Part 3: Notch-Sparing γ-Secretase Inhibitors: SAR Studies of 2-Substituted Aminopyridopyrimidinones

Abstract

In search for novel lead compounds as γ-secretase inhibitors, analogs of aminopyrido[2,3-d]pyrimidin-7-ones (I) were synthesized and evaluated for inhibitory effects on amyloid –β-peptide production and cleavage of the Notch1 receptor mediated by γ-secretase. Selected pyridopyrimidines, such as 1, 8, 9, 10, 11 and 16 are γ-secretase inhibitors that did not have an effect on Notch1 receptor processing.

Graphical Abstract

Alzheimer’s disease (AD) is a progressive and fatal neurodegenerative disease that currently affects over 5 million Americans.¹ AD is characterized by slow but progressive impairment in cognition, behavior and daily-life function. The main pathological hallmarks of AD are the extracellular deposition of amyloid-β (Aβ) peptides and the formation of intracellular neurofibrillary tangles.² To date, treatment of AD involves drugs that can only delay the inevitable AD symptoms without tackling the main neuropathological causes of the disease. Scientific advances in understanding AD etiology has led to the identification of the oligomerization of Aβ peptides in the brain as a key causal factor for AD.³ Accumulation of Aβ oligomers begins in the brain long before the onset of AD as a result of reduced Aβ clearance or increased Aβ production.⁴ Aβ peptides are produced from the amyloid precursor protein (APP) through sequential proteolytic cleavages catalyzed by β- and β-secretases.⁶ Hence, suppressing Aβ production by inhibiting γ-secretase has been proposed as one of the disease-modifying approaches for AD therapy.⁷

γ-Secretase is a protein complex composed of nicastrin, Aph-1, Pen-2, and either presenilin-1 (PS-1) or presenilin-2 (PS-2).⁸ Cleavage of Notch receptors, required for their cell signaling, is also catalyzed by γ-secretase.⁹ Interference with Notch receptor signaling, particularly from Notch1, can lead to severe side effects, including intestinal goblet cell hyperplasia, thymus atrophy, decrease in lymphocytes, and alterations in hair color.¹⁰ Our strategy for seeking AD therapeutics has been Notch-sparing gamma-secretase inhibitors. Greengard and colleagues discovered that Aβ production by γ-secretase in cells is dependent on adenosine triphosphate (ATP).¹¹ It was reported that compounds such as Gleevec™ (imatinib mesylate) and PD173955 (Figure 1) compete with ATP for binding to target kinases and can also inhibit Aβ production by γ-secretase with selectivity over the cleavage of Notch1. Coincidentally, both Gleevec and PD173955 incorporate a phenyl aminopyrimidine pharmacophore.

Figure 1.

Structure of Imatinib and PD173955

Our early screening focus for this series of compounds was evaluating permutations at the C-2 pyrimidine position of the template of PD173955. In Table 1, when R¹ is hydrogen (as in PD173955) and R² is 4-OCH₃-phenyl (1, 47%) and 4-SCH₃-phenyl (2, 47%), overall better activity and selectivity was observed. There also was no change observed for Notch1 processing for either of these compounds. Interestingly, the 4-thiomethylphenyl analog (2) showed Aβ40 inhibition whereas the 3-thiomethylphenyl analog PD173955 (1) showed no Aβ40 inhibition. In addition, 4-CF₃-phenyl analog (4) showed a significant decrease in activity.

Table 1.

2-Amino-6-(2,6-dichlorophenyl)-pyrido[2,3-d]-pyrimidin-7-ones

Entry	R1	R2	Aβ40 (%) Inhibition a	Notch1 Processing b
PD173955	H	3-(CH3S)Ph-	0	No change
1	H	4-(CH3O)Ph-	47	No change
2	H	4-(CH3S)Ph-	47	No change
3	H	Ph-	26	n.t.
4	H	4-(CF3)Ph-	12	n.t.
5	H	i-Pr	0	No change
6	H	n-Hexyl	53	Inhibition
7	H	PhCH2-	0	n.t.
8	H	(2-Pyridinyl)CH2-	59	No change
9	CH3	Ph(CH2)2-	48	No change
10	i-Pr	PhCH2-	90	No change
11	i-Pr	(3,4-Cl2Ph)CH2-	50	No change
12	i-Pr	(3,5-F2Ph)CH2-	73	Inhibition

n.t = not tested.

^a,bSee References and notes.^{15, 20}

^aCompounds were tested at the concentration of 100 μM. The inhibition of Aβ40 production is a percentage of DMSO control

^bCompounds were tested at the concentration of 100 μM. The effects on Notch1 processing are compared to DMSO control; No change: Notch1 processing was not affected by the tested compounds at the concentration of 100 μM.

Phenyl and benzyl analogs (3 and 7) and small alkyl (R² = isopropyl, 5) led to decreased or no activity (26%, 0%, 0%, respectively). However, a longer alkyl chain analog (R² = n-hexyl, 6) and a methylpyridinyl analog (8) showed a significant increase in potency (53%, 59%, respectively), with (6) also showing inhibition of Notch1 processing. Additionally, activity increased substantially when the hydrogen of (7) was replaced with isopropyl group to give 10 (0% to 90%). Good activity of substituted benzyl analogs 11 and 12 (3,4-diCl, 50% and 3,4-diF-, 73%, respectively) was maintained, although not as active as the unsubstituted benzyl analog 10 (90%).

Examples of replacing R¹ and R² of (I) with a 4-substituted piperazine ring are illustrated in Table 2. Activity was increased significantly with a 4-F-phenyl-substituted piperazine ring (16, 68%) and a 2,6-dichloro-phenyl group at C-6.

Table 2.

2-Piperazinyl-pyridopyrimidin-7-ones

Entry	R1	R2	R3	Aβ40 (%) Inhibition a	Notch1 Processing b
13	4-F-Ph-	H	t-Bu	2	n.t.
14	4-F-Ph-	H	Cl	0	n.t
15	4-F-Ph-	F	H	20	Inhibition
16	4-F-Ph-	Cl	H	68	No change
17	CH3	Cl	H	0	No change
18	4-F-Ph	H	H	14	Inhibition
19	2-tetrahydrofuranmethyl	Cl	H	0	n.t.
20	2-pyridinyl	Cl	H	0	n.t.
21	2-pyrimindyl	Cl	H	0	n.t.
22	3-pyridinyloxyl	Cl	H	0	n.t.

^a,bSee Table 1 and References and notes. ^{15, 20}; n.t = not tested.

Similar to phenyl pyridopyrimidinone 10, this piperazinyl pyridopyrimidinone 16 also had very good Aβ40 inhibition. Evaluation of selected drug-like properties (e.g., solubility, LogD, plasma protein binding, permeability, and human and rodent microsomal stability) of each of these two compounds was performed. These results indicated that 16 had a slightly better drug-like profile and thus, was further tested in an Aβ42 cellular assay along with other analogs. Pyridopyrimidinone 16 showed inhibition in cells (Aβ42) with an IC₅₀ = 2.7 μM. (See Supplemental Data).

Additional piperazinyl pyridopyrimidinone analogs were synthesized as illustrated in Table 2. An unsubstituted phenyl ring (18, 14%) or other substitutions such as 2,6-diF- (15, 20%), 4-Cl- (14, 0%), and 4-t-butyl- (13, 2%) on the phenyl ring at C-6 diminished activity. Additionally, replacing the 4-F-phenyl ring (R¹) of 16 with a simple methyl group (17) and with various heterocycles (19-22) all yielded 0% inhibition of Aβ40.

The general synthetic routes to obtain 2-aminopyrido[2,3-d]pyrimidin-7-one analog (I) are described by several research teams.^12-14 According to these reported methods, some of our compounds were prepared as shown in Scheme 1. The preparation of these pyridopyrimidinones typically requires an eight-step synthesis beginning with commercially available 4-chloro-2-thiomethyl-5-pyrimidine-carboxylate ethyl ester (II) which is first converted to the corresponding methylamine and then subsequently reduced with lithium aluminum hydride to yield alcohol (III). Oxidation of (III) with manganese oxide gave intermediate 2-thiomethyl-pyrimidine-5-carboxaldehyde (IV). Condensation of aldehyde (IV) with 2,6-dichlorophenyl-acetonitrile provided pyrido[2,3-d]pyrimidin-7-ylideneamine (V). Acylation of (V) gave 7-N-acetylimine (VI) which was readily hydrolyzed to yield methylsulfide intermediate (VII). Subsequent oxidation of (VII) with m-chloro-perbenzoic acid provided sulfonyl pyridopyrimidinone (VIII). Refluxing (VIII) with selected amines in DMF or neat led to the production of the desired 2-amino pyridopyrimidinones (I) in good yields, ranging from 50% (6) to 95% (4).

Scheme 1.

Synthesis of 2-amino substituted 6-(2,6-dichlorophenyl)-pyrido[2,3-d]pyrimidinones

This original eight-step reported synthesis illustrated in Scheme 1 worked well, but the route was inconvenient for further variation on the C-6 position. To simplify the synthesis, we developed a novel method to prepare key intermediate pyridopyrimidinone (VII) in four steps rather than 5 steps. Methylsulfide intermediate (VII) was easily obtained by the condensation of 5-pyrimidinecarboxyaldehyde (IV) with readily available substituted phenylacetates in DMF in the presence of Cs₂CO₃ at room temperature or under refluxing conditions. This improved synthetic method, a total of 6 steps rather than eight steps, was used to prepare compounds 3 and 4 (Table 1) and 13-22 (Table 2).

As a means of reducing the molecular weight of the target pyridopyrimidinone compounds as well as to explore the role of the C-6 aryl group, compound 23 was synthesized from 8-methyl-2-(methylsulfonyl)pyrido[2,3-d]pyrimidin-7(8H)-one (IX, Scheme 2). Pyridopyrimidinone (IX) was prepared from intermediate (IV) according to a reported method.¹⁴

Scheme 2.

Synthesis of Compound 23

This series of pyridopyrimidinones were evaluated for their inhibition of γ-secretase-mediated Aβ40 production by ELISA using purified human γ-secretase complexes and recombinant APP substrate C100Flag.^15-21 Effects of these compounds on γ-secretase-mediated processing of Notch1 were examined by Western Blot analysis using recombinant substrate N100Flag as previously reported.²¹ Compounds with at least a 50% inhibition in the Aβ40 ELISA were chosen for evaluation of Notch1 processing.

Contrary to the previous report,¹¹ initial biological studies of compound PD173955 showed no effect on γ-secretase-mediated production of Aβ40 at the testing concentration (100 μM) in our assay conditions. Nonetheless, many other 2-amino substituted pyridopyrimidin-7-ones still displayed the desired Notch-sparing γ-secretase inhibition profiles. For example, the results shown in Table 1 illustrate reasonable activity of pyridopyrimidinone analogs that have a 4-substituted phenyl ring at C-2 regardless if R¹ = H or a small alkyl group (e.g., 1, 2, 9, 11 and 12, Table 1). However, the most active compound 10 (90% inhibition of Aβ40) has R¹ = isopropyl and an unsubstituted phenyl ring for R². Compound 10 also had no effect on Notch1 processing. Substitution on the phenyl ring of 10 with 3,4-diCl and 3,5-diF gave a slight decrease in activity (11, 50% and 12, 73%, respectively).

The introduction of a cyclic amine such as a 4-substituted piperazine yielded novel inhibitors as well as a possible avenue to improve the drug-like character, e.g., increasing solubility, of these pyridopyrimidinones. The results showed that the R₁ substitution at the N position of the piperazinyl ring significantly influenced the inhibitory effects of analogs X (Table 2) on Aβ production mediated by γ-secretase. The 4-F-phenyl substituent on the piperazine ring yielded the most active of these analogs, 16, with a 68% inhibition of Aβ40 production and no change in Notch1 processing. Adding molecular weight to the already quite large pyridopyrimidine core was not ideal for solubility and brain penetration. However, when this 4-F-phenyl group of 16 was replaced with a 4-methyl group, (17), activity was completely abolished (0%) as it was also for other 4-aryl groups, such as 2-tetrahydrofuran methyl (19), 2-pyridinyl (20), 2-pyrimidinyl (21) and 2-pyridinyloxyl (22). Hence, the 4-F-phenyl group of (16) on the piperazine ring perhaps could behave like a lipophilic handle that is able to reach necessary interactions that influence the inhibition of Aβ40. Of the various substituents studied for R² and R³ in structure (X) (Table 2), 2,6-dichloro-phenyl was preferred and yielded 68% inhibition (16) of Aβ40 production whereas other phenyl substitutions gave slight to no inhibition (15, 2,6-di-F, 20%; 13, 4-t-butyl, 2%; 14 4-Cl, 0%).

Modifications at position 6 of pyridopyrimidinone (X) led to data (Table 2) that indicated the importance of the 6-(2,6-dichlorophenyl) group for activity. The dichloro-substitution on the phenyl ring also appears to be contributing to Notch selectivity when coupled with the 4-fluorophenyl piperazine ring (16 compared to 15 and 18). Analogs with this same 2,6-dichloro-substitution pattern (19-22), but with various heterocycles replacing 4-fluorophenyl on the piperazine ring showed no activity.

Interestingly, since 15 and 18 each have the 4-F-piperazinyl substituent at R¹, but do not have 2,6-dichloro phenyl at C-6, it is speculated that perhaps the substitution pattern on the C-6 aryl ring (R² and R³) could be influencing the Notch selectivity of the pyridopyrimidine analog since both compounds inhibited Notch. The dihedral angle between the phenyl ring and the pyridopyrimidinone template may be important for Notch1-sparing activity. If so, then it appears that the dihedral angle generated with the 2,6-dichlorophenyl group in 16 (i.e., in which a coplanar orientation would be excluded) is preferred for both Aβ inhibition and Notch1-sparing selectivity over that of an unsubstituted phenyl or 2,6-F-phenyl ring (18 and 15 respectively). When the 6-(2,6-dichlorophenyl) group of 16 was completely removed to give 23, inhibition of Aβ production mediated by γ-secretase was completely lost (Table 3). This may indicate that in addition to the 2,6-dichloro-phenyl group of 16 being important for Notch selectivity, a bulky aryl group such as this at C-6 is also necessary for Aβ40 inhibition.

Table 3.

2-(4-Fluorophenyl)piperazinyl pyridopyrimidin-7-ones

Entry	R	Aβ40 (%) Inhibition a	Notch1 Processing b
16	2,6-dichlorophenyl	68	No change
23	H	0	n.t.

^a,bSee Table 1 and References and notes. ^{15, 20}; n.t = not tested.

In summary, Notch-sparing inhibition of γ-secretase can be achieved from molecules bearing the 6-(2,6-dichlorophenyl)-pyrido[2,3-d]pyrimidin-7-one motif. The data obtained in this study indicates the possibility that a dihedral angle between the plane of C-6-aryl and the pyridopyrimidinone template could influence the Notch-sparing property of these types of molecules. The inhibitory activity of these compounds was significantly affected by a substituted piperazinyl moiety at C-2 of the pyrido[2,3-d]pyrimidin-7-one core. These findings could facilitate further design and synthesis of this series of Notch-sparing γ-secretase inhibitors.