A short de novo synthesis of nucleoside analogs

See allHide authors and affiliations

Science  07 Aug 2020:
Vol. 369, Issue 6504, pp. 725-730
DOI: 10.1126/science.abb3231

Short path to a complex ring

Nucleotide analogs are valuable tools and therapeutics because of their ability to interfere with processes such as DNA synthesis, which are vital to rapidly dividing cells and replicating viruses. These molecules are challenging to synthesize chemically. Meanwell et al. developed a “ribose last” synthetic strategy in which a fluorinated acyclic nucleic acid is formed by an l- or d-proline–catalyzed aldol reaction (see the Perspective by Miller). This intermediate can then be cyclized to yield the nucleic acid analog in one pot with control of anomeric conformation based on cyclization conditions. Nucleotide analogs accessible by this strategy include those with modifications at C2′ and C4′, purines and pyrimidines, and locked and protected products.

Science, this issue p. 725; see also p. 623


Nucleoside analogs are commonly used in the treatment of cancer and viral infections. Their syntheses benefit from decades of research but are often protracted, unamenable to diversification, and reliant on a limited pool of chiral carbohydrate starting materials. We present a process for rapidly constructing nucleoside analogs from simple achiral materials. Using only proline catalysis, heteroaryl-substituted acetaldehydes are fluorinated and then directly engaged in enantioselective aldol reactions in a one-pot reaction. A subsequent intramolecular fluoride displacement reaction provides a functionalized nucleoside analog. The versatility of this process is highlighted in multigram syntheses of d- or l-nucleoside analogs, locked nucleic acids, iminonucleosides, and C2′- and C4′-modified nucleoside analogs. This de novo synthesis creates opportunities for the preparation of diversity libraries and will support efforts in both drug discovery and development.

As fundamental biomolecules, nucleosides play key roles in diverse cellular processes, ranging from cell signaling to metabolism (1). Not surprisingly, synthetic nucleoside analogs (NAs), which are designed to mimic their natural counterparts, are widely exploited in medicinal chemistry (26) and used as tool compounds in chemical biology. NAs have been in use for more than half a century for the treatment of cancer (2, 5), and they represent the largest class of small molecule antivirals (e.g., 1 to 3, Fig. 1) (3, 4). Mechanistically, after in vivo phosphorylation, NAs operate as toxic antimetabolites (4) and can inhibit enzymes that are crucial to cancer cell growth or viral replication (e.g., DNA or RNA polymerases, ribonucleotide reductases, or nucleoside phosphorylases) (2, 4). Recently, NAs have also demonstrated promise as epigenetic modulators, and NA inhibitors of DNA methyltransferase have been approved for cancer therapy (4).

Fig. 1 Nucleoside analogs: objectives and obstacles.

WHO, World Health Organization; e.r., enantiomeric ratio.

The discovery and development of anticancer and antiviral NAs builds on several decades of medicinal chemistry (5, 7, 8). In the early 1980s, C2′ fluorination was found to improve metabolic stability and influence furanose conformation [PSI-6206 (9): 1] (3, 7, 10). Methylation at C2′ and modifications to C3′ and C4′ functionalization (e.g., methyl, azido, and alkynyl) can also influence furanose conformation and attenuate reactivity of the C3′ and C5′ alcohols toward chain extension (MK-8591: 2) (8, 11). Modified nucleobases can often improve NA potency or drug-like properties, and five-membered ring heterocycles [e.g., ribavirin (3)] (12) have been shown to mimic structurally related intermediates that are involved in de novo purine nucleotide biosynthesis (13). In recent years, there has also been increased interest in unnaturally configured NAs (14, 15). For example, the β-l-NA lamivudine (3TC) has found widespread use in the treatment of HIV-1 and AIDS, and several α-d-thymidine analogs have demonstrated promise as antimalarials (16).

Despite intense efforts, NA synthesis still presents many challenges (17). First, NAs are often synthesized from naturally occurring carbohydrates, which limits patterns of substitution and furanose stereochemistry. Second, the addition of nucleobases to activated ribose derivatives (the Vorbrüggen reaction) often fails or proceeds with poor diastereoselectivity with C2′- or C4′-modified NAs (18, 19). Third, modifications at C2′ often require multistep protecting group manipulations, and efficient strategies for producing C4′-modified NAs do not exist. In fact, a recent summit of key opinion leaders highlighted the synthesis of noncanonical nucleosides as an “emerging area of high potential impact” (20). Although efforts to develop de novo NA syntheses have aimed to address these challenges, the resulting processes are often lengthy and target specific, as highlighted by the 16-step process required to produce the C4′-modified NA MK-8591 (2) (19). As a notable example of de novo NA synthesis, MacMillan and colleagues have reported the synthesis of C2′-modified NAs, including the core of sofosbuvir (1), using a sequence that involved a Mukaiyama aldol coupling between a ketene acetal (5) and an α-oxyaldehyde (6) (21). In this work, we disclose a straightforward NA synthesis that involves a one-pot, proline-catalyzed α-fluorination and aldol reaction (αFAR) of heteroaryl-substituted acetaldehydes 9 followed by reduction or organometallic addition and annulative fluoride displacement (AFD). This concise (two- to three-step) process addresses several major and longstanding challenges in NA synthesis by enabling direct access to C3′- and C5′-protected NAs 10 (and hence C2′-modified NAs), providing flexibility in nucleobase substitution, and offering a direct route to C4′-modified NAs. We expect this strategy will become a powerful tool that will enable and inspire drug synthesis.

In a proposed prebiotic synthesis of DNA (22), couplings between nucleobase-type enamines 11 (Fig. 2A) and glyceraldehyde form a nucleobase iminium ion 12 before the furanose in a ribose-last approach. Our experiences in ribose synthesis from chlorohydrins (23) have suggested that N and Cl hemiaminals are too unstable to serve as precursors to related nucleobase iminium ions. Thus, as a synthetic equivalent to 12, we proposed the fluorinated acyclic NA 13 on the basis of the broad utility of glycosyl fluorides as sugar donors. Control over both relative and absolute stereochemistry in 13 may be possible through an organocatalytic aldol reaction of a dihydroxyacetone derivative (e.g., 8) (24) and the α-fluoroaldehyde 14 (2527). This approach to NAs would require (i) harnessing the reactivity of notoriously unstable α-fluoroaldehydes (28) coupled with the additional challenge of a nucleobase connected via an N and F hemiaminal (e.g., 14) and (ii) developing an AFD reaction to form the ribose ring in the last step.

Fig. 2 A short synthesis of the pyrazolyl NA 17.

(A) Ribose-last approach in a proposed prebiotic synthesis of nucleosides inspired a synthetic strategy to NAs. (B) Proline-catalyzed αFAR, followed by reduction and AFD. (2.5x), 2.5 equivalence. (C) Mechanistic studies of the AFD process. NFSI, N-fluorobenzenesulfonimide; DMF, dimethylformamide; MeCN, acetonitrile; OTf, triflate; equiv, equivalence; ND, not determined.

To explore this process, we investigated the α-fluorination (27) of α-pyrazolyl aldehyde 15 (Fig. 2B) and found that a combination of l-proline and N-fluorobenzenesulfonimide (NFSI) in dimethylformamide (DMF) (29) provided an α-fluorohydrate (table S1) as the sole product. Although we were unable to dehydrate this material, direct addition of dioxanone 8 in acetonitrile (MeCN) afforded the fluorohydrins 16a and 16b in good yield and enantioselectivity (Fig. 2B, entry 2; see table S1 for full optimization details). Fluorohydrins 16a and 16b were formed as an ~1.4:1 mixture of epimers at the pseudo-anomeric carbon (indicated with an asterisk in Fig. 2B) that do not interconvert during the reaction or isolation. This result is consistent with a slow epimerization of α-pyrazolyl-α-fluoroaldehydes that precludes a dynamic kinetic resolution (23) or indicates that the transition structures for the proline catalyzed aldol reaction between dioxanone 8 and (R)- or (S)-α-pyrazolyl-α-fluoroaldehydes are energetically similar. Nevertheless, we anticipated that the cyclization event would proceed through the formation of a transient azacarbenium cation (30), which would render the fluoromethine configuration inconsequential. Toward cyclization (AFD), reduction of the fluorohydrins 16a and 16b provided a mixture of 1,3-syn diols, and subsequent reaction with the fluorophilic Lewis acid Sc(OTf)3 afforded the NA 17 as a single β-anomer (entry 4). Alternatively, treatment of the diols 18a and 18b with base (NaOH) resulted in the formation of a mixture of α- and β-anomeric NAs that varied in composition depending on reaction conditions (entries 5 and 6). Using an excess of NaOH (entry 6), the β-anomer 17 was formed as the exclusive product in excellent yield. To examine the mechanism of cyclization, diols 18a and 18b were separated and their relative stereochemistry assigned by J-based configurational analysis and/or x-ray analysis (see supplementary materials). Subjecting the purified syn-fluorohydrin 18a to NaOH (Fig. 2C) promoted a clean cyclization to the β-anomer 17 (an SN2 process). Similarly, the anti-fluorohydrin 18b cyclized to the α-anomer 19, again through stereochemical inversion. Fortuitously, under these same basic reaction conditions, the α-anomer 19 epimerizes to afford the more stable and naturally configured β-anomer 17 (14, 31). Thus, both fluorohydrin aldol products can be transformed together into a single naturally configured β-d-NA through this straightforward reaction sequence. It is notable that the enantiomeric purity of 17 (entry 6) represents an average of the enantiomeric purities of the epimeric fluorohydrins 16.

To assess the general utility of this NA synthesis, we prepared a collection of acetaldehyde derivatives through the alkylation of several heterocycles with bromoacetaldehyde diethyl acetal (20) (Fig. 3A; see supplementary materials for details). Using either Selectfluor or NFSI as the electrophilic fluorinating agent (F+), aldehydes 21 underwent proline-catalyzed αFAR with dioxanone 8 to provide fluorohydrins 22 functionalized with one of the heterocycles uracil, thymine, triazole, deazadenine, pyrazole, phthalimide, adenine, or 2,6-dichloropyrimidine. In the case of the adenine-containing fluorohydrin, the enantiomeric purity was lowered by competing (nonproline) catalysis in the αFAR. Each of the fluorohydrins was isolated as a mixture of epimers at the fluoromethine center that subsequently underwent a 1,3-syn selective carbonyl reduction and cyclization promoted by either base (Fig. 2B) or a Lewis acid (Fig. 2C), as indicated. We found that several heterocycles were compatible with this process and that uracil-, thymine-, or adenine-substituted acetaldehydes could be exploited in short (four-step total) de novo syntheses of the endogenous ribonucleosides uridine (U: 24), 5-methyluridine (m5U: 25), and adenosine (A: 31). Generally, the optimal Lewis acids for promoting AFD reactions were InCl3 or Sc(OTf)3, whereas pyrazole- and uracil-derived fluorohydrins cyclized using NaOH. With the exception of triazole 28, trifluoromethyluracil 29, and deazaadenines 32 and 33, the NAs were produced as an approximate average of the enantiomeric purities of the precursor fluorohydrins 22. Thus, most NAs examined undergo epimerization after AFD (14), which provides a straightforward means to convert mixtures of epimeric aldol products into naturally configured β-d-NAs. For the trifluoromethyl uracil 29 and deazaadenines 32 and 33, αFAR products (e.g., 22) were reduced, separated, and treated individually with Sc(OTf)3 or InCl3 (Fig. 3C). For trifluoromethyl uracil, only the anti-fluorohydrin underwent AFD to form 29, which did not epimerize under the reaction conditions. With deazaadenine, both the syn-fluorohydrin and anti-fluorohydrin cyclized to provide the β- and α-anomers 32 and 33, respectively, by direct fluoride displacement.

Fig. 3 Scope of nucleoside and NA synthesis.

(A) A simple four-step reaction sequence converts readily available starting materials into enantioenriched and naturally configured β-d-NAs. (B) AFD to produce uracil, thymine, pyrazolyl, and 5-pyrimidinyl nucleosides; and NAs can be promoted by NaOH. (C) AFD to produce trifluoromethyl uracil, triazolyl, phthalimidyl, deazaadenine, and adenosine nucleosides; and NAs can be promoted by the Lewis acids Sc(OTf)3 or InCl3. (D) NAs protected at both the C3′ and C5′ alcohol functions are now readily available. (E) Using d-proline to catalyze the αFAR reaction, unnatural nucleosides (l-enantiomers) are now readily available. (F) Exploiting this convenient process, we prepared a small collection of C2′-modified NAs. aTEMPO, BAIB, CH2Cl2 (92% from 34). bi) thiocarbonyldiimidazole, THF; ii) Bu3SnH, azobisisobutyronitrile (55% over two steps from 35). ci) TEMPO, BAIB, CH2Cl2; ii) MeMgBr, THF, −78°C (80% over two steps from 34). dDAST, CH2Cl2 then HCl, methanol (MeOH) (53% from 35). TEMPO, 2,2,6,6-Tetramethylpiperidin-1-yl)oxyl; BAIB, bis(acetoxy)iodobenzene; THF, tetrahydrofuran; DAST, diethylaminosulfur trifluoride; aq, aqueous; rt, room temperature; SI, supplementary materials; d.r., diastereomeric ratio.

To evaluate the practical utility of these processes, several of the αFARs were demonstrated on a >10-g scale [e.g., 25, 28, 29, 30, and 32 (Fig. 3)] and proceeded without complication. C-linked NA 27 could be prepared using this reaction sequence (HetAr = 4,6-dichloropyrimidine), which further extends the utility of this strategy to an additional and important class of NAs (32). In this case, however, the major product of the aldol reaction cyclizes to an α-d-NA and then undergoes a second cyclization event to form the tricycle 27. In addition to naturally configured NAs, this strategy can be adapted for the synthesis of enantiomeric (l-configured) NAs (Fig. 3E) by using d-proline in the αFAR. Thus, l-uridine (ent-24) and the l-configured NA ent-28 were accessed in a straightforward manner. Crude reaction products were generally treated with aqueous acid to remove the acetonide protecting group, but eliminating this step allowed us to isolate C3′- and C5′-protected NAs directly (e.g., 34 and 35, Fig. 3D). To demonstrate that these acetonide-protected NAs can be further derivatized using standard protocols, several C2′-modified NAs were prepared, including C2′-oxo (36), C2′-deoxy (37), C2′-3° alcohol (38), and C2′-epi (39) (Fig. 3F).

Having ready access to a range of αFAR products, we anticipated that the addition of organometallic reagents (rather than hydride reduction) would lead directly to C4′-modified NAs. Toward this goal, we examined reactions of fluorohydrin 41 with a range of organometallic reagents under a variety of reaction conditions (see supplementary materials). Grignard reagents in CH2Cl2 proved to be most compatible, and executing the reactions at −78°C proved necessary because higher reaction temperatures promoted 1,2-hydride shift and fluoride displacement. These reactions generally gave mixtures of tertiary alcohols with a preference for addition to the re face (24), and when the crude reaction mixture was allowed to warm to room temperature, the intermediate magnesium alkoxide 42a underwent AFD to provide the C4′-modified NA 43 directly. This short sequence enables access to enantiomerically enriched C4′-modified NAs in only three steps from simple achiral heterocycles and bromoacetaldehyde diethyl acetal. Alternatively, quenching the mixture of magnesium alkoxides 42a and 42b with ammonium chloride followed by a subsequent Lewis acid–promoted AFD using InCl3 gave the anomeric α-d-NA 44. In this case, the magnesium alkoxides 42a and 42b cyclize selectively using base or Lewis acid to afford access to α-l– and α-d–configured NAs, which are both unnaturally configured NAs of contemporary interest to medicinal chemists (1416).

These results led us to examine the reaction of several additional Grignard reagents with aldol adducts containing triazole, deazaadenine, thymine, pyrazole, or trifluoromethyluracil functions (Fig. 4A). We found that diastereoselectivity depends on both the solvent and the heterocycle, and reactions in tetrahydrofuran (THF) generally gave mixtures of tertiary alcohols of different composition from those generated in CH2Cl2. The addition of MeMgI to ketofluorohydrins substituted with triazole gave predominantly 1,3-syn-diols that cyclized to the naturally configured NA 48. Thus, subtle differences in chelation structures involving the heterocycle and/or β-alkoxide function and organomagnesium reagents play a notable stereodetermining role in these 1,2-addition reactions. Controlling this aspect of the process to selectively access β-d–configured NAs will be the subject of future studies. In this study, a collection of deazaadenine-substituted NAs 43 to 47 were readily accessed as both α- and β-anomers. In general, and as noted in Fig. 3, base-promoted cyclizations resulted in C3′-OH– and C5′-OH–protected NAs (e.g., 49 to 54), whereas Lewis acid–promoted cyclizations resulted in deprotection or protecting group migration (e.g., 44 and 47). Densely functionalized l-configured C4′-modified NAs could be rapidly accessed from the corresponding ketofluorohydrin aldol adducts, including NAs substituted with methyl, cyclopropyl, aryl, and alkynyl groups (Fig. 4). From this study, it is clear that larger collections of C4′-modified NAs (e.g., focused screening libraries) are now readily available. It is worth highlighting that each of the C4′-methyl, cyclopropyl, p-methoxyphenyl, p-chlorophenyl, and alkynyl NAs 43 to 54 were prepared in only three or four steps total, which compares favorably to contemporary syntheses.

Fig. 4 Rapid synthesis of C4′-modified and other NAs.

(A to E) Diversification and large-scale production of αFAR products provide access to C4′-modified NAs, iminonucleosides, and LNAs to support medicinal chemistry efforts. aYield from keto-fluorohydrin aldol adduct. bCombined yield of diastereomers. cProduct after heating of crude reaction mixture to 50°C with camphorsulfonic acid (CSA) and dimethoxyacetone. dProduct after treatment of crude reaction mixture with aqueous HCl. eStarting from a single fluorohydrin 59.

Considering the potential for this process to affect the large-scale production of NAs, we examined the synthesis of the d-uridine derivative 56 starting with 900 g of uracil. Without additional optimization, we were able to generate ~380 g of the aldol adduct 55 (Fig. 2B), which was converted into the protected uridine 56 in excellent yield. Oxidation of the C2′-OH function followed by deprotection and addition of MeMgBr in THF gave the tertiary alcohol 57. This later compound is a previously reported intermediate in the large-scale production of MK-3682 (uprifosbuvir: 58) (33), an HCV NS5B RNA polymerase inhibitor developed for the treatment of hepatitis C virus (HCV).

We also briefly assessed the utility of this process for accessing an unusual class of NAs known as iminonucleosides, wherein the furanose oxygen is replaced by a nitrogen atom. In a single example (Fig. 4C), we observed that reductive amination of the fluorohydrin aldol adduct 59 (isolated as a single diastereomer, as shown) using benzyl amine and followed by a basic work-up led directly to the β-d–configured iminonucleoside 60 in good yield. To further demonstrate the advantages of this strategy, we prepared a C4′-modified deoxy NA (Fig. 4D). In this case, C4′-allyl ribothymidine 61 was accessed through the addition of allylmagnesium bromide to the fluorohydrin 59 followed by base-promoted AFD. A Barton-McCombie deoxygenation then gave the C4′-allyl NA 62 in only six steps from thymine.

Finally, we aimed to exploit this facile C4′-functionalization strategy in the preparation of locked nucleic acids (LNAs) (34). These conformationally restricted NAs demonstrate improved stability, and their incorporation in antisense oligonucleotides can lead to substantial increases in specificity and potency. We evaluated the addition of alkynylmagnesium chloride to the thymine-containing aldol adduct 59 and found that the reaction gave two diastereomeric addition products 63 and 64. The major product was transformed directly into the unusual LNA 65 by simply reacting with NaOH, which promoted both the AFD reaction and a subsequent cyclization between the free alcohol function and alkyne. This four-step total synthesis is comparable to the 23-step route reported for the analogous uracil LNA 67 (35). We were also able to generate the unusual alkyne-functionalized LNA 68, an unreported scaffold in nucleoside chemistry, by simply effecting an AFD of the 1,2-addition product 64. From here, formation of the 2,2′-anhydrothymidine followed by deprotection and treatment with base in warm DMF (36) gave the LNA 68. This scaffold is primed for further diversification through standard click or Sonagashira coupling reactions.

The demonstration of several αFARs on scales >10 g and up to 400 g supports the use of this strategy in the process scale production of NAs. Several generations of medicinal chemistry and total synthesis have provided reliable templates for nucleoside and NA synthesis; however, this study provides opportunities that should influence the construction of diversity libraries and support future efforts in both drug discovery and development.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S14

Tables S1 to S12

Liquid Chromatography Chromatograms

NMR Data

References (3752)

MDAR Reproducibility Checklist

References and Notes

Acknowledgments: The authors thank D. McKearney (Simon Fraser University) and J. A. Newman (Merck & Co., Inc.) for carrying out x-ray crystallographic analyses. Funding: R.B. acknowledges support from the Canadian Glycomics Network (Strategic Initiatives Grant, CD-46), the Natural Sciences and Engineering Research Council (NSERC) of Canada (Discovery Grant, RGPIN-2019-06468), and Merck & Co., Inc. M.M. was supported by an NSERC CGSD2 scholarship, and J.L. was supported by the Deutsche Forschungsgemeinschaft (DFG). Author contributions: M.M. and S.M.S. developed the methodology; M.M., S.M.S., J.L., B.A., and Y.W. performed the experiments; R.C. developed a protocol for assigning stereochemistry by nuclear magnetic resonance (NMR) spectroscopic and computational methods; L.-C.C. and R.B. were responsible for project administration; R.B. conceptualized the research; L.-C.C. and R.B. supervised the execution of experiments; and R.B. wrote the original draft. Competing interests: Simon Fraser University and Merck & Co., Inc. have filed a patent application describing the synthesis of nucleoside analogs via the process presented in this manuscript—U.S. provisional patent application no. 62/994,349. Data and materials availability: All data are available in the manuscript or the supplementary materials. X-ray structures are deposited at the Cambridge Crystallographic Data Centre under reference numbers 1955427, 2008890, and 1955420.

Stay Connected to Science


Navigate This Article