Omipalisib inspired macrocycles as dual PI3K/mTOR inhibitors

Article abstract of DOI:10.1016/j.ejmech.2020.113109

Activation of the phosphatidylinositol 3-kinase (PI3K)/mammalian target of rapamycin (mTOR) signaling pathway occurs frequently in a wide range of human cancers and is a main driver of cell growth, proliferation, survival, and chemoresistance of cancer cells. Compounds targeting this pathway are under active development as anticancer therapeutics and some of them have reached advanced clinical trials or been approved by the FDA. Dual PI3K/mTOR inhibitors combine multiple therapeutic efficacies in a single molecule by inhibiting the pathway both upstream and downstream of AKT. Herein, we report our efforts on the exploration of novel small molecule macrocycles (MCXs) as dual PI3K/mTOR inhibitors. Macrocyclization is an attractive approach used in drug discovery, as the semi-rigid character of these structures could provide improved potency, selectivity and favorable pharmacokinetic properties. Importantly, this strategy allows access to new chemical space thus obtaining a better intellectual property position. A series of MCXs based on GSK-2126458, a known clinical PI3K/mTOR inhibitor is described. These molecules showed potent biochemical and cellular dual PI3K/mTOR inhibition, demonstrated strong antitumoral effects in human cancer cell lines, and displayed good drug-like properties. Among them, MCX 83 presented remarkable selectivity against a panel of 468 kinases, high in vitro metabolic stability, and favorable pharmacokinetic parameters without significant CYP450 and h-ERG binding inhibition. This profile qualified this compound as a suitable candidate for future in vivo PK-PD and efficacy studies in mouse cancer models.

Full text of DOI:10.1016/j.ejmech.2020.113109

Journal Pre-proof
Omipalisib Inspired Macrocycles as Dual PI3K/mTOR Inhibitors
Rosa M. Alvarez, Ana Belén García, Concepción Riesco-Fagundo, José I. Martín,
Carmen Varela, Antonio Rodríguez Hergueta, Esther González Cantalapiedra, Julen
Oyarzabal, Bruno Di Geronimo, Milagros Lorenzo, M Isabel Albarrán, Antonio Cebriá,
David Cebrián, Sonia Martínez-González, Carmen Blanco-Aparicio, Joaquín Pastor
PII:
S0223-5234(20)31081-3
DOI:
https://doi.org/10.1016/j.ejmech.2020.113109
EJMECH 113109
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 11 September 2020
Revised Date: 25 November 2020
Accepted Date: 13 December 2020
Please cite this article as: R.M Alvarez, A.B. García, C. Riesco-Fagundo, J.I. Martín, C. Varela, A.R.
Hergueta, E.G. Cantalapiedra, J. Oyarzabal, B. Di Geronimo, M. Lorenzo, M.I. Albarrán, A. Cebriá,
D. Cebrián, S. Martínez-González, C. Blanco-Aparicio, J. Pastor, Omipalisib Inspired Macrocycles
as Dual PI3K/mTOR Inhibitors, European Journal of Medicinal Chemistry, https://doi.org/10.1016/
j.ejmech.2020.113109.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
©
2020 Elsevier Masson SAS. All rights reserved.

Omipalisib Inspired Macrocycles as Dual PI3K/mTOR Inhibitors
a
a
a
a,1
Rosa Mª Alvarez , Ana Belén García , Concepción Riesco-Fagundo , José I. Martín ,
a
a,2
a
Carmen Varela , Antonio Rodríguez Hergueta , Esther González Cantalapiedra , Julen
a,3
a
a
a
Oyarzabal , Bruno Di Geronimo , Milagros Lorenzo , Mª Isabel Albarrán , Antonio
a
a,4
a
a
Cebriá , David Cebrián , Sonia Martínez-González , Carmen Blanco-Aparicio ,
a,*
Joaquín Pastor
a
Experimental Therapeutics Programme, Spanish National Cancer Research Centre
(CNIO), C/ Melchor Fernández Almagro 3, E-28029 Madrid, Spain
Corresponding Author
*
Joaquín Pastor. Phone: +34 917328000, fax: +34 917328051. E-mail address:
jpastor@cnio.es.
Present addresses
1
Medicinal Chemistry. Diseases of the developing world, GSK. C/ Severo Ochoa, 2, E-
2
8760 Science to Business Technology Park, Spain.
2
Centro de Investigación Lilly S.A., Avda. de la Industria, 30, Alcobendas, E-28108,
Madrid, Spain.
3
Small Molecule Discovery Platform. Molecular Therapeutics Program, Center for Applied
Medical Research, CIMA, University of Navarra, Avda. Pío XII, 55; 31008 Pamplona,
Spain.
4
Pharmacology unit. Diseases of the developing world, GSK. C/ Severo Ochoa, 2, E-28760
Science to Business Technology Park, Spain.

Abbreviations
ADME-T, absorption, distribution, metabolism, excretion and toxicity; ADP, adenosine
diphosphate; APCI; atmospheric pressure chemical ionization; CNS, central nervous
system;
DCE,
dichloroethane;
DCM,
dichloromethane;
DIPEA,
N,N-
diisopropylethylamine; DMAP, 4-dimethylaminopyridine; DME, dimethoxyethane; DMF,
N,N-dimethylformamide; DMSO, dimethyl sulfoxide; EC50, half maximal effective
concentration; ESI, electrospray ionization; EtOAc, ethyl acetate; FDA, food and drug
administration; HATU, hexafluorophosphate azabenzotriazole tetramethyl uronium; cHex,
cyclohexane; HOAt, 1-hydroxy-7-azabenzotriazole; HPLC, high performance liquid
chromatography; HTRF homogeneous time resolved fluorescence; I.V., intravenous;
LCMS, liquid chromatography mass spectrometry; MCX(s) (macrocycle(s)); MeOH,
methanol; NMP, N-methylpyrrolidone; NMR, nuclear magnetic resonance; PDB, protein
data bank; PKB, protein kinase B; P.O., per oral; oral aministration; PyBOP, (benzotriazol-
1
-yl-oxytripyrrolidinophosphonium hexafluorophosphate); SAR, structure activity
relationship; TEA, triethylamine; TFA, trifluoroacetic acid.

Abstract
Activation of the phosphatidylinositol 3-kinase (PI3K)/mammalian target of rapamycin
(mTOR) signaling pathway occurs frequently in a wide range of human cancers and is a
main driver of cell growth, proliferation, survival, and chemoresistance of cancer cells.
Compounds targeting this pathway are under active development as anticancer therapeutics
and some of them have reached advanced clinical trials or been approved by the FDA. Dual
PI3K/mTOR inhibitors combine multiple therapeutic efficacies in a single molecule by
inhibiting the pathway both upstream and downstream of AKT.
Herein, we report our efforts on the exploration of novel small molecule macrocycles
(MCXs) as dual PI3K/mTOR inhibitors. Macrocyclization is an attractive approach used in
drug discovery, as the semi-rigid character of these structures could provide improved
potency, selectivity and favorable pharmacokinetic properties. Importantly, this strategy
allows access to new chemical space thus obtaining a better intellectual property position.
A series of MCXs based on GSK-2126458, a known clinical PI3K/mTOR inhibitor is
described. These molecules showed potent biochemical and cellular dual PI3K/mTOR
inhibition, demonstrated strong antitumoral effects in human cancer cell lines, and
displayed good drug-like properties.
Among them, MCX 83 presented remarkable selectivity against a panel of 468 kinases,
high in vitro metabolic stability, and favorable pharmacokinetic parameters without
significant CYP450 and h-ERG binding inhibition. This profile qualified this compound as
a suitable candidate for future in vivo PK-PD and efficacy studies in mouse cancer models.
Keywords: Dual PI3K/mTOR inhibitors; anticancer therapies; macrocycles; drug-like
properties; selectivity; in vivo PK studies.

Highlights:
•
•
•
•
•
Design and synthesis of a new series of MCXs as dual PI3K/mTOR inhibitors.
Macrocyclization strategy used to tune dual PI3K/mTOR inhibition.
MCXs with potent biochemical, cellular activity and kinase selectivity.
MCXs with good drug-like properties.
MCX 83, highly potent, selective and orally bioavailable PI3K/mTOR inhibitor.
Graphical Abstract

1
. Introduction
The PI3K/AKT/mTOR signaling pathway is implicated in several cellular functions
including cell growth, proliferation, differentiation, motility, survival, angiogenesis and
intracellular trafficking. Activation of this pathway contributes to the development of
tumors such as breast, colorectal and lung cancers and hematologic malignancies and it is
involved in resistance to anticancer therapies. PI3Ks is a group of lipid kinases, consisting
of two subunits: p85 or p55 regulatory subunits, and p110 catalytic subunit. The PI3K
family is divided into classes I, II, and III. Class I PI3Ks in turn are divided into class IA
and IB. Class IA PI3K, a heterodimer of p85/p110 subunits, is the type most related to
human cancer. Class IA PI3K are divided into p110
α
, p110
β
and p110
δ
isoforms, while
subunit, is
p110 belongs to class IB PI3K. PIK3CA, the gene encoding the p110
γ
α
frequently mutated or amplified in many human cancers, being the most common mutations
E545K, H1047R and E542K. PI3Ks are activated by receptor tyrosine kinases and Ras and
Rho family GTPases. Then, they phosphorylate the 3-hydroxyl position of PIP2
(phosphatidylinositol 4,5-diphosphate) to generate PIP3 (phosphatidylinositol 3,4,5-
triphosphate), a secondary messenger that participates in the activation of several
downstream effectors, including AKT (PKB) kinase. Subsequently, activated AKT triggers
a signal transduction cascade that ultimately stimulates mTOR activity [1].
mTOR is a class IV PI3K protein kinase and is part of two distinct complexes (mTORC1
and mTORC2 both sharing the catalytic activity of mTOR) and plays an integral key role in
regulating PI3K/AKT activation and signaling through positive and negative feedback
loops [2-4].
Targeting the PI3K/AKT/mTOR pathway is an important therapeutic approach to treat
cancer. In this regard, small molecule inhibitors of PI3K include dual PI3K/mTOR

inhibitors, pan-isoforms PI3K inhibitors, and isoform-selective PI3K inhibitors. Several of
these molecules have reached advanced clinical trials and the FDA has approved two of
them (www.fda.org) [5]. This is the case for the PI3K-
101, CAL-101) and the selective PI3K- inhibitor Copanlisib (BAY-80-6946).
Dual inhibitors of PI3K and mTOR target the active catalytic sites of both holoenzymes,
δ specific inhibitor Idelalisib (GS-
1
α/δ
inhibiting the pathway both upstream and downstream of AKT. The rationale for
PI3K/mTOR dual inhibition as a therapy was based on the observation that persistent
mTOR signaling was reported to confer resistance to PI3K inhibition, while selective
targeting of the mTOR complex 1 (TORC1) initiates positive feedback leading to PI3K
hyperactivation. Dual PI3K/mTOR inhibitors with strong activity toward all p110 isoforms
and mTOR combine multiple therapeutic efficacies in a single molecule. Pan-isoform PI3K
inhibitors are expected to reduce the risk of drug resistance that might occur in case of
treatment with compounds targeting a single p110 isoform and, together with mTOR
inhibition, could prevent feedback loop activation of AKT following mTOR inhibition [6,
7
].
Macrocyclic versions of small molecules have been used in drug discovery, providing
some benefits over the corresponding “open compounds”. There are several examples
where the fixation of the bioactive conformation of certain inhibitors by macrocyclization
has led to the improvement of their on-target potency. Moreover, macrocyclization can
modulate the selectivity against off-targets or, less common, to add a desired target activity.
This approach can also modulate ADME-T properties and brain penetration for CNS
targeted therapies as it is the case of Crizotinib based MCXs. Last but not least, the design
of novel MCXs has been used in the field of drug discovery to access new chemical space,
thereby gaining intellectual property [8-12].

It is interesting to highlight that among the plethora of PI3K/mTOR inhibitors reported to
date, none of them is a small molecule MCX. In this context, we want to report our efforts
on the discovery and exploration of macrocyclic dual PI3K/mTOR inhibitors.
2
. Results and discussion
2
.1. Design and Synthesis of Inhibitors
As a chemical starting point for the discovery and design of our MCXs we selected the
known clinical PI3K/mTOR inhibitor GSK-2126458 (Omipalisib) (Fig.1) [13].
The SAR of this class of inhibitors, and the available crystallographic information have
demonstrated the importance of the acceptor N of the central scaffold, which established a
key interaction with Val882 in the hinge area of PI3K-γ (Val851 in PI3K-α). Additionally,
the “pyridosulfonamide” moiety filled the so-called affinity pocket and produced
interactions with Lys833 via the sulfonamide group and with Asp841/Tyr867 through a
water bridge with its pyridyl N. On the other hand, the C-4 substituent at the quinoline
scaffold, which points to the solvent front of PI3K, was less important and even expendable
to achieve good activity in this class of inhibitors (Fig. 1A). The structure of GSK-2126458
bound to the kinase domain of mTOR has not been published, however the docking of the
inhibitor revealed a similar binding mode as in PI3K (Fig. 1B). The quinoline moiety
established contacts with the hinge region of mTOR showing a key interaction with
Val2240 and pi-stacking interactions with the exclusive mTOR residue Trp2239. The
“pyridosulfonamide” moiety filled the back pocket area, and established interactions with
residues such as Lys2187 and Gln2167.
The pyridazine and di-fluoro substituted phenyl fragments of GSK-2126458 are located
in the solvent front in both kinases (Fig. 1), and could be used as linkages to design MCXs

capable of retaining the main interactions described above and, therefore, PI3K/mTOR
activity.
Fig. 1. GSK-2126458. (A) X-ray co-crystal structure of PI3K-
γ with GSK-2126458 (PDB
3
L08). (B) Docking of GSK-2126458 in mTOR from PDB 4JSX (crystal structure of
mTOR-Torin 2 complex from Ref. 14). Dash blue lines highlight hydrogen bond
interactions, and dashed green lines represent Pi-interactions.
Based on this information, we designed a series of MCXs using the quinoline core
present in GSK-2126458, and alternative bicyclic scaffolds that hold the acceptor N in a
similar position. Several aryl, heteroaryl and heterocycloalkyl moieties were used as
substituents of these scaffolds, and connected to the 2,4-diF-Phenyl (or Ph) group via amide
linkers to obtain the corresponding macrocyclic derivatives (Fig. 2).

Hinge-Val882
N
N
Bicyclic
scaffolds
several linkers
water
several solvent
fragment replacements
N
N
C: Cycle Aryl/
Heteroaryl/
Heterocyclic
O
O
O
O
C
NH
HN
O
N
N
S
S
F
several hinge
interacting scaffolds
Lys833
O
F
Alkyl amide
linker
solvent exposed
area
NH
Back pocket
R
O
GSK-2126458
MCXs
Fig. 2. Macrocyclic series design. Schematic representation of key interactions of GSK-
2
126458 in PI3K-γ. General structure of resulting MCXs.
2
.2. Chemistry
The synthetic strategy to prepare the designed MCXs and the necessary intermediates is
depicted in Schemes 1 and 2.

c
BicyclicN
Scaffold
BicyclicN
Scaffold
N
S
Y
BicyclicN
Scaffold
N
O
O
a
b
1
2
Y
NH
O
R = Me, R = F; 11a
X
1
2
O
R = H, R = H; 12 (a-c)
2
R
Y
1
2
1-7
NH2
10 (a-g)
a-g: scaffolds
R = Me, R = H; 13d
1
OR
2
R
X, Y: leaving groups
O
d
C: Aryl, Heteroaryl,
Heterocycloalkyl
BicyclicN
Scaffold
N
d or e
f)
(
O
C
NH2
b, (f)
14-24
BOCHN
BicyclicN
Scaffold
BicyclicN
Scaffold
BicyclicN
Scaffold
N
N
S
N
S
O
O
O
O
i
C
C
h
O
O
C
NH
O
NH
O
NH
S
O
H N
BOCHN
OH
2
N
OH
H
O
O
O
71-94
47-70
25-46
Scheme 1. Synthesis of MCXs 71-94. Reagents and conditions: (a) (5-Amino-6-
methoxypyridin-3-yl)boronic acid pinacol ester, Pd(PPh ) or Pd(dppf)Cl or PdCl (PPh ) ,
3
4
2
2
3 2
K CO or Na CO in H O, dioxane or DME, 80 ºC-110 ºC. (b) Benzenesulfonyl chloride
2
3
2
3
2
reagents, pyridine. (c) Boronic acid of [2-methoxy-pyridin-3-ylsulfamoyl]-benzoic acid
methyl ester, Pd(PPh or PdCl (PPh , K CO or Na CO in H O, dioxane or DME, 80
ºC-100 ºC. (d) Ar/Heteroaryl boronic acid derivatives, Pd(PPh ) or PdCl (PPh ) or
3
)
4
2
3
)
2
2
3
2
3
2
3
4
2
3 2
Pd(dba) /Ph P, K CO or Na CO in H O, dioxane or DME, 80 ºC-115 ºC or 110 ºC-150 ºC
2
3
2
3
2
3
2
under microwave irradiation. (e) Heterocyclic amines, NMP, 150 ºC. (f) LiOH.H O in
2
MeOH or K CO in dioxane at 110 ºC. (h) TFA-DCM or HCl 4N in dioxane. (i) DIPEA,
2
3

PyBOP/DMAP or HATU/HOAt as coupling agents, DMF high dilution. List of MCXs in
Tables 1, 2.
The syntheses of MCXs were accomplished starting from the commercial available
functionalized bicycles 1-9: 6-bromo-4-chloroquinoline (1), trifluoro-methanesulfonic acid
8
-chloro-[1,5]naphthyridin-2-yl ester (2), 4-chloro-6-iodo-thieno[2,3-d]pyrimidine (3), 4-
chloro-2-iodo-1H-pyrrolo[2,3-b]pyridine (4), 3-bromo-6-iodoimidazo[1,2-a]pyridine (5), 3-
bromo-5-chloro-pyrazolo[1,5-a]pyrimidine (6), 2-bromo-5-iodo-imidazo[2,1-b][1,3,4]
thiadiazole (7), 6-bromo-4-chloro-quinazoline (8), and 3-bromo-6-chloro-imidazo[1,2-
b]pyridazine (9). Compounds 1-7 reacted, according to Scheme 1, with (5-amino-6-
methoxypyridin-3-yl) boronic acid pinacol ester under palladium-catalyzed coupling
conditions to give their respective intermediates 10 (a-g). Next, we were able to obtain
compounds 25-46 following a sequence of two consecutive steps that can be inverted: first,
sulfonylation of the amino group of the 2-methoxypyridinyl fragment of 10, and then
introduction of the aryl/heteroaryl/heterocyclic (C) moiety; or on the other way around.
Thus, using the appropriate substituted benzenesulfonyl chlorides in dry pyridine as solvent
we obtained the corresponding sulfonamides derivatives 11-13, which further reacted with
different aryl/heteroaryl boronic/boronates via palladium-catalyzed coupling or with excess
of heterocyclic amines by nucleophilic substitution. The reverse sequence of reactions
involved first the Suzuki palladium coupling of 10 with different boronic/boronates to give
intermediates 14-24, which later reacted with the corresponding benzenesulfonyl chloride
reagents to form the expected compounds. An additional ester basic hydrolysis step was
required for those compounds that contained this group.

Boc group deprotection of 25-46 under acidic conditions afforded key precursors 47-70
that rendered final macrocyclic compounds (71-94) by classical amide formation reactions
under high dilution conditions to favor the desired macrocyclization.
The synthesis of MCXs 119-124 derived from 6-bromo-4-chloro-quinazoline (8) and 3-
bromo-6-chloro-imidazo[1,2-b]pyridazine (9) followed the conditions described in Scheme
2
.
BicyclicN
Scaffold
BicyclicN
Scaffold
BicyclicN
Scaffold
N
S
BicyclicN
Scaffold
N
c
d
a or b
X
O
O
O
C
X
C
C
NH
O
Y
NH
2
8
, 9
BOCNH
OH
BOCNH
BOCHN
X, Y: leaving groups
O
C: Aryl, Heteroaryl,
Heterocycloalkyl
95-100
101-106
107-112
e
BicyclicN
Scaffold
BicyclicN
Scaffold
N
N
S
O
O
O
O
C
C
f
NH
NH
O
S
O
H N
N
H
2
OH
O
O
119-124
113-118
Scheme 2. Synthesis of MCXs 119-124. Reagents and conditions: (a) Aryl/Heteroaryl
boronic acids derivatives, Pd(PPh or Pd(dppf)Cl or PdCl (PPh , K CO or Na CO in
O, dioxane or DME, 80 ºC-100 ºC or 120 ºC-140 ºC under microwave irradiation. (b)
3
)
4
2
2
3
)
2
2
3
2
3
H
2
Heterocyclic amines, TEA, DCE, 0 ºC to rt. (c) (5-Amino-6-methoxypyridin-3-yl)boronic
acid pinacol esther, Pd(PPh ) or Pd(dppf)Cl or PdCl (PPh ) , K CO or Na CO in H O,
3
4
2
2
3 2
2
3
2
3
2

dioxane or DME, 80 ºC-120 ºC. (d) 3-(Chlorosulfonyl)benzoic acid, pyridine, 0 ºC and rt.
e) TFA, DCM. (f) DIPEA, HATU, HOAt, DMF high dilution. See structure and
(
numbering of compounds in Table 2.
Reaction of functionalized bicycles 8 at C-4, and 9 at C-3 position with different
aryl/heteroaryl boronic acid derivatives or with heterocyclic amines at the indicated
conditions rendered compounds 95-100. Then, Suzuki coupling reaction was used to insert
the 5-amino-6-methoxypyridin-3-yl moiety affording 101-106. Subsequent sulfonylation
with 3-(chlorosulfonyl)benzoic acid in pyridine yielded compounds 107-112, which were
transformed into the corresponding “amino/acid” intermediates 113-118 after BOC-
deprotection under acidic conditions. The final amide macrocylization reactions, following
conditions described above rendered desired MCXs 119-124.
2
2
.2. Biological evaluation.
.2.1. Biochemical activity
We evaluated in dose response experiments the inhibitory activity of the prepared MCXs
against PI3K-
α and mTOR (IC50 value, half-maximal inhibitory concentration) [15]. In
general, we obtained highly potent biochemical PI3K-α/mTOR inhibitors with IC values
5
0
within the low nanomolar and even picomolar range (Tables 1, 2), and different inhibition
ratios for both kinases under these experimental conditions.
GSK-2126458 was reported as a strong dual inhibitor with slight preference for PI3K-
α
vs. mTOR [13]. This dual activity was also observed under our experimental conditions
(
PI3K IC50= 1.0 nM; mTOR IC50= 1.2 nM). The quinoline based MCXs (Table 1) showed
α

good PI3K-
bearing 3-pyridinyl and di-F-phenylsulfonamide fragments. The removal of both F atoms
72) led to a subtle change of activity (~ 2-fold; IC = 2.0 nM). The replacement of the 3-
α activity in the range of 0.8-23.2 nM, being the most active the MCX 71,
(
50
pyridine by a Ph group (77) yielded a ~ 6-fold less potent MCX although retained a still
potent activity of 6.1 nM. The corresponding de-fluorinated analogue (79) showed similar
drop of activity observed between MCXs 71 and 72. Noteworthy, the exchange of 3-
pyridine by the 4-pyridine regioisomer produced a 10-fold decrease of PI3K-
α activity (72
vs. 80). The introduction of a 3-pyrazole moiety together with an additional methylene
group in the linker gave a potent MCX (76, IC50 = 4.6 nM), and comparable to the 3-
pyridine analogue 72. Interestingly, the ortho-fluorinated phenyl linker analogues of 79,
MCXs 75 and 81, yielded opposite outcome in terms of PI3K-
α activity. Thus, the “2-F-5-
substituted analogue” (75) resulted to be slightly more potent (IC = 3.8 nM) whereas the
5
0
“
2-F-3-substituted derivative” (81) decreased its activity (IC = 23.2 nM) with respect to
50
7
9. Furthermore, the replacement of the benzylamine portion of the linker in 79 by a
inhibitor (IC50 = 2.2
tetrahydroisoquinoline fragment (73) afforded a more potent PI3K-
α
nM), and demonstrated that the NH of the amide group in the linker was expendable to
obtain potent activity. Importantly, those MCXs with heterocycloalkyl linkers such as
piperidine 74 and 78 were also active (IC50 = 2.9 nM and 11.5 nM) suggesting that
additional chemical diversity in the linker region of these molecules was compatible with
PI3K-α activity.
MCXs 71-81 displayed diverse inhibitory activity against mTOR, ranging from 0.3 nM to
55 nM. The most potent MCX was 73 (IC = 0.3 nM) followed by 74 (IC = 4.1 nM),
6
5
0
50
both with a tertiary amide linker in their structure. Noteworthy, MCX 78, the NH analogue

of 74, was 18-fold less potent (IC = 72.2 nM) pointing to a beneficial effect of the N-
5
0
Methylation of this linker to gain mTOR activity. 3-pyrazole- (76, IC50 = 20.9 nM) and 3-
pyridyl- linkers (71 and 72, IC50 = 22.3 and 63.8 nM) were well accommodated in the
binding site of mTOR. The comparison of MCXs 71 and 72 demonstrated the positive
influence of the fluorine atoms of the phenylsulfonamide ring for mTOR potency. This
effect was already observed for the inhibition of PI3K-α. The better mTOR activity of
MCX 77 over 79 further supported this trend (IC50 = 122 vs. 655 nM). The “fluorine effect”
was also observed in the F-substitution of the Ph-linker of 79. This modification was able to
recover mTOR potency in MCXs 75 and 81 (IC ~ 250 nM). Finally, the 4-pyridil linker of
5
0
MCX 80 (IC ~ 250 nM) led to a similar improvement mTOR activity with respect to the
5
0
Ph-linker analogue 79.
From this exploration, we concluded that some of our MCXs displayed potent dual
PI3K/mTOR inhibition (74, 76), whereas others (e.g. 71, 72, 75, and 77) were more PI3K-
selective inhibitors (mTOR/PI3K- ratio 20). Interestingly, the tetrahydroisoquinoline
linker MCX 73 behaved as a strong dual inhibitor with preference for mTOR
mTOR/PI3K- ratio = 0.13).
The interpretation of the observed SAR in Table 1, and in Table 2 where other scaffolds
α
α
≥
(
α
have been explored, could be rather speculative, unless a detailed computational study
and/or generation of X-ray co-crystal structures is achieved [16]. These studies, out of the
scope of the current work, are underway and their results will be reported in due course.
However, preliminary docking studies have been performed with one of our selected
compounds, MCX 83, confirming a similar binding mode than Omipalisib (Fig. 6).

The SAR interpretation for MCXs is often complex and less straightforward than for
open small molecules”. The coupling of bond rotations and intramolecular interactions
“
enables the transmission of three-dimensional information from one side of a MCX to the
other. On other words, making relatively small structural modifications to a MCX can
result in local conformational changes that propagate along the ring to affect distal
structural features [17]. Thus, modifications of the (C)-substituent (linkers), and scaffolds
in our MCXs could lead to distinct preferred conformations and affect their binding in PI3K
and/or mTOR, reinforcing or weakening key interactions in an unexpected manner.
Comparatively, the SAR for the same modifications in analogous “open small molecules”
could be interpreted attending to more local and classical effects of the substituents, well
known for medicinal chemists.
Besides the conformational influence of the explored modifications in our MCXs, we can
enumerate other factors, which could play (or not) a role to the observed SAR.
1
. The structural information available for GSK-2126458 (PDB 3L08) [13], and a
related morpholinyl analogue (PDB 3S2A) [18], showed the absence of interactions of the
C)-substituent at C-4 of the quinoline scaffold in the binding site of PI3K- , and by
homology with PI3K-α.
(
γ
Additionally, a diverse set of C-4 substituents in this series was tolerated without
significant changes in their PI3K activity [18].
2
. The (C)-substituent at the explored scaffolds could affect their acceptor capacity to
establish H-bond interactions with Val851/Val2240 in PI3K- /mTOR. As well, the intrinsic
α
capacity of the explored bicyclic scaffolds to interact with these key residues should be
taken into account.

3
. The contribution of the different MCXs (mainly the scaffolds) to achieve more or less
productive pi-stacking interactions with Trp2239 in the binding site of mTOR should be
considered to further rationalize the SAR observed against this kinase.
Table 1
a
IC s (nM) of PI3K-
α
, mTOR, and mTOR/PI3K-
α
ratios.
5
0
Nr
Structure
PI3K mTOR ratio
Nr
Structure
PI3K mTOR ratio
N
N
N
N
O
O
O
O
O
O
7
7
7
1
NH
N
0.8
22.3
63.8
0.3
27.8
31.7
0.13
77
NH
6.1
122
72.2
655
20.3
S
S
NH
NH
F
F
O
O
F
F
N
N
N
N
N
O
O
O
O
2
3
O
2
78
79
11.5
13.4
6.3
NH
N
O
NH
S
S
O
NH
N
O
H
N
N
N
N
O
O
O
O
2.2
O
48.9
NH
NH
S
S
O
NH
N
O
O
N
N
N
N
N
N
O
O
O
O
7
7
4
5
2.9
3.8
4.1
1.41
65.9
80
81
O
20
223
249
11.1
10.7
O
NH
NH
S
S
O
NH
N
O
N
N
N
N
F
F
O
O
O
O
O
250
O
23.2
NH
NH
S
S
NH
NH
O
O

N
N
N
N
N
S
O
O
O
7
6
NH
N
N
4.6
20.9
4.52 GSK
NH
O
N
1.0
1.2
1.2
O
S
F
O
NH
F
O
a
The reported values are the average of two independent data points. PI3K-α activity was
TM
TM
determined using ADP-Hunter Plus, and mTOR with LanthaScreen assay.
GSK, GSK-2126458.
We explored alternative scaffolds, bearing the required acceptor N atom, to establish key
H-bond interactions with Val851 in PI3K- and Val2240 in mTOR. The initial quinoline
α
scaffold was replaced by 1,5-naphthyridine (82-84), thieno[2,3-d]pyrimidine (85-86), 1H-
pyrrolo[2,3-b]pyridine (87), imidazo[1,2-a]pyridine (88-89), pyrazolo[1,5-a]pyrimidine
(90-92), imidazo[2,1-b][1,3,4]thiadiazole (93-94), quinazoline (119-122) and imidazo[1,2-
b]pyridazine (123-124). The linkers for this exploration were selected from those used for
the quinoline MCXs. The compounds were synthesized following the general procedure
depicted in Scheme 1 and 2. MCXs (82-124) were tested against PI3K-
α and mTOR
kinases yielding the results summarized in Table 2.
The quinazoline MCXs 119-122 were potent PI3K-
below 3.0 nM. The comparison with their corresponding quinoline pairs afforded similar or
even better PI3K- activity for the quinazolines series (~5-fold for 119, 121, and 122). The
α inhibitors showing IC50 values
α
mTOR inhibition was improved between 3 to 20-fold in all the MCXs in comparison with
their quinoline analogues.
The replacement of the quinoline ring by 1,5-naphthyridine afforded highly potent dual
inhibitors (82-84). The improvement of mTOR activity within this series was significant in

comparison with quinoline analogues. Thus, MCX 82 achieved a 40-fold increase in
potency vs. analogue 79 (IC50 = 16.5 vs. 655 nM). Once again, the presence of fluorine
atoms in the phenylsulfonamide fragment (83) improved both PI3K-
α and mTOR activities
(
IC = 0.8 and 3.3 nM) vs. phenylsulfonamide analogue 84 (IC = 3.7 and 23.4 nM).
5
0
50
The improvement of mTOR activity of quinazoline and 1,5-naphthyridine MCXs might
be highly related to the establishment of more favorable pi-stacking interactions of these
more electron deficient scaffolds with Trp2239 side chain, among other factors cited above.
Next, we explored thieno[2,3-d]pyrimidine MCXs 85 and 86. This scaffold reduced the
PI3K-α potency of these MCXs in comparison with their quinazoline analogues, although
MCX 85 still showed an interesting activity (IC50 = 20.4 nM). Similar results were
reported for a pair of non-macrocyclic inhibitors, where the thieno[2,3-d]pyrimidine
scaffold produced a ~ 20-fold decrease in PI3K-
α activity when compared to the
corresponding quinazoline analogue [19]. The argument to explain such decrease in activity
was the possible lone pair repulsion between the sulfur in the thienopyrimidine core and the
backbone carbonyl oxygen of Glu880 in PI3K-
γ
(surrogate of the homologous PI3K-
α).
Conversely, other general reports could point to a productive
σ
-hole type interaction
between the S atom and the carbonyl oxygen of Glu880 [20]. As alternative, we cannot
discard just a less H-bond acceptor character of thienopyrimidine scaffold vs. quinazoline
to explain the PI3K-α SAR.
The influence of the thienopyrimidine scaffold was much more pronounced in mTOR,
yielding nearly inactive MCXs. The reduction of mTOR activity was evident in comparison
with quinazoline analogues (85 vs. 119 and 86 vs. 120; IC = 5790 nM vs. 30.5 nM, and >
5
0
1
0000 nM vs. 16.7 nM).

A potential steric repulsion between the S atom and Tyr2245 was argued to justify a
decline in m-TOR activity when a quinazoline scaffold was replaced by thieno[2,3-
d]pyrimidine in related open analogues [19]. Furthermore, we could also suggest
unfavorable pi-stacking interactions of more electron rich thieno[2,3-d]pyrimidine scaffold
with Trp2239 as contributor for such results, besides other conformational factors cited
above.
The 1H-pyrrolo[2,3-b]pyridine MCX 87 was designed to get an additional H-bond with
the backbone carbonyl group of Glu849 in PI3K-
inhibitor of PI3K- in our series (IC50 = 445.0 nM) and proven to be inactive against
mTOR. The poor activity against PI3K- indicated a negligible formation of the postulated
α. Unfortunately, 87 was the weakest
α
α
bidentate H-bond, perhaps due to conformational limitations to achieve a good fitting in the
binding site of both kinases.
The electron rich character of this scaffold could also be detrimental for the mTOR
activity of MCX 87 due to unfavorable interactions with Trp2239.
The exploration of several MCXs using 5-6 bicyclic scaffolds was also achieved.
Pyrazolopyrimidine MCXs (90-92) gave potent PI3K-
inhibitors (IC50 < 40 nM). Moreover, the structurally similar imidazopyridazine MCXs 123
and 124 afforded again potent PI3K- inhibition (IC ~ 5.0 nM) although slightly less
α (IC50 < 4.1 nM) and mTOR
α
5
0
potent than their direct analogues 91 and 92. Regarding mTOR, the imidazopyridazine
MCXs followed the same trend observed for their pyrazolopyrimidine counterparts.
Compound 123 showed a well-balanced PI3K-α and mTOR activity with ratio of 1, and in
the case of 3-pyrazole derivative (124) the ratio was ~14. The removal of the six membered
ring N from the imidazopyridazine scaffold led to the imidazopyridine analogues 88 and

8
8
1
9. Compound 89 showed similar potency against PI3K-
α
and mTOR than 124, but MCX
8 was 2.5-fold less potent for PI3K-
23.
α, and 9-fold less potent for mTOR than its analogue
Finally, we explored a 5,5-bicyclic scaffold such as imidazothiadiazole with the
corresponding phenyl and 3-pyridine linkers. MCXs 93 and 94 exerted potent PI3K-
inhibition (IC50 = 8.9 and 16.2 nM) although their mTOR activity was significantly
α
weaker (IC50 = 361 and 1060 nM).
The arguments described to justify previous SAR cases could also be applied for MCXs
derived from 5-6 and 5-5 membered rings scaffolds.
Table 2
a
PI3K-α and mTOR IC s (nM).
50
Nr
Structure
PI3K
mTOR
30.5
ratio
12.3
Nr
Structure
PI3K
4.6
mTOR
5.4
ratio
1.2
N
N
N
N
N
N
N
O
O
N
O
O
NH
NH
NH
1
1
1
19
20
21
O
2.5
3.0
1.3
123
S
NH
S
O
HN
O
NH
O
N
N
N
N
N
N
N
N
O
O
O
O
N
N
O
16.7
5.5
3.8
124
5.0
67.6
13.6
S
NH
O
S
O
N
NH
H
O
N
N
N
N
N
N
S
O
O
N
NH
N
N
4.9
88
O
11.5
49
4.3
O
S
O
HN
O
NH
O
O

N
N
N
N
N
N
N
O
O
N
O
O
NH
NH
NH
NH
N
1
22
2.1
5.0
0.8
3.7
5.6
16.5
3.3
2.7
3.3
4.1
6.3
89
90
91
92
7.7
2.9
1.0
4.1
66.3
8.7
1.6
4.6
9.6
O
S
O
NH
O
S
O
N
H
N
H
N
N
N
N
S
N
N
N
O
O
O
O
8
2
3
4
O
4.7
NH
S
O
HN
O
NH
O
N
N
N
N
S
N
N
N
O
O
O
N
O
8
NH
N
4.4
S
O
O
HN
O
NH
F
O
F
N
N
N
N
N
N
N
O
O
O
O
N
N
8
O
23.4
39.9
S
NH
N
O
S
O
N
NH
H
O
N
S
N
N
S
N
N
O
O
O
N
N
HN
S
O
O
O
8
8
8
5
6
7
O
S
20.4
85.3
445
5790
284
>117
>22
93
94
8.9
361
40.4
65.4
N
H
NH
S
O
O
O
N
N
N
S
N
N
O
O
O
N
N
N
HN
O
S
O
O
O
S
>10000
>10000
N
16.2
1060
N
H
NH
H
N
N
N
O
HN
O
O
S
N
NH
a
Average values of two independent experiments. PI3K-
α
activity was determined using
TM
TM
ADP-Hunter Plus, and mTOR with LanthaScreen assay.

Several conclusions were extracted from the exploration described in Table 2. Most of
the MCXs displayed potent PI3K- inhibition (IC50 < 20 nM), except those with
thienopyrimidine and pyrrolopyridine scaffolds. The different linkers explored in potent
MCXs inhibitors did not afford drastic changes in PI3K- activity (< 10-fold with IC50
values in the low or sub-nanomolar range). It was observed a more pronounced inhibition
of PI3K- than mTOR along the explored MCXs, although many MCXs could be cataloged
as dual PI3K- /mTOR inhibitors (ratios < 10, and IC50 values in nanomolar range).
Conversely, several MCXs were more selective PI3K- inhibitors than GSK-2126458
mTOR/PI3K- ratio of 1.24) such as imidazothiadiazoles 93 and 94 with ratio > 30, and
the thienopyrimidine PI3K- inhibitor 85 with ratio = 284.
α
α
α
α
α
(
α
α
2
.2.2. Cellular activity
Next, the cellular activity of the compounds with interesting dual biochemical activity
PI3K- /mTOR IC values < 10 nM / < 50 nM respectively) was evaluated in U2OS cells
(
α
5
0
by measuring their effects on the phosphorylation of AKT on Ser473. This phosphorylation
event is dependent on the orchestrated activities of PI3K, mTORC1, and mTORC2. The
inhibition of PI3K inhibits mTORC1 activity, and through a feedback loop mechanism
activates mTORC2, the kinase responsible of the AKT phosphorylation on Ser473. Besides,
mTORC1 inhibition triggers PI3K activation via RTKs modulation. Therefore, the
Ser473
inhibition of p-AKT
is a good measurement of the efficacy of dual PI3K/mTOR
inhibitors to control the PI3K/mTOR/AKT signaling pathway.
Ser473
In general, it was observed a good inhibition of p-AKT
with EC < 100 nM for all
50
MCXs, except for the pyrazolopyrimidine 90 (Table 3). The activity of MCXs 73 and 83

was even better than the reference compound GSK-2126458 under our experimental
conditions.
Table 3
a
Biochemical (IC ) and Cellular activity (p-AKT, EC ) in nM.
5
0
50
p-
AKT
p-
AKT
Nr
Structure
PI3K
2.2
mTOR
0.3
Nr
Structure
PI3K mTOR
N
N
N
N
N
N
O
O
O
O
7
8
3
3
O
0.10
0.17
92
N
N
4.1
0.8
2.5
39.9
22.3
30.5
38.9
10.7
47.9
NH
NH
S
S
O
O
N
N
H
O
N
N
N
N
N
O
O
O
O
NH
N
0.8
2.1
3.3
5.6
71
NH
N
S
S
O
O
NH
NH
F
F
O
O
F
F
N
N
N
N
N
N
N
O
O
O
O
1
22
4.58 119
O
O
NH
NH
S
S
O
NH
N
H
O
N
N
N
N
N
O
O
O
8
4
O
3.7
4.6
23.4
9.00
10.8
76
82
NH
N
N
4.6
5.0
20.9
16.5
61.9
72.3
O
NH
N
S
S
O
NH
NH
O
O
N
N
N
N
N
N
N
O
O
O
1
23
N
5.4
O
NH
NH
S
S
O
HN
O
O
NH
O

N
N
N
N
N
N
N
O
O
O
O
7
4
O
2.9
1.3
4.1
4.9
18.3 120
O
3.0
2.9
16.7
4.7
99.4
203
NH
NH
S
O
S
NH
N
O
N
N
N
N
N
N
N
O
O
O
1
21
O
NH
N
N
20.6
90
NH
S
S
O
O
HN
O
NH
O
N
N
N
N
N
N
N
S
O
O
O
O
9
1
N
1.0
4.4
37.1 GSK
NH
O
N
1.0
1.2
0.6
NH
S
F
F
O
HN
O
a
Average values of two independent experiments.
GSK: GSK-2126458.
2
.2.3. ADME-T properties of MCXs
We were intrigued about the potential transference of drug-like properties of our starting
point of design, the clinical inhibitor GSK-2126458, to the newly discovered macrocyclic
inhibitors. First, the metabolic stability of MCXs with better cellular potency (p-AKT EC50
<
50 nM) was measured using human, mouse and rat liver microsomes. The results are
summarized in Table 4.
Table 4
ADME-T data.
Nr
hLM mLM
rLM CYP1A2 CYP2C19 CYP2C9 CYP2D6 CYP3A4
a
a
a
(%)
(%)
88
(%) IC50 µM IC50 µM
IC50 µM IC50 µM IC50 µM
7
1
89
97
>30
>30
8.3
>30
>30

7
7
8
8
9
9
1
1
1
3
4
3
4
1
2
69
79
100
77
90
66
57
64
97
77
85
61
74
76
63
63
63
72
>30
>30
14.7
>30
>30
>30
>30
>30
>30
>30
16.8
19.7
3.1
>30
>30
>30
>30
9.9
19.6
>30
>30
>30
>30
>30
>30
>30
>30
6.5
>30
>30
>30
>30
>30
>30
>30
0.9
100 >30
75
92
57
91
86
70
90
>30
>30
>30
>30
>30
>30
>30
>30
>30
>30
>30
>30
19 97
21 80
22 74
23 93
1
6.5
a
Percentage of compound remaining after 15 min incubation in human, mouse and rat liver
microsomes.
In general, the tested MCXs were rather metabolically stable displaying high levels of
intact parent compounds after 15 minutes of incubation. The main exceptions were the
quinoline MCX with a tetrahydroisoquinoline linker 73 and the pyrazolopyrimidine MCX
9
2 bearing a pyrazole linker, which showed weaker stability in all species (~ 60% of
compound remaning).
It is well known that combinatorial therapy is widely used in cancer treatments. In order
to be safe in these therapeutic scenarios a drug should be free of potential drug-drug
interactions. CYP450 inhibition has been identified as a major source for drug-drug
interactions in combinatorial therapy [21]. Therefore, we tested the inhibitory activity of
our MCXs in a panel of CYP450 enzymes (Table 4). All compounds displayed excellent
profiles with negligible inhibition of the CYP450 enzymes or ample therapeutic windows

against their primary activity. The exception was MCX 122 which showed significant
inhibition of CYP2C9 with IC50 below 1 µM.
The blockage of the human “Ether-à-go-go-Related Gene” channel (h-ERG) is associated
with cardiotoxicity since it can lead to a heart related fatal disorder called “Long QT
syndrome”. It is characterized by ventricular arrhythmias, particularly torsades de pointes
(
TdP) [22]. The early detection of this liability within drug discovery programmes is crucial
to progress drug candidates towards the clinic. We tested the binding of our MCXs against
h-ERG, and none of them displayed significant binding (IC50 > 30 M).
µ
2
.2.4. In vivo Pharmacokinetics
We assessed the pharmacokinetic properties of selected MCXs in BALB-C mice after
I.V. and P.O. administration. We selected some dual PI3K/mTOR inhibitors based on
potency, cellular activity (p-AKT EC50 20 nM), microsomal stability higher than 75% in
≤
all species tested, and without any significant ADME-T alert. Compound 73 was also
included due to its picomolar cellular activity. These compounds are representatives of
three different scaffolds. The results of these studies are summarized in Table 5. The tested
MCXs showed oral bioavailability with Fabs ranging from 27% to 85%. The in vivo
clearance (Cl) was below 10% of mouse hepatic blood flow rate for all MCXs (Cl: 0.22-
0
.56 L/h/Kg) except for 83, with moderate values (Cl= 1.89 L/h/Kg; 35% of hepatic blood
flow). The volume of distribution (Vd) was moderate for the compounds with low
clearance (0.28-0.59 L/Kg; accounting for 47-98% of the total body water content of the
animals), indicating that the compounds are increasingly distributed in total body
compartments. By contrast, MCX 83 showed higher Vd (5.98 L/Kg) suggesting much

higher distribution in peripheral tissues, which could be relevant in the treatment of solid
tumors. In this regard, the volume of distribution reported for GSK-2126458 (Table S1)
was of 1 L/Kg [13].
Table 5
a
Pharmacokinetic (PK) profile.
I.V.
P.O.
Nr
Dose AUCinf T 1/2 Cl
Vd Dose Cmax
Tmax T 1/2 AUCinf F %
7
7
8
8
1
1
3
3
4
5
1
5
5
9467.6 0.57 0.53 0.28 10
4455.1 2.01 0.22 0.58 3
3389.3 0.25 2.78 7681
40.6
27.3
84.6
28,2
47.6
962.3 0.50 3.35 3648
1565.8 0.50 5.22 4475
1635.2 0.25 1.81 5029
1266.3 0.16 5.68 8761
2644.4 2.59 1.89 5.98 10
8930.8 0.99 0.56 0.41 10
9203.9 2.91 0.54 0.59 10
21 5
a
PK studies performed I.V./P.O. at the indicated doses during 8 h. The data were adjusted to a non-
compartmental model using Winnolin software. Units of PK parameters: Dose (mg/Kg); Area
Under Curve AUC (h.ng/ml); half life T 1/2 (h); Clearance Cl (L/h/Kg); Volume of distribution Vd
(
L/ Kg); Maximun concentration Cmax (ng/ml). Time of maximum concentration Tmax (h).
All MCXs showed good oral exposures and achieved plasma concentrations (measured
Ser473
up to 8 h) that fairly exceeded the EC values for inhibition of p-AKT
in cells (Fig.
5
0
3
). The best ratios were observed for MCXs 73 (3 mg/Kg dose) followed by 83 (10 mg/Kg
dose) mainly due to their high cellular activity (Fig. 3).