A simple network of synthetic replicators can perform the logical or operation

Article abstract of DOI:10.1021/ol100404g

Figure presented A small network of synthetic replicators is capable of responding to instructional inputs such that the output of the network is an excess of one of the replicators whenever the input contains either or both of the replicators, mirroring

Full text of DOI:10.1021/ol100404g

ORGANIC
LETTERS
2
010
Vol. 12, No. 9
920-1923
A Simple Network of Synthetic Replicators
Can Perform the Logical OR Operation
1
Victoria C. Allen, Craig C. Robertson, Simon M. Turega, and Douglas Philp*
EaStCHEM and Centre for Biomolecular Sciences, School of Chemistry, UniVersity of
St Andrews, North Haugh, St. Andrews, Fife KY16 9ST, United Kingdom
d.philp@st-andrews.ac.uk
Received February 17, 2010
ABSTRACT
A small network of synthetic replicators is capable of responding to instructional inputs such that the output of the network is an excess of
one of the replicators whenever the input contains either or both of the replicators, mirroring the OR boolean logic operation.
4
We have become interested in the exploitation of synthetic
strategies that encompass replication processes for facilitat-
exploiting selection and amplification through emergent
system behavior.
1
ing the fabrication of molecular architectures at the nano-
meter scale. In order to evolve synthetic machinery that is
capable of directing its own synthesis and cooperating with
As an initial step toward this ambitious goal, we explored
a system in which a small network of interdependent
replicating systems cooperated to exhibit defined emergent
behavior. In order to achieve this aim, we must first identify
a family of structurally similar self-replicating templates that
possess the correct steric and electronic properties to allow
them to interact with each other. Previously, we described
1
the kinetic behavior of template T , which was constructed
from nitrone A and maleimide B (Figure 1). This template
is the major product of the reaction between A and B and
was shown to be capable of increasing the rate of its own
formation by around 3-fold, at the expense of its catalytically
2
other similar systems to create an organized hierarchy, it is
important to develop a fundamental understanding of rec-
ognition-mediated processes that allow molecules to function
as specific and efficient templates for the formation of
themselves (autocatalysis) and other templates (cross cataly-
sis). Such an understanding should permit the development
of efficient protocols that allow us to establish and manage
replication, organization, and evolution within synthetic
5
3
molecular and supramolecular assemblies, so-called “sys-
tems chemistry”. Ultimately, the creation of large molecular
and supramolecular assemblies can be programmed by
(
3) (a) Kindermann, M.; Stahl, I.; Reimold, M.; Pankau, W. M.; von
Kiedrowski, G. Angew. Chem., Int. Ed. 2005, 44, 6750–6755. (b) Nitschke,
J. R. Acc. Chem. Res. 2007, 40, 103–112. (c) Ludlow, R. F.; Otto, S. Chem.
Soc. ReV. 2008, 37, 101–108. See also: (d) Prins, L. J.; Scrimin, P. Angew.
Chem., Int. Ed. 2009, 48, 2288–2306.
(
1) (a) Patzke, V.; von Kiedrowski, G. ARKIVOC 2007, 5, 293–310.
b) Vidonne, A.; Philp, D. Eur. J. Org. Chem. 2009, 593–610.
2) (a) Strogatz, S. H. Nature 2001, 410, 268–276. (b) Ravasz, E.;
Somera, A. L.; Mongru, D. A.; Oltvai, Z. N.; Barabasi, A. L. Science 2002,
97, 1551–1555. (c) Girvan, M.; Newman, M. E. J. Proc. Natl. Acad. Sci.
U.S.A. 2002, 99, 7821–7826. (d) Guimer a` , R.; Amaral, L. A. N. Nature
005, 433, 895–900. (e) Perez, J. J. Chem. Soc. ReV. 2005, 34, 143–152.
f) Dobson, C. M. Nature 2004, 432, 824–828. (g) Lavigne, J. J.; Anslyn,
E. V. Angew. Chem., Int. Ed. 2001, 40, 3118–3130.
(
(
(4) (a) Dadon, Z.; Wagner, N.; Ashkenasy, G. Angew. Chem., Int. Ed.
2008, 47, 6128–6136. (b) Wagner, N.; Ashkenasy, G. J. Chem. Phys. 2009,
130, 164907. (c) Wagner, N.; Ashkenasy, G. Chem.sEur. J. 2009, 15, 1765–
1775. (d) Ashkenasy, G.; Jagasia, R.; Yadav, M.; Ghadiri, M. R. Proc.
Natl. Acad. Sci. USA 2004, 101, 10872. (e) Ashkenasy, G.; Ghadiri, M. R.
J. Am. Chem. Soc. 2004, 126, 11140.
2
2
(
(5) Allen, V. C.; Philp, D.; Spencer, N. Org. Lett. 2001, 3, 777–780.
1
0.1021/ol100404g  2010 American Chemical Society
Published on Web 04/14/2010

Figure 1
diastereoisomeric templates, T
and two cross-catalytic cycles.
.
(a) Nitrone A and maleimides B and C are combined to create two self-replicating templates T
1 2
and T . The corresponding
1
′ and T ′, are catalytically inert. (b) Templates T and T cooperate through a network of two autocatalytic
2
1
2
inactive diastereoisomer T
1
′. Building on this discovery, we
′, as suitable
conversion represent the baseline behavior for this system
in the absence of any recognition-mediated reactions. The
reaction of A with C affords a 11.4:1 ratio of T /T ′ at 42%
identified template T , and its diastereoisomer T
2
2
candidates from which to construct the simple network of
interdependent catalytic cycles shown in Figure 1. Electronic
structure calculations led us to expect that, in common with
2
2
conversion after 10 h. The important role of recognition in
this system was demonstrated by the reaction of A with C
in the presence of the competitive inhibitor benzoic acid
(8.4:1 ratio of T /T ′; 23% conversion after 10 h). Finally,
template T
considering this network, there are two autocatalytic cycles
in which each template, T and T , directs its own formation.
Additionally, because T and T differ by only one CH
group, we might expect that two cross-catalytic cycles would
also exist in which T directs the formation of T and vice
1 2
′, T ′ would be catalytically inactive. Therefore,
2
2
1
2
we demonstrated that T is a template for its own formation
2
1
2
2
through an experiment in which 40 mol % of T was added
2
at the start of the reaction between A and C. Under these
1
2
conditions, the rate of formation of T is enhanced signifi-
2
versa. Therefore, the products of a reaction between A-C
should depend on the relative efficiencies of each of these
four cycles.
cantly (1.6-fold) and the ratio of T
conversion after 10 h. The template T
recognition-mediated behavior in these reactions.
Having demonstrated that template T is capable of
directing its own formation, we set about establishing the
cross-catalytic behavior of these templates. In these experi-
ments, the building blocks required to construct one template,
2
/T
2
′ is 16.8:1 at 54%
2
′ does not show any
Initially, it was important to demonstrate that template T
is capable of directing its own formation. We have described
2
2
6
protocols for accomplishing this task previously, and these
methods were applied to this system. All reactions were
performed from a starting concentration of 25 mM of each
either T
reaction mixture was then doped with 40 mol % of one of
the templates, either T or T . Each reaction mixture was
1 2 3
or T , were dissolved in CDCl at 25 mM. This
3
reagent at 10 °C in CDCl and were monitored by 400 MHz
1
H NMR spectroscopy. The control reaction between A and
the methyl ester of C affords a 3.8:1 ratio of the methyl esters
1
2
then incubated at 10 °C for 16 h, and the progress of the
of T
2
and T
2
′ at 11% conversion after 10 h. This ratio and
1
reaction was followed by 400 MHz H NMR spectroscopy.
The results of this study are presented in Figure 2.
(
6) (a) Pearson, R. J.; Kassianidis, E.; Slawin, A. M. Z.; Philp, D.
Chem.sEur. J. 2006, 12, 6829–6840. (b) Kassiandis, E.; Pearson, R. J.;
Philp, D. Chem.sEur. J. 2006, 12, 8798–8812. (c) Kassianidis, E.; Philp,
D. Angew. Chem., Int. Ed. 2006, 45, 6344–6348. (d) del Amo, V.; Slawin,
A. M. Z.; Philp, D. Org. Lett. 2008, 10, 4589–4592.
As expected, the results confirm that each template is
capable of directing its own formation. The addition of T
1
7
engenders a 1.6 times enhancement in the maximum rate
Org. Lett., Vol. 12, No. 9, 2010
1921

Figure 2
of T and (b) the formation of T
40 mol % added at the start of the reaction) is measured by
.
Effect of the addition of T
1 2
and T on (a) the formation
1
2
. The effect of the added template
(
log(relative rate). This measure is calculated by taking the logarithm
of the ratio of the maximum rates of the reactions in the experiments
with and without added template.
Figure 3
for (a) forming T
(b) forming T on a T
1
.
Calculated (B3LYP/6-31G(d,p)) transition-state structures
on a T template (a ) 2.11 Å; b ) 2.12 Å) and
template (a ) 2.07 Å; b ) 2.16 Å). The
1
1
of formation of itself. T
2
is a somewhat poorer template for
2
dihedral angle about the starred bond is 61°. Atom types are
represented by shading dark to light in the order carbon, oxygen,
nitrogen, and hydrogen. Hydrogen bonds are represented by dotted
lines and partial bonds in the transition states by dashed lines. Most
hydrogen atoms have been removed for clarity.
its own formation, engendering a 1.3 times increase in the
maximum rate for its formation. It is clear that the behavior
of the templates toward each other is markedly different.
Template T
increase in the maximum rate of T
inhibits the formation of T slightly (the maximum rate
.8 times that in the absence of added template). This
behavior implies that, although cross-catalytic cycle 1 (Figure
2
1
also directs the formation of T (1.4 times
1
production), whereas
T
1
2
1
complex [A·C·T ] shows significant distortion (Figure 3b)
0
and forces maleimide C to adopt an unfavorable conforma-
tion (starred bond, Figure 3b) in order to access the transition
state. This distortion is a result of the shorter overall length
1
b) operates efficiently, cross-catalytic cycle 2 does not.
We found these results surprising given the structural
of the T
itself to accommodate the transition state leading to T
A·B·T ], the shorter and more rigid structure of T
compression of the T
A·C·T ]. These results are in complete accord with those
1 2
template. Although the T template can compress
similarity of the templates. We therefore turned to electronic
structure calculation of the transition states of the autocata-
lytic and cross-catalytic ternary complexes (Figure 3) in order
to gain some insight into the subtle structural mismatch
between the templates. Calculations were performed at the
B3LYP/6-31G(d,p) level of theory, and transition states
leading to the appropriate products were located successfully
for each of the four ternary complexes.
1
within
1
forces
[
2
2
transition state when accessed from
[
1
observed experimentally and identify the small structural
mismatch that gives rise to the observed behavior.
It is therefore clear that, in a reaction mixture containing
A-C, only three of the four possible catalytic cycles will
be operative. Hence, we might expect that this network will
display emergent behavior analogous to the logical OR
operation. If we imagine that system inputs are the identities
of the templates added and a reaction mixture containing A,
The results of these calculations are revealing. For the
autocatalytic ternary complexes [A·B·T
1
] and [A·C·T
2
], the
transition states leading to [T ·T ] and [T
1
1
2
·T ] are supported
2
on the preformed template without distortion and the
transition states are bound to the preformed templates by
four hydrogen bonds. The structure of [A·B·T ] (Figure 3a)
1
is representative of the transition states where the templates
are matched. In the case of the cross-catalytic ternary
1 2
B and C and the system output is the T /T ratio, we can
use our knowledge of the kinetic behavior of the templates
in isolation to predict the overall network topology. In the
absence of any added template, the replicators are of similar
efficiency and should therefore coexist. In the presence of
complex [A·B·T
2
], the transition state leading to [T
1 2
·T ] is
also supported on the preformed template without distortion
by four hydrogen bonds. However, the calculated structure
of the transition state accessed by the cross-catalytic ternary
added template T
since T is a catalyst for its own formation and an inhibitor
for the formation of T . In the presence of added template
, T will be amplified through the cross-catalytic action
of T , and the more T is produced, the more dominant
autocatalytic cycle 1 will become and the more the formation
1 1 2
, T will be amplified and T suppressed
1
2
T
2
1
(7) Maximum rates (velocities) of reaction were calculated by determin-
ing the largest value of the first derivative of the polynomial function
describing the concentration time profile for each reaction. For bimolecular
reactions, this metric is equivalent to the initial rate. For autocatalytic
reactions, this represents the point of inflection of the sigmoidal curve.
2
1
of T
2
will be inhibited. In the presence of both templates,
1922
Org. Lett., Vol. 12, No. 9, 2010

T
1
will be amplified and T
catalyst for its own formation and an inhibitor for the
formation of T . This behavior represents the logical OR
operation: the presence of either template or both templates
together results in the enhanced formation of T (T /T
).
With this expectation in mind, we conducted a series of
four experiments, in each of which the starting concentrations
of A-C were all 25 mM and 20 mol % template T and/or
was added as appropriate at the start of the reactions. All
reactions were carried out in CDCl at 10 °C, and the reaction
2
suppressed because T
1
is a
realizing that T
of T . Therefore, cross-catalytic cycle 1 (Figure 1b) will
generate significant quantities of T . The production of T
1
will, in turn, allow autocatalytic cycle 1 (Figure 1b) to
become operative and, in addition, impose an inhibitory effect
on the operation of cross-catalytic cycle 2 (Figure 1b). The
2
is, in fact, also a catalyst for the formation
1
2
1
1
1
2
>
1
net effect is amplification of T
both templates are added together (experiment D, Figure 4),
is once again amplified strongly at the expense of T . In
1 2
at the expense of T . When
1
T
1
2
T
2
this case, autocatalytic cycle 1 and cross-catalytic cycle 1
are the dominant pathways in the system.
This series of experiments confirms our expectation that
this small network of simple synthetic replicating systems
is capable of acting in concert to perform the logical OR
3
8
1
mixtures were assayed after 4 and 8 h by 400 MHz H NMR
spectroscopy (see the Supporting Information for details).
The results are summarized in Figure 4.
operation based on the input template. Template T
1
is
enhanced when the template input is T or T or both.
1
2
Although the amplification effects reported here are not
particularly large, we believe that this can be remedied by a
careful computationally assisted replicator design. However,
it is striking that two templates whose molecular weights
2
are around 400 and which differ by a single CH unit can
exhibit the behavior observed here. The results presented here
bode well for the development of more complex recognition-
9
mediated reaction networks that rely on multiple recognition
1
0
events, such as a combination of minimal and reciprocal
replicators. Such systems can potentially generate and express
more complex programmed responses to chemical inputs
through template-directed processes. These strategies are
currently under development in our laboratory.
Figure 4
template is T
.
Ratio of [T
OR T
1
2
]/[T ] is higher when the identity of the input
1
2
.
Acknowledgment. The financial support of the Engineer-
ing and Physical Sciences Research Council and the Bio-
technology and Biological Sciences Research Council is
gratefully acknowledged.
The results of these experiments confirm our predictions.
Taking experiment A (Figure 4) as the baseline, the formation
of T
is added (experiment B, Figure 4). Clearly, in this case,
autocatalytic cycle 1 is dominant. When only T is added
experiment C, Figure 4), T is once again amplified at the
expense of T . This outcome can be rationalized readily by
1 1
is enhanced significantly, by 1.8 times, when only T
Supporting Information Available: Experimental pro-
cedures and characterization for compunds A-C, T
1
, and
2
T
2
and procedures for conducting and assaying kinetic
(
1
experiments. This material is available free of charge via
the Internet at http://pubs.acs.org.
2
(
8) Analysis of complex systems by NMR spectroscopy relies on there
OL100404G
1
being sufficient dispersion of the resonances utilized in the assay. H NMR
spectroscopy is not ideal for this purpose because H chemical shifts span
1
a relatively small range. In the system presented here, dispersion is not an
issue. However, in cases where signal dispersion does become a problem,
F NMR spectroscopy provides a suitable alternative (see ref 9a for an
(9) (a) Sadownik, J. W.; Philp, D. Angew. Chem., Int. Ed. 2008, 47,
9965–9970. (b) Kassianidis, E.; Pearson, R. J.; Wood, E. A.; Philp, D.
Faraday Discuss. 2010, 145, 235–254.
19
example).
(10) Kassianidis, E.; Philp, D. Chem. Commun. 2006, 4072–4074.
Org. Lett., Vol. 12, No. 9, 2010
1923