Identification of Dihydropyrazolo[1,5-a]pyrazin-4(5H)-ones as Cyclic Products of β-Amidomethyl Vinyl Sulfone Alphavirus Cysteine Protease Inhibitors

Optimized syntheses of (E)-5-(2-ethoxyphenyl)-N-(3-(methylsulfonyl)allyl)-1H-pyrazole-3-carboxamide (RA-0002034, 1), a promising antiviral covalent cysteine protease inhibitor lead, were developed. The syntheses avoid the contamination of 1 with the inactive cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2, which is formed by the intramolecular aza-Michael reaction of the vinyl sulfone warhead under basic conditions and slowly at pH 7.4 in phosphate buffer. The pure cysteine protease inhibitor 1 could be synthesized using either modified amide coupling conditions or through the introduction of a MOM-protecting group and was stable as a TFA or HCl salt. Although acyclic 1 demonstrated poor pharmacokinetics with high in vivo clearance in mice, inactive cyclic 2 showed improved plasma exposure. The potential use of cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-ones as prodrugs for the acyclic β-amidomethyl vinyl sulfone warhead was demonstrated by GSH capture experiments with an analog of 2.


Introduction
Alphaviruses, a group of widespread, enveloped, single-stranded positive sense RNA viruses, are transmitted by Aedes aegypti and Aedes albopictus mosquitoes, posing a significant threat to public health [1].These viruses are divided into two categories based on their geographical emergence: Old World alphaviruses, including Chikungunya virus (CHIKV), Ross River virus (RRV), and O'nyong-nyong virus (ONNV) that typically present with rash, fever, and prolonged arthralgia that can persist for months post-infection [2]; and New World alphaviruses, including Venezuelan (VEEV), Western (WEEV), and Eastern (EEEV) Equine Encephalitis viruses and Mayaro virus (MAYV), that often result in encephalitis-like neurological symptoms, accompanied by fever, headache, and nausea, which can be fatal, with 30-50% of EEEV cases resulting in mortality [3].Despite the severity of these diseases, there are currently no FDA-approved drugs for any alphavirus-caused disease, highlighting the urgent need for the development of alphavirus therapeutics.
The largest non-structural protein in the alphavirus genome, nsP2, is essential for viral replication [4].nsP2 contains a C-terminal cysteine protease that uses a catalytic dyad of cysteine and histidine residues to catalyze substrate cleavage.Two covalent inhibitors of alphavirus nsP2 protease that contain a common β-amidomethyl vinyl sulfone warhead have been recently disclosed (Figure 1).Compound #11 was reported as a micromolar inhibitor of VEEV nsP2 protease activity with antiviral activity [5].Likewise, we reported the discovery of RA-0002034 (1) as a potent covalent inhibitor of the CHIKV nsP2 protease with IC 50 = 60 nM [6].Vinyl sulfone 1 inhibited VEEV and CHIKV replication with have been recently disclosed (Figure 1).Compound #11 was reported as a micromolar inhibitor of VEEV nsP2 protease activity with antiviral activity [5].Likewise, we reported the discovery of RA-0002034 (1) as a potent covalent inhibitor of the CHIKV nsP2 protease with IC50 = 60 nM [6].Vinyl sulfone 1 inhibited VEEV and CHIKV replication with EC50 = 0.3 and 0.01 µM, respectively, and decreased viral titer across a wide range of New and Old World alphaviruses [6].Notably, the 5-arylpyrazole in 1 conferred an increase in potency for nsP2 protease inhibition compared to the 1,2-dihydroquinoline in Compound #11, demonstrating the importance of the heterocyclic amide substituent in molecular recognition by the viral protease.Vinyl sulfones have broad utility as covalent inhibitors of cysteine proteases beyond viral nsP2 [7].However, the cysteine reactivity of these warheads must be balanced with concerns of toxicity due to off-target activity or poor pharmacokinetics due to systemic GSH reactivity [8].During the resynthesis of 1, we observed the formation of a cyclic byproduct 2 that effectively masked the vinyl sulfone warhead, rendering it inactive as an nsP2 protease inhibitor.In this report, we document methods to synthesize pyrazole-substituted β-amidomethyl vinyl sulfones, such as 1, devoid of contamination from cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-ones.We also explored the reversibility of the cyclization reaction as a potential prodrug strategy for the cysteine-reactive acyclic β-amidomethyl vinyl sulfone.
have been recently disclosed (Figure 1).Compound #11 was reported as a micromolar inhibitor of VEEV nsP2 protease activity with antiviral activity [5].Likewise, we reported the discovery of RA-0002034 (1) as a potent covalent inhibitor of the CHIKV nsP2 protease with IC50 = 60 nM [6].Vinyl sulfone 1 inhibited VEEV and CHIKV replication with EC50 = 0.3 and 0.01 µM, respectively, and decreased viral titer across a wide range of New and Old World alphaviruses [6].Notably, the 5-arylpyrazole in 1 conferred an increase in potency for nsP2 protease inhibition compared to the 1,2-dihydroquinoline in Compound #11, demonstrating the importance of the heterocyclic amide substituent in molecular recognition by the viral protease.Vinyl sulfones have broad utility as covalent inhibitors of cysteine proteases beyond viral nsP2 [7].However, the cysteine reactivity of these warheads must be balanced with concerns of toxicity due to off-target activity or poor pharmacokinetics due to systemic GSH reactivity [8].During the resynthesis of 1, we observed the formation of a cyclic byproduct 2 that effectively masked the vinyl sulfone warhead, rendering it inactive as an nsP2 protease inhibitor.In this report, we document methods to synthesize pyrazole-substituted β-amidomethyl vinyl sulfones, such as 1, devoid of contamination from cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-ones.We also explored the reversibility of the cyclization reaction as a potential prodrug strategy for the cysteine-reactive acyclic β-amidomethyl vinyl sulfone.
Formation of 2 is proposed to occur under the basic conditions of the amide coupling by a formal aza-Michael conjugate addition [9,10] of the pyrazole N2 into the β-carbon of the vinyl sulfone (Scheme 2).The resulting dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2, which masks the reactive vinyl sulfone warhead within its cyclic structure, was stable under normal laboratory conditions with no propensity to revert to the acyclic 1 upon storage as a solid or as a 10 mM DMSO stock solution.Unsurprisingly, 2 was inactive as an nsP2 protease inhibitor at concentrations up to 200 µM (Table S1).Formation of 2 is proposed to occur under the basic conditions of the amide coupling by a formal aza-Michael conjugate addition [9,10] of the pyrazole N2 into the β-carbon of the vinyl sulfone (Scheme 2).The resulting dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2, which masks the reactive vinyl sulfone warhead within its cyclic structure, was stable under normal laboratory conditions with no propensity to revert to the acyclic 1 upon storage as a solid or as a 10 mM DMSO stock solution.Unsurprisingly, 2 was inactive as an nsP2 protease inhibitor at concentrations up to 200 µM (Table S1).Formation of 2 is proposed to occur under the basic conditions of the amide coupling by a formal aza-Michael conjugate addition [9,10] of the pyrazole N2 into the β-carbon of the vinyl sulfone (Scheme 2).The resulting dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2, which masks the reactive vinyl sulfone warhead within its cyclic structure, was stable under normal laboratory conditions with no propensity to revert to the acyclic 1 upon storage as a solid or as a 10 mM DMSO stock solution.Unsurprisingly, 2 was inactive as an nsP2 protease inhibitor at concentrations up to 200 µM (Table S1).

Stability of 1 in Neutral Phosphate Buffer
To assess if cyclization of 1 into 2 could occur during the assessment of cellular antiviral activity, the stability of 1 in neutral pH phosphate buffer was investigated.To achieve this, a stock solution of 1 was prepared in DMSO-d 6 and diluted into pH 7.4 phosphate buffer (with 10% D 2 O) to a final concentration of 2 mM in the presence of maleic acid as an internal standard.The solution was analyzed by 1 H NMR spectroscopy over 48 h at room temperature (Figure 4).At t = 1 h, vinyl sulfone 1 was present in solution at the expected 2 mM concentration.After 24 h, a reduction in the abundance of 1 by 10% and the appearance of approximately 9% of its cyclic isomer 2 were observed.After 48 h, 81% of 1 remained, with approximately 19% of 2 present in the phosphate buffer.

Stability of 1 in Neutral Phosphate Buffer
To assess if cyclization of 1 into 2 could occur during the assessment of cellular antiviral activity, the stability of 1 in neutral pH phosphate buffer was investigated.To achieve this, a stock solution of 1 was prepared in DMSO-d6 and diluted into pH 7.4 phosphate buffer (with 10% D2O) to a final concentration of 2 mM in the presence of maleic acid as an internal standard.The solution was analyzed by 1 H NMR spectroscopy over 48 h at room temperature (Figure 4).At t = 1 h, vinyl sulfone 1 was present in solution at the expected 2 mM concentration.After 24 h, a reduction in the abundance of 1 by 10% and the appearance of approximately 9% of its cyclic isomer 2 were observed.After 48 h, 81% of 1 remained, with approximately 19% of 2 present in the phosphate buffer.These results demonstrated that partial cyclization of 1 into 2 might occur under standard cell culture conditions.Whether this decrease in the effective concentration of 1 would influence its efficacy as an antiviral nsP2 inhibitor would depend on the frequency of dosing and the time course of the bioassay.

Optimized Synthesis of β-Amidomethyl Vinyl Sulfone 1
Alternative amide coupling conditions [11] were explored for the synthesis of acyclic β-amidomethyl vinyl sulfone 1 that would avoid the formation of cyclic byproduct 2 and the subsequent preparative HPLC separation (Table 1).The original conditions using HBTU as the coupling agent and DIPEA as the base in DMF yielded a 60:40 mixture of 1 and 2 by analytical UPLC analysis (entry 1, Table 1).Switching the coupling agent to hexafluorophosphate azabenzotriazole tetramethyl uronium (HATU) or benzotriazol-1-yl-oxytripyrrolidino-phosphonium hexafluorophosphate (PyBOP) resulted in a similar ratio of 1 and 2 (entries 2 and 3).Use of n-propanephosphonic acid anhydride (T3P) as the coupling agent with triethylamine (TEA) as the base also yielded a 60:40 ratio of 1 and 2, as did N,N′-diisopropylcarbodiimide (DIC) with 4-dimethylaminopyridine (DMAP) (entries 4 and 5).Each of these amide coupling reactions (entries 1-5) occurred in the presence of strongly basic amines with pKa 9.7-10.9.Switching to 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) in combination with hydroxybenzotriazole (HOBt), where amide coupling occurs without the addition of an amine base, resulted in an improved 70:30 ratio of 1 and 2 (entry 6).This result prompted us to trial the amide synthesis with the original benzotriazole tetramethyl uronium coupling agent but as a tetrafluoroborate salt (TBTU) These results demonstrated that partial cyclization of 1 into 2 might occur under standard cell culture conditions.Whether this decrease in the effective concentration of 1 would influence its efficacy as an antiviral nsP2 inhibitor would depend on the frequency of dosing and the time course of the bioassay.

Optimized Synthesis of β-Amidomethyl Vinyl Sulfone 1
Alternative amide coupling conditions [11] were explored for the synthesis of acyclic β-amidomethyl vinyl sulfone 1 that would avoid the formation of cyclic byproduct 2 and the subsequent preparative HPLC separation (Table 1).The original conditions using HBTU as the coupling agent and DIPEA as the base in DMF yielded a 60:40 mixture of 1 and 2 by analytical UPLC analysis (entry 1, Table 1).Switching the coupling agent to hexafluorophosphate azabenzotriazole tetramethyl uronium (HATU) or benzotriazol-1yl-oxytripyrrolidino-phosphonium hexafluorophosphate (PyBOP) resulted in a similar ratio of 1 and 2 (entries 2 and 3).Use of n-propanephosphonic acid anhydride (T3P) as the coupling agent with triethylamine (TEA) as the base also yielded a 60:40 ratio of 1 and 2, as did N,N ′ -diisopropylcarbodiimide (DIC) with 4-dimethylaminopyridine (DMAP) (entries 4 and 5).Each of these amide coupling reactions (entries 1-5) occurred in the presence of strongly basic amines with pK a 9.7-10.9.Switching to 1-ethyl-3-(3dimethylaminopropyl)carbodiimide (EDC) in combination with hydroxybenzotriazole (HOBt), where amide coupling occurs without the addition of an amine base, resulted in an improved 70:30 ratio of 1 and 2 (entry 6).This result prompted us to trial the amide synthesis with the original benzotriazole tetramethyl uronium coupling agent but as a tetrafluoroborate salt (TBTU) in pyridine as a solvent.Under these less basic conditions (entry 7), exclusive formation of 1 was observed by UPLC analysis.An alternate synthesis of β-amidomethyl vinyl sulfone 1 was also developed to avoid the liability of cyclization during the amide bond formation (Scheme 3).MOM-protection of pyrazole 3 occurred exclusively at the N1 position as expected [12,13] and was confirmed by 1 H- 13 C HMBC NMR analysis.The coupling of MOM-protected pyrazole 5 using the TBTUpyridine protocol gave exclusively amide 6 with no byproducts resulting from cyclization.Acid-mediated cleavage of the MOM protecting group yielded pure 1 as an HCl salt without the need for chromatography.The HCl salt of 1 was stable upon storage as a solid or as a 10 mM DMSO stock solution and was routinely checked for purity prior to bioassay.However, researchers are cautioned that commercial samples of 1 or other heterocyclic βamidomethyl vinyl sulfones (ChemSpace, Enamine) may contain undetermined quantities of the respective dihydropyrazolo[1,5-a]pyrazin-4(5H)-ones unless careful quality control by 1 H NMR and UPLC analysis has been performed.
in pyridine as a solvent.Under these less basic conditions (entry 7), exclusive formation of 1 was observed by UPLC analysis.An alternate synthesis of β-amidomethyl vinyl sulfone 1 was also developed to avoid the liability of cyclization during the amide bond formation (Scheme 3).MOM-protection of pyrazole 3 occurred exclusively at the N1 position as expected [12,13] and was con firmed by 1 H- 13 C HMBC NMR analysis.The coupling of MOM-protected pyrazole 5 using the TBTU-pyridine protocol gave exclusively amide 6 with no byproducts resulting from cyclization.Acid-mediated cleavage of the MOM protecting group yielded pure 1 as an HCl salt without the need for chromatography.The HCl salt of 1 was stable upon storage as a solid or as a 10 mM DMSO stock solution and was routinely checked for purity prior to bioassay.However, researchers are cautioned that commercial samples of 1 or other heterocyclic β-amidomethyl vinyl sulfones (ChemSpace, Enamine) may contain undeter mined quantities of the respective dihydropyrazolo[1,5-a]pyrazin-4(5H)-ones unless care ful quality control by 1 H NMR and UPLC analysis has been performed.

Optimized Synthesis of Dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2
Conditions for controlled cyclization of the HCl salt of β-amidomethyl vinyl sulfone 1 to dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 by intramolecular aza-Michael reaction were explored (Table 2).Reactions were performed at room temperature for 2 h in the presence of different bases, with the conversion to 2 monitored by UPLC.K2CO3 in ethano gave efficient cyclization (entry 1).However, no cyclization occurred with K2CO3 in water due to the limited aqueous solubility of 1 (entry 2).Na2CO3 in an aqueous dioxane mixture resulted in clean cyclization to 2 (entry 3).The less basic NaHCO3 in methanol or water was not as effective as K2CO3 (entries 4 and 5) within the 2 h reaction time.The amine base TEA in methanol was also less effective (entry 6).The stronger amine bases, DIPEA or DBU, showed faster conversion to 2 but were still not as rapid as the inorganic bases (en tries 7 and 8).Na2CO3 in an aqueous dioxane (entry 3) was chosen as the optimal cycliza tion conditions for the synthesis of 2 since it was fast, resulted in a simple work-up, and avoided the need for chromatography.2).Reactions were performed at room temperature for 2 h in the presence of different bases, with the conversion to 2 monitored by UPLC.K 2 CO 3 in ethanol gave efficient cyclization (entry 1).However, no cyclization occurred with K 2 CO 3 in water due to the limited aqueous solubility of 1 (entry 2).Na 2 CO 3 in an aqueous dioxane mixture resulted in clean cyclization to 2 (entry 3).The less basic NaHCO 3 in methanol or water was not as effective as K 2 CO 3 (entries 4 and 5) within the 2 h reaction time.The amine base TEA in methanol was also less effective (entry 6).The stronger amine bases, DIPEA or DBU, showed faster conversion to 2 but were still not as rapid as the inorganic bases (entries 7 and 8).Na 2 CO 3 in an aqueous dioxane (entry 3) was chosen as the optimal cyclization conditions for the synthesis of 2 since it was fast, resulted in a simple work-up, and avoided the need for chromatography.

Pharmacokinetic Properties of 1 and 2
Although vinyl sulfones have been used as covalent warheads for inhibition of a wide range of cysteine proteases [7,8], there are relatively few reports of their use in vivo [14].To explore the potential of 1 as a drug lead for the treatment or prevention of alphavirus infections, pharmacokinetic experiments were performed in mice following a 10 mg/kg i.v.dose (Figure 5).

Pharmacokinetic Properties of 1 and 2
Although vinyl sulfones have been used as covalent warheads for inhibition of a wide range of cysteine proteases [7,8], there are relatively few reports of their use in vivo [14].To explore the potential of 1 as a drug lead for the treatment or prevention of alphavirus infections, pharmacokinetic experiments were performed in mice following a 10 mg/kg i.v.dose (Figure 5).Unfortunately, β-amidomethyl vinyl sulfone 1 had very rapid clearance in mice, with plasma levels falling below the limit of MS detection after 1 h and a half-life of only ~10 min.As a control, the pharmacokinetics of the cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 were determined (Figure 2).In comparison to 1, the cyclic pyrazole 2 showed reduced plasma clearance with an i.v.half-life of ~30 min and a 4-fold greater plasma exposure.Reanalysis of the plasma samples from the dosing of 1 showed no detectable levels of 2, demonstrating that the rapid clearance of 1 in vivo was not due to extensive cyclization to 2. Instead, the rapid clearance of 1 was likely due to its β-amidomethyl vinyl warhead, which appeared to be a liability for in vivo exposure in mice.Despite the fact that extensive interconversion between 1 and 2 was not observed in vivo, the improved pharmacokinetics of the cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-one raised the question of whether it might still function as a prodrug if low levels of the reactive vinyl sulfone could be formed in the presence of the viral enzyme.

Reversibility of the Aza-Michael Reaction
Cyclization of 1 to 2 was favored under basic conditions but occurred only slowly at physiological pH.We were eager to determine if the cyclization was reversible and under what conditions the dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 could revert to the acyclic β-amidomethyl vinyl sulfone 1 by a retro-Michael reaction (Scheme 4).Unfortunately, β-amidomethyl vinyl sulfone 1 had very rapid clearance in mice, with plasma levels falling below the limit of MS detection after 1 h and a half-life of only ~10 min.As a control, the pharmacokinetics of the cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 were determined (Figure 2).In comparison to 1, the cyclic pyrazole 2 showed reduced plasma clearance with an i.v.half-life of ~30 min and a 4-fold greater plasma exposure.Reanalysis of the plasma samples from the dosing of 1 showed no detectable levels of 2, demonstrating that the rapid clearance of 1 in vivo was not due to extensive cyclization to 2. Instead, the rapid clearance of 1 was likely due to its β-amidomethyl vinyl warhead, which appeared to be a liability for in vivo exposure in mice.Despite the fact that extensive interconversion between 1 and 2 was not observed in vivo, the improved pharmacokinetics of the cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-one raised the question of whether it might still function as a prodrug if low levels of the reactive vinyl sulfone could be formed in the presence of the viral enzyme.

Reversibility of the Aza-Michael Reaction
Cyclization of 1 to 2 was favored under basic conditions but occurred only slowly at physiological pH.We were eager to determine if the cyclization was reversible and under what conditions the dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 could revert to the acyclic β-amidomethyl vinyl sulfone 1 by a retro-Michael reaction (Scheme 4).The stability of the dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 under standard laboratory conditions suggested that equilibrium with the acyclic β-amidomethyl vinyl sulfone 1 lay almost entirely toward the cyclic form.For example, 1 was below the limits of UPLC detection in 10 mM DMSO stock solutions of 2 even after prolonged storage over several months.Since low levels of the electrophilic β-amidomethyl vinyl sulfone could theoretically still be present, we decided to test whether 1 could be captured using glutathione (GSH).Incubation of 1 with a 100-fold excess of GSH in phosphate buffer at 30 °C led to complete conversion to GSH adduct 9 within 8 h, as monitored by LCMS (Table 3 and Figure S1).In contrast, incubation of dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 with excess GSH did not result in the formation of the GSH adduct 9 even after 24 h.The GSH capture data demonstrated that no detectable level of the acyclic β-amidomethyl vinyl sulfone 1 was present in phosphate buffer, making it extremely unlikely that dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2, despite its improved pharmacokinetic properties, would be useful as a prodrug for 1. Notably, during structure-activity studies of the pyrazole β-amidomethyl vinyl sulfones [15], a phenylsulfonamide-5-substituted pyrazole analog 7 was synthesized that appeared to be less prone to intramolecular cyclization.Under cyclization conditions of Na-HCO3 (1.0 eq.) in MeOH, 1 was 100% converted to 2 in 36 h, but under the same conditions, only ~50% of 7 cyclized to 8.These results suggested that the structure of the pyrazole could influence the rate of aza-Michael cyclization and possibly the propensity for the reverse reaction.To explore this hypothesis, an additional series of GSH capture experiments was performed.Incubation of β-amidomethyl vinyl sulfone 7 with a 100-fold excess of GSH in phosphate buffer at 30 °C led to the formation of its corresponding GSH adduct 10, although conversion was slower than was seen with 1 and required 24 h to complete.More importantly, incubation of the corresponding dihydropyrazolo[1,5-a]pyrazin-4(5H)one 8 with GSH resulted in the formation of 3% of the GSH adduct 10 after 8 h and 6% after 24 h (Figure S1).These GSH capture experiments demonstrate that the cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 8 exists in equilibrium with the acyclic β-amidomethyl vinyl sulfone 7 in phosphate buffer.Unfortunately, the vinyl sulfone 7 was not sufficiently active as an nsP2 protease inhibitor (IC50~4 µM [15]) to allow us to test whether its formation from the cyclic form 8 would result in antiviral activity.However, our demonstration of the reversibility of the aza-Michael reaction at physiological pH adds credence to the potential use of dihydropyrazolo [1,5-a]pyrazin-4(5H)-ones as prodrugs for their corresponding cysteine reactive β-amidomethyl vinyl sulfones.
In The stability of the dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 under standard laboratory conditions suggested that equilibrium with the acyclic β-amidomethyl vinyl sulfone 1 lay almost entirely toward the cyclic form.For example, 1 was below the limits of UPLC detection in 10 mM DMSO stock solutions of 2 even after prolonged storage over several months.Since low levels of the electrophilic β-amidomethyl vinyl sulfone could theoretically still be present, we decided to test whether 1 could be captured using glutathione (GSH).Incubation of 1 with a 100-fold excess of GSH in phosphate buffer at 30 • C led to complete conversion to GSH adduct 9 within 8 h, as monitored by LCMS (Table 3 and Figure S1).In contrast, incubation of dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 with excess GSH did not result in the formation of the GSH adduct 9 even after 24 h.The GSH capture data demonstrated that no detectable level of the acyclic β-amidomethyl vinyl sulfone 1 was present in phosphate buffer, making it extremely unlikely that dihydropyrazolo[1,5a]pyrazin-4(5H)-one 2, despite its improved pharmacokinetic properties, would be useful as a prodrug for 1. Notably, during structure-activity studies of the pyrazole β-amidomethyl vinyl sulfones [15], a phenylsulfonamide-5-substituted pyrazole analog 7 was synthesized that appeared to be less prone to intramolecular cyclization.Under cyclization conditions of NaHCO 3 (1.0 eq.) in MeOH, 1 was 100% converted to 2 in 36 h, but under the same conditions, only ~50% of 7 cyclized to 8.These results suggested that the structure of the pyrazole could influence the rate of aza-Michael cyclization and possibly the propensity for the reverse reaction.To explore this hypothesis, an additional series of GSH capture experiments was performed.Incubation of β-amidomethyl vinyl sulfone 7 with a 100-fold excess of GSH in phosphate buffer at 30 • C led to the formation of its corresponding GSH adduct 10, although conversion was slower than was seen with 1 and required 24 h to complete.More importantly, incubation of the corresponding dihydropyrazolo[1,5a]pyrazin-4(5H)-one 8 with GSH resulted in the formation of 3% of the GSH adduct 10 after 8 h and 6% after 24 h (Figure S1).These GSH capture experiments demonstrate that the cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 8 exists in equilibrium with the acyclic β-amidomethyl vinyl sulfone 7 in phosphate buffer.Unfortunately, the vinyl sulfone 7 was not sufficiently active as an nsP2 protease inhibitor (IC 50 ~4 µM [15]) to allow us to test whether its formation from the cyclic form 8 would result in antiviral activity.However, our demonstration of the reversibility of the aza-Michael reaction at physiological pH adds credence to the potential use of dihydropyrazolo[1,5-a]pyrazin-4(5H)-ones as prodrugs for their corresponding cysteine reactive β-amidomethyl vinyl sulfones.
In conclusion, (E)-5-(2-ethoxyphenyl)-N-(3-(methylsulfonyl)allyl)-1H-pyrazole-3carboxamide (RA-0002034, 1) is a covalent inhibitor of nsP2 cysteine proteases with potent antiviral activity against New and Old World alphaviruses.Although 1 was prone to intramolecular cyclization to a cyclic dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 under basic conditions, two modified procedures were developed for the synthesis of the pure acyclic β-amidomethyl vinyl sulfones as their TFA or HCl salts that can be employed in analog development for structure-activity studies of nsP2 protease inhibitors.The primary liability of 1 as an anti-alphavirus drug lead was its very high clearance in mice, which prompted us to explore the use of the inactive dihydropyrazolo[1,5-a]pyrazin-4(5H)-one 2 as a potential prodrug.Although cyclic 2 showed no evidence for a retro aza-Michael reaction to give 1, the phenylsulfonamide-5-substituted pyrazole analog 8 formed detectable levels of its acyclic β-amidomethyl vinyl sulfone 7, as evidenced by capture with GSH.These results demonstrate that the dihydropyrazolo[1,5-a]pyrazin-4(5H)-one chemotype can function as a masked form of the cysteine-reactive β-amidomethyl vinyl sulfone.Synthesis of dihydropyrazolo[1,5-a]pyrazin-4(5H)-one analogs with substituents that further favor equilibrium with their acyclic vinyl sulfones may provide a new prodrug strategy for covalent inhibition of viral nsP2 cysteine proteases.

GSH Capture Assay
A 10 mM solution of GSH (Sigma Aldrich Cat# G4251 (St. Louis, MO, USA)) was prepared in a pH 7.4 phosphate buffer.A 10 mM DMSO solution of the test compound was diluted in phosphate buffer to give a solution at 100 µM with 1% DMSO.At time zero (t = 0), 50 µL of the 100 µM test compound solution was added to an Eppendorf tube containing 50 µL phosphate buffer and 50 µL of 10 mM GSH solution.The final concentrations of the compound and GSH were maintained at 50 µM and 5 mM, respectively.The Eppendorf tube was vortexed, and then the sample was transferred to a high-recovery autosampler vial for LCMS analysis.Analysis was performed at 8 and 24 h time points, and the percentage of GSH adduct formation was calculated using Agilent LCMS software (OpenLab CDS Version 2.7).

Figure 5 .
Figure 5. Pharmacokinetics of 10 mg/kg i.v.doses of 1 or 2 in mice.(A) Pooled plasma concentrations.In vivo formation of 2 from 1 was not detected.(B) Calculated pharmacokinetic parameters.

Figure 5 .
Figure 5. Pharmacokinetics of 10 mg/kg i.v.doses of 1 or 2 in mice.(A) Pooled plasma concentrations.In vivo formation of 2 from 1 was not detected.(B) Calculated pharmacokinetic parameters.

Table 1 .
Optimization of Amide Coupling between 3 and 4 a .

Table 1 .
Optimization of Amide Coupling between 3 and 4 a .

Table 2 .
Optimization of Intramolecular Cyclization of 1 to 2 a .

Table 2 .
Optimization of Intramolecular Cyclization of 1 to 2 a .

Table 3 .
Reversibility of the aza-Michael cyclization by GSH capture of the acyclic vinyl sulfones.

Table 3 .
Reversibility of the aza-Michael cyclization by GSH capture of the acyclic vinyl sulfones.