Development of a DNA photoaffinity probe for the analysis of 8-oxodG-binding proteins in a human proteome

Article abstract of DOI:10.1002/cbic.201000244

(Figure Presented) Finding DNA repair factors: A comparative study with DNA probes containing 8-oxodG and bearing different photo-crosslinking moieties found that a triethyleneglycol-trifluoromethylphenyldiazirine system was excellent for the labeling of DNA repair factors. This probe was used for the activity-based and selective labeling of different DNA repair factors in a human HCT-116 nuclear proteome.

Full text of DOI:10.1002/cbic.201000244

DOI: 10.1002/cbic.201000244
Development of a DNA Photoaffinity Probe for the Analysis of 8-OxodG-
Binding Proteins in a Human Proteome
Malte Winnacker, Veronika Welzmiller, Ralf Strasser, and Thomas Carell*[a]
Our genome is constantly damaged by exogenous and endog-
enous events.[1] Particularly problematic is oxidative DNA
damage caused by the reaction of reactive oxygen species
(ROS) with the genetic material.[2] The most common oxidative
lesions are 8-oxodG and FaPy-dG.[3] Repair of both lesions is
essential for the survival of cells; unrepaired lesions induce cell
death or cause mutations.[3] Understanding of the chemistry
that leads to the formation of these lesions has improved tre-
mendously during the last decade[3a] and we now understand
the principles of how these lesions are repaired.[1b] Several crys-
tal structures provide detailed insight into the lesion-recogni-
tion and -repair processes.[4]
for comparison—a series of other possible photoaffinity labels,
is shown in Scheme 1.
The synthesis of 1 started with TBDMS protection of iodo-
deoxyuridine (IdU, 3), which provided TBDMS-IdU (4). Sonoga-
shira coupling with the alkinole 5[9] furnished the triethylene-
glycol-substituted nucleoside 6, which was coupled to the di-
azirine iodide 7[10] to provide the nucleoside 8. Deprotection
furnished the target nucleoside 1. Final conversion of 1 into
the phosphoramidite 2 via the DMT-protected nucleoside 9
was accomplished by standard procedures.
Incorporation of
2 into oligonucleotides was possible
through phosphoramidite chemistry. For the photoaffinity ex-
periment we prepared DNA strands containing the diazirine
trap next to a biotin (Bio) label for final affinity purification of
crosslinked proteins and a fluorescein label (Fl) for detection.
Finally an 8-oxodG lesion (G) was incorporated into the com-
plementary counterstrand. Control strands lacking either the 8-
oxodG lesion or the photo-crosslinker were prepared. All pre-
pared DNA single strands were characterized by MALDI-TOF-
MS and HPLC prior to hybridization. Finally they were hybrid-
ized to form the double-stranded DNA probes P1–P8 (Support-
ing Information). Subsequent UV melting point studies con-
firmed that the oligonucleotides exist as undisturbed duplexes
(Supporting Information).
Despite these achievements, our understanding of the pro-
cesses that occur inside the cell nucleus at a lesion site is still
in its infancy. However, new quantitative mass spectrometry
techniques are have recently been developed that revolution-
ize our ability to study cellular processes.[5] With the help of
special trapping molecules (crosslinkers) that covalently con-
nect the involved proteins to their targets it is possible to char-
acterize the involved proteins. Here we report the develop-
ment of the new photo-crosslinker 1 (Scheme 1) that, when
incorporated in DNA,[6] is able to capture proteins involved in
processing the main oxidative DNA lesion 8-oxodG.
We selected a diazirine unit as the photoreactive group re-
sponsible for forming the covalent bond with the protein.[7]
The diazirine is small and upon irradiation at a comparatively
long wavelength (l=365 nm) it forms a reactive carbene,
which binds efficiently to proteins. The diazirine was attached
to the 5-position of the nucleobase uridine so that it should
not disturb the DNA duplex structure. This was important be-
cause we wanted to reduce binding of proteins that might
only recognize a disturbed duplex structure. Finally we decided
to link the diazirine to the nucleobase through a triethylene-
glycol linker. This spacer molecule seemed the best choice for
reducing unspecific binding to the probe itself. In addition, the
chemical nature of the spacer reduced the chances of putative
hydrophobic collapse of the reactive group with the DNA
duplex.[8] The synthesis of the photoaffinity nucleoside 1 and
of the corresponding phosphoramidite 2, together with the
prepared oligonucleotide probes (P1–P8) containing 1 and—
We first analyzed the properties of crosslinker 1 within P2 in
comparison with probes (P1, P3–7) containing other photo-
affinity labels.[7c] For this study we simply replaced the trap
molecule 1 in P2 either by a diazirine, which we attached to
the nucleobase through a different spacer (P1),[11] with a ben-
zophenone (P3, P7), or with nucleobases, which simply carry
halogen atoms (P4–P6; Scheme 1).
For the study we used a special E. coli cell lysate prepared
from E. coli cells that overproduce the Lactococcus lactis forma-
midopyrimidine glycosylase protein (LlFpg, MW=33 kDa). This
protein is one of the main repair enzymes that specifically rec-
ognizes and repairs 8-oxodG lesions.[4c] For the photoaffinity
experiments we added the DNA probes (P1–P8) individually to
the E. coli cell lysate in a special photoaffinity labeling buffer
(Supporting Information) in an Eppendorf tube stored on ice
(conditions: 1 mm DNA probe, 1 mgmLꢀ1 total protein concen-
tration). The solutions were irradiated with 365 nm light (2ꢀ
15 W) in the cases of the diazirine- and benzophenone-carrying
probe duplexes and with 312 nm light (2ꢀ15 W) if a halogen-
containing nucleobase was present as the crosslinking unit.
The solutions were subsequently concentrated and analyzed
by SDS-PAGE.
[a] M. Winnacker, V. Welzmiller, R. Strasser, Prof. Dr. T. Carell
Center for Integrated Protein Science (CiPSM),
Department of Chemistry and Biochemistry
Ludwig-Maximilians-University
Butenandtstrasse 5–13, 81377 Munich (Germany)
Fax: (+49)89-2180-77756
E-mail: thomas.carell@cup.uni-muenchen.de
The results of the studies are depicted in Figure 1. All
probes formed intensive crosslinks in the gel region where
proteins with a weight between 30 kDa and 50 kDa are found.
The observed band indeed represents the expected DNA-LlFpg
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201000244.
ChemBioChem 2010, 11, 1345 – 1349
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1345

Scheme 1. Synthesis of the diazirine phosphoramidite 2 and of the DNA probes P1–P8 used in this study. a) TMDMSCl, DMAP, DMF, RT, 4 h, 95%; b) 5, Pd-
(Ph3)2Cl2, CuI, DMF, iPr2NEt3, RT, 12 h, 64%; c) 7, NaH, THF, RT, 3.5 h, 67%; d) TBAF, THF, RT, 2 h, 100%; e) DMTCl, pyridine, RT!08C, 2 h, 40%; f) (iPr)2NO-
CH2CH2CN, diisopropylammonium tetrazolate, CH2Cl2, RT, 3 h, 55%.
crosslink (large box).[11] In view of the fact that the probe du-
plexes contain several other unnatural units (Fl, Bio, crosslinker)
next to the 8-oxodG lesion, the observed selectivity is rather
surprising. Despite this, the crosslinking efficiencies of the
probes are clearly different. The most intense crosslinks were
obtained with the probes P2, P3, P5, and P7. The results ob-
tained with the control probes, which lacked 8-oxodG (G=dG)
are also depicted in Figure 1 in the (ꢀ) lanes. In these cases
1346
www.chembiochem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1345 – 1349

lesion (lane 3). Control experi-
ments performed without a DNA
probe (lane 4) or with heat-dena-
tured cell lysate (lanes 5 and 6)
again revealed no bands. For
protein identification, we added
avidin-agarose-beads to the re-
maining bigger part of the irradi-
ated samples and isolated the
formed crosslinks with the aid of
the biotin unit. The beads were
washed several times and the
protein material was cleaved
from the beads by heating the
beads in buffer (958C, 5 min;
Supporting Information). Analy-
sis of the solution by gel electro-
Figure 1. SDS-PAGE gels showing the results of the photo-crosslinking experiments with DNA probes P1–P8 (fluo-
rescence detection). Conditions: DNA probe (1 mm), 1 mgmLꢀ1 protein concentration in photoaffinity labeling
buffer (assay volume 100 mL; Supporting Information). +) Probe with 8-oxodG, ꢀ) without 8-oxodG. The ratio
value describes the intensity difference with and without 8-oxodG in the probe. The yield value describes the
amount of crosslink detected.
phoresis
(lanes 7–10)
again
showed crosslink formation with
the DNA probe containing 8-
oxodG (lane 8) but also to the
minor amounts of crosslinks are formed with other proteins in
the same mass range, or some unspecific binding of the FPG
protein to undamaged DNA might be detected. Comparison of
the (+) lanes with the (ꢀ) lanes shows that the benzophe-
none-substituted DNA probes 3 and 7 provide rather large un-
specific signals. Particularly favorable are the signal-to-noise
ratios for P2 and P5. Direct comparison of these two probes
show that P2 is able to label a significantly larger number of
proteins in a very interesting mass range (small box in
Figure 1); the crosslinker 1 in P2 is an excellent trapping
device. It should be noted that the DNA control strand P8,
which contains no photo-crosslinker, gave no DNA–protein
crosslink (lane 16); this indicates that protein association with
the trap-unit itself is indeed a minor issue and that only the
photocrosslinker moieties are able to induce DNA-protein
crosslink formation. If the experiments were performed with a
heat-denatured protein lysate, again no crosslinks were found
(data not shown); this shows that only properly folded pro-
teins are recognized by the DNA probes. Particularly notewor-
thy is the large difference observed between probes 1 and 2;
this demonstrates that the nature of the spacer unit has a tre-
mendous effect on the photo-crosslinking yield and efficiency.
With these result in hand we next started to use P2 to iden-
tify proteins that would bind specifically and directly to 8-
oxodG in a human HCT-116 cell nuclear lysate. The nuclear ex-
tract (nuclear proteome) was chosen because it should contain
the repair protein in an enriched state. The experiment was
performed by incubation of P2 (1 nmol) with the nuclear lysate
(2 mg) for 30 min. The solution was subsequently irradiated for
30 min (1 mm DNA concentration). A small amount of the assay
solution was removed, concentrated, and analyzed by analyti-
cal SDS-PAGE (Figure 2, lanes 2–6). The formation of crosslinks
is clearly evident (lane 2). Crosslinks were also obtained, how-
ever, with the control DNA strand containing no 8-oxodG
Figure 2. SDS-PAGE analysis of the photo-crosslinking experiments: analyti-
cal (lanes 2–6) and preparative (lanes 7–10) photoaffinity labeling experi-
ments with DNA probe 2, containing 8-oxo-dG (lanes 2, 5, 8, 10), and with
undamaged controls (lanes 3, 6, 9) with use of a human HCT-116 cell nuclear
lysate (lanes 5, 6, 10: heat controls).
probe lacking 8-oxodG (lane 9). Again, the control experiment
with heat-denatured cell lysate provided no crosslink (lane 10).
We next prepared slices of the preparative gel (areas a–g).
The gel slices were washed several times, the proteins were in-
dividually digested in-gel with trypsin, and the peptide mix-
tures were subsequently analyzed by nano-HPLC-ESI-MS/MS.
The obtained data set was corrected by subtraction of the
data obtained in a similar experiment performed in the ab-
sence of any probe (background subtraction, lane 7).
The corrected MS-data sets were bioinformatically compared
to the human protein database with the help of the Bio-
Works3.3.1. program with use of appropriate filters (Supporting
Information). We first discovered that the peptide fragments
ChemBioChem 2010, 11, 1345 – 1349
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1347

obtained from lane 8 and from control lane 9 were surprisingly
different (for a complete list see the Supporting Information).
We found proteins that were present in both lanes, due to
unspecific binding, and in addition a set of proteins that were
only present in lane 8 and hence specifically associated with 8-
oxodG. The observed specific proteins are compiled in Table 1.
PARP to DNA duplexes containing small oxidative lesions was
indeed observed.[16]
Other proteins that are not listed in Table 1 were found to
bind to the 8-oxodG-containing and -noncontaining DNA with
similar affinities. These unspecific proteins include, for example,
the FUSE-binding protein (P value 3.0ꢀ10ꢀ13, eight pep-
tides),[17] the ATP-dependant DNA helicase II (P value 2.5ꢀ10ꢀ11
,
12 peptides), and the elongation factor 2b (P value 9.2ꢀ10ꢀ6,
13 peptides). All these proteins most likely bind in general to
DNA or to other structures present in the DNA probes, such as
the photoaffinity trap, the fluorescence tag, or the biotin label.
In summary, we have developed a new DNA photo-crosslink-
er capable of efficient crosslinking of DNA-binding proteins to
DNA in photoaffinity labeling experiments. The crosslinker
shows a good selectivity profile. With DNA duplexes contain-
ing an 8-oxodG lesion next to the crosslinker it was possible to
identify a set of proteins in the human nuclear proteome that
are involved in processing the lesion. As well as proteins such
as hOGG1 that are known to recognize the lesion, other inter-
esting proteins such as PARP or the APEX nuclease were also
found to be involved in lesion processing. We believe that this
photo-crosslinker, in combination with other DNA lesions and
mass spectrometry methods,[18] should allow scientists to char-
acterize proteins that are specifically involved in DNA repair
processes in detail.
Table 1. Proteins in human HCT-116 cell nuclear lysate that were photo-
crosslinked to the DNA probe 2. The values are the output of the Bio-
Works3.3.1. software. The sequence coverage values were found to be
between 10 and 50%. P-value: artifact probability; NCBI-ID: Identification
number in the NCBI database. MS/MS spectra of some selected peptides
from these proteins are shown in the Supporting Information.
Gel area Protein
P value
NCBI-ID
N[a] MW [kDa]
a, d
d
d, e
PARP-1
5.1ꢀ10ꢀ11 gij130781
2.3ꢀ10ꢀ11 gij219474
1.0ꢀ10ꢀ12 gij35053
5
4
8
113
36
35
APEX nuclease
uracil DNA
glycosylase
HMG-1
hOGG1 isoforms 6.8ꢀ10ꢀ12 gij8670532 10
e
e
4.9ꢀ10ꢀ8
gij968888
10
33
36–45
[a] Number of peptides found
Gel area e in lane 8 contains the crosslinked human protein
hOGG1. This protein is a known 8-oxodG repair protein.[12] The
finding that this protein provides such a strong signal shows
that the method, and in particular P2 with the new crosslinker
1, provides important and meaningful results. hOGG1 was not
detected in the gel slice (e) obtained from lane 9. This shows
that the photoaffinity trap is able to specifically capture pro-
teins associated with the lesion.
Acknowledgements
We thank the excellence cluster CiPSM and the DFG (SFB 646 and
SFB 749) for generous support. We thank Claudia Grꢀf for provid-
ing the HCT-116 cell nuclear lysate.
Other interesting proteins that are specifically associated
with the 8-oxodG are the high mobility group 1 proteins (HMG-
1). They recognize aberrant structures such as DNA kinks.[13]
HMG proteins have previously been recognized to be involved
in the binding of cisplatin lesions.[6c] Here we have shown that
the rather small structural changes introduced by the 8-oxodG
lesion are obviously sufficient to induce statistically relevant
binding of the HMG proteins as well.
Keywords: 8-oxodG
· diazirines · mass spectrometry ·
photoaffinity labeling · proteomics
[1] a) T. Lindahl, Nature 1993, 362, 709–715; b) E. C. Friedberg, G. C. Walker,
W. Siede, DNA Repair and Mutagenesis, ASM Press, Washington D.C.,
1995.
[2] a) J. Cadet, S. Bellon, M. Berger, A.-G. Bourdat, T. Douki, V. Duarte, S.
Frelon, D. Gasparutto, E. Muller, J.-L. Ravanat, S. Sauvaigo, Biol. Chem.
2002, 383, 933–943; b) S. S. David, V. L. O’Shea, S. Kundu, Nature 2007,
447, 941–950.
[3] a) C. J. Burrows, J. G. Muller, Chem. Rev. 1998, 98, 1109–1151; b) O. D.
Schꢂrer, Angew. Chem. 2003, 115, 3052–3082; Angew. Chem. Int. Ed.
2003, 42, 2946–2974.
[4] a) A. Banerjee, W. L. Santos, G. L. Verdine, Science 2006, 311, 1153–1157;
b) A. Banerjee, W. Yang, M. Karplus, G. L. Verdine, Nature 2005, 434,
612–618; c) F. Coste, M. Ober, T. Carell, S. Boiteux, C. Zelwer, B. Cas-
taing, J. Biol. Chem. 2004, 279, 44074–44083; d) F. Coste, M. Ober, Y. V.
Le Bihan, M. A. Izquierdo, N. Hervouet, H. Mꢃller, T. Carell, B. Castaing,
Chem. Biol. 2008, 15, 706–717; e) J. C. Fromme, G. L. Verdine, J. Biol.
Chem. 2003, 278, 51543–51548; f) J. M. Krahn, W. A. Beard, H. Miller,
A. P. Grollman, S. H. Wilson, Structure 2003, 11, 121–127.
[5] a) R. Aebersold, M. Mann, Nature 2003, 422, 198–207; b) B. F. Cravatt,
G. M. Simon, J. R. Yates III, Nature 2007, 450, 991–1000; c) S. A. Sieber,
B. F. Cravatt, Chem. Commun. 2006, 2311–2319.
[6] a) J. Brunner, Annu. Rev. Biochem. 1993, 62, 483–514; b) K. M. Meisen-
heimer, T. H. Koch, Crit. Rev. Biochem. Mol. Biol. 1997, 32, 101–140;
c) E. R. Guggenheim, D. Xu, C. X. Zhang, P. V. Chang, S. J. Lippard, Chem-
BioChem 2009, 10, 141–157; d) J. T. Reardon, A. Sancar, Mol. Cell. Biol.
2002, 22, 5938–5945; e) E. A. Maltseva, N. I. Rechkunova, L. C. Gillet,
Furthermore, in the gel areas d and e we detected statistical-
ly relevant binding of uracil glycosylase, which is a major
repair glycosylase, and of the APEX nuclease. This result is
interesting because the APEX protein is part of the cellular
response to oxidative stress and protects cells from the geno-
toxic and cytotoxic effect of oxidizing agents.[14]
MS analysis of the protein digest of gel area a provided an-
other surprise. The PARP protein also shows statistically rele-
vant selectivity for the 8-oxodG lesion and in this study it was
not found in the experiment performed with DNA probes lack-
ing 8-oxodG. This protein associates with DNA breaks and with
large aberrant DNA structures such as those induced by cispla-
tin.[6c] The protein binds to the lesioned site and starts to poly-
merize NAD. This is thought to be a signal that recruits repair
factors to the damaged site.[15] Our experiments now show
that this protein may also be involved in recognition of small
non-helix-disturbing lesions such as 8-oxodG. This observation
is supported by a recent in vitro study in which binding of
1348
www.chembiochem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1345 – 1349

I. O. Petruseva, O. D. Schꢂrer, O. I. Lavrik, Biochim. Biophys. Acta Gen.
Subj. 2007, 1770, 781–789; f) M. Liebmann, F. Di Pasquale, A. Marx,
ChemBioChem 2006, 7, 1965–1969; g) C. X. Zhang, P. V. Chang, S. J. Lip-
pard, J. Am. Chem. Soc. 2004, 126, 6536–6537; h) O. I. Lavrik, R. Prasad,
R. W. Sobol, J. K. Horton, E. J. Ackerman, S. H. Wilson, J. Biol. Chem.
2001, 276, 25541–25548; i) U. Schweizer, T. Hey, G. Lipps, G. Krauss, Nu-
cleic Acids Res. 1999, 27, 3183–3189.
[12] J. P. Radicella, C. Dherin, C. Desmaze, M. S. Fox, S. Boiteux, Proc. Natl.
Acad. Sci. USA 1997, 94, 8010–8015.
[13] M. Stros, D. Launholt, K. D. Grasser, Cell. Mol. Life Sci. 2007, 64, 2590–
2606.
[14] G. Fritz, Int. J. Biochem. Cell Biol. 2000, 32, 925–929.
[15] S. A. Andrabi, N. S. Kim, S.-W. Yu, H. Wang, D. W. Koh, M. Sasaki, J. A.
Klaus, T. Otsuka, Z. Zhang, R. C. Koehler, P. D. Hurn, G. G. Poirier, V. L.
Dawson, T. M. Dawson, Proc. Natl. Acad. Sci. USA 2006, 103, 18308–
18313.
[7] a) M. Hashimoto, Y. Hatanaka, Eur. J. Org. Chem. 2008, 2513–2523;
b) U. K. Shigdel, J. Zhang, C. He, Angew. Chem. 2008, 120, 96–99;
Angew. Chem. Int. Ed. 2008, 47, 90–93; c) J. J. Tate, J. Persinger, B. Bar-
tholomew, Nucleic Acids Res. 1998, 26, 1421–1426.
[16] J. L. Harris, B. Jakob, G. Taucher-Scholz, G. L. Dianov, O. J. Becherel, M. F.
Lavin, Human Mol. Genet. 2009, 18, 4102–4117.
[17] L. R. Benjamin, H.-J. Chung, S. Sanford, F. Kouzine, J. Liu, D. Levens, Proc.
Natl. Acad. Sci. USA 2008, 105, 18296–18301.
[18] M. Mann, Nat. Rev. Mol. Cell Biol. 2006, 7, 952–958.
[8] L. Ballell, K. J. Alink, M. Slijper, C. Versluis, R. M. J. Liskamp, R. J. Pieters,
ChemBioChem 2005, 6, 291–295.
[9] A. Baccaro, S. H. Weisbrod, A. Marx, Synthesis 2007, 13, 1949–1954.
[10] J. Brunner, H. Senn, F. M. Richards, J. Biol. Chem. 1980, 255, 3313–3318.
[11] M. Winnacker, S. Breeger, R. Strasser, T. Carell, ChemBioChem 2009, 10,
109–118.
Received: April 19, 2010
Published online on June 10, 2010
ChemBioChem 2010, 11, 1345 – 1349
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1349