Identification of NVP-BKM120 as a Potent, Selective, Orally Bioavailable Class I PI3 Kinase Inhibitor for Treating Cancer

Abstract

Phosphoinositide-3-kinases (PI3Ks) are important oncology targets due to the deregulation of this signaling pathway in a wide variety of human cancers. Herein we describe the structure guided optimization of a series of 2-morpholino, 4-substituted, 6-heterocyclic pyrimidines where the pharmacokinetic properties were improved by modulating the electronics of the 6-position heterocycle, and the overall druglike properties were fine-tuned further by modification of the 4-position substituent. The resulting 2,4-bismorpholino 6-heterocyclic pyrimidines are potent class I PI3K inhibitors showing mechanism modulation in PI3K dependent cell lines and in vivo efficacy in tumor xenograft models with PI3K pathway deregulation (A2780 ovarian and U87MG glioma). These efforts culminated in the discovery of 15 (NVP-BKM120), currently in Phase II clinical trials for the treatment of cancer.

The phosphoinositide-3-kinase (PI3K) family of lipid kinases is involved in a diverse set of cellular functions, including cell growth, proliferation, motility, differentiation, glucose transport, survival, intracellular trafficking, and membrane ruffling.¹ PI3K’s can be categorized into class I, II, or III, depending on their subunit structure, regulation, and substrate selectivity.² Class IA PI3K’s are activated by receptor tyrosine kinases and consist of a regulatory subunit (p85) and a catalytic subunit (p110). There are three catalytic isoforms: p110α, β, and δ. A single class IB PI3K, activated by GPCRs, consists of only one member: a p110γ catalytic subunit and a p101 regulatory subunit. The primary in vivo substrate of the class I PI3K’s is phosphatidylinositol (4,5) diphosphate (PtdIns(4,5)P2), which upon phosphorylation at the 3-position of the inositol ring to form phosphatidylinositol triphosphate (3,4,5)P3 (PIP3) serves as a second messenger by activating a series of downstream effectors that mediate the cellular functions mentioned above. The PI3K isoforms have different distributions and share similar cellular functions, which are context dependent. In particular, p110α pathway deregulation has been demonstrated in ovarian, breast, colon, and brain cancers.^3,4 Inhibitors of PI3Kα represent an intriguing therapeutic modality for these indications, and as such, there is much interest in generating suitable molecules to test this hypothesis in the clinic.⁵⁻¹⁰

We have previously reported on a series of 6-hydroxyphenyl-2-morpholino pyrimidines,¹¹ as potent pan class I PI3K inhibitors that exhibit high selectivity toward protein kinases (serine/threonine and tyrosine kinases). We have further reported on non-phenol containing heterocyclic, morpholino pyrimidines¹² such as compound 1 which demonstrate in vivo PI3K pathway modulation and modest tumor growth inhibition. Described herein are our efforts to identify potent morpholino pyrimidinyl inhibitors of class I PI3Ks that exhibit potency and pharmacokinetic properties which allow for maximal pathway modulation in vivo and have druglike properties suitable for clinical development. These efforts culminated in the identification of 15, NVP-BKM120.

Aminopyrimidine 1 and analogues such as 3 (Figure 1) exhibit low or sub-nanomolar biochemical potency and sub-micromolar cellular potency against PI3Kα. Even with high rodent CL values, such analogues can demonstrate PI3K pathway modulation in mouse xenograft models.¹² During our exploration of the C₆ position, it was noted that C₆ aminopyridine analogue 4, while being less potent than 3 against PI3Kα (>10× potency loss), exhibited a markedly reduced (>9×) rat CL value, increased %F, and increased oral AUC. Thus, superior pharmacokinetic properties were achievable within this scaffold and the challenge remaining was to retain this kind of pharmacokinetic profile while optimizing all the other attributes (potency, solubility, permeability, safety) necessary for advancement. To address this challenge, the approach taken focused on modifying the electronic properties of the C₆ aminopyridine and C₆ aminopyrimidine moieties. It was hypothesized that electron withdrawing substituents would improve potency in the aminopyridines.¹³ In parallel, a series of substituted pyrimidines was surveyed, assessing the impact of substitution on PK.

Figure 1.

PI3Kα enzymatic potency and rat PK properties of 6-substituted, 4-(aminopyrid-3-yl), 2-morpholino pyrimidines.

To guide the C₆ aminoheterocycle modifications, we looked to the cocrystal structure of aminopyrimidine 2 (X = CH, R = OCH₃) in PI3Kγ to gain an understanding of the aminopyrimidine binding interactions.¹⁴ Given the high homology between the α and γ isoforms and the nanomolar potency of 2 against the two isoforms,¹⁵ p110γ was used as a surrogate for p110α. The cocrystal structure of 2 in the ATP binding site of PI3Kγ (Figure 2) indicates the key binding contacts being made by the aminopyrimidine as well as the morpholine group. The aminopyrimidine interacts via hydrogen bond interactions with Asp836, Asp841, and Tyr867. The C₄′ position of the aminopyrimidine appears to provide a vector to a region of the active site where small groups would be tolerated. The morpholine oxygen forms a known hydrogen bond to the hinge Val882 NH.^16,17 The central pyrimidine C₄ substituent extends out toward solvent and does not appear to make any specific hydrogen bonds. These latter two features are consistent with earlier structures of morpholino pyrimidines.^11,12

Figure 2.

Structure of 2 in PI3Kγ.

With this structural insight, our strategy to find optimal C₆ aminoheterocycles was to substitute the C₄′ position of the C₆ aminopyrimidine or the C₄′ position of the C₆ aminopyridine with small groups that could modulate the electronic properties of the heterocycle. Additionally, it was envisioned that the C₄′ substitution would enforce the nonplanar conformation between the central pyrimidine and the C₆ heterocycle, which could disrupt intermolecular crystal packing and improve aqueous solubility. Finally, upon identification of preferred C₆ aminoheterocycles, further optimization at the central pyrimidine C₄ position to modulate druglike properties and maintain potency was anticipated, since the cocrystal structure indicated this position was partially solvent exposed and would tolerate a range of substituents.

C₄′ modified, C₆ pyridyl or pyrimidyl substituted 2-morpholino 4-aminoquinolyl pyrimidines, synthesized as previously described,¹² were initially screened in biochemical PI3K assays, and compounds with PI3Kα IC₅₀ values < 100 nM were tested in the A2780 ovarian carcinoma cell line (where the PI3K pathway is deregulated due to PTEN deletion) for inhibition of cell proliferation and phosphorylation of AKT^Ser473 as target modulation readout.

The results of the C₄′ modified, C₆ substituent survey (Table 1) indicate that the biochemical potency of the C₆ aminopyridine can be improved by introduction of an electron withdrawing group at the C₄′ position, as is evident in the 3×, 12×, and 20× biochemical potency improvement in CF₃, CN, and Cl replacement of H. Additionally, substitution at the C₄′ position of the aminopyridine does not compromise the rat PK properties, as the CL and AUC are minimally impacted with the CF₃ substitution, 7 vs 4. In contrast to the trends in C₆ aminopyridines, substitution at the C₄′ position of the C₆ aminopyrimidine does not markedly increase the enzymatic or cellular potency. However, substitution at the C₄′ position of the C₆ aminopyrimidine can improve the PK properties, as is evident in 10, where the rat CL is reduced 3-fold and oral AUC increased 3-fold relative to 3.

Table 1. Selected C₆ Aminopyridyl/Pyrimidyl Analogue SAR.

entry	X	R	PI3Kα IC50a (μM)	prolif EC50b (μM)	pAKTSer473 EC50b (μM)	CLc	AUCd	Vsse
3	N	H	<0.002 ± 0.004 (15)	0.146 ± 0.10 (6)	0.042 ± 0.01 (4)	37 ± 5.7	7 ± 2.6	1.9 ± 0.2
4	CH	H	0.11 ± 0.15 (16)	0.54 ± 0.19 (10)	0.31 ± 0.09 (3)	5 ± 0.8	92 ± 22	0.8 ± 0.1
5	CH	Cl	0.0024 ± 0.0004 (2)	0.28 ± 0.14 (4)	0.29
6	CH	CN	0.0041 ± 0.0001 (2)	0.19 ± 0.05 (4)	1.3 ± 0.78 (2)
7	CH	CF3	0.021 ± 0.010 (9)	0.083 ± 0.24 (17)	0.45 ± 0.22 (7)	8 ± 0.3	114 ± 19	2.6 ± 0.4
8	N	NH2	0.0026 ± 0.0001 (2)	0.17	0.18
9	NH	O	0.0074 ± 0.0008 (2)	3.2 ± 1.2 (4)
10	N	CF3	0.0022 ± 0.002 (8)	0.58 ± 0.49 (12)	0.43 ± 0.11 (4)	11. ± 3 2	26 ± 3.4	2.8 ± 0.8
11	N	CH3	0.008 ± 0.009 (8)	0.19 ± 0.11 (8)	0.09

^aATP depletion assay; the value in parentheses is the number of replicates.

^bA2780 cell line.

^cRat, IV, 5 mg/kg dose, mL/(min·kg).

^dRat, oral, 20 mg/kg dose, μM·h.

^eRat, IV, 5 mg/kg dose, L/kg.

While the biochemical and PK parameters can be improved with C₄′ substitution on the C₆ aminoheterocycle, all compounds with the C₄ aminoquinoline exhibit low aqueous solubility, as well as low Caco-2 permeability.¹⁸ The low solubility and permeability potentially could play a role in the lack of correlation between the biochemical potency and cellular activity. Additionally, this solubility range would not be sufficient for further development if concentrations in excess of the cell activity (0.1–1 μM) would need to be achieved in vivo for efficacy, as well as not sufficient to assess the safety profile of the compound.

With these considerations in mind, we further explored C₄ position SAR with the preferred C₄′ substituted C₆ aminoheterocycles, to arrive at potent PI3K inhibitors with improved solubility. From earlier work¹² it was known that the C₄ position could tolerate a wide range of moieties. Additionally, it was known that a morpholine group at the C₄ position maintained reasonable potency while improving solubility relative to the aminoquinoline (compound 12, Table 2). As such, compounds which contained the C₄ morpholine and the C₄′ substituted C₆ aminopyridines or aminopyrimidines were prepared. The properties of these analogues are shown in Table 2.

Table 2. Selected 2,4-Bismorpholino Pyrimidyl Analogue SAR.

entry	X	R	PI3Kα IC50a (μM)	pAKTSer473 EC50b (μM)	prolif EC50b (μM)	CLc	AUCd	Vsse	%F	solf
12	N	H	0.0184 ± 0.001 (2)	0.10	0.93 ± 0.11 (2)					25
13	CH	Cl	0.034 ± 0.006 (6)	0.11	1.77 ± 0.68 (4)	22 ± 5	29 ± 6	1.11 ± 0.7	69 ± 14	45
14	CH	CN	0.068 ± 0.014 (5)	0.10 ± 0.10 (3)	0.51 ± 0.48 (3)	11 ± 2	86 ± 5	1.6 ± 0.1	98 ± 6	11
15	CH	CF3	0.030 ± 0.017 (24)	0.055 ± 0.02 (11)	0.52 ± 0.50 (30)	3 ± 0.4	178 ± 3	3.1 ± 0.4	50 ± 7	132
16	N	NH2	0.023 ± 0.011 (10)	0.079 ± 0.03 (11)	1.46 ± 2.0 (33)	9 ± 0.1	64 ± 7	4.5 ± 0.9	68 ± 14	8
17	NH	O	0.015 ± 0.008 (8)	0.073 ± 0.01 (2)	0.68 ± 0.22 (4)	24 ± 4	24 ± 3	2.8 ± 1.2	56 ± 15	22
18	N	CF3	0.014 ± 0.006 (14)	0.070 ± 0.03 (3)	0.45 ± 0.10 (3)	11 ± 2	62 ± 5	2.1 ± 0.2	83 ± 3	61

^aATP depletion assay; the value in parentheses is the number of replicates.

^bA2780 cell line.

^cRat, IV,¹⁹ mL/(min·kg).

^dRat, oral,¹⁹ μM·h.

^eRat, IV,¹⁹ L/kg.

^fSolubility at pH 7, μM

All bis morpholino compounds demonstrate biochemical activity against PI3K in the nanomolar range, mechanism modulation <300 nM, and inhibition of cell proliferation approximately between 0.5 and 1.5 μM. Additionally, the solubility at pH 7 was markedly improved for all bis morpholino compounds (at least 10–1000×) relative to the C₄ aminoquinoline analogues. This increased solubility, along with the high permeability, vide infra, may account for the comparable cellular potency of the bismorpholino compounds in Table 2 relative to the more biochemically potent but less soluble aminoquinoline compounds in Table 1. The additional morpholine at the C₄ position did not compromise the rat PK parameters,¹⁹ as, for both C₆ substituted aminopyridines as well as C₆ substituted aminopyrimidines in Table 2, low to moderate CL and moderate to high %F were observed. All bismorpholino compounds in Table 2 exhibit high stability in both rat and human liver microsomes, with Cl_int ≤ 9 μL/min/mg in both species.

Compound 15 was of particular interest, as the solubility was the highest (170 μM on crystalline material) and it was among the most potent bismorpholine compounds in cell based assays (50 nM target modulation; 500 nM cell proliferation). Figure 3 shows the cocrystal structure of 15 in PI3Kγ.²⁰ The binding mode was as expected with one of the morpholines binding to the hinge at Val882 and the aminopyridine group interacting via hydrogen bonds to Asp836, Asp841, and Tyr867.

Figure 3.

Structure of 15, NVP-BKM120, in PI3Kγ.

The biochemical activity of 15 was assessed across class I PI3K’s, related lipid kinases, and against more than 200 protein kinases. Some of the data are shown in Table 3. Compound 15 exhibited 50–300 nM activity for class I PI3K’s, including the most common p110α mutants. Additionally, 15 exhibited lower potency against class III and class IV PI3K's, where 2, 5, >5, and >25 μM biochemical activity was observed for inhibition of VPS34, mTOR, DNAPK, and PI4K, respectively. No significant activity was observed against the protein kinases tested.

Table 3. Biochemical Activity of 15, NVP-BKM120, against Class I, III, IV, PI3, and Related Kinases.

class	enzyme	IC50 (μM)	enzyme	IC50 (μM)
Class I PI3K'sa	p110α	0.052 ± 0.037	p110β	0.166 ± 0.029
	p110α–H1047R	0.058 ± 0.002	p110δ	0.116
	p110α–E545K	0.099 ± 0.006	p110γ	0.262 ± 0.094
Class III PI3K'sb	VPS34	2.4 ± 1.5
Class IV PI3K's	mTORc	4.6 ± 1.9	DNAPKd	>5
PI4Kb	PI4Kβ	>25

^aFilter binding assay.

^bATP depletion assay.

^cTR-FRET assay.

^dPromega SignaTECT.

In vitro evaluation of 15 across a range of PI3K deregulated cell lines from a variety of tumor types, including ovarian, glioblastoma, breast, and prostate, was conducted (Table 4). Across all cell lines, pathway modulation and antiproliferative activity was consistent with cellular PI3K inhibition.

Table 4. Cellular Activity of 15 across Multiple Cell Lines with PI3K Pathway Deregulation.

cell line	genotype	pAKTSer473 EC50 (μM)	GI50 (nM)
A2780	PTEN del	0.074	0.635
U87MG	PTEN del	0.13	0.698
MCF7	E545K–PIK3CA	<0.100	0.158
DU145	LKB1 mutant	0.073	0.435

The behavior consistent with selective in vitro inhibition of class I PI3K’s translated to in vivo settings in two models of PI3K-AKT pathway driven cancers: the A2780 ovarian carcinoma and the U87MG glioma model, which carry a PTEN deletion. In A2780 xenograft tumors (Figure 4), oral dosing of 15 at 3, 10, 30, 60, and 100 mg/kg resulted in a dose dependent modulation of pAKT^Ser473. Partial inhibition of pAKT^ser473 was observed at 3 and 10 mg/kg, and near complete inhibition was observed at doses of 30, 60, or 100 mg/kg, respectively. Inhibition of pAKT (normalized to total AKT) tracked well with both plasma and tumor drug exposure. pAKT modulation was also time dependent, with >90% target modulation achieved with the 60 and 100 mg/kg dose at the 10 h time point when the plasma and tumor exposure was ca. 2 μM.²¹

Figure 4.

pAKT^ser473 inhibition and plasma exposure of 15, NVP-BKM120, in an A2780 xenograft model.

Consistent with the mechanism modulation observed, significant tumor growth inhibition was obtained in the A2780 tumor model at 60 mg/kg upon multiple dosing (Figure 5). Efficacy was associated with significant inhibition of the PI3K pathway, as assessed by reduction in pAkt^Ser473, for up to 16 h.

Figure 5.

Efficacy of compound 15, NVP-BKM120, in an A2780 xenograft model.

As was the case in vitro, 15 displays in vivo activity across a range of PI3K pathway deregulated tumor xenograft models.²² In the established U87MG glioma model, significant single agent activity was obtained with 15 at daily oral doses of 30 and 60 mg/kg (Figure 6) in a well tolerated manner.²³ This activity in the U87MG model, coupled with the high permeability and lack of efflux exhibited by 15,²⁴ suggests that 15 may have utility in PI3K-driven gliomas.

Figure 6.

Antitumor activity of 15, NVP-BKM120, against the subcutaneous U87MG glioma tumor model.

With these encouraging rodent pharmacology activities, compound 15 was studied further. Profiling indicated that 15 exhibited no reversible or time dependent CYP450 (3A4, 2C9, 2D6) inhibition up to 50 μM or CYP3A4 induction up to 25 μM, exhibited high permeability with no propensity for efflux,²³ demonstrated no cardiotoxicity potential,²⁵ and showed a clean profile (>10 μM) against enzymes, receptors, and transporters included in internal safety and the external MDS Pharma Services panels. The melting point of 15 is 153 °C, its log D (pH 7.4) is 2.9, and its pK_a is 5.1. The synthesis of 15 is straightforward, proceeding in four steps for the research route (Scheme 1).

Scheme 1. Synthesis of 15.

The pharmacokinetic properties of 15 were evaluated across multiple species (Table 5). Low to moderate CL was observed for 15 across species, as CL values of 11, 3, 13, and 7 mL/(min kg) were observed in mouse, rat, dog, and monkey, respectively. Additionally, 15 exhibited medium to high oral bioavailability across species as 80%, 50%, 44%, and 100% was observed in mouse, rat, dog, and monkey, respectively.

Table 5. PK Properties of 15 across Species²⁶.

	IV			PO
species	t1/2a	CLb	Vssc	%F
mouse	1.6	11	1.4	71–89
rat	11	3 ± 0.4	3.1 ± 0.4	50
dog	6.6 ± 0.8	7.3 ± 1.5	3.1 ± 0.4	>100
monkey	3.6	13	3.1	44

^aUnits of hour.

^bUnits of mL/(min kg).

^cUnits of L/kg.

With the favorable cellular potency, kinase selectivity, preclinical pharmacology, rodent, dog, and monkey pharmacokinetics, physical properties, and preclinical safety profile, 15 was advanced into clinical trials in 2008 and has since shown clinical activity in patients with cancer.²⁷ Specifically promising activity has been observed in those patients with an activated PI3K pathway.²⁸

In summary, we have described the structure guided optimization of a series of 6-aminoheterocyclic, 4-substituted, 2-morpholino pyrimidines which exhibited high CL and low aqueous solubility into a compound suitable for clinical development. Modification of the aminoheterocycle with small groups that modulated the ring electronics either improved potency or reduced the in vivo CL values. Incorporation of a morpholine group at the C₄ central pyrimidine position increased the aqueous solubility while retaining sufficient potency, selectivity, and favorable in vivo properties. The combination of these modifications led to the discovery of a series of substituted 6-aminoheterocyclic, 2,4-bismorpholino pyrimidines. From this series, compound 15 (NVP-BKM120) has advanced into humans and is currently being assessed in phase II trials.²⁹

Glossary

Abbreviations

PI3K

phosphoinositide-3-kinase

DNAPK

DNA dependent protein kinase

PI4K

1-phosphatidylinositol-4-kinase

mTOR

mammalian target of rapamycin

PTEN

phosphatase and tensin homologue

NBS

N-bromosuccinimide

DIEA

diisopropylethylamine

Pd(dppf)Cl₂-DCM

dichloro[1,1′-bis(diphenylphosphino)ferrocene] palladium(II) dichloromethane adduct

DIEA

diisopropylethylamine

DME

dimethoxyethane

CYP

cytochrome P450

CL

clearance

PK

pharmacokinetics

Supporting Information Available

Experimental details for the synthesis and characterization of all compounds, biological assay, and pharmacology model procedures. This material is available free of charge via the Internet at http://pubs.acs.org.