Scaffold-hopping identifies furano[2,3- d ]pyrimidine amides as potent Notum inhibitors

The carboxylesterase Notum is a key negative regulator of the Wnt signaling pathway by mediating the depalmitoleoylation of Wnt proteins. Our objective was to discover potent small molecule inhibitors of Notum suitable for exploring the regulation of Wnt signaling in the central nervous system. Scaffold-hopping from thienopyrimidine acids 1 and 2 , supported by X-ray structure determination, identified 3-methylimidazolin-4-one amides 20-24 as potent inhibitors of Notum with activity across three orthogonal assay formats (biochemical, extra-cellular, occupancy). A preferred example 24 demonstrated good stability in mouse microsomes and plasma, and cell permeability in the MDCK-MDR1 assay albeit with modest P-gp mediated efflux. Pharmacokinetic studies with 24 were performed in vivo in mouse with single oral administration of 24 showing good plasma exposure and reasonable CNS penetration. We propose that 24 is a new chemical tool suitable for cellular studies to explore the fundamental biology of Notum. performed in vivo in mouse with single oral administration of 24 showing good plasma exposure and reasonable CNS penetration. We propose that 24 23 is a new chemical tool suitable for cellular studies to explore the fundamental biology of Notum. Amide 24 has complementary properties to CNS restricted acid 1 and irreversible inhibitor ABC99 , and so represents a valuable addition to the Notum inhibitor chemical toolbox. hydrogen donor; MLM, mouse liver microsomes; MPO, multiparameter optimization; OPTS, trisodium 8-octanoyloxypyrene-1,3,6-trisulfonate; P-gp, P-glycoprotein; SAR, structure SBDD, structure TPSA, topological polar surface area; UPLC-MS, ultra performance liquid chromatography spectrometer. HBTU, O -(1 H -Benzotriazol-1-yl)- N,N,N',N' -tetra-methyluronium hexafluorophosphate.


Abstract
The carboxylesterase Notum is a key negative regulator of the Wnt signaling pathway by mediating the depalmitoleoylation of Wnt proteins. Our objective was to discover potent small molecule inhibitors of Notum suitable for exploring the regulation of Wnt signaling in the central nervous system. Scaffold-hopping from thienopyrimidine acids 1 and 2, supported by X-ray structure determination, identified 3-methylimidazolin-4one amides 20-24 as potent inhibitors of Notum with activity across three orthogonal assay formats (biochemical, extra-cellular, occupancy). A preferred example 24 demonstrated good stability in mouse microsomes and plasma, and cell permeability in the MDCK-MDR1 assay albeit with modest P-gp mediated efflux. Pharmacokinetic studies with 24 were performed in vivo in mouse with single oral administration of 24 showing good plasma exposure and reasonable CNS penetration. We propose that 24 is a new chemical tool suitable for cellular studies to explore the fundamental biology of Notum.
The Wnt signaling pathway regulates several aspects of brain development and function, and dysregulation of Wnt signaling has been implicated to play a role in neurodegenerative diseases such as Alzheimer's disease (AD). 1 Cognitive impairments, characteristic of AD, correlate closely with the loss of synapses and evidence suggests that excess amyloid-ß (A) causes synapse dysfunction by impairing synapse maintenance, at least in part, through causing dysfunction of Wnt signaling. 2,3 Compromised Wnt signaling may also be associated with AD through loss of blood-brain barrier (BBB) integrity 4 and A generation through -secretase (BACE1) expression. 5 Signal transduction by Wnt proteins is tightly regulated by a range of mechanisms including post translational modifications. For example, O-palmitoleoylation of Wnt proteins is required for efficient binding to Frizzled (Fzd) receptors and the subsequent signal transduction. 6 The carboxylesterase Notum is a key negative regulator of the Wnt signaling pathway by specifically mediating the O-depalmitoleoylation of Wnt proteins. 7,8 The role of Notum in the mammalian central nervous system (CNS) has yet to be established although Notum is expressed and upregulated in endothelial cells in the hippocampus of APPPS1 mice and AD patients compared to control. 9 In a disease setting, it follows that inhibition of Notum could restore Wnt signaling with potential benefit in disease where Wnt deficiency is an underlying cause.
The search for Notum inhibitors has identified acids 1 and 2 which have shown utility in mouse models of bone growth and found to increase cortical bone thickness. 10,11 Although 1 demonstrates good oral bioavailability, recent pharmacokinetic studies in mouse showed CNS penetration of 1 is very low with brain:plasma concentration ratio of just 0.01. 12 Additional compounds include irreversible inhibitor ABC99 used to show the role of Notum in the regeneration of aged intestinal epithelium, 13,14 and phenoxyacetamide 3 identified through optimisation of an X-ray fragment screening hit. 15 However, it is unlikely that these compounds will be suitable for in vivo studies where CNS penetration is an essential requirement. Hence, our objective was to discover potent small molecule inhibitors of Notum suitable for exploring the regulation of Wnt signaling in the CNS.
In order to identify new small molecule inhibitors of Notum, we elected to explore if 1 and 2 could be modified to deliver a CNS penetrant tool by capping off the acid as an amide. However, prior art had established that similar carboxamides exhibited poor metabolic stability. 10 Our initial investigations into amide derivatives of 1 somewhat confirmed this result but also showed that judicious choice of the amine partner could significantly improve metabolic stability as measured in liver microsomes. 12 At the outset, we wished to use structure based drug design (SBDD) to accelerate our progress towards the discovery of potent inhibitors by effective binding with Notum. Crystals of C-terminal his-tagged Notum(Ser81-Thr451 Cys330Ser) were soaked with acids 1 and 2, and the crystal structures solved to elucidate their inhibitor binding modes (Figure 1). Notum has a well-defined, large (ca. 380 Å 3 ), hydrophobic active-site pocket adjacent to the catalytic triad (Ser232, His389, Asp340) that accommodates the palmitoleate group of Wnt (PDB:4UZQ). 7 Both 1 and 2 place the thienopyrimidine group into this pocket with the acid forming the only polar interactions through a network on H-bonds to the backbone with Trp128, Gly127 and Ala233, and also a H-bond to the sidechain of His389 ( Figure S1). The position of the thiophene ring differs slightly between 1 and 2 to accommodate the substituents which sit on opposite sides of the inhibitor, but the remainder of the molecules adopted a similar position in the pocket. Overlays of the structures of 1 and 2 with O-palmitoleoyl serine show all three structures effectively fill this pocket ( Figure S2). From a design perspective, these structures show significant solvent exposed space at the mouth of the palmitoleate pocket to accommodate a suitable group as an amide derivative of 1 and 2. The SARs were initially directed at exploring two principle areas of the structure: (1) the amide group (4,5) (Tables 1, 2 and S1); and (2) the pyrimidine heterocyclic group that binds in the palmitoleate pocket (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19) ( Table 3). Combinations of preferred amides and heterocycles were then prepared (20-24) ( Table 4).

Inhibition of Notum carboxylesterase activity of 4-24 (Tables 1-4) was routinely measured in a biochemical
assay where test compounds were incubated with Notum(81-451 Cys330Ser) and trisodium 8octanoyloxypyrene-1,3,6-trisulfonate (OPTS) as the substrate for 1 h and fluorescence recorded. 15 Compounds were then assessed for metabolic stability in mouse liver microsomes (MLM) and for cell permeability by measuring transit performance across a MDCK-MDR1 monolayer. Selected compounds were screened for inhibition of Notum activity in a Wnt/-catenin signaling pathway TCF/LEF Reporter (Luciferase) HEK293 cell line and Notum occupancy in a FP-biotin competition assay. 15,18 Initial SAR studies with amides 4 and 5 (derived from 1 and 2 resp.) suggested that the Notum activity was largely driven by the heterocycle binding in the palmitoleate pocket with the amide moiety offering minimal contribution, although poor choice of amine partner could disrupt the binding; this is consistent with X-ray structures and docking studies ( Table 1). 19,20 From this set of matched pairs, three amide series 4d/5d, 4e/5e and 4f/5f emerged as having potent Notum inhibition (IC50 < 10 nM), moderate MLM stability and cell permeability although they were all substrates for P-gp mediated efflux to some degree. The challenge then became to retain Notum inhibition activity, further improve metabolic stability whilst developing cell permeability without efflux. A wider range of amides around 5d-f were then prepared in the thieno [2,3-d]pyrimidine series 5 as this template offered the advantage of slightly lower lipophilicity when compared to 4 (5 vs 4, cLogP = -0.5).
(Tables 2 and S1). One approach to reduce P-gp mediated efflux is to remove HBD or, if the HBD is essential for binding to the primary target, to partially mask the HBD group by placing a flanking group in close proximity.
N-Alkylation of the piperazin-2-one 5d with either a Me (5g) or Et (5h) group retained potency although impact on efflux was contradictory. C-Alkylation of 5d at the -position with one or two Me groups reduced potency (5i, 5j) and combining these two modifications into ring gave 5k which was inferior to 5d in all aspects. N-Methylation of imidazolidin-4-one 5e proved to more beneficial with 5l showing potent activity, improved MLM stability and high cell permeability with minimal efflux (ER 1.2). N-Substitution with larger alkyl groups such as Et (5m), cPr (5n) and CH2CF3 (5o) retained potent Notum inhibition but eroded MLM stability, and the instability tracked with increased compound lipophilicity.
Modifications to the triazolo [4,3-a]pyrazine amide 5f proved to be detrimental. Alkylation of the available C3 position of the triazole ring with small lipophilic groups (5p-s: Me, Et, cPr, CF3) proved to be progressively detrimental to activity. Switching to the triazole isomer triazolo [1,5-a]pyrazine amides 5t-v retained potency and substitution was tolerated at C2 but offered little advantage. These triazolopyrazine amides 5p-v were at least 3-fold weaker than 5f and failed to improve MLM or the efflux ratio.
At this point, mouse pharmacokinetic data for 5l was generated in vivo to determine the extent of plasma exposure and to check the correlation of the in vitro ADME data with in vivo outcomes. Imidazolidin-4-one 5l (cLogP 2.6; LogD7.4 1.8) was selected as a representative example from this set as it combined good aqueous solubility (77 g/mL; 215 M) and cell permeability with moderate microsomal stability. Following single oral dose (p.o.) of 10 mg/kg, plasma exposure for 5l was low (Cmax 120 ng/mL; AUC(0→inf) 70 ng.h/mL) which highlighted the need to further improve metabolic stability. The next phase of SAR was to explore the pyrimidine heterocyclic group that binds in the palmitoleate pocket (Table 3). This series of SARs was performed with the carboxylic acids with a view to introducing the amide group once preferred heterocycles had been identified. Thienopyrimidines 1 and 2 are potent inhibitors and so a range of alternative substituents on the thiophene ring (6)(7)(8)(9)(10)(11)(12) were investigated. Substituents were selected to optimise binding interactions with Notum and reduce overall lipophilicity through removal of lipophilic groups and/or introduction of polar groups. Deletion of the 5-Cl from 2 to give 6 (5-H) resulted in a significant drop in activity whereas direct replacement of 5-Cl with a 5-Me 7 retained activity. Further modification of the 7 scaffold by the addition of a 2-Me group (8) led to a dramatic decrease in potency and so substitution at C2 was not investigated further. Introduction of a CF3 group at either the 5-or 6-positions (9, 10 resp.) proved to be beneficial whereas the application of a 6-CN as a non-traditional bioisostere for a halogen (11,12) was detrimental to activity on this occasion. 21 Alternative fused 6,5-ring systems (13)(14)(15)(16)(17)(18)(19) were also explored with the objective of replacing the thiophene with a more polar N or O containing heterocycle. Pyrrolopyrimidine 13 was a weak inhibitor although activity was improved when combined with substituents at either the 5-or 7-positions (14,15). A 7-Bn group (16) could be accommodated but there was no significant improvement over 7-H (13), and this was at a significant penalty in added lipophilicity. Pyrazolopyrimidine 17 proved to be the most active inhibitor from these N heterocycles although still 1000-fold weaker than 1. In contrast, the furano [2,3-d]pyrimidines proved to be more successful when combined with optimal substituents. 5,6-Dimethyl furan 18 was 10-fold weaker than the corresponding thiophene analogue (18 vs 7) but replacement of the 6-Me of 18 with a 6-CF3 gave 19 which restored potent Notum inhibition activity in this more polar template (19 vs 9; cLogP = -0.6). Table 3: Notum inhibition of thieno- (6)(7)(8)(9)(10)(11)(12), pyrrolo-(13-16), pyrazolo-(17) and furanopyrimidine acids (18)(19).

6-19
Compound Notum  The strategy of combining the superior acid heterocycles (1,7,9,10,19) with the preferred 3methylimidazolin-4-one amine produced amides 20-24 all with potent Notum inhibition (IC50 < 10 nM) in the biochemical OPTS assay ( Table 4). In general, the Notum inhibition activity of these amides tracked closely to the activity of their corresponding acid, and with the same rank order, again suggesting the amide moiety offered minimal contribution (or disruption) to the binding with Notum.

IC50 (nM) a
These inhibitors 20-24 were screened in the cell-based TCF/LEF reporter gene (Luciferase) assay to assess their ability to restore Wnt/-catenin signaling when activated by exogenous rWnt3a (100 ng/mL) in the presence of Notum (500 ng/mL) ( Evaluation of 20-24 in MLM showed 20-23 to have moderate metabolic stability and offered no significant advantage over 5l. Only 24 demonstrated high metabolic stability in MLM with the potential advantage of low metabolic clearance in vivo. Furthermore, 24 was stable in mouse plasma with no degradation observed after 120 min and did not inhibit CYP450 enzymes ( Table 5). Compound 24 displayed a modest efflux ratio (ER = 2.4) in the MDCK-MDR1 permeability assay which suggests some cognition by P-gp mediated efflux transport.
However, the ER for 24 was perceived to be within acceptable limits based on established precedent. 22 Representative inhibitors 5l and 24 were tested in a Notum occupancy assay using FP-biotin, 18 a serine hydrolase activity-based probe, whereby labelling of Ser232 of Notum with FP-biotin can be blocked by an inhibitor occupying the active site of Notum (Figure 2). Both inhibitors showed an ability to prevent labelling by FP-biotin, confirming they competitively bind to Notum, with potency equivalent to 1.

Figure 2:
Notum activity-based occupancy assay was performed with FP-biotin (FP)(2 M) and test compounds 1, 3, 5l and 24 (10 M) for 10 minutes in conditioned media from HEK293S cells stably transfected with a Notum lentiviral construct. Relative occupancy was calculated by optical density of fluorescent band detecting the level of biotinylation of Notum using Image Studio Lite 5.2, compared to the control-treated sample which was set to 1. N = 3 with s.d.
Hence, on balance, 24 emerged has having a superior profile from this set and was selected for further evaluation in mouse pharmacokinetic studies.  Pharmacokinetic data for 24 was generated in vivo in mouse to evaluate plasma exposure and CNS penetration ( demonstrates reasonable CNS penetration with a brain:plasma concentration ratio of 0.29 based on AUC(0→inf).
The incomplete CNS penetration was probably due to some element of P-gp efflux transport recognition as evidenced by the ER in the MDCK-MDR1 cell line. The combination of incomplete CNS penetration along with preferential binding to brain tissue resulted in (moderate) brain exposure of Cmax [brain] ≈ X.X M (free drug) but this still (exceeded) the Notum EC50 by X-fold at this dose (TBC).
Hence, 24 has potential utility in mouse models of disease under carefully designed experimental protocols where the required site of action, route of administration, dose and duration of action requirements are understood; i.e. the pharmacokinatic-pharmacodynamic relationship is to be established.
In summary, scaffold-hopping from thienopyrimidine acids 1 and 2, supported by X-ray structure determination, identified 3-methylimidazolin-4-one amides 20-24 as potent inhibitors of Notum with activity across three orthogonal assay formats. A preferred example 24 demonstrated good stability in MLM and mouse plasma, and cell permeability in the MDCK-MDR1 assay albeit with modest P-gp mediated efflux. PK studies with 24 were performed in vivo in mouse with single oral administration of 24 showing good plasma exposure and reasonable CNS penetration. We propose that 24 23 is a new chemical tool suitable for cellular studies to explore the fundamental biology of Notum. Amide 24 has complementary properties to CNS restricted acid 1 and irreversible inhibitor ABC99, and so represents a valuable addition to the Notum inhibitor chemical toolbox.  Med. Chem. 2012, 55, 4877-4895. 23. 3-Methyl-1-(2-((5-methyl-6-(trifluoromethyl)furo [2,3-d] In brief, scaffold-hopping from thienopyrimidine acids 1 and 2, supported by X-ray co-structure determination, identified 3-methylimidazolin-4-one amides 20-24 as potent inhibitors of Notum with activity across three orthogonal assay formats. A preferred example 24 demonstrated good stability in MLM and mouse plasma, and cell permeability in the MDCK-MDR1 assay albeit with modest P-gp mediated efflux. PK studies with 24 were performed in vivo in mouse with single oral administration of 24 showing good plasma exposure and reasonable CNS penetration. We propose that 24 is a new chemical tool suitable for cellular studies to explore the fundamental biology of Notum. Amide 24 has complementary properties to CNS restricted acid 1 and irreversible inhibitor ABC99, and so represents a valuable addition to the Notum inhibitor chemical toolbox.
The Letter describes two new Notum-inhibitor x-ray structures with LP-922056 (1) and 2. These structures will be deposited in the PDB when the Letter has been provisionally accepted for publication, as is common accepted practice.
A related Article from this group published on the preprint server Beilstein Arch. discloses an improved synthesis of LP-922056 (1) and includes our preliminary investigations into amide derivatives of 1; these are presented as a  Page S2 Figure S1. Interaction maps of 1 and 2 in the Notum binding pocket.
Page S3 Figure S2. Overlays of compounds 1, 2 and O-palmitoleoyl serine in the Notum binding pocket.

Animal Testing Declaration:
In vivo mouse pharmacokinetic data was generated at GVK Biosciences, Hyderabad, India. GVK Biosciences is accredited by AAALAC and OLAW for the ethical treatment of animals. All the animal experiments are conducted in accordance with IAEC approved protocols. See: https://www.gvkbio.com/biology-services/invivo-pharmacology/ Additional thieno [2,3-d]pyrimidines amides 5w-5uu were prepared by the general methods as described and data presented here for a more complete analysis of the SARs.