Synthesis and Evaluation of Prodrugs of α-Carboxy Nucleoside Phosphonates

A range of lipophilic prodrugs of α-carboxy nucleoside phosphonates, potent inhibitors of HIV-1 reverse transcriptase without requiring prior phosphorylation, were synthesized to evaluate their in vivo potency against HIV in cell culture. A series of prodrug derivatives bearing a free carboxylic acid where the phosphonate was masked with bispivaloyloxymethyl, diisopropyloxycarbonyloxymethyl, bisamidate, aryloxyphosphoramidate, hexadecyloxypropyl, CycloSal, and acycloxybenzyl moieties were synthesized, adapting existing methodologies for phosphonate protection to accommodate the adjacent carboxylic acid moiety. The prodrugs were assayed for anti-HIV activity in CEM cell cultures—the bispivaloyloxymethyl free acid monophosphonate prodrug exhibited some activity (inhibitory concentration-50 (IC50) 59 ± 17 μM), while the other prodrugs were inactive at 100 μM. A racemic bispivaloyloxymethyl methyl ester monophosphonate prodrug was also prepared to assess the suitability of the methyl ester as a carboxylic acid prodrug. This compound exhibited no activity against HIV in cellular assays.


■ INTRODUCTION
Recently, we described the design, synthesis, and evaluation of a novel class of α-carboxy nucleoside phosphonates (α-CNPs), e.g., 1 (Figure 1), which are potent inhibitors of HIV-1 reverse transcriptase (RT) in cell-free assays. 1−5 In addition, this class displays activity against a range of viral polymerases including Herpes virus DNA polymerases. 4 Significantly, these phosphonates do not require phosphorylation to exhibit activity, in sharp contrast to other phosphononucleosides, which have been designed as nucleoside monophosphate mimics and thus require further phosphorylation to the triphosphate in the host cells before they can be active. 6 We have demonstrated that the presence of the phosphonoacetic acid moiety is crucial to their activity, 7 and indeed structural biology investigations have shown that the phosphonoacetic acid moiety binds to the metal site in HIV-1 RT mimicking the triphosphate binding in the natural nucleotide triphosphates. 2 However, these molecules, containing both the phosphonic acid and carboxylic acid moieties, are too polar to efficiently migrate across the cell membrane as they are ionized at physiological pH.
There has been intense activity in the last few decades to develop prodrugs for the phosph(on)ate moiety to facilitate transport of these highly charged molecules into the cell. A number of recent accounts document the tremendous growth in this field. 8−11 In spite of the efforts expended, to date, the only monophosphonate prodrugs clinically approved for HIV treatment are the diisopropyloxycarbonyloxymethyl (POC) ester of tenofovir fumarate and the alafenamide derivative of tenofovir (TAF). 12−15 The bispivaloyloxymethyl (POM) prodrug of adefovir has been approved for the treatment of hepatitis B virus (HBV) infection ( Figure 2). 16,17 Naturally, these two prodrug approaches (POM, POC) which have well-documented clinical efficacy aroused our interest when seeking viable lipohilic forms for α-CNPs. Other strategies such as the ProTide (aryloxyphosphoramidate) approach developed by McGuigan, 18−23 utilized in the prodrug Sofosbuvir (Figure 2), 24 and the related bisamidate prodrugs, 25 which release innocuous amino acids as byproducts, were also attractive targets. The highly lipophilic alkoxyalkyl prodrugs [e.g., hexadecyloxypropyl (HDP)] pioneered by Hostetler were also considered as potential protecting groups for the phosphonate moiety of the α-CNPs. 26−32 Additionally, the cyclic, salicyl alcohol-derived "CycloSal" prodrugs described by Meier were of particular interest given that the CycloSal moiety undergoes a pH-dependent chemical hydrolysis to unmask the phosphonate rather than enzymatic hydrolysis; the stability of the CycloSal prodrugs can be altered by modification of substituents on the aromatic ring. 33−39 A further prodrug moiety of interest, also developed by Meier, was the acyloxybenzyl phosphonate. Acyloxybenzyl prodrugs have been used with nucleoside phosphates, 40 diphosphates, 41,42 and triphosphates, 43 and also with phosphonates. 44,45 In this case, the deprotection of the phosphonate function is triggered by enzymatic hydrolysis of the acyl group, which is remote from the carboxyphosphonate, followed by spontaneous loss of the resulting hydroxybenzyl groups to reveal the free phosphonate.
Most of the above approaches rely on enzymatic hydrolysis of the lipophilic phosphonate esters once they are within the cell to release the phosphonates. However, the majority of this work was developed utilizing phosphonates with an unsubstituted methylene linking the phosphorous and oxygen. Hence, one of the key issues to be addressed was whether the phosphonate esters of the α-CNPs would undergo enzymatic hydrolysis or if the efficiency of release of the free phosphonate would be affected by the presence of the carboxylic acid moiety at this carbon. Accordingly, Meier's CycloSal protecting strategy was of particular interest. The various prodrug approaches are summarized in Figure 3.
In addition to protecting the polar phosphonate moiety, protecting the carboxylic acid was also briefly explored, to enhance the lipophilicity of the highly polar α-CNPs. Prodrugs for carboxylic acids have not been as widely documented, with POM-derivatized prodrugs being the most widely known (e.g., Pivampicillin). 46 More recent advances in this area have been the development of the dioxolenone moiety reported by Cheng and co-workers 47 and the N-alkyl-N-alkyloxycarbonylaminomethyl (NANAOCAM) derivatives of carboxylic acids described by Majumdar. 48 ■ RESULTS AND DISCUSSION Our investigations of the prodrug strategies mostly utilized the racemic thymine α-CNP 1, which displayed excellent inhibitory activity against HIV-RT, as a model compound. While we have demonstrated that the activity in the α-CNPs resides predominantly in the L-series, 1,2 for the prodrug investigations, we employed the racemic compounds. As with our original α-CNPs, each of the compounds is generated as an equimolar mixture of diastereomers at the center adjacent to the phosphonate. Extension to other α-CNPs can be envisaged building on the methodologies developed below.
The principal synthetic challenge was to modify the established protocols to generate lipophilic phosphonate  prodrugs for use with the α-CNPs bearing the α carboxylic acid function. As the inclusion of the carboxylic acid moiety adjacent to the phosphonate is not very common, existing prodrug strategies have not been developed to take on board the potential impact of the carboxylic acid moiety on either the introduction of the lipophilic group or the hydrolysis of the prodrug in vivo to release the active compound�both are likely to be impacted by the reactive carboxylic acid proximal to the phosphonate. A small number of prodrug derivatives of phosphonoacetates 49 and other phosphono-carboxylic acid compounds 50 have been described in recent years.
Synthesis and Evaluation of Bis-POM Methyl Ester Monophosphonate Prodrug 2. Our initial focus centered on compound 2, the bis-POM methyl ester monophosphonate prodrug of 1. Because of the ubiquitous presence of esterases in blood, tissues, and organs, we were hopeful that the bis-POM prodrug 2 would be sufficiently lipophilic to penetrate the cell membrane whereupon the methyl ester would undergo hydrolysis by esterases to give 3. The pivaloyloxymethyl moiety could then detach to unmask the phosphonate to give the fully deprotected α-CNP 1 (Scheme 1). 51 It may also be possible that the activation steps are reversed i.e., the POM moieties are first hydrolyzed to generate 4 and subsequent methyl ester hydrolysis affords 1. In parallel, investigation of the free acid prodrug 3 was envisaged, to obviate the necessity for methyl ester hydrolysis to release 1 inside the cell, with the expectation that 3 would still be sufficiently lipophilic to pass into cells. The synthesis of compound 5, a key intermediate in the production of α-CNPs has already been described by us. 1 Microwave-mediated hydrolysis of 5 in the presence of trimethylsilyl bromide 1 generated the TMS-phosphonate 6, which was subsequently hydrolyzed in situ by stirring in methanol−water for 30 min to give 4, which was reacted directly with POM iodide and Hunig's base in tetrahydrofuran (THF) to furnish the racemic bis-POM target 2 in 94% yield from 5 after chromatography. When POM chloride was employed in place of POM iodide, much lower yields were obtained. A similar approach was employed to convert the 5fluorouracil CNP analogue 7 to the bis-POM derivative 8 (Scheme 2).
Compound 2 was found to be inactive against HIV-1 in cellular assays, indicating a need to make prodrugs with the free carboxylic acid to avoid the need for multiple enzymatic hydrolyzes to release the α-CNPs. Attempts to selectively hydrolyze the methyl ester of 2 while preserving the prodrug appendages by both chemical and enzymatic (using commercially available esterases) means to isolate 3 were unsuccessful.
Synthesis of Free Acid Prodrugs. Undeterred, we revisited our strategy for making "free acid" prodrugs of 1, preferring early introduction of a benzyl ester in the sequence, which would allow a late-stage unmasking of the carboxylic acid by hydrogenolysis (Scheme 3). We were hopeful that the hydrophobic prodrug attachments would make the α-CNP sufficiently lipophilic to transport into the cell, despite the presence of the free carboxylic acid.
Thus, commercially available 7 was converted in two steps to 11 via the dimethyl benzyl ester 10. 52 Diazo transfer to 10 was achieved using 4-dodecylbenzenesulfonyl azide (DBSA) to give the diazo phosphonoester 11 (27% yield over two steps). 53 Use of DBSA for the diazo transfer was preferable to use of 4-acetamidobenzenesulfonylazide (ABSA) due to the ease of separating the reaction products by chromatography (the sulfonamide byproduct of ABSA co-elutes with 11). We then employed a rhodium-catalyzed O−H insertion reaction, which we have developed for the synthesis of α-CNPs and related compounds, 1,7,54 to access the allylic acetate 13 in 68% yield. In an improvement over our previously reported conditions, this reaction was carried out using rhodium pivalate, in dichloromethane (DCM) rather than benzene. Replacement of the methyl ester with the benzyl ester had only a modest impact on the efficiency of the key O−H insertion step. Introduction of the nucleobase thymine to 13 was facilitated by a palladium-catalyzed Tsuji−Trost reaction. 55 Thus, an aqueous acetonitrile solution of the sodium salt of thymine was treated with a degassed solution of 13, bis(dibenzylideneacetone)palladium(0) Pd(dba) 2 and bis-(diphenylphosphino)butane (dppb) in acetonitrile under microwave irradiation to generate the insertion product 14 in 27% yield. The benzyl dimethyl phosphonate ester 14 is a key intermediate in this sequence as it allows access to several of the "free acid" prodrugs. In contrast to the saturated phosphonate 5, the amine base lutidine is required along with TMSBr to effect selective hydrolysis of the phosphonate esters of 14 which have an alkene present in the ring. 1 Bis-POM modification was carried out as described earlier for 5 to give the bis-POM benzyl ester compound 15 in 14% yield following chromatography. Finally, concomitant hydrogenation of the cyclopentyl alkene and hydrogenolysis of the benzyl ester of 15 was achieved under a balloon of hydrogen in the presence of 5% Pd/C to furnish the racemic bis-POM prodrug 3 bearing the free carboxylic acid moiety. Under similar conditions, using POC-iodide, the bis-POC prodrug 19 was likewise prepared in two steps from 14.
We were also able to prepare 15 and 16 from 14 in a convenient one-pot reaction. 56 Heating a solution of 14, sodium iodide and POM-Cl or POC-Cl in acetonitrile in a sealed tube for 48 h gave 15 (46%) and 16 (36%) respectively, after chromatography. This method is preferable since it does not require the hydrolysis and isolation of the intermediate phosphonic acid which can be time-consuming and laborintensive; additionally, this reaction sequence proceeds using the chlorides of the carbonyloxymethyl reagents, thus the additional steps to prepare the analogous iodides are not necessary.
The bisamidate prodrug 20 was also synthesized in two steps from 14. Heating the intermediate TMS-phosphonate Scheme 3. Synthesis of Bis-POM, Bis-POC, Bisamidate, and Aryloxyamidate Prodrugs a a prepared from 14 with L-Ala-O i Pr in the presence of 2,2dithiodipyridine (Aldrithiol-2) and triphenylphosphine in pyridine at 90°C gave the bis-substituted product 17 in 34% yield, which is in line with the modest yields reported in the literature. 25 As before, simultaneous saturation of the alkene and hydrogenolysis of the ester using Pd/C under a balloon of hydrogen was carried out to afford the free acid bisamidate prodrug 20. The phenoxyamidate 18 was prepared in a similar manner to 17 except that phenol was added along with L-Ala-O i Pr (14/phenol/Ala 1:4.8:2.1 ratio) in the reaction of the TMS derivative of 14. Hydrogenation/hydrolysis of 18 gave the phenoxyamidate monophosphonate prodrug 21 in 65% yield, as a complex diastereomer mixture. When the phenoxyamidate reaction was carried out on a fivefold scale (14/phenol/Ala 1:5.1:2.0 ratio) it was possible to isolate both 17 (7%) and 18 (21%) by chromatography.
We had initially also hoped to install the bis-HDP moiety immediately after the base insertion step (i.e., from compound 14), however, several attempts following literature procedures were unsuccessful. Thus, we resolved to introduce the HDP group earlier in the sequence to overcome this issue (Scheme 4). Reaction of the O−H insertion product 13 with TMSBr and lutidine under microwave conditions generated the TMSphosphonate 22 which was not isolated, but treated with oxalyl chloride and dimethylformamide (DMF) (cat.) in DCM. Evaporation of the solvent gave the dichloridate 23 which in the presence of HDPOH and diisopropylethylamine at 0°C gave the desired bis-HDP phosphonate 24 in 30% yield following chromatography. 57 The conversion of 13 to 24 via the TMS-phosphonate 22 and dichloridate 23 could be easily monitored by 31  The palladium-mediated base insertion reaction was carried out under microwave conditions as previously described to give 25 in 34% yield. Concomitant hydrogenation of the cyclopentyl alkene and removal of the benzyl group by hydrogenolysis under an atmosphere of hydrogen in the presence of 5% Pd/C afforded the target prodrug 26 in 90% yield.
CycloSal Prodrugs. We next decided to examine the CycloSal prodrug moiety, developed by Meier and co-workers, although CycloSal phosphonates have been found to be somewhat more labile than similar phosphates. 33 The latestage hydrogenation approach seemed unlikely to be compatible with the CycloSal function so we adopted a different approach to this compound, involving the use of a tert-butyl ester which could be cleaved under acidic conditions. Notably, Meier described the trifluoroacetic acid (TFA) deprotection of monomethoxytrityl (MMTr)-protected PMEA CycloSal derivatives, indicating that the CycloSal phosphonate moiety can survive acidic conditions late in the sequence. 33 Thus, the synthesis started from commercially available tertbutyl dimethylphosphonoacetate, which was converted to the corresponding diazo compound 27 and reacted with the acetoxy alcohol 12 to afford the O−H insertion product 28 (Scheme 5). While this O−H insertion reaction could be catalyzed by Rh 2 (OAc) 4 or Rh 2 (piv) 4 , both of these catalysts also led to formation of a substantial amount of an unidentified side-product, which hampered purification, but the desired product was isolated in 66% yield. Use of Cu(OTf) 2 instead of rhodium catalysts afforded the desired product without forming the undesired side-product, although the eventual yield was slightly lower (41%). Purification of the product from the copper-catalyzed reaction was much more straightforward on a multi-gram scale, so this was the preferred method of preparation of 28. With the intermediate 28 in hand, the base insertion reaction with thymine was carried out, to give 29. In this case the reaction was most conveniently performed in DMF under thermal conditions, rather than under microwave irradiation, typically affording 50−60% yield following evaporation of the solvents and purification by chromatography. Product 29 could also be isolated by extractive workup, although sometimes the product was isolated with a diastereomer ratio far from 1:1; in these cases, the Scheme 4. Synthesis of Bis(hexadecyloxypropyl) Prodrug a a diastereomers could be equilibrated using a catalytic amount of triethylamine. Hydrogenation of 29 to afford the key intermediate 30 was straightforward using a balloon of hydrogen and 5% palladium on carbon. The CycloSal group was installed in a similar manner to that described by Meier et al.; 33 the dimethyl phosphonate group was deprotected using TMSBr, in the presence of 2,6-lutidine to prevent cleavage of the tert-butyl ester, and the resulting bis-TMS intermediate was treated with oxalyl chloride and a catalytic amount of DMF to give the dichloridate, which was then reacted with the relevant salicyl alcohol 31−33 to give the penultimate intermediates 34−36 in low to moderate yields and with variable diastereomer ratios, which could be observed in the NMR spectra of the products. Three salicyl alcohols were used, 3tert-butyl 31, 3-methyl 32, and 3,5-dimethyl 35, while the attempted preparation of the 5-chloro analogue was unsuccessful. Treatment of 34−36 with TFA in CH 2 Cl 2 , starting in an ice bath and allowed to warm slowly overnight, afforded the free acid compounds 37−39 in essentially quantitative yields after evaporation of the volatiles and drying in vacuo.
Bis(Acyloxybenzyl)phosphonate Prodrug. The last prodrug derivative examined in this study was the 4nonanoyloxybenzyl phosphonate 43 (Scheme 6). As discussed above, the cleavage of the acyloxybenzyl group is initiated by enzymatic hydrolysis of the acyl groups, which leads to spontaneous expulsion of quinone methides to leave the free phosphonate. This was attractive to us since the initial hydrolysis would occur remotely from the α-CNP function, in principle overcoming concerns that the α-CNP group itself may be incompatible with cellular enzymes. The bis-(nonanoyloxybenzyl) α-CNP derivative 42 was prepared in a similar way to the bis-POM derivative 2 using the one-pot procedure, from the tert-butyl dimethyl phosphonate 30 by reaction with 4-nonanoyloxybenzyl chloride 41 (prepared from 4-hydroxybenzyl alcohol via acylation to 40 followed by conversion to the chloride using thionyl chloride) and sodium iodide in acetonitrile. The carboxylic acid was revealed by TFA hydrolysis to give the desired product 43 in 40% yield; the yield of 43 from the hydrolysis reaction was a balance between incomplete reaction and hydrolysis of the phosphonate function; difficulties in purification meant that the isolated product 43 was approximately 80% pure, judged by the NMR spectra.
Throughout this investigation, the lipophilic phosphonate prodrugs are generally isolated as mixtures of diastereomers; in the NMR spectra, the signals for the PCH (δ H ∼ 4.5 doublet, J ∼ 20 Hz, and δ C ∼ 74 doublet J ∼ 150 Hz) are particularly characteristic and enabled the assignment of the diastereomeric ratios.
Biochemical Evaluation of Free Acid Prodrugs. Unfortunately, almost all of the prodrug derivatives were inactive against HIV-1 in CEM cell cultures, although 3 displayed limited activity (inhibitory concentration-50 (IC 50 ) 59 ± 17 μM) against HIV-1 in the cellular assays.
The reason for such poor, if any, activity of the prodrugs with the free acids may be explained by (a) lack of sufficient lipophilicity to facilitate significant transportation of the molecule into the cell in the presence of the free carboxylic acid or (b) release of the α-CNP 1 in the cell is hindered by the aforementioned acid function. In fact, to date, no activity of nucleoside phosphonates in which an additional negative charge is present near the phosphonate moiety has been documented, to our knowledge. However, more data is required to fully elucidate the mechanistic basis for the lack of activity. Clearly, with the methyl ester in place the prodrugs were similarly inactive which presumably indicates that the enzymatic release of the free α-CNP is hindered by the presence of the ester moiety relative to the standard phosphononucleoside POC and POM derivatives. Interestingly, in our earlier investigations, the methyl esters of the α-CNPs, both the saturated and unsaturated T-α-CNPs, were inactive against HIV-1 RT, 7 indicating that enzymatic ester hydrolysis to release the free acid might not proceed in vivo, and is a prerequisite for any eventual antiviral activity in drugexposed cells. These results indicate the need for a separate prodrug for the carboxy group in conjunction with the phosphonate protection or a tether to link the phosphonate and carboxy groups to facilitate cell penetration.

■ CONCLUSIONS
In an effort to evaluate the potency of α-CNPs against HIV-1 in vivo (intact virus-infected cells), we synthesized a number of prodrugs, demonstrating methodologies for the attachment of a range of lipophilic groups to the phosphonate, with and without protection at the adjacent carboxylic acid, building on earlier work by a number of teams using simple phosphonates, without the complication of an adjacent carboxylic acid moiety. Key to this was the use of the α-diazophosphonates bearing either a benzyl or tert-butyl ester enabling selective deprotection at this group late in the synthetic sequence. The bis-POM methyl ester prodrugs 2 and 8 were inactive against HIV-1. Among the "free acid" prodrugs 3, 19−21, 26, 37−39, and 43, only the bis-POM derivative 3 displayed a certain degree of poor activity against HIV-1 in cellular assays. ■ EXPERIMENTAL SECTION General Procedures. Solvents were distilled prior to use as follows: dichloromethane was distilled from phosphorus pentoxide and/or calcium hydride; ethyl acetate was distilled from potassium carbonate. Benzene was dried before use with activated 4 Å molecular sieves. For O−H and base insertion reactions, solvents were degassed by purging with nitrogen. Organic phases were dried using anhydrous magnesium sulfate. All commercial reagents were used without further purification. Microwave reactions were carried out in closed vessels using a CEM Discover SP in conjunction with Synergy software; reaction temperatures were measured by IR sensor. 1 H, 13 C, and 31 P spectra were recorded at 20°C on 300, 400, or 600 MHz spectrometers. 1 H and 13 C chemical shifts are given in ppm (δ), referenced to solvent signals. 31 P chemical shifts are referenced to H 3 PO 4 (external standard), and 19 F chemical shifts are referenced to C 6 F 6 . Coupling constants (J) are given in hertz (Hz). In some cases, in 13 C NMR spectra of diastereomer mixtures, several resonances with phosphorus coupling were very closely spaced and have been designated as multiplets (m) with a chemical shift range. Highresolution mass spectra (HRMS) were recorded on a time-of-flight spectrometer in electrospray ionization (ESI) mode. Column chromatography was performed using silica gel 60. Thin-layer chromatography (TLC) was carried out on precoated silica gel plates (60 PF254). Visualization was achieved by UV (254 nm) detection and/or staining with vanillin, permanganate, or ceric ammonium molybdate. cis-4-Hydroxy-2-cyclopentenyl acetate 12, 3,58−60 ABSA, 61 HDP-OH, 62 POM-I, 63 POC-Cl, 64 and POC-I 65 and the salicyl aclohols 31−33 66−68 were prepared by literature methods. Trimethyl phoshonoacetate and tert-butyl dimethylphosphonoacetate were purchased from TCI; 4-hydroxybenzyl alcohol was purchased from Merck.
Cellular HIV-RT Assays. The procedure to determine the anti-HIV-1(IIIB) activity in human lymphocytic CEM cell cultures has been described before. 19,33 Briefly, 200 μL of CEM cell cultures (250,000−300,000 cells/mL) was seeded in the wells of 96-well microtiter trays and exposed to 100 CCID 50 (cell culture-infective dose, 50) HIV-1 (strain IIIB)/mL. The wells contain a serial dilution (5-fold) of the test compounds at 100 μM as the highest compound concentration. After 4 days of incubation at 37°C, HIV-1-induced syncytia formation in the absence (control) or presence (test) of the test compounds was microscopically examined. The inhibitory concentration-50 (IC 50 ) was defined as the test compound concentration required to inhibit virus-induced syncitium formation by 50%.

i s -1 -( 4 -( ( C a r b o x y ) b i s ( p i v a l o y l o x y m e t h y l )phosphonomethoxy)cyclopentan-1-yl)thymine 3 (Bis-POM Tα-CNP Free Acid). cis-1-(4-((Benzyloxycarbonyl)bis-
(pivaloyloxymethyl)phosphonomethoxy)cyclopentan-1-yl)thymine 15 (46 mg, 0.069 mmol) was dissolved in methanol (5 mL) and flushed with nitrogen. Palladium on carbon 5% (5 mg) was added, and the suspension was stirred under a hydrogen-filled balloon at room temperature for 24 h. The reaction mixture was filtered using a syringe filter and concentrated to give product 3 as a colorless film (39 mg, ∼quantitative) as a roughly equal mixture of diastereomers. 1

cis-1-(4-((Carboxy)bis(isopropoxycarbonyloxymethyl)phosphonomethoxy)cyclopentan-1-yl)thymine 19 (Bis-POC Tα-CNP Free Acid).
A mixture of 16 (59 mg, 0.088 mmol) and palladium on carbon (5 mg, 10%) in methanol (5 mL) was stirred under a balloon of hydrogen at atmospheric pressure for 18 h at room temperature. After this time, analysis of a small sample by 1 H NMR spectroscopy revealed reduction of the double bond but the continued presence of the benzyl ester. Fresh catalyst (5 mg) was added, and the mixture was stirred for an additional 6 h under a hydrogen balloon, after which time the reaction was found to be complete. The mixture was filtered through a syringe filter and concentrated to give 19 as a colorless film (44 mg, 86%). 1

i s -1 -( 4 -( C a r b o x y -b i s ( 3 -( m e t h o x y ) p r o p y l )phosphonomethoxy)cyclopent-2-en-1-yl)thymine 26 (Bis-HDP T-α-CNP Free Acid).
A mixture of 25 (27 mg, 0.027 mmol) and 5% palladium on carbon (5 mg) in methanol (5 mL) and CH 2 Cl 2 (1 mL) was stirred under a balloon of hydrogen at atmospheric pressure for 18 h at room temperature. The mixture was filtered through a syringe filter and concentrated to afford 22 mg (90%) of 26 as a colorless waxy solid. 1 (50 mL) and Et 2 O (50 mL) were added. The organic layer was separated, and the aqueous layer was extracted with Et 2 O (5 × 20 mL). The combined organic phases were washed with brine (10 mL), dried over MgSO 4 , concentrated, and eluted over a short plug of silica gel (50% EtOAc/hexanes) to afford 8.2 g (65%) of 27 as a yellow oil with spectroscopic properties in accordance with the literature. 70 (30 mL) was purged with nitrogen gas for 10 min, after which activated 4 Å molecular sieve beads (ca. 1 mL) were added, and the mixture was stirred gently. After 1 h, Rh 2 (piv) 4 (34 mg, 56 μmol, 1 mol %) was added, the mixture was placed in a preequilibrated bath set to 60°C, and stirring was continued under reflux overnight. The mixture was cooled, decanted, concentrated under reduced pressure, and the residue was purified by flash chromatography (SiO 2 , 80% EtOAc/hexane) to afford the desired product 28 as a colorless oil. Yield 1.15 g (66%).

i s -1 -( 4 -( ( t e r t -B u t o x y c a r b o n y l ) d i m e t h y lphosphonomethoxy)cyclopent-2-en-1-yl)thymine 29.
A solution of aq. sodium carbonate (2 M, 2.3 mL, 4.6 mmol) was added to a stirring solution of thymine (0.53 g, 4.22 mmol) and allylic acetate 28 (1.40 g, 3.84 mmol) in N,N-dimethylformamide (45 mL), and the mixture was purged with nitrogen gas for 20 min. Solid Pd(dba) 2 (163 mg, 0.283 mmol) and 1,4-bis(diphenylphosphino)butane (163 mg, 0.384 mmol) were added, and the mixture was placed in a preheated heating block at 65°C. The reaction was monitored by TLC, and after 30 min, no acetate remained. The DMF was removed in vacuo, and the residue was dissolved in CH 2 Cl 2 , filtered, concentrated, and purified by flash chromatography (SiO 2 , 5% MeOH/CH 2 Cl 2 ) to afford the desired product 29 as an off-white foam (1.05 g, 63%).
Alternative Extractive Workup: The reaction was carried out as above, starting with 28 (1.0 g, 2.74 mmol), thymine (0.38 g, 3.01 mmol), Na 2 CO 3 (2 M, 1.65 mL, 3.30 mmol), Pd 2 (dba) 3 (93 mg, 0.10 mmol), and dppb (116 mg, 0.27 mmol) in DMF (30 mL). The cooled reaction mixture was partitioned with Et 2 O and 5% LiCl (150 mL each) and filtered over Celite. The phases were separated, and the aqueous phase was washed with Et 2 O. The combined organic phases were back-extracted with water, and the combined aqueous phases were then extracted with CH 2 Cl 2 (5 × 20 mL). The solvents were evaporated, and the remaining DMF was removed as an azeotrope with toluene. The residue was purified by flash chromatography (SiO 2 , 5% MeOH/CH 2 Cl 2 ) to afford 0.65 g (55%) of 29 as a white solid, shown by 1 H NMR to be ca. 4:1 dr. This was dissolved in CH 2 Cl 2 (25 mL) and three drops of Et 3 N were added. After standing for 24 h at room temperature, the mixture was evaporated and the residue was passed through a short plug of SiO 2 , eluting with 5% MeOH/CH 2 Cl 2 . Recovery 0.61 g, dr 1:1 (overall yield 52%). 1

cis-1-(4-((tert-Butoxycarbonyl)dimethylphosphonomethoxy)cyclopentan-1-yl)thymine 30.
A solution of the alkene 29 (0.91 g) in methanol (50 mL) was placed under nitrogen, and 5% palladium on carbon (110 mg) was added. The flask was flushed with hydrogen, and the mixture was stirred overnight under a balloon of hydrogen. The mixture was filtered over Celite, concentrated, and the residue was passed over a short plug of silica gel eluting with 5% MeOH/CH 2 Cl 2 to afford the desired product 30 as a white solid (0.83 g, 91%). 1  The Journal of Organic Chemistry pubs.acs.org/joc Article mg, 3.47 mmol) in MeCN (3.5 mL) was added dropwise to bromotrimethylsilane (381 μL, 442 mg, 2.89 mmol). The resulting solution was irradiated (50 W, 50°C) for 30 min, after which the volatiles were removed in vacuo. The residue was redissolved in CH 2 Cl 2 (5 mL), and two drops of DMF were added, followed by the dropwise addition of oxalyl chloride (2 M in CH 2 Cl 2 , 0.87 mL, 1.73 mmol). The resulting mixture was stirred at room temperature for 3 h and again evaporated in vacuo. The residue was redissolved in CH 2 Cl 2 (5 mL), cooled in ice, and treated dropwise with a solution of 2-(tertbutyl)-6-(hydroxymethyl)phenol 31 (113 mg, 0.629 mmol) and triethylamine (177 μL, 129 mg, 1.27 mmol) in CH 2 Cl 2 (1 mL); after the addition, the mixture was allowed to warm slowly to room temperature overnight. Silica gel (ca. 1 g) was added, the mixture was concentrated under reduced pressure, and the residue was purified by flash chromatography (80% EtOAc/hexanes) to afford the desired product 34 as a colorless film that was a roughly equal mixture of four diastereomers (47 mg, 15%). 1