Fluorination of Naturally Occurring N6-Benzyladenosine Remarkably Increased Its Antiviral Activity and Selectivity

Recently, we demonstrated that the natural cytokinin nucleosides N6-isopentenyladenosine (iPR) and N6-benzyladenosine (BAPR) exert a potent and selective antiviral effect on the replication of human enterovirus 71. In order to further characterize the antiviral profile of this class of compounds, we generated a series of fluorinated derivatives of BAPR and evaluated their activity on the replication of human enterovirus 71 in a cytopathic effect (CPE) reduction assay. The monofluorination of the BAPR-phenyl group changed the selectivity index (SI) slightly because of the concomitant high cell toxicity. Interestingly, the incorporation of a second fluorine atom resulted in a dramatic improvement of selectivity. Moreover, N6-trifluoromethylbenzyladenosine derivatives (9–11) exhibited also a very interesting profile, with low cytotoxicity observed. In particular, the analogue N6-(3-trifluoromethylbenzyl)-adenosine (10) with a four-fold gain in potency as compared to BAPR and the best SI in the class represents a promising candidate for further development.


Introduction
For many years, natural products (NPs) have been a leading source for the majority of the approved drugs, and their structures are a valuable source of inspiration for medicinal chemists [1]. Interestingly, only 36% of the new chemical entities discovered between 1981 and 2010 were developed without inspiration from a natural product [2].
Among natural products, the development of nucleosides is by far the most fruitful field of investigation. About one hundred drugs derive from nucleoside structures: the vast majority of them were developed as antiviral drugs, and a consistent proportion as antitumor drugs. Natural nucleosides are isolated from DNA, RNA, nucleotides, and coenzymes of various natural sources. Nowadays, the nucleoside library consists of about 550 compounds, and is a promising pool for the development of new biologically active compounds [3][4][5].
N 6 -Modified purine nucleosides (cytokinin nucleosides) are an important group of biologically active natural compounds with a unique spectrum of biological activities [3]. Cytokinin nucleosides contain a hydrophilic ribofuranose moiety and a purine heterocyclic scaffold modified with a hydrophobic residue at the N 6 position. tRNA contains N 6 -isopentenyladenosine and some related nucleosides [6,7]. N 6 -Substituted adenosines are naturally present in plants [8][9][10] and bacteria [11].
In 2008, Arita and co-workers found that N 6 -benzyladenosine (BAPR) exhibited a pronounced antiviral activity against the replication of human enterovirus 71 (EV71) [12]. EV71 is a non-enveloped, single-stranded, positive-sense RNA virus belonging to the Enterovirus genus within the Picornaviridae family. EV71 commonly causes hand-, foot-, and mouth disease (HFMD), a mild and self-limiting illness mostly affecting children under the age of five. In some patients, EV71 has been associated with severe neurological complications including encephalitis, aseptic meningitis, and acute flaccid paralysis [13][14][15]. EV71 is prevalent worldwide, but most of the large outbreaks of neurotropic EV71 have occurred in the Pacific-Asia area [15][16][17]. However, in recent years, such epidemic tracts have been reported also in America and in Europe [15,17]. The World Health Organization has placed EV71 as one of the next biggest worldwide threats to public health, especially to young children, due to the lack of effective antiviral treatments [18,19].
Recently, we showed that, similarly to BAPR, two other naturally occurring plant cytokinin nucleosides, namely N 6 -isopentenyladenosine and N 6 -furfuryladenosine (kinetin riboside), possessed a potent and selective antiviral effect on EV71 [20,21]. Unfortunately, these compounds were rather cytotoxic, with CC 50 values in the low micromolar range (4-8 µM). We were able to improve the selectivity of this group of compounds by modifying the size and the nature of the linker. In particular, a modified BAPR with a two-to-three atom-long linker had a very pronounced antiviral activity, and a 50-fold improvement of the selectivity index (SI) as result of a lower cytotoxicity [21].
The introduction of fluorine in order to improve the pharmacological properties of a drug is a modern trend in medicinal chemistry. Currently, there are about 200 fluorinated drugs on the market (~20% of all pharmaceuticals), with even higher figures for agrochemicals (up to 30%) [22,23]. Therefore, in the present study, we report on the modification of natural BAPR by the substitution in the phenyl ring with fluoro-, difluoro-, and trifluorometyl groups to evaluate the eventual improvement in the antiviral profile of these fluorinated compounds in the context of EV71 replication ( Figure 1). contain a hydrophilic ribofuranose moiety and a purine heterocyclic scaffold modified with a hydrophobic residue at the N 6 position. tRNA contains N 6 -isopentenyladenosine and some related nucleosides [6,7]. N 6 -Substituted adenosines are naturally present in plants [8][9][10] and bacteria [11]. In 2008, Arita and co-workers found that N 6 -benzyladenosine (BAPR) exhibited a pronounced antiviral activity against the replication of human enterovirus 71 (EV71) [12]. EV71 is a nonenveloped, single-stranded, positive-sense RNA virus belonging to the Enterovirus genus within the Picornaviridae family. EV71 commonly causes hand-, foot-, and mouth disease (HFMD), a mild and self-limiting illness mostly affecting children under the age of five. In some patients, EV71 has been associated with severe neurological complications including encephalitis, aseptic meningitis, and acute flaccid paralysis [13][14][15]. EV71 is prevalent worldwide, but most of the large outbreaks of neurotropic EV71 have occurred in the Pacific-Asia area [15][16][17]. However, in recent years, such epidemic tracts have been reported also in America and in Europe [15,17]. The World Health Organization has placed EV71 as one of the next biggest worldwide threats to public health, especially to young children, due to the lack of effective antiviral treatments [18,19].
Recently, we showed that, similarly to BAPR, two other naturally occurring plant cytokinin nucleosides, namely N 6 -isopentenyladenosine and N 6 -furfuryladenosine (kinetin), possessed a potent and selective antiviral effect on EV71 [20,21]. Unfortunately, these compounds were rather cytotoxic, with CC50 values in the low micromolar range (4-8 μM). We were able to improve the selectivity of this group of compounds by modifying the size and the nature of the linker. In particular, a modified BAPR with a two-to-three atom-long linker had a very pronounced antiviral activity, and a 50-fold improvement of the selectivity index (SI) as result of a lower cytotoxicity [21].
The introduction of fluorine in order to improve the pharmacological properties of a drug is a modern trend in medicinal chemistry. Currently, there are about 200 fluorinated drugs on the market (~20% of all pharmaceuticals), with even higher figures for agrochemicals (up to 30%) [22,23]. Therefore, in the present study, we report on the modification of natural BAPR by the substitution in the phenyl ring with fluoro-, difluoro-, and trifluorometyl groups to evaluate the eventual improvement in the antiviral profile of these fluorinated compounds in the context of EV71 replication ( Figure 1).

Chemistry
Recently, we have developed a new useful and versatile approach for the preparation of N 6modified adenosine derivatives by the regioselective N 6 -alkylation of N 6 -acetyl-2′,3′,5′-tri-Oacetyladenosine with alcohols under Mitsunobu reaction conditions or with alkyl halides promoted by a base [20,21,24,25]. The main advantage of our method is the possibility to use both alkyl halides and alcohols for N 6 -modification. This is important, especially in the case when an amine is not stable

Chemistry
Recently, we have developed a new useful and versatile approach for the preparation of N 6 -modified adenosine derivatives by the regioselective N 6 -alkylation of N 6 -acetyl-2 ,3 ,5 -tri-Oacetyladenosine with alcohols under Mitsunobu reaction conditions or with alkyl halides promoted by a base [20,21,24,25]. The main advantage of our method is the possibility to use both alkyl halides and alcohols for N 6 -modification. This is important, especially in the case when an amine is not stable or is hardly available. Using this methodology, several hundred N 6 -substituted adenosines have been synthesized in one of our laboratories.
To simplify the separation procedure, we used 2 ,3 ,5 -tri-O-acetyl-6-chloropurine riboside directly in the substitution reactions. The acetyl groups are completely preserved in the reaction with aniline, and the protected intermediate can be isolated by silica gel chromatography and characterized. After the removal of the acetyl groups by ammonolysis, N 6 -phenyladenosine was obtained in overall high yield [21]. On the other hand, the reaction of 2 ,3 ,5 -tri-O-acetyl-6-chloropurine riboside with benzylamines was accompanied by the formation of by-products due to the partial removal of the protective groups, which complicated the chromatographic control of the reaction, and required a large excess of amines for the full conversion of the starting compound. Therefore, we decided to study the stability of different O-acyl groups to select the one optimal for our purposes. The results of the O-deacylation experiments of 2 ,3 ,5 -tri-O-acylinosine are summarized in Table 1. or is hardly available. Using this methodology, several hundred N 6 -substituted adenosines have been synthesized in one of our laboratories. The traditional approach for the preparation of N 6 -alkylated or N 6 -arylated adenosines is the substitution of the chlorine atom in commercially available 6-chloropurine riboside with alkyl-or arylamines [26,27]. 6-hloropurine riboside can be readily prepared by the deacetylation of 2′,3′,5′-tri-O-acetyl-6-chloropurine riboside [28].
To simplify the separation procedure, we used 2′,3′,5′-tri-O-acetyl-6-chloropurine riboside directly in the substitution reactions. The acetyl groups are completely preserved in the reaction with aniline, and the protected intermediate can be isolated by silica gel chromatography and characterized. After the removal of the acetyl groups by ammonolysis, N 6 -phenyladenosine was obtained in overall high yield [21]. On the other hand, the reaction of 2′,3′,5′-tri-O-acetyl-6chloropurine riboside with benzylamines was accompanied by the formation of by-products due to the partial removal of the protective groups, which complicated the chromatographic control of the reaction, and required a large excess of amines for the full conversion of the starting compound. Therefore, we decided to study the stability of different O-acyl groups to select the one optimal for our purposes. The results of the O-deacylation experiments of 2′,3′,5′-tri-O-acylinosine are summarized in Table 1. According to the data in Table 1, the acetyl group is rather labile under basic conditions, and the benzoyl group is the most stable. The properties of the iso-butyroyl group exhibited the best behavior for our aims, since it is fairly resistant to the action of benzylamines, and its synthesis is more straightforward than that of the benzoyl derivatives. The compound 2′,3′,5′-tri-O-isobutyroyl-6chloropurine riboside (4) has then been used as the starting substrate in the reactions with a small excess of benzylamines with fluoro-and trifluoromethyl groups (Scheme 1). The protective groups were removed in the presence of MeNH2/EtOH at room temperature with the subsequent chromatographic purification of the resulting products. Compounds 5-11 were obtained in overall good yield (50-98%). It should be mentioned that some of these compounds were previously prepared starting from 6chloropurine riboside [26]. or is hardly available. Using this methodology, several hundred N 6 -substituted adenosines have been synthesized in one of our laboratories. The traditional approach for the preparation of N 6 -alkylated or N 6 -arylated adenosines is the substitution of the chlorine atom in commercially available 6-chloropurine riboside with alkyl-or arylamines [26,27]. 6-hloropurine riboside can be readily prepared by the deacetylation of 2′,3′,5′-tri-O-acetyl-6-chloropurine riboside [28].
To simplify the separation procedure, we used 2′,3′,5′-tri-O-acetyl-6-chloropurine riboside directly in the substitution reactions. The acetyl groups are completely preserved in the reaction with aniline, and the protected intermediate can be isolated by silica gel chromatography and characterized. After the removal of the acetyl groups by ammonolysis, N 6 -phenyladenosine was obtained in overall high yield [21]. On the other hand, the reaction of 2′,3′,5′-tri-O-acetyl-6chloropurine riboside with benzylamines was accompanied by the formation of by-products due to the partial removal of the protective groups, which complicated the chromatographic control of the reaction, and required a large excess of amines for the full conversion of the starting compound. Therefore, we decided to study the stability of different O-acyl groups to select the one optimal for our purposes. The results of the O-deacylation experiments of 2′,3′,5′-tri-O-acylinosine are summarized in Table 1. According to the data in Table 1, the acetyl group is rather labile under basic conditions, and the benzoyl group is the most stable. The properties of the iso-butyroyl group exhibited the best behavior for our aims, since it is fairly resistant to the action of benzylamines, and its synthesis is more straightforward than that of the benzoyl derivatives. The compound 2′,3′,5′-tri-O-isobutyroyl-6chloropurine riboside (4) has then been used as the starting substrate in the reactions with a small excess of benzylamines with fluoro-and trifluoromethyl groups (Scheme 1). The protective groups were removed in the presence of MeNH2/EtOH at room temperature with the subsequent chromatographic purification of the resulting products. Compounds 5-11 were obtained in overall good yield (50-98%). It should be mentioned that some of these compounds were previously prepared starting from 6chloropurine riboside [26]. or is hardly available. Using this methodology, several hundred N 6 -substituted adenosines have been synthesized in one of our laboratories. The traditional approach for the preparation of N 6 -alkylated or N 6 -arylated adenosines is the substitution of the chlorine atom in commercially available 6-chloropurine riboside with alkyl-or arylamines [26,27]. 6-hloropurine riboside can be readily prepared by the deacetylation of 2′,3′,5′-tri-O-acetyl-6-chloropurine riboside [28].
To simplify the separation procedure, we used 2′,3′,5′-tri-O-acetyl-6-chloropurine riboside directly in the substitution reactions. The acetyl groups are completely preserved in the reaction with aniline, and the protected intermediate can be isolated by silica gel chromatography and characterized. After the removal of the acetyl groups by ammonolysis, N 6 -phenyladenosine was obtained in overall high yield [21]. On the other hand, the reaction of 2′,3′,5′-tri-O-acetyl-6chloropurine riboside with benzylamines was accompanied by the formation of by-products due to the partial removal of the protective groups, which complicated the chromatographic control of the reaction, and required a large excess of amines for the full conversion of the starting compound. Therefore, we decided to study the stability of different O-acyl groups to select the one optimal for our purposes. The results of the O-deacylation experiments of 2′,3′,5′-tri-O-acylinosine are summarized in Table 1. According to the data in Table 1, the acetyl group is rather labile under basic conditions, and the benzoyl group is the most stable. The properties of the iso-butyroyl group exhibited the best behavior for our aims, since it is fairly resistant to the action of benzylamines, and its synthesis is more straightforward than that of the benzoyl derivatives. The compound 2′,3′,5′-tri-O-isobutyroyl-6chloropurine riboside (4) has then been used as the starting substrate in the reactions with a small excess of benzylamines with fluoro-and trifluoromethyl groups (Scheme 1). The protective groups were removed in the presence of MeNH2/EtOH at room temperature with the subsequent chromatographic purification of the resulting products. Compounds 5-11 were obtained in overall good yield (50-98%). It should be mentioned that some of these compounds were previously prepared starting from 6chloropurine riboside [26].
a The reagent was used in at least 400-fold excess.
According to the data in Table 1, the acetyl group is rather labile under basic conditions, and the benzoyl group is the most stable. The properties of the iso-butyroyl group exhibited the best behavior for our aims, since it is fairly resistant to the action of benzylamines, and its synthesis is more straightforward than that of the benzoyl derivatives. The compound 2 ,3 ,5 -tri-O-isobutyroyl-6-chloropurine riboside (4) has then been used as the starting substrate in the reactions with a small excess of benzylamines with fluoro-and trifluoromethyl groups (Scheme 1). The protective groups were removed in the presence of MeNH 2 /EtOH at room temperature with the subsequent chromatographic purification of the resulting products. Compounds 5-11 were obtained in overall good yield (50-98%). It should be mentioned that some of these compounds were previously prepared starting from 6-chloropurine riboside [26].  Table 1).
The structure of the obtained compounds was confirmed by NMR and mass spectroscopy. The presence of fluorine atoms in the phenyl residue was confirmed by spin-spin coupling constants between 19 F and 1 H in 1 H-NMR spectra (JH-F) and between 19 F and 13 C in 13 C-NMR spectra (JC-F). The 1 H-NMR spectra of the fluorinated N 6 -benzyladenosine analogues (5)(6)(7)(8) in the low field region were complicated by the presence of 19 F-1 H couplings: 3 JH-F-8.0-9.0 Hz, 4 JH-F-6.7-5.5 Hz, and 5 JH-F ˂ 2.0 Hz. In the 13 C-NMR spectra, three types of coupling constants JC-F were present, which are characteristic of fluorinated aromatic compounds [29]: 1 JC-F-240-248 Hz, 2 JC-F-12-24 Hz, and 3 JC-F-7.5 Hz. The presence of trifluoromethyl residue in nucleosidic derivatives (9-11) was confirmed by low-intensive quartet with a coupling constant of ~30 Hz in 13 C-NMR spectra. This constant was consistent with the literature data for trifluoromethylated aromatic compounds [29]. Despite the majority of the synthesized compounds having been characterized by NMR earlier, their detailed analysis and the assignment of all chemical shifts and coupling constants has not been presented. Therefore, we provided in the Supplementary section a detailed NMR analysis for each analogue produced.

Biological Activity on EV71 and Structure-Activity Relationship (SAR)
We have shown earlier that three natural cytokinin nucleosides (compound 1-3) exerted a potent antiviral effect on the replication of EV71 with an EC50 of 0.3-1.4 μM, but exhibited also a rather high cytotoxicity [20,21] (Table 2). As previously mentioned, modifications of the N 6 -substituent (linker) of the BAPR scaffold led to a remarkable increase of selectivity [21]. Here, we produced a series of BAPR analogues to evaluate the effect of the fluorination of BAPR on the replication of EV71. A cytopathic effect (CPE) reduction assay was performed with the newly synthetized analogues (compounds 5-11) in rhabdomyosarcoma (RD) cells. BAPR, N 6 -isopenthenyladenosine, and N 6 -furfuryladenosine were included in the screening, and the toxicity of all of the aforementioned compounds was evaluated in parallel on treated-uninfected cells. (ii) MeNH 2 /EtOH, room temperature., 24 h, 50-98% (overall yields); (The structure of R is given in Table 1).
The structure of the obtained compounds was confirmed by NMR and mass spectroscopy. The presence of fluorine atoms in the phenyl residue was confirmed by spin-spin coupling constants between 19 F and 1 H in 1 H-NMR spectra (J H-F ) and between 19 F and 13 C in 13 C-NMR spectra (J C-F ). The 1 H-NMR spectra of the fluorinated N 6 -benzyladenosine analogues (5)(6)(7)(8) in the low field region were complicated by the presence of 19 F-1 H couplings: 3 J H-F -8.0-9.0 Hz, 4 J H-F -6.7-5.5 Hz, and 5 J H-F < 2.0 Hz. In the 13 C-NMR spectra, three types of coupling constants J C-F were present, which are characteristic of fluorinated aromatic compounds [29]: 1 J C-F -240-248 Hz, 2 J C-F -12-24 Hz, and 3 J C-F -7.5 Hz. The presence of trifluoromethyl residue in nucleosidic derivatives (9-11) was confirmed by low-intensive quartet with a coupling constant of~30 Hz in 13 C-NMR spectra. This constant was consistent with the literature data for trifluoromethylated aromatic compounds [29]. Despite the majority of the synthesized compounds having been characterized by NMR earlier, their detailed analysis and the assignment of all chemical shifts and coupling constants has not been presented. Therefore, we provided in the Supplementary section a detailed NMR analysis for each analogue produced.

Biological Activity on EV71 and Structure-Activity Relationship (SAR)
We have shown earlier that three natural cytokinin nucleosides (compound 1-3) exerted a potent antiviral effect on the replication of EV71 with an EC 50 of 0.3-1.4 µM, but exhibited also a rather high cytotoxicity [20,21] (Table 2). As previously mentioned, modifications of the N 6 -substituent (linker) of the BAPR scaffold led to a remarkable increase of selectivity [21]. Here, we produced a series of BAPR analogues to evaluate the effect of the fluorination of BAPR on the replication of EV71. A cytopathic effect (CPE) reduction assay was performed with the newly synthetized analogues (compounds 5-11) in rhabdomyosarcoma (RD) cells. BAPR, N 6 -isopenthenyladenosine, and N 6 -furfuryladenosine were included in the screening, and the toxicity of all of the aforementioned compounds was evaluated in parallel on treated-uninfected cells.  Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50  Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50  Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50 6.0 ± 0. 6 1.0 ± 0.  Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50  Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50  Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50   Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50 2.7 ± 0.9 0.14 ± 0.05 19 8 N 6 -(2,6-difluorobenzyl) adenosine   Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50   Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50  Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50  Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC50 of 13.3 ± 3.7 μM for the monofluorinated analogue to a CC50 Overall, the incorporation of fluoro-and trifluoromethyl groups significantly improved the selectivity index of BAPR (Table 2). In particular, the monofluorination of the phenyl group (compounds 5-7) slightly changed the SI because of the concomitant cytotoxicity of such compounds. Surprisingly, the incorporation of a second fluorine atom resulted in a substantial improvement over the selectivity. In particular, compound 8 presented an EC 50 comparable to BAPR with a dramatic reduction of cell toxcity: from a CC 50 of 13.3 ± 3.7 µM for the monofluorinated analogue to a CC 50 higher than 250 µM for the difluorinate counterpart. We wanted also to assess the effect of a trifluoromethyl group on the phenyl ring of BAPR on EV71 replication. Compounds 9, 10 and 11 did not show any cytotoxicity at the highest concentration tested, and the analogue N 6 -(3-trifluoromethylbenzyl)-adenosine (compound 10) exhibited also a four-fold improvement in potency as compared to BAPR.
Previous reports showed that the halogenation (and, in particular, the addition of I or Cl atoms) on a BAPR scaffold increased its selectivity by reducing the cell toxicity in cancer cell lines [26,30]. In line with these findings, we observed that the gain in selectivity in our model was mostly due to a decreased cell toxicity. In particular, only the analogues containing two fluorine or a trifluoromethyl group dramatically improved the cytotoxicity. In spite of our interest in understanding compound-driven cell toxicity, addressing this question was beyond our scientific scope. Future works on the optimization of this class of analogues may shed light on the mechanism of action and their metabolization within an infected cell.
Altogether, our data revealed that the introduction of at least two fluorine atoms or a trifluoromethyl group on the phenyl ring of BAPR dramatically improved its selectivity by reducing the cytotoxicity, and in case of compound 10, also by increasing the potency.

General
The solvents and materials were reagent grade and were used without additional purification. Column chromatography was performed on silica gel (Kieselgel 60 Merck, Germany, 0.063-0.200 mm). TLC was performed on an Alugram SIL G/UV254 (Macherey-Nagel, Düren, Germany) with UV visualization. The melting points were determined with Electrothermal Melting Point Apparatus IA6301 and are uncorrected. The 1 H and 13 C (with complete proton decoupling) NMR spectra were recorded on a Bruker (Karlsruhe, Germany) AMX 400 NMR instrument at 303 K. The 1 H-NMR-spectra were recorded at 400 MHz and the 13 C-NMR-spectra at 100 MHz. The chemical shifts in ppm were measured relative to the residual solvent signals as internal standards (CDCl 3 , 1 H: 7.26 ppm, 13 C: 77.1 ppm; DMSO-d 6 , 1 H: 2.50 ppm, 13 C: 39.5 ppm). Spin-spin coupling constants (J) are given in Hz. The high resolution mass spectra (HRMS) were registered on a Bruker Daltonics (Manning Park, Billerica, MA, USA) micrOTOF-Q II instrument using electrospray ionization (ESI). The measurements were done in positive ion mode. Interface capillary voltage: 4500 V; mass range from m/z 50 to 3000; external calibration (Electrospray Calibrant Solution, Fluka); nebulizer pressure: 0.4 Bar; flow rate: 3 µL/min; dry gas: nitrogen (4 L/min); interface temperature: 200 • C. Samples were injected into the mass spectrometer chamber from the Agilent 1260 HPLC system equipped with an Agilent (Santa Clara, CA, USA) Poroshell 120 EC-C18 (3.0 × 50 mm; 2.7 µm) column: the flow rate was 200 µL/min; and the samples were injected from the acetonitrile-water (1:1) solution and eluted in a linear gradient of acetonitrile concentrations (50→100%).