A versatile toolbox for variable DNA functionalization at high density

Article abstract of DOI:10.1021/ja051725b

To broaden the applicability of chemically modified DNAs in nano- and biotechnology, material science, sensor development, and molecular recognition, strategies are required for introducing a large variety of different modifications into the same nucleic acid sequence at once. Here, we investigate the scope and limits for obtaining functionalized dsDNA by primer extension and PCR, using a broad variety of chemically modified deoxynucleotide triphosphates (dNTPs), DNA polymerases, and templates. All natural nucleobases in each strand were substituted with up to four different base-modified analogues. We studied the sequence dependence of enzymatic amplification to yield high-density functionalized DNA (fDNA) from modified dNTPs, and of fDNA templates, and found that GC-rich sequences are amplified with decreased efficiency as compared to AT-rich ones. There is also a strong dependence on the polymerase used. While family A polymerases generally performed poorly on "demanding" templates containing consecutive stretches of a particular base, family B polymerases were better suited for this purpose, in particular Pwo and Vent (exo-) DNA polymerase. A systematic analysis of fDNAs modified at increasing densities by CD spectroscopy revealed that single modified bases do not alter the overall B-type DNA structure, regardless of their chemical nature. A density of three modified bases induces conformational changes in the double helix, reflected by an inversion of the CD spectra. Our study provides a basis for establishing a generally applicable toolbox of enzymes, templates, and monomers for generating high-density functionalized DNAs for a broad range of applications.

Full text of DOI:10.1021/ja051725b

Published on Web 10/08/2005
A Versatile Toolbox for Variable DNA Functionalization at
High Density
Stefan Ja¨ger, Goran Rasched, Hagit Kornreich-Leshem, Marianne Engeser,
Oliver Thum,† and Michael Famulok*
Contribution from the Kekuke´-Institut fu¨r Organische Chemie und Biochemie, UniVersita¨t Bonn,
Gerhard-Domagk-Strasse 1, D-53121 Bonn, Germany
Received March 18, 2005; E-mail: m.famulok@uni-bonn.de
Abstract: To broaden the applicability of chemically modified DNAs in nano- and biotechnology, material
science, sensor development, and molecular recognition, strategies are required for introducing a large
variety of different modifications into the same nucleic acid sequence at once. Here, we investigate the
scope and limits for obtaining functionalized dsDNA by primer extension and PCR, using a broad variety
of chemically modified deoxynucleotide triphosphates (dNTPs), DNA polymerases, and templates. All natural
nucleobases in each strand were substituted with up to four different base-modified analogues. We studied
the sequence dependence of enzymatic amplification to yield high-density functionalized DNA (fDNA) from
modified dNTPs, and of fDNA templates, and found that GC-rich sequences are amplified with decreased
efficiency as compared to AT-rich ones. There is also a strong dependence on the polymerase used. While
family A polymerases generally performed poorly on “demanding” templates containing consecutive stretches
of a particular base, family B polymerases were better suited for this purpose, in particular Pwo and Vent
(exo-) DNA polymerase. A systematic analysis of fDNAs modified at increasing densities by CD spectroscopy
revealed that single modified bases do not alter the overall B-type DNA structure, regardless of their chemical
nature. A density of three modified bases induces conformational changes in the double helix, reflected by
an inversion of the CD spectra. Our study provides a basis for establishing a generally applicable toolbox
of enzymes, templates, and monomers for generating high-density functionalized DNAs for a broad range
of applications.
Introduction
For further acceleration of these developments and to expand
the scope of applications of nucleic acids in nanobiotechnology,
The nanosciences currently face a growing interest in applying
nucleic acids and analogues as building blocks for the bottom-
up construction of self-assembling functional objects in nano-
meter dimensions. The advantage of utilizing nucleic acids for
these purposes lies in the logics and predictability of nucleobase
pairing through complementary hydrogen-bond donors and
acceptors. The encoded self-assembly of complementary nucleic
acid strands to geometrically well-defined duplex structures has
led to the generation of a variety of two- or three-dimensional
topologies, supramolecular assemblies, and nanomechanical
objects.1 Moreover, nucleic acids possess supreme template
properties that allow for their enzymatic replication and
amplification by polymerases and the application of in vitro
evolution technologies to select for nucleic acid-based sensors,2
diagnostics,3 therapeutics,4 and catalysts.5
it is necessary to introduce additional functionalities into
oligonucleotides. This can be achieved either by utilizing bound
cofactors6 or by direct chemical derivatization of the building
blocks. It has been shown that the introduction of modified
DNA-monomers can significantly alter the physical and geo-
metrical properties of nucleic acids.7 Furthermore, additional
chemical diversity can result in oligonucleotides containing
sequence- and site-specific modifications that might exert
additional functions or interactions that go beyond simple
Watson-Crick pairing. These can be used, for example, as
probes for elucidating functional and structural properties of
DNA polymerases,8 for RNA structure-function analysis,9 for
(4) Nimjee, S. M.; Rusconi, C. P.; Sullenger, B. A. Annu. ReV. Med. 2005,
56, 555-583.
(5) (a) Eaton, B. E.; Pieken, W. A. Annu. ReV. Biochem. 1995, 64, 837-863.
(b) Perrin, D. M. Comb. Chem. High Throughput Screening 2000, 3, 243-
269. (c) Vaish, N. K.; Fraley, A. W.; Szostak, J. W.; McLaughlin, L. W.
Nucleic Acids Res. 2000, 28, 3316-3322. (d) Santoro, S. W.; Joyce, G.
F.; Sakthivel, K.; Gramatikova, S.; Barbas, C. F., III. J. Am. Chem. Soc.
2000, 122, 2433-2439. (e) Tarasow, T. M.; Kellogg, E.; Holley, B. L.;
Nieuwlandt, D.; Tarasow, S. L.; Eaton, B. E. J. Am. Chem. Soc. 2004,
126, 11843-11851. (f) Sidorov, A. V.; Grasby, J. A.; Williams, D. M.
Nucleic Acids Res. 2004, 32, 1591-1601. (g) Famulok, M.; Mayer, G.
ChemBioChem 2005, 6, 19-26.
† Present address: Degussa Care Specialties B-CS RD, Goldschmidt AG,
Goldschmidtstrasse 100, D-45127 Essen, Germany.
(1) (a) Seeman, N. C. Trends Biotechnol. 1999, 17, 437-443. (b) Mirkin, C.
A. Inorg. Chem. 2000, 39, 2258-2272. (c) Seeman, N. C. Nature 2003,
421, 427-431. (d) Niemeyer, C. M. Angew. Chem., Int. Ed. 2001, 40,
4128-4158. (e) Gugliotti, L. A.; Feldheim, D. L.; Eaton, B. E. Science
2004, 304, 850-852. (f) Wengel, J. Org. Biomol. Chem. 2004, 2, 277-
280.
(2) Hesselberth, J.; Robertson, M. P.; Jhaveri, S.; Ellington, A. D. J. Biotechnol.
2000, 74, 15-25.
(3) Jayasena, S. D. Clin. Chem. 1999, 45, 1628-50.
(6) (a) Manimala, J. C.; Wiskur, S. L.; Ellington, A. D.; Anslyn, E. V. J. Am.
Chem. Soc. 2004, 126, 16515-16519. (b) Roth, A.; Breaker, R. R. Proc.
Natl. Acad. Sci. U.S.A. 1998, 95, 6027-6031.
9
10.1021/ja051725b CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 15071-15082
15071

A R T I C L E S
Ja¨ger et al.
template-directed synthesis,10 for controlled self-assembly,11 for
facilitating charge transfer in DNA,12 to detect the presence of
abasic sites in a DNA,13 and other purposes.14 Another interest-
ing field of application of functionalized nucleic acids is the in
vitro evolution of aptamers, DNAzymes, or ribozymes for
molecular recognition and catalysis.5 The goal in this research
area is to generate molecules that are both easy to copy and
vary, like DNA, and chemically adept, like proteins.
Chemical modifications can be incorporated into oligonucle-
otides via the heterocyclic nucleobases, in the sugar unit, or at
the backbone level. While short highly modified DNA strands
with less than 50-60 nucleotides are accessible through
chemical solid-phase synthesis, the generation of longer modi-
fied oligonucleotides following this route is either very difficult,
or impossible. In particular, the construction of longer DNA
strands that contain various different modifications throughout
the molecule remains elusive. Thus, strategies that allow one
to introduce a large variety of different modifications into the
same nucleic acid sequence at once might further expand the
applicability of chemically modified DNAs. The ability to
enzymatically replicate a DNA or RNA molecule that is
functionalized at high density in a template-directed fashion
would provide access to longer sequences that are difficult to
obtain synthetically. This is useful, for example, for generating
combinatorial libraries of DNA molecules with expanded chem-
ical functionality from which aptamers and catalysts with ex-
panded structural and functional properties can be isolated by
in vitro evolution.5 Such methodology is also required for multi-
plex sequencing,15 and other purposes such as the incorporation
of artificial base pairs for extending the genetic alphabet.16
It has been shown that up to two different modified
nucleotides can be enzymatically polymerized by primer exten-
sion reactions or polymerase chain reactions (PCR) on a DNA
template.15,17 Earlier, we have reported the first enzymatic
synthesis of a high-density functionalized DNA (fDNA) in
which every base position was equipped with a different
functional group. This fDNA was generated by primer extension
using a natural model template of almost equal base distribu-
tion.18 Recently, we could show that a high-density fDNA could
in turn serve as a template for the polymerase-mediated
generation of a fully functionalized DNA double-strand and can
be amplified by PCR under certain conditions.19
To further expand the scope of the enzymatic generation and
amplification of functionalized DNA and to also explore
potential limitations of this approach, we have now synthesized
an increased set of modified dNTPs and investigated their
template-directed enzymatic incorporation into fDNA. We
provide a systematic analysis of the sequence requirements by
investigating more demanding DNA templates that consist of
multiple nucleotide repetitions. Likewise, we assayed a variety
of polymerases for their capability to incorporate modified
dNTPs in different sequence contexts and identified the most
generic ones. We show that a broad variety of chemical
modifications can be tolerated by the identified DNA poly-
merases under certain conditions, regardless of their chemical
nature. We further investigated whether functionalized DNA
double-strands can be amplified from the demanding templates
by PCR. We also performed an analysis of possible structural
alterations of functionalized DNA duplexes as a function of the
nature and density of the modification. With this analysis, we
have established a platform for the generation and amplification
of DNAs functionalized at various densities with different chem-
ical modifications for a large variety of potential applications.
(7) (a) Sorensen, M. D. K., L.; Bryld, T.; Hakansson, A. E.; Verbeure, B.;
Gaubert, G.; Herdewijn, P.; Wengel, J. J. Am. Chem. Soc. 2002, 124, 2164-
2176. (b) Declercq, R.; Van Aerschot, A.; Read, R. J.; Herdewijn, P.; Van
Meervelt, L. J. Am. Chem. Soc. 2002, 124, 928-933. (c) Summerer, D.;
Marx, A. J. Am. Chem. Soc. 2002, 124, 910-911. (d) Tanaka, K.; Tengeiji,
A.; Kato, T.; Toyama, N.; Shionoya, M. Science 2003, 299, 1212-1213.
(e) Zhu, L.; Lukeman, P. S.; Canary, J. W.; Seeman, N. C. J. Am. Chem.
Soc. 2003, 125, 10178-10179. (f) Sorensen, M. D.; Petersen, M.; Wengel,
J. Chem. Commun. 2003, 2130-2131. (g) Brotschi, C.; Leumann, C. J.
Angew. Chem., Int. Ed. 2003, 42, 1655-1658. (h) Pallan, P. S.; Wilds, C.
J.; Wawrzak, Z.; Krishnamurthy, R.; Eschenmoser, A.; Egli, M. Angew.
Chem., Int. Ed. 2003, 42, 5893-5895. (i) Liu, H.; Lynch, S. R.; Kool, E.
T. J. Am. Chem. Soc. 2004, 126, 6900-6905. (j) Liu, H. B.; Gao, J. M.;
Maynard, L.; Saito, Y. D.; Kool, E. T. J. Am. Chem. Soc. 2004, 126, 1102-
1109. (k) Sun, B. W.; Babu, B. R.; Sorensen, M. D.; Zakrzewska, K.;
Wengel, J.; Sun, J. S. Biochemistry 2004, 43, 4160-4169. (l) Buchini, S.;
Leumann, C. J. Angew. Chem., Int. Ed. 2004, 43, 3925-3928. (m) Zahn,
A.; Brotschi, C.; Leumann, C. J. Chemistry 2005, 11, 2125-2129. (n)
Brotschi, C.; Mathis, G.; Leumann, C. J. Chemistry 2005, 11, 1911-1923.
(8) (a) Kool, E. T. Curr. Opin. Chem. Biol. 2000, 4, 602-608. (b) Tae, E. L.;
Wu, Y.; Xia, G.; Schultz, P. G.; Romesberg, F. E. J. Am. Chem. Soc. 2001,
123, 7439-7440. (c) Summerer, D.; Marx, A. J. Am. Chem. Soc. 2002,
124, 910-911. (d) Summerer, D.; Marx, A. Angew. Chem., Int. Ed. 2001,
40, 3693. (e) Strerath, M.; Summerer, D.; Marx, A. ChemBioChem 2002,
3, 578-580. (f) Strerath, M.; Cramer, J.; Restle, T.; Marx, A. J. Am. Chem.
Soc. 2002, 124, 11230-11231. (g) Berger, M.; Luzzi, S. D.; Henry, A. A.;
Romesberg, F. E. J. Am. Chem. Soc. 2002, 124, 1222-1226. (h) Henry,
A. A.; Romesberg, F. E. Curr. Opin. Chem. Biol. 2003, 7, 727-733. (i)
Kool, E. T., Acc. Chem. Res. 2002, 35, 936-943. (j) Chaput, J. C.; Ichida,
J. K.; Szostak, J. W. J. Am. Chem. Soc. 2003, 125, 856-857. (k) Chaput,
J. C.; Szostak, J. W. J. Am. Chem. Soc. 2003, 125, 9274-9275.
Results and Discussion
1. Design and Synthesis of Modified Nucleotides. We
synthesized various deoxynucleoside triphosphate (dNTP) ana-
logues of all four natural nucleotides bearing basic, acidic, or
(9) Grasby, J. A.; Gait, M. J. Biochimie 1994, 76, 1223-34.
(15) (a) Brakmann, S.; Lo¨bermann, S. Angew. Chem., Int. Ed. 2001, 40, 1427-
1429. (b) Brakmann, S. Methods Mol. Biol. 2004, 283, 137-144.
(16) (a) Piccirilli, J. A.; Krauch, T.; Moroney, S. E.; Benner, S. A. Nature 1990,
343, 33-37. (b) Switzer, C. Y.; Moroney, S. E.; Benner, S. A. Biochemistry
1993, 32, 10489-10496. (c) Atwell, S.; Meggers, E.; Spraggon, G.; Schultz,
P. G. J. Am. Chem. Soc. 2001, 123, 12364-12367. (d) Henry, A. A.; Olsen,
A. G.; Matsuda, S.; Yu, C.; Geierstanger, B. H.; Romesberg, F. E. J. Am.
Chem. Soc. 2004, 126, 6923-6931. (e) Johnson, S. C.; Marshall, D. J.;
Harms, G.; Miller, C. M.; Sherrill, C. B.; Beaty, E. L.; Lederer, S. A.;
Roesch, E. B.; Madsen, G.; Hoffman, G. L.; Laessig, R. H.; Kopish, G. J.;
Baker, M. W.; Benner, S. A.; Farrell, P. M.; Prudent, J. R. Clin. Chem.
2004, 50, 2019-2027. (f) Hutter, D.; Benner, S. A. J. Org. Chem. 2003,
68, 9839-9842.
(10) (a) Li, X.; Liu, D. R. Angew. Chem., Int. Ed. 2004, 43, 4848-4870. (b)
Kanan, M. W.; Rozenman, M. M.; Sakurai, K.; Snyder, T. M.; Liu, D. R.
Nature 2004, 431, 545-549. (c) Sakurai, K.; Snyder, T. M.; Liu, D. R. J.
Am. Chem. Soc. 2005, 127, 1660-1661.
(11) (a) Niemeyer, C. M.; Adler, M. Angew. Chem., Int. Ed. 2002, 41, 3779-
3783. (b) Niemeyer, C. M.; Ceyhan, B.; Hazarika, P. Angew. Chem., Int.
Ed. 2003, 42, 5766-5770. (c) Waybright, S. M.; Singleton, C. P.; Wachter,
K.; Murphy, C. J.; Bunz, U. H. J. Am. Chem. Soc. 2001, 123, 1828-1833.
(d) Scheffler, M.; Dorenbeck, A.; Jordan, S.; Wu¨stefeld, M.; von Kied-
rowski, G. Angew. Chem., Int. Ed. 1999, 38, 3311-3315. (e) Matsuura,
K.; Hibino, M.; Ikeda, T.; Yamada, Y.; Kobayashi, K. Chemistry 2004,
10, 352-359. (f) Vogel, S.; Rohr, K.; Dahl, O.; Wengel, J. Chem. Commun.
2003, 1006-1007.
(17) (a) Sakthivel, K.; Barbas, C. F., III. Angew. Chem., Int. Ed. 1998, 37, 2872.
(b) Perrin, D. M.; Garestier, T.; He´le`ne, C. Nucleosides Nucleotides 1999,
18, 377-391. (c) Gourlain, T.; Sidorov, A.; Mignet, N.; Thorpe, S. J.; Lee,
S. E.; Grasby, J. A.; Williams, D. M. Nucleic Acids Res. 2001, 29, 1898-
1905. (d) Roychowdhury, A.; Illangkoon, H.; Hendrickson, C. L.; Benner,
S. A. Org. Lett. 2004, 6, 489-492. Mehedi Masud, M.; Ozaki-Nakamura,
A.; Kuwahara, M.; Ozaki, H.; Sawai, H. ChemBioChem 2003, 4, 584-
588.
(12) (a) Weizman, H.; Tor, Y. J. Am. Chem. Soc. 2001, 123, 3375-3376. (b)
Weizman, H.; Tor, Y. J. Am. Chem. Soc. 2002, 124, 1568-1569. (c) Giese,
B. Curr. Opin. Chem. Biol. 2002, 6, 612-618. (d) Tanaka, K.; Yamada,
Y.; Tengeiji, A.; Kato, T.; Toyama, N.; Takezawa, Y.; Yori, M.; Shiro,
M.; Shionoya, M. Nucleic Acids Res. Suppl. 2003, 121-122. (e) Behrens,
C.; Carell, T. Chem. Commun. 2003, 1632-1633. (f) Giese, B.; Carl, B.;
Carl, T.; Carell, T.; Behrens, C.; Hennecke, U.; Schiemann, O.; Feresin,
E. Angew. Chem., Int. Ed. 2004, 43, 1848-1851.
(18) Thum, O.; Ja¨ger, S.; Famulok, M. Angew. Chem., Int. Ed. 2001, 40, 3990-
(13) Greco, N. J.; Tor, Y. J. Am. Chem. Soc. 2005, 127, 10784-10785.
(14) Verma, S.; Ja¨ger, S.; Thum, O.; Famulok, M. Chem. Rec. 2003, 3, 51-60.
3993.
(19) Ja¨ger, S.; Famulok, M. Angew. Chem., Int. Ed. 2004, 43, 3337-3340.
9
15072 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005

A Toolbox for DNA Functionalization at High Density
A R T I C L E S
Figure 1. Chemical structures of the synthesized modified nucleosid-5′-O-triphosphates.
lipophilic groups. The functionalities were attached to the
nucleobases via palladium-catalyzed coupling of the corre-
sponding halogenated nucleoside with unsaturated alkynes or
alkenes (Figure 1). Scheme 1 shows a representative synthesis
of the purine dA2 and the pyrimidine dT2. The base-modified
nucleosides were transferred to the 5′-O-triphosphates. Full
reaction schemes and spectroscopic data can be found in the
Supporting Information.
For the purines, the N7- and C8-positions appear to be suitable
for modification. 8-Modified deoxyadenosines are known to
destabilize dsDNA secondary structure, presumably due to
favoring the syn-conformation of the nucleotide.21 However, an
8-modified adenosine has been used as surrogate for dATP in
polymerase chain reaction.5b,17b Furthermore, a CD- and UV-
study showed that a C8-alkyne derivative of 2′-deoxyadenosine
causes only a slight decrease in the melting temperature of
dsDNA and does not disturb the overall B-DNA conformation.22
On the other hand, the substituents of 7-modified 7-deazapurines
The modifications of the functionalized 2′-deoxynucleotides
have to be positioned so as to ensure compatibility of these
derivatives with enzymatic transformations necessary for template-
directed replication of the corresponding oligonucleotides,
independent of their sequence and base composition. For the
pyrimidine derivatives, position C5 had been shown to be
optimal for positioning additional functional groups without
disturbing the helical structures.20
(20) (a) Znosko, B. M.; Barnes, T. W., III; Krugh, T. R.; Turner, D. H. J. Am.
Chem. Soc. 2003, 125, 6090-6097. (b) Sa´gi, J.; Szemzo, A.; Ebinger, K.;
Szabolcs, A.; Sa´gi, G.; Ruff, E.; O¨ tvo¨s, L. Tetrahedron Lett. 1993, 34,
2191-2194.
(21) (a) Saenger, W. Principles of Nucleic Acid Structure; Springer-Verlag: New
York, 1984. (b) Rao, S. N.; Kollman, P. A. J. Am. Chem. Soc. 1986, 108,
3048-3053.
(22) Tierney, M. T.; Grinstaff, M. W. Org. Lett. 2000, 2, 3413-3416.
9
J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005 15073

A R T I C L E S
Ja¨ger et al.
Scheme 1. Synthesis of a C7-Modified N7-Deaza-2′-deoxyadenosine Triphosphate and a C5-Modified Deoxyuridine Triphosphatea
a Conditions: (a) pent-4-ynoic acid methyl ester, Pd(PPh3)4, CuI, Et3N, DMF; (b) (i) POCl3, proton-sponge, OP(OMe)3; (ii) (nBu3NH)2H2P2O7, nBu3N;
(iii) 1 M TEAB; (c) aqueous NaOH; (d) 3-methyl-N-prop-2-ynylbutyramide, Pd(PPh3)4, CuI, Et3N, DMF (DMF ) N,N-dimethylformamide; TEAB )
triethylammonium bicarbonate).
Figure 2. Sequences of the middle part of the templates (40 nt) that are flanked by a 19 nt and a 20 nt long constant primer region.
Table 1. Summary of the Product Formation by Family A and B Type DNA Polymerases in Primer Extension Experiments with Modified
Nucleotides (dA2, dC1, dG1, dT3) and Different Templates (M79, A79, C79, G79, T79, P79)a
templates
polymerases
M79
A79
C79
G79
T79
P79
family A
Taq
Tth
KF (exo-)
Pwo
Tgo
(+++)
+++
+++
+++
+++
+++
+++
(-)
(++)
+++
+++
+++
+++
+++
+++
(++)
+++
+
+++
+++
+++
+++
(+)
(-)
-
-
+
-
+
+
family B
+++
+++
+++
+++
+++
+++
+++
+++
+++
+
Pfu (exo-)
Vent (exo-)
+
+++
a Amount of full-length product generated with the modified nucleotides as compared to a control reaction with four natural nucleotides and the designated
template: (-) < 5%, (+) ) 5-40%, (++) ) 40-70%, (+++) > 70%. The values for Taq polymerase are shown in parentheses because the length of the
primer extension product is slightly shorter than the full-length product (see Figure 3A).
are well accommodated in the DNA major groove23 and their
triphosphates can efficiently replace natural dNTPs in PCR.17c
Moreover, 7-modified purine dideoxy derivatives are used as
chain terminators in cycle sequencing.24
2.1. Suitability of Polymerases for High-Density fDNA
Synthesis by Primer Extension. Encouraged by our previous
finding that Tth DNA polymerase is able to generate a modified
DNA single strand in which every base is substituted by the
modified nucleotides dA2, dC1, dG1, and dT3 on a model
template,18 we investigated the range of other available DNA
polymerases with this ability. Therefore, we initially performed
primer extension experiments with a 5′-radiolabeled primer P3′
using a 79 nucleotide long model template DNA M79 (Figure
2) of almost equal base distribution.
We tested DNA polymerases from the A and B polymerase
families (Table 1), specifically the family A polymerases from
Bacillus stearothermophilus (Bst), Thermus aquaticus (Taq),
Thermus thermophilus (Tth), the Klenow fragment (exo-), and
Sequenase V 2.0. Among the family B DNA polymerases, we
(23) (a) Seela, F.; Zulauf, M. Chemistry 1998, 4, 1781-1790. (b) Seela, F.;
Zulauf, M. HelV. Chim. Acta 1999, 82, 1878-1898. (c) Ramzaeva,
N.; Mittelbach, C.; Seela, F. Nucleosides Nucleotides 1999, 18, 1439-
1440.
(24) Prober, J. M.; Trainor, G. L.; Dam, R. J.; Hobbs, F. W.; Robertson, C. W.;
Zagursky, R. J.; Cocuzza, A. J.; Jensen, M. A.; Baumeister, K. Science
1987, 238, 336-341.
9
15074 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005

A Toolbox for DNA Functionalization at High Density
A R T I C L E S
Figure 4. Generation of fDNA by primer extension using 5′-[32P]-labeled
primer and (A) Taq or (B) Pwo DNA polymerase with the designated
template, analyzed by 8% denaturing PAGE. Lane P, primer; lane (-),
reactions with dA2, dC1, dG1, dT3, and no template; lane 1, the natural
dNTPs and the designated template; lane 2, dA2, dC1, dG1, dT3, and the
designated template.
charged amino and guanidinium groups of dT3 and dC1 that
can potentially form highly stable Coulomb interactions with
the phosphate backbone or with the carboxyl group of dA2.
Highly stabilized secondary and tertiary interactions can be
expected in these oligonucleotides. In contrast, Taq polymerase
repeatedly yielded a product with enhanced electrophoretic
mobility, indicating that extension to full-length product was
not achieved by this polymerase (Figure 3A, lane 5).
Figure 3. Primer extension experiments with 5′-[32P]-labeled primer P3′
and template M79 with modified dNTPs and the denoted DNA polymerase,
analyzed by 8% denaturing PAGE. (A) Reactions with Taq DNA-
polymerase. Lanes: P, primer P3′; M, primer extension with four natural
dNTPs and template M79; 1, reactions with modified nucleotides dC1, dG1,
dT3; 2, dA2, dG1, dT3; 3, dA2, dC1, dT3; 4, dA2, dC1, dG1; 5, dA2,
dC1, dG1, dT3. (B) Reactions with Pwo DNA polymerase: lanes as
described for (A).
used the ones from Pyrococcus furiosus (Pfu), Pyrococcus woesi
(Pwo), Thermococcus gorgonarius (Tgo), and Thermococcus
litoralis [Vent (exo-)].
2.2. Suitability of Templates: Is Polymerization Sequence
Dependent? Having shown that polymerases from both families
A and B are able to accept functionalized dNTPs of all four
DNA bases when using a template with almost equal base
distribution, we next investigated more deeply whether the
template-directed polymerization of modified nucleotides is
dependent on the sequence context of the DNA template.
Therefore, we designed more demanding templates that include
repetitions of small stretches of the same base in their sequence.
This template design results in an accumulation of steric and
electronic requirements of one analogue component during the
polymerase-mediated synthesis of the fDNA. Four 79-mer
templates were used containing 40 nt long A- (A79), C- (C79),
G- (G79), or T-rich (T79) regions with consecutive repetitions
of the same nucleobase from two to four adjacent positions.
The fifth template (P79) was composed of a fully randomized
region of 40 bases, mimicking a combinatorial pool, typically
used for in vitro selection experiments (Figure 2). We also used
the previously described19 shorter template M59 as a positive
control (data not shown).
In our hands, Bst DNA polymerase and Sequenase did not
accept the modified deoxynucleotides as substrates because no
primer extension to full-length product was obtained (data not
shown). However, as shown in Table 1, all other enzymes were
able to generate full-length fDNA from modified nucleotides
dA2, dC1, dG1, dT3 (Figure 1) using template M79 at levels
of more than 70% as compared to positive control primer
extensions with dNTPs.
Figure 3 shows a representative example of these experiments,
including the negative control experiments in which one dNTP
was left out. We also did the same experiment with natural
dNTPs to compare the length of natural and modified DNAs.
The quantity of elongated product (lane 5) is similar to the
amount of DNA generated with natural dNTPs (lane M). No
full-length product was observed in control experiments in which
one of the nucleotides was omitted from the reaction mixtures
(lanes 1-4). As expected, the polymerase stops either directly
at the position where the missing base is encountered in the
template, or one base later. With Pwo polymerase, a product
was obtained that shows electrophoretic mobility similar to that
of the DNA yielded with natural dNTPs (Figure 3B, lane 5).
The observed smearing of the fDNA product band (Figure 3B,
lane 5) is likely due to secondary structures or aggregates that
were not resolved under the conditions of denaturing PAGE.
This might be attributed to the incorporation of positively
We performed primer extensions using these five templates
and the seven DNA polymerases from the A and B families. In
this experiment, we used dA2, dC1, dG1, and dT3 or the
corresponding natural nucleotides for the control reaction. Figure
4 shows two representative examples of primer elongation
experiments using Taq (Figure 4A) and Pwo DNA polymerase
(Figure 4B), respectively; Table 1 summarizes the results for
all tested DNA polymerases and templates.
9
J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005 15075

A R T I C L E S
Ja¨ger et al.
In contrast to the results obtained when using the M79 DNA
as template, the family A DNA polymerases Taq, Tth, and KF
(exo-) were not able to generate full-length fDNA on all
templates. In the case of Taq DNA polymerase, the elongation
was inhibited when three or four consecutive adenosines were
present in the template (Figure 4A, A79, lane 2). Because Taq
polymerase is highly Mg2+-dependent, we repeated the reaction
at higher Mg2+ concentrations to exclude possible quenching
of Mg2+-ions by modified dNTPs. However, even at higher
Mg2+ concentrations in the reaction buffer, formation of full-
length product could not be observed (data not shown). In primer
elongations using the T-rich T79, Taq polymerase paused
significantly after encountering the three or four nucleotide long
T-stretches in the sequence (Figure 4A, T79, lane 2). Overall,
the primer elongation with modified nucleotides resulted in
significantly decreased product yields as compared to the
reactions with natural dNTPs. Tth polymerase showed a similar
behavior, and no full-length product could be observed for the
A- and T-rich template (Table 1). In addition, KF (exo-) failed
to elongate the primer to full-length with the A-rich template.
These data show that family A polymerases performed
acceptably well with the C- and G-rich templates as compared
to the control reactions. The T-rich T79 and the random template
P79 were moderately accepted, whereas A79 was not. This
finding is in accordance with a previous result from the Benner
laboratory, which showed that polymerases from the sequence
family A were unable to incorporate a C5-modified dUTP-
derivative carrying a protected thio-group via an alkynyl linker
at multiple adjacent adenosine positions in a template.25 Whether
these distinctive behaviors of polymerases are due to differences
in the structure and/or to different incorporation kinetics of
individual modified dNTPs remains to be investigated.
Figure 5. Primer extension reactions with Vent (exo-) DNA polymerase
using 5′-[32P]-labeled primer P3′, template M79, and different combinations
of modified nucleotides. Lanes: lane P, primer P3′; lane M, positive control
with four natural nucleotides dATP, dCTP, dGTP, TTP; lanes 1-17,
analogues denoted below the lane.
2.3. Enzymatic Polymerization of Different Combinations
of Modified dNTPs. Our aim was to establish a “toolbox”
containing various functionalized building blocks for the
modular construction of high-density functionalized DNA. So
far, we could show that Pwo and Vent (exo-) DNA polymerases
are optimal enzymes for generation of fDNA with one set of
modified nucleotides (dA2, dC1, dG1, dT3). To explore the
modularity of this approach, we investigated whether other
combinations of modified dNTPs could be enzymatically
polymerized, despite their different electronic and steric proper-
ties, in a template-directed fashion.
In contrast, the polymerase of sequence family B: Vent (exo),
Tgo, Pfu (exo-), and Pwo DNA polymerase performed much
better and produced full-length product from the A-, C-, T-,
and G-rich template in yields comparable to those obtained with
natural dNTPs (Figure 4B and Table 1). While Tgo and Pfu
(exo-) DNA polymerases showed only weak elongation of the
primer with the randomized template P79, Pwo and Vent (exo-)
DNA polymerases led to almost complete extension of the
primer (Figure 4B, P79, lane 2 and Table 1). Thus, Pwo and
Vent (exo-) DNA polymerases proved to be the best enzymes
for the generation of a broad range of fDNA sequences.
Therefore, we performed primer extension experiments with
Vent (exo-) DNA polymerase, template M79, using the modified
analogues dA2, dC1, dG1, dT3 as basis for further incorpora-
tion studies with the other analogues. Specifically, we tested
each modified TTP (dT1-dT7) in combination with dA2, dC1,
dG1; the modified dC2 in combination with dA2, dG1, dT3;
and the modified ATP derivatives (dA1-dA7) in combination
with dC1, dG1, dT3 (Figure 5). To rule out the possibility of
full elongation of the primer due to misincorporation under the
applied conditions, appropriate control reactions were performed
in the absence of the respective nucleotide (Figure 5, lanes 1,
9, 11). As a positive control, a reaction was carried out with
the four natural dNTPs (Figure 5, lane M).
Importantly, the integrity of the resulting fM59 ssDNA,
generated by primer extension with Vent (exo-) DNA poly-
merase, was confirmed by MS analysis (Supporting Information,
part F, Figure S4A). The integrity of the modified M79
template,18 and the modified templates A79, T79, and G79,
generated by primer extension with Vent (exo-) DNA poly-
merase was confirmed by Sanger sequencing (Supporting
Information, part F, Figure S4B).
Taken together, these data show that family A polymerases
exhibit a marked sequence dependence in their ability to yield
full-length oligonucleotides when incorporating modified dNTPs
into the random template or templates that contain multiple
consecutive nucleotides. Among the family B polymerases,
which generally performed better, only Vent (exo-) and Pwo
DNA polymerases show virtually no sequence dependence.
We found that every tested set with modified TTPs (Figure
5, lanes 2-8), modified dCTPs (lane 10), and 7-deaza-modified
dATPs (lanes 12-15) produced full-length product. The yields
were comparable to the control performed with natural dNTPs
(Figure 5, lane M). In fact, the simultaneous incorporation of
various charged residues (dA2, dA4, dA5, dA7, dC1, dT1, dT3,
dT4, dT5, dT7), or nonpolar (dA3, dA6, dC2, dG1, dT2) and
bulky modifications (dC2, dT1, dT2), did not appear to exhibit
a substantial influence on product formation.
(25) Held, H. A.; Benner, S. A. Nucleic Acids Res. 2002, 30, 3857-3869.
9
15076 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005

A Toolbox for DNA Functionalization at High Density
A R T I C L E S
Interestingly, all full-length products derived from modified
dNTPs (Figure 5, lanes 2-8, 10, 12-15), show a lower
electrophoretic mobility as compared to natural DNA (lane M).
This might be due to the increase in molecular weight of the
functionalized oligonucleotides. Remarkably, the incorporation
of the derivative dT3, which bears a guanidinium group with a
rigid propynyl linker appears to lead to bands with decreased
resolution and showed a strong tendency to smear (Figure 5,
lanes 6, 10, 12-15). However, oligonucleotides containing the
same guanidinium functionality attached via a more flexible,
saturated linker to TTP (dT4) showed a band with improved
resolution (Figure 5, lane 7). This observation is also supported
by Roig et al., who reported recently that nucleotide dT3 had
a strong stabilizing effect on a DNA duplex.26
alkynyl-modified 7-deaza purines23 or alkynyl C5-modified
pyrimidines26 in dsDNA increases the melting temperature. For
these reasons, we increased the temperature of the denaturing
step from 95 to 99 °C and used the extremely thermostable Pwo
DNA polymerase in the presence of PCR additives that facilitate
amplification of GC-rich DNAs sequences. To exclude the
unintended amplification of contaminating biotinylated natural
DNA templates, which might be present in small amounts as
elution byproducts from the streptavidin column, the first 10
cycles of the PCR were performed only with primer P5′. In
fact, this primer anneals only to the generated fDNA templates,
excluding any unintentional amplification of unmodified DNA
contaminations. Next, 5′-[32P]-labeled primer P3′ was added,
and 8 cycles of PCR were performed. Equal amounts of fDNA
or DNA templates were used in each experiment.
The C8-modified deoxyadenosines dA6, dA7 could not be
incorporated in these primer extension experiments (Figure 5,
lanes 16, 17). The band pattern looks similar to the termination
band of the reaction in which dATP derivatives were omitted
(Figure 5, lane 11). Primer elongation experiments using Vent
(exo-) DNA polymerase with C8-modified deoxyadenosine
derivatives and natural dCTP, dGTP, TTP showed a band pattern
(data not shown) similar to that of reactions without any dATPs.
This suggests that 8-alkynylated derivatives were not recognized
as substrates by the enzyme.
As shown in Figure 6B (lanes 2), the amplification of DNA
from all different fDNA templates was successful (lanes 2)
although with a lower efficiency than that from natural DNA
templates (lanes 1). This might be attributed to an extensive
formation of secondary structures within the single-stranded
fDNAs or to an increased melting temperature of the corre-
sponding fDNA/DNA hybrid.23,26 Interestingly, the amplification
of G-rich and C-rich templates (C79, G79, lane 2) appears to
be somewhat less efficient than that of the A- and T-rich
templates (T79, A79, lane 2). A similar correlation between
sequence and efficiency of amplification is also observed in
the natural DNA control reactions, although the effect appears
to be less pronounced.
3. PCR Amplification of fDNA. To achieve fully modified
systems that are replicated and amplified by DNA polymerases,
we further investigated the performance of functionalized
templates and functionalized dNTPs in PCR.
3.1. PCR Amplification of Natural dsDNA, Using Fully
Functionalized Templates and Natural dNTPs. The easiest
way to retransform the sequence information of fDNA back to
natural DNA is by a PCR with natural dNTPs. In this case, a
polymerase would have to incorporate a natural dNTP opposite
to a complementary modified base in the functionalized
template. Earlier, we provided an example showing that a fully
functionalized DNA sequence could in turn serve as a template
in a PCR with natural dNTPs for the sequence-specific
amplification of natural DNA with Pwo DNA polymerase.18
Nevertheless, the data in Figure 6 show that there is a
nonnegligible degree of sequence dependence in the in vitro
replication of fDNA templates. Whether this can have an effect
on the outcome of SELEX experiments when using high-density
modified oligonucleotides remains to be investigated. In this
context, it is interesting to note that the pool template P79
yielded a significantly higher amount of PCR product (∼60%)
than the amount obtained on average with the A-, C-, G-, and
T-rich templates (∼40%). This indicates that the sequences in
the “demanding” templates might reflect fairly extreme situa-
tions of oligonucleotide compositions in an average template
library.
To further explore the sequence dependence of fDNA-
amplification, we generated additional fDNA templates for use
in PCR experiments. The single-stranded modified templates
were prepared by primer extension with 5′-biotinylated natural
DNA templates A79, C79, G79, T79, P79 (Figure 2) using 5′-
[32P] labeled primer P3′ and the modified nucleotides dA2, dC1,
dG1, dT3 (Figure 6). The resulting fDNA/DNA-hybrids were
subsequently immobilized on streptavidin-agarose. The radio
labeled fDNA was then recovered by elution with 0.1 N NaOH
and used as a template in PCRs with natural dNTPs. As a
positive control, single-stranded natural DNA was generated in
the same way by primer extension using natural dNTPs. As
additional controls, we immobilized biotinylated templates on
streptavidin-agarose, washed and eluted as described above, and
performed a PCR with the eluate.
3.2. PCR Amplification of High-Density Modified dsDNA,
Using Natural Templates and Modified dNTPs. The ultimate
step is the generation and direct amplification of functionalized
dsDNA. The most difficult case for a DNA polymerase would
be to copy and to directly amplify a functionalized DNA strand
with the corresponding modified dNTPs as the only monomer
source in a PCR. Previously, we could show that the Pwo DNA
polymerase can amplify a double-stranded functionalized DNA
from template M79 in the presence of nucleotides dA2, dC1,
dG1, dT2 under distinct PCR conditions.19
Here, we investigated how generally applicable these condi-
tions are by using the more demanding sequences (Figure 2) as
templates. We performed PCR experiments with the templates
A79, C79, G79, T79, P79, M79, the modified nucleotides dA2,
dC1, dG1, dT2, and primers P3′ and 5′-radiolabeled primer
P5′ (Figure 7A). As a positive control, a parallel set of PCRs
was performed under the same conditions but in the presence
of natural dNTPs instead of the modified ones. Reactions were
stopped and analyzed after 18 PCR cycles.
As mentioned before, standard PCR conditions, using dif-
ferent DNA polymerases, natural dNTPs, and an fDNA tem-
plate, did not result in any PCR-product,18 presumably due to
incomplete melting of the fDNA/DNA-hybrid under the condi-
tions employed. Others have also observed that the presence of
(26) Roig, V.; Asseline, U. J. Am. Chem. Soc. 2003, 125, 4416-4417.
9
J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005 15077

A R T I C L E S
Ja¨ger et al.
Figure 6. PCR amplification from fully functionalized templates and natural dNTPs yielding natural dsDNA. (A) Schematic for the generation and separation
of functionalized DNA single strands (red) from biotinylated natural DNA templates (blue) with modified dNTPs (fdNTPs) and their PCR amplification with
natural dNTPs. (B) PCR experiments with natural single-stranded DNA or fDNA templates, derived from primer extension with templates A79, C79, G79,
T79, P79, primers P5′, 5′-[32P]-labeled P3′, and Pwo DNA polymerase, analyzed by 2.5% agarose gel stained with ethidium bromide. Lane L, DNA ladder;
lane 1, amplification from natural DNA templates; lane 2, amplification from fDNA templates; lane 3, negative control, amplification of eluates from
immobilized biotinylated template. The same samples were loaded on a 10% denaturing polyacrylamide gel, and the bands were quantified (% of elongated
primer) by phosphorimaging.
As shown in Figure 7, PCR experiments with C- and G-rich
templates C79 and G79 gave the lowest yields (<10%).
Experiments with A-rich template A79 and the random sequence
template P79 yielded a comparable amount of product (∼20%),
while T-rich template T79 showed a 2-fold increase in product
formation (∼50%). In general, a decrease in amplification
products was observed in comparison to using natural dNTPs
and the fDNA template. This is not surprising because the
reaction is performed under conditions that are far from standard.
Obstacles might be, for example, the possible formation of
extensive secondary structures that can interfere with primer
annealing or hamper the primer elongation of the polymerase,
or a decreased processivity of the polymerase depending on the
modified dNTP and the sequence context. Another likely reason
for reduced amplification efficiency might be incomplete melting
of the fDNA/fDNA double strands during the denaturing PCR
step. This is supported by the data in Supporting Information
Figure S3 (Supporting Information, Part E).
The integrity of the PCR products was investigated by
sequencing (Supporting Information, Part G). For the A- and
T-rich templates, sequences resulting from both modified strands
were correct, demonstrating that the PCR amplification occurs
without detectable replication errors. However, the results
we obtained for the C- and G-rich templates were different.
When the C-rich template was sequenced using the 3′-primer
P3′, the same sequence as for the unmodified control DNA was
obtained. With the 5′-primer P5′, however, sequence information
is lost within a stretch of ∼14-15 bases. For the G-rich
template, both primers P3′ and P5′ indicated the same loss of
sequence information in this region (Supporting Information,
Part G). This loss of sequence information in a central part of
the C- and G-rich sequences correlates with the low yields for
these templates during PCR with modified nucleotides (Figure
7).
Overall, these data show that while amplification of a
completely functionalized DNA double strand is possible under
9
15078 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005

A Toolbox for DNA Functionalization at High Density
A R T I C L E S
Figure 7. PCR amplification from natural templates and modified dNTPs yielding fully functionalized dsDNA. (A) Schematic for the generation of double-
stranded fDNA (red) from natural DNA templates (blue) and modified dNTPs (fdNTPs). (B) PCRs with various templates A79, C79, G79, T79, P79, M79,
primer P3′, 5′-radiolabeled primer P5′, PCR-additives, and Pwo DNA polymerase analyzed by 10% denaturing PAGE. Lane L, DNA ladder; P, primer P5′;
lane 1, PCR with modified dNTPs dA2, dC1, dG1, dT2 with the respective template; lane 2, PCR with natural dNTPs; (-), control reactions without a
template with modified dNTPs (lane 1) or natural dNTPs (lane 2). Quantification of the product bands was done by phosphorimaging (% of elongated
primer).
certain PCR conditions, it is also obvious that amplification
efficacy, and maintenance of sequence information, strongly
depends on the sequence context. Sequences that are extremely
rich in G or C are amplified with about 8-fold reduced efficiency
with some loss of sequence information as compared to the
positive control using natural dNTPs, and with 1.5-3-fold
reduced efficiency with A- or T-rich sequences.
to a left-handed form, possibly resembling structures similar to
a Z-type DNA.
To elucidate this effect in greater depth, and to study the
possible structural alterations of functionalized DNA duplexes
as a function of the nature and the density of modifications, we
prepared a series of fDNA duplexes (ds fM59, duplexes A-E,
G) in which the natural dNTPs are substituted with modified
ones. For example, in duplex A, we replaced all A’s by dA2,
in duplex B, all C’s were replaced with dC1, and so on (Figure
8). We also prepared an additional duplex (E), in which three
bases, A, C, and T, were replaced with their modified analogues
dA2, dC1, dT2. The ds fM59 duplexes were synthesized by
PCR amplification from a 59mer DNA template (M59) with
about equal base distribution, primers P3′, P5′, and natural or
modified dNTPs (dA2, dC1, dG1, and dT2), using Pwo DNA
polymerase.
4. Circular Dichroism Analyses of fDNAs Modified at
Different Densities. Recently, we have observed that the CD
spectrum of high-density modified M59 dsDNA is inverted, as
compared to a standard B-type DNA double helix.19 The same
effect was observed in the Brakmann laboratory when all of
the pyrimidine residues in a double helix were substituted with
fluorescently labeled analogues.15 These results suggested that
high-density incorporation of modifications into double-stranded
DNA enforces a transition from the right-handed B-type DNA
9
J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005 15079

A R T I C L E S
Ja¨ger et al.
Figure 8. CD spectra of DNA double strands with the sequence of dsM59 containing different densities of modifications. Duplex G is the unmodified
dsDNA M59; in duplex A, all As were substituted with dA2; and, respectively, for duplex B, dC1; duplex C, dT2; duplex D, dG1; duplex E, dA2, dC1,
dT2; and duplex F, dA2, dC1, dT2, dG1.19 Measured in phosphate buffer 10 mM, pH 7.0.
CD spectra of the duplexes containing one of the bases
substituted with modified nucleotides (duplexes A-C) are
presented in Figure 8. As can be seen, there are positive bands
in the range of 270-280 nm as well as negative bands at 250
and 210-220 nm. The general appearance of the CD signal
between 200 and 300 nm resembles that of the unmodified M59,
which shows a typical B-DNA spectrum (Figure 8G). The CD
spectrum of duplex D (containing dG1) includes additional
maxima at 250 nm, which may result from electronic transitions
of the extended π-system of the dG1 phenyl residue interacting
with the transitions of the chirally oriented surrounding bases
(Figure 8D). Similar CD behavior, in particular around 250 nm,
was observed by He and Seela27 when incorporating various
base pairs with propynyl residues into oligonucleotide duplexes.
Nevertheless, the spectra of D seem to resemble a right-helix
DNA with the typical B-type bands shifted to a longer
wavelength.
considerably in comparison to the nonmodified M59 duplex (G)
or to duplexes modified at lower density (A-D). The band
pattern of E includes mainly two minima (low intensity band
at 304 nm and high-intensity band at 244 nm). The same band
pattern is also obtained in the duplex in which all nucleotides
were modified (F),19 except that in E the bands are shifted to
longer wavelengths. These alterations in the CD spectra appear
to correlate with the density of functionalization.
In general, the differences in the CD spectra appear to
originate from the effect that modified nucleobases have on base
pairing and base stacking. The major groove is presumably filled
with up to 4 types of modifications all of which contain the
hydrophobic propynyl moiety with some also having altered
overall base polarizability. In this study, we demonstrate that
incorporation of one type of modified base does not alter the
overall B-type DNA structure of the M59 duplex, regardless
of the type of modification used. However, three modifications
are already enough for inducing a conformational change, that
is, apparently, a step toward the manifestation of the inversion
of the fully functional duplex.
Interestingly, at a density of three modifications, dA2, dC1,
and dT2, different CD spectra (E) were obtained, which vary
(27) He, J. L.; Seela, F. Nucleic Acids Res. 2002, 30, 5485-5496.
9
15080 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005

A Toolbox for DNA Functionalization at High Density
A R T I C L E S
8% denaturing PAGE (8 M urea, 50-60 °C) and visualized and
quantified by autoradiography or phosphorimaging.
Conclusions
Chemical modifications of nucleic acids present vast op-
portunities for extending the structural and functional properties
of these biomolecules. In this study, we report the synthesis of
an expanded set of modified nucleotides. We also provide a
detailed investigation of the scope and limits for the enzymatic
generation of DNA molecules that are functionalized at high
density, by primer extension and PCR. On the basis of a large
variety of chemically modified deoxynucleotide triphosphates,
we can now define reaction conditions that allow the substitution
of the natural nucleobases in each strand with up to four different
base-modified analogues. We show that every base position, in
various sequence contexts, can be addressed by different modi-
fications and thus may facilitate a precise control of the three-
dimensional placement of additional chemical functionalities.
Among the various polymerases used here, only family B
polymerases were able to replicate “demanding” templates that
contained consecutive stretches of a particular base, regardless
of whether the template was a natural DNA or a high-density
modified one. Nevertheless, it is remarkable that natural DNA
polymerases exist that are able to recognize and incorporate a
modified base opposite the complementary functionalized base
in the template. This result suggests that it might be possible to
increase the diversity of functional groups in a single DNA
strand even further, by using modified artificial base pairs, as
accomplished by the groups of Benner,28 Schultz,29 and Romes-
berg.30 With this study, we have established a versatile toolbox
of enzymes, templates, and monomers for generating high-
density functionalized DNAs that might be suited for many
different purposes in disciplines such as drug discovery, drug
delivery, nanoelectronics, and nanocomputing.
Preparation of Single-Stranded Functionalized Templates. For
separation of single-stranded fDNA, 10 pmol of a 5′-[32P]-labeled primer
P3′ was extended on 30 pmol of the appropriate 5′-biotinylated template
with nucleotides dA2, dC1, dG1, dT3 (final concentration 50 µM each),
0.4 U thermostable inorganic pyrophosphatase, and 2 U Vent (exo-)
DNA polymerase in a final reaction volume of 40 µL of 1X polymerase
reaction buffer. The mixture was heated for 2 h to 72 °C. After the
reaction, 10 µL of 5X buffer (750 mm NaCl, 0.5 mm EDTA, 250 mm
HEPES, pH 7.0) was added, and the mixtures were immobilized on 50
µL of streptavidin-agarose (Ultra-Link Plus, Pierce, Rockford, IL), pre-
washed with five 100 µL volumes of wash buffer (150 mM NaCl, 0.1
mm EDTA, 50 mM HEPES, pH 7.0). The column was washed with
15 × 100 µL wash buffer at 25 °C. The fDNA was eluted with two 50
µL volumes of ice-cooled elution buffer (0.1 M NaOH, 150 mM NaCl).
Reactions were neutralized with 7 µL of 5% acetic acid. The
concentration of the fDNA templates was determined by scintillation
counting.
PCR Reactions with Functionalized DNA Templates and Natural
dNTPs. PCR experiments were performed on an Eppendorf Master-
cycler gradient. The reactions were done in an overall reaction volume
of 50 µL in 1X GC-rich reaction buffer (GC-rich PCR System, Roche)
containing 1 M GC-rich resolution solution (GC-rich PCR System,
Roche) using natural dNTPs (final concentration 200 µM each), 2.5 U
Pwo DNA polymerase, and the modified template (about 0.1 pmol per
reaction). Initially, 10 PCR cycles were performed with only 5′-[32P]-
labeled primer P5′ (50 pmol) with the following conditions: initial
denaturing at 99 °C for 3 min, denaturing at 99 °C for 3 min, primer
annealing at 50 °C for 2 min, and extension at 72 °C for 5 min.
Subsequently primer P3′ (50 pmol) was added to the reaction mixture
and an additional 8 cycles of PCR were performed with conditions:
denaturing 99 °C for 1 min, annealing 50 °C for 1 min, extension 72
°C for 1 min, and a final extension step of 5 min at 72 °C. The PCR
reactions were analyzed on a 2.5% agarose gel stained with ethidium
bromide (5 µL sample). Additionally, 5 µL of the PCR reaction mixture
was mixed with 15 µL of stop solution, denatured for 5 min at 99 °C,
and the products separated on an 8% denaturing PAGE (8 M urea,
50-60 °C) and quantified by phosphorimaging.
PCR Reactions with Natural DNA Templates and Modified
dNTPs. The reactions were done in an overall volume of 25 µL in
Pwo reaction buffer (1X) containing 10% DMSO, 5% formamide, 0.75
M betaine, and 50 mM TMAC using modified nucleotides dA2, dC1,
dG1, dT2 (final concentration 200 µM each), 25 pmol of each primer
P3′ and 5′-[32P]-labeled primer P5′, 0.5 pmol of the respective template,
Pwo DNA polymerase (1.25 U), and thermostable inorganic pyrophos-
phatase (0.2 U). As a control, PCRs were performed under the same
conditions using 200 µM natural dNTPs instead of the modified
nucleotides. Amplification was performed through an initial denaturing
at 99 °C for 3 min, followed by 18 cycles of denaturing at 99 °C for
2 min, primer annealing at 50 °C for 2 min, and extension at 72 °C for
2 min. A negative control without template was also carried out. 5′-
[32P]-labeled and denatured 100 bp DNA ladder (peqLab) served as a
length standard. An aliquot of 10 µL of the PCR reaction mixture was
mixed with 30 µL of formamide/water (4:1), containing 20 mM EDTA,
and denatured at 99 °C for 10 min. 5 µL of each reaction mixture was
analyzed by 10% denaturing PAGE (8 M urea, 50-60 °C), and
visualized and quantified by phosphorimaging.
Materials and Methods
Polymerases and PCR Additives. Tth, Pwo, and Tgo DNA
polymerase were purchased from Roche Diagnostics (Mannheim,
Germany). Bst, Vent (exo-) DNA polymerase, and Klenow Fragment
(exo-) were from New England Biolabs (Frankfurt, Germany). Taq
DNA polymerase was from Promega (Madison, WI). Pfu (exo-) DNA
polymerase was purchased from Stratagene (Heidelberg, Germany).
Sequenase V. 2.0 was from Amersham Biosciences (Freiburg, Ger-
many).
The PCR additives Betaine (BioChemika, anhydrous), formamide
(MicroSelect for molecular biology), tetramethylammonium chloride
(Ultra for molecular biology), dimethyl sulfoxide (Ultra for molecular
biology) were purchased from Fluka.
Primer Extension Reactions. 5′-[32P]-labeled primer P3′ (2 pmol)
was annealed to the appropriate template (6 pmol) in 1X polymerase
reaction buffer (provided by the supplier of the DNA polymerase) by
heating the mixture to 95 °C for 5 min and subsequently allowing the
solution to cool over 1 h to room temperature. Thermostable inorganic
pyrophosphatase (0.2 U), DNA polymerase (1 U), and dNTPs (final
concentration 50 µM) were added to a final volume of 20 µL in 1X
polymerase reaction buffer. The reactions were immediately performed
by heating the mixture to the optimal temperature of the enzyme in a
thermocycler for 30 min. Reactions were stopped by addition of 60
µL of stop-solution (80% formamide solution containing 20 mM EDTA)
and heating at 99 °C for 10 min. Reaction products were separated by
Circular Dichroism. CD experiments were performed with a Jasco
810 spectrophotometer. CD spectra of the duplex solutions (0.4 OD
base concentration of duplexes A, C, D, E, 0.6 OD of duplex B, and
0.22 OD of duplex G, in 10 mM phosphate buffer and 1 M NaCl, pH
7.0 at 25 °C) and were recorded using a 0.2 mm quartz cell. A spectrum
of the buffer was measured separately and subtracted from the spectra
resulting from the samples. On average, seven spectra were recorded
for each experiment. The samples for CD (ds fM59 (duplexes A-E,
(28) Sismour, A. M.; Lutz, S.; Park, J. H.; Lutz, M. J.; Boyer, P. L.; Hughes,
S. H.; Benner, S. A. Nucleic Acids Res. 2004, 32, 728-35.
(29) Meggers, E.; Holland, P. L.; B., T. W.; Romesberg, F. E.; Schultz, P. G.
J. Am. Chem. Soc. 2000, 122, 10714-10715.
(30) Henry, A. A.; Olsen, A. G.; Matsuda, S.; Yu, C.; Geierstanger, B. H.;
Romesberg, F. E. J. Am. Chem. Soc. 2004, 126, 6923-31.
9
J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005 15081

A R T I C L E S
Ja¨ger et al.
G)) were generated by large-scale PCR in an overall volume of 5 mL,
using the same conditions and modified nucleotides as described in
the earlier paragraph but with M59 template (primers P3′ and P5′).
Following the PCR, the duplexes were precipitated with cold EtOH
and purified by anion exchange chromatography or by native PAGE
(15%). The full details can be found in the Supporting Information.
is grateful to the Minerva foundation for a postdoctoral fellow-
ship.
Supporting Information Available: Details on synthetic pro-
cedures, spectroscopic characterization, primer extension experi-
ments, melting behavior of fDNA, MS-characterization and se-
uencing data of modified DNA primer extension products, and
sequencing data of modified DNA PCR products. This material
is available free of charge via the Internet at http://pubs.acs.org.
Acknowledgment. We thank Heike Hupfer and Konrad
Sandhoff for access to their Q-TOF ESI MS spectrometer. This
work was supported by the Deutsche Forschungsgemeinschaft
(SFB 624) and the Fonds der Chemischen Industrie. H.K.-L.
JA051725B
9
15082 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005