Preliminary investigation of 6,7-dihydropyrazolo[1,5-a]pyrazin-4-one derivatives as a novel series of mGlu₅ receptor positive allosteric modulators with efficacy in preclinical models of schizophrenia

Abstract

As part of our efforts to identify a suitable back-up compound to our recently disclosed mGlu₅ positive allosteric modulator (PAM) clinical candidate VU0490551/JNJ-46778212, this letter details the investigation and challenges of a novel series of 6,7-dihydropyrazolo[1,5-a]pyrazin-4-one derivatives. From these efforts, compound 4k emerged as a potent and selective mGlu₅ PAM displaying overall attractive in vitro (pharmacological and ADMET) and PK profiles combined with in vivo efficacy in preclinical models of schizophrenia. However, further advancement of the compound was precluded due to severely limiting CNS-related side-effects confirming the previously reported association between excessive mGlu₅ activation and target-related toxicities.

In the search for improved schizophrenia therapeutics, restoration of dysfunctional N-methyl-d-aspartate (NMDA) receptor-mediated glutamatergic neurotransmission has gained increased attention as a potentially advantageous alternative to the modulation of the dopaminergic system exerted by current antipsychotic medications.^1–3 In this context, positive allosteric modulators (PAMs) of the metabotropic glutamate receptor subtype 5 (mGlu₅) hold high promise by virtue of the close physical and functional relationship between mGlu₅ and NMDA receptors in relevant brain areas.^4–8 Despite the recent association of excessive mGlu₅ receptor activation with target-related liabilities (neuronal necrosis⁹ and seizure induction¹⁰), the observation that in vivo efficacy in several schizophrenia models has been confirmed for multiple mGlu₅ PAMs,^11–16 combined with the identification of potential alternatives (reduced cooperativity with glutamate,^17,18 signal bias¹⁶ or reduced residence-time and duration of target engagement¹⁹) to mitigate these burdens, suggests that further investigations on this novel mechanism are warranted. Resulting from our pioneering industrial-academic collaboration (Janssen Research and Development and the Vanderbilt Center for Neuroscience Drug Discovery (VCNDD))²⁰ we have recently reported on the identification²¹ and preclinical characterization¹⁶ of the mGlu₅ PAM advanced lead 1 (VU0409551/JNJ-46778212). Unfortunately, the progression of 1 was stopped as a consequence of toxicology findings in IND-enabling studies^16,21 and we initiated a research campaign aimed to identify novel mGlu₅ PAMs either with a different pharmacological profile (vide supra) or from a different chemotype that could circumvent these issues. Preliminary efforts towards the identification of a suitable back-up compound within the same chemotype,²² as well as within other exocyclic amide subseries,¹⁹ failed to provide the combination of all of the desired attributes in a single molecule. As an alternative strategy, we turned our attention to other mGlu₅ PAM chemotypes reported by our labs.^23,24 These chemical classes, represented by general structures 2 and 3, share a structurally-related carbonyl-containing central scaffold, in contrast to the exocyclic amide moiety in 1. SAR investigation of these series had already revealed aryls as preferred eastern and western substituents for potent mGlu₅ PAM activity and a prominent role of the central core for the modulation of physicochemical properties. As a continuation of this approach, we considered the 6,7-dihydropyrazolo[1,5-a]pyrazin-4-one scaffold 4 as a potential alternative core; in analogy with series 2 and 3, this scaffold presents a similar geometry, a lactam-type carbonyl and an aromatic southern nitrogen. Our strategy involved a multidimensional optimization campaign focused on a broad range of modifications including the eastern and western aryl moieties, the ether linker, as well as the central core (Fig. 1).

Figure 1.

Structure of the clinical candidate 1 (VU0409551/JNJ-46778212). General structures of endocyclic amide-containing mGlu₅ PAM chemotypes and SAR exploration strategy for 6,7-dihydropyrazolo[1,5-a]pyrazin-4-ones 4.

For this SAR exploration, different synthetic routes were developed. Compounds modified at the eastern and/or western aryl substituent were prepared following two different approaches. In the first one (Scheme 1), the western aryl was introduced at the first step of the synthesis from commercially available ketones 5. Thus, in situ generated sodium enolates of 5 were reacted with diethyl carbonate to afford 2,4-diketone intermediates 6. Further treatment of 6 with hydrazine afforded pyrazoles 7. N-alkylation of 7, via either a Mitsunobu or a nucleophilic substitution reaction, afforded a mixture of regioisomeric pyrazoles which after chromatographic separation provided the major regioisomers 8 pure. Boc protecting group removal in acidic media, followed by exposure to aqueous sodium carbonate, yielded bicyclic lactams 9. Finally, N-arylation of 9 via copper (4a–m, 4w) or palladium (4n) mediated couplings with the corresponding aryl halides furnished desired derivatives.

Scheme 1.

Reagents and conditions: (a) (CO₂Et)₂, Na, EtOH, 0 °C to rt, 16 h, 38% (5a) and 30% (5b); (b) N₂H₄·H₂O, EtOH, 80 °C, 16 h, 98% (7a) and 44% (7b); (c) BocNH (CH₂)₂OH, DBAD, PPh₃, THF, 80 °C, 16 h; (d) BocHN(CH₂)₂Br, K₂CO₃, DMF, rt, 16 h; (e) HCl, 1,4-dioxane, rt, 3 h; (f) Na₂CO₃, 1,4-dioxane/H₂O, rt, 16 h, 84% (9a) and 43% (9b) 3 steps; (g) ArX, CuI, (MeHNCH₂)₂, K₂CO₃, toluene, 120 °C, 16 h, 60–97%; (h) 2-Cl-5-F-pyrimidine, Pd(OAc)₂, XPhos, Cs₂CO₃, 1,4-dioxane, 120 °C, 16 h, 28%.

In the second approach (Scheme 2), symmetric diethyl 1H-pyrazole-3,5-dicarboxylate 10 was alkylated in basic media to afford intermediate 11 which after a deprotection–cyclization sequence analogous to the one described above, provided intermediate 12. Eastern aryls were installed at this point via Goldberg-type coupling, resulting in intermediates 13. Western aryls were introduced after reduction of the ester moiety in 13, via either Mitsunobu reactions with the corresponding phenols or palladium- or copper-catalyzed coupling reactions with the corresponding aryl halides to give derivatives 4o–v.

Scheme 2.

Reagents and conditions: (a) Br(CH₂)₂NHBoc, Cs₂CO₃, DMF, rt, 16 h, 100%; (b) HCl, 1,4-dioxane, rt, 1.5 h; (c) Na₂CO₃, 1,4-dioxane/H₂O, rt, 3 h, 99% 2 steps; (d) ArBr, K₂CO₃, CuI, (MeHNCH₂)₂, toluene, 130 °C, 16 h, 92%; (e) NaBH₄, THF/MeOH, 0 °C to rt, 2 h, 48%; (f) 4o–4r, ArOH, DBAD, PPh₃, THF, 0 °C to rt, 16 h, 4–34%; (g) 4s,v, ArX, Cs₂CO₃, Pd(OAc)₂, BiPhPtBu₂, toluene, 120 °C, 16 h, 22 and 65%; (h) 4t–4u, ArX, CuI, Me₂NCH₂CO₂H, Cs₂CO₃, 1,4-dioxane, 120 °C, 16 h, 11–22%.

Compounds containing a –O–CH₂– linker were prepared following a similar procedure (Scheme 3). 1H-pyrazole-derivative 15 was protected on the hydroxy substituent and N-alkylated to provide regiospecifically (likely due to the steric hindrance exerted by the TBDMS protecting group on the adjacent nitrogen) intermediate 16. OH deprotection and alkylation with benzyl bromide furnished pyrazole 18, which, after Boc-deprotection and base-promoted intramolecular amidation, yielded 6,7-dihydropyrazolo [1,5-a]pyrazin-4-one 19. Eastern aryls were again introduced using copper-catalyzed couplings to afford derivatives 4x,y.

Scheme 3.

Reagents and conditions: (a) TBDMSCl, imidazole, DCM, 0 °C, 1 h, 59%; (b) BocNH(CH₂)₂OH, DBAD, PPh₃, THF, rt, 16 h, 76%; (c) TBAF, THF, rt, 16 h, 83%; (d) BnBr, Cs₂CO₃, DMF, rt, 16 h; (e) HCl, 1,4-dioxane, rt, 3 h; (f) Na₂CO₃, 1,4-dioxane/H₂O, rt, 16 h, 95% for 2 steps; (f) ArX, CuI, (MeHNCH₂)₂, K₂CO₃, toluene, 120 °C, 16 h, 83–99%.

Analogs in which a methyl group was introduced on the –CH₂–O– spacer or carried modifications on the 6,7-dihydropyrazolo[1,5-a]pyrazin-4-one scaffold were prepared following similar procedures to the ones described herein (see Supplementary material for further details).

Gratifyingly, this approach proved successful, affording multiple potent mGlu₅ PAMs within this novel chemotype. Representative eastern and western SAR for this series is summarized in Table 1. Keeping the western unsubstituted phenyl group constant, a fluorine walk around the eastern aryl resulted in potent (<100 nM) and highly efficacious mGlu₅ PAMs with some preference for para substitution (4c) over the ortho and meta congeners (4a,b). The introduction of an additional fluorine atom had no noticeable impact either in potency or in efficacy, as assessed by % maximal glutamate response (Glu Max%) (4d,e). Replacement of fluorine by methyl or methoxy groups resulted in comparable potencies (4f,g vs 4c), yet both analogs displayed slightly reduced efficacy. The replacement of the phenyl by pyridyl led to mGlu₅ PAMs with moderate potency and good efficacy, with the pyridyl-2-yl (4h) the preferred regioisomer with a ~3-fold increased potency over the 3- and 4- isomers (4i,j). The activity was further improved by the introduction of a fluorine atom (4k) or a methyl group (4l) at the 5-position of the pyridine, while the introduction of a methoxy substituent at the same position had negligible effect (4m vs 4h). Finally, the replacement of pyridine by pyrimidine resulted in a ~6-fold decrease in the mGlu₅ activity (4n vs 4k). Selectivity screen of selected examples versus the other mGlu family members (mGlu_1–4,6–8)²⁵ revealed, similarly to imidazopyrimidinones 3²⁴ and a closely related tetranaphthridinone series,¹⁸ potent mGlu₃ antagonist activity for the phenyl-containing congeners (e.g., 4c,f mGlu₃ EC₅₀ = 74 and 150 nM, respectively, full antagonism). Gratifyingly, the replacement by pyridine decreased the mGlu₃ antagonism (4k,l mGlu₃ EC₅₀ = 6760 and >10,000 nM, respectively) resulting in >50 fold functional selectivity for mGlu₅ for 4k and 4l. Furthermore, none of the compounds tested showed relevant activities at any of the other mGlus (EC₅₀ > 10 µM vs mGlu_1,2,4,6,7,8). A focused surveillance of the western aryl group SAR was performed, fixing a 4-fluorophenyl group as the eastern amide substituent (Table 1). Akin to the eastern aryl exploration, the introduction of a fluorine atom yielded potent mGlu₅ PAMs with modest influence of the substitution pattern (4o–q). Replacement of fluorine by methyl (4q vs 4r) resulted in a loss in both potency and efficacy. Finally, the introduction of pyridine had a clear detrimental effect on potency (4s–u) which, in this case, was not compensated by the introduction of a 3-fluoro substituent (4v). In contrast to the amide substituent, the nature of the eastern aryl had no noticeable impact on the mGlu₃ antagonist activity with representative examples (e.g., 4q,s mGlu₃ ant. EC₅₀ = 34 and 290 nM, respectively, full antagonism) showing even more potent mGlu₃ than mGlu₅ activity. Additionally, despite the combination of one of the most potent western substituent (3-fluorophenyl) with the 5-fluoro-2-pyridyl amide in 4w, resulting in a highly potent mGlu₅ PAM, the introduction of an eastern pyridyl moiety did not lead to improved selectivity versus mGlu₃ (EC₅₀ = 140 nM).

Table 1.

Structures and activities of analogs 4

Entry	Ar1	Ar2	pEC50a (±SEM)	EC50 (nM)a	Glu Max%a (±SEM)
4a			6.80 ± 0.11	160	86 ± 5
4b			6.90 ± 0.08	130	81 ± 6
4c			7.19 ± 0.02	65	79 ± 3
4d			7.00 ± 0.03	100	101 ± 6
4e			7.06 ± 0.02	87	81 ± 4
4f			7.15 ± 0.06	71	65 ± 5
4g			6.94 ± 0.03	110	58 ± 8
4h			6.12 ± 0.03	760	69 ± 8
4i			5.68 ± 0.11	2090	64 ± 10
4j			5.68 ± 0.15	2090	72 ± 3
4k			6.90 ± 0.04	130	72 ± 5
4l			6.73 ± 0.03	190	76 ± 4
4m			6.21 ± 0.06	620	74 ± 6
4n			6.12 ± 0.05	760	72 ± 6
4o			6.83 ± 0.02	150	66 ± 8
4p			7.00 ± 0.02	100	70 ± 3
4q			6.98 ± 0.04	100	72 ± 5
4r			6.52 ± 0.07	300	58 ± 11
4s			6.31 ± 0.05	490	68 ± 3
4t			5.40 ± 0.12	3980	69 ± 2
4u			6.16 ± 0.08	690	69 ± 4
4v			5.1 ± 0.17	7940	52 ± 12
4w			7.04 ± 0.03	91	62 ± 9

^aCalcium mobilization assay using HEK293 cells expressing human mGlu₅; values are the average of three or more independent determinations.

Other relevant SAR for this chemotype concerning modifications in the spacer and the central core is summarized in Figure 2. Replacement of the –CH₂–O– linkage by the regioisomeric –O–CH₂– resulted in a minor decrease in mGlu₅ potency (~2.5 fold) and efficacy for the match-pair 4x–4c but in an improved selectivity over mGlu₃ (mGlu₃ ant. EC₅₀ = 3390 nM). By contrast, the combination of this linker with the 4-F-2-pyridyl amide substituent resulted in an unexpected complete loss of activity (4y vs 4k). Likewise, the introduction of a methyl group on the –CH₂–O– linker (4z) significantly diminished the mGlu₅ potency by >100 fold. Finally, a small set of central core modifications were also surveyed. While increasing the ring size from piperazinone to homopiperazinone (20) resulted in a slight loss of activity at mGlu₅ (~3 fold) and a modestly improved selectivity (mGlu₃ EC₅₀ = 540 nM), modifications in the pyrazole unit proved to be more deleterious. Thus, either replacement of the pyrazole by a triazole (21a), or the introduction of methyl or chlorine substituents at the 3 position of the 6,7-dihydropyrazolo-[1,5-a]pyrazin-4-one core (21b,c) resulted in a remarkable loss of mGlu₅ potency.

Figure 2.

Structures and mGlu₅ activity of 6,7-dihydropyrazolo[1,5-a]pyrazin-4-one derivatives modified at the spacer and central core.

Since dual mGlu₅/mGlu₃ activities may mask or confound mGlu₅ associated pharmacology and/or toxicology, our selection criteria focused on compounds with potent mGlu₅ PAM activity combined with significant selectivity over mGlu₃ antagonist activity. Therefore, based on these selection principles, 4k and 4x stood out as candidates worthy of further characterization in support of eventual assessment in our primary rodent (rat) in vivo pharmacodynamic paradigm, the amphetamine-induced hyperlocomotion (AHL) model.¹⁷ Comparative data for both compounds, included in Table 2, show that both compounds presented a favorable cLogP for a CNS drug,²⁶ >20 fold selectivity versus mGlu₃ antagonist activity and no relevant activities at any of the other mGlu receptors (EC₅₀s > 10 µM). Furthermore, a moderate in vitro predicted rat hepatic clearance (rCL_hep < 35 mL min⁻¹ kg⁻¹) and a good fraction unbound in rat brain warranted the in vivo evaluation of both compounds. Thus 4k and 4x were evaluated in the AHL challenge model using an oral screening dose of 10 mg/kg (20% β-cyclodextrin (HP-β-CD), homogeneous suspension).¹⁷ Interestingly, while both compounds had terminal unbound brain concentrations (n = 6 animals, T₉₀) above their corresponding in vitro mGlu₅ EC₅₀s, a robust reversal of AHL could only be demonstrated for 4k (50%). This striking difference could be attributable to the lower PAM efficacy at mGlu₅ receptors for 4x (54% vs 72% for 4k) which may render this compound unable to effectively reverse the effects of amphetamine at the tested dose. Similar results have also been reported for other low efficacy mGlu₅ PAMs.^13,14

Table 2.

Relevant comparative profiling data for analogs 4k and 4x

Entry	c Log P	mGlu3 ant EC50a (nM)	mGlu3/mGlu5 EC50 ratio	mGlu1,2,4,6,7,8 EC50b (lM)	Rat CLhepc (mL/min/ kg)	Rat fu brain (%)	AHLd (% reversal)	[Brain]u,90 mine (nM)
4k	1.37	6760	52	>10	34.5	5.5	50	313
4x	2.90	3390	21	>10	34.6	6.5	12	376

^aCalcium mobilization assay using HEK293 cells expressing human mGlu₃; values are the average of three or more independent determinations.

^bCalcium mobilization assay using HEK293 cells expressing human expressing human mGlu₁, mGlu₂, mGlu₇, or mGlu₈ receptors and [³⁵S]-GTPγS functional assay using CHO cells expressing rat mGlu₆ receptors; values are the average of three or more independent determinations.

^cPredicted hepatic clearance based on microsomal CL_int data.

^dAmphetamine-induced locomotion% reversal after a 10 mg/kg p.o. dose, vehicle 20% HP-β-CD.

^eBased upon rat brain homogenate binding (equilibrium dialysis) fraction unbound and total terminal (90 min) brain levels.

Thus, based on its promising preliminary profile as well as its good activity in our pharmacodynamic screen, 4k was advanced to additional pharmacological and ADMET profiling, discrete PK (rat and dog), full dose–response studies in AHL to determine its minimal effective dose (MED), and evaluation in a preclinical cognition model, the contextual fear conditioning (CFC) paradigm.¹⁶ Further pharmacological profiling of 4k in glutamate concentration response curve (CRC) fold-shift (FS) experiments in human mGlu₅ expressing cells confirmed cooperativity at mGlu₅ receptors (7.7) (FS = 11 for 1).¹⁶ Additionally, 4k was assessed in a broad ancillary screening panel of distinct GPCRs, ion channels and transporters²³ in which only weak interaction (57% inhibition of binding at 10 µM) with the human platelet activating factor (PAF) was found. Moreover, other relevant in vitro ADMET properties were also favorable, including moderate predicted human hepatic clearance (hCL_hep = 8.4 mL min⁻¹ kg⁻¹), clean P450 inhibition profile (1A2 2C9, 2D6 and 3A4 IC₅₀s ≥ 30 µM), and low plasma protein binding (f_u (h,r): 8.2, 6.8). Additionally, and despite low solubility (FaSSIF = 13 µg/mL), 4k was found to be well absorbed either from a 20% HP-β-CD solution or suspension (vide infra). In vitro preliminary toxicity assessment also confirmed a benign profile for 4k in AMES, high content cytotoxicity assay (>100 µM), and in GSH/CN trapping studies, with no significant interactions on cardiovascular relevant ion channels (hERG, Ca, Na channels, IC₅₀s > 10 µM). The in vivo PK profile for 4k was assessed in rat and dog (Table 3), and moderate clearance and volume of distribution for both species were noted, with resulting moderate half-lives. A longer time to reach peak levels (T_max) in the rat likely reflects solubility-limiting absorption for this species. Oral bioavailability was good for both species (>45%) with an excellent 88% in the rat.

Table 3.

Pharmacokinetic parameters for 4k

Parameters	Rata	Doga
Dose (mg kg−1) iv/p.o.	2.5b/10c	0.5b/5c
CLp (mL/min/kg) iv	19	25
Vdss (L/kg) iv	1.6	2.6
t1/2 (h) iv	1.6	2.4
Cmax (ng/mL) p.o.	730	519
Tmax (h) p.o.	4	0.7
AUC0–inf (ng h/mL) p.o.	6668	1536
F (%) p.o.	88	48

^aData reported as average of three animals.

^bDosed as 20% HP-β-CD solution.

^cDosed as 20% HP-β-CD suspension.

In full dose–response studies, 4k produced a dose- and concentration-dependent reversal of amphetamine-induced hyperlocomotion upon oral administration with a MED of 3 mg/kg (Fig. 3) and full reversal at 100 mg/kg. Terminal (1.5 h) plasma and brain concentrations in this study were linear for doses from 3 to 10 mg/kg and sub-linear from 30 to 100 mg/kg). Furthermore, in addition to its antipsychotic-like activity, 4k also induced a dose-dependent enhancement of CFC acquisition in rats (MED = 0.3 mg/kg p.o.) demonstrating efficacy in an established measure of hippocampal-dependent cognitive function (Supplementary material) which may be suggestive of efficacy across multiple domains of schizophrenia. In virtue of its attractive in vitro and in vivo properties, 4k was considered for advancement as a potential back-up candidate for 1. Unfortunately, during a toxicology rat MTD study (25, 75, 150, 300 and 600 mg/kg, p.o.) evidence for signs for CNS-related side-effects (compulsive behavior (chewing and head shaking), pedaling movements of the forepaws and tremors) starting at the lowest tested dose, combined with the death of 2/6 animals at the highest dose, precluded further development and confirmed again the subtle relationship between mGu₅ activation and target related toxicities. As a consequence, and in order to identify suitable back-up candidates, our efforts have now shifted toward identifying compounds with an even further reduced efficacy at mGlu₅.

Figure 3.

4k has antipsychotic-like activity in male, Sprague Dawley rats. 4k dose-dependently (3–100 mg/kg, p.o.) reverses AHL. Vehicle is 20% HP-β-CD (3–10 mg/kg solution, 30–100 mg/kg suspension).

In conclusion, as part of our efforts towards a back-up compound for the mGlu₅ PAM clinical candidate 1 (VU0409551/JNJ-46778212) from a different chemotype, a scaffold hopping strategy resulted in the identification of a series of 6,7-dihydropyrazolo[1,5-a]pyrazin-4-ones as potent mGlu₅ PAMs. SAR investigations within this chemotype resulted in the identification of compound 4k as a potential back-up candidate. Despite 4k possessing attractive in vitro and in vivo pharmacological, ADMET and PK properties, rat toxicology MTD studies deemed the compound inadequate for further progression as a consequence of side-effects likely due to excessive mGlu₅ activation. Efforts aimed to identify less efficacious mGlu₅ PAMs, within this and other chemotypes will be reported in due course.