Cladribine Analogues via O6-(Benzotriazolyl) Derivatives of Guanine Nucleosides

Cladribine, 2-chloro-2′-deoxyadenosine, is a highly efficacious, clinically used nucleoside for the treatment of hairy cell leukemia. It is also being evaluated against other lymphoid malignancies and has been a molecule of interest for well over half a century. In continuation of our interest in the amide bond-activation in purine nucleosides via the use of (benzotriazol-1yl-oxy)tris(dimethylamino)phosphonium hexafluorophosphate, we have evaluated the use of O6-(benzotriazol-1-yl)-2′-deoxyguanosine as a potential precursor to cladribine and its analogues. These compounds, after appropriate deprotection, were assessed for their biological activities, and the data are presented herein. Against hairy cell leukemia (HCL), T-cell lymphoma (TCL) and chronic lymphocytic leukemia (CLL), cladribine was the most active against all. The bromo analogue of cladribine showed comparable activity to the ribose analogue of cladribine against HCL, but was more active against TCL and CLL. The bromo ribose analogue of cladribine showed activity, but was the least active among the C6-NH2-containing compounds. Substitution with alkyl groups at the exocyclic amino group appears detrimental to activity, and only the C6 piperidinyl cladribine analogue demonstrated any activity. Against adenocarcinoma MDA-MB-231 cells, cladribine and its ribose analogue were most active.


Introduction
Cladribine, 2-chloro-2′-deoxyadenosine, has been a molecule of interest for well over five decades. The history of this compound dates back to 1960, when it was used in the synthesis of 2′-deoxyguanosine and 2′-deoxyinosine [1]. A decade later, cladribine was shown to be a poor substrate for adenosine deaminase that underwent phosphorylation by deoxycytidine kinase, finally resulting in the triphosphate, and inhibiting DNA synthesis rather than RNA synthesis [2,3]. Later still, it was shown to be a substrate for deoxyguanosine kinase, which is responsible for phosphorylation of purine nucleosides in mitochondria [3]. Several mechanisms have been proposed by which cladribine can cause mitochondrial damage and apoptotic cell death [4][5][6][7][8].
In contemporary medicine, cladribine is used in the treatment of lymphoid malignancies, most notably for its efficacy against hairy cell leukemia [9]. Cladribine is also being evaluated against several other indolent lymphoid malignancies, also in combination with other drug candidates [10,11].
The synthesis of cladribine has primarily relied on three major methods: (a) glycosylation reactions of a nucleobase with a sugar [12][13][14][15][16][17][18], and its variations; (b) deoxygenation of the C2′ hydroxyl group of a suitable nucleoside derivative [12,15,19,20]; (c) enzymatic glycosyl transfer reactions [21][22][23][24]; and (d) conversion of readily available nucleoside precursors (some utilizing nucleosides for glycosyl transfer reactions) [21,22,[24][25][26]. Each of these methods has been used with varying levels of convenience and success. Among the many approaches, one convenient method is the selective displacement of a leaving group (chloride or aryl sulfonate) from the C6 position of a suitable purine nucleoside precursor. Despite the availability of this selective SNAr displacement, we find that no other N6-substituted cladribine analogues have been synthesized by such a method. Because of our interest in broadening the utilities of O 6 -(benzotriazol-1-yl)purine nucleoside derivatives, we elected to evaluate the synthesis of N6-substituted cladribine analogues via amide-bond activation of guanine nucleosides with (benzotriazol-1yloxy)tris(dimethylamino)phosphonium hexafluorophosphate (BOP).

Results and Discussion
For the modification of the nucleobases of inosine, and 2′-deoxyinosine (3 examples), BOP had been used for in situ activation of the amide moieties in these substrates, followed by SNAr displacement with amines [27,28]. It was proposed that reaction of the amide group with BOP proceeded via a phosphonium intermediate, which could be directly captured by a reactive amine [27]. On the other hand, with less reactive amines, the O-(benzotriazol-1-yl) intermediate can be formed by competitive capture of the phosphonium intermediate by the benzotriazol-1-yloxy anion [27].
In 2007, we first reported the isolation of O 6 -(benzotriazol-1-yl)inosine and -2′-deoxyinosine by amide-bond activation with BOP, in the absence of a nucleophile [29], an observation that was later reconfirmed by others [30,31]. These electrophilic nucleosides, which are stable to storage, are exceptionally good partners in SNAr reactions with oxygen, nitrogen, and sulfur nucleophiles [29]. Subsequently, we demonstrated that O 6 -(benzotriazol-1-yl)inosine and -2′-deoxyinosine can also be prepared via the use of PPh3/I2 and 1-hydroxybenzotriazole [32]. The amide-bond activation protocol was then modified to tether the O 6 -(benzotriazol-1-yl) nucleosides onto a polymer support for high-throughput type of applications [33]. Interestingly, the amide bond activation when applied to the urea functionality of O 6 -benzyl-protected 2′-deoxyxanthosine did not yield the O 2 -(benzotriazol-1-yl) derivative but rather terminated in an isolable and synthetically useful phosphonium salt [34]. We also showed that guanosine and 2′-deoxyguanosine undergo facile reactions with BOP, and that O 6 -(benzotriazol-1-yl) guanine nucleosides are effective substrates for SNAr reactions as well [35]. In the combined course of these investigations we had ascertained plausible operative mechanisms of these amide-activation reactions, results that were later applied to a one-pot etherification protocol for purine nucleosides and pyrimidines [36].
In 2001, diazotization reactions leading to cladribine and other halo derivatives have been performed on unprotected 2,6-diaminopurine 2′-deoxyribonucleoside [45]. However, no substituents other than NH2 were introduced into the C6 position, and the synthesis of the precursor is not particularly convenient. For the present study, we anticipated the need for saccharide protection, and both acetyl and t-butyldimethysilyl (TBS) protecting groups were considered. Between these, TBS was selected because acetyl groups can be susceptible to cleavage with amines, which could be a complicating problem in the method development stage. Thus, nucleosides 1a and 1b were silylated to give the corresponding products 2a and 2b, which were converted to the O 6 -(benzotriazol-1-yl) guanosine derivatives 3a and 3b, respectively (Scheme 1).
We opted for diazotization-chlorination conditions using t-BuONO/TMSCl, a reagent combination initially introduced for nucleoside modification in 2003 [46]. We [47] and others [44] have previously used non-aqueous conditions for halogenation at the C2 position of purine nucleoside derivatives. Under these conditions, reactions of substrate 3a proceeded modestly. However, the obtained product was contaminated with ~30% of the C2 protio O 6 -(benzotriazol-1-yl)-3′,5′-di-O-(t-butyldimethylsilyl)-2′deoxyinosine (Table 1, entries 1 and 2). With the combination of t-BuONO/TMSCl/(BnNEt3) + Cl − , no C2 protio product was observed, but only a low product yield was obtained (entry 3). These results compelled us to consider other conditions. SbCl3 and SbBr3 have previously been used for the halogenation of nucleosides [48,49] Thus, the next series of experiments involved SbCl3, (BnNEt3) + Cl − , and combinations of these reagents (entries [4][5][6][7][8][9][10][11]. Whereas most experiments yielded only 35%-39% of product 4a, a reasonable yield improvement was observed in entry 9. It was noted that efficient filtration of the reaction mixture after workup is critical to obtaining a good product recovery due to the pasty nature of the mixture (see the Experimental Section). On larger scales, better product recoveries were observed (entries 10 and 11). By comparison, diazotization/chlorination of the ribose derivative 3b proceeded well with both t-BuONO/SbCl3 (entry 12) and t-BuONO/TMSCl (entry 13), with the latter providing a better yield of compound 4b. No C2 protio product was apparent in the reactions of precursor 3b.
The next stage in the chemistry involved SNAr reactions at the C6 positions of substrates 4a and 4b. Cladribine and its ribose analogue were prepared by reactions with aqueous ammonia (see the Experimental Section for details). In order to prepare other analogues, reactions were conducted with 1.5 equiv each of methylamine, dimethylamine, pyrrolidine, piperidine, morpholine, and N,N,N′-trimethylethylenediamine (products, reaction times, and yields are shown in Figure 1). Most reactions proceeded smoothly and in good to high yields. Methylamine (2 M in THF) was used for the synthesis of 6a and 6b, whereas a 40% aqueous solution of dimethylamine was used for the synthesis of 7a and 7b. As we have previously shown, water is not generally detrimental to reactions of these benzotriazolyl purine nucleosides [35]. In reactions of 4a and 4b with N,N,N′-trimethylethylenediamine, yields were lower. In each ~10%-15% of compounds 7a and 7b were isolated as byproducts. The source of dimethylamine is currently unknown but its origin can be linked to N,N,N′trimethylethylenediamine.
Finally, for biological testing, desilylation was performed. Because we did not anticipate decomposition of starting materials or products, we only tested the use of KF (2 equiv/silyl group) in MeOH at 80 °C (Scheme 2). The results of the desilylation reactions are shown in Table 2.  Because there is currently no known method for the diazotization/bromination of compound 3a, we investigated a route similar to that for chlorination. Results from these experiments are listed in Table 3. What is notable with the diazotization/bromination, in contrast to the chlorination, was that use of 3.5 equiv of t-BuONO led to incomplete conversion over 2 h at −10 to −15 °C. Addition of another 3.5 equiv of t-BuONO then led to complete conversion over an additional 1 h.
A similar conversion of 3b led to the ribose analogue 19b in a comparable 64% yield (Table 3, entry 3). Compound 19a was then converted to the bromo analogue of cladribine as shown in Scheme 3. SNAr displacement with aqueous NH3 proceeded in 83% yield producing compound 20a, which was finally desilylated with KF in anhydrous MeOH at 80 °C (28 h) to yield 2-bromo-2-deoxyadenosine (21a) in 66% yield. Corresponding conversions of the ribose analog 19b, via intermediate 20b, gave compound 21b. Yields for these conversions were comparable to the deoxyribose series. Notably, previously unknown compound 19a and 19b are relatively easily prepared, new bifunctional reactive nucleosides that can undergo SNAr reactions at the C6 and metal-mediated reactions at the C2 position. To the extent that C-Cl bonds can be activated by metal catalysts compounds 4a and 4b also offer this type of orthogonal reactivity. Results from the orthogonal reactivities of these new halo nucleosides will be reported in the future.

Results of Tests against HCL, TCL, CLL, and MDA-MB-231 Breast Cancer Cells
The newly synthesized compounds as well as cladribine and its bromo analogue were tested against HCL, TCL, and CLL. Data from these assays are shown in Table 4.
From these data, across the entire series, cladribine (12a) was best. The bromo analogue of cladribine (21a) showed lower activities. A comparison of the ribose analogue of cladribine (12b) and the bromo analogue of cladribine (21a) is interesting. Both compounds 12b and 21a show comparable activities against HCL, but the latter shows higher activities against TCL and CLL. The bromo ribose derivative 21b showed activity but was inferior to compounds 12a, 12b, and 21a. In this series, the only other compound to demonstrate any activity was the piperidinyl derivative 16a.  The compounds were also tested against adenocarcinoma MDA-MB-231 breast cancer cells. These cells were treated with three-fold serial dilutions of the compounds ranging from 0-1.8 mM for 24 h. Viable cells were fixed with cold methanol and nuclei were stained with 0.4% propidium iodide (PI). Cell viability was calculated as the percentage of surviving cells after treatment as measured by differences in fluorescence units between treated and untreated wells ( Figure 2). IC50 values were obtained from dose response curve fittings using the non-linear regression function of GraphPad Prism ® (La Jolla, CA, USA). Dashed horizontal line represents 50% cell viability. Columns represent means ± SEM of at least three independent experiments. Significant differences are described with * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001 compared to control.
From the IC50 values shown in Table 5, the two compounds that emerged as most promising were cladribine 12a and its ribose analogue 12b, followed by the bromo derivatives 21a and 21b, which were about 10 times less active. Table 5. IC50 values (mM) of the compounds synthesized on MDA-MB-231 breast cancer cells.

General Considerations
Thin-layer chromatography was performed on 200 μm aluminum-foil-backed silica gel plates for the 2′-deoxynucleosides and on Merck 60F254 (Merck, Billerica, MA, USA) for the ribose analogues. Column chromatographic purifications were performed on 100-200 mesh silica gel. CH2Cl2 for the chlorination reactions was distilled over CaH2. Precursors 3a and 3b were prepared as described previously [35]. The yield of 3a was 71% on a 2.52 mmol scale and the yield of 3b was 71% on a 2.1 mmol scale. TMSCl was redistilled prior to use and all other commercially available compounds were used without further purification. 1 H-NMR spectra were recorded at 500 MHz or at 400 MHz in the solvents indicated under the individual compound headings and are referenced to residual protonated solvent resonances. 13 C-NMR spectra were recorded at 125 MHz or at 100 MHz in the solvents indicated under the individual compound headings and are referenced to the solvent resonances (Supplementary Materials). Chemical shifts (δ) are reported in parts per million (ppm), and coupling constants (J) are in hertz (Hz). Standard abbreviations are used to designate resonance multiplicities (s = singlet, d = doublet, t = triplet, dd = doublet of doublet, ddd = doublet of doublet of doublet, quint = quintet, m = multiplet, br = broad, app = apparent). The saccharide carbons of the nucleoside are numbered 1′ through 5′ starting at the anomeric carbon atom and proceeding via the carbon chain to the primary carbinol carbon atom. The purinyl proton is designated as H-8 and the saccharide protons are designated on the basis of the carbon atom they are attached to.
A mixture of compound 3a (1.0 g, 1.63 mmol) and SbCl3 (520.6 mg, 2.28 mmol) in dry CH2Cl2 (16.3 mL) was cooled to −15 °C using dry ice and acetone, in a nitrogen atmosphere. t-BuONO (0.678 mL, 5.70 mmol) was added dropwise and the mixture was stirred at −10 to −15 °C for 3.5 h, at which time TLC indicated the reaction to be complete. The reaction mixture was poured into ice-cold, saturated aqueous NaHCO3 (25 mL) with stirring. The mixture was filtered using vacuum (note: use of vacuum for this filtration is critical for maximizing product recovery) and the residue was washed with CH2Cl2 (25 mL). The organic layer was separated and the aqueous layer was back extracted with CH2Cl2 (2 × 15 mL). The combined organic layer was washed with water (15 mL) and brine (15 mL), dried over anhydrous Na2SO4, and evaporated under reduced pressure. Purification of the crude material on a silica gel column sequentially eluted with hexanes, 5% EtOAc in hexanes, followed by 20% EtOAc in hexanes gave 0.67 g (65% yield) of compound 4a as a white foam. Rf (SiO2 and 30% EtOAc in hexanes) = 0.55.

2-Chloro
To a solution of compound 4a (126.0 mg, 0.2 mmol) in 1,2-DME (2 mL), 28%-30% aqueous ammonia (32 μL) was added, and the mixture was stirred at room temperature for 1.5 h. The mixture was diluted with EtOAc (5 mL) and washed with 5% aqueous NaCl (5 mL). The organic layer was separated and the aqueous layer was back extracted with EtOAc (5 mL). The combined organic layer was dried over anhydrous Na2SO4 and evaporated under reduced pressure. The crude material was chromatographed on a silica gel column by sequential elution with 20% EtOAc in hexanes followed by 5% MeOH in CH2Cl2 to afford 85.0 mg (82% yield) of compound 5a as a white solid. Rf

General Procedure for the Synthesis of Cladribine Analogues
To a solution of compound 4a (126.0 mg, 0.2 mmol) in 1,2-DME (2 mL) the appropriate amine (1.5 equiv) was added, and the mixture was stirred at room temperature. The mixture was diluted with EtOAc (5 mL for compounds 6a, 9a, 11a, and 15 mL for compounds 7a, 8a, 10a) and washed 5% aqueous NaCl (5 mL for compounds 6a, 9a, 11a, and 15 mL for compounds 7a, 8a, 10a). The organic layer was separated and the aqueous layer was back extracted with EtOAc (5 mL for compounds 6a, 9a, 11a, and 15 mL for compounds 7a, 8a, 10a). The combined organic layer was dried over anhydrous Na2SO4 and evaporated under reduced pressure. The crude material was chromatographed on a silica gel column; see individual compound headings for details.

General Procedure for the Synthesis of Ribose Analogues of Cladribine
To a solution of compound 4b (500 mg, 0.655 mmol) in 1,2-DME (8 mL) was added the appropriate amine (1.5 equiv) and the mixture was stirred at room temperature. The mixture was diluted with EtOAc (25 mL) and washed with 5% aqueous NaCl (15 mL). The organic layer was separated and the aqueous layer was back extracted with EtOAc (15 mL). The combined organic layers were dried over anhydrous Na2SO4 and evaporated under reduced pressure. The crude material was chromatographed on a silica gel column. See individual compound headings for details.

General Procedure for the Desilylation of Cladribine Analogues
To a 0.1 M solution of the silylated compound in anhydrous MeOH, KF (2 equiv/silyl group) was added. The mixture was heated at 80 °C for 20-26 h, cooled, and silica gel was added. The mixture was evaporated to dryness and the compound-impregnated silica gel was loaded onto a wet-packed silica gel column. The products were obtained by elution with appropriate solvents (see the individual compound headings for details).

General Procedure for the Desilylation of Ribose Cladribine Analogues
To a 0.1 M solution of the silylated compound in anhydrous MeOH, KF (2 equiv/silyl group) was added. The mixture was heated at 80 °C for 24 h, cooled, and silica gel was added. The mixture was evaporated to dryness and the compound-impregnated silica gel was loaded onto a wet-packed silica gel column. The products were obtained by elution with appropriate solvents (see the individual compound headings for details).

O 6 -(Benzotriazol-1-yl)-2-bromo-9-[2-deoxy-3,5-di-O-(t-butyldimethylsilyl)-β-D-ribofuranosyl]purine
(19a): A mixture of compound 3a (300.0 mg, 0.489 mmol) and SbBr3 (247.7 mg, 0.685 mmol) in dry CH2Br2 (4.9 mL) was cooled to −15 °C using dry ice and acetone, in a nitrogen atmosphere. t-BuONO (203.8 μL, 1.713 mmol) was added dropwise and the mixture was stirred at −10 °C to −15 °C for 2 h. Because TLC indicated the presence of starting material, another aliquot of t-BuONO (203.8 μL, 1.713 mmol) was added and the reaction proceeded for 1 h at −15 °C, at which time TLC indicated the reaction to be complete. The reaction mixture was poured into ice-cold, saturated aqueous NaHCO3 (5 mL) with stirring. The mixture was filtered using vacuum (note: use of vacuum for this filtration is critical for maximizing product recovery) and the residue was washed with CH2Cl2 (5 mL). The organic layer was separated and the aqueous layer was back extracted with CH2Cl2 (2 × 5 mL). The combined organic layer was washed with water (5 mL) and brine (5 mL), dried over anhydrous Na2SO4, and evaporated under reduced pressure. Purification of the crude material on a silica gel column sequentially eluted with hexanes, 5% EtOAc in hexanes, and 30% EtOAc in hexanes gave 208.5 mg (63% yield) of compound 19a as a white foam. Rf (SiO2 and 30% EtOAc in hexanes) = 0.60. 1  3 mg, 0.20 mmol) in 1,2-DME (2 mL), 28%-30% aqueous ammonia (48.6 μL) was added, and the mixture was stirred at room temperature for 45 min. The mixture was diluted with EtOAc (15 mL) and washed with 5% aqueous NaCl (10 mL). The organic layer was separated and the aqueous layer was back extracted with EtOAc (15 mL). The combined organic layer was dried over anhydrous Na2SO4 and evaporated under reduced pressure. The crude material was chromatographed on a silica gel column by sequential elution with hexanes, 20% EtOAc in hexanes and 40% EtOAc in hexanes to afford 93.1 mg (83% yield) of compound 20a as a white foam. Rf (SiO2 and EtOAc) = 0. 50  Because TLC indicated the presence of starting material, another aliquot of t-BuONO (280 μL, 2.352 mmol) was added and the reaction was allowed to progress for 1 h at −15 °C, at which time TLC indicated the reaction to be complete. The reaction mixture was poured into ice-cold, saturated aqueous NaHCO3 (10 mL) with stirring. The mixture was filtered using vacuum (note: use of vacuum for this filtration is critical for maximizing product recovery) and the residue was washed with CH2Cl2 (15 mL). The organic layer was separated and the aqueous layer was back extracted with CH2Cl2 (2 × 15 mL). The combined organic layer was washed with water (10 mL) and brine (10 mL), dried over anhydrous Na2SO4, and evaporated under reduced pressure. Purification of the crude material on a silica gel column sequentially eluted with hexanes, 5% EtOAc in hexanes, and 30% EtOAc in hexanes gave 350 mg (64% yield) of compound 19b as a white foam. Rf (SiO2 and 30% EtOAc in hexanes) = 0.80.

Conclusions
We have demonstrated, for the first time, the synthesis of O 6 -(benzotriazol-1-yl)-2-chloro-9-[2deoxy-3,5-di-O-(t-butyldimethylsilyl)-β-D-ribofuranosyl]purine (4a) by diazotization-chlorination of O 6 -(benzotriazol-1-yl)-3′,5′-di-O-(t-butyldimethylsilyl)-2′-deoxyguanosine (3a) with t-BuONO and SbCl3 in CH2Cl2. This procedure afforded better yields than the chlorination using t-BuONO and Me3SiCl. This compound and its ribose analogue, O 6 -(benzotriazol-1-yl)-2-chloro-9-[2,3,5-tri-O-(tbutyldimethylsilyl)-β-D-ribofuranosyl]purine, both undergo smooth reactions with ammonia, and primary, and secondary amines to produce cladribine (12a), its N-modified analogues (13a-18a), and the corresponding ribose derivatives (12b-18b), after a simple desilylation with KF in MeOH. These compounds were tested against HCL, TCL, and CLL, but none of the new compounds was more active than cladribine itself. The bromo as well as ribose analogues of cladribine displayed activity but the bromo analogue of cladribine was more active against TCL and CLL as compared to both the ribose equivalent and the bromo ribose analogue of cladribine. The compound containing both the bromine atom and a ribose ring was least active among the compounds possessing a primary amino group at the C6 position. Thus, it appears that a free amino group at this location is critical to the activity of cladribine. Interestingly, the C6 piperidinyl analog of cladribine showed low activity. Tests against MDA-MB-231 breast cancer cells showed that only cladribine and its ribose analogue showed some activity. The bromo analogues were about 10 times less active and all others showed no potential. Despite the lack of a major improvement in the activity of cladribine, or the identification of new compounds with activity against breast cancer, this work has provided a route to four doubly functionalizable nucleoside derivatives. The orthogonal reactivities of these compounds, i.e., SNAr at the C6 position and metal catalysis at the C2 position, can be used for development of novel nucleoside analogues. We anticipate pursuing further work along these lines in the future.

Author Contributions
Sakilam Satishkumar, Prasanna K. Vuram, and Siva Subrahmanyam Relangi had equal contributions to the synthesis portion of this work. Sakilam Satishkumar and Prasanna K. Vuram (City College of New York) performed optimization of the synthetic procedures, executed syntheses of the deoxyribose series, performed spectroscopic analyses of the compounds, and produced a part of the experimental section. Messrs. Relangi and Venkateshwarlu Gurram (GVKBIO) executed syntheses of the ribose series, performed spectroscopic analyses of the compounds, and produced a part of the experimental section. Hong Zhou and Robert J. Kreitman (National Cancer Institute) tested the compounds against HCL, TCL, and CLL. Michelle M. Martínez Montemayor tested the compounds against breast cancer cell lines. Lijia Yang (City College of New York) performed HRMS analysis of all compounds synthesized in Lakshman's laboratories. Narender Pottabathini (GVKBIO) was responsible for the oversight of the work performed in his laboratories. Mahesh K. Lakshman (City College of New York) conceived and designed the research, assisted with data analysis, and wrote a significant portion of this manuscript.