Macrocyclization of Organo-Peptide Hybrids through a Dual Bio-orthogonal Ligation: Insights from Structure-Reactivity Studies

Article abstract of DOI:10.1002/cbic.201200579

Macrocycles constitute an attractive structural class of molecules for targeting biomolecular interfaces with high affinity and specificity. Here, we report systematic studies aimed at exploring the scope and mechanism of a novel chemo-biosynthetic strategy for generating macrocyclic organo-peptide hybrids (MOrPHs) through a dual oxime-/intein-mediated ligation reaction between a recombinant precursor protein and bifunctional, oxyamino/1,3-amino-thiol compounds. An efficient synthetic route was developed to access structurally different synthetic precursors incorporating a 2-amino- mercaptomethyl-aryl (AMA) moiety previously found to be important for macrocyclization. With these compounds, the impact of the synthetic precursor scaffold and of designed mutations within the genetically encoded precursor peptide sequence on macrocyclization efficiency was investigated. Importantly, the desired MOrPHs were obtained as the only product from all the different synthetic precursors probed in this study and across peptide sequences comprising four to 15 amino acids. Systematic mutagenesis of the "i-1" site at the junction between the target peptide sequence and the intein moiety revealed that the majority of the 20 amino acids are compatible with MOrPH formation; this enables the identification of the most and the least favorable residues for this critical position. Furthermore, interesting trends with respect to the positional effect of conformationally constrained (Pro) and flexible (Gly) residues on the reactivity of randomized hexamer peptide sequences were observed. Finally, mechanistic investigations enabled the relative contributions of the two distinct pathways (side-chain→C-end ligation versus C-end→side-chain ligation) to the macrocyclization process to be dissected. Altogether, these studies demonstrate the versatility and robustness of the methodology to enable the synthesis and diversification of a new class of organo-peptide macrocycles and provide valuable structure-reactivity insights to inform the construction of macrocycle libraries through this chemo-biosynthetic strategy.

Full text of DOI:10.1002/cbic.201200579

CHEMBIOCHEM
FULL PAPERS
DOI: 10.1002/cbic.201200579
Macrocyclization of Organo-Peptide Hybrids through
a Dual Bio-orthogonal Ligation: Insights from Structure–
Reactivity Studies
John R. Frost, Francesca Vitali, Nicholas T. Jacob, Micah D. Brown, and Rudi Fasan*[a]
Macrocycles constitute an attractive structural class of mole-
cules for targeting biomolecular interfaces with high affinity
and specificity. Here, we report systematic studies aimed at
exploring the scope and mechanism of a novel chemo-biosyn-
thetic strategy for generating macrocyclic organo-peptide hy-
brids (MOrPHs) through a dual oxime-/intein-mediated ligation
reaction between a recombinant precursor protein and bifunc-
tional, oxyamino/1,3-amino-thiol compounds. An efficient syn-
thetic route was developed to access structurally different syn-
thetic precursors incorporating a 2-amino- mercaptomethyl-
aryl (AMA) moiety previously found to be important for macro-
cyclization. With these compounds, the impact of the synthetic
precursor scaffold and of designed mutations within the ge-
netically encoded precursor peptide sequence on macrocycli-
zation efficiency was investigated. Importantly, the desired
MOrPHs were obtained as the only product from all the differ-
ent synthetic precursors probed in this study and across pep-
tide sequences comprising four to 15 amino acids. Systematic
mutagenesis of the “iꢀ1” site at the junction between the
target peptide sequence and the intein moiety revealed that
the majority of the 20 amino acids are compatible with MOrPH
formation; this enables the identification of the most and the
least favorable residues for this critical position. Furthermore,
interesting trends with respect to the positional effect of con-
formationally constrained (Pro) and flexible (Gly) residues on
the reactivity of randomized hexamer peptide sequences were
observed. Finally, mechanistic investigations enabled the rela-
tive contributions of the two distinct pathways (side-chain!C-
end ligation versus C-end!side-chain ligation) to the macro-
cyclization process to be dissected. Altogether, these studies
demonstrate the versatility and robustness of the methodolo-
gy to enable the synthesis and diversification of a new class of
organo-peptide macrocycles and provide valuable structure–
reactivity insights to inform the construction of macrocycle
libraries through this chemo-biosynthetic strategy.
Introduction
Peptide-containing macrocycles represent an attractive class of
compounds for the development of chemical probes and ther-
apeutics.[1] Growing interest in this structural class has been
fueled by the identification of several bioactive cyclopeptides
occurring in nature[2] as well as the recognition of the potential
benefits conferred by backbone cyclization in terms of proteo-
lytic stability,[3] cell permeability,[4] and protein-binding proper-
ties.[5] Methods for generating libraries of macrocyclic peptides
can thus provide valuable sources of relevant ligand diversity,
in particular toward the identification of agents that could
target extended biomolecular interfaces.[1]
exhibit the desired target-binding properties. A widespread ap-
proach used to restrict the conformational flexibility of riboso-
mal peptides is the formation of disulfide bridges between cys-
teine residues flanking randomized target sequences.[6] Alterna-
tively, cysteine- or amine-reactive agents have been applied to
introduce covalent crosslinks within displayed polypeptide
sequences to generate macrocyclic or bicyclic peptide struc-
tures.[7] Other notable approaches in this area have involved
the generation of cyclic peptides by means of split inteins,[8]
through chemoenzymatic[9] or in vitro translation[10] methods,
or by manipulating the enzymes involved in the biosynthesis
of naturally occurring cyclopeptides.[11] Within these strategies,
however, the pool of building blocks available for library con-
struction is limited to the 20 natural amino acids, and thus re-
mains considerably smaller than that potentially accessible
through chemical synthesis.[12] This limitation has been in part
addressed by adapting these protocols to enable the introduc-
tion of unnatural amino acids at specific positions within the
peptide sequence.[7c,10b,11b,13] Despite these advances, none of
current methodologies allows for the straightforward incorpo-
ration of synthetic, nonproteinogenic elements of arbitrary
design into genetically encoded peptide frameworks.
In this regard, generating conformationally constrained pep-
tides from genetically encoded precursors offer several poten-
tial advantages. There is the opportunity to access vast chemi-
cal diversity (106–1010 molecules) rapidly through combinatorial
mutagenesis combined with the possibility of coupling library
creation with powerful selection methods (e.g., phage-, yeast-,
mRNA-, or ribosome-display) to isolate the member(s) that
[a] J. R. Frost, Dr. F. Vitali, N. T. Jacob, M. D. Brown, Prof. R. Fasan
Department of Chemistry, University of Rochester
Hutchison Hall, Rochester, NY 14620 (USA)
E-mail: fasan@chem.rochester.edu
To provide a way to combine the advantages of genetic en-
coding with the versatility of chemical synthesis, our group has
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201200579.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 147

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
been exploring methodologies for constructing organo-pep-
tide macrocycles (macrocyclic organo-peptide hybrids or
MOrPHs) through a dual ligation between bifunctional synthet-
ic molecules (referred to as “synthetic precursors” or SPs) and
recombinant intein-fused proteins (referred to as “biosynthetic
precursors” or BPs) containing an unnatural amino acid.[14] In
a first implementation of this concept, MOrPHs were obtained
through a CuI-catalyzed alkyne/azide cycloaddition-/hydrazide-
mediated ligation.[14a] More recently, we introduced a conven-
ient, catalyst-free approach to MOrPH synthesis that exploits
a tandem ligation between a oxyamino/amino-thiol compound
and an intein-fused protein precursor incorporating a side-
chain keto group from a genetically encoded p-acetylphenyla-
lanine residue.[14b] Interestingly, in this previous study, only SP3
(1, Scheme 1) was found to be a viable precursor for inducing
the desired macrocyclization reaction, whereas bifunctional
molecules containing a similar alkyloxyamine functionality but
a different amino-thiol group (i.e., either cysteine- or 2-amino-
thiophenol-based moiety) failed to do so. These findings point-
ed at the importance of the 2-amino-mercaptomethyl-aryl
(AMA) moiety in SP3 (dotted box in Scheme 1) as a critical
structural element for directing macrocycle formation within
this approach.
Scheme 1. Chemical structures of the oxyamino-/2-amino-mercaptomethyl-
aryl (AMA)-based synthetic precursors described in this study. The AMA
moiety is highlighted by the dotted box.
To investigate this aspect, a panel of four different AMA-
based synthetic precursors were designed and synthesized
(compounds 2–5, Scheme 1). SP4 (2) provides the most com-
pact scaffold, with the oxyamine functionality being directly
connected to the AMA moiety through a single methylene
unit. SP5 (3), SP6 (4), and SP7 (5) present more extended struc-
tures incorporating aryl (3, 5) or triazolyl (4) linker groups.
Compared to SP3, these second-generation synthetic precur-
sors are expected to possess a lower degree of conformational
freedom as given by the presence of two (SP4), four (SP6, SP7),
and five (SP5) rotatable bonds separating the AMA moiety
from the oxyamine oxygen atom as opposed to six in 1. These
compounds were also meant to present a range of different
spacing distances between the two ligation points within the
molecule (i.e., the oxyamino group and the amino group in
the AMA moiety, Scheme 2) as determined based on calculated
energy-minimized conformations (6.5 ꢁ (SP4), 8.5 ꢁ (SP5), 11 ꢁ
(SP6), 12.5 ꢁ (SP7)). These features would allow us to test the
possibility of accessing macrocycles of variable ring size in ad-
dition to incorporating different synthetic moieties.
Building upon these preliminary investigations, we report
here a series of systematic studies aimed at expanding and ex-
ploring in more depth the scope of this methodology towards
the synthesis and diversification of this new class of hybrid
organo-peptide macrocycles. A first point of interest concerned
the amenability of this methodology to permit the use of
structurally diverse AMA-containing synthetic precursor scaf-
folds as a way to modulate the architecture and topology of
the final MOrPH products. We then investigated the effect of
amino acid substitutions within the precursor peptide se-
quence on the accessibility and efficiency of macrocyclization,
thereby gaining structure–reactivity insights also at the level of
the genetically encoded portion of the MOrPHs. Finally, mecha-
nistic studies were performed to elucidate the relative contri-
bution of the two distinct pathways envisioned to mediate the
formation of the hybrid macrocycles through this chemo-bio-
synthetic strategy (Scheme 2).
To synthesize 2–5, a different synthetic scheme (Scheme 3)
was implemented from that used to prepare SP3,[14b] as the
latter was found to lack both versatility and scalability. In par-
ticular, our goal was to develop an efficient, versatile, and scal-
able route that could give rapid access to a variety of structur-
ally different AMA-containing synthetic precursors. The new
strategy embodies these features and hinges on the prepara-
tion of the key intermediate 7, which can be afforded in high
yield (72%) from commercially available methyl 3-amino-4-
methylbenzoate (6) through a three-step sequence involving
N-Boc protection followed by photocatalyzed benzylic bromi-
nation with NBS and substitution of the benzylic bromide with
tritylmercaptan. Compound 7 can be used to conveniently link
the N-Boc-S-trityl-protected AMA moiety to a variety of oxya-
mine-containing organic scaffolds. Accordingly, basic hydrolysis
of the benzoate group in this molecule followed by amide
coupling with the aniline derivative 11 and final TFA deprotec-
tion yielded SP7 (5). To generate SP4 (2), 7 was converted into
the mesylate derivative 8, which was then substituted with N-
Boc-hydroxylamine followed by TFA deprotection. To obtain
SP6 (4), the mesylate group in 8 was substituted with sodium
Results and Discussion
Design and synthesis of structurally diverse AMA-based
synthetic precursors
In SP3 (Scheme 1), the oxyamino functionality was connected
to the 2-amino-mercaptomethyl-aryl (AMA) moiety by a flexible
four-methylene linker. This design was initially useful for com-
paring the relative efficiency of different amino-thiol moieties
for macrocyclization.[14b] The ability to use more compact and/
or conformationally rigid scaffolds to direct MOrPH formation
would be desirable though, as these compounds could induce
a greater degree of rigidification in the backbone of the result-
ing macrocycle. However, these features could have a potential-
ly negative effect on the efficiency of the macrocyclization.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 148

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Scheme 2. Overview of the chemo-biosynthetic method to generate MOrPHs through an oxime-/AMA-mediated dual ligation. The precursor protein (BP) is
composed of an N-terminal sequence (R’=chitin binding domain), a genetically encoded p-acetyl-phenylalanine (pAcF), a target peptide sequence of variable
length (n=2–13), and the GyrA mini-intein from M. xenopi (N198A variant). In the oxyamine/amino-thiol synthetic precursor (SP), X corresponds to a variable
linker structure (see structures in Scheme 1). The two possible reaction pathways discussed in the text are outlined. In species “b”, the position of the iꢀ1 res-
idue is indicated.
Scheme 3. Synthesis of synthetic precursors 2–5. a) LiAlH4, THF, 95%; b) Ms-Cl, DIPEA, CH2Cl2, 88%; c) NaN3, DMF, 100%; d) LiAlH4, THF, 95%; e) LiOH, THF/H2O,
100%; f) 11, DCC, DMAP, CH2Cl2, 26%; g) TIPS, TFA/CH2Cl2, 100%; h) HONHBoc, DBU, DMF, 89%; i) TIPS, TFA/CH2Cl2, 100%; j) 12, CuSO4, NaAsc, CH2Cl2/H2O,
72%; k) TIPS, TFA/CH2Cl2, 100%; l) 13, HBTU, DIPEA, CH2Cl2, 55%; m) TIPS, TFA/CH2Cl2, 100%.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 149

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
azide, and the resulting product (9) was subjected to CuI-cata-
lyzed alkyne–azide cycloaddition with N-Boc-propargyloxya-
mine (12) followed by final deprotection with TFA. Finally, the
azido-containing intermediate 9 was reduced to the benzyla-
mine counterpart 10, which was then coupled to the benzoic
acid derivative 13 followed by TFA treatment to yield SP5 (3).
To test the ability of this scheme to yield gram quantities of
a desired synthetic precursor, the synthesis of 2 was repeated
on a larger scale; this resulted in the isolation of 1.3 g of this
compound in 55% overall yield. Altogether, these studies dem-
onstrated both the versatility and the scalability of the newly
developed synthetic scheme for generating oxyamine-/AMA-
containing synthetic precursors.
buffer (pH 7.5) at room temperature and under reducing con-
ditions (20 mm tris-(2-carboxyethyl)phosphine hydrochloride
(TCEP)). These mixtures were then analyzed after 5 and 12 h by
SDS-PAGE followed by gel densitometry to quantify the extent
of protein splicing; MALDI-TOF mass spectrometry was used to
monitor product formation. In the presence of SP4, the protein
precursors showed between 61 and 78% splicing after 5 h
across all the target sequence lengths (Figure 1, Figure S1 in
Macrocyclization reactions with target peptide sequences of
variable length
Next, we tested the ability of compounds 2–5 to induce
MOrPH formation according to Scheme 2. We used a set of
seven model biosynthetic precursors comprising, from the N
to the C terminus, an affinity tag (chitin binding domain, CBD),
the unnatural amino acid p-acetylphenylalanine (pAcF), and
a variable target peptide sequence (vide infra) fused to Mxe
GyrA intein (species “a”, Scheme 2). These constructs were pro-
duced in Escherichia coli by coexpressing an orthogonal, pAcF-
specific tRNACUA/aminoacyl-tRNA synthetase pair[15] to intro-
duce pAcF into the protein in response to an amber stop
codon. Altogether, the seven constructs comprised a set of
target sequences consisting of four, five, six, eight, ten, 12, and
15 amino acids (Table 1); these enabled us to test the accessi-
Figure 1. Extent of protein splicing for biosynthetic precursors CBD4(pAcF)
&
&
to CBD15(pAcF) after 5 hours of incubation with compound 2 ( ), 3 ( ), 4
&
&
( ), or 5 ( ), as determined by densitometric analysis of SDS-PAGE protein
gel (see also Figure S1).
the Supporting Information). After 12 h, nearly quantitative
splicing (88–96%) was observed for the protein constructs
(Figure S1). Similarly, all the other synthetic precursors were
found to induce splicing of the biosynthetic precursors at com-
parable or slightly higher degree than SP4 after 12 hours of
incubation. However, significantly more splicing (about 80 to
95%) was observed with these compounds at the shorter time
point (5 h; Figure 1). Also in these cases, the extent of precur-
sor protein splicing did not appear to be affected by the
length of the target sequence.
Table 1. Length and composition of the target peptide sequences in the
biosynthetic precursors used in this study.
Protein
Length
Amino acid sequence
CBD4(pAcF)
CBD5(pAcF)
CBD6(pAcF)
CBD8(pAcF)
CBD10(pAcF)
CBD12(pAcF)
CBD15(pAcF)
tetramer
pentamer
hexamer
octamer
decamer
TGST
TGSGT
TGSYGT
TGSAEYGT
TGSKLAEYGT
TGSWGKLAEYGT
TGSHNRWGKLAEYGT
In addition to the desired macrocyclization, potential side re-
actions that would result in splicing of the protein precursors
are the spontaneous hydrolysis of the intein-mediated thioest-
er to give the linear product H2N-CBD-pAcF-(target sequence)-
COOH or incomplete cyclization after SP-induced splicing due
to failure of the oxime-mediated ring closure resulting in the
linear intermediate “d” in Scheme 2. All these by-products can
be distinguished from the desired macrocycles by their differ-
ent masses. Remarkably, for all the reactions with SP4, MALDI-
TOF MS analysis revealed the presence of a single peak consis-
tent with the m/z ([M+H]+ ion) of the desired MOrPH after
both 5 and 12 h (Figure 2). Similarly, the other synthetic precur-
sors SP5, SP6, and SP7 were able to undergo efficient macro-
cyclizations across all the peptide target sequence lengths
tested, as indicated by the observation of a single product
with a mass consistent with that of the expected macrocycle
(Figures 3 and S2–S4).
dodecamer
pentadecamer
bility of macrocyclization with the different synthetic precur-
sors across peptide moieties of varying length. With the excep-
tion of the last amino acid residue prior to the intein (the “iꢀ1
site”, Scheme 2), the composition of the sequences was ran-
domly chosen and included a representative subset of the 20
natural amino acids. The N-terminal chitin binding domain, on
the other hand, served the double purpose of reproducing the
settings of a typical display system (in which the MOrPHs
would be generated as tethered to the C terminus of a carrier
protein) and that of providing a convenient handle for affinity
purification.
Test reactions were carried by mixing SP4–SP7 (15 mm) with
each of the biosynthetic precursors (100 mm) in phosphate
To confirm the occurrence of the desired S!N acyl transfer
at the level of the AMA moiety in these products, the samples
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 150

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
were briefly exposed to iodoacetamide (20 mm,
10 min). In each case, this led to the complete disap-
pearance of the MOrPH product peak and the ap-
pearance of a new peak with m/z of +57. This signal
is consistent with the expected mass for the S-alky-
lated MOrPH adduct, thereby indicating that the
thiol group in the synthetic moiety of the macrocycle
is available for reaction with the electrophilic reagent
and, thus, that the desired S!N acyl transfer had oc-
curred.
Overall, the results from the above experiments
showed that the methodology is remarkably tolerant
to variations in both the synthetic precursor struc-
ture and the length of the genetically encoded pep-
tide sequence, thus allowing the construction of
MOrPHs of variable backbone and ring size (Figures 2
and 3).
Synthesis and stability of MOrPHs in cell lysates
In view of future applications of this method for gen-
erating and screening MOrPH libraries against bio-
logical targets, it was important to establish the bio-
orthogonality and stability of the macrocycles in
a complex biological milieu. Owing to the presence
of many metabolites, other biomolecules, and pro-
teases that could prevent the reaction or degrade
the product, a most stringent test is to perform the
macrocyclization in cell lysate. Accordingly, synthetic
precursor 2 was directly added to a clarified lysate
solution of E. coli cells expressing the pentamer bio-
synthetic precursor CBD5(pAcF). After 2-hour, 5-hour,
and overnight incubation, the reaction products
were isolated from the mixture by affinity chroma-
tography on chitin beads. This procedure would
allow isolation of the desired CBD-tethered MOrPH
as well as of by-products resulting from hydrolysis of
the intein, incomplete cyclization (C-terminal ligation
but no oxime formation), and proteolytic degrada-
tion of the peptide moiety embedded in the MOrPH.
Importantly, MALDI-TOF MS analysis of the chitin
bead eluates reproduced the results obtained in the
case of the purified biosynthetic precursors, with the
desired CBD-tethered MOrPH occurring as the only
observable species (Figure S5). These data demon-
strated the excellent bio-orthogonality of the ligation
reactions involved in MOrPH formation as well as the
suitability of using this approach to produce MOrPHs
in biological media.
Figure 2. MALDI-TOF spectra and chemical structures of MOrPHs
generated upon treatment of SP4 (2) with biosynthetic precur-
sors CBD4(pAcF) to CBD15(pAcF). Calculated (c) and observed
(o) m/z values corresponding to the [M+H]+ adduct of the mac-
rocyclic product (m) are indicated. R’=chitin binding domain.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 151

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Influence of the “iꢀ1” residue on macrocyclization
Next, we sought structure–reactivity insights into
the effect of substitutions at the level of the geneti-
cally encoded moiety of the macrocycles. The amino
acid residue preceding the N-terminal cysteine of an
intein (the “iꢀ1” position) has been shown to influ-
ence the splicing behavior of these proteins.[16] Ac-
cordingly, this position was expected to play a major
role in affecting the reactivity of the biosynthetic
precursor toward MOrPH formation. In general, the
ability to mutate this position with as many residues
as possible is desirable in order to increase diversity
in the macrocycle libraries generated through this
method. At the same time, certain substitutions
could decrease the reactivity of the thioester linkage
(due to steric hindrance or by disfavoring the intein-
catalyzed N!S acyl transfer reaction) or favor pre-
mature splicing of the biosynthetic precursor during
expression, both of which are undesirable as they
would have a negative impact on the accessibility of
the precursor polypeptide for the macrocyclization
reaction. Knowledge of the most and least favorable
mutations at this site would thus be useful during
library design in order to maximize the content of
the desired products.
In the panel of biosynthetic precursors described
earlier (Table 1), a threonine was kept constant at
the iꢀ1 site as this residue was reported to cause
minimal premature splicing of GyrA fusion pro-
teins.[16d] In order to investigate the effect of the iꢀ1
residue in a systematic manner, this position was
mutated to each of the other 19 natural amino acids
in the pentamer target sequence construct, CBD5-
(pAcF). After 12 hours of expression in E. coli at
278C, the corresponding biosynthetic precursors
were purified by Ni-affinity chromatography using
a polyhistidine tag fused to the C terminus of the
intein. This procedure allows the isolation of both
the full-length protein and spliced intein, thus ena-
bling the quantification of the extent of premature
splicing by means of SDS-PAGE. These analyses re-
vealed that as many as nine out of the 20 constructs
(iꢀ1=T, S, G, P, A, F, W, Y, and N) showed little or no
premature splicing (0–20%; Figure 4A). Seven con-
structs (iꢀ1=C, V, I, M, E, Q, and R) exhibited moder-
ate in vivo splicing (25–50%), and only four variants
(iꢀ1=D, H, K, or L) were isolated in just 30 to 5%
full-length form (Figure 4A). These results confirmed
the importance of the iꢀ1 residue in affecting the
reactivity and stability of the GyrA intein-catalyzed
thioester toward hydrolysis; in most cases, they can
be rationalized on the basis of the reactivity of the
iꢀ1 residue side-chain functionality. For example,
the high level of premature splicing for the con-
structs with D, H, and K at this site is likely caused
by the corresponding side-chain group favoring the
Figure 3. MALDI-TOF spectra and chemical structures of MOrPHs generated upon treat-
ment of biosynthetic precursor CBD5(pAcF) with SP4 (2), SP5 (3), SP6 (4), and SP7 (5).
Calculated (c) and observed (o) m/z values corresponding to the [M+H]+ adduct of the
macrocyclic product (m) are indicated. R’=chitin binding domain.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 152

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Thr (70–75% overall yield) represent the optimal choices at
this site, providing an ideal combination of low susceptibility
to hydrolysis and high reactivity toward SP-induced cyclization.
Following in terms of performance, there is the group consist-
ing of Trp, Asn, Arg, and Gln (50–60% overall yield) and that
comprising Met, Ser, Cys, and Glu (40% overall yield). Finally,
these analyses revealed the least favorable amino acid substi-
tutions for this position, due to their effect on dampening the
reactivity of the precursor protein toward SP-induced cycliza-
tion (Gly, Pro), promoting premature splicing (Asp, Lys, His), or
both (Leu, Val, Ile). Collectively, these studies indicated that 12
out of the 20 possible amino acids at the iꢀ1 site are compati-
ble with MOrPH formation, thus leading to the identification of
the subsets of most and least favorable residues for this posi-
tion.
In light of the results, it was also interesting to compare and
contrast the efficiency of AMA-mediated ligations in the con-
text of the various iꢀ1 variants here (given by the y values in
Figure 4A) with that of native chemical ligation (NCL) reactions
in the presence of peptide thioesters containing variable termi-
nal amino acids.[17] In both cases, aliphatic residues (L, V, I) and
proline have the most negative impact on the ligation rates. In
contrast, aromatic residues at the terminal site allow for fast
AMA-induced ligations but work only moderately well in the
case of NCL. However, a most striking difference between the
two systems concerns Gly and Thr. In NCL, thioesters with ter-
minal Gly and Thr exhibit, respectively, among the fastest and
slowest ligation rates,[17] whereas a reversion of this trend is
observed with the SP-mediated ligations carried out here.
Thus, other factors play a role in modulating the reactivity of
the thioester bond in the latter system beyond steric effects,
which appear to predominate in the context of NCL reactions.
Intein GyrA has been the focus of detailed structural and
mechanistic studies,[18] and, in order to rationalize these obser-
vations, we inspected the available crystal structure of the
GyrA(Cys1Ser) variant. Interestingly, the b-carbon of the iꢀ1
residue (Ala) was found to lie within 3–7 ꢁ of other residues in
the active site and adjacent protein surface (Figure 5).[18a] Thus,
interactions are likely to occur between these residues that
could modulate the reactivity of the neighboring thioester link-
age. These interactions could be the basis of the unexpectedly
higher performance of threonine compared to the other b-
branched amino acids (Ile, Val; Figure 4), the former being the
only one possessing a side chain capable of forming H-bonds
(e.g., with the proximal Ser179, Figure 5). Another interesting
observation concerns the optimal performance of the protein
variants containing Tyr and the structurally related Phe at iꢀ1,
both of which show little reactivity toward hydrolysis but high
reactivity toward AMA-mediated ligation. Intriguingly, in the
natural protein containing the GyrA intein used here (Mycobac-
terium xenopi DNA gyrase subunit A), tyrosine occurs at the
terminal position of the N-extein.[19] It is therefore tempting to
speculate that this residue might have been selected during
evolution because it can provide an optimal performance in
the context of protein splicing, a property also exhibited in the
context of the present application.
Figure 4. Reactivity of the 20 biosynthetic precursor iꢀ1 variants. A) Plot of
the amount of full-length protein after purification from E. coli versus yield
of SP4-induced macrocyclization after 5 hours of incubation. Errors are
ꢂ10%. B) Overall yield for the MOrPH-forming reactions as given by the %
of full-length protein ꢂ % of SP-induced macrocyclization yield from full-
length precursor protein.
hydrolysis of the neighboring thioester through either direct
nucleophilic attack or through a general base-catalyzed mech-
anism. This notwithstanding, these experiments showed that
the majority (80%) of the iꢀ1 substitutions resulted in excel-
lent to good amounts of the full-length precursor protein (50–
100%). At least for these variants, further increase in the full-
length content could be achieved under less stringent expres-
sion conditions (i.e., lower temperature and/or shorter cultur-
ing time).
To test the reactivity of these 16 variants toward macrocycli-
zation, each construct was incubated with 2 under standard re-
action conditions (15 mm SP, 20 mm TCEP, RT). Interestingly,
high levels of 2-induced splicing (75–100%) after 5 h were ob-
served for eight of these constructs (iꢀ1=T, A, F, Y, W, E, Q,
and R; Figure 4A, Figure S6). At the same time point, four con-
structs (iꢀ1=S, C, M, N) exhibited moderate splicing (40–
70%), whereas four (iꢀ1=G, V, I, P) show only little reactivity
(10–30% splicing; Figure 4A, Figure S6). Importantly, for all of
the variants, the desired MOrPH was obtained as the only
product, as illustrated by the MALDI-TOF MS spectra in Fig-
ure S7.
Since the macrocycle was the only observable product of
these reactions (Figure S7), the percentage of SP-induced splic-
ing (as determined by densitometric analysis of SDS-PAGE gels)
provides a measure of the yield of macrocyclization starting
from the full-length protein precursor (Figure 4A). The data
above could therefore be combined to calculate overall yields
(=% of full-length protein x% MOrPH yield) to define the most
and least optimal amino acid substitutions at position iꢀ1 for
maximizing the amount of MOrPH formed (Figure 4B). These
analyses revealed that Phe, Tyr (ꢁ90% overall yield) and Ala,
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 153

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Figure 6. Amount of SP4-induced protein splicing for the proline-scan and
glycine-scan biosynthetic precursor libraries and the reference X5T library.
The target sequence, in which X corresponds to a fully randomized position
(NNK codon), is specified.
Figure 5. View of the GyrA(C1S) active site illustrating the proximity of the
iꢀ1 amino acid to other residues of the protein (from PDB ID: 1AM2).[18a] Dis-
tances (····) between the a-carbon atom of Ala(iꢀ1) and the side chains of
the nearest amino acids are indicated.
insertion (Figure 6). Consistently, the largest effects for both
sets of libraries were observed when positions “iꢀ2” and “iꢀ3”
were substituted with either proline or glycine. At the iꢀ2 po-
sition, the insertion of both of these residues improved the ef-
ficiency of SP4-induced splicing (69–72%, 5 h) as compared to
the reference library X5T (55%, 5 h). The “iꢀ3” site also ap-
peared to be sensitive to changes in the conformational flexi-
bility of the corresponding residue, but in this case the intro-
duction of proline resulted in reduced biosynthetic precursor
reactivity (26% splicing, 5 h), whereas replacement with Gly
increased it (70% splicing, 5 h). For all the other positions, the
insertion of neither proline nor glycine altered the reactivity of
the biosynthetic precursor libraries at a significant level com-
pared to the reference library X5T.
Effect of conformationally constrained/flexible residues
within the target peptide sequence
Experiments were then carried out to gain further structure–re-
activity insights with respect to the other positions of the pep-
tide target sequence. Among the natural amino acids, proline
and glycine possess unique conformational properties because
of the restricted rotation around the NꢀC(a) bond (f angle) as
a result of the cyclic structure in the former and the absence
of a side-chain group in the latter. Accordingly, these residues
were chosen to probe the effect of altering the conformational
flexibility of the genetically encoded peptide moiety on the re-
activity of the precursor protein toward SP-induced macrocycli-
zation. A “proline scan” set of biosynthetic precursor libraries
was constructed by inserting proline at each of five positions
in a hexamer target sequence. A threonine was kept constant
at the iꢀ1 position in these constructs to eliminate biases due
to the large effect of substitutions at this site on the biosyn-
thetic precursor reactivity, as determined by the studies above.
To average the sequence-dependent effect of the remaining
positions not occupied by proline, these sites were fully
randomized by using a comprehensive degenerate codon
(NNK). A similar set of “glycine scan” libraries was constructed.
For comparison purposes, a biosynthetic precursor library con-
taining a fully randomized hexamer sequence with Thr at iꢀ1
(i.e., CBD-pAcF-X5T-GyrA, where X is any amino acid; referred
to as “X5T”) was also prepared.
A trend emerging from these analyses is that the positional
effect of the Pro/Gly substitution on the target sequence reac-
tivity correlates with its proximity to the intein and thus to the
reactive thioester bond. This result supports the critical role of
the C-terminal ligation step during MOrPH formation, as deter-
mined by mechanistic studies (vide infra). Based on the experi-
mental evidence accumulated here and during our earlier
investigations with biosynthetic precursors with randomized
pentamer and octamer target sequences,[14b] it is reasonable to
assume that most (>90%) of the randomized target sequen-
ces that undergo C-terminal ligation lead to the desired macro-
cycle product. Thus, these experiments showed the possibility
of accessing MOrPH libraries with possibly reduced or in-
creased peptide backbone flexibility through designed Pro/Gly
mutations within the genetically encoded peptide moiety of
the MOrPHs. Proline, in particular, has been exploited to gener-
ate conformationally biased peptide libraries.[6e,g] The demon-
strated compatibility of Pro substitutions with most of the
positions of an hexamer peptide moiety suggest that this
approach could provide a viable strategy to bias the conforma-
tional properties of the MOrPHs, complementary to the use of
different synthetic precursor scaffolds.
About 20000 recombinants from each of these libraries
were pooled together, and the corresponding biosynthetic pre-
cursors were expressed and purified as a mixture. As expected
with a Thr residue always at iꢀ1, premature splicing was found
to occur at comparably low levels for all the libraries (13% for
X5T compared to an average of 12 and 5% for the Pro and Gly
scan libraries, respectively). Upon treatment with SP4, signifi-
cant differences in the biosynthetic precursor reactivity were
observed for some of the positions as a result of the Pro/Gly
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 154

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Mechanism of the MOrPH-forming reaction
(Figure S8); thus suggesting that all these synthetic precursors
share a common mechanism. Based on these results, we con-
cluded that, whereas both reaction pathways can participate in
the MOrPH-forming process, the contribution of path B pre-
dominates, accounting for ꢁ90% of the macrocyclic product
formed over time.
Finally, we conducted studies to elucidate the mechanism of
the macrocyclization orchestrated by the AMA-based synthetic
precursors. As illustrated in Scheme 2, two possible reaction
pathways to the MOrPH product can be envisioned. The first
would proceed through the formation of an oxime SP–BP
adduct followed by intramolecular AMA-mediated cyclization
(“path A”), a process reminiscent of that involved in the forma-
tion of MOrPHs through CuAAC/hydrazide-mediated ligatio-
n.[14a] Alternatively, SP-induced intein splicing could occur first,
followed by an intramolecular oxime ligation (“path B”).
The observation that the bimolecular ligation reaction at the
C-terminal end of the target sequence is considerably faster
than that at the side-chain level is reasonable given the slow
kinetics of ketoxime formation at neutral pH.[20] The derived
mechanism is also consistent with the larger effect of Pro/Gly
substitutions close to the C terminus of the target sequence
on the precursor protein reactivity toward SP4 (Figure 6) as
well as with the low dependence of the rate of SP-induced
protein splicing on the target sequence length for constructs
with an identical iꢀ1 residue (Figure 1). Indeed, a larger de-
pendence on this parameter would be expected if the reaction
proceeded largely through path A, as observed for our MOrPH
synthesis through CuAAC/hydrazide-mediated ligations, which
relies exclusively on a side-chain!C-end tandem ligation
mechanism.[14a] Furthermore, as no acyclic intermediate “d” is
observed even at the shortest time points, we deduce that the
oxime ligation must be greatly accelerated when proceeding
intramolecularly rather than intermolecularly. This rate accelera-
tion, combined with the fact that both pathways remain avail-
able for macrocyclization, can thus account for the remarkably
clean outcome of these reactions.
To investigate this aspect, the products formed from the re-
action between SP4 (2) and the pentamer biosynthetic precur-
sor, CBD5(pAcF), were analyzed by MALDI-TOF MS after 30, 60,
120, and 180 min. At each time point, the only observed spe-
cies corresponded to the macrocycle. Similar results were ob-
tained for SP5 (3) and SP6 (4), whereas in the presence of SP7
(5), a certain amount of the linear acyclic intermediate CBD-
(pAcF)-TGSGT-5 (species “d” in Scheme 2) was also observed at
shorter times (30, 60 min). However, after 180 min, only the 5-
containing MOrPH was observed. In general, these results are
consistent with “path A”, in which a branched SP–protein
adduct is formed by oxime ligation, and this is followed by
macrocyclization through the SP amino-thiol moiety. At the
same time, a direct attack on the intein thioester by the AMA
moiety in the synthetic precursor followed by a fast oxime liga-
tion is not excluded.
In light of these mechanistic insights, it is also possible to
explain the marked difference in MOrPH-forming ability be-
tween the AMA-based SPs reported here and the cysteine- or
2-amino-thiophenol-based synthetic precursors described ear-
lier.[14b] Because path A appears to be a minor contributor in
the MOrPH-forming reaction, fast and efficient ligation at the
C-terminal end of the target sequence becomes crucial, there-
by highlighting the critical role of the AMA moiety and its su-
perior intein-splicing properties for the success of the macro-
cyclization.
To discern the relative contributions of paths A and B to
macrocyclization, the extent of protein splicing induced by SP4
was monitored over time in parallel reactions with CBD5(pAcF)
and the analogous construct CBD5(OpgY), in which pAcF is
replaced with O-propargyl-tyrosine, which lacks the side-chain
ketone required for oxime ligation. This allows measurement
of the rate at which the protein is spliced solely by action of
the amino-thiol moiety in the synthetic precursor (i.e., the first
step of path B). It emerged that the amount of spliced protein
in the reaction with CBD5(OpgY) was 90–94% of that observed
in the presence of CBD5(pAcF) across the various time points
(Figure 7). The same experiments were repeated with SP5–SP7,
and trends similar to that observed with SP4 were observed
Conclusions
In summary, we have developed an efficient, versatile, and scal-
able synthetic route to afford structurally different bifunctional
precursors for the synthesis of MOrPHs through a chemo-bio-
synthetic strategy based on a tandem oxime-/AMA-mediated
ligation. With these compounds, we demonstrated that a varie-
ty of nonproteogenic structures can be efficiently incorporated
into the final macrocyclic products. This feature is of particular
interest given the range of different scaffolds found to be ame-
nable to this process, and thus the important changes they are
expected to impose on the overall topology and conformation-
al properties of the resulting MOrPHs, especially in the context
of the shorter target sequences (i.e., tetra- to octamer). Overall,
the high yields and remarkably clean outcomes of the MOrPH-
forming reactions across this panel of compounds and target
sequence lengths of four to 15 amino acids demonstrate the
high degree of versatility and modularity of the methodology
for the preparation and diversification of this new class of
Figure 7. Time course of SP4-induced splicing of precursor protein CBD5 in-
&
^
corporating either pAcF ( ) or the ketone-free amino acid OpgY ( ). Error
bars are from experiments carried out in duplicate. Results from the same
experiment with synthetic precursors SP5–SP7 are provided in Figure S8.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 155

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
1.55 ppm (d, J=11.2 Hz, 9H); 13C NMR (CDCl3, 126 MHz): d=166.9,
152.8, 136.4, 132.6, 130.3, 128.9, 124.9, 121.8, 80.8, 52.0, 28.3,
17.9 ppm.
organo-peptide macrocycles. Furthermore, the demonstrated
possibility to carry out these cyclization reactions in complex
biological mixtures (i.e., cell lysate) adds to the value of this
strategy for molecular discovery campaigns.
This compound (6.63 g, 25 mmol) was dissolved in carbon tetra-
chloride (100 mL), and N-bromosuccinamide (4.89 g, 27.5 mmol,
1.1 equiv) was added. The reaction vessel was equipped with
a reflux condenser and irradiated with UV light for 3 h, then cooled
to room temperature; the mixture was filtered. The filtrate was dis-
solved in CH2Cl2 (100 mL), washed with saturated K2CO3 (aq) and
brine, then dried over anhydrous MgSO4. Volatiles were removed
to afford methyl 4-(bromomethyl)-3-((tert-butoxycarbonyl)amino)-
benzoate (6.7 g, 78% yield) as an orange–white solid. 1H NMR
(CDCl3, 500 MHz): d=8.47 (s, 1H), 7.73 (dd, J=8.0, 1.7 Hz, 1H), 7.36
(d, J=8.0 Hz, 1H), 6.75 (s, 1H), 4.50 (s, 2H), 3.91 (s, 3H), 1.55 ppm
(s, 9H); 13C NMR (CDCl3, 126 MHz): d=28.2, 29.9, 52.3, 81.3, 123.8,
125.1, 130.0, 131.5, 131.7, 136.9, 152.6, 166.2 ppm.
Our structure–reactivity analyses at the level of the geneti-
cally encoded moiety of these compounds showed the critical
role of the “iꢀ1” residue in influencing the reactivity and ac-
cessibility of the precursor proteins toward macrocyclization.
Importantly, a considerable fraction (60%) of the 20 possible
amino acid substitutions at this site was found to be compati-
ble with MOrPH formation. In addition, specific residues (F, Y, T,
A) that can maximize the yield of the macrocyclization reaction
were identified. These structure–reactivity data, together with
those gathered in the context of the Pro/Gly scanning experi-
ments, will be valuable to guide the design and construction
of high-quality and, optionally, conformationally biased MOrPH
libraries for screening purposes.
Methyl
4-(bromomethyl)-3-((tert-butoxycarbonyl)amino)benzoate
(6.7 g, 19.59 mmol), tritylmercaptan (6.5 g, 23.5 mmol, 1.2 equiv)
and potassium carbonate (3.25 g, 23.5 mmol, 1.2 equiv) were dis-
solved in dry DMF (100 mL), and the mixture was stirred under
argon at room temperature for 15 h, concentrated by rotary evapo-
ration, then dissolved in CH2Cl2. The solution was washed with ice-
cold H2O, with saturated NaHCO3 (aq), and finally with brine. The
organic layer was dried over anhydrous MgSO4 and filtered, and
volatiles were removed to afford methyl 3-((tert-butoxycarbonyl)-
amino)-4-((tritylthio)methyl)benzoate (7) as a golden yellow solid
(10.24 g, 97% crude yield). 1H NMR (CDCl3, 500 MHz): d=8.41 (s,
1H), 7.65 (d, J=7.9 Hz, 1H), 7.48 (d, J=8.0 Hz, 5H), 7.34 (t, J=
7.8 Hz, 6H), 7.25 (t, J=7.3 Hz, 5H), 7.18 (d, J=8.0 Hz, 1H), 6.72 (s,
1H), 3.88 (s, 3H), 3.21 (s, 2H), 1.56 ppm (d, J=2.5 Hz, 9H); 13C NMR
(CDCl3, 126 MHz): d=166.7, 152.8, 144.1, 136.9, 130.7, 129.3, 128.2,
126.9, 124.9, 123.1, 80.8, 67.4, 52.1, 34.1, 28.4 ppm. This material
was carried forward without further purification.
Finally, we have shed light on the mechanism of the macro-
cyclization and defined the relative contribution of the two re-
action pathways participating in the MOrPH-forming process.
These studies unveiled that the mechanism involving C-end!
side-chain ligation predominates, thus highlighting the critical
role of the AMA moiety in promoting MOrPH formation in this
method. These findings explained the structure–reactivity
trends observed here as well as previous observations on
other amino-thiol moieties.[14b] They also have a number of
other important implications. For example, they suggest that
even faster macrocyclization rates could be achieved by in-
creasing the rate of the ligation reaction at the side-chain
level. Furthermore, given the proven efficiency of the AMA
group in mediating efficient ligations at the level of polypep-
tide thioesters under catalyst-free conditions, we envision that
this moiety could prove useful in a number of other relevant
applications beyond MOrPH synthesis, such as protein conju-
gation and auxiliary-assisted peptide/protein ligation.
Synthesis of N-Boc-S-trityl-3-amino-4-(mercaptomethyl)benzyl
mesylate (8): Compound 7 (20.32g, 48 mmol) was dissolved in
anhydrous THF (400 mL), then the solution was cooled to 08C. A
solution of LiAlH4 in THF (1m, 52.8 mL, 52.8 mmol, 1.1 equiv) was
added slowly. The reaction mixture was stirred under argon at 08C
for 3 h, the reaction was quenched by slow addition of cold H2O
(3 mL) and NaOH (4n, 1 mL) at 08C, then the mixture was stirred
for 10min at room temperature. The mixture was concentrated
under reduced pressure, suspended in EtOAc/sat. NaHCO3 (10:1,
330 mL), then filtered through celite. The filtrate was washed once
with saturated NaHCO3, then with brine. The organic layer was
dried with anhydrous MgSO4, and volatiles were removed to afford
a yellow solid, which was purified by flash column chromatography
(silica gel, hexanes/EtOAc 7:3) to afford a yellow oil (18 g, 95%
Experimental Section
Reagents and analytical methods: Chemical reagents and sol-
vents were purchased from Sigma–Aldrich, Acros Organics, and
Fluka. Silica gel chromatography purifications were carried out by
using AMD Silica Gel 60 230–400 mesh. 1H and 13C NMR spectra
were recorded on Bruker Avance spectrometers by using solvent
peaks as reference. LC-MS analyses were performed on a Thermo
Scientific LTQ Velos ESI/ion-trap mass spectrometer coupled to an
Accela U-HPLC. MALDI-TOF spectra were acquired on a Bruker Au-
toflex III MALDI-TOF spectrometer by using a stainless steel MALDI
plate and sinapinic acid as matrix.
1
yield). H NMR (CDCl3, 500 MHz): d=7.78 (s, 1H), 7.49 (d, J=7.3 Hz,
5H), 7.34 (t, J=7.7 Hz, 5H), 7.26 (t, J=3.0 Hz, 5H), 7.13 (d, J=
7.8 Hz, 1H), 7.01 (d, J=7.8 Hz, 1H), 6.73 (s, 1H), 4.63 (s, 2H), 3.17
(s, 2H), 1.54 ppm (s, 9H); 13C NMR (CDCl3, 126 MHz): d=153.1,
144.3, 141.5, 136.8, 130.9, 129.3, 128.2, 126.9, 124.5, 122.2, 120.4,
80.5, 67.2, 65.1, 33.9, 28.4 ppm.
Synthesis of N-Boc-S-trityl-3-amino-4-(mercaptomethyl)benzoate
methyl ester (7): Methyl 3-amino-4-methylbenzoate (9.7 g,
58.7 mmol) and di-tert-butyl dicarbonate (17 mL, 74 mmol,
1.2 equiv) were dissolved in dry THF (200 mL), and the mixture was
heated under reflux for 72 h. The solvent was removed by rotary
evaporation to afford a pink–white solid that was suspended in
ice-cold hexanes (30 mL) and filtered to afford methyl 3-((tert-bu-
This compound (9.3 g, 18.19 mmol) was dissolved in anhydrous
CH2Cl2 (100 mL), and the solution was cooled to 08C. Methanesul-
fonylchloride (1.8 mL, 23.66 mmol, 1.3 equiv) and N,N-diisopropyl-
ethylamine (DIPEA; 4.2 mL, 23.66 mmol, 1.3 equiv) were added,
and the reaction mixture was stirred under argon at 08C for 2 h.
The mixture was then dissolved in CH2Cl2, washed twice with satu-
rated NaHCO3 (aq), and then once with brine. The organic layer
was dried over anhydrous MgSO4, and volatiles were removed to
toxycarbonyl)amino)-4-methylbenzoate as
yield). H NMR (CDCl3, 500 MHz): d=8.45 (s, 1H), 7.69 (d, J=7.9 Hz,
1H), 7.21 (d, J=7.9 Hz, 1H), 6.29 (s, 1H), 3.90 (s, 3H), 2.30 (s, 3H),
a white solid, (99%
1
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 156

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
afford 8 as a yellow solid (9.42 g, 88% yield). 1H NMR (CDCl3,
500 MHz): d=7.88 (s, 1H), 7.49 (d, J=7.3 Hz, 5H), 7.34 (t, J=7.7 Hz,
5H), 7.26 (d, J=14.6 Hz, 5H), 7.16 (d, J=7.8 Hz, 1H), 7.04 (d, J=
9.5 Hz, 1H), 6.75 (s, 1H), 5.18 (s, 2H), 3.17 (s, 2H), 2.90 (s, 3H),
1.54 ppm (s, 9H); 13C NMR (CDCl3, 126 MHz): d=152.8, 144.1, 137.3,
133.7, 131.3, 129.3, 128.2, 126.9, 126.3, 123.8, 121.9, 80.8, 71.3, 67.3,
38.4, 33.9, 28.4 ppm. The material was carried forward without fur-
ther purification.
ganic layer was dried over anhydrous MgSO4, filtered, and concen-
trated under reduced pressure to afford a yellow oil, which was
purified on silica gel with hexanes/EtOAc (1:1) as eluent to afford 9
1
as a yellow oil (2.3 g, quant.). H NMR (CDCl3, 400 MHz): d=7.80 (s,
1H), 7.50 (t, J=4.38 Hz, 6H), 7.34 (t, J=7.64 Hz, 6H), 7.25 (t, J=
7.28 Hz, 3H), 7.14 (d, J=7.80 Hz, 1H), 6.94 (dd, J=7.98, 1.70 Hz,
1H), 6.76 (s, 1H), 4.28 (s, 2H), 3.17 (s, 2H), 1.55 ppm (s, 9H);
13C NMR (CDCl3, 126 MHz): d=152.9, 144.2, 137.2, 135.8, 131.2,
129.3, 128.2, 126.9, 123.2, 121.3, 80.6, 67.2, 54.4, 33.9, 28.4 ppm;
MS-ESI: calcd for C32H32N4O2S: 559.68 [M+Na]+; found: 559.22.
Synthesis of SP4 (2): Compound 8 (1.06 g, 1.8 mmol) was dis-
solved in dry CH3CN (18 mL) and cooled to 08C. tert-Butyl-N-hy-
droxycarbamate (0.32 g, 2.4 mmol, 1.3 equiv) then 1,8-diazabicy-
cloundec-7-ene (DBU; 0.37 mL, 2.4 mmol, 1.3 equiv) were slowly
added. The reaction was stirred at 08C for 1 h, then warmed to am-
bient temperature, and stirred under argon overnight. Volatiles
were removed and the resulting crude mixture was dissolved in
CH2Cl2, washed with saturated K2CO3 (aq), and then with brine. The
organic layer was dried over anhydrous MgSO4 then concentrated
to afford a yellow oil, which was purified through flash chromatog-
raphy (silica gel, hexanes/EtOAc 7:3) to afford a yellow oil (1.00 g,
Synthesis of SP6 (4): Compounds 9 (0.1 g, 0.186 mmol) and 12
(0.127 g, 0.745 mmol, 4 equiv) were dissolved in THF/H2O (1:1,
6 mL). CuSO4 (0.045 g, 0.28 mmol, 1.5 equiv) and sodium ascorbate
(0.147 g, 0.745 mmol, 4 equiv) were added, and the reaction mix-
ture was stirred at room temperature for 30 min, then dissolved in
CH2Cl2 and washed twice with concentrated ammonium hydroxide,
once with saturated NaHCO3 (aq), and once with brine, then dried
over anhydrous MgSO4. Volatiles were removed under reduced
pressure, and the resulting material was purified on silica gel (hex-
1
1
89% yield). H NMR (CDCl3, 400 MHz): d=7.81 (s, 1H), 7.49 (d, J=
anes/EtOAc 7:3) to yield protected 4 (0.094 g, 72% yield). H NMR
7.6 Hz, 6H), 7.32 (q, J=7.6 Hz, 6H), 7.24 (t, J=7.2 Hz, 3H), 7.13 (d,
J=7.6 Hz, 2H), 7.02 (dd, J=8, 1.6 Hz, 1H), 6.74 (s, 1H), 4.79 (s, 2H),
3.17 (s, 1H), 1.54 (s, 9H), 1.46 ppm (s, 4H); 13C NMR (CDCl3,
126 MHz): d=156.6, 152.9, 144.2, 136.9, 136.3, 130.8, 129.3, 128.1,
126.9, 125.4, 124.2, 122.3, 81.6, 80.4, 77.9, 67.2, 33.9, 28.4,
27.5 ppm; MS-ESI: calcd for C37H42N2O5S: 649.79 [M+Na]+; found:
649.33.
(CDCl3, 400 MHz): d=7.77 (brs, 1H), 7.54 (s, 1H), 7.47 (d, J=4 Hz,
6H), 7.38 (s, 1H), 7.33 (t, J=8 Hz, 6H), 7.26–7.23 (m, 3H), 7.11 (d,
J=8 Hz, 1H), 6.68–6.83 (m, 1H), 6.76 (s, 1H), 5.47 (s, 2H), 4.96 (s,
2H), 3.15 (s, 2H), 1.53 (s, 9H), 1.45 ppm (s, 1H); MS-ESI: calcd for
C40H45N5O5S: 730.87 [M+Na]+; found: 730.26.
The protected precursor (0.094 g, 0.133 mmol) was deprotected
with TFA in CH2Cl2, as described above for 2, to afford 4 (0.065 g,
quant.). 1H NMR (CD3OD, 500 MHz): d=8.00 (s, 1H), 7.08 (d, J=
8 Hz, 1H), 6.71 (d, J=1.5 Hz, 1H), 6.63 (dd, J=8, 1.5 Hz, 1H), 5.468
(s, 2H), 4.933 (s, 2H), 3.671 ppm (s, 2H); 13C NMR (CD3OD,
126 MHz): d=146.8, 143.4, 136.4, 131.0, 127.3, 126.0, 118.9, 116.8,
68.9, 54.9, 25.6 ppm; MS-ESI: calcd for disulfide C22H28N10O2S2:
529.66 [M+H]+; found: 529.18.
The protected precursor (0.551 g, 0.88 mmol) was dissolved in an-
hydrous CH2Cl2 (9 mL), and the solution was cooled to 08C. Triiso-
propylsilane (TIPS; 0.45 mL, 2.2 mmol) was added followed by tri-
fluoroacetic acid (TFA; 2 mL). The reaction mixture was stirred
under argon at 08C for 30 min, then warmed to ambient tempera-
ture and concentrated under reduced pressure to afford a solid;
this was washed with ice-cold hexanes to afford 2 as an off-white
solid (0.366 g, quantitative yield). 1H NMR (CD3OD, 500 MHz): d=
7.06 (d, J=8 Hz, 1H), 6.76 (d, J=1.5 Hz, 1H), 6.65 (dd, J=8, 1.5 Hz,
1H), 4.56 (s, 2H), 3.69 (s, 2H), 1.38 ppm (s, 1H); 13C NMR (CD3OD,
126 MHz): d=146.5, 136.6, 130.4, 126.6, 118.9, 117.3, 78.9,
25.8 ppm; MS-ESI: calcd for disulfide C16H22N4O2S2: 367.51 [M+H]+;
found: 367.53.
Synthesis of N-Boc-S-trityl-3-amino-4-(mercaptomethyl)benzyl
amine (10): Compound 9 (2.3 g, 4.29 mmol) was dissolved in dry
THF (30 mL), and the mixture was cooled to 08C. A solution of
LiAlH4 in THF (1m, 5.16 mL, 5.16 mmol) was added slowly. The reac-
tion mixture was stirred at 08C under argon for 2 h, the reaction
was quenched by the slow addition of cold H2O (3 mL) and NaOH
(4n, 1 mL) at 08C, then the mixture was stirred for 10 min at room
temperature. Volatiles were removed under reduced pressure, and
the resulting material was dissolved in a mixture of ethyl acetate/
sat. NaHCO3 (100:15 mL). The solution was filtered through celite.
The filtrate was concentrated under reduced pressure to afford
a light brown solid that was purified by flash chromatography
(silica gel, hexanes/EtOAc 8:2!7:3) to afford 10 as a white solid
Synthesis of N-Boc propargyloxyamine (12): Propargyl bromide
(80% by weight in toluene; 1.6 g, 13.44 mmol) was dissolved in
dry MeCN (40 mL), and the mixture was cooled to 08C. tert-Butyl-
N-hydroxycarbamate (2.32 g, 17.47 mmol, 1.3 equiv) and DBU
(2.61 mL, 17.47 mmol, 1.3 equiv) were added. The reaction mixture
was stirred for 20 min at 08C, then warmed to ambient tempera-
ture, and stirred for another 1 h. Volatiles were removed under
reduced pressure, and the resulting yellow oil was suspended in
CH2Cl2, washed twice with saturated NaHCO3 (aq) and once with
brine, then dried over anhydrous MgSO4. Volatiles were removed
under reduced pressure, and the resulting crude material was puri-
fied on silica gel (hexanes/EtOAc 8:1!7:3) to give 12 (1.5 g, 65%
1
(2 g, 95% yield). H NMR (CDCl3, 400 MHz): d=7.74 (s, 1H), 7.49 (d,
J=7.70 Hz, 6H), 7.34 (t, J=7.71 Hz, 6H), 7.26–7.25 (m, 4H), 7.10 (d,
J=8.70 Hz, 1H), 6.39 (d, J=7.63 Hz, 1H), 6.73 (s, 1H), 3.80 (s, 2H),
3.16 (s, 2H), 1.54 ppm (s, 9H); 13C NMR (CDCl3, 126 MHz): d=153.1,
144.3, 143.9, 136.9, 130.9, 129.4, 128.2, 126.2, 122.4, 120.5, 80.4,
67.1, 46.3, 33.9, 28.43 ppm. MS-ESI: calcd for C32H34N2O2S: 1021.47
[2M+H]+; found: 1021.51.
1
yield). H NMR (CDCl3, 400 MHz): d=7.39 (s, 1H), 4.48 (d, J=2 Hz,
2H), 2.5 (s, 1H), 1.49 ppm (s, 1H); 13C NMR (CDCl3, 126 MHz): d=
Synthesis of tert-butyl 3-(carboxybenzyl)oxycarbamate (13): 3-
Bromobenzyl benzoate (0.5 g, 2.18 mmol) was dissolved in dry
MeCN (12 mL), and the mixture was cooled to 08C. tert-Butyl N-hy-
droxy carbamate (0.39 g, 2.9 mmol), then DBU (0.45 mL) were
slowly added. The reaction mixture was stirred at 08C for 2 h and
was then warmed to ambient temperature and left under argon
overnight. Volatiles were removed under reduced pressure, and
the resulting material was dissolved in CH2Cl2. This solution was
156.5, 82.1, 78.3, 75.6, 63.7, 28.2 ppm.
Synthesis of N-Boc-S-trityl-3-amino-4-(mercaptomethyl)benzyl
azide (9): Compound 8 (2.5 g, 4.24 mmol) and sodium azide
(0.56 g, 8.6 mmol) were dissolved in anhydrous DMF (30 mL), and
the mixture was stirred under argon at ambient temperature for
12 h. The reaction mixture was then dissolved in CH2Cl2 (150 mL)
and washed with saturated NaHCO3 (aq) and with brine. The or-
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 157

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
washed once with saturated K2CO3 (aq) and once with brine, then
dried over anhydrous MgSO4 and concentrated under reduced
pressure to afford a white solid (0.392 g, 64% yield). 1H NMR
(CDCl3, 400 MHz): d=8.06 (s, 1H), 8.02 (d, J=7.75 Hz, 1H), 7.61 (d,
J=7.63 Hz, 1H), 7.45 (t, J=7.68 Hz, 1H), 7.14 (s, 1H), 4.90 (s, 2H),
3.93 (s, 3H), 1.48 ppm (s, 9H).
acetate (50 mL) was added, and the solution was filtered through
celite. The filtrate was concentrated to afford the crude product as
yellow oil, which was purified by flash chromatography (silica gel,
hexanes/EtOAc 7:3) to afford 11 as a yellow oil (0.081 g, 92%
yield). H NMR (CDCl3, 400 MHz): d=7.18 (d, J=8.31 Hz, 2H), 7.05
(s, 1H), 6.67 (d, J=8.31 Hz, 2H), 4.73 (s, 2H), 1.48 ppm (s, 9H).
1
This material (0.173 g, 0.615 mmol) was dissolved in THF (5 mL),
and LiOH (1n, 1.24 mL) was added. The reaction mixture was
stirred under argon at ambient temperature overnight, then con-
centrated under reduced pressure, dissolved in ethyl acetate, and
washed once with HCl (aq) (0.25m) then once with brine. The
organic layer was dried over anhydrous MgSO4, filtered, then con-
centrated under reduced pressure to afford 13 as a white solid
(0.054 g, 79% yield). MS-ESI: calcd for C13H17NO5: 290.27 [M+Na]+;
found: 290.23.
Synthesis of SP7 (5): Compound 7 (1.6 g, 2.96 mmol) was dis-
solved in THF (37 mL). LiOH (1.0m, 7.54 mL) was added, and the re-
action mixture was stirred under argon at ambient temperature for
48 h. Volatiles were removed under reduced pressure, and the
resulting material was dissolved in ethyl acetate, washed with HCl
(0.25m) and brine, then dried over anhydrous MgSO4. Volatiles
were again removed to yield the carboxylic acid derivative of 7 as
an off-white solid (1.6 g, quant.). 1H NMR (CD3OD, 400 MHz): d=
7.99 (s, 1H), 7.67 (dd, J=7.97, 1.62 Hz, 1H), 7.43 (q, J=3.13 Hz,
6H), 7.31 (t, J=7.46 Hz, 6H), 7.23 (t, J=7.31 Hz, 3H), 7.09 (d, J=
8.07 Hz, 2H), 3.33 (s, 2H), 1.49 ppm (s, 9H).
Synthesis of SP5 (3): Compounds 10 (0.263 g, 0.98 mmol) and 13
(0.487 g, 0.98 mmol) were dissolved in dry CH2Cl2 (8 mL). 2-(1H-
Benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophospho-
nate (HBTU; 0.558 g, 1.47 mmol, 1.5 equiv) and DIPEA (0.43 mL,
2.45 mmol, 2.5 equiv) were added. The reaction mixture was stirred
at ambient temperature under argon for 3 h, then dissolved in
CH2Cl2, and washed with water and brine. The organic layer was
dried over anhydrous MgSO4 and concentrated under reduced
pressure to afford a yellow oil, which was purified by flash chroma-
tography (silica gel, hexanes/EtOAc 8:2) to afford a protected 3 as
a yellow oil (0.4 g, 55% yield). 1H NMR (CDCl3, 400 MHz): d=7.80
(d, J=7.4 Hz, 2H), 7.64 (d, J=7.8 Hz, 1H), 7.5 (s, 1H), 7.48 (d, J=
7.5 Hz, 6H), 7.41 (t, J=7.65 Hz, 1H), 7.33 (t, J=7.5 Hz, 6H), 7.25 (d,
J=5.7 Hz, 3H), 7.23 (s, 1H), 7.19 (brs, 1H), 7.11 (d, J=7.8 Hz, 1H),
7.01 (d, J=7.8 Hz, 1H), 6.73 (s, 1H), 4.87 (s, 2H), 4.57 (d, J=5.5,
2H), 3.16 (s, 2H), 1.53 (s, 9H), 1.44 ppm (s, 9H); 13C NMR (CDCl3,
126 MHz): d=166.9, 156.7, 153.0, 144.2, 138.5, 136.9, 136.4, 134.7,
131.8, 131.2, 129.3, 128.7, 128.2, 127.3, 127.1, 126.9, 123.5, 121.5,
81.9, 80.6, 77.8, 67.2, 43.9, 33.9, 28.4, 28.1 ppm; MS-ESI: calcd for
C45H49N3O6S: 782.33 [M+Na]+; found: 782.35.
The carboxylic acid (1.03 g, 2 mmol) and 11 (0.619 g, 2.6 mmol,
1.3 equiv) were dissolved in anhydrous CH2Cl2 (4 mL). HBTU
(1.1379 g, 3 mmol, 1.5 equiv) and DIPEA (2.6 mL, 2.5 equiv) were
added. The solution was heated to 308C and stirred under argon
for 48 h. CH2Cl2 (100 mL) was added, and the solution was washed
with water and saturated NaHCO3, and then dried over anhydrous
MgSO4. Volatiles were removed to yield a crude product, which
was purified by flash chromatography (silica gel, hexanes/EtOAc
7:3) to afford fully protected 5 as a yellow oil (0.536 g, 26% yield).
1H NMR (CDCl3, 400 MHz): d=8.23 (s, 1H), 8.08 (s, 1H), 7.63 (d, J=
8.4 Hz, 2H), 7.56 (dd, J=7.6, 1.2 Hz, 1H), 7.49 (d, J=7.6 Hz, 6H),
7.35 (t, J=7.6 Hz, 6H), 7.28–7.21 (m, 6H), 7.11 (s, 1H), 6.82 (s, 1H),
4.82 (s, 2H), 3.21 (s, 1H), 1.56 (s, 9H), 1.49 ppm (s, 9H); 13C NMR
(CDCl3, 126 MHz): d=165.2, 156.6, 153.0, 143.9, 138.2, 136.6, 134.9,
131.4, 131.1, 129.7, 129.2, 128.1, 126.9, 123.3, 120.2, 81.4, 80.9, 77.8,
67.3, 33.8, 28.3, 28.1 ppm; MS-ESI: calcd for C44H47N3O6S: 768.32
[M+Na]+; found: 768.26.
Protected 5 (0.268 g, 0.36 mmol) was deprotected with TFA in
CH2Cl2, as described for 2, to afford 5 (0.199 g, quantitative yield).
1H NMR (CD3OD, 400 MHz): d=7.69 (d, J=8 Hz, 2H), 7.37 (d, J=
8 Hz, 2H), 7.28 (s, 1H), 7.22–7.20 (m, 2H), 4.67 (s, 2H), 3.76 (s, 2H),
3.33 ppm (s, 2H); 13C NMR (CD3OD, 126 MHz): d=169.2, 146.9,
139.8, 136.1, 134.6, 130.5, 130.0, 122.1, 117.9, 116.2, 78.6, 25.7 ppm;
MS-ESI: calcd for disulfide C30H32N6O4S2: 605.75 [M+H]+; found:
605.57.
The protected precursor (0.186 g, 0.244 mmol) was deprotected
with TFA in CH2Cl2, as described for 2, to afford 3 as a white solid
1
(0.136 g, quant.). H NMR (CD3OD, 400 MHz): d=7.83 (s, 1H), 7.78
(d, J=7.5 Hz, 1H), 7.51 (d, J=7.5 Hz, 1H), 7.44 (t, J=8 Hz, 1H), 6.96
(d, J=8 Hz, 1H), 6.74 (s, 1H), 6.65 (dd, J=8, 1.5 Hz, 1H), 5.50 (s,
1H), 4.70 (s, 2H), 4.48 (s, 2H), 3.64 ppm (s, 2H); 13C NMR (CD3OD,
126 MHz): d=169.9, 147.2, 140.9, 139.5, 135.9, 132.8, 132.4, 129.7,
128.1, 127.9, 121.2, 118.2, 116.3, 78.2, 44.4, 40. 7 ppm; MS-ESI: calcd
for disulfide C32H36N6O4S2: 633.80 [M+H]+; found: 633.60.
Cloning and plasmid construction: Construction of the plasmids
for the expression of CBD4(pAcF) through CBD12(pAcF) was de-
scribed earlier.[14b] The plasmid for CBD15(pAcF) was prepared by
overlap extension PCR with CBD12(pAcF) as template. The PCR
product (750 bp) was cloned into a pET22b(+) vector (Novagen)
by using NdeI and XhoI restriction enzymes. The glycine and pro-
line scan libraries were prepared in a similar manner by using
pCBD6 vector as template and forward primers containing NNK
codons corresponding to the randomized positions. The plasmids
encoding for the iꢀ1 variants were prepared by using forward pri-
mers of general sequence 5’-AGACAG GATCCG GCXXXT GCATCA
CGG-3’, 5’-GCTAGT TATTGC TCAGCGG-3’ as reverse primer, and
pCBD5 vector as template. The PCR product was cloned into the
BamHI/XhoI cassette of the pCBD5 vector. The plasmid constructs
were confirmed by DNA sequencing (Functional Genomics Center
of the University of Rochester).
Synthesis of tert-butyl-(4-aminobenzyl)oxycarbamate (11): 1-
(Bromomethyl)-4-nitrobenzene (0.5 g, 2.3 mmol) was dissolved in
dry MeCN (12 mL) and cooled to 08C. tert-Butyl-N-hydroxycarba-
mate (0.616 g, 4.6 mmol, 2 equiv) and DBU (0.71 mL, 4.6 mmol,
2 equiv) were added. The reaction mixture was stirred at 08C for
2 h, then warmed to ambient temperature overnight. Volatiles
were removed under reduced pressure, and the resulting material
was dissolved in CH2Cl2, washed with saturated K2CO3 (aq) and
with brine, then dried over anhydrous MgSO4 and filtered. The vol-
atiles were removed to yield crude tert-butyl-(4-nitrobenzyl)oxycar-
bamate (0.432 g, 70% crude yield), which was carried forward with-
1
out further purification. H NMR (CDCl3, 500 MHz): d=8.21 (dd, J=
6.83, 1.93 Hz, 2H), 7.58 (dd, J=6.63, 2.04 Hz, 2H), 7.41 (s, 1H), 4.97
(s, 2H), 1.48 ppm (s, 9H).
tert-Butyl-(4-nitrobenzyl)oxycarbamate (0.1 g, 0.37 mmol) was com-
bined with 10% Pd/C (0.032 g). The mixture was suspended in dry
methanol (4 mL), then purged with hydrogen gas for 30 min; ethyl
Protein expression and purification: Proteins were expressed in
E. coli BL21(DE3) cells co-transformed with the plasmid encoding
for the biosynthetic precursor and pEVOL_pAcF vector, which en-
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 158

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
codes for an orthogonal tRNACUA, and pAcF-tRNA synthetase[15] for
the incorporation of pAcF through amber stop codon suppression.
Overnight cultures were grown in lysogeny broth containing ampi-
cillin (50 mgLꢀ1) and chloramphenicol (34 mgLꢀ1) and were used
[2] a) Y. Hamada, T. Shioiri, Chem. Rev. 2005, 105, 4441–4482; b) N. H. Tan,
J. Zhou, Chem. Rev. 2006, 106, 840–895.
[3] a) T. Satoh, S. Li, T. M. Friedman, R. Wiaderkiewicz, R. Korngold, Z.
Huang, Biochem. Biophys. Res. Commun. 1996, 224, 438–443; b) D. P.
Fairlie, J. D. Tyndall, R. C. Reid, A. K. Wong, G. Abbenante, M. J. Scanlon,
D. R. March, D. A. Bergman, C. L. Chai, B. A. Burkett, J. Med. Chem. 2000,
43, 1271–1281; c) D. Wang, W. Liao, P. S. Arora, Angew. Chem. 2005, 117,
6683–6687; Angew. Chem. Int. Ed. 2005, 44, 6525–6529.
[4] a) L. D. Walensky, A. L. Kung, I. Escher, T. J. Malia, S. Barbuto, R. D.
Wright, G. Wagner, G. L. Verdine, S. J. Korsmeyer, Science 2004, 305,
1466–1470; b) T. Rezai, B. Yu, G. L. Millhauser, M. P. Jacobson, R. S.
Lokey, J. Am. Chem. Soc. 2006, 128, 2510–2511; c) O. S. Gudmundsson,
D. G. Vander Velde, S. D. Jois, A. Bak, T. J. Siahaan, R. T. Borchardt, J. Pept.
Res. 1999, 53, 403–413.
[5] a) Y. Q. Tang, J. Yuan, G. Osapay, K. Osapay, D. Tran, C. J. Miller, A. J.
Ouellette, M. E. Selsted, Science 1999, 286, 498–502; b) F. Al-Obeidi,
A. M. Castrucci, M. E. Hadley, V. J. Hruby, J. Med. Chem. 1989, 32, 2555–
2561; c) M. A. Dechantsreiter, E. Planker, B. Matha, E. Lohof, G. Holze-
mann, A. Jonczyk, S. L. Goodman, H. Kessler, J. Med. Chem. 1999, 42,
3033–3040; d) N. R. Graciani, K. Y. Tsang, S. L. McCutchen, J. W. Kelly,
Bioorg. Med. Chem. 1994, 2, 999–1006; e) R. L. Dias, R. Fasan, K. Moehle,
A. Renard, D. Obrecht, J. A. Robinson, J. Am. Chem. Soc. 2006, 128,
2726–2732; f) R. M. Cardoso, F. M. Brunel, S. Ferguson, M. Zwick, D. R.
Burton, P. E. Dawson, I. A. Wilson, J. Mol. Biol. 2007, 365, 1533–1544;
g) L. K. Henchey, J. R. Porter, I. Ghosh, P. S. Arora, ChemBioChem 2010,
11, 2104–2107.
to inoculate M9 medium (0.4 L) containing ampicillin (50 mgLꢀ1
)
and chloramphenicol (34 mgLꢀ1) and supplemented with 1% gly-
cerol. At OD600 =0.6, protein expression was induced by adding
0.05% l-arabinose, isopropyl-b-d-thiogalactopyranoside (IPTG;
0.25 mm), and pAcF (1 mm). Cultures were grown for an additional
16 h at 278C and harvested by centrifugation at 3399g. Frozen
cells were resuspended in Tris (50 mm), NaCl (300 mm), and imida-
zole buffer (20 mm; pH 7.4) and lysed by sonication. Proteins were
purified through Ni-NTA affinity chromatography by using Tris
(50 mm), NaCl (150 mm), and imidazole (300 mm; pH 7.4) for the
elution. The protein samples were concentrated, buffer-exchanged
with potassium phosphate buffer (50 mm, pH 7.5) containing NaCl
(150 mm), and stored at ꢀ808C. The OpgY-containing biosynthetic
precursors were prepared in a similar manner with only difference
being that cells contained a vector encoding for a tRNACUA/AARS
pair specific for OpgY, and OpgY was added to the culture medium
during protein expression.[14a] The biosynthetic precursor libraries
were prepared by pooling together E. coli colonies from the cor-
responding LB plates after transformation of the DNA ligation
mixture.
[6] a) E. Koivunen, D. A. Gay, E. Ruoslahti, J. Biol. Chem. 1993, 268, 20205–
20210; b) N. C. Wrighton, F. X. Farrell, R. Chang, A. K. Kashyap, F. P. Bar-
bone, L. S. Mulcahy, D. L. Johnson, R. W. Barrett, L. K. Jolliffe, W. J.
Dower, Science 1996, 273, 458–464; c) L. Oligino, F. D. Lung, L. Sastry, J.
Bigelow, T. Cao, M. Curran, T. R. Burke, Jr., S. Wang, D. Krag, P. P. Roller,
C. R. King, J. Biol. Chem. 1997, 272, 29046–29052; d) W. J. Fairbrother,
H. W. Christinger, A. G. Cochran, G. Fuh, C. J. Keenan, C. Quan, S. K.
Shriver, J. Y. Tom, J. A. Wells, B. C. Cunningham, Biochemistry 1998, 37,
17754–17764; e) G. R. Nakamura, M. A. Starovasnik, M. E. Reynolds, H. B.
Lowman, Biochemistry 2001, 40, 9828–9835; f) K. Deshayes, M. L. Schaff-
er, N. J. Skelton, G. R. Nakamura, S. Kadkhodayan, S. S. Sidhu, Chem. Biol.
2002, 9, 495–505; g) J. Stamos, C. Eigenbrot, G. R. Nakamura, M. E. Rey-
nolds, J. Yin, H. B. Lowman, W. J. Fairbrother, M. A. Starovasnik, Structure
2004, 12, 1289–1301; h) C. D. Shomin, E. Restituyo, K. J. Cox, I. Ghosh,
Bioorg. Med. Chem. 2011, 19, 6743–6749.
[7] a) S. W. Millward, T. T. Takahashi, R. W. Roberts, J. Am. Chem. Soc. 2005,
127, 14142–14143; b) C. Heinis, T. Rutherford, S. Freund, G. Winter, Nat.
Chem. Biol. 2009, 5, 502–507; c) Y. V. Guillen Schlippe, M. C. T. Hartman,
K. Josephson, J. W. Szostak, J. Am. Chem. Soc. 2012, 134, 10469–10477;
d) S. Chen, J. Morales-Sanfrutos, A. Angelini, B. Cutting, C. Heinis, Chem-
BioChem 2012, 13, 1032–1038.
[8] a) C. P. Scott, E. Abel-Santos, M. Wall, D. C. Wahnon, S. J. Benkovic, Proc.
Natl. Acad. Sci. USA 1999, 96, 13638–13643; b) T. M. Kinsella, C. T.
Ohashi, A. G. Harder, G. C. Yam, W. Li, B. Peelle, E. S. Pali, M. K. Bennett,
S. M. Molineaux, D. A. Anderson, E. S. Masuda, D. G. Payan, J. Biol. Chem.
2002, 277, 37512–37518; c) J. A. Kritzer, S. Hamamichi, J. M. McCaffery,
S. Santagata, T. A. Naumann, K. A. Caldwell, G. A. Caldwell, S. Lindquist,
Nat. Chem. Biol. 2009, 5, 655–663; d) G. Deschuyteneer, S. Garcia, B.
Michiels, B. Baudoux, H. Degand, P. Morsomme, P. Soumillion, ACS
Chem. Biol. 2010, 5, 691–700.
Cyclization reactions: Reactions were carried out by adding the
synthetic precursor (15 mm) to a solution of protein precursor
(100 mm) in potassium phosphate buffer (50 mm, NaCl 150 mm,
pH 7.5) in the presence of TCEP (20 mm). For SDS-PAGE analyses,
and aliquot (5 mL) of the reaction mixture was removed at the indi-
cated time point(s), diluted in DTT-free 4ꢂ loading buffer, and ana-
lyzed on 18% polyacrylamide gels. The extent of protein splicing
was measured and quantified by densitometry analysis using the
NIH Image Software. The percentage of SP-induced protein splicing
was calculated based on the difference between the amount of
spliced protein at t=0 and at the time of the analysis. MALDI-TOF
analyses of the low-molecular-weight products (8–10 kDa) of the
reactions were carried out on a Bruker Autoflex III MALDI-TOF
spectrometer. Prior to analysis, protein samples were diluted in
50% acetonitrile/H2O (0.1% TFA), and this solution was mixed with
a sinapinic acid solution (10 mg mLꢀ1 in 50% acetonitrile/H2O with
0.1% TFA). The samples were analyzed in reflectron positive (RP)
mode and calibration by using low-molecular-weight (2–15 kDa)
protein standards.
Acknowledgements
This work was supported by the U.S. National Science Foundation
grant CHE-1112342. N.T.J. acknowledges the NSF REU program
for financial support. MS instrumentation was supported by the
U.S. National Science Foundation grants CHE-0840410 and CHE-
0946653. We are grateful to Peter G. Schultz for the plasmid en-
coding for tRNACUA/pAcF-t-RNA synthetase.
[9] a) R. M. Kohli, C. T. Walsh, M. D. Burkart, Nature 2002, 418, 658–661; b) T.
Hamamoto, M. Sisido, T. Ohtsuki, M. Taki, Chem. Commun. 2011, 47,
9116–9118.
[10] a) Y. Goto, A. Ohta, Y. Sako, Y. Yamagishi, H. Murakami, H. Suga, ACS
Chem. Biol. 2008, 3, 120–129; b) T. Kawakami, A. Ohta, M. Ohuchi, H.
Ashigai, H. Murakami, H. Suga, Nat. Chem. Biol. 2009, 5, 888–890.
[11] a) M. S. Donia, B. J. Hathaway, S. Sudek, M. G. Haygood, M. J. Rosovitz, J.
Ravel, E. W. Schmidt, Nat. Chem. Biol. 2006, 2, 729–735; b) Y. X. Shi, X. A.
Yang, N. Garg, W. A. van der Donk, J. Am. Chem. Soc. 2011, 133, 2338–
2341; c) S. J. Pan, A. J. Link, J. Am. Chem. Soc. 2011, 133, 5016–5023;
d) T. A. Knappe, F. Manzenrieder, C. Mas-Moruno, U. Linne, F. Sasse, H.
Kessler, X. Xie, M. A. Marahiel, Angew. Chem. 2011, 123, 8873–8876;
Angew. Chem. Int. Ed. 2011, 50, 8714–8717.
Keywords: bio-orthogonal ligation · cyclization · intein ·
macrocycles · organo-peptide hybrids
[1] a) L. A. Wessjohann, E. Ruijter, D. Garcia-Rivera, W. Brandt, Mol. Diversity
2005, 9, 171–186; b) E. M. Driggers, S. P. Hale, J. Lee, N. K. Terrett, Nat.
Rev. Drug Discovery 2008, 7, 608–624; c) J. A. Robinson, S. Demarco, F.
Gombert, K. Moehle, D. Obrecht, Drug Discovery Today 2008, 13, 944–
951; d) E. Marsault, M. L. Peterson, J. Med. Chem. 2011, 54, 1961–2004.
[12] a) Z. J. Gartner, B. N. Tse, R. Grubina, J. B. Doyon, T. M. Snyder, D. R. Liu,
Science 2004, 305, 1601–1605; b) R. Fasan, R. L. Dias, K. Moehle, O.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 159

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Zerbe, D. Obrecht, P. R. Mittl, M. G. Grutter, J. A. Robinson, ChemBio-
Chem 2006, 7, 515–526; c) S. B. Shin, B. Yoo, L. J. Todaro, K. Kirshen-
baum, J. Am. Chem. Soc. 2007, 129, 3218–3225; d) Y. U. Kwon, T. Koda-
dek, Chem. Commun. 2008, 5704–5706; e) V. S. Fluxa, J. L. Reymond,
Bioorg. Med. Chem. 2009, 17, 1018–1025; f) D. G. Rivera, O. E. Vercillo,
L. A. Wessjohann, Org. Biomol. Chem. 2008, 6, 1787–1795; g) S. A.
Fowler, D. M. Stacy, H. E. Blackwell, Org. Lett. 2008, 10, 2329–2332; h) A.
Isidro-Llobet, T. Murillo, P. Bello, A. Cilibrizzi, J. T. Hodgkinson, W. R. Gal-
loway, A. Bender, M. Welch, D. R. Spring, Proc. Natl. Acad. Sci. USA 2011,
108, 6793–6798; i) R. Hili, V. Rai, A. K. Yudin, J. Am. Chem. Soc. 2010,
132, 2889–2891; j) S. Ghosh, L. A. Ingerman, A. G. Frye, S. J. Lee, M. R.
Gagne, M. L. Waters, Org. Lett. 2010, 12, 1860–1863; k) V. Dewan, T. Liu,
K. M. Chen, Z. Qian, Y. Xiao, L. Kleiman, K. V. Mahasenan, C. Li, H.
Matsuo, D. Pei, K. Musier-Forsyth, ACS Chem. Biol. 2012, 7, 761–769.
[13] a) S. W. Millward, S. Fiacco, R. J. Austin, R. W. Roberts, ACS Chem. Biol.
2007, 2, 625–634; b) T. S. Young, D. D. Young, I. Ahmad, J. M. Louis, S. J.
Benkovic, P. G. Schultz, Proc. Natl. Acad. Sci. USA 2011, 108, 11052–
11056; c) M. D. Tianero, M. S. Donia, T. S. Young, P. G. Schultz, E. W.
Schmidt, J. Am. Chem. Soc. 2012, 134, 418–425.
[16] a) S. Chong, F. B. Mersha, D. G. Comb, M. E. Scott, D. Landry, L. M. Vence,
F. B. Perler, J. Benner, R. B. Kucera, C. A. Hirvonen, J. J. Pelletier, H.
Paulus, M. Q. Xu, Gene 1997, 192, 271–281; b) T. T. Hoang, Y. Ma, R. J.
Stern, M. R. McNeil, H. P. Schweizer, Gene 1999, 237, 361–371; c) T. C.
Evans, Jr., J. Benner, M. Q. Xu, Protein Sci. 1998, 7, 2256–2264; d) M. W.
Southworth, K. Amaya, T. C. Evans, M. Q. Xu, F. B. Perler, Biotechniques
1999, 27, 110–120.
[17] T. M. Hackeng, J. H. Griffin, P. E. Dawson, Proc. Natl. Acad. Sci. USA 1999,
96, 10068–10073.
[18] a) T. Klabunde, S. Sharma, A. Telenti, W. R. Jacobs, Jr., J. C. Sacchettini,
Nat. Struct. Biol. 1998, 5, 31–36; b) A. Romanelli, A. Shekhtman, D. Cow-
burn, T. W. Muir, Proc. Natl. Acad. Sci. USA 2004, 101, 6397–6402; c) S.
Frutos, M. Goger, B. Giovani, D. Cowburn, T. W. Muir, Nat. Chem. Biol.
2010, 6, 527–533.
[19] A. Telenti, M. Southworth, F. Alcaide, S. Daugelat, W. R. Jacobs, Jr., F. B.
Perler, J. Bacteriol. 1997, 179, 6378–6382.
[20] W. P. Jencks, J. Am. Chem. Soc. 1959, 81, 475–481.
[14] a) J. M. Smith, F. Vitali, S. A. Archer, R. Fasan, Angew. Chem. 2011, 123,
5181–5186; Angew. Chem. Int. Ed. 2011, 50, 5075–5080; b) M. Satyanar-
ayana, F. Vitali, J. R. Frost, R. Fasan, Chem. Commun. 2012, 48, 1461–
1463.
[15] L. Wang, Z. Zhang, A. Brock, P. G. Schultz, Proc. Natl. Acad. Sci. USA
2003, 100, 56–61.
Received: September 10, 2012
Please note: Minor changes have been made to Figure 2 since this manu-
script was published in ChemBioChem EarlyView. The Editor.
Published online on November 30, 2012
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 147 – 160 160