Asymmetric Synthesis of (−)-6′-β-Fluoro-aristeromycin viaStereoselective Electrophilic FluorinationGyudong Kim,† Ji-seong Yoon,† Dnyandev B. Jarhad,† Young Sup Shin,† Mahesh S. Majik,†

Varughese A. Mulamoottil,† Xiyan Hou,† Shuhao Qu,† Jiyong Park,‡ Mu-Hyun Baik,‡

and Lak Shin Jeong*,†

†Research Institute of Pharmaceutical Sciences, College of Pharmacy, Seoul National University, Seoul, 151-742, Korea‡Center for Catalytic Hydrocarbon Functionalizations, Institute for Basic Science (IBS), Daejeon, 34141, Korea

*S Supporting Information

ABSTRACT: (−)-6′-β-Fluoro-aristeromycin (2), a potentinhibitor of S-adenosylhomocysteine (AdoHcy) hydrolase, hasbeen synthesized via stereoselective electrophilic fluorinationfollowed by a purine base build-up approach. Interestingly,purine base condensation using a cyclic sulfate resulted in asynthesis of (+)-5′-β-fluoro-isoaristeromycin (2a). Computa-tional analysis indicates that the fluorine atom controlled theregioselectivity of the purine base substitution.

(−)-Aristeromycin (1),1 isolated from Streptomyces citricolor, isa representative carbocyclic nucleoside that exists in Nature(Figure 1). This compound is a potent inhibitor of S-

adenosylhomocysteine (AdoHcy) hydrolase2 and shows potentantiviral activities against several RNA and DNA viruses. It wasfirst synthesized as a racemate by Clayton and his co-worker,3

and its enantioselective synthesis4 has since been reported.Based on a bioisosteric rationale, racemic mixtures of 6′-substituted aristeromycin analogues were also synthesized andevaluated for their inhibitory activity against AdoHcy hydro-lase.5a

Among these, (±)-6′-β-fluoro-aristeromycin (2) exhibitedthe most potent inhibitory activity against AdoHcy hydrolase.Thus, the synthesis of optically pure (−)-6′-β-fluoro-arister-omycin (2) is an interesting objective because the biologicalactivity is generally only found in one enantiomer, the D-isomer.As illustrated in Scheme 1, previous work demonstrated thatracemic (±)-25a as well as optically active (−)-25b could besynthesized via the key fluoroazide 5, which was synthesized bytreating the azido alcohols 4a and 7a with triflic anhydridefollowed by fluorination with tris(dimethylamino)sulfur

(trimethylsilyl)difluoride (TASF).5 The key fluoroazide 5 wasconverted to the desired product 2 using a the linear base build-up approach. However, these fluorination-containing ap-proaches required many steps to prepare the substrates 4aand 7a for fluorination and had low overall yields. Furthermore,the opening of racemic epoxide 3 with sodium azide was notregioselective (4a:4b = 1:1) and the opening of chiral epoxide 6was only moderately regioselective (7a:7b = 4:1).5b Thus, itwas highly desirable to develop a stereoselective fluorinationmethod with a substrate that could be efficiently prepared. Inthe present work, stereoselective electrophilic fluorination ofthe silyl enol ether 13 was employed for the synthesis of the keyfluoroketone 9β, which could be elaborated to the desired(−)-2. It was also previously reported that the 6-α-fluoroepoxide derived from 3 was opened with an adenine anion, butthis method showed poor regioselectivity (2:1 ratio) and lowyield.5a We thought that these drawbacks might be overcomeby employing a cyclic sulfate6 as a carbafuranosyl donor,because of its high electrophilicity and its capacity forregioselectivity under mild conditions.Herein, we report a stereoselective synthesis of optically pure

(−)-6′-β-fluoroaristeromycin (2) via a stereocontrolled electro-philic fluorination and the regioselective synthesis of 5′-β-fluoro-isoaristeromycin using cyclic sulfate chemistry, startingfrom D-ribose.As shown in our retrosynthetic plans (Scheme 2), our

priority (route I) was to try a Mitsunobu condensation of 12a(R = H) or a direct purine base condensation of 12b (R = Tf),

Received: August 10, 2017Published: October 13, 2017

Figure 1. Structures of (−)-aristeromycin (1) and its analogue 2.

Letter

pubs.acs.org/OrgLett

© 2017 American Chemical Society 5732 DOI: 10.1021/acs.orglett.7b02470Org. Lett. 2017, 19, 5732−5735

Cite This: Org. Lett. 2017, 19, 5732-5735

to access (−)-2. The second synthetic plan (route II) was tocondense the highly electrophilic cyclic sulfate 116,7 with apurine base. Compound 11 could be derived from 9β via theregioselective cleavage of the acetonide as a key step. The laststrategy (route III) would utilize the linear purine base build-upapproach from amino intermediate 10, which could easily bederived from 12a or 12b. It was envisaged that the β-

fluoroketone 9β could be prepared through a stereoselectiveelectrophilic fluorination of silyl enol ether 13, which could beeasily accessed from 8. Compound 8 could be prepared from14 by a Michael reaction using a Gilman reagent,8 and 14 couldbe efficiently derived from D-ribose with our previouslypublished procedure.9

To begin our synthesis (Scheme 3), D-ribose was convertedto known 14 in five steps, according to a modified known

procedure9 using ring-closing metathesis of 15 with NeolystM2 followed by PDC oxidation. Michael addition of 14 with aGilman reagent afforded known compound 88 in 70% yield.With 8, the substrate for fluorination in hand, we tried tointroduce the β-fluorine at the 6-position. After manyexperiments, the key fluorination was achieved using theelectron-rich silyl enol ether 13 with an electrophilic fluorinesource such as Selectfluor or N-fluorobenzenesulfonimide(NFSI). The silyl enol ether 13 was prepared by trapping 8with TESCl in the presence of LiHMDS. The best result for β-fluorination was obtained using Selectfluor in DMF at 0 °C(Table 1, entry 4). The ratio of β-fluoroketone 9β and α-

fluoroketone 9α depended on the reaction temperature andsolvent, giving ratios between 1.5:1 to 14:1 (Table 1).However, reagent-controlled asymmetric fluorination usingquinine or quinidine alkaloids did not give any desiredproduct.10 Tactics other than the use of Selectfluor were

Scheme 1. Comparison of Previous and Present Synthesis of2

Scheme 2. Retrosynthetic Analysis of (−)-2

Scheme 3. Synthesis of β-Fluoroketone 9β

Table 1. Stereoselective Electrophilic Fluorination

entry solvent temp (°C) time (h) ratio (9α:9β)a yield (%)b

1 CH3CN −10 48 1:1.5 622 CH3CN rt 12 1:2 603 CH3CN 50 1 1:14 364 DMF 0 12 1:5.2 915 DMF rt 4 1:4.3 82

aDetermined by crude 1H NMR. bIsolated total yield after silica gelcolumn chromatography.

Organic Letters Letter

DOI: 10.1021/acs.orglett.7b02470Org. Lett. 2017, 19, 5732−5735

5733

problematic. For instance, fluorination using NFSI afforded thedesired β-fluoroketone 9β in low yield (15%).With the key β-fluorinated substrate 9β in hand, we turned

our attention to direct SN2 base condensation, as shown inScheme 4. Compound 9β was reduced with NaBH4 to give 12a,

which was converted to triflate 12b by treatment with triflicanhydride. 12a and 12b were condensed with 6-chloropurine orN6-Boc-adenine under the Mitsunobu conditions and directSN2 conditions, respectively, but to our disappointment,Mitsunobu condensation of 12a gave only recovery of startingmaterial and direct SN2 condensation of 12b resulted in theelimination product instead of giving desired condensedproduct 16 or 17.The failure of the SN2 displacement of the triflate 12b with

purine nucleophiles can be explained by Hale’s rules for SN2displacement of pyranoside and furanoside triflates, whichemphasize the great importance of opposing α- and β-stericeffects and unfavorable dipolar repulsions in developing SN2transition states in sulfonate ester systems.11 Although 12bexhibits a favorable “Vicinal Triflate Effect”, wherein a stronglyelectron-withdrawing OTf lowers adjacent repulsive permanent

dipoles (F and OR) in a developing SN2 transition state, top-side steric hindrance caused by the adjacent C6−F and thebulky β-trans pseudoaxial-CH2OtBu groups impedes the attackof incoming sterically bulky and weakly nucleophilic purinebases, hindering the development of the colinear SN2 transitionstate needed for displacement, resulting in the SN2 reaction notproceeding.11 In contrast, the SN2 displacement of triflate 12bwith NaN3 occurred readily since the azide was a linear, lessbulky, and a more powerful nucleophile capable of attaining therequisite colinear SN2 transition state, which is made morefavorable by the aforementioned dipole-lowering “VicinalTriflate Effect”, which inductively reduces the adjacentpermanent opposing dipoles associated with the C6−F andthe C2−O groups. These findings show that Hale’s dipole-lowering “Vicinal Triflate Effect” extends to carbafuranosetriflates with vicinal dipolar electronegative groups.The azide obtained above was reduced with LAH to afford

amine 10, which was converted to the final (−)-2 using thepurine base build-up approach.5 Compound 10 was treatedwith 5-amino-4,6-dichloropyrimidine5 to give a base precursor,which was cyclized with CH3C(O)OCH(OEt)2

5 to yield 16.Treatment of 16 with tert-butanolic ammonia followed by theremoval of protective groups with 67% aq trifluoroacetic acid(TFA) yielded the final product, (−)-6′-β-fluoro-aristeromycin(2).To overcome the failure of the direct SN2 purine base

condensation, a cyclic sulfate was proposed as a glycosyl donor,as shown in Scheme 5. Regioselective cleavage of the

isopropylidene group of 12a with trimethylaluminum12

afforded diol 19. Treatment of 19 with thionyl chloride gavethe cyclic sulfite, which was immediately oxidized to cyclicsulfate 11.6,7 To our surprise, condensation of 11 with 6-chloropurine in the presence of NaH and 18-crown-6 producedisomeric 5′-β-fluoro-isoaristeromycin analogue 20 as a singleregioisomer after acidic hydrolysis. The structural determi-nation of 20, using 2D NMR experiments, showed that the C2-position had been substituted with 6-chloropurine (see ppS66−S68 in Supporting Information (SI)). This result could bealso explained by Hale’s rules.11 Nucleophilic attack at C1

Scheme 4. Synthesis of (−)-2 Using the Purine Base Build-up Approach

Scheme 5. Synthetic Approach to (+)-2a Using Cyclic SulfateChemistry

Organic Letters Letter

DOI: 10.1021/acs.orglett.7b02470Org. Lett. 2017, 19, 5732−5735

5734

would lead to greater dipolar repulsive and steric repulsiveinteractions in the developing SN2 transition state, but thesetransition state destabilizing interactions would not be observedfor the same reaction at C2. As a result, the C2 displacementproceeded regioselectively.Density functional theory (DFT) calculations were utilized

to elucidate the origin of the observed regioselectivity. Thereaction energy profile of the SN2 attack of the nucleophile (6-chloropurine anion) was computed on the C1 and C2 positionsof the cyclic sulfate (see Figure 1S in SI). The nucleophilicsubstitution steps are rate-determining, bearing the highestenergy barriers to the activation: 32.3 kcal/mol for thesubstitution at C1 and 29.0 kcal/mol for the substitution atC2. The substitution at the 2′-position is favored by 3.3 kcal/mol, which is in excellent agreement with the experimentaloutcome where the C2 substituted product is the sole product.To explain the energy preference shown in Figure 1S, fragmentanalysis was also carried out. The distortion energy penalty forthe cyclic sulfate substrate in the transition state for C1-substitution was +6.3 kcal/mol higher in energy than that forC2-substitution, which confirmed the preference on 2′-substitution (see Figure 2S in SI).Treatment of 20 with tert-butanolic ammonia followed by

deprotection with 67% aqueous TFA produced the final iso-nucleoside (+)-2a.The final compound (−)-2 was assayed for inhibitory activity

against AdoHcy hydrolase. As expected, it exhibited potentinhibitory activity (IC50 = 0.37 μM).In conclusion, enantiomerically pure (−)-6′-β-fluoro-arister-

omycin (2) was synthesized from D-ribose. The β-fluoro unitwas prepared by stereoselective electrophilic fluorination with areadily available silyl enol ether. Our method provides a β-fluorinated sugar, which can be extensively utilized for thestructure−activity relationship studies of fluorocarbocyclicnucleosides. We also developed a regioselective synthesis of(+)-5′-β-fluoro-isoaristeromycin (2a) using cyclic sulfatechemistry, whose displacement outcome was supported byDFT calculations. It is believed that all chemistry employed inthis study can be extensively applied to the synthesis offluorinated carbocyclic nucleosides.

■ ASSOCIATED CONTENT

*S Supporting Information

The Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.or-glett.7b02470.

Experimental procedures; copies of 1H, 13C, and 19FNMR spectra; and DFT energy calculation (PDF)

■ AUTHOR INFORMATION

Corresponding Author

*E-mail: lakjeong@snu.ac.kr.

ORCID

Mu-Hyun Baik: 0000-0002-8832-8187Lak Shin Jeong: 0000-0002-3441-707XNotes

The authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by a grant from the MidcareerResearch Program (370C-20160046) of the National ResearchFoundation, Korea.

■ REFERENCES(1) Kusaka, T.; Yamamoto, H.; Shibata, M.; Muroi, M.; Kishi, T.;Mizuno, K. J. Antibiot. 1968, 21, 255.(2) (a) De Clercq, E.; Cools, M. Biochem. Biophys. Res. Commun.1985, 129, 306. (b) Marquez, V. E.; Lim, M. Med. Res. Rev. 1986, 6, 1.(c) De Clercq, E.; Bergstrom, D. E.; John, A. H.; Montgomery, J. A.Antiviral Res. 1984, 4, 119. (d) De Clercq, E. Nat. Rev. Drug Discovery2002, 1, 13.(3) (a) Shealy, Y. F.; Clayton, J. D. J. Am. Chem. Soc. 1966, 88, 3885.(b) Shealy, Y. F.; Clayton, J. D. J. Am. Chem. Soc. 1969, 91, 3075.(c) Shealy, Y. F.; Thorpe, M. C.; Coburn, W. C., Jr.; Clayton, J. D.Chem. Pharm. Bull. 1980, 28, 3114.(4) (a) Arita, M.; Adachi, K.; Ito, Y.; Sawai, H.; Ohno, M. J. Am.Chem. Soc. 1983, 105, 4049. (b) Yoshikawa, M.; Okaichi, Y.; Cha, B.C.; Kitagawa, I. Tetrahedron 1990, 46, 7459. (c) Wolfe, M. S.; Lee, Y.;Bartlett, W. J.; Borcherding, D. R.; Borchardt, R. T. J. Med. Chem.1992, 35, 1782. (d) Madhavan, G. V.; Martin, J. C. J. Org. Chem. 1986,51, 1287.(5) (a) Madhavan, G. V. B.; McGee, D. P. C.; Rydzewski, R. M.;Boehme, R.; Martin, J. C.; Prisbe, E. J. J. Med. Chem. 1988, 31, 1798.(b) Yin, X.-Q.; Schneller, S. W. Tetrahedron Lett. 2005, 46, 7535.(6) For a review of cyclic sulfites and cyclic sulfates, see: (a) Lohray,B. B. Synthesis 1992, 1992, 1035. (b) Lohray, B. B.; Bhushan, V. Adv.Heterocycl. Chem. 1997, 68, 89. (c) Byun, H.-S.; He, L.; Bittman, R.Tetrahedron 2000, 56, 7051.(7) (a) Gao, Y.; Sharpless, K. B. J. Am. Chem. Soc. 1988, 110, 7538.(b) Naik, S. D.; Chandra, G.; Sahu, P. K.; Kim, H.-R.; Qu, S.; Yoon, J.-S.; Jeong, L. S. Org. Chem. Front. 2016, 3, 1472. (c) Lee, S.; Lee, S.;Park, H. J.; Lee, S. K.; Kim, S. Org. Biomol. Chem. 2011, 9, 4580.(8) (a) Gilman, H.; Jones, R. G.; Woods, L. A. J. Org. Chem. 1952, 17,1630. (b) Song, G. Y.; Paul, V.; Choo, H.; Morrey, J.; Sidwell, R. W.;Schinazi, R. F.; Chu, C. K. J. Med. Chem. 2001, 44, 3985.(9) (a) Choi, W. J.; Park, J. G.; Yoo, S. J.; Kim, H. O.; Moon, H. R.;Chun, M. W.; Jung, Y. H.; Jeong, L. S. J. Org. Chem. 2001, 66, 6490.(b) Moon, H. R.; Choi, W. J.; Kim, H. O.; Jeong, L. S. Tetrahedron:Asymmetry 2002, 13, 1189. (c) Mulamoottil, V. A.; Nayak, A.; Jeong, L.S. Asian J. Org. Chem. 2014, 3, 748.(10) Shibata, N.; Suzuki, E.; Asahi, T.; Shiro, M. J. Am. Chem. Soc.2001, 123, 7001.(11) (a) Hale, K. J.; Hough, L.; Manaviazar, S.; Calabrese, A. Org.Lett. 2014, 16, 4838. (b) Hale, K. J.; Hough, L.; Manaviazar, S.;Calabrese, A. Org. Lett. 2015, 17, 1738.(12) (a) Takano, S.; Ohkawa, T.; Ogasawara, K. Tetrahedron Lett.1988, 29, 1823. (b) Lee, J. A.; Kim, H. O.; Tosh, D. K.; Moon, H. R.;Kim, S.; Jeong, L. S. Org. Lett. 2006, 8, 5081.

Organic Letters Letter

DOI: 10.1021/acs.orglett.7b02470Org. Lett. 2017, 19, 5732−5735

5735