Isostable DNA

Article abstract of DOI:10.1021/ja074209p

The high fidelity detection of multiple DNA sequences in multiplex assays calls for duplexes whose stability is independent of sequence (isostable DNA), forming under universally stringent conditions. Nature did not evolve DNA to form isostable duplexes.

Full text of DOI:10.1021/ja074209p

Published on Web 11/14/2007
Isostable DNA
Carolin Ahlborn, Karsten Siegmund, and Clemens Richert*
Contribution from the Institut fu¨r Organische Chemie, UniVersita¨t Karlsruhe (TH), Karlsruhe,
Germany
Received June 22, 2007; E-mail: cr@rrg.uka.de
Abstract: The high fidelity detection of multiple DNA sequences in multiplex assays calls for duplexes
whose stability is independent of sequence (isostable DNA), forming under universally stringent conditions.
Nature did not evolve DNA to form isostable duplexes. Here we report how probe strands can be modified
so that an all-A/T target strand is bound with the same or slightly higher affinity than the corresponding
all-G/C strand with the same sequence of purines and pyrimidines. We refer to these probes that feature
covalently attached ligands as “decorated nucleic acids”. Caps, intercalators, and locks were used to stabilize
A/T duplexes, and N4-ethylcytosine residues were employed to tune down the stability of G/C duplexes
without significantly affecting base pairing selectivity. Near-isostability was demonstrated in solution and
on microarrays of high and low density. Further, it is shown that hybridization results involving decorated
probes on microarrays can be predicted on the basis of thermodynamic data for duplex formation in solution.
Predictable formation of isostable DNA not only benefits microarrays for gene expression analysis and
genotyping, but may also improve the sequence-specificity of other applications that rely on the massively
parallel formation of Watson-Crick duplexes.
according to the temperature of their environment9 by tuning
the G/C content of their genome. This would be impossible, if
Introduction
Nature evolved DNA that features two types of base pairs.
Guanine and cytosine form strong base pairs with three hydrogen
bonds and stabilizing secondary electrostatic interactions.1
Adenine and thymine form weak base pairs with two hydrogen
bonds and no favorable secondary interactions. The binding
constant is approximately 3 orders of magnitude larger for G:C
base pairs than for A:U base pairs.2 The differences in base-
pair stability serve important roles in cells. For example, A/T-
rich sequence motifs (“TATA boxes”) are found at the start
locus of genes, where strands have to be separated.3 Other
A/T-rich elements are sites for preferred recombination in
plants.4 Sequences that are rich in G and C are found in
regions of the genome critical for regulating gene
expression,5 where stable duplexes are advantageous. Other
motifs with high G/C content are inverted repeats6 and GC
islands.7 Patterns of AA/TT dinucleotides and GC dinucleotides
induce a curvature in DNA that favors nucleosome formation.8
Finally, some bacteria and archea adjust duplex stability
their genetic material was made up of base pairs with equal
stability.
While beneficial for its natural roles, the dependence of
duplex stability on G/C content can be problematic for biomedi-
cal applications of DNA. Such applications include use as
hybridization probe, primer, sequencing probe,10 antisense11 or
antigene agent,12 and as component for nanotechnology and
material sciences.13-15 For many uses of DNA, sequence
selectivity is critical. Perhaps most demanding is the massively
parallel detection of target strands on DNA microarrays. On
current high-density microarrays, >105 different duplexes have
to form simultaneously.16 The growth in total spot number and
density on microarrays continues unabated, and more demanding
applications are being tackled with microarrays,17,18 including
SNP genotyping,19-21 which requires discrimination between
target strands differing in a single nucleotide.
(9) Nelson, K. E.; Paulsen, I. T.; Heidelberg, J. F.; Fraser, C. M. Nat.
Biotechnol. 2000, 18, 1049-1054.
(10) Shendure, J.; Porreca, G. J.; Reppas, N. B.; Lin, X.; McCutheon, J. P.;
Rosenbaum, A. M.; Wang, M. D.; Zhang, K.; Mitra, R. D.; Church, G. M.
Science 2005, 309, 1728-1732.
(1) Pranata, J.; Wierschke, S. G.; Jorgensen, W. L. J. Am. Chem. Soc. 1991,
113, 2810-2819.
(2) Kyogoku, Y.; Lord, R. C.; Rich, A. Biochim. Biophys. Acta 1969, 179,
10-17.
(11) Uhlmann, E.; Peyman, A. Chem. ReV. 1990, 90, 543-584.
(12) Praseuth, D.; Guieysse, A. L.; He´le`ne, C. Biochim. Biophys. Acta 1999,
1489, 181-206.
(3) Voet, D.; Voet, J. G. Biochemistry, 2nd ed.; J. Wiley & Sons, New York,
1995.
(13) Park, S. H.; Yin, P.; Liu, Y.; Reif, J. H.; LaBean, T. H.; Yan, H. Nano
Lett. 2005, 5, 729-733.
(4) Mu¨ller, A. E.; Kamisugi, Y.; Gru¨neberg, R.; Niedenhof, I.; Ho¨rold, R. J.;
Meyer, P. J. Mol. Biol. 1999, 291, 29-46.
(14) Rothemund, P. W. K. Nature 2006, 440, 297-302.
(15) Feldkamp, U.; Niemeyer, C. M. Angew. Chem., Int. Ed. 2006, 45, 1856-
1876.
(5) Hapgood, J. P.; Riedemann, J.; Scherer, S. D. Cell Biol. Intern. 2001, 25,
17-31.
(6) Ruskin, B.; Fink, G. R. Genetics 1993, 133, 43-56.
(7) Welm, A. L.; Mackey, S. L.; Timchemko, L. T.; Darlington, G. J.;
Timchenko, N. A. J. Biol. Chem. 2000, 275, 27406-27413.
(8) Istvan, A.; Marvich, T. N.; Tomsho, L. P.; Qi, J.; Zanton, S. J.; Schuster,
S. C.; Pugh, B. F. Nature 2007, 446, 572-576.
(16) Pennisi, E. Science 2003, 302, 211.
(17) Waldman, M. Nature 2006, 444, 256.
(18) Schumacher, A.; Kapranov, P.; Kaminsky, Z.; Flanagan, J.; Assadzadeh,
A.; Yau, P.; Virtanen, C.; Winegarden, N.; Cheng, J.; Gingeras, T.; Petronis,
A. Nucleic Acids Res. 2006, 34, 528-542.
9
15218
J. AM. CHEM. SOC. 2007, 129, 15218-15232
10.1021/ja074209p CCC: $37.00 © 2007 American Chemical Society

Isostable DNA
A R T I C L E S
Even as “enabling technique”, DNA microarray assays suffer
from challenges, such as reproducibility and calibration,22-25
as highlighted in comparative studies.26-28 Perhaps the most
important challenge is low fidelity caused by cross-hybridization,
i.e., binding of target strands to probes that are not fully
complementary.29 This phenomenon causes false positive signals
and/or exaggerated signals at spots with related sequences.30
Cross-hybridization also draws target strands away from the spot
with the fully complementary probes, lowering signal intensity
at this spot. Minimizing false positive results by using more
stringent hybridization and washing conditions necessarily
increases the likelihood of false negative results for duplexes
of low stability. The different stability of A:T and G:C base
pairs is thus the molecular basis of a key factor that limits the
fidelity of massively parallel detection.
To counter the effect of base pairing strength, probe design
for microarrays usually includes the selection of sequences with
an appropriate G/C content, using computer algorithms,31 but
this narrows the accessible sequence space, particularly for
members of gene families formed through gene duplication. In
a confined sequence space, cross-hybridization becomes more
likely. On some commercial microarrays, control sequences
differing by a single nucleotide are employed, but using the
signal they produce for background subtraction is known to be
problematic.32 Single mismatches cause different drops in free
energy of binding, depending on sequence context and position
in a duplex.33,34 That the physical limits of selective molecular
recognition are readily reached can be demonstrated in simula-
tions. The spread of melting points between probe/target
duplexes can be >20 °C for a commercial DNA microarray,35
corroborating concerns that melting points will necessarily differ
over a wide range in mixtures with sequences of varying G/C-
content.36
Figure 1. Cartoons of binding curves for probes containing a high G/C
content or a high A/T content on (a) a conventional and (b) a chemically
optimized DNA microarray. Curves for perfectly matched duplexes (PM,
solid lines) and those with a single mismatch (MM, broken lines) are shown.
An intermediate hybridization temperature in panel a leads to a false negative
result for the A/T-rich sequence and to a lack of base pairing fidelity for
the G/C-rich sequence. For panel b, with modified residues (A*, C*, G*,
and T*) PM targets of any sequence are bound with similar affinity, and
any mismatched base in a target leads to a drop in affinity.
e20 °C drop in melting point caused by a single mismatch.
This is true even when the mismatch is at the center of the
sequence. No temperature exists at which both sequences bind
their targets single base selectively, and at an intermediate
temperature neither the all-G/C sequence nor the all-A/T
sequence shows single-base specific binding (Figure 1a).
Given the limitations of natural DNA, it is desirable to adjust
duplex stability by chemically modifying probes, rather than
confining oneself in the available sequence space. Oligonucle-
otides that give duplex stabilities independent of sequence
(isostable DNA) can be expected to be superior hybridization
probes in multiplexing experiments. With binding curves of the
kind shown in Figure 1b, at an optimized temperature, all
perfectly matched targets will be bound, but mismatched targets
will remain essentially unbound for all spots of a microarray.
The “ideal chip” scenario also requires accurate prediction of
binding curves, so that conditions that ensure maximum fidelity
can be chosen without iteration.37
The extent to which detection of target strands can be affected
by base composition may be demonstrated for a specific
sequence of purines and pyrimidines, such as 5′-TAATTTT-
TAATAAT-3′ and 5′-CGGCCCCCGGCGGC-3′, whose du-
plexes melt more than 40 °C apart from each other (vide infra).
This difference in melting points is much greater than the
(19) Kennedy, G. C.; et al. Nat. Biotechnol. 2003, 21, 1233-1237.
(20) Matsuzaki, H.; et al. Genome Res. 2004, 14, 414-425.
(21) Gresham, D.; Ruderfer, D. M; Pratt, S. C.; Schacherer, J.; Dunham, M. J.;
Botstein, D.; Kruglyak, L. Science 2006, 311, 1932-1936.
(22) Lee, M. L. T.; Kuo, F. C.; Whitmore, G. A.; Sklar, J. Proc. Natl. Acad.
Sci. U.S.A. 2000, 97, 9834-9839.
(23) Jenssen, T. K.; Langaas, M.; Kuo, W. P.; Sørensen, B. S.; Myklebost, O.;
Hovig, E. Nucleic Acids Res. 2002, 30, 3235-3244.
(24) Kothapalli, R.; Yoder, S. J.; Mane, S.; Loughran, Jr., T. P. BMC
Bioinformatics 2002, 3, 22.
(25) Draghici, S.; Khatri, P.; Eklund, A. C.; Szallasi, Z. Trends Genet. 2006,
22, 101-109.
(26) Iacobuzio-Donahue, C. A.; Ashafq, R.; Maitra, A.; Adsay, N. V.; Sheng-
Ong, G. L.; Berg, K.; Hollingsworth, M. A.; Cameron, J. L.; Yeo, C. J.;
Kern, S. E.; Goggins, M.; Hurban, R. H. Cancer Res. 2003, 63, 8614-
8622.
Despite numerous biological studies involving DNA microar-
rays,38 much of the chemical research on microarrays has
focused on the linkers to the surface and protecting groups for
photolithographic syntheses.39,40 Few reports on chemically
adjusting the target affinity of DNA probes to isostability can
be found in the literature.41,42 To the best of our knowledge,
truly isostable DNA has not yet been demonstrated, nor has
(27) Tan, P. K.; Downey, T. J.; Spitznagel, E. L.; Xu, P.; Fu, D.; Dimitrov, D.
S.; Lempicki, R. A.; Raaka, B. M.; Cam, M. C. Nucleic Acids Res. 2003,
31, 5676-5684.
(28) Cao, W.; et al. BMC Genomics 2004, 5, 26.
(29) Marshall, E. Science 2004, 306, 630-631.
(30) Plutowski, U.; Richert, C. Angew. Chem., Int. Ed. 2005, 44, 621-625.
(31) Li, X. Y.; He, Z. L.; Zhou, J. Z. Nucleic Acids Res. 2005, 33, 6114-6123
and references therein.
(32) Irizarry, R. A.; Bolstad, B. M.; Collin, F.; Cope, L. M.; Hobbs, B.; Speed,
T. P. Nucleic Acids Res. 2003, 31, e15.
(37) Urakawa, H.; El Fantroussi, S.; Smidt, H.; Smoot, J. C.; Tribou, E. H.;
Kelly, J. J.; Noble, P. A.;. Stahl D. A. Appl. EnViron. Microbiol. 2003, 69,
2848-2856.
(33) Hong, B. J.; Sunkara, V.; Park, J. W. Nucleic Acids Res. 2005, 33, e106.
(34) Zhang, L.; Wu, C.; Carta, R.; Zhao, H. Nucleic Acids Res. 2007, 35, e18.
(35) Siegmund, K.; Ahlborn, C.; Richert, C., submitted for publication. The
program ChipCheckII can be accessed at http://chipcheck.chemie.uni-
karlsruhe.de/ and can be made available for download upon request.
(36) Southern, E.; Mir, K.; Shchepinov, M. Nat. Genet. 1999, 21, 5-9.
(38) Dalma-Weiszhausz, D. D.; Warrington, J.; Tanimoto, E. Y.; Miyada, G.
Methods Enzym. 2006, 410, 3-28, and references therein.
(39) Niemeyer, C. M.; Blohm, D. Angew. Chem., Int. Ed. 1999, 38, 2865-
2869.
(40) Pirrung, M. C. Angew. Chem., Int. Ed. 2002, 41, 1277-1289.
9
J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007 15219

A R T I C L E S
Ahlborn et al.
such DNA been employed on microarrays. Here we show how
duplex stability can be adjusted such that an all-A/T
tetradecamer sequence is bound with the same or slightly higher
affinity than its all-G/C counterpart, leading to signals of similar
intensity on DNA microarrays.
tures.55,56 We term such ligand-enforced oligonucleotides “deco-
rated nucleic acids.” They have the advantage of being fully
compatible with automated DNA synthesis, a feature deemed
critical for application in industrial microarray fabrication. This
ensures that substituents may be strategically placed anywhere
in a probe sequence via established methods possibly including
photolithographic syntheses.57,39 The decorated DNA can be
expected to engage in more extensive interactions with the target
structure than unmodified DNA, interrogating its structure (and
thus its sequence) more thoroughly.
Target strands without modifications (other than fluorophore
labeling at the terminus) were to be employed in our study to
ensure that DNA from biological sources or amplified via
inexpensive PCR with unmodified dNTPs may be detected. To
generate A/T-rich duplexes with sufficiently high thermal
stability, we combined duplex-stabilizing modifications in the
interior with “molecular caps”58 at either terminus of the
hybridization probes. Caps are non-nucleosidic substituents that
increase duplex stability and base-pairing fidelity at the termini
by bridging canonical terminal base pairs much better than
mismatched base pairs.59,60,61 These were to be combined with
LNA residues, introduced via commercially available phos-
phoramidite building blocks.62
When selecting replacements for thymidine residues, we drew
on the results of our recent combinatorial search on duplex-
stabilizing substituents protruding into the major groove of
duplex DNA.63 A pyrenylbutyramidopropargyl ligand attached
to the 5-position of 2′-deoxyuridine residues had been identified
as the substituent giving the largest melting point increase. Thus
far, the deoxyuridine residues carrying the substituent had to
be coupled to fully assembled oligonucleotides on solid support
via cross coupling of 2′-deoxy-5-iodouridine residues.63 This
procedure was unattractive for regular use in microarray
syntheses. We therefore synthesized phosphoramidite 1 (Scheme
1), which allows for chain assembly via automated DNA
synthesis without any unconventional steps. The synthesis of 1
started with coupling of pyrenylbutyric acid (2) to propargy-
lamine (3). Amide 4 was then used in a Sonogashira coupling
with 5′-protected 2′-deoxy-5-iodouridine 5. The resulting alkynyl
nucleoside 6 was then phosphitylated to give 1 in 66% overall
yield.
Results
In the first phase of our study, we focused on enhancing the
stability of T:A base pairs. Several approaches for stabilizing
A:T and T:A base pairs are known from the literature. An
additional hydrogen bond can be introduced in these base pairs
by employing a diaminopurine derivative as replacement for
adenine, as demonstrated by Seela and co-workers.43 A similar
effect is difficult to realize for thymidine, as adenines in the
target strands do not offer a third hydrogen-bonding functionality
on their Watson-Crick base-pairing face. Backbone-modified
oligonucleotides are known to increase duplex stability, includ-
ing peptide nucleic acids (PNA),44,45 but these are difficult to
combine freely with unmodified DNA. Locked nucleic acids
(LNA)46 do not have this disadvantage47 and have been
recommended for gene expression profiling applications.48 Other
DNA backbone modifications can give duplexes with increased
thermal stability, but are less readily incorporated during
conventional synthesis cycles and are not readily available
commercially. These include N3′-P5′-phosphoramidates,49 guani-
dinium nucleic acids (DNG),50 and 1′,5′-anhydrohexitol nucleic
acids.51 For RNA as target strand, data on backbone-modified
oligonucleotide derivatives that form duplexes of increased
stability are available, mostly from studies on antisense inhibi-
tion of gene expression,11,52,53,54 but it is not clear how easily
they can be applied to microarrays.
We selected DNA derivatives that retain the entire framework
of natural DNA and are solely stabilizing through additional
interactions mediated by substituents (rather than replacements
of the DNA backbone or nucleobases) to avoid the high
propensity of non-ionic analogues to form alternative struc-
(41) (a) Nguyen, H.-K.; Fuornier, O.; Assaline, U.; Dupret, D.; Thuong, N. T.
Nucleic Acids Res. 1999, 27, 1492-1498. (b) Nguyen, H.-K.; Bonfils, E.;
Auffray, P.; Costaglioli, P.; Schmitt, P.; Asseline, U.; Durand, M.; Maurizot,
J.-C.; Dupret, D.; Thuong, N.T. Nucleic Acids Res. 1998, 26, 4249-4258.
(c) Nguyen, H.-K.; Auffray. P.; Asseline, U.; Dupret, D.; Thuong, N. T.
Nucleic Acids Res. 1997, 25, 3059-3065.
(42) Fidanza, J. A.; McGall, G. H. Nucleosides Nucleotides Nucleic Acids 1999,
18, 1293-1295.
We then tested to what extent the pyrenyldeoxyuridine
residues act synergistically with molecular caps (Scheme 2).
We chose a trimethoxystilbene cap for the 5′-terminus,
readily introduced at the end of DNA syntheses via phosphora-
midite 7,61 which is commercially available. The three-
dimensional structure of a duplex capped by this 5′-substituent
is known.64 Stilbenes had previously been shown to be excellent
(43) (a) He, J.; Seela, F. Nucleic Acids Res. 2002, 30, 5485-5496. (b) Seela,
F.; Becher, G. Nucleic Acids Res. 2001, 29, 2069-2078. (c) Seela, F.;
Zulauf, M. HelV. Chim. Acta 1999, 82, 1878-1898.
(44) Nielsen, P. E. Annu. ReV. Biophys. Biomol. Struct. 1995, 24, 167-183.
(45) Uhlmann, E.; Peyman, A.; Breipohl, G.; Will, D. W. Angew. Chem. Int
Ed. 1998, 37, 2797-2823.
(46) (a) Christensen, N. K.; Petersen, M.; Nielsen, P.; Jacobsen, J. P.; Olsen, C.
E.; Wengel, J. J. Am. Chem. Soc. 1998, 120, 5458-5463. (b) Koshkin,
A.A.; Singh, S. K.; Nielsen, P.; Rajwanshi, V. K.; Kumar, R.; Meldgaard,
M.; Olsen, C. E.; Wengel, J. Tetrahedron 1998, 54, 3607-3630. (c) Obika,
S.; Nanbu, D.; Hari, Y.; Andoh, J.I.; Morio, K.I.; Doi, T.; Imanishi, T.
Tetrahedron Lett. 1998, 39, 5401-5404.
(55) Richert, C.; Roughton, A. L.; Benner, S. A. J. Am. Chem. Soc. 1996, 118,
4518-4531.
(47) Nielsen, K. E.; Singh, S. K.; Wengel, J.; Jacobsen, J. P. Bioconjugate Chem.
2000, 11, 228-238.
(48) Tolstrup, N.; Nielsen, P. S.; Kolberg, J. G.; Frankel, A. M.; Vissing, H.;
Kauppinen, S. Nucleic Acids Res. 2003, 31, 3758-3762.
(56) Benner, S. A.; Hutter, D. Bioorg. Chem. 2002, 30, 62-80.
(57) (a) Fodor, S. P. A.; Read, J. L.; Pirrung, M. C.; Stryer, L.; Lu, A. T.; Solas,
D. Science 1991, 251, 767-773. (b) Pirrung, M. C. Chem. ReV. 1997, 97,
473-488. (c) Gao, X. L.; Gulari, E.; Zhou, X. C. Biopolymers 2004, 73,
579-596.
(49) (a) Gryaznov, S. M.; Chen, J.-K. J. Am. Chem. Soc. 1994, 116, 3143-
3144. (b) Chen, J.-K.; Schultz, R. G.; Lloyd, D. H.; Gryaznov, S. M. Nucleic
Acids Res. 1995, 23, 2661-2668. (c) Gryaznov, S. M.; Lloyd, D. H.; Chen,
J.-H.; Schultz, R. G.; DeDionisio, L. A.; Ratmeyer, L.; Wilson, W. D. Proc.
Natl. Acad. Sci. U.S.A. 1995, 92, 5798-5802.
(58) Richert, C.; Gru¨nefeld, P. Synlett 2007, 1-18.
(59) Bleczinski, C. F.; Richert, C. J. Am. Chem. Soc. 1999, 121, 10889-10894.
(60) Narayanan, S.; Gall, J.; Richert, C. Nucleic Acids Res. 2004, 32, 2901-
2911.
(50) Linkletter, B. A.; Szabo, I. E.; Bruice, T. C. J. Am. Chem. Soc. 1999, 121,
3888-3896.
(61) Dogan, Z.; Paulini, R.; Rojas Stu¨tz, J. A.; Narayanan, S.; Richert, C. J.
Am. Chem. Soc. 2004, 126, 4762-4763.
(51) Hendrix, C.; Rosemeyer, H.; De Bouvere, B.; Van Aerschot, A.; Seela, F.;
Herdewijn, P. Chem. Eur. J. 1997, 3, 110-120.
(62) For reviews on LNAs, see: (a) Wengel, J. Acc. Chem. Res. 1999, 32, 301-
310. (b) Jepsen, J. S.; Wengel, J. Curr. Opin. Drug DiscoV. DeV. 2004, 7,
188-194.
(52) Freier, S. M.; Altmann, K. H. Nucleic Acids Res. 1997, 25, 4429-4443.
(53) Manoharan, M. Biochim. Biophys. Acta 1999, 1489, 117-130.
(54) Summerton, J. Biochim. Biophys. Acta 1999, 1489, 141-158.
(63) Kottysch, T.; Ahlborn, C.; Brotzel, F.; Richert, C. Chem. Eur. J. 2004, 10,
4017-4028.
9
15220 J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007

Isostable DNA
A R T I C L E S
Scheme 1. Synthesis of Phosphoramidite 1a
a Conditions: (a) HOBt, EDC, DMF, 76%; (b) Pd(PPh3)4, CuI, NEt3, DMF, 92%; (c) NCC2H4O-P(NiPr2)Cl, DIEA, CH3CN, 95%.
Scheme 2. Synthesis of “Highly Decorated”, Lysine-Terminated Oligonucleotide AT2CapPyLNA (12)a
a Controlled pore glass 967 was prepared as reported: (a) 1. 12-trityloxylauric acid, HBTU, HOBt, DIPEA, DMF, 2. TFA/CH2Cl2 (1:1); (b) 1. Boc-Lys(TFA)-OH, HBTU,
HOBt, DIPEA, DMF, 2. TFA/CH2Cl2; (c) 1. 2,2-dimethyl-3-trityloxypropionic acid, HBTU, HOBt, DIPEA, DMF, 2. TFA/CH2Cl2; (d) tetrazole, CH3CN, followed by oxidation
and capping; (e) conventional DNA synthesis, including 1, 7, 10, and 11; (f) NH4OH.
bridges for termini of duplexes.65 Oligonucleotides with the
trimethoxystilbene cap show excellent mismatch discrimination
at the terminus, both in solution and on microarrays.61 A cholic
acid residue amide-linked to the 5′-terminus59 was also tested
9
J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007 15221

A R T I C L E S
Ahlborn et al.
Figure 2. Oligonucleotides binding an all-A/T target sequence with an increasing number of duplex-stabilizing ligands or LNA residues. The TL and AL
in AT2CapPyLNA (12) denote the LNA analogues of thymidine and deoxyadenosine residues, respectively; DP-Lys denotes the dimethylpropionic acid-
lysine linker. See AT2CapPyLNA (12) in Scheme 2 for the structure of the 3′-terminus.
and gave similar results, both in solution and on microarrays,
but was not pursued further, as it requires a 5′-terminal 5′-amino-
2′,5′-dideoxynucleotide and amide-forming steps during solid-
phase syntheses. A pyrenylmethylpyrrolidinol 5′-cap60 produced
oligonucleotides that were difficult to purify to homogeneity
when additional lipophilic residues were incorporated.
For the 3′-terminus, a 2′-attached anthraquinonecarboxamido
group (AQ) was chosen, which has recently been shown to
increase duplex stability on DNA-microarrays.66 This substituent
is also known to induce satisfactory mismatch discrimination
at the terminus.60 The synthesis of the phosphoramidite 8 and
Figure 3. Oligonucleotides LNACap (19), LNACapcon (20), and com-
pLNACap (21) used in melting curve experiments (compare Table 1).
(64) Tuma, J.; Paulini, R.; Rojas Stu¨tz, J. A.; Richert, C. Biochemistry 2004,
43, 15680-15687.
(65) Lewis, F. D.; Liu, X.; Wu, Y.; Miller, S. E.; Wasielewski, M. R.; Letsinger,
its use as reagent for generating oligonucleotides with a linker
R. L.; Sanishivili, R.; Joachimiak, A.; Tereshko, V.; Egli, M. J. Am. Chem.
Soc. 1999, 121, 9905-9906.
(66) Al-Rawi, S.; Ahlborn, C.; Richert, C. Org. Lett. 2005, 7, 1569-1572.
(67) Schwope, I.; Bleczinski, C. F.; Richert, C. J. Org. Chem. 1999, 64, 4749-
4761.
for 3′-immobilization on aldehyde-bearing glass surfaces has
been reported.66 The lysine-terminated linker67 has been used
for microarrays previously.68,61
(68) Dombi, K.; Griesang, N.; Richert, C. Synthesis 2002, 816-824.
(69) For the oligonucleotides, a short-hand naming system is being used in
addition to the numbering to make the text easier-to-read. For example,
AT, ATCap, AT2Cap, and AT2CapPy denote oligonucleotides of the
all-A/T series without substituents, with one cap, two caps, and two caps
plus pyrenyl substituents, respectively. Names starting with GC refer to
compounds from the all-G/C series. The prefix comp stands for the
complementary strand, and the endings con, Cy, and LNA refer to control
compounds, cyanine dye-labeled compounds, and LNA-containing com-
pounds, respectively. In the GC series, Et and I indicate the presence of
ethylcytosine and inosine residues in the oligonucleotide with numbers
indicating how many are found in the sequence. Strands with a mismatched
nucleobase contain an ending highlighting this nucleobase, e.g., compGC_A
for a complementary strand with a mismatched adenine residue. See the
Supporting Information for a separate listing of the short-hand convention.
(70) Koshkin, A. A.; Singh, S. K.; Nielsen, P.; Rajwanshi, V. K.; Kumar, R.;
Meldgaard, M.; Olsen, C. E.; Wengel, J. Tetrahedron, 1998, 54, 3607-
3630.
Starting from linker-bearing controlled pore glass 9 (Scheme
2) the synthesis of oligonucleotide AT2CapPyLNA (12) was
performed, using phosphoramidites 1, 7, 8 and LNA-phosphora-
midites 10 and 11, as well as the phosphoramidites of unmodi-
fied nucleosides (Scheme 2).69 A series of oligonucleotides
featuring only subsets of the duplex-stabilizing ligands [AT (13),
ATCap (14), AT2Cap (15), and AT2CapPy (16)] was also
prepared (Figure 2), using the analogous strategy but fewer
nonstandard phosphoramidites. Further, the DNA target se-
quence fully complementary to 12-16, namely compAT, 5′-
ATTATTAAAAATTA-3′ (17) was prepared, together with its
Cy-labeled analogue compATCy, 5′-Cy3-ATTATTAAAAAT-
TA-3′ (18).
(71) (a) Kaur, H.; Arora, A.; Wengel, J.; Maiti, S. Biochemistry 2006, 45, 7347-
7355. (b) You, Y.; Moreira, B. G.; Behlke, M. A.; Owczarzy, R. Nucleic
Acids Res. 2006, 34, e60.
(72) Reich, N. O.; Sweetnam, K. R. Nucleic Acids Res. 1994, 22, 2089-2093.
9
15222 J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007

Isostable DNA
A R T I C L E S
Figure 4. Oligonucleotides prepared with an all-G/C-sequence and different numbers of substituents. In the sequence, CEt is indicating N-4-ethylcytidine
and I indicates inosine residues.
Table 1. UV-Melting Points of Duplexes of Oligonucleotide
LNACap (19) or LNACapcon (20) with Their Complementary
Strand and Mismatch-Containing Derivatives of compLNACap
(21a-t)
Table 2. UV-Melting Points of Duplexes of GC (22) or GC3Et (23)
and Target Strands with or without a Mismatched Nucleobase
Opposite to or Next to a Modified Cytosine Residue
Duplexa:
5
′
-CGGNC CNC GGNGGC-DP-Lys3
′
GC (22) or GC3Et (23)
3
′
-GCCGGB1B2B3CCGCCG-5
′ compGC (27, 27a-e)
Duplexa:
5
′
-YGGTTGAC-3 LNACap (19) or LNACapcon (20)
′
3
′
-BCCAACTG-5′ compLNACap (21a-t)
150 mM NaCl
1 M NaCl
Tm
C)d
150 mM NH4OAc
Tm
C)d
1 M NH4OAc
Tm
C)d
Tm
C)c
∆
Tm
hyperchr
(%)e
Tm
C)c
∆
(°
hyperchr
(%)e
Nb
B1B2B3
(
°
(°
C)d
(
°
∆
hyperchr
(%)e
Tm
C)c
∆
(°
hyperchr
(%)e
Yb
B
Tm
(
°
C)c
(
°
(
°
C
GGG 72.4 ( 0.4
-
14
12
11
9
12
10
15
14
12
11
11
10
75.8 ( 0.4
15
13
12
8
13
10
18
14
12
11
11
11
A
A
A
A
T
27.3 ( 0.4
20
17
19
23
18
17
17
20
35.7 ( 0.3
21
18
19
24
18
17
18
20
C
C
C
C
C
GAG 58.3 ( 0.8 -14.1
GCG 54.9 ( 0.4 -17.5
GTG 57.2 ( 0.7 -15.2
GGT 53.5 ( 0.8 -18.9
TGG 56.7 ( 0.5 -15.7
63.3 ( 0.4 -12.5
61.5 ( 0.3 -14.3
62.2 ( 0.2 -13.6
59.4 ( 0.4 -16.4
61.0 ( 0.1 -14.8
66.9 ( 0.4
C
A
G
T
C
A
G
26.0 ( 0.6 -1.3
26.8 ( 0.7 -0.5
34.4 ( 0.5 -1.3
34.9 ( 0.7 -0.8
28.0 ( 0.6
28.7 ( 0.5
24.3 ( 0.7 -4.4
24.8 ( 0.4 -3.9
25.7 ( 0.6 -3.0
0.7
36.5 ( 0.7
37.1 ( 0.4
32.9 ( 0.6 -4.2
33.3 ( 0.5 -3.8
33.7 ( 0.5 -3.4
0.8
AL
AL
AL
AL
CEt GGG 63.1 ( 0.7
-
CEt GAG 51.9 ( 0.9 -11.2
CEt GCG 47.6 ( 1.3 -15.5
CEt GTG 48.9 ( 0.8 -14.2
CEt GGT 45.1 ( 0.8 -18.0
CEt TGG 46.8 ( 1.0 -16.3
55.3 ( 0.3 -11.6
53.7 ( 0.5 -13.2
54.0 ( 0.5 -12.9
49.7 ( 0.9 -17.2
51.3 ( 0.7 -15.6
a Conditions: 1.5 µM strand concentration for each oligonucleotide,
NH4OAc-buffer, pH 7.0. b AL denotes LNA A residue. c Average of four
melting points ( standard deviation. d Difference of melting points accord-
ing to fully complementary duplex. e Average of hyperchromicities (hy-
perchr) from 4 curves, measured at 260 nm.
a Conditions: 1 µM strand concentration (each), salt concentration 150
mM NaCl, 10 mM sodium phosphate buffer (pH7) and 150 mM NaCl; 1
M NaCl, 10 mM sodium phosphate buffer (pH7) and 1 M NaCl. b CEt
denotes N-4-ethylcytidine c Average of four melting points ( SD. d ∆Tm
) melting point difference with respect to the perfectly matched duplex.
e Average of hyperchromicities (hyperchr) at 260 nm upon duplex dissocia-
tion from 4 curves.
Experiments were then performed to evaluate whether a
molecular cap provides better base pairing fidelity at the
terminus than an LNA residue. The results from a series of
melting curve experiments with oligonucleotide LNACap (19)
featuring a 5′-terminal LNA-nucleoside or unmodified control
LNACapcon (20) with target strands compLNACap (21a-t)
featuring a matched or mismatched nucleobase at their 3′-
terminal position (Figure 3) showed very modest decreases in
melting point for the mismatches. The same terminal mismatches
in duplexes of 5′-TMS-BGGTTGAC-3′, where TMS is a
trimethoxystilbene cap and B is A, C, G, or T give melting
point depressions of between 6.5 and 23.4 °C, justifying the
use of caps, rather than using LNA only. Further, the melting
point increase for the fully complementary duplex is much
greater for the TMS cap (up to +12.2 °C for an octamer)61 than
for a terminal LNA residue (Table 1). As a consequence, LNA
nucleosides were only used in the interior of the sequence. For
the pyrenyl-substituent at the 5-position of dU residues, satisfac-
tory mismatch discrimination had been demonstrated previ-
ously,63 and the same is true for the 3′-anthraquinone substit-
uent.60 For LNA oligonucleotides where the LNA residues are
in the interior of the sequence, data on base pairing selectivity70
and stabilizing effects are known from the literature.71 These
9
J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007 15223

A R T I C L E S
Ahlborn et al.
Table 3. UV-Melting Points of Duplexes of GC (22) (control) or
GC3I (24) and Target Strands with or without a Mismatched
Nucleobase Directly Facing the Deoxyinosine Residue or Next to It
AT2CapPyLNA (12), AT (13), ATCap (14), AT2Cap (15),
and AT2CapPy (16). Inosine can form only two hydrogen
bonds, compared to three for guanine, and the ethyl chain at
the N4-position reduces base-pair stability based on an entropic
effect (only one of the two conformers is capable of pairing
with G). The synthesis of compounds GC (22), GC3Et (23),
GC3I (24), GC3EtCI (25), and GC8Et (26) started from linker-
bearing cpg 9 (Scheme 2), so that strands suitable for im-
mobilization on microarray surfaces were obtained. The fully
complementary target strand compGC 5′-GCCGCCGGGGGC-
CG-3′ (27), its fluorophore-labeled counterpart compGCCy 5′-
Cy3-GCCGCCGGGGGCCG-3′ (28), and mismatch-containing
derivatives (27a-i) were also prepared via conventional chain-
assembly strategies.
Duplexa:
5
′
-CNGCCCC CN G CGNC
−
DP-Lys-3
′ GC (22) or GC3I (24)
3
′
-GCCGGGG B3B4B5GCCG-5
′
compGC (27, 27e-i)
150 mM NaCl
Tm
C)d
1 M NaCl
Tm
C)d
Tm
C)c
∆
hyperchr
(%)e
Tm
C)c
∆
(°
hyperchr
(%)e
Nb B3B4B5
(
°
(
°
(
°
G
G
G
G
G
G
I
I
I
I
I
GCC 72.4 ( 0.4
GAC 63.5 ( 0.3
14
14
14
14
14
11
13
15
15
15
14
10
75.8 ( 0.4
66.9 ( 0.3
15
14
13
14
15
13
16
15
16
16
15
12
-8.9
-8.9
GGC 61.9 ( 0.2 -10.5
65.1 ( 0.3 -10.7
GTC 62.9 ( 0.6
GCA 66.3 ( 0.5
TCC 54.2 ( 0.4 -18.2
GCC 59.5 ( 0.8
-9.5
-6.1
66.8 ( 0.3
70.3 ( 0.4
-9.0
-5.5
59.2 ( 0.7 -16.6
64.0 ( 0.3
GAC 58.9 ( 0.3
GGC 53.2 ( 0.2
GTC 54.4 ( 0.4
GCA 53.8 ( 0.4
-0.6
-6.3
-5.1
-5.7
64.3 ( 0.3 + 0.3
57.0 ( 0.3
58.8 ( 0.3
58.6 ( 0.5
-7.0
-5.2
-5.4
To test the effect of the base modification on base pairing
fidelity, melting curves with complementary strands containing
a mismatched nucleobase directly opposite the inosine/N4-
ethylcytosine or immediately next to the base pair it forms
were performed. The results are compiled in Tables 2 and 3.
It can be discerned that an N4-ethylcytosine residue retains
the base pairing specificity of cytosine, with fairly similar
drops in melting point for mismatch-containing duplexes of at
least 11.2 °C. The same cannot be said of inosine as replacement
for guanine. Particularly, a mismatched adenine facing the
inosine in duplex, GC3I:compGC_A (24 : 27f) leads to an
unusually stable duplex, whose melting point rivals that of the
perfectly matched counterpart. As a consequence, oligonucleo-
tides GC3I (24) and GC3EtCI (25) (Figure 4) were not used
for further experiments, even though the latter had shown the
required drop in melting point for the duplex with its comple-
mentary target strand (the Tm of duplex GC3EtCI:compGC
(25:27) is 47.8 ( 0.8 °C at 150 mM NaCl and 52.9 ( 0.6 °C
at 1 M NaCl).
I
TCC 38.3 ( 0.6 -21.2
45.0 ( 0.7 -19.0
a Conditions: 1 µM strand concentration, salt concentration 150 mM
NaCl, 10 mM sodium phosphate buffer (pH7) and 150 mM NaCl; 1 M
NaCl, 10 mM sodium phosphate buffer (pH7) and 1 M NaCl. b I denotes
deoxyinosine residue. c Average of four melting points ( SD. d ∆Tm
)
melting point difference with respect to the perfectly matched duplex.
e Average of hyperchromicities (hyperchr) at 260 nm upon duplex dissocia-
tion from 4 curves.
are generally more favorable than those of the corresponding
values for unmodified DNA.
Next, approaches for reducing the thermal stability of the all-
G/C counterpart of AT (13) without loss of base-pairing fidelity
was tested. For this, the tetradecamers shown in Figure 4 were
synthesized and employed in UV-melting experiments. They
include control compound GC (22) with unmodified nucleobases
and sequences with deoxyinosine72 and 2′-deoxy-N4-ethylcy-
tidine residues41 replacing deoxyguanosine and deoxycytidine
residues, respectively. The sequences GC (22), GC3Et (23),
GC3I (24), GC3EtCI (25), and GC8Et (26) feature the same
order of purines and pyrimidines as in the all-A/T sequences
The next focus of our study was on measuring the extent to
which the modifications introduced to the all-A/T sequence
motif and the all-G/C sequence motif produced isostable DNA.
Table 4. UV-Melting Points of Duplexes of AT (13), ATCap (14), AT2Cap (15), AT2CapPy (16), AT2CapPyLNA (12) and GC (22), GC3Et
(23), GC8Et (26) with Fully Complementary Target Strands
salt
Tm
C)c
∆
Tm
hyperchr
(%)e
oligonucleotide sequencea,b
target stranda,b
(
°
(°
C)d
150 mM NaCl
TAATTTTTAATAAT (13)
ATTATTAAAAATTA (17)
31.4 ( 0.4
37.4 ( 0.5
41.9 ( 0.4
44.9 ( 0.3
54.9 ( 0.3
72.4 ( 0.4
63.1 ( 0.7
48.3 ( 0.4
35
36
33
26
26
14
15
15
TMS-TAATTTTTAATAAT (14)
TMS-TAATTTTTAATAAUAQ (15)
TMS-TAATTTUPyTAAUPyAAUAQ (16)
TMS-TAAUPyTLTLUPyTLALALUPyAAUAQ (12)
CGGCCCCCGGCGGC (22)
6.0
10.5
13.5
23.5
GCCGCCGGGGGCCG (27)
ATTATTAAAAATTA (17)
CGGCEtCCCEtCGGCEtGGC (23)
CEtGGCEtCEtCEtCEtCEtGGCEtGGCEt (26)
-9.3
-24.1
1 M NaCl
TAATTTTTAATAAT (13)
40.7 ( 0.6
46.1 ( 0.5
50.3 ( 0.5
51.6 ( 0.3
56.7 ( 0.3
75.8 ( 0.4
66.9 ( 0.4
48.4 ( 0.5
36
36
32
25
26
15
18
14
TMS-TAATTTTTAATAAT (14)
TMS-TAATTTTTAATAAUAQ (15)
TMS-TAATTTUPyTAAUPyAAUAQ (16)
TMS-TAAUPyTLTLUPyTLALALUPyAAUAQ (12)
CGGCCCCCGGCGGC (22)
5.4
9.6
10.9
16.0
GCCGCCGGGGGCCG (27)
CGGCEtCCCEtCGGCEtGGC (23)
CEtGGCEtCEtCEtCEtCEtGGCEtGGCEt (26)
-8.9
-27.8
a Conditions: 1 µM strand concentration, salt concentration at 150 mM NaCl, 10 mM sodium phosphate buffer, pH 7, 150 mM NaCl at 1 M NaCl, 10
mM sodium phosphate buffer, pH 7, and 1 M NaCl. b Sequences are given from 5′- to 3′-terminus; TMS denotes a trimethocystilbene cap, UAQ denotes
deoxyuridine with an anthraquinone cap; UPy denotes deoxyuridine with a pyrenylbutyramidopropargyl ligand, TL and AL denote LNA residues, and CEt
denotes N-4-ethyl-2′-deoxycytidine. See Figures 2 and 4 for structures. c Average of four melting points ( SD. d ∆Tm ) melting point difference with
respect to the unmodified duplex. e Average of hyperchromicities (hyperchr) at 260 nm upon duplex dissociation from 4 curves.
9
15224 J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007

Isostable DNA
A R T I C L E S
Figure 5. Fluorescence image of a microarray with low-density spots after hybridization and washing. Sequences of immobilized oligonucleotides are given
above and below. Each oligonucleotide was immobilized in duplicate, with spots vertically above each other to demonstrate reproducibility, using spotting
solutions containing 10 µM DNA, 2 × SSC, 0.2% SDS. The resulting spot size is approximately 2 mm. Hybridization was performed with a solution
containing equimolar amounts of target oligonucleotides compATCy (18, 5′-Cy3-ATTATTAAAAATTA-3′) and compGCCy (28, 5′-Cy3-GCCGC-
CGGGGGCCG-3′) at 10 µM each, in 0.1 M MES buffer, pH 6.3, for 1 h at 48 °C.
First, duplex stabilities were measured in homogeneous solution,
using equimolar amounts of AT2CapPyLNA (12), AT (13),
ATCap (14), AT2Cap (15), AT2CapPy (16), or GC (22),
GC3Et (23), GC8Et (26), and target strands compAT (17) or
compGC (27) (Table 4). Of the all-A/T sequences with
decorating substituents only (AT2CapPy:compAT, 16:17)
brought the melting point within the range of that for the all-
G/C motif with N4-ethylcytosines throughout (GC8Et:com-
pGC, 26:27). At 150 mM NaCl, the former duplex melted 3.4
°C below the latter; at 1 M salt, its melting point was 3.2 °C
above that of the latter (Table 4). Incorporating five LNA
residues on top of the duplex-stabilizing ligands (oligonucleotide
AT2CapPyLNA, 12), raised the melting point for the duplex
with all-A/T target compAT (17) by another 5.1-10 °C, so
that the goal of isostable duplexes was in fact exceeded,
producing duplexes for an all-A/T motif that meet well above
that for the tuned-down all-G/C counterpart GC8Et:compGC
(26:27).
9
J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007 15225

A R T I C L E S
Ahlborn et al.
observation from the in-solution experiments that duplex
stability can be tuned beyond isostability. As expected, target
affinity increased with an increasing number of duplex-
stabilizing substituents, though AT2CapPy (16) bound slightly
less of its target strand than AT2Cap (15) lacking the pyrenyl
residues. This effect is not significant, however, considering
the experimental error.
High fidelity multiplex detection of DNA on microarrays
requires well chosen hybridization and washing conditions,73
and predicting hybridization results has remained challenging
thus far.74 We therefore tested to what extent the fluorescence
intensities measured agreed with theoretical binding curves,
predicted with the computer program ChipCheck.75,35 ChipCheck
employs ∆H and ∆S values for predicting duplex formation.
For the current duplexes involving modified strands, enthalpy
and entropy of duplex formation were calculated from
melting curves using MeltWin 3.076 (see Table S1, Supporting
Information). The resulting theoretical binding curves in the
temperature range 0-100 °C are shown in Figure 7. All
input parameters for the ChipCheck simulation were those of
the microarray experiment (30 µL hybridization solution, 10
µM target strand, 2.5 mm2 spot size), with an estimated probe
density of 1011 molecules/cm,2,40 that is, 4 × 10-15 mol probes/
spot.
The position of the curves in Figure 7 reflects the order of
melting points (compare Table 4), with similar cooperativity
for the individual transitions. The experimental values for each
probe strand on the microarray spots (percent of the fluorescence
signal at saturation) were entered as large symbols. Excellent
agreement between calculated and experimental values was
observed for the more tightly binding probe strands, down to
the duplexes AT2CapPyLNA:compAT (12:17) and GC8Et:
compGC (26:27), whose melting points are close to the
temperature at which the microarray experiment was performed
(48 °C). For duplexes with lower melting points, the experi-
mental signal was stronger than predicted, suggesting a certain
level of unspecific binding on the surface, probably due to the
unoptimized surface chemistry of our custom-made microarrays.
Finally, a series of exploratory assays were performed with
high-density microarrays. These involved the same immobiliza-
tion method, but the spotting was performed by a robot suitable
for producing high spot densities. The sequences of the
oligonucleotides used, a representative fluorescence image after
hybridization with the target strands and washing, and the
integration of the fluorescence intensity for two different target
concentrations are shown in Figure 8.
Figure 6. Fluorescence intensities generated through integration of the spot
intensities of the type of image shown in Figure 5. The columns represent
the mean of the two spots each from five independent arrays displaying
the hybridization probe whose compound number is given below each bar.
The short-hand codes are: AT (13), ATCap (14), AT2Cap (15), AT2CapPy
(16), AT2CapPyLNA (12), GC8Et (26), GC3Et (23), and GC (22).
Our study thus showed that a decorated duplex based
exclusively on T:A and A:T base pairs can have a higher thermal
stability than the corresponding duplex containing only G:C and
CEt:G base pairs (Table 4). Furthermore, it is feasible to adjust
target affinities stepwise by an increasing number of substituents
introduced in the probe strands.
We then turned to microarray-based experiments. The oli-
gonucleotide probes with their 3′-terminal lysine residue were
covalently immobilized on aldehyde-coated glass slides using
an optimized version of a known procedure68 involving Schiff
base formation, reduction, and capping of the remaining
aldehyde groups (Scheme S1, Supporting Information). The first
series of experiments involved manual microarray generation
with a micropipette. Each oligonucleotide was spotted in
duplicate, and the remaining surface was subsequently coated
with a triethyleneglycol methyl ether terminating in an amino
group.68 A detailed study was performed to optimize the
conditions for hybridizing to target strands compATCy (18)
and compGCCy (28), including variation of the salt content,
temperature, incubation time, and washing procedure. Further,
control experiments with Cy3- and Cy5-labeled target strands
were performed that confirmed the absence of detectable cross-
hybridization between the target strands for the two different
sequence motifs (all-A/T and all-G/C). The final experiments
were performed with target strands featuring the same label
(namely cyanine dye Cy3) to avoid a bias introduced by different
fluorescence properties of the labels.
Figure 5 shows a typical fluorescence image of a microarray
displaying probe strands AT2CapPyLNA (12), AT (13),
ATCap (14), AT2Cap (15), AT2CapPy (16) and GC (22),
GC3Et (23), GC8Et (26) after hybridization at 48 °C and
washing at 48 and then 25 °C. The microarray assay was
performed five times under the same conditions. The aggregate
results from the quantitative fluorescence analyses are compiled
in Figure 6. At the target strand concentration chosen (10 µM
each), the signal for spots with the most strongly binding probe
strands GC (22) and GC3Et (23) was near saturation. (The
saturation signal level had also been independently confirmed
in a series of microarray experiments involving probe strands
AT (13) and GC (22) and from target strands compATCy (18)
and compGCCy (28) (from 0.01 to 80 µM), see Figure S1,
Supporting Information). More importantly, the most strongly
enforced all-A/T strand (AT2CapPyLNA, 12) bound more of
its target (compATCy, 18) than the most “tuned-down” version
of the all-G/C sequence (GC8Et, 26) bound its corresponding
all-G/C target strand (compGCCy, 28). This confirmed the
Again, the fluorescence intensities (Figure 8c/d) show that
the affinity of the most strongly enforced oligonucleotide
AT2CapPyLNA (12) for its all-A/T target compATCy (18) is
in the same range as that of the tuned-down oligonucleotide
probes GC3Et (23) and GC8Et (26) for their all-G/C target
compGCCy (28), as required for a microarray with (roughly)
isostable DNA. Further, an increasing number of duplex-
stabilizing ligands was again found to lead to a stepwise increase
in duplex stability for the all-A/T motif, while the introduction
(73) Han, T.; Melvin, C. D.; Shi, L. M.; Branham, W. S.; Moland, C. L.; Pine,
P. S.; Thompson, K. L.; Fuscoe, J. C. BMC Bioinformatics 2006, 7, S17.
(74) Pozhitkov, A.; Noble, P. A.; Domazet-Loso, T.; Nolte, A. W.; Sonnenberg,
R.; Staehler, P.; Beier, M.; Tautz, D. Nucleic Acids Res. 2006, 34, e66.
(75) Siegmund, K.; Steiner, U. E.; Richert, C. J. Chem. Inf. Comput. Sci. 2003,
43, 2153-2162.
(76) McDowell, J. A.; Turner, D. H. Biochemistry 1996, 35, 14077-14089.
9
15226 J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007

Isostable DNA
A R T I C L E S
Figure 7. Binding curves for target strands compAT (17) and compGC (27), hybridizing to the probe strands listed in the legend on the right-hand side
on microarray spots of 2.5 mm2 size, as calculated with the computer program ChipCheck. Values from experimental data, obtained from fluorescence
readings in microarray experiments with the same probe strands and labeled target strands compATCy (18) and compGCCy (28) (compare Figures 2 and
4) are entered at the hybridization temperature (48 °C) as large symbols. See Table 4 for sequences.
of ethylcytosine residues induces the necessary reduction in
target affinity of the probes of the all-G/C motif. The agreement
between predicted and observed binding was even better than
for the low-density microarray (see Figure S2, Supporting
Information).
Discussion
Besides technical issues, such as further reducing the
background on microarrays, the current results open up questions
regarding the concept of high fidelity DNA microarrays and its
limitation. Fundamentally, the fidelity of molecular recognition
should only be limited by the intrinsic binding energy,77 offered
by the nucleotides of the target strand. In natural DNA, the
complexes formed between probe strands and target strands are
fairly open. Compared to the active site of polymerases and
kinases,78-80 where a recognition from either π-surface and most
of the hydrogen-bonding edges are realized, the extent to which
a target strand is recognized in DNA duplexes is limited. If
nature had had a need for maximum fidelity, rather than a
compromise between fidelity, accessibility, and kinetic lability,
duplexes with more fully engulfed nucleobases could certainly
have been evolved. To what extent this compromise can be
changed toward isostable DNA through chemical modifications,
without affecting other physicochemical properties of DNA
unfavorably, is not immediately clear, though.
It is therefore interesting to ask what type of chemical
modification proved suitable for generating isostable DNA in
our study and which did not. All successful modifications
employed here are enforcing the duplex by bridging, by reducing
its flexibility, or by filling up what would otherwise be a larger
groove. The caps bridge the termini, thereby reducing fraying.
Figure 8. Results of microarray experiments with a high-density pattern
The pyrenyl moieties attached via alkynyl linkers to the
5-position of deoxyuridines can be expected to intercalate,63
again stiffening the duplex structure81 both through bridging
and through reduced dynamics along the helix axis. The ethyl
substituents at position 4 of the cytosines will be located in the
of spots. (a) Pattern spotted in individual areas of the microarray. The
numbers on the spots correspond to the sequences given on the right-hand
side. Spotting solutions containing 10 µM DNA, 2 × SSC buffer, 0.2%
SDS. Each compound was spotted twice to demonstrate reproducibility.
(b) Fluorescence image of a section of the microarray featuring three spotting
patterns after hybridization and washing. The spot size is approximately
80 µm in diameter. Hybridization was performed with a solution containing
target strands oligonucleotide compATCy (18, 5′-Cy3-ATTATTAAAAAT-
TA-3′) and compGCCy (28, 5′-Cy3-GCCGCCGGGGGCCG-3′), 0.01 µM
each, in 2 × SSC, 0.2% SDS for 1 h at 48 °C. (c and d) Fluorescence
intensities, as obtained through integration over areas of 80 µm diameter
of individual spots, the values are the means of 16 patterns on one chip.
Error bars show the standard deviation: (c) incubation with 0.01 µM target
strands compATCy (18) and compGCCy (28); (d) incubation with 1 µM
target strands.
(77) For the concept of intrinsic binding energy and its limits, see: Fersht, A.
Enzyme structure and mechanism, 2nd ed.; W. H. Freeman: New York,
1985; pp 303; 348-350.
(78) Doublie, S.; Tabor, S.; Long, A. M.; Richardson, C. C.; Ellenberger, T.
Nature 1998, 391, 251-258.
(79) For a review, see: Rothwell, P. J.; Waksman, G. AdV. Protein Chem. 2005,
71, 401-440.
(80) Bellinzoni, M.; Haouz, A.; Grana, M.; Munier-Lehmann, H.; Shepard, W.;
Alzari, P. M. Protein Sci. 2006, 15, 1489-1493.
(81) (a) Lerman, L. S. J. Mol. Biol. 1961, 3, 18-30. (b) Saenger, W. Principles
of Nucleic Acid Structure Springer, New York: 1984, p. 350.
9
J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007 15227

A R T I C L E S
Ahlborn et al.
Figure 9. Cartoon highlighting differences in conformational flexibility in a natural duplex and a duplex involving a decorated nucleic acid probe. Compared
to its unmodified counterpart the decorated duplex is expected to show a reduced propensity to form alternative, mismatched structures. Alternative structures
may also be less well stabilized by caps, less well bridged by intercalators, and less well accommodating conformational locks.
major groove, where they take up some of the space otherwise
filled by solvent. Finally, the methoxy locks of the LNA residues
linking 2′- and 4′-position reduce the conformational flexibility
of the deoxyriboses, again creating a stiffer structure82 and
increasing the number of contacts within the duplex.83 Inosine,
on the other hand, the only modification that failed to uphold
base-pairing fidelity, is smaller than the guanine it replaces,
leaving behind a hole that can be filled in a number of different
ways, most of which open up ways for alternative base pairs
that compete with the desired complex. Neither of the structures
tested in our quest for high fidelity probes, which includes the
current results and those of earlier studies,59,61,63,60,84 was
successful as a fidelity-enhancing element unless it created a
larger structure than that of natural DNA. Apparently, the fewer
alternative conformations are accessible and the more extensive
the number of contacts with the target strands, the more scarce
the number of alternative base-pairing arrangements, and the
more likely that a high fidelity probe is being generated that
binds only the fully complementary strand. A single accessible
conformation with exquisitely sculptured surface that fits a single
sequence in DNA structure space is the ultimate realization in
this scenario (Figure 9, above).
To what extent do alternative structures that have been
explored as hybridization probes on microarrays fit into this
picture? Peptide nucleic acids (PNAs), for example, have been
tested as probes, but have been found to be less uniform or less
predictable in their hybridization properties than DNA.85,86 PNA
has high affinity for target strands because its nonionic backbone
is not electrostatically repelled by the target strand. It features
an acyclic backbone, though, for which more conformations are
accessible than those found for DNA. In this more open
structure, alternative complexes can form more readily. The lack
of repulsion between like charges in the backbone also encour-
ages backfolding of the PNA chains, a property they share with
polypeptide chains, a class of biopolymers known for their
ability to form diverse folded structures.87 Other nonionic nucleic
acid analogues also show a propensity to fold back onto
themselves,55 making them problematic for high fidelity rec-
Figure 10. Overlay of UV-melting curves of duplexes without and with
adjusted base-pairing strength. The duplexes of the all-A/T and all-G/C
sequence motif without chemical modifications (13:17 and 22:27, respec-
tively) are shown together with the curves of the most strongly modified
versions of either sequence motif (12:17 and 26:27). Relative, rather than
absolute, hyperchromicity was plotted to more easily judge how similar
the cooperativity of the transitions is.
ognition of target strands from a diverse set of sequences by a
single, specific type of molecular interactions. The decorated
nucleic acids presented here not only avoid a pruning of the
backbone, they also have very few additional hydrogen-bonding
sites. Instead, they feature substituents whose shape comple-
mentarity and effect on stiffness enforce specific molecular
recognition without attracting alternative base-pairing arrange-
ments, and the like charges of their phosphodiester anions help
to reduce back-folding and aggregation.
We would like to emphasize that our approach for adjusting
duplex stability can be applied directly to existing platforms
for microarray generation, including those involving photo-
lithographic syntheses, since all building blocks employed are
phosphoramidites. Further, it is clear from the extensive
literature data on base-pairing fidelity at the terminus and the
results for LNA residues at the terminus presented in Table 1
that the same effect cannot be achieved with LNAs alone. Still,
it is important to ask to what extent ideal binding curves have
been achieved with the chemical approach presented here. Figure
10 shows an overlay of the melting curves of the unmodified
and most heavily modified duplexes of both the all-A/T and
the all-G/C sequence motif. It can be discerned that the
cooperativity of the melting transitions has not suffered from
the covalent modifications introduced. Cooperative transitions
are necessary for achieving high fidelity, as broad transitions
favor situations where mismatched duplexes are partially formed
while perfectly matched duplexes are still not fully formed at a
given temperature. The results obtained on microarrays further
confirm that equilibrium binding data obtained in solution are
useful predictors for detection on chip surfaces, despite the fact
that washing steps with the potential to bias those equilibria
(82) Other conformationally restricted DNA analogues exist; see, for example:
(a) Renneberg, D.; Leumann, C. J. J. Am. Chem. Soc. 2002, 124, 5993-
6002. (b) Steffens, R.; Leumann, C. J. J. Am. Chem. Soc. 1997, 119, 11548-
11549.
(83) Egli, M.; Minasov, G.; Teplova, M.; Kumar, R.; Wengel, J. Chem. Commun.
2001, 651-652.
(84) Connors, W. H.; Narayanan, S.; Kryatova, O. P.; Richert, C. Org. Lett.
2003, 5, 247-250.
(85) Igloi, G. L. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 8562-8567.
(86) Weiler, J.; Gausepohl, H.; Hauser, N.; Jensen, O. N.; Hoheisel, J. D. Nucleic
Acids Res. 1997, 25, 2792-2799.
(87) Branden, C.; Tooze, J. Introduction to Protein Structure, 2nd ed.; Garland:
New York, 1999.
9
15228 J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007

Isostable DNA
A R T I C L E S
ing O-benzotriazol-1-yl-N,N,N′,N′-tetramethyluronium hexafluorophos-
phate (HBTU) and 1-hydroxybenzotriazole (HOBT) as well as Boc-
Lys-(TFA)-OH, were from Advanced ChemTech (Louisville, KY) and
were used without purification. The 2-cyanoethyl-N,N-diisopropyl-
chlorophosphoramidite for phosphitylations was obtained from TRC
(North York, Canada). Underivatized long-chain alkylamine controlled
pore glass (LCAA cpg) with a loading capacity of 78 µmol/g was from
Controlled Pore Glass, Inc. (Lincoln Park, NJ). Anhydrous solvents
were ordered and stored over molecular sieves and used without further
purification. Standard phosphoramidites (dABz, dCBz, T, dGdmf), the
phosphoramidites of locked nucleic acid monomers and the chemicals
for DNA synthesis, including dicyanoimidazole (DCI) activator, were
from Proligo (Hamburg, Germany), except for dGdmf-loaded controlled
pore glass (cpg), which was from ABI (Warrington, U.K.). Further,
the phosphoramidites of N-4-ethyl-2′-deoxycytidine and 2′-deoxy-5′-
dimethoxytrityl-5-iodouridine were obtained from ChemGenes (Wilm-
ington, MA), whereas the phosphoramidite of Cy3 and 2′-deoxyinosine
were from Glen Research (Sterling, VA). The 3′-O-yl-cyanoethyl-N,N-
diisopropylphosphoramidite of 2′-(anthraquinone-N-methylcarboxa-
mido)-2′-deoxy-5′-O-(4,4′-dimethoxytrityl)uridine (8) was synthesized
as reported.66 The synthesis of the phosphoramidites of trimethoxystil-
bene61 and 3,6,9-trioxydecylamine68 also followed known protocols.
The LNA phosphoramidites, namely compound 10 and 11, were
obtained from Glen Research (Sterling, VA), and were used without
modification. Their synthesis has been described in the literature.46b
Aldehyde-coated glass slides were from Genetix Ltd. (New Milton,
U.K.). Phosphate buffered saline solution (PBS, pH 7), 20-fold
concentrated saline sodium citrate (SSC, pH 7.4), and 2-(N-morpholino)-
ethanesulfonic acid buffer (MES, pH 6.3) were prepared using standard
protocols.88
The unmodified oligonucleotides compGC 27, 27a, 27b, 27c, 27d,
27e, 27f, 27g, 27h, and 27i and the Cy3-labeled oligonucleotide
compGCCy 28 were purchased from Operon Biotechnologies GmbH
(Cologne, Germany) or VBC-Genomics (Vienna, Austria) in HPLC-
purified form. The buildings blocks for the assembly of the DP-Lys-
loaded cpg and the loaded support itself were generated following
protocols published earlier.67 Oligonucleotides were purified via
reversed phase HPLC on Nucleosil columns (C4 or C18; both 250 mm
× 4.6 mm by Macherey Nagel, Du¨ren, Germany). Unless otherwise
noted, this involved a gradient of CH3CN (B) in 0.1 M triethylammo-
nium acetate (pH 7.0) and detection at 260 nm. Yields of oligonucle-
otides are based on the integration of the HPLC trace of crude products.
The integration was not corrected for the absorbance caused by the
solvent front. Extinction coefficients for oligonucleotides were calcu-
lated through linear combination of the extinction coefficients of the
nucleotides and are uncorrected for hyperchromicity effects. For
modified oligonucleotides, the extinction coefficients were calculated
as the sum of the extinction coefficient of the unmodified oligonucle-
otide and the extinction coefficient of the covalently linked chro-
mophores. NMR spectra were acquired on a Bruker AC 250, AVANCE
400, or AVANCE 500 spectrometer. MALDI-TOF-MS spectra were
acquired on a Bruker REFLEX IV spectrometer in negative, linear mode
using the software XACQ 4.0.4 and XTof 5.1. The matrix was a mixture
of 2,4,6-trihydroxyacetophenone (THAP, 0.2 M in ethanol) and
diammonium citrate (0.1 M in water) (2:1, v/v). Calculated masses are
average masses, and m/z found are those of the unresolved isotope
envelopes of the pseudomolecular ions ([M - H]-). DNA sequences
are given 5′- to 3′-terminus.
through different dissociation kinetics are involved. Our results
were obtained with short incubation times, suggesting that the
added substituents of the decorated probes do not slow down
the kinetics of duplex formation to a point that requires adjusting
of protocols.
One may, of course, want to increase base-pairing stability
to the point where duplexes of any sequence are formed with
the stability of those between all-G/C sequences. While in
medicinal chemistry, tighter binding is almost universally hailed
as an improvement, the same is not necessarily the case for
probing specific sequences in genomic DNA on microarrays.
If one was to increase the affinity such that even short stretches
of nucleotides suffice to form stable duplexes close to the boiling
point of an aqueous buffer, false positives would again be
difficult to suppress, as these short sequences are not sufficiently
unique in a target genome to ensure the selectivity required for
detecting typical genes, let alone genes coding for members of
families of homologous proteins. Since it takes approximately
16 nucleotides in length to be statistically unique in the human
genome, hybridization probes shorter than those employed here
cannot be expected to be suitable for genome-wide analyses.
Oligonucleotide probes are used for binding target regions
against a genomic sequence background in biomolecular ap-
plications ranging from those frequently discussed for microar-
rays, such as gene expression profiling, SNP detection, DNA
copy number determination, alternative splicing analysis, and
epigenomics to those involving duplex formation in solution,
such as antisense, antigene, chimeraplasty, and RNA interfer-
ence. Isostable DNA may prove beneficial for several of these
applications, as it may reduce the need to optimize binding
properties each time a new sequence is being targeted. But,
additional experiments are needed to confirm the validity of
this approach.
The current results provide a concept for undertaking such
experiments, as they predict that a combination of (i) additional
ligands interrogating the target strand, (ii) a stiffening of the
duplex, and (iii) retention of the negatively charged DNA
backbone is beneficial for high fidelity detection of target strands
via isostable duplexes. It is hoped that our concept can serve
as a blueprint for future designs and offers a framework for
theoretical predictions.
Conclusions
In conclusion, the current results show that it is possible to
adjust duplex stabilities for oligodeoxynucleotides to and beyond
the same melting point with decorated nucleic acids, even for
the extreme case of all-A/T and all-G/C sequences. This was
demonstrated in solution and on microarrays of low and high
spot density. In our approach, only one of the two strand is
being modified, so that unmodified target strands may be
detected. Isostable DNA duplexes allow for the use of univer-
sally stringent conditions that should lead to higher fidelity in
the sequence-specific detection of diverse target strands under
multiplexing conditions. Predictable hybridization equilibria
should allow for in silico optimization of hybridization condi-
tions, further reducing the need for conventional, low-fidelity
experiments.
1-N-(Propin-3-yl)-4-(pyren-1-yl)butyramide (4). To a stirred solu-
tion of 1-pyrene-4-butyric acid (2) (640 mg, 2.22 mmol) in DMF (5
mL) were added HOBt (408 mg, 2.66 mmol) and EDC (467 µL, 413
mg, 2.66 mmol). After 10 min at room temperature, propargylamine
(171 µL, 147 mg, 2.66 mmol) was added slowly. The mixture was
Experimental Section
General. Unless noted otherwise, chemicals were purchased from
Acros (Geel, Belgium), Aldrich/Fluka/Sigma (Deisenhofen, Germany),
or Merck (Darmstadt, Germany). Reagents for amide coupling, includ-
(88) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular cloningsa Laboratory
Manual, 2nd ed.; Cold Spring Harbor Laboratory: Cold Spring Harbor,
New York, 1989.
9
J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007 15229

A R T I C L E S
Ahlborn et al.
stirred for 16 h, taken up in CH2Cl2 (70 mL), and washed thrice each
with aqueous HCl (1 M, 70 mL), aqueous NaOH (1 M, 70 mL), and
brine (70 mL). The organic layer was dried over Na2SO4 and evaporated
in vacuo. Recrystallization from EtOH gave the title compound as a
yellow solid (548 mg, 1.9 mmol, 76%). Rf ) 0.8, CHCl3/MeOH 9:1.
1H NMR (250 MHz; CDCl3): δ (ppm) ) 2.13 (m, 5H), 3.30 (t, J )
7.4 Hz, 2H), 3.96 (m, 2H), 5.52 (bs, 1H), 7.72-8.26 (m, 9H). 13C NMR
(100.6 MHz; CDCl3): δ (ppm) ) 27.20, 29.21, 32.65, 35.64, 71.63,
79.56, 123.37, 124.82, 124.97, 125.00, 125.12, 125.90, 126.77, 127.40,
127.47, 127.49, 128.81, 130.01, 130.92, 131.44, 135.69, 172.17. HR-
FAB-MS (3-NBA) calcd for C23H19NO, 325.1467; found, 325.1469.
the latter value was employed for LNA phosphoramidites, as recom-
mended by the suppliers.
Manual Chain Extension (General Protocol A). The solid support
(cpg) and the phosphoramidite (20-30 equiv) were dried at 0.1 Torr
in a polypropylene vessel. A solution of 1-H-tetrazole (0.45 M in
CH3CN, 50-200 µL, depending on the scale) was added, and the
reaction mixture was shaken on a vortexer at low speed for 1 h at room
temperature. After washing with CH3CN and CH2Cl2, oxidizer solution
for DNA synthesis (0.5 mL) was added, followed by shaking for 15
min at room temperature. The cpg was washed with CH3CN, DMF,
MeOH, and CH2Cl2, and was dried at 0.1 Torr, followed by continuation
of the DNA synthesis or deprotection.
Deprotection and Cleavage of Oligonucleotides from cpg (Gen-
eral Protocol B). At the end of DNA syntheses, the cpg was briefly
dried (0.1 Torr) and treated with saturated aqueous NH3 (1 mL) for 16
h at room temperature or 3 h at 55 °C. The supernatant was transferred
to a new vessel, and the solid support was washed with water (2 × 0.3
mL). Excess ammonia in the combined aqueous solution was removed
with a gentle stream of compressed air. The solution was filtered and
then used directly for HPLC purification.
5′-O-(4,4′-Dimethoxytrityl)-5-(3-(4-(pyren-1-yl)butyramido)propin-
1-yl)-2′-deoxyuridine (6). 5′-O-(4,4′-Dimethoxytrityl)-5-iodo-2′-deox-
yuridine (312 mg, 0.48 mmol) was dissolved in DMF (2.5 mL). The
solution was degassed in vacuo and the flask flushed with argon. Then,
NEt3 (321 µL, 231 mg, 2.29 mmol), compound 4 (465 mg, 1.43 mmol),
tetrakis(triphenylphosphine)palladium (55 mg, 48 µmol), and copper
iodide (20 mg, 0.10 mmol) were added. In each case, the mixture was
stirred until a clear solution was obtained before the next compound
was added. After 4 h at room temperature, the mixture was diluted
with ethyl acetate (70 mL) and washed thrice with aqueous EDTA
solution (5%, 70 mL) and brine (70 mL). The organic layer was dried
over Na2SO4 and the solvent was removed in vacuo. The yellow residue
was purified by column chromatography (silica, pretreated with eluant
containing 0.5% NEt3, CH2Cl2/MeOH 93:7, Rf ) 0.29) to give the title
Detritylation of Oligonucleotides During Purification (General
Protocol C). The removal of DMT or MMT groups on the 5′-terminal
nucleotide was achieved on a C-18 cartridge (Sep-Pak Vac 3 cm3, 500
mg, Milford, MA). The cartridge was first washed with CH3CN (4 mL)
and TEAA buffer (triethylammonium acetate, 0.1 M, pH 7, 6 mL).
The cartridge was loaded with the oligonucleotide dissolved in water
or TEAA buffer. Then, TEAA buffer (2 mL), trifluoroacetic acid
solution (0.2% in H2O, 4 mL), and TEAA buffer (4 mL) were passed
through the cartridge. The oligonucleotide was eluted with CH3CN/
TEAA (3:2, v/v, 4 mL) and CH3CN/TEAA (9:1, v/v, 4 mL). Fractions
were analyzed by MALDI-TOF-MS, and those containing pure product
were combined and repeatedly lyophilized from water to remove the
buffer.
1
compound as a slightly yellow solid (375 mg, 439 µmol, 92%). H
NMR (500 MHz; CDCl3): δ (ppm) ) 1.96 (t, J ) 7.2 Hz, 2H), 2.05
- 2.13 (m, 2H), 2.19-2.56 (m, 1H), 2.41-2.48 (m, 1H), 2.70 (bs,
1H), 3.25 (dd, J ) 10.8 Hz, J ) 3.3 Hz, 1H), 3.28-3.32 (m, 2H), 3.39
(dd, J ) 10.8 Hz, J ) 2.7 Hz, 1H), 3.72 (s, 6H), 3.85-3.99 (m, 2H),
4.03-4.07 (m, 1H), 4.43-4.48 (m, 1H), 5.35-5.40 (m, 1H), 6.28 (t,
J ) 6.3 Hz, 1H), 6.76-6.81 (m, 4H), 7.16-7.20 (m, 1H), 7.22-7.27
(m, 6H), 7.33-7.37 (m, 2H), 7.82 (d, J ) 7.9 Hz, 1H), 7.95-8.01 (m,
3H), 8.06-8.10 (m, 2H), 8.12-8.18 (m, 3H), 8.26 (d, J ) 9.1 Hz,
1H), 9.11 (bs, 1H). 13C NMR (100.6 MHz; CDCl3): δ (ppm) ) 27.06,
29.94, 32.67, 35.37, 41.57, 55.27, 63.43, 72.15, 74.12, 85.74, 86.64,
87.06, 89.55, 99.56, 113.36, 123.38, 124.78, 124.90, 124.93, 125.04,
125.84, 126.70, 127.00, 127.32, 127.38, 127.46, 127.84, 128.07, 128.75,
129.90, 129.97, 130.87, 131.37, 135.33, 135.49, 135.83, 142.98, 144.51,
149.09, 158.58, 158.61, 161.73, 172.04. HR-FAB-MS (3-NBA) calcd
for C53H47N3O8, 853.3363; found, 853.3365.
AT2CapPyLNA, 5′-TMS-TAAUPyTLTLUPyTLALALUPyAAUAQ
-
DP-Lys-3′ (12). Starting from solid support 9, a manual coupling cycle
with 8 according to general protocol A was performed, followed by
automated DNA synthesis with standard phosphoramidites and 1, 10,
and 11. Next, phosphoramidite 7 was coupled manually according to
general protocol A, followed by cleavage of the oligonucleotide from
solid support according to general protocol B. Crude 12 was dissolved
in 0.1 M LiOH with 20% CH3CN and HPLC purified on a C-18 column
with a gradient of CH3CN (20-70% in 40 min; Rt ) 23 min), yield of
40%. MALDI-TOF MS m/z [M - H]- calcd for C258H273N53O108P15,
6308.5; found, 6307.5.
AT, 5′-TAATTTTTAATAAT-DP-Lys-3′ (13). Prepared on solid
support 9 by standard DNA synthesis, deprotection (general protocol
B), and HPLC purification (C-18 column) with a gradient of 0% CH3CN
for 5 min, to 20% in 55 min; Rt ) 49 min, yield of 35%. MALDI-TOF
MS m/z [M - H]- calcd for C151H197N48O90P14, 4558.1; found, 4559.1.
ATCap, 5′-TMS-TAATTTTTAATAAT-DP-Lys-3′ (14). Synthe-
sized on support 9 with phosphoramidite 7 (general protocol A),
followed by cleavage (general protocol B) and HPLC purification (C-
18 column) with a gradient of 0% CH3CN for 5 min, to 25% in 30
min; Rt ) 31 min, yield of 66%. MALDI-TOF MS m/z [M - H]-
calcd for C172H221N49O97P15, 4991.5; found, 4989.2.
AT2Cap, 5′-TMS-TAATTTTTAATAAUAQ-DP-Lys-3′ (15). Start-
ing from solid support 9 phosphoramidite 8 was coupled (general
protocol A), followed by standard DNA synthesis, coupling of
phosphoramidite 7 (general protocol A), and cleavage (general protocol
B). HPLC purification (C-18 column) with a gradient of 0% B for 5
min, to 35% in 40 min; Rt ) 32 min, yield of 45%. MALDI-TOF MS
m/z [M - H]- calcd for C187H228N50O100P15, 5240.7; found, 5240.1.
AT2CapPy, 5′-TMS-TAATTTUPyTAAUPyAAUAQ-DP-Lys-3′ (16).
Synthesized identically to 15, except that phosphoramidite 1 was also
used in manual coupling steps (general protocol A). HPLC (C-18
column) with a gradient of 0% B for 5 min, to 35% in 40 min; Rt )
5′-O-(4,4′-Dimethoxytrityl)-5-(3-[4-(pyren-1-yl)butyramido]propin-
1-yl)-2′-deoxyuridin-3′-O-yl-cyanoethyl-N,N-diisopropylphosphora-
midite (1). Compound 6 (150 mg, 0.175 mmol) was dried for 16 h at
0.1 Torr, suspended in CH3CN (900 µL), and treated with DIEA (92
µL, 68 mg, 0.526 mmol). To the stirred suspension, 2-cyanoethyl-N,N-
diisopropylchlorophosphoramidite (51 µL, 54 mg, 0.228 mmol) was
added. After 15 min at room temperature, a clear, slightly yellow
solution formed, which was diluted with CH2Cl2 (30 mL) and washed
twice with saturated aqueous NaHCO3 (30 mL) and brine (30 mL),
respectively. The organic layer was dried over Na2SO4 and concentrated
to a volume of approximately 0.5 mL. The solution was added dropwise
to cyclohexane (20 mL), leading to an off-white precipitate. After
centrifugation, the supernatant was aspired, and the solid was purified
by column chromatography (silica, pretreated with eluant containing
0.5% NEt3, elution with acetone/cyclohexane 1:1, Rf ) 0.25) to give 1
as a slightly yellow solid (175.6 mg, 0.166 mmol, 95%). 31P NMR
(202 MHz; CDCl3): δ (ppm) ) 149.3, 149.7. MALDI-TOF MS
(THAP, linear positive mode) 1050.6.
Automated DNA Chain Assembly. The assembly of the unmodified
portion of oligodeoxynucleotide was performed on 1 µmol scale,
following the protocol recommended by the manufacturer of the
synthesizer (8909 Expedite synthesizer, Perseptive Biosystems, system
software 2.01), using a polypropylene reaction chamber for DNA
synthesis (Prime Synthesis, Aston, PA). Phosphoramidites of modified
residues were coupled with elongated coupling times of 3-9 min, where
9
15230 J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007

Isostable DNA
A R T I C L E S
40 min, yield of 24%. MALDI-TOF MS m/z [M - H]- calcd for
GC3EtCI, 5′-CIGCEtCCCEtCIGCEtGIC-DP-Lys-3′ (25). Starting
from 9, 25 was synthesized via standard DNA synthesis involving the
phosphoramidite building blocks of 2′-deoxyinosine and 2′-deoxy-N-
4-ethylcytidine followed by cleavage (general protocol B) and HPLC
purification (C-18 column) with a gradient of 0% B for 5 min, to 21%
in 35 min; Rt ) 31 min, yield of 41%. MALDI-TOF MS m/z [M -
H]- calcd for C149H198N53O88P14, 4573.7; found, 4571.2.
GC8CEt, 5′-CEtGGCEtCEtCEtCEtCEtGGCEtGGCEt-DP-Lys-3′ (26).
The synthesis was analogous to that of 23, except that the 5′-terminal
DMT group was removed on-cartridge according to general protocol
C subsequent to HPLC purification. HPLC (DMT-on, C-18 column),
gradient of 0% B for 5 min, to 35% in 45 min; Rt ) 43 min, yield of
70%. MALDI-TOF MS m/z [M - H]- calcd for C159H221N56O88P14,
4758.4; found, 4756.0.
UV-Melting Curves. UV-melting curves were measured on a Perkin-
Elmer Lambda 10 spectrophotometer with detection at 260 nm and
gradients of 1 °C/min. Melting temperatures were determined with UV
Winlab 2.0 (Perkin-Elmer), based on the extrema of the first derivative
of the 91-point smoothed curves, and are averages of four measure-
ments. Hyperchromicities were determined by calculating the difference
in absorption between high- and low-temperature baseline and dividing
by the low-temperature absorption. Thermodynamics of duplex forma-
tion were calculated by fitting with Meltwin.75
C231H258N52O102P15, 5859.2; found, 5857.0.
compAT, 5′-ATTATTAAAAATTA-3′ (17). Assembled via stan-
dard DNA synthesis, cleaved from solid support according to the general
protocol B, and HPLC purified (C-18 column) with a gradient of 0%
B for 5 min, to 18% in 35 min; Rt ) 38 min, yield of 66%. MALDI-
TOF MS m/z [M - H]- calcd for C140H175N52O80P13, 4268.9; found,
4268.4.
compATCy, 5′-Cy3-ATTATTAAAAATTA-3′ (18). Synthesized
via standard DNA synthesis followed by manual coupling of the
Cy3-phosphoramidite (general protocol A) and removal of the MMT
group (general protocol C). After cleavage from solid support
(general protocol B), HPLC purification (C-4 column) with a gradient
of 0% B for 5 min, to 30% in 25 min; Rt ) 29 min, yield of 46%.
MALDI-TOF MS m/z [M - H]- calcd for C169H211N54O84P14, 4776.5;
found, 4773.6.
LNACap, 5′-ALGGTTGAC-3′ (19). Chain assembly by standard
DNA synthesis, followed by manual coupling of phosphoramidite 10
(general protocol A), deprotection (general protocol B), and HPLC
purification (C-18 column) with a gradient of 0% B for 5 min, to 18%
in 25 min; Rt ) 30 min, yield of 82%. MALDI-TOF MS m/z [M -
H]- calcd for C80H99N32O47P7, 2477.6; found, 2472.8.
LNACapcon, 5′-AGGTTGAC-3′ (20). Synthesized via standard
DNA synthesis, cleavage (general protocol B), and HPLC purification
(C-18 column) with a gradient of 0% B for 5 min, to 18% in 25 min;
Rt ) 29 min, yield of 72%. MALDI-TOF MS m/z [M - H]- calcd for
C79H99N32O46P7, 2449.7; found, 2449.3.
compLNACap_A, 5′-GTCAACCA-3′ (21a). Synthesized analo-
gously to 20, HPLC purification (C-18 column) with a gradient of 0%
B for 5 min, to 18% in 30 min; Rt ) 30 min, yield of 75%. MALDI-
TOF MS m/z [M - H]- calcd for C77H98N31O44P7, 2378.6; found,
2378.1.
compLNACap_C, 5′-GTCAACCC-3′ (21c). Synthesized analo-
gously to 20, HPLC purification (C-18 column) with a gradient of 0%
B for 5 min, to 18% in 30 min; Rt ) 29 min, yield of 81%. MALDI-
TOF MS m/z [M - H]- calcd for C76H98N29O45P7, 2354.6; found,
2354.4.
DNA Microarrays.
Slides. The slides were cleaned from dust in a stream of argon,
washed with CH2Cl2 (50 mL, 10 min), ethanol (50 mL, 10 min), and
distilled water (50 mL, 10 s), and dried in a stream of argon.
Manual Spotting. For immobilization, a solution of lysine-
terminated oligonucleotides (1 µL, 10 µM DNA in 0.1 M MES buffer,
0.1 M glycerol) was applied to the slide surface without touching the
surface. After 2 h at room temperature in a humid chamber, a solu-
tion of NaBH3CN (0.35 µL, 50 mM in PBS buffer) was added to the
drops. After 1 h at room temperature, the droplets were removed by
rinsing with buffer (10 mL, 1 × SSC/0.2% SDS) followed by washing
by immersion into buffer (50 mL, 1 × SSC/0.2% SDS, 1 min and 50
mL, PBS, 1 min). The slide was used directly for passivation, without
drying.
compLNACap_G, 5′-GTCAACCG-3′ (21g). Synthesized analo-
gously to 20, HPLC purification (C-18 column) with a gradient of 0%
B for 5 min, to 18% in 30 min; Rt ) 30 min, yield of 74%. MALDI-
TOF MS m/z [M - H]- calcd for C77H98N31O45P7, 2394.6; found,
2394.2.
compLNACap, 5′-GTCAACCT-3′ (21t). Synthesized analogously
to 20, HPLC purification (C-18 column) with a gradient of 0% B for
5 min, to 18% in 30 min; Rt ) 30 min, yield of 89%. MALDI-TOF
MS m/z [M - H]- calcd for C77H99N28O46P7, 2369.6; found, 2368.9.
GC, 5′-CGGCCCCCGGCGGC-DP-Lys-3′ (22). Assembled on
support 9 via standard DNA synthesis, followed by cleavage (general
protocol B) and HPLC purification (C-18 column) with a gradient of
0% B for 5 min, to 20% in 40 min; Rt ) 38 min, yield of 20%. MALDI-
TOF MS m/z [M - H]- calcd for C143H189N56O88P14, 4534.0; found,
4527.3.
GC3Et, 5′-CGGCEtCCCEtCGGCEtGGC-DP-Lys-3′ (23). Prepared
on support 9 via a standard DNA synthesis, including the 3′-
phosphoramidite of 2′-deoxy-N-4-ethylcytidine, followed by cleavage
(general protocol B) and HPLC purification (C-18 column) with a
gradient of 0% B for 5 min, to 30% in 35 min; Rt ) 25 min, yield of
38%. MALDI-TOF MS m/z [M - H]- calcd for C149H201N56O88P14,
4618.2; found, 4617.5.
GC3I, 5′-CIGCCCCCIGCGIC-DP-Lys-3′ (24). Synthesized on
support 9 via automated synthesis involving the 3′-phosphorami-
dite of 2′-deoxyinosine, followed by cleavage (general protocol
B) and HPLC purification (C-18 column) with a gradient of 0% B for
5 min, to 19% in 35 min; Rt ) 32 min, yield of 64%. MALDI-TOF
MS m/z [M - H]- calcd for C143H186N53O88P14, 4489.0; found,
4487.9.
Automated Spotting. Solutions of the lysine-terminated oligonucle-
otides (1 µL, 10 µM DNA in 0.1 M MES buffer, 1 M glycerol) were
applied to the slide surface using a genemachine Omnigrid 100 with
SMP 3 pins (Genemachines/Genomic Solutions, Ann Arbor, MI).
Approximately 0.25 nL was applied per spot. After incubation for 16
h at room temperature in a humid chamber, the droplets were removed
by rinsing with buffer (10 mL, 1 × SSC/0.2% SDS) followed by
washing with buffer (50 mL, 1 × SSC/0.2% SDS, 1 min and 50 mL,
PBS, 1 min). The slides were then used directly for passivation without
drying.
Background Passivation. Remaining aldehyde groups on the surface
were reacted with 3,6,9-trioxydecylamine68 (10 mM solution in PBS
buffer), and the resulting imino linkages reduced with NaBH3CN (50
mM in the solution applied). After incubation for 3 h at room
temperature, the slide was washed with buffer (50 mL, 1 × SSC/0.2%
SDS, 1 min), twice with water (50 mL each, 10 s), and dried in a stream
of argon.
Hybridization. For hybridization, a solution containing the Cy3-
labeled target strands (30 µL, 10 µM DNA, 2 × SSC buffer, 0.2%
SDS) was applied to the slide surface under a cover slip. The slide
was stored in an incubation chamber (Peqlab-Biotechnologie,
Erlangen, Germany) at the desired temperature with sufficient humidity
to avoid drying. The slide was washed by immersion into buffer (50
mL, 1 × SSC/0.2% SDS) at the same temperature as that used for
hybridization. A second washing step involved buffer (50 mL, 1 ×
SSC) at room temperature. The slide was then immediately dried in a
stream of argon.
Fluorescence Scans and Data Analysis. The read-out of the
fluorescence signals from the spots of the slides was performed with
9
J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007 15231

A R T I C L E S
Ahlborn et al.
an ArrayWoRxe Biochip Reader (GeneScan, Freiburg, Germany). The
exposure time was set to 0.10 s. Fluorescence of Cy3 was detected by
using the following filters: 546 ( 5 nm (excitation); 575 ( 30 nm
(emission). The fluorescence reading for Cy5-labeled oligonucleotides
performed during control studies involved the following settings:
excitation at 640/20 nm and detection at 680/30 nm. The analysis of
for helpful discussions, to S. Lumpp and A. Hochgesand for
expert technical assistance, and to P. Gru¨nefeld, K. Imhof, E.
Kervio, and M. Ro¨thlingsho¨fer for a review of the manu-
script. This study was supported by DFG (FOR 434 and RI
1063/4-4).
e
images was performed using the ArrayWoRx analysis software. The
diameter of the area analyzed for each spot was set to a value slightly
below the real diameter (e.g., 0.9 mm diameter for manually spotted
arrays) to avoid integration of background signal. Unless noted
otherwise, the signal of the background was subtracted from the signals
for individual spots.
Supporting Information Available: Complete refs 19, 20,
and 28; reaction scheme for immobilization of probe strands
on slides; thermodynamics of duplex formation; data on
saturation for microarrays; calculated binding curves and
experimental data for high-density microarray; NMR spectra;
and MALDI-TOF mass spectra. This material is available free
of charge via the Internet at http://pubs.acs.org.
Acknowledgment. The authors are grateful to M. Bauer for
spotting high-density microarrays, to S. Al-Rawi for providing
synthetic intermediates, to A. Patra for mass spectrometric data,
to J. Rojas Stu¨tz, U. Plutowski, P. Baumhof, and K. Giessler
JA074209P
9
15232 J. AM. CHEM. SOC. VOL. 129, NO. 49, 2007