Probing the Effects of Pyrimidine Functional Group Switches on Acyclic Fleximer Analogues for Antiviral Activity

Due to their ability to inhibit viral DNA or RNA replication, nucleoside analogues have been used for decades as potent antiviral therapeutics. However, one of the major limitations of nucleoside analogues is the development of antiviral resistance. In that regard, flexible nucleoside analogues known as “fleximers” have garnered attention over the years due to their ability to survey different amino acids in enzyme binding sites, thus overcoming the potential development of antiviral resistance. Acyclic fleximers have previously demonstrated antiviral activity against numerous viruses including Middle East Respiratory Syndrome coronavirus (MERS-CoV), Ebola virus (EBOV), and, most recently, flaviviruses such as Dengue (DENV) and Yellow Fever Virus (YFV). Due to these interesting results, a Structure Activity Relationship (SAR) study was pursued in order to analyze the effect of the pyrimidine functional group and acyl protecting group on antiviral activity, cytotoxicity, and conformation. The results of those studies are presented herein.

In parallel, a secondary series featuring a methoxy at the 2-position on the pyrimidine was developed in order to better analyze the effect of hydrogen bonding motifs on antiviral activity. This series was also tested against the aforementioned viruses; however, these analogues were not more active than compound 1. Furthermore, the addition of an acetate protecting group (1-Ac) led to a decrease in antiviral activity, but also a large decrease in cytotoxicity (unpublished results). In order to explore the effects of different hydrogen bond donor and acceptor groups on the fleximer pyrimidine ring, as well as the effect of various acetate protecting groups on antiviral activity, an SAR study was pursued. Herein, we report the synthesis and antiviral evaluation of series 1-5 ( Figure 3). Most of the recent studies within the Seley-Radtke lab have focused on the development of fleximer analogues for use in antiviral and anticancer therapeutics [11,17,20]. In particular, fleximer analogues based on the structure of the Food and Drug Administration (FDA)-approved drug Acyclovir have demonstrated great promise, with compound 1 demonstrating low micromolar activity against coronaviruses [17], filoviruses [20], and most recently, flaviviruses (  Most of the recent studies within the Seley-Radtke lab have focused on the development of fleximer analogues for use in antiviral and anticancer therapeutics [11,17,20]. In particular, fleximer analogues based on the structure of the Food and Drug Administration (FDA)-approved drug Acyclovir have demonstrated great promise, with compound 1 demonstrating low micromolar activity against coronaviruses [17], filoviruses [20], and most recently, flaviviruses ( Figure 2, unpublished results).  [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20].
In parallel, a secondary series featuring a methoxy at the 2-position on the pyrimidine was developed in order to better analyze the effect of hydrogen bonding motifs on antiviral activity. This series was also tested against the aforementioned viruses; however, these analogues were not more active than compound 1. Furthermore, the addition of an acetate protecting group (1-Ac) led to a decrease in antiviral activity, but also a large decrease in cytotoxicity (unpublished results). In order to explore the effects of different hydrogen bond donor and acceptor groups on the fleximer pyrimidine ring, as well as the effect of various acetate protecting groups on antiviral activity, an SAR study was pursued. Herein, we report the synthesis and antiviral evaluation of series 1-5 ( Figure 3). In parallel, a secondary series featuring a methoxy at the 2-position on the pyrimidine was developed in order to better analyze the effect of hydrogen bonding motifs on antiviral activity. This series was also tested against the aforementioned viruses; however, these analogues were not more active than compound 1. Furthermore, the addition of an acetate protecting group (1-Ac) led to a decrease in antiviral activity, but also a large decrease in cytotoxicity (unpublished results). In order to explore the effects of different hydrogen bond donor and acceptor groups on the fleximer pyrimidine ring, as well as the effect of various acetate protecting groups on antiviral activity, an SAR study was pursued. Herein, we report the synthesis and antiviral evaluation of series 1-5 (

Chemistry
Synthesis of these analogues began with modifying the acyclic sugar moiety 6 by the addition of different acyl protecting groups (Scheme 1). The synthesis of 6 and 7 was previously published by our group [17,20], and all intermediates were obtained in a similar fashion, with the addition of the appropriate anhydride followed by triethylamine and diisopropylamine (DIPEA) in anh. CH2Cl2 to give 8-10 as clear to pale yellow oils in high yields (78-99%).

Chemistry
Synthesis of these analogues began with modifying the acyclic sugar moiety 6 by the addition of different acyl protecting groups (Scheme 1). The synthesis of 6 and 7 was previously published by our group [17,20], and all intermediates were obtained in a similar fashion, with the addition of the appropriate anhydride followed by triethylamine and diisopropylamine (DIPEA) in anh. CH 2 Cl 2 to give 8-10 as clear to pale yellow oils in high yields (78-99%).

Chemistry
Synthesis of these analogues began with modifying the acyclic sugar moiety 6 by the addition of different acyl protecting groups (Scheme 1). The synthesis of 6 and 7 was previously published by our group [17,20], and all intermediates were obtained in a similar fashion, with the addition of the appropriate anhydride followed by triethylamine and diisopropylamine (DIPEA) in anh. CH2Cl2 to give 8-10 as clear to pale yellow oils in high yields (78-99%).
Once the 4-methoxypyrimidine series 3 was successfully synthesized, the 2-aminopyrimidine series 4 was pursued. Similar to the previous series, synthesis of series 5 started with commercially available 2-amino-5-iodopyrimidine (14) and bis(tributyltin) to give the stannyl pyrimidine 15 in good yields (86%, Scheme 4). Following purification of the stannyl pyrimidine, Stille palladium catalyzed cross-coupling methods were employed to realize compounds 4 through 4t BuAc in high yields compared to the previous series (>50% for all analogues compared to an average of 15% for series 1).
Finally, the 4-aminopyrimidine series 5 was originally pursued using similar methods, starting with commercially available 4-aminopyrimidine (16, Scheme 5). Using N-iodosuccinimide in glacial acetic acid, 4-amino-5-iodopyrimidine 17 was realized, followed by the addition of bis(tributyltin) to give the stannyl coupling partner 18. Subsequent coupling with 6, 7, and 8 gave very low yields (<10%) and gave rise to purifications issues, as it was difficult to isolate the pure product without residual tin impurities. Products 5 to 5-EtAc were also fairly soluble in water; thus, using a 1-M potassium fluoride (KF) wash to convert the soluble tin impurity to the insoluble SnBu3F was unsuccessful. Furthermore, the addition of the acyl protecting group had no effect on tin impurity removal, thus the deprotection of 5-Ac to give 5 with methanolic ammonia was ineffective as well. As such, a protecting group for the 4-aminopyrimidine was envisioned in attempts to make purification more facile and increase the overall yields.
First, protection of the 4-amino group was tried using two equivalents of di-tert-butyl dicarbonate to give a di-boc protected amine, which could easily be deprotected after coupling to the sugar using mild acidic conditions (Scheme 6).  Using N-iodosuccinimide in glacial acetic acid, 4-amino-5-iodopyrimidine 17 was realized, followed by the addition of bis(tributyltin) to give the stannyl coupling partner 18. Subsequent coupling with 6, 7, and 8 gave very low yields (<10%) and gave rise to purifications issues, as it was difficult to isolate the pure product without residual tin impurities. Products 5 to 5-EtAc were also fairly soluble in water; thus, using a 1-M potassium fluoride (KF) wash to convert the soluble tin impurity to the insoluble SnBu 3 F was unsuccessful. Furthermore, the addition of the acyl protecting group had no effect on tin impurity removal, thus the deprotection of 5-Ac to give 5 with methanolic ammonia was ineffective as well. As such, a protecting group for the 4-aminopyrimidine was envisioned in attempts to make purification more facile and increase the overall yields.
First, protection of the 4-amino group was tried using two equivalents of di-tert-butyl dicarbonate to give a di-boc protected amine, which could easily be deprotected after coupling to the sugar using mild acidic conditions (Scheme 6).
Protection of the 5-iodo intermediate 17 using di-tert-butyl dicarbonate was successful in producing the di-boc compound 19; however, subsequent attempts at installing the stannyl component (20) were unsuccessful. Furthermore, attempts at installing the boronic ester (21) for use in more mild Suzuki coupling conditions were also unsuccessful. Final attempts at employing the di-boc protecting group tried protecting the stannyl pyrimidine 18; however, this method also proved ineffective. As such, attention turned toward using an in situ protection of the amine group followed by Negishi conditions for the 5-i PrAc and 5-t BuAc compounds, as these conditions had previously been successful for other fleximers developed in our lab (Scheme 7) [22].
impurity to the insoluble SnBu3F was unsuccessful. Furthermore, the addition of the acyl protecting group had no effect on tin impurity removal, thus the deprotection of 5-Ac to give 5 with methanolic ammonia was ineffective as well. As such, a protecting group for the 4-aminopyrimidine was envisioned in attempts to make purification more facile and increase the overall yields.
First, protection of the 4-amino group was tried using two equivalents of di-tert-butyl dicarbonate to give a di-boc protected amine, which could easily be deprotected after coupling to the sugar using mild acidic conditions (Scheme 6). Protection of the 5-iodo intermediate 17 using di-tert-butyl dicarbonate was successful in producing the di-boc compound 19; however, subsequent attempts at installing the stannyl component (20) were unsuccessful. Furthermore, attempts at installing the boronic ester (21) for use in more mild Suzuki coupling conditions were also unsuccessful. Final attempts at employing the diboc protecting group tried protecting the stannyl pyrimidine 18; however, this method also proved ineffective. As such, attention turned toward using an in situ protection of the amine group followed by Negishi conditions for the 5i PrAc and 5t BuAc compounds, as these conditions had previously been successful for other fleximers developed in our lab (Scheme 7) [22].
Compound 17 was first protected in situ with chlorotrimethylsilane to give intermediate 23, which was subsequently complexed with ZnCl2 to give intermediate 24. Dropwise addition of the zinc complex to either 9 or 10 gave 5i PrAc and 5t BuAc, respectively. While the coupling itself went well, later purification of these analogues proved difficult. However, the addition of 10% NH4OH to CH3OH in the chromatography gradient greatly decreased streaking and allowed for purification of these products (yields ~48%).

Discussion
As the different series were completed, they were sent for testing to collaborators at the Centers for Disease Control (CDC) for testing against EBOV. Briefly, a recombinant reporter virus, EBOV-Makona expressing ZsGreen (rEBOV/ZsG), was used to infect compound-treated Huh7 cells at a multiplicity of infection (MOI) of 0.3. Three days later, ZsGreen fluorescence was measured, and EC50 values were determined. Concurrently, the compound effects on cell viability were determined by measuring the cellular ATP content of compound-treated, mock-infected cells [20,23,24]. The results of this study are summarized in Table 1. None of the analogues tested demonstrated any cytotoxicity up to 100 µM, and while compound 1 remained the most active against EBOV, interesting trends emerged regarding the acyl protecting groups as well as the moieties on the pyridmine ring.
When the parent compounds of each series were tested, it was found that changing the hydrogen bond-donating amino group at the 2-poistion (compound 1) to a hydrogen bond-accepting methoxy Compound 17 was first protected in situ with chlorotrimethylsilane to give intermediate 23, which was subsequently complexed with ZnCl 2 to give intermediate 24. Dropwise addition of the zinc complex to either 9 or 10 gave 5-i PrAc and 5-t BuAc, respectively. While the coupling itself went well, later purification of these analogues proved difficult. However, the addition of 10% NH 4 OH to CH 3 OH in the chromatography gradient greatly decreased streaking and allowed for purification of these products (yields~48%).

Discussion
As the different series were completed, they were sent for testing to collaborators at the Centers for Disease Control (CDC) for testing against EBOV. Briefly, a recombinant reporter virus, EBOV-Makona expressing ZsGreen (rEBOV/ZsG), was used to infect compound-treated Huh7 cells at a multiplicity of infection (MOI) of 0.3. Three days later, ZsGreen fluorescence was measured, and EC50 values were determined. Concurrently, the compound effects on cell viability were determined by measuring the cellular ATP content of compound-treated, mock-infected cells [20,23,24]. The results of this study are summarized in Table 1. None of the analogues tested demonstrated any cytotoxicity up to 100 µM, and while compound 1 remained the most active against EBOV, interesting trends emerged regarding the acyl protecting groups as well as the moieties on the pyridmine ring.
When the parent compounds of each series were tested, it was found that changing the hydrogen bond-donating amino group at the 2-poistion (compound 1) to a hydrogen bond-accepting methoxy group (compound 2) decreased activity (1.49 µM compared to 23.44 µM, p = 0.0082). Furthermore, when there is a hydrogen atom at the 2-position (compound 3), a decrease in activity compared to compound 1 was also observed (1.49 µM compared to 78.88 µM, p = 0.0061). Compounds 4 and 5, which feature only an amino group in the 2-position and 4-position, respectively, did not display any activity against EBOV. Interestingly, activity could be recovered for series 4 and 5, as well as improved for series 3, by altering the acyl protecting group. For instance, the addition of the acetate protecting group with 3-Ac greatly increased activity compared to the parent analogue 3 (35.36 µM compared to 78.88 µM, p = 0.0162). However, the addition of the acyl protecting group seemed to affect each series differently, as compound 1-Ac was not active against EBOV. In contrast, the parent compound 1 was very active, but 4-Ac was active against EBOV (EC 50 = 52.89 µM), whereas the parent compound 4 was not active. For series 2, the activity of 2-Ac appeared to increase, but did not reach statistical significance (p = 0.7922).
For series 2 and 4, there was a loss in activity with the ethyl acyl (EtAc) analogues compared to analogues with only an acetate group, such as 2-Ac compared to 2-EtAc as well as 4-Ac compared to 4-EtAc. However, activity returned with the isopropyl acyl ( i PrAc) protected compounds. Finally, the tert-butyl acyl ( t BuAc) protected analogues were variable in activity across the series, though for the most part, these analogues were not significantly different in activity from the acetate (Ac). The adenosine series 5 were not active except for the t BuAc, which demonstrated weak activity. These results, along with the potent activity observed for compound 1, which also bears a 4-methoxy group, suggests that the 4-methoxy group is an important determinant for antiviral activity.
The difference in activity between the Ac, EtAc, and i PrAc is an interesting trend, as the EtAc analogues were consistently the least active analogues. In order to analyze if this trend was due to a change in lipophilicity or due to branching, the logP values were determined for each analogue ( Table 2).
LogP values are a common measure of lipophilicity, and the greater the value, the more lipophilic the compound. Not surprisingly, all the parent analogues (R group = H) were the least lipophilic of the series, with the amino series 4 and 5 demonstrating the lowest lipophillicity. On average, the difference between the parent analogues and the acetate analogues (R = Ac) was not substantial; however, the ethyl acyl analogues (R = EtAc) were substantially more lipophilic than the acetate analogues (for example, −0.14 compared to 0.52 for series 5). Not surprisingly, the increased branching also led to an increase in lipophilicity, as the t BuAc analogues were the most lipophilic of their series. While increased lipophillicity is typically associated with a better drug candidate, there is a limit where increased lipophillicity affects potency. For instance, the isopropyl acyl analogue 2-i PrAc demonstrated an EC 50 of 31.96 ± 9.12 µM against EBOV and a logP of 2.15, whereas 2-t BuAc has an EC 50 of 81.34 ± 10.15 µM and a logP of 2.85. Furthermore, the isopropyl acyl analogue 3-i PrAc demonstrates an EC 50 of 38.29 ± 28.15 µM and a logP of 1.76, whereas 3-t BuAc has an EC 50 of 51.40 ± 10.01 µM and a logP of 2.47. This suggests that compounds with logP values over 2.8 are too lipophilic, and thus are less active against EBOV. Interestingly, compound 4-i PrAc is less active than the more branched 4-t BuAc and is less lipophilic (logP = 0.88 vs 1.59), suggesting that compound with logP values greater than 1 may be more active. All together, these results suggest that lipophillicity and not branching affects antiviral activity.
Previous studies with the Seley-Radtke lab found that in solution, ribose fleximers preferred the typical anti-conformation found with nucleosides [21]. As such, studies were pursued in order to determine if the acyclic fleximer analogues also preferred anti-conformation in solution. 1 H-NMR spectra were recorded on 300-MHz NMR spectrometer at 3-mM concentration in 1% dimethylsulfoxide (DMSO)-d 6 /D 2 O.
Using well-resolved NMR signals of H6, H8, and H11 aromatic protons ( Figure 4A), an assessment of the conformational preferences along the C11-C10-C5-C6 torsion angle was possible through the use of 1D Nuclear Overhauser Effect Spectroscopy (NOESY) experiments. Previous studies with the Seley-Radtke lab found that in solution, ribose fleximers preferred the typical anti-conformation found with nucleosides [21]. As such, studies were pursued in order to determine if the acyclic fleximer analogues also preferred anti-conformation in solution. 1 H-NMR spectra were recorded on 300-MHz NMR spectrometer at 3-mM concentration in 1% dimethylsulfoxide (DMSO)-d6/D2O. Using well-resolved NMR signals of H6, H8, and H11 aromatic protons ( Figure 4A), an assessment of the conformational preferences along the C11-C10-C5-C6 torsion angle was possible through the use of 1D Nuclear Overhauser Effect Spectroscopy (NOESY) experiments.
Selective inversion of the H11 proton resulted in strong NOE enhancement of 4.3% at H1', whereas other NOE enhancements were relatively weak ( Figure 4B). The lack of strong NOE Selective inversion of the H11 proton resulted in strong NOE enhancement of 4.3% at H1', whereas other NOE enhancements were relatively weak ( Figure 4B). The lack of strong NOE enhancements between H11 and H6 suggests that anti-orientation along the C11-C10-C5-C6 torsion angle is the predominant conformation of compound 1 in solution. Additionally, inversion of the H6 proton showed no NOE enhancements ( Figure 4C), which confirms that H6 and H11 are spatially apart, overall supporting anti-conformation.
These studies also found that compounds 1-Ac, 2, and 2-Ac also favored anti-conformation along the C11-C10-C5-C6 torsion angle. Interestingly, selective inversion of the H11 proton of compound 5 resulted in relatively strong NOE enhancement of 4.6% at H1', as well as moderate NOE enhancement of 2.8% at H6 ( Figure 5B). These results suggest that syn orientation along the C11-C10-C5-C6 torsion angle is predominant. This conformation was also determined to be prominent in the acetate prodrug 5-Ac.  These results are particularly interesting due to the lack of antiviral activity seen with series 5 against EBOV. The difference in the conformations between compounds 1, 2, and 5 may offer a reason for the drastic differences in antiviral activity observed. The differences in orientation are particularly highlighted in Figure 6.
While compounds 1, 1-Ac, 2, 2-Ac, 5, and 5-Ac have been analyzed in these studies, it will be interesting to see if the 4-methoxy series 3 or the 2-amino series 4 also demonstrate different orientations compared to series 1 or 5. Due to the antiviral activity found with series 3 and 4, it is expected that these analogues also are predominately found in the anti-conformation.

Chemistry General Information
All reactions were performed using oven-dried glassware under a nitrogen atmosphere with magnetic stirring. Reagents were purchased from Sigma-Aldrich (Rockville, MD, USA), Alfa Aesar (Haverhill, MA, USA), and Combiblocks (San Diego, CA, USA). Solvents were either purchased as anhydrous or were dried using the MBRAUN solvent purification system (MB-SPS) (MBRAUN, Stratham, NH, USA). Reactions were monitored by thin layer chromatography (TLC) using EMD silica gel 60 F254 coated glass-backed TLC plates (Milliporesigma, Rockville, MD, USA) and These results are particularly interesting due to the lack of antiviral activity seen with series 5 against EBOV. The difference in the conformations between compounds 1, 2, and 5 may offer a reason for the drastic differences in antiviral activity observed. The differences in orientation are particularly highlighted in Figure 6.  These results are particularly interesting due to the lack of antiviral activity seen with series 5 against EBOV. The difference in the conformations between compounds 1, 2, and 5 may offer a reason for the drastic differences in antiviral activity observed. The differences in orientation are particularly highlighted in Figure 6.
While compounds 1, 1-Ac, 2, 2-Ac, 5, and 5-Ac have been analyzed in these studies, it will be interesting to see if the 4-methoxy series 3 or the 2-amino series 4 also demonstrate different orientations compared to series 1 or 5. Due to the antiviral activity found with series 3 and 4, it is expected that these analogues also are predominately found in the anti-conformation.

Chemistry General Information
All reactions were performed using oven-dried glassware under a nitrogen atmosphere with While compounds 1, 1-Ac, 2, 2-Ac, 5, and 5-Ac have been analyzed in these studies, it will be interesting to see if the 4-methoxy series 3 or the 2-amino series 4 also demonstrate different orientations compared to series 1 or 5. Due to the antiviral activity found with series 3 and 4, it is expected that these analogues also are predominately found in the anti-conformation.

Chemistry General Information
All reactions were performed using oven-dried glassware under a nitrogen atmosphere with magnetic stirring. Reagents were purchased from Sigma-Aldrich (

Conclusions
In this study, an SAR was pursued in order to increase antiviral activity against EBOV and decrease cytotoxicity. While none of the analogues analyzed were cytotoxic up to 100 µM, none of the analogues were more active against EBOV than compound 1. Regardless, it was determined that the 4-methoxy series were overall more active than the 2-amino series, suggesting that the 4-methoxy moiety may be more important for activity. Furthermore, compounds with logP values greater than 1 but less than 2.4 offer the necessary lipophilicity for activity. It was also found that compounds within series 1 and 2 adopt the typical anti-conformation, whereas compounds within series 5 adopt a syn conformation, thus offering another possible explanation for the differences in antiviral activity. With these initial results in hand, the synthesis of additional analogues, especially those with a methoxy group at the 4 position will be considered, in order to continue to try and further explore antiviral activity. The results of those studies will be reported elsewhere as they become available.