Expanding the chemistry of DNA for in vitro selection

Article abstract of DOI:10.1021/ja908035g

Six new 5-posltlon modified dUTP derivatives connected by a unique amide linkage were synthesized and tested for compatibility with the enzymatic steps of In vitro selection. Six commercially available DNA polymerases were tested for their ability to effi

Full text of DOI:10.1021/ja908035g

Published on Web 03/04/2010
Expanding the Chemistry of DNA for in Vitro Selection
Jonathan D. Vaught,†,‡ Chris Bock,‡ Jeff Carter,‡ Tim Fitzwater,‡ Matt Otis,‡
Dan Schneider,‡ Justin Rolando,† Sheela Waugh,‡ Sheri K. Wilcox,‡ and
Bruce E. Eaton*,†
Department of Chemistry and Biochemistry, UniVersity of Colorado, Boulder, Colorado
80309-0215, and SomaLogic, Inc., 2945 Wilderness Place, Boulder, Colorado 80301
Received September 22, 2009; E-mail: bruce.eaton@colorado.edu
Abstract: Six new 5-position modified dUTP derivatives connected by a unique amide linkage were
synthesized and tested for compatibility with the enzymatic steps of in vitro selection. Six commercially
available DNA polymerases were tested for their ability to efficiently incorporate each of these dUTP
derivatives during PCR. It was not possible to perform PCR under standard conditions using any of the
modified dUTP derivatives studied. In contrast, primer extension reactions of random templates, as well
as defined sequence templates, were successful. KOD XL and D. Vent DNA polymerases were found to
be the most efficient at synthesizing full-length primer extension product, with all of the dUTP derivatives
tested giving yields similar to those obtained with TTP. Several of these modified dUTPs were then used
in an in vitro selection experiment comparing the use of modified dUTP derivatives with TTP for selecting
aptamers to a protein target (necrosis factor receptor superfamily member 9, TNFRSF9) that had previously
been found to be refractory to in vitro selection using DNA. Remarkably, selections employing modified
DNA libraries resulted in the first successful isolation of DNA aptamers able to bind TNFRSF9 with high
affinity.
tion,20 Diels-Alder cycloaddition,21-25 Michael additions,26
Introduction
nucleophilic substitutions,27,28 porphyin metalations,29 and
RNA and DNA in vitro selection methods1-3 have yielded
specific sequences that function as biopolymer ligands (aptam-
ers) and catalysts for a wide range of applications in therapeutics4,5
and diagnostics.6-9 Both DNA and RNA aptamers in their native
form have been selected to specifically bind proteins and small
molecules.10-13 In addition, a growing list of organic reactions
are catalyzed by specific nucleic acid sequences. These include
acyl transfers,14-18 aldol condensation,19 amide bond forma-
enzyme cofactor synthesis.30,31
There are limited data on the use of modified nucleotides in
either DNA or RNA to select modified sequences with functions
beyond that of native DNA and RNA. In most of biology,
proteins are the predominant catalysts (e.g., enzymes) and
affinity molecules (e.g., antibodies), although there are notable
exceptions such as the ribosome peptidyl transferase, which is
catalyzed by RNA.48 A common explanation for the functional
dichotomy between nucleic acids and proteins is that amino
† University of Colorado.
‡ SomaLogic, Inc.
(1) Ellington, A. D.; Szostak, J. W. Nature 1990, 346, 818–822.
(2) Tuerk, C.; Gold, L. Science 1990, 249, 505–510.
(3) Ellington, A. D.; Szostak, J. W. Nature 1992, 355, 850–852.
(4) Lee, J. F.; Stovall, G. M.; Ellington, A. D. Curr. Opin. Chem. Biol.
2006, 10, 282–289.
(16) Lee, N.; Bessho, Y.; Wei, K.; Szostak, J. W.; Suga, H. Nat. Struct.
Biol. 2000, 7, 28–33.
(17) Illangasekare, M.; Sanchez, G.; Nickles, T.; Yarus, M. Science 1995,
267, 643–647.
(18) Kumar, R. K.; Yarus, M. Biochemistry 2001, 40, 6998–7004.
(19) Fusz, S.; Eisenfuhr, A.; Srivatsan, S. G.; Heckel, A.; Famulok, M.
Chem. Biol. 2005, 12, 941–950.
(5) Nimjee, S. M.; Rusconi, C. P.; Sullenger, B. A. Annu. ReV. Med. 2005,
56, 555.
(6) Chu, T. C.; Shieh, F.; Lavery, L. A.; Levy, M.; Richards-Kortum, R.;
Korgel, B. A.; Ellington, A. D. Biosens. Bioelectron. 2006, 21, 1859–
1866.
(20) Wiegand, T. W.; Janssen, R. C.; Eaton, B. E. Chem. Biol. 1997, 4,
675–683.
(21) Tarasow, T. M.; Tarasow, S. L.; Eaton, B. E. Nature 1997, 389, 54–
57.
(7) Collett, J. R.; Cho, E. J.; Lee, J. F.; Levy, M.; Hood, A. J.; Wan, C.;
Ellington, A. D. Anal. Biochem. 2005, 338, 113–123.
(8) Jhaveri, S.; Rajendran, M.; Ellington, A. D. Nat. Biotechnol. 2000,
18, 1293–1297.
(22) Seelig, B.; Jaschke, A. Chem. Biol. 1999, 6, 167–176.
(23) Tarasow, T. M.; Tarasow, S. L.; Tu, C.; Kellogg, E.; Eaton, B. E.
J. Am. Chem. Soc. 1999, 121, 3614–3617.
(9) Brody, E. N.; Willis, M. C.; Smith, J. D.; Jayasena, S.; Zichi, D.;
Gold, L. Mol. Diagn. 1999, 4, 381–388.
(24) Stuhlmann, F.; Jaschke, A. J. Am. Chem. Soc. 2002, 124, 3238–3244.
(25) Tarasow, T. M.; Tarasow, S. L.; Eaton, B. E. J. Am. Chem. Soc. 2000,
122, 1015–1021.
(10) Osborne, S. E.; Ellington, A. D. Chem. ReV. 1997, 97, 349–370.
(11) Famulok, M.; Szostak, J. W. Angew. Chem., Int. Ed. Engl. 1992, 31,
979–988.
(26) Sengle, G.; Eisenfuhr, A.; Arora, P. S.; Nowick, J. S.; Famulok, M.
Chem. Biol. 2001, 8, 459–473.
(12) Eaton, B. E. Curr. Opin. Chem. Biol. 1997, 1, 10–16.
(13) Gold, L.; Polisky, B.; Uhlenbeck, O.; Yarus, M. Annu. ReV. Biochem.
1995, 64, 763–797.
(27) Wecker, M.; Smith, D.; Gold, L. RNA 1996, 2, 982–994.
(28) Wilson, C.; Szostak, J. W. Nature 1995, 374, 777–782.
(29) Li, Y. F.; Sen, D. Nat. Struct. Biol. 1996, 3, 743–747.
(30) Huang, F. Q.; Bugg, C. W.; Yarus, M. Biochemistry 2000, 39, 15548–
15555.
(14) Lohse, P. A.; Szostak, J. W. Nature 1996, 381, 442–444.
(15) Suga, H.; Lohse, P. A.; Szostak, J. W. J. Am. Chem. Soc. 1998, 120,
1151–1156.
(31) Jadhav, V. R.; Yarus, M. Biochimie 2002, 84, 877–888.
9
10.1021/ja908035g  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 4141–4151 4141

A R T I C L E S
Vaught et al.
acids have a wider array of chemical functionality than
nucleotides. Few examples have been reported where the affinity
or catalytic activity of selected aptamers or nucleic acid catalysts
could be enhanced by adding chemical diversity to nucleobases.
Most of these experiments have focused on RNA, and the results
have been mixed. It remains an open scientific question as to
which biopolymer, RNA or DNA, provides the better framework
(for catalysis or aptamers) and if chemical modification of the
nucleobases can be generally beneficial. Two important chal-
lenges for the field are the synthesis of modified nucleotides
and their enzymatic incorporation into nucleic acids for in vitro
selection. Herein we describe the possibilities for creating the
same type of 5-position uridine modification chemistry in DNA
that has been successful in RNA selections.
The majority of data reported on modified aptamers is from
RNA modified at the 2′-position of ribose, a modification used
to increase the in vivo stability of RNA.32 Modified DNA has
been used to select both catalysts and aptamers. Elegant
landmark examples have been reported where base-modified
deoxyuridine and deoxyadenosine were incorporated into DNA
catalysts for phosphodiester bond cleavage that, unlike their
unmodified RNA counterparts, did not require divalent metal
cations.35 Examples of modified DNA aptamers are also known.
Recently, the selection of an enantioselective thalidomide
aptamer was reported that contained a base-modified deoxyuri-
dine, which may prove to be useful as an enantiospecific
bioanalytic reagent.40 However, not all modified chemistries for
nucleic acids have proven beneficial. In 1994, a thrombin
aptamer was selected using a modified deoxyuridine.41 In
comparison to a previously known unmodified DNA aptamer,
the affinity of the modified aptamer for thrombin was signifi-
cantly worse. It appears that the functional group of the modified
DNA, and perhaps the chemistry of the linkage to the nucleo-
base, may be important in determining if a functional group
has any advantage in an aptamer for high affinity binding to a
protein.42
To be useful for in vitro selection experiments, modified
triphosphates must serve as substrates for polymerases used to
replicate the oligonucleotide under investigation. In the case of
RNA, the kinetics of enzymatic incorporation by T7 RNA
polymerase for a series of 5-position carboxamid-modified UTP
derivatives has been previously reported.38 Two of these UTP
derivatives have been used successfully for in vitro selection
of RNA catalysis for organic reactions that include nontemplated
amide bond formation,20 a Diels-Alder cycloaddition with a
low reactivity acyclic diene,21 and urea bond formation with
stereoselectivity for tripeptide substrates.39 In addition to the
modified deoxynucleotides used to select a thalidomide aptamer,
thrombin aptamer, and a catalyst for phosphodiester bond
cleavage, several groups have synthesized modified deoxy-
Figure 1. New dUTP derivatives prepared as described in Scheme 1.
nucleotides using conventional alkyne linkage chemistries to
the nucleobase. These are proven substrates for certain DNA
polymerases but have yet to be reported as successful for in
vitro selection experiments for aptamers to protein targets.33–37
To better understand what modifications in DNA could yield
improved aptamers and catalysts, more chemical examples are
needed. The restricted rotation of the amide bond and its ability
to serve as both hydrogen bond acceptor and donor appeared
to be attractive attributes for the linker since new H-bonding
contacts may be possible with protein targets. Furthermore,
minimizing the number of rotable bonds between the nucleobase
and the pendant modification seemed desirable to allow for
precise orientation at the aptamer protein interface.
To efficiently survey new DNA chemistries required a
streamlined chemical modification synthetic procedure. Herein
we describe adaptation of our previously reported methods for
preparing modified uridines to the synthesis of 5-position
modified 2′-deoxyuridines, preparing the corresponding dUTP
derivatives, testing these as substrates with six DNA poly-
merases, and establishing viable in vitro selection methods.
Finally, we will describe using these dUTP derivatives to form
DNA libraries and how these libraries were successful for in
vitro selection to yield a high affinity aptamer sequence to a
challenging human protein target TNFRSF9.
Results and Discussion
Synthesis of 5-Position Carboxamid-2′-deoxyuridines and
Triphosphates. In order to expand the repertoire of DNA, it
was desirable to attach functional groups to the 5-position of
2′-deoxyuridine by an amide linkage that may provide new
H-bonding motifs and concomitant 3D structures. Further, it
was of interest to synthesize a range of hydrophobic groups,
both aromatic and aliphatic, to increase the hydrophobic content
(Figure 1) of DNA to potentially mimic the types of side chains
found in the binding domains of antibodies. These functional
groups were added to 2′-deoxyuridine in a single step by
palladium-catalyzed carboxyamidation of 5-iodo-2′-deoxyuridine
under mild conditions (Scheme 1). The palladium-catalyzed
5-position modification of uridines can also be accomplished
(32) Aurup, H.; Williams, D. M.; Eckstein, F. Biochemistry 1992, 31, 9636–
9641.
(33) Sakthivel, K.; Barbas, C. F. Angew. Chem., Int. Ed. 1998, 37, 2872–
2875.
(34) Jager, S.; Rasched, G.; Kornreich-Leshem, H.; Engeser, M.; Thum,
O.; Famulok, M. J. Am. Chem. Soc. 2005, 127, 15071–15082.
(35) Perrin, D. M.; Garestier, T.; Helene, C. Nucleosides, Nucleotides
Nucleic Acids 1999, 18, 377–391.
(36) 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.
(37) Benner, S. A. Acc. Chem. Res. 2004, 37, 784–797.
(38) Vaught, J. D.; Dewey, T.; Eaton, B. E. J. Am. Chem. Soc. 2004, 126,
11231–11237.
(40) Shoji, A.; Kuwahara, M.; Ozaki, H.; Sawai, H. J. Am. Chem. Soc.
2007, 129, 1456–1464.
(41) Latham, J. A.; Johnson, R.; Toole, J. J. Nucleic Acids Res. 1994, 22,
2817–2822.
(39) Nieuwlandt, D.; West, M.; Cheng, X. Q.; Kirshenheuter, G.; Eaton,
B. E. Chembiochem 2003, 4, 651–654.
(42) Tarasow, T. M.; Eaton, B. E. Biopolymers 1998, 48, 29–37.
9
4142 J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010

Expanding the Chemistry of DNA for in Vitro Selection
A R T I C L E S
Scheme 1. Synthesis of 6a-fa
a Conditions: (i) 1.1 equiv of DMTrCl, pyridine, rt, 16 h; (ii) 10 equiv
of Ac2O, pyridine, rt, 16 h; (iii) 10 mol % of Pd(PPh3)4, 5 equiv of TEA,
3 equiv of RNH2, DMA/THF, 50 psi CO, 70 °C, 48-72 h; (iv) 1 equiv of
TCA, MeOH/CH2Cl2 1:1, rt, 2-4 h; (v) 1.1 equiv of 2-chloro-4H-1,3,2-
benzodioxaphosphorin-4-one, DMF/pyridine/dioxane 1:1:4, rt, 10 min; (vi)
1.5 equiv of tributylammoniun pyrophosphate, tributylamine, DMF, rt, 10
min; (vii) 0.02 M I2, THF/pyridine/H2O 70:20:10, rt. 15 min; (viii) concd
NH4OH, 1 h.
Figure 2. Some possible orientations of 5-position modifications on dUTP
that lead to new H-bonding motifs. Atoms and bonds shown in blue indicate
highly similar regions where a modified dUTP could present itself almost
identically to each of the native nucleotides.
hydrocarbon chains frequently used in other synthetic schemes
for making base-modified nucleotides. Also, the amide bond
introduces two new hydrogen-bonding contacts, a hydrogen
bond donor and hydrogen bond acceptor (Figure 2). This amide
is cross-conjugated with the pyrimidine ring which has signifi-
cant electronic and structural effects on the pyrimidine ring
system (Eaton, B. E. and Rolando, J., unpublished results) in
addition to the effects of the appended functional group
connected by the amide. For example, the tautomerization of
dUTP to make the hydroxyl at C4 could be facilitated by the
presence of the amide hydrogen forming an intermediate H-bond
with the oxygen at C4. Another possibility is that, depending
on the orientation of the amide with respect to the ring, the
modified uridine base could display hydrogen-bonding patterns
similar to those observed in Hoogsteen binding, where the bases
are dA, dG, or dC. In the context of a folded ssDNA, alternative
base pairings (e.g., duplex, triplex, etc.) could be advantageous,
providing alternative hydrogen-bonding motifs that stabilize
unique three-dimensional structures. Another key feature of the
C5-modification chemistry described herein is that the number
of rotable bonds between the functional group and the base is
minimized, making these types of connections to the modifica-
tions more like those found connecting the functional groups
found in proteins.42 Lastly, the types of functional groups
appended are similar to some of the most common critical
residues in antibody CDR regions, mainly aromatic groups with
varying degrees of hydrophobicity and polarizability.
using stannanes to provide a ketone linkage to the functional
group.43 The utility of stannanes for incorporation of functional
groups is limited, however, because stannanes are difficult to
prepare. An alternative palladium-catalyzed reaction is the
carbonylative coupling using amines.44 In contrast to stannanes,
numerous amines are readily available. However, when applying
these 5-position carbonylative coupling methods to 2′-deox-
yuridines, different nucleoside protection and coupling proce-
dures must be followed to achieve good yields.
Starting with commercially available 5-iodo-2′-deoxyuridine,
it was necessary to first protect the 5′-hydroxyl with a DMT
group to give 1 followed by acetylation of the 3′-hydroxyl to
give the differentially protected nucleoside 3. Linkage of the
various functional groups was accomplished by palladium
(Pd[PPh3]4, 10 mol %)-catalyzed carbonylative coupling with
primary amines and CO (50 psi) in DMF/THF at 70 °C, using
5 equiv to scavenge the HI produced in forming 4a-f.
Following purification of 4a-f by column chromatography, the
5′-DMT protecting group was removed with trichloroacetic acid
(1 equiv, in methanol/CH2Cl2, rt). With the 5-hydroxyl depro-
tected, triphosphate synthesis was accomplished using the
Eckstein45 triphosphate method to give 6a-f in good to
moderate (30-60%) yield.
While there are other chemistries available for introducing
modifications to the 5′-position of 2′-deoxyuridine, this chem-
istry is unique in that the modifications are linked through an
amide bond directly to the pyrimidine base at the C5-position,
which is much less hydrophobic than alkyne linkers and
DNA Polymerase Screening for Modified DNA Synthesis
by Primer Extension. In order to determine whether these new
modified nucleotides could be incorporated enzymatically into
DNA, we tested the ability of commercially available DNA
polymerases to incorporate the dUTP derivatives 6a-f in primer
extension reactions. It should be noted that several groups have
previously reported modified DNA synthesis using 5-position
modified dUTP derivatives with various DNA polymerases
during PCR and primer extension reactions. However, none of
(43) Dewey, T. M.; Mundt, A. A.; Crouch, G. J.; Zyzniewski, M. C.; Eaton,
B. E. J. Am. Chem. Soc. 1995, 117, 8474–8475.
(44) Dewey, T. M.; Zyzniewski, M. C.; Eaton, B. E. Nucleosides Nucle-
otides 1996, 15, 1611–1617.
(45) Ludwig, J.; Eckstein, F. J. Org. Chem. 1989, 54, 631–635.
9
J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010 4143

A R T I C L E S
Vaught et al.
Figure 4. Fluorescent image of Cy5-labeled primer extension products
from a 92 nucleotide ssDNA template with a 40N random region using
TTP (lanes 1 and 9) or 6a-f (lanes 2-7 and 10-15, respectively), and
Taq (lanes 1-8), or D. Vent (exo-) (lanes 9-16). Lanes 8 and 16 are control
lanes with no TTP or dUTP.
Figure 3. Fluorescent image of primer extension products from a 92
nucleotide biotinylated ssDNA template using TTP or 6a, and KOD XL
(lanes 1 and 2), Pfu (exo-) (lanes 3 and 4), D. Vent (exo-) (lanes 5 and 6),
Tth (lanes 7 and 8), or Taq (lanes 9 and 10). All reactions were incubated
with streptavidin before gel loading. In lanes 1-10, odd numbered lanes
are with 6a and even numbered lanes are with TTP. Lanes 11 and 12 are
control lanes, with no TTP and no enzyme, respectively. Lane 13 is a dsDNA
ladder of 10 base increments starting with 20 bases.
Table 1. Primer Extension Efficiencies of Various DNA
Polymerases with 6a-f (Percentages Obtained from the Ratio of
the Amount of Full-Length Product from Primer Extensions with
6a-f to the Amount of Product Obtained with TTP)
triphosphate KOD XL Pfu (exo-) D. Vent (exo-)
Tth
Taq
KF (exo-)
the modifications used in these previously reported studies
contained an amide linkage connected to the 5-position. In
addition, longer linkers and even specific linker chemistries had
been proposed33,46 to be required to allow 5-position modified
dUTP analogues to work as substrates for DNA polymerases.
Further, it was typically required to adjust the PCR buffer and
primer extension conditions to get useful yields of full-length
DNA. These previous reports on the enzymatic incorporation
of C5-position modified deoxynucleotide triphosphates made
us concerned that the new triphosphates we had prepared would
be only poorly incorporated into DNA, if at all.
Two groups of three polymerases each from Family A and
Family B were screened using the dUTP analogue 6a in addition
to dATP, dCTP, and dGTP. Family A DNA polymerases
included Thermus aquaticus (Taq), Thermus thermohilus (Tth),
and the Klenow fragment (exo-). Family B DNA polymerases
included Pyroccus furiosus (Pfu, exo-), Pyrococcus species
GB-D (D. Vent), and Thermococcus kodakaraensis KOD1
(KOD XL).47 Each polymerase was used in the commercially
supplied buffer. To simulate what would be required for an in
vitro selection, a 92-mer template was used containing a 40N
random region flanked by fixed sequence regions for primer
annealing. The template used for these studies was synthesized
by automated DNA synthesis with biotin on the 5′-end so that
the primer extension product modified DNA could be separated
from the template strand by incubating the primer extension
mixture with streptavidin subsequent to analysis by denaturing
polyacrylamide gel electrophoresis. Visualization of DNA bands
was achieved by staining the gel with SYBR Gold. A repre-
sentative polyacrylamide gel of this screening reaction is shown
in Figure 3. All polymerases except Klenow fragment (exo-)
were able to incorporate 6a, but there were significant differ-
ences between them and the utility of their incorporation was
unclear. It appeared that Family B DNA polymerases were
somewhat superior to Family A with regard to incorporation of
this modified dUTP analogue.
TTP
6a
6b
6c
6d
6e
100.0% 100.0%
100.0%
80.9%
64.1%
47.5%
100.0%
51.6%
99.4%
100.0% 100.0% 100.0%
16.0% 15.5% 1.1%
85.9%
83.6%
84.6%
88.0%
88.1%
90.3%
39.3%
41.3%
28.4%
61.5%
34.4%
59.5%
3.7%
6.1%
5.8%
6.4%
7.1%
8.9%
0.4%
0.2%
0.7%
16.5% 14.3% 1.5%
30.1% 32.1% 1.4%
6f
It was next of interest to determine if 6a was unique in its
ability to be tolerated as a substrate by D. Vent polymerase
and not well-tolerated by Taq polymerase. Each of the modified
dUTP derivatives 6a-f was tested with these two polymerases.
In these primer extensions, a primer was used that had a 5′-
Cy-5 fluorescent dye attached so that only the product DNA
would be visualized during analysis by polyacrylamide gel
electrophoresis (Figure 4). As seen previously for 6a, Taq
polymerase gave low to modest yield of full-length DNA
product under the primer extension conditions used. In contrast,
D. Vent gave good to excellent yields of full-length DNA
product and appeared less sensitive to the type of modification
attached to the dUTP. Primer extension experiments were then
performed on all of the DNA polymerases of interest under the
following conditions: DNA polymerase (1.5 U), dATP, dCTP,
dGTP, TTP, or one of the modified dUTP derivatives 6a-f (final
concentration 200 µM each dNTP). Reactions were carried out
at 70 °C for 30 min, except for KF reactions, which were carried
out at 37 °C. Table 1 summarizes these primer extension
screening results. The percent yield for each polymerase was
obtained from the ratio of product from each primer extension
reaction with 6a-f relative to the amount of product from primer
extension with TTP. Generally, Family B DNA polymerases
gave greater yields of full-length DNA when incorporating 6a-f
than the Family A DNA polymerases. Surprisingly, KOD XL
and D. Vent (exo-) were able to incorporate all of the modified
dUTP derivatives 6a-f with similar or better yields than with
TTP. It will be noted that these yields were under screening
conditions comparable to those that would be used during in
vitro selection and should not be considered as relative Vmax
etc.
(46) Lee, S. E.; Sidorov, A.; Gourlain, T.; Mignet, N.; Thorpe, S. J.; Brazier,
J. A.; Dickman, M. J.; Hornby, D. P.; Grasby, J. A.; Williams, D. M.
Nucleic Acids Res. 2001, 29, 1565–1573.
Interestingly, the isobutyl-modified dUTP 6b gave the lowest
yields, as opposed to the uridines modified with larger aromatic
groups such as indole or napthyl. As expected as a result of
increased mass to charge ratio, all primer extension product
DNA incorporating 6a-f showed a decrease in electrophoretic
mobility compared to DNA where TTP was incorporated.
(47) KOD XL is approximately 98% exo- and 2% native KOD. See:
Nishioka, M.; Mizuguchi, H.; Fujiwara, S.; Komatsubara, S.; Kitaba-
yashi, M.; Uemura, H.; Takagi, M.; Imanaka, T. J. Biotechnol. 2001,
88, 141–149.
(48) Polacek, N.; Mankin, A. S. Crit. ReV. Biochem. Mol. Biol. 2005, 40,
285–311.
9
4144 J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010

Expanding the Chemistry of DNA for in Vitro Selection
A R T I C L E S
Scheme 2. Generic DNA In Vitro Selection
Figure 5. SYBR Gold stained gel image of PCR using fixed ssDNA
template (5′-ATA TAT ATG ATG TGA GTG TGT GAC GAG CTC AAT
TAT GGT GTT GCC TCT TCC TAG TCG TAC TGA TGC ATC CTC
TTC CAC CAC AAC CGA GAC ACA AAA AAA A-3′), KOD XL, dATP,
dGTP, dCTP, and modified dUTPs 6a-e (lanes 2-6, respectively) or TTP
(lane 1).
PCR Amplification of 6a-f. A typical procedure for perform-
ing DNA in vitro selection to identify new catalysts and
aptamers is shown in Scheme 2. The selection cycle begins with
the chemical synthesis of a template DNA strand library (ca.
1014 sequences) composed of a random sequence flanked by
fixed sequence regions required for PCR. PCR amplification of
this template library in the presence of a 3′-primer containing
a biotin at its 5′-end yields biotinylated dsDNA. Note that this
step helps ensure that the sequences to go forward into selection
are amplifiable. The dsDNA is then treated with streptavidin
immobilized on beads, and the complement strand is removed
from the beads by heating or raising the pH to generate a ssDNA
library in solution. The desired selection pressure is then applied
to the ssDNA for binding a target or performing a chemical
reaction. The selected ssDNA sequences are then subjected to
the same PCR procedure with the 3′-primer containing a biotin
at its 5′-end to yield a dsDNA library enriched in the sequences
suitable to perform the target binding or catalysis. In addition,
a significant number of background sequences will also be
carried forward so that the cycle must be repeated several times
to isolate the best sequences as per the selection conditions.
DNA in vitro selection is especially attractive because,
compared to RNA in vitro selection, only one enzymatic step
(PCR) is required compared to three (transcription, reverse
transcription, and PCR). Given the success of the primer
extension studies using 6a-f, it was of interest to determine if
the Family A and Family B polymerases discussed above could
function in PCR so that a DNA in vitro selection cycle could
be demonstrated. Furthermore, since a pool of sequences was
used in the primer extension experiments, it was unclear if
incorporation of multiple 5-position modified 2′-deoxyuridines
within a short sequence would result in lower yields of full-
length DNA. A test sequence complementary to the sequence
shown in Figure 5 was designed and prepared by automated
DNA synthesis. Real-time PCR using 6a-f was performed using
the complement to the sequence in Figure 5 as the input DNA
template. This amplified DNA sequence would contain every
triplet combination encoding TTP or modified dUTP insertion,
except TTT, and is designed to be challenging for a polymerase.
After 30-35 cycles Pfu (exo-), D. Vent (exo-), and KOD XL
appeared to generate PCR product by real-time analysis.
However, upon analysis by denaturing polyacrylamide gel
electrophoresis, only short fragments and traces of the full-length
DNA were being produced (Figure 5). Disappointingly, none
of the PCR enzymes tested were able to generate full-length
product using any of the modified dUTP derivatives. In the
positive control experiment, Pfu (exo-), Taq, D. Vent (exo-),
or KOD XL with the same template and TTP were all able to
generate full-length PCR product between 5 and 10 cycles using
real-time analysis, which was confirmed by denaturing poly-
acrylamide gel electrophoresis analysis. The lack of enzymatic
incorporation of these new modified dUTP derivatives was of
serious concern, and it was unclear if these new triphosphates
were of any use at all.
It seemed reasonable to propose that PCR failed because the
polymerase must use a modified DNA template for subsequent
cycles of PCR amplification after the first extension step, in
contrast to the primer extension reactions already discussed
above where the polymerases are only required to use an
unmodified DNA template and only need to contend with the
modified dUTP substrate. It appeared possible that using a
modified template was much more difficult for these poly-
merases than incorporation of these modified dUTP derivatives
using an unmodified DNA template.
Primer Extension of Modified Templates Containing 6a-f.
In order to test the possibility that primer extension could be
successful with TTP instead of modified dUTP to generate
unmodified DNA using a modified template, primer extension
was performed on six modified DNA templates prepared by
KOD primer extension using 6a-f. Because KOD XL and D.
Vent were able to effectively incorporate 6a-f during primer
extension, they were tested for their ability to read a modified
92-mer template containing a 40N random region and synthesize
the unmodified complement DNA strand. A biotinylated 3′-
primer was used so that the newly synthesized unmodified DNA
complement could be treated with streptavidin and separated
from the modified template strand on a polyacrylamide gel by
electrophoresis. Both KOD XL and D. Vent were able to
synthesize the unmodified complement strand. However, it was
found that increasing the time of primer extension to 30 min
was required to observe excellent yields of full-length DNA
from these modified templates. The data obtained from poly-
acrylamide gel electrophoresis and staining with Sybr Gold are
summarized in Table 2. Modified templates containing the
napthyl (6c), and imidazole (6f) groups gave the lowest yield
of full-length DNA for both KOD XL and D. Vent polymerase.
In addition, the template containing the indole group (6d) proved
to be a difficult template for KOD giving a modest yield of
9
J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010 4145

A R T I C L E S
Vaught et al.
Table 2. Yields of Full-Length Products from Primer Extension
Reactions on Modified DNA Templates Derived from 6a-f (Yields
Represent the Ratio of Streptavidin-Shifted Product to Modified
Template As Quantified by Fluorescence Using a Fluorescent Gel
Imager)
triphosphate
D. Vent (exo-)
KOD XL
TTP
6a
6b
6c
6d
6e
87%
89%
107%
42%
82%
89%
61%
86%
71%
65%
42%
44%
83%
45%
6f
Scheme 3. Revised DNA Selection Scheme To Allow Use of
Modified DNA
Figure 6. Fluorescent image of Sybr Gold stained primer extension products
from a 92 nucleotide ssDNA 40N random template attached to agarose
beads using either KOD XL or D. Vent (exo-) polymerase. dT is the TTP
control lane. Lanes 6a-6e are primer extension products using modified
dUTP derivatives 6a-f shown in Figure 1.
successful if the first PCR step was given a 30 min extension
time. Further, as discussed above, primer extension reactions
in solution were observed to work well with KOD XL and D.
Vent polymerases. However, Scheme 3 requires not only that
primer extension occur using modified dUTP derivatives but
also that it works using a template strand bound to a bead.
Therefore, we prepared a biotinylated template strand (92-mer,
40N random sequence) attached to a streptavidin-coated agarose
bead (Pierce), as described in Scheme 3. Primer extension
reactions with 6a-f (or TTP), dATP, dCTP, dGTP, and either
D. Vent or KOD XL were then performed on the agarose bead
immobilized antisense DNA strand. The modified DNA library
was then eluted off of the beads using NaOH (20 mM),
neutralized with HCl (80 mM), and analyzed by polyacrylamide
gel electrophoresis using Sybr Gold staining and fluorescent
imaging (Figure 6). Both KOD XL and D. Vent were able to
effectively incorporate all of the modified dUTP derivatives
6a-f during these bead-bound primer extension reactions. In
accord with the initial primer extension experiments, KOD XL
and D. Vent were able to generate full-length products at similar
yields to TTP with all six modified dUTP analogues. Taken
together, these results presented here set the stage for new DNA
in vitro selection experiments utilizing the chemical modifica-
tions described herein.
Selection of DNA versus Modified DNA Aptamers. With the
in vitro selection methods established as described above, it was
of interest to determine if, in fact, modified DNA aptamers to
proteins could be selected. To make a challenging test case
comparison between DNA and modified DNA, we chose tumor
necrosis factor receptor super family member 9 (TNFRSF9), a
difficult protein target that in previous DNA selections failed
to yield an aptamer with even poor binding (Kd > 100 nM; S.
Wilcox, unpublished results). TNFRSF9 is a member of the TNF
receptor super family and contributes to the clonal expansion,
survival, and development of human T cells. It has a pI of 8.16
and a molecular weight of 26.2 kDa. It is an important
therapeutic target for a variety of diseases in cancer and
inflammation.49 Notably, a RNA aptamer for this target has been
reported (Kd ) 4 × 10-8 M), but DNA aptamers for this
important human protein have not.50 As a positive control for
full-length DNA. Overall, these data suggest that D. Vent is
somewhat better than KOD XL in primer extension to produce
DNA from these modified templates.
Revised DNA In Vitro Selection Cycle. Because primer
extension was shown to work on these modified templates to
give good to excellent yields of full-length DNA, it seemed
plausible to construct a revised DNA selection scheme that
obviated the need to perform PCR with the modified dUTP
derivatives 6a-f (Scheme 3). As in Scheme 2, the selection
cycle of Scheme 3 begins with a DNA template library prepared
by automated DNA synthesis, and Scheme 3 is identical to
Scheme 2 in steps taken until the heat or NaOH wash step to
remove the DNA from the beads. At this point in the selection
cycle, in contrast to Scheme 2, the free ssDNA would be
discarded, leaving the beads with the complementary ssDNA
library attached. The beads would then be used for a primer
extension step using modified dUTP, dATP, dCTP, and dGTP.
This would then generate a modified DNA strand that could
subsequently be removed from the beads by another heat or
NaOH wash step to isolate the modified ssDNA library. We
have demonstrated that the amide linkage of these modified
dUTP derivatives is stable to basic conditions for extended
periods of time at ambient temperature. This modified ssDNA
library could then undergo the selection step for binding or
catalysis. The selected sequences would then go into PCR with
TTP, dATP, dCTP, and dGTP, avoiding PCR amplification
using the modified dUTP. Using the same 3′-primer containing
a 5′-biotin would create an enriched library ready to begin the
next cycle of selection.
(49) Croft, M. Nat. ReV. Immunol. 2009, 9, 271–285.
(50) McNamara II, J. O.; Kolonias, D.; Pastor, F.; Mittler, R. S.; Chen, L.;
Giangrande, P. H.; Sullenger, B.; Gilboa, E. J. Clin. InVest. 2008,
118, 376–386.
From the results presented above, it was clear that PCR
amplification of a modified template with TTP would be
9
4146 J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010

Expanding the Chemistry of DNA for in Vitro Selection
A R T I C L E S
Table 3. Measured Kd Values for Selected DNA Pools Containing
Either TTP, 6a, 6b, or 6d
target protein
TTP (Kd, M)
6a (Kd, M)
6b (Kd, M)
6d (Kd, M)
TNFRSF9
TACSTD2
>1 × 10-7
6 × 10-9
>1 × 10-7
4 × 10-9
9 × 10-9
3 × 10-9
5 × 10-9
5 × 10-10
the DNA in vitro selection methods described above, tumor-
associated calcium signal transducer 2 (TACSTD2) was also
used as a protein target in a parallel experiment. TACSTD2 is
known to bind random DNA pools with a Kd < 10-7 M and
appeared to be a good target for selection of DNA aptamers,
although it should be noted that no DNA aptamers had been
reported for this target. The starting random DNA libraries were
prepared using standard automated phosphoramidite synthesis
on an ABI 3900 synthesizer to give a base composition ratio of
ca. 1:1:1:1 A:C:G:T in the 40N random region. These aptamer
selection experiments on the protein targets TNFRSF9 and
TACSTD2 were monitored by C0t curve analysis for conver-
gence of the sequence pools. Protein concentrations for the first
cycle of selection were on the order of 1 µM and DNA library
concentration 1 nM. Selection for affinity binding was driven
by decreasing protein concentration during the course of the
selection. Both selection experiments were stopped at 8 cycles
based on convergence.
Figure 7. Plot of filter binding data for Kd measurements of the in vitro
selected pools and the starting random libraries using either TTP, 6a, 6b,
or 6d for TNFRSF9. Note that for TTP “selected” and TTP random, no
binding was observed in either case. Fraction bound was calculated as the
fraction of radiolabeled DNA bound to protein immobilized on beads as a
function of protein concentration and were normalized to the amount of
DNA bound at the curve plateau.
The PCR product of the cycle 8 pool was converted into the
ssDNA aptamer pool by primer extension methods using KOD
as described above. Analysis of these experiments began by
measuring the Kd of the DNA and modified DNA aptamer pools
using radiolabeled pools in a filter binding assay. A cutoff of
100 nM was used for the Kd measurement with a measured Kd
above this value being considered a failure. In support of the
notion that TACSTD2 was a suitable control protein, all of the
DNA selections (TTP, 6a, 6b, and 6d) yielded aptamers. Good
(Kd ) 9 nM) to excellent (Kd ) 0.5 nM) binding affinity was
observed for this target protein (Table 3). As had been observed
previously, the DNA library derived from TTP resulted in a
failed selection for TNFRSF9. The selection using the DNA
library derived from 6b also failed on this target. Gratifyingly,
selections performed with DNA libraries derived from 6a and
6d yielded sequence pools with Kd values well below the cutoff
limit of 100 nM (10-7 M). Perhaps most surprisingly, the DNA
aptamer pool derived from 6d had a significantly lower Kd for
TNFRSF9 than that reported for the unmodified RNA aptamer
(Table 3 and Figure 7). These data clearly show the advantage
of using DNA base modifications at the 5-postion of the uridine
ring in aptamers.
However, it could not be ruled out from the pool data that
the types of modifications used for the aptamers described herein
were simply hydrophobic and that the specific structure of the
5-position modification was less important. Moreover, without
sequencing and performing base composition analysis on the
clones, it was unclear if the modified dUTP analogues were
even tolerated in the structures or of any importance whatsoever.
To test this possibility, a representative “hit” sequence was
examined for binding to the target protein TNFRSF9.
Figure 8. Plot of filter binding data for Kd measurements of the in vitro
selected DNA sequence clone 1684-40 and the evolved cycle 8 pool for
the protein target TNFRSF9 using either TTP, 6a, or 6d for enzymatic
preparation. Fraction bound was calculated as the fraction of radiolabeled
DNA bound to protein immobilized on beads as a function of protein
concentration and normalized to the amount of DNA bound at the curve
plateau.
sequence was C C T G C A C C C A G T G T C C C C G A
C G G G G C C C T C T A G C C G T A C T C T G T A A
T G G C G G A T G C T G A C G G A G A G G A G G A
C G G, where the letters in bold italics indicate the sequence
evolved from the random region. The base composition of this
sequence was representative of the aptamers in this family at
31% C, 21% T (modified dU), 31% G, and 17% A. Note that
the sequence was prepared enzymatically by primer extension
using KOD XL as described above, which means that only
positions labeled italicized T will contain a modification within
the DNA sequence.
To determine if the binding to TNFRSF9 was merely a
consequence of adding some hydrophobic groups to clone 1684-
40 or a more specific structural effect unique to the modification
used in the selection of this sequence, primer extension reactions
were performed using either 6d (indole dUTP analogue used
in the selection), 6a (benzyl), or 6c (napthyl) dUTP analogues.
In addition, clone 1684-40 was also transcribed with TTP to
determine if this sequence required modification at all to bind
TNFRSF9. Employing the same filter binding assay as used to
study the evolved DNA aptamer pool binding, and including a
control experiment with the cycle 8 aptamer pool, the Kd values
were compared and contrasted, as shown in Figure 8. No Kd
was observed within the 100 nM cutoff limits for the DNA
aptamer sequence 1684-40 binding TNFRSF9 when prepared
from TTP (red squares). Similarly, clone 1684-40 prepared by
Cloning and sequencing of the cycle 8 aptamer pool from
the 6d selection on the TNFRSF9 target was performed and
revealed two distinct families. From a sequencing perspective,
this selection resulted in a high degree of convergence, making
analysis of the major family descriptive of the outcome of the
in vitro selection. DNA clone 1684-40 was chosen for further
study because it was a member of a major sequence family. Its
9
J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010 4147

A R T I C L E S
Vaught et al.
enzymatic transcription with either 6a or 6c gave no binding to
TNFRSF9 within the 100 nM cutoff. In contrast, enzymatic
preparation of 1684-40 with 6d gave a Kd of approximately 5
nM in binding TNFRSF9, within experimental error, the same
as that observed for the evolved cycle 8 pool. It is somewhat
surprising that substitution of 6c (napthyl) for another bicyclic
aromatic 6d (indole) did not give an aptamer with at least weak
binding to TNFRSF9. Perhaps these aromatic groups confer
different aptamer structures, or these apparent subtle differences
in pendant aromatic groups confer more significant changes at
the aptamer protein interface, or both. Of course, it is true that
indole is not just another aromatic and can form another H-bond
to a target, whereas napthyl cannot. It may be that the indole
groups in clone 1684-40 form important contacts with TN-
FRSF9.
Further spectroscopic work is underway to determine the
molecular basis for the apparent binding improvements conferred
by these modified DNA aptamers for a refractory protein target
TNFRSF9, as well as the control protein target TACSTD2. In
addition, we hope to learn if aptamer structures may be altered
by the presence of these modifications. In any event, we are
encouraged by these successful in vitro selections against
refractory protein targets, and significant effort is now underway
to select aptamers for a wide range of targets in the human
proteome.
of this effort will show the extent to which these new DNA
aptamers can serve as useful reagents in this endeavor, the
ultimate goal of which is to discover proteomic biomarkers to
help detect, diagnose, and treat human disease. It is anticipated
that modified DNA aptamers will be generated by the chem-
istries and methods described herein to greater than 1000 human
protein targets by the end of 2010.
Experimental Section
General. All compounds were prepared from reagent grade
starting materials as purchased from Aldrich unless otherwise noted.
Tetrakis[triphenylphosphine]palladium(0) was used as received from
Strem Chemicals. NMR spectra were recorded on Varian 300 or
400 MHz spectrometers and are reported as δ values, referenced
to solvent resonances. ESI mass spectra were recorded at Soma-
Logic (Boulder, CO).
5′-Primer (5′-ATATATATGATGTGAGTGTGTGACGAG-3′)
was obtained from IDT and used without further purification.
Biotinylated primer (5′-B-TTTTTTTTGTGTCTCGGTTGTGGTG-
3′), Cy-5-labeled primer (5′-Cy5-TTTTTTTTGTGTCTCGGT-
TGTGGTG-3′), 40N random template (5′-B-TTTTTTTTG-
TGTCTCGGTTGTGGTG40NCTCGTCACACACTCACATCAT-
ATATAT), and triplet dU template (TTTTTTTTGTGTCTCGGT-
TGTGGTG-GAA GAG GAT GCA TCA GTA CGA CTA GGA
AGA GGC AAC ACC ATA ATT GAG-CTCGTCACACACTCA-
CATCATATATAT), where B and Cy-5 represent either a biotin group
or Cy-5 group appended during DNA synthesis using biotin phos-
phoramidite or Cy-5 phosphoramidite from Glen research, were
prepared using standard phosphoramidite chemistry on an ABI 394,
reversed-phase HPLC purified, then gel purified. Streptavidin used for
gel shift analysis of primer extension products was dissolved in
phosphate buffered saline. UltraLink Immobilized Streptavidin Plus
agarose beads (Pierce) were washed 3 times with 1× wash buffer (100
mM NaCl, 40 mM HEPES pH 7.5, 5 mM KCl, 1 mM MgCl2, and
0.25% Tween-20) and resuspended to their original volume as a 50%
slurry immediately before use. All gels of unlabeled DNA were
visualized and quantitated by staining with SYBR Gold nucleic acid
gel stain (Invitrogen) and imaged on a FujiFilm 5100 fluorescent
imager.
Synthesis of Nucleoside Precursors, 5a-5f. In a heavy-walled
glass vessel (200 mL) equipped with a Teflon vacuum stopcock,
5-iodo-5′-O-DMTr-3′-O-acetyl-2′-deoxyuridine (0.72 mmol), 3
equiv of amine, 5 equiv of triethylamine, 0.1 equiv of tetrakis-
[triphenylphosphine]palladium(0), and 15 mL of 1:1 DMA/THF
(freshly distilled off of sodium/potassium alloy and benzophenone)
were added. The vessel was evacuated and filled with 50 psi of
CO three times, then sealed and heated at 70 °C for 48-72 h. The
reaction was cooled, vented, and evaporated under high vacuum
to a bubbly yellow solid. The resulting solid was purified with flash
silica gel using 1:20 MeOH/CH2Cl2. The nucleoside (ca. 95% pure
by 1H NMR) was then dissolved a 1:1 solution of MeOH/CH2Cl2;
1.1 equiv of 3% TCA (in CH2Cl2) was added, and the solution
was stirred at room temperature for 2-4 h. The reaction mixture
was evaporated to dryness under high vacuum, and the resulting
bright orange residue was purified with flash silica gel using 5-10%
MeOH in CH2Cl2. Products were then used as recovered or
recrystallized from ethyl acetate, yielding white to slightly yellowish
crystals.
Conclusions
Streamlined synthetic methods are now available to prepare
5-position modified dUTP derivatives analogous to the UTP
derivatives previously used to successfully identify new RNA
catalysts for organic reactions via in vitro selection. PCR
amplification typically used for DNA in vitro selection was
unsuccessful. Fortunately, the dUTP derivatives of this study
were incorporated efficiently in a primer extension step, and a
modified in vitro selection method was developed that enables
the use of these dUTP derivatives for in vitro selection. Of the
DNA polymerases studied, the Family B polymerases, D. Vent
and KOD XL, were most successful in generating full-length
DNA. The modified DNA template could also be used in primer
extension using TTP to create unmodified DNA suitable for
standard PCR using these polymerases. The ability to perform
primer extension with 6a-f on streptavidin agarose beads
demonstrates the final step for a revised DNA in vitro selection
that avoids performing PCR with modified dUTP derivatives.
With these new DNA modification chemistries established,
it will be possible to perform in vitro selection experiments using
DNA that were previously successful with RNA. There are many
apparent (2′-hydroxyl) and less apparent (pseudorotaion of
ribose) differences between RNA and DNA structure and
dynamics. It is unknown if these structural differences make
RNA a superior catalyst or aptamer when compared to DNA.
However, for the protein target TNFRSF9, the modified DNA
aptamer described herein had a lower Kd when compared to
that of the published RNA aptamer (approximately 5 versus 48
nM). Having these chemistries and in vitro selection methods
defined expands the types of modifications available for the
DNA aptamer field and paves the way to compare these two
related nucleic acid biopolymer platforms.
Spectroscopic Data for Nucleosides 5a-f.
1
5a: H NMR (400 MHz, DMSO-d6) δ 11.90 (s, 1H), 9.11 (s,
1H), 8.78 (s, 1H), 7.27 (m, 5H), 6.14 (t, J ) 7 Hz, 1H), 5.22 (s,
1H), 5.19 (s, 1H), 4.47 (d, J ) 6 Hz, 2H), 4.09 (s, 1H), δ 3.61 (s,
2H), 2.34 (m, 2H), 2.05 (s, 3H); 13C NMR (300 MHz, DMSO-d6)
δ 170.7, 164.0, 163.8, 150.7, 146.7, 141.6, 138.1, 128.9, 127.5,
127.2, 105.9, 87.0, 86.1, 75.4, 61.8, 38.3, 21.5.
The DNA modification chemistry and associated in vitro
selection methods described herein are now being applied in a
large-scale effort to select aptamers to the estimated 3500
proteins at the core of the circulating human proteome to use
in multiplexed arrays for human proteome analysis. The results
5b: 1H NMR (300 MHz, DMSO-d6) δ 8.75 (s, 1H), 8.71 (s, 1H),
6.11 (t, J ) 7 Hz, 1H), 5.20 (s, 1H), 5.15 (s, 1H), 4.06 (s, 1H),
3.58 (s, 2H), 3.30 (s, 1H), 3.07 (t, J ) 6 Hz, 2H), 2.32 (m, 2H),
9
4148 J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010

Expanding the Chemistry of DNA for in Vitro Selection
A R T I C L E S
2.02 (s, 3H), 1.73 (m, 1H), 0.84 (d, J ) 6 Hz, 6 H); 13C NMR (300
MHz, DMSO-d6) δ 170.7, 164.0, 162.0, 150.2, 146.5, 106.3, 86.1,
86.0, 75.4, 61.8, 46.4, 38.3, 28.8, 21.5, 20.7.
Hz, 1H), 4.64 (s, 1H), 4.58 (s, 2H), 4.44 (m, 1H), 4.07 (d, J ) 4
Hz, 2H), 3.99 (m, 2H), 2.29 (m, 2H); 31P NMR (400 MHz, D2O)
δ -9.80 (d, J ) 46 Hz, 1P), -10.40 (d, J ) 50 Hz, 1P), -22.33
(t, J ) 44 Hz, 1P); HRMS (ESI) calcd for C16H20N4O15P3 (M-)
601.0143, found 601.0169.
1
5c: H NMR (400 MHz, DMSO-d6) δ 11.92 (s, 1H), 9.14 (t, J
) 6 Hz, 1H), 8.80 (s, 1H), 8.09 (d, J ) 8 Hz, 1H), 7.94 (d, J ) 8
Hz, 1H), 7.85 (m, 1H), 7.74 (m, 1H), 7.53 (m, 3H), 6.13 (t, J ) 7
Hz, 1H), 5.22 (m, 1H), 4.93 (d, J ) 6 Hz, 2H), 4.10 (d, J ) 4 Hz,
1H), 3.61 (s, 2H), 2.35 (m, 2H), 2.05 (s, 3H); 13C NMR (400 MHz,
DMSO-d6) δ 170.7, 163.9, 162.0, 150.2, 146.8, 135.7, 134.0, 131.5,
129.3, 128.2, 127.1, 126.6, 126.3, 126.2, 124.0, 121.2 106.1, 86.2,
86.0, 75.4, 61.9, 38.3, 21.5.
1
6f: UV λmax 276 nm ε 13 500 cm-1 M-1; H NMR (400 MHz,
D2O) δ 8.36 (s, 1H), 8.27 (s, 1H), 7.04 (s, 1H), 6.05 (t, J ) 7 Hz,
1H), 4.47 (m, 1H), 4.09 (d, J ) 2 Hz, 1H), 4.02 (dd, J ) 10, 7 Hz
2H), 3.49 (ddd, J ) 9, 6, 3 Hz, 2H), 2.82 (t, J ) 6 Hz, 2H), 2.28
(m, 2H); 31P NMR (400 MHz, D2O) δ -8.14 (d, J ) 48 Hz, 1P),
-10.32 (d, J ) 47 Hz, 1P), -21.82 (t, J ) 49 Hz, 1P); LRMS
(ESI) calcd for C15H21N5O15P3 (M-) 604.3, found 604.3
1
5d: H NMR (300 MHz, DMSO-d6) δ 11.87 (s, 1H), 10.81 (s,
1H), 8.80 (t, J ) 8 Hz, 1H), 8.74 (s, 1H), 7.56 (d, J ) 8 Hz, 1H),
7.32 (d, J ) 8 Hz, 1H), 7.14 (d, J ) 2 Hz, 1H), 7.05 (t, J ) 7 Hz,
1H), 6.95 (t, J ) 8 Hz, 1H), 6.13 (t, J ) 7 Hz, 1H), 5.22 (s, 1H),
5.18 (t, J ) 8 Hz, 1H), 4.07 (s, 1H), 3.61 (s, 2H), 3.54 (dd, J ) 13,
Primer Extension Reactions for Enzyme Screening with
6a-f. Pfu (exo-) and KF (exo-) were purchased from Stratagene;
Taq was purchased from Abgene; KOD XL was purchased from
Novagen; Deep Vent (exo-) was purchased from New England
Biolabs, and Tth was purchased from Epicenter Biotechnologies.
Reaction buffers were used as received by the supplier of the DNA
polymerase except for KF, which was prepared according to the
test conditions described in the product insert and consisted of 66
mM KHPO4 (pH 7.4), 10 mM MgCl2, and 1 mM ꢀ-mercaptoet-
hanol. 5′-Primer either labeled with biotin or with Cy-5 (final
concentration 3.75 µM) and 40N template (final concentration 0.75
µM) was mixed in 1× reaction buffer and heated to 95 °C for 1
min and then cooled to room temperature over 10 min. DNA
polymerase (1.5 U), dATP, dCTP, dGTP, TTP or one of the
modified dUTP derivatives 6a-f (final concentration 200 µM each
dNTP), and MgCl2 (to Pfu, Tth, and Taq as specified in each product
insert) were added to a final volume of 15 µL. Reactions were
carried out at 70 °C for 30 min, except for KF reactions, which
were carried out 37 °C. For reactions analyzed by gel shift,
streptavidin was immediately added to each reaction at 10-fold
molar excess to template. After 1 min, 15 µL of denaturing gel-
loading buffer (8 M urea, 20 mM Tris pH 8.0, 1 mM EDTA, and
0.05% bromophenol blue) was added to each reaction and heated
to 70 °C for 2 min. Reactions were analyzed by 8% PAGE (19:1
acrylamide/bisacrylamide, 8 M urea, 45-50 °C), visualized, and
quantified by fluorescence.
7 Hz, 2H), 2.89 (t, J ) 7 Hz, 2H), 2.31 (m, 2H), 2.05 (s, 3H); 13
C
NMR (300 MHz, DMSO-d6) δ 170.7, 163.7, 162.0, 150.3, 146.7,
141.4, 136.9, 127.8, 123.5, 123.3, 121.6, 119.0, 112.2, 112.1, 106.4,
86.1, 86.0, 75.4, 61.9, 38.3, 25.9, 21.5.
1
5e: H NMR (400 MHz, DMSO-d6) δ 11.90 (s, 1H), 9.09 (t, J
) 6 Hz, 1H), 8.45 (m, 2H), 8.41 (s, 1H), 7.57 (m, 2H), 6.10 (t, J
) 7 Hz, 1H), 5.20 (s, 1H), 5.17 (s, 1H), 4.44 (d, J ) 6 Hz, 2H),
4.09 (d, J ) 4 Hz, 1H), 3.51 (s, 2H), 2.31 (m, 2H), 2.05 (s, 3H);
13C NMR (300 MHz, DMSO-d6) δ 170.7, 164.0, 163.8, 150.7,
146.7, 141.6, 138.1, 128.9, 127.5, 127.2, 105.9, 87.0, 86.1, 75.4,
61.8, 38.3, 21.5.
1
5f: H NMR (400 MHz, DMSO-d6) δ 11.85 (s, 1H), 9.44 (s,
1H), 8.53 (s, 1H), 8.30 (s, 1H), 7.49 (s, 1H), 6.77 (s, 1H), 6.15 (t,
J ) 8 Hz, 1H), 5.21 (d, J ) 3 Hz, 2H), 4.01 (s, 1H), 3.59 (ddd, J
) 16, 12, 4 Hz, 2H), 3.44 (dd, J ) 13, 7 Hz, 2H), 2.66 (t, J ) 7
Hz, 2H), 2.25 (m, 2H), 2.04 (s, 3H); 13C NMR (400 MHz, DMSO-
d6) δ 170.7, 164.5, 163.8, 154.0, 144.9, 135.3, 129.4, 106.3, 94.6,
85.8, 85.4, 79.9, 75.6, 62.0, 38.1, 29.9, 21.6.
Synthesis of UTP Derivatives 6a-f. All 5-position modified
dUTP derivatives were prepared as described by the method of
Eckstein.36
Spectroscopic Data for dUTPs 6a-f.
1
6a: UV λmax 280 nm ε 13 700 cm-1 M-1; H NMR (400 MHz,
Primer Extension Reactions with 6a-f on Streptavidin
Agarose Beads. A 40N random template or triplet dU template
was amplified by PCR with 3′-biotinylated primer and 5′-primer.
The resulting dsDNA was immobilized on streptavidin agarose
beads (Pierce) by adding 15 µL of a 50% bead slurry per 50 µL
PCR reaction. Reactions were combined and shaken at room
temperature for 5 min before being spun in a centrifuge for 20 s.
Supernatant was removed, and then the beads were washed three
times with 1× wash buffer by shaking, spinning, and removing
supernatant. ssDNA was then eluted from the beads using 80 mM
NaOH and neutralized with 20 mM HCl. The beads were washed
three times with 1× wash buffer and resuspended in 1× primer
extension buffer consisting of 1× reaction buffer (as supplied by
polymerase manufacturer), 1.5 U KOD XL or D. Vent, 5′-primer
(final concentration 4 µM), dATP, dCTP, dGTP, and TTP or one
of the modified dUTP derivatives 6a-f (final concentration 400
µM each dNTP) at a final volume of 50 µL. Two 50 µL PCR
reactions (30 µL of orginal 50% bead slurry) were used per single
50 µL primer extension reaction. Reactions were then heated to 95
°C for 30 s, annealed at 55 °C for 15 s, heated at 70 °C for 30 min
with shaking, and then spun in a centrifuge for 20 s. Supernatant
was removed and each reaction washed three times with 1× wash
buffer as above. Modified ssDNA was then eluted from the beads
using 80 mM NaOH and neutralized with 20 mM HCl. Beads were
then washed three times with 1× wash buffer before being used
again. Beads were used four times for primer extension reactions
with no detectable decrease in product yield. Eluted products were
then diluted in 2× denaturing gel-loading buffer (8 M urea, 20
mM Tris pH 8.0, 1 mM EDTA, and 0.05% bromophenol blue),
analyzed by 8% denaturing PAGE, and visualized by fluorescence.
D2O) δ 8.38 (s, 1H), 7.19 (m, 5H), 6.06 (t, J ) 7 Hz, 1H), 4.46
(m, 1H), 4.39 (d, J ) 6 Hz, 2H), 4.08 (dd, J ) 7, 4 Hz, 1H), 4.02
(m, 2H), 2.28 (dd, J ) 10, 4 Hz, 2H); 31P NMR (400 MHz, D2O)
δ -9.64 (d, J ) 51 Hz, 1P), 10.40 (d, J ) 52 Hz, 1P), 22.28 (t, J
) 48 Hz, 1P); HRMS (ESI) calcd for C17H21N3O15P3 (M-)
600.0191, found 600.0197.
1
6b: UV λmax 276 nm ε 10 200 cm-1 M-1; H NMR (400 MHz,
D2O) δ 8.41 (s, 1H), 6.09 (t, J ) 7 Hz, 1H), 4.47 (m, 1H), 4.10
(m, 1H), 4.04 (m, 2H), 2.27 (m, 2H), 1.71 (m, 2H), 0.75 (d, J ) 6
Hz, 6H); 31P NMR (400 MHz, D2O) δ -9.64 (d, J ) 51 Hz, 1P),
-10.40 (d, J ) 52 Hz, 1P), -22.28 (t, J ) 48 Hz, 1P); HRMS
(ESI) calcd for C14H23N3O15P3 (M-) 566.0347, found 566.0354.
1
6c: UV λmax 281 nm ε 20 000 cm-1 M-1; H NMR (400 MHz,
D2O) δ 8.22 (s, 1H), 7.79 (d, J ) 8 Hz, 1H), 7.74 (d, J ) 7 Hz,
1H), 7.66 (d, J ) 8 Hz, 1H), 7.39 (m, 2H), 7.29 (m, 2H), 5.93 (t,
J ) 7 Hz, 1H), 4.70 (dd, J ) 27, 16 Hz, 2H), 4.42 (m, 1H), 4.08
(m, 1H), 4.02 (m, 2H), 2.26 (m, 1H), 2.18 (m, 1H); 31P NMR (400
MHz, D2O) δ -10.22 (d, J ) 46 Hz, 1P), -10.62 (d, J ) 50 Hz,
1P), -22.67 (t, J ) 44 Hz, 1P); HRMS (ESI) calcd for
C21H23N3O15P3 (M-) 650.0347, found 650.0365.
1
6d: UV λmax 279 nm ε 13 100 cm-1 M-1; H NMR (400 MHz,
D2O) δ 8.22 (s, 1H), 7.79 (d, J ) 8 Hz, 1H), 7.74 (d, J ) 7 Hz,
1H), 7.66 (d, J ) 8 Hz, 1H), 7.39 (m, 2H), 7.29 (m, 2H), 5.93 (t,
J ) 7 Hz, 1H), 4.70 (dd, J ) 27, 16 Hz, 2H), 4.42 (m, 1H), 4.08
(m, 1H), 4.02 (m, 2H), 2.26 (m, 1H), 2.18 (m, 1H); 31P NMR (400
MHz, D2O) δ -10.22 (d, J ) 46 Hz, 1P), -10.62 (d, J ) 50 Hz,
1P), -22.67 (t, J ) 44 Hz, 1P); HRMS (ESI) calcd for
C20H24N4O15P3 (M-) 653.0456, found 653.0450.
1
6e: UV λmax 277 nm ε 12 100 cm-1 M-1; H NMR (400 MHz,
D2O) δ 8.44 (m, 2H), 8.40 (s, 1H), 7.57 (m, 2H), 6.04 (t, J ) 7
9
J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010 4149

A R T I C L E S
Vaught et al.
Primer Extension Reactions with dATP, dCTP, dGTP,
and TTP on Templates Modified with 6a-f. Modified 40N
template was enzymatically synthesized on beads and purified as
described above, mixed with 3′-biotinylated primer in 1× reaction
buffer (as supplied by polymerase manufacturer), heated to 95 °C
for 1 min, cooled to room temperature over 10 min, and then
subjected to reaction conditions identical to those described above
for the primer extension reactions used for the enzyme screening,
using only KOD XL and D. Vent (exo-).
Real-Time PCR with 6a-f. PCR was performed on a Biorad
MyIQ real-time PCR thermal cycler with single-color detection.
Reactions were carried out in 50 µL 1× reaction buffer, MgCl2
(an additional 2 mM to that contained in the supplied reaction buffer
or an additional 2 mM to the minimal amount specified for the
reaction buffers without MgCl2), 1.25 U polymerase, dATP, dCTP,
dGTP, and TTP or one of the modified dUTP derivatives 6a-f
(final concentration 200 µM each dNTP), 5′-primer and 3′-
biotinylated primer (final concentration 2 µM each), 0.05 pmol
triplet dU template, and SYBR Green I (Invitrogen). Thermal
cycling conditions consisted of an initial denaturing step at 95 °C
for 1 min, 55 °C for 15 s, 70 °C for 1 min, then repeated cycling
through these steps only cutting back the denaturing step at 95 °C
to 30 s. Products were then diluted in 2× denaturing gel-loading
buffer (8 M urea, 20 mM Tris pH 8.0, 1 mM EDTA, and 0.05%
bromophenol blue), analyzed by 8% denaturing PAGE, and
visualized by fluorescence.
a 0.5 µM candidate DNA mixture was prepared in 40 µL of
SB1T. Ten microliters of a protein competitor mixture was added
to each DNA mix (0.1% HSA, 10 µM casein, and 10 µM
prothrombin in SB1T).
Binding reactions were performed by adding 50 µL target protein-
coated beads or (His)6-coated beads (5 mg/mL in SB1T) to the
DNA mixture and incubating at 37 °C for 15 min with mixing.
The DNA solution was removed, and the beads were washed five
times at 37 °C with SB1T containing 0.1 mg/mL herring sperm
DNA (Sigma-Aldrich). Unless indicated, all washes were performed
by resuspending the beads in 100 µL wash solution, mixing for
30 s, separating the beads with a magnet, and removing the wash
solution. Bound aptamers were eluted from the beads by adding
100 µL of SB1T + 2 M guanidine-HCl and incubating at 37 °C
for 5 min with mixing. The aptamer eluate was transferred to a
new tube after magnetic separation. After the first two selection
rounds, the final two of five target bead washes were done for 5
min instead of 30 s.
Primer beads were prepared by immobilizing biotinylated reverse
PCR primer (5′-ABABTTTTTTTTCCGTCCTCCTCTCCGTC-3′)
to streptavidin-coated paramagnetic beads (MyOne-SA, Dynal).
Five milliliter MyOne-SA beads (10 mg/mL) were washed once
with NaClT (5 M NaCl, 0.01% Tween-20) and resuspended in 5
mL of biotinylated reverse PCR primer (5 µM in NaClT). The
sample was incubated at 25 °C for 15 min, washed twice with 5
mL of NaClT, resuspended in 12.5 mL of NaClT (4 mg/mL), and
stored at 4 °C.
In Vitro Selection. DNA and modified DNA libraries were
prepared with dATP, dGTP, 5-methyl-dCTP (MedCTP), and either
dTTP or one of three dUTP analogues: 6a, 6b, or 6d. Candidate
mixtures were prepared by polymerase extension of a primer
annealed to a biotinylated template (5′- ABABCCGTCCTC-
CTCTCCGT-40N-GGGACACTGGGTGCAGG-3′), where B in-
dicates a biotin incorporated during DNA synthesis and 40N
indicates a 40 nucleotide random region. For each candidate mixture
composition, 4.8 nmol forward PCR primer (5′-ATATATATCCT-
GCACCCAGTGTCCC-3′) and 4 nmol template were combined
in 100 µL 1× KOD DNA polymerase buffer (Novagen), heated to
95 °C for 8 min, and cooled on ice. Each 100 µL primer/template
mixture was added to a 400 µL extension reaction containing 1×
KOD DNA polymerase buffer, 0.125 U/µL KOD DNA polymerase
(Novagen), and 0.5 mM each dATP, MedCTP, dGTP, and dTTP
or dUTP analogue and incubated at 70 °C for 30 min. Double-
stranded product was captured via the template strand biotins
by adding 1 mL of streptavidin-coated magnetic beads (Magna-
Bind Streptavidin, Pierce, 5 mg/mL in 1 M NaCl + 0.05%
Tween-20) and incubating at 25 °C for 10 min with mixing.
Beads were washed three times with 0.75 mL of SB1T buffer
(40 mM HEPES, pH 7.5, 125 mM NaCl, 5 mM KCl, 1 mM
MgCl2, 1 mM CaCl2, 0.05% Tween-20). The aptamer strand was
eluted from the beads with 1.2 mL of 20 mM NaOH, neutralized
with 0.3 mL of 80 mM HCl, and buffered with 15 µL of 1 M
HEPES, pH 7.5. Candidate mixtures were concentrated with a
Centricon-30 to approximately 0.2 mL and quantified by UV
absorbance spectroscopy.
The extracellular domains of TNFRSF9 and TACSTD2 fused
to IgG1 and 6-His were purchased from R&D Systems. TNFRSF9
and TACSTD2 were immobilized on Co2+-NTA paramagnetic
beads (TALON, Dynal). Target proteins were diluted to 0.2 mg/
mL in 0.5 mL of B/W buffer (50 mM Na-phosphate, pH 8.0, 300
mM NaCl, 0.01% Tween-20) and added to 0.5 mL of TALON
beads (prewashed three times with B/W buffer and resuspended to
10 mg/mL in B/W buffer). The mixture was rotated for 30 min at
25 °C and stored at 4 °C until use. TALON beads coated with
(His)6 peptide were also prepared and stored as above. Prior to use,
beads were washed three times with B/W buffer, once with SB1T,
and resuspended in SB1T.
Twenty-five microliter primer beads (4 mg/mL in NaClT) were
added to the 100 µL aptamer solution in guanidine buffer and
incubated at 50 °C for 15 min with mixing. The aptamer solution
was removed, and the beads were washed five times with SB1T.
Aptamer was eluted from the beads by adding 85 µL of 20 mM
NaOH and incubating at 37 °C for 1 min with mixing. Eighty
microliters of aptamer eluate was transferred to a new tube after
magnetic separation, neutralized with 20 µL of 80 mM HCl,
and buffered with 1 µL of 0.5 M Tris-HCl, pH 7.5.
Selected aptamer DNA was amplified and quantified by QPCR.
Forty-eight microliters of DNA was added to 12 µL of QPCR Mix
(5X KOD DNA polymerase buffer, 25 mM MgCl2, 10 µM forward
PCR primer, 10 µM biotinylated reverse PCR primer, 5× SYBR
Green I, 0.125 U/µL KOD DNA polymerase, and 1 mM each
dATP, dCTP, dGTP, and dTTP) and thermal cycled in an ABI5700
QPCR instrument with the following protocol: 1 cycle of 99.9 °C,
15 s, 55 °C, 10 s, 70 °C, 30 min; 30 cycles of 99.9 °C, 15 s, 72
°C, 1 min. Quantification was done with the instrument software,
and the number of copies of DNA selected with target beads
and (His)6 beads was compared to determine signal/background
ratios.
Following amplification, the PCR product was captured on
MyOne-SA beads via the biotinylated antisense strand; 1.25 mL
MyOne-SA beads (10 mg/mL) were washed twice with 0.5 mL
of 20 mM NaOH, once with 0.5 mL of SB1T, resuspended in
2.5 mL of 3 M NaCl, and stored at 4 °C. Twenty-five microliter
MyOne-SA beads (4 mg/mL in 3 M NaCl) were added to 50 µL
double-stranded QPCR product and incubated at 25 °C for 5
min with mixing. The beads were washed once with SB1T, and
the “sense” strand was eluted from the beads by adding 200 µL
of 20 mM NaOH and incubating at 37 °C for 1 min with mixing.
The eluted strand was discarded, and the beads were washed
three times with SB1T and once with 16 mM NaCl. Aptamer
sense strand was then prepared with the appropriate nucleotide
composition as described above and the whole cycle repea-
ted.
Kd Measurements. Affinities of the enriched libraries were
measured using TALON bead partitioning. DNA was renatured by
heating to 95 °C and slowly cooling to 37 °C. Complexes were
formed by mixing a low concentration of radiolabeled DNA (∼1
× 10-11 M) with a range of concentrations of target protein (1 ×
10-7 to 1 × 10-12 M final) in SB1 buffer and incubating at 37 °C.
Affinity selections were performed separately with each
candidate mixture, comparing binding between target protein
beads (signal) and (His)6 beads (background). For each sample,
9
4150 J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010

Expanding the Chemistry of DNA for in Vitro Selection
A R T I C L E S
A portion of each reaction was transferred to a nylon membrane
and dried to determine total counts in each reaction. Twenty-five
micrograms of MyOne TALON beads (Invitrogen) was added to
the remainder of each reaction and mixed at 37 °C for 1 min. A
portion was then passed through a MultiScreen HV plate under
vacuum to separate protein-bound complexes from unbound DNA
and washed with 100 µL of SB1 buffer. The nylon membranes
and MultiScreen HV plates were phosphorimaged, and the amount
of radioactivity in each sample was quantified using a FUJI FLA-
3000. The fraction of captured DNA was plotted as a function of
protein concentration, and a nonlinear curve-fitting algorithm was
used to extract equilibrium binding constants (Kd values) from the
data.
Acknowledgment. We thank the U.S. Department of Energy
(DOE), Office of Basic Energy Science (BES), and the National
Science Foundation (CHE 0414527) for partial financial support
of this work.
JA908035G
9
J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010 4151