N -4 Alkyl Cytosine Derivatives Synthesis: A New Approach N -4 Alkyl Cytosine Derivatives Synthesis: A New Approach

: The selective N -4 alkylation of cytosine plays a critical role in the synthesis of biologically active molecules. This work focuses on the development of practical reaction conditions toward a regioselective synthesis of N -4-alkyl cytosine derivatives. The sequence includes a direct and selective sulfonylation at the N -1 site of the cytosine, followed by the alkylation of the amino site using KHMDS in CH 2 Cl 2 /THF mixture, providing a fast and efﬁcient approach consistent with pyrimidine-based drug design. Abstract: The selective N -4 alkylation of cytosine plays a critical role in the synthesis of biologically active molecules. This work focuses on the development of practical reaction conditions toward a regioselective synthesis of N -4-alkyl cytosine derivatives. The sequence includes a direct and selective sulfonylation at the N -1 site of the cytosine, followed by the alkylation of the amino site using KHMDS in CH 2 Cl 2 /THF mixture, providing a fast and efficient approach consistent with pyrimi-dine-based drug design.


Introduction
Integrase (IN) catalyzes the insertion of viral DNA [1] into the genome of infected cells and acts as a cofactor for reverse transcription [2].
In the context of HIV-1 infection, IN was successfully targeted for drug development [3]. Raltegravir (MK-0518) [4,5] was approved by the Food and Drug Administration in 2007, and other integrase inhibitors (INI), including Elvitegravir (GS-9137) [6,7], are progressing through clinical development [8]. The breakthrough of INI has produced a great impulse in the use of multiple drugs that act on different viral targets, known as Highly Active Antiretroviral Therapy (HAART) [9]. Important examples of this class are the lens epithelium-derived growth factor (LEDGF) inhibitors [10][11][12] (Figure 1).

Introduction
Integrase (IN) catalyzes the insertion of viral DNA [1] into the genome of infected cells and acts as a cofactor for reverse transcription [2].
In the context of HIV-1 infection, IN was successfully targeted for drug development [3]. Raltegravir (MK-0518) [4,5] was approved by the Food and Drug Administration in 2007, and other integrase inhibitors (INI), including Elvitegravir (GS-9137) [6,7], are progressing through clinical development [8]. The breakthrough of INI has produced a great impulse in the use of multiple drugs that act on different viral targets, known as Highly Active Antiretroviral Therapy (HAART) [9]. Important examples of this class are the lens epithelium-derived growth factor (LEDGF) inhibitors [10][11][12] (Figure 1).   Unfortunately, the development of resistance is a constant and inevitable threat to the application of therapies; there is always a need for new antiviral drugs with high activity and low cytotoxicity to assist and sometimes also substitute previously utilized drugs.
Reactions 2022, 3 193 Molecules acting on the IN HIV-1 are not immune to this problem [13]. This has prompted the research of more efficient and inexpensive new drugs. In this context is the design and synthesis of new cytosine-based antiretroviral (ARV) compounds, which are able to inhibit IN HIV-1.
Current studies of structure-activity relationships (SAR) on the above mentioned INI structures have identified two common regions [14]: a region with two metal-binding motifs critical to all members of this class of active site binders and a region with a hydrophobic site that requires a substituted benzyl group [15,16].
Taking into consideration these findings, we exploited the commercially available cytosine scaffold to synthetize new integrase strand transfer inhibitors (INSTIs) [1,3,4,17,18]. In detail, starting from a preliminary docking analysis [19], which clarified that chelation motif N-(aryl/alkyl sulfonyl) amide could selectively fill the binding site, we set out to investigate an original and efficient strategy for the synthesis of type 1 nucleobases ( Figure 2). Reactions 2022, 3, FOR PEER REVIEW 2 activity and low cytotoxicity to assist and sometimes also substitute previously utilized drugs.
Molecules acting on the IN HIV-1 are not immune to this problem [13]. This has prompted the research of more efficient and inexpensive new drugs. In this context is the design and synthesis of new cytosine-based antiretroviral (ARV) compounds, which are able to inhibit IN HIV-1.
Current studies of structure-activity relationships (SAR) on the above mentioned INI structures have identified two common regions [14]: a region with two metal-binding motifs critical to all members of this class of active site binders and a region with a hydrophobic site that requires a substituted benzyl group [15,16].
Likewise, the Borch reductive alkylation method [36,37] and the titanium (IV), which also catalyzed [35], required an excess of amine to favor the formation of the iminium intermediates, thereby hampering the dissolution in the solvent that was usually used.
The methodology described herein shows the regioselective formation of our new compounds under conditions consistent with the stability of future drug moieties.

Materials and Methods
All reagents (Aldrich, St. Louis, MA, USA and Merck, KGaA, Darmstadt, Germany) were acquired at the highest purity available and used without further purification. Thinlayer chromatographies were performed with silica gel plates Merck 60 F254, and the display of the products on TLC was performed with a lighting UV lamp, solutions of ninhydrin (0.2% in CH3OH mol), and molecular iodine. The column chromatographies were carried out using silica gel 70-230 mesh (Merck, KGaA, Darmstadt, Germany). Elemental analyses were performed on a FlashSmart V Elemental Analyzer (ThermoFisher Scientific, Waltham, MA, USA). The 1 H and 13 C NMR spectra were recorded on spectrometers: Cytosine derivatives are versatile intermediates in the synthesis of biologically and pharmaceutically active molecules [20][21][22][23][24][25][26][27] and are widely used as antineoplastic [6], antiviral [24], and anti-AIDS agents [27]. Some groups have very recently focused their attention on N-4 alkyl analogue, which improves uptake and bioavailability of gemcitabine, a worldwide chemotherapeutic cytidine analogue [20].
Likewise, the Borch reductive alkylation method [36,37] and the titanium (IV), which also catalyzed [35], required an excess of amine to favor the formation of the iminium intermediates, thereby hampering the dissolution in the solvent that was usually used.
The methodology described herein shows the regioselective formation of our new compounds under conditions consistent with the stability of future drug moieties.

Materials and Methods
All reagents (Aldrich, St. Louis, MA, USA and Merck, KGaA, Darmstadt, Germany) were acquired at the highest purity available and used without further purification. Thinlayer chromatographies were performed with silica gel plates Merck 60 F254, and the display of the products on TLC was performed with a lighting UV lamp, solutions of ninhydrin (0.2% in CH 3 OH mol), and molecular iodine. The column chromatographies were carried out using silica gel 70-230 mesh (Merck, KGaA, Darmstadt, Germany). Elemental analyses were performed on a FlashSmart V Elemental Analyzer (ThermoFisher Scientific, Waltham, MA, USA). The 1 H and 13 C NMR spectra were recorded on spectrometers: Bruker DRX (400 MHz) and Varian Inova Marker (500 MHz) in CDCl 3 solution unless otherwise specified. The chemical shifts are reported in ppm (δ) and the J in Hz.

Synthesis of 4-amino-1-((4-chlorophenyl)sulfonyl) pyrimidin-2(1H)-one (2)
Sodium hydride (118 mg; 4.9 mmol) at 0 • C under nitrogen atmosphere was added to a stirring solution of cytosine (500 mg, 4.5 mmol) in dry DMF (38 mL). After 2 h, 4chlorobenzenesulfonyl chloride (1.4 g, 6.8 mmol) was added and stirring was continued over a period of 30 min. The resulting solution was then allowed to warm to room temperature. After 1.5 h, the reaction was quenched with methanol (0.60 mL). The solvent was evaporated under reduced pressure, replaced with chloroform and washed with brine, and then dried (Na 2 SO 4 ). The evaporation of the solvent under reduced pressure gave a crude mixture that was purified by column chromatography (CHCl 3 /MeOH 95:5) to yield compound 2 (0.96 g, 75%). 1

General Procedure Synthesis of N-4 Alkyl Cytosine Derivatives
Derivative 2 (0.5 eq) was dissolved in dry CH 2 Cl 2 :THF (1:1, 5 mL), followed by the addition of 0.5 M KHMDS in THF (0.75 eq) at −40 • C under nitrogen atmosphere. After 1 h, electrophile (0.6 eq) was added and the reaction was allowed to warm to 5 • C within 24 h. TLC monitored the progress of the reaction. The mixture was then treated with methanol (0.5 mL) and further stirred for 10 min at rt. The solvent was evaporated under reduced pressure, replaced with ethyl acetate and washed with brine, and then dried (Na 2 SO 4 ). The evaporation of the solvent under reduced pressure gave a crude mixture that was purified by PLC (1:1 Hexane/Ethyl Acetate) to yield the pairs 1a-3a, 1b-3b, 1e-3e, and 1h-3h.

Results and Discussion
We started with the synthesis of the suitable precursor of our target, namely the cytosine sulfonylate derivate 2, obtained by exploiting the well-known good reactivity of the N-1 site [31,32,[40][41][42]. Indeed, as shown in Scheme 1, the commercially available cytosine was selectively sulfonylated with 4-chlorobenzene-1-sulfonyl chloride in DMF, the solvent required to overcome the known low solubility of the starting material. It is noteworthy that a temperature of 0 • C was mandatory to avoid a competitive reaction in favor of the exocyclic amine group. Under these conditions, compound 2 was obtained with 78% yield, as confirmed by NMR.

Results and Discussion
We started with the synthesis of the suitable precursor of our target, namely the cytosine sulfonylate derivate 2, obtained by exploiting the well-known good reactivity of the N-1 site [31,32,[40][41][42]. Indeed, as shown in Scheme 1, the commercially available cytosine was selectively sulfonylated with 4-chlorobenzene-1-sulfonyl chloride in DMF, the solvent required to overcome the known low solubility of the starting material. It is noteworthy that a temperature of 0 °C was mandatory to avoid a competitive reaction in favor of the exocyclic amine group. Under these conditions, compound 2 was obtained with 78% yield, as confirmed by NMR.

Scheme 1. N-4 alkyl model reaction.
Then, in our exploratory studies, we experimented with the representative N-4 alkylation (Scheme 1) with benzyl bromide as an electrophile under different conditions in terms of base, time, solvent, and temperature.
Firstly, the mixture of 2 and benzyl bromide was dissolved in DMSO and left at 25 °C for 4 h, then it was allowed to reach 80 °C and was kept under these conditions for a further 20 h, but no reaction took place and the starting materials remained completely unconsumed ( Table 1, entry 1). Next, the exploitation of non-nucleophilic bases was investigated. In detail, pyridine (Pyr) and triethylamine (TEA) were found to be ineffective ( Table 1, entries 2-4) and only a trace of the desired product was achieved when bicyclic amide (1,8-diazabiciclo(5.4.0)undec-7ene, (DBU)) [43,44]-which is able to form a charge transfer complex-was employed. The low nucleophilicity of the nitrogen atom, as well as the steric hindrance on the same nitrogen, resulting in a stalled reaction, could be clarified by the supposed complex reported in Figure 3. Then, in our exploratory studies, we experimented with the representative N-4 alkylation (Scheme 1) with benzyl bromide as an electrophile under different conditions in terms of base, time, solvent, and temperature.
Firstly, the mixture of 2 and benzyl bromide was dissolved in DMSO and left at 25 • C for 4 h, then it was allowed to reach 80 • C and was kept under these conditions for a further 20 h, but no reaction took place and the starting materials remained completely unconsumed ( Table 1, entry 1). Next, the exploitation of non-nucleophilic bases was investigated. In detail, pyridine (Pyr) and triethylamine (TEA) were found to be ineffective ( Table 1, entries 2-4) and only a trace of the desired product was achieved when bicyclic amide (1,8-diazabiciclo(5.4.0)undec-7ene, (DBU)) [43,44]-which is able to form a charge transfer complex-was employed. The low nucleophilicity of the nitrogen atom, as well as the steric hindrance on the same nitrogen, resulting in a stalled reaction, could be clarified by the supposed complex reported in Figure 3. 1 Reactions were performed using cytosine sulfonylate 2 (1 eq), bases (1.5 eq), and benzyl bromide (1.2 eq). 2 TLC monitored the progress of the reaction. When the concept of strongest base (lithium diisopropylamide (LDA), hexamethyldisilazane lithium (LiHMDS), and hexamethyldisilazane potassium (KHMDS)) was explored, positive results were produced.
Remarkably, the N-1 substituted cytosines that participated as an acidic compound (with pKa lower than that of KHMDS) reacted with the base. Thus, the formed anion of the substrate could act as a nucleophile in reaction with benzyl halides. In fact, as reported in Table 1, when using KHMDS in DMF at −40 °C (entry 10) the reaction was completed within 4 h and workup afforded the expected monobenzylated product 1a as the major compound with 40% yield, together with a minor side-product 3a (with 30% yield, Scheme 1). As is well known for enolates, our products increased the separation of the metal cation from the anion with the larger alkali metals, which leads to a more reactive but less stable anionic intermediate.
Attempts to optimize the reaction through modification of the ratio between 2 and BnBr proved unsuccessful, and we did not find any effects of the ratio between 2 and bases on the reaction in terms of yield.
Therefore, the promising approach of the protocol prompted us to evaluate the substrate scope. As shown in Table 2 (entries b-h), a wide range of benzyl bromides containing both electron-donating (EDG) and electron-withdrawing (EWG) substituents were well tolerated with good conversion. However, at this stage, the results are difficult to rationalize. In relation to entry i, the reactivity of bromomethane is definitely higher compared to that of primary alkyl bromides, and the fact that there is more than one nucleophilic center on the cytosine substrate results in byproduct formation that is not valuable. When the concept of strongest base (lithium diisopropylamide (LDA), hexamethyldisilazane lithium (LiHMDS), and hexamethyldisilazane potassium (KHMDS)) was explored, positive results were produced.
Remarkably, the N-1 substituted cytosines that participated as an acidic compound (with pKa lower than that of KHMDS) reacted with the base. Thus, the formed anion of the substrate could act as a nucleophile in reaction with benzyl halides. In fact, as reported in Table 1, when using KHMDS in DMF at −40 • C (entry 10) the reaction was completed within 4 h and workup afforded the expected monobenzylated product 1a as the major compound with 40% yield, together with a minor side-product 3a (with 30% yield, Scheme 1). As is well known for enolates, our products increased the separation of the metal cation from the anion with the larger alkali metals, which leads to a more reactive but less stable anionic intermediate.
Attempts to optimize the reaction through modification of the ratio between 2 and BnBr proved unsuccessful, and we did not find any effects of the ratio between 2 and bases on the reaction in terms of yield.
Therefore, the promising approach of the protocol prompted us to evaluate the substrate scope. As shown in Table 2 (entries b-h), a wide range of benzyl bromides containing both electron-donating (EDG) and electron-withdrawing (EWG) substituents were well tolerated with good conversion. However, at this stage, the results are difficult to rationalize. In relation to entry i, the reactivity of bromomethane is definitely higher compared to that of primary alkyl bromides, and the fact that there is more than one nucleophilic center on the cytosine substrate results in byproduct formation that is not valuable.
However, as in the model reaction, a mixture of two different N-alkylated products, namely 1 and 3, were obtained. The mono/di-alkylation ratio ranged from 6:4 to 7:3, as determined by the integration of characteristic protons for each product in the 1 HNMR spectra of the concentrated reaction mixtures.
A combination of homo-and heteronuclear 2D NMR experiments (DQF-COSY, 13 C-1 H HSQC, and HMBC, NOESY) were used to assign all the spin systems of 1a and 3a. In detail, the proton resonances of all systems were obtained by the COSY technique and were used to assign the carbon resonance in the HSQC spectra. The 13 C-1 H HMBC spectrum of 3a (see Supplementary Materials) shows a correlation between CH 2 at δ = 4.54 ppm and the nitrogen-bearing carbon C4 signal at δ = 164.1 ppm, as well as a comparable correlation between CH 2 at δ = 4.96 ppm and the same C4. However, the first CH 2 is also correlated to carbons at δ = 126.2 ppm and at δ = 135.0 ppm, whereas the second CH 2 shows a correlation to carbons at δ = 128.7 ppm and δ = 136.0 ppm. These values, together with the NOE contact, are diagnostic of benzyl groups on different nitrogen atoms, as depicted in the structure of 3a.
Based on the entire experimental outcome and the reported literature [45][46][47], we postulated that the undesired dibenzylated byproduct 3a might be due to the competitive pathway illustrated [47] in Scheme 2, where the nucleophilic substitution of benzyl bromide first occurs by the NH 2 (N4) group and then by the cytosine N3 site of the bidentate nucleophile.  However, as in the model reaction, a mixture of two different N-alkylated product namely 1 and 3, were obtained. The mono/di-alkylation ratio ranged from 6:4 to 7:3, a determined by the integration of characteristic protons for each product in the 1 HNM spectra of the concentrated reaction mixtures.
A combination of homo-and heteronuclear 2D NMR experiments (DQF-COSY, 13 C 1 H HSQC, and HMBC, NOESY) were used to assign all the spin systems of 1a and 3a. I  However, as in the model reaction, a mixture of two different N-alkylated produc namely 1 and 3, were obtained. The mono/di-alkylation ratio ranged from 6:4 to 7:3, determined by the integration of characteristic protons for each product in the 1 HNM spectra of the concentrated reaction mixtures.
A combination of homo-and heteronuclear 2D NMR experiments (DQF-COSY, 13 1 H HSQC, and HMBC, NOESY) were used to assign all the spin systems of 1a and 3a.  However, as in the model reaction, a mixture of two different N-alkylated product namely 1 and 3, were obtained. The mono/di-alkylation ratio ranged from 6:4 to 7:3, determined by the integration of characteristic protons for each product in the 1 HNM spectra of the concentrated reaction mixtures.
A combination of homo-and heteronuclear 2D NMR experiments (DQF-COSY, 13 C 1 H HSQC, and HMBC, NOESY) were used to assign all the spin systems of 1a and 3a.  However, as in the model reaction, a mixture of two different N-alkylated produc namely 1 and 3, were obtained. The mono/di-alkylation ratio ranged from 6:4 to 7:3, determined by the integration of characteristic protons for each product in the 1 HNM spectra of the concentrated reaction mixtures.
A combination of homo-and heteronuclear 2D NMR experiments (DQF-COSY, 13 1 H HSQC, and HMBC, NOESY) were used to assign all the spin systems of 1a and 3a. 13 96 and 3.82 D, respectively), and they do not participate in hydrogen bonding. Their high polarity allows them to dissolve charged species such as various anions used as nucleophiles. The lack of hydrogen bonding in the solvent means that the latter is relatively "free" in the solution, making it more reactive.
Moreover, THF and CH 2 Cl 2 as borderline polar aprotic solvents have moderately higher dielectric constants (7.52 and 8.93, respectively) and small dipole moments (1.75 and 1.60 D, respectively). The intermediate polarity makes them good "general purpose" solvents for a wide range of reactions where they serve only as the medium (for example, in the Grignard reaction and for enolate formation).
Thus, to reduce the reactivity of the nitrogen ring in an attempt to increase the efficiency of the behavior, the model reaction was carried out using DMF in a 1:1 mixture with THF as co-solvent [48,49]; we obtained an interesting result, which drove us to perform the reaction in a new 1:1 mixture of CH 2 Cl 2 /THF. As we postulated, these conditions led to very efficient results (Table 3, entry 4).
Reactions 2022, 3, FOR PEER REVIEW 8 bromide first occurs by the NH2(N4) group and then by the cytosine N3 site of the bidentate nucleophile.

Scheme 2. Postulated mechanism.
DMSO and DMF have large dielectric constants (47.24 and 38.25, respectively), large dipole moments (3.96 and 3.82 D, respectively), and they do not participate in hydrogen bonding. Their high polarity allows them to dissolve charged species such as various anions used as nucleophiles. The lack of hydrogen bonding in the solvent means that the latter is relatively "free" in the solution, making it more reactive.
Moreover, THF and CH2Cl2 as borderline polar aprotic solvents have moderately higher dielectric constants (7.52 and 8.93, respectively) and small dipole moments (1.75 and 1.60 D, respectively). The intermediate polarity makes them good "general purpose" solvents for a wide range of reactions where they serve only as the medium (for example, in the Grignard reaction and for enolate formation).
Thus, to reduce the reactivity of the nitrogen ring in an attempt to increase the efficiency of the behavior, the model reaction was carried out using DMF in a 1:1 mixture with THF as co-solvent [48,49]; we obtained an interesting result, which drove us to perform the reaction in a new 1:1 mixture of CH2Cl2/THF. As we postulated, these conditions led to very efficient results (  1 Reactions were performed using cytosine sulfonylate 2 (1 eq), bases (1.5 eq), and benzyl bromide (1.2 eq).
In our mind, the CH 2 Cl 2 -THF mixture could synergistically ensure the solubility of our polar substrate, resulting in the best interaction of the latter with the base as well as the electrophile. CH 2 Cl 2 did not change the regiochemical outcome (N-enamine type versus N-imine type) of mono-alkylation.
Under the optimized reaction conditions, the scope of various para-substituted benzyl bromides was again investigated, and the results are summarized in Table 4. our polar substrate, resulting in the best interaction of the latter with the base as well as the electrophile. CH2Cl2 did not change the regiochemical outcome (N-enamine type versus N-imine type) of mono-alkylation.
Under the optimized reaction conditions, the scope of various para-substituted benzyl bromides was again investigated, and the results are summarized in Table 4.