Multicomponent reactions in nucleoside chemistry

Summary This review covers sixty original publications dealing with the application of multicomponent reactions (MCRs) in the synthesis of novel nucleoside analogs. The reported approaches were employed for modifications of the parent nucleoside core or for de novo construction of a nucleoside scaffold from non-nucleoside substrates. The cited references are grouped according to the usually recognized types of the MCRs. Biochemical properties of the novel nucleoside analogs are also presented (if provided by the authors).


Introduction
Chemical modifications of natural ribose or 2'-deoxyribose nucleosides resulted in the development of a group of compounds referred to as nucleoside analogs ( Figure 1). The essential role of nucleoside analogs in medicine is reflected by the fact that currently thirty-six compounds from this class are used throughout the world in the therapy of viral or cancer diseases [1]. Moreover, several novel nucleoside analogs (including those embedded in versatile conjugate or pronucleotide scaffolds) are under clinical or preclinical trials [1]. Recent studies have also revealed a potential of nucleoside analogs as radiopharmaceuticals [2][3][4][5][6], antibiotics [7][8][9], anti-infective agents [10][11][12], or molecular probes [13,14]. Taking into account the importance of nucleoside analogs in medicine and biotechnology, there is a considerable interest in the development of simple and efficient synthesis of these compounds.
Multicomponent reactions (MCRs) represent an excellent tool for the generation of libraries of small-molecule compounds, for instance they are indispensable for the structure-activity relationship (SAR) studies. Many excellent comprehensive reviews on MCRs have been published. The reviews have covered the significant topics in this field, such as: (a) the applications of MCRs in the drug discovery process [15][16][17][18][19][20], or in the total synthesis [21,22]; (b) strategies developed for the construction of new structural frameworks [23]; (c) the use of specific building blocks [24][25][26][27][28], reagents [29][30][31][32], catalysts [33], reaction conditions [34,35], or preparative techniques [36] in MCRs; (d) methods for the design of new MCRs [37,38]; or (e) higherorder MCRs [39]. However to date, the application of MCRs in the chemistry of nucleoside analogs has not been methodically discussed. To the best of our knowledge, the only review arti- cles in this field were published from the Dondoni research group [40,41] or from the Torrence research group [42,43], and they were limited to the results obtained by these groups.
The present review covers reports published up to October 2013, and is devoted to the employment of MCRs in the synthesis of nucleoside analogs. The references were selected in accordance with the definition of a MCR given by Ugi et al.: "a multicomponent reaction comprises reactions with more than two starting materials participating in the reaction and, at the same time, the atoms of these educts contribute the majority of the novel skeleton of the product" [44]. In this review, we understand educts as compounds that contribute carbon atoms to the MCR product [45]. By the analogy to nucleosides included in the DNA/RNA nucleic acids, this review is limited to MCRs involving furanosyl nucleosides as (i) reaction components, or (ii) products obtained from non-nucleoside substrates. The cited references are grouped according to the usually recognized types of the MCRs [46].

Review 1. The Mannich reaction
The classical Mannich reaction yields β-aminoaldehydes or β-aminoketones and involves: an aldehyde, a primary (or a secondary) amine, and an enolizable aldehyde (or ketone) (Scheme 1a) [47,48]. The use of a hydrogen active component other than an enolizable aldehyde or ketone leads to a variety of structurally diverse products (Scheme 1b). The Mannich reaction products (commonly named as Mannich bases) can serve as starting materials in the syntheses of a variety of compounds. The employment of a nucleoside as the hydrogen active component has been one of the most common variants of the Mannich reaction. Treatment of uracil (or 2-thiouracil) nucleosides 1 with aq formaldehyde and a secondary amine (i.e., dimethylamine [49,50], diethylamine [51,52], N-methylbenzylamine [49], pyrrolidine [53,54], or piperidine [55,56]) at temperatures ranging from 60 °C to 100 °C afforded the corresponding 5-(alkylaminomethyl)pyrimidine nucleosides 2 (Scheme 2). Compounds 2 served as precursors to a variety of compounds. The transformations leading to thymidine or its derivatives 3 involved: (a) the metal-catalyzed hydrogenolysis of products 2 [51,52,54,55] (or their 5-(4-tolylthio)methyl derivatives [57]), or (b) the reduction of methylammonium iodides derived from compounds 2 with sodium borohydride [53]. Compounds 4 were achieved by treatment of the corresponding methylammo-nium iodides with an organic nucleophile [56,[58][59][60]. As studies on the synthesis of 5-taurinomethyluridine showed [60], this two-step procedure was much more efficient than a direct Mannich reaction involving taurine, formaldehyde and 2',3'-Oisopropylideneuridine [61]. Watanabe et al. described the synthesis of 7-(morpholinomethyl)tubercidin 5 by heating tubercidin, 37% aq formaldehyde and morpholine at 90 °C overnight (Scheme 3) [62]. Compound 5 was converted into the natural nucleoside toyocamycin in five steps. efficient conversion of compounds 6 to the 7-(morpholinomethyl) derivatives 7 required the use of acetic acid as a co-solvent. However, in the case of 7-deaza-2'-deoxyguanosine (8) the regioselectivity of the reaction changed from the C-7 to the C-8 position of the 7-deazapurine system (Scheme 4). The formation of product 9 could be explained by the influence of the electron-donating properties of the C-2 amino group stabilizing the σ-complex formed during the electrophilic attack at the C-8 carbon atom. Since the attempted acylation of the guanine amino group of 8 did not succeeded in the formation of the C-7-substituted guanosine 10, the compound was obtained in three steps from derivative 7b by conventional protectinggroup manipulations (Scheme 4).
The use of 3'-ethynylnucleoside 11 as the alkyne-derived hydrogen active component was described by Dauvergne et al. (Scheme 5) [64]. Treatment of compound 11 with paraformaldehyde and diisopropylamine in the presence of cuprous bromide in refluxing THF afforded the Mannich base 12 in 81% yield. The deprotection of compound 12 with tetrabutyl- Examples of the Mannich reaction employing a nucleoside as the aldehyde-bearing component are rather limited. Zhang et al. obtained a series of pyrimidine nucleoside-thazolidinone hybrids 15 from 5-formyl-3',5'-di-O-acetyl-2'-deoxyuridine (14), an arylamine and mercaptoacetic acid (Scheme 6) [65]. The reactions were performed in a ionic liquid ([bmim]PF 6 ). Products 15 were obtained in good to moderate yields. Antiparasitic activities of the hybrid compounds 15 were evaluated; some of them showed moderate activities against trypomastigote forms of Trypanosoma brucei brucei GVR 35 (e.g., IC 50 = 25 µM for Ar = C 6 H 4 -Cl-4).
The Mannich reaction was also used to construct nucleoside scaffolds from non-nucleoside substrates (Schemes 7-9). Filichev et al. used pyrrolidine 16, paraformaldehyde and uracil for the preparation of the Mannich base 17, which is considered as an 1'-aza-analog of pseudouridine (Scheme 7) [66]. Information on application of compound 17 was not given.
Compounds 25 and 26, prepared by Chen et al. [71], can be considered as analogs of reversed nucleosides [72] with the thiazolidin-4-one mimic of a nucleobase (Scheme 10). The compounds were obtained from condensation of aminosugar 24, arylaldehydes and mercaptoacetic acid in the presence of DMAP and DCC at room temperature. The reaction proceeded  with almost no stereoselectivity for the majority of these aldehydes, i.e., two diasteroisomers were isolated in ratios from 0.8 to 1.35. A modest stereoselectivity was observed in the case of 2-chlorobenzaldehyde with the 25a:26a ratio of 3.73. Com-pounds 25a and 25b, in contrast to their isomers 26, showed moderate activity against human cervical cancer cells at the concentration of 100 µM. Recently, the same group has developed the synthesis of D-glucopyranose-derived counterparts of compounds 25 and 26 [73]. The formation of an intermediate imine from a sugar azide and an aldehyde by Staudinger/aza-Wittig reaction was the key step of the synthesis.

The Kabachnik-Fields reaction
The Kabachnik-Fields reaction (Scheme 11) proceeds in a three-component system involving a carbonyl compound (aldehyde or ketone), amine, and a hydrophosphoryl compound (mainly alkyl/aryl phosphite) [74,75]. The reaction products, commonly termed as α-aminophoshonates, display properties of industrial and/or medical interest.
An example of the application of the Kabachnik-Fields reaction in nucleoside chemistry represents the preparation of α-arylaminophosphonates 28 and 29 by Zhang et al.
(Scheme 12) [76]. The reactions between 5-formyl-2'-deoxyuridine 27 (or its 3',5'-di-O-acetyl derivative 14), an aniline and dimethyl phosphite were carried out under solvent-free conditions at 60 °C (for 14) or at 80 °C (for 27). Products 28 and 29 were obtained in good to excellent yields as 1:1 diastereoisomeric mixtures arising from the generation of a stereogenic center at the aminophosphonate chain. The mixtures were not separated. Activity of hybrid compounds 28 and 29 against VZV and CMV viruses, as well as against Leishmania donovani promastigotes, was evaluated. Unfortunately, none of them showed any activity up to 250 μM.

The Ugi reaction
The Ugi reaction allows for a facile synthesis of a bisamide from a ketone (or an aldehyde), an amine, an isocyanide, and a carboxylic acid (Scheme 13) [77,78]. The Ugi MCRs involving a nucleoside as the substrate bearing the formyl, amino, or isocyano group have been reported.
The four-component Ugi reaction employing 3',5'-di-O-acetyl-5-formyl-2'-deoxyuridine (14) as the key substrate afforded nucleosides 30 bearing a N-acyl α-amino acid amide moiety at the uracil C-5 carbon atom (Scheme 14) [79]. The variant of the reaction with trimethylsilyl azide (TMS-N 3 ) in place of the carboxylic acid gave the tetrazole-substituted nucleosides 31 [79]. Products 30 and 31 were obtained as 1:1 diastereoisomeric mixtures owing to the formation of the new stereogenic center at the amino acid residue. In most cases, the diastereoisomeric mixtures of compounds 30 were separated through column chromatography due to the large differences in the polarity of the diastereoisomers. Anti-leishmanial activity of  Boehm and Kingsbury reported a facile synthesis of N-methylated di-and tri-peptide polyoxins by the Ugi reaction (Scheme 16) [81]. The aldehyde 36, aq methylamine, racemic isonitrile 37, and (S)-N-(benzyloxycarbonyl)phenylalanine were combined in MeOH to produce 38 as a mixture of four possible diastereoisomers in a total yield of 45%. The cyclohexylidene protecting group was then removed in refluxing aq AcOH. The resulting diastereoisomers 39 were separated by reversed phase HPLC to yield two pure isomers and the remaining two as an inseparable 1:1 mixture. These were further deprotected by hydrogenolysis under the hydrogen transfer conditions using the Pd black-formic acid system. Only one of the two pure isomers 40 was found to bind to chitin synthase.
Another approach to the solid-phase synthesis of nucleoside analogs was developed by Sun and Lee (Scheme 19) [85]. The library of 1344 compounds 49 was obtained for antibacterial screening. In this report, 5'-azidothymidine or 5'-azido-2'deoxyuridine was linked to a polystyrene butyldiethylsilane resin and subsequently reduced to the polymer-supported thymidinyl (R = CH 3 ) or 2′-deoxyuridinyl (R = H) aminonucleoside 47. The library synthesis was executed in 96-well plates, with one of the two amines 47, 12 carboxylic acids, 8 aldehydes, and an isocyanide per plate. The products 49 were cleaved from the support with HF/pyridine in THF. As expected, the Ugi products 49 were obtained as ca. 1:1 mixtures of diasteroisomers (based on HPLC and 1 H NMR analysis). Members of this library were claimed to show promising biological activity, however details were not given.
Muraymycins (MRYs) are a class of naturally occurring nucleoside-lipopeptide antibiotics with excellent antibacterial activity. Matsuda and coworkers envisaged that MRYs complex molec-Scheme 20: Reagents and reaction conditions: i. MeOH, 76%; ii. 80% aq TFA, 100%. ular structure could be efficiently assembled with the help of the Ugi reaction as the key step at the end of their synthesis. This approach was first exercised with a ring-opened muraymycin D2 analogue (Scheme 20) [86]. The reaction of carboxylic acid 50, 2,4-dimethoxybenzylamine, isovaleraldehyde, and isonitrile-substituted nucleoside 51 in methanol yielded the desired product as a 1:1 mixture of diastereoisomers, which were fully deprotected using aq TFA to furnish the muraymycin analogue 52.
This successful route to the MRYs was then applied in the total synthesis of muraymycin D2 and its epimer (Scheme 21) [87]. After completion of the synthesis of the urea dipeptide 53 bearing the cyclic moiety found in muraymycin D2, the fourcomponent condensation was performed similarly as in [86] to yield the protected product 54 as a 1:1 diastereomeric mixture. Functional group manipulation and HPLC separation completed the total synthesis. This approach was further developed in the synthesis of a number of MRY analogues in the following paper from the same research group [88].
More recently, the Ugi reaction was applied at a late stage of the synthesis of 3′-hydroxypacidamycin D (Scheme 22) [89]. The urea dipeptide 55, 2,4-dimethoxybenzylamine, the protected (S)-2-(methylamino)propanal, and isonitrile 56 were simply combined in ethanol at ambient temperature for 48 h. The expected compound 57 and its epimer were obtained in rea-sonable yields, and were separated by column chromatography. The syntheses of 3′-hydroxypacidamycin D and its epimer were then accomplished in four steps from intermediates 57 or epi-57, including selective deprotection of the N-methyl-Boc group, coupling with N-Boc-L-alanine, and global deprotection. This strategy was also applicable to the synthesis of a considerable number of pacidamycin analogues.

The Biginelli reaction
The Biginelli reaction (Scheme 26) consists in the three-component condensation of a 1,3-dicarbonyl compound, an aldehyde, The Dondoni group developed Lewis acid-promoted reactions employing the sugar derivatives 77 acting as: the component bearing the urea function (77a), the aldehyde function (77b), or the β-ketoester function (77c) (Scheme 29) [101,102]. In contrast to the N-1-substituted homo-C-nucleosides 78, the C-4 or C-6-substituted C-nucleosides (i.e., compounds 79 or 80, respectively) were obtained with the diastereoisomeric excess varied from 33% to 50%. The diastereoisomers were separated and their absolute configuration was determined using X-ray crystallography and circular dichroism spectroscopy. The stereochemical outcome of the synthesis of compounds 79 and 80 was suggested to result from some internal asymmetric induction of the chiral residue of the sugar aldehyde 77b or the sugar β-ketoester 77c, respectively. The debenzylated forms of C-nucleosides 78, 79 and 80 (as single diastereoisomers) were evaluated in vitro and in vivo as antimitotic agents [41]. They appeared to be less active than the reference (4S)-monastrol. Pyranose-derived nucleoside analogs were also prepared by these methods [101,102].
Sharma et al. used 2,4,6-trichloro [1,3,5]triazine (TCT) as the source of hydrogen chloride to promote the reactions leading to C-4-substituted C-nucleosides 81 with the high (ca. 7:1) diastereoisomeric ratio (Scheme 30) [103]. The products were isolated as single diastereoisomers. Since the reactions conducted in the presence of molecular sieves (4Å) were unsuccessful, the authors suggested that traces of moisture present in the reaction system played the key role in the release of hydrogen chloride from TCT. A pyranose-derived nucleoside analog was also prepared by this method.

Very recently, Figueiredo et al. synthesized C-nucleosides 83
with the C-4 substituted 3,4-dihydropyrimidin-2(1H)-thione as a nucleobase (Scheme 31) [104]. The products were obtained as the C-4-(R) single diastereoisomers. The use of microwave irradiation allowed the authors to perform these reactions with ten times smaller volume of the solvent than that employed in the reactions carried out under conventional heating conditions. Compound 83b showed promising activity against acetylcholinesterase at a concentration of 100 µmol/L.
The Hantzsch reaction involving the sugar-derived aldehydes 84, ethyl acetoacetate and ammonium acetate was applied by Sharma et al. in the synthesis of nucleoside analogs 85, bearing the 1,4-DHP nucleobase at the C-4-or C-1 carbon atom of the sugar (Scheme 33) [107]. Analogously to the previously reported Biginelli reaction [103], compounds 85 were obtained in high yields under the TCT-catalysis conditions. A pyranosederived nucleoside analog was also prepared by this method.
Using compound 87 as an example (Scheme 34), the Dondoni group demonstrated that the C-nucleosides with the C-4-substituted 1,4-DHP nucleobase can be efficiently obtained from the three-component reaction between the sugar aldehyde 86, ethyl acetoacetate, and ethyl 3-aminocrotonate [108,109]. The course of the reaction in the presence of various additives was examined in detail. The best results were obtained in the presence of 4 Å molecular sieves. The analysis of the reaction products showed that ytterbium triflate induced partial 1,2-elimination of benzyl alcohol from the ribosyl residue of the starting aldehyde 86, consequently leading to the 1',2'-didehydro-derivative of the target product 87. Pyranose-derived nucleoside analogs were also prepared by this method [108,109].
This approach involving a sugar aldehyde, 3-oxoester, and an ester of 3-aminocrotonic acid was then extended by the Dondoni group to 2,5-deoxyhexose-derived aldehydes 88 (Scheme 35) [110]. The best results were obtained when the reaction was performed with an excess (1.5 equiv) of methyl acetoacetate and methyl 3-aminocrotonate under L-proline- catalyzed conditions. In contrast to other catalysts tested (ytterbium triflate, D-proline, (S)-5-(pyrrolidin-2-yl)-1H-tetrazole, or (S)-1-(pyrrolidin-2-ylmethyl)pyrrolidine/TFA system), the catalytic effect of L-proline resulted in an increase in the reaction yield. Moreover, epimerization on the C-1 carbon atom of the starting aldehyde 88 was also suppressed. The latter effect was attributed to the preferential activation of methyl 3-aminocrotonate by L-proline via the corresponding enamine as compared to the activation of the sugar aldehyde.
The preliminary studies of the Dondoni group on the synthesis of C-nucleosides bearing the unsymmetrical 1,4-dihydropyridine nucleobase showed that the internal asymmetric induction by the sugar moiety played a crucial role in the formation of compounds 91 (Scheme 36) [110]. the same conditions. The absolute configuration of the C-4 carbon atom of compound 91 or ent-91 was not determined.
The approach involving an enamine (i.e., compound 92) as one of the reaction components was also used by Tewari et al. for the preparation of C-nucleosides 93 (Scheme 37) [111]. The reactions were carried out in the presence of tetrabutylammonium hydrogen sulfate as a phase-transfer catalyst. The yield of products 93 varied from 90% to 98%. As the authors suggested on the basis of comparative experiments performed without the catalyst, tetrabutylammonium hydrogen sulfate facilitated dehydration and cyclization steps of the reaction owing to its acidic properties. The reaction variant involving the corresponding sugar aldehyde 75, 4-aminopent-3-en-2-one and ethyl 3-oxobutanoate allowed to obtain unsymmetrical products 93. Galactose-6'-aldehyde-derived counterparts of the symmetrical nucleosides 93 were also prepared by this method.

The carbopalladation of dienes
A reaction of an aryl halide, an unsaturated alkene (diene, allene), and an amine catalyzed by Pd(0) species, referred to as carbopalladation of dienes, results in the three-component assembly of an unsaturated amine (Scheme 38) [112].
The palladium-catalyzed reactions of 5-iodopyrimidines, various acyclic or cyclic dienes, and amines were optimized by Larock et al. [113]. Thus, coupling of 5-iodo-2'-deoxyuridine After an extensive search for optimal reaction conditions, the authors found that the best yields could be achieved in the presence of zinc salts, in particular with secondary amines. In some cases, protection of the hydroxy groups in 94a was also neces- sary. The reactions between 3',5'-di-O-acetyl-5-iodo-2'deoxyuridine (94b), long-chain 1,ω-dienes (e.g., deca-1,9-diene or tetradeca-1,13-diene) and morpholine afforded products as mixtures of regioisomers resulting from the addition of the nucleoside moiety to the C-1 or C-2 carbon atom of the C=C double bond.
The three-component reactions of nucleoside-derived (uridine or thymidine) allenes 97, a range of aryl iodides, and 1-adamantylamine was accomplished smoothly under the palladium-catalyzed conditions (Scheme 40, the uridine example is shown) [114]. The coupling products 98 were obtained as (Z)stereoisomers for studies related to the drug discovery against the hepatitis C virus.
The methodology shown in Scheme 40 [114] was further elaborated on reactions of polyfunctional iodide 99 with four equiva-lents of nucleoside-derived allenes 97 (the uridine example shown), and a number of amines 100 (four equivalents, Scheme 41). The polyfunctional products 101 were obtained with excellent (Z)-stereoselectivity. The authors noticed a pronounced relationship between pK a of the amine and the isolated yield of the product, i.e., 1-adamantylamine provided the highest yield.

The Bucherer-Bergs reaction
The three-component Bucherer-Bergs reaction provides 5-mono-or 5,5-disubstituted hydantoins from the condensation of a carbonyl compound with potassium cyanide and ammonium carbonate (Scheme 42) [115]. The chemistry of hydantoins attracted a considerable attention because of their importance in medicine and industry [116,117]. N-Nucleoside analogs with (thio)hydatantoin scaffold as a nucleobase mimic were also extensively investigated [118].  cycloaddition of the ylide to N-methylmaleimide. The formation of compound 107 as the only product was rationalized using semi-empirical calculations. In the same contribution, the cascade reactions starting from uracil polyoxin C 106 were described (Scheme 44). Decarboxylative formation of azomethine ylides from 106 and an aldehyde (or ketone), followed by reaction of the ylide with maleimide afforded mixtures of cycloadducts 108 and 109 in molar ratios varied from 1:1 to 12:1. Compounds 108 were inactive against Aspergillus fumigatus or Candida albicans at concentration of 125 µg/mL. The Montmorillonite K-10 clay-microwave irradiation reaction system was also used by Yadav [126]. This method does not strictly comply with the Ugi's definition of MCRs because of the sequential addition of the substrates. However, in our opinion the method is worth noting since it represents an interesting extension of the Vorbrüggen N-glycosylation process. Thus, the reaction sequence leading to nucleosides 119 was initiated by the titanium(IV) chloride-promoted alkylation of 2,3-dihydrofurane 117 with ethyl pyruvate at −78 °C (1 hour), followed by the coupling of the resulting oxocarbenium ion with the silylated nucleobase 118. Compounds 119 were obtained as single diastereoisomers. The similar (not shown) reaction employing the silylated thymine and ethyl glyoxalate gave the corresponding product as 1:1 mixture of isomers at the C-2'a carbon atom.

Conclusion
In this comprehensive review application of multicomponent reactions (MCRs) in nucleoside chemistry has been presented.
In recent years, growing interest in the construction of novel nucleoside scaffolds by MCR has been observed. This conclusion is supported by the fact that 23 out of 60 original works cited in this review appeared within the last five years. Up to date, much more efforts were devoted to the preparation of novel nucleoside scaffolds by a structural modification of the parent nucleosides (37 examples) than by their de novo construction from non-nucleoside substrates (23 examples). A majority of the reported modifications of the parent nucleosides concerned their nucleobase moiety (27 examples). However, the number of reports on modifications of the purine nucleobase was limited (4 examples). Among reports on the de novo construction of nucleosides from non-nucleoside substrates, the ones dealing with the construction of a non-natural nucleobase predominated (18 examples). Interestingly, a combinatorial solid-phase approach has not been extensively exploited (2 examples). The findings concerning the syntheses of nucleoside antibiotic analogs or 1'-aza-analogs of immucilins are interesting in view of both organic synthesis and potential applications. The trends of a great research potential in this field could be identified from the presented literature survey. The most recent reports were mainly directed to: (i) the employment of novel reaction techniques, such as microwave irradiation, ionic liquids or inorganic supports, or (ii) the development of novel MCRs leading to nucleoside analogs bearing an unconventional nucleobase. As reports dealing with these issues revealed, a combination of both these trends may result in the preparation of structurally interesting compounds. An intensification of studies on the structure-activity relationship of these compounds would provide valuable data on their potential applications. We hope that continued efforts in this field will result in novel nucleoside drug candidates.