A stable blue-coloured form of the enzyme carrying a neutral FADH radical cofactor can be interpreted as an intermediate analogue of the light-driven DNA repair reaction and can be reduc
Trang 1photolyase monitored by Fourier transform infrared
spectroscopy
Erik Schleicher1,*, Benedikt Heßling2, Viktoria Illarionova1, Adelbert Bacher1, Stefan Weber3,
Gerald Richter1,† and Klaus Gerwert2
1 Lehrstuhl fu¨r Organische Chemie und Biochemie, Technische Universita¨t Mu¨nchen, Germany
2 Lehrstuhl fu¨r Biophysik, Ruhr-Universita¨t-Bochum, Germany
3 Freie Universita¨t Berlin, Fachbereich Physik, Berlin, Germany
Cyclobutane pyrimidine dimers (Pyr<>Pyr) and
pyri-midine–pyrimidone (6–4) photoproducts are the
predo-minant structural modifications resulting from exposure
of DNA to ultraviolet light [1,2] The structure of
Pyr<>Pyr was elucidated by Blackburn and Davies
already 40 years ago [3,4] Both photoproducts result from 2p+2p cyclo-additions The potentially mutagenic
or lethal modifications [5] must be repaired in order to ensure cell survival and genetic stability This can be effected by excision-repair or by photoreactivation
Keywords
DNA photolyase; DNA repair; FT-IR;
pyrimidine dimer; stable-isotope labelling
Correspondence
G Richter, School of Biological and
Chemical Sciences, University of Exeter,
Stocker Rd, Exeter, EX4 4QD, UK
Fax: +44 1392 26 3434
Tel: +44 1392 26 3494
E-mail: g.richter@exeter.ac.uk
K Gerwert, Lehrstuhl fu¨r Biophysik,
Ruhr-Universita¨t-Bochum, Universita¨tsstr.
150, 44780 Bochum, Germany
Fax: +49 2343 21 4238
Tel: +49 2343 22 4461
E-mail: gerwert@bph.ruhr-uni-bochum
*Present address
Freie Universita¨t Berlin, Fachbereich Physik,
Arnimallee 14, 14195 Berlin, Germany
†Present address
School of Biological and Chemical Sciences,
University of Exeter, UK
(Received 9 December 2004, revised 10
February 2005, accepted 16 February 2005)
doi:10.1111/j.1742-4658.2005.04617.x
Cyclobutane-type pyrimidine dimers generated by ultraviolet irradiation of DNA can be cleaved by DNA photolyase The enzyme-catalysed reaction
is believed to be initiated by the light-induced transfer of an electron from the anionic FADH) chromophore of the enzyme to the pyrimidine dimer
In this contribution, first infrared experiments using a novel E109A mutant
of Escherichia coli DNA photolyase, which is catalytically active but unable
to bind the second cofactor methenyltetrahydrofolate, are described A stable blue-coloured form of the enzyme carrying a neutral FADH radical cofactor can be interpreted as an intermediate analogue of the light-driven DNA repair reaction and can be reduced to the enzymatically active FADH)form by red-light irradiation Difference Fourier transform infra-red (FT-IR) spectroscopy was used to monitor vibronic bands of the blue radical form and of the fully reduced FADH) form of the enzyme Preliminary band assignments are based on experiments with 15N-labelled enzyme and on experiments with D2O as solvent Difference FT-IR mea-surements were also used to observe the formation of thymidine dimers by ultraviolet irradiation and their repair by light-driven photolyase catalysis This study provides the basis for future time-resolved FT-IR studies which are aimed at an elucidation of a detailed molecular picture of the light-driven DNA repair process
Abbreviations
DTT, dithiothreitol; FT-IR, Fourier transform infrared; MTHF, 5,10-methenyltetrahydrofolylpolyglutamate; Pyr<>Pyr, cyclobutane pyrimidine dimmer.
Trang 2mediated by DNA photolyases Specifically, photolyases
catalyse the light-driven cleavage of the cyclobutane ring
of tricyclic pyrimidine dimers, and (6–4) photolyases
cleave the pyrimidine (6–4) pyrimidone photoproduct
[6,7] Both enzymes have similar sequences [7,8] The
protein family also includes the cryptochromes which
participate in the regulation of circadian rhythms but
appear to be devoid of DNA repair activity [9–11]
The 3D structures of DNA photolyases (EC 4.1.99.3)
from Escherichia coli [12], Anacystis nidulans [13]
and Thermus thermophilus [14] have been determined
by X-ray crystallography All enzymes use anionic
reduced FADH)as redox-active cofactor [15–17] Both
5,10-methenyltetrahydrofolylpolyglutamate (MTHF)
and 8-hydroxy-5-deazaflavin serve as light-harvesting
cofactors in DNA photolyases [18–20]
DNA photolyase of E coli is typically isolated as a
blue-coloured protein carrying a neutral flavin radical,
FADH•, as a chromophore This catalytically inactive
form can be converted to the enzymatically active form
by photoreduction Tryptophan 306 is believed to serve
as the electron donor for this reaction on basis of
site-specific mutagenesis studies [21], time-resolved
electron paramagnetic resonance [22] and transient
optical absorption experiments [23]
Photolyase in the catalytically active FADH) form
binds light-damaged DNA in a light-independent step
with high affinity [24,25] Subsequent to
photoexcita-tion of the FADH) cofactor by direct absorption of
near-ultraviolet or visible light or by Fo¨rster-type
energy transfer from the MTHF antenna chromophore
[26], the excited-state FADH)chromophore is believed
to donate an electron to the pyrimidine dimer in the
DNA, thus generating a substrate radical anion and a
neutral FADH•radical [17,22,27] The dimeric pyrimi-dine radical anion splits into pyrimipyrimi-dine monomers, and the excess electron is transferred back to the FADH• cofactor to regenerate the initial redox state
of the flavin, FADH)(Fig 1)
This paper describes the first examination of DNA photolyase by Fourier transform infrared (FT-IR) spectroscopy Specific infrared bands observed in dif-ference FT-IR spectra are assigned to various photo-processes in this experimental system Hence, this study provides the basis for future time-resolved FT-IR studies which are aimed at an elucidation of a detailed molecular picture of the light-driven DNA repair process
Results
Construction of a DNA photolyase E109A mutant MTHF, the second cofactor of E coli DNA photo-lyase, acts as a light-harvesting antenna However, the protein has a relatively low affinity for this cofactor which is therefore partially lost during purification [28] Thus, individual wild-type enzyme batches typi-cally differ in their MTHF content Heterogeneity of the enzyme with respect to the chromophores, how-ever, is a serious handicap for spectroscopic studies
In order to obtain enzyme batches with reproducible absorption properties, we therefore decided to con-struct a mutant protein that does not bind MTHF but
is nevertheless enzymatically active
X-ray structure analysis has shown that the posi-tion-2 amino group and the position-3 imino group of the pteridine moiety of MTHF form hydrogen bonds
Fig 1 Putative repair reaction mechanism
of DNA photolyase.
Trang 3with the c-carboxylic group of glutamate residue 109
[12] Therefore we replaced the glutamate codon by a
codon specifying alanine using PCR-driven
site-direc-ted mutagenesis A recombinant Bacillus subtilis strain
carrying the resulting plasmid p602E109A expressed
DNA photolyase to a level of about 15% of total cell
protein Purification by published procedures afforded
the blue radical form of the mutant enzyme The yield
of isolated protein was about twofold higher than that
obtained with the recombinant E coli strain described
earlier [29]
The blue radical forms of wild-type photolyase and
the E109A mutant protein show similar absorption
spectra in the visible range above 450 nm (Fig 2A)
At shorter wavelengths, however, the absorbance of
the mutant protein is substantially lower than that of
the wild-type enzyme The absorbance difference
between the wild-type and the mutant enzyme (Fig 2B) closely resembles the spectrum of enzyme-bound MTHF [30]
Both the blue radical form of wild-type and mutant protein could be converted into the catalytically active form by photoreduction [15] The photobleached wild-type and mutant protein forms were both devoid of significant absorbance at wavelengths above 500 nm (Fig 2) In the short-wavelength range, the absorbance
of the mutant protein was again substantially lower than that of the wild-type enzyme (Fig 2A), and the absorbance difference between the proteins under study was again similar to the spectrum of MTHF (Fig 2B) These data show that the mutant protein is devoid of MTHF, and its long-wavelength absorption
is exclusively due to the flavin chromophore All subse-quent experiments were performed with the catalyti-cally active mutant protein [catalytic activity was measured by absorbance changes of UV-irradiated oligo-(dT)18 DNA at 260 nm (data not shown)] which appears as a valid model for the study of the DNA photorepair process
Photoactivation of the catalytically blue radical form of DNA photolyase
Overexpression strains of E coli can generate large amounts of recombinant DNA photolyase in the cata-lytically active dihydroflavin form, but the typical iso-lation procedures are conducive to the conversion of the enzyme into a catalytically inactive form character-ized by strong optical absorption in the range 400–
650 nm That blue-coloured species contains the flavin chromophore in the neutral radical form as shown in some detail by EPR analyses [20,29,31] The catalyti-cally active pale yellow dihydroflavin form can be easily regenerated by photoreduction of the radical form in the presence of an appropriate electron donor such as dithiothreitol (see Fig 3)
With regard to its electronic state, the stable but cata-lytically inactive blue radical form of the enzyme appears as a valid model of the transient flavin radical species that is believed to be involved in the catalytic
Fig 2 UV ⁄ vis spectra of E coli DNA photolyase at different redox
states (A) Dashed line, wild-type DNA photolyase in the blue
radi-cal form; dotted line, wild-type DNA photolyase in the reduced
form; solid line, E109A DNA photolyase in the blue radical form;
short dotted line, E109A DNA photolyase in the reduced form (B)
Solid line, difference spectrum of wild-type and E109A DNA
photo-lyase both in the radical form; dashed line, difference spectrum of
wild type and E109A DNA photolyase both in the fully reduced
Trang 4DNA-repair cycle (Fig 1) Furthermore, a flavin radical
is also involved in the light-driven photoreduction of the
blue radical enzyme species We therefore decided to
study this photoreduction of the stable blue radical form
of the enzyme to the catalytically competent FADH)
form by FT-IR spectroscopy
Infrared spectra of 1.2-mm solutions of blue radical
enzyme were measured at 4C in the dark The
enzyme samples were then irradiated for 3 min with
red light (k > 530 nm) Infrared spectra were again
obtained and were subtracted from the respective
pre-irradiation spectra affording the difference spectrum
shown in Fig 4A Positive as well as negative
differ-ence bands with relative intensities up to 0.1% were
observed Positive signals represent vibrational transi-tions characteristic of the enzymatically active FADH) form, and negative bands indicate vibrational transi-tions of the blue radical form
The reproducibility of the measurements was excel-lent As an example, the traces A and A¢ in Fig 4 were obtained with independently prepared enzyme batches The close similarity between the infrared characteristics
of the two samples is illustrated by subtraction of trace A¢ from trace A affording the double difference spec-trum shown as trace D in Fig 4
The most salient features in the difference spectra (Fig 4A) were bands at 1532 and 1396 cm)1, and changes in the amide-I (1600–1700 cm)1) region The frequencies of infrared bands can be modulated
by isotope substitution Growth of the recombinant
E coli strain used for production of photolyase on minimal medium supplemented with 15NH4Cl as the sole source of nitrogen afforded enzyme with 15N sub-stitution of most amino acids (with the exception of tryptophan, lysine, threonine and methionine which were added to the culture medium in unlabelled form; whereas they may be partially 15N-labelled by reversi-ble transamination, their 15N abundance has not been determined) Moreover, since the production strain is autotrophic with respect to riboflavin biosynthesis, the flavin chromophore of the biosynthetically labelled enzyme is also rendered universally 15N labelled Photoreduction of the 15N-labelled blue radical enzyme afforded difference infrared spectra with a sig-nificantly modified pattern of absorption bands attrib-uted to the blue radical form (negative bands of trace
B in Fig 4) and to the catalytically active FADH) form obtained after photoreduction (positive bands of trace B in Fig 4) The difference spectrum is qualita-tively similar to trace A, but the intense negative band
at 1532 cm)1 in trace A is shifted to 1524 cm)1 and the positive band at 1396 cm)1 in trace A has disap-peared
A more detailed assessment of the impact of 15N substitution is possible by inspection of the double dif-ference in trace E which is obtained by subtraction of trace B from trace A in Fig 4 In contrast to trace D
in Fig 4, the difference bands do not cancel out This indicates that numerous vibration bands have shifted
as a consequence of the universal15N labelling Major differences are especially observed in the region between 1500 and 1700 cm)1
Acidic protons in the protein can easily be exchanged by dialysis against D2O The photoreduc-tion of such treated enzyme sample afforded difference infrared spectra indicating frequency modulation of a considerable number of vibration modes The
photo-Fig 4 FT-IR difference spectra of DNA photolyase (A, A¢)
Photore-duction of DNA photolyase (two different batches of protein) (B)
Photoreduction of [U- 15 N]-DNA photolyase (C) Photoreduction of
DNA photolyase in D 2 O-containing buffer; double differences are
shown in lanes D–F (D) Subtraction of A¢ from A (E) Subtraction
of (B) from (A) (F) Subtraction of (C) from (A) (DA ¼ absorbance
difference [absorbance units], DDA ¼ double absorbance difference
[absorbance units]).
Trang 5reduction of the radical form in D2O buffer is shown
in trace C in Fig 4, which is again qualitatively similar
to trace A, but the negative band at 1532 cm)1 has
shifted to 1530 cm)1, and the band at 1396 cm)1 in
trace A appears with substantially reduced intensity
The residual intensity at this frequency (trace C of
Fig 4) can be attributed to incomplete H«D
exchange
Again, the impact of deuterium replacement of
acidic protons is best observed after subtraction of
trace C from trace A affording the double-difference
spectrum shown as trace F in Fig 4 As in the case
with 15N substitution, the partial deuteration has
affected the frequencies of numerous signals, notably
in the range between 1500 and 1700 cm)1
Photodamaging of thymidine oligonucleotides
Oligo-(dT)18DNA was used to monitor the formation
of thymidine dimers by difference FT-IR spectrometry
An excimer laser with its emission at 308 nm was used
to irradiate a 4-mm solution of oligothymidine placed
inside the infrared spectrometer The subtraction of an
infrared spectrum acquired prior to UV-irradiation
from a spectrum obtained after irradiation afforded
the difference spectrum shown as trace A in Fig 5
The photoreaction results in positive difference bands
at 1464, 1396 and 1302 cm)1 which belong to the
photodamaged form of DNA Negative difference
bands are observed at 1483, 1424 and 1289 cm)1 and
belong to undamaged DNA In summary, photodam-age afforded highly characteristic and reproducible changes in the vibrational spectrum of DNA
Photodamaging of 5-fluoro-uridine oligonucleotides
Similar irradiation experiments were performed with dodecameric deoxyoligonucleotide where the methyl group is replaced by fluorine (deoxy-5-fluoro-uridine) The difference spectrum observed with this oligo-nucleotide (Fig 5B) is similar to that observed for the photodamage of oligo-deoxythymidine The spectrum
of the irradiated deoxy-5-fluorouridine oligonucleotide shows major positive difference bands at 1741, 1460 and 1392 cm)1 and negative bands at 1715, 1410, 1364 and 1274 cm)1
DNA photorepair The subsequent experiments addressed the enzyme-mediated repair of UV-damaged DNA which had been prepared by broadband ultraviolet irradiation of the oligo-(dT) DNA substrate Permanganate titration of the irradiated DNA showed that about 50% of the bases had been converted to dimers (data not shown) Samples containing a mixture of photodamaged DNA and blue radical enzyme at an approximate fourfold excess of thymidine dimers with respect to enzyme molecules were irradiated in a two-step procedure Initially, the enzyme was photoreduced to the catalyti-cally active form by irradiation with red light (> 530 nm) This reaction was followed by difference FT-IR spectrometry which afforded a difference spec-trum closely similar to that shown as trace A in Fig 4 (data not shown) and confirmed that photorepair of DNA had not occurred This is in agreement with published data indicating that photorepair requires irradiation in the wavelength range below 530 nm [32] The sample was then irradiated with white light for
a period of 2 min During this irradiation period, infrared spectra were recorded at intervals Subtracting the spectrum obtained before the white-light irradia-tion from each of the subsequent spectra afforded a series of difference spectra shown in Fig 6A These difference spectra comprise numerous positive as well
as negative bands
A plot at various amplitudes vs time indicates that absorption differences at specific wavelengths progress with significantly different kinetics (Fig 6B) More specifically, a number of bands reach saturation levels within a period of about 10 min (e.g bands at 1464,
1396, 1302 and 1244 cm)1, whereas other bands
Fig 5 FT-IR difference spectra of DNA (A) Oligo-(dT)18 DNA
photodamage with UV radiation in the absence of photolyase (B)
Oligo-(deoxy-5-fluorouracil) 12 DNA photodamage with UV radiation
in the absence of photolyase (DA ¼ absorbance difference
[absor-bance units]).
Trang 6required up to about twice as much time to reach
saturation levels (e.g bands at 1540 and 1520 cm)1)
For a preliminary interpretation of the infrared
dif-ference bands accompanying the light-driven enzymatic
repair of photodamaged DNA, the traces in Fig 6A
can be compared with the difference spectrum
describ-ing the UV-light driven formation of thymidine dimers
(trace A in Fig 5) For ease of viewing, trace A of
Fig 5 is depicted again in Fig 7 as trace A, and the
time trace after 20 min of white-light illumination in
Fig 6A is depicted again in Fig 7 as trace B It is
obvious that a number of difference bands appear in
these traces with opposite signs and essentially cancel
out upon summation of traces A and B affording trace
C Notably, the bands that cancel out in this way are
essentially those that reach saturation at early times in
the photorepair experiments shown in Fig 6B This
suggests that these bands are characteristic of
thymi-dine dimers which are either formed by UV radiation
or consumed in the enzyme-mediated photorepair experiments
Discussion
The study of presteady-state kinetics has been pre-dominantly the domain of absorption and fluorescence spectroscopy in the visible and ultraviolet ranges These methods combine high sensitivity and selectivity with excellent time resolution down to the level of fem-toseconds However, many enzyme substrates and reaction intermediates are devoid of appropriate chro-mophoric groups Moreover, it is difficult to assign optical transients to specific intermediate structures due to the paucity of structural information in the visi-ble and ultraviolet frequency ranges
Infrared spectroscopy combines the advantages of sensitivity and high time resolution with a wealth of spectroscopic information on the reacting species and can be applied to virtually any reactant However, the interpretation is hampered by the fact that virtually all
Fig 6 Repair FT-IR difference spectra measured at time intervals
of 2 min (A) The relative change of selected bands with time (B)
(DA ¼ absorbance difference [absorbance units]).
Fig 7 FT-IR difference spectra of DNA photolyase and DNA (A) Oligo-(dT)18DNA photodamage with UV radiation in the absence of photolyase (B) Photoreactivation of DNA photolyase followed by DNA photorepair (after 20 min white-light irradiation) Addition of spectra A and B is shown in lane C (DA ¼ absorbance difference [absorbance units]).
Trang 7components of the reaction mixture contribute to the
infrared absorption This lack of selectivity in the
vibration frequency range can be addressed in different
ways Notably, selective stable-isotope labelling can be
used as a basis for band assignments
Several aspects of DNA photolyase are favourable
for an in depth presteady-state kinetic analysis (a) By
ultra-short laser pulses the enzyme reaction can be
triggered with a high quantum yield (b) The FAD
cofactor and the DNA substrate can be observed in
the visible and⁄ or ultraviolet ranges as well as in the
IR range (c) Selective stable-isotope labelling is
feasi-ble for the FAD chromophore, the apoenzyme and the
DNA substrate
This study was designed to explore the potential of
infrared spectroscopy for this enzymatic system The
data show that several chemical processes can be
observed with high reproducibility in the infrared
fre-quency range Most notably, we were able to monitor
the enzymatic repair of DNA Moreover, it was shown
that stable-isotope labelling can be used for the
pur-pose of signal assignment to specific molecular
vibra-tions Clearly, selective labelling of the flavin cofactor,
the substrate and of specific amino acid types in the
apoenzyme should be able to generate a wealth of
information at a molecular level
Although all molecular components present in the
samples used in this work are expected to contribute
to the infrared envelope, the photochemical processes
studied influence predominantly the structures of the
flavin chromophore and the pyrimidine moiety of
DNA Changes in these structural motifs are therefore
more likely to afford difference infrared bands of
sig-nificant intensity as compared to the apoprotein With
these assumptions, some tentative signal assignments
can be made These are discussed below
Photoactivation of DNA photolyase
Red-light irradiation selectively induced the
one-elec-tron reduction of the blue radical enzyme and did not
cause any changes in the DNA (neither photodamage
nor photorepair) The accompanying 1532 cm)1
differ-ence band was not affected by the presdiffer-ence of intact
or photodamaged DNA (data not shown) Universal
15N labelling or replacement of acidic protons by
deu-terium caused bathochromic shifts of this band of 8
and 2 cm)1, respectively (traces B and C in Fig 4)
Previous resonance Raman experiments on E coli
DNA photolyase [33] showed an intense band at
1528 cm)1 which experienced bathochromic shifts of 8
or 2 cm)1 in samples which were labelled with 15N or
which have been treated with D2O, respectively This
band was also observed in more recent resonance Raman experiments [34] Albeit located at 1529 cm)1,
no significant shift was observed after D2O treatment Assuming that the slight offset between the Raman and infrared bands (1528⁄ 1529 cm)1 vs 1532 cm)1) is due to calibration uncertainties, we propose that this band can be attributed to the flavin chromophore in the blue radical form on the basis of the resonance Raman activity
In photoreduction experiments with photolyase in buffer containing D2O (trace C in Fig 4), the absorp-tion signal at 1396 cm)1 (attributed to the FADH) form) showed significantly reduced intensity; the resi-dual intensity at 1396 cm)1 was attributed to incom-plete H«D exchange This band can be tentatively assigned to H(5) in plain rocking mode of FADH) Deuterium substitution of the chromophore would be expected to shift this band to the frequency range around 900 cm)1; however, the detection of the hypothetical band was not possible due to the insuffi-cient transparency of the sample in this frequency range A new difference band observed after H«D exchange at 1423 cm)1is indicative of a coupled vibra-tion mode at 1396 cm)1 Additional support for this assumption comes from uniform isotopic 15N labelling
of DNA photolyase Photoreduction results in a split-ting of the former absorbance at 1396 cm)1 into two new lines at 1382 and 1405 cm)1 indicating the contri-bution of at least two modes (Fig 4B) We find dis-tinct absorbance changes in the range of amide-I vibrations (1675⁄ 1660 ⁄ 1644 ⁄ 1625 cm)1) showing some variation in their relative intensity As no absorbance change is observed above 1700 cm)1, the C¼O stretch-ing range of protonated carbonyls, a protonation or environmental change of carbonyl groups during photoreduction is excluded
Photodamage of DNA The ultraviolet irradiation of DNA afforded several positive as well as negative difference bands which can be attributed to the consumption (negative difference bands at 1425, 1326 and 1289 cm)1) and the formation (positive difference bands at 1464, 1396 and
1302 cm)1) of thymidine dimers, respectively
An experiment with DNA carrying fluorouracil instead of thymidine afforded a qualitatively similar difference spectrum, but minor shifts in the bands appeared that are qualitatively reproduced by model calculations of the vibration modes of thymidine as compared to fluorouridine: Tavan and coworkers have recently calculated an approximate 20-cm)1 blue shift both of the C(2)¼O(2) and C(4)¼O(4) carbonyl-stretch
Trang 8vibrations due to replacement of fluorouracil with
thy-midine [35] Hence, this shift is required to disentangle
the carbonyl stretch vibrations of the thymidine dimer
from those of carbonyl vibrations from the protein
and dominant water vibrations By exploiting this
fre-quency shift, the strong vibration band at 1741 cm)1is
assigned to the C(4)¼O(4) carbonyl-stretch vibration
of the fluorouracil dimer and the strong vibration
band at 1715 cm)1 to the C(4)¼O(4) carbonyl-stretch
vibration of fluorouracil monomer The examination
of fluorouracil-containing substrate therefore provides
an important tool for the investigation of carbonyl
vibrations involved in the DNA-repair process
Enzyme-catalysed photorepair of photodamaged
DNA
For a preliminary interpretation of the infrared
differ-ence bands accompanying the light-driven enzymatic
repair of photodamaged DNA, the traces in Fig 6A can
be compared with the difference spectrum describing the
UV-light driven formation of thymidine dimers (trace A
in Figs 5 and 7) The major difference bands between
1270 cm)1and 1470 cm)1(Figs 6A and 7B) can be
asso-ciated with thymidine-dimer repair under
quasi-steady-state conditions Although experimental conditions vary
between data reported by Jorns and coworkers, the
overall rate for thymidine repair is in the same range
[36] and clearly shows the potential of FT-IR
spectro-scopy for direct measuring of kinetic rate constants
However, the photorepair is also accompanied by
certain additional difference bands in the spectral
range of 1520–1540 cm)1 A preliminary kinetic
analy-sis shows that these bands appeared at a slower rate as
compared to those which can be clearly associated
with DNA repair Hence, the bands in this range
represent a slower secondary process which cannot yet
be assigned to a specific molecular process on the basis
of the available data
The antisymmetric PO2 stretching vibration is a
characteristic marker for nucleic-acid backbone
confor-mation and is located between 1220 and 1240 cm)1,
depending on the helical conformation [37] When
irra-diating oligo-(dT)18 DNA, the conformation of the
backbone of a single-strand DNA is not expected to
change dramatically It is well known from
foot-printing, crystallographic and NMR studies that the
backbone conformation is substantially distorted in
double-stranded DNA containing a single Pyr<>Pyr
[38–40] However, this should be different in
single-stranded DNA, which is known to be much more
flexible in solution Therefore, no major difference band
is expected in this frequency region (trace A in Fig 5) If
thymidine dimer repair occurs, the conformation of the backbone should, on the other hand, change while the enzyme–product complex decays (or the enzyme– substrate complex is formed), because the chemical environment of the backbone phosphate is altered: In
an enzyme–DNA complex, electrostatic interaction of backbone phosphate and basic residues of the DNA photolyase significantly contribute to the DNA binding
of the enzyme [41–44] Therefore, an additional difference band at 1244⁄ 1224 cm)1 can be detected in Fig 7B Interestingly, the kinetics of the formation of the positive band at 1224 cm)1, which can be assigned
to the formation of enzyme-unbound oligo-(dT)18 DNA, are different to that of the bands assigned to thymidine dimer repair (Fig 6B)
Time constants under quasi-steady-state conditions for photolyase binding to UV-damaged DNA in the millisecond range have been reported by direct methods using stopped-flow experiments [34] They represent the rate determining steps of substrate-to-enzyme binding, which is expected to vary depending on the experimental conditions used Given the highly viscous buffer solu-tion [50% (v⁄ v) glycerol], the low temperature (4 C) at which our FT-IR experiments were carried out, and the higher relative substrate concentration, it is not surpris-ing that significantly longer time constants (approximate
350 s) are observed in our studies (Fig 6B)
In summary, characteristic infrared bands assigned
to the enzyme as well as the DNA substrate can be associated with DNA photorepair These observations can form the basis for time-resolved single turnover experiments, which require time-resolved FT-IR experiments on a picosecond timescale
Experimental procedures
Materials
Restriction enzymes and DNA ligase were from New Eng-land BioLabs (Frankfurt am Main, Germany) and from Roche Diagnostics (Mannheim, Germany) Taq DNA poly-merase was from Eurogentec (Seraing, Belgium) Dithio-threitol was from Sigma Oligonucleotides were custom-synthesized by MWG Biotech (Ebersberg, Germany) 15
NH4Cl was from Cambridge Isotope Laboratories (And-over, MA, USA) Microorganisms and plasmids are sum-marized in Table 1
Site directed mutagenesis
A procedure modified after Marini et al [45] was used for site-directed mutagenesis Plasmid pEPHR [29] was used as template All primers used are shown in Table 1
Trang 9The general scheme of mutagenic PCR involved two
rounds of amplification cycles using one mismatch and two
flanking primers (primers M1, Rct and Ply5, Table 1)
Dur-ing the first round, five amplification cycles were carried
out with the respective mismatch primer and with one of
the flanking primers The second flanking primer was then
added and the reaction was continued for 10 additional
cycles The PCR fragment was cleaved with the restriction
endonucleases EcoRI and BamHI and was then ligated into
the expression vector pNCO113 yielding the plasmid
pE109A For methylation of DNA, this construct was
electroporated into the E coli strain XL1-Blue and plated
on Luria–Bertani (LB) agar containing 150 mgÆL)1
ampicil-lin The plasmid was reisolated and electroporated into the
expression strain M15[pREP4], which was then plated on
LB agar containing 150 mgÆL)1 ampicillin and 15 mgÆL)1
kanamycin Transformants were monitored for expression
of DNA photolyase
Construction of an expression plasmid
Plasmid pE109A was digested with restriction
endonu-cleases EcoRI and BamHI The resulting 1416-bp fragment
was isolated and was ligated into the vector p602-CAT
The resulting plasmid p602E109A was electrotransformed
into E coli M15[pGB3] cells, which were plated on LB agar
containing 150 mgÆL)1 ampicillin and 15 mgÆL)1
kanamy-cin The plasmid p602E109A was reisolated and
electro-transformed into B subtilis BR151 cells [46] which were
plated on LB agar containing 15 mgÆL)1 kanamycin and
10 mg L)1erythromycin
Cultivation of bacterial cells
The recombinant B subtilis strain harbouring plasmids p602E109A and pBL1 was cultured in baffled 2-L Erlen-meyer flasks containing 700 mL LB medium supplemented with 15 mg L)1 kanamycin and 10 mg L)1 erythromycin The cultures were incubated at 32C with shaking At
an optical density of 0.7 (600 nm), isopropylthio-b-d-galactopyranoside was added to a final concentration of
1 mm, and incubation was continued overnight The cells were harvested by centrifugation and stored at)20 C
Preparation of15N-labelled DNA photolyase
The recombinant B subtilis strain harbouring the plasmids p602E109A and pBL1 was cultured in baffled 2-L Erlenmeyer flasks with 700 mL mineral medium containing (L)1), 6 g Tris, 0.35 g K2HPO4, 5 g glucose, 0.138 g MgSO4, 1 g 15NH4Cl, 5.55 mg CaCl2, 4 mL vitamin con-centrate, 1 mL trace metal mix, 40 mg tryptophan, 40 mg threonine, 40 mg lysine, 40 mg methionine, 15 mg kana-mycin, 10 mg erythromycin The pH was adjusted to 7.4 by the addition of 2 m hydrochloric acid Vitamin concentrate contained (per L) 20 mg pyridoxamine hydrochloride, 10 mg thiamine hydrochloride, 20 mg p-aminobenzoic acid, 20 mg calcium pantothenate, 5 mg biotin, 10 mg folic acid, 15 mg nicotinic acid, 100 lg cyanocobalamine Trace metal mix contained (per L) 16.0 g MnCl2Æ4H2O, 1.5 g CuCl2Æ2H2O, 27.0 g of CoCl2Æ6H2O, 37.5 g FeCl3Æ6H2O, 3.3 g H3BO3, 8.4 g zinc acetate, 40.8 g sodium citrate, 5 g EDTA
Cultures were incubated at 32C with shaking At an optical density of 0.7–0.9 at 600 nm, isopropylthio-b-d-galactopyranoside was added to a final concentration of
1 mm, and incubation was continued for 10 h The cells were then harvested by centrifugation and stored at)20 C
Isolation of DNA photolyase
DNA photolyase was prepared essentially as described pre-viously [29] The ammonium sulphate precipitation step was performed after chromatography on Heparin Sephar-ose Enzyme concentration was monitored photometrically (e580¼ 4800 m)1cm)1) [47]
Buffer exchange
Samples were transferred into the desired buffer [usually containing 50 mm Hepes pH 7.0, 100 mm NaCl, 10 mm dithiothreitol, 50% (v⁄ v) glycerol] by repeated dilution and ultrafiltration through C30 microconcentrators (Pall Gel-man, Dreieich, Germany) at 4C Experiments in D2O
Table 1 Bacterial strains, plasmids and primers.
Strain or
plasmid
Genotype or relevant
E coli
M15[pREP4] lac,ara,gal,mtl,recIA + , uvr +
[pREP4,lacI, kanr]
[49]
M15[pGB3] lac,ara,gal,mtl,recIA + , uvr +
[pGB3,lacI, bla r ]
[50]
XL1-Blue recA1, endA1, gyrA96, thi-1, hsdR17,
supE44, relA1, lac[F¢, proAB,
lacI q Z?M15, Tn10 (tet r )]
[51]
B subtilis
BR151[pBL1] trpC2,lys-3,metB10 [pBL1,lacI,eryr] [46]
Plasmids
pNCO113 expression vector for E coli [50,52]
pEPHR pNCO113 with the phr gene of E coli [29]
pE109A pNCO113 with the phr gene of E coli
with mutation Glu109Ala
This study p602-CAT expression vector for B subtilis [53]
p602E109A p602-CAT with the phr gene of
E coli with mutation Glu109Ala
This study Primers (5¢-3¢)
M1 (forward) GAGCGGATAACAATTTCACACAG
Rct (reverse) ACAGGAGTCCAAGCTCAGCTAATT
Ply5 (mismatch) CCCGGGCCCGCGCATTCACTTCATACTG
Trang 10were carried out at pH 7.0 (uncorrected glass electrode
reading) The dilution⁄ concentration cycle was repeated five
times to give a final D2O enrichment of 95–99%
Preparation of substrate
A 4-mm solution of single-strand oligo-(dT)18DNA was
irra-diated for 45 min using a 254 nm G8W UV-lamp (Sylvania,
Cordes, Delmenhorst, Germany) placed at a distance of
5 cm The reaction was monitored photometrically (260 nm)
Monitoring of enzyme activity
Following the procedure developed by Jorns et al [36], the
enzyme activity was measured by monitoring the repair of
cyclobutane pyrimidine dimers by DNA photolyase as a
function of time by using UV–vis spectroscopy
UV-irra-diated single-strand oligo-(dT)18 DNA was used as a
sub-strate The photorepair was performed by illuminating the
mixture with 365-nm light from a dual wavelength UV
lamp, and the repair of the cyclobutane pyrimidine dimers
by DNA photolyase was followed by changes of the
absorption at 260 nm
Monitoring of irradiation damage
Aliquots (2–10 lL) of solutions containing photodamaged
oligothymidine in water were mixed with 6 lL 1 m potassium
phosphate buffer pH 7.0 and 30 lL 20 mm KMnO4[48] and
water was added to a final volume of 600 lL under an inert
atmosphere at room temperature After 5 min, the reaction
mixture was centrifuged at 10 000 g for 1 min The
inte-grated absorbance of the supernatant was monitored in the
range 460–590 nm Reaction mixtures containing no
oligonu-cleotide (100% yield) and undamaged DNA (0% yield) were
used as references The consumption of KMnO4is equivalent
to undamaged thymidine Typically, about 50% of the bases
were damaged as analysed by this method
FT-IR sample preparation
Stock solutions contained 1.2 mm DNA photolyase and
4 mm DNA (native or damaged), respectively Equal
volumes of the stock solutions were mixed as required for
experiments Reaction mixtures were transferred into a
cuv-ette equipped with calcium fluoride windows and a 5-lm
spacer under a nitrogen atmosphere in the dark Prior to
measurements, the samples were thermally equilibrated in
the spectrometer
FT-IR instrumentation
FT-IR spectra were recorded with infrared spectrometers
(IFS 66, 66 V, 66 VS or 88) from Bruker Instruments
(Bremen, Germany) These instruments were all equipped with highly sensitive MCT-detectors and are similar in their optical layout, but are equipped with different light sources for irradiation of samples with visible or UV light Red-light irradiation was performed with a 100-W halogen lamp (Spindler & Hoyer, Go¨ttingen, Germany) using an optical
OG 530 filter (Schott, Mainz, Germany) Pulsed excimer lasers (LPX 240i, LPX 305, Lambda Physik, Go¨ttingen, Germany) were used for irradiation at 308 nm in the evacu-ated IFS 66VS and 66 V spectrometers Sample irradiation was invariably performed inside the spectrometer in order
to avoid physical handling of the cuvettes during IR experi-ments
All spectra were recorded with a bandwidth of 2 cm)1 Typically, 100 scans were accumulated and Fourier-trans-formed with the apodization function Happ–Genzel weak
Acknowledgements
This work was supported by the Deutsche Forschungs-gemeinschaft (SFB-533, TP A5 and SFB-498, TP A2),
by the Fonds der Chemischen Industrie and by the Hans-Fischer-Gesellschaft We thank Dr Chris Kay for stimulating discussions and Richard Feicht for excellent technical assistance
References
1 Friedberg EC (1995) DNA Repair and Mutagenesis ASM Press, Washington, D.C
2 Sancar GB (1990) DNA photolyases–physical-proper-ties, action mechanism, and roles in dark repair Muta-tion Res 236, 147–160
3 Blackburn GM & Davies RJH (1966) Structure of DNA-derived thymine dimer Biochem Biophys Res Commun 22, 704–706
4 Blackburn GM & Davies RJH (1965) Structure of thy-mine photo-dimer Chem Commun 215–216
5 Otoshi E, Yagi T, Mori T, Matsunaga T, Nikaido O, Kim ST, Hitomi K, Ikenaga M & Todo T (2000) Respective roles of cyclobutane pyrimidine dimers, (6–4) photoproducts, and minor photoproducts in ultraviolet mutagenesis of repair-deficient xeroderma pigmentosum
A cells Cancer Res 60, 1729–1735
6 Todo T, Takemori H, Ryo H, Ihara M, Matsunaga T, Nikaido O, Sato K & Nomura T (1993) A new photo-reactivating enzyme that specifically repairs ultraviolet light-induced (6–4) photoproducts Nature 361, 371–374
7 Nakajima S, Sugiyama M, Iwai S, Hitomi K, Otoshi
E, Kim ST, Jiang CZ, Todo T, Britt AB & Yama-moto K (1998) Cloning and characterization of a gene (UVR3) required for photorepair of 6–4 photopro-ducts in Arabidopsis thaliana Nucleic Acids Research
26, 638–644