Potent Inhibitors of Phosphatidylinositol3 (PI3) Kinase that have Antiproliferative Activity Only When Delivered as Prodrug Forms

Article abstract of DOI:10.1002/cmdc.201200583

Prodrugs for PI3K: A series of substituted analogues of the phosphatidylinositol3 kinase (PI3K) inhibitor LY294002 were prepared and found to potently inhibit the isolated enzyme but not MCF7 cell proliferation. Two tetrazolyl-substituted analogues were further derivatized as prodrugs resulting in restoration of cell-based activity. These data provide a conceptual model for development of tumor-targeting prodrug forms of cell-impermeable PI3K inhibitors. Copyright

Full text of DOI:10.1002/cmdc.201200583

CHEMMEDCHEM
COMMUNICATIONS
DOI: 10.1002/cmdc.201200583
Potent Inhibitors of Phosphatidylinositol 3 (PI3) Kinase
that have Antiproliferative Activity Only When Delivered
as Prodrug Forms
Nathan J. O’Brien,[a] Syazwani Amran,[b] Jelena Medan,[a] Ben Cleary,[a] Leslie W. Deady,[a]
Ian G. Jennings,[b] Philip E. Thompson,[b] and Belinda M. Abbott*[a]
Phosphatidylinositol 3 kinase (PI3K) enzymes have emerged as
important therapeutic targets in recent years as PI3K isoforms
are involved in numerous cell signaling pathways associated
with a range of diseases.[1] Prominent among these is cancer,
with numerous instances of genetic mutations that enhance
the PI3K signal identified, such as mutations in the PI3KCA
gene that expresses PI3Ka, and downregulation of phospha-
tase PTEN that attenuates the PI3K signal.
LY294002 is a small-molecule PI3K inhibitor that has provid-
ed much of the pharmacological characterization of PI3K func-
tions.[2] It inhibits all of the four class I isoforms of PI3K, with
a Ki value of approximately 1 mm at each. It was also co-crystal-
lized with the PI3Kg isoform in 2001, providing one of the first
inhibitor-bound structures of the enzyme.[3] LY249002 has been
shown to inhibit tumor cell proliferation, angiogenesis and mi-
gration via inhibition of PI3K. However, LY249002 had not pro-
gressed as a therapeutic due to a range of limitations, such as
low potency and solubility.
In recent years, LY294002 has been superseded by a range
of new inhibitors with improved potency, isoform selectivity
and/or pharmaceutical profiles, and these agents are currently
under clinical investigation.[1] They include GDC0941, TGX-221,
and NVP-BEZ235, among others (Figure 1a).
Others have sought to improve upon LY294002 using a tar-
geted prodrug approach (Figure 1b). SF1126 is a prodrug that
conjugates LY294002 to the amino terminus of the Arg–Gly–
Asp–Ser integrin binding tetrapeptide via a morpholinium
methylsuccinate, and it was shown to have potent antitumor
activity against several human tumor types.[4] The conjugated
form exhibits improved cellular permeability, which enhances
the delivery of the active compound into the cell. In a compara-
ble manner, TGX-221, a potent PI3Kb inhibitor, has been pur-
sued as a targeted prodrug, with TGX-D1 composed of an anti-
human epidermal growth factor receptor 2 (HER2)-specific pep-
Figure 1. a) Known and b) reported phosphatidylinositol 3 kinase (PI3K)
inhibitors.
tide linked to an N-hydroxyethyl-TGX-221 analogue via a pros-
tate-specific antigen (PSA) substrate linker group.[5] The target-
specific conjugate exhibits increased cellular uptake compared
with the parent drug and improved selectivity for HER2 cancer
cells. More recently, an analogue of LY294002 has also been
developed as a peptidic prodrug to target PSA protease.[6]
In light of recent developments, our research has also exam-
ined derivatization of LY294002 and the ways in which com-
pounds might be effectively delivered as prodrugs. In our prior
studies of LY294002, we identified that elaboration through
the 6-position with a methyl or methoxy substituent was toler-
[a] N. J. O’Brien, J. Medan, B. Cleary, Dr. L. W. Deady, Dr. B. M. Abbott
Department of Chemistry
La Trobe University Institute for Molecular Science
La Trobe University
Bundoora, Victoria 3086 (Australia)
E-mail: b.abbott@latrobe.edu.au
[b] S. Amran, Dr. I. G. Jennings, Dr. P. E. Thompson
Medicinal Chemistry
Monash Institute of Pharmaceutical Sciences
Monash University
Parkville, Victoria 3052 (Australia)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201200583.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 914 – 918 914

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
ated, and our examination of the co-crystal with PI3Kg (PDB:
1E7V[3]) showed that substituents in the 6-position would proj-
ect in a comparable position to the second phosphate group
of ATP (PDB: 1E8X[3]), which interacts with Ser806 of the
enzyme.[7] On that basis, we considered that incorporation of
polar or negatively charged groups might lead to increased
affinity.
Hydrolysis of compounds 5 and 6 using sulfuric acid and mild
heating gave desired carboxamides 7 and 8 in good yields.
These products were found to have very poor solubility in or-
ganic solvents and required heating in deuterated dimethyl
sulfoxide before NMR analysis.
Attempts to hydrolyze morpholinyl carboxamide 7 to the
corresponding carboxylic acid (9) were unsuccessful due to
concomitant hydrolysis occurring at the conjugated amine
connected to the morpholine ring. Pyridinyl analogue 10a did
not pose the same difficulties and was obtained in 75% yield.
With carboxylic acid 10a in hand, methyl (10b), ethyl (10c),
and propyl (10d) esters were synthesized by Fischer esterifica-
tion in yields of 66%, 62% and 44%, respectively. The corre-
sponding benzyl ester (10e) was prepared via the active imida-
zole ester in 46% yield.[9] The same activated intermediate was
also treated with benzylamine to yield benzylamide 10 f in
46% yield. The nitrile precursors 5 and 6 were also subjected
to [2+3] cycloaddition with sodium azide to yield the corre-
sponding tetrazole derivatives (11 and 12).[10] The NMR spectra
of compounds 11 and 12 were obtained in deuterated dimeth-
yl sulfoxide at 340 K due to solubility problems at lower tem-
peratures.
Here, we describe a range of LY294002 analogues with sub-
stitutions at the 6-position that are potent inhibitors of purified
recombinant PI3Ka. These compounds were inactive against
cancer cells, which we attributed to very poor cell permeability.
We then developed prodrug forms that show no intrinsic PI3K
activity, but effectively inhibit cancer cell proliferation, which
we postulate is due to intracellular release of the active com-
pounds. We believe these studies provide a platform for the
development of new therapeutics that are exquisitely targeted
to tumor tissue, as they can only express activity when deliv-
ered as a prodrug because premature hydrolysis results in
a cell-impermeable compound.
Syntheses of various 6-substituted analogues of LY294002
and its pyridinyl analogue were pursued via key nitrile precur-
sors 5 and 6 (Scheme 1). Precursors were achieved via path-
ways similar to those previously described.[7a,8] First, 3-acetyl-5-
bromo-4-hydroxybenzonitrile (2) was prepared from 4-cyano-
phenol (1) in three steps and 52% yield. Elaboration of 2-hy-
droxyacetophenone 2 through to the 6-cyano analogue of
LY294002 (5) was achieved in 19% yield. Synthesis of the anal-
ogous 2-(4-pyridinyl) derivative (6) was achieved from 2 in
14% yield.
The products from these syntheses were first evaluated as
inhibitors of PI3Ka by screening at 10 mm. As shown in
Figure 2, a range of potencies was displayed, with four com-
pounds (7, 10a, 11 and 12) showing near-complete blockade.
These compounds were selected for full dose–response assays,
with compounds 11 and 12 exhibiting in IC50 values of 20 nm
and 25 nm, respectively. Compounds 10a and 7 were some-
what less active, with IC50 values of 450 nm and 1.7 mm, respec-
tively. All four compounds were also found to have similar
Functional group manipulation at the nitrile furnished car-
boxamides, a carboxylic acid, and a series of ester derivatives.
Scheme 1. Reagents and conditions: a) AcCl, Et3N, THF, 08C!RT, 12 h; b) AlCl3, 1808C, 3 h; c) NBS, CH3CN, RT, 12 h; d) LiHMDS, ꢀ788C!08C, 1 h; then mor-
pholinecarbonyl chloride, ꢀ788C!RT, 12 h; e) Tf2O, CH2Cl2, RT, 12 h; f) LiHMDS, ꢀ788C!08C, 1 h; then methyl isonicotinate ꢀ788C!RT, 12 h; then H2SO4,
AcOH, 808C, 3 h; g) Pd(OAc)2, PhB(OH)2, DMF, 808C, 12 h; h) 80% H2SO4, 558C, 1 h; i) 80% H2SO4, NaNO2, 558C, 20 min; j) R’H, H2SO4, reflux, 12 h; k) 1,1’-car-
bonyldiimidazole (CDI), 1,4-dioxane, then PhCH2OH, reflux, 24 h; l) CDI, 1,4-dioxane, then PhCH2NH2, reflux, 12 h; m) NaN3, NH3Cl, DMF, reflux, 3 d.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 914 – 918 915

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
Figure 2. Results of an in vitro assay for inhibition of phosphatidylinositol 3
kinase (PI3K) enzyme activity. Inhibitors were assayed at 10 mm for inhibition
of PI3Ka.
Scheme 2. Reagents and conditions: a) cat. ZnCl2, CH2Cl2, 08C!258C, 2.5 h;
b) K2CO3, CH3CN, reflux, 4 d.
potency across all four class I PI3K isoforms (Supporting Infor-
mation).
Compounds 11 and 12 were then tested for their ability to
inhibit proliferation of MCF7 cells, a model breast cancer cell
line. They were found to be profoundly inactive at concentra-
tions up to 20 mm. The same was found against cultured
HCT116 colon cancer cells. The lack of cellular activity of the
tetrazolyl analogues was hypothesized to be due to the com-
pounds inability to cross the cell membrane. Tetrazoles, while
well known as carboxylic acid isosteres, have also been identi-
fied as having limitations with respect to oral activity, support-
ing the explanation that such compounds might not be able
to traverse the cell membrane.
Both prodrug forms were inactive against the isolated enzyme
but found to be stable to mildly acidic aqueous conditions
that might be used for analysis (i.e., 1n HCl, 0.1% aq trifluoro-
acetic acid). Both 14 and 15 were then assessed as substrates
for pig liver esterase, and the prodrugs were efficiently con-
verted into the free tetrazolyl products with a half-life of ap-
proximately 60 min (Figure 3a and Supporting Information).
Finally, prodrugs 14 and 15 were tested as inhibitors of cell
proliferation under identical conditions to the previously inac-
tive tetrazole compounds (11 and 12). Gratifyingly, both pro-
drugs were found to be effective inhibitors of MCF7 cell prolif-
eration (Figure 3b), consistent with a sequence of cellular
uptake, intracellular hydrolysis and inhibition of PI3K. Pyridinyl
compound 15 had an IC50 value of 2 mm, while morpholinyl
compound 14 was less potent with an IC50 value of 18 mm
(Table 1). The known cell-permeable inhibitor GDC0941 had an
IC50 value of 50 nm under the same assay conditions.[14] Com-
pounds 11 and 12 are markedly more potent (50- and 40-fold,
respectively) than the IC50 value of 1 mm for LY294002 against
p110a[15] and were comparable in potency to GDC0941, as
shown in Table 1.
A prodrug strategy was pursued to address the putative cell
impermeability of 11 and 12.[11] We postulated that a prodrug
form would be taken up into cells and that the prodrug group
would be removed by in situ cleavage. As a model system, we
chose a known pivaloyl ester group described previously in
a number of examples including tetrazoles.[12]
The prodrug precursor 1-chloro-2-methylpropyl pivalate (13)
was prepared by reaction of isobutyraldehyde and pivaloyl
chloride in the presence of catalytic zinc chloride. The alkyla-
tion reactions followed the procedure of Ryono et al.[11] Tetra-
zole 11 was refluxed with 13 in acetonitrile. After four days at
reflux, substantial starting material remained but the desired
pivalate (14) was retrieved in 34% yield after HPLC purification
(Scheme 2). It is well established that 5-aryltetrazoles (exclud-
ing ortho-substituted derivatives) are alkylated preferentially at
N2.[13] A trace of the alternate 1-substituted regioisomer was
also observed but not isolated.
This work reports two key findings. Firstly, we have shown
that inclusion of a prosthetic acidic or polar group at the 6-po-
sition of LY294002 results in a dramatic increase in affinity, at-
tributable to a specific interaction with the serine residue
(Ser806 in g isoform), which is a conserved residue of the
active site that typically binds the second phosphate of ATP.
This interaction could well be able to be further exploited in
other PI3K inhibitors to enhance potency.
Alkylation of pyridinyl analogue 12 was performed under
similar conditions, and the major regioisomer (15) was ob-
tained in just 8% yield. It should be noted that the NMR spec-
trum of 15 was poorly resolved with respect to the pyridinyl
Secondly, we have shown that tetrazolyl inhibitors 11 and
12, while potent inhibitors of purified enzyme activity in vitro,
are inactive in cellular assays, possibly due to poor permeabili-
ty across cell membranes. However, functional inhibition of
PI3K can be achieved by conversion of the tetrazolyl com-
pounds to a prodrug form. We have shown that prodrug 14
and 15 are susceptible to a model esterase, and are cleaved at
a comparable rate. The two parent forms have approximately
equivalent inhibitory potency against purified enzyme but
1
13C and H nuclei, with missing or broad peaks attributed to
the rotation of the pyridine ring causing deshielding effects
from the 8-phenyl ring. Variable temperature experiments sup-
ported this hypothesis.
With prodrug candidates 14 and 15 in hand, in vitro PI3K in-
hibition and the stability to various conditions were assayed.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 914 – 918 916

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
show an order of magnitude difference in cell-based assays,
and GDC0941, which has similar potency against PI3K, has
a further tenfold increase in cellular potency. To explain these
results, we believe that the efficiency of uptake of the prodrug
forms is likely to be a key issue, and that the prodrugs might
also be degraded extracellularly diminishing the exposure of
the cells to the active drug form. Intracellular degradation, or
possibly other metabolic processes, could also alter the ability
of the inhibitors to block cell proliferation. Nonetheless, there
is significant encouragement in this demonstration of cellular
activity of the prodrug form.
These findings could have important consequences in future
therapeutic design, as PI3K is ubiquitously expressed and the
prospect of side effects due to nonselective tissue targeting is
a significant concern expressed in the literature.[1] More exten-
sive studies on the effects of these inhibitors on cell signaling
pathways, for example pAkt activation, will be required to con-
firm that the mechanism of action of these inhibitors is truly
intracellular PI3K inhibition as a result of prodrug hydrolysis.
Nonetheless, our experiments together with the results from
compounds SF1126 and TGX-D1 suggest that there might be
a mechanism for using specific targeting prodrugs to gain
access to target tissues, such as tumors.[4,5] Successful targeting
will deliver the therapeutic into the cells, hopefully to produce
the desired efficacy. Unsuccessful delivery and prodrug degra-
dation yields a cell-impermeable product that has no opportu-
nity to enter nontarget cells and thus has decreased side effect
potential. As compared with the other examples of targeting
prodrugs, which use cell-permeable active species, this strat-
egy offers a distinct class of compounds for analysis.
*
Figure 3. a) Time course of prodrug digestion of compounds 14 ( ) and 15
&
( ) by purified bovine pig liver esterase with percentage digestion, as deter-
mined by RP-HPLC (see Supporting Information), plotted against incubation
^
*
)
time. b) Inhibition of MCF7 cell proliferation by compounds 14 ( ), 15 (
~
and GDC0941 ( ). Compounds were tested at the indicated concentrations
for effect on cell proliferation measured as described in the Supporting In-
formation. Duplicate determinations are shown. Curves were fitted using
GraphPad Prism software.
Acknowledgements
The authors warmly thank Dr. Oleg Schmidt-Kittler (Kimmel
Cancer Center, John Hopkins University) for testing compounds 7,
10a, 11 and 12 against the four class I PI3K isoforms. This re-
search was funded by the Australian Federal Government
through the provision of an Australian Postgraduate Award to
S.A.
Table 1. Effect of prodrug and active product on PI3K enzyme activity in-
hibition and MCF7 cell proliferation.
Keywords: inhibitors
3-kinase · PI3K · prodrugs · tetrazoles
·
LY294002
·
phosphatidylinositol
Inhibitor
IC50 [mm]
p110a[a]
MCF7[b]
GDC0941
LY294002
7
0.036
1.0[15]
1.7
0.05ꢁ0.01
n.d.
n.d.
[1] S. J. Shuttleworth, F. A. Silva, A. R. L. Cecil, C. D. Tomassi, T. J. Hill, F. I.
Raynaud, P. A. Clarke, P. Workman, Curr. Med. Chem. 2011, 18, 2686–
2714.
[2] C. J. Vlahos, W. F. Matter, K. Y. Hui, R. F. Brown, J. Biol. Chem. 1994, 269,
5241–5248.
[3] E. H. Walker, M. E. Pacold, O. Perisic, L. Stephens, P. T. Hawkins, M. P.
Wymann, R. L. Williams, Mol. Cell 2000, 6, 909–919.
[4] J. R. Garlich, P. De, N. Dey, J. D. Su, X. Peng, A. Miller, R. Murali, Y. Lu,
G. B. Mills, V. Kundra, H. K. Shu, Q. Peng, D. L. Durden, Cancer Res. 2008,
68, 206–215.
[5] W. Tai, R. S. Shukla, B. Qin, B. Li, K. Cheng, Mol. Pharmaceutics 2011, 8,
901–912.
[6] D. Baiz, T. A. Pinder, S. Hassan, Y. Karpova, F. Salsbury, M. E. Welker, G.
Kulik, J. Med. Chem. 2012, 55, 8038–8046.
[7] a) B. Abbott, P. Thompson, Aust. J. Chem. 2003, 56, 1099–1106; b) A. D.
Robertson, S. P. Jackson, V. Kenche, C. L. Yaip, H. Prabaharan, P. E.
10a
11
12
14
15
0.45
n.d.
0.025
0.026
>10
>20
>20
17.7ꢁ2.3
2.1ꢁ1.1
>10
[a] PI3K p110a enzyme activity; IC50 values were determined from graphs
of activity versus inhibitor concentration using the GraphPad Prism soft-
ware; data represent the mean (n=2–4 independent experiments); the
individual IC50 values were within 20% of each other. [b] MCF7 cell
growth inhibition; IC50 values were determined from graphs of cell
number versus inhibitor concentration using the GraphPad Prism soft-
ware; data represent the meanꢁ SEM (n=3–6 independent experi-
ments).
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 914 – 918 917

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
Thompson, (Thrombogenix Pty Ltd, Toorak, Australia), Int. Pat. Appl.
WO/2001/053266 A1 (Eur. Pat. Specif. EP 1257537 B1), 2001; c) B. M.
Abbott, P. E. Thompson, Bioorg. Med. Chem. Lett. 2004, 14, 2847–2851.
[8] a) S. P. Jackson, A. D. Robertson, V. Kenche, P. Thompson, H. Prabaharan,
K. Anderson, B. Abbott, I. Goncalves, W. Nesbitt, S. Schoenwaelder, D.
Saylik (Kinacia Pty Ltd, Box Hill, Australia), Int. Pat. WO/2004/016607,
2004; b) S P. Jackson, A. D. Robertson, V. Kenche, P. Thompson, H. Pra-
baharan, K. Anderson, B. Abbott, I. Goncalves, W. Nesbitt, S. Shoe-
nwaelder, D. Saylik, (AstraZeneca AB, Sodertalje, Sweden), US Pat.
US 7,598,377, 2009 (Priority data: 60/403,639 on Aug 16, 2002 and 60/
428,283 on Nov 22, 2002).
Petrillo, J. R. Powell, T. Schaeffer, E. R. Spitzmiller, R. E. White, Bioorg.
Med. Chem. Lett. 1994, 4, 201–206.
[12] M. T. Obermeier, S. H. Chong, S. A. Dando, A. M. Marino, D. E. Ryono, A.
StarrettArroyo, G. C. DiDonato, B. M. Warrack, R. E. White, R. A. Morrison,
J. Pharm. Sci. 1996, 85, 828–833.
[13] a) L. Huff, R. A. Henry, J. Med. Chem. 1970, 13, 777–779; b) R. J. Spear,
Aust. J. Chem. 1984, 37, 2453–2468.
[14] C. O’Brien, J. J. Wallin, D. Sampath, D. GuhaThakurta, H. Savage, E. A.
Punnoose, J. Guan, L. Berry, W. W. Prior, L. C. Amler, M. Belvin, L. S. Fried-
man, M. R. Lackner, Clin. Cancer Res. 2010, 16, 3670–3683.
[15] Z. H. Zheng, S. I. Amran, P. E. Thompson, I. G. Jennings, Mol. Pharmacol.
2011, 80, 657–664.
[9] L. W. Deady, M. L. Rogers, L. Zhuang, B. C. Baguley, W. A. Denny, Bioorg.
Med. Chem. 2005, 13, 1341–1355.
[10] M. E. Zwaagstra, H. Timmerman, M. Tamura, T. Tohma, Y. Wada, K.
Onogi, M. Q. Zhang, J. Med. Chem. 1997, 40, 1075–1089.
Received: December 19, 2012
[11] D. E. Ryono, J. Lloyd, M. A. Poss, J. E. Bird, J. Buote, S. Chong, T. Dejneka,
K. E. J. Dickinson, Z. Gu, P. Mathers, S. Moreland, R. A. Morrison, E. W.
Published online on April 8, 2013
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 914 – 918 918