DNA methyltransferase-moderated click chemistry

Article abstract of DOI:10.1021/ol0504749

(Chemical Equation Presented) Biological methylation plays a vital role in regulatory mechanisms of gene transcription. Methylation of both promoter sequences within the genome, as well as protein substrates, has a profound impact upon gene transcription.

Full text of DOI:10.1021/ol0504749

ORGANIC
LETTERS
2
005
Vol. 7, No. 11
141-2144
DNA Methyltransferase-Moderated Click
Chemistry
2
Rachel L. Weller and Scott R. Rajski*,‡
†
The School of Pharmacy and Department of Chemistry, UniVersity of Wisconsin,
Madison, Wisconsin 53705
srrajski@pharmacy.wisc.edu
Received March 4, 2005
ABSTRACT
Biological methylation plays a vital role in regulatory mechanisms of gene transcription. Methylation of both promoter sequences within the
genome, as well as protein substrates, has a profound impact upon gene transcription. Yet, few tools exist by which to identify sites of
biological methylation in complex biological mixtures. We have generated a novel adenosine-derived N-mustard that serves as an efficient
synthetic cofactor and allows for subsequent “click” chemistry involving the modified nucleic acid substrate.
Manifolds for the sequence-specific recognition and modi-
fication of DNA have proven to be vital for structural and
functional studies of DNA, as well as the pursuit of rationally
designed therapeutic agents. Successful strategies for sequence-
selective targeting of DNA include triplex-forming oligode-
oxyribonucleotides (ODNs), designer zinc finger proteins,
directing delivery of the new alkyl group through association
with individual target sequences. The transferred methyl
group is derived from S-adenosyl-L-methionine (SAM).2
SAM is routinely coined “mother nature’s methyl iodide”
for good reason. The agent is a potent and potentially
nonspecific alkylating agent. However, under the direction
of MTases, SAM is a highly selective alkylating agent
capable of DNA, RNA, and protein modification and is an
important biosynthetic tool for secondary metabolite produc-
1
and most notably, synthetic polyamides. The mechanisms
of recognition in each of these examples are unique but
contingent upon a series of temporary interactions between
the modifying agent and the substrate. We report here a
method for covalent site-specific DNA alkylation mediated
by methyltransferases (MTases) that enables subsequent
chemoselective ligation via the Huisgen [2 + 3] cycloaddi-
tion.
3
tion. Indeed, SAM-dependent methylation of DNA and
proteins is now a major theme in epigenomics. Understanding
the role that biological methylation plays, particularly in gene
transcription, has been hampered by a general lack of tools
for identification of MTase substrates. The largest impedi-
ment arguably stems from the methyl group’s relative
absence of functionality, which renders it difficult to identify
in complex biological environments.
DNA MTases combine sequence-specific recognition with
covalent modification to act as molecular “third parties”,
†
Department of Chemistry.
The School of Pharmacy.
‡
(2) Cheng, X.; Roberts, R. J. Nucleic Acids Res. 2001, 29, 3784-3795.
(3) Blackburn, G. M.; Gamblin, S. J.; Wilson, J. R. HelV. Chem. Acta
2003, 86, 4000-4006.
(
1) Uil, T. G.; Haisma, H. J.; Rots, M. G. Nucleic Acids Res. 2003, 31,
6
064-6078.
1
0.1021/ol0504749 CCC: $30.25
© 2005 American Chemical Society
Published on Web 05/05/2005

Figure 1. Denaturing polyacrylamide gel electrophoresis (DPAGE) of DNA alkylation reactions with synthetic oligonucleotides and 4
relative to 1a). â ) native unmodified DNA, and $ ) DNA alkylated with 1a or 4. Each series of reactions/lanes 1-11 are as follows:
, DNA-only control; 2, DNA + 100 µM 1a (no enzyme added); 3, DNA + 1 µM 1a + M.TaqI (to 1 µM); 4, DNA + 10 µM 1a +
M.TaqI; 5, DNA + 50 µM 1a + M.TaqI; 6, DNA + 100 µM 1a + M.TaqI. Lanes 7-11 were same as 2-6 except that alkyne cofactor
was used. Right-hand panel lanes are identical in loading to left-hand loadings with the exception that M.EcoRI was used instead of
(
1
4
M.TaqI. Please note that for all reactions, the concentration of enzyme used was 1 µM. Controls accounting for cofactor solvent and/or
appropriate MTase did not result in electrophoretically retarded materials (data not shown).
Aziridine adenylates 1a-c take part in MTase-dependent
activation toward nucleophiles but also cation-π interactions
DNA alkylations (Scheme 1).4 Moreover, azido adenylates
-6
between cofactor and MTase, which are a trademark of
8
SAM-MTase complexes. We sought to exploit this phe-
nomenon in the design of a novel synthetic cofactor devoid
of the inherently labile aziridine moiety of 1a-c. Hydro-
chloride salt 4 was envisioned to rapidly form aziridinium 5
in situ, thus avoiding synthetic difficulties associated with
intact aziridines. This intermediate was expected to be more
reactive and more amenable to MTase-promoted chemistry
by virtue of 5′ amine quaternization via aziridination instead
of a potentially reversible protonation (as has been proposed
for materials such as 1a-c). Iodide 4 was designed with the
propargyl substituent to allow for possible bioconjugation
of the DNA substrate postalkylation.
Scheme 1. 5′-Aziridine Adenylates as MTase-Dependent
DNA-Modifying Agents
In evaluating the hypothesis that 4 (and related agents)
could serve as an effective cofactor for MTases, we were
initially highly skeptical due to the unknown impact of the
alkynyl substituent upon cofactor:MTase interactions. Indeed,
1
b and 1c allow the conversion of DNA MTases into
6
azidonucleoside transferases. Substrates of azidation are the
same as those ordinarily acted upon by MTases. Unlike the
methyl group, azides provide a chemically unique handle to
which other probes (radioisotopes, affinity matrix handles,
etc.) can be linked under biologically amenable conditions.
The sequence selectivity of DNA azidation is accomplished
by DNA MTases, and substrates modified with either 1b or
1
ligation chemistry.
Aziridines 1a-c are believed to undergo quaternization
of the 5′ amine followed by MTase binding, delivery to the
site of methylation, and subsequent aziridinium ring opening
with concomitant substrate alkylation. Initial generation of
the positively charged aziridinium is consistent not only with
4
has so far proven to be completely devoid of DNA-
damaging activity in the presence of the cytosine C5 MTase
M.HhaI. In contrast, we have found that 4 is highly amenable
to use by two different N6 adenine MTases.
7
Figure 1 shows that the activity of 4 closely parallels that
of 1a with both M.TaqI and M.EcoRI. To evaluate M.TaqI
activity, the ODN 5′-TGAATCTCGAGCACCC-3′ was 5′-
c can be elaborated postenzymatically via Staudinger
32
end labeled with P, gel purified, and annealed to comple-
6
mentary ODN 3′-AAACTTAGAGCTCGTGGG-5′. M.TaqI
ordinarily methylates each adenine N6 (italicized) within the
palindromic sequence shown in bold. Evaluation of M.EcoRI
activity called for 5′-end labeling of 5′-TGAATGAATTC-
GACCC-3′ followed by gel purification and annealing to the
complement 3′-AAACTTACTTAAGCTGGG-5′. M.EcoRI
ordinarily methylates each adenine N6 (italicized) within the
bold-faced palindrome. Duplex substrates were incubated
with the corresponding enzyme and either the aziridine 1a
(
4) Pignot, M.; Siethoff, C.; Linscheid, M.; Weinhold, E. Angew. Chem.,
Int. Ed. Engl. 1998, 37, 2888-2891.
5) Pignot, M.; Siethoff, C.; Linscheid, M.; Weinhold, E. European Patent
WO 00/06578, 2000.
6) Comstock, L. R.; Rajski, S. R. Nucleic Acids Res. 2005, 33, 1644-
652.
7) (a) Budisa, N. ChemBioChem. 2004, 5, 1176-1179. (b) Koehn, M.;
Breinbauer, R. Angew. Chem., Int. Ed. 2004, 43, 3106-3116.
(
(
1
(8) Fauman, E. B.; Blumenthal, R. M.; Cheng, X. In S-Adenosyl-
methionine-Dependent Methyltransferases: Structures and Function; Cheng,
X., Blumenthal, R. M., Eds.; World Scientific: Singapore, 1999; pp 1-13.
(
2142
Org. Lett., Vol. 7, No. 11, 2005

or hydrochloride 4. Incubations were then quenched via
addition of tRNA (40 µg/reaction) and repetitive EtOH
precipitation. Notably, a second, more slowly moving DNA
adduct is observed in reactions of M.EcoRI. This observation
is consistent with work previously reported by Jeltsch and
co-workers in which M.EcoRI was found to display signifi-
cant promiscuity. The precise origin and sequence selectivity
of modification of this very slow mobility adduct is under
active investigation.
4
as reductants for CuSO . The number of examples with
nucleic acids is still relatively small presumably because of
the well-known role that Cu(I) can play in Haber-Weiss
1
1,12
redox cycling to produce hydroxyl radical.
However, it
was not obvious that the Cu(I)-TTA complex can promote
Haber-Weiss redox cycling. Moreover, if Cu(I)-TTA does
permit hydroxyl radical formation, it is likely that t-BuOH
present in the reaction would immediately sequester radical
intermediates, thus circumventing DNA strand scission. As
such, we envisioned the scenario in Scheme 3 where DNA
9
Scheme 2. Proposed Aziridinium Ion Formation
Scheme 3. Proposed Huisgen Cycloaddition with Alkyne
Modified DNA
To ensure that the regiochemistry of DNA alkylation with
4
is the same as that normally observed for SAM-dependent
methylation, we performed experiments involving prem-
ethylated substrates (Supporting Information). Alkylation
reactions involving 4 were carried out on duplexes enzymati-
cally methylated with SAM prior to treatment with fresh
MTase and 4. Confirmation that methylation of both M.TaqI
and M.EcoRI sequence-bearing duplexes had taken place was
accomplished by subjection of these DNAs to TaqI and
EcoRI restriction endonucleases. DNAs subjected to SAM
and MTase failed to act as sites of endonucleolytic cleavage;
all other DNAs did, thus, confirming the integrity of substrate
methylation. Similarly, methylated duplexes failed to undergo
reaction when presented with fresh MTase and 50 µM 4.
That substrate DNA methylation precluded alkylation by 4
indicates that MTase-driven substrate alkylations by 4 and
SAM share the same regiochemistry. This efficient and
predictable site-specific alkylation of nucleic acids paved the
way for novel elaboration of the modified sites by virtue of
the propargyl substituent of 4.
Bioconjugation via alkyne moieties is ideally suited to the
Huisgen [2 + 3] cycloaddition. This abiotic chemoselective
ligation of azides and alkynes affords highly stable triazole
moieties under biologically relevant conditions.10 The reac-
tion has been extensively used in a number of biological
motifs; the most effective incarnation calls for the use of
Cu(I) and a unique tris-triazaloylamine (TTA) ligand along
with either ascorbate or tris-carboxyethylphosphine (TCEP)
lesion 7 could render triazole 8 following reaction with alkyl
azides.
The viability of the coupling indicated by Scheme 3 is
supported by Figure 2. DNA modified with 4 (in an
M.EcoRI-dependent fashion) underwent extremely facile
coupling to azide 9 under aqueous conditions.13 The slow
mobility adduct formed (ø) is proposed to be the result of
Huisgen [2 + 3] cycloaddition since the product is not
apparent in reactions lacking alkyne or azide. This is best
reflected in comparing lanes 3 to 5 and then 5 to 6 for both
timepoints. Subjection of DNA modified with 1a to azide, a
1
:1 complex of Cu-TTA, and TCEP afforded no new
radiolabeled product. The same conditions applied to DNA
modified with alkyne 4 rendered the new DNA product noted
(11) (a) Seo, T. S.; Li, Z.; Ruparel, H.; Ju, J. J. Org. Chem. 2003, 68,
609-612. (b) Seo, T. S.; Bai, X.; Ruparel, H.; Li, Z.; Turro, N. J.; Ju, J.
Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 5488-5493. (c) Poulin-Kerstein,
A. T.; Dervan, P. B. J. Am. Chem. Soc. 2003, 51, 15811-15821.
(12) Chen, C. B.; Milne, L.; Landgraf, R.; Perrin, D. M.; Sigman, D. S.
(
9) Jeltsch, A.; Christ, F.; Fatemi, M.; Roth, J. J. Biol. Chem. 1999, 274,
9538-19544.
10) Breinbauer, R.; K o¨ hn, M. ChemBioChem 2003, 4, 1147-1149.
ChemBioChem. 2001, 2, 735-740.
1
(13) Lee, J. W.; Jun, S. I.; Kim, K. Tetrahedron Lett. 2001, 42, 2709-
(
2711.
Org. Lett., Vol. 7, No. 11, 2005
2143

Figure 3. DPAGE analysis of click reactions of 4-modified DNA
and azides 9-11. Lane A contained DNA-4 conjugate standard,
and lanes 1-3 contained reactions of 4-modified DNA with 2.5
mM 9, 10, and 11, respectively
Figure 2. DPAGE analysis of click reactions between pyrene azide
and synthetic oligonucleotides modified either with 1a or 4. Lane
9
A is a DNA-4 conjugate control. Lanes 1 and 2 are reactions of
DNA treated with M.EcoRI but no cofactor. Lanes 3 and 4 are
reactions involving DNA subjected to M.EcoRI and 1a. Lanes 5
and 6 are reactions using DNA treated with M.EcoRI and alkyne
cofactor 4. Reactions assayed in lanes 1-6 all contained 80 nM
differences among the variously modified DNAs. Subsequent
click reactions performed with 10 and 11 versus 9 (all other
factors the same) afforded products again denoted by “ø” in
lanes 1-3. Particularly striking is that coupled product
bearing the piperazine moiety of 11 is significantly more
reduced in electrophoretic mobility than are products formed
with 9 or 10. Protonation of the piperazine nitrogen during
electrophoresis is the likely cause of this difference. Com-
parison of the same reactions carried out for shorter times
revealed very similar rates of product formation (data not
shown). Importantly, these results refute the possibility that
the data from Figure 2 were dependent upon the pyrene
moiety of 9 and potential redox activities of the polycyclic
moiety.
3 4
DNA duplex, 20 mM NaHCO , 1 mM CuSO :TTA complex, 2
mM TCEP, and 5% t-BuOH. All reactions were conducted at room
temperature, and the time of reaction is noted above each series of
lanes. Reactions monitored in odd-numbered lanes contained 2.5
mM azide 9; even-numbered lanes/reactions lacked azide. In this
manner, it was deduced that the generation of slow mobility material
designated with ø is azide dependent.
by ø. Similarly, subjection of DNA modified with 4 to
reaction conditions lacking 9 failed to render ø. We believe
this to be the first entirely solution-phase1 example of
nucleic acid-based click chemistry that does not call for
1a
These data highlight two important concepts. First, syn-
thetic cofactors need not be restricted only to 5′-aziridines,
which have proven to be rather elusive synthetic targets. By
demonstrating the viability of N-mustards such as 4, we hope
to hasten the rate at which these agents undergo further
development as biochemical tools. In this vein, it is
significant that 4 has been used very effectively to protect
linearized pUC19 DNA from R.TaqI digestion following
brief treatments with M.TaqI (unpublished data). Thus,
MTase-dependent deliVery of 4 to DNA is not restricted to
small duplexes. Second, DNA can take part in click
chemistry, and synthetic cofactor 4 represents a novel means
by which to perform such chemistry in a sequence-selective
manner.
extensive heating1 or templating effects. It is significant
that during the course of EtOH precipitations following
cycloaddition, we did not observe any isotope loss to
supernatant, consistent with little or no Cu(I)-mediated DNA
strand scission. This, in combination with the appearance of
very clean DNA bands, suggests that, indeed, DNA can serve
as an effective participant in Cu(I)-catalyzed click chemistry.
We have also noted that ascorbate can be effectively
substituted for TCEP without appreciable nucleic acid
degradation (data not shown).
1b
11c
Coupling of alkyne-modified DNA and 9 proceeded very
quickly as evidenced by the amount of product observed in
lane 5 of Figure 2. Preliminary experiments (data not shown)
suggest this is largely a function of Cu(I) availability. For
solubility reasons, we generated a 1:1 complex of TTA ligand
and CuSO prior to cycloaddition. This contrasts with most
4
click chemistry examples in which TTA and a Cu(I) source
are added independently. In our hands, independent addition
Acknowledgment. We thank Ms. Lindsay R. Comstock
for TTA, Derek Robinson (New England Biolabs) for
samples of M.TaqI, and University of Wisconsin for support
of this research.
4
of CuSO and TTA had a profoundly detrimental impact on
Supporting Information Available: Experimental pro-
cedures and characterization for substances 4 and 9-11 and
biological protocols relating to Figures 1-3. This material
is available free of charge via the Internet at http://pubs.acs.org.
coupling efficiency relative to the use of premixed CuSO
TTA.
4
:
The generality of the cycloaddition is depicted in Figure
3. Azides 10 and 11 were prepared with the expectation that
size and charge differences would result in visible mobility
OL0504749
2144
Org. Lett., Vol. 7, No. 11, 2005