Amplified Self-Immolative Release of Small Molecules by Spatial Isolation of Reactive Groups on DNA-Minimal Architectures

Article abstract of DOI:10.1002/anie.202001123

Triggering the release of small molecules in response to unique biomarkers is important for applications in drug delivery and biodetection. Due to low quantities of biomarker, amplifying release is necessary to gain appreciable responses. Nucleic acids have been used for both their biomarker-recognition properties and as stimuli, notably in amplified small-molecule release by nucleic-acid-templated catalysis (NATC). The multiple components and reversibility of NATC, however, make it difficult to apply in vivo. Herein, we report the use of the hybridization chain reaction (HCR) for the amplified, conditional release of small molecules from standalone nanodevices. We couple HCR with a DNA-templated reaction resulting in the amplified, immolative release of small molecules. We integrate the HCR components into single nanodevices as DNA tracks and spherical nucleic acids, spatially isolating reactive groups until triggering. This could be applied to biosensing, imaging, and drug delivery.

Full text of DOI:10.1002/anie.202001123

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Amplified Self-Immolative Release of Small Molecules by Spatial
Isolation of Reactive Groups on DNA-Minimal Architectures
Authors: Hanadi Farouk Sleiman, Alexander Prinzen, Daniel Saliba,
Christopher Hennecker, Trinh Tuan, and Anthony Mittermaier
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202001123
Link to VoR: https://doi.org/10.1002/anie.202001123

10.1002/anie.202001123
Angewandte Chemie International Edition
RESEARCH ARTICLE
Amplified Self-Immolative Release of Small Molecules by Spatial
Isolation of Reactive Groups on DNA-Minimal Architectures
Alexander L. Prinzen, Daniel Saliba, Christopher Hennecker, Tuan Trinh, Anthony Mittermaier and
Hanadi F. Sleiman*
require separate components that are able to reversibly bind the
target template in order to generate turnover, making it primarily
feasible in vitro.[6] Due to differences in pharmacokinetics and
tissue distribution of different components, it would be desirable
to combine the amplification system into a single nanodevice for
in vivo applications.
Integrating reactive functional groups into a single device,
however, poses the significant challenge of preventing the groups
from reacting non-specifically until they are triggered. Some
natural systems have evolved unique mechanisms to spatially
isolate different reactive groups within the same structure.[7]
Enzymes such as plasminogen[8] and vascular endothelial growth
factors C & D[9], have precisely placed intramolecular reactive thiol
and disulfide moieties, which are brought in close proximity upon
allosteric recognition. In DNA nanotechnology, reactive functional
groups can be effectively separated for sequential transfer
reactions in the form of a DNA walker.[10]
Abstract: Triggering the release of small molecules in response to
unique biomarkers is important for applications in drug delivery and
biodetection. Due to typically low quantities of biomarker, amplifying
release is necessary to gain appreciable responses. Nucleic acids
have been used for both their biomarker recognition properties and as
stimuli, notably in amplified small molecule release via nucleic acid-
templated catalysis (NATC). The multiple components and
reversibility of NATC, however, make it difficult to apply in vivo. Here,
we report the first use the hybridization chain reaction (HCR) for the
amplified, conditional release of small molecules from standalone
nanodevices. We first couple HCR with a DNA-templated reaction
resulting in the amplified, immolative release of small molecules.
Moreover, we integrate the HCR components into single nanodevices
as DNA tracks and spherical nucleic acids, spatially isolating reactive
groups until triggering. Overall, this work translates the amplification
of HCR into small molecule release without the use of multiple
components, and will aid its application to biosensing, imaging and
drug delivery.
In addition to DNA-templated chemical reactions,
amplification methods that rely on toehold mediated strand
displacement have been extensively investigated.[11] For example,
enzyme-free isothermal amplification methods, such as the
12]
hybridization chain reaction (HCR)[4a,
or catalytic hairpin
assembly (CHA)[4b] rely on metastable hairpins for amplification.
Furthermore, the different component molecules necessary for
these methods have been combined into single metastable
nanodevices using DNA nanotechnology approaches.[13] HCR in
particular has been used for bio-sensing assays[14], bio-imaging[15]
bio-medicine[16], and small molecule synthesis[17]. While previous
HCR-small molecule hybrids have been generated that release
through non-specific hydrolysis[16b] or diffusion[16a, 18], to our
knowledge HCR has not been developed for the conditional
release of small molecules.
Here, we set out to develop the hybridization chain reaction
for the conditional, amplified release of pro-functional small
molecules. Combining HCR with a DNA-templated reaction and
strategically positioning reactive functional groups on each hairpin
allows these reactive groups to be kept separated under high
dilution conditions. Once polymerization is induced, the reactive
groups increase in effective concentration and are able to trigger
Introduction
,
Concealing the activity of functional small molecules, such
as fluorescent or therapeutic structures, and triggering their
release in active form in response to a biological stimulus has
important applications in a variety of different areas including
smart therapeutics, biological detection and bio-imaging.[1]
Nucleic acids are favorable biological molecules to trigger the
release of pro-functional small molecules, because of their protein
recognition properties (e.g., aptamers) and unique spatiotemporal
expression profiles (e.g., mRNA, miRNA). They have thus been
used as both triggers and recognition components in a variety of
stimuli-responsive nanodevices.[2] However, many of these
devices respond to a stoichiometric amount of stimulus, and thus,
the typically small amounts of biological targets present in vitro or
in vivo would not be sufficient to elicit a sizable response.[3]
To address this issue, many amplification methods have
been developed.[4] Notably, the release of functional small
molecules via nucleic acid-templated catalysis has emerged as a
powerful tool to amplify nucleic acid signals into detection and
drug release responses.[4c, 5] This approach uses two DNA-small
molecule conjugates, which perform a DNA-templated reaction
with each other upon binding a nucleic acid target. The reaction
releases the small molecule payload from the conjugate, allowing
it to elicit its function.[5b] While this is a powerful method, it does
a
self-immolative mechanism, releasing functional small
molecules in an amplified manner.
Importantly, this method operates using toehold-mediated
strand displacement with metastable hairpins, allowing their
combination into complex structures. By using different DNA
nanostructures, such as polymerizable tracks and spherical
nucleic acids, we show that both HCR hairpin components can be
integrated into a single structure, while effectively separating
reactivity until triggering. Using DNA nanotechnology allows for
the organization of different functional components with high
precision and structural control.
In this contribution, HCR is the tool used to achieve
amplification, but the innovation in our approach lies in the spatial
separation of chemically reactive functional groups within the
same nanostructure, for triggered amplified release of small
molecules. This required careful synthetic and kinetic studies,
which will be valuable for the diagnostic and therapeutic
communities. We anticipate that the development of these
systems can lead to new analytical tools for biosensing and
[*]
A. L. Prinzen, D. Saliba, C. Hennecker, T. Trinh, Prof. A. Mittermaier,
Prof. H. F. Sleiman
Department of Chemistry, McGill Univerisity
801 rue Sherbrooke West, Montreal, QC, H3A 0B8 (Canada)
E-mail: hanadi.sleiman@mcgill.ca
Supporting information for this article is given via a link at the end of the
document.
1
This article is protected by copyright. All rights reserved.

10.1002/anie.202001123
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 1. (a) Design of H1-Fl/ H1-CT and H2-SH for HCR (b) Mechanism of HCR resulting in the release of pro-functional molecules: i. initiator hybridizes to the
toehold domain of H1 and performs a toehold mediated strand displacement. ii. Loop domain of H1 hybridizes to the toehold domain of H2 and performs a toehold
mediated strand displacement. iii. DNA templated reaction occurs, releasing the pro-functional molecule connected to a biodegradable linker. iv. Immolative linker
degrades to release the functional molecule in its active form. v. The HCR process repeats polymerizing H1 and H2, amplifying the release of pro-functional molecule.
(c) Mechanism of DNA templated disulfide exchange, resulting in the release of the pro-functional molecule. (d) Mechanism of immolative linker degradation,
resulting in the release of active molecule.
bioimaging as well as stimuli-responsive drug delivery vehicles.
Our approach is a combination of two areas of research: DNA-
templated catalysis and DNA-based amplification methods like
HCR. In principle, it can be applied to any DNA based toehold
mediated amplification methods such as CHA.[4b]
stem domain and revealing its loop domain (Figure 1b-ii). This
process repeats until all of H1 and H2 are consumed, generating
amplification with long polymers of H1 and H2 (Figure 1b-v).
To address the second process, each of the hairpins in the
HCR process was modified with unique design features. To the
stem domain of H1 was added a reactive group (E), connected
through a linker group (R) to the small functional molecule (M)
(Figure 1a.). H2 was modified between the stem and loop
domains with another reactive group (N). The groups (E) and (N)
serve the purpose of reacting together in a bond cleavage process.
After polymerization these groups would be strategically placed in
close proximity on each monomer to increase their effective
concentration and react as a DNA templated reaction (Figure 1b-
iii). We chose to use a DNA-templated sulfide nucleophilic attack
for the bond cleavage reaction (Figure 1b). DNA-templated thiol-
disulfide exchange has typically been used as a DNA ligation
reaction[19], but also in nucleic acid templated catalysis[20]. In our
design, the (E) domain of H1 is a disulfide that can be cleaved by
the thiol group (N) of H2, releasing the (R) and (M) groups.
Interestingly, this DNA-templated reaction is accompanied by
covalent bond formation between (E) and (N) in the resulting
polymer (Figure 1c). To our knowledge this would be the first time
that HCR is accompanied by component ligation. In addition to
providing an additional method to monitor the process using
Results and Discussion
Design, Synthesis and Characterization of HCR. In order
to use HCR for conditional release, two processes are required,
1) HCR polymerization for amplification and 2) a DNA mediated
bond cleavage reaction that occurs after polymerization to liberate
the small molecules.
In the simplest version of HCR there are two hairpins (H1 &
H2) that are designed to have a toehold domain (T), a stem
domain (S) and a loop domain (L). Both the toehold and loop
domains on opposite hairpins are complementary to each other,
but are kinetically trapped and do not hybridize (Figure 1a.). The
HCR process begins when an initiator strand (I) is introduced,
binding to the toehold domain of H1, displacing the stem domain,
and revealing the loop domain (Figure 1b-i). From here the loop
domain of H1 binds to the toehold domain of H2, displacing its
2
This article is protected by copyright. All rights reserved.

10.1002/anie.202001123
Angewandte Chemie International Edition
RESEARCH ARTICLE
denaturing gel electrophoresis (see below), this results in a more
stable, less dynamic polymer which makes it a useful material,
rather than only a side-product.
hybridizing H2-SH (H2-2-SH), resulting in minimal release (Figure
S12 and Figure S14). Therefore, it was determined that leakage
is primarily due to spontaneous polymerization between H1-FL
and H2-SH when no initiator is present. Leakage was further
reduced to <10% over 24hr in the second generation by
increasing the stem length of the hairpins from 12 to 18 bases,
(see below).
In addition to the bond cleavage reaction, we also required
a method to conceal and then recover the activity of the small
molecule (M). Therefore, we introduced a self immolative linker
(R) between (E) and (M). Using the self-immolative linker (R)
gives us a traceless degradable linker that is triggered by the bond
cleavage reaction (Figure 1b iv). More specifically, while (M) and
(R) are connected to (E) its activity is sequestered until the DNA
templated bond cleavage reaction occurs and the self immolative
linker undergoes a series of intramolecular cyclizations to release
the small molecule (M), recovering its activity (Figure 1d). We
introduced an extra diethylamine linkage into (R) to form
carbamate linkages, as opposed to carbonate linkages, as this
had been shown previously to reduce non-specific hydrolysis of
the functional molecule from the linker.[21] To monitor the extent
and rate of small molecule release in real time by fluorescence
spectroscopy we initially chose (M) to be a methoxyfluorescein
molecule (Figure S1), that is non-emissive when connected to the
self immolative linker.[22]
Modified H1 and H2 strands were synthesized, and reacted
to generate methoxyfluorescein conjugated H1 (H1-FL) and thiol
modified H2 (H2-SH). These strands were then characterized by
gel electrophoresis and LC/MS (Detailed synthesis and
characterization SI-VI.). For H2-SH, it was non-trivial to install an
internal thiol modification, as complex synthesis and purification
factors must be considered and there are limited commercial
reagents available (Detailed in SI-VI.). Briefly, to overcome these
challenges we functionalized H2 with a serinol amino modification
in the desired location and labelled it on the solid support with an
activated ester molecule containing a disulfide, before the final
deprotection. Stability of the H1-FL strand was confirmed by
fluorescence and high performance liquid chromatography
(HPLC), with no appreciable degradation over 24 hrs (Table S3
and Figure S14).
We are able to monitor the system using three methods 1)
the degree of monomer consumption, as determined by native
agarose gel electrophoresis gel electrophoresis (AGE), which
informs on the efficiency and rate of the HCR process (Figure 2a
and Figure 2c) 2) the degree of ligation as determined by
denaturing AGE, which informs on the yield and kinetics of
release of the pro-methoxyfluorescein molecule (Figure 2b and
Figure 2c) and 3) fluorescence intensity increase of the
fluorophore, which informs on the yield and rate of the self-
immolative cyclization’s to produce the active molecule (Figure
2c). First, to determine the extent of amplification, we performed
our polymerizations at a constant hairpin concentration of 750nM,
while varying the amount of initiator between 0 and 1 equiv. (Table
S3, Figure S12 and Figure S14).
In order to verify that our system follows the intended HCR
mechanism (Figure 1b), the kinetics of assembly were analyzed
according to the scheme shown in Figure 2e. The initiator, I, and
polymer I(H1H2)n bind to hairpin H1, and the IH1 dimer and
polymer I(H1H2)nH1 bind to H2 according to the second-order
rate constant k1. These steps were assumed to have identical
kinetics, since the toe-hold and stem loop regions are identical in
H1 and H2. The ligation of adjacent H1 and H2 fragments (and
concomitant release of the pro-functional molecule) obeyed first-
order kinetics with rate constant k2. The immolative cyclization
reaction was described with the first-order rate constant k3.
Theoretical kinetic traces were obtained by numerically
integrating the coupled differential equations (see Supporting
Information). The native AGE, denaturing AGE, and fluorescence
data were fit to this kinetic scheme by non-linear least squares
minimization, yielding the rate constants k1, k2, and k3. In addition,
we found it necessary to account for a small amount of unreacted
material present at the end of the reactions, which can be
attributed to misfolded secondary structures, spurious synthesis
errors, and/or stoichiometric concentration differences between
the two DNA strands (Figure 2a).[24] We therefore included an
adjustable parameter in the fits describing the concentration of
inactive or “misfolded” H1 and H2 strands and assumed equal
quantities of both.
Kinetics were assessed at 0.1eq of initiator at 25°C, and the
calculated curves were in excellent agreement with the
experimental data, giving us a high degree of confidence that this
system follows an HCR mechanism (Figure 2d). The total amount
of misfolded material was found to be 144 ± 5 nM, reducing the
effective starting concentration of each hairpin from 750 nM to
about 606 nM. The association rate constant (k1) was found to be
1·104 ± 1·103 M-1 s-1. This is slightly slower than reported in
previous studies of hairpin-hairpin interactions, possibly due to
the additional functional groups present in the H1-FL and H2-SH
strands (Figure 2d and Figure S17).[25] We found that the ligation
step was much faster than H1/H2 association, meaning that the
rate of pro-functional molecule release is governed almost
exclusively by k1. It was therefore not possible to extract a precise
value for k2, though a lower bound of ≳1·10-3 s-1 was obtained.
Finally, the rate of the immolative cyclisation’s (k3) was found to
be 2.1·10-4 ± 3·10-5 s-1, making it the slowest step of the reaction.
It is likely also the slowest step in templated HCR, as templation
can accelerate HCR by two orders of magnitude.[13d]
It was found that between 1 and 0.1 equiv. of initiator, our
system polymerized to ≈90% (Figure 2a). At 0.1 equiv. of initiator
the polymerization afforded approx. 640 nM of released
methoxyfluorescein which is significantly more than the
concentration of initiator (75 nM), indicating amplified release
(Figure 2c and Figure S14). This represents an 85% yield over
three steps (Polymerization, ligation & linker degradation) with
respect to the concentration of H1-FL (Figure 2c). Even at 25nM
initiator concentration we were able to release 340 nM of
methoxyfluorescein which approaches the concentration of some
relevant biomarkers (Figure S14).[23] The ligation product of H1-
FL and H2-SH was further confirmed by LC-MS (Figure S15). We
confirmed that the disulfide exchange reaction was still viable over
time and that the overall yield was not reduced due to non-specific
thiol oxidation by performing a series of polymerizations initiated
at different time points (Figure S16).
To showcase the versatility of our system to release
different functional molecules we generated another H1 hairpin
making (M) a camptothecin prodrug (H1-CT) (Figure S2 for
detailed synthesis and structure). The amount of release from H1-
CT, was monitored through native/denaturing agarose and HPLC
analysis. Release was found to be comparable the H1-FL
releasing 80-95% of the camptothecin prodrug (Figure S18 and
Table S6).
Looking towards future applications of our system, we
analysed the release of small molecule from H1-FL in response
to biologically relevant reducing conditions. We exposed H1-FL to
both extracellular (0.01mM) and intracellular (1mM), levels of free
thiols, in the form of DTT.[26] Release was monitored by
fluorescence and HPLC analysis. Minimal release was found
when H1-FL was exposed to extracellular levels of reducing agent
over 24hrs, while intracellular levels of reducing agent cleaved the
small molecule from H1-FL (Figure S19 and Table S7). Given the
wealth of HCR sequences which respond to various extracellular
signals through the use of aptamers, we anticipate that our
system could operate through extracellular recognition and
release events followed by drug internalization.[16b, 27] This may be
For the sequences used, circuit leakage at 0 equiv. of
initiator was found to be only 5% after 1hr up to a maximum of
approx. 20% over 24hrs (Figure 2c and Figure S14). We
determined that circuit leakage was not due to non-specific
reactions between H1-FL and H2-SH by exposing H1-FL to a non-
3
This article is protected by copyright. All rights reserved.

10.1002/anie.202001123
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 2. Characterization of HCR between H1-FL and H2-SH after 24hr, by (a) Native agarose and (b) Denaturing agarose: L: Ladder, Lane 0: H1-FL + H2-SH,
Lane 0.1: Lane 0 + 0.1 equiv. I-1, Lane 1: Lane 0 + 1 equiv. I-1. d. (c) Degree of monomer consumption, ligation and released methoxyfluorescein after 24 hr, of
H1-FL and H2-SH with 0, 0.1 and 1 equiv. I-1. (d) HCR kinetics, experimental data is represented as circles and simulated kinetics are solid lines. Error bars
correspond to the standard deviation of triplicate measurements. Native-AGE (green) is reported as the average value of H1 and H2 strands, denatured-AGE (blue)
is reporting on ligated H1H2 complexes, fluorescence measurements (red) report on the completion of immolative cyclization. (e) Sequential mechanism of the HCR
reaction with ligation and immolative cyclization steps, the colours of each species correspond to their traces in d, species that were not measured are represented
in grey (see SI-VIII. Figure S17 for more detail.).
a more viable method without the need to internalize the DNA
component which is always a challenge for oligonucleotide
therapeutics.[28]
to hybridize to a DNA scaffold (B’) via adapter strands S1 & S2
with sequence (A-B) (Figure 3a.). Sequence’s B1 and B2 were
designed to be two turns of DNA, orienting the hairpins on the
scaffold in the same direction. Additionally, we added a spacer
region (X) next to the loop region (L) of the hairpins as this had
been shown previously to allow templated HCR to occur without
steric constraints.[13c] We calculated that, for our system, adding
11 T spacers was sufficient to allow the loop region of one hairpin
unit to hybridize with the overhang region of the next hairpin unit,
once initiated (Figure 3a).
Polymerization of the single unit occurs by an intermolecular
initiation of H1 (Figure 3d-i) on the template followed by an
intramolecular hybridization to H2 (Figure 3d-ii). Here, a nucleic
acid templated reaction can occur, releasing the cargo (Figure 3d-
iii). Additionally, H2 is available for an intermolecular hybridization
to H1 on another unit, the whole process then repeats growing the
polymer (Figure 3d-v and Figure 3d-vi).
For this design two types of DNA templates were generated
for HCR, one with blunt ends on the template (BBB) and the other
with short sticky ends (SBB) (Figure 3b). Inspired by the
supramolecular scaffold method of localized HCR, we
hypothesized that the short sticky ends could improve
propagation of the HCR unit, by reversibly associating the
individual units together, giving overall enhanced kinetics for
release.
The modified hairpins were synthesized using the general
procedure to generate H1-2-FL and H2-2-SH (Detailed in SI-VI.).
Before moving forward with the templated HCR, we first examined
the effect of the changes made to H1-FL and H2-SH had on
leakage and polymerization without using a template (NT). By
increasing the stem (S) of these hairpins from 12 to 18 bps, the
resulting leakage was ≤10% over 24hrs. Next, we introduced the
hairpins to 1 equiv. (750nM) of initiator strand (I-2) and found that
over 24 hrs, approximately 560 nM of the conjugate was released,
representing a 75% yield over 3 steps. With 0.1 equiv. of initiator
(75 nM) approximately 490 nM of the conjugate was released,
indicating that amplification was once again achieved, and
Design, Synthesis and Characterization of DNA
Templated HCR. Having shown that we could indeed amplify the
release of functional molecules using HCR, we then investigated
whether both components could be integrated into a single
nanodevice. Localized HCR has been recently shown to
accelerate the rate of HCR on single stranded and DNA origami
templates.[13c, 13d] This strategy operates by anchoring the hairpins
to a single scaffold, increasing their effective concentration. A
major advantage of anchoring the hairpins is that a robust
nanostructure is formed which is more desired for in vivo
applications. However, the hairpins of the circuit in this previous
strategy are all unique in sequence and required to be precisely
placed on the scaffolds, hindering scalability. Additionally, the
extent of polymerization was controlled by the size of the scaffold,
and while this may be beneficial for creating defined circuits, limits
amplification to the size of the scaffold. Another recent strategy to
localize HCR has been to selectively recruit the HCR hairpins with
a supramolecular scaffold, using a short reversible hybridization
region.[29] In this case, the hairpins no longer needed to be
precisely placed and the extent of polymerization was no longer
confined to the template, however, this results in a weakly held
construct that is more suited to in vitro than in vivo applications
Keeping these two strategies in mind, we developed a
design that allows us to generate a robust structure with two
hairpins anchored to a short DNA scaffold. This results in a two-
hairpin monomer that can polymerize upon analyte recognition
(Figure 3a). Using a design like this means that 1) HCR is not
limited to the size of the scaffold, and 2) only two hairpins need to
be precisely placed. To achieve this, our design makes some
changes to the sequences of the original HCR system.
First, to avoid circuit leakage, we increased the stem length
(S) of the hairpins from a 12 to an 18 b.p. hybridization region.
Next we extended the 5’ and 3’ ends of the H1 and H2 hairpins in
order to provide a hybridization region (A) that could then be used
4
This article is protected by copyright. All rights reserved.

10.1002/anie.202001123
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 3. (a) Design of Blunt (BA) end assembly with H1-2-FL and H2-2-SH hairpins. (b) Native PAGE of the assembly of BA and SA. (c) Release of
methoxyfluorescein from the HCR of H1-2-FL and H2-2-SH with 0, 0.1, and 1 equiv. of I-2, on No Template (NT), SA, and BA after 24hr, fluorescence was measured
at λex= 470 nm, λem= 515 nm. (d) Mechanism of HCR by the BA: i. Initiator hybridizes intermolecularly to T1 of H1-2-FL and performs a strand displacement,
exposing the L1 region. ii. L1 hybridizes intramolecularly to T2 on H2-2-SH and performs a strand displacement to expose the L2 region. iii. a DNA templated
reaction occurs releasing the pro-functional molecule connected to the immolative linker. iv. Degradation of the immolative linker to release the active functional
molecule. v. L2 region hybridizes intermolecularly to the T1 region of another BA unit. vi. Polymerization continues, amplifying the release of active molecule.
representing a 65% yield over 3 steps (Figure 3c). Analysis of the
native agarose gel reveals that the polymerized species seem to
have lower average molecular weight distributions (Figure S20a,
lane NT-0.1), than the original counterparts (Figure S12a. lane 5).
We suspect that with the changes made to the hairpins, the
propagation rate of polymerization has been slowed down (H1
hybridizing to H2), while the initiator binding step is still rapid (I
hybridizing to H1).
Confirming that HCR still works with these changes, we then
moved on to generate the sticky (SA) and blunt (BA) ended
assemblies. To make the assemblies we pre-annealed each of
the hairpins separately for correct folding, and then hybridized
them with each template at room temperature (Figure 3b, detailed
in SI-IX). Control experiments without initiator showed the
absence of non-specific reactivity between the H1-2-FL disulfide
and the H2-2-SH thiol, as leakage was comparable to the non-
templated hairpins (Figure 3c and Figure S20-S22). Upon adding
1 equiv. and 0.1 equiv. of initiator strand (I-2) to each of the
assemblies, we found that the yields and amplification were
maintained compared to the non-templated hairpins, 75% and
65% respectively, amplifying approximately 6.5x (Figure 3c).
Comparing the sticky and blunt ended assemblies when 1
equiv. of initiator is added (Figure S20, lanes SA-1 & BA-1), the
blunt ended assemblies remain more defined and do not appear
to polymerize as the sticky ended assemblies. We hypothesize
that this is due to a combination of slow propagation and a pre-
organization effect with the sticky ended assemblies. If, in solution
the sticky ended assemblies are pre-organized, then when
initiator is introduced instead of polymerizing uni-directionally, the
polymerization can occur in two directions (Figure S24).
Statistically this can leave un-hybridized hairpins in the middle of
the pre-organized polymer, which then can only hybridize
intermolecularly. Indeed, when we look at the atomic force
microscopy (AFM) of the sticky, and blunt ended assemblies and
no template with 0.1equiv of initiator we find that there is more
bundling of the sticky ended assemblies, than the blunt ended
assemblies and no template, indicating increased intermolecular
branching between units (Figure S23).
Ultimately, we found that the rate of release between the
sticky and blunt ended assemblies and no template remained
unchanged (Figure S22). This is consistent with remaining
intermolecular hybridizations between units, and the slowest step
being the immolative cyclization’s to activate the fluorophore.
Based on the native gel and AFM analysis we can infer that the
intramolecular hybridizations do in fact occur faster than
intermolecular hybridizations, however more detailed studies on
strictly the rate of polymerization would be needed to confirm.
Overall, it was successfully established that we could in fact
integrate a set of reactive hairpins into a single nanodevice, and
selectively activate the amplified release of pro-functional small
molecules in response to a DNA signal. Interestingly, the HCR
process also results in a cross-linked and rigid polymeric
5
This article is protected by copyright. All rights reserved.

10.1002/anie.202001123
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 4. (a) Design of Spherical nucleic acid (SNA) assemblies for HCR between H1-2-FL and H2-2-SH. (b) Native agarose of the SNA assembly. (c) Release of
methoxyfluorescein from the HCR of H1-2-FL and H2-2-SH with 0, 0.1, and 1 equiv. of I-2, on No Template (NT) and SNA assembly after 24hr, fluorescence was
measured at λex= 470 nm, λem= 515 nm. (d) Mechanism and AFM characterization of the release of functional molecules from the SNA assembly, resulting in
clustering of the SNA’s.
nanostructure that is partially ligated, and the presence of this
structure at e.g., the disease site could possibly contribute a
therapeutic effect. This approach would be similar to enzyme-
triggered peptide self-assembly, which has been shown to
selectively kill cancer cells.[30]
Design, Synthesis and Characterization of Spherical
Nucleic acid Templated HCR. Having shown the integration of a
pair of hairpins into a single device, we next wanted to investigate
integrating multiple hairpins into a single nanodevice. Spherical
nucleic acids have been extensively studied for their cellular
uptake and nuclease resistance properties and provide us a 3D
platform for templation of the HCR reaction.[31]
native/denaturing & fluorescence (Figure 4c. and Figure S25-
S27).
The amount of release was then monitored in response to 1
and 0.1equiv of initiator (I-2). Here we found that the yields were
slightly improved from 75% to 82% for the 1equiv and from 66%
to 74% for 0.1equiv of initiator, over three steps (Figure 4c.). Once
again we were able to amplify the signal of 75 nM by releasing
560 nM of methoxyfluorescein, giving us an overall amplification
of approximately 7.5x. Additionally, we characterized the HCR
polymerization by dynamic light scattering (DLS) and atomic force
microscopy (AFM). By both of these techniques we observed that
the particles clustered together when 1 or 0.1 equiv. of initiator
strand was added, generating large aggregates of SNA’s (Figure
4d and Figure S28). These changes in morphology in response to
a low amount of DNA signal may be of further interest for future
dynamic nanotechnology studies. In contrast to the approach in
Fig. 3, the SNA carries multiple hairpins, rather than just two, and
each SNA nanostructure can release a relatively large number of
small molecules. Using spherical nucleic acids as the platform for
HCR allows for localization of multiple HCR components in a
single device and is more suited for future in vivo studies, where
distribution of different components may differ.
In our lab we have developed sequence controlled spherical
nucleic acids that are made through the introduction of dodecane
phosphoramidites, providing
a hydrophobic block for self-
assembly.[32] For our purposes here, we generated two
amphiphilic nucleic acids with opposite directionalities using this
method. Each of these strands is then able to hybridize to the
overhang region (A) on the previously generated H1-2-FL and H2-
2-SH hairpins, giving them the correct directionality for the HCR
process to occur (Figure 4a). The assemblies were generated in
a stepwise fashion by first generating the spherical nucleic acid
and sequentially adding each hairpin (Figure 4b, detailed in SI-X.).
After the assemblies were generated, we found that there was
minimal leakage (≤10%) of the system over 24hrs by HPLC,
6
This article is protected by copyright. All rights reserved.

10.1002/anie.202001123
Angewandte Chemie International Edition
RESEARCH ARTICLE
Bioorg. Med. Chem. 2001, 9, 2501-2510; c) D. Chang, E.
Lindberg, N. Winssinger, J. Am. Chem. Soc. 2017, 139, 1444-
1447; d) K. K. Sadhu, N. Winssinger, Chemistry – A European
Journal 2013, 19, 8182-8189; e) H. Wu, S. C. Alexander, S.
Jin, N. K. Devaraj, J. Am. Chem. Soc. 2016, 138, 11429-
11432; f) K. T. Kim, S. Angerani, D. Chang, N. Winssinger, J.
Am. Chem. Soc. 2019, 141, 16288-16295.
L. Holtzer, I. Oleinich, M. Anzola, E. Lindberg, K. K. Sadhu, M.
Gonzalez-Gaitan, N. Winssinger, ACS Central Science 2016,
2, 394-400.
J. Chiu, P. J. Hogg, The Journal of biological chemistry 2019,
294, 2949-2960.
D. Butera, T. Wind, A. J. Lay, J. Beck, F. J. Castellino, P. J.
Hogg, The Journal of biological chemistry 2014, 289, 2992-
3000.
Conclusion
In this work we first demonstrate that by rationally placing
disulfide and thiol functionalities on opposite hairpins of HCR we
can amplify a DNA signal to selectively release small molecules.
This is the first time HCR has been used for the triggered release
of small molecules which can be monitored in three ways 1) by
the degree of monomer consumption, 2) by the degree of ligation
and 3) by fluorescence intensity. Unlike previous HCR methods
where small molecule cleavage is non-specific, our construct is
built to only carry out reactions in the presence of an initiator,
significantly decreasing non-specific cargo release. Our synthetic
approach can also be applied to other DNA toehold-mediated
amplification processes, such as CHA. Additionally, we were able
to take advantage of the metastability of the HCR hairpins and
integrate both reactive hairpins into single unit nanodevices. We
used both DNA and SNA assemblies which, importantly, were
able to separate reactive functional groups and isolate reactivity,
preventing release, and amplify release when triggered. The
nanostructures not only release small molecules, but in addition
[6]
[7]
[8]
[9]
J. Chiu, J. W. H. Wong, M. Gerometta, P. J. Hogg,
Biochemistry 2014, 53, 7-9.
[10] Y. He, D. R. Liu, Nat. Nanotechnol. 2010, 5, 778.
[11] D. Y. Zhang, G. Seelig, Nature Chem. 2011, 3, 103-113.
[12] H. M. T. Choi, V. A. Beck, N. A. Pierce, ACS Nano 2014, 8,
4284-4294.
[13] a) K. Ren, Y. Xu, Y. Liu, M. Yang, H. Ju, ACS Nano 2018, 12,
263-271; b) G. Chatterjee, N. Dalchau, R. A. Muscat, A.
Phillips, G. Seelig, Nat. Nanotechnol. 2017, 12, 920; c) H. Bui,
V. Miao, S. Garg, R. Mokhtar, T. Song, J. Reif, Small 2017, 13,
1602983; d) H. Bui, S. Shah, R. Mokhtar, T. Song, S. Garg, J.
Reif, ACS Nano 2018, 12, 1146-1155; e) J. Wang, D.-X.
Wang, J.-Y. Ma, Y.-X. Wang, D.-M. Kong, Chemical Science
2019.
conditionally generate
a partially ligated, crosslinked DNA
polymer, which could have added therapeutic effects, in analogy
to enzyme-driven peptide assembly. Given the abundance of
DNA templated reactions and different HCR sequences, future
work will focus on increasing sensitivity and accelerating the
kinetics, and adapting this method for in vitro/vivo release
applications.
[14] C. R. Park, S. J. Park, W. G. Lee, B. H. Hwang, Biotechnol.
Bioprocess Eng. 2018, 23, 355-370.
[15] H. M. T. Choi, J. Y. Chang, L. A. Trinh, J. E. Padilla, S. E.
Fraser, N. A. Pierce, Nat. Biotechnol. 2010, 28, 1208-1212.
[16] a) G. Zhu, J. Zheng, E. Song, M. Donovan, K. Zhang, C. Liu,
W. Tan, Proceedings of the National Academy of Sciences
2013, 110, 7998-8003; b) Y.-M. Wang, Z. Wu, S.-J. Liu, X.
Chu, Anal. Chem. 2015, 87, 6470-6474.
.
[17] W. Meng, R. A. Muscat, M. L. McKee, P. J. Milnes, A. H. El-
Sagheer, J. Bath, B. G. Davis, T. Brown, R. K. O'Reilly, A. J.
Turberfield, Nature Chem. 2016, 8, 542.
Acknowledgements
This work was supported by NSERC, FRQNT, CFI, and the
Canada Research Chairs Program. A.P. thanks NSERC for a
graduate scholarship. H.F.S thanks the Canada Council for the
Arts for a Killam Fellowship. H.F.S is a Cottrell Scholar of the
Research Corporation. H.F.S. and A.P. also thank the Mass
Spectroscopy Facility (McGill University) for their help with the
HPLC, and LC-MS experiments.
[18] T. Jiang, L. Zhou, H. Liu, P. Zhang, G. Liu, P. Gong, C. Li, W.
Tan, J. Chen, L. Cai, Anal. Chem. 2019, 91, 6996-7000.
[19] a) M. De Stefano, K. Vesterager Gothelf, ChemBioChem 2016,
17, 1122-1126; b) M. L. McKee, A. C. Evans, S. R. Gerrard, R.
K. O'Reilly, A. J. Turberfield, E. Stulz, Org. Biomol. Chem.
2011, 9, 1661-1666; c) D. Li, X. Wang, F. Shi, R. Sha, N. C.
Seeman, J. W. Canary, Org. Biomol. Chem. 2014, 12, 8823-
8827.
[20] V. Patzke, J. S. McCaskill, G. v. Kiedrowski, Angew. Chem.
Int. Ed. 2014, 53, 4222-4226.
[21] a) Y. Meyer, J.-A. Richard, M. Massonneau, P.-Y. Renard, A.
Romieu, Org. Lett. 2008, 10, 1517-1520; b) Y. Meyer, J.-A.
Richard, B. Delest, P. Noack, P.-Y. Renard, A. Romieu, Org.
Biomol. Chem. 2010, 8, 1777-1780.
Keywords: Dynamic DNA Nanotechnology
•
Isothermal
Amplification
Catalysis
• Self-Immolative • Nucleic Acid Templated
[22] K. Dan, A. T. Veetil, K. Chakraborty, Y. Krishnan, Nat.
Nanotechnol. 2019, 14, 252-259.
[23] Andrew D. Bosson, Jesse R. Zamudio, Phillip A. Sharp, Mol.
Cell 2014, 56, 347-359.
[24] a) X. Chen, N. Briggs, J. R. McLain, A. D. Ellington,
Proceedings of the National Academy of Sciences 2013, 110,
5386-5391; b) D. Y. Zhang, E. Winfree, Nucleic Acids Res.
2010, 38, 4182-4197.
[25] D. Y. Zhang, A. J. Turberfield, B. Yurke, E. Winfree, Science
2007, 318, 1121-1125.
[1]
[2]
M. Di Pisa, O. Seitz, ChemMedChem 2017, 12, 872-882.
a) D. M. Kolpashchikov, Acc. Chem. Res. 2019, 52, 1949-
1956; b) Y. Yu, B. Jin, Y. Li, Z. Deng, Chemistry – A European
Journal 2019, 25, 9785-9798.
[3]
a) H. H. Fakih, J. J. Fakhoury, D. Bousmail, H. F. Sleiman,
ACS Appl. Mater. Interfaces 2019, 11, 13912-13920; b) K. E.
Bujold, J. C. C. Hsu, H. F. Sleiman, J. Am. Chem. Soc. 2016,
138, 14030-14038; c) F. Nagatsugi, S. Nakayama, S. Sasaki,
Nucleosides, Nucleotides Nucleic Acids 2007, 26, 799-803; d)
A. Okamoto, K. Tanabe, T. Inasaki, I. Saito, Angew. Chem. Int.
Ed. 2003, 42, 2502-2504; e) K. E. Bujold, J. Fakhoury, T. G.
W. Edwardson, K. M. M. Carneiro, J. N. Briard, A. G. Godin, L.
Amrein, G. D. Hamblin, L. C. Panasci, P. W. Wiseman, H. F.
Sleiman, Chemical Science 2014, 5, 2449-2455.
a) R. M. Dirks, N. A. Pierce, Proc. Natl. Acad. Sci. U. S. A.
2004, 101, 15275-15278; b) P. Yin, H. M. T. Choi, C. R.
Calvert, N. A. Pierce, Nature 2008, 451, 318; c) Z. Ma, J. S.
Taylor, Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 11159-11163.
a) K. Gorska, A. Manicardi, S. Barluenga, N. Winssinger,
Chem. Commun. 2011, 47, 4364-4366; b) Z. Ma, J.-S. Taylor,
[26] S. Mura, J. Nicolas, P. Couvreur, Nat. Mater. 2013, 12, 991-
1003.
[27] a) Q. Xu, G. Zhu, C.-y. Zhang, Anal. Chem. 2013, 85, 6915-
6921; b) L. Wang, W. Li, J. Sun, S.-Y. Zhang, S. Yang, J. Li, J.
Li, H.-H. Yang, Anal. Chem. 2018, 90, 14433-14438.
[28] A. Lacroix, E. Vengut-Climent, D. de Rochambeau, H. F.
Sleiman, ACS Central Science 2019, 5, 882-891.
[29] W. Engelen, S. P. W. Wijnands, M. Merkx, J. Am. Chem. Soc.
2018, 140, 9758-9767.
[4]
[5]
[30] J. Zhou, X. Du, N. Yamagata, B. Xu, J. Am. Chem. Soc. 2016,
138, 3813-3823.
7
This article is protected by copyright. All rights reserved.

10.1002/anie.202001123
Angewandte Chemie International Edition
RESEARCH ARTICLE
[31] a) X. Tan, B. B. Li, X. Lu, F. Jia, C. Santori, P. Menon, H. Li, B.
Zhang, J. J. Zhao, K. Zhang, J. Am. Chem. Soc. 2015, 137,
6112-6115; b) X. Tan, X. Lu, F. Jia, X. Liu, Y. Sun, J. K.
Logan, K. Zhang, J. Am. Chem. Soc. 2016, 138, 10834-10837;
c) C. H. Kapadia, J. R. Melamed, E. S. Day, BioDrugs 2018,
32, 297-309.
[32] a) J. J. Fakhoury, T. G. Edwardson, J. W. Conway, T. Trinh, F.
Khan, M. Barłóg, H. S. Bazzi, H. F. Sleiman, Nanoscale 2015,
7, 20625-20634; b) T. G. W. Edwardson, K. M. M. Carneiro, C.
J. Serpell, H. F. Sleiman, Angew. Chem. Int. Ed. 2014, 53,
4567-4571; c) D. Bousmail, L. Amrein, J. J. Fakhoury, H. H.
Fakih, J. C. C. Hsu, L. Panasci, H. F. Sleiman, Chemical
Science 2017, 8, 6218-6229.
8
This article is protected by copyright. All rights reserved.

10.1002/anie.202001123
Angewandte Chemie International Edition
RESEARCH ARTICLE
Entry for the Table of Contents
Spatially Isolated and Ready to Amplify: Here, we show the self-immolative, amplified release of small molecules in response to a
nucleic acid signal on 1D, 2D, and 3D assemblies using the hybridization chain reaction (HCR). Current methods of small molecule
release using a nucleic acid signal rely on reversible hybridization to generate amplification, and as such have not been integrated
into standalone nanodevices. We demonstrate the ability of our platforms to effectively separate the reactive groups and
subsequently release small molecules in an amplified manner when triggered, for biosensing, imaging and drug delivery.
9
This article is protected by copyright. All rights reserved.