Synthesis, dynamic combinatorial chemistry, and PCR amplification of 3'-5' and 3'-6' disulfide-linked oligonucleotides

Article abstract of DOI:10.1002/anie.201405761

Disulfide dithymidines linked 3'-5' or 3'-6' were synthesized and incorporated into oligonucleotides through a combined phosphotriester and phosphoramidite solid-phase oligonucleotide synthesis approach. The disulfide links are cleaved and formed reversibly in the presence of thiols and oligonucleotides. This link was shown to be sequence-adaptive in response to given templates in the presence of mercaptoethanol. The artificial 3'-5' and 3'-6' disulfide link was tolerated by polymerases in the polymerase chain reaction (PCR). By using sequencing analysis, we show that single mutations frequently occurred randomly in the amplification products of the PCR.

Full text of DOI:10.1002/anie.201405761

Angewandte
Chemie
DOI: 10.1002/anie.201405761
Reversible DNA Ligation
Synthesis, Dynamic Combinatorial Chemistry, and PCR Amplification
of 3’–5’ and 3’–6’ Disulfide-linked Oligonucleotides**
Dennis Jul Hansen, Ilenia Manuguerra, Michael Brøndum Kjelstrup, and
Kurt Vesterager Gothelf*
Abstract: Disulfide dithymidines linked 3’–5’ or 3’–6’ were
synthesized and incorporated into oligonucleotides through
a combined phosphotriester and phosphoramidite solid-phase
oligonucleotide synthesis approach. The disulfide links are
cleaved and formed reversibly in the presence of thiols and
oligonucleotides. This link was shown to be sequence-adaptive
in response to given templates in the presence of mercapto-
ethanol. The artificial 3’–5’ and 3’–6’ disulfide link was
tolerated by polymerases in the polymerase chain reaction
(PCR). By using sequencing analysis, we show that single
mutations frequently occurred randomly in the amplification
products of the PCR.
They utilized a preformed peptidic backbone with reversible
thioester linkages to the nucleobases, and in this manner they
were able to exchange the nucleobases in response to given
templates.
Applications of reversibly linked oligonucleotides could
include formation of randomized libraries of oligonucleotides.
If the artificial backbone is tolerated by polymerases, such
libraries could hereafter generate peptide libraries by phage
display.[14,15] If the randomized sequences are made up of
stable nucleotide trimers, stop codons could principally be
omitted.[16] We have therefore also investigated the behavior
of the disulfide-linked oligonucleotides toward polymerases
by employing these as templates in a primer extension
reaction.
Two different disulfide linkages were designed; the 3’–5’
linkage contains the same carbon skeleton as natural nucleo-
sides, whereas the 3’–6’ linkage contains the same number of
atoms in the internucleosidic linkage as natural nucleosides
(Scheme 1). Patzke et al. modeled the 3’–5’ disulfide link and
found that the internucleotidic geometry is comparable with
the natural phosphodiester link.[9]
T
he disulfide bond is common in nature, in which it
contributes significantly to the tertiary structure and stability
of extracellular proteins.[1] Disulfides are known to exist in
dynamic exchange equilibria in the presence of thiolate,
a process known to alter the structure of proteins,[2] and are
commonly exploited within the field of dynamic combinato-
rial chemistry to generate novel ligands and receptors.[3–7]
Disulfide-linked dinucleosides have previously been synthe-
sized by Witch et al.[8] Only recently was the disulfide linkage
incorporated into oligonucleotides as an internucleosidic
linkage.[9] Patzke et al. utilized a DNA-templated reaction
between an activated disulfide and a thiol to generate a fast
ligation reaction. We set out to study if disulfide-linked
oligonucleotides can exist in dynamic equilibria that would
form different products under the influence of different
templates.
The thiols 1 and 2, and the disulfide 3 were synthesized
from thymine by modified literature procedures (see the
Supporting Information, SI).[17–21] The coupling of thiols, 1 and
2, with pyridyl disulfide, 3, was achieved with triethylamine in
CH2Cl2 over few minutes, resulting in the protected disulfide
dithymidines, 4 and 5, in up to 75% yield. Triethylamine
It has been argued that in early evolution it would be
advantageous for pre-oligonucleotides to have reversible
backbone linkages that could be formed and cleaved in the
absence of enzymes, and synthetic models of such systems
have, for example, included the imine,[10] the glyoxylate,[11]
and the boronic ester[12] linkage. An alternative approach
employing reversible linkages was devised by Ura et al.[13]
[*] D. J. Hansen, I. Manuguerra, M. B. Kjelstrup, Prof. K. V. Gothelf
Danish National Research Foundation: Center for DNA Nano-
technology
Department of Chemistry and iNANO
Gustav Wieds Vej 14, 8000 Aarhus C (Denmark)
E-mail: kvg@chem.au.dk
Homepage: http://www.cdna.dk
[**] This work was supported by the Danish National Research
Foundation (Grant number DNRF81) and Aarhus University, Faculty
of Science. Dr. Thomas Tørring is acknowledged for design of
oligonucleotide sequences.
Scheme 1. Formation of the disulfide-linked dinucleoside and subse-
quent preparation of the 2-cyanoethylphosphate esters. DMTr=dime-
thoxytrityl.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201405761.
Angew. Chem. Int. Ed. 2014, 53, 1 – 5
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

.
Angewandte
Communications
Table 1: Sequences of the oligodeoxyribonucleotides synthesized con-
taining the 3’–5’ or 3’–6’ disulfide linkage.
trihydrofluoride facilitated the removal of the tert-butyldi-
methylsilyl (TBS) protecting group in 82–85% yield.
Incorporation of the disulfide dithymidines into oligonu-
cleotides was attempted by standard phosphoramidite
chemistry. Preparation of the phosphoramidite of 7 was,
however, unsuccessful apparently due to reduction of the
disulfide moiety by PIII in the phosphoramidite, as was evident
from 31P NMR measurements. This is surprising, because
disulfide linkers are commonly employed in phosphorami-
dite-based solid-phase synthesis.[22] Instead we turned to the
phosphotriester method for oligonucleotide synthesis, in
which the monomer contains PV, and therefore would not be
prone to undesired redox reactions. However, the phospho-
triester method typically involves 2-chlorophenyl protection
groups for the phosphate moiety, but this protection group
requires oximate reagents in the final deprotection.[23,24] We
therefore decided to employ the 2-cyanoethyl protection
group. The resulting intermediate is the same as that obtained
from the phosphoramidite coupling cycle; therefore, only one
final deprotection step is required. The 2-cyanoethyl-pro-
tected phosphate monomers could thus be prepared by
a mesitylene sulfonyl chloride mediated condensation of the
respective 3’ alcohol and pyridinium 2-cyanoethylphosphate.
Oligonucleotides bearing the disulfide-linked backbone
modification could hereafter be synthesized as shown in
Scheme 2. We synthesized six different oligonucleotides
(ODN-1–ODN-6, Table 1).
Name
Sequence
ODN-1 (aSSb)
d(CTTTCATCTTACTTATCTACATCCCGACCGC
TCGT6’SSTTAACCGCAACCTCACCATACAACC)
d(TCTCACCTCACCATCCCTACTACACTAACT6’SSTCG
CGAGGCATTATCACACTACACTTCACACTTCACATCA)
d(GTTCCGCCAACTTACGACCGCTCGT6’SSTCGC
GAGGCATGCCCTCAGGACCAGCACGGATAAC)
d(AACCCTACACTAACT6’SSTCGCGAGG
ODN-2 (cSSd)
ODN-3
ODN-4
CATGCCCTCAGGACCAGCACGGATAAC)
ODN-5
d(GTTCCGCCAACTTACGACCGCTCGT5’SSTCGC
GAGGCATGCCCTCAGGACCAGCACGGATAAC)
d(AACCCTACACTAACT5’SSTCGCGAGG
ODN-6
CATGCCCTCAGGACCAGCACGGATAAC)
Figure 1. Denaturing PAGE analysis of the reductive cleavage of 3’–5’
or 3’–6’ disulfide-linked oligonucleotides.
to bring these disulfide-linked oligonucleotides into a dynamic
equilibrium under thermodynamic control.[4] In this manner it
should be possible to favor the formation of a certain
oligonucleotide, if a complementary oligonucleotide is pres-
ent in the solution. To investigate this concept, the oligonu-
cleotides ODN-1 (aSSb) and ODN-2 (cSSd) were brought
into dynamic exchange equilibrium by the addition of ME, as
illustrated in Figure 2. The dynamic equilibrium was inves-
tigated by denaturing PAGE, without a template, and with the
sequences complementary to the central regions surrounding
the disulfide linkage a’b’, c’d’, a’d’, c’b’ (ODN-11–ODN-14,
SI).
In the presence of ME the oligos are cleaved as expected
(Figure 2, lanes 2, 4, and 6). In the presence of templates and
ME, it is observed that the resulting sequence of the
oligonucleotides adapts a sequence complementary to the
template. Control experiments confirm that in the absence of
ME this is not observed. ME is present in 10000-fold excess
(2.5 mm) and the disulfides present are a mixture of all mixed
disulfides of a, b, c, d, and in particular of ME. However, in
the presence of ME and template a’b’, we observe complete
Scheme 2. Oligonucleotide synthesis employing a mixed phosphorami-
dite and phosphotriester strategy for the incorporation of disulfide
dithymidines 16 and 17. MSNT=1-(2-mesitylensulfonyl)-3-nitro-1H-
1,2,4-triazol; 1-MeIm=1-methylimidazole.
Since the oligonucleotides contain a disulfide group in the
backbone at one single position, we expected that the
oligonucleotides could be site-specifically cleaved under
reductive conditions to yield two thiol-terminated oligonu-
cleotides. This was indeed the case, as shown in Figure 1.
Treatment of the disulfide-linked oligonucleotides with an
excess of tris(2-carboxyethyl)phosphine (TCEP), dithiothrei-
tol (DTT), or mercaptoethanol (ME) resulted in two shorter
oligonucleotides of the expected molecular size as seen by
denaturing polyacrylamide gel electrophoresis (PAGE).
Inspired by the concepts of dynamic combinatorial
chemistry of disulfides, we speculated that it may be possible
2
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1 – 5
These are not the final page numbers!

Angewandte
Chemie
two DNA polymerases to read the disulfide linkage was
assessed through primer extension reactions with three
different primers (ODN-7–ODN-9, SI) for the oligonucleo-
tides ODN-4 and ODN-6 containing the 3’–6’ and 3’–5’
disulfide linkage, respectively. The primers were designed to
bind at different distances relative to the disulfide of the
oligonucleotide, with primer 1 binding directly next to the
disulfide modification, and primer 2 and 3 binding 5 and 10
bases, respectively, downstream of the disulfide. Primer
extension was hereafter attempted utilizing two different
polymerases; the Q5 polymerase and the Taq polymerase. In
all cases we observed an amplicon of the expected molecular
size with primers 2 and 3, whereas primer 1 only yielded
a significant amount of amplicon in the case of the 3’–5’ linked
ODN-6 with the Q5 polymerase. In the remaining cases either
no amplicon was observed or only inefficient amplification
was observed.
The results shown for primer extension of ODN-6
containing the 3’–5’ disulfide linkage shows that the poly-
merase can indeed read through the artificial disulfide
backbone (Figure 3). Almost identical results were observed
for primer extension of ODN-4 containing the 3’–6’ disulfide
linkage (SI). For both oligonucleotides it is also observed that
the polymerase does not bind efficiently to the oligonucleo-
tide when the primer is extended from the position of the
disulfide. These experiments do, however, not give insight
into the nature of the amplicon. It has previously been
Figure 2. A) Denaturing PAGE analysis of mercaptoethanol (ME)-medi-
ated dynamic equilibration of disulfide-linked oligonucleotides in the
presence of various templates. In all cases experiments were per-
formed with ODN-1 (aSSb) and ODN-2 (cSSd). B) The dynamic
equilibrium and the template-induced equilibrium shift.
disappearance of cSSd, whereas a band for the formation of
aSSb is observed (Figure 2, lane 8). In the presence of ME and
c’d’ we observe disappearance of aSSb and appearance of aSH
and HSb, whereas the band for cSSd appears (Figure 2,
lane 10). The same templating effect is observed for the
crossover products aSSd and cSSb when the respective
templates are applied. As observed in Figure 2, lane 12, the
presence of ME and template a’d’ leads to the formation of
aSSd and the amounts of aSH or dSH are greatly reduced. In
the presence of ME and c’b’ the last and shorter crossover
product cSSb appears as a new weak band (Figure 2, lane 14).
It is therefore evident that the various templates induce
a thermodynamically controlled equilibrium shift to form
sequences complementary to the given template. We believe
the template-directed formation of disulfide-linked oligonu-
cleotides is mediated by the template because the product is
thermodynamically highly favored by hybridization to the
template. This dynamic equilibration was observed at ME
concentrations down to 250 mm. At lower concentrations no
reaction took place, which is probably caused by significantly
lower rates of the bimolecular reactions.
For the analysis of potential combinatorial libraries of
oligonucleotides containing one or more disulfides in the
backbone it is crucial that the disulfide linkage can be read by
polymerases for amplification and sequencing. The ability of
Figure 3. A) The primer extension setup with ODN-6 as template and
three different primers. B) Denaturing PAGE analysis of primer exten-
sion using the Q5 polymerase.
Angew. Chem. Int. Ed. 2014, 53, 1 – 5
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

.
Angewandte
Communications
observed, that for some artificial backbones, the polymerase
skips a nucleoside around the unnatural linkage,[25] whereas
for other backbones the nucleosides are amplified as if they
were completely natural oligonucleotides.[26] The polymerase
chain reaction (PCR) amplicons were therefore analyzed by
DNA sequencing. For this purpose, we utilized oligonucleo-
tides ODN-3 and ODN-5 to be able to design a set of primers,
ODN-9 and ODN-10, binding to the oligonucleotide distant
to the disulfide linkage. Cloning into plasmids and replication
in E. coli resulted in plasmids with the PCR product inserted
that could be analyzed by DNA sequencing. For each
disulfide modification and polymerase, ten colonies were
sequenced. For 39 out of 40 clones we observed that the two
thymidines surrounding the disulfide were indeed amplified
as two thymidines. We did, however, also observe a relatively
high rate of point mutations occurring during the PCR
amplification (SI). In particular, nonsystematic deletions of
nucleotides in the sequence surrounding the disulfide modi-
fication were observed. Insertions or substitutions of nucle-
otides were also observed, although not at a frequency as high
as for the deletions. The two different linkages present in
ODN-3 (3’–6’) and ODN-5 (3’–5’) displayed similar error
rates. One clone, in which the two thymidines surrounding the
disulfide linkage were amplified as one thymidine, was
therefore ascribed to such a nonsystematic mutation. In the
context of the application of such experiments for in vitro
evolution, these point mutations could in fact prove benefi-
cial, as these disulfide-linked oligonucleotides would most
certainly exhibit significantly faster evolution than naturally
occurring oligonucleotides.
We have developed a synthetic method that allows for the
incorporation of disulfide-linked dinucleoside monomers into
oligonucleotides and the disulfide-linked oligonucleotides
display an interesting new class of reversibly backbone-linked
oligonucleotides. Whether the oligonucleotides are linked 3’–
5’ or 3’–6’ does not seem to alter the properties significantly.
The sequence of such oligonucleotides is able to adapt to
a given template, and the disulfide moiety is tolerated by
DNA polymerases although with a relatively high mutation
rate of the sequence surrounding the disulfide linkage. The
methodology potentially allows for the incorporation of
several disulfide linkages in one sequence, which would
enable the formation of complex libraries of disulfides.
Keywords: disulfides · dynamic combinatorial chemistry ·
nucleic acids · primer extension · solid-phase synthesis
.
[1] C. S. Sevier, C. A. Kaiser, Nat. Rev. Mol. Cell Biol. 2002, 3, 836 –
847.
[2] H. F. Gilbert, Adv. Enzymol. 1990, 63, 69 – 172.
[3] S. Ladame, A. M. Whitney, S. Balasubramanian, Angew. Chem.
Int. Ed. 2005, 44, 5736 – 5739; Angew. Chem. 2005, 117, 5882 –
5885.
[4] S. Otto, R. L. E. Furlan, J. K. M. Sanders, Drug Discovery Today
2002, 7, 117 – 125.
[5] S. Otto, R. L. E. Furlan, J. K. M. Sanders, Science 2002, 297,
590 – 593.
[6] P. T. Corbett, J. Leclaire, L. Vial, K. R. West, J. Wietor, J. K. M.
Sanders, S. Otto, Chem. Rev. 2006, 106, 3652 – 3711.
[7] S. Otto, R. L. E. Furlan, J. K. M. Sanders, J. Am. Chem. Soc.
2000, 122, 12063 – 12064.
[8] E. M. Witch, R. Cosstick, Nucleosides Nucleotides 1997, 16,
1555 – 1558.
[9] V. Patzke, J. S. McCaskill, G. von Kiedrowski, Angew. Chem. Int.
Ed. 2014, 53, 4222 – 4226.
[10] X. Li, Z. J. Zhan, R. Knipe, D. G. Lynn, J. Am. Chem. Soc. 2002,
124, 746 – 747.
[11] H. Bean, F. L. Anet, I. Gould, N. Hud, Origins Life Evol.
Biospheres 2006, 36, 39 – 63.
[12] A. R. Martin, I. Barvik, D. Luvino, M. Smietana, J. Vasseur,
Angew. Chem. Int. Ed. 2011, 50, 4193 – 4196; Angew. Chem.
2011, 123, 4279 – 4282.
[13] Y. Ura, J. M. Beierle, L. J. Leman, L. E. Orgel, M. R. Ghadiri,
Science 2009, 325, 73 – 77.
[14] G. P. Smith, V. A. Petrenko, Chem. Rev. 1997, 97, 391 – 410.
[15] M. Hamzeh-Mivehroud, A. A. Alizadeh, M. B. Morris, W.
Bret Church, S. Dastmalchi, Drug Discovery Today 2013, 18,
1144 – 1157.
[16] P. H. Faix, M. A. Burg, M. Gonzales, E. P. Ravey, A. Baird, D.
Larocca, BioTechniques 2004, 36, 1018.
[17] D. K. Baeschlin, M. Daube, M. O. Blꢀttler, S. A. Benner, C.
Richert, Tetrahedron Lett. 1996, 37, 1591 – 1592.
[18] J. Huang, E. B. McElroy, T. S. Widlanski, J. Org. Chem. 1994, 59,
3520 – 3521.
[19] M. Xie, T. S. Widlanski, Tetrahedron Lett. 1996, 37, 4443 – 4446.
[20] J. W. Gaynor, M. M. Piperakis, J. Fisher, R. Cosstick, Org.
Biomol. Chem. 2010, 8, 1463 – 1470.
[21] L. Lin, J. Sheng, R. K. Momin, Q. Du, Z. Huang, Nucleosides
Nucleotides Nucleic Acids 2009, 28, 56 – 66.
[22] M. J. Conrad, S. Coutts, J. P. Hachmann, D. S. Jones, D. A.
Livinsgton, WO 1993002093, 1993.
[23] C. B. Reese, R. C. Titmas, L. Yau, Tetrahedron Lett. 1978, 19,
2727 – 2730.
[24] C. B. Reese, L. Zard, Nucleic Acids Res. 1981, 9, 4611 – 4626.
[25] A. El-Sagheer, T. Brown, J. Am. Chem. Soc. 2009, 131, 3958 –
3964.
Received: May 29, 2014
Revised: September 22, 2014
Published online: && &&, &&&&
[26] A. H. El-Sagheer, A. P. Sanzone, R. Gao, A. Tavassoli, T. Brown,
Proc. Natl. Acad. Sci. USA 2011, 108, 11338 – 11343.
4
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1 – 5
These are not the final page numbers!

Angewandte
Chemie
Communications
Reversible DNA Ligation
D. J. Hansen, I. Manuguerra,
M. B. Kjelstrup,
K. V. Gothelf*
&&&& — &&&&
Synthesis, Dynamic Combinatorial
Chemistry, and PCR Amplification of 3’–5’
and 3’–6’ Disulfide-linked
Internucleosidic 3’–5’ and 3’–6’ disulfide
linkages allows cleavage and template-
directed formation of desired disulfides
in the presence of mercaptoethanol. The
artificial disulfide backbone is tolerated
by polymerases and the sequences can be
amplified by polymerase chain reaction.
Oligonucleotides
Angew. Chem. Int. Ed. 2014, 53, 1 – 5
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!