Toward fluorinated aminoglycosides: Structural studies of phenylhydrazine condensation with carbohydrate derivatives

LETTER ▌249 letter Toward Fluorinated Aminoglycosides: Structural Studies of Phenylhydrazine Condensation with Carbohydrate Derivatives Phenylhydrazine Condensation with Carbohydrate Derivatives Antonio Franconetti,* Pastora Borrachero, Manuel Gómez-Guillén, Francisca Cabrera-Escribano* Departamento de Química Orgánica, Facultad de Química, Universidad de Sevilla, Apartado de Correos No. 1203, 41071 Sevilla, Spain Fax +34(95)4624960; E-mail: afrancone[email protected]; E-mail: [email protected] Received: 28.10.2012; Accepted: 15.11.2012 Abstract: The reaction of phenylhydrazine with a sugar dialdehyde in water, as a key step for the synthesis of the 3-amino-3-deoxy-D- glucose moiety contained in kanamycin, has been revisited. Structural studies (IR and NMR as well as a simple theoretical model based on energy-minimization calculations and MD calculations) reported herein support the observed stereo- and regioselectivity. Efforts to improve the reproducibility and viability of the process as part of a convenient approach towards fluorinated kanamycin are also now presented. Key words: carbohydrates, condensation, phenylhydrazine, stereo- and regioselectivity, green chemistry Aminoglycosides are the most commonly used antibiotics worldwide thanks to the combination of their high efficacy with low cost. These compounds act on gram-positive and gram-negative bacteria as protein synthesis inhibitors.1 However, they are nephrotoxic and ototoxic in some cases, thus limiting their use. Kanamycin is a representative member of the aminoglycoside antibiotic family that shows multiple therapeutic applications and contains a 3-amino-3-deoxy-D-glucose moiety as a structural unit (Scheme 1). In general, substitution of an OH group by a fluorine atom is an important resource in medicinal chemistry.2 In addition to the bioavailability issue, the fluorination of bioactive molecules can increase their biological activity and improve their lipid permeability and metabolic stability. Furthermore, it is known that the presence of fluorine in the carbohydrate moiety contained in aminoglycoside antibiotics reduces their toxicity, especially when the fluorine atom is adjacent to the amino group. Among a large number of strategies or agents to introduce fluorine into carbohydrates, diethylaminosulfur trifluoride (DAST) is commonly used for the direct replacement of a hydroxyl group by fluorine under very mild conditions.3,4 With the aim of developing a potentially efficient route to fluorinated kanamycin, we set out to synthesize the 3-ami- no-3-deoxy-D-glucose moiety and study its subsequent fluorination with DAST at position 2. Therefore, as a previous step of kanamycin synthesis, the formation of 3- amino-3-deoxy-D-glucose from commercial available products has to be carried out. Condensation of phenylhydrazine with the sugar-derived dialdehyde 2 in water, as a key step, has been used5 as an elegant and simple synthetic strategy (Scheme 1). We have revisited this process and we report herein new structural NMR and IR data of compounds 2 and 4 as well as a simple theoretical model for explaining the result of the key condensation process. In addition to this, we describe novel improved reaction conditions and our recent progress toward 2-fluorinated 3-amino-3-deoxy-D-glu- cose. The synthesis of methyl 4,6-O-benzylidene-3-deoxy-3- phenylazo-α-D-glucopyranoside (4) was carried out in water by using phenylhydrazine.5 It has been proposed that, in aqueous solution, an equilibrium (Scheme 2) exists between the dialdehyde 2 and the hemialdal 3, obtained by periodate oxidation of methyl 4,6-O- benzylidene-α-D-glucopyranoside (1). This reaction was carried out for a week at room temperature (Table 1, method A) in aqueous solution to obtain the oxidized product in 70% yield, but this proved irreproducible. When we used sodium periodate in a MeOH–H2O mixture to accelerate the reaction (Table 1, method B), a white solid, oxidized product was obtained in 98% yield.6 Scheme 1 O OO O O Ph OMe O O O PhN2HO OMe Ph O O O O NH2 HO HO H2N H2N NH2 HO NH2 HO OH OH kanamycin B 2reduction to amine protecting group 4 SYNLETT 2013, 24, 0249–0253 Advanced online publication: 11.12.2012 0936-52141437-2096 DOI: 10.1055/s-0032-1317783; Art ID: ST-2012-D0925-L © Georg Thieme Verlag Stuttgart · New York Downloaded by: Universidad de Sevilla. Copyrighted material. 250 A. Franconetti et al. LETTER Synlett 2013, 24, 249–253 © Georg Thieme Verlag Stuttgart · New York We have performed IR spectra of compounds obtained by using both methods (Scheme 2, Table 1) and we did not observe a carbonyl absorption (ca. 1600 cm–1) in either case. This indicates that the periodate oxidation product of compound 1 has the structure of the hemialdal 3. Furthermore, the product obtained by the periodate method showed 1H NMR and 13C NMR spectra (in DMSO-d6) with no signals for the aldehydic protons nor carbonyl carbons, respectively. This information reinforced the hemialdal structure 3 assigned to the periodate oxidation product of compound 1, not only in the solid state as deduced from the IR spectrum, but also in solvents other than water, since the equilibrium between 2 and 3 occurs only in aqueous solution. Therefore, we used D2O as the solvent for recording the 1H NMR spectrum of compound 3 under presaturation conditions; in this case, the aldehydic proton signal was clearly observed at δ = 9.97 ppm. Analogously, the aldehydic carbonyl signal appeared in the 13C NMR spectrum (in D2O) at δ = 196.8 ppm [other relevant signals: δ = 135.5–126.4 ppm (aromatic carbons), 101.2 ppm (PhCH), 96.6 ppm (C-1), and 48.9 ppm (CH3)], confirming the presence of the dialdehyde 2 in the sample. Next, we have investigated the condensation reaction of phenylhydrazine with dialdehyde 2 to obtain compound 4 (Scheme 3). We have tried different reaction conditions as indicated in Table 2. Under the conditions studied, the results were not reproducible. Regarding compound 4, the best results were obtained using conditions A and E (Table 2).7 From the 1H NMR spectrum we realized that the crude product obtained under conditions A (Table 2) was a 3:1 mixture of compounds 4 and 4a. The presence of both regioisomers could be inferred from the two diagnostic signals for the anomeric OMe groups at δ = 3.50 and 3.52 ppm, and two singlets corresponding to H-7 of the benzylidene groups at δ = 5.77 and 5.74 ppm. Recrystallization of the crude solid product obtained under conditions E (Table 2),7 which was carried out under argon atmosphere and absence of light, afforded highly pure 4. Its 1H NMR spectrum (300 MHz, CDCl3) included aromatic proton signals (10 H) at δ = 7.74–7.29 ppm, two singlets at δ = 5.77 and 3.50 ppm, respectively, assigned to the benzylic (H-7) and OMe protons, a broad singlet at δ = 4.72 ppm (anomeric proton), two triplets at δ = 5.48 (10.2 Hz) and 4.10 ppm (10.0 Hz, H-3 and H-4), and a Scheme 2 O O O O O Ph OMe O HO OH O O Ph OMe O 23 O O O Ph OMe 1 HO OH NaIO4 H2O–MeOH Table 1 Reaction Conditions to Obtain Dialdehyde 2 or Hemialdal 3 Method Solvent Time Other conditions Yield (%) AH 2O 1 w darkness 70 B MeOH–H2O 20 h darkness, cooling 98 Scheme 3 O O O Ph OMe 4 PhN2OH O O O O O Ph OMe 2 Table 2 O O O Ph OMe HO N2Ph 4a + PhNHNH2 Table 2 Reaction of Phenylhydrazine with Dialdehyde 2 Conditions Temp (°C) Reagent Reagent (equiv) Other conditions Products (yield, %)a A 85 PhNHNH21.5 – 4 (15) + 4a (7) B 80 PhNHNH22.5 not cooling 5 (78) C 70 PhNHNH3Cl/NaOAc 2 darkness, ice (0 °C) 5 (–b) D 83 PhNHNH22.5 ice/NaCl (–15 °C) 5 (68) E 92 NaOAc/PhNHNH3Cl 2 darkness, ice/NaCl (–15 °C), argon 4 (16) a After recrystallization. b Not quantified, only detected by TLC. Downloaded by: Universidad de Sevilla. Copyrighted material. LETTER Phenylhydrazine Condensation with Carbohydrate Derivatives 251 © Georg Thieme Verlag Stuttgart · New York Synlett 2013, 24, 249–253 double doublet at δ = 4.75 ppm (3J1,2 = 4.5 Hz), assigned to H-2. Nevertheless, compound 4 [Rf = 0.88 (49:1 CH2Cl2–acetone)], was unstable, and its rapid decomposition into products of lower Rf was observed by TLC. Additionally, the low yields obtained for this compound could be explained by a competitive acid hydrolysis of 2 at the anomeric position in the aqueous medium to give 2,4-O- benzylidene-D-erythrose and glyoxal, which subsequently reacted with excess phenylhydrazine to afford glyoxal bisphenylhydrazone 5 (Figure 1). This compound, a useful product as promoter for metal-catalyzed cross-coupling reactions,8 was isolated in most conditions (conditions B, C, and D, Table 2) as an orange solid and identified according to the literature.5 Figure 1 To justify the higher proportion of isomer 4, we used a simple theoretical model with ChemBio3D and Discovery Studio 3.1. First, we applied an energy-minimization calculation using MM2 field force. Aromatic stacking interactions are widespread in nature, generally in biomolecules (e.g., DNA). In this context, two aromatic residues are considered to interact if the distance between phenyl centroids is less than 7 Å.9 From our calculations, a stacking interaction between the two phenyl groups present in the molecule of 4 was indicated (Figure 2). By using the same conditions to minimize energy, we observed an increased theoretical energy in the isomer 4a, since the additional stabilization (–12.1 kJ mol–1) of an aromatic stacking interaction is not possible in this case. Figure 2 Modeling results: (a) isomer 4 showing aromatic interaction between aromatic ring residues, and (b) isomer 4 showing integrity of carbohydrate chair; (c) isomer 4a where aromatic interaction is not observed Additionally, molecular-dynamics calculations10 were performed for isomers 4 and 4a. In our simulation, the target temperature was held at 300 K, and the step and frame intervals were 2.0 and 10 fs, respectively. With this information, we could build a model for isomer interaction to check that aromatic stacking is not broken (Figure 3). Figure 3 (a) Model of isomer 4, and (b) model of isomer 4a based on MD calculations Crude compound 4 obtained under conditions E (Table 2) was reduced in MeOH by hydrogen in the presence of Adam’s catalyst (PtO2) to give methyl 3-amino-4,6-O-ben- zylidene-3-deoxy-α-D-glucopyranoside (6) in quantitative yield pure enough to be used for further transformation, but too unstable to be purified by column chromatography (Scheme 4). Scheme 4 Prior to the 2-OH fluorination of compound 6, we considered the possibility of selective protection of the 3-amino group by acetylation under the mild conditions described by Sharma et al.11 for related compounds. Under these conditions (Ac2O–EtOH, 0 °C, Scheme 4), only trace amounts of the 3-acetamido compound 7 were obtained along with the 2,3-N,O-diacetate 8 in poor yield (15%). Moreover, conventional acetylation in Ac2O–pyrdine of compound 6 also turned out to be ineffective, with 8 being isolated in 20% yield. All these compounds presented characteristic 1H NMR features; the anomeric protons displayed a coupling constant 3J1,2 = 4.5 Hz, in agreement with that expected for α anomers, and a singlet was observed at δ = ca. 4.5 ppm corresponding to the anomeric OMe. NHN 5 NNH O O O Ph OMe 4 PhN2OH H2, PtO2 MeOH O O O Ph OMe 6 H2NOH O O O Ph OMe AcHN OH O O O Ph OMe 8 (15%) AcHN OAc + 7 (traces) (20%) Ac2O–Py r.t. Ac2O–EtOH 0 °C quant. yield Downloaded by: Universidad de Sevilla. Copyrighted material. 252 A. Franconetti et al. LETTER Synlett 2013, 24, 249–253 © Georg Thieme Verlag Stuttgart · New York While these reactions constitute a highly stereo- and regioselective route toward aminoglycosides that can be considered an example of a green chemistry strategy12 the lack of reproducibility and low yields made it necessary to seek alternative synthesis routes to 3-amino-3-deoxy-D- glucose13 (Scheme 5). Accordingly, hydrolysis of the diacetal 914 (HCl–MeOH, reflux, 7 h) followed by the conventional protection of the 4- and 6-OH functions as a benzylidene acetal gave the anomeric 2:1 mixture of 10/11, whose spectroscopic data were identical to those previously described in the literature.13a After separation by column chromatography, the major α-D-anomer 10 was dissolved in MeCN and treated with DAST under reflux for six hours. Workup and purification by column chromatography on silica gel led to the desired pure 2-fluorinated compound 12 (25%),15 along with a diastereomeric mixture of rearranged, ring-con- tracted products 13 and 14 (63%)4 and unreacted starting material 10 (9%). Usually, in 1H NMR spectra of fluorinated sugars, the geminal coupling constants (2JF,H) are 43–59 Hz in pyranoses and 46–66 Hz in furanose rings.16 The vicinal coupling constants (3JF,H) depend on the electronegativity of neighboring substituents. In our case, the observed 2J2,F = 48.2 Hz and 3J3,F = 29.0 Hz values are in agreement with the proposed structure 12. The required disposition of the hydrogen atoms on the 4C1 pyranose conformation for these coupling constants is shown in Scheme 5. Compound 12, having the backbone of an important moiety of kanamycin and a fluorine atom at C-2, presents a suitable feature to potentially reduce the toxicity of this aminoglycoside antibiotic. This compound has a latent amino functionality, the azide group, reduction of which would lead to the desired amine after hydrolysis of the benzylidene group (Figure 4). The azide group could be readily reduced with hydrogen in the presence of Pd/C catalyst or by Staudinger reaction with triphenylphosphine, and the benzylidene group could be hydrolyzed in a mixture AcOH–H2O (7:3)17 or p-tolu- enesulfonic acid.18 In summary, the aqueous condensation of phenylhydrazine and the dialdehyde obtained by periodate oxidation of methyl 4,6-O-benzylidene-α-D-glucopyranoside (1), described in the literature as an elegant and easy synthesis of the corresponding 3-amino-3-deoxy derivative 6, has poor reproducibility in our hands. Spectroscopic studies and a simple theoretical model based on energy-minimi- zation and molecular-dynamics calculations support the stereo- and regioselectivity of the process in favor of the 3-phenylazo isomer 4. In view of our unsuccessful efforts to improve the reproducibility and viability of this approach toward fluorinated kanamycin, we present the preparation of fluorinated precursor subunit 12 starting from commercially available diacetone-D-glucose, via the 3-azido-3-deoxy derivative 9. Acknowledgment We thank the AECID (Projects A/023577/09 and A/030422/10) and the ‘Junta de Andalucía’ (FQM 142 and Project P09-AGR-4597) for financial support. We also thank D. Lara for invaluable collaboration. References and Notes (1) (a) Li, J.; Chang, T. Anti-Infect. Agents Med. Chem. 2006, 5, 255. (b) Corey, E. J.; Czakó, B.; Kürti, L. Molecules and Medicine; John Wiley and Sons: Hoboken, 2007. Scheme 5 Alternative synthetic route to fluorinated aminoglycoside derivatives and coupling constants of compound 12 O O O Ph OMe 10 N3OH O O O N3 O O 9 i. HCl/MeOH, reflux 48 h ii. PhCHO, ZnCl2 80% O O O Ph OMe 11 N3OH + 2:1 DAST MeCN, reflux O N3 F 13 + 14 (63%) F H MeO HO O O O Ph OMe N3F + 12 (25%) O O O Ph N3 F OMe H2 H1 H3 3J1,F = 7.8 Hz 3J3,F = 29.0 Hz 2J2,F = 48.2 Hz Figure 4 O O O Ph OMe N3Ftoxicity reduction facile reduction to amino group O HO HO O H2NOH kanamycin moiety Downloaded by: Universidad de Sevilla. Copyrighted material. LETTER Phenylhydrazine Condensation with Carbohydrate Derivatives 253 © Georg Thieme Verlag Stuttgart · New York Synlett 2013, 24, 249–253 (2) Kirk, K. L. Org. Process Res. Dev. 2008, 12, 305. (3) Ferret, H.; Déchamps, I.; Gomez Pardo, D.; Van Hijfte, L.; Cossy, J. ARKIVOC 2010, (viii) 1, 26. (4) Vera-Ayoso, Y.; Borrachero, P.; Cabrera-Escribano, F.; Carmona-Asenjo, A.; Gómez-Guillén, M. Tetrahedron: Asymmetry 2004, 15, 429; and references cited therein. (5) Patroni, J.; Stick, R. Aust. J. Chem. 1985, 38, 947; and references cited therein. (6) Synthesis of Compound 2 (Method B) NaIO4 (7.7 g, 36 mmol) was dissolved in H2O (95 mL). After cooling at 5 °C, a solution of methyl 4,6-O-benzylidene-α-D- glucopyranoside (5 g, 17.73 mmol) in MeOH (95 mL) was added. The mixture was stirred in the dark for 20 h at r.t. EtOH was added to precipitate iodates, and the filtered solution was evaporated under vacuum and dried to obtain dialdehyde 2 (98% yield). 13C NMR (125.7 MHz, D2O): δ = 196.8 (CHO), 135.5–126.4 (Ph), 101.2 (CHPh), 96.6 (C-1), 75.9 (C-4), 71.3 (C-6), 64.6 (C-5), 48.9 (OCH3) ppm. HRMS (CI): m/z calcd for C13H15O5 [M–CHO]: 251.0919; found: 251.0913; and m/z calcd for C13H13O5 [M–OMe]: 249.0763; found 249.0766. (7) Preparation and Most Relevant Data of Compound 4 Conditions A5 Phenylhydrazine (250 μL, 2.59 mmol) was added to a solution of compound 2 (0.5 g, 1.79 mmol) in H2O (125 mL) at 85 °C. On shaking the yellow mixture rapidly with icewater, a yellow solid precipitated. This crude product was filtered off and purified by recrystallization from a small volume of butan-1-ol to give a yellow solid, identified as a mixture of 4 (95.2 mg, 15%) and 4a (27.7 mg, 7%). Attempted column chromatography on silica gel resulted in rapid decomposition. Conditions E A solution of compound 2 (0.33 g, 1.18 mmol) in H2O (100 mL) was heated at 92 °C, under argon in darkness. The solution was cooled, then NaOAc (0.39 g, 4.73 mmol) was added followed by slow addition of phenylhydrazine hydrochloride (0.34 g, 2.36 mmol). After rapid cooling to – 15 °C and shaking of the mixture, a yellow solid precipitated. This product was filtered off, washed with cool H2O, recrystallized from butan-1-ol and dried under vacuum to give a yellow solid which was identified as pure methyl 4,6-O-benzylidene-3-deoxy-3-phenylazo-α-D- glucopyranoside (4, 68.6 mg, 16%). Rf = 0.88 (CH2Cl2– acetone = 49:1). 1H NMR (300 MHz, CDCl3): δ = 7.74–7.29 (m, 10 H, Ph), 5.77 (s, 1 H, H-7), 5.48 (t, 1 H, H-3, J3,4 = J3,2 = 10.2 Hz), 4.75 (dd, 1 H, H-2, J2,1 = 4.5, J2,3 = 10.6 Hz), 4.72 (br s, 1 H, H-1), 4.46 (dd, 1 H, J6ec,5 = 4.6 Hz, J6ec,6ax = 10.2 Hz, H-6eq), 3.32 (m, 2 H, H-5 and H-6ax), 4.10 (t, 1 H, H-4, J4,3 = J4,5 = 10.0 Hz), 3.50 (s, 3H, OCH3) ppm. HRMS (CI): m/z calcd for C20H22N2O5: 370.1529; found: 370.1522. (8) (a) Liu, Y.-H.; Li, G.; Yang, L.-M. Tetrahedron Lett. 2009, 50, 343. (b) Trofimov, B. A.; Myachina, G. F.; Ermakova, T. G.; Kuznetsova, N. P.; Volkova, L. I.; Sultangareev, R. G.; Larina, L. I.; Klyba, L. V.; Sukhanov, G. T.; Sakovich, G. V. Russ. J. Org. Chem. 2009, 45, 1683. (9) Hunter, C. A.; Lawson, K. R.; Perkins, J.; Urch, C. J. J. Chem. Soc., Perkin Trans. 2 2001, 651. (10) (a) Rode, M. B.; Hofer, S. T.; Kugler, D. M. The Basics of Theoretical and Computational Chemistry; Wiley-VCH: Weinheim, 2007. (b) Young, D. Computational Chemistry; John Wiley and Sons: New York, 2001. (11) Sharma, M.; Bernacki, R. J.; Hillman, M. J.; Korytnyk, W. Carbohydr. Res. 1993, 240, 85. (12) Anastas, Y.; Warner, J. C. Green Chemistry: Theory and Practice; Oxford University Press: New York, 1998. (13) (a) Albert, R.; Dax, K.; Pleschko, R.; Stutz, A. E. Carbohydr. Res. 1985, 137, 282. (b) Faghih, R.; Cabrera- Escribano, F.; Castillon, S.; Garcia, J.; Olesker, A.; That Thang, T. J. Org. Chem. 1986, 51, 4558. (14) Brimacombe, J. S.; Bryan, J. G. H.; Husain, A.; Stacey, M.; Tolley, M. S. Carbohydr. Res. 1967, 3, 318. (15) Preparation of Compound 12 A solution of compound 10 (109 mg, 0.355 mmol) in dry MeCN (6.5 mL) was treated with DAST (236 μL, 1.785 mmol) at 0 °C under argon. After several minutes, the cooling bath was removed, and the mixture was heated to reflux for 6 h, monitoring the reaction by TLC. On completion, after evaporation of the solvent, the residue was treated with CH2Cl2 (20 mL) and cold sat. aq NaHCO3 (50 mL). The aqueous layer was extracted with CH2Cl2 (3 × 20 mL), and the combined organic layers were washed with sat. aq NaCl (50 mL), dried (Na2SO4), filtered, and concentrated. The residue was purified by column chromatography on silica gel (1:5 to 1:3 gradient hexane–Et2O with 1% Et3N) to give unreacted 10 (9%), compound 12 (25 mg, corresponding to 25% yield from converted substrate) and the epimeric mixture 13/14 [63% yield, characterized by transformation into the corresponding (1R and 1S)-2,5- anhydro-3-azido-6-O-tert-butyl-diphenylsilyl-3,4-dideoxy- 1,4-difluoro-1-O-methyl-D-talitols.4] Selected Analytical Data Compound 12: solid; mp 56–60 °C; Rf = 0.44 (Et2O–hexane = 1:3). 1H NMR (300 MHz, CD3COCD3): δ = 7.49–7.36 (m, 5 H, Ph), 5.81 (s, 1 H, CHPh), 4.91 (dd, 1 H, 3J1,F = 7.8 Hz, J1,2 = 1.7 Hz, H-1), 4.78 (ddd, 1 H, 2J2,F = 48.2 Hz, J2,3 = 2.3 Hz, H-2), 4.27 (dd, 1 H, J6,6′ = 11.7 Hz, J5,6 = 1.7 Hz, H-6), 4.12 (ddd, 1 H, J3,4 = 10.6 Hz, J4,5 = 9.0 Hz, 4J4,F = 1.6 Hz, H-4), 3.92 (ddd, 3J3,F = 29.0 Hz, H-3), 3.89–3.83 (m, 2 H, H- 5 and H-6′), 3.44 (s, 3 H, OCH3) ppm. 13C NMR (75.8 MHz, CD3COCD3): δ = 138.8–127.2 (Ph), 102.5 (CHPh), 99.5 (d, 2JC1,F = 31.0 Hz, C-1), 89.8 (d, 1JC2,F =180.0 Hz, C-2), 74.6 (C-4), 69.2 (C-6), 64.9 (C-5), 59.7 (d, 2JC3,F = 16.6 Hz), 55.7 (OCH3) ppm. HRMS (CI): m/z calcd for C14H16FN3O4 + H: 310.1203; found: 310.1207. (16) Michalik, M.; Hein, M.; Frank, M. Carbohydr. Res. 2000, 327, 185. (17) Baer, H. H.; Gan, Y. Carbohydr. Res. 1991, 210, 233. (18) Vera-Ayoso, Y.; Borrachero, P.; Cabrera-Escribano, F.; Gómez-Guillén, M.; Caner, J.; Farrás, J. Synlett 2010, 271. Downloaded by: Universidad de Sevilla. Copyrighted material.