Improved synthesis of the PAR-1 thrombin receptor antagonist RWJ-58259

A significant improvement on the synthesis of the PAR-1 antagonist RWJ-58259 is described, which involves a base-related two-fold yield increase in the indazole N-alkylation, and an easier purification and a nine-fold yield increase in the urea formation, by using triphosgene/propylene oxide as urea coupling traceless reagents.


Introduction
In addition to the key role of thrombin in blood coagulation processes, this serine protease mediates multiple cellular responses, such as platelet aggregation and cell proliferation in fibroblasts, neurons, and endothelial, smooth muscle, and tumor cells. 1 These thrombin effects on cells are mediated by activation of the denominated protease-activated receptors (PARs).Among these, PAR-1 is the principal thrombin-activated receptor involved in platelet aggregation and in endothelial and tumor cell proliferation.Human PAR-1 is activated by the thrombin-mediated cleavage of the N-terminal extracellular domain at the Arg 41 /Ser 42 peptide bond, unveiling the recognition sequence SFLLRN, which acts as a tethered activation ligand.It has been proposed that a PAR-1 antagonist could have potential for treating thrombosis, atherosclerosis, inflammation and cancer metastasis, without altering thrombin's role in hemostasis. 1 The first potent and in vivo effective PAR-1 antagonists were SFLLRN-based peptidomimetics, which provided the first in vivo proof-of-concept of the therapeutic potential of PAR-1 antagonists.1a,2 Among these, the indazole-derived urea RWJ-58259 (Scheme 1, 7) reached advanced preclinical studies in different animal models 3,4 and, in spite of its low oral bioavailability, 3 it is considered an standard reference in pharmacological studies on PAR-1 receptors. 4,5n the context of a research project on novel peptidomimetic PAR-1 antagonists, we needed a reference antagonist as standard for the biological evaluation of new compounds.Up to now, no PAR-1 antagonist is commercially available and we could not get a sample from their respective developing companies.So, we decided to prepare RWJ-58259 following the reported methodology.2a,6 As shown in Scheme 1, this methodology involved a convergent solution-phase formation of the urea derivative 6 from the two corresponding moieties: the 6-amino-indazole fragment 4 and the dipeptide fragment 5.However, when we tried to reproduce the patented procedure 6 we obtained very low yields both in the step of the N-alkylation of the indazole (a) and in the formation of the urea (c), which made it impractical to get the final compound RWJ-58259 (7) in a reasonable yield for our studies.Consequently, we undertook and describe herein the optimization of these two steps.

N-Alkylation of the indazole derivative 1
The alkylation of indazole 1 to the 1-(2,6-dichlorobenzyl) derivative 2 has been described in 33% yield using one equiv of 2,6-dichlorobenzyl bromide and KOH as base.2a,6 When we first attempted to reproduce this alkylation, compound 2 was obtained in a similar low yield (34%) and the TLC and HPLC (Novapak C 18 ) analyses of the crude reaction mixture showed the presence of a peak (30%) with the same R f and t R of the starting material 1. Trying to increase the yield of 2, we repeated the reaction using a 30% excess of 2,6-dichlorobenzyl bromide, but, contrary to what we expected, the yield of 2 decreased and we could only isolate the bromide of the product of dialkylation (in the indazole and in the pyrrolidine) 3, which coeluted with the starting indazole 1 on TLC and on the HPLC Novapak C 18 column.In view of this result, we evaluated by HPLC the influence of different bases (KOH, Cs 2 CO 3 , TEA, NaH) on the mono/dialkylation rate.As shown in Table 1, the best yield of the monoalkylation product 2 was obtained with Cs 2 CO 3 (entry 2), although, the reaction was slower and required 24 h.TEA (entry 3) not only produced higher ratio of the dialkylation product 3, as also happened with NaH, but led to a more complex reaction mixture.Taking into consideration these results, the alkylation was carried out with one equiv of 2,6-dichlorobenzyl bromide in the presence of one equiv of Cs 2 CO 3 .Under these conditions, the monoalkylated indazole 2 was chromatographically isolated in a 66% yield (double compared to the described procedure 2a,6 ).

Urea formation from the indazole and dipeptide units 4 and 5
This coupling was described in 26% yield, by using 4-nitrophenyl chloroformate as ureacoupling reagent in the presence of diisopropylethylamine at -20 ºC.2a,6 When we applied this methodology, the urea 6 was obtained in a yield lower than 10%, what made it impractical for our objective and which moved us to study alternative procedures for its improvement.It is worth mentioning that, as the urea 6 and the 6-aminoindazole 4 coeluted on TLC and on our initially used HPLC column (Novapak C 18 , 4 µM, 3.9×150 mm), HPLC-MS (Xbride C 18 , 3.5 µM, 2.1×100 mm column) was used to analyze the results of this study.First, we observed that the low yield of the described methodology was not significantly affected by the reaction temperature (-40, -20 or 0 ºC).Then, we decided to evaluate a methodology that had been successfully used in our laboratory for the synthesis of asymmetric ureas, 7 which involved the in situ formation of the isocyanate of one of the two amino moieties, by reaction with bis(trichloromethyl)carbonate (triphosgene) in the presence of Et 3 N as HCl acceptor, followed by reaction with the other amine.When this methodology was applied to the formation of the 6aminoindazole-derived isocyanate 8 (Scheme 2), previously to the reaction with 0.5 equiv of the dipeptide 5, the urea 6 precipitated from the reaction medium in a 50 % yield, but, the 1 H NMR spectrum of this precipitate showed that Et 3 N•HCl had coprecipitated along with urea 6.Then, we tried to remove the ammonium salt by washing the precipitate with H 2 O and several organic solvents, but all attempts were unsuccessful.
In view of the significant improvement in the yield of 6 achieved using triphosgene, and to avoid the formation of the ammonium chloride, we studied the methodology using a traceless HCl-acceptor, such as propylene oxide.This replacement gave an excellent result, as the urea 6 was isolated by precipitation from the reaction medium in 91% yield and in a HPLC purity higher than 95%.The excess of the starting 6-aminoindazole 4 was recovered from the filtrate of the reaction mixture.Trying to optimize the reaction efficacy, the influence of the molar ratio 4/5 (1/0.5, 1/0.7 and 1/0.9) was also studied.The results showed that while the reaction yield was not significantly affected, when the molar proportion of the dipeptide 5 was higher than 0.5, this compound coprecipitated with the urea 6, and the purification of the reaction mixture required chromatography.
The final N-Boc-deprotection was carried out quantitatively by treatment with a 3 N solution of HCl in MeOH.This last step did not require chromatographic purification, as in the previously described deprotection using TFA/anisole.2a, 6 The hydrochloride of RWJ-58259 (7) was obtained in 91% overall yield (from 5, nine-fold higher than by the described methodology) and with higher than 95% HPLC purity.

Conclusions
In summary, we have developed a significant improvement of the described synthesis of the PAR-1 antagonist RWJ-58259 at laboratory scale (0.1-1 g) required to obtain enough quantity to be used as a standard reference in our pharmacological studies.First, replacement of KOH by Cs 2 CO 3 has allowed to double the yield of the N-alkylation of the indazole derivative 1.Second, by using triphosgene in the presence of propylene oxide as traceless reagents for the urea formation we have achieved a nine-fold increase in the yield of this key step, with respect to that obtained using 4-nitrophenyl chloroformate and diisopropylethylamine.Finally, the N-Boc deprotection by 3N HCl solution in MeOH allowed the quantitative isolation of the corresponding hydrochloride, avoiding a chromatographic purification in this step.

Experimental Section
General Procedures.All reagents were of commercial quality.Solvents were dried and purified by standard methods.Analytical TLC was performed on aluminum sheets coated with a 0.2 mm layer of silica gel 60 F 254 .Silica gel 60 (230-400 mesh) was used for flash chromatography.Analytical RP-HPLC was performed on a Novapak C 18 (3.9×150mm, 4µm) column, with a flow rate of 1mL/min, and using a tunable UV detector set at 214 nm.Mixtures of CH 3 CN (solvent A) and 0.05 % TFA in H 2 O (solvent B) were used as mobile phases.HPLC-EMS was performed on an XBride C 18 (2.1×100mm, 3.5 µm) column at 30 ºC, with a flow rate of 0.25 mL/min.5-80% Gradients of CH 3 CN with 0.08% of formic acid (solvent A) in 0.1% of formic acid in H 2 O (solvent B) were used as mobile phases. 1 H NMR spectra were recorded at 300, 400, or 500 MHz, using TMS as reference, and 13 C NMR spectra were recorded at 100, or 125 MHz.ESI-MS spectra were performed, in positive mode, using MeOH as solvent.

Supplementary Material Avalaible
HPLC-EMS and 1 H NMR spectra of ureas 6 and 7.

Table 1 .
Influence of the base on the N-alkylation of 1 a Determined by HPLC (Novapak C 18 , 4 µM, 3.9×150 mm).b Described yield of isolated compound.c Yield of isolated compound.