Potent, selective small molecule inhibitors of type III phosphatidylinositol-4-kinase α-but not β-inhibit the phosphatidylinositol signaling cascade and cancer cell proliferation

Article abstract of DOI:10.1039/c3cc48391f

Two series of inhibitors of type III phosphatidylinositol-4-kinase were identified by high throughput screening and optimised to derive probe compounds that independently and selectively inhibit the α- and the β-isoforms with no significant activity towards related kinases in the pathway. In a cellular environment, inhibition of the α-but not the β-subtype led to a reduction in phosphatidylinositol-4-phosphate and phosphatidylinositol-4,5- bisphosphate concentration, causing inhibition of inositol-1-phosphate formation and inhibition of proliferation in a panel of cancer cell lines. the Partner Organisations 2014.

Full text of DOI:10.1039/c3cc48391f

Featuring the work of Dr Michael J. Waring et al. from
AstraZeneca, Alderley Park, UK and Cancer Research
Technology, UK.
As featured in:
Potent, selective small molecule inhibitors of type III
phosphatidylinositol-4-kinase α- but not β-inhibit the
phosphatidylinositol signaling cascade and cancer cell
proliferation
Two series of inhibitors of type III phosphatidylinositol-4-kinase
are presented that selectively inhibit the α- and the β- isoforms
with no significant activity towards related pathway kinases. In a
cellular environment, inhibition of the α- but not the β- subtype
reduced the concentration of proximal biomarkers, causing
inhibition of proliferation in a panel of cancer cell lines.
See Michael J. Waring et al.,
Chem. Commun., 2014, 50, 5388.
This paper is the first publication of work arising from a strategic
alliance between AstraZeneca and Cancer Research Technology
focused on bringing forward novel cancer therapies.
www.rsc.org/chemcomm
Registered charity number: 207890

ChemComm
View Article Online
View Journal | View Issue
COMMUNICATION
Potent, selective small molecule inhibitors of
type III phosphatidylinositol-4-kinase a- but not
b-inhibit the phosphatidylinositol signaling
cascade and cancer cell proliferation†
Michael J. Waring,*a David M. Andrews,a Paul F. Faulder,a Vikki Flemington,a
Jennifer C. McKelvie,b Sarita Maman,c Marian Preston,a Piotr Raubo,a Graeme R. Robb,a
Karen Roberts,a Rachel Rowlinson,a James M. Smith,c Martin E. Swarbrick,c Iris Treinies,b
Jon J. G. Wintera and Robert J. Woodc
Cite this: Chem. Commun., 2014,
50, 5388
Received 1st November 2013,
Accepted 12th December 2013
DOI: 10.1039/c3cc48391f
www.rsc.org/chemcomm
Two series of inhibitors of type III phosphatidylinositol-4-kinase
were identified by high throughput screening and optimised to
derive probe compounds that independently and selectively inhibit
the a- and the b-isoforms with no significant activity towards related
kinases in the pathway. In a cellular environment, inhibition of the
a- but not the b-subtype led to a reduction in phosphatidylinositol-
4-phosphate and phosphatidylinositol-4,5-bisphosphate concentra-
tion, causing inhibition of inositol-1-phosphate formation and inhibition
of proliferation in a panel of cancer cell lines.
Phosphatidylinositol-4-kinases (PI4Ks) catalyse the phosphorylation
of phosphatidylinositol (PI) at the D4 position of the inositol head-
group to synthesise phosphatidylinositol-4 phosphate (PI4P). PI4P
acts as a binding partner for a wide range of proteins with pleckstrin
Scheme 1 The PI4K cascade.
homology (PH) or related lipid-binding domains and has the generation of Golgi-derived carriers. In addition to the Golgi
reported roles in membrane trafficking. PI4P is also an impor- apparatus, PI4Kb has also been visualised on lysosomes where it
tant signalling molecule as a precursor to other phosphoinositides functions to maintain lysosomal membrane integrity.2
such as phosphatidylinositol-4,5-bisphosphate [PI(4,5)P2] and
PI4Ks have been implicated in human disease. There are several
phosphatidylinositol-3,4,5-trisphosphate [PI(3,4,5)P3] (Scheme 1).1 reports of increased mRNA expression of the different PI4K isoforms
There are four PI4Ks in mammals, termed type II (PI4KIIa in tumours and tumour cell lines.3,4 In addition it has been reported
and b) and type III (PI4KIIIa and b, hereafter referred to as that over expression of the Golgi-associated PI4P-binding protein
PI4Ka and b). Each shows differential sub-cellular distributions GOLPH3 results in tumour growth and transformation by activating
resulting in compartmentalised PI4P synthesis. PI4Ka is the main the mTOR signalling pathway.5 The gene encoding GOLPH3 is
isoform responsible for PI4P generation at the plasma membrane located in a specific region of chromosome 5p13 frequently amplified
and is also localised at the endoplasmic reticulum. PI4Kb is in solid tumours. There are also several reports of an important role
mainly associated with the Golgi complex where it functions in for the PI4Ks in the replication and secretion of multiple viruses.6–8
Pharmacological manipulation of cellular PI4P levels has been a
challenge due to the non-specificity and lack of potency of known
PI4K inhibitors such as wortmannin, LY294002 and phenylarsine
a AstraZeneca, Mereside, Alderley Park, Macclesfield, Cheshire SK10 4TG, UK.
E-mail: mike.j.waring@astrazeneca.com; Tel: +44 (0)1625 230942
oxide;9 although more recently, reports have emerged describing
PI4K inhibitors from several groups.6–8 Hence, suitable chemical
probes that are more potent and selective towards other enzymes in
the pathway, with particular emphasis on the downstream kinases
PIP5K and PI3K, were sought. These probes would elucidate the
roles of PI4Ka and b in the PI signalling pathway and investigate
the utility of inhibition of these enzymes in disease modification.
b Cancer Research Technology, Wolfson Institute for Biomedical Research,
University College London, Gower Street, London WC1E 6BT, UK
c Cancer Research Technology, Jonas Webb Building, Babraham Research Campus,
Cambridge CB22 3AT, UK
† Electronic supplementary information (ESI) available: Details of experimental
procedures, assay protocols and full data for growth inhibition. CCDC 969955.
For ESI and crystallographic data in CIF or other electronic format see DOI:
10.1039/c3cc48391f
5388 | Chem. Commun., 2014, 50, 5388--5390
This journal is ©The Royal Society of Chemistry 2014

View Article Online
Communication
ChemComm
Table 1 Optimisation of PI4Kb inhibitor probes
activity relationship (SAR). Removal of the substituent as in 5
showed a loss of potency and selectivity. This lack of selectivity is
perhaps unsurprising since this core motif is precedented to give
rise to potent PI3K inhibitors.11
PI4Ka PI4Kb PI3Ka PIP5Kg
pIC50 pIC50 pIC50 pIC50
Compound Structure
With the substituent removed, other replacements for the 3-pyridyl
ring were investigated. Replacement with N-methylpyridone 6 led to
an increase in potency (for both PI4Ka and b) without increasing PI3K
activity. Introduction of an N-substituent on the pyridone showed that
marked gains in potency and selectivity could be achieved once the
substituent reached a certain size. Most notably, a significant increase
in PI4Ka potency was achieved with the N-cyclopropylmethyl
derivative 7, which was a potent (pIC50 8.2) inhibitor of PI4Ka
with only weak activity against PI4Kb, PI3Ka or PIP5Kg (pIC50s 5.9,
5.2 and o4.0 respectively).
1
8.0
8.3
8.5
6.2
2
3
5.3
5.1
7.2
7.8
6.1
4.7
6.2
o4.0
The observed SAR was rationalised using protein structures of
PI3K (pdb code 2CHZ), PI4Ka (model based on 4AOF) and PI4Kb
(model based on 2CHZ). Unselective 1 can dock reasonably into all
three models (Fig. 1a and b). In addition to the hinge interaction, the
crucial interaction for this compound is between the anionic nitrogen
of the sulfonamide (pKa = 5.6) and the nearby Lys833 residue (PI3Kg).
The equivalent Lys is also present in both PI4K isoforms but is far less
accessible in the a-isoform. On altering the pyridine substitution
pattern as in 2, this nitrogen is no longer charged, the preferred
ligand conformation is altered and the interaction with Lys is lost.
In PI4Kb, a new interaction is formed between the oxygen of the
sulfonamide and the Lys (Fig. 1d) but this cannot be accommodated
in PI3Kg, hence the greater selectivity of 2.
A set of roughly 100 000 compounds enriched in substructures
likely to bind to lipid kinases was selected from the AstraZeneca
compound collection and screened against recombinant PI4Kb.
Active compounds from this screen were selected for follow-up IC50
determination against PI4Ka and b. This process revealed acyl-
aminothiazole 1 (Table 1), which was a potent inhibitor of both
PI4Ka and b (pIC50 8.0 and 8.3 respectively) but was more active
against PI3Ka (pIC50 8.5) and had some residual activity against
PIP5Kg (pIC50 6.2).10 A series of analogues based on this lead revealed
that PI4Ka, PI3K and PIP5K activities could all be diminished
significantly by changing the position of the nitrogen in the central
pyridine ring albeit with a 1.1 log drop in PI4Kb potency (illustrated
by 2). Library synthesis varying the thiazole amide showed that both
potency and selectivity could be optimised by modifications in this
region. As part of this exercise, 3 was revealed as a potent (pIC50 7.8)
inhibitor of PI4Kb with good lipid kinase selectivity (pIC50 5.1,
4.7 and o4.0 vs. PI4Ka, PI3Ka and PIP5Kg respectively).
The b-selective 3 docks well into PI4Kb (Fig. 1d) but not PI4Ka
under the same protocol. Comparison with the less selective 2 (Fig. 1c)
shows that the acylpiperidine group of 3 would clash with the protein
The same high throughput screen described above also identi-
fied 4 (Table 2), which was appealing as a starting point for a PI4Ka
probe as it exhibited greater potency for PI4Ka (pIC50 7.0) than
PI4Kb (pIC50 6.4). Hence, structural modifications were explored
with the aim of improving PI4Ka potency and selectivity. Truncation
of the ethoxy-substituent of 4 revealed a quite exquisite structure
Table 2 Optimisation of PI4Ka inhibitor probes
PI4Ka PI4Kb PI3Ka PIP5Kg
Compound Structure
pIC50
pIC50
pIC50
pIC50
4
7.0
6.4
5.8
o4.0
5
5.8
6.2
5.8
4.5
6
7
6.4
8.2
6.5
5.9
5.6
5.2
o4.0
o4.0
Fig. 1 Protein homology models with ligands (highlighted with green carbons)
bound. Protein Van der Waals surface is shown in yellow; positive directional
interactions, including hydrogen-bonds are shown as dotted lines.
This journal is ©The Royal Society of Chemistry 2014
Chem. Commun., 2014, 50, 5388--5390 | 5389

View Article Online
ChemComm
Communication
in the a-isoform while the same group fits well and makes an
additional interaction with a nearby Lys residue in the b-isoform.
This is likely due to the Pro residue adjacent to the hinge in PI4Ka: the
restricted geometry constrains the protein conformation and thus
blocks this part of the binding pocket. The equivalent residue in
PI4Kb (Val) imposes no such restrictions. Even for 2, the a-isoform
scores the pose less well, fitting with the b-bias of the whole series.
The a-selective 7 docks well in the PI4Ka model (Fig. 1e) but
fails to dock into PI4Kb. The similar, but less-selective compound
5 docks well in the b-isoform (Fig. 1f) showing that the pocket in
PI4Kb is not large enough to accommodate the cyclopropylmethyl
group; this group fits well in PI4Ka. Critically, Arg1765 in PI4Ka
defines the subpocket that accepts the cyclopropylmethyl group
and reduces the overall volume of the site and preventing the
larger compounds 2 and 3 from binding well.
Compounds 3 and 7 were tested at 1 mM in the Millipore kinase
panel consisting of 259 different kinases.12 Compound 3 showed
no activity against any kinase in the panel, 7 only showed activity
against three targets (FGR 98%, ZIPK 72% and STK17A 68%, for
full results see ESI†). Critically for subsequent experiments, neither
compound showed significant activity against PDGFRb.
In order to assess the cellular activity of 3 and 7, the effect of
the compounds on inositol monophosphate (IP1) was measured.
PI4P is phosphorylated by the PIP5Ks to synthesise PI(4,5)P2.
PI(4,5)P2 is hydrolysed by phospholipase Cs (PLCs) to generate
inositol triphosphate and subsequently IP1. PLCg can be activated
to hydrolyse PI(4,5)P2 by stimulation of receptor tyrosine kinases.
NIH3T3 cells engineered to overexpress PDGFRb were stimulated
with PDGF resulting in a dose-dependent increase in IP1 concen-
tration. Treatment of the cells with 3 and 7 prior to PDGF
stimulation revealed that 7 (PI4Ka) inhibited the accumulation
of IP1 (pIC50 6.3) whereas 3 (PI4Kb) did not.
In order to examine a more proximal marker of inhibition,
tandem mass spectrometry was used for the direct measure-
ment of PIP, PIP2 and PIP3. NIH3T3-PDGFRb cells were treated
with 7 and 3 (30 mM for 2 hours) in serum-free media followed
by subsequent stimulation with PDGF for 5 minutes. These
studies revealed that the PI4Ka inhibitor 7 caused a qualitative
reduction in cellular PIP, PIP2 and PIP3 levels (Table 3 and ESI†).13
The PI4Kb inhibitor 3 showed no effect.
Since these mass spectrometric measurements are not isomer
specific, effects on PI(4,5)P2, were determined using a fluorescence-
based method consisting of a fusion of the PH domain of PLCd1,
known to specifically bind PI(4,5)P2, with red fluorescent protein
overexpressed in U2OS cells.14 Treatment of this system with 3 and
7 at 30 mM showed that 7 (PI4Ka) inhibited the formation of basal
PI(4,5)P2 at the membrane (19% inhibition). The PI4Kb inhibitor 3
showed no effect.
Fig. 2 Heat map of pGI50 values for 3 and 7 in a panel of 183 cancer cell
lines. Colour shows pGI50 from 5.0 (yellow) to 6.1 (red), grey o5.0.
Both 3 and 7 were tested for growth inhibition in a panel of
183 cancer cell lines. PI4Ka inhibitor 7 inhibited cell growth with
pGI50 values of >5.0 in 91 of the cell lines. In contrast, PI4Kb
inhibitor 3 showed a pGI50 >5.0 in only 18 lines (Fig. 2 and ESI†).
These studies highlight the value of high quality small molecule
inhibitors as probes to interpret complex biological cascades. The
exquisite selectivity of the chemical probes allows their pharma-
cology to be ascribed to their intended targets with a high degree
of confidence and should be of further utility in the study of
PI4K biology and the PI pathway in general.‡
We would like to acknowledge the contributions of a number of
colleagues who contributed to the wider project of which this work was
a part. In particular, we acknowledge Sabina Cosulich for initiation
and early leadership of the project, Robert Garcia for generation of IP1
data, the AZ cell panel team for generation of the GI50 data and Nullin
Divecha for the provision of the U2OS cells used in the PI(4,5)P2 assay.
Notes and references
‡ Samples of compounds 3 and 7 are available on request for further
biological study.
1 T. R. Graham and C. G. Burd, Trends Cell Biol., 2011, 21, 113.
2 S. Sridhar, B. Patel, D. Aphkhazava, F. Macian, L. Santambrogio,
D. Shields and A. M. Cuervo, EMBO J., 2013, 32, 324.
3 C. Curtis, S. P. Shah, S. F. Chin, G. Turashvili, O. M. Rueda, M. J.
¨
Dunning, D. Speed, A. G. Lynch, S. Samarajiwa, Y. Yuan, S. Graf, G. Ha,
G. Haffari, A. Bashashati, R. Russell, S. McKinney, A. Langerød, A. Green,
E. Provenzano, G. Wishart, S. Pinder, P. Watson, F. Markowetz, L. Murphy,
´
I. Ellis, A. Purushotham, A.-L. Børresen-Dale, J. D. Brenton, S. Tavare,
C. Caldas and S. Aparicio, Nature, 2012, 486, 346.
4 A. Sanchez-Palencia, M. Gomez-Morales, J. A. Gomez-Capilla, V. Pedraza,
´
L. Boyero, R. Rosell and E. Farez-Vidal, Int. J. Cancer, 2011, 129, 355.
5 K. L. Scott, O. Kabbarah, M. C. Liang, E. Ivanova, V. Anagnostou,
J. Wu, S. Dhakal, M. Wu, S. Chen, T. Feinberg, J. Huang, A. Saci,
H. R. Widlund, D. E. Fisher, Y. Xiao, D. L. Rimm, A. Protopopov,
K.-K. Wong and L. Chin, Nature, 2009, 459, 1085.
6 M. J. LaMarche, J. Borawski, A. Bose, C. Capacci-Daniel, R. Colvin,
M. Dennehy, J. Ding, M. Dobler, J. Drumm, L. A. Gaither, J. Gao,
X. Jiang, K. Lin, U. McKeeever, X. Puyang, P. Raman, S. Thohan,
R. Tommasi, K. Wagner, X. Xiong, T. Zabawa, S. Zhu and B. Wiedmann,
Antimicrob. Agents Chemother., 2012, 56, 5149.
7 A. Bianco, V. Reghellin, L. Donnici, S. Fenu, R. Alvarez, C. Barhuffa,
F. Peri, M. Pagani, S. Abrignani, P. Neddermann and R. De Fran-
cesco, PLoS Pathog., 2012, 8, e1002576.
8 A. Leivers, M. Tallant, J. B. Shotwell, S. Dickerson, M. R. Leivers, O. B.
McDonald, J. Gobel, K. L. Creech, S. L. Strum, A. Mathis, S. Rogers, C. B.
Moore and J. Botyanszki, J. Med. Chem., 2013, DOI: 10.1021/jm400781h.
9 T. Balla and A. Balla, Trends Cell Biol., 2006, 16, 351.
10 Multiple isoforms of PI3K and PIP5K were screened but the data for
PI3Ka and PIP5Kg are indicative of wider activity.
11 M. M. Stec, K. L. Andrews, S. K. Booker, S. Caenepeel, D. J. Freeman,
J. Jiang, H. Liao, J. McCarter, E. L. Mullady, T. San Miguel, R. Subramanian,
N. Tamayo, L. Wang, K. Yang, L. P. Zalameda, N. Zhang, P. E. Hughes and
M. H. Norman, J. Med. Chem., 2011, 54, 5174.
Table 3 Effects of 3 and 7 on various endpoints in the PI pathway
Inhibition at 30 mM
12 www.millipore.com.
PI(4,5)P2
13 This is contradictory to literature reports that depletion of PI4P does
not affect steady state levels of PI(4,5)P2, see N. Jethwa, N. Fili and
B. Larijani, J. Chem. Biol., 2012, 5, 137 and references therein.
Compound
IP1 pIC50
PIP
PIP2
PIP3
(basal) (%)
3
7
o4.5
À
À
À
0
19
´
14 P. Varnai and T. Balla, J. Cell Biol., 1998, 143, 501.
6.3
+
+
+
5390 | Chem. Commun., 2014, 50, 5388--5390
This journal is ©The Royal Society of Chemistry 2014