbituricacid in methanol containing excess triethylamine.Identical analytical data were found for the two samples. Compound 6 Yield 1.5 g (35%). 'H NMR (CDCl,): 6 = 0.86-0.93 (6 H,t, J = 7.2 Hz); 1.12-1.18 (9 H, t, J = 7.2 Hz); 1.24-1.32 (4 H, m); 1.56 (4 H,m); 2.17-2.19 (2 H, m); 2.62-2.65 (2 H, m); 3.03-3.12 (6 H, q, J = 7.2 Hz); 4.29 (4 H, m); 4.58 (2 H, m); 6.56-6.62 (1 H, d, J = 13.8 Hz); 7.61-7.81 (4 H, m); 7.96-8.14 (4 H, m); 8.32-8.36 (1 H, d; J = 8.2 Hz). FAB MS (nitrobenzyl alcohol matrix): 699 (MH'); 597 [M(C,H,),NH+]. UV-VIS (C,H,OH): A,,Jnm = 572 (&,,Jdm3 mol-' cm-' = 164000). FTIR (KBr disc): 1630; 1670 cm- '(CO stretch). 'H NMR spectra were recorded with Bruker AC250 or General Electric GN-500 FT-NMR instruments with TMS as internal standard.Absorption spectra were recorded with a Hitachi U3210 spectrophotometer and fluorescence spectra were recorded with a fully corrected Perkin-Elmer LS5 spec- trofluorimeter. Solutions for fluorescence studies were adjust- ed to possess an absorbance of <0.05 at the excitation wavelength. Singlet excited-state lifetimes were measured by time-correlated, single-photon-counting techniques using a mode-locked, synchronously pumped, cavity-dumped Rho-damine 6G dye laser. The excitation wavelength was 565 nm and fluorescence was isolated from scattered laser light with a high-radiance monochromator. The instrumental response function was ca. 60 ps and was deconvoluted from the experi- mental decay profile prior to data analysis.The fluorescence lifetime for oxonol 5 was measured with a synchronous streak camera following excitation at 532 nm with a 30 ps laser pulse. ca. 500 individual laser shots were averaged and analysed by computer iteration after deconvolution of the instrument response. Flash photolysis studies were made with a frequency-doubled Quantel YG481 Nd : YAG laser (pulse width 10 ns; pulse energy 70 mJ). Solutions were adjusted to possess an absorbance of ca. 0.2 at 532 nm and were purged with N,, 0, or air. Transient differential absorption spectra were recorded point-by-point with five individual laser shots being averaged at each wavelength. Kinetic studies were made at fixed wavelength with 50 individual laser shots being aver- aged and analysed by computer non-linear, least-squares iter- ative procedures.Where appropriate, the laser intensity was attenuated with crossed-polarizers. In several cases it was necessary to restrict the intensity of the monitoring beam to a low level in order to avoid photolysis of the photoi~orner.~ This attenuation was achieved by placing appropriate neutral density filters before the sample cell. Differential absorption coefficients for the excited triplet state were measured by the energy-transfer method using anthracene as donor. Anthra- cene was excited at 355 nm and the triplet state was moni- tored at 422 nm assuming a differential absorption coefficient of 52000 dm3 mol-' cm-'." The derived differential absorption coefficients for the excited triplet states at the respective maxima were as follows: 1, 45000; 2, 42000; 3, 44OOO; 4,45000; 5,28000 and 6,41000 dm3 mol-' cm-'.Differential absorption coefficients for the unstable isomers were measured by the complete bleaching method in 0,- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 saturated ethanol, with due allowance for triplet population. The values derived for the absorption peak maxima were as follows: 1, 15000; 2,17000; 3,16000; 4, 18000; 5,15000 and 6, 17000 dm3 mol-' cm-I. In all cases, the laser intensity was calibrated using zinc meso-tetraphenylporphyrin in benzene as standard.' '3'' The temperature was controlled using a heated metal block and was measured with a thermo-couple in direct contact with the solution.Time-resolved thermal lensing studies were made with a conventional setup2' using a Melles Griot 40 mW laser diode emitting at 835 nm as the probing beam. Output from the diode laser was separated from the excitation pulse with a narrow bandpass filter and directed through a 0.3 mm pinhole into the entrance slits of a New Focus low-noise silicon photocell. For each measurement, the intensity of the excitation laser pulse was varied using crossed-polarizers and the intensity of the probing beam was attenuated with glass neutral density filters immediately before entering the detec- tor. The instrument was calibrated using tris(2,2'-bipyridyl) ruthenium(I1) dichloride in deoxygenated ethanol, for which the triplet energy was taken as 196 kJ mol-', as measured by luminescence spectroscopy.Solutions were adjusted to possess an absorbance of 0.40at 532 nm and 40 individual traces were averaged before computer analysis. Partition coefficients were determined by dissolution of 1 mg of dye in a mixture of chloroform (10cm3) and a 2% (w/w) aqueous solution of human serum albumin (10 cm3). After sonication and standing for 24 h, a further 10 cm3 of chloroform were added and the layers separated. The concen- tration of dye in each layer was measured by absorption spectroscopy and the partition coefficient was defined as the molar ratio of dye in water relative to chloroform. Diffusion coefficients were measured in 50% (w/w) aqueous glycerol using the fluorescence photofading method developed by Axelrod et aL2' The solution was examined with a Zeiss Axiovert 35 inverted microscope and excitation was provided by a 10 ns laser pulse at 532 nm (70 mJ) focussed through the optics of the microscope.Recovery of the initial fluorescence signal, due to Brownian motion, was monitored at 610 nm. A suspension of Daudi cells (1 x lo6 cells ~m-~) in growth medium was mixed with a merocyanine dye (20 pg cm-3). After 30 min of incubation at 37"C, the suspension was placed in Falcon petri dishes (35 x 10 mm) and exposed to 514 nm light delivered with an argon ion laser. After irradia- tion for varying times, the cells were washed twice with RMPI- 1640 culture medium, resuspended in growth medium, and incubated overnight at 37°C.After 22 h of incubation, cell viability was determined by the trypan blue exclusion method.22 Comparative experiments were made with dye only and light only controls. The cell-killing efficacy is report- ed in terms of the log (reduction), which is defined as the logarithm of the number of surviving cells divided by the number of cells exposed to irradiation. For fluorescence microscopy studies, Daudi cells were stained with dye, washed and aliquots were placed on microsope slides with a cover slip on top. This wet mount was examined by polarized fluorescence spectroscopy under a Zeiss Axiovert 35 inverted microscope with 514 nm excitation and 600 nm emission wavelengths.The observed fluorescence intensity, in arbitary units, was compared to that for Merocyanine 540under iden- tical conditions. Results Ground-state Conformation of 4 A detailed understanding of the isomerization processes can be made only if the conformation of the ground state is known with certainty. Previous molecular mechanics MM2 calculations and resonance Raman results3 suggested that the ground state of MC540 possessed an all-trans conformation, but convincing structural data are lacking. A full structural analysis was undertaken, therefore, for merocyanine 4 in CDCl, solution using high-resolution and two-dimensional NMR techniques. The similarity of absorption and fluores- cence spectra and the close comparability of the photo- physical properties of merocyanines 1-4 and 6 suggest to us that all of these dyes possess identical conformations. High-field (500 MHz) 'H NMR COSY spectra were recorded for 4 in CDCl, and are shown in Fig.1 and 2. Com- plete assignment of all the protons in the molecule was pos- sible and the derived chemical shifts and coupling constants are collected in Table 1. In particular, the alkenic protons could be clearly identified and assigned. Thus, H(9) was unambiguously identified from its characteristic chemical shift (6 = 5.59) and doublet pattern. For this proton the mea- sured coupling constant, J, was found to be 12.5 Hz. Three other resonances were observed to possess very similar J values; 6 = 7.80 (J = 13.0 Hz), 6 = 7.87 (J = 12.9 Hz) and 6 = 7.96 (J = 13.0 Hz) (Fig.1). These latter three resonances are assignable, therefore, to the remaining three alkenic protons; the observed patterns of two triplets and a doublet being consistent with this assertion. Individual peak assign- ment could be made from the two-dimensional spectra which indicated mutual coupling between adjacent protons (Fig. 2). The second doublet (6 = 7.96) was clearly due to H(12) while the two triplets were assignable to H(10) (6 = 7.87) and H(11) (6 = 7.80). Since the coupling constants found for the alkenic protons were essentially the same, we can conclude that a single type of double bond prevails. On this basis, the ground-state conformation must be either all-trans or all-cis.Furthermore, since an all-cis arrangement cannot be accom- modated for merocyanine 4, we conclude that the only acceptable ground-state conformation is all-trans. Detailed analysis of all the resonances indicated that only this single isomer was present to the observable limit (~95%). The average magnitude of the coupling constant for the alkenic protons (J = 12.9 Hz) is significantly lower than the expected value for an isolated trans double bond (J NN 17 Hz); the corresponding isolated cis double bond has an expected coupling constant of about 10 Hz.*~ This finding is consistent with each of the carbon atoms in the polymethine bridge pos- sessing partial double-bond character, as expected for a merocyanine dye which can exist in zwitterionic resonance forms.Indeed, the coupling constants for individual protons Table 1 'H NMR spectral properties recorded for merocyanine 4 in CDCl, solution atom label proton shift," 6 coupling constant,b J/Hz 16, 16' 0.884.91 7.3, 2.0 8 0.93-0.96 7.3 -15, 15' 1.3 1-1.35' 7 1.36-1.41' 14, 14 1.62-1.68' -6 1.72-1.78' 5 3.86-3.89 7.3 13, 13' 4.39-4.42 7.8 9 5.54-5.64 12.5 1 7.09-7.1 1 7.7 3 7.23-7.26 7.8, 1.0 2 7.28-7.31 7.7, 1.2 4 7.39-7.41 8.0 11 7.77-7.82 13.0 10 7.84-7.89 12.9 12 7.95-7.98 13.0 ~ ~~ a kO.1 ppm referenced to TMS. kO.1 Hz. 'Too much resonance overlap to evaluate coupling constant. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 ;3y 15’ 16 7\ 8 ’* . rL 0 b-,LL1 0t ‘1 0 6 5 4 3 26 High-resolution(500 MHz) twodimensional COSY NMR spectrum recorded for 4 in CDCl, showing the atom labelling Fig.1 are seen to increase as that proton nears the thiobarbiturate subunit, owing to an increased electron density. This effect can be interpreted in terms of zwitterionic structures formed by electron donation from the N atom in the benzoxazole subunit to one of the carbonyl groups in the thiobarbiturate subunit.24 -o R ao&rfS Y :+ R In the all-trans conformation, the carbonyl groups on the thiobarbiturate subunit are coplanar with the polymethine bridge and are well positioned for intramolecular hydrogen bonding. Indeed, FTIR spectra indicate the existence of such hydrogen bonding in the solid state, as evidenced by a broad absorption centred at 3400 cm-’.The carbonyl groups appear as a broad peak centred at 1630 cm-’and a some-what less intense and sharper peak centred at 1670 an-’.For a vinylogous amide carbonyl group in the absence of hydro- gen bonding, we would expect2’ to observe the CO stretching band at 1670 cm-’.Hydrogen bonding, of medium strength, is expectedZS to lower this frequency by about 30 cm-’. On s Fig. 2 The (500 MHz) twodimensional COSY NMR spectrum recorded for 4 in CDCl, showing the expanded aromatic and alkenic regions J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 this basis, the carbonyl band centred at 1670 cm-' is assign- ed to a non-hydrogen-bonded group and the corresponding peak centred at 1630 cm-' is attributed to the hydrogen- bonded counterpart.Photophysical Properties of 3 in Ethanol The photophysical behaviour of the various merocyanine dyes 1-4 followed a common pattern, as illustrated below for the hexyl derivative 3. All measurements were made in ethanol solution at 22°C under an atmosphere of air, nitrogen or oxygen. Some of the photophysical properties of the butyl derivative 2 (i.e. Merocyanine 540) have been reported previo~sly~-'~and the major goal of this study is to evaluate the effect of structural changes on the rates of isomerization. Ultimately, these results should be of value for the design of improved photosensitizers for use in photodynamic therapy. Note that previous work has established that minor structur- al changes can be accompanied by significant differences in the efficacy for light-induced cytotoxicity.26 Absorption and fluorescence spectral profiles recorded for 3 in dilute ethanol solution are shown in Fig.3. The absorp- tion spectrum shows a sharp peak centred at 560 nm [Emm = (178000 2 12000) dm3 mol-' cm-'1 while the fluorescence excitation spectrum gave a good match to the absorption spectrum over the entire visible spectral range. The Stokes' shift, measured in dilute ethanol solution, was (660 f40) cm-', indicating the absence of a substantial geometric change upon promotion to the first excited singlet state. The fluoresence quantum yield (@J, measured relative to Rho- damine 101 in ethanol,,' was found to be (0.16 f0.02) and the fluoresence lifetime (zs) was measured to be (415 20) ps.The radiative lifetime calculated from the Strickler-Berg expression2' (zo = 2.7 ns) was in good agreement with that determined experimentally (zo = = 2.6 ns). Laser flash photolysis studies carried out in N,-saturated ethanol solution indicated the presence of two transient species after excitation with a 10 ns laser pulse at 532 nm (Fig. 4). The shorter-lived transient, which aborbs predomi- nantly around 660 nm, was assigned to the triplet excited state of the dye on account of its reactivity towards molecular oxygen. The longer-lived transient, which absorbs principally at 595 nm, did not react with oxygen and, on the basis of wavelengthlnm Fig.3 Absorption and fluoresence spectra recorded for merocya- nine 3 in dilute ethanol solution. The excitation wavelength used for the fluorescence spectrum was 540 nm. previous ' was assigned to a geometric isomer formed by rotation around one of the polymethine double bonds. The differential absorption spectrum of this species (Fig. 4) shows that the isomer absorbs somewhat to the red of the ground state. The isomer was observed to be extremely photolabile and absorbed both the exciting and analysing pulses. Under low-intensity illumination, the lifetime of the unstable isomer was found to be (6.1 f0.5) ms. The differen- tial molar absorption coefficient for the isomer at 595 nm was estimated to be (1.6 f0.2) x lo4 dm3 mol-' cm-', by com- plete conversion in 0,-saturated solution, and this allowed determination of the quantum yield for formation of the isomer (0as (0.45 & 0.06).Isomerization occurs exclusively from the first excited singlet state. The triplet state decayed with a lifetime (z,) of (700 _+ 50) ps in thoroughly deaerated ethanol solution and was quenched by 0, with a bimolecular rate constant of (1.3 f0.3) x lo9 dm3 mol-' s-'. The triplet state was also populated via energy transfer from triplet anthracene in ethanol solution following laser excitation at 355 nm. From these latter studies, the bimolecular rate constant for triplet energy trans- fer was found to be (4 f1) x lo9 dm3 mol-' s-' and the differential molar absorption coefficient for the triplet stateof 3 at 680 nm was found to be (4.4 _+ 0.5) x lo4 dm3 mol-' cm-'.The triplet state, as formed by energy transfer, exhibited the same differential absorption spectrum asthat generated by direct excitation of 3 and decayed cleanly to the prepulse baseline. On the basis of actinometric laser flash photolysis ~tudies,''~~~ the quantum yield for population of the triplet excited state (0,)was determined to be (0.0030 f0.0006). wavelengthlnm t \II I 1 I I I I 1 I I I-aw! 4505#)5506006Hl7#]7sl8#) wavelengthlnm Fig. 4 Differential absorption spectra recorded 1 ps after excitation of merocyanine 3 in (a) deoxygenated and (b) oxygenated ethanol solution with a 10 ns laser pulse at 532 nm. (a) Spectra for both triplet state and unstable isomer; (b) spectrum of the isomer alone. Table 2 Photophysical properties measured for the various merocyanine dyes in dilute ethanol solution 1 560 0.163 430 0.0035 775 0.42 6.9 2 560 0.160 410 0.0030 820 0.40 6.7 3 560 0.170 415 0.0031 495 0.45 6.1 4 560 0.150 380 0.0030 210 0.47 3.7 5 540 0.092 125 0.00084 635 0.65 0.012 6 572 0.180 530 0.0038 780 0.24 42.0 Absorption maximum, +2.+_ 12%.' &20ps. +50 ps. f10%. Comparison of the Photophysical Properties of Merocyanines Comparable photophysical properties were observed for the other merocyanine dyes in ethanol solution and there were no major differences in the magnitude of any of the measured parameters.Thus, the absorption maximum of the ground- state dye remained at 560 nm and the Stokes' shift was invariant at (660 & 60)cm-'. The fluorescence quantum yield and excited-singlet-state lifetime remained insensitive to the length of the alkyl substituent on the thiobarbiturate subunit and to the presence of the sulfonato group (Table 2). Simi-larly, the quantum yield for formation of the triplet excited state and of the unstable isomer did not change, within experimental error. Lifetimes measured for the triplet and unstable isomer were similar, if not identical, throughout the series (Table 2). It is clear, therefore, that these structural changes are not manifest in a significant variation in the photophysical properties, at least in ethanol at 22 "C.Photophysical Properties of Compound 5 Compound 5 is an oxonol that is somewhat soluble in water. The absorption and fluorescence spectra recorded for 5 in ethanol are shown in Fig. 5. The absorption maximum, which is centred at 540 nm, is situated some 20 nm to the blue of that found for merocyanines 1-4. The fluoresence spectrum shows reasonable mirror symmetry with the absorption spec- trum, while the Stokes' shift of (620 _+ 30) cm-' indicates little structural difference between ground and excited states. Both the quantum yield of fluorescence and the fluorescence lifetime are significantly lower than those of the merocyanine dyes (Table 2). In oxygenated ethanol, the unstable isomer was formed in high quantum yield (Table 2).The differential absorption spectrum recorded for the unstable isomer is shown in Fig. 6 and is similar to those observed for the merocyanine dyes, although the maximum absorption now , lDL,\11 1 I I wavelength/nm Fig. 5 Absorption and fluorescence spectra recorded for oxonol5 in dilute ethanol solution. The excitation wavelength used for the fluo- rescence spectrum was 520 nm. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1 I D I 1 D h Q, C me -m P W I I I D I I I I I I I I 4505005506m69m'150800 wavelength/nm Fig. 6 Differential absorption spectra recorded (0)1 ps and (A)60 ps after excitation of oxonol 5 in deoxygenated ethanol solution with a 10 ns laser pulse at 532 nm. The shorter-lived species is attrib- uted to the unstable isomer and the longer-lived species is believed to be the triplet excited state.appears at 570 nm and the spectral features are considerably broader. The isomer was found to possess a relatively short lifetime compared with those of the merocyanine dyes (Table 2). In contrast, the triplet excited state, which was formed in extremely low quantum yield in deoxygenated ethanol (Table 2), retained a long lifetime. The differential absorption spec- trum recorded for the triplet (Fig. 6) differs from those found for the merocyanine dyes in that the absorption peak lies further into the near-infrared region and the differential molar absorption coefficient measured at the absorption maximum = (2.8 & 0.6) x lo4 dm3 mol-' cm-'1 is sig-nificantly lower.Photophysical Properties of Merocyanine 6 The general properties determined for 6 correspond very closely to those recorded for merocyanines 1-4. Extension of the benzoxazole subunit causes a modest red shift in the absorption and fluorescence maxima and exerts a small influ- ence on the various photophysical properties (Table 2). It appears that the increased size of the naphthoxazole subunit results in a decreased rate of isomerization and this effect is manifest in increased fluorescence and triplet yields and a sig- nificant decrease in the quantum yield for formation of the unstable cis isomer. The cis isomer possesses a relatively long lifetime, indicative of the increased size of the rotor.Activation Energies and Energetics for Merocyanines 4 and 5 The enthalpy difference between the ground-state trans isomer and the unstable cis isomer was measured for merocy- anine 4 and oxonol 5 by time-resolved thermal lensing tech- niques.20 Using the <Di values given above, the enthalpy changes accompanying conversion of the cis isomer to the trans isomer were found to be (75 f5) and (120 f10) kJ mol-', respectively, for 4 and 5 in ethanol solution. Clearly, the cis isomer for 5 is considerably less stable than that of 4. Activation energies for conversion of the cis isomers to trans isomers (EL) were measured by recording the relevant rate constants at different temperatures in ethanol solution and expressing the results in terms of conventional Arrhenius plots.The derived values for these thermal processes were (120 f10) and (60 f6) kJ mol-', respectively, for 4 and 5. The corresponding activation energies for photoinduced trans to cis isomerization (El) were measured by recording the J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 fluorescence lifetime as a function of temperature and were found to be (24 & 3) and (19 f3) kJ mol-', respectively, for 4 and 5. It is seen that the excited-state barriers are very much less than those for the thermal reactions. The activation energies can be considered as linear com- binations of several terms. For merocyanine 4, where zwitter- ions persist, the measured activation energy will contain activation energy terms associated with solvent viscosity (E,), solvent polarity as expressed by the EA30) parameter, and twisting of the molecule (EJ.E, = E,, -p[Ed30) -301 + E, Here, expresses the fraction of the E-,-(30) parameter that is involved in a polarity-dependent activation energy.29 For oxonol 5, where zwitterions are not expected, the activation energy is assumed to be independent of solvent polarity. For ethan01,~' E, = 9.2 kJ mol-' and [Ed30) -30) = 91.5 kJ mol-'. Previously, the /lterm was measured for a structur-ally related merocyanine dye" and found to be 0.11 and -0.09, respectively, for the photoinduced and thermal isom- erization processes. On this basis, the activation energy associated with rotation around the central double bond (EJ has values of 25 and 10 kJ mol-', respectively, for light- induced trans to cis isomerization of 4 and 5. The corre- sponding values for thermal cis to trans isomerization of 4 and 5, respectively, are 103 and 51 kJ mol-'.It is clear from these values that both isomerization processes are much easier in 5 than in 4. Light-induced Cytotoxicity Studies The general photophysical behaviour of merocyanines 14 is quite similar, although the lipophilicity changes among the various compounds. This latter feature was measured experi- mentally by determining the partition coefficient (P) for dye distributed between aqueous human serum albumin and CHC1,; 1 (log P x 4.0), 2 (log P = 1.18), 3 (log P = 0.36), 4 (log P = -2.28), 5 (log P = 1.41) and 6 (log P = -1.88).The dyes demonstrate a clear preference for the aqueous protein, except for 4 and 6 which show markedly increased lipo- philicity relative to Merocyanine 540 (2). The diffusion coeffi- cients (D) were measured in aqueous glycerol, using the fluorescence photofading method, and the values are com- piled in Table 3. It can be seen that, under these conditions, the values remain very similar. The cellular concentration of merocyanine dye after incu- bation with Daudi cells was determined by polarized fluores- cence microscopy and the values are collected in Table 3. The more lipophilic derivatives are seen to exhibit the highest preference for localization within the cellular system while the most hydrophilic derivative shows little tendency to assimi- late within the cell.The percentage of cells surviving after Table 3 Partition coefficients, diffusion coefficients, relative in situ concentrations and cell-killing efficacies measured for the various dyes compound log P D/107an's-' [dye]" 10g(red[40])~ log(redC901)' 1 4.0 2.2 14 0.08 0.10 2 1.18 2.0 100 0.37 1.30 3 0.36 1.9 115 0.40 1.60 4 -2.28 2.1 290 1.22 5.40 5 1.41 2.0 - - - 6 -1.88 1.7 260 1 .oo 4.30 Relative in situ dye concentration measured by polarized fluorescence. Light-induced cell-killing efficacy in terms of log (reduction) as measured after exposure to 40 J cn-*. Light-induced cell-killing efficacy in terms of log (reduction) as measured after exposure to 90 J exposure to 40 and 90 J cm-' illumination at 514 nm was also measured for each of the dyes and the results are col- lected in Table 3.It is seen that, within experimental limi- tations, there is a direct correlation between the in situ dye concentration and the efficacy for light-induced cytotoxicity. Discussion It is apparent from the results collected in Table 2 that the photophysical properties of merocyanine 540 show little dependence on the nature of the alkyl substituents appended to the thiobarbiturate group. Similarly, the sulfonate group does not influence the photophysical properties of the dye in ethanol solution. In particular, lifetimes measured for the excited singlet state and for the unstable isomer remain com- parable, within experimental limitations, for merocyanines 14.This finding means that the rates of both light-induced (forward) and thermal (reverse) isomerization steps are essen- tially independent of the nature of these substituents. Some- what similar results have also been observed16 for thermal isomerization of merocyanine dyes based on benzimidazole, where the unstable isomer is much shorter lived than those found for merocyanines 1-4. We must conclude, therefore, that isomerization does not involve large-scale torsional motion of the thiobarbiturate subunit and that the sulfonate group does not associate closely with a solvent cluster. These findings are consistent with a model in which the thiobarbiturate subunit is held in place by hydrogen bonding between a carbonyl group and a proton in the polymethine bridge.This has the effect of inhibiting rotation around the C(5)-C(6) bond in the transition state and effectively pre- vents isomerization at this bond. It is expected that rotation around the C(l)-C(2) bond will be partially inhibited by steric interaction between the substituent attached to the ben- zoxazole N atom and H(10)and, in any case, isomerization at this bond seems unlikely to account for the observed optical absorption changes. Instead, rotation of the benzoxazole subunit around the C(3)-C(4) bond, leading to formation of the corresponding cis isomer (Scheme I), appears to give a Y IIhv li S H H--0 R Scheme 1 satisfactory account of all our experimental data.In the absence of specific solvation at the sulfonate group, the molar volumes of the rotors in merocyanines 1-4 are not too dis-similar; the molar volumes of the rotating groups31 in 1 and 4, respectively, being ca. 135 and ca. 120 cm3 mol-'. The slightly faster rates of isomerization found for 4 relative to 1-3 are consistent with this modest change in volume of the rotor. The photophysical properties observed for merocya-nine 6 also appear to be consistent with this scheme. Thus, the molar volume3' of the rotor in 6 is increased to ca. 180 cm3 mol-' and this has the effect of decreasing the rates of isomerization relative to merocyanines 1-3. For the oxonol 5 the negative charge should be considered as being delocalized over all four oxygen atoms, giving rise to the two extreme resonance forms shown in Scheme 2.Again, the non-ionized thiobarbiturate subunit is locked in place by hydrogen bonding, as evidenced by FTIR spectroscopy, and this has the effect of stabilizing the trans conformation. However, because of extensive electron delocalization, each of the C-C bonds in the polymethine bridge can be considered as being of 1.5 bond order. Photoisomerization occurs at the C(3)-C(4) bond, forming the corresponding cis isomer (Scheme 2). The rates of both forward and reverse isomer-izations are significantly faster than found for merocyanines 1-3 despite the fact that the molar volume of the rotor in 5 (V z 140 cm3 mol-') is comparable to that of 1-3 (V z 135 em3 mol-1).31 The enhanced rates of isomerization found for the oxonol most probably reflect the decreased bond order since, despite the presence of zwitterion~,~~the bond order of the isomerizing bond in merocyanines 1-3 must be closer to 2.0 than 1.5.The activation energies for isomerization, after correction for changes in viscosity and polarity, are signifi-cantly less for 5 than for 4. Again, such differences can be attributed to changes in the bond order of the isomerizing bond. The lower activation energies observed for the pho-toinduced processes relative to the corresponding thermal processes arise simply because of the higher potential energy available for the former reactions. s R R 11hv S 1I Scbeme 2 J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 I 1 1 1 I I I 3J 20 -to oa Lo 20 311 4.0 log P Fig. 7 Correlation between the measured partition coefficient and the cell-killingefficacy measured after exposure to 40J (m) and 90 J cm-' (A) illumination at 514 nm for the various merocyanine dyes The photophysical properties and diffusion coefficients recorded for compounds 1-6 remain comparable, but the effi-cacy for light-induced cytotoxicity towards Daudi cells varies significantly among the compounds. For the merocyanine dyes, where direct comparison is more meaningful, there is a clear correlation between the quantity of dye assimilated into the cell and the cell-killing effect (Table 3). There is a further correlation between the efficacy for light-induced cytotoxicity and the partition coefficient (Fig. 7, Table 3) and it is clear that, for the limited range of compounds studied, increased lipophilicity leads to both enhanced localization within the cell and more effective cell killing.This is an important quan-titative structure-reactivity criterion since it allows more rational design of improved photosensitizers for leukemia therapy. This work was supported by the National Institutes of Health (CA 53619). The CFKR is supported by the Uni-versity of Texas at Austin. References 1 C. J. Gomer, Photochem. Photobiol., 1991,54, 1093. 2 F. Sieber, Photochem. Photobiol., 1987,46, 1035. 3 A. Harriman, L. C. T. Shoute and P. Neta, J.Phys. Chem., 1991, 95,2415. 4 B. Kalyanaraman, J. B. Feix, F. Sieber, J. P. Thomas and A. W. Girotti, Proc. Natl. Acad. Sci. USA, 1987,84,2999. 5 J. Davila, A. Harriman and K. S. Gulliya, Photochem. Photobiol., 1991,53, 1. 6 M. Hoebeke, J. Piette and A. Van der Vorst, J. Photochem. Pho-tobiol. B: Biol., 1991,9,281. 7 J. Davila, A. Harriman and K. S. Gulliya, J. Chem. SOC., Chem. Commun., 1989,1215. 8 P. H. Aramendia, M. Krieg, C. Nitsch, E. Bittersman and S. E. Braslavsky, Photochem. Photobiol., 1988,48,187. 9 N. S. Dixit and R. A. Mackay, J. Am. Chem. SOC.,1983, 105, 2928. 10 R. J. Singh, J. B. Feix, T. J. Pintar, A. W. Girotti and B. Kalya-naraman, Photochem. Photobiol., 1991,53, 345. 11 A. Harriman, J. Photochem. Photobiol.A: Chem., 1992,6579. 12 Y. Onsager, M. Yin, D. R. Bessire and E. L. Quitevis, J. Phys. Chem., 1993,97,2344. 13 S. P. Velsko, D. H. Waldeck and G. R. Fleming, J. Chem. Phys., 1983,78,249. 14 P. R. Dragsten and W. W. Webb, Biochemistry, 1978,17,5228. 15 A. C. Benniston, K. S. Gulliya and A. Harriman, Photochem. Photobiol., submitted. 16 A. C. Benniston, K. S. Gulliya and A. Harriman, Photochem. Photobiol., submitted. 17 I. Carmicheal and G. L. Hug, J. Phys. Chem. Re$ Data, 1986, 15, 1. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 961 18 L. Pekkarinen and H. Linschitz, J. Am. Chem. Soc., 1960, 82, 25 A. J. Gordon and R. A. Ford, The Chemist’s Companion, Wiley-2407. Interscience, Chichester, 1972, p. 184. 19 J. K. Hurley, N. Sinai and H. Linschitz, Photochem. Photobiol., 26 W. H. H. Gunther, R. Searle and F. Sieber, Phosphorus, Sulfur, 1983,38,9. and Silicon, 1992,67, 417. 20 G. Rossbroich, N. A. Garcia and S. E. Braslavsky, J. Photo-27 S. R. Meech and D. Phillips, J. Photochem., 1983,23, 193. chem., 1985,31,37. 28 S. J. Strickler and R.A. Berg, J. Chem. Phys., 1962,37,814. 21 D. Axelrod, D. E. Koppel, J. Schlessinger, E. Elson and W. W. 29 C. Reichardt, Solvent Effects in Organic Chemistry, Verlag-Webb, Biophys. J., 1976,16, 1055. Chemie, Weinheim, 1979. 22 K. S. Gulliya, J. L. Mathews, J. W. Fay and R. M. Dowben, 30 J. M. Hicks, M. T. Vandersall, E. V. Sitzmann and K. B. Cancer Chemother. Pharmacol., 1988,22,211. Eisenthal, Chem. Phys. Lett., 1987, 135,413. 23 A. E. Derome, Modern NMR Techniques for Chemistry 31 A. Bondi, J. Phys. Chem., 1964,68,441. Research, Pergamon, Oxford, 1987. 24 L. G. S. Brooker, A. C. Craig, D. W. Heseltine, P. W. Jemkins Paper 3/06287B; Received 21st October, 1993 and L. L. Lincoln, J. Am. Chem. SOC.,1965,87,2443.