Identification and optimisation of a 4′,5-bisthiazole series of selective phosphatidylinositol-3 kinase alpha inhibitors

Article abstract of DOI:10.1016/j.bmcl.2015.06.078

Abstract Exploring the affinity-pocket binding moiety of a 2-aminothiazole (S)-proline-amide-urea series of selective PI3Kα inhibitors using a parallel-synthesis approach led to the identification of a novel 4′,5-bisthiazole sub-series. The synthesis and optimisation of both the affinity pocket and (S)-proline amide moieties within this 4′,5-bisthiazole sub-series are described. From this work a number of analogues, including 14 (A66) and 24, were identified as potent and selective PI3Kα inhibitor in vitro tool compounds.

Full text of DOI:10.1016/j.bmcl.2015.06.078

Bioorganic & Medicinal Chemistry Letters 25 (2015) 3569–3574
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Identification and optimisation of a 40,5-bisthiazole series of selective
phosphatidylinositol-3 kinase alpha inhibitors
⇑
Robin A. Fairhurst , Patricia Imbach-Weese, Marc Gerspacher, Giorgio Caravatti, Pascal Furet,
Thomas Zoller, Christine Fritsch, Dorothea Haasen, Joerg Trappe, Daniel A. Guthy, Dorothee Arz,
Jasmin Wirth
Novartis Institutes for BioMedical Research, Global Discovery Chemistry, Novartis Pharma AG, Werk Klybeck, Postfach, CH-4002 Basel, Basel, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history:
Available online 27 June 2015
Exploring the affinity-pocket binding moiety of a 2-aminothiazole (S)-proline-amide-urea series of selec-
tive PI3Ka
inhibitors using a parallel-synthesis approach led to the identification of a novel 40,5-bisthia-
zole sub-series. The synthesis and optimisation of both the affinity pocket and (S)-proline amide moieties
within this 40,5-bisthiazole sub-series are described. From this work a number of analogues, including 14
Keywords:
Oncology
Kinase inhibitor
Phosphatidylinositol-3-kinase-alpha
(A66) and 24, were identified as potent and selective PI3K
a
inhibitor in vitro tool compounds.
Ó 2015 Elsevier Ltd. All rights reserved.
Of the lipid kinase family the class I phosphatidylinositol-3-ki-
nases (PI3Ks): PI3K , PI3Kb, PI3K and PI3Kd have been the most
(SAR) starting from the corresponding pan-PI3K inhibitor
2-aminothiazole-acetamide analogues, as exemplified by 1 leading
to 2 in Figure 1. Following on from this activity we have already
described the optimisation of the closely related starting point 3,
in which the pyrimidyl affinity-pocket binding moiety was opti-
mised to provide the clinical candidate alpelisib (NVP-BYL719),
Figure 2.6 In this Letter we describe a parallel approach from 3 that
was taken to optimise the (S)-proline-amide aminothiazole series
to more broadly understand the affinity-pocket and proline SAR
surrounding PI3K inhibition.
a
c
extensively investigated to date. The work in this area has high-
lighted the class I PI3K’s to be potential targets for the treatment
of a wide range of diseases.1 In an effort to fulfil these therapeutic
opportunities, the search for isoform selective agents has become
increasingly important to supersede the first generation of pan-
PI3 K inhibitors.2 The main drive for increasing isoform selectivity
being to enhance tolerability, by dialing out unnecessary activities,
and to facilitate a better understanding of PI3K signalling path-
ways. In particular, PI3K
a
is amplified, or overexpressed, in a range
In the early stages of the project the affinity pocket SAR had
been developed primarily around substituted phenyl and 4-pyrim-
idyl residues, as exemplified by the 3-fluoro-4-methylsulpho-
of cancers, and has been identified as one of the most commonly
mutated genes in human cancer.3 Additionally, these mutations
typically lead to enhanced, or constitutive, PI3K
a
kinase-activity.
nylphenyl in
2
and 2-tertꢀbutylpyrimidyl in 3. Additionally,
As a result, identifying selective PI3K inhibitors has become an
a
within these residues a range of substituents were found to be tol-
area of active research to explore for the treatment of a range of
erated in the positions meta and para to the point of attachment to
cancers.4
Previously, an (S)-proline-amide aminothiazole-urea series had
been identified within Novartis and shown to have a high level of
selectivity for the PI3Ka
isoform.5 The origin of this selectivity was
N
H
N
SO2Me
N
O
determined to be derived from the proline amide-moiety targeting
an interaction with the non-conserved glutamine 859 residue, sit-
uated at the entrance to the ATP-pocket of PI3Ka. This series was
identified using a parallel synthesis approach which explored the
S
N
H
SO2Me
S
N
F
O
F
O
2-amino thiazole-substituent structure activity relationships
H2N
1
2
Figure 1. Structures of the N-acetyl pan-PI3K inhibitor 1 and the (S)-proline-amide-
urea PI3K selective inhibitor 2.
⇑
Corresponding author. Tel.: +41 61 69 64294; fax: +41 61 6966246.
E-mail address: robin.fairhurst@novartis.com (R.A. Fairhurst).
a
http://dx.doi.org/10.1016/j.bmcl.2015.06.078
0960-894X/Ó 2015 Elsevier Ltd. All rights reserved.

3570
R. A. Fairhurst et al. / Bioorg. Med. Chem. Lett. 25 (2015) 3569–3574
PI3Ka, or to modify the physicochemical properties of the inhibi-
N
O
N
O
tors without negatively impacting the ATP-site interaction.
Towards the end of the project, the modelling used to generate
the above hypotheses could be shown to be in good agreement
with a single-crystal X-ray structure of alpelisib bound into the
N
N
H
N
H
N
S
S
N
N
N
CF3
ATP pocket of PI3Ka 12
.
O
O
To explore the above possibilities, alternative synthetic
approaches were investigated to improve upon the flexibility of
the C–H arylation route used in the identification of 8. The
sequence that was identified is shown in Scheme 2, and proved
to be more flexible with respect to the range of 2-substituents
(R2) that could be readily introduced into the 40-linked thiazole
moiety.13 Starting from commercially available 5-acetyl-2-(N-
acetylamino)-4-methylthiazole 9, the derived bromoketone was
readily cyclised to the corresponding thiazoles 10 upon reaction
with a range of thioamides and thioureas, catalysed by ammonium
phosphomolybdate.14 Acetamide hydrolyses of 10 were followed
by activation as the acylimidazole intermediates which could be
conveniently isolated by filtration from the reaction mixtures
when dichloromethane was used as the solvent. These shelf-stable
intermediates were then reacted with the corresponding proline
amide derivatives to give the final products 11–33.
H2N
H2N
3
alpelisib
Figure 2. Structures of the PI3K
a selective inhibitors 3 and alpelisib.
the thiazole 5-position. Therefore, to rapidly expand the SAR in this
region, the feasibility of a parallel synthesis approach was investi-
gated that would enable the introduction of a range of aromatic
residues into the 5-position of the thiazole core. The synthetic
route that was identified to support this approach was a C–H ary-
lation of the unsubstituted 5-position of 2-acetamido-4-methylth-
iazole 4.7 This reaction enabled a range of aryl chlorides, bromides
and iodides to be coupled to give the N-acylated aminothiazole
analogues 5, as shown in Scheme 1. The aryl halides used to pre-
pare the analogues 5 were selected from internal and commercially
available sources. At this point, any interesting aryl groups would
be anticipated to be pan-PI3K inhibitors, based upon analogy with
1, and biochemical screening was carried to assess the extent of
inhibition of PI3K
exhibiting an interesting level of activity (<10
hydrolysed and converted to the corresponding (S)-proline-
amide-urea analogues 6, via the N-acylimidazole intermediates.
Following the above approach a small number of heterocycles
were identified as 5-thiazole substituents with PI3K biochemical
Table 1 shows the biochemical and cellular data for the refer-
ence compounds and the representative analogues 11 to 20 from
the series in which the affinity pocket SAR is explored through
variation of the 2-substituent (R2) in the 40-linked thiazole moi-
ety.15 From the analogues incorporating alkyl residues, 11 to 15,
the most active examples have been shown to contain a quaternary
centre at the point of attachment to the 40-thiazole 2-position. Both
the tertꢀbutyl and methylcyclopropyl analogues, 14 and 15, have
a
, PI3Kb, PI3K
c
and PI3Kd.8 The analogues 5
M) were then
l
been shown to give the highest levels of PI3Ka activity and selec-
tivity. Selectivities of greater than 20-fold were determined with
the PI3Kd-isoform typically being the closest off-target to the
activities <1 lM, including: 2-substituted 4-thiazoyl, 5-substituted
2-pyrazinyl, 6-substituted 2-pyrazinyl, and 6-substituted 4-pyrim-
idinyl. Of these, the 4-(2-isopropylthiazoyl) was selected as a
particularly interesting opportunity for further optimisation based
PI3Ka activity. This SAR is in line with our findings in the related
pyrimidyl and pyridyl series, as exemplified by the tertꢀbutyl ana-
logue 3.6 For the most active examples in Table 1 the biochemical
activities translated well into cellular activities, with IC50 shifts of
less than 10-fold, and the cellular data were used as the primary
assays to drive the PI3K SAR. To better understand the origin of
the cellular potency shifts, the variation in plasma protein binding
(PPB) within the series was assessed using a chromatography
method measuring the affinity for immobilised human and rat
serum albumin (HSA and RSA),16 selected physicochemical and
in vitro pharmacokinetic (PK) data are shown in Table 2. In the case
of alpelisib a good correlation was observed between PPB predicted
by HSA and RSA binding compared with PPB measured directly by
a rapid equilibrium dialysis method (94.4% and 94.3%).6 Analysing
the data, the similar and modest cellular IC50 shifts observed across
the series are consistent with the predict less than 2-fold variation
in unbound fraction from the HSA and RSA data. Interestingly, the
tertꢀbutyl analogue 14 has been identified from our published
claims by Vogt et al, and shown independently to be a potent
upon the pan-PI3K and PI3K
a activities of 7 and 8, Figure 3 and
Table 1. Additionally, when the selection was made some
precedent already existed for the 40,5-bisthiazole biaryl core
from a related series of pan-PI3K inhibitors.9 Based upon these
data compound 8 was selected as the starting point for further
optimisation. More recently a related 40,5-bisthiazole series of
PI3K
c selective inhibitors has been reported originating from a
high throughput screening hit.10
Docking the 40,5-bisthiazole 8 into a homology model of PI3K
a,
derived from a single-crystal X-ray structure of an aminothiazole
based inhibitor bound into PI3Kc 11
,
suggested that modifying the
nature of the 2-substituent in the 4-linked thiazole moiety could
further enhance the interaction within the affinity pocket. More
specifically, the 2-isopropyl group was determined not to fully
occupy the portion of the affinity pocket formed by the residues
I800, I848, P778 and K802, and highlighted the possibility for lar-
ger groups to better fill this region, as depicted in Figure 4.
Additionally, opportunities were seen for substituting the proline
ring to either: increase the interaction with the ATP site of
and selective PI3Ka
-inhibitor.17 As a result, 14 has been assigned
the identifier A66, and has found use as a reference PI3Ka-selective
N
N
O
N
O
Ar
H
N
i)
ii), iii)
S
Ar
N
H
H
S
S
N
N
O
O
4
5
6
H2N
Scheme 1. Library approach to exploring the SAR of the thiazole 5-position. Reagents and conditions: (i) Ar-X (X = Cl, Br, I), Pd(OAc)2, P(t-Bu)3HBF4, Cs2CO3, DMF, 160 °C
Biotage InitiatorÒ microwave reactor, 1 h; (ii) 1.25 M HCl in MeOH, 50 °C, 17 h; (iii) carbonyl diimidazole (CDI), Et3N, DMF, 50 °C, 17 h, then (S)-proline amide, DMF, 50 °C,
17 h.

R. A. Fairhurst et al. / Bioorg. Med. Chem. Lett. 25 (2015) 3569–3574
3571
Table 1
PI3K , PI3Kb, PI3K
a
c
and PI3Kd activities for the reference compounds and for the 40-thiazole 2-substituted analogues 11–20 exploring the affinity pocket SAR
2-Thiazole substituent R2
N
S
H
S
N
N
R2
11
12
13
14
15
(A66)
N
O
O
F
F
F
N
N
N
F
H2N
16
17
18
19
20
Compound
Biochemical IC50
PI3Kb
(
l
M)
Cellular IC50 (lM)
PI3K
a
PI3K
c
PI3Kd
PI3K
a
PI3Kb
PI3Kd
1
2
3
0.18
0.28
0.007
0.005
0.21
0.020
0.070
0.033
0.088
0.015
0.014
0.55
0.044
0.039
0.16
1.6
>9.1
1.9
1.2
1.8
>9.1
8.5
7.0
>9.1
6.5
4.9
0.87
>10
0.23
0.25
1.2
n.d.
6.3
6.1
>10
8.0
n.d.
>10
3.0
>10
n.d.
2.9
0.41
4.7
n.d.
>10
n.d.
>10
3.1
n.d.
>10
1.5
1.2
1.9
0.38
0.29
0.62
n.d.
1.5
1.6
1.8
1.5
n.d.
8.1
0.039
0.074
0.99
0.51
0.52
0.36
0.61
0.20
0.37
1.0
Alpelisib
7
8
11
12
13
14 (A66)
15
16
17
18
19
20
2.2
0.95
>10
>10
>10
>10
>10
>10
>10
>10
>10
>10
>10
>10
6.3
>10
>10
>10
>10
>10
>10
>10
>10
4.2
>9.1
8.6
>9.1
8.9
0.63
7.4
n.d.
0.81
0.44
1.0
8.2
0.026
5.5
0.18
n.d. not determined.
Aryl 2-substituents have also been found to be well tolerated, as
exemplified by 17 and 18, leading to activities within 2-fold of
the most active quaternary alkyl substituted examples. The
dimethylaminomethyl analogue 19 led to a much greater 50-fold
shift between biochemical and cellular activity, and was represen-
tative of our findings when a basic group, of sufficient pKa to be
extensively protonated at physiological pH, was present within
the 2-substituent. Introduction of fluorine into the most active
example 15 was well tolerated, as demonstrated by the trifluo-
romethylcyclopropyl analogue 20 retaining an equivalent level of
potency and selectivity.
N
S
N
H
H
S
S
N
N
S
N
N
N
O
O
O
7
8
H2N
Figure 3. Structure of the 40,5-bisthiazole PI3K inhibitors 7 and 8.
Evaluating the biophysical profiles of the analogues from
Table 1 with the most interesting levels of PI3K
showed high levels of permeability,19 but with limited solubility
(<130
M).20 In particular the aryl-substituted analogues, 17 and
18, were found to be very poorly soluble (<4 M). Unsurprisingly,
a activity: all
l
l
the tertiary amine containing 19 was shown to possess a much
higher solubility, but as discussed above, this was accompanied
by an unacceptably low level of cellular activity, and also a reduced
level of permeability. In the case of the tertꢀbutyl analogue 14, an
in vivo PK study in the rat was representative of our findings for
the series of compounds exemplified within Table 1. When orally
dosed as a suspension (3.0 mg kgꢁ1), a bioavailability of only 2%
was determined for compound 14.21 Interpreting these data, a
metabolic first-pass effect was anticipated to contribute to some
extent to the measured low oral-bioavailability. This was based
upon the rat in vivo clearance approximating to 75% of liver blood
flow, which for 14 was consistent with the rat in vitro microsomal
clearance.22 High microsomal clearance was also observed for the
most interesting examples in Table 1. Only the poorly soluble
2-aryl analogue 18 and weakly active 2-dimethylaminomethyl 19
exhibited modest stability in rat liver microsomes. However, the
limited aqueous solubility was also considered to be a major con-
tributor to the low oral bioavailability, at what was considered to
Figure 4. Model of compound 8 bound into the ATP pocket of PI3Ka. Key hydrogen
bonds are represented by dashed lines. The purple arrow indicates the region of the
affinity pocket formed by the side chains of residues I800, I848, P778 and K802
which is incompletely filled by the 2-isopropyl group.
inhibitor.18 The dimethylamino analogue 16, isosteric with the iso-
propyl starting point 8, has been shown to be 10-fold less active,
and this result focused the attention on C-substituted analogues.

3572
R. A. Fairhurst et al. / Bioorg. Med. Chem. Lett. 25 (2015) 3569–3574
N
S
O
i), ii)
iii), iv), v)
N
O
N
O
H
N
S
S
N
N
R2
S
S
HN
HN
N
R1
R2
O
O
11 33
-
9
10
H2N
Scheme 2. Synthesis of the 40,5-bisthiazole analogues 11–33. Reagents and conditions: (i) Br2, 1,4-dioxane, 50 °C, 18 h (62–75%); (ii) R2C(S)NH2, 0.1 equiv (NH4)3PMo12O40
EtOH, reflux, 48 h (16–76%); (iii) conc. HCl, EtOH, reflux, 3 h (83–95%); (iv) CDI, DCM, 25 °C, 6 h (47–93%); (v) proline-amide derivative, DMF, 25 °C, 18 h (22–90%).
,
Table 2
Physicochemical and in vitro PK data for the reference compounds and selected 40-thiazole 2-substituted analogues
Compound
clogP/PSA (Å2)
Sol. pH 6.8 (
lM)
HDM FA (%)
Caco-2 A-B/B-A (10ꢁ6 cmsꢁ1
)
Rat microsome Cl (
l
l minꢁ1 mgꢁ1
)
HSA/RSA (%)
2
0.8/122
2.1/101
2.1/101
2.2/101
3.2/101
2.5/101
2.2/101
3.0/101
0.5/104
3.1/101
3.1/101
2.3/101
1.4/121
0.7/121
2.4/101
1.5/121
1.5/121
2.4/104
2.1/104
2.4/104
2.3/101
435
53
285
113
124
90
27
<4
977
5
151
29
141
711
304
10
258
>1000
>1000
973
167
28
89
82
81
96
96
95
93
39
99
99
92
27
29
91
25
46
48
41
31
91
1.4/7.7
3.8/18.3
n.d.
n.d.
n.d.
12.3/15.6
n.d.
n.d.
19
29
n.d.
Alpelisib
8
12
13
14 (A66)
15
18
19
21
23
24
25
26
27
28
29
30
31
90.0/92.7
84.4/n.d.
90.8/92.8
92.2/94.0.
n.d.
92.2/93.5
n.d.
n.d.
n.d.
n.d.
93.9/93.5
n.d.
n.d
132
136
283
126
68
18
30
140
89
62
67
n.d
70
51
70
139
52
91
n.d.
14.5/17.4
11.1/17.4
n.d.
0.9/12.4
n.d.
5.5/20.7
0.7/8.4
1.2/20.8
n.d.
92.2/n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
32
33
n.d.
90.2/91.8
375
clogP was calculated using clogP version 7.4, BioByte Corporation.
be a sub-efficacious exposure in the rat PK studies. As a result,
approaches were sought to increase the solubility and metabolic
stability of the most interesting analogues from Table 1, as a means
to increase oral exposure, whilst retaining, or improving, their
The 2-methyl analogues, 21 and 22, both retained comparable
PI3K activity and selectivity compared to the parent compounds
a
14 and 20. Similarly, a cis or trans, methyl or hydroxyl group in
the 3-position, entries 23 to 26, were also shown to be well toler-
ated. Of these analogues, the cis-3-methyl derivative 24 provided a
4-fold increase in the cellular potency, whilst maintaining a high
level of selectivity for the other PI3K isoforms, and proved to be
the most cellular active example prepared within the series. The
trans-3-hydroxy analogue 25 also provided a similar favourable
favourable PI3Ka potency and selectivity profiles.
One area which was explored, as a means to improve upon the
profiles of the analogues 11 to 20, was the introduction of sub-
stituents into the proline moiety. Previous work had shown the
primary amide to be essential for PI3Ka activity and the corre-
sponding 4-membered azetidine analogues to be less active.6 The
larger 6-membered analogues were found to be chemically unsta-
ble, and unsuitable for pharmacological evaluation.23 Therefore,
the effort was focused on modifications of the pyrrolidine nucleus,
and Table 3 shows the biochemical and cellular data for the ana-
logues 21 to 33 bearing substituents in the proline 2-, 3- and 4-po-
sitions. Substituting the 5-position was not explored because
molecular modelling indicated insufficient space to be available
to accommodate a substituent on either face of the pyrrolidine ring
PI3Ka activity and isoform selectivity profile, comparable to
the parent compound 14. Fluoro and hydroxyl substituents
in the 4-position were also well tolerated, entries 27 to 29.
Interestingly, the cis-4-hydroxy analogue 29 has been shown to
lead to a decrease in the relative PI3Kb cellular selectivity, whilst
maintaining a similar level of selectivity versus the other isoforms.
Introduction of a dimethylamino group into the 4-position on
either face of the proline ring was tolerated to a greater extent than
the introduction of a basic group in the 40-thiazole 2-position, but
when docked within the ATP pocket of PI3K
a. The SAR for the ser-
still resulted in a 5-fold loss of cellular PI3Ka activity for the more
ies is exemplified in Table 3 using the three most promising 2-thi-
azole substituents (R2): tertꢀbutyl, methylcyclopropyl and
trifluoromethylcyclopropyl. The substituted proline-amide build-
ing-blocks used to prepare 21 to 33 were readily synthesised from
commercially-available proline-analogues, or as outlined below,
and introduced using the synthetic sequence in Scheme 2.13 The
non-commercial intermediates were: the 2-methylproline ana-
logue, which was prepared by the alkylation of proline using the
self-reproduction of chirality concept;24 and the cis-3-methylpro-
line analogue, which was synthesised using an amino–zinc-eno-
late-cyclisation.25
potent trans analogue 30. The 2,3-cyclopropyl analogue 33 showed
no PI3K activity advantage over the corresponding compounds
mono-methylated at the proline 2- and 3-positions.
Evaluating the biophysical profiles of the analogues from
Table 3: of the methylated prolines, only the trans-3-methyl ana-
logue 23 potentially led to some degree of increased solubility
(150 lM), compared to the unsubstituted parent compound 14.
However, the methylated analogues showed no improvement in
microsomal clearance, as might be anticipated from their increased
lipophilicity. This increased metabolic instability was evident
in vivo in a mouse pharmacokinetic study with 21, where plasma

R. A. Fairhurst et al. / Bioorg. Med. Chem. Lett. 25 (2015) 3569–3574
3573
concentrations were found to fall rapidly, to below 50 nM, within
2 h of dosing (bolus intravenous administration, 10 mg kgꢁ1).21
The hydroxylated analogues also led to modest increases in solu-
against a panel of 28 receptor, ion channel, transporter and enzyme
assays. No significant inhibition was observed at concentrations up
to 10 lM in these screens, supporting a favourable level of speci-
bility (up to 710
l
M), with slightly improved levels of in vitro
ficity for the series. Of particular interest, and in contrast to a num-
ber of reported class 1 PI3K inhibitors, the absence of mTor activity
determined for 24 was also reproduced for other 40,5-bisthiazole
analogues tested from Tables 1 and 3, and appeared to be charac-
teristic of the series.
metabolic stability compared to the unsubstituted parent com-
pounds. However, introducing the secondary hydroxyl group
resulted in a reduced level of permeability, which was likely driven
primarily by the increased polarity (polar surface area of 121 Å2).19
Additionally, data for the hydroxylated analogues in Caco-2 mono-
layers showed a 10-fold higher basolateral to apical flux, support-
ing the low passive permeability from the hexadecane membrane
(HDM) assay, and indicating a significant level of efflux to be oper-
ating. This permeability profile was only evident to a marginal
extent in the cellular assays, as assessed by the biochemical to cel-
lular potency shifts for 25, 26, 28 and 29, but was anticipated to be
a further limitation to achieving an acceptable level of oral
bioavailability. The dimethylamino analogues, 30, 31 and 32, as
anticipated, led to a higher level of solubility (>1 mM), and a rat
PK with compound 31 showed that an improvement in oral expo-
sure was possible for the series. Oral dosing of a suspension of 31
(3.0 mg kgꢁ1) resulted in 59% bioavailability, in addition to an
improved terminal half-life of 1.9 h.21 The increased half-life being
a result of a lower clearance (27 ml minꢁ1 kgꢁ1), consistent with
To better understand the origin of the above activity changes
the analogues 11–33 were modelled in the ATP pocket of PI3K
a
derived from the X-ray structure solved with alpelisib in which
the urea-H and adjacent thiazole-N make a bidentate H-bonding
interaction with V851 in the hinge region.12 From these insights
the higher potency of 24 can be rationalised through favourable
contributions from both the cis-3-methylproline-amide and
the microsomal data,22 and
a larger volume of distribution
(3.3 L kgꢁ1) compared to other non-basic members of the series.
To further characterise the potential of the series, the most
potent analogue 24 was tested against an internal panel of 35
kinase biochemical assays, including the lipid kinases phos-
phatidylinositol-4-kinase beta (PI4Kb) Vps34 and mTor. No signif-
icant inhibition at concentrations up to 10
these assays, with the exception of PI4Kb, where an IC50 of
0.25 M was determined. Indicating a similar level of selectivity
lM were measured in
l
over PI4Kb as compared to the selectivity over the other class 1
PI3K’s, which was found to be a consistent finding across the
40,5-bithiazole series. In addition, compound 24 was also tested
Figure 5. Proposed binding mode of compound 24 in the ATP pocket of PI3Ka. Key
hydrogen bonds are indicated as dashed lines.
Table 3
PI3K
a, PI3Kb, PI3K
c
and PI3Kd activities for the substituted proline-amide derivatives 21–33
Urea substituents R
N
N
N
N
N
N
O
S
O
O
O
O
O
H
R
S
N
HO
N
R2
H2N
H2N
H2N
N
H2N
H2N
21
22
23
24
25
N
N
N
N
N
N
N
F
HO
O
HO
O
N
O
O
O
O
O
HO
H2N
26
H2N
H2N
H2N
H2N
H2N
H2N
27
28
29
30, 31
32
33
Compound
R2
Biochemical IC50
(lM)
Cellular IC50 (lM)
PI3Ka
PI3Kb
PI3K
c
PI3Kd
PI3K
a
PI3Kb
PI3Kd
21
22
23
24
25
26
27
28
29
30
31
32
33
tertꢀBu
CF3cPr
tertꢀBu
CF3cPr
tertꢀBu
CF3cPr
tertꢀBu
tertꢀBu
tertꢀBu
tertꢀBu
MecPr
tertꢀBu
tertꢀBu
0.030
0.10
5.0
2.1
8.1
4.6
7.1
>9.1
8.0
>9.1
>9.1
>9.1
>9.1
>9.1
>9.1
n.d.
n.d.
n.d.
2.5
>10
5.0
n.d.
n.d.
n.d.
1.5
>10
3.9
3.8
4.0
7.9
9.2
>10
n.d.
3.3
0.20
0.40
0.14
5.6
>10
8.1
>10
6.2
>10
5.8
4.5
1.3
>10
>10
>10
>10
>10
>10
3.4
3.9
>10
9.9
3.6
2.5
5.2
>10
>10
>10
>10
0.022
0.014
0.024
0.043
0.039
0.071
0.084
0.17
0.047
0.18
0.32
0.45
0.12
0.15
0.98
3.0
8.3
7.9
>10
>10
>10
n.d.
>10
0.52
1.2
0.18
5.5
0.54

3574
R. A. Fairhurst et al. / Bioorg. Med. Chem. Lett. 25 (2015) 3569–3574
14. Das, B.; Reddy, V. S.; Ramu, R. J. Mol. Catal. 2006, 252, 235.
trifluoromethylcyclopropyl moieties, as highlighted in Figure 5.
More specifically, the proline cis-3-methyl substituent is assumed
to make van der Waals contact with the side chain of H855, and
the trifluoromethylcyclopropyl group makes further van der
Waals contacts with the side chains of residues I800, I848 and
P778. A hydrogen bond between one of the fluorine atoms and
the side chain amino group of K802 is also hypothesised.
15. The cellular assays measure the compounds ability to inhibit phosphorylation
of Akt (Ser473) in Rat1 cell lines overexpressing either PI3K
a, PI3Kb or PI3Kd,
and are run in the presence of 0.5% fetal calf serum. These assays are described
in further detail in: Maira, S.-M.; Pecchi, S.; Huang, A.; Burger, M.; Knapp, M.;
Sterker, D.; Schnell, C.; Guthy, D.; Nagel, T.; Wiesmann, M.; Brachmann, S.;
Fritsch, C.; Dorsch, M.; Chène, P.; Shoemaker, K.; De Pover, A.; Menezes, D.;
Martiny-Baron, G.; Fabbro, D.; Wilson, C. J.; Schlegel, R.; Hofmann, F.; García-
Echeverría, C.; Sellers, W. R.; Voliva, C. F. Mol. Cancer Ther. 2012, 11, 317.
16. Reilly, J.; Etheridge, D.; Everatt, B.; Jiang, Z.; Aldcroft, C.; Wright, P.; Clemens, I.;
Cox, B.; Press, N. J.; Watson, S.; Porter, D.; Springer, C.; Fairhurst, R. A. J. Liq.
Chromatogr. Relat. Technol. 2011, 34, 317.
In conclusion, exploring the affinity pocket SAR of an (S)-pro-
line-amide aminothiazole-urea series of selective PI3Ka inhibitors
led to the identification a 40,5-bisthiazole sub-series. The most
active examples from this sub-series contained a quaternary alkyl
2-substituent in the 40-thiazole affinity-pocket moiety with a cis-3-
methyl substituent in the proline ring, and achieved comparable
levels of activity and PI3K-isoform selectivity to the previous best
(S)-proline-amide aminothiazole-urea examples. However, further
profiling revealed minimal levels of oral exposure for the most
interesting 40,5-bisthazole examples, and low aqueous solubility
was assigned as a primary reason for this limitation. Attempts to
address this pharmacokinetic deficiency identified no examples
with the required combination of potency, solubility, permeability
and metabolic stability to move forward into more extensive pre-
clinical profiling. As a result, our attention turned to alternative
ways to improve the series, and identify analogues with the tar-
geted profiles.26
17. Sun, M.; Hillmann, P.; Hofmann, B. T.; Hart, J. R.; Vogt, P. K. Proc. Natl. Acad. Sci.
U.S.A. 2010, 107, 15547.
18. The analogue 14 (A66), CAS number [1166227-08-2], has been used as a
reference compound in: (a) Wang, X.; Li, J.; Yang, Y.; Meng, L. Acta Pharmacol.
Sin. 2013, 34, 1201; (b) Smith, G. C.; Ong, W.-K.; Costa, J. L.; Watson, M.;
Cornish, J.; Grey, A.; Gamble, G. D.; Dickinson, M.; Leung, S.; Rewcastle, G. W.;
Han, W.; Shepherd, P. R. FEBS J. 2013, 280, 5337; (c) Stengel, C.; Jenner, E.; Meja,
K.; Mayekar, S.; Khwaja, A. Br. J. Haematol. 2013, 162, 283; (d) Jamieson, S.;
Flanagan, J. U.; Kolekar, S.; Buchanan, C.; Kendall, J. D.; Lee, W.-J.; Rewcastle, G.
W.; Denny, W. A.; Singh, R.; Dickson, J.; Baguley, B. C.; Shepherd, P. R. Biochem.
J. 2011, 438, 53.
19. Passive membrane permeability was assessed using a hexadecane membrane
(HDM) permeability assay, as described in: Wohnsland, F.; Faller, B. J. Med.
Chem. 2001, 44, 923; Data are expressed as the predicted fraction absorbed (FA)
from the gastrointestinal tract following oral administration. The permeability
for selected compounds was measured using Caco-2 monolayers, as described
in: Pade, V.; Stavchansky, S. J. Pharm. Sci. 1998, 87, 1604.
20. Solubility was measured using a high throughput equilibrium method at pH
6.8, as described in: Glomme, A.; März, J.; Dressman, J. B. J. Pharm. Sci. 2005, 94,
1. No assessment of the physical form (amorphous or crystalline) was made for
the solid samples run in this high throughput assay.
Acknowledgments
21. Rat pharmacokinetics: were conducted in female animals of the Sprague-
Dawley strain; blood levels were assessed form 5 min to 48 h post dosing;
animals were conscious, permanently-canulated with free access to food;
doses of 1.0 and 3.0 mg kgꢁ1 were applied in the intravenous and oral arms
respectively. Mouse pharmacokinetics: were conducted in female animals of
the athymic nude mouse strain; assessed plasma levels 1 and 2 h post dosing;
animals were conscious with free access to food.
22. In vitro metabolic stability was measured, using the compound depletion
approach, in rat liver microsomes from pooled rat hepatic microsome
preparations, microsomes (protein concentration 0.5 mg/mL) were incubated
The authors would like to thank Francesca Blasco, Peter Wipfli
and Christian Schnell for conducting pharmacokinetic studies,
and Aurore Roustan, Romain Gambert, Peter Loewenberg and
Esther Roehn-Carnemolla for their excellent technical assistance.
References and notes
at 37 °C with the test compounds (1 lM) and the co-factor NADPH. Aliquots
were removed after incubation times of 0, 5, 15 and 30 min. Reactions were
stopped by addition of acetonitrile and samples were subsequently analysed
by LC–MS/MS after protein precipitation. The data were analysed as percentage
disappearance of parent relative to the zero time sample, from which hepatic
extraction ratios were determined.
1. (a) Burke, J. E.; Williams, R. L. Trends Biochem. Sci. 2015, 40, 88; (b)
Vanhaesebroeck, B.; Stephens, L.; Hawkins, P. Nat. Rev. 2012, 13, 195.
2. Bauer, T. M.; Patel, M. R.; Infante, J. R. Pharmacol. Ther. 2015, 146, 53.
3. Polivka, J., Jr.; Janku, F. Pharmacol. Ther. 2014, 142, 164.
4. Flanagan, J. U.; Shepherd, P. R. Biochem. Soc. Trans. 2014, 42, 120.
5. Bruce, I.; Akhlaq, M.; Bloomfield, G. C.; Budd, E.; Cox, B.; Cuenoud, B.; Finan, P.;
Gedeck, P.; Hatto, J.; Hayler, J. F.; Head, D.; Keller, T.; Kirman, L.; Leblanc, C.; Le
Grand, D.; McCarthy, C.; O’Connor, D.; Owen, C.; Oza, M. S.; Pilgrim, G.; Press, N.
E.; Sviridenko, L.; Whitehead, L. Bioorg. Med. Chem. Lett. 2012, 22, 5445.
6. Furet, P.; Guagnano, V.; Fairhurst, R. A.; Imbach-Weese, P.; Bruce, I.; Knapp, M.;
Fritsch, C.; Blasco, F.; Blanz, J.; Aicholz, R.; Hamon, J.; Fabbro, D.; Caravatti, G.
Bioorg. Med. Chem. Lett. 2013, 23, 3741.
23. As a representative example of the stability of the 6-membered ring containing
analogues: a product consistent with the targeted piperazine analogue 34
could be detected in the final urea-forming step, following Scheme 2. However,
attempts to isolate 34 in a pure form at room temperature were not possible,
due to the formation a major byproduct over time. The isolated byproduct was
consistent with the further reaction of 34 to form the internally cyclised
hydantoin 35, as shown below.
7. Chiong, H. A.; Daugulis, O. Org. Lett. 2007, 9, 1449.
8. Biochemical activities were measured using Kinase-GloÒ and AdaptaÒ assays,
as outlined in Ref. 6.
N
N
Ar
O
N
9. (a) Quattropani, A.; Dorbais, J.; Covini, D.; Desforges, G.; Rueckle, T. Thiazole
derivatives and use thereof. WO 2006/125805, Nov. 30, 2006.; (b) Quattropani,
A.; Rueckle, T.; Schwarz, M.; Dorbais, J.; Sauer, W.; Cleva, C.; Desforges, G.
Thiazole derivatives and use thereof. WO 2005/068444, July 28, 2005.
10. Oka, Y.; Yabuuchi, T.; Fujii, Y.; Ohtake, H.; Wakahara, S.; Matsumoto, K.; Endo,
M.; Tamura, Y.; Sekiguchi, Y. Bioorg. Med. Chem. Lett. 2012, 22, 7534.
11. Knight, Z. A.; Gonzalez, B.; Feldman, M. E.; Zunder, E. R.; Goldenberg, D. D.;
Williams, O.; Loewith, R.; Stokoe, D.; Balla, A.; Toth, B.; Balla, T.; Weiss, W. A.;
Williams, R. L.; Shokat, K. M. Cell 2006, 125, 733.
H
N
S
Ar
N
O
S
N
O
NH2
Boc
N
O
N
34
35
Boc
24. Seebach, D.; Boes, M.; Naef, R.; Schweizer, W. B. J. Am. Chem. Soc. 1983, 105,
5390.
12. The single-crystal X-ray structure of alpelisib bound into PI3K
deposited with the Protein Data Bank ID: 4JPS.
a has been
25. Lorthiois, E.; Marek, I.; Normant, J. F. J. Org. Chem. 1998, 63, 2442.
26. (a) Fairhurst, R. A.; Gerspacher, M.; Imbach-Weese, P.; Mah, R.; Caravatti, G.;
Furet, P.; Fritsch, C.; Schnell, C.; Blanz, J.; Blasco, F.; Desrayaud, S.; Guthy, D. A.;
Knapp, M.; Arz, D.; Wirth, J.; Roehn-Carnemolla, E.; Luu, V. H. Bioorg.
Med. Chem. Lett. 2015, 25, 3575; (b) Gerspacher, M.; Fairhurst, R. A.; Mah, R.;
Roehn-Carnemolla, E.; Furet, P.; Fritsch, C.; Guthy, D. A. Bioorg. Med. Chem. Lett.
2015, 25, 3582.
13. Experimental details for the syntheses of the final products and intermediates
are described. In: (a) Fairhurst, R. A.; Furet, P.; Gerspacher, M.; Mah, R.
Preparation of carboxamidecycloaminoureathizole derivatives for use as PI3-K
inhibitors. WO 2011/000905, Jan. 6, 2011. (b) Fairhurst, R. A.; Imbach, P.
Preparation of N1-bithiazoyl pyrrolidinedicarboxamides and related
compounds as phosphatidylinositol-3-kinase inhibitors. WO 2009/080705,
July 2, 2009.