Mechanism of the Decay of Thymine Triplets in DNA Single Strands

Letter pubs.acs.org/JPCL

Mechanism of the Decay of Thymine Triplets in DNA Single Strands Bert M. Pilles,† Dominik B. Bucher,† Lizhe Liu,† Pascale Clivio,§ Peter Gilch,‡ Wolfgang Zinth,*,† and Wolfgang J. Schreier*,† †

Lehrstuhl für BioMolekulare Optik, Fakultät für Physik and Munich Center for Integrated Protein Science CIPSM, Ludwig-Maximilians-Universität München, Oettingenstr. 67, 80538 München, Germany ‡ Institut für Physikalische Chemie, Heinrich-Heine-Universität Düsseldorf, Universitätsstr. 1, 40225 Düsseldorf, Germany § Institut de Chimie Moléculaire de Reims, CNRS UMR 7312, Université de Reims Champagne Ardenne, UFR de Pharmacie, 51 rue Cognacq-Jay, 51096 Reims Cedex, France S Supporting Information *

ABSTRACT: The decay of triplet states and the formation of cyclobutane pyrimidine dimers (CPDs) after UV excitation of the all-thymine oligomer (dT)18 and the locked dinucleotide TLpTL were studied by nanosecond IR spectroscopy. IR marker bands characteristic for the CPD lesion and the triplet state were observed from ∼1 ns (time resolution of the setup) onward. The amplitudes of the CPD marker bands remain constant throughout the time range covered (up to 10 μs). The triplet decays with a time constant of ∼10 ns presumably via a biradical intermediate (lifetime ∼60 ns). This biradical has often been invoked as an intermediate for CPD formation via the triplet channel. The present results lend strong support to the existence of this intermediate, yet there is no indication that its decay contributes significantly to CPD formation. SECTION: Spectroscopy, Photochemistry, and Excited States

transfer from higher- to lower-lying triplet states is strongly favored and triplet states might tend to be localized on thymine bases.19,22 The latter would make them prone to be hotspots for mutations, yet time-resolved studies on triplet energy transfer and localization on thymine bases are still missing. Experiments with nanosecond laser flash photolysis revealed that triplet states in TMP dissolved in water are formed with a yield ΦISC of ∼0.014 after UV excitation.23 Yet, performing the same experiment on the oligonucleotide (dT)20, no triplet state absorption was observed within the time resolution of 200 ns. Assuming that kinetics and yields of intersystem crossing (ISC) and thereby triplet formation are similar in the monomer and the oligomer, it was suggested that in the oligomer the triplet states are effectively quenched in a photochemical reaction yielding CPD lesions on a picosecond or nanosecond time scale, not resolved in the experiment.23 In line with this, more recent femtosecond pump−probe experiments and timeresolved fluorescence emission in the UV and visible spectral range point to the possibility of reactive triplet states as precursors for CPD lesion formation.24 In this study, we search for a triplet pathway toward the CPD by means of UV-pump IR-probe spectroscopy on the nano- to microsecond time scale. (See Material and Methods in the

DNA photodamage is among the most important external hazards for the integrity of cells exposed to UV radiation.1 In particular, neighboring pyrimidine bases are prone to form UVinduced DNA photolesions and are known to be mutational hotspots.2,3 The photolesion with the highest abundance is the intrastrand [2 + 2] photoaddition between two thymine bases on a DNA strand forming a cyclobutane pyrimidine dimer (CPD).2 Although the formation of the CPD lesion was described as early as 1960,4,5 there is still a continuing debate about the multiplicity of the dimer precursor state and the predominant reaction pathways. (See Figure 1.) A series of early experiments on thymine nucleotides in frozen matrices and aggregates in solution was indicative of a direct singlet reaction channel that allows for dimer formation in an ultrafast photoreaction for properly oriented thymine bases.6−8 These results could only recently be validated by time-resolved experiments with femtosecond UV-pump and IRprobe pulses (TRIR).9−11 Using the high structural sensitivity of infrared (IR) spectroscopy, evidence was given that in the all-thymine DNA single strand (dT)18 CPD lesions are formed within ∼1 ps after UV excitation.9 This was interpreted as a concerted CPD formation from the initially excited singlet state with ππ* character.9,12−15 It is well known that triplet photosensitization can lead to CPD formation in DNA strands.16−19 The lowest excited triplet state of the four different DNA bases (adenine, guanine, cytosine, thymine) is the one of thymine.20,21 It has been proposed that energy © 2014 American Chemical Society

Received: February 19, 2014 Accepted: April 15, 2014 Published: April 15, 2014 1616

dx.doi.org/10.1021/jz500364g | J. Phys. Chem. Lett. 2014, 5, 1616−1622

The Journal of Physical Chemistry Letters

Letter

Figure 1. Schemes for UV-induced CPD formation. (a) For bases in a reactive prearrangement CPD lesions are formed within 1 ps from an excited singlet state. (b) For bases that are too distant for an ultrafast reaction, CPD formation could occur via a long-lived triplet state (red) after intersystem crossing (ISC) of the initially populated singlet ππ* state (blue).

Figure 2. Time-resolved IR absorption changes after 266 nm excitation of (dT)18 (left) and TLpTL (right). (a) Contour plot of the experimental data. (b) Transient difference spectra at the indicated delay times. (c) Difference spectra for CPD lesions as obtained from steady-state illumination experiments. Broken lines denote the marker bands of CPD lesions. To improve the signal-to-noise ratio of the transient spectra in panel b, we have averaged the data in time (10 data points) and frequency (3 data points). Note that different ΔA scales are used for (dT)18 and TLpTL.

data have been corrected for the contribution of solvent heating. (See the Supporting Information.) Comparison of the CPD difference spectra with the contour representations and the transient difference spectra in Figure 2 shows that the spectral patterns of the CPD lesions are present in the time-resolved data. This applies to the marker bands in the range of 1300−1500 cm−1 (broken lines) as well as to the carbonyl region (1600−1750 cm−1) with dominant groundstate bleaches. The patterns are evident from the earliest delay times (∼1 ns) onward. Throughout the time range covered, there is no gain in amplitude for these spectral signatures. The amplitude for this CPD signature is substantially higher for TLpTL. Taking the IR absorption coefficients and the excitation conditions into account, the amplitude for TLpTL surmounts the one for (dT)18 by a factor of three. This matches the ratio between the CPD yields ΦCPD determined in steady-state experiments.11 Both data sets lend to the interpretation that the

Supporting Information.) We hereby profit from three recent findings: CPDs exhibit distinct marker bands in the spectral range of 1300−1500 cm−1;9,11 the local triplet state of thymine shows up in the IR via a carbonyl resonance located around 1600 cm−1;25 and the triplet yield ΦISC of TMP increases with decreasing wavelength,26 allowing for an enhanced triplet signal. To connect with our previous studies9,11 in the pico- to nanosecond time range, the samples looked upon ((dT)18 and the locked nucleic acid dinucleotide TLpTL) were excited with UV pulses at 266 nm. For a closer look at the fate of the triplet state, the (dT)18 sample was studied with 250 nm excitation taking profit from the increased triplet yield. Time-resolved absorption changes obtained for (dT)18 and TLpTL after excitation with 266 nm UV pulses are given in Figure 2. The data shown here cover a range from 1300 to 1800 cm−1 and a time range from 3 ns to 1 μs. Thereby, the transient 1617



Letter

Figure 3. Time-resolved IR absorption changes of (dT)18 after 250 nm excitation. For this wavelength, a higher triplet yield as compared with the excitation at 266 nm (see Figure 2) occurs. Left: Transient absorption difference spectra. Right: Transient absorption difference spectra corrected for the absorption change obtained at late delay times (double-difference spectra ΔΔA(ν)). This highlights spectroscopic changes that are not due to CPD. (a) Contour plot of the experimental data. (b) Transient difference spectra at the indicated delay times. The red arrow marks a prominent band of the local triplet state. (c) Contour plot of double-difference spectra. (d) Transient double-difference spectra at the indicated delay times. The broken lines indicate the positions of triplet (red) and biradical (blue) bands and the recovery of the starting material (black). To improve the signalto-noise ratio of the absorption changes in panels b and d, the data have been averaged in time (10 data points) and frequency (3 data points).

CPD lesion is formed on time scales shorter than 1 ns and does not gain population afterward. Inspecting the data up to 100 ns, one might come to the conclusion that the CPD population decreases on that time scale. For (dT)18 and TLpTL, broad positive absorption changes in the range 1300−1500 cm−1, and bleaches lose amplitude up to ∼100 ns. The relative effect is more pronounced for (dT)18. We will see later in experiments with increased triplet yield that these features are related to triplet states. In this experiment, we rely on the fact that excitation at 250 nm leads to a two-fold increase (relative to 266 nm) of the triplet yield ΦISC of TMP.26 The respective transient difference spectra are shown in Figure 3 (left). Again, the pattern characteristic for CPD lesions is observed throughout the complete time range covered, but the spectroscopic changes within the first 100 ns are now more pronounced. The spectral characteristics of these changes are better perceived after the following procedure. The difference spectrum recorded at late delay times, at 10 μs after photoexcitation (essentially due to CPD formation), is subtracted from all difference spectra. The resulting doubledifference spectra ΔΔA(ν) (Figure 3, right) are zero within noise level for ∼100 ns to 10 μs. Earlier on, they differ markedly from the difference spectra for the CPD lesion (cf. Figure 2). Thus, the spectroscopic changes seen during the first 100 ns are not due to a (partial) decay of the CPD population. Within these first 100 ns, there are bands decaying rapidly as the one at 1596 cm−1. Other bands, for example, the one at 1624 cm−1, first rise and then decay. Concomitant with the second decay, the ground-state bleach at 1665 cm−1 recovers. The time traces plotted in Figure 4 highlight these features. Time constants for the kinetic processes were determined via a global analysis using a sum of two exponentials convoluted with the instrumental response function and an offset. The procedure yielded two time constants τ1 = 12 ns and τ2 = 59 ns and the respective decay-associated spectra DAS A1(ν) and A2(ν). The quality of the fit can be judged from Figure 4. In the resulting

Figure 4. Time-resolved absorption changes after UV excitation (250 nm) of (dT)18 at the indicated wavenumbers (cf. Figure 3). The constant absorption change at late delay times was subtracted. The data were analyzed in terms of a sum of two exponentials convoluted with the instrumental response function and an offset. Resulting fits on the data are given as solid lines. As indicated by the scheme in the lowest panel the transient signals can be interpreted as follows: decay of species I with a 12 ns time constant; formation and decay of species II with a 12 ns and a 59 ns time constant, respectively; recovery of the starting material with the decay of species II.

kinetic scheme (cf. Figure 4 lower panel), a first transient species I decays with a time constant τ1, thereby populating species II. The decay of the latter species with the time constant τ2 results in a recovery of the (dT)18 starting material. Longer-lived singlet excited states of thymine including local nπ* states or charge-transfer states are expected to decay on the 100 ps to 1 ns time scale.9,27−30 In contrast, species I exhibits a significantly longer lifetime of ∼10 ns, making the assignment to a singlet excited state unlikely. Spectrally, species I is characterized by a prominent absorption band around 1618



Letter

Figure 5. Comparison between decay associated difference spectra and calculated (DFT) difference spectra for species I and II. Schematic illustrations of the postulated structures are given on top. DAS A1(ν) and A2(ν) are given in panels a and c. The computed difference spectra A3LT(ν) − A3BR(ν) and A3BR(ν) − AG(ν) are shown in panels b and d, respectively. Arrows highlight the correlation between experimentally observed and computationally determined bands.

1600 cm −1 . A similar spectral feature was observed experimentally for the triplet state of thymine and thymidine in deuterated acetonitrile.25 In the latter study, a characteristic absorption signature in the infrared spectral range was observed consisting of a pronounced absorption band peaking at 1603 cm−1 and other IR bands with smaller amplitude between 1300 and 1400 cm−1. Preliminary experiments performed on the monomer TMP in aqueous solution yielded similar results concerning the position of the triplet marker band. On the basis of DFT computations, the band around 1600 cm−1 was assigned to the C4O stretch vibration of the triplet state.25,31 Thus, experimental and quantum-chemical data for the thymine triplet state are in good agreement with the spectral properties of species I. Species I will be denoted in the following as the lowest triplet state 3LT. Concerning species II, there are abundant suggestions in the literature that triplet excited thymines in vicinity to a groundstate thymine decay via a triplet biradical intermediate 3BR (e.g., see refs 12 and 32−34). In such a biradical (cf. Figure 5), the two C6 atoms of the thymines are covalently linked. Formally, the two unpaired electrons reside on the C5 atoms (1,4 biradical). To our knowledge, neither experimental nor quantum-chemical data on the IR spectrum of 3BR are available. To check for the plausibility of a 3BR involvement, we computed IR difference spectra and compared them with the respective DAS. It is thereby assumed that TpT is a suitable model for the spectroscopic changes occurring in (dT)18. IR spectra of TpT in its ground state AG(ν), of the local triplet state A3LT(ν), and of the triplet biradical A3BR(ν) were computed using DFT methods. (See the Supporting Information.) The computations were performed using the B3LYP density functional35,36 with the 6-311G** basis set as

implemented in the Gaussian 03 (Revision E.01) quantumchemical package.37 The DFT method in combination with the hybrid functional was chosen based on the good results obtained for the triplet state of thymidine in ref 25. Additionally, the quality of the calculations was verified by comparing the computed spectra of TpT with experimental data. Following the geometry optimization, computations of harmonic vibrational wavenumbers, normal modes, and IR intensities were carried out. To account for the deuterated solvent, we replaced labile hydrogen atoms by deuterium atoms.38 Optimized structures including spin-density distributions are given in the Supporting Information. For the calculation of difference spectra, as shown in Figure 5, the computed stick spectra were convoluted with a Lorentzian (fwhm of 25 cm−1). Additionally, the IR wavenumbers were scaled by 0.96.39 Provided that the first process with the time constant τ1 is the decay of the local triplet state and the concomitant formation of 3 BR, the DAS A1(ν) should be given by A3LT(ν) − A3BR(ν). For the second process, the DAS A2(ν) needs to be compared with A3BR(ν) − AG(ν). Both comparisons (see Figure 5) lend credibility to the previous assignment. Looking at the DAS of species I and starting from high wavenumbers, one can clearly identify two positive absorption changes between 1650 and 1700 cm−1. This signature is followed by a bleach (∼1625 cm−1) and another positive feature at 1600 cm−1. To lower frequencies, mainly smaller negative features are observed. The same ordering of positive and negative features is seen in the calculations (indicated by arrows in Figure 5 (left)). The positions of the higher lying bands in the region of the CO double bonds match very well (1650−1750 cm−1), only the 1619



Letter

Irrespective of the molecular interpretation of species I and II and the assignment of the time constants τ1 and τ2, our experimental data show unequivocally that within noise level no CPD is formed in the time range 1 ns to 10 μs. The upper limit for the CPD formation yield ΦCPD,T via a triplet channel can be estimated as follows: For (dT)18 and 250 nm excitation (Figure 3), the relative noise level with respect to the CPD signal is smaller than 0.1. The overall quantum yield ΦCPD for the CPD formation is on the order of ∼0.05.26 Accordingly, the yield ΦCPD,T is smaller than ∼0.005. This is in agreement with a recent study by Banyasz et al. concluding that the contribution of 3ππ* states to CPD formation in a (dT)20 single strand should be lower than 10%.26 The yield ΦCPD,T depends on the triplet yield ΦISC and the fraction ηT of triplets forming CPDs, ΦCPD,T = ΦISC·ηT. Assuming a triplet yield ΦISC of ∼0.03,26 as found for TMP in water, the fraction ηT is calculated to be ∼0.15 or smaller. This is in line with other 1,4 biradicals where cyclization efficiencies much smaller than unity were reported.46 The low fraction ηT is in accordance with findings of low efficiencies for CPD formation from triplet states of monomeric thymine in aqueous solution. In the study by Whillans and Johns, it was calculated that only one-fifth of triplet thymine in solution quenched by the parent molecule lead to a CPD lesion.47 An even lower ratio was found by Wagner and Bucheck.32,48 Using cis-1,3-pentadiene as a triplet counter and acetonitrile as solvent, they found that only ∼0.02 of the thymine triplet−ground state interactions resulted in CPD formation. Triplet sensitization experiments mentioned in the introduction are also in line with a small fraction ηT. The yields ΦCPD,SEN for the sensitized dimerization are small, on the order of 0.01 or lower (cf. table 1 in ref 19). For efficient ISC of the sensitizer and efficient triplet energy transfer to thymine, the CPD yield ΦCPD,SEN ought to approach the fraction ηT. So, also sensitization experiments point to a low propensity of thymine triplets to form CPDs. Direct investigations of the fate of triplet states in native DNA are hampered by the small quantum yields and the multitude of other interactions between different bases and base pairs, yet, the lowest triplet excited state of the four DNA bases is the one of thymine. Therefore, one might expect that triplet states indeed get localized on thymine bases in double strands as well as in the investigated model systems. Thus, the characteristic signature of the involved reaction species obtained in this study will be the basis for further investigations in larger systems. In this study, the decay of thymine triplet states and the formation of cyclobutane pyrimidine dimers (CPDs) after UV excitation were studied by nanosecond IR spectroscopy. The all thymine oligomer (dT)18 and the locked dinucleotide TLpTL were investigated in buffered D2O covering the nanosecond to microsecond time range. For either sample, distinct IR marker bands for CPD lesions were observed from

Mechanism of the Decay of Thymine Triplets in DNA Single Strands

Recommend Documents