NSAID-derived γ-secretase modulators. Part III: Membrane anchoring

Article abstract of DOI:10.1016/j.bmcl.2009.10.035

Selective lowering of Aβ₄₂ levels with small-molecule substrate targeting γ-secretase modulators (sGSMs), such as some non-steroidal anti-inflammatory drugs, is a promising therapeutic approach for Alzheimer's disease. Here we present N-substit

Full text of DOI:10.1016/j.bmcl.2009.10.035

Bioorganic & Medicinal Chemistry Letters 19 (2009) 6986–6990
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
NSAID-derived
c-secretase modulators. Part III: Membrane anchoring
Stefanie Baumann a, Nicole Höttecke a, Robert Schubenel b, Karlheinz Baumann b, Boris Schmidt a,
*
a Clemens Schöpf-Institute of Chemistry and Biochemistry, Technische Universität Darmstadt, Petersenstr. 22, D-64287 Darmstadt, Germany
b F. Hoffmann-La Roche Ltd, Pharmaceuticals Division, Preclinical Research CNS, Bldg. 70/345, CH-4070 Basel, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history:
Selective lowering of Ab42 levels with small-molecule substrate targeting
c
-secretase modulators
Received 18 September 2009
Revised 6 October 2009
Accepted 7 October 2009
Available online 13 October 2009
(sGSMs), such as some non-steroidal anti-inflammatory drugs, is a promising therapeutic approach for
Alzheimer’s disease. Here we present N-substituted carbazole- and O-substituted fenofibrate-derived
sGSMs and their activity data. Seven out of 19 screened compounds exhibited promising activity against
Ab42 secretion at a low micromolar level. We presume that the sGSMs interact with lys624 at the mem-
brane interface and that the lipophilic substituents anchor the compound orientation in the membrane.
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
Alzheimer’s disease
Gamma-secretase modulator
Carprofen
Membrane anchor
c
-Secretase is one of two aspartyl proteases held responsible for
in the low micromolar range and little or no effect on the
tase cleavage at the -site. Furthermore, cell-based studies with
Ab42-lowering compounds have revealed that sGSMs do not affect
c-secre-
the generation of the Alzheimer’s disease causing pathology; amy-
loid b-peptide (Ab) aggregation to amyloid plaques. The under-
standing of the precise mechanism of Ab generation is crucial for
the development of drugs targeting the disease. Small-molecule
e
the cleavage of other
c-secretase substrates, such as Notch and
others.3,6,9–12
substrate targeting
c
-secretase modulators (sGSMs) have shown
Biotinylated, photoactivatable probes were developed to iden-
tify the target of flurbiprofen- and fenofibrate-derived sGSMs.11
Surprisingly, these photoprobes did not label the core proteins of
promise in therapeutic treatment of Alzheimer’s disease.1 These
sGSMs seem to interfere with substrate recognition/cleavage and
shift the precision of
42 to the beta-amyloid 38 site to generate more Ab38 and less
Ab42 This activity is summarized as straight modulation of
-secretase. The development of modulators will benefit from
c
-secretase cleavage from the beta-amyloid
the c-secretase complex, but instead labeled the b-amyloid pre-
cursor protein (APP), APP carboxy-terminal fragments (CTFs) and
Ab peptide in human neuroglioma H4-cells. sGSM interaction
was localized to the residues 29–36 of Ab, a region critical for
.
c
precise information on the binding site(s), but up to now this
information is rather limited.
O
O
OH
Selected NSAIDs, for example, ibuprofen, sulindac sulfide, and
flurbiprofen were found to modulate the secretion of Ab in vitro
and in vivo.2–4 Some cyclooxygenase-2 (COX-2) specific inhibitors,
for example, celecoxib, and the peroxisome proliferator-activated
Cl
O
O
Cl
N
H
N
H
O
1 Carprofen
2 Carbazole
3 Fenofibrate
receptor-
to increase Ab42 levels selectively, resulting in inverse modulation
of
-secretase activity.5,6
We recently described N-substituted carprofen derivatives and
carbozolyloxyactetic acids as
-secretase modulators (Fig. 1).7,8
c (PPAR-c) antagonist, fenofibrate (3, Fig. 1) were found
OH
O
c
O
O
OH
c
Cl
N
O
S
The introduction of a lipophilic substituent, which may vary from
arylsulfone to alkyl, turned 2-carbazolyloxyacetic acids into
O
N
decyl
c-secretase modulators (e.g., 4: IC50 (Ab42) = 7.5
lM, 5: IC50
5 BSc3041
(Ab42) = 2.9 M). The most active compounds displayed activity
l
4 BSc2912
* Corresponding author. Fax: +49 6151 163278.
E-mail address: schmidt_boris@t-online.de (B. Schmidt).
Figure 1. Carprofen (1), carbazole (2), fenofibrate (3) and the most active
derivatives BSc2912 (4) and BSc3041 (5).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.10.035

S. Baumann et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6986–6990
6987
Table 1
Compounds 8–19
aggregation. This substrate targeting by sGSMs results in alteration
of Ab42 production and inhibition of Ab aggregation.13 Substrate
labeling has been competed by other sGSMs, and labeling of an
APP c-secretase substrate was more efficient than of a Notch sub-
Entry Compds BSc
R
Substituent
number
1
2
3
4
5
6
7
8
8a
8b
8c
8d
8e
8f
3770
3853
3854
3915
3955
3956
3984
3958
R1 Octadecyl
strate. The established binding site of flurbiprofen-derived sGSMs
resides close to the membrane surface, and implies high lipophilic-
ity in combination with an acidic functional group is necessary for
binding. This combination is unusual for a therapeutic drug, but
common to amphiphilic surfactants, thus creating an obstacle for
drug development.
Fenofibrate (3) is used as lipid-regulating agent in humans. In
H4-cells, 3 raises Ab42 by over 300% and decreases Ab38 by up to
60% in a dose-dependent fashion, while Ab40 levels are not
altered.5,14 At these concentrations, fenofibrate did not show toxic
effects in the lactose dehydrogenase (LDH) and 3-(4,5-dimethyl-
thiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays.
Notably, the active metabolite of fenofibrate, the corresponding
R1 trans-Oleic acid
R1 Decacarbonyl
R1 cis-Oleic acid
R1 8-(N,N-Dimethyl)-4-sulfonylnaphthalene
R1 40-(N,N-Dimethyl)-1-sulfonylazabenzene
8g
8h
R1 40-Methylbiphenyl-3-carbonitrile
R1 3-(3-Methylphenyl)-3-(trifluoromethyl)-
3H-diazirine
9
10
11
12
13
14
15
16
17
18
19
17a
17b
17c
17d
17e
17f
17g
18a
18b
18c
19
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3989
R2 Octyl
R2 Nonyl
R2 Decyl
R2 Undecyl
R2 Dodecyl
R2 Tetradecyl
R2 Hexadecyl
R3 Octyl
fenofibric acid, did not raise Ab42 at doses up to 250 lM.
R3 Nonyl
The results with the N-substituted carprofen derivatives and
carbazolyloxyacetic acids prompted us to investigate the detailed
effects of the lipophilic substituents of N-substituted carbazolyl-
oxyacetic acids. Therefore, lipophilic substituents like trans- and
cis-oleic acid were introduced. Cis-configurated oleic acid is a com-
ponent of biological lipid bilayers and should incorporate into the
cellular membrane readily. Sterically more demanding substitu-
ents like aza-benzene, naphthalene, or biphenyl were linked to
the carbazolic backbone testing space availability. These carbazol-
yloxyacetic acid derivatives have been synthesized according to
Scheme 1 and are summarized in Table 1. One derivative (8h) fea-
tures a compact photoreactive subunit aiming to minimize the
tether used in previous crosslinking experiments. Such probes
can be useful to identify binding sites on proteins either by mass
spectrometry or via additional labeling.13 The synthesis is outlined
in Scheme 2 and described in the supplement.
R3 Decyl
R4 2,4-Dinitrophenylhydrazyl
O
NOH
CF3
NOTs
CF3
a
b
CF3
9
10
11
c
HN NH
CF3
N
N
N
N
e
d
CF3
Br
CF3
14
13
12
In addition, we tried to convert the inverse c-secretase modu-
f
lating effects of fenofibrate into straight modulation. The corre-
sponding O-alkylated oximes (Table 1) were readily accessible by
straightforward synthesis (Scheme 3, 11a–g).
O
O
O
O
The ongoing discussion on the existence of two (or even more)
binding sites in the c-secretase complex advocates for an investi-
OH
g
N
Ot-Bu
N
N
N
N
gation into the potential distance between these sites. Two modu-
lating monomer units linked by a variable spacer may form a more
active dimer interacting with the two potential binding sites. The
length of these spacers would thus provide information about
the distance between both binding sites. Fenofibrate dimers with
variable alkyl tethers offer access to such distance mapping. The
conversion of the fenofibrate with 2,4-dinitrophenylhydrazine
provided the hydrazone 19 (Scheme 3). The UV-activity and intrin-
sic color of 19 may be exploited in cellular mechanistic studies of
N
F3C
15
8h
F3C
Scheme 2. Synthesis of a carbazole-derived sGSM 8h equipped with a photoreac-
tive unit. Reagents and conditions: (a) NH2OHꢀHCl, py, EtOH, reflux; (b) TsCl, DIEA,
DMAP, CH2Cl2, 0 °C to rt; (c) NH3, Et2O, ꢁ78 °C to rt; (d) I2, MeOH, Et3 N, pH >7; (e)
NBS, AIBN, CCl4, reflux; (f) 7, KOt-Bu, THF, 0 °C to rt, 12 h; (g) TFA, CH2Cl2, 0 °C to rt,
2–8 h.
To evaluate the compounds 8a–h, 17a–g, 18a–c, 19 for their po-
tency to modulate Fb secretion, we used the Ab liquid phase elec-
trochemiluminescence (LPECL) assay to measure Ab isoforms.7 Cell
viability was measured by a colorimetric cell proliferation assay
(CellTiter 96TM AQ assay, Promega) utilizing the bioreduction of
MTS (Owen’s reagent) to formazan. The results are summarized
in Table 2 and the dose-dependent curves of the most active com-
pounds are shown in Figure 2.
c
-secretase modulation.
t-Bu
O
OH
a
O
O
O
N
H
N
H
6
7
OH
O
Introduction of a double bond by substitution with cis- and
trans-configurated oleic acid did not show any differences in
potency. Both isomers, E (8b) and Z (8d) inhibited Ab42 secretion
b, c
N
R1
with an IC50 of 13 lM. Remarkably, there is a twofold increase in
potency compared to the saturated analogue 8a (BSc3770). The
introduction of an E- or Z-double bond seems to stabilize the mem-
brane orientation of the lipophilic anchor resulting in increased po-
tency. In addition linear lipophilic alkyl chains, even amides (8c)
and sulfonamides are tolerated. We suggest that linear, lipophilic
8a−g
Scheme 1. Synthesis of carbazole-derived sGSMs. Reagents and conditions: (a) t-
butyl 2-bromoacetate, K2CO3, acetone, reflux; (b) KOt-Bu, R1X, THF, 0 °C to rt; (c)
20% TFA, CH2Cl2.

6988
S. Baumann et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6986–6990
O
Cl
O
O
O
OH
O
Cl
Cl
Cl
O
O
a
b ,c
O
O
N
O
R2
O
N
NOH
17a−g
16
3
e,c
d,c
O
O
Cl
O
OH
OH
N
O
R3
N
R4
19
O
Cl
N
O
18a−c
HO
O
Scheme 3. Synthesis of fenofibrate derived sGSMs. Reagents and conditions: (a) NH2OHꢀHCl, py, EtOH, 80 °C, 4 h; (b) NaH, R2X, DMF, 0 °C to rt, 6–12 h; (c) 2 N NaOH, MeOH,
rt, 4–12 h; (d) 2,4-dinitrophenylhydrazine, H2SO4, MeOH, rt, 2 h; (e) NaH, X–R3–X (0.5 equiv), DMF, 0 °C to rt, 6–12 h.
Table 2
Modulation and cell toxicity results for compounds 8–19
Entry
Compds
BSc number
Cell toxicity (lM)
EC50 Fb38
(lM)
IC50 Fb40
(l
M)
IC50 Fb42 (lM)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
8a
8b
8c
8d
8e
8f
8g
8h
17a
17b
17c
17d
17e
17f
17g
18a
18b
18c
19
3770
3853
3854
3915
3955
3956
3984
3958
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3989
—
1.4
9.3
12.6
9.2
>40
7.6
>80
>80
16.1
>40
>80
>40
>40
>80
>80
47.6
>40
>40
47.6
>80
>40
>40
>40
>40
>80
>80
>80
47.6
>40
>80
>40
>40
>80
>80
37.3
>40
>40
37.3
26.0
13.0
22.4
13.1
>40
9.0
>80
>80
36.2
>40
36.1
>40
>40
>80
>80
23.7
>40
>32.6
17.0
n.t.a
n.t.
—
80
n.t.
—
—
80
—
—
—
80
—
—
—
80
80
80
a
n.t. = not tested.
substituents are useful anchors, given that para-substituted ben-
zenes (8f) are modulators in contrast to meta-substituted benzenes
(8e, g, h) showing no activity at all.
is visible with all other alkylated derivatives 17b–g, however, only
at higher concentrations. Hydrazone 19 (BSc3989) inhibited total
Ab. The dimers 18a–c showed modulatory effects at high concen-
trations. Some compounds showed cellular toxicity at a concentra-
The activities of the fenofibrate derivatives have been deter-
mined at an initial concentration of 40
l
M (see Supplementary
tion of 80 lM in H4-cells. As fenofibrate is an approved drug, the in
data). As expected, fenofibrate (BSc3931) and the oxime 16 showed
inverse modulation, the fenofibrate acid (BSc3932) displayed no
activity at all (data not shown). The O-substituted compounds
17a (BSc3934) and 17c (BSc3936) displayed modulating effects.
silico parameters are established and adequate. The introduction of
the lipophilic substituent caused a dramatic increase of the clog P
value (fenofibrate: tPSA: 52.6, clog P: 5.2; fenofibrate-C8-oxime:
tPSA: 57.1, clog P: 10.2). We presume that the lipophilic anchor
of the compounds is essential for orientation within the mem-
brane. The maximum alkyl chain length should therefore not ex-
ceed the length of a natural phospholipid (see 8a–d, Table 2).
The results of the fenofibrate dimers indicate that the maximum
chain length has been reached with C12. Furthermore, our results
do not indicate an interaction with more than one binding site.
We compared the structures of highly active compounds (see
supplement of Ref. 13) to rationalize the binding mode of straight
sGSMs. Apparently, a carboxylic acid moiety is essential for potency.
At a concentration of 40 lM the Ab42 level was reduced by 60%,
while the Ab38 level was increased. The modulating effects of the
dimers 18a, b are not as strong, but still significantly stronger than
those of flurbiprofen or indomethacin (see Supplementary data),
which are benchmarks for GSM activity. The C12 dimer 18c
(BSc3943) does not show any effect on Ab secretion.
Two out of eleven fenofibrate derivatives showed modulating
effects at low concentrations with the most active being the
O-alkylated C8 derivative 17a (BSc3934). The modulating tendency

S. Baumann et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6986–6990
6989
300
200
100
0
300
a
c
b
Aβ 38
Aβ 40
Aβ 42
200
100
0
-7
-6
-5
-4
-7
-6
-5
-4
conc. / log M
conc. / log M
175
150
125
100
75
d
150
100
50
50
25
0
0
-7.0
-6.5
-6.0
-5.5
conc. / log M
-5.0
-4.5
-4.0
-7.0
-6.5
-6.0
-5.5
-5.0
-4.5
-4.0
conc. / log M
200
e
150
100
50
0
-7.0
-6.5
-6.0
-5.5
-5.0
-4.5
-4.0
conc. / log M
Figure 2. Dose–response curves for the most active N-substituted carbazolyloxyacetic acid and fenofibrate derivatives; Fb (% of control). (a) Compound 8d. (b) Compound 8c.
(c) Compound 8f. (d) Compound 17a (e) Compound 18c.
We assume that this functionality interacts with a lysine (maybe
lys624) of the substrate APP, which is located next to the
membrane interface. Thereby, the lipophilic substituents of the
sGSMs serve as membrane anchors. Binding to the substrate may
avoid substrate dimerization involving the GAIIG motif (see Sup-
plementary data), thereby driving the cleavage shift away from
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2009.10.035.
References and notes
Fb42
.
1. Schmidt, B.; Baumann, S.; Narlawar, R.; Braun, H. A.; Larbig, G. Neurodegen. Dis.
2006, 3, 290.
2. Avramovich, Y.; Amit, T.; Youdim, M. B. H. J. Biol. Chem. 2002, 277, 31466.
3. Weggen, S.; Eriksen, J. L.; Das, P.; Sagi, S. A.; Wang, R.; Pietrizik, C. U.; Findlay, K.
A.; Smith, T. E.; Murphy, M. P.; Bulter, T.; Kang, D. E.; Marquez-Sterling, N.;
Golde, T. E.; Koo, E. H. Nature 2001, 414, 212.
4. Eriksen, J. L.; Sagi, S. A.; Smith, T. E.; Weggen, S.; Das, P.; McLendon, D. C.; Ozols,
V. V.; Jessing, K. W.; Zavitz, K. H.; Koo, E. H.; Golde, T. E. J. Clin. Invest. 2003, 112,
440.
5. Kukar, T.; Murphy Michael, P.; Eriksen Jason, L.; Sagi Sarah, A.; Weggen, S.;
Smith Tawnya, E.; Ladd, T.; Khan Murad, A.; Kache, R.; Beard, J.; Dodson, M.;
Merit, S.; Ozols Victor, V.; Anastasiadis Panos, Z.; Das, P.; Fauq, A.; Koo Edward,
H.; Golde Todd, E. Nat. Med. 2005, 11, 545.
In summary, the introduction of a lipophilic substituent turned
fenofibrate- and carbazole-derived carboxylic acids into sGSMs,
which display selective reduction of Fb42 and an increase of the
less aggregatory Fb38 fragment. The most active compounds dis-
play activity on APP overexpressing cell lines in the low micromo-
lar range. Some of these sGSMs do not target the
c-secretase
complex directly, but rather bind to its substrate APP, thereby
changing the cleavage pattern. The dimerization of APP may be
influenced by this sGSM binding. The lipophilic substituents an-
chor the compound in the membrane.
6. Gasparini, L.; Rusconi, L.; Xu, H.; del Soldato, P.; Ongini, E. J. Neurochem. 2004,
88, 337.
7. Narlawar, R.; Perez Revuelta, B. I.; Baumann, K.; Schubenel, R.; Haass, C.;
Steiner, H.; Schmidt, B. Bioorg. Med. Chem. Lett. 2007, 17, 176.
8. Narlawar, R.; PerezRevuelta, B. I.; Haass, C.; Steiner, H.; Schmidt, B.; Baumann,
K. J. Med. Chem. 2006, 49, 7588.
9. Takahashi, Y.; Hayashi, I.; Tominari, Y.; Rikimaru, K.; Morohashi, Y.; Kan, T.;
Natsugari, H.; Fukuyama, T.; Tomita, T.; Iwatsubo, T. J. Biol. Chem. 2003, 278,
18664.
Acknowledgements
We thank the DFG (B.S. and S.B., SPP1085 (SCHM1012-3-1/2),
the EU (B.S. and S.B., APOPIS contract LSHM-CT-2003-503330)
BMBF (N.H., KNDD) for support of this work.

6990
S. Baumann et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6986–6990
10. Beher, D.; Clarke, E. E.; Wrigley, J. D. J.; Martin, A. C. L.; Nadin, A.; Churcher, I.;
Shearman, M. S. J. Biol. Chem. 2004, 279, 43419.
11. Morihara, T.; Chu, T.; Ubeda, O.; Beech, W.; Cole, G. M. J. Neurochem. 2002, 83,
1009.
12. Weggen, S.; Eriksen, J. L.; Sagi, S. A.; Pietrzik, C. U.; Golde, T. E.; Koo, E. H. J. Biol.
Chem. 2003, 278, 30748.
13. Kukar, T. L.; Ladd, T. B.; Bann, M. A.; Fraering, P. C.; Narlawar, R.; Maharvi, G.
M.; Healy, B.; Chapman, R.; Welzel, A. T.; Price, R. W.; Moore, B.; Rangachari, V.;
Cusack, B.; Eriksen, J.; Jansen-West, K.; Verbeeck, C.; Yager, D.; Eckman, C.; Ye,
W.; Sagi, S.; Cottrell, B. A.; Torpey, J.; Rosenberry, T. L.; Fauq, A.; Wolfe, M. S.;
Schmidt, B.; Walsh, D. M.; Koo, E. H.; Golde, T. E. Nature 2008, 453, 925.
14. Gebel, T.; Arand, M.; Oesch, F. FEBS Lett. 1992, 309, 37.