Synthesis of New Acadesine (AICA-riboside) Analogues Having Acyclic d-ribityl or 4-Hydroxybutyl Chains in Place of the Ribose

The antiviral activity of certain acyclic nucleosides drew our attention to the fact that the replacement of the furanose ring by an alkyl group bearing hydroxyl(s) could be a useful structural modification to modulate the biological properties of those nucleosides. Herein, we report on the synthesis of some novel acadesine analogues, where the ribose moiety is mimicked by a d-ribityl or by a hydroxybutyl chain.


Introduction
A complete understanding of the interactions of the complex metabolic network and of its numerous and in some cases unclear regulator mechanisms still today constitutes a challenge for many researchers in the biological and biomedical fields. This is especially significant if it refers to cells that are affected by diseases and survive with certain modified metabolic pathways. The nucleosides, and OPEN ACCESS their structurally related biomolecules have a very important role in the metabolism acting as synthetic precursors and regulatory agents and being involved in signal transduction. They can be agonists or antagonists of central enzymes in normal or altered metabolic pathways and can be useful tools to demonstrate or affect metabolic rewiring. In most cases they have become important drugs [1][2][3][4][5][6]. In summary they are molecules that stimulate intense research aimed at the development of new structural analogues possessing potential regulatory or pharmacological activities [7][8][9][10][11].
In this context, 5-aminoimidazole-4-carboxamide riboside (acadesine or AICAR, Figure 1) has central role, acting as both a purine biosynthetic precursor and as a modulator of a very high number of biological properties. AICAR, after its 5'-phosphorylation to ZMP, is involved in important metabolic pathways through the activation of the AMP-activated protein kinase (AMPK). In the cells AICAR is phosphorylated to ZMP (Figure 1) that is a mimic of adenosine 5'-monophosphate (AMP) [12,13]. The direct binding of ZMP to an allosteric site of AMPK causes its phosphorylation and activation by a cellular kinase, resulting in a series of important metabolic events, including the inhibition of the basal and insulin-stimulated glucose uptake [14,15], the inhibition of lipid synthesis and the activation of certain ATP-generating processes such as glycolysis and fatty acid oxidation [16]. In the treatment of ischemia, the cardio-protective effect of AICAR has been attributed to the stimulation of the release of extracellular adenosine levels as well as to the activation of AMPK [13,17]. The AMPK pathway is also implicated in the regulation of cell proliferation and activation by AICAR could result in pro-apoptotic effects [18,19]. In particular, AICAR has been revealed to also be an antagonist of the protein Hsp90, a chaperone that regulates the correct interaction between proteins [20]. In such tumors Hsp90 is over-expressed, promoting aberrant cell survival and reproduction even in hostile environments [21].
Recently, it has been established that mutations affecting the Ca 2+ releasing channel RYR1 are associated with a broad spectrum of human disorders, including malignant hyperthermia, central core disease and core-rod myopathy [22]. By using a mouse model of malignant hyperthermia having a mutation in the RYR1 gene, Lanner et al. have recently demonstrated that AICAR can inhibit Ca 2+ leakage through RYR1 by a mechanism independent from AMPK activation, thus preventing heat-induced sudden death in the mutated mouse [22].
Nevertheless, AICAR is far from being a good drug lead because it has a short half-life in cells and is not strictly specific for the AMPK enzyme; furthermore, it suffers from a number of side effects: it increases uric acid production and favors lactic acidosis [15]. In light of the fact that AICAR can enter into AMPK-dependent or -independent processes [23][24][25], the design and synthesis of novel AICAR derivatives/analogues could be useful to better understand how the related metabolic pathways work and how to obtain new drug candidates.
The discovery of the antiviral activity of acyclovir and the acyclic nucleoside phosphonates [6,26,27] has emphasized that the replacement of the furanose ring by an alkyl group bearing hydroxyl(s) could be an interesting structural modification to induce new properties in the biological activity of the nucleosides. In most cases, the activity of these sugar-modified nucleosides has been attributed to the conformational freedom adopted by the alkyl chain that could, in principle, promote the recognition by the nucleoside related enzymes and prevent the development of viral resistance [27]. Furthermore, the conformational flexibility of acyclic nucleosides and nucleotides could also influence their base-pairing properties. In fact, Van Aerschot et al., with the aim of discovering universal nucleoside analogues, have inserted novel acyclic nucleoside derivatives in oligonucleotide strains, evaluating their hybridizing properties [28].
In recent years, we have focused our attention on the preparation of new base-and sugar-modified nucleosides and nucleotides both by classic solution chemistry and, more recently, by a solid-phase approach [29][30][31][32], enlarging the collection of new potential antimetabolites. Herein, we report on the synthesis of a small set of 5-aminoimidazole-4-carboxamides (AICAs) carrying D-ribityl or 4-hydroxybutyl chains at the N1-imidazole position, as well as 5-hydroxypentyl chains at the 4N carboxamide position.

Results and Discussion
The preparation of imidazole nucleoside analogues having non-glycosidic linkages has usually been accomplished by the condensation of 2-amino-2-cyanoacetate with a suitable chiral or achiral aminoalcohol in the presence of triethylorthoformate, obtaining compounds that show potent adenosine deaminase activities [33]. Other authors have reported on the synthesis of 1-(4-O-methyl-2deoxy-D-ribityl)-5-amino-4-carboxamide imidazole and its derivatives coupling the imidazole portion with a fully protected and activated polyhydroxyalkyl chain by S N 2 displacement [28].
In accordance with Hirota's procedure [34], in the first part of our work we attempted the preparation of derivative 5 (Scheme 1), in which a ribityl chain replaces the ribose moiety, by the direct reductive cleavage of the C1'-O4' bond of AICAR and of 2',3'-O-isopropylidene AICAR. However, this approach failed, resulting only in complex reaction mixtures in both cases. Therefore, we designed a new synthetic route, on the assumption that the construction of this kind of open-ribose AICAR could be performed starting from a suitable purine nucleoside, from which the 5-amino-4carboxamide moiety could be obtained through the degradation of the purine ring. In the last years we have reported some synthetic strategies to obtain AICAR, ZMP and their 4N-alkyl derivatives, through the purine ring degradation of suitable inosine or 5'-phosphate inosine precursors, carrying a strong electron-withdrawing group (2,4-dinitrophenyl, DNP) at the N1 of the purine ring [29][30][31][32]35].
Considering the instability of the DNP group under the reductive conditions necessary to open the ribose, its introduction at the N1 base position was performed after the reductive cleavage of the C1'-O4' bond of the 2',3'-O-isopropylideneinosine [34] by reaction with DIBAL-H in dry THF that furnished the ribitylinosine 1 in 68% yield. The hydroxyls of 1 were then protected by acetylation, yielding compound 2 (Scheme 1). Compound 2 was then reacted with 2,4-dinitrochlorobenzene (DNClB) in the presence of K 2 CO 3 to give the intermediate 3. This reaction served to activate the C2 position of the purine, rendering it susceptible to nucleophilic attack by the amines, which induces the cleavage of the N3-C2 purine bond. In particular, we have previously demonstrated that when a 1,ω-diaminoalkane is employed, the fate of the open intermediate depends on the length of the alkyl chain separating the two amino groups. If the diamine is composed of two or three methylene groups, AICAR is formed in a high yield [35]. Therefore, compound 3 was treated with a solution of ethylendiamine (EDA) in DMF and compound 4 was obtained (71% yield) by purine-ring opening/degradation and the concomitant deacetylation of the 4'-5' hydroxyl groups of the ribityl moiety. Finally, the isopropylidene group removal on 4 was performed by 10% trifluoroacetic acid (TFA) treatment, affording the N1-ribityl AICA 5 quantitatively.

Scheme 1. Synthesis of compounds 5 and 8.
In order to obtain further AICAR derivatives we probed the reactivity of 3 with the 5-amino-pentan-1-ol. As expected this reaction, following a mechanism that starts with the scission of the N3-C2 bond and then proceeds with the reclosure of the purine ring, furnished the N1-hydroxyalkyl-inosine 6 (85% yield). During this reaction a concomitant deacetylation of the hydroxyl groups at 4' and 5' ribityl moiety was observed.
It is well known that N1-alkylated inosines are susceptible to purine ring opening/degradation when treated with alkali [36], affording the corresponding 4N-alkyl AICARs [31,32,35]. In accordance with these data, 6 was refluxed with 5M NaOH solution in ethanol, giving a good yield (80%) of 7. The reaction was followed by UV spectrophotometry, because TLC monitoring was difficult to perform. After 5 h the disappearance of the purine band at λ max 249 nm (pH = 7) and the concomitant appearance of the imidazole band at λ max 268 nm (pH = 7) confirmed the end of the reaction. The isopropylidene group from 7 was removed as for 4, providing 8 in an almost quantitative yield.
In the second part of our synthetic work we planned to tune a synthetic procedure to obtain AICA derivatives bearing hydroxyalkyl chains at the AICA N1 and at the 4N positions [37,38]. To achieve this goal, we identified 6-chloropurine as a useful precursor on which to introduce the suitable 4-hydroxybutyl chain, followed by its transformation into the AICA derivative 10. Specifically, in a previous paper we demonstrated that 6-chloropurine could be readily transformed into 9 with a good yield by the following sequence of reaction: base-mediated alkylation on N9, acidic hydrolysis of chlorine, acetylation of the primary hydroxyl and reaction of the resulting hypoxanthine derivative with DNClB (Scheme 2) [39]. As for 3, the treatment of 9 with EDA in DMF produced the N1-hydroxybutyl AICA 10 in 69% yield. Alternatively, the treatment of 9 with 5-amino-pentan-1-ol produced 11 in a 78% yield, which, unexpectedly, still retained the acetyl group, as detected by spectroscopic analyses. This fact was not detrimental for the completion of the synthesis. In fact, as for 6, by refluxing 11 with 5 M NaOH solution in ethanol, 12 was obtained (in 75% yield), after the purine ring degradation and hydrolysis of the acetyl group.
The synthesis of compound 12 opened the way to the preparation of novel AICA analogues, combining sugar alterations with base modifications on the natural skeleton of the AICA riboside.

Conclusions
We have here reported useful synthetic procedures to introduce alternative moieties (D-ribityl or 4-hydroxybutyl) into the ribose portion of the AICA-riboside. The obtained AICAR analogues have the imidazole (AICA) without modifications (compounds 5 and 10) or bear the 5-hydroxypentyl chain on a carboxamide function (compounds 8 and 12). We believe that these synthetic pathways could enlarge the toolbox of the reactions operating on the AICA riboside (acadesine) and could furnish new "tuneable" AICAR analogues in terms of their molecular size, flexibility and hydrogen bond formation for their interactions with metabolic enzymes. Works are in progress to evaluate their activity into AMPK-dependent and -independent processes in order to better understand the metabolic pathways involving AICAR and ZMP.