Angewandte
Chemie
achieved in a few steps[4b,9] from the protected parent amino
acids. Propargylamine, a precursor for GlyY[Tz]Xaa, is even
commercially available (Gly = G = glycine, Xaa = any a-
amino acid). Because Gly is one of the most common and
well-distributed residues in proteins, retrosynthetic discon-
nections at Gly-Xaa sites are particularly attractive. The
incorporation of 1,4-disubstituted triazoles into the backbone
of long peptides has been reported and was based on a
classical SPPS approach involving triazole-containing pseudo-
dipeptides as building blocks.[8c] To our knowledge, the few
examples of post-SPPS backbone engineering by triazole
formation involve protected peptide fragments.[12–14] Most
examples concern small cyclic peptides. Given that CuAAC is
extremely robust, orthogonal to most other reactions, and
tolerant of an extensive variety of functional groups, including
unprotected peptide side chains, we decided to explore the
potential of CuAAC to assemble unprotected peptides into
backbone-engineered protein analogues.
An iterative version of peptidomimetic triazole ligation
(PTL) is crucial for the synthesis of medium-sized (> 80
amino acids) bioactive proteins: the temporary masking of a
terminal alkyne with a silyl protecting group[15] enables
multiple successive CuAAC steps under the mildest condi-
tions compatible with complex biomacromolecules. We and
others have shown that triisopropylsilyl (TIPS) is a group of
choice for multiple linear CuAAC steps,[16] and we have
demonstrated its full compatibility with the conditions of
fluorenylmethoxycarbonyl (Fmoc) SPPS.[16a]
consensus QXVXG motif (Figure 1b, loop L1), and a C-
terminal b-hairpin loop (Figure 1b, loop L2).[21] Together they
form a large, wedge-shaped binding surface that inserts into
the cathepsin active-site cleft.[22]
From a retrosynthetic point of view, the 97 amino acid
sequence of cystatin A has been dissected into three sec-
tions,[23] to ensure an optimal length of the fragments for their
synthesis by SPPS (Figure 1, green, red, and blue parts). The
C-terminal propargylamide peptides 1 and 2a were obtained
using a general approach based on a backbone-amide-
linker[24,25] (BAL) strategy (Scheme 1 in the Supporting
Information). Propargylamine and 3-(triisopropylsilyl)prop-
2-yn-1-amine[16a] were subjected to reductive amination with
4-(4-formyl-3,5-dimethoxyphenoxy)butanoic acid, and the
resulting secondary amine was subsequently Fmoc-protected
and loaded onto aminomethyl ChemMatrix resin through
standard amide coupling. Fmoc deprotection furnished the
solid-supported secondary amine, which was efficiently cou-
pled to the next amino acid using HATU (N-[(dimethylami-
no)-1H-1,2,3-triazolo[4,5-b]pyridin-1-yl-methylene]-N-meth-
ylmethanaminium hexafluorophosphate N-oxide) as the cou-
pling reagent. The SPPS elongations of the three peptide
segments 1, 2a, and 3 were achieved using the Fmoc/tBu
strategy. The main challenge at this stage was to obtain crude
peptides with a high standard purity and yield. Residue 1Met,
which only minimally participates in the inhibitory properties
of the protein,[26,27] was substituted by an isosteric norleucyl
residue in peptide segment 1.
For the proof of concept of a multiple-PTL approach, we
focused on human cystatin A (also known as stefin A), a
natural cysteine-protease inhibitor about 100 amino acids
long and without any cysteine[17] (see Figure 1a for its amino
acid sequence).
Cystatin A potently inhibits papain-like cysteine pro-
teases of the family 1, including cysteine cathepsins. Besides
their primary role in end-stage breakdown of endocytosed
proteins in lysosomal compartments, these proteases are also
associated with numerous major pathological disorders;[18,19]
this association makes the proteases emerging therapeutic
targets.[20] The inhibitory site of cystatin A involves three
noncontiguous regions comprising an N-terminal part with a
conserved glycyl residue, a central b-hairpin loop containing a
Although the synthesis of 3 was straightforward, the low
yields of the first elongations of peptide 1 and 2a carried out
with HBTU (N-[(1H-benzotriazol-1-yl)(dimethylamino)-
methylene]-N-methylmethanaminium hexafluorophosphate
N-oxide) as the coupling reagent were very disappointing.
We exploited recent advancements in SPPS by combining the
use of pseudoproline dipeptide derivatives,[28] a polar resin,[29]
and HATU as a powerful coupling reagent, and we conducted
a meticulous step-by-step optimization of the synthesis of 1
and 2a (see the Supporting Information for details). The
synthesis of 2a was particularly demanding,[30] and its
purification by conventional reversed-phase (RP)-HPLC
resulted in very low yields. A careful analysis of the crude
peptide revealed the main coproducts to be N-truncated
peptides bearing terminal acetamides, which would not
interfere with the PTL (see the Supporting Information).
We thus decided to directly engage crude, lyophilized 2a in
the CuAAC step, whereas peptide segments 1 and 3 were
purified by RP-HPLC before ligation.
We then determined appropriate reaction conditions to
carry out the first ligation (Scheme 2) between 1 and 2a. We
chose the most standard CuSO4/sodium ascorbate protocol
for the generation of the active CuI species, and all the triazole
ligations were conducted in an argon atmosphere in a
carefully degassed phosphate buffer at pH 8. A slight excess
of 2a was engaged in a CuAAC with 1 (ca. 1 mm) in a 1:1
HFIP (hexafluoroisopropyl alcohol)/buffer mixture.[31] Grat-
ifyingly, under these conditions, peptide 1 was totally
consumed in less than 15 min to cleanly yield the triazolopep-
tide 4a (Figure 2a, trace 2 and the Supporting Information).
Further treatment with an excess of tetra-n-butylammonium
Figure 1. a) Amino acid sequence of human cystatin A. M=Met=me-
thionine, I=Ile=isoleucine, P=Pro=proline, L=Leu=leucine,
S=Ser=serine, E=Glu=glutamic acid, A=Ala=alanine, K=Lys =ly-
sine, T=Thr=threonine, Q=Gln=glutamine, V=Val =valine,
D=Asp=aspartic acid, Y=Tyr =tyrosine, N=Asn=asparagine,
R=Arg=arginine, H=His=histidine, F=Phe=phenylalanine.
b) Schematic three-dimensional structure of human cystatin A.
Angew. Chem. Int. Ed. 2012, 51, 718 –722
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
719