Synthesis of benzo-, pyrido-, thieno- and imidazo-fused N- hydroxy-4-oxopyrimidine-2-carboxylic acid derivatives

Article history: N-Hydroxy-4-oxoquinazoline-2-carboxamide derivatives (cyclic hydroxamic acids) and related Received ...................... pyridoand thieno-substituted analogues, as well as a N-hydroxyhypoxanthine-2-carboxamide, Revised ........................ were synthesised for the first time, by means of a four-step sequence that involves a smooth Accepted ...................... reaction of aminohydroxamates with methyl trimetoxiacetate. Other strategies were unsuccessful. Available on line ............... © 2010 Elsevier Ltd. All rights reserved. ______________________________

Raltegravir (Isentress, MK-0518), developed by Merck, is the first HIV-1 integrase inhibitor (INI) approved by the FDA. 1 Raltegravir and its congeners (Figure 1), 1b,c elvitegravir (Gilead) and S/GSK candidates have similar pharmacophoric elements and mechanism of action. 1d,e The interest of some of us in hydroxamates 2a and the wellknown chelating power of hydroxamic acids, 2b prompted us to propose alternative candidates based on the replacement of the C-OH group of raltegravir by a N-OH group, which would enhance the acidity of the OH proton and would improve the coordination with magnesium ions. Thus, we designed N-hydroxypyrimidinone derivatives shown in Figure 1 (bottom). The challenge was to develop a general approach, as simple as possible, to these unknown molecules.
The strategies analyzed by us at the beginning of our project are summarised in Scheme 1. In the first approach (a), we envisaged the N-oxidation of N3 of 4-aminopyrimidines and of N1 of adenine, 3 where the π-electrondonating ability of the amino group is crucial for the formation of the N-oxide. The subsequent deamination, either enzymatically or via diazotization, 4 should afford the desired N-hydroxy pyrimidones (cyclic hydroxamic acids). While 2-cyano derivatives of adenine and adenosine (Scheme 1, top, X = CN) 5 gave the N-oxide in acceptable conversions (with a large excess of m-CPBA in CH2Cl2, as the reagent of choice among eight reagents examined), the corresponding methyl ester (X = COOMe) and amide (X = CONHCH2C6H4F = CONH-4-FBn) did not. In other words, a slight increase in the steric hindrance at position 2 hampered the N-oxidation. Although we achieved the conversion of Scheme 1. Retrosynthetic analysis of N-hydroxypyrimidin-4-ones. X = CN to X = COOMe with NaOMe in MeOH at 50 ºC, the methoxy group could not be replaced by the 4fluorobenzylamino group (a substitution that was very easy when the N-oxide substituent was absent). Moreover, whereas deamination of adenine derivatives (to obtain hypoxanthine derivatives) was quite efficient in our hands, either with adenosine deaminase or via nitrosation, the Noxide derivatives did not react. In short, the route via formation of N-oxides is not feasible.
Approach b involved a ring opening-ring closing process by using hydroxylamine (or a protected derivative of it, such as NH2OBn or NH2OTr) to attack the oxazinones of Scheme 1. We prepared the corresponding imidazo-oxazinone 6 by treatment of the o-amino carboxylic acid with a large excess of ClCOCOOMe and DIPEA in CH3CN. Opening with hydroxylamine derivatives gave mixtures coming from the attack of the hydroxylamines at the diverse electrophilic positions. Moreover, heating the open compounds in Ac2O gave rise to degradation compounds.
According to disconnection c, the hydroxamates (PG = Bn or PG = 4-methoxybenzyl = PMB), prepared from the amino ester, benzylamine or 4-methoxybenzylamine, and LiN(SiMe3)2 (LiHMDS), were heated with dimethyl oxalate and NaOEt, 7 in search of direct cyclisation. Only starting material was recovered. Heating with MeOCO-CONH-4-FBn was also unsuccessful. With ClCOCOOMe and DIPEA or in pyridine, mainly the O-acylation occurred. On the other hand, with K2CO3 or with DMAP, NHCOCOOMe derivatives were obtained, but their cyclisation to N-hydroxypyrimidinones under several dehydrating conditions did not take place.
Among all the approaches that we attempted only that one indicated as d in Scheme 1 turned out to be productive. Although the reaction of vicinal amino carboxamides with alkyl orthoformates such as CH(OR)3 is classical 8 and orthoesters such as methyl 2,2,2-trimethoxyacetate 9 have been used with vicinal diamines or aminophenols, 9d,e there are no reported precedents, according to a SciFinder search, of the reaction of amino hydroxamates with this oxalic acid orthoester. We first examined the conversion of amino ester 1a (methyl anthranilate) to benzohydroxamate 2a (Scheme 2). An excess of a strong base such as LiHMDS was required. 10 With a stronger base such as LDA the reaction was even faster. 10 Without strong base the reaction did not progress, even in refluxing 1,4-dioxane. The more general conditions starting from amino esters 1a-e, the analogous pyridines 1g and 1h and the methyl 5-aminoimidazole-4carboxylate 1k 11 are shown in Scheme 2, which gave rise to good-to-excellent yields of hydroxamates of type 2 (in the case of 1e, a mixture was obtained, but 2e and 2f were readily separated by column chromatography). On the other hand, owing to the decomposition of amino esters 1i and 1j in strong basic media, we prepared 2i and 2j by the standard coupling of the corresponding amino acids with O-(4methoxybenzyl)hydroxylamine (NH2OPMB) and 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC). 12 We then subjected 2a and (MeO)3CCOOMe excess to an screening of solvents, acid catalysts, temperatures, and reaction times. Most relevant results are shown in Table 1.  Acid catalysis was required, as expected for an orthoester, to achieve the complete conversion of 2a to 3a. Not only TsOH and Lewis acids were active, in several solvents (see entries 2-10 of Table 1), but also weaker acids such as AcOH could be used (entry 11). With 2 mol% of TsOH in refluxing CH3CN (entry 12), the cyclisation occurred in 4 h.
This cyclisation or condensation is general, as it could be applied to 2-aminohydroxamates 2b-f, to aminopyridines 2g and 2h and to thiophene derivatives 2i and 2j, as indicated in Table 2, with some adaptations. Afterwards, the ester groups of 3a-j were transformed to carboxamides 4a-j, 13 which were subjected to the cleavage of the O-PMB bond with TFA, to give the desired cyclic hydroxamic acids, 5a-j. Table 2. From amino esters 2a-j to quinazolinones 5a-f, pyridopyrimidinones 5g and 5h and thienopyrimidinones 5i and 5j a Entry Step 1 Yield (%) Step 2 Yield (%) b Step 3 Yield (%) Finally, we obtained N-hydroxyhypoxanthine 5k 14 in three high-yielding steps from 2k as detailed in Scheme 3.
Samples of the molecules thus prepared (5a, 5e, 5f and 5k) were tested as inhibitors of the 3'-processing and strand transfer activities of HIV-1 integrase at a single 10 µM concentration, with regard to raltegravir; 15a unfortunately, no activity was observed. They also appeared to be inactive for concentrations up to 10 µM in an HIV-1 antiviral, singleround-of-infection assay. 15b Scheme 3. From 2k to 5k. Chemical shifts (dH in regular type, dC in italics). Assignments confirmed by 2D NMR (HSQC).
In summary, we have synthesised new benzo-, pyrido-, thieno-and imidazo-fused N-hydroxypyrimidinones (quinazolinone, piridopyrimidinone, thienopyrimidinone and hypoxanthine ring systems) with carboxyl derivatives at position 2. The optimized sequence involves the reaction of ester groups with NH2OPMB and a strong base (LDA or LiHMDS) as well as a crucial acid-catalyzed condensation of these amino hydroxamates with methyl trimethoxyacetate, which has been carefully investigated to provide excellent yields of the cyclic compounds under very mild conditions. Overall, the protocol reported here for the first time for the synthesis of N-hydroxypyrimidinones 5a-k (4 steps, most of them in 85-98% yields, up to 76% overall yield) appears to be of wide scope. In short, in the unsuccessful search for new antiviral drugs we have developed the best procedure to date for reaching hitherto unknown pyrimidinone-2-carboxylic acid derivatives, which may enjoy other applications. 16