The role of organic linkers in directing DNA self-assembly and significantly stabilizing DNA duplexes

Article abstract of DOI:10.1021/ja3033197

We show a simple method to control both the stability and the self-assembly behavior of DNA structures. By connecting two adjacent duplexes with small synthetic linkers, factors such as linker rigidity and DNA strand orientation can increase the thermal denaturation temperature of 17 base-pair duplexes by up to 10 °C, and significantly increase the cooperativity of melting of the two duplexes. The same DNA sequence can thus be tuned to melt at vastly different temperatures by selecting the linker structure and DNA-to-linker connectivity. In addition, a small rigid m-triphenylene linker directly affects the self-assembly product distribution. With this linker, changes in the orientation of the linked strands (e.g., 5′3′ vs 3′3′) can lead to dramatic changes in the self-assembly behavior, from the formation of cyclic dimer and tetramer to higher-order oligomers. These variations can be readily predicted using a simple strand-end alignment model.

Full text of DOI:10.1021/ja3033197

Article
pubs.acs.org/JACS
The Role of Organic Linkers in Directing DNA Self-Assembly and
Significantly Stabilizing DNA Duplexes
Andrea A. Greschner, Violeta Toader, and Hanadi F. Sleiman*
Department of Chemistry and Center for Self-Assembled Chemical Structures (CSACS), McGill University, 801 Sherbrooke Street
West, Montreal, QC H3A 2K6, Canada
S
* Supporting Information
ABSTRACT: We show a simple method to control both the
stability and the self-assembly behavior of DNA structures. By
connecting two adjacent duplexes with small synthetic linkers,
factors such as linker rigidity and DNA strand orientation can
increase the thermal denaturation temperature of 17 base-pair
duplexes by up to 10 °C, and significantly increase the cooperativity
of melting of the two duplexes. The same DNA sequence can thus
be tuned to melt at vastly different temperatures by selecting the
linker structure and DNA-to-linker connectivity. In addition, a small
rigid m-triphenylene linker directly affects the self-assembly product distribution. With this linker, changes in the orientation of
the linked strands (e.g., 5′3′ vs 3′3′) can lead to dramatic changes in the self-assembly behavior, from the formation of cyclic
dimer and tetramer to higher-order oligomers. These variations can be readily predicted using a simple strand-end alignment
model.
which give dimers for all connectivities (Scheme 1c). We
propose a mechanism where the assembly is directed by strand-
end orientation, and we use this mechanism to successfully
predict the assembly outcome of a system with shorter DNA
strands.
Changing the duplex connectivity permitted further tuning of
the TM’s of each dimerup to 5 °C, an effect that could be
used to create DNA nanostructures with varied melting
temperatures without changing the sequences.
Previous reports from the groups of Mirkin, Nguyen, and
Schatz have shown that gold nanoparticles19−22 or poly-
mers14,21−24 decorated with multiple neighboring DNA strands
can display increased cooperativity in their melting transitions
and increased melting temperatures. Recently as well, ‘caged
dimers’ that are linked together by longer and more flexible
organic linkers displayed increased cooperativity in their
melting transitions.12,25,26 The present study demonstrates
that, when smaller and more rigid organic linkers with the
proper strand orientation are used to link two DNA strands,
significant stabilization can be obtained, comparable now to
aggregates with multiple DNA strands. In addition, the nature
and connectivity of the linker can directly influence the DNA
self-assembly outcome in a predictable manner.
INTRODUCTION
■
DNA is a powerful template to organize nanomaterials with
precisely programmed features. Most current approaches in
DNA nanotechnology, such as DNA tile assembly1−3 or DNA
origami,4,5 use only DNA strands to guide the assembly
process. We6−8 and others9−12 have demonstrated an
alternative strategy that uses synthetic molecules as corner
units and DNA strands as arms. This strategy allows the
combination of the diverse structures and functions of organic
molecules or transition metal complexes with the programm-
ability of DNA. It has led to new methods of building DNA
nanostructures that are DNA-economic and intrinsically
dynamic, and that can display redox, photophysical, photo-
chemical or catalytic activity.13−18
Here, we demonstrate that the structure and connectivity of
synthetic organic linkers can play a major role in guiding the
DNA self-assembly process, and can significantly stabilize the
DNA duplexes in the resulting nanostructures (Scheme 1 and
Chart 1). By attaching two identical DNA strands to short
linkers in a 5′ to 3′ manner, self-assembled DNA dimers come
together with significant increase in their thermal denaturation
temperature (TM) as compared to separate DNA strands. A
short and rigid triphenylene linker gives the greatest
stabilizationa near 10 °C increase in TM, and almost full
cooperativity in the melting of its two duplexes.
This study provides new structural parameters that can be
readily used in DNA nanotechnology. Instead of relying on a
large number of different DNA sequences to prevent self-
assembly errors, the structure of the organic linker and its
connectivity can be used to tune the assembly process, while
A rigid organic linker also provides the ability to tune and
dramatically modify the self-assembly outcome. 5′-3′ con-
nectivity leads to clean dimer formation, while 5′-5′
connectivity does not give dimer at all, but instead leads to
higher-order cycles and oligomeric assemblies. (Scheme 1b).
The outcome is different when flexible corner units are used,
Received: April 10, 2012
Published: August 8, 2012
© 2012 American Chemical Society
14382
dx.doi.org/10.1021/ja3033197 | J. Am. Chem. Soc. 2012, 134, 14382−14389

Journal of the American Chemical Society
Article
Scheme 1. (a) With Organic Linkers Connecting Two
Identical 17-Base DNA Strands in a 5′-3′ Manner, Assembly
Gives a Flat Dimer; (B) Triphenylene Linker 3 Gives
Different Structures for Different Connectivities; (C)
Flexible Linkers 1 and 2 Assemble to Give Dimers
Regardless of the Connectivity
Chart 1. Four Linkers Studied in This Work: (1) Hexane-
Diol Linker 1 (2) Four-Thymine DNA Linker 2, (3) m-
Triphenylene Rigid Organic Linker 3, (4)
Diphenylphenanthroline Linker 4; (Right) Three Possible
DNA-to-Linker Connectivities, 5′-3′, 5′-5′, and 3′-3′
keeping the DNA strands identical in sequence. Varying the
structure and connectivity of the linker also allows the thermal
denaturation temperature of the same DNA sequence to be
tuned to widely different values.
Fundamentally, this work and the work cited above
emphasize that using common databases to estimate the
melting temperatures of DNA duplexes in more complex
assemblies (based only on sequence and salt conditions) can be
misleading. Depending on their connectivity, duplexes in close
proximity can strongly interact with each other, leading to large
differences in their thermal denaturation behavior from that
estimated for individual duplexes.
RESULTS AND DISCUSSION
■
Dimer Assembly with 5′3′ Connectivity. For this series
of experiments, two DNA duplexes were connected with linkers
1−4, where 1 is a flexible hexane-diol (C6) linker, 2 is a DNA-
based four-thymidine (T4) linker, 3 is a rigid m-triphenylene
organic linker, and 4 is a rigid 2,9-diphenyl-1,10-phenanthroline
linker (Chart 1).17,27 Each linker was attached to two 17-base
single-stranded DNA arms of identical sequence (A′), with 5′3′
connectivity. When combined with its complement (A″), only
one structure was observed via PAGE analysis, with a mobility
corresponding to a dimer (Figure 1). The identity of the bands
in Figure 1 as fully double stranded dimers was confirmed with
control experiments (see Supporting Information [SI]). Due to
the use of identical DNA strands, there can be two possible
assembly outcomes for dimers of 5′3′ connectivitycyclic
dimers (cycA) and flat dimers (flatA) as shown in Scheme 2.
Figure 1. Assembly of 5′3′-connected dimers with each of 4 linkers.
Far left: A′ monomer. Lane 1: hexane diol linker 1. Lane 2: T4 DNA-
based linker 2. Lane 3: Rigid triphenylene linker 3. Lane 4: Rigid
diphenylphenanthroline linker 4.
A number of experiments were used to distinguish which
dimer was formed. First, a FRET experiment was designed.
Strand 2A′ was modified with an Epoch Eclipse quencher at the
3′ end and a Yakima Yellow fluorophore at the 5′ end (F2A′,
Figure 2, see SI for more detail).28 In the flat dimer, the
fluorophore and quencher would be rigidly oriented at opposite
ends of the structure, resulting in unquenched fluorescence. In
contrast, in the cyclic structure, the fluorophore would be
spatially oriented near the quencher, resulting in a decrease in
14383
dx.doi.org/10.1021/ja3033197 | J. Am. Chem. Soc. 2012, 134, 14382−14389

Journal of the American Chemical Society
Article
ligands are oriented toward one anotheras in the flat dimer
Scheme 2. Two Possible Dimer Formations When the 17-
Base DNA Arms Are Identical (A system), a Cyclic (cycA),
and a Flat (flatA) Dimer
(Figure 3a). In the cyclic orientation (Scheme 2), the two
Figure 3. (a) In the flat dimer orientation, the diphenylphenanthroline
moieties are near enough to bind copper(I). (b) Molecular modeling
of the flat dimer indicates that the diphenylphenanthrolines are
positioned such that copper can bind (see SI). (c) Circular dichroism
shows in increase at ∼330 nm, characteristic of copper(I) binding.
phenanthroline linkers are separated by the length of a 17 base
pair duplex, and copper coordination to both would be
unfavorable. The binding of copper [Cu(CH3CN)4][PF6] to
pre-annealed structures was monitored at room temperature
using circular dichroism (CD). Figure 3c clearly shows an
increase in the CD trace at 330 nm, characteristic of copper
binding,29 suggesting flat-dimer assembly. In contrast, when we
generated a flat dimer where the phenanthroline linker is face-
to-face with the 4-thymidine linker 2, no copper coordination
was detected by CD (see SI).
Effect of Linker on Dimer Stability. Having determined
the assembly outcome of 5′3′ linker connectivity, we proceeded
to study the relative effect of each linker on DNA stability.
Results of the thermal denaturation analysis are given in Figure
4 and Table 1. It is clear that linking the two duplexes into a
dimer results in a substantial increase in stability in comparison
Figure 2. FRET measurements indicate that the 5′3′ dimer assembly
most closely resembles a flat dimer. The black curve represents a
control consisting of a rigidifed F2A′ strand. The red curve is the
dimer F2A. The green curve is a dimer of F2A′ and 4A″. The blue
curve is a dimer of F2A′ and 1A″. The yellow curve is a dimer of F2A′
and 3A″. The gray curve is F2A′.
overall fluorescence (Figure 2 inset). Two controls were used in
this experiment. First, the F2A′ strand was hybridized with two
strands that are complementary to its two arms (Figure 2, black
trace). This is expected to separate the FRET pair to a distance
comparable to that of a flat dimer. Second, the F2A′ strand in
its single-stranded form was used to estimate a closer distance
for the FRET pair and increased quenching, due to its flexible
nature (Figure 2, grey trace). Against these two controls, the
17mer strands with 5′3′ connectivity showed higher
fluorescence values in the range expected for a flat dimer, for
all linkers. A decrease in fluorescence intensity with dimers that
place small linkers 1 and 3 face-to-face with the larger T4 linker
in F2A′ is likely due to increased bending as a result of the
linker size mismatch, thus bringing the fluorophore closer to
the quencher. Overall, this first set of experiments is consistent
with formation of flat dimers as the preferred assembly
outcome for the 5′3′ connectivity.
In a second experiment, we attempted to generate an
authentic sample of a cyclic dimer from linker 3 (see below),
using DNA strands of different sequence. Unlike the clean
dimer formation observed earlier (Figure 1, lane 3), this
experiment yielded product mixtures, which further supports
the identity of the structures in Figure 1 as flat dimer.
Figure 4. Thermal denaturation of the A system with each of the four
linkers. The black curve represents a single duplex (control). The red
curve is the dimer with linker 2. The blue curve is the dimer with
linker 1. The yellow curve is the dimer with linker 3, and the green
curve is the dimer with linker 4.
Finally, we took advantage of the metal-binding properties of
linker 4. This linker is capable of binding copper(I) when two
14384
dx.doi.org/10.1021/ja3033197 | J. Am. Chem. Soc. 2012, 134, 14382−14389

Journal of the American Chemical Society
Article
that its two duplexes are melting almost simultaneously. Linkers
3 and 4 can undergo π−π stacking when they are oriented in a
face-to-face manner in the dimer (Scheme 2), thus increasing
the hybridization enthalpy, as well as further mediating the
interaction between the two duplexes and contributing to
allosteric cooperativity. In addition, their rigidity and small size
result in a large degree of preorganization, which contributes to
chelate cooperativity. Because the TM of a duplex is
logarithmically related to the initial concentration of single-
stranded DNA, a ΔTM of 10 °C for 3flatA implies a large
effective concentration of the single-strands of intermediate
structure Ai (estimated to be approximately 800 μM, when 5
μM strands A′ and A″ are used, see SI). The small and rigid m-
triphenylene linker 3 significantly increases chelate coopera-
tivity, by preorganizing the two single strands of intermediate Ai
and confining them to a highly reduced effective volume.
Significant TM increases have been demonstrated for
aggregates linked with multiple duplexes down to a minimum
of three.26 Recently, the groups of Nguyen and Schatz26
demonstrated that trimers linked by aromatic vertices and
assembled in a face-to-face (flat) manner result in higher TM’s
than the same trimers assembled as caged duplexes. The
observed increases were attributed primarily to π−π stacking
interactions between the linkers.
Interestingly, in the present work, we found that flexible
linkers 1 and 2 both significantly raised the melting
temperature of the system over that of the single duplex,
with significant increases in cooperativity. In particular, small
aliphatic linker 1, that is unable to π-stack, shows a significant
increase of 8 °C (Table 1). These results imply that, although
π-stacking of linkers can play an important role, it is not
required for stabilization of two linked duplexes. A compact
linker, even when nonaromatic, can significantly contribute to
both allosteric and chelate cooperativity, resulting in increases
in TM and fwhm.
Table 1. Thermal Denaturation Results for Linkers
ΔTM
(°C)
fwhm
(°C)
ΔH
linker
TM (°C)
(kJ/mol)
single 17mer duplex
64.6 0.8
72.5 0.6
71.6 0.7
0
8.5
4.2
4.6
3.2
4.7
3.1
540
4
1
7.9
780 20
830 20
1050 20
810 10
1250 20
2
7.0
3
74
1
9.6
4
74.6 0.3
79.0 0.5
10.0
14.4
single 34 mer duplex
to a single duplex. For all linkers, we observe a significant
increase in TM, accompanied by a drastic decrease in full width
at half-maximum (fwhm), which is a measure of melting
cooperativity.12,19,21,23 In particular, use of the rigid linkers 3
and 4 result in the greatest increase in TM of up to 10 °C with a
corresponding decrease in fwhm from 8 °C → 3 °C and 8 °C
→ 4.7 °C respectively.
The melting behavior of the two duplexes can be understood
as a composite of two types of phenomena: allosteric and
chelate cooperativity.30 We hypothetically consider the melting
of dimer A, or the reverse association of strands A′ and A″ as a
two-step process, as shown in Scheme 3.
Scheme 3. Hypothetical Two-Step Melting of Flat DNA
Duplex Dimers
Allosteric cooperativity refers to a favorable interaction
between the two duplexes in dimer A, which can make
dissociation of the first duplex (step 1) more difficult than
dissociation of the second duplex (step 2). The small linkers
used in this study appear to allow a favorable interaction
between the two duplexes, such that their melting begins to
resemble that of a 34mer duplex instead of two separate 17mer
duplexes. This can explain both the observed increase in TM
and the narrowing of the melting curve, as measured by the full
width at half-maximum. The first duplex begins denaturation at
a higher temperature, because of the additional stabilization
mediated by the small linker (flatA, Scheme 3). Now the
system is at a much higher temperature than that required to
melt the second duplex, and with the linkers still attached to
already melted strands, the second duplex will denature very
rapidly (step 2, Scheme 3), resulting in a narrow fwhm (Table
1).
Chelate cooperativity can best be understood by following
the reverse association process in Scheme 3. As strands A′ and
A″ come together, a transient structure flatAi is formed.
Association of the second pair of complementary strands in
flatAi is now intramolecular and depends on the effective
concentration of these two strands. This effective concentration
is highly sensitive to the geometry of the linkers between the
duplexes. The smaller and more rigid the linkers, the more
preorganized the two strands in this intermediate structure.
This leads to a higher effective concentration of the two
strands, and a greater proportion of the fully bound dimer A.
Aromatic, rigid linkers 3 and 4 give the greatest stabilization
with ΔTM = 9.6 and 10 °C respectively. 3 is arguably the best,
exhibiting the steepest melting curve (fwhm = 3.2 °C) and
greatest change in enthalpy (ΔH = 1050 kJ mol−1), indicating
Overall, this implies that using small and structurally well-
defined linkers provides a simple method to significantly
increase the stability and melting cooperativity of DNA
duplexes without modifying the sequences of the strands.
Effect of DNA-to-Linker Connectivity on Stability. To
investigate the importance of DNA-to-linker connectivity on
assembly outcome, DNA strands with different connectivities
were made. The connectivity is named according to the DNA
strand-end that is attached to the linker, i.e. 5′5′ refers to a
strand with the directionality 3′-5′-L-5′-3′, where L = 1, 2, or 3
(Chart 1). See Table S1 (SI) for sequences used. Each
complementary pair of strands is capable of assembling into a
wide array of structures, including dimer, trimer, tetramer, etc.
As shown earlier, all molecules with 5′3′ connectivity and
identical strands preferentially assemble into flat dimers. In
contrast, the 5′5′ or the 3′3′ strands are not capable of forming
flat dimers, and because of their strand polarity they only give
cyclic structures when all their available strands are bound.
Thus, in order to better compare these connectivities, we
moved to a system B containing nonidentical strands of similar
melting temperature, such that all connectivities uniformly lead
to cyclic structures when fully bound, and flat dimer is
prevented, as shown in Scheme 4 (see SI for system A with
5′5′/3′3′ connectivity and identical sequences).
First, B′ and B″ strands with the desired connectivities were
annealed, and the resulting structures were analyzed by PAGE
(Figure 5). This shows that flexible linkers 1 and 2 consistently
form cyclic dimers for all connectivities. Interestingly, the
14385
dx.doi.org/10.1021/ja3033197 | J. Am. Chem. Soc. 2012, 134, 14382−14389

Journal of the American Chemical Society
Article
in a flat configuration, no strain would be expected upon
binding. However, by tethering the duplexes at both ends with
structurally small linkers, strain is likely induced, and would be
relieved by over- or under-winding of the duplexes, and/or
fraying of the duplex ends. Scheme 5 shows a schematic model
Scheme 4. Assembly of 17 Base Strands with Flexible
Linkers; Each DNA Arm Is Unique and Flexible Linkers
Assemble to Form Cyclic Dimers
a
Scheme 5. (a) 17bp System; (b) Optimized Models for a
b
Cyclic Dimer (17 Bases Per Duplex, with Linker 3)
Figure 5. PAGE analysis of dimers of linkers 1 and 2 with each
possible connectivity. Lanes: (1) 1A 5′3′, (2) 2A 5′3′, (3) 1A 5′5′, (4)
2A 5′5′, (5) 1A 3′3′, (6) 2A 3′3′.
a
The red DNA strand-ends are ideally aligned at the top of the duplex.
Underneath, directional arrows show how the DNA strand-ends would
align (blue DNA). Models minimized using AMBER forcefield,
b
product distribution for linker 3 changes dramatically with
differing connectivities, which will be discussed later. Thermal
denaturation information for DNA-to-linker connectivity
studies (Table 2) will thus focus on linkers 1 and 2. Because
Hyperchem 8.0.
of the strand-end alignment, illustrating how two linkers would
subtend the top and bottom duplex ends (circles with arrows).
Each duplex was rotated about its long axis such that the top
linker could easily bridge the distance between the two
duplexes. A bottom view shows how the DNA strand-ends
would then be oriented for binding the second linker. For 17bp
duplexes (Scheme 5), these models show that 5′3′ connectivity
possesses a better DNA strand-end alignment for linker
binding. Conversely, 3′3′ and 5′5′ dimers would need to
distort to allow the linker to bridge the duplex ends. Further
evidence for the role of strand-end alignment is given below,
using the rigid linker 3.
Effect of Rigid Linker and DNA-to-Linker Connectivity
on DNA Self-Assembly. Based on the strand-end alignment
problem, a greater degree of duplex strain and distortion is
expected for the 5′5′ and 3′3′ connectivity, as is manifested by
the thermal denaturation data and PAGE analysis for linkers 1
and 2. Both linkers are relatively flexible, and would be
conformationally mobile enough to somewhat relieve this
strain, leading to the high dimer yields in Figure 5. On the
other hand, linker 3 is much more rigid and would therefore
not be as readily distorted to relieve this strain. 3 is smaller than
any previously reported synthetic linker, allowing it to hold the
duplexes closer together and increasing their interactions.
To promote formation of cyclic structures for all
connectivities of linker 3, we again constructed a B system
with two distinct 17 base arms (see SI for PAGE and thermal
denaturation results for connectivities with identical DNA
strands).
Table 2. Effect of Connectivity on Thermal Stability
fwhm
(°C)
ΔH
Linker
Connectivity
TM (°C)
(kJ/mol)
single duplex CA
single duplex CB
1B
N/A
N/A
5′3′
5′5′
3′3′
5′3′
5′5′
3′3′
64.6 0.8
61.6 0.3
67.5 0.2
67.2 0.6
65.6 0.3
67.4 0.4
66.6 0.2
62.9 0.1
7.8
8.2
472
467
6
4
4.5
580 10
5.4
400
518
7
9
5.9
2B
4.6
590 20
6.0
321
359
6
5
>7.8
the two DNA strands in B′ and B″ have different sequences,
the results of Table 2 should be considered qualitatively (SI
provides the full thermal denaturation results for the 5′5′ and
3′3′ connectivities with identical DNA strands).
As the connectivity is changed from 5′3′ to 5′5′ and 3′3′, a
steady decrease in TM, increase in fwhm, and decrease in ΔH
are observed, to a point where the 3′3′ system melts more like
two independent duplexes (see also SI). This decrease in
stability is also apparent in the PAGE analysis in Figure 5. In
the 5′3′ connectivity, formation of dimer is quantitative for
both linkers. However, both 3′3′ and 5′5′ connectivities are
lower yielding, with some nonhybridized monomers and
higher-order structures as side products.
One factor that can possibly contribute to the destabilization
of some of the cyclic dimers is DNA strand-end alignment. If
the two duplexes were only connected by a linker at one end, or
Interestingly, we find that linker 3 gives a dramatically
different assembly outcome, as compared to the more flexible
14386
dx.doi.org/10.1021/ja3033197 | J. Am. Chem. Soc. 2012, 134, 14382−14389

Journal of the American Chemical Society
Article
linkers 1 and 2 (Scheme 6). PAGE analysis (Figure 6) shows
that the 5′5′ connectivity for 3 gives no dimers. Instead, larger,
Scheme 6. (Top) Assembly Outcome for the Rigid
Triphenylene Linker 3 in the 17-Base System; Different
Connectivities Result in Different Structural Distributions;
(Bottom) Assembly Outcome for the Rigid Triphenylene
Linker 3 in the 13 Base System; Structural Distribution Has
Changed
Figure 6. Structural distribution for assemblies of rigid vertex 3, with:
Lanes 1, 5′5′ connectivity. Lanes 2, 5′3′ connectivity. Lanes 3, 3′3′
connectivity. Top system B with distinct DNA 17 base arms. Bottom
system B with 13 base arms.
nongel penetrating oligomers are observed. The result for 3′3′
shows a well-resolved ladder of distinct structures ranging from
dimer to octamer with the bulk of the assemblies (∼40%)
present as nonpenetrating oligomer. The characterization of the
discrete structures as cyclic, closed assemblies is supported by
experiments shown in the SI. The 5′3′ connectivity for linker 3
gives a number of cyclic structures, although dimer is the main
product (∼30%), and very little nongel-penetrating oligomer is
formed. Linker 3 thus appears to favor the formation of higher-
order oligomers and prevents the formation of small cyclic
assemblies for the 5′5′ and 3′3′ connectivities. While it gives a
more significant proportion of dimer for the 5′3′ connectivity,
the assembly outcome is very different from the flexible linkers
1 and 2, which result in dimers for all connectivities.
Scheme 7. Strand-End Alignment Analysis Predicts That 3′3′
DNA-to-Linker Connectivity Will Be Favored in a 13bp
System
To explain these findings, we performed preliminary
molecular modeling studies of the cyclic dimers with linker 3
(Scheme 5b). These suggest that the 5′3′ connectivity results in
less duplex distortion than the 5′5′ and 3′3′ dimers, consistent
with better strand-end alignment (Figure 5b). However there is
still fraying of the duplexes with this more favorable 5′3′
connectivity, due to the reduced size and lack of conformational
mobility in the triphenylene linker 3. These studies confirm
that, by choosing a specific linker structure and connectivity,
one can dramatically change DNA self-assembly.
To further investigate how strand-end alignment influences
self-assembly products, a new, shorter system was developed,
SB, this time with 13 base-pairs per duplex, reducing the total
number of helical turns (Schemes 6 and 7). The 13bp duplex
length was predicted to show optimal strand-end alignment for
the 3′3′ connectivity, rather than the 5′3′ connectivity
opposite the trend of the 17bp rigid linker system (Scheme 5a).
Again, the rigid linker 3 was used since it is sensitive to
changes in connectivity. On the basis of simple models
(Scheme 7), the 3′3′ connectivity appears most favorable in
allowing the linkers to bridge the top and bottom strand-ends.
If DNA strand-end orientation is indeed a driving factor in
determining self-assembly products, a shift in product
distribution toward smaller structures should be observed for
14387
dx.doi.org/10.1021/ja3033197 | J. Am. Chem. Soc. 2012, 134, 14382−14389

Journal of the American Chemical Society
Article
this connectivity. Interestingly, a significant shift is observed. By
changing the number of bases, product distribution was
changed from higher-order structures to only two small
products, namely the dimer and the tetramer (Figure 6,
bottom, Scheme 6). In addition, where before only 17% of the
structures formed were dimer and the major product (41%)
was oligomeric, 3′3′ dimer is now one of the two principal
products (55% yield). Conversely, where 5′3′ connectivity
previously yielded smaller cyclic structures, we now see the
defined structures disappear completely, leaving only oligo-
meric assemblies. The assignment of the gel electrophoresis
bands as closed dimers and tetramers is supported by a number
of experiments that are described in the Supporting
Information.
This data supports the conclusion that, for the small linker
systems, strand-end-orientation is an important factor in
determining which DNA-to-linker connectivity provides the
greatest increase in stability. This can be readily predicted by
simple modeling such as that shown in Schemes 5 and 7.
Understanding the effect of DNA-to-linker connectivity
provides an extra dimension of tunability when designing
DNA structures. Connectivity can be used to fine-tune Tm by
an additional 7 °C, as well as provide a method to obtain
different self-assembled products, such as dimers or tetramers
rather than oligomers.
AUTHOR INFORMATION
Corresponding Author
hanadi.sleiman@mcgill.ca
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank FQRNT, NSERC, CFI, CIHR, and CSACS for
financial support, and Faisal Aldaye for the basic design of the
symmetrical assembly system.
REFERENCES
■
(1) Lin, C.; Liu, Y.; Rinker, S.; Yan, H. ChemPhysChem 2006, 7,
1641−1647.
(2) Seeman, N. C. J. Theor. Biol. 1982, 99, 237−247.
(3) Sobey, T. L.; Renner, S.; Simmel, F. C. J. Phys.: Condens. Matter
2009, 21, 034112.
(4) Andersen, E. S.; Dong, M.; Nielsen, M. M.; Jahn, K.; Lind-
Thomsen, A.; Mamdouh, W.; Gothelf, K. V.; Besenbacher, F.; Kjems,
J. ACS nano 2008, 2, 1213−1218.
(5) Andersen, E. S.; Dong, M.; Nielsen, M. M.; Jahn, K.; Subramani,
R.; Mamdouh, W.; Golas, M. M.; Sander, B.; Stark, H.; Oliveira, C. L.
P.; Pedersen, J. S.; Birkedal, V.; Besenbacher, F.; Gothelf, K. V.; Kjems,
J. Nature 2009, 459, 73−U75.
(6) Yang, H.; Altvater, F.; de Bruijn, A. D.; McLaughlin, C. K.; Lo, P.
K.; Sleiman, H. F. Angew. Chem., Int. Ed. 2011, 50, 4620−4623.
(7) McLaughlin, C. K.; Hamblin, G. D.; Sleiman, H. F. Chem. Soc.
Rev. 2011, 40, 5647−5656.
(8) Aldaye, F. A.; Sleiman, H. F. J. Am. Chem. Soc. 2007, 129, 13376−
13377.
(9) Shi, J.; Bergstrom, D. E. Angew. Chem., Int. Ed. 1997, 36, 111−
113.
(10) Zimmermann, J.; Cebulla, M. P. J.; Monninghoff, S.; von
Kiedrowski, G. Angew. Chem., Int. Ed. 2008, 47, 3626−3630.
(11) Scheffler, M.; Dorenbeck, A.; Jordan, S.; Wustefeld, M.; von
Kiedrowski, G. Angew. Chem., Int. Ed. 1999, 38, 3311−3315.
(12) Eryazici, I.; Prytkova, T. R.; Schatz, G. C.; Nguyen, S. T. J. Am.
Chem. Soc. 2010, 132, 17068−17070.
(13) Yoshizawa, M.; Fujito, M. Pure Appl. Chem. 2005, 77, 1107−
1112.
(14) Gibbs, J. M.; Park, S.-J.; Anderson, D. R.; Watson, K. J.; Mirkin,
C. A.; Nguyen, S. T. J. Am. Chem. Soc. 2005, 127, 1170−1178.
(15) Lo, P. K.; Karam, P.; Aldaye, F. A.; McLaughlin, C. K.; Hamblin,
G. D.; Cosa, G.; Sleiman, H. F. Nat. Chem. 2010, 2, 319−328.
(16) Seeman, N. C. Nature 2003, 421, 427−431.
(17) Yang, H.; McLaughlin, C. K.; Aldaye, F. A.; Hamblin, G. D.;
Rys, A. Z.; Rouiller, I.; Sleiman, H. F. Nat. Chem. 2009, 1, 390−396.
(18) Yang, H.; Metera, K. L.; Sleiman, H. F. Coord. Chem. Rev. 2010,
254, 2403−2415.
(19) Jin, R.; Wu, G.; Li, Z.; Mirkin, C. A.; Schatz, G. C. J. Am. Chem.
Soc. 2003, 125, 1643−1654.
(20) Long, H.; Kudlay, A.; Schatz, G. C. J. Phys. Chem. B 2006, 110,
2918−2926.
(21) Lytton-Jean, A. K. R.; Gibbs-Davis, J. M.; Long, H.; Schatz, G.
C.; Mirkin, C. A.; Nguyen, S. T. Adv. Mater. 2009, 21, 706−709.
(22) Park, S. Y.; Gibbs-Davis, J. M.; Nguyen, S. T.; Schatz, G. C. J.
Phys. Chem. B 2007, 111, 8785−8791.
(23) Gibbs-Davis, J. M.; Schatz, G. C.; Nguyen, S. T. J. Am. Chem.
Soc. 2007, 129, 15535−15540.
(24) Kudlay, A.; Gibbs, J. M.; Schatz, G. C.; Nguyen, S. T.; Olvera de
la Cruz, M. J. Phys. Chem. B 2007, 111, 1610−1619.
(25) Stepp, B. R.; Gibbs-Davis, J. M.; Koh, D. L. F.; Nguyen, S. T. J.
Am. Chem. Soc. 2008, 130, 9628−9629.
(26) Eryazici, I.; Yildirim, I.; Schatz, G. C.; Nguyen, S. T. J. Am.
Chem. Soc. 2012.
CONCLUSIONS
■
We have shown a simple method to control both the stability
and the self-assembly behavior of DNA structures. By using
small, synthetic linkers that connect two adjacent duplexes,
factors such as linker size, rigidity and connectivity can increase
the thermal denaturation temperature of 17-base pair duplexes
by up to 10 °C and significantly narrow the melting profile of
the two duplexes. For the same DNA sequence, one can now
tune the melting temperature to vastly different values by
selecting the linker structure and DNA-to-linker connectivity.
Furthermore, a small rigid linker can be used to directly affect
the self-assembly product distribution. Because of the strict
requirements that it imposes, subtle changes in the orientation
of the linked strands (e.g., 5′3′ vs 3′3′) can now lead to
dramatic changes in the self-assembly behavior. These
variations can be readily predicted using a simple strand-end
alignment model.
̈
̈
Incorporation of these linkers into DNA strands is a very
simple, on-column, and high-yielding process. We anticipate the
usefulness of this method in DNA nanotechnology, where the
melting temperature of the same DNA sequence can now be
rationally varied and controlled with simple structural
modifications of the duplex linkers.
Fundamentally, this study contributes further insight into
DNA interduplex interactions, which are important in
chromosome packaging and homologous recombination. In
addition, the large increase in stability and melting cooperativity
of short duplexes will find a number of applications in
biotechnology, such as in more sensitive DNA detection and
diagnostics.
ASSOCIATED CONTENT
■
S
* Supporting Information
Detailed methods, DNA sequences, quantitative analysis,
additional figures. This material is available free of charge via
the Internet at http://pubs.acs.org.
(27) Aldaye, F. A.; Sleiman, H. F. J. Am. Chem. Soc. 2007, 129,
10070−10071.
14388
dx.doi.org/10.1021/ja3033197 | J. Am. Chem. Soc. 2012, 134, 14382−14389

Journal of the American Chemical Society
Article
(28) Lukhtanov, E. A.; Metcalf, M.; Reed, M. W. Am Biotechnol Lab
2001, 19, 68−69.
(29) Yang, H.; Altvater, F.; de Bruijn, A. D.; McLaughlin, C. K.; Lo,
P. K.; Sleiman, H. F. Angew. Chem. 2011, 50, 4620−4623.
(30) Hunter, C. A.; Anderson, H. L. Angew. Chem., Int. Ed. 2009, 48,
7488−7499.
14389
dx.doi.org/10.1021/ja3033197 | J. Am. Chem. Soc. 2012, 134, 14382−14389