Design, synthesis, and evaluation of potential carbamate prodrugs of 5′-methylthioadenosine (MTA)

5′-Methylthioadenosine (MTA) is a natural substrate of MTA phosphorylase (MTAP) and is converted to adenine via a salvage pathway for AMP production in normal healthy cells. The lack of MTAP expression in many solid tumors and hematologic malignancies compared to normal healthy cells has been explored in a potential therapeutic strategy to selectively target tumor cells using antimetabolites such as 5-fluorouracil (5-FU) and 6-thioguanine (6-TG) while protecting normal healthy cells with MTA. Herein, a series of carbamate prodrugs, namely the N-(alkyloxy)carbonyl-MTA derivatives 2a-f, was designed, synthesized, and evaluated as potential prodrugs of MTA. All carbamate prodrugs were stable in phosphate buffer, pH 7.4 at 37 °C. In the presence of mouse liver microsomes, the prodrugs were converted to MTA at varying rates with the hexyl and butyl carbamates 2a and 2b most readily activated (t1/2 of 1.2 and 9.4 h, respectively). The activation was shown to be mediated by carboxyesterases present in mouse liver microsomes.

and is a substrate of methylthioadenosine phosphorylase (MTAP) in a process that generates adenine and 5-methylthioribose-1-phosphate [2,3]. The latter is then converted in several steps to methionine in the methionine salvage pathway. Adenine is converted to AMP by the ubiquitous enzyme adenine phosphoribosyltransferase (APRT) with phosphoribosyl-1-pyrophosphate (PRPP) serving as the donor of the phosphoribosyl group. Although AMP may also be produced in cells by de novo purine biosynthesis, MTAP is responsible for the generation of essentially all free adenine in normal healthy cells [4].
Many solid tumors and hematologic malignancies are characterized by a deficiency of MTAP and are unable to salvage adenine from MTA. The difference in the lack of expression of MTAP in tumor cells compared to normal healthy cells has been explored in a therapeutic strategy for selectively killing tumor cells [4,5]. Antimetabolites such as 5-fluorouracil (5-FU), 6-thioguanine (6-TG) or antifolates disrupt DNA replication and inhibit de novo purine synthesis through the release of cytotoxic nucleotides generated via phosphorylation using PRPP. One major drawback of these cytotoxic nucleotides is that they produce side effects on the growth and proliferation of replicating normal healthy cells. To reduce the cytotoxicity of these chemotherapeutic drugs and thus increase their therapeutic index, administration of MTA or MTA analogs in combination with these antimetabolites could prevent damage to healthy MTAPcompetent cells [6,7]. This is because adenine generated in the above adenine salvage pathway by MTAP in healthy cells will block the conversion of 5-FU/6-TG to their cytotoxic nucleotides by competing for the rate-limiting pools of PRPP. Since no adenine is produced in tumor cells due to lack of MTAP, PRPP is present in sufficient levels and the coadministered drug can be readily converted to its toxic metabolite. Thus, a high degree of selectivity can be achieved. Indeed, MTA protection was shown to be quite remarkable in vitro with at least 10-100-fold changes in IC 50 values when co-administered with 6-TG, suggesting that higher doses of 6-TG than currently administered could be given before any side effects are seen [3-5, 8, 9]. MTA in high doses has been shown to prevent liver damage in experimental models [10,11]. Bertino et al. also reported that co-administration of 100 mg/kg MTA protected nude mice from a lethal dose of 5-FU, raising its maximum tolerated dose from 75 to 200 mg/kg [5]. In a clinical study, MTA was dosed daily at 600 mg for 1 month to 50 volunteers and at 1600 mg daily for 1 month to another 10 volunteers without reports of any toxicity in either group [12,13]. These proof of concept preclinical and clinical studies suggest that MTA can produce a beneficial protective effect to healthy cells and thus reduce systemic side effects during cancer chemotherapy using these antimetabolites.
There are several MTA prodrugs disclosed in a US patent by modifying the ribosyl hydroxyl groups in the form of carbamates, phosphates and esters to deliver MTA to protect MTAP-competent cells [14]. No prodrugs of MTA with a promoiety attached to the C6-amino group has been reported despite the existence of a similar and yet very successful prodrug capecitabine, that was patented in 1992 and approved for clinical use in 1998 [15]. As shown in Fig. 1, capecitabine is activated in a three-step process to release the final active drug 5-FU starting with hepatic carboxylesterase-mediated hydrolysis in the first step to form 5′-deoxy-5-fluorocytidine (5′-DFCR) [15,16]. Inspired by the design and the first-step hydrolytic activation of capecitabine, we report, herein, the design of new prodrugs of MTA by attaching promoieties to the 6-amino group of adenine for activation by carboxylesterases to release MTA in an effort to overcome the potential stability and bioavailability issues of MTA.

Prodrug design
We focused our design of prodrugs on the 6-NH 2 of MTA that could potentially be activated by a carboxylesterase Fig. 1 The three-step activation of capecitabine by hepatic carboxylesterase, cytidine deaminase and thymidine phosphorylase [15,16] enzyme [17,18]. Carboxyesterases are ubiquitously present in most animals including humans, and several types are present in blood, liver and other tissues. The initial design of our MTA prodrugs was based on the activation of the marketed prodrug of 5-FU known as capecitabine. Clinically, capecitabine passes intact through the intestinal mucosa and is converted to its active form, 5-FU, by a 3step enzymatic activation in liver and tumor tissues (Fig. 1). Its activation is initiated by carboxylesterase-mediated hydrolysis of the carbamate moiety to 5′-deoxy-5-fluorocytidine primarily in the liver with a reported half-life of 0.55-0.87 h [19,20].
We used a similar strategy to design the carbamate or N-(alkyloxy)carbonyl-MTA prodrugs 2a-f as depicted in Fig. 2 that can potentially be activated by the liver carboxylesterases for systemic release of MTA. In the process of designing MTA carbamate prodrugs 2a-f, we varied the alkyl groups of the promoiety from 2 to 6 carbons to discern what effect the alkyl group would have on the rate of cleavage of the carbamate moiety by the liver carboxylesterase. The carbamate functional group can impart chemical stability and since carbamates do not ionize, they can also provide better oral bioavailability for the prodrugs. Upon hydrolysis, the prodrugs would release alcohol and carbamic acid that would be unstable under physiological pH and decompose spontaneously to the parent amine drug (in our case, MTA) and CO 2 .

Chemical synthesis
The syntheses of N-(alkyloxy)carbonyl-MTA prodrugs 2a-f are straightforward and were accomplished in two steps as shown in Scheme 1. In step 1, treatment of commercially available MTA with an excess amount (6 eq.) of commercially available alkyl chloroformates in the presence of a base in anhydrous dichloromethane gave the intermediates 3a-f with a carbamate on 6-NH 2 of adenine and two carbonates on the two hydroxyl groups of ribose. The carbonates were more easily and selectively hydrolyzed using 0.1 N NaOH to afford the desired prodrugs 2a-f.

Evaluation of N-(alkyloxy)carbonyl-MTA Prodrugs 2a-f
We first evaluated the stability of the carbamate or N-(alkyloxy)carbonyl-MTA prodrugs 2a-f in 50 mM phosphate buffer, pH 7.4 at 37°C using an HPLC assay that is capable of separating the prodrugs and the parent molecule MTA on a reverse phase C 18 column. As shown in Table 1, all the N-(alkyloxy)carbonyl-MTA prodrugs 2a-f of MTA demonstrated great stability under these conditions with <10% change in measured peak area over 72 h. Then we determined their activation in the presence of mouse or human liver microsomes (HLM). In the presence of mouse liver microsomes, the prodrugs were activated at varying rates in the order of hexyl (2a) > butyl (2b) > propyl (2e) > 2-methoxyethyl (2f) > isopropyl (2c) ≈ ethyl (2d) with the hexyl and butyl carbamate prodrugs 2a and 2b exhibiting a half-life of 1.2 and 9.4 h, respectively. As shown in Fig. 3A and 3B, the activation of prodrugs 2a and 2b in the presence of MLM was accompanied with the formation of MTA, parent drug. It is also clear from the MTA decomposition phase shown in Fig. 3B vs. 3A that overall degradation of MTA was slower when the prodrug activation was slower in the presence of MLM, suggesting that prodrug 2b significantly increased the "stability" of MTA because of the extended release of MTA from the prodrug which validates our design principle. However, as shown in Table 1, the activation of the N-(alkyloxy)carbonyl-MTA prodrugs in the presence of HLM was much slower with t 1/2 all >48 h; prodrugs 2a and 2b showed about 44 and 19% change, respectively, in the peak area of the prodrugs measured over the 48 h incubation with HLM.
To confirm the type of enzyme involved in the metabolic activation of N-(alkyloxy)carbonyl-MTA prodrugs, we performed the activation of 2b using mouse liver microsomes in the presence of 1 mM bis(4-nitrophenyl)-phosphate (BNPP), an inhibitor of carboxyesterases, or 1 mM of nicotinamide adenine dinucleotide phosphate (NADPH), a cofactor needed for oxidative metabolism. As shown in Fig. 4, the BNPP completely prevented the activation of 2b by mouse liver microsomes while NADPH had no effect on the activation of 2b, confirming that the activation of N-(alkyloxy)carbonyl-MTA prodrug 2b is mediated by a carboxylesterase(s), not by oxidoreductases, in mouse liver microsomes.  It is not surprising that MLM seems to preferentially hydrolyze carbamate prodrugs with longer chain length and that HLM seems to hydrolyze the carbamate prodrugs at much slower rates as compared to MLM. Such differences in substrate preferences of carboxylesterases from different species and tissues in hydrolyzing esters with different alkyl chain length has been reported [21][22][23]. Similar data were obtained when we tested the activation of the prodrugs 2a and 2b in both mouse and human liver S9 fractions (data not shown), suggesting that the prodrugs are good substrates of mouse liver enzymes but not the human liver enzymes.
In summary, we designed, synthesized, and conducted activation studies of carbamate prodrugs of 5′-methylthioadeosine (MTA), in the form of N-(alkyloxy)carbonyl-MTA. All of the carbamate prodrugs obtained were stable in phosphate buffer, pH 7.4 at 37°C. In the presence of mouse liver microsomes, the prodrugs were converted to MTA at varying rates with the hexyl and butyl carbamates 2a and 2b most readily activated exhibiting t 1/2 of 1.2 and 9.4 h, respectively. The activation was shown to be mediated by carboxyesterases present in mouse liver microsomes. The activation of these prodrugs in the presence of HLM was much slower with t 1/2 all >48 h. Consistent with the activation studies in the presence of mouse liver microsomes, the hexyl and butyl carbamates 2a and 2b were shown to have the highest level of conversion at 44% and 19%, respectively. Further studies will be needed to fully explore the potential of these carbamate prodrugs.

Experimental section
All reactions were performed in oven-dried glassware. All solvents used were either ACS reagent grade or HPLC grade. All reactions were performed using anhydrous solvents. Moisture sensitive reactions were performed under N 2 atmosphere. Air sensitive reagents were transferred to a reaction flask using a cannula under N 2 atmosphere. All reactions were monitored by Thin Layer Chromatography (TLC) using 0.25 mm Whatmann precoated silica gel plates. TLC plates were visualized by either UV absorbance or potassium permanganate solution. Purification was done by using automated flash column chromatography using Teledyne ISCO CombiFlash Companion System.
Analytical LC-MS was obtained using Agilent 1200 Series LC system equipped with G6140A Quadruple MS on a C 18 column. Solvent A was 0.1% formic acid/H 2 O and solvent B was 0.1% formic acid in methanol and the gradient used was 10-90% B in 5 mins at a flow rate of 0.8 ml/ min. NMR spectra were recorded on a Bruker Ultrashield 400 MHz at ambient temperature. High Resolution Mass Spectrometry data was obtained at the Center for Integrative Proteomics Research, Rutgers University.
General procedure for the synthesis of N-(alkyloxy) carbonyl-MTA prodrugs (2a-f) 5′-MTA was suspended in 10-20 ml anhydrous dichloromethane. 1-Methylimidazole (8 eq.) was added and the suspension was stirred at 0°C for 5 mins. An alkyl chloroformate (6 eq.) was added dropwise to the cold suspension over 5 mins while maintaining the temperature at 0°C. The reaction mixture was allowed to warm to room temperature and was stirred for an additional 3 days. The reaction progress was monitored using TLC and LCMS. The reaction mixture was diluted with dichloromethane (20 mL) and washed successively twice each with saturated NaHCO 3 (10 ml) and brine (10 mL). The organic phase was separated and dried over Na 2 SO 4 and filtered. The solvent was evaporated, and the crude product was purified by column  Fig. 4 Activation of N-(alkyloxy)carbonyl-MTA prodrug 2b after 4 h incubation with mouse liver microsomes in the presence of 1 mM BNPP or NADPH. The relative peak area of each compound was measured using HPLC of aliquots taken 4 h from incubations at 37°C in the presence of 0.5 mg/ml mouse or human liver microsomes + 1 mM BNPP or NADPH chromatography using methanol and dichloromethane as an eluent to give the intermediate 3a-f as an oil. To a solution of intermediate 3a-f in 5 mL ethanol was added 0.1 N NaOH (1 eq.) and the resulting reaction mixture was stirred at room temperature for 12 h. On completion of the reaction, 1 N HCl was added to the reaction mixture until the pH was neutral. Dichloromethane (10 mL) was then added to the reaction mixture followed by washing with brine (5 mL). The organic layer was dried over Na 2 SO 4 , filtered and the solvent was evaporated. The crude product was purified using column chromatography in methanol in dichloromethane to give final target compounds 2a-f as white solids.

Stability assay of MTA prodrugs in phosphate buffer
The stock solution of the prodrug (10 mM in DMSO) was diluted tenfold to give a 1 mM working solution in 10% DMSO. 100 μL of the 1 mM solution was added to 875 μL of 50 mM phosphate buffer containing 1 mM EDTA, pH 7.4 and incubated at 37°C for 5 min in an Eppendorf vial. The vial was then centrifuged for 5 min and the supernatant was split into 3 fractions of 292.5 μL, one for stability in phosphate buffer, one for activation in HLM and one for activation in mouse liver microsomes.
The fraction for analyzing stability was mixed with 7.5 μL of deionized water and incubated at 37°C for 3 days. Time points were collected after every 0, 3, 5, 10, 15, 30, 45 min, 1, 3, 5, 8, 10, 12, 24, 48, 72 h. At each time point, 20 μL aliquots were taken and diluted with 40 μL cold acetonitrile and centrifuged for 5 mins. 40 μL of the supernatant were taken and diluted with 40 μL deionized water and frozen before injecting on the HPLC. The samples were analyzed using a Perkin Elmer Series 2000 HPLC with a Phenomenex Kinetix C18 (50 × 4.6 mm) column and a gradient of 10-90% acetonitrile containing 0.1% TFA in 10 min at a flow rate of 0.6 mL/min was used with detection wavelength of 220 and 254 nm.
In vitro activation assay in human and mouse liver microsomes [20] The mouse and HLM were obtained from Xeno Tech with a protein concentration of 20 mg/ml and stored in Eppendorf vials at −70°C. Stock solution of 1 mM compound in 1 mL DMSO was diluted tenfold to give a 1 mM solution containing 10% DMSO. 100 μL of the 1 mM solution was added to 875 μL of 50 mM phosphate buffer containing 1 mM EDTA, pH 7.4 and incubated at 37°C for 5 min in an Eppendorf vial. The vial was then centrifuged for 5 min and the supernatant was split into 3 fractions of 292.5 μl, one for stability in phosphate buffer, one for activation in HLM and one for activation in mouse liver microsomes. The fraction for analyzing activation was mixed with 7.5 μL of either the human or mouse liver microsomes and incubated at 37°C for 2 days. Time points were collected after every 0, 3, 5, 10, 15, 30, 45 min, 1, 3, 5, 8, 10, 12, 24, and 48 h. At each time point, 20 μL aliquots were taken and quenched with 40 μL cold acetonitrile and stored on ice for 30 mins before centrifugation for 5 mins. 40 μL of the supernatant were taken and diluted with 40 μL deionized water and frozen before injecting on the HPLC. The samples were analyzed using a Perkin Elmer Series 2000 HPLC with a Phenomenex Kinetix C18 (50 × 4.6 mm) column and a gradient of 10-90% acetonitrile containing 0.1% TFA in 10 min at a flow rate of 0.6 mL/min was used with a detection wavelength of 220 and 254 nm. The half-life was calculated from the first order rate equation of the conversion of prodrug to MTA.
Oxidative activation assay in presence of NADPH Stock solution (10 mM in DMSO) was diluted tenfold to give a solution of 1 mM in 10% DMSO. 100 μL of the 1 mM solution was added to 875 μL of 50 mM phosphate buffer containing 1 mM EDTA, pH 7.4 and incubated at 37°C for 5 min in an Eppendorf vial. The vial was then centrifuged for 5 min and the supernatant was split into 2 fractions, each of 292.5 μL, one for oxidative activation in mouse liver microsomes and one for HLM. Each of the fractions was mixed with 20 μL of NADPH and either 20 μL of mouse liver microsomes or HLM and incubated at 37°C for 4 h. At the end of incubation, 20 μL aliquots were taken and quenched with 40 μL cold acetonitrile and stored on ice for 30 mins before centrifugation for 5 mins. 40 μL of the supernatant were taken and diluted with 40 μL deionized water and frozen prior to HPLC analysis. The samples were analyzed using a Perkin Elmer Series 2000 HPLC with a Phenomenex Kinetix C18 (50 × 4.6 mm) column and a gradient of 10-90% acetonitrile containing 0.1% TFA in 10 min at a flow rate of 0.6 mL/min was used with a detection wavelength of 220 and 254 nm. The half-life was calculated from the first order rate equation of the conversion of prodrug to MTA.

Compliance with ethical standards
Conflict of interest The authors declare no competing interests.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.