Spectroscopy and photophysics of styrylquinoline-type HIV-1integrase inhibitors and its oxidized forms studied by steady state andtime resolved absorption and Ñuorescence

Rolande Burdujan,a Jean dÏAngelo,b Didier Fatima Zouhiri,b Patrick Tauc,cDesmae� le,bJean-Claude Brochon,c Christian Auclair,d Mouscadet,d Pascal Pernot,eJean-FrancÓ oisFrancis TÐbel,a Mironel Enescua and Marie-Pierre Fontaine-Aupart*a

a L aboratoire de Photophysique UPR 3361 CNRS, Paris-Sud,Mole� culaire, Universite�91405 Orsay Cedex, France. E-mail : marie-pierre.fontaine-aupart=ppm.u-psud.fr

b Centre dÏEtudes Pharmaceutique, L aboratoire de Chimie Organique, UPRES-A 8076 CNRS,Paris-Sud, 92296 Chatenay Malabry, FranceUniversite�

c L aboratoire de Biotechnologies et de Pharmacologie UMR 8532,Ge� ne� tique Applique� e,ENS Cachan, 92435 Cachan, France

d Physicochimie et Pharmacologie des Biologiques, UMR 8532 CNRS,Macromole� culesInstitut Gustave Roussy, V illejuif, France

e L aboratoire de Chimie Physique, Paris-Sud, 91405 Orsay Cedex, FranceUniversite�

Received 19th March 2001, Accepted 10th July 2001First published as an Advance Article on the web 7th August 2001

The geometric and electronic structure in solution of a new 2-styrylquinoline-type inhibitor of theHIV1-integrase was examined for the Ðrst time by following changes in the photophysical properties of thechromophore using steady state as well as time resolved absorption and Ñuorescence methods. The resultsobtained under di†erent conditions of pH and solvent revealed that there are at least two rotamers in theground state which exhibit di†erent photophysical and photochemical properties. Picosecond Ñuorescence andabsorption measurements gave evidence for a very short (D20È30 ps) singlet excited state lifetime for oneconformer and a much longer one for the other conformer, from a few hundreds of picoseconds up tonanoseconds, depending on the solvent characteristics. At physiological pH, the longer lived conformer canalso undergo oxygen oxidation or photooxidation giving rise to the formation of the semiquinone radical andultimately to a stable orthoquinone species. The role played by the singlet excited state of the rotamer on thephotooxidation process is also detailed using picosecond and nanosecond absorption measurements.

Introduction

The replicative life cycle of the HIV virus requires the integra-tion of viral DNA into genomic DNA of the host cell, aprocess which is mediated by the viral protein integrase (IN).The development of inhibitors of IN activity constitutes a newchallenge for the treatment of HIV-1 infection.1 Several inhibi-tors of HIV-1 integrase have been now identiÐed, but only afew display antiviral activity in cells. A promising class of suchantiviral agents based on a 2-styrylquinoline-like structure(Scheme 1) has attracted widespread interest since these mole-cules inhibit HIV integrase function both in vitro and in vivoand are devoid of cellular toxicity.2h4

No contribution has appeared up to now about thephysico-chemical properties of compound 1 and its analogs(Scheme 1) which would allow an understanding of the bio-logical function of these molecules at the molecular level,including the nature of the interactions with the biologicaltargets. Since compound 1 is a trans-2-styrylquinoline-likechromophore, di†erent conformational geometries (tautomersand rotamers) of the drug should exist in solution.5h7 Thus,the physico-chemical as well as the photochemical behavior of1 are the results of the di†erent contributions of these individ-ual conformers5h10 and it is of importance to know which isthe predominant form of the drug in the various environ-

mental conditions and in its biological binding site. For thispurpose, we have carried out a detailed study of the steadystate and time resolved absorption and Ñuorescence propertiesof 1. The spectroscopic properties of the drug in various sol-vents and in aqueous solutions at various pH values werestudied with the aim to obtain information on the inÑuence ofthe environmental conditions on its stability, its geometricand electronic structure in solution. Absorption and emissionspectra, Ñuorescence decays and quantum yields have beeninterpreted in terms of the existence of an equilibrium betweenat least two rotamers depending on the protonation state ofthe drug and the protic or aprotic character of the solvent.

Furthermore, we have also proved the ability of 1 to beoxidized by oxygen or as a result of photonic excitation andcharacterized the intermediates involved in such a process bysubpicosecond and nanosecond laser Ñash photolysis. Thepicosecond formation of a semiquinone radical was demon-strated and a mechanism for the formation of the more stableorthoquinone species was proposed.

Materials and methodsSamples

The compound 1 and the analogs 2 and 3 were prepared inthe Laboratoire de Chimie Organique (Chatenay Malabry), as

DOI: 10.1039/b102555b Phys. Chem. Chem. Phys., 2001, 3, 3797È3804 3797

This journal is The Owner Societies 2001(

Publ

ishe

d on

07

Aug

ust 2

001.

Dow

nloa

ded

by C

arle

ton

Uni

vers

ity o

n 01

/06/

2014

23:

21:2

8.

View Article Online / Journal Homepage / Table of Contents for this issue

Scheme 1 Structure of 1 and analogs 2 and 3

described in the literature.2 The drugs were Ðrst dissolved(10~1 M) in DMSO solution (Prolabo, Normapur) and thendiluted in bu†ers prepared from deionized water (passedthrough an ion exchange column, Bioblock ScientiÐc type U3)twice distilled in an all-quartz two stage Heraeus set-up. TheÐnal percentage of DMSO never exceeded All the sol-2&.vents were of spectroscopy grade from Merck. The percentageof water in ethanol was 5%. For the pH inÑuence study, com-pound 1 was diluted in HCl at pH 2.0, in 20 mM TrisÈHCl or20 mM phosphate bu†er for pH ranging(KH2PO4/K2HPO4)from 7.0 to 8.5 and in 20 mM glycine bu†er for pH [ 8.5.Some results were obtained in 50 mM borate bu†er at pH 8.0.Salts used were of analytical grade (Merck).

The samples were deaerated by bubbling argon through thesolution prior to the experiments. The oxygenated sampleswere obtained by maintaining 1 atm upon the solution. AllO2the gases were from Alphagaz with a purity of 99.99%. All themeasurements were carried out at room temperature (298 K).

Steady state experiments

The absorption and Ñuorescence steady state spectra weremeasured respectively with a Cary 210 (Varian) spectropho-tometer and a Perkin-Elmer MPF3L Ñuorimeter. The spectrawere recorded with home-made computer-controlled acquisi-tion systems. In Ñuorescence experiments, the absorbance wasusually less than 0.1 at the excitation wavelength. Excitationspectra were corrected for the Ñuctuations of the Xe-lampintensity in the range 270È380 nm. Under these conditions,the spectral resolution of the Ñuorescence excitation spectrawas the same as that of the absorption spectra.

The concentration of each compound was determined byweighing and used to estimate the molar absorption coeffi-cient of the chromophore. The Ñuorescence quantum yielddetermination was carried out using 9,10-diphenylanthracene(DPA) in cyclohexane as a reference nm;(jexc \ 366 Uf \ 1under deaerated conditions)11 according to the followingrelationship :

/fx \ /fs(FsAx/Fx As)(ns2/nx2)where the subscript x refers to the unknown and the subscripts to the standard, A is the absorbance at the excitation wave-length, F is the integrated emission across the band and n therefractive index of the solvent.

Time resolved Ñuorescence set-up

Fluorescence decays were measured by the time-correlatedsingle photon counting technique.12,13 The excitation light

pulse source was a Ti-sapphire subpicosecond laser (Tsunami,Spectra Physics) associated with a third harmonic generatortuned at 320 nm and was vertically polarized. The repetitionof the laser was set down to 4 MHz and the pulse durationwas 1 ps. Fluorescence emission was detected through amonochromator (Jobin-Yvon H10) set at 460 nm for dioxaneexperiments and at 540 nm for the other conditions (*j\ 16nm), by a microchannel plate photomultiplier (HamamatsuR1564U-06) connected to an ampliÐer, Phillips ScientiÐc 6954(gain 50). The excitation light pulse was triggered by a Hama-matsu photodiode (S4753). The pulse signal was ampliÐed,shaped and connected to the stop input of the TAC (TimeAmplitude Converter, Ortec 457) through a discriminator(Tenelec 453). The function of the instrumental response (100ps) was recorded by detecting the light scattered by a watersolution. The time scaling was 8.14 ps or 34.6 ps per channeland 2048 channels were used. Total Ñuorescence intensity, i(t),was recorded by orientating the emission polarizer at the““magic ÏÏ angle of 54.75¡. The Ñuorescence decay and theinstrumental response proÐle were collected alternately during90 s and 30 s, respectively, over at least 20 periods until thetotal count for the Ñuorescence decay reached 12È16 millions(giving approximately 105 counts in the peak channel). Toreduce the pileup errors from the timeÈamplitude converter,the counting rate never exceeded 10 kHz. The microcell(volume 50 ll) was thermostated with a Haake type-F3 circu-lating bath.

Analysis of Ñuorescence decay i(t) was performed by thequantiÐed maximum entropy method (MEM).14,15 Timeresolved Ñuorescence decay data were Ðtted to a function thatis a sum of 100 discrete exponentials. The explored lifetimedomain ranges from 0.02 to 5 ns in a logarithm scale. Therecovered distribution shows rather narrow peaks (data notshown), the positions of which are summarized in Table 1.The time resolution obtained after pulse convolution is D20ps.

Time-resolved absorption experiments

Subpicosecond laser photolysis was performed by a pumpÈprobe method described previously.16 BrieÑy, the laser systemwas the second harmonic of a Nd: YAG-pumped dye laserwhich delivers 0.5 ps pulses at a repetition rate of 10 Hz. Thelaser was used at 630 nm and frequency doubled to provide 15lJ excitation pulses at 315 nm. The non-converted fraction ofthe laser beam (j \ 630 nm) was focussed into water to gener-ate a white light continuum used as a probing beam. The twoprobing beams, sample and reference, were simultaneouslyanalyzed with a polychromator (Jarrell-Ash, entrance slit 100lm) and an OMA Spec 4000 System (EG&G, PrincetonApplied Research) equipped with a CCD detector. The timeresolution of the system was less than 1 ps. The polarizationof the pump light was set at the magic angle (54.7¡) relative tothe polarization of the probe light. In a given experiment,transient spectra were determined for a set-up of 20 chosendelay times of the probing beam with respect to the pumpingbeam (from [4 ps to ]400 ps) which were automaticallyscanned at least 30 times. For each delay, the data were accu-mulated over 50 laser shots every scan. All of the wavelengthsdo not arrive at the sample at the same time owing to disper-sion e†ects (group velocity and continuum generation), thusthe transient spectra were corrected for this wavelengthdependence of the delay times.

The excitation source of the nanosecond photolysis systemwas a Nd: YAG laser (Quantel, YG 441) of 3 ns full-width athalf-maximum with third harmonic (355 nm) generation. The355 nm beam was directed onto one side of a 10 mm squaresilica cell containing the sample. The transient transmissionvariations were monitored at right angles to the excitation ina cross beam arrangement using a xenon Ñash lamp, a mono-

3798 Phys. Chem. Chem. Phys., 2001, 3, 3797È3804

Publ

ishe

d on

07

Aug

ust 2

001.

Dow

nloa

ded

by C

arle

ton

Uni

vers

ity o

n 01

/06/

2014

23:

21:2

8.

View Article Online

Table 1 Singlet excited state decay parameters and Ñuorescence quantum yields of compound 1 in di†erent solvents and at di†erent pH. Theexcitation wavelength is 320 nm and the drug concentration 15 lM

q1/ps A1 q2/ps A2 q3/ns A3(^15%) (^2%) (^10%) (^2%) (^10%) (^2%) s2 /f (^0.01)

HCl pH 2.0 20 100 1.5Tris-HCl pH 7.0 30 30 230 70 1.75 0.02Tris-HCl pH 8.0 30 40 220 60 1.8 0.02Borate pH 8.0 20 100 2.5Dioxane 910 52 1.6 48 1.44 0.11DMSO 50 7 650 31 3.0 62 1.47 0.14Ethanol (5 50 45 290a 12 1.7 12 1.3 0.04% water) 780 31

a This component is due to the of 5% of water in the ethanol used.

chromator, a photomultiplier (HTV R928, response time 1 ns)and a digitized oscilloscope (Tektronix 2440) controlled by amicrocomputer. The Ñuence of the incident laser pulse in thesample was obtained by calibration of the joulemeter usinganthracene in deaerated cyclohexane as a triplet actinom-eter.17

The transient absorption data can be expressed as a sum ofcontributions from the spectrally active species :

*A(j, t)\ ;i

Ei(j)D

i(t)

where is the absorption spectrum of species i, andEi(j) D

i(t)

the corresponding kinetic evolution functions. Then, a kineticscheme, using a compartmentalÈglobal analysis method, isdesigned where kinetic parameters and absorption spectra aresimultaneously recovered.18 The kinetic parameters are deter-mined by non-linear iterative convolution method minimizingthe sum of weighted-least-squares,18 whereas the absorptionspectra are calculated for each set of kinetic parameters by aleast-squares algorithm.18 In the absence of a normalizationreference, no attempt is made to determine the spectralweights for each species.

Drug irradiations

Steady state photolysis of drug 1 in Tris bu†er solution (pH 7)was performed by irradiating the samples with an OSRAM-75W1 xenon lamp whose output was passed through a Ðlter(Schott WG 345) to select wavelengths higher than 315 nm.The photochemical e†ects on the drug were also induced byexciting 1 with nanosecond laser pulses using the Ñash pho-tolysis set-up (vide supra). These laser experiments allowed usto check the monophotonic character of the photoinducede†ects.

Results and discussion

Spectral evidence for the existence of rotamers in the groundstate

Compound 1 is a trans-2-styrylquinoline-like molecule andthus likely to exist as di†erent conformational isomers involv-ing rotation around the quasi-single bond between the styryland quinoline moieties.5h7 Detailed information on the exis-tence of such ground state rotamers can be obtained fromtheir absorption, excitation and emission spectra as well as thekinetic parameters of excited state deactivation.

The absorption spectra of compound 1 at pH 7.0 is illus-trated in Fig. 1. Two distinct electronic transitions areobserved, the highest one centered at 330 nm with a shoulderat 355 nm and a minor absorption band at 260 nm. The simi-larity between the absorption spectra of 1 and ofstyrylquinolines7,8 allows a phenomenological attribution ofthe absorption bands to the di†erent parts of the molecule.

The transition centered at 260 nm corresponds to the elec-tronic transition of the catechol part of the chromophore. Itcan be seen (Fig. 1) that the absorption spectrum of 2 closelyresembles that of 1 with only a slight hyperchromic e†ect,reÑecting the minor inÑuence of the replacement of a ÈOHgroup by a methoxy group on 3@-C. In contrast, the structuralchanges in 3 (substitution of the ethylenic bond by a saturatedbond) cause the loss of conjugation between the quinolinepart and the catechol group of the molecule which results in asigniÐcant increase of the molar absorption coefficient of thecatechol group of this compound by comparison with thevalues obtained for the molecules 1 and 2.

The n-styrylquinolines7h10,19 are also characterized by atransition around 350 nm, attributed to the quinoline ring ofthe molecule and a distinct absorption transition around 315nm that is characteristic of the stilbene-like chromophore. Thestructural modiÐcations of the quinoline parts of compounds1 and 2 by comparison with the 2-styrylquinolines lead to ared-shift of the latter absorption transition which is superim-posed on the less intense band of the quinoline (Fig. 1). Thehigh molar absorption coefficient values of this transitionstrongly suggest that the lowest excited singlet state of themolecules is of p,p* character, as for styrylquinolines.8,19 Thisis further conÐrmed by the red-shift observed in the emissionspectra with increasing solvent polarity (vide infra).

Both the emission and excitation spectral shapes of 1 areindependent of (from 270 to 450 nm) and (from 480 tojex jem560 nm) respectively (Fig. 2) ; the fact that the maximum emis-sion is centered at 540 nm indicates that the excitation energyis largely localized in the quinoline moiety of the molecule.Furthermore, the normalized emission intensity changesobserved upon excitation at di†erent wavelengths indicatethat more than one species is implicated in the absorptionspectra. This was conÐrmed by the Ñuorescence excitation

Fig. 1 Absorption spectra of compound 1 (ÈÈ) and its analogs 2(- - -) and 3 (ÉÉÉÉÉÉ) observed in Tris bu†er pH 7.0 ; drug concentrationD40 lM. The inset shows the veriÐcation of the BeerÈLambert lawfor 1.

Phys. Chem. Chem. Phys., 2001, 3, 3797È3804 3799

Publ

ishe

d on

07

Aug

ust 2

001.

Dow

nloa

ded

by C

arle

ton

Uni

vers

ity o

n 01

/06/

2014

23:

21:2

8.

View Article Online

Fig. 2 Normalized Ñuorescence emission (right) and correctedexcitation (left) spectra of 1 (12 lM) in Tris bu†er pH 7.0, at di†erentexcitation and emission wavelengths at room temperature. The nor-malization is done with respect to the number of absorbed photons.The absorption spectrum (left, dashed line) is also reported for com-parison.

spectrum of 1 (Fig. 2), which is less structured compared tothe absorption spectrum, with only one peak around 350 nm,and could not be superimposed on the latter. Furthermore,the Ñuorescence decay of 1 was deconvoluted in two well-separated components, assigned to a shorter lived (30 ps) anda longer lived (230 ps) species (Table 1).

A di†erent hypothesis can be proposed to account for theseresults. We have Ðrst veriÐed that the absorbance of the drugsolutions followed BeerÈLambertÏs law in the concentrationrange 5È100 lM (Fig. 1, inset), revealing the absence of self-association of the drug in its ground state. The di†erencebetween absorption and Ñuorescence excitation spectra couldalso be explained by a 1trans*È1cis* singlet excited state pho-toisomerisation of 1 as observed for styrylquinolines6,8,19,20and 2-styrylanthracene.10 Such a process can be analyzed bychanges in the absorption spectrum of the chromophore uponirradiation of a deoxygenated solution.8,19 In our case, theabsorption spectral evolution corresponds to an oxidationand not a photoisomerisation process (vide infra). Accordingto the protonation state of the drug at physiological pH (videinfra), the existence of tautomers can also be considered on theo-hydroxybenzoic part of the chromophore. However, it isadmitted that the neighborhood of hydroxy and carboxy sub-stituents favors the formation of a divalent hydrogen atomlink21 and that the enol form prevails in solution.22 Therefore,the relation of absorption, Ñuorescence excitation and emis-

sion properties of 1 strongly suggests the presence of at leasttwo conformers of the molecule in the ground state. Byanalogy with the case of 2-styrylquinolines,7 these two con-formers can be assigned to two rotamers originating from therotation of the quinoline moiety around the quasi-single bondbetween the quinoline and ethylenic carbon atom (Scheme 2).The relative absorption spectra of these two rotamers cannotbe obtained, the stationary Ñuorescence emission of one ofthem being too weak to be detected. But, considering that theabsorption spectrum of 1 corresponds to the overlappingabsorption of the two rotamers in equilibrium and that theabsorption maximum of one of them peaks at 340 nm, as(R1)revealed by its excitation spectrum, the second rotamer (R2)presumably has an absorption maximum at a shorter wave-length (D330 nm), consistent with the lower Ñuorescenceintensity measured for nm. This blue-shiftedjexc \ 340absorption of by comparison with indicates a largerR2 R1steric repulsion between the hydrogen atoms in the R1rotamer. Assuming that the radiative lifetime of the two rota-mers are of the same order, it may be shown, by using therelationship that is the component with theUf \ qf] kr , R1longer emission lifetime (230 ps) and a Ñuorescence quantumyield of 0.02 and that has a Ñuorescence lifetime of 30 psR2and a quantum yield 10] lower than that for R1.

Solvent dependence of the absorption and Ñuorescenceproperties

pH e†ects. Compound 1 also undergoes pH-dependentstructural changes which can be followed from the absorptionspectra (Fig. 3). In acidic media, the absorption spectrum pre-sents three main bands with peaks at 285, 340 and 415 nm. ApH increase from 2.0 to 7.0 considerably a†ects the spectrum.The two bands centered at 340 and 415 nm have collapsedinto one major band nm) ; a blue-shift (from 285 to(jmaxB 330270 nm) and a decrease of the maximum of the catechol tran-sition are also observed with increasing pH. In the pH range7.0È8.5, the spectra are quite similar with only changes in thepeak intensities. Beyond pH 8.5, a gradual shift of the 330 nmmaximum to 370 nm is observed, associated with an hyper-chromic e†ect on the catechol transition (275 nm).

The drug 1 possesses various sites of protonation : at N-1and at the oxygen atoms C-7 and C-8 of the quinoline moietyand C-3@ and C-4@ of the catechol part. Accurate determi-nation of the di†erent of the chromophore by absorptionpKameasurements alone (lack of Ñuorescence in acidic conditions)was impossible. However, referring to the literature, the proto-

Scheme 2 Rotamer equilibrium between the two rotamers and and the (photo)oxidation mechanism at pH 7.0.R1 R2 ,

3800 Phys. Chem. Chem. Phys., 2001, 3, 3797È3804

Publ

ishe

d on

07

Aug

ust 2

001.

Dow

nloa

ded

by C

arle

ton

Uni

vers

ity o

n 01

/06/

2014

23:

21:2

8.

View Article Online

Fig. 3 Absorption spectra of compound 1 as a function of pH. Thedrug concentration range is 30È40 lM.

nation state of the molecule may be inferred from the pKavalues for each of its moieties.The of the ground state of several n-styrylquinolinespKawhich possess only one protonation site on the nitrogen atom

of the pyridine ring range between 4.8 and 5.8.8,23,24 Further-more, this value does not seem to be inÑuenced by structuralmodiÐcations of either the quinoline24 or benzene23 parts ofthe chromophore. Thus, similar values can be expectedpKafor the N-1 nitrogen atom of the drug. In 1, the carboxy andhydroxy substituents lie at adjacent positions, C-7 and C-8exactly as in o-hydroxybenzoic acid which has two ofpKas2.97 and 13 for and OH/O~ respectively.21h22CO2H/CO2~Previous work25 also reported two of 9.5 and 12.8, forpKas,the catechol group. These values should be valid for com-pound 3 for which the quinoline and catechol moieties areindependent. In contrast, lower values should be expected forthe catechol part of 1 due to the conjugation of the molecule.This hypothesis was conÐrmed by the results concerning thepossibility of oxidation of the drug at physiological pH (videinfra).

The Ñuorescence properties of 1 as a function of pH werealso investigated and the results reported in Table 1. It isknown that the of the carboxy group in the singletpKaexcited state is rather close to the value in the groundpKastate22 while both the acidity of the hydroxy group and thebasicity of the ring nitrogen are enhanced in the excitedstate.24 However, N-1 proton uptake or HO-8 proton ejectionreactions, generally coupled with intramolecular electrontransfer, occur on a longer time scale24 than the lifetime of thesinglet excited state of the drug. Consequently, it can beadmitted that the singlet excited state protonation equilibriumis not attained and that the Ñuorescence properties of thequinoline part of the drug appear to be dependent on theprotonation state of the ground state. The singlet excitedpKastate of the catechol group can reasonably be expected to besigniÐcantly lower than that of the ground state,26 corres-ponding to an excited protonated form at pH 2.0 and a totallydeprotonated form at pH P 7.0.

At pH 2.0 (the drug is then totally protonated), the station-ary Ñuorescence emission of 1 is too weak to be detected.Nevertheless, the time resolved Ñuorescence data reveal asingle exponential decay with a decay time of 20 ps (Table 1)revealing the existence of only one type of conformer. Similarresults are obtained by dissolving compound 1 in boratebu†er (pH 8.0). In this case, boric acid may form a complexwith the o-hydroxybenzoic part of the molecule leading toprotonation of the N-1 nitrogen atom.27 This protonationresults in a higher bond order of the single bond between thestyryl and quinoline moieties and thus favors a conformationof the drug with a short Ñuorescence lifetime such as the R2rotamer. Beyond pH 7.0 (deprotonation of N-1 and of the

catechol part), the dynamic Ñuorescence data (biexponentialdecay, Fig. 2, Table 1) are consistent with the presence of twoemissive species (no data were obtained for pH[ 8.5 due tosigniÐcant oxidation of drug 1 under such conditions (videinfra)). This pH study revealed the inÑuence of the proton-ation state of N-1 on the occurrence of the equilibriumbetween the two rotamers and of the drug.R1 R2

Polarity e†ects. The absorption, excitation and Ñuorescencespectra of compound 1 were investigated in the non-polarsolvent dioxane, and in solvents of increasing polarity,ethanol, DMSO and Tris bu†er pH 7.0, in order to examinesolvent e†ects on the photophysical properties and conforma-tional equilibrium. Inspection of the absorption spectra showsthat the catechol transition (260 nm) is practically una†ectedby the solvent nature while the maximum of the main band(330È360 nm) is blue-shifted in ethanol, DMSO and waterwith respect to dioxane (Fig. 4(a). In contrast, the excitationspectrum of the drug in polar solvents undergoes a small red-shift compared to the excitation spectrum measured indioxane (Figs. 2 and 4, (b)È(d)).

The Ñuorescence properties of 1 are also greatly inÑuencedby solvent properties. The Ñuorescence Stokes-shift is muchlarger ([100 nm) in polar solvents and increases with increas-ing solvent polarity (Fig. 4(a)). These spectral evolutions areassociated with variations of the Ñuorescence quantum yieldsand lifetimes (Table 1). The Ñuorescence quantum yieldsdecrease with increasing solvent polarity, except in DMSO.Similar weights of shorter (a few tens of picoseconds) andlonger lived (a few hundreds of picoseconds) components areobserved in ethanol and water while in dioxane Ñuorescenceoccurs only from two longer lived components. While DMSOis a protic solvent, the longer lived decay components aresimilar to those for dioxane.

The photophysical properties of 1 in dioxane, DMSO andethanol by comparison with the results obtained in water indi-cate that the conformational equilibrium of the drug changeswith solvent properties. Referring to the absorption and exci-tation spectra, it appears that the protic character of thesolvent favors the occurrence of an equilibrium distribution ofconformers of the drug. Considering the Ñuorescence lifetimes,the shorter lived component observed in ethanol and DMSO(20È50 ps) can be attributed to the rotamer and theR2hundreds of picoseconds one (650È780 ps) to the rotamer,R1already described in the case of water. On a longer time scale,

Fig. 4 (a) Absorption spectra (left) and Ñuorescence spectra (right) ofcompound 1 in dioxane, ethanol, DMSO and Tris bu†er pH 7.0. Thedrug concentration is 1 lM in dioxane (È É È É) and DMSO (ÉÉÉÉÉ), 5lM in ethanol (È È È)and 12 lM in aqueous solution (ÈÈ). The Ñuo-rescence spectra are normalized to the same absorbance at the excita-tion wavelength (360 nm.). (b), (c), (d), Comparison of the absorption(ÈÈ) and excitation Ñuorescence (È È È) spectra of 1, in respectively,dioxane nm), DMSO nm) and ethanol(jem \ 440 (jem\ 520 (jem \

nm).540

Phys. Chem. Chem. Phys., 2001, 3, 3797È3804 3801

Publ

ishe

d on

07

Aug

ust 2

001.

Dow

nloa

ded

by C

arle

ton

Uni

vers

ity o

n 01

/06/

2014

23:

21:2

8.

View Article Online

the Ñuorescence emission analysis in DMSO and ethanol ismore complex than in water, an additional component in thenanosecond time range is observed. Due to the superpositionof the Ñuorescence excitation spectra at various emissionwavelengths, the existence of a new conformer in the funda-mental state can be reasonably excluded both in ethanol andDMSO. Therefore, the nanosecond Ñuorescence componentprobably arises from a singlet excited state reaction, thenature of which falls beyond the scope of the present paper.The excitation spectrum of 1 in dioxane is roughly the imageof its absorption spectrum with a maximum at 350 nm (Fig.4(b)) and the absence of the tens of picoseconds Ñuorescencelifetime suggests the presence of only the rotamer in thisR1solvent. The nanosecond Ñuorescence component can beexplained in the same way as in ethanol and DMSO. Thissolvent polarity study reveals that even though the con-R2former has a larger steric repulsion than the conformer itsR1,formation is favored in protic solvents, a result alreadyobserved for 2-styrylquinoline.7

Spectral evidence for the formation of an oxidized form inaqueous solution

The absorption spectra of 1 in Tris bu†er pH 7.0 under atmo-spheric pressure was found to evolve with time after prep-aration while kept in the dark (Fig. 5). It is seen that thelower-energy absorption band centered at 330 nm decreaseswith time while the second band increases and is blue-shiftedfrom 270 to 260 nm and a new transition appears at 440 nm.Moreover, the presence of an isosbestic point at 395 nmstrongly suggests the formation of a new species in the groundstate.

The following facts suggest that 1 can be oxidized : (i) nospectral evolution was observed in anaerobic conditions or atpH 2.0 (even in the presence of oxygen) corresponding to atotally protonated form of the drug (see above), (ii) theabsorption spectrum evolution increased on the one handwith increasing pH (for pH [ 7.0) in correlation with depro-tonation of the catechol group of the molecule (vide supra) andon the other hand with oxygen pressure (from atmosphericpressure to 1 atm of and (iii) the spectral evolution is lessO2)pronounced with compound 2 which possesses only one OHgroup on its benzene part and is absent for 3 which is stillprotonated on its catechol part in the pH range 7.0È9 (videsupra) The oxidation depends on how long the solutionsare exposed to air and on the protonation state of the drug

Fig. 5 Absorption spectral changes of 1 solutions (45 lM) in deaer-ated Tris bu†er, pH 7.0, as a function of Ñuence upon irradiation witha xenon lamp (j [ 325 nm) or with a nanosecond laser source : (È È È)before irradiation, (ÈÈ) 6 ] 104 J m~2, (È É È É È) 1.5] 105 J m~2.The spectrum obtained after 2 h (----) in the absence of irradiation butunder aerated conditions is superimposed on that obtained upon irra-diation with a Ñuence of 6] 104 J m~2.

(for pHP 10, oxidation was initiated as soon as the drug wasdiluted into the bu†er).

Upon steady state irradiation of deoxygenated solutions of1, a similar evolution of the absorption spectrum is observed(Fig. 5) as in the presence of oxygen, but it takes place muchfaster. These results indicate that, upon light activation, pho-tooxidation of the drug also occurs.

A likely mechanism proposed for the reaction of oxygenwith a catechol17 is the formation of a semiquinone radical byone-electron oxidation followed by dismutation of this radicalgiving rise to an orthoquinone species which typically absorbsat k B 420 nm. Such an oxidation pathway can account forthe spectral evolution observed with 1 in the presence ofoxygen. It must be noted that the orthoquinone formed is alsounstable ; after resting for one day in the dark, the spectrum ofthe oxidized drug has undergone further changes.

Since similar spectral evolutions are observed under pho-tolysis of 1 in anaerobic conditions and in the presence ofoxygen (Fig. 5), we can predict that the semiquinone radical isalso the main intermediate upon light activation of the drug.To verify this hypothesis and to investigate the photooxida-tion mechanism of 1, time resolved absorption spectroscopyexperiments have been performed.

Nanosecond and subpicosecond laser photolysis

Fig. 6 shows the transient spectra obtained from 355 nm laserexcitation of 1 in or saturated Tris bu†er solu-N2O, N2 O2tion at pH 7.0, recorded at di†erent delay times after the endof the pulse. The di†erence spectrum recorded at the end ofthe laser pulse is characterized by a positive broad band witha maximum at 430 nm and a negative band due to groundstate depletion around 330 nm. The dependence of the signalat 430 nm on the laser energy is linear up to an incidentÑuence of 40 mJ cm~2, as expected for a monophotonicprocess. By increasing the energy of the excitation pulse, thetransient absorbance greatly increased and was followed by arapid decay (only a few nanoseconds) of the signal, which isconsistent with the occurrence of an additional photophysicalprocess. Thus, the excitation of compound 1 was performedunder energy conditions such that this second transient wasnot signiÐcantly formed.

The transient spectrum measured at the end of the laserpulse in is coincident both in shape and intensity withN2those obtained in the presence of efficient hydrated electronscavengers and a result which leads to the conclu-(N2O O2),sion that photoionization is not involved in the laser photoly-

Fig. 6 (a) Transient di†erence absorption spectra measured on 355nm photolysis of compound 1 (40 lM) in 20 mM Tris bu†er pH 7.0,under saturation : delay time 10 ns, 400 ns, (*) 10 ls.N2O (…) (>)Inset : Time proÐle observed at di†erent characteristic wavelengths.(b) Spectra of A and B species according to the analytical(=) (…)model. The dashed line spectrum is adjusted to the absorbance spec-trum obtained at 400 ps delay by subpicosecond laser photolysis.

3802 Phys. Chem. Chem. Phys., 2001, 3, 3797È3804

Publ

ishe

d on

07

Aug

ust 2

001.

Dow

nloa

ded

by C

arle

ton

Uni

vers

ity o

n 01

/06/

2014

23:

21:2

8.

View Article Online

sis of the drug. Furthermore, typical triplet quenchers suchas oxygen (under 1 atm), piperylene (0.1 M ) or 1-naphalenemethanol (10~3 M) had no e†ect on the transientabsorption spectrum, revealing that the corresponding speciesis not an excited triplet state. In agreement with this conclu-sion, the spectral absorbance evolution is characterized by theformation, on the nanosecond time scale, of a new transient(build-up of the absorbance for wavelengths between 400 and430 nm) (Fig. 6, inset), followed by a decrease of the transientabsorbance over all the spectral range on the microsecondtime scale.

The absorbance changes (Fig. 6(a)) can be analyzed at eachwavelength on the basis of two consecutive reactions corre-sponding, respectively, to second order and Ðrst order kineticsaccording to the following scheme:

A] A] B] A0B] C

where A corresponds to the transient species observed at theend of the laser pulse, to the drug in its fundamental state,A0B to the product of dismutation and C to the species gener-ated by the disappearance of B. In this kinetic scheme, wesuggest that A corresponds to a radical species. In view of thestructure of 1, two mechanisms can be proposed : a photo-decarboxylation involving the formation of a carbanion onthe quinoline part,28h30 or the formation of a semiquinoneradical from the deprotonated catechol group of the chromo-phore. To check a pathway involving decarboxylation, laserphotolysis experiments were carried out with drug 1 as a func-tion of pH (pH 2.0, 7.0, 8, 8.5) and with the compounds 2 and3 at pH 7.0 (using energy laser conditions corresponding to amonophotonic process). The results reveal that (i) no transientsignal was observed for the fully protonated form of 1 (pH 2.0)nor for compound 3 and (ii) similar transient absorptions weredetected for compounds 1 and 2 at pH P 7.0 (correspondingto partial deprotonation of the catechol part of thechromophores). A photodecarboxylation process should beindependent of both these pH conditions and structural modi-Ðcations between the di†erent compounds studied and thussuch a process can reasonably be excluded. Therefore thetransient absorption spectrum measured at the end of thenanosecond laser pulse can be attributed to the semiquinoneradical that originates from a charge transfer reactionoccurring on a time scale faster than nanoseconds (vide infra).This was conÐrmed by experiments performed in the presenceof various amounts of ascorbic acid, a quencher of semi-quinone radicals.25 An efficient radical quenching with a rateconstant of 2.5 ] 107 M~1 s~1 was observed.

The semiquinone radical was found to evolve by a dis-mutation reaction (k \ 3 ] 1010 M~1 s~1) leading simulta-neously to the formation of the ground state of the drug (A0)and of a species (B) absorbing predominantly at 420 nm (Fig.6(a)), assigned to an orthoquinone. The analytical modelallows the determination of the absorption spectra of bothsemiquinone and orthoquinone (Fig. 6(b)). The orthoquinoneÐrst formed is unstable and decays (k \ 4.5^ 0.3] 105 s~1)to give the more stable photoproduct (C) already observed bysteady state absorption and not yet identiÐed (Fig. 5).

As stated above, the hypothesis of a semiquinone radicalformation requires a charge transfer reaction as the precursorstep, a process which probably occurs on the picosecond timescale. This was conÐrmed by experiments performed usingsubpicosecond laser excitation.

The transient absorption spectrum obtained at 1.5 ps by320 nm subpicosecond laser photolysis of compound 1 ischaracterized by one main band, centered at 460 nm. (Fig. 7).The transient absorbance develops out within the laser pulseduration, as illustrated in Fig. 8(a). Since both and rota-R1 R2mers can be excited at 320 nm, the transient spectrum is likely

Fig. 7 Transient di†erence absorption spectra measured on 320 nmphotolysis of compound 1 ([1]\ 40 lM) in 20 mM Tris bu†er pH 7.0as a function of time.

to be due to the sum of the excited singlet state absorptionS1of both rotamers. The kinetics of this di†erential absorptionreveals a fast component of 8 ps ^2 ps (Fig. 8(b)). Accord-ingly, a spectral evolution was also identiÐed which can bequantitatively represented by the Ðrst momentum of theabsorption spectrum deÐned by :

l6 \/ lA(l) dl/ A(l) dl

where l is the wavenumber and A the measured transientabsorption. A spectral displacement (*lB 2 ] 1013 s~1(*j B 13 nm) was thus found at the end of the fast process(Fig. 8(c)).

The time constant of this process closely matches the fastdecay of the Ñuorescence intensity (30 ps) (taking into accountthe experimental accuracy in the time resolved Ñuorescencemeasurements) indicating that it could correspond to decayS1of the rotamer. From these assignments, the transientR2absorption spectrum measured at 50 ps nm) can(jmaxB 445be attributed to the absorption of the rotamer whichS1 R1has not started to evolve because of its much longer Ñuores-cence lifetime (Table 1).

Clearly (Fig. 7, 8(c)) at times later than 50 ps, the decay ofthe band centered at 445 nm is accompanied by the rise of aband centered at 430 nm with the same time constant of 200ps. The existence of an isosbestic point at 454 nm indicatesthat the rotamer in its singlet excited state interconvertsR1into a new transient species. One notes a close similaritybetween the spectra obtained at 360 ps delay in subpicosecond

Fig. 8 (a) Rise of the di†erence absorbance of 1 in aqueous solutionat pH 7.0 upon 320 nm laser excitation. (b) Decay of the di†erenceabsorbance of the same solution. (c) Plot of the Ðrst momentum of theabsorption spectrum against the delay time.

Phys. Chem. Chem. Phys., 2001, 3, 3797È3804 3803

Publ

ishe

d on

07

Aug

ust 2

001.

Dow

nloa

ded

by C

arle

ton

Uni

vers

ity o

n 01

/06/

2014

23:

21:2

8.

View Article Online

measurements and at the end of the nanosecond laser excita-tion (Fig. 6(b)), meaning that the semiquinone radical isalready formed at 400 ps. It may be noted that the maximumof the nanosecond spectrum is less intense than that of thepicosecond one (Fig. 6(b)) due to the fact that the dismutationreaction has begun to occur on the subnanosecond time scale.

The semiquinone radical formation requires an intramole-cular electron transfer between the catechol and quinolineparts of the drug which may occur consecutively to energyredistribution in the excited states as reported for suchcomplex molecules.29,30 This electron transfer process can bemediated by singletÈtriplet intersystem crossing. However, acommon feature for a number of stilbene9 and styrylderivatives6,31 is the very low intersystem crossing quantumyield. Thus, is therefore more likely to be the reactiveS1excited state.

Conclusion

In this study, it has been shown that compound 1 is present insolution as a mixture of two stable rotamers with di†erentphotophysical and photochemical properties. The equilibriumbetween the conformers is strongly pH and solvent dependent.In acidic medium, only the rotamer (Scheme 2) is presentR2while in aprotic solvents (dioxane) only the rotamer exists.R1The structural di†erences between the two conformers aremainly reÑected in their singlet excited state lifetimes. The R2rotamer Ñuoresces in only a few picoseconds while the R1rotamer has a longer excited state lifetime, ranging from a fewhundreds of picoseconds to nanoseconds, depending on thesolvent characteristics. This study reveals that the physico-chemical properties of 1 are clearly related to its spectroscopicproperties which are quite sensitive to its environment.

It was also shown that the rotamer can undergo oxida-R1tion or photooxidation involving a semiquinone radical as thekey intermediate. Upon photolysis, this radical is formed fromthe excited singlet state which deactivates through intramole-cular electron transfer from the catechol to the quinoline moi-eties. The electronic conjugation of the molecule is the keycondition for the occurrence of such a (photo)oxidationprocess. The possible occurrence of such oxidation at physio-logical pH can be a limiting factor for its biological activity.This Ðnding has prompted the synthesis of novel drugs forwhich oxidation is precluded while the cellular activity isretained.

References1 D. J. Hazuda, P. Felock, M. Witmer, A. Wolfe, K. Stillmock, J. A.

Grobler, A. Espeseth, L. Gabryelski, W. Schleif, C. Blau andM. D. Miller, Science, 2000, 287, 646.

2 K. Mekouar, J. F. Mouscadet, D. F. Subra, H. Leh, D.Desmae� le,C. Auclair and J. dÏAngelo, J. Med. Chem., 1998, 41,Savoure� ,

2846.3 F. Zouhiri, J. F. Mouscadet, K. Mekouar, D. D.Desmae� le,

H. Leh, F. Subra, M. Le Bret, C. Auclair and J.Savoure� ,dÏAngelo, J. Med. Chem., 2000, 43, 1533.

4 M. Ouali, C. Laboulais, H. Leh, D. Gill, D. K.Desmae� le,Mekouar, F. Zouhiri, J. dÏAngelo, C. Auclair, J. F. Mouscadetand M. Le Bret, J. Med. Chem., 2000, 43, 1949.

5 D. H. Waldeck, Chem. Rev., 1991, 91, 415.6 U. Mazzucato and F. Momicchioli, Chem. Rev., 1991, 91, 1679.7 S. C. Shim, D. W. Kim and M. S. Kim, J. Photochem. Photobiol.

A, 1991, 56, 227.8 G. Galiazzo, G. Gennari and P. Bortolus, J. Photochem., 1983,

23, 149.9 G. Gennari, P. Bortolus and G. Galiazzo, J. Mol. Struct., 1991,

249, 189.10 A. Spalletti and G. Bartocci, Phys. Chem. Chem. Phys., 1999, 1,

5623.11 S. R. Meech and D. Phillips, J. Photochem., 1983, 23, 193.12 Ph. Wahl, in New T echniques in Biophysics and Cell Biology, ed.

M. Pain and B. Smith, John Wiley & Sons, Chichester, 1975,vol. 2, p. 233.

13 D. V. OÏConnor and D. Philips, in T ime-correlated Single PhotonCounting, Academic Press, London, 1984.

14 J.-C. Brochon, Methods Enzymol., 1994, 240, 262.15 A. K. Liversey and J.-C. Brochon, Biophys. J., 1987, 52, 693.16 M. Enescu, M. P. Fontaine-Aupart, B. Soep and F. TÐbel, J.

Phys. Chem. B, 1998, 102, 3631.17 R. C. Bensasson, C. Land and R. G. Truscott, in Excited States

and Free Radicals in Biology and Medicine, Oxford UniversityPress, New York, 1993, pp. 80È82 and 236È239.

18 M. P. Fontaine-Aupart, E. Renault, C. Videlot, F. TÐbel, R.Pansu, M. Charlier and P. Pernot, Photochem. Photobiol., 1999,70, 829.

19 G. Gennari, G. Galiazzo and P. Bortolus, J. Photochem., 1986,35, 177.

20 R. Zamboni, M. Belley, E. Champion, L. Charette, R. DeHaven,R. Frenette, J. Y. Gauthier, T. R. Jones, S. Leger, P. Masson,C. S. McFarlane, K. Metters, S. S. Pong, H. Pierachuta, J.Rokach, M. H. W. R. Williams and R. N. Young, J.The� rien,Med. Chem., 1992, 35, 3832.

21 Handbook of Chemistry and Physics, The Chemical RubberCompany, Boca Raton, FL, p. D86.

22 G. S. Denisov, N. S. Golubev, V. M. Schreiber, Sh. S. Shajakhme-dov and A. V. Shurukhina, J. Mol. Struct., 1997, 437, 153.

23 S. W. Wang and T. I. Ho, Chem. Phys. L ett., 1997, 268, 434.24 E. Bardez, A. Chatelain, B. Larrey and B. Valeur, J. Phys. Chem.,

1994, 98, 2357.25 Z. B. Alfassi and R. H. Shuler, J. Phys. Chem., 1985, 89, 3359.26 C. A. Parker, in Photoluminescence of Solutions, Elsevier,

Amsterdam, 1968, ch. 4.27 T raite de Chimie de Pascal, volume of borine, Masson,mine� rale

1933, pp. 220È222 and 304È308.28 F. Bosca and M. A. Miranda, Photochem. Photobiol., 1999, 70,

653.29 S. Encinas, M. A. Miranda, G. Marconi and S. Monti, Photo-

chem. Photobiol., 1998, 68, 633.30 S. Sortino and J. C. Scaiano, Photochem. Photobiol., 1999, 69, 167.31 G. G. Aloisi, F. Elisel and L. Latterini, J. Chem. Soc., Faraday

T rans., 1992, 88, 2139.

3804 Phys. Chem. Chem. Phys., 2001, 3, 3797È3804

Publ

ishe

d on

07

Aug

ust 2

001.

Dow

nloa

ded

by C

arle

ton

Uni

vers

ity o

n 01

/06/

2014

23:

21:2

8.

View Article Online