Exploration of Diverse Reactive Diad Geometries for Bifunctional Catalysis via Foldamer Backbone Variation

Article abstract of DOI:10.1021/jacs.8b05869

What is the best spatial arrangement of a pair of reactive groups for bifunctional catalysis of a chemical transformation? The conformational versatility of proteins allows reactive group geometry to be explored and optimized via evolutionary selection, but it has been difficult for chemists to identify synthetic scaffolds that allow broad comparative evaluation among alternative reactive group geometries. Here we show that a family of helices, adopted predictably by oligomers composed partially or exclusively of β-amino acid residues, enables us to explore a range of orientations for a pair of pyrrolidine units that must work in tandem to catalyze a crossed aldol reaction. Thus, the crossed aldol reaction serves as an assay of reactive diad efficacy. We have chosen a test reaction free of stereochemical complexity in order to streamline our study of reactivity. The best geometry enhances the initial rate of product formation by two orders of magnitude. Our findings raise the possibility that rudimentary catalysts involving an isolated secondary structure might have facilitated the development of prebiotic reaction networks.

Full text of DOI:10.1021/jacs.8b05869

Article
pubs.acs.org/JACS
Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Exploration of Diverse Reactive Diad Geometries for Bifunctional
Catalysis via Foldamer Backbone Variation
Zebediah C. Girvin and Samuel H. Gellman*
Department of Chemistry, University of Wisconsin, 1101 University Avenue, Madison, Wisconsin 53706, United States
S
* Supporting Information
ABSTRACT: What is the best spatial arrangement of a pair of reactive
groups for bifunctional catalysis of a chemical transformation? The
conformational versatility of proteins allows reactive group geometry to
be explored and optimized via evolutionary selection, but it has been
difficult for chemists to identify synthetic scaffolds that allow broad
comparative evaluation among alternative reactive group geometries.
Here we show that a family of helices, adopted predictably by oligomers
composed partially or exclusively of β-amino acid residues, enables us to
explore a range of orientations for a pair of pyrrolidine units that must
work in tandem to catalyze a crossed aldol reaction. Thus, the crossed
aldol reaction serves as an assay of reactive diad efficacy. We have chosen
a test reaction free of stereochemical complexity in order to streamline
our study of reactivity. The best geometry enhances the initial rate of
product formation by two orders of magnitude. Our findings raise the possibility that rudimentary catalysts involving an isolated
secondary structure might have facilitated the development of prebiotic reaction networks.
develop a new method for conducting crossed aldol reactions,
but rather to use the crossed aldol reaction to compare
alternative reactive diad geometries and identify an optimum in
terms of bifunctional catalysis.
INTRODUCTION
■
Life depends upon poly-α-amino acid catalysts (enzymes) that
promote a wide array of reactions, frequently with extra-
ordinary rate accelerations relative to the uncatalyzed
processes.1 Despite extensive study, however, the origins of
the large catalytic rate enhancements that are commonly
encountered remain unclear, in terms of both specific
mechanisms and the evolutionary path from prebiotic catalysts
to the efficient enzymes that abound in biology.2,3 Consid-
erable enzyme-inspired research has been directed toward
simpler systems that display properties thought to be
important for enzymatic catalysis. This approach has been
motivated by the prospect that decreasing catalyst complexity
should facilitate mechanistic analysis4−6 and by the desire for
new and efficient methods to synthesize organic molecules.7−10
Proper spatial organization of two or more reactive groups
appears to be a critical feature of many enzyme mechanisms,2,3
and bifunctional or multifunctional catalysis has been explored
in smaller synthetic systems.4−12
Here we introduce a new strategy for evaluating diverse
orientations of a pair of reactive groups, with the goal of
identifying arrangements that enable coordinated catalytic
action. Our approach makes use of “foldamers”, protein-
inspired oligomers that feature unnatural backbones and
display discrete conformational preferences.13−15 We employ
foldamers that contain β-amino acid residues and that adopt
distinct helical secondary structures. These helices are used to
position pairs of pyrrolidine units in different three-dimen-
sional arrangements that are assessed for bifunctional catalysis
of a crossed aldol reaction. The goal of this work is not to
RESULTS AND DISCUSSION
■
̈
Our experiments build on careful kinetic studies by Erkkila and
Pihko that established that pyrrolidine plays a dual role in
catalyzing crossed aldol condensations involving formaldehyde
as the electrophile.16 The nucleophilic aldehyde is activated via
enamine formation, and formaldehyde is activated via iminium
formation. This finding encouraged us to look to foldamers
containing pairs of pyrrolidine-derived β-amino acid residues
as a basis for probing the relationship between spatial
organization of a reactive diad and catalysis of the crossed
aldol condensation. We focused on hydrocinnamaldehyde as
the nucleophile.
Cyclically constrained β-amino acid residues enable tuning
of foldamer secondary structure preference and stability via
control of ring size and stereochemistry.17−20 trans-2-Amino-
cyclopentanecarboxylic acid (ACPC) residues (Figure 1), for
example, support formation of a β-peptide helix characterized
by CO(i)···H-N(i+3) H-bonds.21 Combining (S,S)-ACPC
residues with L-α-amino acid residues in varying proportions
and patterns engenders a family of related helical secondary
structures that feature CO(i)···H-N(i+3) or CO(i)···H-
N(i+4) H-bonds.22−24 The pyrrolidine-based APC residue
Received: June 10, 2018
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.8b05869
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 3. The crossed aldol reaction used in these studies.
Figure 1. Cyclic β-amino acid ACPC and pyrrolidine derivative APC.
(Figure 1) displays conformational propensities indistinguish-
able from those of ACPC, as established by NMR analysis of
diverse foldamers and numerous α/β-peptide crystal structures
in the Protein Data Bank.25
reaction completion) under a set of conditions suggested by
precedent (Figure 3).16 Product formation was monitored by
UPLC. The two aldehyde substrates were used in equimolar
quantities. Reactions were conducted in isopropanol at 37 °C,
with 4 vol % water and 2 equiv of triethylamine and propionic
acid relative to the aldehyde starting materials.
The first set of studies involved β-peptides 1−5 (Figure 4A).
β-Peptide 1, containing a single APC residue, was used as the
reference for this series. Comparison of 1 with pyrrolidine
revealed that the secondary amine within an APC residue is
intrinsically less effective for crossed aldol catalysis relative to
the secondary amine within pyrrolidine itself: the initial rate of
product formation was 216-fold larger for pyrrolidine than for
1. This difference in reactivity may reflect the presence of two
electron-withdrawing substituents on the APC ring.
Figure 2 shows three examples from the substantial set of
ACPC-containing foldamer crystal structures in the Cambridge
Structure Database. These examples illustrate helical secondary
structures containing CO(i)···H-N(i+3) H-bonds formed by
three backbones, one containing exclusively β residues (Figure
2A),21 a second with a 1:1 α:β repeat (Figure 2B),23 and a
third with a 1:2 α:β repeat (Figure 2C).24 Collectively, these
three helical secondary structures should provide access to
diverse arrangements of a pyrrolidine diad, because ACPC →
APC replacements are not expected to cause conformational
changes.25,26
We constructed a series of APC-containing oligomers based
on each of the three foldamer families illustrated in Figure 2 to
explore distinct pyrrolidine diad geometries for catalysis of the
selected crossed aldol condensation (Figure 3). Each oligomer
contains a C-terminal β3-homotyrosine (β3-hTyr) residue to
facilitate concentration determination via UV absorbance, and
the remaining β subunits are derived from ACPC or APC.
Catalytic activities for the α-methylenation of hydrocinnamal-
dehyde were compared by assessing relative initial rates (≤1%
β-Peptides 2−5 contain two APC residues with varied
sequential spacing, which leads to different three-dimensional
orientations of the secondary amines upon formation of the
CO(i)···H-N(i+3) H-bonded helix.21 The catalytic efficacies
of the different APC diads were compared in terms of initial
reaction rates normalized to the initial rate observed with
mono-APC β-peptide 1. Specifically, the initial rate of product
formation measured for each bis-APC β-peptide (2−5), at 1
mol % β-peptide with reference to each of the aldehyde
Figure 2. Crystal structures illustrating helical secondary structures adopted by ACPC-containing foldamers with different β residue content: (A)
pure β (CSD: WELNOQ), (B) 1:1 α/β (CSD: OGAVAU), (C) 1:2 α/β (CSD: PUCDEX). Yellow and cyan represent carbon atoms in α and β
residues, respectively. Blue represents nitrogen atoms, and red represents oxygen atoms. Hydrogen atoms other than those bound to nitrogen are
omitted for clarity. Hydrogen bonds are depicted with dashed lines. (D) Three-dimensional relationships expected for the pair of side chain
nitrogen atoms in foldamers containing two APC residues with varied sequential separation, based on the crystal structures shown in (A−C).
Distances were measured between carbon-4 in the cyclopentane rings of appropriately spaced ACPC residues. For each ACPC residue, we defined
a vector based on the positions of carbon-4 and carbon-2. The angle between these two vectors, when the structure is viewed along the helix axis, is
given. Values refer to average distance or average angle based on all possible measurements in the given structure; standard deviations can be found
in Table S12. *Only one possible i, i+6 diad can be measured in 1:1 α/β-peptide (B).
B
DOI: 10.1021/jacs.8b05869
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Significant enhancements in initial rates were observed for β-
peptides 3 and 4. The larger effect (νREL = 62) was measured
for β-peptide 4, which features i, i+3 spacing of the APC
residues. This sequence relationship corresponds to a little
more than one turn of the expected helix; thus, helical folding
of 4 should cause a convergent orientation of the pyrrolidine
ring nitrogen atoms. A smaller enhancement was observed for
i, i+2 APC spacing (3, νREL = 30). For i, i+4 APC spacing,
significantly greater than one helical turn, the enhancement
was more modest (5, νREL = 14). Collectively, the initial rate
data for this series support our premise that a well-established
foldamer secondary structure represents a scaffold that can be
used to compare the catalytic propensities of distinct reactive
group diad geometries.
Series 6−9 (Figure 4B) features 1:1 α:β residue alternation.
Crystallographic studies of this α/β-peptide family, including
the example shown in Figure 2B, indicate that oligomers in the
length range of 6−8 favor a helix containing CO(i)···H-N(i
+3) H-bonds and ∼3 residues per turn.23 (Longer 1:1 α/β-
peptides, such as 9, can also access a different helix containing
CO(i)···H-N(i+4) H-bonds and ∼4.5 residues per turn.23)
Mono-APC α/β-peptide 6 was used as a reference for assessing
cooperative catalysis by bis-APC analogues 7−9. No initial rate
enhancement was detected when APC residues were placed as
close together as possible in terms of sequence (i, i+2; 7). A
modest enhancement was observed for i, i+4 APC spacing (8;
νREL = 21), which corresponds to a little more than one turn of
the expected helix. Further sequential separation of the APC
residues (i, i+6) caused a diminution in initial rate enhance-
ment (9; νREL = 7). The general pattern of an increase in νREL
followed by a decrease as the APC residues are moved farther
apart in sequence, with a maximum catalytic effect for spacing
near one helical turn, is common to this 1:1 α/β-peptide series
and the β-peptide series 2−5. However, the maximum νREL is
higher for the β-peptide series, which suggests that the helix
formed by the pure β backbone can achieve a more favorable
geometry for the secondary amine diad than is accessible with
the 1:1 α:β backbone.
α/β-Peptides 10−14 (Figure 4C) have a 1:2 α:β backbone
repeat, and available crystal structures indicate that these
foldamers favor a helix containing CO(i)···H-N(i+3) H-
bonds and ∼3 residues per turn.22,24 In this conformation, i, i
+3 spacing should place APC residues almost exactly one turn
apart. Initial rate data for the 1:2 α/β-peptide series suggest
that this secondary amine diad arrangement is particularly
favorable for promoting the crossed aldol condensation, as 13
displayed an initial rate enhancement relative to mono-APC α/
β-peptide 10 (νREL = 143) that was larger than any seen in the
previous two peptide series. In contrast, placing the two APC
residues adjacent in sequence, as in 11, offered no rate
enhancement, behavior that matches observations with the
other two foldamer backbones. An APC pair with i, i+2 spacing
was also catalytically ineffective (12). Lengthening the
separation between APC residues beyond one helical turn, to
i, i+4 (14), resulted in very limited reactivity (νREL = 8).
We conducted further studies with α/β-peptide 13 because
this foldamer appears to provide a particularly favorable
secondary amine diad geometry for bifunctional catalysis of the
crossed aldol condensation. To ask whether catalytic efficacy
was specific to hydrocinnamaldehyde, we examined the crossed
aldol reaction between hexanal and formaldehyde. Based on
comparison with the mono-APC α/β-peptide 10, α/β-peptide
13 displayed a relative initial rate of 154 8, which is very
Figure 4. Foldamers used to evaluate different APC diad geometries.
(A) β-peptides, (B) 1:1 α/β-peptides, and (C) 1:2 αβ-peptides. vREL
represents average initial rate
standard deviation relative to the
initial rate for the mono-APC peptide in each series (i.e., 1, 6, or 10),
based on a minimum of three independent measurements for each
value. APC residues are highlighted in red, and N-methyl APC
residues are highlighted in blue (for control compounds 15 and 16).
substrates, was divided by the initial rate observed with 2 mol
% 1 to generate νREL. This approach ensured that the
concentration of APC units was constant across these
measurements. The data for β-peptide 2 show that placing
two APC residues adjacent in sequence has no effect on νREL
,
which suggests that two pyrrolidine rings juxtaposed in this
way cannot work cooperatively to catalyze the crossed aldol
condensation. This conclusion is expected based on the helical
conformation established for these β-peptides,21 which has
∼2.5 residues per turn and should induce a divergent
orientation of the pyrrolidine ring nitrogen atoms of the
sequentially adjacent APC residues.
C
DOI: 10.1021/jacs.8b05869
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
similar to the value measured for hydrocinnamaldehyde
(Figure 4). Thus, catalytic efficacy of 13 does not seem to
depend on the identity of the nucleophilic aldehyde.
Our studies were not motivated by preparative consid-
erations, but we nevertheless established that 13 is competent
to drive the crossed aldol reaction to completion. After 22 h, a
98% yield of the α-methylenation product was obtained in the
presence of 5 mol % 13. In contrast, only a 6% yield was
obtained after the same period in the presence of 10 mol %
mono-APC α/β-peptide 10. Yield data for other α/β-peptides
can be found in Table S11.
Altering the dipeptide linking segment between the i, i+3-
spaced APC residues caused modest but significant changes in
the initial rate of product formation. Replacing the intervening
ACPC residue of 13 with (S)-β3-hLeu (Figure S11) caused a
decline in relative initial rate (νREL = 73). This β residue
replacement would be expected to decrease helical propensity,
based on comparison of cyclic and β3 residues in other α/β-
peptide backbones.25,27 In contrast, an increase in initial rate
(νREL = 185) was observed upon replacement of the
intervening Aib residue with L-Ala, a change that should
enhance conformational freedom. The α/β-peptide containing
both changes showed intermediate reactivity (νREL = 114).
Removal of all side chains, that is, replacement of ACPC-Aib
with β-hGly-Gly, resulted in intermediate reactivity as well
(νREL = 102). The variation in crossed aldol reactivity observed
among 13 and the analogues with “relaxed” linkers spans <3-
fold difference in νREL, but these findings suggest that
modulating the conformational mobility of the segment
between the reactive sites offers a path to improving the
efficacy of bifunctional foldamer catalysts.
The crossed aldol condensation is first order in catalyst:
varying the concentration of α/β-peptide 13 between 1 mol %
and 10 mol % caused a linear change in the initial rate of
product formation (Figures S20 and S21). Reactions run with
1 mol % 13 relative to formaldehyde and varying amounts of
hydrocinnamaldehyde revealed that the initial rate of product
formation increased as hydrocinnamaldehyde was increased
from 1 mol % to 100 mol % (relative to formaldehyde);
however, further increases in the amount of hydrocinnamalde-
hyde had little effect on initial rate (Figures S22 and S23).
Reactions run with 1 mol % 13 relative to hydrocinnamalde-
hyde and varying amounts of formaldehyde, between 50 mol %
and 1000 mol %, showed that rising formaldehyde
concentration exerts a mild inhibitory effect on the initial
rate of product formation (Figures S24 and S25). This
inhibition may indicate that formaldehyde reacts more avidly
than does hydrocinnamaldehyde with the APC side-chain
nitrogen atoms.
Figure 5. UPLC data (220 nm) depicting reaction progress. Red,
blue, and green boxes highlight the regions containing peaks for
catalyst 13, hydrocinnamaldehyde + peptide intermediates, and the
crossed aldol product, respectively. Reaction mixture (A) at 0.5 min
after mixing, (B) at 7.5 min, and (C) at 17 h, after the starting
materials have been fully consumed. Catalyst 13 rapidly disappears
after exposure to the reaction conditions (A → B), and the peak for
the catalyst reappears upon reaction completion (C).
possibility that is supported by the observations discussed
below.
Our experimental observations are accommodated by the
catalytic cycle proposed in Figure 6A. The bis-APC foldamer
catalyst 13 is represented generically by structure A. The
hypothetical catalytic mechanism involves condensation of one
APC nitrogen with hydrocinnamaldehyde and the other APC
nitrogen with formaldehyde to generate enamine-iminium
intermediate B. Intramolecular attack of the enamine on the
iminium forms iminium C, which contains a transient cross-
link. C could tautomerize to D, followed by N-protonation and
C−N bond scission to generate E. Alternatively, C could be
transformed directly to E. Hydrolysis of the iminium in E
would liberate the product and regenerate A. The cross-linked
intermediate proposed above could correspond to C or D, or a
mixture of the two. The proposed mechanistic cycle
corresponds to intramolecular dual covalent catalysis.
We tested this hypothesis by preparing isomeric α/β-
peptides 15 and 16 (Figure 4C), each of which is derived from
13 and bears a methyl group on one of the APC ring nitrogens.
Replacing either of the secondary amino groups in 13 with a
tertiary amino group precludes formation of enamine-iminium
intermediate B. Neither 15 nor 16 displays a significant
increase in initial rate relative to mono-APC α/β-peptide 10.
These observations support our hypothesis that the crossed
aldol rate enhancement observed for 13 emerges from dual
covalent catalysis.
We sought further support for the mechanism proposed in
Figure 6A by introducing NaBH4 after partial completion of
the crossed aldol condensation catalyzed by 13. This reagent
should reduce the proposed iminium intermediates to tertiary
amines, which are stable and therefore amenable to detection
via mass spectrometry. MS analysis of reaction mixtures
generated via reductive trapping revealed species with diverse
m/z values, including 969.263, which for z = +1 could
correspond to the cross-linked α/β-peptide F (from reaction of
C with NaBH4), the alkene G (from reaction of E with
NaBH4), or a mixture of F and G (Figure 6B). We
distinguished among these possibilities by conducting the
Monitoring of reaction progress by UPLC (equimolar
aldehydes, 1 or 10 mol % 13) revealed that the α/β-peptide
peak disappears upon mixing with the starting materials and
reappears only after product formation is complete (Figure 5;
reaction conducted with 10 mol % 13). In contrast, α/β-
peptide 13 is fully detectable by UPLC when mixed with 100
equiv of either formaldehyde or hydrocinnamaldehyde
(Figures S41 and S42). These observations support the
hypothesis that catalysis requires the coordinated action of
two secondary amine units, and they raise the possibility that
the catalytic mechanism proceeds via a relatively stable
intermediate that is formed only in the presence of both
aldehyde substrates. This hypothetical intermediate might
contain a transient cross-link between the APC residues, a
D
DOI: 10.1021/jacs.8b05869
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 6. (A) Proposed catalytic cycle for the foldamer-catalyzed crossed aldol reaction. Hydrocinnamaldehyde-derived atoms are shown in red,
and formaldehyde-derived atoms are shown in blue. (B) Possible products from reductive trapping of the catalytic reaction mixture. Circled in red
in intermediate F is the α-hydrogen derived from hydrocinnamaldehyde. Circled in blue in intermediate H is the α-deuterium derived from α,α-
dideutero-hydrocinnamaldehyde.
reductive trapping experiment with α,α-dideutero-hydro-
cinnamaldehyde (Figure 7).
that a particularly stable intermediate is formed under the
reaction conditions. We speculate that a transient cross-link, as
in C or D, explains this level of stability. Since α/β-peptide 13
can be detected via UPLC after complete consumption of the
aldehyde starting materials, we hypothesize that the proposed
cross-linked intermediate is not sufficiently stable to inhibit
catalysis.
Exposure of α/β-peptide 13 to the enal product under the
reaction conditions (in the absence of formaldehyde or
hydrocinnamaldehyde), followed by reductive trapping and
MS analysis, led to detection of an adduct between the enal
and 13. This observation motivated us to explore the
possibility that the cross aldol reaction process is fully
reversible via a crossover experiment. α/β-Peptide 13,
hydrocinnamaldehyde, and the enal formed from hexanal
were combined under the reaction conditions. However, no
crossover product (i.e., hexanal or the enal expected from
hydrocinnamaldehyde) was detected (after 70 min). Moreover,
when this reaction was monitored by LC-MS, the peak for 13
did not disappear, which suggests that the reaction between
catalyst 13 and an enal, which was detected via reductive
trapping, does not lead to formation of the cross-linked
In this case, the cross-linked product generated via reduction
of C would be predicted to retain one deuterium (species H),
while the alkene generated via reduction of E would be
predicted to contain no deuterium, leading to a difference
between these products of 1 amu. MS analysis revealed that
both the cross-linked α/β-peptide H (monodeuterated) and
alkene G are formed (Figure S43).
We conducted analogous reductive trapping studies after
exposure of α/β-peptide 13 to 100 equiv of only one of the
aldehyde substrates under the reaction conditions. In each
case, MS analysis of the product mixture detected the presence
of mono- and dialkylated derivatives of 13 (e.g., when
formaldehyde was used, m/z values consistent with mono-
and dimethyl derivatives of 13 were observed). Thus, failure of
UPLC to detect iminium/enamine derivatives of 13 after
exposure to 100 equiv of one aldehyde substrate or the other,
described above, suggests that these types of intermediates
hydrolyze rapidly under the chromatographic conditions. The
ability of UPLC to detect transiently modified forms of α/β-
peptide 13 in the presence of both aldehyde substrates suggests
E
DOI: 10.1021/jacs.8b05869
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
therefore renders the crossed aldol process particularly useful
for assessment of reactivities among catalyst candidates.29,30
Variations in the sequential spacing between APC units
strongly affect catalytic efficacy, as would be predicted if
distinct sequential juxtapositions of reactive groups are
translated by folding into distinct spatial relationships of
those groups. The best bis-APC α/β-peptide catalysts enhance
the initial rate of product formation by two orders of
magnitude relative to a mono-APC control compound. This
observation is encouraging in terms of ongoing exploration of
foldamer-based catalysis.
Our work complements other efforts to develop foldamer
catalysts by highlighting the way that access to multiple related
secondary structures, via predictable sequence-design strat-
egies, enables evaluation of alternative reactive group geo-
metries.31,32 β3-Homoleucine oligomers have been used in
place of α-leucine oligomers to promote asymmetric enone
epoxidation with hydrogen peroxide; exposed helix termini are
presumed to be sites of H-bond-mediated catalysis.33 Attach-
ment of a catalytically active nitroxyl unit to a helical peptoid
enabled kinetic resolution of 1-phenylethanol via preferential
oxidation of one enantiomer.34 In this case, the chiral foldamer
presumably plays a steric role, selectively impeding approach of
one alcohol enantiomer to the nitroxyl unit. Two examples of
catalysis by self-assembling helical β-peptides have been
reported. One involves a retro-aldol reaction that depends
upon nucleophilic reactivity of a β3-hLys side chain.35 The
other case involves activated ester hydrolysis that requires
terminal α-histidine residues.36 Recently described helical
oligo-urea foldamers promote the asymmetric conjugate
addition of dimethyl malonate to nitroalkenes, apparently via
H-bond-mediated activation of the acceptor.37 In conceptually
related work, rigid polycyclic molecular scaffolds have been
used to develop catalysts for reactions such as aldol
condensations.6 Oligomers based on four different xeno-
nucleic acid backbones that display endoribonuclease activity
have been selected from large candidate pools generated via
engineered polymerase enzymes.38 This type of selection-based
catalyst discovery is presently impossible with peptidic
foldamers.
Our work is distinctive relative to these precedents because
of our focus on and evidence for bifunctional catalysis. We
have shown that the availability of a family of distinct but
related helices (Figure 2) enables rapid assessment of
alternative spatial organizations of the reactive diad. This
aspect of our work highlights the value of continued efforts to
identify new foldamer scaffolds. Previous catalyst development
efforts with foldamers have typically been based on a single
folding pattern.34−37 Analogous designs with conventional
peptides have focused on the α-helix,39,40 although selections
of short α-peptide catalysts from combinatorial libraries have
shown that nonhelical conformations can be effective for
reactive group organization.41 Our foldamer-based approach is
advantageous because multiple helical scaffolds with comple-
mentary shapes are available among β- and α/β-peptides, and
the intrinsic modularity of peptides makes it straightforward to
alter diad geometry within each scaffold via sequence changes.
Much of the prior work with synthetic α-peptide and
foldamer catalysts has emphasized stereo- and regioselectivity
in product formation,7−12,29,34,39−41 while the studies reported
here focus on reactivity. The best catalytic enhancements we
observed are small relative to those of highly evolved natural
enzymes;1 however, our approach has necessarily been limited
Figure 7. MALDI-TOF MS spectra depicting reduced intermediates.
(A) Reaction carried out with hydrocinnamaldehyde as the
nucleophile. Observed mass of 969.263 corresponds to F, G, or
both. (B) Reaction carried out with α,α-dideutero-hydrocinnamalde-
hyde as the nucleophile. Mass corresponding to G is observed, as well
as 970.300, corresponding to deuterated macrocyclic intermediate H.
intermediate. These results led us to conclude that the crossed
aldol reaction is not reversible under our conditions.
The proposed crossed aldol mechanism (Figure 6A) raises
the possibility that a bifunctional catalyst would be effective for
an aldol-based cyclization. Indeed, under the conditions
employed for the crossed aldol reaction, combining 1,8-
octanedialdehyde with 10 mol % of the best catalyst, α/β-
peptide 13, generates a 75% yield of cycloheptene-1-
carbaldehyde after 140 min (Figure 8). Under identical
Figure 8. Aldol-based cyclization of 1,8-octanedialdehyde.
conditions, isomeric α/β-peptide 12 generates an 8% yield of
the cyclic enal. Previously, this intramolecular aldol reaction
has been achieved by refluxing the dialdehyde in dichloro-
methane for 2 days with 76 mol % proline.28
CONCLUSIONS
■
We have employed a crossed aldol condensation to compare
the abilities of different helical scaffolds to achieve pyrrolidine
(APC) diad geometries that support bifunctional catalysis. The
lack of stereochemical complexity in this crossed aldol reaction
streamlines initial rate analysis via product formation and
F
DOI: 10.1021/jacs.8b05869
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(6) Zhao, Q.; Lam, Y.; Kheirabadi, M.; Xu, C.; Houk, K. N.;
Schafmeister, C. E. J. Org. Chem. 2012, 77, 4784−4792.
(7) Miller, S. J. Acc. Chem. Res. 2004, 37, 601−610.
(8) Doyle, A. G.; Jacobsen, E. N. Chem. Rev. 2007, 107, 5713−5743.
(9) Park, Y.; Harper, K. C.; Kuhl, N.; Kwan, E. E.; Liu, R. Y.;
Jacobsen, E. N. Science 2017, 355, 162−166.
to short, isolated secondary structures as scaffolds for reactive
group organization (six or seven residues). In contrast, all
known enzyme active sites are embedded within tertiary
structures. This distinction is significant, because concave
active sites allow control over substrate solvation,42,43 but an
active site created along the side of a single helix is solvent-
exposed. Noteworthy progress in de novo enzyme design has
been reported, but these efforts have not yet achieved the rate
enhancements manifested by proficient enzymes.44 The
favorable reactivity we observe for an aldol-based cyclization
with the optimal foldamer suggests that even rudimentary
catalysts involving an isolated secondary structure might have
facilitated the maturation of prebiotic reaction networks.
Enzymes typically contain hundreds of α-amino acid
residues, which are necessary to generate globular folds with
enclosed active sites. It is unlikely, however, that such long
polypeptides existed in the prebiotic period. We therefore
speculate that exploring the capabilities of isolated unnatural
secondary structures as scaffolds for bifunctional catalysis
represents a first step, inspired by prebiotic hypotheses
regarding the origins of enzymes,45 toward development of
larger foldamers that adopt discrete tertiary structures and
approach enzyme-like reactivities.
(10) DiRocco, D. A.; Ji, Y.; Sherer, E. C.; Klapars, A.; Reibarkh, M.;
Dropinski, J.; Mathew, R.; Maligres, P.; Hyde, A. M.; Limanto, J.;
Brunskill, A.; Ruck, R. T.; Campeau, L.; Davies, I. W. Science 2017,
356, 426−430.
(11) Wiesner, M.; Revell, J. D.; Wennemers, H. Angew. Chem., Int.
Ed. 2008, 47, 1871−1874.
(12) Liu, L.; Cotelle, Y.; Avestro, A.; Sakai, N.; Matile, S. J. Am.
Chem. Soc. 2016, 138, 7876−7879.
(13) Gellman, S. H. Acc. Chem. Res. 1998, 31, 173−180.
(14) Hill, D. J.; Mio, M. J.; Prince, R. B.; Hughes, T. S.; Moore, J. S.
Chem. Rev. 2001, 101, 3893−4012.
(15) Guichard, G.; Huc, I. Chem. Commun. 2011, 47, 5933−5941.
̈
(16) Erkkila, A.; Pihko, P. M. Eur. J. Org. Chem. 2007, 2007, 4205−
4216.
(17) Cheng, R. P.; Gellman, S. H.; DeGrado, W. F. Chem. Rev. 2001,
101, 3219−3232.
(18) Fernandes, C.; Faure, S.; Pereira, E.; Thery, V.; Declerck, V.;
Guillot, R.; Aitken, D. J. Org. Lett. 2010, 12, 3606−3609.
(19) Winkler, J. D.; Piatnitski, E. L.; Mehlmann, J.; Kasparec, J.;
Axelsen, P. H. Angew. Chem., Int. Ed. 2001, 40, 743−745.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.8b05869.
(20) Martinek, T. A.; Fulop, F. Chem. Soc. Rev. 2012, 41, 687−702.
̈
̈
■
(21) Appella, D. H.; Christianson, L. A.; Klein, D.; Powell, D. R.;
Huang, X.; Barchi, J. J.; Gellman, S. H. Nature 1997, 387, 381−384.
(22) Schmitt, M. A.; Choi, S. H.; Guzei, I. A.; Gellman, S. H. J. Am.
Chem. Soc. 2006, 128, 4538−4539.
S
Experimental procedures, initial rate plots, circular
dichroism, calibration curves, UPLC spectra, and
MALDI-TOF spectra (PDF)
(23) Choi, S. H.; Guzei, I. A.; Spencer, L.; Gellman, S. H. J. Am.
Chem. Soc. 2008, 130, 6544−6550.
(24) Choi, S. H.; Guzei, I. A.; Spencer, L.; Gellman, S. H. J. Am.
Chem. Soc. 2008, 130, 6544−6550.
(25) Johnson, L. M.; Gellman, S. H. Methods Enzymol. 2013, 523,
407−429.
AUTHOR INFORMATION
Corresponding Author
*gellman@chem.wisc.edu
■
̈
(26) Erkkila, A.; Pihko, P. M. J. Org. Chem. 2006, 71, 2538−2541.
(27) Price, J. L.; Hadley, E. B.; Steinkruger, J. D.; Gellman, S. H.
Angew. Chem., Int. Ed. 2010, 49, 368−371.
ORCID
(28) Miller, S. A.; Bobbitt, J. A.; Leadbeater, N. E. Org. Biomol.
Chem. 2017, 15, 2817−2822.
Zebediah C. Girvin: 0000-0001-5338-1319
(29) Alford, J. S.; Abascal, N. C.; Shugrue, C. R.; Colvin, S. M.;
Romney, D. K.; Miller, S. J. ACS Cent. Sci. 2016, 2, 733.
(30) Giuliano, M. W.; Miller, S. J. Top. Curr. Chem. 2015, 372, 157.
(31) Metrano, A. J.; Abascal, N. C.; Mercado, B. Q.; Paulson, E. K.;
Hurtley, A. E.; Miller, S. J. J. Am. Chem. Soc. 2017, 139, 492−516.
(32) Schnitzer, T.; Wennemers, H. J. Am. Chem. Soc. 2017, 139,
15356−15362.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This research was supported in part by the U.S. National
Science Foundation (CHE-1565810). Additional support was
provided by the Office of the Vice Chancellor for Research and
Graduate Education at the University of Wisconsin-Madison
with funding from the Wisconsin Alumni Research Founda-
tion.
(33) Coffey, P. L.; Drauz, K. H.; Roberts, S. M.; Skidmore, J.; Smith,
J. A. Chem. Commun. 2001, 2330−2331.
(34) Maayan, G.; Ward, M. D.; Kirshenbaum, K. Proc. Natl. Acad.
Sci. U. S. A. 2009, 106, 13679−13684.
̈
(35) Muller, M. M.; Windsor, M. A.; Pomerantz, W. C.; Gellman, S.
H.; Hilvert, D. Angew. Chem., Int. Ed. 2009, 48, 922−925.
(36) Wang, P. S. P.; Nguyen, J. B.; Schepartz, A. J. Am. Chem. Soc.
2014, 136, 6810−6813.
DEDICATION
■
This paper is dedicated to the memory of Professor Ronald
Breslow.
(37) Becart, D.; Diemer, V.; Salaun, A.; Oiarbide, M.; Nelli, Y. R.;
Kauffmann, B.; Fischer, L.; Palomo, C.; Guichard, G. J. Am. Chem.
Soc. 2017, 139, 12524−12532.
REFERENCES
■
(38) Taylor, A. I.; Pinheiro, V. B.; Smola, M. J.; Morgunov, A. S.;
Peak-Chew, S.; Cozens, C.; Weeks, K. M.; Herdewijn, P.; Holliger, P.
Nature 2015, 518, 427−430.
(39) Akagawa, K.; Kudo, K. Acc. Chem. Res. 2017, 50, 2429−2439.
(40) Kinghorn, M. J.; Valdivia-Berroeta, G. A.; Chantry, D. R.;
Smith, M. S.; Ence, C. C.; Draper, S. R. E.; Duval, J. S.; Masino, B. M.;
Cahoon, S. B.; Flansburg, R. R.; Conder, C. J.; Price, J. L.; Michaelis,
D. J. ACS Catal. 2017, 7, 7704−7708.
(1) Wolfenden, R. Annu. Rev. Biochem. 2011, 80, 645−667.
(2) Herschlag, D.; Natarajan, A. Biochemistry 2013, 52, 2050−2067.
(3) Khersonsky, O.; Tawfik, D. S. Annu. Rev. Biochem. 2010, 79,
471−505.
(4) Desper, J. M.; Breslow, R. J. Am. Chem. Soc. 1994, 116, 12081−
12082.
(5) Cramer, K. D.; Zimmerman, S. C. J. Am. Chem. Soc. 1990, 112,
3680−3682.
G
DOI: 10.1021/jacs.8b05869
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(41) Lewis, C. A.; Gustafson, J. L.; Chiu, A.; Balsells, J.; Pollard, D.;
Murry, J.; Reamer, R. A.; Hansen, K. B.; Miller, S. J. J. Am. Chem. Soc.
2008, 130, 16358−16365.
(42) Narlikar, G.; Gopalakrishnan, V.; McConnell, T. S.; Usman, N.;
Herschlag, D. Proc. Natl. Acad. Sci. U. S. A. 1995, 92, 3668−3672.
(43) Warshel, A. J. Biol. Chem. 1998, 273, 27035−27038.
(44) Jiang, L.; Althoff, E. A.; Clemente, F. R.; Doyle, L.;
Rothlisberger, D.; Zanghellini, A.; Gallaher, J. L.; Betker, J. L.;
Tanaka, F.; Barbas, C. F., III; Hilvert, D.; Houk, K. N.; Stoddard, B.
L.; Baker, D. Science 2008, 319, 1387−1391.
(45) Romero, M. L.; Rabin, A.; Tawfik, D. S. Angew. Chem., Int. Ed.
2016, 55, 15966−15971.
H
DOI: 10.1021/jacs.8b05869
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX