Triazole-Containing Macrocyclic Protease Inhibitors
FULL PAPER
Table 1. In vitro inhibition data for macrocyclic compounds 1–11 and acyclic compounds 12–15.
b-strand geometry, almost uni-
formly known to favor binding
to proteases.
Compound b-Strand
P2 amino Ring
Cysteine proteases
IC50 [nm]
Proteasome
IC50 [nm][b,c]
CP-L
conformation[a] acid
size
Calpain II[b,d] Cat L[e] Cat S[e] CT-L
T-L
Not all the macrocycles are
1
2
3
4
5
6
7
8
no[f]
no
Leu
Leu
Leu
Leu
Leu
Leu
Leu
Phe
Ile
18
19
21
15
15
20
21
21
21
20
17
–
1020[f]
360
190
1200
1900
310
920
35
35
23
12
47
480
34
87
47
32
22
87
39
10
27
3000 >25000 >25000
>25000 >25000 >25000
3600 >25000 >25000
>25000 >25000 >25000
>25000 >25000 >25000
970 >25000 >25000
250 >25000 >25000
310 >25000 >25000
360 >25000 >25000
250 >25000 >25000
constrained into
a b-strand,
with 4–11, but not 1–3, adopting
this geometry. In accordance
with previous studies on deriva-
no[f]
yes[g]
yes[g]
yes[f]
yes[f]
yes[f]
yes
940[f]
582[g]
355[g]
137[f]
97[f]
tives of Cat 0811 (Figure 1)[1c]
a
2.3
3.0
1.6
2.7
3.9
19-membered macrocycle, as in
2, does not constrain the back-
bone into a b-strand. Interest-
89[f]
9[h]
10
11
12
13
14
15
410
yes[g]
yes
Ile
390[g]
697
Leu
Leu
Leu
Phe
Ile
20
>25000 >25000 >25000 ingly, however, the 20- and 21-
nd[i]
nd[i]
nd[i]
nd[i]
780[f]
1030[f]
490
2.5
4.6
1.5
2.2
54 >25000 >25000
membered macrocycles 6–10
containing a polar triazole ring
–
–
–
150 >25000 >25000
nd[i] >25000 >25000
at P3 and a more hydrophobic
tyrosine at P1, do adopt this ge-
ometry. In general, the com-
pounds with b-strand conforma-
tion were more potent towards
Cat L (6–10), Cat S (4, 6–11)
and calpain II (6–8).
A comparison of the inhibito-
ry activity of the macrocycles
6–9 with that of their acyclic
azide–alkyne analogues 12–15
provides some insight into the
effect of macrocyclization on
390
20 >25000 >25000
3
[a] b-Strand conformation was determined on the basis of JNHCaH coupling constants from 1H NMR spectra.
[b] Values are the mean of three experiments and variation between experiments is < ꢂ5%. [c] Final concen-
tration of substrates was 50 mm. Km values were 48 mm for Suc-LLVY-AMC (CT-L),[26a] >500 mm for Boc-LRR-
AMC (T-L),[26b] and 120 mm for Cbz-LLE-AMC (CP-L).[26c] [d] The final concentration of BODIPY-FL casein
was ꢀ0.09 mm. Km value for calpain II was 0.64 mm.[26d] [e] Data were calculated from experiments with five dif-
ferent inhibitor concentrations. IC50 values were obtained by nonlinear regression, with standard errors <20%
except Cat S values for 6 (24%), 8 (29%), 9 (21%) and 13 (22%). Assays were performed with the chromo-
genic substrate Cbz-Phe-Arg-pNA (pNA =p-nitroanilide) at a final concentration of 100 mm. Km values were
17 mm for Cat L and 118 mm for Cat S.[24c–e] [f] Taken from ref. [6a]. [g] Taken from ref. [15]. [h] Contains
ꢀ10% of epimer (a to the aldehyde) based on NMR spectroscopic analysis. [i] nd=not determined.
15, against calpain II, cathepsin L (Cat L), cathepsin S
(Cat S), and the three activities of the 20S proteasome (i.e.,
its chymotrypsin-like (CT-L), caspase-like (CP-L), and tryp-
sin-like (T-L) activities) were investigated and the results
are shown in Table 1. All compounds were potent inhibitors
of Cat S, with derivatives 6–10 and 12–15 exhibiting IC50
values of less than 5 nm. A number of these compounds (6–
10, 12, 14, and 15) were also highly active against Cat L,
with IC50 values in the range of 10–50 nm. There is some cor-
relation between activity and the ability of the macrocycles
to adopt a b-strand geometry as discussed below. The mac-
rocyclic aldehydes 6, 7, and 8 were the most active com-
pounds of the series against calpain II, with IC50 values of
137, 97, and 89 nm, respectively. The 20- and 21-membered
macrocyclic compounds 6–10, all of which adopt a b-strand
geometry, were good inhibitors of CT-L activity with an IC50
<1 mm. Very strong inhibition of CT-L was also observed for
the acyclic azido–alkynes 12 and 15, which exhibit IC50
values of 54 and 20 nm, respectively. Interestingly, all com-
pounds were inactive against CP-L and T-L (IC50 >25 mm).
All the compounds investigated showed the greatest activity
against Cat S, particularly so in the case of macrocycles 3, 4,
11, and 14. Several classes of highly potent inhibitors for
Cat S have been reported recently that contain an electro-
philic warhead that interacts with the active-site cysteine.[10a]
Nitrile-based compounds have attracted particular interest,
some of which exhibit Ki values toward Cat S in the subna-
nomolar range.[23] Our study is unique in that it addresses
the role of constraining the backbone of an inhibitor into a
potency. The macrocycles of 6–9 generally enhances the po-
tency of inhibition for both Cat L and calapin II compared
with the acyclic analogue. Interestingly, potency against CT-
L is greatest for the acyclic azide–alkyne analogues, whereas
the data for Cat S suggests that constraining the backbone
into a b-strand geometry with a macrocycle has little effect
with all derivatives being highly potent. This second obser-
vation may reflect in part the inability of the assay to dis-
criminate between these particularly potent compounds, as
their single-digit nanomolar IC50 values are in the same
range as the concentration of Cat S in the assay. Future ki-
netic studies are needed to fine-tune the structure–activity
relationships of these macrocyclic Cat S inhibitors. Whereas
a b-strand geometry is almost universally known to favor
ligand binding to a protease,[6a] there is some recent sugges-
tion that a macrocycle is not favored for the proteasome.[16]
Whereas inhibitors of the proteasome reportedly adopt hy-
drogen bonds with the protease that are characteristic of
binding in this geometry, unlike other proteases, the P2
group does not seem to form important contacts with the
active site.[1a]
Some insight into the influence of the key P1 substituent
on the inhibition of proteases is apparent on comparing
macrocycles 7 and 11. An aromatic Tyr at P1 as in macrocy-
cle 7 considerably enhances inhibitory potency against cal-
pain II (7-fold), Cat L (14-fold), Cat S (7-fold), and CT-L
(>100-fold) relative to a triazole at P1 as in 11. Thus an aro-
matic (nontriazole) residue at P1 appears to contribute to
binding within the active site of these proteases. Further in-
Chem. Eur. J. 2013, 00, 0 – 0
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
5
&
ÞÞ
These are not the final page numbers!