Pro-Nucleotides - Recent Advances in the Design of Efficient Tools for the Delivery of Biologically Active Nucleoside Monophosphates

A summary of the most recent advances to the design of pronucleotides will be presented. Approaches that have been designed to be activated by enzymes such as carboxyesterases [bis(POM)-, bis(POC)-, bis(SATE)-, bis(AB) phosphotriesters and the arylphosphoramidates] or by reductases [bis(SDTE) approach] will be discussed as well as the amino acids phosphoramidate diester concept with its still unknown delivery mechanism and the cycloSal approach that releases the nucleotides by an induced tandem reaction. Nucleoside analogues, e.g. 2',3'-dideoxy-2',3'-didehydrothymidine 1 (d4T), 3'-azido-2',3'-dideoxythymidine 2 (AZT) or 5-fluoro-2'deoxyuridine 3 (5-FdU), are structurally different as compared to the corresponding natural DNA or RNA nucleosides with regard to modification of the glycon as well as the aglycon residue. Due to this modified structure, these compounds are widely used as antiviral or antitumor drugs in chemotherapy (Figure 1)(1). Since the discovery of AZT 2 as the first nucleoside drug for the treatment of AIDS, considerable efforts have been made to develop new nucleoside analogues that would be more active, less toxic inhibitors of the HIV-1 reverse transcriptase (RT)(2). The general mode of action of nucleoside analogues is the inhibition of the HIV-1 RT by acting as competitive inhibitors or as DNA chain terminators. To act as DNA chain termination agents or RT inhibitors, intracellular conversion of the nucleoside analogues into their 5'-mono-, 5'-dior 5'-triphosphates is a prerequisite after cell penetration(3). The enormous disparity in antiHIV activity that is evident for a large number of dideoxynucleoside analogues belies their apparent structural similarity. Due to these structural differences as compared to natural nucleosides the metabolization to the corresponding dideoxynucleoside triphosphates is often inefficient and consequently the therapeutic efficacy is sometimes limited(4). For example, in the case of the anti-HIV active dideoxynucleoside analogue d4T 1 (Stavudine, Zerit®)(5) the first phosphorylation to d4T 5'-monophosphate 4 catalyzed by thymidine kinase (TK) is the rate-limiting step in human cells(6). More striking, however is 2',3'-dideoxyuridine triphosphate (ddUTP) which is one of the most powerful and selective inhibitors of HIV reverse transcriptase (Ki = 0.05 μM) while the parent nucleoside 2',3'-dideoxyuridine 5 (ddU) is virtually ineffective at blocking HIV infection in cultured cells. Biochemical and pharmacological studies in three different human T cell lines (CEM, ATH8 and Molt-4) showed that ddU 5 itself was not anabolized to the 5'-monophosphate, most apparently because it was a poor substrate for cellular nucleoside kinases because of the considerable substrate specificity of these enzymes(7). In contrast, in a few cases the limited efficacy is also due to a catabolic enzymatic reaction. For example, 2',3'-dideoxyadenosine 6 (ddA) is rapidly intracellularly deaminated to ddI 7 by adenosine deaminase (ADA)(8,9). As a consequence, ddI 7 has to be converted into its ultimate bioactive metabolite ddATP via ddAMP 8 by five enzymatic steps (5'nucleotidase, adenylosuccinate synthase, adenylosuccinate lyase and two nucleotide kinases)(8,9). Finally, the resistance of the human immunodeficiency virus to the clinically used antiviral dideoxynucleoside AZT 2 (Zidovudine, Retrovir®)(10) is on the one hand directly related to multiple point mutations within the HIV-1 reverse transcriptase gene of the virus but may on the other hand be also due to lower susceptibility of resistant target cells to the drug related with a decreased activity or inability of the enzyme thymidine kinase to phosphorylate AZT 2 to the dideoxynucleoside monophosphate AZTMP 9(11). Consequently, direct administration of the nucleotides d4TMP 4, ddAMP 8 and AZTMP 9 should bypass these limiting steps and hence has advantages for the biological activity. Unfortunately, because of the high polarity of the nucleoside monophosphates (nucleotides), these compounds are not able to easily penetrate cellular membranes or the blood-brain barrier. However, the phosphate moiety offers a suitable site to attach degradable lipophilic carrier residues. As a result, one effort to improve the therapeutic potential of nucleoside analogues is the delivery of the corresponding nucleotide from neutral, membranepermeable prodrugs (Pro-Nucleotide Approach; Figure 2)(12). T hi s do cu m en t w as d ow nl oa de d fo r pe rs on al u se o nl y. U na ut ho riz ed d is tr ib ut io n is s tr ic tly p ro hi bi te d. 234 C. Meier SYNLETT So, a lipophilic phosphotriester may penetrate into the target cell where first partial and at the end complete hydrolysis delivers the nucleotide. A suitable nucleotide prodrug has to fulfill two requirements: i) it has to be lipophilic enough for passive diffusion of the membrane and bloodbrain barrier; ii) furthermore, it should be able to deliver the nucleoside hydrolytically or enzymatically leaving a non-toxic masking group(13). In principle, two different concepts for prodrug design are known: bipartate and tripartate prodrugs. In the former concept the drug is modified by a one-component masking group. In this form the drug is biologically inactive. After a simple cleavage of the mask, the active drug is liberated (Figure 3). In the latter concept, the drug is modified by a two-component masking group. Again, the drug is biologically inactive in this bound form. The mechanism of liberation involves a first chemical or enzymatic reaction under cleavage of part I of the masking moiety. The drug is still inactive but the effect of this first reaction is an activation of the remaining masking group II with the consequence of a spontaneous successive cleaving reaction releasing the now bioactive drug (Figure 3)(14). In the case of a nucleotide prodrug one should take into account that under physiological conditions two negatively charged phosphate oxygen's have to be masked in order to obtain a neutral, lipophilic phosphate ester. Consequently, not only one masking group is necessary but two. So, the efficient intracellular delivery of nucleotides from a prodrug requires the existence of a specific delivery mechanism or different rates of conversion of the prodrug to the drug intracellularly versus extracellularly. One comment with respect to toxic side events should be given. Neutral phosphorus derivatives with a good leaving group attached to the phosphorus are known to be toxic suicide inhibitors of acetylcholinesterase(15). The anti-acetylcholinesterase activity of phosphorus derivatives is an inverse function of the pKa of the leaving group on the phosphorus atom and parallels the rate of spontaneous hydrolysis by P-O bond cleavage(16). In order to circumvent the possible problem of anti-cholinesterase activity, neutral phosphate ester prodrugs should be designed to undergo heterolytic cleavage of the C-O bond rather than the P-O bond of the ester. Many strategies have been developed to achieve this goal. As a general motive, uncharged nucleotide triesters are used as membrane-permeable nucleotide precursors(12). The major differences of these approaches are the delivery mechanisms to the nucleotides. First attempts have been made with simple dialkyl phosphotriesters. These compounds generally belong to the class of bipartate prodrug systems. After a first, sometimes selective hydrolysis of the phosphotriester via a nucleophilic reaction at the phosphorus center, the resulting phosphodiester is often extremely stable against a further chemical hydrolysis due to the charge at the phosphate which prevents a second nucleophilic reaction(17). Even if the subsequent hydrolysis is possible, one should take the pseudorotation phenomenon into account that excludes a selective delivery of the nucleotide(18). As a consequence, almost all approaches based on chemical hydrolysis reported so far were unable to deliver the nucleotide selectively except the cycloSal approach that will be discussed later. For this reason, the newer pro-nucleotide approaches are based on the concept of a tripartate prodrug system(14) and are based on the general idea of a selective chemical or enzymatic reaction within the masking group which leads to a second, spontaneous successive reaction yielding the charged phosphate ester. These approaches utilize and exploit the differences in reducing potentials, enzyme activity, and pH value. The concepts working with enzymatic trigger processes (bis(POM)-, bis(POC)-, bis(DTE)-, bis(SATE)-, bis(AB)and the arylphosphoramidate concept) as well as the cycloSal approach based on a pH-driven degradation have demonstrated the successful intracellular delivery of free nucleotides from lipophilic precursors. It should be added, that it is not the intention of the author to give an entire overview of the pro-nucleotide field. The current review highlights some approaches that have been designed to deliver the nucleotide selectively by a special mechanism. With this selection, the author wants to point out that designing a delivery mechanism and bypassing a certain limiting metabolization the inherent biological potential of an already known nucleoside analogue could be used to a higher extent and consequently, the improvement in antiviral activity could be, in many cases, better than synthesizing new potential nucleoside analogues. Furthermore, this review is restricted to the delivery mechanism and so the synthesis of the compounds may be gleaned from the original literature. Chris Meier, born 1962 in Berlin, was trained in Chemistry at the University of Marburg. He passed his Diploma and Ph.D. thesis in the group of Prof. G. Boche in Organic Chemistry. Then he moved as postdoc fellow to the Pasteur-Institute in Paris where he started his work in nucleoside and oligonucleotide chemistry. In 1991 he joined the group of Prof. J.W. Engels at the University of Frankfurt/Main starting his Habilitation which he finished in July '96 and in May '97 he was appointed as an associate professor at the University of Würzburg. Biographical Sketch T hi s do cu m en t w as d ow nl oa de d fo r pe rs on al u se o nl y. U na ut ho riz ed d is tr ib ut io n is s tr ic tly p ro hi bi te d. March 1998 Pro-Nucleotides Recent Advances in the Design of Efficient Tools 235 Bis(POM)and Bis(POC)-Nucleotides The first approach reported by D. Farquhar et al. towards a new class of tripartate pro-nucleotides are the bis(pivaloyloxymethyl)[POM] phosphotriesters of general type 10 (shown exemplified as the ddU derivative)(19,20). This approach utilizes a carboxyesterase-catalyzed cleavage of the pivaloyl ester within the POM-masking group to yield the highly reactive hydroxymethyl phosphotriester 11 which subsequently eliminates spontaneously formaldehyde to give the mono(POM) phosphodiester 12. The carboxyesterase which is used for this activation process may be more prevalent inside the cells. To obtain the free nucleotide, this enzymatic activation has to be repeated via 13 or, alternatively, a phosphodiesterase cleaves the phosphodiester 12 directly to yield the nucleotide (Figure 4). This approach has been applied successfully to the delivery of 2',3'dideoxyuridine monophosphate from bis(POM)-ddUMP(21). Furthermore, this concept has been used for the delivery of the anti-HIV drug AZTMP(22), the anti-herpes and anti-HIV drug phosphonomethoxyethyladenine (PMEA)(23) and for the antitumor active drug 5-fluoro-2'-deoxyuridine monophosphate (5-FdUMP)(24). In the case of the phosphonate analogue PMEA a highly improved bioavailability was obtained. Moreover, the bis(POM) derivatives of (R)-9-(2-phosphonomethoxypropyl)adenine (PMPA) and (R)-9-(2phosphonomethoxypropyl)diaminopurine (PMPDAP) were found to be 9to 23-fold more active than the parental compounds(23). However, the cytotoxicity of the bis(POM) analogues was also increased by virtually similar degrees. Nevertheless, studies with radiolabelled bis(POM)PMEA showed a considerable increase (100-fold) of the cellular uptake of PMEA by using the phosphonate diester(23). In an independent study, ten different PMEA prodrugs tested in vivo for oral bioavailability in cynomolgus monkeys. From the results of this study, bis(POM)-PMEA was selected for human clinical trails. In these clinical trails, the oral bioavailability of bis(POM)-PMEA from a single 500 mg dose was greater than 40% in fed subjects(25). The in vitro antitumor tests of the bis(POM)-5-FdUMP triester demonstrated high growth inhibition in cell lines that were resistant to the usually used drug 5-fluorouracil. Furthermore, this triester derivative showed in vivo bioactivity after intraperitonal application against P 388 leukemia in mice. Beside the antitumor activity, these results verified also the penetration of the pronucleotide through cellular membranes(24). A drawback of this approach is the requirement of a second identical activation reaction because the intermediate phosphodiester is a significantly poorer substrate for the activating carboxyesterases. Moreover, the delivery of one molecule of the nucleotides results in the liberation of two equivalents of potentially toxic formaldehyde and pivalinic acid (Figure 4). Additionally, it has been shown that the bis(POM) phosphotriesters 10 were chemically unstable and highly susceptible to serum-mediated hydrolysis, factors which limit their potential utility for intracellular drug delivery(22,24a). A modification of the bis(POM)-approach and a way to overcome the limitations observed with these compounds are bis(isopropyloxycarbonyloxymethyl)[bis(POC)] nucleotides of general type 14 (shown as the PMPA derivative) that have been published very recently by A. Fridland et al.(26) and L. Naesens et al.(27). This modification uses a carbonate diester within the masking group. The mechanism of action is again a carboxyesterase-catalyzed cleavage of the isopropyl ester to give isopropanol and in the shown example the intermediate phosphonate diester 15 that fragmentates via 16 into carbon dioxide and formaldehyde and finally to the mono(POC) phosphonate ester 17. The mono(POC) ester 17 is subsequently degraded to yield PMPA or generally the nucleotide after a second carboxyesterase activation or by phosphodiesterase cleavage (Figure 5). In contrast to the bis(POM)-approach, the bis(POC) modification avoids the formation of two equivalents of pivalinic acid that accumulate in the cells and potentially cause toxicity. The bis(POC)-approach was applied to the anti-HIV drug PMPA in order to increase the low bioavailability of this phosphonate derivative (Figure 5). Bis(POC)-PMPA is chemically stable at low pH and has shown 30% bioavailability in dogs with minimal toxicity in repeat 5-day dosage of 60 mg/kg/day. The antiHIV activity of bis(POC)-PMPA 14 in human peripheral blood lymphocytes and in dentritic T-cell coculture system was 35and 16fold, respectively, higher than that of PMPA. Bis(POC)-PMPA 14 was non-toxic at concentrations that completely suppressed viral replication. Studies of the metabolism of [3H]-bis(POC)-PMPA showed that is was readily taken into the human cells, hydrolyzed to PMPA, and phosphorylated to the monoand diphosphate derivatives. These results show that bis(POC)-PMPA 14 is a membrane-permeable form of PMPA and shows promise as a drug for the treatment of HIV infections(26). Bis(SDTE)and Bis(SATE)-Nucleotides Two further approaches that are also based on enzymatic activation have been reported by J.-L. Imbach and G. Gosselin(28). They designed the bis(S-[2-hydroxyethylsulfidyl]-2-thioethyl)[bis(SDTE)] 18 and the bis(S-acyl-2-thioethyl)[bis(SATE)] nucleotides 19. The former concept was constructed to take advantage of the greater reducing potential within the cells to liberate the nucleotide into the cytosol. After T hi s do cu m en t w as d ow nl oa de d fo r pe rs on al u se o nl y. U na ut ho riz ed d is tr ib ut io n is s tr ic tly p ro hi bi te d. 236 C. Meier SYNLETT the reductase cleavage of the disulfide bond in 18 with formation of thioethanol and thioethyl phosphotriester 20, this compound eliminates spontaneously episulfide to yield the intermediate phosphodiester 21. As in the case of the bis(POM)(10) and the bis(POC) phosphotriester 14, the bis(SDTE) derivative 18 requires a second identical enzymecatalyzed activation step (Figure 6). The initial idea of this concept was, that the "soft" thiol-nucleophile in 20 will attack the "soft" electrophilic α-carbon atom rather than the "hard" phosphorus atom in the elimination reaction avoiding possible pseudorotation processes(18) because all reactions take place within the masking group without involving the phosphate ester moiety. This concept belongs also to the class of tripartate prodrugs. It could be shown that the bis(SDTE) triester of the nucleotides ddUMP and AZTMP 9 were cleaved 30-fold faster by reductases in cell extracts than in culture medium(28c). Furthermore, the bis(SDTE) approach has been applied to the nucleoside 5-FdU 3(29) and to the phosphonate nucleotide PMEA(30). Whereas an improvement in antiviral activity most probably due to an enhanced bioavailability for the bis(SDTE)-PMEA derivative 18 compared with PMEA was found, the bis(SDTE)-5-FdUMP failed to show better antitumor potency as compared to the free nucleoside(29). The major limitation of this approach is the restricted chemically stability and high susceptibility to serum-mediated hydrolysis (Figure 6). In contrast to the bis(SDTE) approach, the bis(SATE) approach is similar to the bis(POM) concept with respect to the activation process. Again, carboxyesterases are used to trigger the nucleotide delivery. Here, the enzyme cleaves a thioester in 19 to yield a carboxylic acid (as in the bis(POM) approach pivalinic acid) as well as the above mentioned thioethyl phosphonate diester 22 which undergoes a fragmentation to episulfide and the phosphonate monoester 23 (Figure 6). Thus, the spontaneous, second step is identical in the bis(SDTE) and the bis(SATE) concepts. Nevertheless, again two independent activation steps are required to yield the free nucleotide. With these two approaches, two equivalents of episulfide, which has been shown to be chronically and acutely toxic in mice and rats, are released besides the nucleotide(31). The toxicity and moreover also mutagenicity has also been observed in vitro(32). J.-L. Imbach and J.-P. Sommadoussi have shown that the introduction of the SDTE and the SATE group do not induce additional cytotoxicity of the phosphotriesters as compared to the parent nucleosides (e.g. AZT) in human myeloid colony-forming cells (human bone marrow cells)(33). It should be mentioned that the stability of the bis(SATE) phosphotriesters could be adjusted by varying the thio ester residue: the lowest stability was obtained for thio esters of acetic acid whereas with increasing alkyl residue of the carboxylic acid the stability was increased. The optimized compounds are bis(SATE) compounds bearing two S-pivaloyl-2-thioethyl side chains as shown in Figure 6 because the highly lipophilic and sterically demanding t-butyl residue protects the masking group from too rapid cleavage as compared to the S-acetyl or S-i-propyl counterparts(28d,30a,31). Again, J.-L. Imbach et al. demonstrated an increased stability of the bis(SATE) derivatives of AZTMP against degradation in culture medium as compared to cellular extracts(34). The bis(SATE) approach has been successfully applied to the thymidine kinase-bypass (TK-bypass) of d4T 1(35) (shown in Figure 6) and AZT 2(34), the adenosine deaminase-bypass (ADA-bypass) of the anti-HIV active nucleoside ddA 6(36) as well as for an improvement of the bioavailability of the anti-herpes and anti-HIV drug PMEA. Furthermore, the corresponding bis(SATE)-ddAMP triesters exhibited a higher stability than the dideoxynucleoside ddA 6 against acid-catalyzed depurination(36a). It should be mentioned that the bis(SATE)-AZTMP derivatives demonstrated a 10-fold decrease in activity from the wild-type CEM/O cell line to the mutant TKCEM cell line(34). This result is puzzling but has also been observed by others. On the other hand, J. Cinatl, Jr. et al.(11a) have shown that the bis(SATE)AZTMP derivatives retained the biological activity of AZT 2 in AZT resistant Molt4/8 cells. Especially bis(S-pivaloyl-2-thioethyl)-AZTMP showed comparable cytotoxic and antiviral activity in sensitive and resistant cells. With these experiments, J. Cinatl et al. demonstrated that Molt4/8rAZT cells exert resistance to the anti-HIV activity of the drug mainly owing to the lack of AZT phosphorylation and that resistance may be bypassed by using AZT monophosphate prodrugs. However, as for the bis(POM) phosphotriesters 10 reported by D. Farquhar, the intermediate mono(SATE) phosphonate esters 21 and 23, respectively, are significantly lower substrates for the activating carboxyesterases due to the negative charge at the phosphonate residue in the vicinity of the enzyme cleavage site. Before hydrolysis takes place, the bis(SDTE) as well as the bis(SATE) phosphotriesters may serve as lipophilic precursors for a passive membrane permeation. J.-L. Imbach et al. presented the comparison of the bis(SATE)-approach to the isomeric S,S'-bis(O-acyl-2-oxyethyl) phosphorodithiolates(37). Surprisingly, these derivatives proved as active as the original bis(SATE) triesters with the nucleosides d4T 1 and ddA 6, although the delivery mechanism could not be the same. Further work is in progress to study the cleavage mechanism. Recently, the same group extended their prodrug design to bis(S-glycopyranosyl-2-thioethyl) [bis(SGTE)] nucleoside phosphotriesters(38). These compounds were designed to deliver the nucleotide by activation of glycosidases. However, no antiviral data have yet been published.