A conformational constraint improves a β-secretase inhibitor but for an unexpected reason

Article abstract of DOI:10.1016/j.bmcl.2009.07.071

During our ongoing efforts to develop a small molecule inhibitor targeting the β-amyloid cleaving enzyme (BACE-1), we discovered a class of compounds bearing an aminoimidazole motif. Initial optimization led to potent compounds that have high Pgp efflux r

Full text of DOI:10.1016/j.bmcl.2009.07.071

Bioorganic & Medicinal Chemistry Letters 19 (2009) 4993–4995
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
A conformational constraint improves a b-secretase inhibitor
but for an unexpected reason
b
a
a
a
Ivory D. Hills a, , M. Katharine Holloway , Pablo de León , Ashley Nomland , Hong Zhu ,
Hemaka Rajapakse a, Tim J. Allison c, Sanjeev K. Munshi c, Dennis Colussi d, Beth L. Pietrak d,
Dawn Toolan d, Sharie J. Haugabook d, Samuel L. Graham a, Shawn J. Stachel a
*
a Department of Medicinal Chemistry, Merck Research Laboratories, Merck & Co., PO Box 4, West Point, PA 19486, USA
b Department of Chemistry Modeling and Informatics, Merck Research Laboratories Merck & Co., PO Box 4, West Point, PA 19486, USA
c Department of Structural Biology, Merck Research Laboratories, Merck & Co., PO Box 4, West Point, PA 19486, USA
d Department of Neuroscience, Merck Research Laboratories, Merck & Co., PO Box 4, West Point, PA 19486, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
During our ongoing efforts to develop a small molecule inhibitor targeting the b-amyloid cleaving
enzyme (BACE-1), we discovered a class of compounds bearing an aminoimidazole motif. Initial optimi-
zation led to potent compounds that have high Pgp efflux ratios. Crystal structure-aided design furnished
conformationally constrained compounds that are both potent and have relatively low Pgp efflux ratios.
Computational studies performed after these optimizations suggest that the introduction of the con-
straint enhances potency via additional hydrophobic interactions rather than conformational restriction.
Ó 2009 Elsevier Ltd. All rights reserved.
Received 24 June 2009
Revised 10 July 2009
Accepted 14 July 2009
Available online 17 July 2009
Keywords:
BACE-1
b-Secretase inhibitor
Conformational constraint
Alzheimer’s disease (AD) is the most common cause of demen-
tia and is expected to become a more prevalent problem in the
aging human population.1 b-amyloid precursor protein cleaving
enzyme (BACE-1) plays an integral role in the formation of b-amy-
loid (Ab) peptides, which subsequently form plaques in the brain.2
Thus, considerable effort has been directed towards developing a
small molecule inhibitor of BACE-1, with the hopes of stopping
or reversing AD progression.3 However, progress has been ham-
pered by the requirement that an inhibitor of BACE-1 be both po-
tent and brain penetrant.
studies suggest that the constraint furnishes more potent com-
pounds by the introduction of additional hydrophobic binding
interactions rather than conformational restriction.
Our initial foray into aminoheterocyclic BACE-1 inhibitors has
been previously described. Optimization of an aminothiazole HTS
hit led to aminoimidazole 1, a potent compound with a high Pgp
efflux ratio (IC50 = 0.47
the discovery of inhibitor 1, we synthesized aminoimidazole 2, a
less potent compound with a lower Pgp efflux ratio (IC50 = 1.8 M;
lM; Pgp (BA/AB) = 20). Concurrent with
l
Pgp (BA/AB) = 4.0) (Fig. 1). Of the many derivatives examined, the
Our research laboratories have disclosed inhibitors bearing ami-
nothiazole and aminoimidazole cores.4,5 These compounds are
inhibitors of BACE-1; however, they also have relatively high Pgp
efflux ratios (BA/AB) and are not anticipated to be brain penetrant.
Herein, we discuss the further development of these aminohetero-
cyclic BACE-1 inhibitors, focusing on simultaneously enhancing
potency and maintaining or lowered Pgp efflux. Of particular inter-
est is the introduction of a conformational constraint inspired by a
crystal structure. Introduction of conformational rigidity is a stan-
dard strategy in drug discovery and we anticipated this modifica-
tion would lead to a significant improvement in potency over
comparable unconstrained analogs.6 Interestingly this was only
true for certain substitution patterns; furthermore, computational
2-methoxy-5-nitro substituents on the benzyl subunit were found
benzyl subunit
benzyl subunit
CF3
MeO
NO2
Me
Me
N
N
N
N
MeO
MeO
NH2
NH2
1
2
BACE-1 IC50 = 0.47 µM
BACE-1 IC50 = 1.8 µM
Pgp (BA/AB) = 20
Pgp (BA/AB) = 4.0
* Corresponding author.
Figure 1. Compound 1 is potent but a substrate for Pgp efflux. Aminoimidazole 2 is
E-mail address: ivory_hills@merck.com (I.D. Hills).
less potent but has a lower Pgp efflux ratio.
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.07.071

4994
I. D. Hills et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4993–4995
to furnish the most potent aminoimidazole inhibitor. However, we
also believed that these functional groups were in part responsible
for our high Pgp efflux ratio and we sought to prepare potent inhib-
itors without this substitution pattern.7 While inhibitor 2 has a sig-
nificantly attenuated Pgp efflux ratio compared to compound 1, we
believed to move forward the potency of compound 2 would need
to be improved. Therefore, we attempted to further optimize this
scaffold, hoping to improve potency with substituents that would
provide BACE-1 inhibitors with relatively low Pgp efflux.
MeO
MeO
introduce
constraint
NO2
Me
NO2
N
N
N
Me
MeO
N
MeO
NH2
NH2
1
3
BACE-1 IC50 = 0.47 µM
BACE-1 IC50 = 3.2 µM
We were aided in this endeavor by the X-ray crystal structure of
inhibitor 1 bound to the BACE-1 active site (Fig. 2). Two crucial
observations were made: (1) the bioactive conformation of 1 places
the hydrogen atom of the central tertiary carbon in close proximity
of an ortho hydrogen on 4-methoxy phenyl subunit’s aromatic ring
and; (2) there appears to be sufficient room in the active site to
accommodate additional structural modification to the inhibitor.
We hypothesized that the introduction of an appropriate con-
formational constraint would lead to a more potent inhibitor and
we sought to covalently attach an ortho-carbon of the phenyl sub-
unit and the central tertiary carbon, consequently leading to a
more rigid molecule pre-organized in the bioactive conformation.
We immediately added an indane-based constraint to inhibitor 2,
since it was the more potent of the two leading analogs (Fig. 3).
Disappointingly, constrained analog 3 was found to be almost sev-
enfold less potent than the unconstrained inhibitor! We were per-
plexed but not discouraged by this result and simple plastic
models led us to hypothesize that perhaps the 2-methoxy-substit-
utent on the benzyl subunit engages in negative steric interactions
with the indane-based constraint. This should in turn lead to a rel-
ative destabilization of the putative bioactive conformation.
Therefore, we were cautiously optimistic when we synthesized
inhibitor 4, a constrained variant of aminoimidazole 2. We were
pleased to find that compound 4 is approximately fivefold more
potent than 2 (Fig. 4). In fact, constrained compound 4 is even
more active than potent non brainpenetrant inhibitor 1, which
bears the 2-methoxy-5-nitro substituted benzyl subunit
Figure 3. An indane-constraint was introduced to furnish a rigid molecule pre-
organized in the bioactive conformation. Initial application of this strategy led to a
less potent compound.
CF3
CF3
F
N
N
N
Me
N
Me
MeO
MeO
NH2
NH2
5
4
BACE-1 IC50 = 0.063 µM
BACE-1 IC50 = 0.35 µM
Pgp (BA/AB) = 3.6
Figure 4. Indane-constrained aminoimidazoles bearing a 4-trifluoromethyl substi-
tuted benzyl subunit are potent and have relatively low Pgp efflux ratios.
With these results in hand, we were interested to see if compu-
tations would support our conjecture that the indane constraint
leads to more potent inhibitors by reducing conformational flexi-
bility. We also wanted to further understand why constrained
inhibitor 3 is less potent than inhibitor 1. Gratifyingly, modeling
of 1 and 3,9 both bound and unbound to BACE-1, clearly shows that
introduction of the indane constraint reduces the percentage of
conformers that resemble the bound conformer of compound 3.
Modeling shows that 18.5% of the conformers of compound 1
adopt a bound-like geometry. However, the bound-like conforma-
tion of 3 is disfavored by 1.2 kcal/mol and only 8.9% of the con-
formers of 3 closely resemble the conformation required for
effective binding to the BACE-1 active site. This is consistent with
the observed decrease in potency and is presumably due to nega-
tive steric interactions. As illustrated in Figure 5, the lowest energy
conformer for compound 3 (A, green) is quite dissimilar from the
conformer believed to be required for binding to BACE-1 (A, white).
We next examined inhibitors 2 and 4. When we studied uncon-
strained analog 2, we found that 52% of the conformers adopt a
(IC50 = 0.35
lM versus IC50 = 0.47 lM). This compound, which
lacks the 2-methoxy-5-nitro substitution, led us to believe that it
would be possible to synthesize a potent aminoimidazole with a
relatively low Pgp efflux ratio. Indeed, additional optimization
led to inhibitor 5, a highly active BACE-1 inhibitor with a relatively
low Pgp efflux ratio (IC50 = 0.063 l
M; Pgp (BA/AB) = 3.6).8
Figure 2. X-ray structure of inhibitor 1 as bound in the BACE-1 active site (PDB ID
code: 3H0B). Residues within 3.5 Å of 1 are shown as gray sticks and H-bonding
interactions with the inhibitor are labelled. The crucial tertiary carbon hydrogen
and the adjacent ortho aromatic hydrogen have been added to highlight their
alignment and proximity.
Figure 5. Comparison of modeled bound poses (white) relative to the lowest
energy conformers (green) for inhibitors 3 and 4. A. In the lowest energy conformer
computed for 3, both the 2-methoxy-5-nitro phenyl and the aminoimidazole rings
are flipped 180° relative to the modeled bound pose. B. By contrast, there are 3 low
energy conformers of 4 that are almost identical to the modeled bound pose.

I. D. Hills et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4993–4995
4995
conformation very similar to the bound geometry. Interestingly, as
can be seen in Figure 5, the lowest energy conformer for compound
4 (B, green) is closely related to the conformer modeled in the
BACE-1 active site (B, white). However, to our surprise, there is a
modest decrease in the population of conformers that naturally
resemble the bound conformation for constrained compound 4
(only 42% of the conformers are similar to the bound pose). This
is likely due to a 0.09 kcal/mol energy penalty between the
bound-like pose and the lowest energy conformer. This observa-
tion alone would lead us to conclude that a small erosion of po-
tency should occur when a constraint is installed to furnish 4.
However, force-field based and empirical scoring methods11 sug-
gest that the observed enhanced potency may be due to a favorable
change in the binding energy, as a result of additional van der
Waals contacts between residues on the BACE flap (e.g., Tyr71)
and the indane constraint. Thus, the addition of a constraint does
lead to a more potent inhibitor. However, computations suggest
this strategy is not successful due to conformational restriction;
rather it is due to gaining additional hydrophobic interactions.
This class of constrained aminoimidazole BACE-1 inhibitors is
synthesized in a straightforward manner (Scheme 1). The synthesis
of compound 5 begins with the fluorination of 6-methoxy-1-inda-
none to furnish 7. The carbonyl of 7 is readily removed via hydro-
genation and subsequent oxidation by CrO3 yields isomeric
indanone 9. Treatment of 9 with TosMIC and KOt-Bu allows for
the isolation of nitrile 10. Standard acidic methanolysis leads to
methyl ester 11, which is then alkylated using NaHMDS and (tri-
fluoromethyl)benzyl chloride in THF to give 12. The methyl ester
is then converted to a methyl ketone through the addition of (tri-
methylsilyl)methyllithium to provide intermediate 13. Formation
of a putative TMS-enolate via NaHMDS and TMSCl, followed by
addition of trimethylphenylammonium tribromide leads to al-
pha-bromo ketone 14. Simple substitution of the bromide by
methylamine furnishes amino ketone 15. Finally treatment of the
amino ketone with 2-ethyl-2-thiopseudourea under basic condi-
tions allows the isolation of indane-constrained aminoimidazole 5.
In summary, we have been able to achieve our initial goal of
producing an inhibitor of BACE-1 with improved potency and a rel-
atively low Pgp efflux ratio. This was accomplished by introducing
an indane-based constraint that was effective for certain analogs
(2 vs 4) but not others (1 vs 3). Computational studies unexpect-
edly suggest that the constraint enhances potency via additional
favorable hydrophobic interactions, rather than conformational
restriction.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2009.07.071.
References and notes
1. Facts and figures concerning Alzheimer’s Disease can be found at: www.alz.org.
2. (a) Sinha, S.; Anderson, J. P.; Barbour, R.; Basi, G. S.; Caccavello, R.; Davis, D.;
Doan, M.; Dovey, H. F.; Frigon, N.; Hong, J.; Jacobson-Croak, K.; Jewett, N.; Keim,
P.; Knops, J.; Lieberburg, I.; Power, M.; Tan, H.; Tatsuno, G.; Tung, J.; Schenk, D.;
Seubert, P.; Suomensaari, S. M.; Wang, S.; Walker, D.; Zhao, J.; McConlogue, L.;
John, V. Nature 1999, 402, 537; (b) Yan, R. Q.; Bienkowski, M. J.; Shuck, M. E.;
Miao, H. Y.; Tory, M. C.; Pauley, A. M.; Brashler, J. R.; Stratman, N. C.; Mathews,
W. R.; Buhl, A. E.; Carter, D. B.; Tomasselli, A. G.; Parodi, L. A.; Heinrikson, R. L.;
Gurney, M. E. Nature 1999, 402, 533.
3. For recent reviews discussing BACE-1 inhibition, see: (a) Stachel, S. J. Drug Dev.
Res. 2009, 70, 101; (b) Ghosh, A. K.; Gemma, S.; Tang, J. Neurotherapeutics 2008,
5, 399; (c) Hills, I. D.; Vacca, J. P. Curr. Opin. Drug Discovery Dev. 2007, 10, 383;
(d) McGaughey, G. B.; Holloway, M. K. Exp. Opin. Drug Discovery 2007, 2, 1129.
4. Stachel, S. J.; Coburn, C. A.; Rush, D.; Jones, K. L. G.; Zhu, H.; Rajapakse, H.;
Graham, S. L.; Simon, A.; Holloway, M. K.; Allison, T. J.; Munshi, S. K.; Espeseth,
A. S.; Zuck, P.; Colussi, D.; Wolfe, A.; Pietrak, B. L.; Lai, M.-T.; Vacca, J. P. Bioorg.
Med. Chem. Lett. 2009, 19, 2977.
O
F
F
b
c
O
R
MeO
MeO
MeO
a
6 R = H
8
9
7
R = F
5. (a) Fuchs, K.; Heine, N.; Eickmeier, C.; Handschuh, S.; Dorner-Ciossek, C.;
Hoerer, S. PCT Appl. WO2009007300, 2009.; (b) Holenz, J.; Kers, A.; Kolmodin,
K.; Rotticci, D.; Oehberg, L.; Hellber, S. PCT Appl. WO2009005471, 2009.; For
recent examples of aminoheterocycle- or guanidine-based BACE-1 inhibitors,
see: (c) Baxter, E. W.; Conway, K. A.; Kennis, L.; Bischoff, F.; Mercken, M. H.; De
Winter, H. L.; Reynolds, C. H.; Tounge, B. A.; Luo, C.; Scott, M. K.; Huang, Y.;
Braeken, M.; Pieters, S. M. A.; Berthelot, D. J. C.; Masure, S.; Bruinzeel, W. D.;
Jordan, A. D.; Parker, M. H.; Boyd, R. E.; Qu, J.; Alexander, R. S.; Brenneman, D.
E.; Reitz, A. B. J. Med. Chem. 2007, 50, 4261; (d) Jennings, L. D.; Cole, D. C.; Stock,
J. R.; Sukhdeo, M. N.; Ellingboe, J. W.; Cowling, R.; Jin, G.; Manas, E. S.; Fan, K. Y.;
Malamas, M. S.; Harrison, B. L.; Jacobsen, S.; Chopra, R.; Lohse, P. A.; Moore, W.
J.; O’Donnell, M.-M.; Hu, Y.; Robichaud, A. J.; Turner, M. J.; Wagner, E.; Bard, J.
Bioorg. Med. Chem. Lett. 2008, 18, 767; (e) Geschwidner, S.; Olsson, L.-L.; Albert,
J. S.; Deinum, J.; Edwards, P. D.; de Beer, T.; Folmer, R. H. A. J. Med. Chem. 2007,
50, 5903; (f) Congreve, M.; Aharony, D.; Albert, J.; Callaghan, O.; Campbell, J.;
Carr, R. A. E.; Chessari, G.; Cowan, S.; Edwards, P. D.; Frederickson, M.;
McMenamin, R.; Murray, C. W.; Patel, S.; Wallis, N. J. Med. Chem. 2007, 50, 1124.
6. For recent reviews describing efforts to synthesize conformationally
constrained compounds in drug discovery, see: (a) Andrej, P.; Kikelj, D. Curr.
Med. Chem. 2006, 13, 1525; (b) Kapurniotu, A. Curr. Med. Chem. 2004, 11, 2845;
(c) Freidiner, R. M. J. Med. Chem. 2003, 46, 5553.
OMe
F
F
d
e
N
O
MeO
MeO
10
11
CF3
h
CF3
f
F
R
F
Br
O
O
MeO
MeO
12 R = OMe
14
g
13
R = Me
CF3
7. All aminoimidazoles examined bearing the 2-methoxy-5-nitro benzyl subunit
CF3
were found to have a BA/AB ratio greater than 12.5.
F
8. The fluoro group was initially discovered by empirical screening. Subsequently
X-ray crystal data was obtained that supports the hypothesis that the fluoro
group occupies a small hydrophobic pocket.
i
j
F
NHMe
N
Me
O
N
MeO
9. Conformational searches were performed within Maestro (Version 8.5.207,
Schrödinger, LLC, New York, NY, www.schrodinger.com) using the MMFF94s
force field,10 an implicit water model, mixed torsional/low mode sampling with
a 20 kcal/mol energy window, and subsequent Boltzmann population analysis.
10. (a) Halgren, T. A. J. Comput. Chem. 1996, 17, 490; (b) Halgren, T. A. J. Comput.
Chem. 1996, 17, 520; (c) Halgren, T. A. J. Comput. Chem. 1996, 17, 553; (d)
Halgren, T. A.; Nachbar, R. B. J. Comput. Chem. 1996, 17, 587; (e) Halgren, T. A. J.
Comput. Chem. 1996, 17, 616.
MeO
NH2
15
5
Scheme 1. Reagents and conditions: (a) AccuflourTM NFTh (1 equiv), MeCN, reflux,
4 h; (b) Pd/C, H2, H2SO4, EtOAc, rt, overnight; (c) CrO3, AcOH/H2O, 0 °C, 5 h; (d)
TosMIC, KOt-Bu, DME, 0–50 °C, overnight; (e) HCl, MeOH, 60 °C, overnight; (f)
NaHMDS, 4-(trifluoromethyl)benzyl chloride, THF, À78 °C, 30 min; (g) (trimethyl-
silyl)methyllithium, THF, rt, 1 h; (h) (1) NaHMDS, TMSCl, THF, À78 °C, 30 min; (2)
trimethylphenylammonium tribromide, THF, rt, overnight; (i) Methylamine, THF, rt,
2 h; (j) 2-ethyl-2-thiopseudourea hydrobromide, NaOH, THF/H2O, rt, overnight.
11. Inhibitor models were energy minimized in the BACE-1 active site using the
MMFF94s force field10 with
a 2r distance dependent dielectric constant.
Modeled bound poses were subsequently rescored with Xscore.12
12. Wang, R.; Lu, Y.; Fang, X.; Wang, S. J. Chem. Inf. Comput. Sci. 2004, 44, 2114.