and AZT-linked 5-(1,3-dioxoalkyl)tetrazoles and

N3 of thymidine and of 3’-azido-3’-deoxythymidine (AZT) has been linked to a tetrazole ring Received .......... by condensation of nucleoside-derived 2-oxonitriles with the lithium salt of 5-acetyl-1-(4- Revised ............. fluorobenzyl)tetrazole (obtained by a “click” reaction). 4-Acetyl-1,2,3-triazole, also prepared Accepted .......... by a Cu-catalysed cycloaddition, has been similarly linked. A route for the conjugation of Available on line ...... NRTIs with pharmacophoric elements of integrase inhibitors (INIs) has thus been disclosed. ______________________________

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The HIV-1 integrase (IN) is a well-established target in the long-lasting fight against AIDS. Many integrase inhibitors (INIs) have been developed since Shionogi and Merck scientists disclosed years ago that derivatives of 2,4-dioxobutanoic acids (so-called diketo acids) blocked the strand transfer capability of the viral integrase. 1 Other small molecules sharing the same pharmacophoric elements, such as bioisosteric 5-(1,3-dioxopropyl)tetrazoles or heteroaromatic substrates were soon designed and synthesised. Famous INIs are depicted in Figure 1 in their Mg 2+ -complexing anionic forms; 2 the outstanding one is yet raltegravir (Isentress, approved by FDA in 2007). 3 Nevertheless, as expected owing to the fast HIV-1 mutations, resistance against raltegravir has appeared as well. 4 Alternative INIs with independent resistance profiles and drug candidates tackling other steps of the HIV life cycle are always a must.
Conjugation of the pharmacophore-a Mg 2+ -chelating moiety with a hydrophobic 4-fluorobenzyl group-with nucleosides could favour the trans-membrane transport of the new chemical entities and might "simplify" their cata-bolism affording less toxic metabolites. We also believed that conjugation of the pharmacophore with a nucleoside reverse transcriptase inhibitor (NRTI) could give rise to molecules of potentially synergic, dual effect (both NRTI and INI). 5 We planned to link tetrazole or 1,2,3-triazole rings, via the formation of a 1,3-dioxo moiety, to nucleosides, without disturbing the key features of the latter. We chose thymidine and 3'-azido-3'-deoxythymidine (AZT, zidovudine) as working models ( Figure 2). If we were able to generate such conjugates, we could apply the concept to a number of cases by linking other nucleosides and NRTIs to related INI moieties (e.g., to the 5-hydroxypyrimidinone substructure of raltegravir and its congeners). We report here our first approach: the synthesis of molecules such as those at the left side of Figure 2, with the nucleoside and the hydrophobic group at opposite sites of the chelating group. In this context, we evaluated first the effect of the attachment of simple (but relatively polar) substituents to appropriate positions of AZT. The activity against wildtype HIV-1 (NL4-3) 6 of the samples shown in Figure 3 (1a-f), prepared in our lab, 7 clearly indicated that polar organic groups attached to position 3 (N3) of the thymine ring of AZT reduced slightly or only moderately its activity (but not by a factor of let's say 10 3 , which would have made our objective nonsense). Thus, the thymidine kinase seems capable of converting these N3-substituted AZTs 8 into their phosphate esters (subsequently phosphorylated up to the triphosphate esters) and incorporate them into the nascent viral DNA, as any other chain-terminating NRTI.
The technical problem of the preparation of tetrazoles C5-substituted with COCH 3 or related electron-withdrawing groups (EWGs), under very mild conditions, was solved by us time ago by means of a Cu 2 (OTf) 2 -catalysed cycloaddition. 9 Moreover, the preparation of 1,2,3-triazole derivatives by Cu(I)-catalyzed alkyne-azide cycloaddition (CuAAC, Sharpless' "click chemistry") is currently a subject of great scope. 10 We applied these two "click" reactions to 4-fluorobenzyl azide (Scheme 1), to prepare 2 and 3. Following own protocols, 7 we prepared nucleoside substrates represented by structure 4. Among various attempted condensation reactions, 11 only the coupling of an excess of lithium enolates of 2 and 3, at low temperature, with appropriate nucleoside derivatives 4 and 5 (with Z = CN) afforded satisfactory results. 12

Scheme 1. Preparation of starting materials.
With all this information in hand, we performed the syntheses summarised in Scheme 2. We obtained aldehydes 4 (Z = H) and 5 (Z = H), for the first time, from the corresponding nucleosides (TBS-diprotected thymidine, as a model for checking the reactions, and TBS-protected AZT) and freshly prepared propynal in the presence of DMAP, in CH 2 Cl 2 -CH 3 CN at rt. We treated them with trimethylsilyl cyanide; the silylated cyanohydrins were oxidised in situ with 2,3-dichloro-5,6-dicyano-p-benzoquinone (DDQ) in moist 1,4-dioxane, 13 to give 4 (Z = CN) and 5 (Z = CN) in 75-85% overall yields. In the most delicate step (the most prone to give rise to decomposition of the nucleoside substrates), these 2-oxonitrils were added drop-wise at -78ºC to 2 and 3 (220 mol%), to afford 6-9 in 44-70% yields. 14 Scheme 2. Synthesis of the AZT derivatives 10 and 11.
The integrase strand-transfer inhibitory activities of 10 and 11 (enzyme inhibitory concentrations, IC 50 ) 16 are collected in Table 1, together with their effective concentrations (EC 50 ) and cytotoxic concentrations (CC 50 ) under the conditions indicated 16 (which differ from those of Ref. 6 but are useful for comparison). Compound 10 showed a low INI activity (see Table 1, entry 2, IC 50 = 350 µM) while that of 11 (entry 3) was undetectable; an additional chelating group or atom (COO -, OH, CONH or heterocyclic N) closer to the enol-enone moiety is probably required. On the other hand, 10 and 11 were almost as active as AZT as far as EC 50 values 16 are concerned (compare entries 2 and 3 to entry 4), a fact that we expected (see the preceding page) but it had to be demonstrated. In this context, we should stress that 11 was more cytotoxic than AZT, so that we will discontinue the optimisation of this triazole derivative (of triazole derivatives in general); since thymidine derivatives also appeared to be cytotoxic (e.g., TBS-deprotected 7, CC 50 = 6 µM), we ruled out them for future improvements as well. Only tetrazole derivative 10, among 6-11, might become a lead. In summary, we have been fortunate to obtain for the first time representative conjugates of nucleosides and 1,3dioxopent-4-enyl derivatives of tetrazole and 1,2,3-triazole. Condensation of 5-acetyl-1-(4-fluorobenzyl)tetrazole and 4-acetyl-1-(4-fluorobenzyl)-1,2,3-triazole with nucleosidederived 2-oxonitriles (never reported so far in the nucleoside field, a reaction only successful with aromatic substrates) was the method of choice. The key synthetic challenges have been solved. The integrase inhibition activity of the most promising "minimalistic" candidate (10) is still insufficient, but we have demonstrated that it keeps the features of AZT derivatives. Syntheses of other conjugates, more advanced prototypes with one further Mg 2+ -chelating group on 10, are under way.

Acknowledgments
This work was started with funds from the FP6 of the European Union Commission, within the TRIoH project (Targeting Replication and Integration of HIV-1); it included a 3.5-year studentship to L.B. Additional support from the Spanish Government, by means of grant SAF2005-24643-E (2006SAF2005-24643-E ( -2007 to J.V. is acknowledged. Work done at Cachan (Table 1)  2813-2819, and references cited therein. 14. Experimental procedure. The methyl ketone (2 or 3, 2.20 mmol) was solved in anhyd. THF (10 mL) and cooled to -78ºC. This solution was added dropwise, under Ar, to a LiHMDS solution (0.2 M, 11 mL) at -78ºC; stirring was maintained for further 5 min. Acyl cyanide (4 or 5, 1.00 mmol), solved in anhyd. THF (5 mL) at -78ºC was then added dropwise to the enolate solution. The reaction mixture was stirred for 2 h while the temperature was allowed to rise up to -10ºC. The resulting dark-red solution was quenched with cold aqueous HCl (0.05 M, 4.5 mmol), to give a yellowish solution. The mixture was diluted with EtOAc (50 mL) and water (20 mL), the layers were separated and the aqueous one extracted twice with further EtOAc (20 mL). The organic extracts were treated with brine (25 mL), dried over Na 2 SO 4 , filtered and evaporated to dryness. Purification of the crude product by silica gel chromatography (hexanes/EtOAc/AcOH, 70:30:1)