Autocatalysis and organocatalysis
with synthetic structures
1
Seiji Kamioka, Dariush Ajami, and Julius Rebek Jr.
The Skaggs Institute for Chemical Biology and Department of Chemistry, The Scripps Research Institute, 10550 North Torrey Pines Road,
La Jolla, CA 92037
Contributed by Julius Rebek, November 5, 2009 (sent for review September 8, 2009)
The discovery of ribozymes led to the proposal of an RNA world,
where a single type of molecule was supposedly capable of self-
replication and chemical catalysis. We show here that both auto-
catalysis and organocatalysis can be engineered into a synthetic
structure. The compound is shown to selectively accelerate its
own formation and catalyze either hydrogenation or nucleophilic
addition to α,β-unsaturated aldehydes. The observed reactivity
indicates that the components of a purported pre-RNA world
conceivably include smaller organic molecules.
proline itself requires extensive synthetic sequences, but creation
of an imidazolidinone in the space between the recognition
elements as in 2a could be contemplated in a single step—the
same step that assembles the autocatalyst. This functional group,
devised by MacMillan and coworkers (16, 17) is a versatile
organocatalyst that operates through reversibly formed covalent
intermediates (18).
In the experiment, 2a could be obtained from condensation of
(
Fig. 1 and SI Appendix). The desired cis 2a was accompanied
imidazolidinone catalysis ∣ self-replication ∣ template effects
by the trans isomer 2c. The N-methyl thymine derivatives 2b
and 2d were also prepared and used as controls because their
inability to “base pair” abrogates molecular recognition during
their reactions with the diaminotriazine subunit.
The earliest evidence that recognition-based reactions were
involved in the final steps of the synthesis came from the yields
of reactions involving closely related structures. Specifically, the
cis 2a (38%) was obtained in higher yield than trans 2c (18%),
whereas the N-methyl 2b (19%) and 2d (15%) were produced
in comparable amounts. The preference for 2a was diminished
in a polar solvent such as MeOH, which competes for hydrogen
bonds, and was eliminated entirely in the reaction of 4a and 4b
tudies of prebiotic chemistry raise questions concerning which
S
functions came nearer the origin of life—genetics or metabo-
lism (1). The discovery of the catalytic activity of ribozymes pro-
vided an answer in which both functions could be accounted for in
an “RNA world” (2). Although no one doubted that RNA could
carry genetic information, some 15 years lapsed before an RNA
molecule was shown to self-replicate (3), and this type of catalysis
has since been developed to an extraordinary level of efficiency
(
4). Short, self-complementary DNA strands were shown to self-
replicate in 1986 (5), and autocatalysis based on molecular recog-
nition was found in simplified organic molecules (6) and even
peptides (7). At the present time, many types of self-replicating
structures are known (8). These systems function like their nu-
cleic acid counterparts in that the self-complementary structures
operate as templates for their own formation. However, unlike
ribozymes, the synthetic self-replicating systems are not known
to catalyze other chemical reactions. The wholly synthetic struc-
tures act as either organocatalysts or autocatalysts, but published
evidence for both activities in a single molecule is limited (9). This
research was undertaken to find molecules that could perform in
both capacities, and we report here an imidazolidinone-based
compound and its properties as a catalyst and autocatalyst.
A consistent and quantitative measure of autocatalysis was
obtained through kinetic studies. The condensation reaction of
3
with 4a was analyzed by HPLC and the results are shown in
Fig. 2 Left. The initial rate for the coupling reaction (blue line)
is shown with that of 0.25 equiv (red line) and 0.50 equiv (green
line) cis 2a added. The yields were corrected for the addition of
the catalysts and are determined with respect to an internal stan-
dard. Clearly, 2a accelerates its own formation. That this process
is grounded in self-recognition was shown by parallel studies
monitoring trans 2c as a function of added 2a (Fig. 2 Right). Here,
no acceleration of the rate occurred; cis 2a is, accordingly, a self-
ish (8) replicator because it catalyzes its own formation but not
that of its structurally related diastereomer trans 2c. By compar-
ison, trans 2c also accelerates its own formation and, unexpect-
edly, the formation of the cis 2a as well (S.I.). Thus, trans 2c is
a cooperative (19) replicator. The trans arrangement of alkyl
groups on an imidazolidinone is not expected to be advantageous
for organocatalysis so trans 2c was not tested in those capacities.
Both cis 2a and trans 2c are stable under the reaction conditions;
they neither interconvert nor racemize.
Results and Discussion
To realize an autocatalytic/organocatalytic molecule, we made
modifications in the structure of compound 1, a molecule
that bears self-complementary hydrogen-bonding subunits—a
diaminotriazine and cyclic imide—that allow for dimerization
in noncompeting organic media (Fig. 1). The arrangement of
these subunits in 1 was earlier shown to allow its action as an
autocatalyst (10): It acts as a template for its own synthesis, gath-
ering two subunits on its surface to facilitate the formation of the
amide bond. The insertion of a functional group known to cata-
lyze organic transformations into 1 was a prospect to be explored.
We initially inserted a thiourea into the framework of 1 (11).
Although the thiourea function is a widely used organocatalyst
Taken together, these results show that autocatalysis based on
recognition—replication—takes place during the formation of
the cis 2a, and the likely process is illustrated in Fig. 3. The two
complementary hydrogen-bonding sites of 2a (diaminotriazine
and thymine) can assemble the starting materials into the
(12), its applications were limited in the present context: The
same hydrogen bonding that accounts for the thiourea’s activity
requires the use of solvents that also lead to tight dimerization
of the template. This effect minimizes the fraction of the mono-
meric compound that operates as the organocatalyst. To avoid
this limitation we sought an organocatalytic function that oper-
ates through reversible formation of covalent bonds, such as
the venerable proline-based catalysts (13–15). Insertion of
Author contributions: S.K., D.A., and J.R. designed research; S.K. and D.A. performed
research; S.K., D.A., and J.R. analyzed data; S.K., D.A., and J.R. wrote the paper.
The authors declare no conflict of interest.
1
To whom correspondence should be addressed. Email: jrebek@scripps.edu.
www.pnas.org/cgi/doi/10.1073/pnas.0912769107
PNAS ∣ January 12, 2010 ∣ vol. 107 ∣ no. 2 ∣ 541–544