Molecular modeling, synthesis, and biological evaluation of macrocyclic calpain inhibitors

Article abstract of DOI:10.1002/anie.200805014

The design and elaboration of a series of macrocyclic templates that exhibit a propensity to adopt a β-strand-like peptide-backbone conformation led to potent and selective inhibitors of calpain 2. Macrocycle 1 retarded calcium-induced opacification in an

Full text of DOI:10.1002/anie.200805014

Angewandte
Chemie
DOI: 10.1002/anie.200805014
Drug Design
Molecular Modeling, Synthesis, and Biological Evaluation of
Macrocyclic Calpain Inhibitors**
Andrew D. Abell,* Matthew A. Jones, James M. Coxon, James D. Morton, Steven G. Aitken,
Stephen B. McNabb, Hannah Y.-Y. Lee, Janna M. Mehrtens, Nathan A. Alexander,
Blair G. Stuart, Axel T. Neffe, and Roy Bickerstaffe
The introduction of a macrocycle into a biologically active
peptide can increase potency[1,2] and selectivity[3] by reducing
the entropic penalty of inhibitor–enzyme binding. The
incorporation of macrocycles in peptides has been used to
mimic secondary structure, such as extended b-strand-like
and bent b-turn-like conformations.[3,4] The challenge is to
devise protocols for the design, evaluation, and synthesis of
macrocyclic templates that provide access to families of
inhibitors.[5] Such templates should 1) have a well-defined
conformation that can be readily assessed by computational
screening, 2) be readily synthesized from natural building
blocks, and 3) be easily modified to target an enzyme for a
Scheme 1. Macrocyclic templates and acyclic peptides. Dihedral angles
F and Y that describe a b strand are shown. Boc=tert-butoxycarbonyl,
Cbz=carbobenzyloxy.
specific biological application.
Herein we present studies on the 16–19-membered
macrocycles 1–4 (Scheme 1) designed to be constrained into
a b-strand-like geometry, a conformation universally adopted
by inhibitors and substrates on binding to a protease.[6] We
report a versatile approach based on ring-closing metathesis
(RCM) to these orthogonally protected templates as well as
computational analysis of their potential to form a b strand
and of their binding to a target protease. The templates were
converted into aldehydes 1d–4d as potential inhibitors of the
calcium-activated cysteine protease calpain. The alcohols 1c–
4c were evaluated to establish whether the macrocycles
negate the need for a reactive warhead. Calpain was chosen as
a challenging target for the design of selective inhibitors,[7]
and because its link to cortical cataracts[8] provides an
opportunity to assess the in vivo efficacy of our approach.
The acyclic tripeptide analogues 5a–d were also investigated
to establish the importance (or unimportance) of the macro-
cyclic constraint for potency and selectivity of inhibition.
Conformational searches[9] were carried out on 1a–d to
5a–d to assess their ability to adopt a b-strand conformation.
The resulting ensembles of low-energy conformers were
examined by XCluster (see the Supporting Information for
detailed results). For the 16-membered macrocycles 1a–d, all
conformers (> 99%) within 12 kJmolÀ1 of their global
minima adopt a b-strand conformation (see Figure 1 for
[*] Prof. A. D. Abell,[+] Dr. M. A. Jones, Prof. J. M. Coxon, Dr. S. G. Aitken,
Dr. S. B. McNabb, Dr. J. M. Mehrtens, Dr. N. A. Alexander,
B. G. Stuart, Dr. A. T. Neffe[#]
Department of Chemistry, University of Canterbury
Private Bag 4800, Christchurch (New Zealand)
Dr. J. D. Morton, Dr. H. Y.-Y. Lee, Prof. R. Bickerstaffe
Agriculture and Life Sciences Division, Lincoln University
Post Office Box 84, Canterbury (New Zealand)
Fax: (+64)3-325-3851
[+] Present address: School of Chemistry & Physics
The University of Adelaide
North Terrace, Adelaide, SA 5005 (Australia)
Fax: (+61)8-8303-4358
E-mail: andrew.abell@adelaide.edu.au
[#] Present address: Institute of Polymer Research
GKSS Research, Centre Geesthacht GmbH
Kantstrasse 55, 14513 Teltow (Germany)
[**] We acknowledge the assistance of Dr. D. Q. McDonald (computa-
tional chemistry), Dr. A. Muscroft-Taylor (synthesis), and M. Muir
(supply of calpain), and financial support from the New Zealand
Public Good Science and Technology Fund, Foundation for
Research Science and Technology, the Australian Research Council,
and Douglas Pharmaceuticals Limited.
Supporting information (confirmation of the structure and purity of
all compounds, details on modeling studies, and biological assay
data) for this article is available on the WWW under http://dx.doi.
org/10.1002/anie.200805014.
Figure 1. Overlaid conformers of the two clusters of 1d.
Angew. Chem. Int. Ed. 2009, 48, 1455 –1458
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1455

Communications
1d). These macrocycles are essentially rigid with little
apparent influence from the C and N termini. The 17- and
18-membered macrocycles 2a–d and 3a–d each have at least
one cluster of b-strand peptide-backbone conformers. Thus,
their backbone conformations are more flexible with some
influence from the C- and N-terminal groups. The flexibility
of the 19-membered rings in 4a–d does not constrain these
macrocycles into a b-strand conformation. The acyclic
analogues 5a–c do not exhibit b-strand conformers. However,
aldehyde 5d has one minor cluster of b-strand peptide-
backbone conformers.
On the basis of this established ability to adopt a b-strand
conformation, we predicted the order of potency of a
macrocyclic inhibitor to be: 16-membered 1 > 17-membered
2 ꢀ 18-membered 3 > 19-membered 4. Each of the macro-
cyclic aldehydes 1d–4d was docked in silico into ovine
calpain 1 and
2 (o-CAPN1 and o-CAPN2) homology
models[10] by using Glide.[11] The results of these docking
studies (see the Supporting Information) suggest that 1d–4d
bind in the same b-strand conformation, with similar hydro-
gen-bonding patterns and an appropriate positioning of the
aldehyde group near the nucleophilic sulfur atom in the active
site (Figure 2). This mode of binding is in agreement with that
observed in the crystal structures of leupeptin (a naturally
occurring inhibitor) and SNJ1715 cocrystallized in rat cal-
pain 1 (r-CAPN1). On the basis of these results, we targeted
macrocycles 1–4 for synthesis and biological evaluation.
Figure 2. A representative pose of macrocycle 2d in the o-CAPN1
model.
Scheme 2. Synthesis of 1a–4a: a) K2CO3, 4-bromobut-1-ene, DMF
(27%); b) NaOH, THF, H2O, MeOH, (85%); c) HATU, DIPEA, Leu-
OMe, DMF (10: 64%; 12: 86%); d) NaOH, THF, H2O, MeOH (11:
100%, 13: 97%); e) HATU, DIPEA, (S)-vinyl-Gly-OMe, DMF, 13 (14:
83%), or HATU, DIPEA, (S)-allyl-Gly-OMe, DMF, 13 (15: 97%), or
HATU, DIPEA, (S)-allyl-Gly-OMe, DMF, 11 (16: 84%), or HATU,
DIPEA, (S)-Cys(S-allyl)-OMe, DMF, 13 (17: 83%); f) Grubbs second-
generation catalyst, 1,1,2-TCE, D, then H2, Pd-C, MeOH (1a: 24%, 2a:
22%, 3a: 29%, 4a: 66%); g) 4m HCl, 1,4-dioxane (quantitative for
1a–4a), then benzyl chloroformate, DIPEA, DMF, (1b: 53%, 2b: 49%,
3b: 79%, 4b: 77%); h) LiAlH4, THF (1c: 83%, 2c: 83%, 3c: 72%,
4c: 70%); i) SO3·pyridine, DIPEA, DMSO, CH2Cl2 (1d: 92%, 2d: 80%,
3d: 90%, 4d: 61%). DIPEA=diisopropylethylamine, DMF=N,N-
dimethylformamide, DMSO=dimethyl sulfoxide, HATU=O-(7-azaben-
zotriazol-1-yl)-N,N,N’,N’-tetramethyluronium hexafluorophosphate,
1,1,2-TCE=1,1,2-trichloroethane
The synthesis of macrocycles 1a–4a was based on the
RCM of dienes 14–17, which were prepared from natural
building blocks (Scheme 2). The reaction of commercially
available N-Boc-Tyr-OMe (6) with 4-bromobut-1-ene gave
the homoallylic ether 7, which was hydrolyzed to give 8.
Separate reactions of 8 and 9 (commercially available) with
Leu-OMe in the presence of HATU gave dipeptides 10 and
12, which were hydrolyzed to give 11 and 13, respectively.
Dienes 14, 15, and 17 were prepared by coupling 13 with (S)-
vinyl-Gly-OMe,[12] (S)-allyl-Gly-OMe, and (S)-Cys(S-allyl)-
OMe, respectively. Diene 16 was similarly prepared from 11
by treatment with Cys(S-allyl)-OMe. RCM of the acyclic
dienes 14–17 gave mixtures of cis and trans alkenes, which
were hydrogenated directly to give macrocycles 1a–4a.
1456
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 1455 –1458

Angewandte
Chemie
The separate treatment of 1a–4a with HCl/dioxane gave
the corresponding hydrochloride salts in quantitative yield.
These salts were treated, without purification, with benzyl
chloroformate to give esters 1b–4b. Reduction of 1b–4b with
lithium aluminium hydride gave the alcohols 1c–4c, which
were oxidized with SO3·pyridine and DSMO[13] to give
aldehydes 1d–4d (Scheme 2).
The aldehydes 1d–4d and alcohols 1c–4c were assayed
against ovine calpain 1 (o-CAPN1) and ovine calpain 2
(o-CAPN2)[14] to determine in vitro potency and isoform
selectivity, and to assess the potential of the constituent
macrocycle to enable noncovalent inhibition in the absence of
an aldehyde (Table 1). Modeling studies suggested that the
alcohol group could interact with the oxyanion hole defined
aldehyde group, is an inhibitor of o-CAPN2 (IC50 = 700 nm).
Again, the 17-membered ring is optimum for activity: 1c, 3c,
and 4c are all significantly less potent (see Table 1). The 16-
and 19-membered macrocyclic structures (1c and 4c) are
almost devoid of activity, as is the acyclic analogue 5c. Thus,
the introduction of an appropriate macrocycle provides a new
class of noncovalent inhibitors of calpains and other cysteine
proteases (see below).
The macrocycles 2c,d and the acyclic tripeptides 5c,d
were also assayed against cathepsin B (a cysteine protease
from the same clan as calpain),[16] papain, pepsin, and a-
chymotrypsin.[17] All were inactive against papain, pepsin, and
a-chymotrypsin in the concentration range of the assays (up
to 50 mm). However, the 17-membered macrocyclic aldehyde
2d (IC50 = 70 nm) and the analogous acyclic aldehyde 5d
(IC50 = 5 nm) are both potent inhibitors of cathepsin B. The
macrocyclic alcohol 2c (IC50 = 300 nm) and the acyclic
analogue 5c (IC50 = 200 nm) are both remarkably potent for
noncovalent inhibitors.
Table 1: In vitro inhibition data.
Docking studies with the o-CAPN1 and o-CAPN2
homology models were carried out with the macrocyclic
alcohols 2c and 3c, which lack a reactive aldehyde but display
activity against o-CAPN1 and o-CAPN2. A representative
pose of each alcohol (see the Supporting Information) shows
that the b-strand peptide-backbone conformations are ori-
entated similarly with respect to the enzyme. In each case, the
methylene carbon atom of the primary alcohol is positioned
close to the sulfur atom of the active-site cysteine residue (at a
distance of 3.1 and 4.5 ꢀ for 2c and 3c, respectively). This
structural feature is analogous to that observed in the crystal
structure of the hemiacetal formed when SNJ1715 was
cocrystallized with r-CAPN1.[15]
We next chose to investigate the potential of our most
potent o-CAPN2 inhibitor, 2d, to retard the development of
calcium-induced cortical cataracts, which have been linked to
an overactivity of this enzyme.[8] Lenses from 9–12-month-old
lambs were incubated for 48 h in EMEM (Eagle Minimum
Essential Medium) culture medium (10 mL) at 378C in 5%
CO2. Inhibitor 2d (1 mm) was added to one lens of each of six
pairs of sheep lenses in the culture medium. After incubation
for 3 h, CaCl2 was added to a final concentration of 5 mm.
Intact ovine lenses (n = 6) were also incubated in EMEM
culture medium as a control. After 6 h, all lenses were
photographed over a grid, and the opacity was graded by
using the software Image-Pro 4.1. Lenses treated with only
calcium showed substantial opacity as associated with cata-
ract formation. The presence of 2d prevented this calcium-
induced opacification, and these lenses remained essentially
transparent after incubation for 6 h (Figure 3). Thus, the
calpain inhibitor 2d in the culture medium was able to
significantly reduce lens opacity (p < 0.005 in a paired t test).
Finally, we were interested in correlating potency with the
ability of each compound (on the basis of the earlier modeling
studies) to adopt a b-strand conformation (16-membered 1 >
17-membered 2 ꢀ 18-membered 3 > 19-membered 4). The 17-
and 18-membered macrocyclic aldehydes 2d and 3d are more
potent than the 16-membered macrocycle 1d (see Table 1),
which was shown by modeling to be particularly rigid. We
suggest that the rigid macrocycle 1d does not have the
Compound
R2
X
n
IC50[a] [nm]
o-CAPN1
o-CAPN2
1c
2c
3c
4c
5c
1d
2d
3d
4d
5d
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
CHO
CHO
CHO
CHO
CHO
CH2
CH2
CH2
S
1
2
3
4
–
1
2
3
4
–
13000
1750
1340
50000
3200
400
220
170
3150
50
31000
700
1100
28000
15600
850
–
CH2
CH2
CH2
S
30
180
1010
130
–
[a] Values are the mean of three experiments. Variation between
experiments is less than Æ10%.
by the side chain of Gln109/99 and the backbone amide
nitrogen atom of Cys115/105 in the ovine calpains.[15]
The 17-membered macrocyclic aldehyde 2d is a partic-
ularly potent inhibitor of o-CAPN2 (IC50 = 30 nm) with
greater than sevenfold selectivity for o-CAPN2 over
o-CAPN1 (IC50 = 220 nm for o-CAPN1). The 18-membered
macrocycle 3d is approximately equally potent against both
proteases (IC50 = 170 and 180 nm for o-CAPN1 and
o-CAPN2, respectively). The 16-membered macrocycle 1d
is
less
potent
(IC50 = 400
and
850 nm
for
o-CAPN1 and o-CAPN2, respectively) with greater than
twofold selectivity for, in this case, o-CAPN1 over
o-CAPN2. The incorporation of a sulfur atom into the 19-
membered macrocycle (in 4d) resulted in significantly
reduced potency with some selectivity for o-CAPN2. By
comparison, the acyclic tripeptide aldehyde 5d is a potent
calpain inhibitor (IC50 = 50 and 130 nm for o-CAPN1 and
o-CAPN2, respectively) with > 2.5-fold selectivity for
o-CAPN1 over o-CAPN2.
Of particular significance is the observation that the 17-
membered macrocyclic alcohol 2c, which lacks the reactive
Angew. Chem. Int. Ed. 2009, 48, 1455 –1458
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1457

Communications
[1] D. Lamarre, P. C. Anderson, M. Bailey, P. Beaulieu, G. Bolger, P.
Bonneau, M. Bꢁs, D. R. Cameron, M. Cartier, M. G. Cordingley,
A.-M. Faucher, N. Goudreau, S. H. Kawai, G. Kukolj, L. Lagacꢂ,
S. R. LaPlante, H. Narjes, M.-A. Poupart, J. Rancourt, R. E.
Sentjens, R. St George, B. Simoneau, G. Steinmann, D. Thi-
beault, Y. S. Tsantrizos, S. M. Weldon, C.-L. Yong, M. Llinꢃs-
Brunet, Nature 2003, 426, 186 – 189.
[2] Z.-D. Shi, K. Lee, H. Liu, M. Zhang, L. R. Roberts, K. M.
Worthy, M. J. Fivash, R. J. Fisher, D. Yang, T. R Burke, Jr.,
Biochem. Biophys. Res. Commun. 2003, 310, 378 – 383.
[3] R. J. Cherney, L. Wang, D. T. Meyer, C.-B. Xue, Z. Wasserman,
K. D. Hardman, P. K. Welch, M. B. Covington, R. A. Copeland,
E. C. Arner, W. F. DeGrado, C. P. Decicco, J. Med. Chem. 1998,
41, 1749 – 1751.
Figure 3. Effect of 2d on calcium-induced cataract formation in
cultured ovine lenses. The values given are the mean opacification
grading score (ÆSEM: standard error of the mean) in each case and
represent the mean opacity of six lenses. Fully opaque lenses have
scores of 80–100, whereas transparent lenses have scores of less than
20. a) Control: (16.0Æ2.1); b) Ca2+ only: (58.8Æ6.7); c) Ca2+ with 2d:
(16.7Æ1.5).
[4] R. C. Reid, M. J. Kelso, M. J. Scanlon, D. P. Fairlie, J. Am. Chem.
Soc. 2002, 124, 5673 – 5683.
[5] S. E. Gibson, C. Lecci, Angew. Chem. 2006, 118, 1392 – 1405;
Angew. Chem. Int. Ed. 2006, 45, 1364 – 1377; E. M. Driggers, S. P.
Hale, J. Lee, N. K. Terrett, Nat. Rev. Drug Discovery 2008, 7,
608 – 624.
[6] J. D. A. Tyndall, T. Nall, D. P. Fairlie, Chem. Rev. 2005, 105, 973 –
999.
[7] M. A. Jones, A. D. Abell, J. D. Morton, J. M. Coxon, S. B.
McNabb, H. Y.-Y Lee, S. G. Aitken, J. M. Mehrtens, L. J. G.
Robertson, A. T. Neffe, S. Miyamoto, R. Bickerstaffe, K. Gately,
J. M. Wood, Bioorg. Med. Chem. 2008, 16, 6911 – 6923.
[8] L. J. G. Robertson, J. D. Morton, J. M. Yamaguchi, R. Bicker-
staffe, T. R. Shearer, M. Azuma, Invest. Ophthalmol. Vis. Sci.
2005, 46, 4634 – 4640.
flexibility necessary for optimum binding, despite its favored
b-strand peptide-backbone conformation. By contrast, the
slightly more flexible 17-membered macrocycle (and to a
lesser extent the 18-membered macrocycle) combine an
ability to adopt a b-strand conformation with an opportunity
to optimize binding. The potent and selective o-CAPN2
inhibitor 2d has potential for the treatment of cortical
cataracts. The flexibility of the 19-membered ring in 4d
does not constrain the macrocycle in a b-strand conformation.
This compound is only moderately potent despite its docking
in both the o-CAPN 1 and o-CAPN2 models in a b-strand
conformation.
In conclusion, a combination of modeling, synthesis, and
biological evaluation of 1c–4c and 1d–4d has identified
versatile macrocyclic templates that exhibit a varying and
predictable propensity to adopt a b-strand backbone con-
formation. Potency against calpain can be correlated to this
conformational preference as well as the flexibility of the
macrocycle. We have presented a versatile synthetic route to
these templates, with an opportunity to prepare a range of
macrocycles and protease inhibitors through modification and
extension at the two termini and of the constituent peptide
sequence of the diene for RCM.
[9] MacroModel, version 9.1, Schrꢁdinger, LLC, New York, NY,
2005.
[10] Preliminary studies were carried out with an r-CAPN1 construct
(T. Moldoveanu, C. M. Hosfield, D. Lim, J. S. Elce, Z. Jia, P. L.
Davies, Cell 2002, 108, 649 – 660). o-CAPN1 and o-CAPN2
homology models[7] were developed from the X-ray crystal
structure of human calpain 1, 1ZCM, by mutations in the active
site to mimic the sequence of o-CAPN1 (Genbank accession
number EU623071) or o-CAPN2 (Genbank accession number
EU161096).
[11] Glide, version 4.0, Schrꢁdinger, LLC, New York, NY, 2005.
[12] A. Krebs, V. Ludwig, J. Pfizer, G. Dꢄrner, M. W. Gꢁbel, Chem.
Eur. J. 2004, 10, 544 – 553.
[13] J. R. Parikh, W. v. E. Doering, J. Am. Chem. Soc. 1967, 89, 5505 –
5507.
[14] V. F. Thompson, S. Saldana, J. Cong, D. E. Goll, Anal. Biochem.
2000, 279, 170 – 178.
[15] D. Cuerrier, T. Moldoveanu, J. Inoue, P. L. Davies, R. L.
Campbell, Biochemistry 2006, 45, 7446 – 7452.
[16] R. Vicik, M. Busemann, K. Baumann, T. Schrimeister, Curr. Top.
Med. Chem. 2006, 6, 331 – 353.
[17] H. Y.-Y. Lee, J. D. Morton, L. J. G. Robertson, J. McDermott, R.
Bickerstaffe, A. Abell, M. A. Jones, J. M. Mehrtens, J. M. Coxon,
Clin. Experiment. Ophthalmol., in press.
Aldehyde 2d (CAT811)[18] proved to be the most potent
and selective o-CAPN2 inhibitor (Table 1). The alcohol 2c
(CAT505),[18] although less potent, is a unique example of an
o-CAPN2-selective noncovalent inhibitor. Aldehyde 2d sig-
nificantly retarded calcium-induced opacification in vitro in
an ovine-lens culture assay and shows promise in slowing the
progression of cortical cataracts in animal trials with a
hereditary-ovine-cataract model. The animal studies will be
reported separately.
[18] A. D. Abell, J. M. Coxon, M. A. Jones, S. G. Aitken, B. G. Stuart,
A. T. Neffe, J. M. Nikkel, S. B. McNabb, M. Klanchantra, J. K.
Duncan, J. D. Morton, R. Bickerstaffe, L. J. G. Robertson, H. Y.-
Y. Lee, M. S. Muir, PCT Int. Appl., WO 2008048121, 2008.
Received: October 14, 2008
Published online: January 14, 2009
Keywords: biological activity · inhibitors · macrocycles ·
.
molecular modeling · peptides
1458
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 1455 –1458