The Solid-State and Solution-State Reassigned Structures of Tagitinin A , a 3 , 10-Epoxy-Germacrolide from Tithonia diversifolia , and the Interconversion of 3 , 10-Epoxy-Germacrolide Conformational Families via a Ring-Atom Flip Mechanism

* e-mail: rglaser@bgumail.bgu.ac.il; delgado@servidor.unam.mx # Dedicated to Prof. Kurt Mislow on the occassion of his 80th birthday and to Prof. Alfonso Romo de Vivar, for his 50 years of research at the Instituto de Química, UNAM. “...his wisdom shall endure...” Pirke Aboth, chapter III, verse 12. The Solid-State and Solution-State Reassigned Structures of Tagitinin A, a 3,10-EpoxyGermacrolide from Tithonia diversifolia, and the Interconversion of 3,10-Epoxy-Germacrolide Conformational Families via a Ring-Atom Flip Mechanism


Introduction
Horeau's rule 1 to determine the absolute configuration of chirotopic stereogenic secondary alcohols is considered to be a well-known and proven method, 2 and has been reviewed by Brewster 3 and Horeau. 4 It was used to determine the (R/S)-absolute configuration and subsequent , -orientation 5,6 of secondary hydroxyl groups in 3,10epoxy germacrolide 6,7-trans-lactone sesquiterpene natural products whose 3,10-oxiranyl oxygen affords a -C(3)-O-C(10)-fragment common to either a 3(2H)furanone [e.g. zexbrevin 7 (1)] or to a cis-fused tetrahydrofurano moiety [e.g. tagitinin A 8 (2)] (both drawn with (6R,7S)-stereochemistry). The skeletons of these compounds are based upon 1-isopropyl-4,8dimethylcyclodecane (germacrane). 9 In this method, a chiral secondary-alcohol will react faster with one of the enantiotopic C 2 H 5 C*H(Ph)C(=O)O-carbonyl groups in excess optically inactive 2-phenylbutyric anhydride [a mixture of d,l and meso-diastereomers] than with its enantiotopic counterpart in a type of kinetic resolution. Isolation of the excess unreacted 2-phenylbutyric acid enables correlation of its sign of optical rotation with the spatial disposition of small, medium, and large local environments around the secondary carbinol carbon.
Zexbrevin (1) has been isolated from the Zexmenia brevifolia plant, and its structure and stereochemistry were reported by Romo de Vivar et al. 7 It was converted in a number of steps to 8-desmethylacryl-hexahydroxyzexbrevin [originally drawn with 4 ,11 -dimethyl groups, but now known by X-ray diffraction analysis 10 to be 4 ,11 and subsequently illustrated as 3] and the stereochemistry of the free 8-OH was ascertained according to Horeau's procedure as follows. The excess recovered (-)--phenylbutyric acid was reported to exhibit [ ] D 27 -12°, representing an optical yield of 53.3%. According to the rule, the small, medium, and large local environments around C (8) are represented as in 4, and the absolute configuration was assigned as (8S). Using a (6R,7S)-6,7-trans-lactone skeleton and the (8S) result, the authors 7 reported an -orientation for the 8-hydroxyl group. However, the X-ray crystallographically determined structures of tetrahydrozexbrevin (5) 10 and 9acetoxyzexbrevin, 11 and phototetrahydrozexbrevin A 12,13 were later reported, and the orientation of the 8isobutyryloxy moiety was then found to be ' ' for all three compounds. Horeau's rule failed to predict the correct -orientation in this case.
Tagitinin A (2) was originally isolated by Pal et al. 14 from Tithonia diversifolia. Due to its similarity to tirotundin (6a) and its similar chemical behavior, Pal et al. 14 proposed structure 6b without specification of stereochemistry at C(1), C(4), and C(8). Herz and de Jong 8 undertook a more extensive study of tagitinin A and related compounds. They determined the stereochemistry of the 4-methyl and 8-isobutyryloxy moieties to be 4 and 8 , respectively, and reported 1 H and 13 C NMR chemical shifts [measured at 270 and 67.9 MHz, respectively, CDCl 3 ], plus some of the J HH coupling constants. 8 They also utilized Horeau's method to determine the stereochemistry of the chirotopic stereogenic secondary hydroxyl fragment at C(1), and found it to be (1S). 8 The excess recovered (-)--phenylbutyric acid was reported to exhibit [ ] D 27 -6.55° (benzene), representing an optical yield of 41.8%. 8 Based upon a (6R,7S)-skeleton for 6b, they proposed an -orientation for the 1-hydroxyl group. 8 Structure 6b depicts the stereochemistry of tagitinin A as illustrated in the report of Herz et al. 8 The failure of Horeau's rule with the deacylzexbrevin derivative 3 prompted us to reinvestigate the tagitinin A stereochemistry at C(1). This paper reports the solid-state twist-chair-boat (TCB) conformation of the nine-membered ring in 2 (as determined by single crystal X-ray diffraction analysis) and the reassignment of a -orientation for the C(1)-hydroxyl. The solution-state stereochemistry (utilizing NMR techniques) is also described herein. Furthermore, the classification of 3(2H)-furanone [e.g. 1] or cis-fused tetrahydrofurano [e.g. 2] 3,10-epoxy-germacrolides into respective oxacyclononane skew-chair-chair (SCC) and twist-chair-boat/skew-chair-boat (TCB/SCB) conformational families, 15 and their theoretical interconversion via a C(9) atom-flip mechanism is also discussed herein.

Results and Discussion
Solid-state stereochemistry of tagitinin A Tagitinin A (2) was isolated from T. diversifolia, and its chemical and physical properties were found to be identical to those described in the literature. 8 T. diversifolia (also known as "Mexican arnica") has been used in Mexican traditional medicine to treat inflammatory ailments. Its ethnopharmacology has been discussed recently. 16 Compound 2 was subjected to X-ray diffraction analysis, but the absolute configuration 17 of the chiral crystal was unable to be determined. 18 The resulting molecular structure (depicted as Ball and Stick 19 model 7) within these crystals showed that the originally proposed -oriented 1-hydroxy group was indeed inverted to a -orientation. From now on, model 7 will refer to the solidstate structure of crystalline 2. Evidently, within the diastereomeric transition-states of the kinetically controlled Horeau reaction, the "medium" versus "large" bulk or steric volume expressed by a particular sub-unit attached to the chirotopic stereogenic secondary carbinol carbon atom may not always be discernible by inspection of simple models. This may rationalize the failure of Horeau's rule to correctly predict both the 1-hydroxyl orientation in 7 and the 8-hydroxyl disposition in 3. Alternatively, one perhaps could argue that the two experimental findings for 7 [(1S)stereochemistry by Horeau's rule based and 1 -relative configuration by X-ray crystallography, as well as (8S)stereochemistry/8 -relative configuration for 1] are not actually mutually exclusive, but that the germacrolide skeleton simply has the opposite (6S,7R)-6,7-trans-lactone stereochemistry since neither the absolute configuration of 7 nor those of the zexbrevin-type compounds (e.g. 3,5) were ever determined by X-ray crystallography. However, this assertion is untenable since numerous X-ray crystallographic determinations of (6R,7S)-absolute configuration for other germacrolide 6,7-trans-lactone natural products are listed in the Cambridge Crystallographic Data Base 20 (CCDB), and one can assume that the biosynthetic pathways determining chirality are all very similar for this class of compounds. A few of the many recent representative CCDB examples of germacrolide X-ray crystallographic absolute configuration determinations were arbitrarily chosen. [21][22][23][24] Thus, it is very reasonable that the absolute configuration of the 6,7-trans-lactone fragment in the tagitinin A and zexbrevin 3,10-epoxy-germacrolide skeletons should also be (6R,7S), leaving us with the conclusion that Horeau's rule failed in both cases.
The O(1), and C(3-10) atoms of 2 define an oxacyclononane ring (8) having a twist-chair-boat 15 (TCB) conformation. Superimposition 25 of all the ring atoms of 2 on corresponding atoms in Density Functional Theory B3LYP/6-31g(d) 26 calculated TCB conformation oxacyclononane (8) and cyclononane 15 (9) models affords small root mean square (RMS) differences of only 0.168 and 0.109 Å., respectively [see comparison of torsion angles in Table 1]. Exchange of a methylene in model 9 into an ether oxygen in model 8 removes a transannular interaction between endo protons on C(2) and C(7), and as a result, brings the oxygen in 8 slightly closer to C(2) [2.838 Å and 2.992 Å O ... C(7) in experimentally determined tagitinin A (7) and in calculated model-8, respectively versus 3.352 Å C(2) ... C(7) in model-9]. The 14° H(6a)-C(6) ... C(7)-H(7a) and -3° H(3a)-C(3) ... C(10)-H(10a) torsion angles in 8 enable both a trans-6,7-lactone closure and 3,10-ethano bridging to proceed without strain. O-ethyl-tirotundin 27 (10), a 1-deoxy-O(2)-ethylated analogue of 2, has been isolated from T. rotundifolia. While coordinates of 10 are not to be found in the CCDB 20 using Conquest 1.4, 28 nor in the article itself, 27 endocyclic torsion angles for the nine-membered ring, the tetrahydrofuran ring, the lactone, as well as other selected torsion angles for 10, are available from Supplementary Material deposited for the paper. A comparison of these angles for 10 with those from 7 shows the same stereochemistry for both compounds (see Table 1). The root mean square (RMS) difference for the list of all 23 angles provided is only 3.0°. A subunit geometry comparison of 7 versus 10 affords RMS differences of 3.9° for the endocyclic angles of the TCB conformation oxacyclononane moiety, 2.7° for the 443 The Solid-State and Solution-State Reassigned Structures of Tagitinin A Vol. 16, No. 3A, 2005 tetrahydrofuran ring, and 2.2° for the lactone. The iconic representation of 10 (as well as others to follow) is a 2Ddimensional projection of the actual 3D-dimensional structure. 29 Anet has developed a very useful general analysis for subsequent assignment of substituent orientation (isoclinal, axial, equatorial) in rings of any size. 30  Dale 31 has defined "corner" positions as medium ring atoms which have identically signed synclinal (gauche, ca. 60°) endocyclic torsion angles on either side. Since corner atoms have two isoclinal ligands pointing outwards, gem-dimethyl groups are commonly found at corner positions of medium rings. Ring atoms C(3) and C(10) in the TCB C 2 -symmetry oxacyclononane and cyclononane models-8,9 are located at "corner" positions. Ligation of an ethano bridge to C (3,10) twists this region of the oxacyclononane ring in 7, but C(3) and C(10) still retain their character as "corner-like" positions [torsion angles for C(3) are 37° and 102°, while those on either side of C(10) are -89° and -59°]. In accord with this, one finds the only two doubly-substituted oxacyclononane ring atoms in 7 to be C(3) [ligated to O(2) hydroxyl and ethano bridge C(2)] and C (10) [ligated to C(14) methyl and ethano bridge C(1)]. Ordinarily, substituents on these two close proximity doubly-substituted ring atoms [C(3,10)] would have suffered severe steric mutual repulsion, but this is removed in 7 by linking the two groups together as an ethano bridge. In the parent TCB C 2 -symmetry cyclononane conformation 9 it is apparent that C(8) is also a "corner" position, since it is homotopic to C(3). However, in the TCB conformation for the cis-fused tetrahydrofurano family of 3,10-epoxygermacrolide lactones it is unlikely that the O(6)oxycarbonyl [-OC(=O)R] moiety ligated to C(8) in 7 would have an -orientation (i.e. syn to the neighboring C(14) methyl). Such a disposition would afford an unfavorable 1,3-cis-diaxial type relationship. On the other hand, the O(6)-oxycarbonyl moiety in the skew-chair-chair (SCC) conformation 3(2H)-furanone family of 3,10-epoxygermacrolide lactones can be either - [32][33][34] or -10-13,35-37 oriented since SCC represents a conformational change involving a flip of ring-atom C (9), and now the C(14) methyl is equatorial. Finally, one can predict that a epimeric C(15) methyl diastereomer of 7 would not be disposed to retain the TCB conformation since it would then suffer a transannular interaction with the inward pointing axial H(9 ).

Solution-state stereochemistry of tagitinin A
The 1 H and 13 C NMR spectral parameters of crystalline 7 dissolved in acetone-d6 are reported in Table 2 .55, respectively. The resolution at low temperature was used to assign H(13endo,13exo) by means of a NOESY spectrum measured at that temperature. In the 2D spectrum, H(8 ) afforded a markedly higher intensity cross-peak to the 5.66 ppm olefinic proton relative to that observed for the 6.08 ppm geminal neighbor. Therefore, the 5.59 and 6.10 signals in the 298° spectrum were assigned to H(13endo) [closer to H (8 )] and H(13exo) [closer to lactone carbonyl O(5)], respectively.
Irradiation of H (9 ) (9 ) is pointing into the interior towards its transannular H(4 ) and H(6 ) neighbors, while it would be pointing away from these protons in the SCC conformation.
For the C(3)-C(6) fragment, the measured 8.1 Hz 3 J(4 -5 ) and ca. 0 Hz 3 J(4 -5 ) values are also consistent with a TCB conformation and not those expected for the SCB. The dihedral angles H(4 )-C(4)-C(5)-H(5 ) and H(4 )-C(4)-C(5)-H(5 ) are respectively, 147° and 97° in structure 7. Therefore, the vicinal proton-proton coupling constants and NOE experiments are all consistent with an acetone-d6 solution-state nine-membered ring conformation that is similar to the TCB found for crystalline 7. However, inspection of the 13 C NMR and DEPT spectra shows the presence of low intensity 79.1, 47.9 (shoulder), and 106.3 methine signals that are ca. 0.1 ppm from methine resonances assigned to the respective C(1), C(2), and C(3) [major:minor ca. 3:2]. Low intensity 13 C NMR signals were not observed from other carbon nuclei in the molecule. Thus, while the nine-membered ring TCB conformation appears to be population biased, some small degree of flexibility appears to exist for the tetrahydrofuran moiety of tagitinin A in solution. The lower magnitudes of the minor component peak intensities are consistent with a slow exchange partner(s) either having a different puckering of the THFmoiety, or involving different rotamers about the C(

Nine-membered ring conformational interconversion and molecular modeling
Medium rings are large enough to undergo conformational interchange by segmental motion. 15 One of the mechanisms for medium ring conformational interchange is the ring atom-flip 42 (also referred to as "wagging"). 43 In this interconversion, one of the ring atoms flips to the other side of the ring, and in so doing, the axial and equatorial disposition of its exocyclic ligands are interchanged. This exchange of axial/equatorial orientations also occurs to exocyclic ligands on ring-atoms located at either side of the flipping atom. 15 Thus, H(8a) is equatorial in SCB type 3 (13) and axially oriented in SCC (14) B3LYP/6-31g(d) models of cyclononene. 15 Flipped ring-atoms conformations have been observed by X-ray diffraction analysis: two ring atom-flipped eight-membered ring conformations superimpose upon each other in conformationally dynamically disordered crystals of nefopam methobromide or methiodide quaternary ammonium salts. 44,45 The above stereochemical analysis can be combined with an earlier observation by Gershenzon et al. 39 who isolated 11b (a 1 -hydroxyl cis-tetrahydrofurano analogue) and the corresponding 1,2-dehydrofurano germacrolide lactone from the Viguiera microphylla plant. The authors J. Braz. Chem. Soc. noted that there were significant differences in the chemical shifts and coupling constants of the H (6,7,8,9 ,9 ) protons measured in their 1H NMR spectra [see 3 J(8 -9 ) and 3 J(8 -9 ) values for 11b noted above]. 39 These were attributed to the presence of a 1 -hydroxyl group or a 1,2-double bond, and also to conformational differences involving the orientation of the 8 -oxycarbonyl moiety. "The side chain appears to have an equatorial orientation in the 1 -hydroxyl compounds and an axial orientation in the 1(2) compounds." 39 We note that the hybridization of C(1) seems to be at the root for (T/S)CB versus SCC conformational preference, and subsequent diequatorial or diaxial disposition for the C(1) atom/8 -oxycarbonyl moiety. As an input structure for DFT B3LYP/6-31g(d) modeling, the 11c skeleton was converted into a 3(2H)furanone having an 8 -oxyformyl group for simplicity. A trigonal C(1) atom appears to be essential for an SCC conformational preference with an axially oriented C(1), while a tetrahedral C(1) affords a preferred SCB type 3 with an equatorial C(1). The 3(2H)furanone SCB type 3 conformational model (15) was found to be 0.87 kcal higher in energy than the SCC model (16). Keeping the C(1) carbonyl intact while changing the C(2)=C(3) double bond to a single bond still afforded an SCB type 3 model (17)   In conclusion, as with 8-desmethylacrylhexahydroxyzexbrevin (3), Horeau's rule also failed to predict the correct / -relative configuration for the 1-hydroxyl group in tagitinin A, and its configuration at C(1) must now be reassigned as 1 . Analysis of the 1 H NMR spectrum of 2 shows tagitinin A to maintain its crystallographically found TCB conformation and "W-like" H(O2)-O(2)-C(3)-C(4)-H(4 ) arrangement in acetone-d6 solution. Finally, the TCB/SCB type 3 conformations of the saturated/unsaturated ninemembered moieties within 3,10-epoxy-germacrolide rings can be interconverted to SCC by means of a C(9) ring atom-flip mechanism which changes the orientation of the C(1) atom and C(8)-oxycarbonyl moiety from diequatorial to diaxial. The stereochemistry of 3,10-epoxygermacrolide lactone structures in the CCDB, and the DFT B3LYP/6-31g(d) modeling results in this work can be interpreted as showing that tetrahedral C(1) atoms stabilize the TCB/SCB type 3 conformations, while their trigonal counterparts stabilize the SCC conformation.

Molecular modeling and graphics
Density Functional Theory B3LYP/6-31g(d) geometry optimized models 8,9,13-20 were produced with the Gaussian-98W revision A-7 program, 26 and all were found 447 The Solid-State and Solution-State Reassigned Structures of Tagitinin A Vol. 16, No. 3A, 2005 to have only positive values for vibrational frequencies. Superimposition of molecular structures was performed with the MacMimic 3.0 program. 25 Ball and stick-type noniconic molecular graphics were drawn with the Ball&Stick 3.8 3 program. 19 2D-iconic projections of the molecular models and X-ray crystallographic 3D-structures were generated using the combination of CS-Chem3D Pro 5.0 and CS-ChemDraw Ultra 5.0 programs. 29 NMR Spectroscopy 1 H and 13 C NMR spectroscopy were recorded at 500 and 125 MHz, respectively, at 298 K on a Varian Unity-Plus500 NMR spectrometer. Samples were measured in acetone-d6 using the deuterio solvent as an internal lock, and tetramethylsilane (TMS) as the internal spectral reference. DEPT (90° and 135° pulse angles) were used to determine the hydrogen multiplicity of the 13 C signals. COSY 2D NMR spectroscopy was used to ascertain the spin-spin coupling systems, and HETCOR 2D-NMR spectroscopy was used to correlate the 13 C and 1 H chemical shifts. NOE experiments were performed using the NOE-Difference technique, as well as by a NOESY 2D spectrum. 1 H spectral simulation was performed using the Gnmr 4.1 program. 41 Crystallography Crystallographic measurements were made on a Bruker Smart Apex automatic diffractometer with a CCD area detector using graphite-monochromated Mo K ( = 0.71073 Å) radiation. A clear, colorless plate crystal of C 19 H 28 O 7 , 2, [grown by slow crystallization from methanol] having approximate dimensions 0.40 x 0.20 x 0.18 mm was chosen, mounted on glass fiber, fixed on a goniometer head, and then placed in the X-ray diffractometer. The SMART 5.625 program 46 was used for centering, indexing, and data collection. Unit cell dimensions were obtained by a leastsquares fit of 3439 carefully centered reflections in the range of 2.27° 30.94°. Cell constants correspond to an orthorhombic system P2 1 2 1 2 1 cell with dimensions at 291(2) K of: a = 9.6580(13) Å, b = 9.9775(13) Å, c = 20.360(3) Å, V = 1961.9(5) Å 3 . For Z = 4 and FW = 368.41, the calculated density is 1.247 g cm -3 . Data were collected at 291(2) K using the scan technique. Space group determination was based upon systematic absences, packing considerations, a statistical analysis of intensity distribution, and the successful solution and refinement of the structure. Data were collected to a maximum value of 24.99° (100% completeness to ) and no significant decay was observed.
The structure was solved by direct methods and refined by full matrix least squares on F 2 using the SHELXTL97 program. 47 Atomic scattering factors were taken from Volume IV of the International Tables for X-ray Crystallography. 48 Non-hydrogen atoms were refined anisotropically, while hydrogens were placed at calculated positions, and refined as riding atoms on their respective attached atom, with the exception of those ligated to O(2) and O(3) which were located and refined as non-hydrogen atoms with a U 1.2 Å 2 thermal isotropic factor from the attached O-atom. At convergence, the final discrepancy indices on F were R(F) = 0.0523, R w (F 2 ) = 0.0990 and GOF on F 2 = 0.902 for the 3459 reflections with I net 2 (I net ) and 245 parameters refined with 0 restraints and 0 constraints. The largest difference peak and hole was 0.149 and -0.130 e.Å -3 .