A template-based approach to inhibitors of calpain 2, 20S proteasome, and HIV-1 protease

Article abstract of DOI:10.1002/cmdc.201300387

Specificity counts: A template-based approach to protease inhibitors is presented using a core macrocycle that presents a generic β-strand template for binding to protease active sites. This is then specifically functionalized at P₂, and the C and N termini to give inhibitors of calpain 2, 20S proteasome, and HIV-1 protease. Copyright

Full text of DOI:10.1002/cmdc.201300387

CHEMMEDCHEM
COMMUNICATIONS
DOI: 10.1002/cmdc.201300387
A Template-Based Approach to Inhibitors of Calpain 2, 20S
Proteasome, and HIV-1 Protease
Seth A. Jones,[a] Paul M. Neilsen,[b] Limei Siew,[b] David F. Callen,[b] Nathan E. Goldfarb,[c]
Ben M. Dunn,[c] and Andrew D. Abell*[a]
A number of protease classes are known, each with a character-
istic mechanism by which member enzymes hydrolyze peptide
substrates. Despite these distinctive mechanistic differences,
protease inhibitors almost universally adopt a b-strand confor-
mation upon active site binding.[1,2] This mode of binding is
primarily dictated by the geometry of the active site subsites
(designated S1–Sn and S1’ to Sn’ according to the nomenclature
of Schecter and Burger)[3] that accommodate the amino acid
side chains of natural peptide substrates that adopt the same
b-strand geometry on binding and hydrolysis.[1]
Macrocyclic inhibitors exploit this by pre-organizing the
structure into a b-strand conformation, thereby decreasing en-
tropy loss associated with ligand–receptor binding while also
enhancing biostability.[1,2] Typically this involves chemically link-
ing the P1 and P3 residues of a peptidomimetic-based protease
inhibitor.[1,2] This approach has driven the development of mac-
rocyclic inhibitors with enhanced potency against various pro-
Figure 1. Examples of macrocyclic protease inhibitors.
teases. For example, compound 1 (Figure 1) is a potent inhibi-
tor of calpain 2 that shows promise in the treatment of cal-
pain-induced cataract formation.[2] As further examples, 2 is
a potent inhibitor of cathepsin S,[4] 3 is a naturally occurring
ACE inhibitor,[5] and 4 is a highly potent macrocyclic inhibitor
of HIV protease,[6] as are a number of other isostere-containing
macrocycles.[1] However, a major drawback of current studies is
that each inhibitor requires the synthesis of a separate and
unique core b-strand structure, with the appropriate function-
ality attached at the N and C termini.
There would be significant benefit in developing generic
macrocyclic templates suitable for elaboration into a range of
inhibitors of different classes of proteases.[7] Herein we describe
the development of one such macrocyclic b-strand template
(see 5 in Figure 2) that can be elaborated into inhibitors of cal-
pain 2 (compound 1), 20S proteasome (6), and HIV protease (7
and 8). These proteases present three different mechanistic
classes that are all of significant therapeutic interest. Calpain is
a ubiquitous cysteine protease, the over-activation of which is
[a] Dr. S. A. Jones, Prof. A. D. Abell
School Chemistry and Physics
The University of Adelaide, Adelaide, SA 5005 (Australia)
E-mail: andrew.abell@adelaide.edu.au
[b] Dr. P. M. Neilsen, L. Siew, Prof. D. F. Callen
Centre for Personalised Cancer Medicine
The University of Adelaide, Adelaide, SA 5005 (Australia)
[c] Dr. N. E. Goldfarb, Prof. B. M. Dunn
Department of Biochemistry and Molecular Biology
University of Florida, Gainesville, FL 32610-0245 (USA)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201300387.
Figure 2. The macrocyclic template 5 and protease inhibitors derived from
it.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1918 – 1921 1918

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
implicated in a variety of human conditions including cataracts
and traumatic brain injury.[8] The proteasome is a complex pro-
tease, inhibitors of which are at the forefront of treatments for
multiple myeloma.[9] This protease contains three separate ac-
tivities: chymotrypsin-like, trypsin-like, and caspase-like. We
focus on the chymotrypsin-like activity, as it plays an important
role[10] in malignancies and is a target of current anticancer
therapeutic strategies.[11] Interestingly, reactive inhibitors of cal-
pain invariably inhibit the proteasome, so selectivity amongst
these proteases is a major ongoing challenge that is also ad-
dressed here.[12] The third protease, HIV protease, is an impor-
tant target for current therapies of HIV-positive patients and is
considered a benchmark aspartic acid protease.[1,13]
The core macrocycle of 5 (Figure 2) has a number of key
structural features that make it an ideal candidate as a general
template for elaboration into a number of different protease
inhibitors.[1,2] The component 17-membered macrocycle, with
its alkylated tyrosine moiety, accurately defines a b-strand con-
formation to the peptide backbone, as is required for active
site binding. This is evidenced by a conformational search (and
a related X-ray structure) on this ring system that reveals
a strong preference for a b-strand-based conformation.[2a] Im-
portantly, the ring is not overly rigid and it has some flexibility
that we postulate as being important to allow optimum bind-
ing to a range of proteases. Second, the template provides an
opportunity to incorporate the macrocycle into either the P1–
P3 (see 1 and 6) or P1’–P3’ positions (see 7 and 8). Third, its syn-
thesis by ring-closing metathesis is straightforward,[2] with an
opportunity to subsequently attach additional functionality at
either the C or N termini in order to target a specific protease.
For example, a warhead can be attached at the C terminus
(see 1 and 6) or an isostere to the N terminus (see 7 and 8).
Fourth, the introduction of a macrocycle decreases the pepti-
dic character of the inhibitor to enhance metabolic stability.[1,13]
Finally, the tripeptide backbone of the template is able to form
critical hydrogen bonds with a protease active site, to further
stabilize binding.[14]
Scheme 1. Synthesis of 5b. Reagents and conditions: a) Grubbs’ 2nd-genera-
tion catalyst, CH2Cl2, reflux, 3 h, (79%); b) H2, Pd/C, CH2Cl2, MeOH, RT, 16 h,
(94%); c) SOCl2, MeOH, 08C!RT, 17 h, (quant).
Our synthesis of the key macrocyclic templates 5a and 5b
involved ring-closing metathesis of an orthogonally protected
tripeptide-based diene as depicted in Scheme 1 for 5b and as
reported for the preparation of 1.[2a] The alkene of 10 was hy-
drogenated to give 11, the N-Boc group of which was re-
moved on treatment with SOCl2 and methanol to give 5b.
These conditions represent an improvement on related pub-
lished methodology (treatment with 4m HCl in 1,4-dioxane)
that can result in competing ester hydrolysis.[15] See the Sup-
porting Information for full details. The analogue 5a was simi-
larly prepared starting with the corresponding P2 Leu-based tri-
peptide.[2a]
Scheme 2. Synthesis of protease inhibitors. Reagents and conditions:
a) benzyl chloroformate, DIPEA, DMF (69%); b) LiAlH4, THF (84%);
c) SO3·pyridine, DMSO, DIPEA, CH2Cl2 (1, 80%; 6, 46%); d) 14 or 15, 4 ꢁ mo-
lecular sieves, MeOH, NaOAc, Na(OAc)3BH (7: 18%, 16: 28%); e) PTSA,
MeOH, 458C (4%). DIPEA=N,N-diisopropylethylamine, DMF=N,N-dimethyl-
formamide, PTSA=para-toluenesulfonic acid, THF=tetrahydrofuran.
The preparation of the protease inhibitors 1 and 6–8 from
5a and 5b then proceeded as shown in Scheme 2. The macro-
cyclic aldehydes (1 and 6) were prepared by first attaching
a benzyloxycarbonyl (Cbz) protecting group to the N termini
of both 5a and 5b to give 12a and 12b, respectively. The
methyl esters of these were separately reduced to the give the
primary alcohols 13a and 13b, which were oxidized to the
corresponding aldehydes under Parikh–Doering[16] conditions
as shown in Scheme 2. The HIV protease inhibitors 7 and 8
were also prepared from 5b, with the key step being a reduc-
tive amination with an appropriate aldehyde. Ile was chosen at
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1918 – 1921 1919

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
P2 because this substituent is known to be favored in inhibitors
of the HIV protease.[8,17] Here, treatment of a mixture of 5b
and 14 with NaOAc and Na(OAc)3BH gave the reduced amide
isostere 7 in an unoptimized yield of 18%. A similar reductive
amination of the aldehyde 15 with the amine 5b gave 16, the
oxizolidone of which was hydrolyzed on treatment with para-
toluenesulfonic acid in methanol to give the hydroxyethyl-
amine isostere-based inhibitor 8. The acyclic diene-based ana-
logue of the macrocycle 7 (see 17 in Scheme 2) was prepared
by a similar reductive amination strategy as detailed in the
Supporting Information.
Table 2. In vitro activity of macrocycles against HIV-1 B protease and
XMRV protease.
Compd
HIV-1 B protease
XMRV protease
Inhib. [%][a]
Ki [nm][b]
Inhib. [%][a]
Ki [nm][b]
7
8
17
86
63
26
98
80
ND[c]
16
74
ND[c]
ND[c]
ND[c]
ND[c]
[a] Percentage inhibition of protease at 1000 nm (representative value).
[b] Based on two biological replicates, SD<5%. [c] Not determined.
The macrocyclic aldehydes 1 and 6 and the corresponding
alcohols (13a and 13b) were assayed against ovine calpain 2
and the chymotrypsin-like activity of the 20S proteasome; the
results are shown in Table 1. Potent inhibitors of both calpain 2
cle 7 was particularly potent, with a determined Ki value of
98 nm. This macrocycle was also a potent inhibitor (Table 2) of
the related aspartic acid protease, XMRV protease, which has
attracted considerable attention recently.[18]
The final aspect of the study involved assessing the ability of
our new macrocyclic proteasome inhibitors (6, 1, and their al-
cohol derivatives) to induce apoptosis or necrosis in three
cancer cell lines, given the known role of the proteasome in
these malignancies.[19] The corresponding IC50 values for cell
death are listed in Table 3 with comparative data given for
human embryonic fibroblasts, which are used as a normal non-
malignant cell line.
Table 1. In vitro activity of key macrocycles against calpain 2 and the 20S
proteasome.
Compd
P2
R
IC50 [nm][a]
20S Proteasome[b]
Calpain 2
6
13b
1
Ile
Ile
Leu
Leu
CHO
CH2OH
CHO
5000
7000
100[c]
700
38
886
1422
inactive[d]
Table 3. IC50 values for cell viability assays on MDA-MB-468, WE-68,
HCT116 and human embryonic fibroblasts.
Compd
IC50 [mm][a]
HCT116[d]
13a
CH2OH
MDA-MB-468[b]
WE-68[c]
Human embryonic
fibroblasts
[a] Based on three biological replicates, SD<5%. [b] IC50 is for the chymo-
trypsin-like activity of 20S proteasome. [c] A value of 30 nm has been re-
ported.[1a] [d]>25000 nm.
6
13b
1
12
62
>80
>80
11
>80
>80
>80
19
>80
57
>80
>80
>80
>80
13a
>80
(compound 1) and the proteasome (see 6) are apparent, with
an unexpected subtle role played by the P2 residue in defining
the specificity of inhibition. In particular, the macrocyclic alde-
hyde 6, with Ile at P2, shows significant selectivity for the chy-
motrypsin-like activity of 20S proteasome over calpain 2 (130-
fold), whereas 1, which differs only in having Leu at P2, has the
opposite selectivity and is a potent inhibitor of calpain 2
(Table 1). The effect of the P2 group on selectivity is mirrored in
the results obtained for the corresponding alcohols 13, where
Ile (13b) favors the proteasome and Leu (13a) calpain 2. Thus
the subtle choice of the P2 amino acid in our macrocycles de-
fines the selectivity profile against these two proteases. Impor-
tantly, acyclic peptidic aldehydes invariably lack selectivity be-
tween calpain and the proteasome.[12]
[a] Based on two biological replicates, SD<5%. [b] Breast cancer. [c] Sar-
coma. [d] Colon cancer.
The most potent proteasome inhibitor (macrocyclic alde-
hyde 6) displayed significant activity against the MDA-MB-468,
WE-68, and HCT116 cancer cell lines (IC50: 12, 11, and 19 mm, re-
spectively), whilst importantly being nontoxic to normal
human embryonic fibroblasts. By contrast, the less potent pro-
teasome inhibitor 1 is only weakly active against the HCT116
cell line (IC50 =57 mm). The less active macrocyclic alcohols 13
were devoid of anticancer activity, except for the reasonable
activity displayed by 13b against MDA-MB-468. Consistent
with this observation, 13b does show in vitro activity against
the chymotrypsin-like activity of 20S proteasome (Table 1). We
speculate that the reduced conformational flexibility displayed
by the macrocyclic aldehydes may decrease nonspecific cyto-
toxicity by minimizing nonspecific biological interactions, as
these compounds showed activity against cancer cell lines in
the absence of cytotoxicity against normal human fibroblasts.
This contrasts results for the benchmark proteasome inhibitors
bortezomib and the acyclic peptidic aldehyde MG132 that
The isostere-containing macrocycles 7 and 8, and also 17
(the acyclic analogue of 7) were assayed against HIV-1 B pro-
tease, the major strain of HIV in the Americas, East Asia, Ocean-
ia, and Western Europe; the results are listed in Table 2. A pre-
liminary screen, based on a percentage inhibition at a fixed
concentration, identified the macrocyclic reduced amide iso-
stere 7 as the most potent inhibitor. The macrocyclic core of
this structure clearly enhances activity, with the acyclic reduced
amide isostere 17 being significantly less active. The macrocy-
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1918 – 1921 1920

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
Chem. Soc. 1996, 118, 3375–3379; c) J. D. A. Tyndall, T. Nall, D. P. Fairlie,
Chem. Rev. 2005, 105, 973–999.
have remarkably limited specificity for cancer cells over normal
cells.[20]
[2] a) A. D. Abell, M. A. Jones, J. M. Coxon, J. D. Morton, S. G. Aitken, S. B.
McNabb, H. Y.-Y. Lee, J. M. Mehrtens, N. A. Alexander, B. G. Stuart, A. T.
Neffe, R. Bickerstaffe, Angew. Chem. 2009, 121, 1483–1486; Angew.
Chem. Int. Ed. 2009, 48, 1455–1458; b) A. Pehere, A. D. Abell, Org. Lett.
2012, 14, 1330–1333; c) A. D. Abell, N. A. Alexander, S. G. Aitken, H.
Chen, J. M. Coxon, M. A. Jones, S. B. McNabb, A. Muscroft-Taylor, J. Org.
Chem. 2009, 74, 4354–4356.
[3] I. Schechter, A. Berger, Biochem. Biophys. Res. Commun. 1967, 27, 157–
162.
[4] A. D. Pehere, C. J. Sumby, A. D. Abell, Org. Biomol. Chem. 2013, 11, 425–
429.
[5] a) E. Marsault, M. L. Peterson, J. Med. Chem. 2011, 54, 1961–2004; b) H.
Kase, M. Kaneko, K. Yamada, J. Antibiot. 1987, 40, 450–454.
[6] A. K. Ghosh, S. Kulkarni, D. D. Anderson, L. Hong, A. Baldridge, Y.-F.
Wang, A. A. Chumanevich, A. Y. Kovalevsky, Y. Tojo, M. Amano, Y. Koh, J.
Tang, I. T. Weber, H. J. Mitsuya, J. Med. Chem. 2009, 52, 7689–7705.
[7] M. P. Glenn, L. K. Pattenden, R. C. Reid, D. P. Tyssen, J. D. A. Tyndall, C. J.
Birch, D. P. Fairlie, J. Med. Chem. 2002, 45, 371–381.
[8] a) Y. Huang, K. K. W. Wang, Trends Mol. Med. 2001, 7, 355–362; b) M.
Pietsch, K. C. H. Krystle, A. D. Abell, Curr. Top. Med. Chem. 2010, 10, 270–
293.
[9] A. M. Ruschak, M. Slassi, L. E. Kay, A. D. Schimmer, J. Natl. Cancer Inst.
2011, 103, 1007–1017.
[10] A. F. Kisselev, A. L. Goldberg, Chem. Biol. 2001, 8, 739–758.
[11] D. Hempel, M. Z. Wojtukiewicz, L. Kozłowski, J. Romatowski, H. Ostrow-
ska, Tumour Biol. 2011, 32, 753–759.
Herein we present a template-based approach to protease
inhibitors based on a core macrocycle 5 constrained into a
b-strand geometry. This core defines a suitable geometry for
active site binding, and an ability to introduce (template) ap-
propriate groups at both the C and N termini to allow the tar-
geting of different proteases. Potent and selective inhibitors of
both calpain 2 and the chymotrypsin-like activity of 20S pro-
teasome were derived from the template by attaching an alde-
hyde warhead to the C terminus and through the subtle
choice of P2 residue. This second point is an important obser-
vation for inhibitors of these two classes of protease, which
are invariably both inhibited by peptidic aldehydes without
a great deal of selectivity.[12] The most potent 20S proteasome
inhibitor 6 was also shown to have selective activity for cancer
cells relative to normal human fibroblasts. A potent inhibitor of
the HIV-1 B protease was also prepared from the template,
with the attachment of a reduced amide isostere to the N ter-
minus. The corresponding acyclic analogue was significantly
less active, demonstrating the importance of the core macrocy-
cle. We propose macrocyclic cores of type 5 as versatile tem-
plates for the preparation of inhibitors of other proteases by
attaching specific functionality at either the N or C termini. A
number of synthetic methods are now available for introduc-
ing a macrocycle into a peptide to further expand a template-
based approach to inhibitor design.[1–3]
[12] G. de Bettignies, O. Coux, Biochimie 2010, 92, 1530–1545.
[13] Advances in Peptidomimetics and Amino Acid Mimics, Vol. 1 (Ed.: A. D.
Abell), JAI Press, London, 1997, p. 302.
[14] B. G. Stuart, J. M. Coxon, J. D. Morton, A. D. Abell, D. Q. McDonald, S. G.
Aitken, M. A. Jones, R. Bickerstaffe, J. Med. Chem. 2011, 54, 7503–7522.
[15] S. A. Jones, PhD Thesis, University of Adelaide, 2010.
[16] J. R. Parikh, W. v. E. Doering, J. Am. Chem. Soc. 1967, 89, 5505–5507.
[17] P. L. Darke, R. F. Nutt, S. F. Brady, V. M. Garsky, T. M. Ciccarone, C.-T. Leu,
P. K. Lumma, R. M. Freidinger, D. F. Veber, I. S. Sigal, Biochem. Biophys.
Res. Commun. 1988, 156, 297–303.
Acknowledgements
[18] M. Li, F. DiMaio, D. W. Zhou, A. Gustchina, J. Lubkowski, Z. Dauter, D.
Baker, A. Wlodawer, Nat. Struct. Mol. Biol. 2011, 18, 227–229.
[19] a) Y. S. Hong, S. W. Hong, S. M. Kim, D. H. Jin, J. S. Shin, D. H. Yoon, K. P.
Kim, J. L. Lee, D. S. Heo, J. S. Lee, T. W. Kim, Int. J. Oncol. 2012, 41, 76–
82; b) T. Nakamura, K. Tanaka, T. Matsunobu, T. Okada, F. Nakatani, R. Sa-
kimura, M. Hanada, Y. Iwamoto, Int. J. Oncol. 2007, 31, 803–811.
[20] P. M. Neilsen, A. D. Pehere, K. I. Pishas, D. F. Callen, A. D. Abell, ACS
Chem. Biol. 2013, 8, 353–359.
Financial support from the Australian Research Council (ARC) is
gratefully acknowledged.
Keywords: inhibitors
templates · b-strands
·
peptidomimetics
·
proteases
·
[1] a) J. D. A. Tyndall, R. C. Reid, D. P. Tyssen, D. K. Jardine, B. Todd, M. Pass-
more, D. R. March, L. K. Pattenden, D. A. Bergman, D. Alewood, S. H. Hu,
P. F. Alewood, C. J. Birch, J. L. Martin, D. P. Fairlie, J. Med. Chem. 2000, 43,
3495–3504; b) D. R. March, G. Abbenante, D. A. Bergman, R. I. Brink-
worth, W. Wickramasinghe, J. Begun, J. L. Martin, D. P. Fairlie, J. Am.
Received: September 30, 2013
Revised: October 7, 2013
Published online on October 15, 2013
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1918 – 1921 1921