Second generation of BACE-1 inhibitors part 3: Towards non hydroxyethylamine transition state mimetics

Article abstract of DOI:10.1016/j.bmcl.2009.03.149

Our first generation of hydroxyethylamine BACE-1 inhibitors proved unlikely to provide molecules that would lower amyloid in an animal model at low oral doses. This observation led us to the discovery of a second generation of inhibitors having nanomolar activity in a cell-based assay and with the potential for improved pharmacokinetic profiles. In this Letter, we describe our successful strategy for the optimization of oral bioavailability and also give insights into the design of compounds with the potential for improved brain penetration.

Full text of DOI:10.1016/j.bmcl.2009.03.149

Bioorganic & Medicinal Chemistry Letters 19 (2009) 3674–3678
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Second generation of BACE-1 inhibitors part 3: Towards non
hydroxyethylamine transition state mimetics
*
Nicolas Charrier, Brian Clarke, Leanne Cutler, Emmanuel Demont , Colin Dingwall, Rachel Dunsdon,
Julie Hawkins, Colin Howes, Julia Hubbard, Ishrut Hussain, Graham Maile, Rosalie Matico, Julie Mosley,
Alan Naylor, Alistair O’Brien, Sally Redshaw, Paul Rowland, Virginie Soleil, Kathrine J. Smith,
Sharon Sweitzer, Pam Theobald, David Vesey, Daryl S. Walter, Gareth Wayne
Neurology and Gastrointestinal Centre of Excellence for Drug Discovery, GlaxoSmithKline R&D, New Frontiers Science Park, Third Avenue, Harlow, Essex CM19 5AW, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
Our first generation of hydroxyethylamine BACE-1 inhibitors proved unlikely to provide molecules that
would lower amyloid in an animal model at low oral doses. This observation led us to the discovery of a
second generation of inhibitors having nanomolar activity in a cell-based assay and with the potential for
improved pharmacokinetic profiles. In this Letter, we describe our successful strategy for the optimiza-
tion of oral bioavailability and also give insights into the design of compounds with the potential for
improved brain penetration.
Received 25 February 2009
Revised 30 March 2009
Accepted 30 March 2009
Available online 5 April 2009
Keywords:
Alzheimer
Ó 2009 Elsevier Ltd. All rights reserved.
BACE-1
Apartic protease
Hydroxyethylamine
In the two preceding papers,1 we have shown that our first
generation of BACE-1 inhibitors was unlikely to deliver a com-
pound capable of lowering amyloid in an animal model at low
(ꢀ10–20 mg/kg) oral doses. We then described the unique proper-
ties (in terms of in vitro potency and selectivity) of a series of com-
pounds bearing a 7,6,5 indole-derived tricyclic non prime side
substituent. Notably, exemplars of these hydroxyethylamine tran-
sition-state mimetics can achieve nanomolar potency in a cell-
based assay without interacting with the prime side of the enzyme.
A potential disadvantage of such compounds was, however,
suggested by the synthetic route used to access them. The prepara-
tion of the tricyclic moiety involves Michael addition of the indole
N–H to a vinyl sulfonamide and we were concerned that a reverse
elimination reaction could take place in vivo, generating a tran-
sient Michael acceptor such as 2 that could be irreversibly trapped
by endogenous thiol-containing nucleophiles (Fig. 1) and poten-
tially lead to a sub-optimal toxicological profile.
a solvent that, after 6 h at 40 °C, 90% of the tricyclic inhibitor was
converted to the opened adduct 3 (ratio determined by LC–MS).
The conditions needed to generate and trap the transient Michael
acceptor 2 were thus significantly harsher than physiological con-
ditions and the potential for toxicity via this mechanism was con-
sidered to be extremely low.
However, this line of thinking led us to assess the potential of
inhibitors incorporating the analogous ‘reverse’ tricycle non prime
side substituent 4 (Scheme 1). These were expected to have a sim-
ilar binding efficiency, but without the potential to generate a
Michael acceptor in vivo. The synthesis of this novel ring system
used a high yielding (93%) intramolecular condensation of an al-
pha-sulfonamide ester to a Vilsmeier–Haack adduct, followed by
decarboxylation and hydrogenation. These reactions could be per-
formed on a 50 g scale. Because of previously developed SAR,3 we
chose to focus only on compounds in which the sulfonamide nitro-
gen was methylated and the indole nitrogen was ethylated.
A comparison of the activity and selectivity of the two tricyclic
series is shown below (Table 1). Surprisingly, the small structural
modification inherent to the new tricycle system led to an increase
in activity in both enzyme and cell-based assays, whilst inhibition
of BACE-2 and Cat-D remained similar (providing an increase in
selectivity). Interestingly, compounds incorporating the new ring
system also appeared to be less lipophilic in general. Overall, nano-
molar potency could be achieved in a cell-based assay with com-
pounds of relatively low (for protease inhibitors)4 molecular
In order to assess this possibity further, we examined the stabil-
ity of the compounds using phenyl methane thiol as a surrogate
nucleophile. The tricyclic indoles2 were stable in the presence of
3 equiv of phenyl methane thiol in ethanol at 80 °C and no decom-
position (in particular, no adduct such as 3) was observed by LC–
MS after 4 h. It was only when phenyl methane thiol was used as
* Corresponding author. Tel.: +44 0 1438764319; fax: +44 0 1438768232.
E-mail address: emmanuel.h.demont@gsk.com (E. Demont).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.03.149

N. Charrier et al. / Bioorg. Med. Chem. Lett. 19 (2009) 3674–3678
3675
O
O
S
R1
O
O
S
N
N
O
O
S
O
R1
N
R1
N
N
N
N
O
O
N
N
S
SH
Enzyme
3
1
2
Figure 1. Potential for the formation of Michael acceptor in the tricyclic series of BACE-1 inhibitors.
O
a-c
O
O
O
O
d, e
Br
S
O
Cl
O
O
O
O
O
N
O2N
N
O
S
f, g
O
O
O
O
O
N
NO2
h-j
N
O
O
S
O
N
O
N
k, l
O
S
O
O
O
O
N
N
4
Scheme 1. Synthesis of novel 7,6,5 indole-like tricyclic non prime side substituent. Reagents and conditions: (a) MeOH, SOCl2, 25 °C, 16 h 95%; (b) DMF, dimethylformamide
dimethyl acetal, 40 °C, 24 h, >95%; (c) MeOH, 10% Pd/C, NH4COOH, 50 °C, 2 h, >90%; (d) Na2SO3, H2O/MeOH, 50 °C, 30 min, >98%; (e) PCl5, 50 °C to 100 °C, 1 h; (f)
ClSO2CH2COOCH3, pyridine, DMAP, CH2Cl2, T < 20 °C, 83%; (g) CH3I, K2CO3, DMF, 16 h, 35–40% (2 steps); (h) POCl3, DMF, 0 °C, 40 min, then 60 °C, 48 h; 93%; (i) NaH, DMF,
C2H5I, 25 °C, 16 h, 81%; (j) NaOH, MeOH/H2O, reflux, 2 h, 57%; (k) 2 N HCl, dioxan, reflux, 16 h, 84%; (l) 10% Pd/C, NH4COOH, EtOH/H2O, 78 °C, 1 h, 77%.
Table 1
Comparison of activity and selectivity of the two tricyclic series of BACE-1 inhibitors
O
CF3
B
C
A
O
O
S
R1
N
X
O
N
N
O
Y
F
E
D
a
b
Compd
X
Y
R1
A
BACE-1 IC50
(lM)
BACE-2 IC50
(l
M)
Cat-D IC50
(l
M)
Ab40 IC50 (lM)
LogD @ pH 7.4
MW
555
CMR
15.2
5
6
7
8
C
N
C
N
C
N
C
N
C
N
C
N
C
N
C
N
C
N
C
N
C
N
C
0.005
0.015
0.004
0.007
0.010
0.022
0.011
0.073
0.023
0.048
0.009
0.020
0.59
0.49
0.075
0.028
0.42
0.43
1.49
3.1
2.17
1.14
0.37
0.263
2.11
6.61
0.57
0.027
0.090
0.003
0.008
0.064
0.144
0.026
0.118
0.073
0.262
0.014
0.016
0.9
1.31
2.59
2.8
1.08
1.52
—
—
0.07
—
B
C
D
E
F
567
527
509
499
511
14.4
14.8
14.1
13.9
14.2
0.89
9
1.19
5.43
34.59
75.5
19.54
36.31
6.79
7.93
10
11
12
13
14
15
16
1.79
2.17
N
a
In all tables, IC50s reported are means of the values of three different experiments. Each IC50 is within threefold of the mean value.
In SHSY5Y wild type cells. See previous paper for details.
b

3676
N. Charrier et al. / Bioorg. Med. Chem. Lett. 19 (2009) 3674–3678
Table 2
Comparison of activity and selectivity of sulfone and sulfonamide tricyclic inhibitors
CF3
O
O
O
S
G
R1
X
N
A
O
N
N
O
D
F
Compd
X
R1
G
BACE-1 IC50
(
lM)
BACE-2 IC50
(l
M)
Cat-D IC50 (lM)
17
18
6
19
12
20
16
21
N–CH3
O
N–CH3
O
N–CH3
O
N–CH3
O
0.002
0.008
0.009
0.086
0.073
0.36
0.059
0.26
0.21
0.45
3.1
3.2
0.263
1.15
1.26
2.16
0.66
A
D
F
0.13
75.51
15.14
7.93
0.020
0.25
2.97
weight (MW) and size (as defined by their calculated molar refrac-
tivity, CMR).
benzylic prime side residue are, unsurprisingly, predicted to be less
likely to be well absorbed due to their bigger volume compared to
inhibitors with a smaller alkyl chain. The chemical space in which
compounds with an acceptable lipophilicity profile (1 < LogD < 3)
are predicted to be well absorbed is now populated, in contrast
to the predictions for our first generation sultams.
Looking in more detail at the compounds with an alkyl non
prime side substituent, three comments can be made. (1) Increas-
ing the size of the alkyl substituent (whilst maintaining the basi-
city of the nitrogen bound to the catalytic aspartates) increases
lipophilicity and CMR in parallel so that the absorption prediction
remains similar (see representative examples Fig. 3). (2) It is possi-
ble to increase the lipophilicity of adducts with similar CMR by
lowering the basicity of the nitrogen bound to the aspartate resi-
dues. This can be achieved either by using a cyclopropyl amine
or by introducing electron withdrawing fluorine atoms into the
prime side residue (Fig. 4), both of these strategies lead to lower
basicity and hence higher log D @ pH 7.4;9 (3) For compounds with
the same CMR, it is possible to modify lipophilicity by the choice of
The significant effect on activity associated with a shift in the
position of a single nitrogen atom in such relatively large inhibitors
can be linked with earlier findings in the initial tricyclic series in
which the N-methyl sulfonamide derivatives were significantly
more potent than the sulfonate counterparts5 (Table 2). The sulfo-
natederivativeswerestableunder theexperimental conditionsused
for both enzyme and cell-based assays, ruling out in situ decomposi-
tion of the sulfonates as the reason for their weaker activity.
Sulfonamides are known to be weaker hydrogen bond acceptors
than sulfones,6 an observation that can be explained by the
absence of resonance between the nitrogen lone pair and the SO2
group, with only the inductive effect of nitrogen making a contri-
bution.7 The greater potency of the sulfonamides compared with
the sulfonates is therefore unlikely to be related to the strength
of the hydrogen bond made with the enzyme.8 Since the tricycle
residues are conformationally very constrained, it is also unlikely
that there are significant conformational differences between the
sulfonates and sulfonamides and the differences in activity may
best be attributed to the impact of these functionalities on the elec-
tronics of the bicyclic aromatic ring.
tricycle non prime side substituent (see
DLogD, Table 1). Increased
lipophilicity can also be achieved by substituting the S1 phenyl
A graphical representation of predictions of oral absorption for
the two tricyclic series is shown in Figure 2. Derivatives bearing a
Scatter Plot
4
4
3.5
O
O
S
NH
3.5
N
R
O
N
N
3
2.5
2
O
3
N
Alkyl prime side
substituent
2.5
Benzylic prime side
substituent
NH
2
1.5
1
1.5
1
NH
NH
Good
Good
Good borderline
Poor borderline
Poor
Good borderline
Poor borderline
Poor
0.5
0
NH
0.5
0
13
14
15
16
17
CMR
13
14
15
16
CMR
17
18
19
Figure 3. Prediction of absorption for inhibitors with increase alkyl prime side
Figure 2. Prediction of absorption for the tricyclic series of inhibitors.
substituent.

N. Charrier et al. / Bioorg. Med. Chem. Lett. 19 (2009) 3674–3678
3677
ring with halogen atoms which has no effect on enzyme or cellular
activity (data not shown).10
Within the set of compounds predicted to be well absorbed,
compounds 7 and 16 proved to have good oral exposure in multi-
ple species (Table 3).11
3
2.5
2
F
F
O
O
S
NH
N
R
O
F
N
N
O
N
F
Not unexpectedly for aspartyl protease inhibitors, CNS penetra-
tion studies comparing MDR knockout mice and wild-type mice
showed that the compounds are Pgp substrates.12 The ratio of
brain to blood concentrations was markedly higher in the knockout
mice, suggesting a high degree of active efflux in the wild-type
mice in the case of inhibitors 7 and 16. (Table 4).
Since these tricyclic inhibitors incorporate extensively opti-
mized non prime side residues and need to form minimal interac-
tions with the prime side of the enzyme, one of the few remaining
options likely to reduce Pgp interactions and increase the likeli-
hood of brain penetration is to modify the transition state mimetic
itself and reduce the number of hydrogen bond donors and accep-
tors present.13 This approach has been used successfully in another
series of inhibitors.14
NH
F
1.5
1
F
NH
NH
Good
Good borderline
Poor borderline
Poor
NH
0.5
0
NH
13
13.5
14
14.5
15
15.5
CMR
Figure 4. Prediction of absorption for inhibitors bearing cyclopropyl or polyfluo-
rinated alkyl prime side substituents.
Compounds in the tricyclic series appear to need a basic nitro-
gen in order to see a good correlation of activities in both the
Table 3
In vivo pharmacokinetics profile of inhibitors 7and 16 in rat and dog
Vda (L/kg)
T1/2 (h)
Cmax
(lM)
%F
a
a
b
Species
Compd
Clblood (mL/min/kg)
Rat n = 2–3
16
7
16
7
72,62
57 12
11,8
3, 2.8
0.184 0.082
0.088 0.03
2.969 0.787
3.586 2.343
17, 22
13 3
11.0 3.6
4.9 0.9
4.8 1.6
3.3 1.5
4.2 0.2
6.8 0.2
c
Dog n = 3
23
14
5
4
79 50
52
9
a
1 mg free base (fb)/kg/h i.v. dose (solution of mesylate salt in 0.9% w/v saline containing 10% w/v Kleptose).
10 mg fb/kg (solution of mesylate salt in 1% v/v Tween 80 and 1% w/v methylcellulose aqueous).
Two dogs had Fpo ꢀ50% whilst the third dog had Fpo markedly higher (>100%) than would have been predicted from the blood clearance.
b
c
Table 4
CNS penetration study data for inhibitors 7 and 16
Compd
Mouse Strain
Css blood (
l
M)
Css brain (lM)
CLb (mL/min/kg)
Brain:blood ratio
Ratio (À/À):(+/+)
16
mdr1a/b (À/À)
mdr1a/b (+/+)
0.397, 0.544
0.490 0.095
0.989, 1.310
0.061 0.009
89, 62
72 14
2.49, 2.41
0.13 0.01: 1
19.4
7
mdr1a/b (À/À)
0.339 0.037
0.381 0.032
0.335 0.045
0.040 0.008
94 11
0.99 0.11:1
0.10 0.02:1
9.5
mdr1a/b (+/+)
85
6
Table 5
Influence of pKa on cellular activity of BACE-1 inhibitors
O
O
S
R1
N
O
N
N
O
X
Y
R1
BACE-1 IC50
(
lM)
Ab40 IC50
(
lM)
IC50s ratioa
pKa
b
Compd
X
Y
22
23
24
16
7
25
26
27
28
C
N
C
N
C
C
C
C
C
N
C
N
C
N
N
N
N
N
(CH2)2CH3
CH2CH2F
CH2CH2F
c-C3H5
(CH2)2CF3
CH2CHF2
CH2CF2CF3
CH2CF3
Phenyl
0.008
0.050
0.040
0.011
0.003
0.020
0.032
0.079
0.005
0.029
0.209
0.245
0.030
0.003
0.051
4.427
2.109
0.739
3.7
4.2
6.2
2.7
1.0
10.58
9.19
9.19
8.76
8.6
7.45
5.7
5.4
2.6
140.0
26.6
147.5
4.6
a
Calculated as Ab40 IC50/BACE-1 IC50
.
b
Reported value of pKa for R1NH2. See Ref. 9 for details.

3678
N. Charrier et al. / Bioorg. Med. Chem. Lett. 19 (2009) 3674–3678
O
O
O
S
N
O
N
N
O
N
Figure 5. Structure and binding mode to the BACE-1active site of inhibitor 29.
4. Eder, J.; Hommel, U.; Cumin, F.; Martoglio, B.; Gerhartz, B. Curr. Pharm. Des.
2007, 13, 271.
enzyme and cell-based assays (Table 5); compounds with amines
of low basicity (pKa < 6) do not generally inhibit amyloid secretion
in cell-based assays.
5. The tricyclic sulfonates where synthesised in an analogous manner to the
sulfonamide, using 7-hydroxy indoles as starting material. See: Charrier, N.;
Demont, E.; Dunsdon, R.; Maile, G.; Naylor, A.; O’Brien, A.; Redshaw, S.;
Theobald, P.; Vesey, D. S.; Walter, D. W. Synthesis 2006, 3467.
6. Abraham, M. H.; Duce, P. P.; Prior, D. V.; Baratt, D. G.; Morris, J. J.; Taylor, P. J. J.
Chem Soc., Perkin Trans 2 1989, 1355.
The alpha-amino ketone 29 (Fig. 5) is a first example of a com-
pound demonstrating that a tricyclic BACE-1 inhibitor without a
hydroxyethylamine transition state mimetic - and with fewer
hydrogen bond donors and acceptors - can achieve excellent po-
tency in cell-based assays (BACE-1 IC50 = 13 nM; Ab40
IC50 = 48 nM). Interestingly, this derivative binds to the enzyme’s
aspartyl residues as a hydrate, whilst is only present as the ketone
in solution (based on NMR studies).15
In summary, in this and the two preceding papers, we have pre-
sented the results and reasoning that led us to discover a second
generation of stable, potent and orally bioavailable BACE-1 inhibi-
tors. These derivatives are Pgp-substrates and have sub-optimal
brain penetration but early findings suggest that non hydroxyeth-
ylamine inhibitors with better likelihood of crossing the blood
brain barrier can be generated from this series. Further results in
this area will be reported in due course.
7. Charton, M. Prog. Phys. Org. Chem. 1981, 13, 119.
8. One may argue that these results may be explained by the sulfonate being a
weaker hydrogen bound acceptor than the sulfonamide, oxygen being more
electronegative than nitrogen (with respective electronegativity values of 3.44
and 3.04; Pauling scale). However, in this specific case, the conjugation of both
systems with an aromatic ring should dilute this effect.
9. For measured pKa of cyclopropyl amine or poly fluoro alkylamines, see: (a)
Kluger, R.; Hunt, J. C. J. Am. Chem. Soc. 1984, 106, 566; (b) Willberg, K. B.; Ross, B.
S.; Isbell, J. J.; McMurdie, N. J. Org. Chem. 1993, 58, 1372; (c) Sakai, T.; Korenaga,
T.; Washio, N.; Nishio, Y.; Minami, S.; Ema, T. Bull. Chem. Soc. Jpn. 2004, 77,
1001.
10. In our hands, 2-F and 2,4-F phenyls are equipotent to the unsubstituted phenyl
derivative.
11. Data for compound 16 have already been published in: Charrier, N.; Clarke, B.;
Cutler, L.; Demont, E.; Dingwall, C.; Dunsdon, R.; Hawkins, J.; Howes, C.;
Hussain, I.; Maile, G.; Matico, R.; Mosley, J.; Naylor, A.; O’Brien, A.; Redshaw, S.;
Rowland, P.; Soleil, V.; Smith, K. J.; Sweitzer, S.; Theobald, P.; Vesey, D. S.;
Walter, D. W.; Wayne, G. J. Med. Chem. 2008, 51, 3313.
12. See for example: Anderson, B. D. Drug Metab. Dispos. 2002, 30, 479. and
references cited herein.
References and notes
13. Doan, K. M. M.; Humphreys, J. E.; Webster, L. O.; Wring, S. A.; Shampine, L. J.;
Serabjit-Singh, C. J.; Adkison, K. K.; Polli, J. W. J. Pharmacol. Exp. Ther. 2002, 303,
1029.
1. See: (a) Charrier, N.; Clarke, B.; Cutler, L.; Demont, E.; Dingwall, C.; Dunsdon, R.;
Hawkins, J.; Howes, C.; Hussain, I.; Maile, G.; Matico, R.; Mosley, J.; Naylor, A.;
O’Brien, A.; Redshaw, S.; Rowland, P.; Soleil, V.; Smith, K. J.; Sweitzer, S.;
Theobald, P.; Vesey, D.; Walter, D. W.; Wayne, G. Bioorg. Med. Chem. Lett. 2009
19, 3664; (b) Charrier, N; Clarke, B.; Demont, E.; Dingwall, C.; Dunsdon, R.;
Hawkins, J.; Hussain, I.; Maile, G.; Matico, R.; Mosley, J.; Naylor, A.; O’Brien, A.;
Redshaw, S.; Rowland, P.; Soleil, V.; Smith, K. J.; Sweitzer, S.; Theobald, P.;
Vesey, D.; Walter, D. W.; Wayne, G. Bioorg. Med. Chem. Lett. 2009, 19, 3669.
2. The tricyclic inhibitor used in this experiment was the hydroxyethylamine
derivative bearing a pyrane as prime side substituent.
3. Clarke, B.; Demont, E.; Dingwall, C.; Dunsdon, R.; Faller, A.; Hawkins, J.;
Hussain, I.; MacPherson, D.; Maile, G.; Matico, R.; Milner, P.; Mosley, J.; Naylor,
A.; O’Brien, A.; Redshaw, S.; Riddell, D.; Rowland, P.; Soleil, V.; Smith, K. J.;
Stanway, S.; Stemp, G.; Sweitzer, S.; Theobald, P.; Vesey, D.; Walter, D. W.;
Ward, J.; Wayne, G. Bioorg. Med. Chem. Lett. 2008, 18, 1011.
14. See for examples: (a) Rajapakse, H. A.; nantermet, P. G.; Selnick, H. G.; Munshi,
S.; McGaughey, G. B.; Lindsley, S. R.; Young, M. B.; Lai, M.-T.; Espeseth, A. S.; Shi,
X.-P.; Colussi, D.; Pietrak, B.; Crouthamel, M.-C.; Tugusheva, K.; Huang, Q.; Xu,
M.; Simon, A. J.; Kuo, L.; Hazuda, D. J.; Graham, S.; Vacca, J. P. J. Med. Chem.
2006, 49, 7270; (b) Stanton, M. G.; Staufferch, S. R.; GrHegro, A. R.; Steinbeiser,
M.; Nantermet, P.; Sankaranarayanan, S.; Price, E. A.; Wu, G.; Crouthamel, M.-
C.; Ellis, J.; Lai, M.-T.; Espeseth, A. S.; Shi, X.-P.; Jin, L.; Colussi, D.; Pietrak, B.;
Huang, Q.; Xu, M.; Simon, A. J.; Graham, S. L.; Vacca, J. P.; Selnick, H. J. Med.
Chem. 2007, 50, 3431.
15. The PDB deposition codes and refinement details for the BACE-1 complex
crystal structures for 29 are 2wf4, 1.8 Å resolution, R = 0.192, Rfree = 0.224. For
another example of amino ketones as BACE-1 inhibitors, see: Uchikawa, O.;
Kazuyoshi, A.; Koike, T.; Tarui, N.; Hirai, K. Int. Appl. WO2004014843A1, 2004.