Stepwise Design of γ-Secretase Modulators with an Advanced Profile by Judicious Coordinated Structural Replacements and an Unconventional Phenyl Ring Bioisostere

Article abstract of DOI:10.1021/acs.jmedchem.0c00909

Starting from RO6800020 (1), our former γ-secretase modulator (GSM) lead compound, we utilized sequential structural replacements to improve the potency (IC50), pharmacokinetic properties including the free fraction (fraction unbound (fu)) in plasma, and in vivo efficacy. Importantly, we used novel CF3-alkoxy groups as bioisosteric replacements of a fluorinated phenyl ring and properties such as lipophilicity, solubility, metabolic stability, and free fraction could be balanced, maintaining low Pgp efflux needed for CNS penetration. In addition, by reducing aromaticity, we prevented phototoxicity. Additional substitution in the triazolopyridine core disturbed the binding to phosphatidylinositol 4-kinase, catalytic β (PIK4CB). We also introduced less lipophilic head heterocycles devoid of covalent binding (CVB) liability. After these changes, further modifications to the trifluoroethoxy bioisosteric replacement allowed rebalancing of properties, such as lipophilicity, and also potency. Our optimization strategy culminated with in vivo active RO7101556 (18B) having excellent properties and being selected as an advanced candidate.

Full text of DOI:10.1021/acs.jmedchem.0c00909

Subscriber access provided by Stony Brook University | University Libraries
Article
Stepwise Design of #-Secretase Modulators with Advanced
Profile by Judicious Coordinated Structural Replacements
and an Unconventional Phenyl Ring Bioisostere
Rosa Maria Rodriguez Sarmiento, Caterina Bissantz, Johan Bylund, Anja Limberg,
Werner Neidhart, Roland Jakob-Roetne, Lisha Wang, and Karlheinz Baumann
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.0c00909 • Publication Date (Web): 24 Jul 2020
Downloaded from pubs.acs.org on July 25, 2020
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the
course of their duties.

Page 1 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Stepwise Design of γ-Secretase Modulators with
Advanced Profile by Judicious Coordinated
Structural Replacements and an Unconventional
Phenyl Ring Bioisostere
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
1
1
1
Rosa María Rodríguez Sarmiento *, Caterina Bissantz , Johan Bylund , Anja Limberg , Werner
1
1
1
1
Neidhart , Roland Jakob-Roetne , Lisha Wang , Karlheinz Baumann
1Pharmaceutical Research and Early Development, F. Hoffmann-La Roche AG, Roche Innovation
Center Basel, Grenzacherstrasse 124, 4070 Basel, Switzerland
ABSTRACT
Starting from RO6800020 (1) our former γ-secretase modulator (GSM) lead compound, we
utilized sequential structural replacements in order to improve the potency (IC50), pharmacokinetic
properties including the free fraction (fu) in plasma, and in vivo efficacy. Importantly, we used
novel CF -alkoxy groups as bioisosteric replacements of a fluorinated phenyl ring and properties
3
such as lipophilicity, solubility, metabolic stability and free fraction could be balanced,
maintaining low Pgp efflux needed for CNS penetration. In addition, by reducing aromaticity we
prevented phototoxicity. Additional substitution in the triazolopyridine core disturbed the binding
to phosphatidylinositol 4-kinase, catalytic beta (PIK4CB). We also introduced less lipophilic head
heterocycles devoid of covalent binding (CVB) liability. After these changes, further modifications
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
to the trifluoroethoxy bioisosteric replacement allowed rebalancing of properties, such as
lipophilicity, and also potency. Our optimization strategy culminated with in vivo active
RO7101556 (18B) having excellent properties and being selected as an advanced candidate.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Keywords: γ-secretase modulator, phototoxicity, covalent binding, biososteric replacement.
INTRODUCTION
Alzheimer´s disease is a severe neurodegenerative condition that is the most common form
of dementia in the elderly, with a prevalence of 5% after 65 years of age which increases to about
3
0% in people aged 85 years or older.1-3 More than 40 million people worldwide are impacted
and, despite this, the current treatments offer only symptomatic relief with no or little affect upon
disease progression.3, 4 Patients diagnosed with Alzheimer’s disease will experience progressive
loss of cognitive function which causes social and psychiatric disability, finally affecting the
independence of the patient.5
Genetic and experimental evidence indicates that β-amyloid peptides (Aβ) are contributors
to the pathogenesis of AD.6, 7 Longer forms of these peptides, in particular Aβ42 and Aβ43, are
8
thought to be neurotoxic and more prone to aggregation , while the shorter peptides (with 38 amino
acids or less) are thought to be innocuous or even protective against formation of aggregates. We
focused on γ-secretase modulation for the treatment of Alzheimer´s Disease in order to obtain a
shift in cleavage of CTFβ (C-terminal fragment after beta cleavage) to make smaller less
amyloidogenic Aβ species and reduce the Aβ42. Aβ42 is made via the cleavage of amyloid
precursor protein in two key steps, by the action of two secretases (β and γ). The second step is the
cleavage of CTFβthrough action of γ-secretase.An imbalance between production and degradation
of these Aβ42 peptides lead to aggregation into oligomers and plaques which is believed to be the
ACS Paragon Plus Environment

Page 3 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
9
key event in the disease. GSMs bind to presenilin, the catalytic subunit of γ secretase, and shift
the production of amyloid beta protein towards non-toxic shorter amyloid peptides such as Aβ37
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
0
and Aβ38 at the expense of amyloidogenic Aβ42/Aβ40. The shorter Aβ peptides do not form
aggregates and may even function as Aβ42 aggregation inhibitors.11, 12
However, γ-secretase is an enzymatic protein complex that is involved in many biological
pathways and cleaves several substrates apart from APP, including the Notch protein.13,14
Concerns about Notch inhibition has been raised. As example, the phase III trial of Semagacestat
(
LY450139) the most advanced γ-secretase inhibitor (GSI), was terminated because of a reduction
of cognition scores and an increase in the risk of skin cancer compared to placebo. It is thought
that many of the side effects from GSIs are due to the large number of substrates cleaved by γ-
secretase and that an increase in skin cancer incidents was associated with inhibition of the γ-
secretase dependent Notch1 signaling pathway, as one of it functions is acting as a tumour
11
suppressor. Consequently, achieving APP selectivity over Notch signaling protein has become
an overarching goal in the design of next generation γ-secretase inhibitors (GSIs) but this has
proven challenging.11, 15 A more attractive alternative is the γ-secretase modulation which does not
interfere with the Notch pathway. GSMs also do not cause a rebound effect on Aβ production
which may be responsible for the worsening of the cognitive function.16
Our former GSM lead compound RO6800020 (1), was derived from several optimization
cycles from patented chemical series which were terminated due to limited in vivo activity together
17
with issues such as Pgp, GSH or drug-drug interactions. RO6800020 (1) showed improved
profile and exhibited promising in vivo efficacy in our double transgenic APP-Swedish mice Aβ42
model (Figure 2). However, RO6800020 showed a phototoxicity flag,18, 19 a covalent binding
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
0
liability and inhibited phosphatidylinositol 4-kinase catalytic beta (PIK4CB) activity. In addition,
compound 1 exhibited very high plasma protein binding (PPB) (fraction unbound in plasma (fup)
for 1 is 0.03% in human and 0.29% in mouse), which resulted in uncertainties regarding the
predictions of human exposure needed for efficacy, also due to the difficulties to measure such
low fu values with accuracy. Considering the potential differences in fu between mouse and
humans, a total plasma concentration above 20 µM was anticipated to be required in order to obtain
efficacy in humans, increasing our safety concerns. Moreover, the observed high PPB
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
1
consequently led to uncertainty regarding the interpretation of in vitro assays results , especially
for the primary potency assay and the hepatocyte metabolic stability assay where 10% fetal calf
serum (FCS) was utilized in the cellular media.
Capitalizing on our internal potent GSM series and starting from our most profiled lead
compound RO6800020 (1), we utilized sequential structural replacements to improve desired
properties and potency (Figure 1). We evaluated many central piperidine replacements such as
bicyclic, bridged, or azaspiro-like piperidines (data not shown). We also revisited our previously
22
reported bridged piperidines as replacement, observing that in particular with the [3.2.1] bridged
piperidine (2), we improved the overall potency, reducing lipophilicity (logD = 4.1) while
maintaining the apparent metabolic stability. Our new GSM (2) had higher potency which resulted
in good in vivo efficacy with lower efficacious free concentrations. Also higher and more reliable
fraction unbound (fu) measurements (fup for 2 in human and mouse is 0.46% and 0.3%). This
lowered the estimated human efficacious doses (Figure 2) and allowed better predictions.
Furthermore, in order to understand the impact of potential unspecific binding for the interpretation
of potency, we measured fu in the cellular assay media for a number of compounds and calculated
their free in vitro IC50 values. Subsequently, for compounds with measurable PPBs, we established
ACS Paragon Plus Environment

Page 5 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a clear correlation between the free in vitro IC50 and free brain IC50 (Figure 2), where the free brain
IC50 was estimated by PKPD analysis corrected for fu plasma and a potential brain gradient based
on observed mouse Pgp efflux ratios.23
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Importantly, in order to eliminate the hyperconjugated system and prevent phototoxicity,
we explored unconventional bioisosteric replacements of the phenyl ring at the core and identified
fluorinated alkoxy groups as potent and non classical bioisosteres for the terminal phenyl ring (5A)
(Figure 1). Using an alkoxy chain moiety as a bioisosteric replacement, we reduced the aromaticity
and consequently, we also decreased the absorption in the UVA region (320–400ꢀnm), preventing
phototoxicity while maintaining potency. The reduction of aromaticity (or the increase of the
fraction of sp3-hybridized carbon atom count) reduced lipophilicity and protein binding and
increased solubility. We also introduced an additional substituent in the amino-triazole core (9B)
to disturb the observed binding (Kd~100 μM) to PIK4CB kinase, improving selectivity. Finally,
we reevaluated the heteroaromatic head groups that we found responsible for the observed covalent
binding (CVB) and introduced new heterocycles with no CVB safety liability (18B).
The CF alkoxy bioisosteric replacement of the phenyl ring dramatically affects compound
3
properties, while also maintaining an excellent metabolic stability and low Pgp efflux. Further
modifications on the trifluoro alkoxy moiety are readily accessible. As a consequence, we could
introduce small changes on the substitution pattern of the alkoxy chain, in order to re-balance
properties such as lipophilicity and potency, in agreement with the core changes and with the
different characteristics of the different heterocyclic head groups introduced to eliminate the
compounds CVB liability (18B) (Figure 1).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
5A
9B
18B
Figure 1. Sequential improvement of potency and properties for 1. Increase of potency and free
fraction (2). Improvement on physicochemical properties and elimination of phototox liability
upon introduction of a phenyl ring bioisostere (5A). Reduction of PIK4CB side activity by
disturbing the kinase hinge binding (9B). Elimination of covalent binding liability with new
heterocyclic head rings and properties-potency rebalance with modifications on the alkoxy
bioisosteric replacement (18B).
RESULTS AND DISCUSSION
Lead optimization. We aimed to obtain an orally available GSM without covalent binding
and phototoxicity. Our most profiled GSM compound 1 has moderate potency (IC50 (Aβ42) =
5
5nM), high lipophicity, very low human plasma free fraction (fu: 0.03%) and low solubility.
Lead compound 1 also showed significant metabolically induced covalent binding (in the
1
4
presence of NADPH, CVB of C-RO6800020 in human liver microsomes (HLM) = 149 pmoles
bound per mg protein) indicating a risk for idiosyncratic toxicity and drug bioactivation via
formation of reactive metabolites.24, 25 Therefore, we decided to explore alternative central
piperidine rings to see if the difluoro piperidine could be involved in the formation of a reactive
ACS Paragon Plus Environment

Page 7 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
metabolite that could react with GSH, may be via enamine formation. By exploring replacements
for the central ring (variously bridged, bicyclic or azaspiro piperidine analogs) we realized that the
best replacement was the reported [3.2.1] bridged piperidine,22 as it did not compromise the
metabilic stability and gave an excellent improvement in potency (2). Compound 2 has improved
potency along with decreased protein binding (hfu: 0.4%) compared to compound 1 (hfu: 0.03%)
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
table, Figure 2). This contributed to a reduction of the estimated efficacious human dose for
compound 2 (7-25 mg/day), as well as an increased confidence in human predictions due to
reduction of unspecific binding on in vitro assays. Furthermore, compound 2 despite having similar
14
covalent binding (105 pmol/mg, measured with the radiolabeled C compound) as compound 1,
shows lower risk for idiosyncratic toxicity, due to the lower predicted human dose (7- 25mg/kg).24
*
Predictions based on in vitro-in vivo extrapolation observed in rodents (correlated for free fractions in hepatocytes incubations
and plasma) and potential differences in potency and fu plasma between human and mouse
*
* Predictions based on the assumption that a steady state plasma concentration of IC50 was needed for sufficient efficacy
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 2. Correlation between estimated free in vitro IC50 (Aβ42) and in vivo free brain IC50
concentrations for 5A and compounds from the same chemical series (side A). Effects of potency
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
IC50) and predicted human pharmacokinetic properties improvement, with a more balanced
plasma protein binding, lowering human efficacious doses (side B, table). Human/mice free IC50
predictions and initial human dose estimations for 50% of Aβ42 reduction at steady state for
compounds 1, 2 (table).
We observed that alternatively bridged (e.g [3.3.1]) and non bridged (e.g bicyclic or aza-
spiro) piperidine analogs, are less potent than the [3.2.1] bridged piperidines (2) on binding to
γ−secretase. The particular potency improvement for the [3.2.1] bridged piperidine series
compared to other replacements including the difluoro piperidines, can be caused by a most
favorable conformational change due to the repulsion between the bridge and the lone electron
pair from the piperidine nitrogen that enforces an almost linear orientation of the two exit vectors
from the piperidine, enabling an optimal binding to γ−secretase (Figure 3).
Figure 3. Conformational differences between compound 1 and 2.
ACS Paragon Plus Environment

Page 9 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Unfortunately, with respect to safety, compound 2 showed in vitro phototoxicity (Figure
) (cytotoxicity after UV irradiation: IC50 UVA: 1.67μM) which was confirmed also in vivo.
4
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. In vitro phototoxicity for compound 2 measured with cells exposed to the compound in
the presence (+Irr) or absence (-Irr) of UVA light. The phototoxic potential (indicated by high
UVA absorption) is expressed through the use of the Photoirritancy Factor (PIF) by comparing the
IC50 +Irr to the IC50 –Irr. PIF >5 indicates phototoxicity.19
We realized that besides compound 2, also compound 1, has a phototoxicity safety risk
due to the high UVA absorption. For this reason, we wanted to reduce the aromaticity decreasing
the phototoxic risk from the hyper-conjugated system. In this regard, we successfully exchanged
the terminal fluorinated phenyl ring with an alternative fluorinated alkoxy chain that behaves as a
2
6
bioisosteric structural moiety. After evaluation of fluorinated alkoxy groups, the trifluoroethoxy
was the most suitable as a non cyclic bioisosteric replacement (Figure 5). The increase of sp3-
hybridized carbon atoms count has a positive impact on physicochemical properties decreasing
lipophilicity (increasing fu), improving solubility and human metabolic stability while maintaining
low P-glycoprotein efflux (compounds 2 and 5A, Table 1).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Moreover, the reduced aromaticity (as illustrated by reduced UVA absorption) abolished
phototoxicity (Figure 5).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
F
F
CF3
O
H
F
S
N
N
H
N
S
N
N
N
N
N
A
N
N
N
N
N
2
5Aor 9A
A(A= N), 9A(A= C)
5
Compound 2
Compound 9A
Figure 5. Fluoro phenyl ring bioisostere. Reduced UV absorption, resulting in no phototoxicity
risk.
Further exchange of the heterocyclic head group influences the affinity at the target while
maintaining the improvement on the mentioned properties (5A vs 9A, Table 1).
Compounds 5A and also 9A have improved solubility, lipophilicity, human metabolic
stability as well as a higher fraction unbound in plasma compared with 2. In addition, they do not
have phototoxicity (Figure 5, Table 1) (IC50 UVA >100 μM, PIF <5). Both compounds have also
very good metabolic stability (low Cl in hepatocytes), low Pgp efflux and high potencies as GSMs,
in particular 9A (with the isothiazole head group with hIC50 Aβ42: 4 nM) (Table 1). The novel
fluoro phenyl ring bioisostere offers promising in vitro activity, additionally better
ACS Paragon Plus Environment

Page 11 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
physicochemical properties and an improved pharmacokinetic profile, which translates into a
strong in vivo effect on Aβ42 reduction (Table 1).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 1. The alkoxy bioisosteric replacement of the phenyl ring affects compound properties.
Compound
2
5A (A=N)
21/12
9A (A=C)
4/3
a
Human IC50 Aβ42 total /free [nM]
4.5/1
4/0.9
Mouse IC50 total/free [nM]
9/5
4/3
In vivo remaining % Aβ42
56-27-22
59-33-15
75-47-*
@
1-3-10 mg/kg p.o.
logD b
4.14
3
3.17
198
3.9
c
Solubility [μg/mL]
F.u. (h/m): [%] d
31
0.4/0.3
6.6/5.4
1.59/1.72
1.67/23
<2/<1
2.8/ 10.1
1.3/4
1.7/1.8
3.3/15
1.0/1.35
>137/1.0
2.9/>0.9
e
Cl heps (h/m): [μL/min/Mcells]
Pgp h/m f
1.3/1.5
443/ 1.1
>10/4.5
IC50 UVA [μM]/ PIF
hERG IC50 / IC20 μM
aIC50 Aβ42 (measured in cell assay at least in 3 independent experiments), free IC50 is
b
calculated based on the measured fu in the cell assay media containing (10% FCS); logD =
c
logarithm of the octanol-water distribution coefficient @ pH 7.4; Solubility expressed in μg/mL
d
and measured in Lysa assay; measured % of free fraction unbound in mouse and human;
e
f
Clearance in human hepatocytes expressed in μL/min/million cells; Pgp efflux. *Not measured.
We observed that for the new alkoxy series, the potency was improved when the thiazole
group was replaced by a more lipophilic isothiazole group (5A, 8A vs 9A, 10A, Table 2). A certain
size of the R alkoxy chain is needed for potency (4A vs 5A, 6A and 7A). The shorter methoxy
3A) group was not a good replacement being the less potent, followed by the OCF (4A) group.
head group, the lipophilicity
1
(
3
With the isothiazole instead of a thiazole as the heterocyclic R
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
increased and the solubility is reduced (e.g 5A vs 9A and 8A vs 10A). Further elongations of the
1
fluorinated alkoxy chains O-R with the isothiazole as a head group (e.g as 7A) are not interesting
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
due to high lipophilicity (logD >3.9) and very low solubility (solubility < 1 μg/mL). Additions of
1
hydrophilic R
1
substitutions on the OR alkoxy chain (such as the oxetane in 11A) improves
solubility however, it tends to reduce the potency (Table 2).
Table 2: Comparison between different alkoxy replacements with thiazole and isothiazole
heterocyclic head groups
*
IC50
Lysac
Fupd
μg/mL] h/m [%]
Clhepe
h/m
PIK4CBf
Kd (μM)
1
R2
Aβ42
ha
logDb
Compd
R O
[
[nM]
3A
4A
5A
231
2.59
3.3
396
102
198
12/16
-/-
10/27
-/-
61
21
3.17
2.8/6.0 1.3/6.4
0.150
-
6
A
14
14
3.36
3.76
1.7
0.2
3/5.21
1.3/21
7
A
A
-
-
--
chiral
16 (27
enant.)
8
3.69
3.90
78
31
2.8//6
3.1/10
9A
4
1.7/1.8 3.3/15
0.120
ACS Paragon Plus Environment

Page 13 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
chiral
6
(7
1
0A
3.81
2.26
6.3
1/3.4
-/-
3.1/36
-/-
0.085
-
enant.)
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
11A
49
456
ahuman IC50 Aβ42 (measured in cell assay at least in 3 independent experiments); logD =
b
c
logarithm of the octanol-water distribution coefficient @ pH 7.4; Solubility expressed in μg/mL
d
and measured in Lysa assay; measured % of free fraction unbound in mouse and human;
e
f
Clearance in human hepatocytes expressed in μL/min/million cells; Kd measured for PIK4CB
phosphatidylinositol 4-kinase, catalytic beta. *IC50 values are suitable for SAR comparison with
compounds exhibiting trustable fu plasma > 0.5%.
Prevention of kinase inhibition.
We realized that some of our more interesting lead candidates such as compounds 5A and 9A,
had side activities against PIK4CB (Table 2). PIK4CB was previously implicated in hepatitis C
virus (HCV) replication. However, in vitro profiling of inhibitor compounds revealed a role of this
kinase in lymphocyte proliferation.20 Therefore, we also tested compound 2 which showed
PIK4CB binding Kd of 120 nM. Although for 9A, or even 5A, we were less concerned about
reaching required selectivity due to their high in vitro potency as modulators, we needed to
eliminate kinase selectivity concerns. We decided to analyze the binding mode for our compounds
by docking compound 2 into the PIK4CB binding site. In fact, the core structure fits well into the
ATP binding site and we observed a possible hinge binding by the amino-triazole moiety (Figure
6
).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Hinge region
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Compound 2
PIK4CB Kd [μM] 0.12
Figure 6. Hinge binding to kinase PIK4CB from the amino-triazolo pyridine.
With the 5-(trifluoromethyl)-8-trifluoroethoxy 2-aminotriazolo pyridine analog 9B we were able
to prevent the binding at the unwanted kinase. We postulated using docking (for 9A and 9B) into
the PIK4CB binding site, that the additional 5-CF3 group disrupts the hinge binding of the
aminotriazolo pyridine moiety. All 5-trifluoromethyl 8-alkoxy aminotriazole pyridine analogs
(
9B) showed high potency as GSMs and no detectable kinase activity (Figure 7).
Hinge region
H
N
S
R
N
N
N
N
N
CF3
O
R = H R = CF
9A
3
Compound
9
B
IC50 Aβ42 [nM]
PIK4CB Kd [μM] 0.12
4
2
>40
Figure 7. Disruption of the hinge binding site by substitution on the 5 position of the 8-alkoxy
aminotriazole pyridine ring.
Compounds 5A and 9A (Figure 5) (with thiazole and isothiazole head groups) have excellent
molecular properties for brain penetration including high solubility, acceptable lipophilicity and
good metabolic stability, together with high in vitro and in vivo potency (compound 9A showed in
vivo 47% remaining A
β
42 @ 3mg/Kg, compound 5A showed 33% remaining Aβ42, see Table 2).
ACS Paragon Plus Environment

Page 15 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
For this reason, they were chosen for further profiling including covalent binding studies with
radiolabeled material as we knew there was a CVB liability for compounds 1 and 2, with a thiazole
as heterocyclic head group. Compound 5A showed only borderline covalent binding CVB: 30
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
pmoL/mg protein). Compound 9A with the isothiazole as head group, showed higher reactivity
(CVB: 96 (pmoL/mg protein)). For these two particular analogs, we were less concerned due to
the significant in vitro potency (hIC50 Aβ42 [nM] 5A: 21 nM (total), 12 nM (free); hIC50 Aβ42
[nM] 9A: 4 nM (total), 3 nM (free)) including promising pharmacokinetics with acceptable free
exposures that translated into a high in vivo efficacy (>50% Aβ42 reduction at 3 and 10 mg/kg p.o.
as indicated in Table 1). Moreover as the CVB liability is related to the efficacious human dose
(which is related to potency and fraction unbound), with compounds having <10-20mg/Kg human
2
4
dose being of no concern. However, it seemed likely that the undesirable compound bioactivation
via formation of GSH adducts, was due to formation of reactive metabolites through reaction at
the thiadiazole and isothiazole heterocyclic rings (manuscript studying the mechanism under
preparation). Therefore, as we wanted to have advanced candidates without a CVB flag, we
explored new heterocyclic “head” groups with lower lipophilicity and lower reactivity, compared
to the thiazole and isothiazole rings.
We used the 8 alkoxy-2-aminotriazolo pyridine core substituted and the non substituted
derivative with a CF group at position 5 (cores A or B, Figure 8).
3
2
R : explored selection of heterocycles in order of hydrophobicity
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
A
B
Figure 8. Explored thiazole and isothiazole heterocyclic replacements according to decreasing
lipophilicity (or hydrophobicity) for cores A and B.
Analogs of compound 9A using the most hydrophilic heteroaromatic replacements such as
1
,3,4 or 1,2,4 oxadiazoles (12A and 14A) lost potency (Table 3). However, they did not have a
CVB flag. We observed that when we increased the compound lipophilicity with heterocyclic head
rings such as pyridine or pyrimidine devoid of CVB safety liability, we increased in vitro potency
(13A). By replacing the isothiazole 9A for more hydrophilic heterocyclic rings R
2
(12A, 14A) we
needed to balance again the compounds potency by modifications on the bioisosteric replacement
(
(
15A vs 12A). By introducing more hydrophobic head groups (Figure 8) we improved potency
13A or 16A vs 12A or 15A, Table 3), to the detriment of solubility and lipophilicity. In this regard,
we revisited a diverse combination of heterocyclic rings R
substitutions OR for the different cores A or B type, in order to rebalance the compound properties
and achieve desired potency ranges. A selection of key representative analogs is shown in Table
2
and aminotriazole pyridine alkoxy
1
3
. For compounds with core type B, more hydrophilic head groups are admitted (12B, 14B vs 12A,
1
4A) with solubility and lipophilicity maintained in very good ranges. To our satisfaction, all
compounds without a thiazole and isothiazole as heterocyclic head groups, were devoid of CVB
flag.
Table 3: Assessment of key heterocyclic head rings R
2
and fluorinated phenyl ring bioisosteric
replacements OR for the different cores A or B type.
1
ACS Paragon Plus Environment

Page 17 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
A
B
IC50Aβ42
Lysa
Fup
Clhep
h/m
PIK4CB
Kd (μM)
1
R2
Compd
R O
logD
h/mouse
[
μg/mL] h/m [%]
[
nM]
9
A
4/4
3.99
31
265
11
1.7/1.8 3.3/15
0.120
1
1
2A
3A
114/>400 1.79
-/9.5
1.2/2.0
-/-
-/4.6
1/9
-/-
-
0.250
-
24/21
108/-
4.04
2.68
14A
346
1
1
5A
6A
20/17
3.57
144
5.5/6.1
1/11
-
7
/3
4.1
4.1
11
0.6
55
77
2/3.2
0.2/-
4/34
2/19
1/8.5
1/1.6
-
9
B
2/2
>40
>40
>40
1
2B
33 /20
27/20
3.79
3.58
1.7/3.1
2.8/4.4
14B
17B
21/11
3.9
3.8
88
82
2.8/1.7 2.3/9.4
23
chiral
1
8B
30 /21
5/4.3
1.1/3.4
>40
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
As we postulated, with the 5-trifluoromethyl 8-alkoxy aminotriazole pyridine core B
analogs we obtained the desired kinase selectivity. In addition, we maintained high potencies and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
acceptable compound lipophilicity, by using the most hydrophilic heterocyclic head groups R
2
such as the oxadiazoles (12B, 14B, 18B) and the pyrimidine (17B) that are beneficial for solubility,
do not have CVB, and offer a good metabolic stability (Table 3).
Three compounds 12B, 17B and 18B with good overall properties and acceptable in vitro
total/free IC50 were selected for further characterization (Table 4). Finally compound 18B, was
selected as a promising candidate for minitox studies in rats. The three compounds and in particular
18B with a lower hERG IC50, do not have any safety flags and they show a very promising in vitro
profile.
Table 4. In vitro profile of selected compounds
1
2B
17B
(RO7103286)
18B
(RO7101556)
Compound
(
RO7077068)
h IC50 Aβ42 total /free [nM]
Mouse IC50 total / free [nM]
h Notch IC50: [nM]
PIK4CB Kd [μM]
33 (free 21)
21 (free 12)
11 (free 7)
> 10000
23
30 (free 19)
21 (free 14)
> 10000
>40
20 (free 12)
> 10000
>40
Solubility [μg/mL]
logD
55
88
82
3.79
3.90
3.85
Thesa/Fassif/Fessif (mg/mL)
P-gp ER (h/m):
5, 12, 92
1.43 / 1.62
3, 13, 466
2.08 / 3.92
85, 234, 390
1.75 / 2.16
ACS Paragon Plus Environment

Page 19 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
F.u. (h/r/m): [%]
Cl heps (h/r/m): [μL/min/Mcells]
hERG IC20 / IC50 [μM]
Phototox IC50 UVA [μM]
GSH (CVB potential)
Ames / MNT
1.7 /4.6 / 3.1
1 / 6.9 / 8.5
3.5/ 8.75
33
2.8 / 1.0/ 1.7
2.3 / 5.8 / 9.4
1.4/ 4.5
5.0 / 5.9 / 4.3
1.1/6.3/3.4
5.4/>20
25
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
31
clean
clean
clean
clean / clean
clean
clean / clean
clean
clean / clean
clean
GSH
EST
No flag
No flag
No flag
We evaluated 18B in a pharmacodynamic (PD) mice model to determine its efficacy
using the double transgenic APP-Swedish mice. The compound was administered orally at single
doses of 10 mg/Kg and 20 mg/Kg which led to strong effects on Aβ42 reduction (Figure 9, with
Aβ42 % control in the graphic) in a time and dose dependent manner. An in vivo IC50 of 40 nM
was determined fitting the time course of exposures and the Aβ42 levels (turnover model).
1
00
50
0
2
0 mg/kg
0
10
20
time after dosing (h)
Figure 9. Time course effect of 18B on Aβ42 reduction (expressed as % of control) in APP swe
tg mouse. Dose and time dependent effect.
In a single dose pharmacokinetic study (SDPK) in rodents and minipig (species considered
for minitox) 18B shows acceptable oral bioavailability (F% = 33%) and good profile (such as low
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
clearance and suitable half life) for a single daily dose oral administration with a promising
prediction (low clearance) in humans (Table 4). The detailed outcome of the minitox studies will
be the subject of a dedicated subsequent paper.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 5. In vivo SDPK profile of RO7101556 (18B) in rat / minipig
Rata
Minipigb
AUCinf_D (po) (h*kg*ng/ml/mg)
Cl (iv) (ml/min/kg)
1748
3.8
8.1
2.2
33
1383
3.35
Terminal t1/2 (iv) (h)
Vss (iv) (l/kg)
F (%)
6
.7
0.83
27
a
b
IV (bolus) 1 mg/kg and PO (gavage) 2 mg/kg; n=2/group (M); sampling over 24h; IV
(
bolus) 1 mg/kg and PO (gavage) 3 mg/kg; n=3/group (M); sampling over 240h.
With respect to the mode of action, 18B behaves as a potent GSM decreasing the amount
of large Aβ40/42 peptides prone to aggregation and increasing the ratio of the shorter Aβ38
peptide, which as mentioned, are thought to be protective against formation of aggregates.
Furthermore, we confirmed that 18B behaves as a GSM having a clear differentiation with respect
to GSIs as it does not inhibit the Notch pathway (Figure 10).
In vitro effect of RO7101556 on the secretion
of the various Aβ species
In vitro effect of RO7101556 on Aβ42 secretion and
Notch processing in human cells
ACS Paragon Plus Environment

Page 21 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 10: Mechanism of action of 18B as a GSM
Chemistry.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
In order to determine the influence of different fluorinated alkoxy chains OR , as phenyl ring
bioisosteres (Scheme 1) we explored final alkylation of an advanced common intermediate (24,
1
Scheme 1). Initially for most alkoxy bioisosteric replacements we could introduce the R group at
the beginning of the synthesis via SN
Bechamp reduction of nitro alkoxy compound using e.g. Fe and HCl or through hydrogenation
with Pd/C catalyst and H to obtain the amine 21. However for upscaling and exploration of diverse
OR1 alkoxy chains, we decided to use a modular synthesis (with OR1a= OCH ) starting with
2
alkylation or Mitsunobu to produce (20), followed by the
2
3
commercially available 20 and using the methyl ether as a protecting group that could be removed
from a common precursor (23-1), to prepare a variety of alkoxy analogs via alkylation of the
advanced intermediate 24. These alternatives provides an accessible route to a big range of alkoxy
triazolo pyridine analogues. For OR1=OCH
available) with ethoxy carbonyl isothiocyanate, gave a suitable precursor 21a for synthesis of a
,2,4-triazole (22) via cyclisation through a concerted mechanism with the elimination of CO
3
, OCF
3
the reaction between amine 21 (commercially
1
2
using ammonia as the nitrogen source in the presence of a base such as diisopropylethylamine.27
The Sandmeyer reaction was used to transform the aminotriazole 22 into a bromotriazole (23).
Removal of the methyl group using BBr from 23-1 gives access to the free bromo triazolo alcohol
3
intermediate 24, which was alkylated to give a large range of fluoro alkoxy intermediates allowing
the exploration of diverse bioisosteric replacements 25-3 to 25-8 (Scheme 1). Alternatively, the
thiazole, isothiazole (26, 28) or a different heterocyclic BOC protected amino bridged piperidine
(30, 32, 34, 36, 38, Scheme 2) were prepared via alkylation, reaction with intermediate 4-methyl-
6
-thioxo-3,6-dihydro-2H-1,3-thiazin-2-one or via Pd or Cu catalyzed couplings with the iodo,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
chloro or bromo heterocycle and the BOC protected amino bridged-piperidine (commercially
4b
available CAS 847862-26-4, or initially prepared as outlined in our patent ). After BOC removal
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
under acidic conditions, the second intermediate II was obtained (Scheme 2).
Buchwald-Hartwig coupling of the corresponding 2-bromo-[1,2,4]triazolo, 8-alkoxy pyridine
intermediates IA (or cores 23, 25) with the heterocyclic amino bridged piperidines (intermediates
II) gave the final compounds 3A-16A in high yield (Scheme 3).
Scheme 1. Synthesis of core 23, 25 (Intermediate IA): a
R1a=
2
0-2
a
2
0-1
c
b
1
9
20
21a
22
2
2
1-1 (R1a=OCH3)
1-2 (R1a=OCF3)
22-1 (R1a=OCH3)
22-2 (R1a=OCF3)
2
1a-1 (R1a=OCH3)
21a-2 (R1a=OCF3)
d
f
e
R1a=
23
2
5
25-3
2
4
2
3-1
R1b=
25-4
23-1 (R1a=OCH3)
23-2 (R1a=OCF3)
2
5-5
2
5-8
2
5-7
25-6
aReagents and conditions: a) Fe/HCl, EtOH, 90°C, overnight, 60%; b) EtO
CNCS (1.1 equiv),
NEt (3 equiv),
2
.
dioxane, 3 hrs, r.t., 96% (21a-1), 105% (21a-2); c) NH
2
OH HCl (5 equiv), i-Pr
2
MeOH/EtOH (1/1), room temperature to 60°C, 4 hrs, 98.2% (22-1), 77.4% (22-2); d) tert-butyl
nitrite (1.5 equiv), CuBr
2
(1.5 equiv), acetonitrile, 75°C, 2 h, 62% (23-1), 77.3% (23-2) ; e) BBr
3
1
(5.5 equiv), CH
Cl
2 2
, 0°C-60°C, 81%; f) R I (1-2 equiv), K
2
CO (2-3 equiv), DMF, 120°C,
3
ACS Paragon Plus Environment

Page 23 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
overnight, 77.4% (25-3), 90.3% (25-4), 86.8% (25-5), 39.6% (25-7), 90.1% (25-8) or 1,1,1-
trifluoropropan-2-yl 1,1,2,2,3,3,4,4,4-nonafluorobutane-1-sulfonate (1.5 equiv), K
2
CO (2 equiv),
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
DMF, 120°C, overnight, 53.4% (25-6).
Scheme 2: Preparation of heterocyclic bridged amino piperidines (Intermediates II): a
b
a
2
7
2
6
c
d
b
e
2
9
2
8
CAS (847795-98-6)
31
3
0
R2=
g
f
3
2
34
3
8
3
6
3
3
37
39
R2=
h
3
3, 35, 37, 39
3
2, 34, 36, 38
35
aReagents and conditions: a) 5-chloro-3-methyl-1,2,4-thiadiazole (1.2 equiv), Et
EtOH, 80°C, overnight, 92% ; b) TFA (10 equiv), CH Cl , r.t., overnight, 94% (27) or HCl (10
3
N (5 equiv),
2
2
equiv), r.t., 2 hrs, 55.3% (29); c) i) 4-methyl-6-thioxo-3,6-dihydro-2H-1,3-thiazin-2-one (1eq), 4-
methylmorpholine (3 equiv), 4-dimethylaminopyridine (0.01eq), dioxane, 80°C, overnight, ii)
DIPEA (4 equiv), I
2
(2 equiv) in dioxane, from 0°C-r.t overnight, 72%; d) BrCN (1.4 equiv),
Na CO (1.4 equiv), EtOH, 2 hrs r.t. to 50°C overnight, 92.5%; e) (z)-N´-hydroxyacetimidamide
2
3
(
1equiv), ZnCl (1.2 equiv), EtOH then HCl (3 equiv), 60°C, 7 hrs, 97%; f) 4-chloro-2-
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
t
(
trifluoromethyl)pyridine (1equiv), Pd(II)OAc (0.08 equiv), Na BuO (1.5 equiv), Hunig´s base (2
equiv), 2-(dicyclohexylphosphino)biphenyl (0.16 equiv), dioxane, microwave 140°C, 45 min,
6.3% (32) or 5-iodo-3-methylisoxazole (1.2 equiv), CuI (0.1 equiv), K PO (2 equiv), N,N-
diethylsalicymide (0.3 equiv), DMF, -5°C, 6 hrs, 22% (36); g) 2-bromo-5-methyl-1,3,4-
oxadiazole (2 equiv), Et N (4 equiv), EtOH, 130°C, overnight, 68.1% (34) or 4-chloro-6-
methylpyrimidine (1.2 equiv), Et N (4 equiv), EtOH, 100°C, overnight, 99% (38); h) HCl (6 or 10
equiv), CH Cl , r.t., overnight, 91.9% (33), 80% (37), 94.2% (39) or TFA (10 equiv) CH
overnight, 80.2% (35).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
3
4
3
3
2
2
2
Cl
2
, r.t.,
Scheme 3: Final coupling step for intermediates IA and II for compounds 3A-16A: a
R1= OCH3
R1=OCF3
R1=OCH2CF3
3A
4A
5A
6A
R1=O(CH2) CF3
2
R1=O(CH2) CF3
7A
8A
3
R1=OCH(CH3)CF3
R1=OCH2CF3
R1=OCH(CH3)CF3(S) 10A
R1=OCH2-oxetane
9A
Intermediate II
11A
a
23 or 25
R1=OCH2CF3
R1=OCH2CF2CF3
12A
15A
R2=
37
2
7
2
9
R2=
Intermediate IA
33
R1= OCH3
R1=OCF3
R1=OCH2CF3
23-1
R1=OCH(CH3)CF3 13A
R1=OCH(CH3)CF3 14A
35
3
1
23-2
25-3
25-4
R1=O(CH2) CF3
2
R1=O(CH2) CF3
25-5
3
R1=OCH(CH3)CF3 25-6
R1=OCH2CF2CF3 25-7
R1=OCH2-oxetane 25-8
R1=OCH2CF2CF3
16A
aReagents and conditions: a) intermediate II (1equiv), sodium phenoxide (1.1-1.6 equiv)
(dba) .
CHCl (0.08 equiv), dioxane, 130-145°C,
or NaOtBu (3 equiv), xantphos (0.16 equiv), Pd
2
3
3
4
5-60 min in microwave, 54% (3A), 80.6% (4A), 66.9% (5A), 51.9% (6A), 63% (7A), 80,1%
(
8A), 60.6% (9A), 65.3% (10A), 81.2% (11A), 59.8% (13A), 63.4% (14A), or intermediate II (1.1
ACS Paragon Plus Environment

Page 25 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
equiv), NaOtBu (2.05 equiv), bromo-(tri-tert-butyl)-phosphine-palladium (I) dimer (0.15 equiv),
dioxane, 110-120°C, 1-2 hrs, 43% (12A), 11% (15A), 27.3% (16A).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
-Bromo-5-substituted-8-alkoxy triazolo-pyridines cores (Intermediates IB from
compounds 9B, 12B, 14B, 17B and 18B), can be prepared as described in Scheme 4 from 3-
bromo-(trifluoromethyl)pyridin-2-amine 41 by following similar synthetic steps as described in
Scheme 1. Compound 41 is obtained by bromination of the 6-(trifluoromethyl) pyridin-2-amine
precursor 40. Intermediate 8-bromo-5-(trifluoromethyl)-[1,2,4]triazolo[1,5-a]pyridin-2-amine
(
43) undergoes Ullmann coupling with a range of OR groups to produce several alkoxy
1
derivatives (44-1, 44-2 and 44-3). Sandmeyer reaction provides the 2-bromo-5-substituted-8
alkoxy triazolo-pyridine in high yields (45).
Similarly as described for intermediate IA, Buchwald-Hartwig coupling of the
corresponding intermediate IB (45) with the heterocyclic amino bridged piperidines (intermediates
II in Scheme 2) give access to compounds 9B, 12B, 14B, 17B, 18B (Scheme 5).
The enantiomerically pure compounds, such as compound 18B, coul be obtained from the racemic
precursor by chromatographic separation using a chiral column. The absolute configuration of the
different enantiomers was determined by using the chiral alkylating agents (step d, Scheme 4) to
obtain the intermediate 44 as single enantiomers and comparing them with the different
enantiomers obtained from chiral separation.
Scheme 4. Synthesis of core B (Intermediates IB): a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
b
c
d
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
R1=
4
0
41
42
43
44
4
4-2
44-1
e
4
4-3
R1=
45-1
45-2
45
45-3
aReagents and conditions: a) Br
CNCS (1.1 equiv), dioxane, 100°C, 5 h, 76.3%; c) NH
(1equiv), CH
Cl
2
, room temperature, overnight, 44%; b)
2
2
.
EtO
2
2
OH HCl (5 equiv), i-Pr
2
NEt (3 equiv),
1
MeOH/EtOH (1/1), room temperature 1h to 60°C, 3 h, 69.1%; d) R OH (1.5 equiv), NaH (1.3
equiv), CuBr (0.01 equiv), DMF, 155°C, 4h, 61.2% (44-1), 65.5% (44-2), 75.2% (44-3); e) tert-
butyl nitrite (1.5 equiv), CuBr
), 89.6% (45-3).
2
(1.5 equiv), acetonitrile, 60°C-75°C, 6h, 82.9% (45-1), 85.5% (45-
2
Scheme 5: Final coupling step for intermediates IB and II for 9B, 12B, 14B, 17B, 18B: a
R1= OCH2CF3
9B
Intermediate II
a
R1= OCH2CF3
12B
14B
R2=
4
5
R1= OCH(CH3)CF3
R2=
31
R1= OCH2CF3,
R1= OCH(CH3)CF3 45-2
R1=OCH2CHF2
45-3
45-1
29
R1= OCH(CH3)CF3 (S) 18B
3
5
R1= OCH2CHF2
17B
3
9
ACS Paragon Plus Environment

Page 27 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
aReagents and conditions: a) intermediate II (1.1 equiv), NaOtBu (2-3 equiv), xantphos
(
(
(
0.16 equiv), Pd
2
(dba) .
3
CHCl (0.08 equiv), 125-145°C, 30 min in microwave, 47.1% (9B), 69%
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
14B), 59.2% (17B), 60.1% (18B) or intermediate II (1.1 equiv), NaOtBu (2.05 equiv), bromo-
tri-tert-butyl)-phosphine-palladium (I) dimer (0.15 equiv), dioxane, 110°C, 1hr, 54% (12B).
CONCLUSIONS
Identification of compound 18B, an advanced GSM modulator with a promising in vitro
and in vivo profile is described starting from RO6800020 (1), our former lead compound. We
applied sequential structural changes to RO6800020 starting on the central piperidine ring in order
to improve potency and increase efficacy. Also we reduced idiosyncratic toxicity risk window by
reducing predicted human dose, increasing potency and balancing the free fraction. The exchange
of a phenyl ring by a nonclassical alkoxy bioisosteric replacement, has an enormous impact on
phototoxicity and key molecular properties such as solubility and lipophilicity while maintaining
potency, low Pgp efflux and metabolic stability. The CF alkoxy bioisosteric replacement of the
3
phenyl ring dramatically affects the compound properties. In addition, small modifications on the
fluoro alkoxy group allowed further balancing of potency and properties.
A judicious structural modification introducing a 5-substituent on the core with the alkoxy
bioisosteric replacement on 8 position avoids hinge binding to PIK4CB which improved
selectivity. This modification, in combination with small changes on the bioisosteric replacement,
allowed the use of alternative less lipophilic head heterocycles devoid of any CVB liability, while
maintaining potency.
This work also highlights the main physicochemical and safety property changes induced
by this novel bioisostere. The CF alkoxy replacement of an aromatic ring was of critical
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 69
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
importance in achieving the desired balance between target modulation and properties such as
solubility, lipophilicity, fu, Pgp and metabolic stability, and also mitigated safety liabilities
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
phototoxicity, CVB).
EXPERIMENTAL SECTION
In vitro cellular Aβ secretion assay.
Human neuroglioma H4 cells stably overexpressing human APP695 isoform with the Swedish
double mutation (K595N/M596L) were plated at 30,000 cells/well/100 µl in 96-well plates in
IMDM media containing 10% FCS, 0.2 mg/l Hygromycin B and incubated at 37°C, 5% CO2. 3-
4 hr post plating, test compounds were diluted in culture media and 50 μl is added to the wells as
1
.5-fold concentrate to achieve the final concentration. Compound incubation is performed for 24
hr. Final doses typically range from 4 µM down to 0.0013 µM in half-log steps resulting in a eight
point dose response curve. Appropriate controls using vehicle only and reference compound were
applied to this assay. The final concentration of Me SO was 0.4%.
2
After incubation at 37°C, 5% CO2, the supernatant was subjected to quantification of secreted
Aβ peptides by the means of an AlphaLisa assay kits (Perkin Elmer). 20 µl of the cell culture
supernatant was transferred to an assay plate. Then 10 µl of a mixture of the AlphaLisa coupled
capture antibody and the biotinylated detection antibody was added and incubated for 3 hours at
room temperature while softly shaking the assay plate. After a further addition of 20 µl of the
Donor beads the assay plate was incubated for 30 min at room temperature and constant shaking
without exposure to direct light. The assay plate was then read on a Paradigm AlphaLisa Reader
using the build-in program with excitation at 680 nm and emission at 570 nm.
ACS Paragon Plus Environment

Page 29 of 69
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The measured signals were then used to calculate IC50 values for inhibition of Aβ peptides
secretion by nonlinear regression fit analysis using XLfit 5.3 software (IDBS).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Notch cellular reporter assay
The cellular Notch reporter assay used a stably transfected HEK293 cell line expressing the human
Notch1 and luciferase reporter. Briefly, the cells were seeded in 96well microtiter plates in cell
culture medium (Iscove without phenolred, plus 10% (v/v) fetal bovine serum) to about 80%
confluency and the compounds were added at a 10x concentration in 1/10 volume of medium
without FCS containing 8% DMSO (final concentration of DMSO was kept at 0.8% v/v). After 20
hrs incubation at 37°C and 5% CO2 in a humidified incubator the microtiter plates were allowed
to equilibrate to room temperature. 45 µl of SteadyGlo Mix (Promega) were added and incubated
another 20’ at room temperature. Chemoluminescence was measured in a SpectraMax Paradigm
Multi-Mode Microplate Reader using a Glow Luminescence cartrige (Molecular Devices).
Animals and Housing Conditions.
Animals were group-housed in holding rooms at controlled temperature (20–22°C), humidity
(55-65%) and under a 12-h light/dark cycle (lights on 0600 h). All animals were allowed ad libitum
access to food and water. The experimental procedures used in the present investigation received
prior approval from the City of Basel Cantonal Animal Protection Committee based on adherence
to federal and local regulations.
Acute treatment of APP-Swe transgenic mice.
APP-Swe transgenic mice aged 2-4 months were acutely treated with different doses as indicated
with group sizes ranging from 3-6 animals. Test compounds were applied per os once in the
ACS Paragon Plus Environment