lh NMR Model Studies of Amphotericin B: Comparison of X-Ray and NMR Stereochemical Data

Transcription

1 MAGNETIC RESONANCE IN CHEMISTRY, VOL. 30, (1992) lh NMR Model Studies of Amphotericin B: Comparison of X-Ray and NMR Stereochemical Data Pawel Sowinski, Jan Pawlak and Edward Borowski* Department of Pharmaceutical Technology and Biochemistry, Technical University of Gdansk, Gdansk, Poland Pierluigi Gariboldi Department of Chemical Sciences, University of Camerino, Camerino (MC), Italy H NMR data for amphotericin B methoxycarbonylmethylamide extracted from DQF-COSY, ROESY and 1D spectra were compared with x-ray data for amphotericin B. The conformation of amphotericin B in solution was identical with that found by x-ray analysis in the solid state. Most of the relay artifacts observed in the ROESY spectrum were recognized, and their pathways of formation were assigned. KEY WORDS Polyene macrolides Amphotericin B Solution-state conformation 2D NMR COSY ROESY INTRODUCTION Polyene macrolides constitute a class of antifungal agents of which over 200 reprentatives have been recognized, and a considerable number have been characterized by their gross structures. However, the elucidation of their stereochemistry is a challenging problem. To our knowledge, only the stereostructures of amphotericin B2 and roxaticin3 have been determined by x-ray analysis, because most of the polyene macrolides exhibit poor crystallinity. On the other hand, NMR studies of polyene macrolides have long been hampered by their low solubility and high tendency for association, for which reason DMSO-d, is the most commonly used solvent. The resonance line width in DMSO-d, was 7-8 Hz? and this almost prevented coupling constant analysis. An insight into the stereochemistry of lienomycin and mycoticin6 has been achieved by a combination of H NMR analysis and chemical degradation or synthesis. The most recent studies reporting the stereostructures of nystatin A,, vacidin A* and pimaricin seem to offer a general protocol for the elucidation of the stereochemistry of polyene macrolides. This prompted us to compare the crystallographic and H NMR data for amphotericin B, which allowed a critical insight into scalar and dipolar coupling analysis. Arnphotericin B (1, Fig. 1) was derivatized into its methoxycarbonylmethylamide (2), which exhibited increased solubility in most NMR solvents. This allowed pyridine-d,-methanol-d, to be chosen as the best solvent for the differentiation of proton chemical shifts. The natural line width in this solvent system was Hz. * Author to whom correspondence should be addressed /92/ $ by John Wiley & Sons, Ltd. RESULTS AND DISCUSSION The DQF-COSY spectrum (Fig. 2) of the amphotericin B derivative 2 displayed a full set of connectivities within four structural blocks, i.e. C-2-C-12, C-14-C-21, C-32-C-37 and the mycosamine moiety. The assignments of the proton chemical shifts are collected in Table 1. The chemical shifts of H-37 (5.85 ppm) and H-2 (2.48 and 2.65 ppm) indicated that blocks C-32-C-37 and C-2-C-12 are linked via a lactone moiety. The remaining blocks were assembled from the H- 12/H-14 and H-l /H-19 ROEs observed in the ROESY spectrum (Fig. 2). It should be mentioned that the connectivities within the heptaene chromophore could not be traced owing to the strongly coupled spin system, but usually such a fragment of a non-aromatic polyene macrolide can be defined by its UV spectrum. The presence of a hemiketal ring in amphotericin B was manifested in the 13C NMR spectrum by a signal at 98.4 ppm. The chair conformation of the six-membered hemiketal ring was demonstrated by the magnitude of the vicinal coupling constants between H-l4b, H-15, H-16 and H-17, which were in the range Hz (Table 2). An H-14b/H-16 ROE was also observed. The C-17/C-19 relative configuration was derived as follows: (i) J(17,18b) = 10.3 Hz and J(17,18a) x 0 Hz assigned an H-17/H-l8b staggered position; (ii) J(18b,19) = 1.5 Hz and J(18a,19) = 4.7 Hz, together with the H-l8b/H- 19 ROE, indicated the synclinal position of H-19 to the C-18 methylene protons; and (iii) the H-2/H-17 ROE defined the mycosamine moiety in relation to the macrolidone plane, thereby assigning the C- 17/C- 19 relative configuration. The /l configuration of the glycosidic linkage and the 4C, conformation of D-mycosamine were reflected by its coupling pattern (Table 2) and also by the H-l /H-3 and H-l /H-5 ROEs (Table 1). The spatial relationship Received 1 June 1991 Accepted (revised) 28 November 1991

2 276 P. SOWINSKI, J. PAWLAK, E. BOROWSKI AND P. GARlBOLDl HO "" 0 Off 0 Arnphotericin B (1) R= Arnphotencin B methoxycarbonylmefhylornide (2) R= NHCH,COOCH, Figure 1. Structures of amphotericin B (1 ) and its methoxycarbonylmethylamide derivative (2). F a- Y 3x3s 1b.10 I T r Fl IPPt4 Figure MHz ROESY (upper triangle) and DQF-COSY (lower triangle) spectra of amphotericin B methoxycarbonylmethyiamlde (1 5 mg ml-', 12% methanol-d, in pyridine-d,) combined along the diagonal

3 'H NMR STUDIES OF AMPHOTERICIN B, AND COMPARISON WITH X-RAY DATA 211 Table 1. 'H cbemical shifts and ROE effects of amphotericin B methoxycarbonylmethylamide (2)' Proton 2a 2b 3 4a 4b 5 6a 6b 7a 7b Oa 1 Ob 11 12a 12b 14a 14b a 18b (ppm) ROE to protons 2b. 4a 2a. 3 2b. 4b. 5, 30 2a, 4b 3, 4a. 5 3, 4b, 6b, 7, 28 (30) 6b 5, 6a. 8 7b 5, 7a. 8, 9 6b. 7b. 10b 7b, 11, 22 (24) lob, 12a 8, 1 Oa, 11 9, lob, 12b. 20, 22 1 Oa, 12b, 14a 11,12a 12a, 14b, 15 14a, 16 14a 14b 20, 2' 78b. 1' 18a, 19 18b, 21, 1' 11, 17 6 Proton (ppm) Me Me Me ' ' ' ' ' 1.62 a Assignments in parentheses may be interchangeable or coexisting. ROE lo protons 19 9 (9/24), (9122) 5 (5130) 3, 5 (5/28) 33 34,37 31, 35, 36, 34-Me 32, 36, 37, 34-Me 33, 36, 34-Me, 36-Me 33, 34, , 34-Me 32, 34, 36, 37-Me 33, 34, 35 35, 36, 37-Me 37, 36-Me 2'. 3'. 5, 18a. 19 l', 3'. 17 l', 2' 6' 1'. 6 5'. 4' of the mycosamine moiety and the C-14-C-19 fragment demonstrated by the H-l'/H-19, H-l'/H-l8a and H-2'/H-17 ROEs (Fig. 2) defined the absolute configurations at C-19, C-17, C-16 and C-15. The position of the C-12 methylene group as a substituent of the hemiketal ring was inferred to be equatorial from the H-14a/ H-12a ROE and from the lack of H-12/H-15 and H-12/H-17 ROEs, which would be observed if the C-12 methylene group were in an axial position. Although the coupling pattern and the ROE connectivities of the C fragment could be matched with several conformational variants,x only a fully stretched conformation of this fragment met the steric requirements imposed by an all-trans heptaene chromophore. Using this reservation J(2a,3), J(3,4a), J(4a,5), J(5,6a), J(7a,8), J(9,1Oa), J(lOa,l I), J(11,12a) and J(2b,3), J(3.4b) J(4b,5), J(5,6b), J(7b,8), 5(8,9), J(9,10b), J(lOb,ll), J(l1, 12b), which were in the range and Hz, Table 2. Experimental 'H-'H vicinal coupling constants (Hz) for amphotericin B methoxycarbonylmethylamide and calculated values for the x-ray structure of amphotericin B Protons Expt Calc Protons Expt Calc 2a-3 2b-3 3-4a 3-4b 4a-5 4b5 54a 54b 7a4 7 M Oa 9-1 Ob 1 Oa-11 1 ob a 11-12b a-15 14b a b 18a-19 18b '-2, 2'-.3' 3'4, 4-5' _

4 278 P. SOWIfiSKI, J. PAWLAK, E. BOROWSKI AND P. GARIBOLDI respectively, defined the relative configuration of the C-2-C- 12 fragment. The H-17/H-20 and H-19/H-21 ROEs combined with J(19,20) = 9.2 Hz defined the spatial relationship of the heptaene chromphore and the C-12-C-19 fragment. On the other hand, the mutual alignment of the C-2-C-12 fragment and the heptaene chromphore was given by the ROEs between the olefinic and the C-3, C-5, C-9 and C-1 1 hydroxymethine protons. These two conditions were met by only one enantiomer of the C-2-C-12 fragment, and this defined the absolute configuration at C-3, C-5, C-8, C-9 and C-1 1. The relative configuration of the C-34-C-37 fragment was derived in a straightforward manner from J(33,34) = 10.0 Hz, J(34,35) = 9.8 Hz, J(35,36) = 2.1 Hz and J(36,37) = 2.2 Hz and from the H-32/H-34, H-33/H-35, H-33/H-36 and H-32/H-37 ROEs. The steric requirement for closing the macrolactone ring allowed only one enantiomer of the C-34-C-37 fragment, thus proving the absolute configurations at C-34, C-35, C-36 and C-37. Hence the absolute configuration of amphotericin B and its conformation in solution were attributed from the 'H NMR spectral data extracted from the DQF- COSY, ROESY and 1 D spectra. The conformation of amphotericin B in solution appeared to be very closely related with that in the solid state, assigned by x-ray analysis. This relationship was quantitatively evaluated by comparison of measured vicinal coupling constants with those calculated from x-ray data.' Since the crystallographic data did not provide hydrogen atom coordinates, these were geometrically calculated from the hybridization of the appropriate carbon atoms. The J couplings were calculated by the simplified version of the extended Karplus equation.' The experimental and measured J couplings are compared in Table 2. The standard deviation (S.D.) of the differences between the measured and calculated J couplings was found to be 0.63 Hz, whereas the S.D. of the errors of the extended Karplus equation was reported to be 0.41 Hz." The difference between these two values is related to the accuracy of the crystallographic data (see Experimental) and the coupling constant measurements (S.D. estimated as 0.13 Hz), rather than to conformation implications. Consequently, the conformations of amphotericin B in solution and in the solid state, determined by 'H NMR and x-ray analysis, respectively, were found to be identical. The resulting interatomic distances (Fig. 3) were in full agreement with the ROEs displayed by the ROESY spectrum (Fig. 2 and Table 1). The ROESY experiment was preferred to NOESY in order to avoid the well known problems with vanishing NOES, which occur in mid-size molecules (me = 1) to which polyene macrolides belong. The ROESY spectrum of amphotericin B displayed, in addition to a very abundant map of connectivities, some artificial cross-peaks originating from relay and Hartmann-Hahn transfer (HHT) artifacts.12 The HHT cross-peaks, recognized by their positive sign, which is oppoiste to ROE, were assigned as H-8/H-9, H-15,'H- 16, H-16/H-17, H-18a/H-19, H-33/H-34, H-3'/H-4 and H-4/H-5'. The maximum efficiencies of Hartmann- Hahn transfer (RA,)l3 for these cross-peaks were found to be in a range 14-95%. In general, the HHT cross-peaks are observed in close proximity to the diagonal and counter-diagonal, where the condition for effective magnetization transfer is met. The remaining three positive cross-peaks, H-l'/H-6, H-l'/H-l8b and H-3'/H-5', could not be Figure 3. Conformation of amphotericin B methoxycarbonylrnethylamide. ROEs displayed by the ROESY spectrum, depicted as bidirectional arrows, are accompanied by the corresponding interatomic distances (A) extracted from the x-ray data of amphotericin B.

5 'H NMR STUDIES OF AMPHOTERICIN B, AND COMPARISON WITH X-RAY DATA ~~ Table 3. Relay effects observed in the ROESY spectrum of amphotericin B methoxycarbonylmethylamide Spin pair H -1 '/H -6' H- 1 '/H-l8b H-3'/H-5' H-l'/H-4' H-2'/H-4' H -2'/H - 5' H -3'/H -6' H-2'/H-16 H-33/H-37 H - 1 9/H - 20 Formation pathway H-l'/H-5' ROE, H-5'/H-6' ROE H-l'/H-l8a ROE, H-18a/H-l8b ROE H-3'/H-4' HHT, H-4'/H-5' HHT H-l'/H-3' ROE, H-3'/H-4' HHT and/or H-l'/H-5' ROE, H-4'/H-5' HHT H-2'/H-3' ROE, H-3'/H-4' HHT H-2'/H-3' ROE, H-3'/H-4' HHT, H-4'/H-5' HHT H-4'/H-6' ROE, H-3'/H-4 HHT H-2'/H-17 ROE, H-16/H-17 HHT H-34/H-37 ROE, H-33/H-34 HHT H-19/H-21 ROE, H-20/H-21 HHT assigned as HHT because the protons were not coupled within each pair. The first two and the third cross-peaks were then classified as ROE/ROE and HHT/HHT relays, respectively (Table 3). The recognition of the HHT cross-peaks created the starting point from which ROE/HHT relayed crosspeaks, exhibiting the same as the ROE negative sign, were traced. However, ROESY spectra measured at shorter mixing times (200, 120 and 40 ms) were diagnostic for the discrimination of relay artifacts, since at 120 ms these cross-peaks, remarkably, vanished. The ROE/HHT cross-peaks observed in the ROESY spectrum measured at a mixing time of 400 ms are listed in Table 3. methanol-d, in pyridine-d, at a concentration of 15 mg ml-', using the pyridine signal at 8.71 ppm as the reference standard. Proton spectra were collected for 1024 scans in data points over a 4.0 khz spectral width (2.05 acquisition time) and zero-filled to 32K before Fourier transformation. The natural line width, measured for the COOCH, signal, was Hz. The spectrum was recorded with a resolution enhancement by a Lorentz-Gaussian transformation. Double-quantum filtered COSY 2D NMR spectra were acquired in the phase-sensitive mode. Data were collected in a matrix (48 scans for each increment) with a spectral width of 1937 Hz and processed in 1K x 1K matrix. Exponential multiplication and Gaussian apodization were applied prior to 2D Fourier transformation (LB = 0.1, AF = 0.12 and LB2 = 0.1, AF2 = 0.06 for the first and second dimensions, respectively). ROESY spectra were collected and processed with the same parameters as the DQ-COSY spectra. The 90"-t 1-~mix(300-t)n-t2 pulse sequence was applied with a time-shared spin-lock (effective value 1.4 khz) within the t, time. Four different ROESY spectra were obtained with z~~~ 40, 120, 200 and 400 ms. Crystallographic data for amphotericin B2 were connected with the hydrogen atom coordinates by application of the Crystal Structure Utility pr~gram.'~ Standard deviations for valence and dihedral angles determined by means of this program were 2.6" and 2.0, respectively. The methoxycarbonylamide of amphotericin B (2) was prepared from amphotericin B (Rh6ne-Poulenc, Paris) by the procedure described previously.'* ~ ~ ~~ EXPERIMENTAL All NMR spectra were recorded at 24 C on a Varian VXR 300 spectrometer. Samples were dissolved in 12% Acknowledgements This work was supported by the Ministry of National Education and the State Committee for Scientific Research (Warsaw, Poland) and by the University of Camerino (Italy). REFERENCES S. Omura and H. Tanaka, in Macrotide Antibiotics: Chemistry. Biology and Practice, edited by S. Omura, pp Academic Press, New York (1984). (a) W. Mechlinski, C. P. Schaffner and P. Ganis, Tetrahedron Left (1970); (b) P. Ganis, G. Avitabile, W. Mechlinski and C. P. Schaffner, J. Am. Chem. SOC. 93,4560 (1 971 ). R. Maehr, R. Yang, L.-N. Hong, C.-M. Liu, M. H. Hatada and L. J. Todaro, J. Org. Chem. 54, (1 989). J. M. Brown and P. J. Sidebottom, Tetrahedron 37, 1427 (1 981 ). J. Pawlak, K. Nakanishi, Y. lwashita and E. Borowski, J. Org. Chem. 52,2896 (1987). (a) S. L. Schreiber and M. T. Goulet. Tetrahedron Lett (1987); (b) S. L. Schreiber, M. T. Goulet and T. Sammakia, Tetrahedron Lett (1987); (c) S. L. Schrieber and M. T. Goulet, J. Am. Chern. SOC. 109,8120 (1 987). J.-M. Lancelin, F. Paquet and J.-M. Beau, Tetrahedron Lett (1 988) (a) P. Sowinski, P. Gariboldi, A. Czerwinski and E. Borowski, J. Antibiot. 42, 1631 (1989); (b) P. Sowinski, P. Gariboldi, J. Pawlak and E. Borowski, J. Antibiot. 42, 1639 (1 989). J.-M. Lancelin and J.-M. Beau, J. Am. Chem. SOC. 112,4060 (1 990). W. Oroshnik and A. D. Mebane, Prog. Chem. Org. Prod. 21, 17 (1963). K. lmai and E. Osawa, Magn. Reson. Chem. 28,688 (1 990). B. T. Farmer, 11, S. Macura and L. R. Brown, J. Magn. Reson. 72,347 (1 987). A. Bax and D. G. Davis, J. Magn. Reson. 63, 207 (1 985). I. VickoviE, CSU (Release 7988). Faculty of Science, University of Zagreb (1 988). A. Czerwinski, W. A. Konig, P. Sowinski and E. Borowski, J. Antibiot. 40, (1 987).