The Structure of β-Lumicolchicine

tion with our investigation of the structure of a-lumi- colchicine6 it is of priniary importance that the struc- ture of 0-lumicolchicine be establish...
0 downloads 0 Views 407KB Size
March 20, 1OG3

STRUCTURE OF

[CONTRIBUTION FROM THE

DEPARTMEST O F CHEMISTRY,

P-LUMICOLCHICINE

503

IOWA STATE UNIVERSITY OF SCIENCE AND TECHNOLOGY, AMES, IOWA]

The Structure of P-Lumicolchicine' BY 0. L. CHAPMAN, H. G. SMITHAND R . W. KING RECEIVED AUGUST20, 1962 T h e structure of p-lurnicolchicine has been rigorously established by nuclear magnetic resonance studies using solvent effects, double resonance techniques and deuteration on p-lurnicolchicine, y-lurnicolchicine and the alcohols obtained by sodium borohydride reduction of these ketones.

Irradiation of colchicine (I) gives three crystalline photoproducts, a-, /3- and y - l ~ m i c o l c h i c i n e , ~ -Struc~ tures I1 and 111 have been suggested for p- and y-

I

CH30

I11 lumicolchicine.*~5 Recently, a-lumicolchicine has been shown to be a dimer of /3-lumicolchicine.6 In connection with our investigation of the structure of a-lumicolchicine6 it is of priniary importance that the structure of 0-lumicolchicine be established beyond question. Evidence previously presented for structure II,*14m5although reasonably interpretable on the basis of this structure, is by no means rigorous. Specifically, the following problems are posed: (1) N o evidence is available t h a t ring B of colchicine remains intact in the colchicine to /3-lumicolchicine transformation.* (2) Evidence for bridging ring C of colchicine to form rings C and D of I1 is limited to the infrared carbonyl absorption of desmethyldihydro-P-lumicolchicine(IV) and the ambiguous observation that a dialdehyde obtained from a glycol derived from desmethyldihydro-P-lumi-

IV

0

V

OCH3

colchicine did not cyclize4 as does the dialdehyde obtained by reduction and cleavage of ring C of colchicine. (3) No evidence is available t o exclude V from consideration. The complex photoisomerizations of simple tropolones (e.g., VI-IX)' show t h a t photoisomerization of colchicine need not be simple, and structure V must be ( 1 ) P a r t I V of t h e Photochemical Transformations series: see 0. L. Chapman. D. J. Pasto, G. W. Borden a n d A. A. Griswold, J. A m . Chem. SOL.,84, 1220 (1962), for p a r t 111. (2) R. p e w e a n d W. Wulf, Be?., 84, 621 (1951). ( 3 ) F.S a n t a v q , B i d . L i s t y , 91,246 (1950). (4) E . J. Forbes, J. Chem. SOC.,3864 (1955). ( 3 ) P. L). Gardner, R. L. Brandon a n d G. R . Haynes, J. A m . Chem. S O L . , 79, 6334 (1957). ( G ) 0. L. C h a p m a n a n d H. G. S m i t h , ibid., 89, 3914 (1901); 0. L. Chapm a n , H . G . Smith and R. W. King, ibid., 85, 806 (1963). (7) W. G . Dauben, K . Koch, 0. L. C h a p m a n a n d S. L. Smith, ibid., 89,

1768 (1961).

seriously considered. (4) The stereochemical assign. ment of structures I1 and I11 to p- and y-lumicolchicine is equivocal.* The nuclear magnetic resonance spectra of /3-lumicolchicine (Fig. 1) and y-lumicolchicine (Fig. 2) integrate for 25 protons and show in each case four methylene protons and one methine (,on the carbon bearing nitrogen) in essentially the same positions as the four methylene protons and the methine proton of ring B in colchicine. Ring B is therefore intact in the photoproducts. The nuclear magnetic resonance spectra of P-lumicolchicine and y-lumicolchicine could accommodate structure 11, I11 or V for either of these compounds. The high-field bridgehead proton which is not visible in the nuclear magnetic resonance spectrum of p-lurnicolchicine can be observed by use of pyridine as solvent (Fig. 3). The diamagnetic anisotropy of this solvent produces remarkable shifts of certain proton^.^ The methyl group of the enol ether is shifted to higher field as is one of the aromatic methoxyl groups. The bridgehead protons are essentially unchanged, and the shift of the enol ether methyl groups clearly reveals the second bridgehead proton a t 6.37 r . The methylene resonance a t 7.29 r in P-lumicolchicine is shifted to lower field as is the proton on nitrogen. The latter shift is probably a result of hydrogen bonding to pyridine. The aromatic proton is shifted to lower field and now overlaps the olefinic proton resonance. These highly specific shifts to

both higher and lower field positions indicate that the ,~ shielding efect of pyridine, as suggested by S l ~ r n parises from coordination of pyridine molecules at specific sites on the P-lumicolchicine molecule, not from a n average effect of the solvent as a whole. Observation of the olefinic proton and the two bridgehead protons provides strong support for bridging of the tropolone ring in the formation of P-lumicolchicine. Rigorous evidence t h a t the olefinic proton is coupled t o a bridgehead proton comes from a double resonance experiment in which saturation of the lower field bridgehead proton collapses the olefinic proton doublet to a singlet (Fig. 4). The rigorous establishment of the nature of rings C and D is now complete. The nuclear magnetic resonance spectrum as well as the chemical evidence, however, still does not permit a choice between structures 11, I11 and V. (8) T h e base-catalyzed epimerization of 8-lumicolchicine t o 7-lumicolchicine' we cannot reproduce, Professor Gardner has informed us t h a t he has been unable t o reproduce this experiment. T h e remaining evidence for t h e stereochemical assignments is based on t h e e f f e c t of concentration o n extinction coefficient in t h e ultraviolet spectra of t h e tetrahydro derivatives of p-lumicolchicine and y-lumicolchicine. T h e stereochemistry of t h e t e t r a hydro derivatives is n o t established a n d has been assigned on t h e basis of arguments concerning t h e mode of reduction.6 T h e demonstration (in this paper) t h a t t h e dihydro derivatives of 8-lumicolchicine a n d y-lumicolchicine both show strong intramolecular hydrogen bonds renders the distinction between t h e tetrahydro derivatives somewhat tenuous. (9) G . Slomp a n d F. MacKellar, J. A m . Chem. SOL.,81, 999 (1960).

O. L. CHAPMAN, H. G. SMITH AND

SI )-I-

R.W. KING

Vol. 83

m 3

4

5

?

a

7

6

a

Fig. 1.-Xuclear magnetic resonance spectrum of p-lumicolchicine. T h e scale is calibrated in r-units.

1

I

I

I

3

4

5

Fig. 2.--Suclear

3

4

Fig. 3.--Nuclear

I

6

8

7

magnetic resonance spectrum of r-lumicolchicine.

A

1

1

5

6

7

--A

B

magnetic resonance spectrum of p-lumicolchicine in pyridine.

Reduction of /3-lumicolchicine with sodium borohydride4 in tetrahydrofuran gives a homogeneous alcohol

X

OCHs

which will be shown (see below) t o be X. Reduction of the ketone should give an alcohol in which hydride has been delivered t o the most accessible side of the molecule. lo In the reduction of /3-lurnicolchicine this would ( I O ) 0 . L. Chapman, D. J. Pasto, G. W. Borden and A . A . Griswold, J . A m . Chem. Soc., 84, 1220 (1862).

lead to the partial stereochemistry XI regardless of the gross structure of the remainder of the molecule. Reduction of y-lumicolchicine with sodium borohydride in tetrahydrofuran gives a mixture of alcohols which we could not separate. Reduction of y-lumicolchicine

805

STRUCTURE OF P-LUMICOLCHICINE

March 20, 1963

Fig. 4.-Double resonance experiment. Up;>er spectrum shows (from left t o right) t h e olefinic proton doublet ( J = 3.2 c.P.s.', the aromatic proton singlet and t h e K-H doublet. Lower spectrum shows the same region with the adjacent bridgehead proton (151 C.P.S. upfield) saturated. Note collapse of the olefinic doublet t o a singlet.

with sodium borohydride in methan01,~however, gives a homogeneous, crystalline alcohol analogous t o that obtained from /3-lumicolchicine. Comparison of the nuclear magnetic resonance spectra of the alcohols derived from /3-lumicolchicine and r-lumicolchicine (Fig.

-

/ &

OCHS

L NaBH,,

XI 5 and 6) show general similarity except that the alcohol from @-lumicolchicineshows an anomalously low-field (relative to P-lumicolchicine, y-lumicolchicine and dihydro- ylumicolchicine) N-H absorption. This lowfield K-H strongly suggests that a new intramolecular hydrogen bond is possible in the alcohol which is not possible in @-lumicolchicine. I t is necessary to establish the intramolecular nature of the hydrogen bond. This can most easily be done by extrapolation of the N-H resonance position (as a function of concentration) t o infinite dilution. Such extrapolation gives 488 C.P.S. (relative t o internal tetramethylsilane a t 60 Mc.) for the resonance of the N-H of the alcohol from /3-lumicolchicine at infinite dilution in both deuteriochloroform and carbon tetrachloride. Similar extrapolation for the N-H resonance of the alcohol derived from ylumicolchicine gives 325 C.P.S. The N-H resonance of (3-lumicolchicinealcohol (X) shows a much smaller shift on dilution (23 c.P.s.) than the N-H of 7-lumicolchicine alcohol ('76 c.P.s.). The difference of 163 C.P.S. in the position of the N-H resonance lines of the two alcohols a t infinite dilution and the small shift of the N-H resonance of @-lumicolchicinealcohol leave no doubt that a strong hydrogen bond exists between the acetamido group and the newly introduced hydroxyl function in the alcohol derived from @-lumicolchicine. Such a hydrogen bond I S possible only af the acetamzdo group and the hydroxyl function are on the Same side of the molecule 2n close prozcmzty. These requirements are met only in structure X. This evidence rigorously eliminates structure V for P-lumicolchicine and a t the same time rigorously establishes the stereochemistry of /3-lumicolchicine as 11. The nature of the actual hydrogen bond is established as CH,-C=O

I

iK-H

H

I I

0

by the following facts: (1) Shaking the sample with D20 exchanges the hydroxyl hydrogen for deuterium without significantly affecting the position of the N-H

I

i

CHs0

5

6

1

678 7

8

Fig. 5.-Upper spectrum shows the nuclear magnetic resonance spectrum of the alcohol derived from @-lumicolchicine at normal concentration. Lower spectrum shows t h e same sample after fourfold dilution. Note t h e large shift in the resonance position of the hydroxyl proton and the much smaller shift of the N-H resonance.

i , 3

I

I

L

5

Fig. 6.-Suclear

, 6

7

8

magnetic resonance spectrum of the alcohol derived from r-lumicolchicine.

resonance. ( 2 ) Extrapolation of the 0-H rescjiiance position t o infinite dilution shows a change in resonance location (268 to 162 c.P.s.) indicative of intermolecular hydrogen bonding of this protor.. The alcohol from 7-lumicolchicine shows an intramolecularly hydrogen bonded hydroxyl function (in contrast t o the alcohol obtained from P-lumicolchiciiie) . This hydroxyl resonance is still a t low field (297 c.P.s.) after extrapolation to infinite dilution. Only threc possible hydrogen bonded structures (XII, XI11 and, XIV) are possible. No definitive choice among these structures is possible, although on the basis of the rigorously established structure of P-lumicolchicine, structure XIY may seem more plausible. The observation of a photochemical conversion of /3- to ylumicolchicine" cannot be considered direct evidence in support (11) G. 0. Schenck, H. J. K u h n a n d 0 -A. Neurnuller, I'elrahedron Letters, 1, 12 (1961).

0 . L. CHAPMAK, H. G. SMITH AND R. W. KISG

SO(i

XI1

XI11

OCH3

XIV of structure XIV in view of the photochemical transformations of simple tropolones.7 The virtual identity of the infrared, ultraviolet and nuclear magnetic resonance spectra of /3-lumicolchicine and y-lumicolchicine coupled with the firm assignment of structure I1 t o 0luniicolchicine supports structure 111 for 7-lumicolchicine and XIV for the alcohol derived from it. This assignment, however, cannot be considered rigorous and the two stereoisomers of V must be considered possibilities.

Experiment a1 Nuclear Magnetic Resonance Spectra.-The nuclear magnetic spectra were run in deuteriochloroform using a Yarian Associates, HI