The role of imidazole in peptide cyclization by transesterification: Parallels to the catalytic triads of serine proteases

Article abstract of DOI:10.1039/c3ob27464k

The improved bioavailability, stability and selectivity of cyclic peptides over their linear counterparts make them attractive structures in the design and discovery of novel therapeutics. In our previous work, we developed an imidazole-promoted preparation of cyclic depsipeptides in which we observed that increasing the concentration of imidazole resulted in the concomitant increase in the yield of cyclic product and reduction in dimerization, but also resulted in the generation of an acyl-substituted side product. In this work, we used transition state analysis to explore the mechanism of the imidazole-catalyzed esterification of one such peptide, Ac-SAFYG-SCH₂φ, and determined the acyl substitution product to be an intermediate in a competing reaction pathway involving acyl substitution of the thioester by imidazole. Our findings indicate that imidazole plays an essential role in this side-chain to C-terminal coupling, and by extension, in transesterifications in general, through a concerted mechanism wherein imidazole deprotonates the nucleophile as the nucleophile attacks the carbonyl. The system under study is identical to the histidine-serine portion of the catalytic triads in serine proteases and it is likely that these enzymes employ the same concerted mechanism in the first step of peptide cleavage. Additionally, relatively high concentrations of imidazole must be used to effectively catalyze reactions in aprotic solvents since the overall reaction involves imidazole acting both as an acid and as a base, existing in solution as an equilibrium distribution between the neutral form and its conjugate acid. The Royal Society of Chemistry 2013.

Full text of DOI:10.1039/c3ob27464k

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
PAPER
The role of imidazole in peptide cyclization by
transesterification: parallels to the catalytic triads of
Cite this: Org. Biomol. Chem., 2013, 11,
2979
serine proteases
Kendall G. Byler,*a Yangmei Li,a Richard A. Houghtena and
Karina Martinez-Mayorga*a,b
The improved bioavailability, stability and selectivity of cyclic peptides over their linear counterparts make
them attractive structures in the design and discovery of novel therapeutics. In our previous work, we
developed an imidazole-promoted preparation of cyclic depsipeptides in which we observed that increas-
ing the concentration of imidazole resulted in the concomitant increase in the yield of cyclic product and
reduction in dimerization, but also resulted in the generation of an acyl-substituted side product. In this
work, we used transition state analysis to explore the mechanism of the imidazole-catalyzed esterification
of one such peptide, Ac-SAFYG-SCH2ϕ, and determined the acyl substitution product to be an intermedi-
ate in a competing reaction pathway involving acyl substitution of the thioester by imidazole. Our
findings indicate that imidazole plays an essential role in this side-chain to C-terminal coupling, and by
extension, in transesterifications in general, through a concerted mechanism wherein imidazole deproto-
nates the nucleophile as the nucleophile attacks the carbonyl. The system under study is identical to the
histidine-serine portion of the catalytic triads in serine proteases and it is likely that these enzymes
employ the same concerted mechanism in the first step of peptide cleavage. Additionally, relatively high
concentrations of imidazole must be used to effectively catalyze reactions in aprotic solvents since the
overall reaction involves imidazole acting both as an acid and as a base, existing in solution as an equili-
brium distribution between the neutral form and its conjugate acid.
Received 20th December 2012,
Accepted 9th March 2013
DOI: 10.1039/c3ob27464k
www.rsc.org/obc
at least a part of their importance in nature to their tendency
to resist degradation by exopeptidases. In addition, cyclization
Introduction
Biologically active cyclic peptides have been found in an tends to increase the rigidity of the three-dimensional struc-
increasing number of plant and animal species. Chain length ture and this can add further to the resistance to peptidases in
can vary widely, as can the mode of cyclization. In homodetic general8 and greatly enhance the intrinsic biological activity of
(‘head-to-tail’) cyclic peptides, the cyclic backbone consists the molecule.9 The beneficial impact of this structural rigidity
exclusively of true peptide bonds, as in the cyclooctapeptide on activity can be further enhanced by the incorporation of
planktocyclin, which is a strong inhibitor of mammalian additional disulfide bridges into the structure, as was recently
trypsin and chymotrypsin.1 While in heterodetic peptides, the shown by Pakkala et al. in cyclic peptides containing disulfide
backbone includes at least one other type of linkage such as bridges with promising prostate anti-cancer activity.10
ester bonds, isopeptide bonds, or disulfide bridges. Examples
The synthesis of cyclic peptides and depsipeptides based
of such structures are cyclic depsipeptides,2 glutathione, the on structures from natural products is well established11 and
conotoxins,3 and the trypsin inhibitor SFTI-1, with a single di- methods of cyclizing linear peptides has been described in the
sulfide bridge.4 The biological and medical significance of literature.12 As the cyclization of biologically-active peptides
cyclic peptides as potential therapeutics is evinced by an has been shown to improve both bioavailability and selectivity,
increasing number of publications describing their antiviral,5 the adoption of these structures in the development of combi-
hemolytic6 and anti-helminthic7 activities. Cyclic peptides owe natorial libraries for drug discovery is gaining importance in
pharmaceutical research initiatives. The design, synthesis and
screening of large combinatorial libraries as mixtures has
enabled us to discover potent and selective ligands for a
variety of biologically relevant targets13 with substantial
improvements in both time and cost over conventional high
aTorrey Pines Institute for Molecular Studies, 11350 Southwest Village Parkway, Port
St. Lucie, FL 34987, USA. E-mail: kmartinez@tpims.org
bInstituto de Química, UNAM, Mexico City 04510, Mexico
This journal is © The Royal Society of Chemistry 2013
Org. Biomol. Chem., 2013, 11, 2979–2987 | 2979

View Article Online
Paper
Organic & Biomolecular Chemistry
throughput screening. Applying these synthetic and screening histidine, and serine. Exhibiting an effective pKa well above
techniques to libraries of cyclic peptides should lead to the the usual values found in proteins,17 the histidine imidazo-
rapid identification of compounds with promising therapeutic lium moiety deprotonates the serine, which then acts as a
profiles. Efficient strategies for the synthesis of combinatorial nucleophile toward the amide carbonyl of a bound peptide
libraries have been well-documented in the literature,14 yet the ligand. This reaction mechanism, highlighted in most bio-
application of these methods to the development of cyclic chemistry texts, has been studied extensively18 through residue
peptide libraries has been met with challenges related to deletion and kinetics studies, as well as by X-ray crystallogra-
peptide cyclization in general; specifically, side reactions, phy of enzymes in complex with bound ligands, and is typi-
dimerization/polymerization, and the effects of ring size and cally depicted graphically as a concerted mechanism, although
inclusion of non-turn-inducing residues on reaction rates and not explicitly described as such in the accompanying texts.
product yields. A recent study shows that, based on their con- Our mechanistic study of the cyclization of depsipeptides
formational properties, cyclotetrapeptides can be viewed as 3D using density functional theory (DFT) presented here proposes
scaffolds.14f Head-to-tail linkages typically involve activation of two isoenergetic cyclization pathways, provides one expla-
the C-terminal carboxyl group, followed by nucleophilic attack nation for the need of high concentrations of imidazole, and
of the activated carbonyl by the neutralized N-terminus. Depsi- ultimately shows how a concerted deprotonation step is ener-
peptide cyclizations, which involve linking an N-terminal getically feasible, coinciding with and implicitly confirming
serine side-chain to the C-terminus of the linear precursor, the histidine-mediated deprotonation step proposed to occur
require nucleophilic attack on a carbonyl by a weak nucleo- in cleavage by serine proteases. These findings provide
phile (the serine hydroxyl) in order to complete the cyclization, insights on the role of imidazole on chemical synthesis in
which, in turn, requires activation of the carbonyl group.
We recently developed an imidazole-promoted macrolacto-
nization of peptide thioesters to prepare cyclic depsipeptides.
The natural cyclic depsipeptide kahalalide B has been syn-
thesized in high yield using this method15 and peptide phenyl
ester ligation reactions by imidazole have been reported by
general.
Results
Experimental conditions for peptide cyclization
others.15b In our search for ideal experimental conditions for Typically, in N-to-C terminus macrolactamizations, a chemi-
depsipeptide synthesis, we found that cyclization only took cally activated C-terminal carbonyl reacts with a neutralized N-
place under high concentration of imidazole, typically around terminal amine to achieve cyclization of the peptide. In the
1 M. Imidazole is known as an acyl transfer catalyst,16 facilitat- present work, the side-chain hydroxyl group of the acylated N-
ing transfer in reactions between amines and dimethylform- terminal serine reacts with the C-terminal carbonyl, which has
amide and in lactam cyclizations that do not otherwise been activated through thioester transesterification, in order
proceed under standard conditions. In the process of synthe- to complete the cyclization. The structure of the pentapeptide
sizing some natural depsipeptides, we detected the presence of studied here, Ac-SAFYG-SCH2ϕ (I), is shown in Scheme 1,
a peptidyl acylimidazole as an intermediate in the reaction along with those of the cyclic monomer and cyclic dimer.
mixture, which allowed us to postulate cyclization pathways
involving the catalytic participation of imidazole.
The impact of imidazole on the macrolactonization of I was
investigated by adding increasing quantities of imidazole to
Key transitions along these pathways are structurally remini- promote cyclization (Table 1). It was found that adding what
scent of the systems found in the catalytic sites of serine pro- should be considered a catalytic amount of imidazole (0.5%
teases, which are responsible for the peptidase (and esterase) mole ratio) did not promote cyclization: conversion of the
activity of these enzymes. Serine proteases possess a triad of linear peptide thioester to the cyclic monomer and dimer was
residues in their active sites that lie in close proximity and very slow (5 days) and with low yields when the concentration
function as a charge relay system to catalyze the cleavage of of imidazole was below 0.5 M. Adding much larger quantities
peptide chains. The most frequently encountered of these cata- of imidazole, beyond expected catalytic amounts, dramatically
lytic triads consists of the three amino-acid residues aspartate, improved the yield of cyclic product (92% product composition
Scheme 1 The cyclization of Ac-SAFYG-SCH2ϕ (I) on the left yields the cyclic monomer (l) and the cyclic dimer (r).
2980 | Org. Biomol. Chem., 2013, 11, 2979–2987
This journal is © The Royal Society of Chemistry 2013

View Article Online
Organic & Biomolecular Chemistry
Paper
incoming nucleophile than does the thiobenzyl group in I,
which may contribute to the observed cyclization. It was pro-
posed, therefore, that the overall reaction was being catalyzed
by imidazole, proceeding via an acyl-substituted intermediate
(Scheme 2, II) that could provide a lower transition barrier to
cyclization versus that of the uncatalyzed pathway.
Table 1 The effects of imidazole concentration on reaction product compo-
sitions, as determined from LC-MS peaks, at room temperature with starting
peptide concentration of 1 mM
Percent composition
Imidazole
concentration (M)
Cyclic
monomer
Cyclic
dimer
Starting
peptide
Reaction
time (h)
Proposed mechanism of imidazole catalysis of peptide
cyclization
0.3
0.5
1.0
1.5
14.1
71.7
84.8
92.3
8.9
6.9
15.2
7.7
77.0
21.4
—
126
120
24
—
24
To rationalize the experimental findings, two reaction path-
ways were postulated involving the catalytic participation of
imidazole in the cyclization reaction of I. Using molecular frag-
at an imidazole concentration of 1.5 M), with the degree of ments to represent the N- and C-terminal ends of the larger
peptide cyclization increasing with increasing imidazole con- peptide, as shown in Fig. 1 and 2, we sought to model the ther-
centration. Maximal effect was observed at an imidazole con- modynamics of the transitions involved with the reactive por-
centration of 1.5 M (a 1000 : 1 mole ratio of imidazole to linear tions of the larger peptides using density functional theory. In
peptide), which is well beyond a conventional catalytic concen- the first pathway, imidazole replaces the thioate in an acyl sub-
tration. It is interesting that the yields of the cyclic monomer stitution reaction, cyclization then occurs with imidazole as
and the cyclic dimer are dependent on the imidazole concen- the final leaving group. In the second pathway, imidazole par-
tration, with greater effect to the formation of the cyclic ticipates in a more auxiliary capacity, stabilizing transition
monomer. This effect shows that the participation of imid- states and the alkoxide intermediate, with the thioate as the
azole in the cyclization mechanism has the same effect in the final leaving group. In both pathways, imidazole acts as a base,
formation of the monomer and the dimer. The dependence on abstracting a proton from the serine hydroxyl (here modeled
such large concentrations of imidazole suggested that imid- using methanol) and as its conjugate acid to protonate the
azole might be acting as a base, removing the hydroxyl proton phenylmethanethiolate leaving group, with no net change in
of serine and making it a much stronger nucleophile. But con- stoichiometry, thus acting as a true catalyst. Protonated imid-
sidering these results strictly in terms of basicity, it is difficult azolium ions were included in transition state complexes and
to rationalize how such a weak base could induce cyclization intermediate complexes to stabilize the carbonyl/alkoxide
and dimerization. The pKa of the serine hydroxyl is approxi- oxygens through hydrogen bonding or ion pairing. In the
mately 13, well above the pKa of imidazole in water (∼7.0): that actual peptide, it is likely that stabilization of the reactive sites
imidazole could abstract a proton from an alcohol, given the occurs through hydrogen bonding interactions with the
relative proton affinities of the species involved, seemed un- peptide backbone.
likely. Although the H-bonding energy of imidazole to the
The first step along the substitution pathway (Fig. 1)
hydroxyl hydrogen is at the upper end of the hydrogen bond involves nucleophilic attack by imidazole on the thioester car-
energy range (the imidazole–water analog has been calcu- bonyl, forming a tetrahedral alkoxide zwitterion. The imidazo-
lated19 to be ∼7 kcal mol−1), imidazole should not be expected lium moiety is deprotonated, leaving a negatively-charged
to abstract a proton from an alcohol in bulk solution.
alkoxide. Deprotonation of the acyl-substituted imidazolium
Another possibility regarding the mode of catalysis involved probably occurs rapidly after, if not during, the substitution,
the nucleophilic substitution of the C-terminal thiobenzyl as the calculated aqueous pKa for the substitution product is
group by imidazole, resulting in a peptidyl acyl imidazole estimated to be 2.86. Bender and Turnquist concluded20 from
(Scheme 2). We had previously observed this type of intermedi- the results of their kinetics study of the imidazole-catalyzed
ate in the reaction mixtures of some cases of difficult to cyclize hydrolysis of p-nitrophenyl acetate that either scenario is
depsipeptides. This suggested the presence of an alternate plausible. An imidazolium cation forms a complex with the
cyclization pathway in which the acyl imidazole structure rep- hemithioacetal oxygen through hydrogen bonding and/or ion
resented a reaction intermediate and which might be more pairing, finally protonating the alkoxide to give the first inter-
energetically favorable. It is also possible that the imidazolium mediate, I1. The subsequent transition is marked by the loss
moiety in (Scheme 2, II) presents slightly less steric inhibition of the thioate leaving group, which may or may not be proto-
to the formation of a van der Waals complex with the nated prior to leaving, to give the experimentally observed
Scheme 2 The peptide Ac-SAFYG-SCH2ϕ (I) on the left and the imidazole substitution product (II) on the right.
This journal is © The Royal Society of Chemistry 2013
Org. Biomol. Chem., 2013, 11, 2979–2987 | 2981

View Article Online
Paper
Organic & Biomolecular Chemistry
Fig. 1 Proposed mechanism of imidazole catalysis in the acyl substituted reaction pathway using model fragments of the larger peptide.
Fig. 2 Proposed mechanism of imidazole catalysis in the unsubstituted reaction pathway using model fragments of the larger peptide.
intermediate, I2. From the production of the acyl-substitution mol−1 above the reference energy at TS3, where a concerted
product, both pathways undergo analogous chemical tran- deprotonation and nucleophilic attack of the acyl-substituted
sitions, beginning with an SN2 attack on the carbonyl by intermediate takes place with a scaled imaginary frequency of
methanol/methoxide. Along the unsubstituted pathway 627.6 cm−1. The overall energetic barrier that must be over-
(Fig. 2), acyl substitution occurs with the thioate as the leaving come following this reaction pathway, then, is 14.8 kcal mol−1
group, while in the substitution pathway, the leaving group is The relative ZPVE’s for structures along the unsubstituted
.
an imidazolium that must be protonated prior to leaving. pathway are shown in Fig. 4. The highest energy barrier along
While the transitions involving the thioate leaving group are this pathway lies at TS3*, which is the equivalent transition in
energetically accessible to the system, ions are not well sol- the substitution pathway, TS3, corresponding to the concerted
vated in polar aprotic solvents such as acetonitrile, and the deprotonation of the alcohol with attack on the carbonyl, with
thioate, with a pKa in DMSO of 15.3,21 should be expected to a scaled imaginary frequency of 660.3 cm−1. The energy
readily abstract a proton from imidazolium cation.
associated with this transition is 9.36 kcal mol−1 above the
reference energy, and only 0.15 kcal mol−1 below that of TS3.
This small energy difference is within the reported error of
Transition state analysis
The zero-point vibrational energies (ZPVE) for the structures 0.2 kcal mol−1 for transition state energies calculated using
along the acyl substitution pathway were plotted with respect our model chemistry and should thus be considered as energe-
to the ZPVE of the initial system and are shown in Fig. 3. The tically equivalent to the corresponding transition along the
largest energy barrier along this pathway, at 12.05 kcal mol−1 substitution pathway and the approximate activation energy
above the reference energy, lies at TS1, where an incoming imid- for both pathways.
azole attacks the carbonyl carbon in the first transition
Transition state energies were lowered substantially when
associated with acyl substitution of the thioester. In this tran- imidazolium cations were allowed to form complexes with the
sition state, an imidazole was not included to stabilize the transitional carbonyl/alkoxide oxygens. The energies of the
forming alkoxide oxygen, but is estimated to confer 2–3 kcal intermediates were lowered as well, with imidazolium cations
mol−1 in stabilization to the complex (from calculations of the protonating the alkoxides and remaining in complex with the
I2 structure with and without complexed imidazole – this is resulting tetrahedral intermediates. The imidazolium cation in
described in more detail in the Discussion section). Adjusting complex with the acyl substitution product, I2, offers approxi-
for the effect of imidazole stabilization places the activation mately −3 kcal mol−1 of stabilization over I2 alone in conti-
energy on par with the next largest energy barrier of 9.51 kcal nuum solution. Transition states corresponding to the serine
2982 | Org. Biomol. Chem., 2013, 11, 2979–2987
This journal is © The Royal Society of Chemistry 2013

View Article Online
Organic & Biomolecular Chemistry
Paper
Fig. 3 Optimized geometries and scaled zero-point vibrational energies relative to that of the starting structure along the acyl substitution pathway, with energies
in kcal mol−1 (light gray bars reference the energies along the unsubstituted pathway).
Fig. 6 Activation energies for transition states TS3 and TS3* from IRC calcu-
lations, relative to the energy of the starting structure, showing a slightly higher
energy of activation for the acyl substituted structure, but leading to a more
Fig. 4 Optimized geometries and scaled zero-point vibrational energies rela-
stable product.
tive to that of the starting structure along the unsubstituted pathway, with ener-
gies in kcal mol−1 (light gray bars reference the energies along the acyl
substitution pathway).
gas phase and in continuum solvation consistently failed to
converge or yielded higher order saddle points that could not
be resolved to first order points on the potential mean force
(PMF) surfaces.
Following the electronic energies along the intrinsic reac-
tion coordinate (IRC) of the two pathways, the unsubstituted
pathway is a slightly more energetically-favorable route, with
an activation energy at TS3* of 8.97 kcal mol−1 above the
energy of the initial system versus 9.50 kcal mol−1 for TS3
(Fig. 6). The activation energy difference between the two tran-
sitions is only 0.53 kcal mol−1, which becomes a difference of
only 0.15 kcal mol−1 when corrected for vibrational contri-
Fig. 5 Transition state geometry for the acyl substituted structure (TS3)
showing the deprotonation of methanol (representing serine) by imidazole in
butions, within the margin of error for the method. Given that
the energy difference between the two pathways is negligible
concert with attack on the carbonyl by methoxide.
and the substitution product is observed experimentally, cycli-
zation should be expected to proceed along both pathways.
hydroxyl attack on the carbonyl, TS3 and TS3*, could not be
In order to explore the possibility that pyridine, the other
resolved until an additional imidazole was included to depro- aromatic nitrogenous heterocycle in the series of bases investi-
tonate the hydroxyl in a concerted mechanism (Fig. 5). Tran- gated experimentally, might also participate in a concerted
sition state calculations for methoxide attack on 1 in both the deprotonation and nucleophilic attack, but perhaps with a
This journal is © The Royal Society of Chemistry 2013
Org. Biomol. Chem., 2013, 11, 2979–2987 | 2983

View Article Online
Paper
Organic & Biomolecular Chemistry
structures. The key steps in the mechanism of catalysis of acyl
substitutions by imidazole must lie in the concerted deproto-
nation by imidazole and SN2 attack on the carbonyl by the
nucleophile and in the ability of imidazole to protonate inter-
mediates and leaving groups. Imidazole is amphoteric and
acts as an aqueous buffer in high concentrations. In aprotic
solvent, however, imidazole must act as a buffer to itself, being
the sole proton donor, as represented in the following equili-
brium expression (eqn (1)).
Fig. 7 Transition state geometry for the structure corresponding to TS3* using
pyridine in place of imidazole showing the deprotonation of methanol (repre-
senting serine) by pyridine in concert with attack on the carbonyl by methoxide.
ð1Þ
higher activation energy, we performed a transition state calcu-
lation with pyridine replacing imidazole as the base deproto-
nating methanol. The transition state geometry for TS3* was
used as the starting geometry for the calculation of the tran-
sition state with pyridine. The resulting transition state geo-
metry is nearly identical to that observed in TS3* (Fig. 7) and
the transition corresponding to the imaginary frequency
(646.2 cm−1) shows the same concerted deprotonation of
methanol as it attacks the carbonyl carbon. Interestingly, the
transition state energy is only 0.76 kcal mol−1 higher than that
of TS3*, suggesting cyclization should occur as readily using
pyridine as imidazole. As it does not, our calculations suggest
that pyridine, with a conjugate acid pKa of 5.3, should also act
as a base to deprotonate the hydroxyl. We expect that, in
aprotic solvent, pyridine is not protonated to a sufficient
degree to act as a proton donor to the intermediate alkoxides
or to the leaving groups in the final transitions, and thus
cannot effectively catalyze cyclization.
In a basic environment, no such stabilization could occur
since the pKa’s of the alkoxide intermediates are significantly
higher than those of the bases used in the study. The pKa of
the first intermediate following acyl substitution has a pre-
dicted aqueous pKa of 11.8, with the imidazole moiety having
a pKa of 6.3 (this structure is partially protonated in mildly
acidic solution). After the attack on the carbonyl by serine
(methanol), the predicted aqueous pKa of the intermediate alk-
oxide is 13.3 for the unsubstituted pathway, but 10.1 for the
intermediate alkoxide along the substituted pathway. In both
cases, the alkoxide is more basic than imidazolium cation and
should readily abstract a proton from imidazolium. The pro-
portion of pyridine’s conjugate acid, however, is exceedingly
small in basic solution and is unlikely to stabilize the inter-
mediates through protonation.
pKa values for imidazolium are available in DMSO21,22 (with
a dielectric constant of ε = 47.2, versus that of acetonitrile, ε =
36.6), which we may use to approximate the pKa in acetonitrile.
The pKa of protonated imidazolium cation is reported as 6.37,
while that of the anion is 18.6. The reported pKa values of
methanol and ethanol are approximately 29 in DMSO, indicat-
ing neutral imidazole as the primary proton source in our
system, which yields an associated imidazolium cation concen-
tration of ∼1 × 10−9 M at an equilibrium concentration of imid-
azole of 1 M. The higher than catalytic concentration of
imidazole required to catalyze this cyclization has also been
reported in the literature.20 Suchy et al. report16a that 2 equiva-
lents of imidazole are required for their imidazole-catalyzed
formylation reaction to proceed within a reasonable period of
time. Acting as both a base and an acid, imidazole may require
some minimum concentration to establish an equilibrium of
the free base and the conjugate acid in order to effectively cata-
lyze these acyl substitution reactions. In the aprotic media
used in this study, imidazole needs to act as a proton donor to
both the alkoxide intermediates and to the leaving groups, in
addition to stabilizing transition states through hydrogen
bonding or ion pairing. It is expected, therefore, that relatively
large quantities of imidazole should be required to produce
sufficient concentrations of the protonated specie to act as
proton donor to the leaving groups and transition states in
both cyclization pathways. In addition, an imidazole concen-
tration of 1 M corresponds to a concentration range observed
to yield a pronounced downfield 1H NMR shift for the imid-
azole N proton (from δ = 10.4 ppm at 0.17 M to δ = 12.75 ppm
in CDCl3), as a result of extensive intermolecular imidazole–
imidazole hydrogen bonding. It is therefore possible that the
concentration of imidazole required to catalyze the cyclization
of this peptide is dependent in some way upon the establish-
ment of a hydrogen-bonded imidazole network surrounding
the reactive sites of the peptide.
Discussion
It is certainly interesting that the nucleophilic attacks on
Imidazole has been reported in the literature as an acyl trans- the carbonyls in TS3 and TS3* involve a concerted deprotona-
fer catalyst, yet the precise mechanism of its participation has tion of the serine (methanol) hydroxyl, which has an aqueous
not been described in detail. From transition state analysis, we pKa of approximately 13, by imidazole, with an aqueous pKa of
observed very similar activation energies between the two reac- ∼7. Precedence for this departure from expected behavior in
tion pathways, indicating no significant energetic preference bulk solution is, nonetheless, found in nature. Many enzy-
for acyl substitution in the overall reactions of the model matic hydrolysis reactions involving imidazole, such as those
2984 | Org. Biomol. Chem., 2013, 11, 2979–2987
This journal is © The Royal Society of Chemistry 2013

View Article Online
Organic & Biomolecular Chemistry
Paper
catalyzed by the phospholipases and serine proteases proceed coordinate (IRC)30 calculations. Zero-point vibrational energies
via a bifunctional mechanism, in which imidazole, in the form (ZPVE) from the frequency calculations for all points along
of the amino acid histidine, acts as both a base and an acid. both pathways were scaled using a factor (0.97) described in
The apparent pKa of the His57 imidazole in complexes of the literature31 for the same functional and basis set and
chymotrypsin with peptidyl trifluoromethyl ketones17c has been plotted relative to the energies of species in the stoichiometric
shown to range from 10–12, very near the pKa of serine. Upon equation. Calculations of aqueous pKa were performed using
closer inspection, transition states TS3 and TS3* follow essen- the pKa prediction module in MarvinSketch32 using 2D input
tially the same mechanism utilized by the catalytic triads of structures.
the serine proteases, such as chymotrypsin and subtilisin, to
Although transition state analysis is typically performed on
cleave peptide chains. By the accepted mechanism, the cata- structures in the gas phase, transitions observed in the gas
lytic sites of these enzymes generate a serine (Ser195) side chain phase may not correspond to first-order saddle points for sol-
oxyanion in situ through a coupled proton transfer to the imid- vated structures. The solvation potential energy surface, or
azole of a neighboring histidine (His57) side chain.18 As the potential of mean force (PMF), lies below the gas phase poten-
anion is formed, it acts as a nucleophile, attacking the amide tial energy surface, but does not necessarily have the same topo-
carbonyl to form a hemiacetal alkoxide intermediate. Peptide logy:33 van der Waals complexes that form between two
cleavage is achieved by formation of a serine-peptide ester that species due to complementary partial charges or dipoles that
is hydrolyzed in subsequent steps, with re-protonation of represent local energetic minima in the gas phase are not local
serine by the histidine imidazole. This is analogous to the minima on the PMF surface, due to already extant interactions
mechanism of cyclization in our system, suggesting strongly with the solvation field. To model the solution chemistry as
that the same concerted deprotonation and nucleophilic closely as possible, transition states were determined using the
attack that occurs in TS3* is likely the same transition that polarized continuum model with integral equation formalism
occurs in these catalytic triads. The mechanism for the initial (IEF-PCM)34 method using radii and terms from the SMD28
step associated with peptide cleavage is generally depicted in solvation model. This model, which uses atomic surface
textbooks18 showing the electron flow that implies a concerted tension terms to model cavitation, dispersion, and solvent
process, our results show, in chemical detail, how this mech- rearrangement effects, is applicable to solvents of any dielec-
anism proceeds in concert.
tric since solvent properties were modeled in the training
process. Since zero-point energies were used in our energetic
analysis of the reaction pathways, some consideration was
given the effect of implicit solvation on calculated vibrational
frequencies (and hence, vibrational energies), since trans-
lations, rotations and vibrations tend to coordinate as libra-
tions in solution. Ribeiro et al. report35 that vibrational
contributions to free energy are generally insensitive to
whether they are calculated in the gas phase or using implicit
solvation, with the differences in the vibrational contributions
to the free energy between the two phases amounting to a
mean unsigned error of approximately 0.2 kcal mol−1 for a
diverse set of neutral and ionic molecules. Since explicit imid-
azoles were also included in transition state and intermediate
Methods
Experimental methods
Macrolactonization of the linear peptide thioester, Ac-SAFYG-
SCH2ϕ, was carried out at a concentration of 1 mM in an-
hydrous acetonitrile with varying amounts of imidazole added
to promote cyclization at room temperature. Aliquots of reaction
mixture were evaluated by LC-MS every 24 h from initiation to
determine the extent of the cyclization.
Computational methods
Fragment structures representing the larger peptides along complexes in conjunction with implicit solvation, the method
both reaction pathways were geometry optimized using AM123 is similar to the cluster-continuum36 method described by
in MOE24 2009.10. The putative transition states were con- Pliego et al.
structed as described below and minimized to first-order
In our DFT study we used the M06-2X hybrid meta-GGA
saddle points with Gaussian25 09 using density functional (generalized gradient approximation37) functional,27 which is
theory26 (DFT). The M06-2X hybrid functional27 with the a meta exchange-correlated functional with 54% HF exchange
6-31+G(d,p) basis set and the SMD28 polarizable continuum sol- that is reported to give good results for activation energies,
vation model using acetonitrile solvent parameters were also with low average mean unsigned errors for charge transfer
used. The Synchronous Transit-Guided Quasi-Newton (STQN) complexes, proton and heavy atom transfers, and nucleophilic
method29 employed in the QST3 procedure in Gaussian was substitutions. It is described as the most accurate method for
used to locate the quadratic region around the transition state, thermochemistry and kinetics across a number of evaluated
then to optimize to the saddle points using an eigenvector-fol- databases.27 Transition state input structures were set up to
lowing algorithm. Frequency calculations were used to follow a Bürgi–Dunitz trajectory38 for nucleophilic attack on
confirm transition states as first-order saddle points on the the carbonyl at angles of approximately 105°. Liotta et al.39
surface representing the potential of mean force by frequency report that, for nucleophilic attacks on carbonyls in general,
calculations, followed by inspection of imaginary frequencies. the softer the nucleophile, the greater is the deviation of the
Further confirmation was sought through intrinsic reaction approach angle from 90° whereas the harder the nucleophile,
This journal is © The Royal Society of Chemistry 2013
Org. Biomol. Chem., 2013, 11, 2979–2987 | 2985

View Article Online
Paper
Organic & Biomolecular Chemistry
the closer the approach angle comes to 90°.39 The resulting
transition state geometries, proceeding along their IRC trajec-
tories, were in good agreement with the general description of
the transition geometry, wherein, as the nucleophile moves
closer to the carbonyl carbon, the carbon moves further out of
the RR′CvO plane.
3 C. J. Armishaw, Toxins, 2010, 2, 1471–1499.
4 J. Austin, R. Kimura, Y.-H. Woo and J. Camarero, Amino
Acids, 2010, 38, 1313–1322.
5 (a) M. Doss, P. Ruchala, T. Tecle, D. Gantz, A. Verma,
A. Hartshorn, E. C. Crouch, H. Luong, E. D. Micewicz,
R. I. Lehrer and K. L. Hartshorn, J. Immunol., 2012, 188,
2759–2768; (b) V. Dewan, T. Liu, K. M. Chen, Z. Q. Qian,
Y. Xiao, L. Kleiman, K. V. Mahasenan, C. L. Li, H. Matsuo,
D. H. Pei and K. Musier-Forsyth, ACS Chem. Biol., 2012, 7,
761–769.
Conclusion
In conclusion, the side-chain to C-terminal cyclization of the
peptide Ac-SAFYG-SCH2ϕ was achieved experimentally through
the use of imidazole as a cyclization catalyst. However, good
yields of cyclic product were not obtained until concentrations
of imidazole above what is considered catalytic quantities were
added. Not until imidazole concentration approached 1 M did
the remaining unreacted peptide disappear from the isolated
product mixture to give a 90% relative yield of the cyclic depsi-
peptide. Using transition state analysis of two proposed reac-
tion pathways, we observed that, in what are the rate-limiting
transition along both pathways, imidazole catalyzes cyclization
through a concerted mechanism wherein imidazole removes a
proton from an alcohol, forming an alkoxide, which, in turn,
acts as a nucleophile toward an ester carbonyl. The concerted
deprotonation by imidazole and SN2 nucleophilic attack on the
carbonyl observed in the calculated cyclization transitions of
both pathways thus represents the mechanism by which imid-
azole is able to deprotonate hydrolizable species with much
higher pKa’s. Furthermore, we anticipate this reaction mechan-
ism is generalizable to other types of nucleophilic acyl
additions and substitutions involving imidazole, including
enzymatic systems. The chemical participants in the described
transition form a system similar to that utilized by the catalytic
triads of serine proteases to cleave peptide bonds, which likely
follow the same mechanism, with histidine imidazoles depro-
tonating serine hydroxyls in concert with nucleophilic attack
on bound peptide carbonyls, the first step of peptide cleavage
subsequent to binding.
6 A. Machado, M. A. Fazio, A. Miranda, S. Daffre and
M. T. Machini, J. Pept. Sci., 2012, 18, 588–598.
7 R. Dahiya and H. Gautam, Marine Drugs, 2011, 9, 71–81.
8 L. Gentilucci, R. D. Marco and L. Cerisoli, Curr. Pharm.
Des., 2010, 16, 3185–3203.
9 D. P. Fairlie, G. Abbenante and D. R. March, Curr. Med.
Chem., 1995, 2, 654–686.
10 M. Pakkala, J. Weisell, C. Hekim, J. Vepsäläinen, E. Wallen,
U.-H. Stenman, H. Koistinen and A. Närvänen, Amino Acids,
2010, 39, 233–242.
11 (a) G. Weltrowska, I. Berezowska, C. Lemieux, N. N. Chung,
B. C. Wilkes and P. W. Schiller, Chem. Biol. Drug Des., 2010,
75, 182–188; (b) L. C. Purington, I. D. Pogozheva,
J. R. Traynor and H. I. Mosberg, J. Med. Chem., 2009, 52,
7724–7731.
12 (a) J. S. Davies, J. Pept. Sci., 2003, 9, 471–501; (b) J. Eichler
and R. A. Houghten, Protein Pept. Lett., 1997, 4, 157–164.
13 (a) R. A. Houghten, C. Pinilla, J. R. Appel, S. E. Blondelle,
C. T. Dooley, J. Eichler, A. Nefzi and J. M. Ostresh, J. Med.
Chem., 1999, 42, 3743–3778; (b) C. Dooley, N. Chung,
B. Wilkes, P. Schiller, J. Bidlack, G. Pasternak and
R.
Houghten,
Science,
1994,
266,
2019–2022;
(c) R. Houghten, C. Dooley and J. Appel, AAPS J., 2006, 8,
E371–E382.
14 (a) M. Lebl, J. Comb. Chem., 1999, 1, 3–24; (b) S. E. Cwirla,
E. A. Peters, R. W. Barrett and W. J. Dower, Proc. Natl. Acad.
Sci., 1990, 87, 6378–6382; (c) J. Scott and G. Smith, Science,
1990, 249, 386–390; (d) J. Devlin, L. Panganiban and
P. Devlin, Science, 1990, 249, 404–406; (e) M. A. Gallop,
R. W. Barrett, W. J. Dower, S. P. A. Fodor and E. M. Gordon,
J. Med. Chem., 1994, 37, 1233–1251; (f) L. Gentilucci,
G. Cardillo, A. Tolomelli, R. DeMarco, A. Garelli,
S. Spampinato, A. Sparta and E. Juaristi, ChemMedChem,
2009, 4, 517–523.
Acknowledgements
This work was supported by the State of Florida, Executive
Office of the Governor’s Department of Economic Opportunity
and funded, in part, by NIDA DA031370 to RAH. KM acknow-
ledge PAPIIT grant IA200513-2. The authors also thank Che-
mAxon for providing MarvinSketch and Dr. Ralph Puchta for
his helpful suggestions with transition states calculations.
15 (a) Y. Li, M. Giulionatti and R. A. Houghten, Org. Lett.,
2010, 12, 2250–2253; (b) G. M. Fang, H. K. Cui, J. S. Zheng
and L. Lui, ChemBioChem, 2010, 11, 1061–1065.
16 (a) M. Suchý, A. A. H. Elmehriki and R. H. E. Hudson, Org.
Lett., 2011, 12, 3952–3955; (b) L. Mandell, J. W. Moncrief
and J. H. Goldstein, Tetrahedron, 1963, 19, 2025–2030.
17 (a) W. W. Bachovchin, W. Y. L. Wong, S. Farr-Jones,
A. B. Shenvi and C. A. Kettner, Biochemistry, 1988, 27,
7689–7697; (b) T. C. Liang and R. H. Abeles, Biochemistry,
1987, 26, 7603–7608; (c) J. Lin, C. S. Cassidy and P. A. Frey,
Biochemistry, 1998, 37, 11940–11948.
References
1 H. I. Baumann, S. Keller, F. E. Wolter, G. J. Nicholson,
G. Jung, R. D. Süssmuth and F. Jüttner, J. Nat. Prod., 2007,
70, 1611–1615.
2 L. J. Cruz, C. Cuevas, L. M. Cañedo, E. Giralt and
F. Albericio, J. Org. Chem., 2005, 71, 3339–3344.
18 L. Hedstrom, Chem. Rev., 2002, 102, 4501–4523.
2986 | Org. Biomol. Chem., 2013, 11, 2979–2987
This journal is © The Royal Society of Chemistry 2013

View Article Online
Organic & Biomolecular Chemistry
Paper
19 T. Kar and S. Scheiner, Int. J. Quantum Chem., 2006, 106, 26 W. Kohn and L. J. Sham, Phys. Rev., 1965, 140, A1133–
843–851. A1138.
20 M. L. Bender and B. W. Turnquest, J. Am. Chem. Soc., 1957, 27 Y. Zhao and D. G. Truhlar, Theor. Chem. Acc., 2008, 120,
79, 1652–1655.
215–241.
21 F. G. Bordwell, Acc. Chem. Res., 1988, 21, 456–463.
22 M. R. Crampton and I. A. Robotham, J. Chem. Res., 1997,
22–23.
23 M. J. S. Dewar, E. G. Zoebisch and E. F. Healy, J. Am. Chem.
Soc., 1985, 107, 3902–3909.
28 A. V. Marenich, C. J. Cramer and D. G. Truhlar, J. Phys.
Chem. B, 2009, 118, 6378–6396.
29 (a) C. Peng and H. B. Schlegel, Isr. J. Chem., 1993, 33, 449;
(b) C. Peng, P. Y. Ayala, H. B. Schlegel and M. J. Frisch, J.
Comput. Chem., 1996, 17, 49.
24 MOE, Chemical Computing Group, Montreal, 2009.
30 K. Fukui, Acc. Chem. Res., 1981, 14, 363–368.
25 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, 31 I. M. Alecu, J. Zheng, Y. Zhao and D. G. Truhlar, J. Chem.
M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, Theor. Comput., 2010, 6, 2872–2887.
B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, 32 MarvinSketch 5.6.0.0, Chemaxon Ltd (http://www.che-
X. Li, H. P. Hratchian, A. F. Izmaylov, G. Z. J. Bloino, maxon.com), 2011.
J. L. Sonnenberg, M. E. M. Hada, K. Toyota, R. Fukuda, 33 C. J. Cramer, in Implicit Models for Condensed Phases, John
J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, Wiley & Sons, Ltd, Chichester, 2005.
H. Nakai, T. Vreven, J. J. A. Montgomery, J. E. Peralta, 34 M. Cossi, G. Scalmani, N. Rega and V. Barone, J. Chem.
F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, Phys., 2002, 117, 43–54.
K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, 35 R. F. Ribeiro, A. V. Marenich, C. J. Cramer and
K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar,
J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene,
D. G. Truhlar, J. Phys. Chem. B, 2011, 115, 14556–
14562.
J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, 36 J. R. Pliego and J. M. Riveros, J. Phys. Chem. A, 2002, 106,
R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, 7434–7439.
R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, 37 J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett.,
K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, 1996, 77, 3865–3868.
J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, 38 B. H. Bürgi, J. D. Dunitz and E. Shefter, J. Am. Chem. Soc.,
J. B. Foresman, J. V. Ortiz, J. Cioslowski and D. J. Fox, 1973, 95, 5065–5067.
GAUSSIAN 09 (Revision A.01), Gaussian, Inc., Wallingford, 39 C. L. Liotta, E. M. Burgess and W. H. Eberhardt, J. Am.
CT, 2009.
Chem. Soc., 1984, 106, 4849–4852.
This journal is © The Royal Society of Chemistry 2013
Org. Biomol. Chem., 2013, 11, 2979–2987 | 2987