1. Structure elucidation by NMR and computer-assisted structure generation of bi- and tri-cyclic products derived from electrocyclic rearrangements in biomimetic synthesis Barbara Odell*, Tim D.W. Claridge, John E. Moses, Sebastian Bruckner, & Jack E. Baldwin The Dyson Perrins Laboratory, Department of Chemistry, University of Oxford, OX1 3QY, UK, Barbara.Odell@chem.ox.ac.uk INTRODUCTION Biomimetic synthesis of natural products is a well established area of chemistry with applications to the pharmaceutical industry. Two areas of interest to the Baldwin group at Oxford University are i) the crispatene family of compounds 1 (I) which are believed be mild cytotoxic agents; and ii) the SNF4435C and SNF4435D family 2 of hexasubstituted bicyclo immunosupressants 2 (II) isolated from the culture broth of a strain of Steptomyces spectabilis. Although very diverse in structure, it is believed that both classes of compound are originally derived from biosynthetic transformations involving propionate and acetate. One interesting aspect is that these proposed biosynthetic schemes are believed to involve polyene intermediates. This prompted the Baldwin group to consider using polyenes as precursors in chemical transformations using thermal, photochemical and heterogeneous catalysis as possible routes to biologically active molecules. Extensive 1D and 2D NMR analysis of the resulting products was employed to elucidate the structures of the rearranged products. The complexities we encountered in determining unambiguous structures led us to consider the application of Computer-Assisted Structure Elucidation (CASE) programs in this task. The process of assigning unique structures consistent with 1 H, 1 H and 1 H, 13 C correlation data proved to be unusually challenging for these systems due to four principle features:  the general lack of vicinal proton-proton couplings  the ambiguity introduced by 2 J CH, 3 J CH, 4 J CH and even 5 J CH correlations in HMBC experiments of these extended conjugated systems  the potential variety and complexity of structures which may, in principle, result from the electrocyclisations  the time required for data analysis and identification of candidate structures In general when seeking unknown structures, there may be several candidates consistent with the available NMR data, but the spectroscopist may only consider one or perhaps a smaller subset of them. Identifying functional groups and smaller fragments is not usually a problem. Connecting them together is where the difficulties can arise. Hence, we wished to employ a computer program which would utilize knowledge of the functional groups together with correlations from 2D NMR data to produce all viable structures consistent with the input data in a short time. The development of expert systems (ES) or Computer-Assisted Structure Elucidation (CASE) programs has flourished over recent years 3-9 : StrucEluc 3, X-PERT 4, RASTR (STREC) 5, COCON 6,7, and LSD 8,9. We chose to examine the Nuzillard program ‘Logic for Structure Determination’ or LSD. This is made freely available by Jean-Marc Nuzillard and is reasonably user-friendly in terms of data input which consists of 1D 13 C, COSY, HMQC/HSQC and HMBC data. 1 H and 13 C NMR spectra for the cyclisation product: note the scarcity of proton resonances and couplings HMBC of the cyclisation product The round circles on the HMBC spectrum indicate 4 J CH (blue, H 15 to C 19 ) and 5 J CH (red, H 23 to C 22 ). Example 1 – Photochemical Rearrangement of the all-(E) Pentaene Summary scheme for the all- (E) Tetraene Conclusions The work reported here shows that the LSD Computer-Assisted Structure Elucidation (CASE) program is a powerful tool for utilizing 1D and 2D NMR data to generate chemical structures with viable connectivities. In cases such as the examples shown above, where long-range heteronuclear 1 H- 13 C NMR correlations are too complex to provide rapid solutions to the assignment problem, we discovered that the LSD ‘ELIM’ command was very useful for overcoming assignment discrepancies arising from ambiguities over very long-range correlations (n>3). We would recommend that use of such structure generation programs be augmented by additional nOe studies to cross-check the proposed structures and to yield stereochemical data. Together they from an integrated approach to 3D chemical structure elucidation yielding consistent structural and stereochemical assignments for the molecule in question. Spectra for the various products analysed were recorded on: a Bruker DRX500 spectrometer equipped with a triple resonance inverse probe, a Bruker AMX500 spectrometer equipped with a broadband probe, and/or a Bruker DPX400 equipped with an inverse probe. The LSD program was run on a Silicon Graphics O2 workstation. Details of the LSD program can be obtained from: http://www.univ-reims.fr/Labos/UPRESA6013/GNOSIE/LSD/ * = traces of impurities S = starting tetraene I = isomerisation intermediate THE NMR STRUCTURAL ASSIGNMENT PROBLEM LSD input file for the cyclisation product The file format is an edited example of the LSD input file for this compound. LSD output file for the cyclisation product ; sb1311;File identity DISP 1;Define output format HIST 1;Display progress during structure generation ELIM 2 5;Define eliminations for n J CH > 3 MULT 1 C 2 0;Define atom numbers, type, hybridisation & multiplicity eg 13 C from DEPT, APT MULT 22 C 3 3 MULT 23 C 3 3 MULT 24 X 3 0;Define heteroatoms or dummy groups eg NO 2 MULT 25 O 2 0 MULT 26 O 3 0 VALE X 1;Define valence of dummy groups HMQC 3 3;Define HMQC connectivities HMQC 5 5 LSD program suggests a crispatene core solution for the cyclisation product COSY 9 11 ;Define COSY correlations ( 2/3 J) COSY 10 12 COSY 14 20 COSY 3 13 HMBC 1 3 ;Define HMBC correlations ( # 13 C-# 1 H) HMBC 1 13 HMBC 1 14. HMBC 22 23 HMBC 23 5 HMBC 23 3 BOND 4 11 ;Define known bonding patterns eg within esters, aromatic rings or other known fragments BOND 4 12. BOND 14 26 BOND 14 20 EXIT ;Terminate file Example 2 – Photochemical and Thermal Rearrangements of the all- (E) Tetraene Photochemical Cyclised product: crispatene core with observed nOes Intermediate: 1 st isomerisation product Starting tetraene Thermal Tricyclic core with observed nOes References 1 J E Moses, J E Baldwin, R Marquez, R Adlington, T D W Claridge and B Odell Organic Letters Vol 5 No 5 2003 661-663. 2 J E Moses, J E Baldwin, R Marquez, R Adlington and A R Cowley Organic Letters Vol 4 No 21 2002 3731-3734. 3 K A Blinov, D Carlson, M E Elyashberg, G Martin E R Martirosian, S Molodtsov and A J Williams Magn Reson Chem, 2003, 41, 399-372. 4 M E Elyashberg, E R Martirosian, Y Z Karasev, H Thiele and H Somberg Anal.Chim. Acta 1997, Vol 337,265. 5. M E Elyashberg, VV Serov, E R Martirosian, L A Zlatina, YZ Karasev, V N Koldashov, Y Y Yampolskiy. J. Mol. Struct. 1991, Vol 230, 191. 6 T Lindel, J Junker, M Köck. Eur. J. Org. Chem. 1999 Vol 3: 579. 7 M Köck, J Junker, W Maier, M Will, T. Lindel Eur. J. Org. Chem. 1999 3 573. 8 J-M Nuzillard and G Massiot Tetrahedron 1991; Vol 47 3655. 9 S V Ley, K Doherty, J-M Nuzillard and G Massiot Tetrahedron 1994; Vol 42 12267. Acknowledgements We would like to thank Dr. Jean-Marc Nuzillard of the University of Reims, France, for making his LSD program available to us and for helpful discussions. We also thank Mr Charles Baker-Glenn for his valuable assistance in the preparation of this poster. Crystal structure of the tetraene showing steric clash of adjacent methyl groups;- a driving force for isomerisation The unusual tricyclic core suggested by NMR via LSD and nOes for the thermal rearrangement was subsequently confirmed by crystallographic analysis of a suitable derivative, which also proved the absolute stereochemistry (stereo view shown). The ELIM command allows the user to define any number of HMBC correlations that may arise over more than 3-bonds. Thus ELIM 2 5 allows structures to be generated that can have up to 2 correlations arising over 4 or 5 bonds. This is useful when no structures can be generated assuming n  3 for all n J CH.