.lug. 20, 1938
4393
INTERCONVERSIONS O F AhIARYLLIDACE.\E .JLKr\LOIDS
[CONTRIBUTION FRO?d TIIE 1,ABORATORY OF CIIEMISTRY O F NATURAL PRODUCTS, XATIONAL INSTITUTES O F HEALTH]
HEARTISSTITUTE,
XATIONAL
Interconversions of Amaryllidaceae Alkaloids by Sodium and Amyl Alcohol' BY H. M. FALES AND W. C. WILDMAN RECEIVED FEBRUARY 13, 1958 Sodium with amyl alcohol has been found effective for interrelating a number of Amaryllidaceae alkaloids. With the aid of this reagent, partial structures have been assigned to the alkaloids falcatine and narcissidine; lycorine and methylpseudolycorine have been converted to caranine and pluviine, respectively. The conversions of powelline and buphanidrine to dihydroepicrinine and buphanisine, respectively, by this reagent, have provided additional proof for the structures assigned earlier to powelline, buphanidrine and buphdnisine.
Through empirical correlations in infrared spec- examined by Claysonll who obtained a 60% yield tra and by the use of certain classification reagents of I1 by the addition of isoamyl alcohol to powdered such as manganese dioxide,lJ selenium dioxide and sodium suspended in a xylene solution of I under mercuric acetate, - j an Amaryllidaceae alkaloid of reflux. Degradative work in our laboratory had unknown constitution may be placed tentatively indicated that a similar relationship existed in the within one of the six fundamental ring systems aromatic substitution of the alkaloids powelline which have been established for the family. (111, R = OCH3) and crinine (111, R = H).'.' Within each given ring system, a number of al- Powelline contains the functional groups of crinine kaloids may stem from simple variations in the type of aromatic and aliphatic ring s u b s t i t ~ t i o n . ~ ~ ~ - ~ Since many of these alkaloids are obtainable only in quantities insufficient for detailed degradative study, chemical methods which would permit the I CHsO I I1 conversion of a scarce alkaloid to an alkaloid of proven structure would be of great value. We and, in addition, one methoxyl group which was have found that the action of sodium and amyl placed in the aromatic ring for the spectral reasons alcohol on certain alkaloids of the family is most mentioned earlier. Since the formation of crinine useful in this respect. This paper reports our ob- by the ar-demethoxylation of powelline would servations concerning the effect of this reagent constitute additional chemical proof of the strucon a number of alkaloids a t our disposal. ture of powelline, i t seemed desirable to study the The methylenedioxy and methylenedioxy- effect of sodium and isoamyl alcohol on powelline. methoxy groups constitute two of the most preva- The infrared spectrum of the crude reaction mixlent types of aromatic substitution within the ture indicated that ar-demethoxylation was essenfamily. These two types of substitution may be tially complete, since only weak absorption was recognized by differences in the ultraviolet absorp- observed a t 6.2 p. However, chromatography of tion spectra and by the presence of an intense, the mixture on alumina afforded no crinine. The sharp band a t 6 . 2 ~in the infrared spectra of al- most strongly adsorbed substance was identified as kaloids containing the methylenedioxymethoxy- dihydroepicrinine (IV). Preceding this material, It has been our observation two isomeric, non-crystalline substances (V) of phenyl f u n ~ t i o n . ~ that alkaloids with the latter type of substitution molecular formula CI6Hl7NO2were eluted. Each give a deep violet color in concentrated sulfuric of these isomers gave (-)-crinane upon the abacid while colors no deeper than orange result from sorption of one equivalent of hydrogen under catalkaloids containing the methylenedioxyphenyl alytic conditions. The nature of these products is group. chemical evidence that (a) powelline is derived The alkaloid derivatives hydrocotarnine (I) and from 5,lob-ethanophenanthridine, the ring systern hydrohydrastinine (11) may be cited as simple ex- of crinine, (b) the methylenedioxy group of powelamples of these types of substitution. The con- line is in the S,9-position as in crinine, and (c) the version of hydrocotarnine to hydrohydrastinine hydroxyl group of powelline is in position 3 as in was reported first by Pyman and Remfry.lo crinine. The reaction products can be explained Nearly fifty years later this conversion was re- readily only on the basis of structure 111 (R = OCH3) for powelline. (1) Paper XI1 in a series on t h e alkaloids of t h e Amaryllidaceae; The isomeric desoxycrinines Va and Vb may be previous paper, W. C . Wildman, THIS JOURNAL, 80, 2567 (1958). derived from the hydrogenolysis of the allylic 3(2) R. 1. Highet and W. C. Wildman, ibid., 77, 4399 (1955). (3) H. M. Fales, E. W. Warrihoff and W. C. Wildman, ibid., 77, 5885 hydroxyl group. The formation of the inter(1955). mediate radical (IX, R = OCH3or H), followed by (4) H. M. Fales, Laura D. Giuffrida and W. C . Wildman. i b i d . , 78, reduction of the resonance hybrid would explain 4145 (1956). the formation of two isomers. Such a mechanism ( 5 ) H. M. Fales a n d W. C. Wildman, ibid., 7 8 , 4151 (19%). (6) Carol K. Briggs, Patricia F. Highet, R. J. Highet and W. C. is supported by the observation that dihydroepiWildman, ibid., 78, 2899 (1956). crinine is the sole product of the action of sodium ( 7 ) W. C. Wildman, Chemistry 6' I n d u s f r y , 1090 (1956). and isoamyl alcohol on dihydroepipowelline (VIII) . ( 8 ) H.-G. Boit, H. Ehmke, S. Uyeo and H. Yajima, Chem. Ber., 90, 383 (1957). The formation of dihydroepicrinine may occur (9) W. C. Wildman a n d Carol J. Kanfman, THIS JOURNAL, 77, 4S07 by one of two paths. The reaction may proceed (1955). (10) F.L. Pyman ant1 F. G . P . R e m f r y , J . Cheru. SOC.,159.5 (1912).
( 1 1 ) D B. Clayson, ibid , 2010 (1949).
4396
\' I
through VI, formed by the action of some carbonyl contaniinant or other hydrogen acceptor upon 111 i n the basic reaction medium.lZ Reduction of VI to the corresponding saturated ketone, followed by the reduction of the 3-carbonyl group would be expected to lead to that alcohol which is thermoclynaniically most ~ t a b 1 e . I ~From the observation'
IS
that oxopowelliiie (VI, K = OCH3) and oxocriiiine (VI, I< = H) are reduced by lithium aluminum hydride or sodium borohydride to epipowelline (VII, R = OCH3) and epicrinine (VII, R = H ) , respectively, it is likely that the hydroxyl groups of the epi series are the more stable and presumably of equatorial conformation. The latter phases of this mechanism have precedent in the conversion of (+)-carvone (X) to (+)-dihydrocanwA (XI). I 4
.()
CHB-C=CHI X
-----f
3 I
/("*
CH3--C=CH2 XI
The deinethoxylation of the aromatic ring occurs during these transformations, but there is no evidence to indicate whether it precedes, follows or is simultaneous with the reactions of ring C.I5 (12) C,f t h e necessity for trace amounts of carbonyl component in t h e equilibration of alcohols; W.E Doering and T. C . ilschner, Tms J O U R N A L , 71, 838 (1949). (13) (a) D . H R Barton, E x p e v i e x f i a , 6, 316 (1950); (b) G. Vavon. Bidi. TOG. chim. ( F r a n c e ) , 141 49, 937 (1931). !14) K . G. Johnson and J . Read, J . Chrm. Soc., 233 (1934). (15) Determination of t h e position of the methoxyl group in the uromatic ring is under investigation in these laboratories It is (rf interest t u note t h a t t h e biogenetic proposals of Barton and Cohen'e require the methuxyl group t o be in the lO-yiJsitiou C f . I.: \I,. \VUIIhuff, ( ' h r n l i i l v y & I N d u s f r y , 1386 ( (16) L). 11, IC=C-N
C-C=N
dromethylpseudolycorine and dehydrolycorllic ant1 may be assigned the structures XXVb and XXVL, respectively, from the following facts. Both COIIIpounds showed hydroxyl absorption in the infrared a t 2.88 ,u (Nujol), and a normal 0-acetyl derivative (A 3.75 p) was obtained from XXVa. Both bases formed normal methosalts, and XXVa fortned a hydroperchlorate which exhibited no bands in the
6.0 p region, characteristic of the
I
> C = N--bond.
+
Both XXVa and XXVb rapidly absorbed one mole of hydrogen over palladium-on-charcoal to afford caranine (XVIII) and pluviine (XX), respectively. The isolation of these products provided chemical proof that dehydrolycorine and dehydromethylpseudolycorine contain the caranine and pluviine nuclei, respectively, less two atoms of hydrogen. Characteristic of conjugated dienes, both XXVa and XXVb showed strong absorption maxima a t 230 mp when the appropriate aromatic chromophores were subtracted from the observed spcctr‘i II(-)\
/?+
I
’
buphanidrine, buphanisine and buphananiine) show well-defined carbon-hydrogen stretching bands at 3.30 p at high concentrations in carbon tetrachloXXVa, RI.Rz = >CH? SS\.I ride. These bands are not present in the correb, Ri,R? = C R j sponding dihydro derivati~es.~’Neither XXIa nor X X I b (nor lycorine diacetate, galanthine, methyl- The alternative structures XXVIa and XXVIb pseudolycorine, falcatine, narcissidine and caranine) can be eliminated on the basis of the wave length shows such a band, thus indicating the probable of these ultraviolet maxima and the fact that the absence of 1,2-, 2,3- or 4,5-unsaturation in these position of these maxima did not shift more than 1 compounds. Of the remaining possible positions mp in acid solution. Furthermore, a compound of for the unsaturation, 3,3a and 3a,4, the latter is the structure XXVIa or XXVIb would be exless likely since X X I a was recovered unchanged pected to dehydrate readily under the conditions after treatment with boiling hydrochloric acid.28 employed in its formation.29 Although XXVb was The structure of a third product, C ~ ~ H I ~ N recovered O~, in 73y0yield when treated with methanwhich was isolated in minute yield from the action olic hydrochloric acid, under more stringent conof sodium and n-amyl alcohol on galanthine has ditions (warm phosphorus oxychloride), i t was posprovided a plausible mechanism for the formation sible to convert XXVb to an anhydro base (XXII, no 0 at C,) .5 (25) K Takeda, K. Kotera and S Mizukami, THISJOURNAL, 80, 2362 (1958) Most convincing evidence that XXVa and XXVb (26) N J Leonard and V W Gash, rbrd., 76, 2781 (1954). are intermediates in the formation of XVIII and (27) W. H Tallent and I. J Siewers, AWQ!.Chcm., 28, 953 (1956). (28) H C Brown, J. H Brewster and H Schechter, THISJOURNAL, (29) Cf.t h e conveniuri 7 6 , 407 (1a;r) corine methosalts.22
c>f
lyrorine methohydroride i o auhydriilv-
ilug. 20, 1955
4399
INTERCONVERSIONS O F hIIARYLLIDACEAE ALKALOIDS
XXIa from lycorine and X X and X X I b from galanthine was obtained in the following manner. A solution of XXVb was treated with sodium and namyl alcohol under the same conditions initially employed for galanthine itself, and the same products, XX and XXIb, were obtained. Under similar conditions, XXVa gave caranine (XVIII) and XXIa. The formation of XXVa and XXVb probably takes place through a simple lJ4-elimination of alkoxyl, acetoxyl or even hydroxyl by the strong base. Conversion of these substances to caranine and pluviine would result from 1,4-reduction of the conjugated diene system, while in the formation of X X I a and XXIb, hydrogenolysis of the 1-hydroxyl group occurs prior to 1,4-reduction of the diene. Simple hydrogenolysis of X I X a and XIXc probably occurs simultaneously with this reaction path since XXVa reacts with sodium and n-amyl alcohol to afford a mixture of XVIII and X X I a having a higher ratio of the latter to the former than does the mixture obtained directly from lycorine. It is important to note that this over-all reaction mechanism involves no optically active centers except those subsequently converted to methylene carbon atoms. Since dihydrolycorine and dihydrocaranine have been shown to have the same stereochemistry a t C1, C3al Cllb and C1lC,30 any alternative mechanism would be required to neither invert nor racemize these centers. The foregoing products and principles have enabled us to come to a conclusion concerning the structure of narcissidine. The alkaloid narcissidine was isolated first from the bulbs of Narcissus poeticus L.31 Subsequent isolation^^^^^^ have shown that the base is present in many Narcissus species. I n our laboratory, i t has been isolated in 0.01570 yield from Hymenocallis amancaes (Ruiz and Pavon) Nichols. Boit and S t e n d e P showed that the alkaloid possesses the expanded molecular formula ClbHlzN(OCH3)3(0H)2 and contains one reducible double bond. Preliminary tests by these workers showed that it was not affected by diazomethane, hydroxylamine, dilute acid or dilute base. We have confirmed these results and have determined by periodate titration that the hydroxyl groups of dihydronarcissidine are vicinal. A considerable amount of rather speculative evidence was available to indicate that narcissidine was an alkaloid of the pyrrolo [delphenanthridine (lycorine) type, yet valid chemical proof of the ring system was difficult to obtain. The infrared spectrum of narcissidine showed many bands in common with methylpseudolycorine (XIXb) and, like many alkaloids of the same basic ring system, it gradually turned yellow in the presence of air and light. However, the Hofmann and Emde reactions were unsuccessful under conditions by which lycorine gave anhydro compounds. Like methylpseudolycorine and galanthine, narcissidine was oxidized by both selenium dioxide and mercuric acetate, but the initially yellow oxidation products darkened rapidly and no useful products could be (30) K. Takeda and R. Rotera, Pharm. Bull., 6, 234 (1957). (31) H.-G. Boit and W. Stender, Chcm. Bcr., 87, 624 (1954). (32) H.-G. Boit, W. D6pke and Anita Beitner, ibid., 90, 2197 (1957). (33) I€.-C.Boit and H. Ehmke, ibid., 89, 103 (1956).
recovered from the black reaction mixture. However, 0,O-diacetyldihydronarcissidinewas oxidized in good yield by potassium permanganate to a neutral, conjugated lactam. This type of reaction is characteristic of alkaloids derived from pyrrolo[delphenanthridine and does not occur with alkaloids containing the 5,lOb-ethanophenanthridine ring system (e. g., crinine). In view of these rather unpromising results, narcissidine was treated with sodium and n-amyl alcohol in the hope that more useful degradation products could be obtained. From the structures of the reaction products, all of which were known from earlier work, it has been possible to assign tentatively the structure XXVII to narcissidine. The reaction products were pluviine (XX), X X I b and XXVb. The isolation of these products made it certain that narcissidine contains the 9,lO-dimethoxypyrrolo [de Iphenanthridine nucleus and that one hydroxyl group of the alkaloid is located in the 1-position. Since the hydroxyl groups of dihydronarcissidine (and, therefore, of narcissidine also) are vicinal, the second hydroxyl must be located in position 2 or l l b . The latter position is not likely since narcissidine is stable toward acid and catalytic hydrogenolysis and forms a diacetyl derivative with ease. The position of the third methoxyl group is ambiguous, since the preceding experiments indicate that either aliphatic or aromatic methoxyl groups may be removed by sodium and amyl alcohol.
XSYII
OH
Ho\n/o CHs
XXIX
Although the ultraviolet spectrum of narcissidine is identical with that of methylpseudolycorine (XIXb), the difference between the spectra of these alkaloids and that of anhalinine (XXVIII) is quite small. To determine the type of aromatic substitution, narcissidine was oxidized by potassium permanganate. The isolation of m-hemipinic acid from this oxidation proves that the aromatic ring of the alkaloid is substituted by methoxyl groups only in the 9- and 10-positions. From the foregoing data, i t may be concluded that the two aromatic methoxyls and the two hydroxyl groups of narcissidine are in the same positions as in methylpseudolycorine (XIXb). The only remaining problem in the structural determination of narcissidine is the placement of the double bond and the aliphatic methoxyl group. The positions of these two functions are mutually
dependent. The ultraviolet spectrum of narcissidine and the infrared spectrum of 0,O-diacetyliiarcissidine hydroperchlorate show that the double bond is neither conjugated with tlie aromatic ring nor CY,@to the nitrogen atom. Since narcissidine can be neither an enol (evidenced by its insolubility in alkali and the formation of a normal 0,O-diacetate (A 5.T3 p ) but no reaction with hydroxylamine) nor an enol ether (shown by stability to dilute acid). only two positions, 3,3a and 3a.4, are possible for the double bond These restrictions limit the possible structures for narcissidine to XXVII and XXIX. ;1 choice between these two structures may be derived from considerations of the biogenesis of the alkaloid, the possible iiiechanisnis by which XX and X X I b are formed by the action of sodium and n-amyl alcohol on the alkaloid, and by analogy with the structures of other alkaloids of the family containing this ring system. .Idouble bond in position 3a,4 (as in XXIX) has not been iound in any alkaloid of the lycorine type. =Ilso, a roliipound of a structure such as XXIX would be expected to isomerize with dilute acidzbto a A? 3d structure and afford a dihj-droxy ketone as a reaction product. A 4 has ~ been mentioned earlier. iiarcissidine is stable to dilute acid. From our previous studies, sodium and n-amyl alcohol v ould be expected to react with X X I X to afford either XXX or X X X I by 1,4-elirnination or hydrogenolysis, respectively. h-either of these coinpounds is a promising intermediate for the observed formation of XX or X X I b 0 1 1 the other hand, XXVII contains the double bond i n the 3,3aposition which is common to the other alkaloids powxssing this ring system, ~ i i t lsuch a structure OF1
011
Tro\A
structure for narcissidine. Other degradation methods are untlrr investigatioli t o coiifiriii this proposal. Acknowledgment.---The authors wish t o thank Ilr. €3. G. Schubert of the Section of Plant Introduction, U. S . Department of Agriculture, for her help in obtaining the plant materials used in this study. \Ye are indebted to IIessrs. D. L. Iiogerson, Jr., and J. D. Link for the isolation of the crude alkaloid mixturc and to l l i s s Elizabeth Kielar for the isolation o f sex-era1 of the pure alkaloids. Thr. skillful assistance of Miss Patricia Wagner and Mrs. L. C . Karren in the instrumental aspects (it' the work is acknowledged gratefully. Experimental3$ Isolation of Alkaloids .--Sarcissidhe WBS isolated frnm tlie ~ T u ~ c i s s hybrid us "Deanna I h r b i n " in yields comparable to those reported by Boit and Ehmke.33 T h e alkaloid also was isolated from Hymemcollis an~ancaes (Ruiz and P a v o \~ ~ Sichols by methods reported earlier.4 Narcissidine, hippeastrine, lycorine and lycoramine were isolated in 0.015. 0.0015, 0.11 and 0 . 0 1 4 ' ~yields, respectively. The melting points and infrared spectra of the isolated alkaloids were compared rr-ith thow of ciuthentic samples, and niixturc melting points gave 110 depression. The isolation and characterization of falcatine, Iycorinc, galanthine, buphanidrinr and powelline have been described in previous papers of thi.: series. Narcissidine (XXV1Ij.---The base crystallized from acetone as colorless prisms, 1n.p. 201-203' dec., [ D I ] ~ -:11'((~ ~ s ~ 1.5); reported31m.p. 218-219' dec., CY]*^^^^ -32.O0(chloroform). In an evacuated capillary, narcissidine showed :1 m.p. 221--223" dec. The ultraviolet absorption spectrum showed an inflection at 227 nip (log t 4.96) and a maxiinuni
.-1nnZ. Calcd. for C15H23SOj:C, 64.85; H, 6.95; S , 4.20; 3 OCH3, 2i.93; neut. equiv., 333.4. Found: C , 04.63; H, 7.08; S ,4.23;OCH3, 25.08; neut. equiv., 333. The base was recovered in SO('; yield after treatment wit11 G S h:-drochloric acid a t 100' for 12 hours and in 50"; yield from the action of ~varinphosphorus oxychloride for ten minutes. It v a s recovered in 80% yield from the action d lithiurn aluniinum hydride in refluxing tetrahydrofuran overiiight. 0,O-Diacetylnarcissidine hydroperchlorate was prepared from 0,O-diacetylnar~issidine~~ and dilute, aqueous perchloric acid. One recrystallization from water afforded Iimg prisms, m.p. 220-23,5° dec. The product exhibited :ihorptiori bands in the infrared spectrum ( S u j o l ) a t ,5.72 :inti 3.7,5 1 due to the 0 - x c t y l groups and a t 6.18 and (;.?ti v. clue t o tlie aromatic ring. .-I 1 1 ~ 1 . Calcd. for CY2H2;NO~.HC10,: C , 51.02; 13, 5.45; C1, 6.8.;. l ' c i n r i d : C, 61.08; H, 3.13; C1, 7.16. Dihydronarcissidine. .I solutiim I J f 350 mg. of narcisiitlirie in glacial acetic acid was hydrogenated a t room teiniierature and atmospheric pressure in the presence of I O 0 Ing. of prereduced platinum oxide. Slightly more than one inole of hydrogen was absorbed in 20 minutes. The solutiiiii wets filtered, evaporated a n d basified with potassiuiii hydroxide. The clear s i ~ l u t i ~ iwas n extracted with cIi111r~1form, and the chloroform extracts mere dried over siicliuin sulfate and evapnratcd. Tile product (280 mg., 80' ) crystallized from eth\.l acetare a s fine, non-birefringent prismi, 1n.11. t38.$-1f%)', i n 1 2 3 j j ~ i - I l l " , [ a ] 2 3 4 : j 6 -241' (C 0 . 2 2 , t i i i n e t l ~ y l f o r i ~ ~ :i r. ~ ~ ~ i ~ l ~ .Inu1. C:rlcd. for C,,S,,SO,: C , 64.46; H , 7.51; OC€II, 2 7 . X ; glycol. l,O(l. F I I ~ I IC~,64.63; : H , 7.68; OCH,, 28.0-4; gl!-col, 0..l)icalof iminc s d t s in the 6 p rcgiim.
Dehydromethylpseudolycorine methiodide was pIcp:ti-ccl b>-allowing methyl iodide to react with the free base i n :ice-
11one.3~
tone, and then one recrystallizatior~ uf the product friiiri ethanol, m.p. 242-2-16" dec. A 4 d . Calcd. for ClaHJ?S031: C, 50.59; 11,
