3414
JEAN
A. H ~ T S U CAND K WILLIAMN.LIPSCOMB
Vol. 85
TABLE I N.M.R.SPECTRA OF COMPOUNDS IIa, I I b , I I c Compound
Aromatic protons
C-H,
IIa
Multiplet, intensities 2 and 3
IIb
a2bz pattern, intensities 2 and 2
IIC
azb2 pattern, intensities 2 and 2
Doublet, intens. 1 r = 5.62 Jsb 8.4 C.P.S. Broad doublet, intens. 1 Broad doublet, intens. 1 T = 4.40 r = 5.49 J a b 6.9 C.P.S. Broad doublet, intens. 1 Broad doublet, intens. 1 7 = 3.69 r = 5.50 Jab 5.5 C . P . S .
Anal. Calcd. for CzoH,&aO (330.38): C , 72.70; H, 5.49; h', 16.96. Found: C , 72.60; H , 5.32; N , 16.66. of P-methylphenylglyoxal was prepared in the The Same manner as 1 1 in ~ goyc yield, m.p. 136-1370; lit.20 a-form m.p. 145', 0-form m.p. 167-168". A n a l , Calcd. for CZIH20sT((328.21): c, 76.79; H, 6.14; h', 17.07. Found: C , 76.91; H , 5.95; N , 16.97. Proton Resonance SPectra.-The structures of the new cornpounds described in this work have been confirmed by integrated n.m.r. spectra.Z1 0-Ketosulfonesz2exhibit a singlet representing the methylene hydrogen atoms in the -CO-CH2-SOz- grouping. However, for the methylene group in the 8-ketosulfoxides Ia-d (--CO-CHz-SO-) a n a b Pattern is observed for the methY1eX1e protons due to the asymmetrj- caused by the sulfoxide group. I n the spectrum of I I a there is in addition to the aromatic protons (multiplet, intensity 5, ortho and meta-para hydrogens separated), a pair of doublets (total intensity 2 ) for the methylene group at 7 = 5.55 and 5.72 ( J a b 14 c.P.s.; relative intensities 0.18 (20) E. Llurio, GQZZ.chim. i t d . , 66, 89 (1935).
THE
Singlet, intens. 3 r = 8.00 Singlets, intens. 3 r = 8.00, 6.14 Singlets, intens. 3 r = 8.05, 7.60
and 1.82),23as well as a singlet a t r = 7.34 (intensity 3) for the methyl group. Similarly, Ib gives the following absorptions in addition t o the aromatic protons (azb2 pattern, intensity 4 ) ; a pair of doublets total (intensity 2 ) a t T = 5.R7 and 5.77 (J.,> 14 c.P.s., relative intensities 0.28 and 1.72) for the methylene group, a singlet a t 7 = 6.12 (intensity 3 ) for the niethoxy group and a singlet a t i = 7.27 (intensity 3) for the methyl gr(iup. Compound IC s h o w an anbzpattern for the aromatic hydrogcn atoms; a pair of doublets, T = lj.50 and 5.81, J A R 15.5 c.p.s. for the metllylene group, and singlets at = 7.25 7.58 for Id tile center the methylsulfinyI groups and p m e t h y l groups. peaks of the a b system for the methylene group are oxill- slightly separated, The n.m.r. spectra of the hemimercaptals I I a , IIb, and IIc exhibit a pair
i
of
doublets for the H,-C-OH,
grouping,
D
~
I
terium exchange in deuterium oxide sblution allowed the assignment of one doublet to the -OHb group. Further n.m.r. data are listed in Table I. ~~
(21) Spectra were t a k e n in chloroform-d a t 60 Mc. with tetramethylsilane as internal standard. (22) H.-D. Becker a n d G. A . Russell. J . Org. C h e m . , 28, 1896 (1563).
[CONTRIBUTION FROM
-CHn
-0Hb
Doublet, intens. 1 7 = 3.87
(23) Calculated; 6. 1,. M. J a c k m a n , "Applications of Nuclear Magnetic Resonance Spectroscopy in Organic Chemistry," Pergamon Presa, Kew York, N. Y . , 19.55.
DEPARTMENT OF CHEMISTRY, HARVARD U N I V E R S I T Y , CAMBRIDGE 38, MASS.]
Molecular and Crystal Structure of the Di-p-bromobenzoate of the Methyl Ester of Gibberellic Acid BY JEAN A. HARTSUCK A N D WILLIAMN. LIPSCOMB RECEIVED M A Y17, 1963 The complete molecular structure and stereochemistry, except for absolute configuration, have been determined by a single crystal X-ray diffraction study of the di-p-bromobenzoate of the methyl ester of gibberellic acid. The lactone ring is shown t o be trans t o the two-carbon bridge. The space group is Co2, and there are four molecules of CsaOeBrzin a unit cell having parameters a = 28.22, b = 7.69, c = 17.63 A , , and 8 = 125' 12'. The value of R = 2 1 1 F, I - I F, 1 1 / Z I F,, I is 0.13 for the 1978 observed diffraction maxima. The 30 H atoms of the molecule were not located in this study.
The gibberellins are active metabolites which have been isolated for about forty years from culture filtrates of the fungus gibberella fujzkuroi. Plants which have been subjected to an excess of a gibberellin grow abnormally rapidly in their early stages but then wilt and die before reaching maturity. On the other hand, gibberellins have also been isolated from healthy plants. Therefore, these natural products are thought to be normal plant growth hormones which cause disease when present in excess. A t least ten different, but similar, such compounds have been characterized. One of these compounds, gibberellic acid, has been most widely studied by chemists because its isolation2 from fungal cultures is well defined: reproducible, and of high yield. The chemical evidence, reviewed by Grove,3 for the molecular structure leaves some doubt as to the configuration of the lactone ring; but we shall show that contrary to chemical expectations the lactone bridge is in an a-orienta(1) P. W. Brian, J. F. Grove, and J. MacMillan, P Y O R Chem V. Ore F a t . Prod., 18, 3.50 (1460). (2) P. J. Curtis and B. E Cross, C h e m . I d . (London), 1066 (1954). (3) J. F. Grove, Qunrf Ret!. (London), 16, 56 (1561).
tion. Preliminary results of an independent X-ray diffraction study by McCapra, Scott, Sim, and Young4 on methyl bromogibberellate appeared after our study was well under way. These authors4 deduced the correct chemical structure of gibberellic acid from the stereospecificity of the bromination reaction which changes the molecular conformation. The present X-ray diffraction study of single crystals of the di-p-bromobenzoate of the methyl ester of gibberellic acid confirms the chemical structure, shows that the lactone ring is indeed trans to the two-carbon bridge and yields for the first time detailed bond distances and angles in the essentially unaltered ring structures of the molecule.
Experimental The di-p-bromobenzoate of the methyl ester of gibberellic acid was prepared by E. J . Corey and S. Barcza. Colorless, irregularly shaped single crystals were obtained from solutions in ethyl acetate. All crystals which were photographed were less than 0.08 mm. in cross section, and hence absorption and extinction corrections were not made. Reciprocal lattice symmetry of CZh, the systematic extinction of reflections for ( 4 ) F. McCapra, A . I. Scott. G . A . Sim, and D. W. Young, Proc. Chem.
Soc., 185 (1962).
~
-
Nov. 5 , 1963
DI-p-BROMOBENZOATE OF GIBBERELLIC ACID
+
which h k is odd, and the expectation t h a t the derivative, like gibberellic acid itself, is optically active uniquely determined the monoclinic space group as C23-C2. This space group was later confirmed as the structure determination progressed. Visual estiinates were made with the use of a scale prepared from timed single crystal reflections. A total of 1978 independent reflections TABLE 1.
LIST OF OBSERVED F’S
3415
H-l9,K,L(l)(-21, )5,4* 10 12 1 1 10 17 36 41 19 39 73 43 8, 35,30,68,31,57,2Z,15,25,2i.-:i4: 14’4*:5 4,3i,z:(3j(-is,jS 15 13,13,62,36,20,26 7* 63,38,21 6b,41 ,I ,-,15,18,21,21, (5j(-lO -2124 25 16 8* 8* 21 17 6* 15 H-20 K i(O)(-il ?)8’10:4*:8 5* 58 58 5* 23 4* 28 14 4* 2 2 , 1 5.38 48,4*. 42,4*, 64.23 -,45.2 5.4; 4; 7’8 12 ) 1 1 i 41 1 6’9 20 22,28,32,11,39,51 84’47 9 35 67’30’16 i9 1 1 15’14 16 i 0 ~ ~ ~ ~ - 1 6 , 2 ~ 1 1 , 1 7 , 1 7 , 1 ~ i, 12 , ~~ , 13 ~, ,~ 1 , ~ 1 , i 4 , ~ ~ , ~ 4 , $ 6 , i 1 ~ ~ 0 , ~ ,
-
in. i n
H-32, .~ k;i(O) (-1 6,-9) 16,16,17,-,4*, l9,18,11,( 2 )(- 16,-8) 1 1,7, 9 12 1 2 1 1 5* 7,12, H:33:K, i(1 !(-i6,-10)5,6,6,13,3*,3*,6, Absent reflections are starred; unobserved reflections are indicated by a dash. If the h even reflections are multiplied by 0.9462 and the h odd reflections are multiplied by 0.9507 these structure factors are on an absolute scale. (Table I ) were obtained from Weissenberg photographs of H k l for 0 5 H I 6 and hK1 for 0 5 K 5 5 , and from precession photographs of the hk0 level. Unit cell parameters of Q = 28.22 i 0.02, b = 7.69 f 0.03, c = 17.63 f 0.09 A . , and p = 125’ 12‘ f 20’were obtained from Weissenberg photographs upon which a powder pattern of S a C l was superimposed. Assumption of four molecules, confirmed below, in the unit cell yields a calculated density of 1.54 f 0.02 g. cm. - 3 . This value is in good agreement with a value of 1.521 f 0.002 g. c m . 3 , measured by pyconometric methods after removal of adsorbed air from the crystal with the aid of a vacuum line. The integrated intensities from the films were correlated by computer method^.^ In all of the later calculations, except for the determination of the scale factors, the 172 absent reflections were included each with a value of one-half of the minimum observed intensity in the appropriate region of the film. The usual corrections were made for Lorentz and polarization factors. In addition t o a single over-all scale factor for all intensities, a separate single scaling parameter was introduced, relating all reflections for which h is even t o all reflections for which h is odd. However, a preliminary value of this additional scaling parameter was obtained from the hkO precession photograph. We chose to introduce this additional parameter in all later refinements rather than depend upon the upper level precession photographs about the c-axis for a more complete correlation. The standard deviations u of the observed intensities were determineds during the correlation procedure. The weight ~ ( 1 assigned ,) t o a particular measurement I , is based upon the number of times t h a t the reflection is observed and on the precision of reflections with the same indices. The relation of u and ~ ( 1 t,o )an average intensity f is then U * = ( I ) z / Z w ( I i ) . Systematic errors, e.g., neglect of anomalous dispersion, have not been included, and no decrease in weights was made in the final correlation between common reflections taken about different crystal axes. Therefore the standard deviations are perhaps small, but we feel that they establish a reasonable basis for the relative weights used below in the least squares refinements. Structure Determination.-The coordinates of the two Br atoms in the moleciile were located from a three-dimensional Patterson function which had been sharpened so that the average intensity is independent of sin 8 . A three-dimensional electron density map based upon the 500 most intense reflections and
~ _ _ _ _ (a) P G
Simpson. P h D Thesis H a r v a r d University, 1968
3416
JEAN
A. HARTSUCK AND WILLIAMN. LIPSCOMB
Fig. 1.-Structure of the di-p-bromobenzoate of gibberellic acid. The absolute configuration was not determined in this study.
OH
system for CsrOsBrz. 112
a
, I
L
Fig. 5.-Structure Fig, 3.-Crystal
Carbons are
H0
17
Fig. 2.--Nutnbering
Fig, 4. -Perspective drawing of gibberellic acid. open circles; oxygens are blacked.
packing in C340SBr2. The origin is surrounded by a diamond.
upon the phases calculated frotn Br atonis failed to yield the structure. We noted that a t least 400 of these most intense reflections were in common with a seperate list of 500 reflections which, at this stage, gave best agreement between observed and calculated intensities ( B r only). Concurrently, we computed a three-dimensional minimum function in which the minimum value was taken a t each point from the superposition of Patterson functions6 translated t o the two Br positions in the molecule and t o the two other Br positions related by the crystallographic twofold axis. From this function 32 peaks were chosen as C atoms, of which four were later shown to be incorrect, and a three-dimensional electron density map was then computed from the 500 most intense reflections, but with phases based upon the Br atoms and these 32 peaks assigned as C atoms. The contours of this function were placed on plastic sheets, frotn which 38 C and 0 atoms were located. At this stage only atoms 20,25,26, and 27 (Fig. 1 and 2 ) failed t o appear. Three additional consecutive electron density maps yielded these additional atoms and also gave improved position pnrameters for all atoms. The structure was refined by three-dimensio ial least squares methods in which off-diagonal terms were included. One cycle of refinement of all distance parameters for Br, C , and 0 of one individual, isotropic temperature on each of the atoms yielded a value of RF = Z [ 1 F, 1 - 1 F, I 1 / Z 1 F, 1 = 0.23. This cycle was then followed by one cycle of refinement of distance parameters and six (anisotropic) thermal parameters on each Br, C, and 0. Because of the limitation of storage of the I B M 7090 this cycle had to be carried out in three separate computer runs in which different sets of atoms were refined, but in which several atoms in cotnmon between different runs were retained. Therefore, not all off-diagonal interactions have been included in this refinement, and also in a later refinement. The value of R F reduced t o 0.15. At this stage, an accurate three-dimensional model was built, arid €I atoms, which could not be found with certainty in difference electron density maps, were then introduced in stereochen~ically reasonable positions. The coordinates of these H atoms were then used to calculate fixed atom contributions in one more cycle of three-dimensional refinement of position and anisotropic thermal parameters of Br, C, and 0 atoms. The final parameters are shown in Table 11, and the final value of R F = 0.13 is compared in Table I11 with other criteria of agreement. The slightly higher value of R F for the h02 reflections is not surprising, because it was the first level estimated, and the data processing program also indicated that these reflections had significantly larger standard deviations than the average. A final three-dimensional electron density map from which Br, C, an
H2
of gibberellic acid ( C ~ Y H Z ?inOthe ~ ) presulnably correct absolute configuration
Br atoms. These peaks are attributed t o lack of convergelice of the difference Fourier series along the relatively short y-axis. S o abnormalities in van der Waals contacts were found, and the tnolecular structure fills the unit cell so that there is no room for additional atoms. The molecular packing in the- crystal is shown in Fig. 3.
Results and Discussion The perspective drawing of the molecule in Fig. 4 is the mirror image of the known'.* absolute cow figuration. Bond distances and angles are listed in Table IV along with standard deviations computed from the variance--covariance matrix including off diagonal terms. .I reasonable limit of three times these standard deviations should, perhaps, be placed on these values, especially in the absence of any reasonable way to treat the unknown systematic errors. As sometimes occurs, the bond distances in the part of the molecule nearest the heavy atoms seem to be less reliable than those elsewhere in the structure. Half of the bond distances in the benzene rings have standard deviations of +0.03, while only ','18 of the rest of the $stances have standard deviations exceeding ~ k 0 . 0 2A. .llso, the Br atoms tend to produce problems of convergence of the electron density in their immediate vicinity, and the largest anisotropic thermal motions in the structure occur in the region of the p Br-benzenoate groups. Hence, we feel that the larger systematic errors are probably present in the benzene rings, and that the bond distances and angles in the interesting part of the molecule can probably be considered reliable to within about twice the calculated standard deviations. The most interesting region of the molecule is a t the junction of the central five-membered ring, the lactone ring, and the cyclohexene ring. The angles a t C p Iare 1.5-21-22 = 10,5', 22-21-32 (lactone 0) = 101', 22-21-28 := 112', while the angle external to the rings is 15~-21-28 = 123'. At C Z 2the angles are 21-22-~31= 99' and 21-22-23 = 101' in the rings, and 23--22-31 = 11iOexternalto therings. Theangle 12-17-16 = 102' is also of interest in connection with possible small values of bond angles due to strain. ( 7 ) G Stork and H. S e w m a n , J A m . Chem Soc., 81, 3168 (1959). ( 8 ) J. F Grove and T . P. C . Mulholland, J . Chem. S O C . , 3007 (1960)
DI-~-BROMORENZOATE OF G I B B E R E I ~ ACID LIC
Nov. 5 , 1963
3417
T A B L EI1 F I X A LS T R U C T U R PA E RAMETERS Atomic posil ions
Atomic positions
r/b 0.673 999 752 755 ,813 ,861 ,876 820 895 ,877 ,956 ,972 1.124 1.093 1.011 0.861 ,808 ,936 1.011 1.079 0.940 ,806 ,719 ,626 ,661 507 ,413 1.057 0.987 ,813 ,698 ,821
%/a
%/a
Z/C
Ylb
H 59 (20) 1 105 462 580 0.376 0 845 711 256 H 60 (22) 406 1 200 075 H 61 (28) 630 329 1 120 679 026 H 62 (29) ,238 c-1 0 775 H 63 (30) 678 000 ,203 c 5 905 864 070 H 64 (44) 269 C 6 990 H 65 (43) 965 190 360 c 7 815 935 405 H 66 (41) 394 C 8 750 845 H 67 (40) 285 ,229 c 9 602 660 180 H 68 (23) ,147 0 10 294 Anisotropic temperature factors 0 11 ,274 c 12 Bnb 821 &a 812 818 828 ,202 C 13 Br 1 - 30 67 - 19 48 630 150 ,135 C 14 2 24 -2 430 99 Br 2 26 ,185 C 15 1 46 24 32 320 130 c 3 ,253 C 16 81 41 190 c 4 17 12 27 .239 c 17 18 - 10 520 25 -7 c 5 76 ,354 C 18 27 45 200 C 6 75 56 31 363 c 19 170 93 26 29 c 7 32 0 ,439 c 20 31 16 -11 41 87 330 C 8 ,125 c 21 -2 88 - 10 28 c 9 35 270 ,185 c 22 -4 -8 87 34 0 10 31 390 ,237 C 23 2 0 11 73 23 490 - 16 26 ,328 C 24 11 c 12 23 95 240 18 30 ,373 0 25 33 C 13 34 40 76 33 340 349 ,665 0 26 52 34 66 C 14 28 100 420 ,439 ,703 c 27 21 24 C 15 19 17 74 260 ,074 ,635 C 28 - 14 15 18 26 C 16 80 300 ,045 ,666 C 29 11 13 C 17 19 76 320 26 ,066 ,687 C 30 22 - 10 - 39 C 18 71 290 25 ,108 ,662 C 31 22 21 24 c 19 0 77 280 ,049 570 0 32 - 24 34 34 c 20 - 37 75 330 039 ,676 ,597 c 33 c 21 - 23 - 23 19 68 260 16 - ,021 ,568 ,581 0 34 24 11 c 22 6 80 150 29 ,137 ,521 690 c 35 2 8 C 23 18 80 180 26 133 750 ,801 0 36 36 c 24 10 19 69 320 20 020 766 ,802 0 37 0 25 - 29 29 20 79 370 17 101 784 807 C 38 7 29 270 32 0 26 19 86 ,176 ,846 ,838 c 39 c 27 58 45 31 75 450 19 ,269 ,867 ,824 C 40 24 C 28 - 32 89 240 29 - 13 ,338 ,924 ,864 C 41 C 29 30 5 98 210 34 25 ,316 962 .928 C 42 C 30 - 1 64 380 25 20 8 ,220 ,940 ,940 c 43 22 C 31 74 6 - 15 210 28 ,150 ,884 .893 c 44 - 12 22 0 32 74 380 19 8 ,470 ,412 ,820 H 45 (8)" 14 c 33 69 350 19 17 - 16 ,414 474 ,920 H 46 (7) - 13 - 15 0 34 78 380 20 25 ,138 ,338 ,815 H 47 ( 5 ) -2 c 35 - 38 81 280 28 30 ,218 ,725 ,275 H 48 (4) 24 - 10 33 - 15 79 0 36 460 ,140 479 1.095 H 49 (13) 0 37 -6 27 63 640 30 - 46 ,210 1,250 ,515 H 50 (13) 11 44 C 38 92 26 34 480 ,075 1.045 ,507 H 51 (14) c 39 8 26 32 81 510 18 ,120 540 1.230 H 52 (14) 12 24 25 -41 c 40 70 440 230 1,100 H 53 (15) 633 -2 C 41 76 -7 32 420 31 ,178 H 54 (17) ,505 0.765 42 110 -5 C 42 140 35 - 18 7 298 H 55 (17) ,560 0.750 93 c 43 -1 21 460 20 ,370 ,663 1.055 H 56 (18) 24 c 44 60 - 16 27 - 16 340 ,405 0,820 ,660 H 57 (18) 500 ,640 1,050 H 58 (20) These hydrogen atom positions were determined from an accurate three dimensional model and were never refined. The carbon atoms to which the hydrogens are attached are indicated in parentheses. .hisotropic temperature factors 0,j in the P3r12 2P12hk 2&kI 2 p l : J d ) are a six-parameter correction for thermal motion. Values expression exp - ( p ~ , h , ~ &?k2 for pi, have been multiplied by lo4. Each number has only two significant figures. Br 1 Br 2 c 3
0 300 1.036 0.342 ,319 ,351 ,409 43% 398 443 425 497 539 523 ,540 ,601 ,611 ,553 642 ,598 ,606 ,619 ,666 ,641 ,688 ,738
-
+
+
+
+
Atom Czzis 0.7 A. out of the plane of the five-membered carbon ring, and 0.6 A. out of the plane of the lactone ring of which it is also a member. The nearplanarity Of the Other atoms Of the lactone ring is in , ~ we agreement with observations of M a t h i e ~ o n and
find that the bond Czl-Oa2 is 0.13 A. longer than bond C33-032, again in substantial agreement with his observations concerning the geometry of lactone rings. (9) A. McL. Mathieson and J. C. Taylor, Tefrohcdron Lellers. 17, 590 (1961).
3418
JEAN
A. HARTSUCK AND WILLIAM N. LIPSCOMB
TABLE 111
RF
=
"
' '
SUMMARY OF AGREEMENT
Fo' - Fc' 1 = 0.13
R F= ~
"I2
'I
ZIFOI
RwF?=
[W _
I;;,F,,;F,/~/~]"2 _ _
-
- 1 F o ' 2 1= 0.23
z Fo2
~ = 0.28
as a function of sin e-Range of sin 0 RF
-RF hkl for
RF
h even h odd 1 even 1 odd hOl
0.11 12 12 .ll .18
Okl
.10
0,OO-O.40 .40.50.60.65.70-
.i5.80.85.go-1
.50 .60 .65 .70 .75 ,80 .85 .90 . 00
0.13 .12 ,13 13 .12 .17 .16 .15 .16 .17
Vol. 85
In the ester groups we find that one benzene ring is tilted about 15' about the C6H6-C02bond relative to the plane of the CO, group, whereas the other benzene ring and its attached COz groups are substantially planar. This molecular feature appears to us to be associated with the requirement for near-parallel stacking of the benzene rings in the crystal, and hence this deviation from coplanarity may not be a property of the molecule in solution. For convenience to the reader we show in Fig. 5 the structure of gibberellic acid in the presumably correct absolute configuration. We close with a comment on the strategy of the investigation. Our intuitive feeling is that superposition procedures are more powerful methods when the ratio of heavy to light atoms is as small as that present here. The ratio Z N 2 / 8 n 2 , where N is the atomic number of the heavy atom and n is the atomic number of the light atoms, is 1.41. We were therefore led to
TABLE IV C-Br
Bond distances," A. C.p'-C.p2
Bond angles) degrees Atoms
Atoms
1-3-4 122 22-3 1-30 107 1-3-8 117 31-30-29 114 3-4-5 121* 30-29-28 124 C-C (benzene) 29-28-2 1 4-3 1.34* 4-5-6 117 118 5-4 1.43* 5-6-7 122 28-21-22 112 6-5 1.41 6-7-8 119 22-2)-32 101 cap2-Cbenzene 7-8-3 119 2 1-32-33 107 7-6 1.34t 8-3 1.41* 9-6 1.49 &3-4 121 * 32-33-31 108 8-7 1.45 39-38 1.49* 7-6-9 124 32-33-34 118 5-6-9 114 34-33-31 134 * 40-39 1.38 c=c 6-9-10 127* 33-3 1-22 102 41-40 1.40* 1.33 20-19 112 22-3 1-35 115 6-9-1 1 42-41 1.41* 29-28 1.34 10-9-11 121 30-3 1-35 112 43-42 1.43* c=o 121 16-23-24 113 44-39 1.43 9-11-12 10-9 1.25 11-12-13 117 22-23-24 113 44-43 1.39 25-24 1.20 11-12-19 110 23-24-25 124 C.p'-Cap' 34-33 1.15 118 23-24-26 111 11-12-17 13-12 1.60 38-37 1.21 12-13- 14 116 25-24-26 125 14-13 1.51 C.pr0 13-14-15 112 24-26-27 118 15-14 1.55 11-9 1.36 14-15-16 112 29-30-36 114 16-15 1.56 26-24 1.30 15-1 6-1 7 112 31-30-36 105 17-12 1.54 33-32 1.42 16-1 7-1 2 102 30-36-38 119 17-16 1.56 38-36 1.34 17-12-13 105 36-38-3 7 126* 18-16 1,58 17-12-1 9 101 36-38-39 112 21-15 1.51 Cspa-O 12-19-20 124 * 37-38-39 121* 22-21 1,53 12-11 1.43 23-16 1.50 27-26 1.49 12-19-18 108 38-39-40 121 23-22 1.59 32-21 1.55 20- 19-18 127 38-39-44 118 31-22 1.55 36-30 1.47 19-18-16 108 39-40-41 121 121 40-41-42 31-30 1.57 18-1 6-1 7 95 35-31 1.51* 15-16-23 105 4 1-42-43 118* 16-23-22 107 42-43-44 122* 23-22-21 101 43-44-39 118 22-21-1 5 105 44-3940 121 21-15-16 107 41-42-2 123 21-15-14 118 43-42-2 120 21-22-3 1 99 Unmarked distances have standard deviations of f 0 . 0 2 A. Those distances indicated by a n asterisk are f 0 . 0 3 A. and by a dagger are fO.O1 A. All standard deviations include the errors of the unit cell dimensions. * The central atom is the vertex of the angle. Unmarked angles have standard deviations of + l o . Those angles indicated by a n asterisk are f2". 3-1 42-2
1.90 1.84
19-12 19-18 24-23 28-21 30-29 33-3 1
1.53 1.46 1.54 1,52 1.42 1.51
Atom CI7is also of special interest. It is the out-ofplane member of the five-membered ring to which i t belongs, and the six-membered ring of which this atom is also a member is in the boat conformation. I t would be of interest to carry out X-ray diffraction studies of other large molecules, particularly those containing five-membered rings and highly fused ring systems. There is n o molecule having sufficiently well established structural parameters for comparison with the results described here.
abandon the usual heavy atom method without a fair trial based upon use of a larger number of reflections and upon weights for the Fourier coefficients which depend upon the reliability of phase determination by the heavy atoms alone. Bromine was selected as a heavy atom in order that the standard deviations for light atoms would be small. Also, two Br atoms were introduced in the molecule in order to reduce the chance that these heavy atoms would produce pseudosymmetry in the crystal, and the benzene rings were added
TOTAL SYNTHESIS
Nov. 5, 1963
in order to facilitatelo the location of known molecular features in order to improve the reliability of location of the gibberellic acid part of this ester. In summary, this strategy was successful, but the disadvantages were the introduction of a relatively large number of extra atoms in the determination, and the occurrence of the strongly anisotropic thermal vibrations normal to the planes of the benzene rings. (10) M (1958)
G
Rossmann
and
W
N
Lipscomb, Tefrahedron, 4, 275
OF
(+)-a-ONOCERIN
3419
Acknowledgment.-We wish to thank Professor E. J. Corey and Mr. S. Barcza for the preparation of this derivative, for helpful discussions, and for the measurement of the density of the crystal. We acknowledge use of the IBM 7090 computers a t the Harvard and the M I T Computation Centers, and support of this research by the National Institutes of Health, the National Science Foundation (including a Fellowship to J. A. H.), and the Air Force Office of Scientific Research.
[CONTRIBUTION FROM THE CHANDLER LABORATORIES OF COLUMBIA UNIVERSITY, NEW YORK 27, N. Y . , A N D THE WEIZMANN INSTITUTE OF SCIENCE, REHOVOTH, ISRAEL]
Total Synthesis of Polycyclic Triterpenes : The Total Synthesis of (+)-a-Onocerin BY GILBERTSTORK, ALEXMEISELS,AND J. E. DAVIES' RECEIVED MAY28, 1963 The total synthesis of the natural ( +)-a-onocerin is described.
The elucidation of the structure of the triterpene cyonocerin (I) by Barton and Overtonj2in 1956, was of considerable importance on two counts. First, this triterpene is an uncomplicated example of the squalene biogenetic hypothesis for the usual pentacyclic triterpenes and steroids, in which the assumed concerted cyclization is starting from both ends of the chain rather than proceeding ~nidirectionally.~Second, Barton and Overton2 were able to cyclize a-onocerin, v i a the @-isomer11, to a pentacyclic triterpene system, y-onocerin (111), which, although it has not been found in nature so far, is simply related to the natural pentacyclic triterpene hydro~yhopanone~ (V). In fact, the dehydration product IV of hydroxyhopanone has been made from onocerin by partial synthesis by Schaffner, et ~ 1 We . ~ now record the details of the total synthesis H
H
of natural a-onocerin. The starting point of the synthesis was the assumption that the Kolbe electrolytic coupling6 of suitable y-ketocarboxylic acids should be a simple solution to one of the problems of synthesizing this particular symmetrical molecule. Our initial experiments showed the feasibility of such a scheme: 2oxocyclohexaneacetic acid, under the usual electrolytic coupling conditions, gave a very good yield of 1,2-di-(2oxocyclohexy1)-ethane (VI). This result allowed the
VI dissection of the synthetic problem into three parts: (a) the construction of the proper ketoacid, e.g., VI1 and its resolution; (b) the coupling of the substance to the symmetrical diketone VI11 which should be identical with the known ozonolysis product from natural onocerin; (c) the transformation of the two carbonyl groups of the dione into the two exocyclic methylene groups of the final product. We will now turn our attention to the first goal, the construction of the (-)-hydroxyketoacid VII. The sequence of vicinal substituents present in VI1 ap-
vIn
VI1
HsC CH3 (1) For a preliminary communication see J . A m . Chem. Soc., 81, 6510 (1959). (2) D . H . R Barton and K . H . Overton, J . Chem. SOL.,2639 (1955). (3) Cf. G. Stork and A . W . Burgstahler, J . A m . Chem. Soc., 77, 5068 ( 1 9 5 5 ; A . Eschenmoser, L. Ruzicka, 0 . Jeger, and D. Arigoni, Helv Chint. A d a , 38, 1890 (1955). ( 4 ) H Fazakerley, T. G. Halsall, and E. R . H. Jones, J . Chem. SOL.,1877 (1959)
QY
0
IX ( 5 ) K . Schaffner, L. Cagliotti, D. Arigoni, and 0 . Jeger, Hriu. C h i m . A d a , 41, 152 (1958). (6) Cf.B C . Id. Weedon in "Advances in Organic Chemistry Methods and Results," Vol. I , Interscience Publishers, Inc , New York, S . Y . , 1960, Chapter I. A related scheme has been used independently by E. J . Corey and R . R . Sauers, J A m . Chem. Soc., 79, 3925 (1957).
