July 5 , 1962
2601
S T R U C T V R E S O F P A R T H E N I N AND h l B R 0 S I N
A?65
1000
3000
2000
1
o
calcd for r: 13
9
4000
SECONDS.
Fig. 2.-Behavior of galvinoxyl (lower curve) and iodine (middle curve) as competitive scavengers for cyanoisopropyl radicals: upper line, sum of lower two curves.
Fig. 3.--Optical densities a t 500 and 765 mfi during competitive scavenging of radicals from azobis-isobutyronitrile by galvinoxyl and iodine.
tion of AIBN, the efficiency of radical production from this initiator is again found to be 62.2% a t 61.65'. When two scavengers a t concentrations x and y are competing for the same reactive free radicals with the large rate constants k, and k,, their relative rates of disappearance are governed by the equations
An attempt to determine in the same way the relative efficiencies of galvinoxyl and iodine toward the tert-butoxy radical from di-tert-buty peroxyoxalate was unsuccessful. Analysis of the curves showed that scavenger was not disappearing as fast as radicals were being formed and that the rate of scavenger disappearance was constantly changing throughout the experiment. This is another instance of the unsuitability of iodine and iodides for quantitative work with peresters, on which we have commented previously.10 Toward the cyanoisopropyl radical, for which both galvinoxyl and iodine are satisfactory scavengers, galvinoxyl has reactivity to spare. If, with the 10-fold difference between these scavengers in mind, we examine the ends of the curves of disappearance of G and 1 2 singly in the presence of AIBN, we see a consistent but small departure from linearity as the last few per cent. of iodine disappear, but no corresponding effect in the case of galvinoxyl. The normal mode of disappearance of free radicals in such a system must be slow compared to either of these scavenging processes. Acknowledgment.-We thank the National Science Foundation for support of this work.
so that
I n the range of optical densities where these quantities can be determined with comparable precision (up to 92y0 consumption of gahinoxyl and l6Y0 consumption of iodine) this function is reasonably linear, with a slope r of about 10. With all values of r higher than this, the experimental curve of a vs. b can be calculated within the experimental uncertainty (Fig. 3 ) . Smaller values of r fail to fit the data.
~
~~
-
[CONTRIBCTIOS F R O M
THE
DEPARTMEST OF CHEMISTRY O F
THE
FLORIDA STATE USIVERSITY,
TALLAHASSEE, F L A .
]
The Structures of Parthenin and Ambrosin1s2 B Y ivERNER
HERZ,HIROSHI WATANABE,
h I A K O T 0 & l I Y A Z A K 1 3 A N D \'UKICHI
KISHIDA
RECEIVED FEBRUARY 2, 1962 Parthenin, the main constituent of Parthenium hysterophorus L., has been shown to be a sesquiterpene lactone and correlated with ambrosin. a constituent of Ambrosia maritima L . Structures for these substances have been established.
According t o extracts of Parthenium hysterophorus L., a bitter herb common in the (1) For preliminary reports on part of this work, see (a) W. Herz, H. Watanabe and M. Miyazaki, J . A m . Chem. SOL.,81, 6088 (1959): (b) W. Herz. M. Miyazaki and Y.Kishida, Tetrahedvon L P ~ ~ PNo. V S2, , 82 (1961). (2) Supported in part by grants from the h'ational Science Foundation (NSF-G 14396) and the United States Public Health Service (RG-5814).
Southeastern United States and the West Indies, have enjoyed some reputation as a folk remedy against various afAictions such as ulcerated sores, certain skin diseases, facial neuralgia, fever and (3) Recipients of Fulbright Travel Awards in 1968-1959 and 19591960, respectively. (4) H. V. Amy, J Pharm., 121 (1890): 169 (1897). Earlier references to the medicinal use of P . hyrlerophovur L. are given in these
papers.
2602
UT.HERZ,H.
WATANABE,
M.
anemia. By extracting the plant with dilute alcohol Arny was able to isolate an apparently non-alkaloidal, non-glycosidic substance of m.p. 168-169' which he named parthenin. ,4rny claimed to have determined the empirical formula, but in fact no formula was ever published. Subsequent references to this plant appear to be limited to a report5 which states that extracts of P. hysterophorus L. gave positive tests with Wagner and Dithmar reagent but that the alkaloid content was very small and that the medicinal properties, if any, must be due to substances other than alkaloids. We became interested in this problem because of the occurrence of the azulenogenic sesquiterpene alcohol partheniol, as the cinnamate ester, in the only other Parthenium species which had been investigated previously. This is P. argentatum Gray, more commonly known as guayule.6 It seemed possible that the main constituent of P. hysterophorus was also a sesquiterpene derivative with a perhydroazulene skeleton. The identification of parthenin as a sesquiterpene lactone with an "abnormal" carbon skeleton and the correlation of parthenin with ambrosin, another naturally occurring sesquiterpene lactone, is the subject of this paper. Extraction of P . hysterophorus gave a crystalline D substance C16H1804, m.p. 163-166', [ C Y ] ~ ~$7.0Z0, whose properties seemed to correspond to those of the material isolated by Arny4J and for which we have therefore retained the name parthenin.8 The infrared spectrum of parthenin had bands a t 1655 and 1592 cm.-' which disappeared on catalytic hydrogenation to tetrahydroparthenin (11) and which were therefore assigned to two double bonds. As in the case of helenaling and balduilin,1° the higher frequency arises from the presence of an exocyclic methylene group conjugated with a y-lactone function (infrared bands a t 1755 and 1408 cm. -I) . l l This is supported by the following evidence. Parthenin formed a pyrazoline12 and had a nearinfrared band a t 1.64 p13 which disappeared on hydrogenation. Comparison of the C-methyl values and the methyl region of the n.m.r. spectra (see Table I) of parthenin and I1 indicated the presence of an additional C-methyl group in the latter. Ozonolysis of parthenin gave rise to form(5) A. J. Loustalot and C. Pagan, EI Crisol, 3, no. 5, 3 (1949);
C. A , , 44,2719 (1950). (6) A. J. Haagen-Smit and C. T.0. Fong, J . A m . Chem. SOC.,70. 2075 (1'348). (7) As noted by Arny, the yield of crystalline material which seems to be concentrated in the leaves varies seasonally and seems to reach a peak in late summer. No attempt was made t o put this on a quantitative basis and the yields 0.14Yo, given in the earlier reference,'Z and 0.33% reported in this paper refer t o yields from whole plants collected at various times during the summers of 1957 and 1959, respectively. (8) Parthenin should not be confused with a substance recently isolated from Chrysanfhemum parthenium (L.) Bernh. by V. Herout, M. Sourek and F. Sorm, Chemi.rfry b' I n d u s f r y , 1069 (1959), and named by them parthenolide. (9) (a) R. Adams and W. H e n , J . A m . Chem. Soc., 7 1 , 2346, 2551, 2554 (1949); (b) G. Biichi and D. Rosenthal, ibid., 7 8 , 3860 (1956). (IO) W. Herz, R. B. Mitra and P. Jaydraman, ibid., 81, 6061 (1959). ( 1 1 ) M.Hordk and J. Pliva, Chemistry & I n d u s t r y , 102 (1960). (12) P.G. Deuel and T. A. Geissman, J . A m . Chem. Soc., 79, 3778 (1957). Washburn and M. S. Mzhoney, ibid., 80, 504 (1958). (13) W.
a.
h!hYAZAKI AND
Y . KISHIDA
Vol. 84
aldehyde. When the ozonolysis was carried out in methanol a t -78', there was also formed in good yield norparthenone (111), C14Hl&,,which contained the partial structure A. I n accordance with the behavior of model compounds, l 4 this oc-ketobutyrolactone was completely enolized (titratable with
05
-OH
dA
dilute sodium hydroxide solution, ferric chloride test), exhibited the appropriate ultraviolet (Xmax 235 mp, e 12000)15and infrared maxima, could be further oxidized with the liberation of oxalic acid and furnished an enol benzoate. The infrared band a t 1592 cm.-l coupled with a strong carbonyl absorption a t 1718 cm. -l suggested the presence of a cyclopentenone grouping in parthenin. The ultraviolet spectrum (Xmax 215 and 340 mp, e 15100 and 22, high intensity a t 206210 mp) was similar to that of helenalin and balduilin and could be interpreted as arising from the overlap of a,P-unsaturated ketone and cr,p-unsaturated y-lactone chromophores. This hypothesis was strongly reinforced by the presence in the infrared spectrum of tetrahydroparthenin of two carbonyl maxima, one a t 1760 cm.-l due to the y-lactone and a second at 1742 ern.-' characteristic of a cyclopentanone, and by the ultraviolet absorption of I1 (Amax 277 mp, e 71). A positive Zimmermann test and an infrared band a t 1410 cm.-l indicated that the keto group of I1 was flanked by a t least one methylene. Chemical evidence for the ketonic function was difficult to adduce, probably because of the ease with which dehydration occurred under the influence of acidic reagents, but the preparation of a t least one crystalline derivative from a degradation product and the conversion to tetrahydroambrosin t o be discussed in the sequel provided satisfactory proof. The main product in the hydrogenation of parthenin was not tetrahydroparthenin (II)] but a substance IV resulting from the uptake of only two atoms of hydrogen, which was resistant to further hydrogenation. A double-strength carbonyl band in the infrared a t 1745 cm.-l indicated that the cyclopentenone chromophore had been reduced, but a relatively strong band a t 1668 cm.-' and the ultraviolet absorption (Xma, 220 and 261 mp, e 14000 and 70) pointed to the continued presence of conjugation. The formation of acetic acid and the absence of formaldehyde, coupled with the appearance of a new sharp signal a t 1.80 p . p m in the n.m.r. spectrum,16 intensity 3 protons, es(14) H . Schinz and M. Hinder, H e b . Chim. Acta, S O , 1349 (1947). See also the ozonolysis product of iresin, C. Djerassi and W. Rittel, J . A m . Chem. Soc., 79, 3528 (1957), and an enolic ketolactone described by C. J. w. Brooks, G. Eglinton and D. s. Magrill, J . Chcm. Soc., 308 (1961). (15) Reference 14 gives XmsI 232 mp, log t 4.1, for or-ketobutyrolactones. In the present instance the absorption is slightly modified by the cyclopentenone chromophore. (16) Spectra were run b y hlr. Fred Boerwinkle of our Department and Mr. L. F. Johnson of Varian Associates on a Varian HR-60 instrument in deuteriochloroform solution. Tetramethylsilane served as
STRUCTURES OF PARTHENIN AND AMBROSIN
July 5 , 1962
tablished the presence of a methyl group on doublybonded carbon. We conclude that the catalyst effected isomerization of C to D in the manner
'C' 0
n b
01u 7
postulated as occurring during the hydrogenation of ambrosin17 and name the new compound dihydroisoparthenin (IV) in accordance with established procedure. 9b,17,18. The fourth oxygen atom of the parthenin formula derives from a hydroxyl group (infrared band at 3430 cm.-l) which was assumed to be tertiary because I1 and IV could not be acetylated with acetic anhydride-pyridine and were not oxidized by chromium oxide a t room temperature. The ease with which parthenin and its derivatives underwent dehydration supported this conclusion. Treatment of parthenin with hot formic acid resulted in the formation of anhydroparthenin (V) which, from its ultraviolet spectrum (Xmas 210 and 296 mp, e 14300 and 125001, was clearly a conjugated dienone. Compound V retained the exocyclic methylene group of parthenin (formation of formaldehyde on ozonolysis, near infrared band a t 1.64 P ) , but had a new vinyl methyl group (singlet methyl proton signal a t 2.03 p.p.m.). Norparthenone on treatment with formic acid formed an analogous anhydro derivative (VI) whose ultraviolet spectrum (Amax 240 and 298 mp, e 9000 and 10900) showed that the dienone system was not conjugated with the enolic a-ketobutyrolactone chromophore. These observations indicate that the tertiary hydroxyl group occupies a position y- or 6 - to the a,P-unsaturated ketone system. T h a t the hydroxyl group is indeed a t the yposition was demonstrated by the facile deoxygenation of parthenin and norparthenone with zinc and acetic acid. l9 Analysis and ultraviolet spectrum (Xmas 295 m,u, e 91) of the product (VIII) from parthenin showed that deoxygenation was accompanied by reduction of the coniugated lactone group. The remaining double bond was triply substituted (e 1750 a t 206 mp) and resistant to hydrogenation. That the zinc-acetic acid reagent is capable of reducing the lactone system separately was also shown by the reduction of coronopilin (IX) 2o to a substance (X) isomeric with tetrahydroparthenin. internal standard and frequencies were calibrated b y the side-band technique. We are grateful t o Dr. M. T. Emerson and Mr. Johnson for assistance with the assignments. The spectrometer at Florida State University was purchased with funds provided by the Institute of Molecular Biophysics. (17) (a) L. Bernardi and G. Biichi, Experienlia, 18, 466 (1957); (b) F. Sorm, M. Such$ and V. Herout, Coil. Cecchosiov. Chem. Commun., 24, 1548 (1959). (18) In the preliminary this substance was referred to as dibydroparthenin and the m.p. was erroneously given as 142-144'; the m.p. of 1V is 2OC-20lo. (19) For a recent example of this type of reduction, see T. G. Halsall, W. J Rodewald and D. Willis, J . Chem. Soc., 2978 (1959). (20) W. Herz and G.H6genauer, J. Org. Chcm., 26, 5011 (1961). (21) The reduction gave a relatively poor yield of crystalline material. Presumably I1 and X are epimeric at CII,with X the thermo-
2603
The deoxygenation product VI1 of norparthenone could be obtained also by dehydration of dihydronorparthenone20 (XI) with thionyl chloride-pyridine a t mom temperature, an observation which renders unlikely the possibility of an acid-catalyzed rearrangement during the conversion of I to VIII, and I11 to VII.
Q. Q 90
0
0
0
g-$ 0 IV
XI1
t
Q OH 1
0
0
0
XI11
f1
c
0
x
11,
0
11
0
0
I
0 V
\
Q
0
0 0
VI11
Q
0 0 VI
0
0
VI1
IX
Other considerations necessary for the placement of the hydroxyl group on the carbon skeleton derive from the following observations. Parthenin, I1 and IV did not consume lead tetraacetate or periodic acid in neutral solution, thus excluding a possible a-keto1 structure. In basic solution, I1 and IV were similarly unreactive toward periodic acid. Hence the tertiary hydroxyl group cannot be placed on a carbon atom vicinal to that carrying the lactone ether oxygen.22 It now remained to deduce the carbon skeleton by dehydrogenation, the usual practice in sesquiterpene chemistry. The presence of a cyclopentenone group suggested that parthenin, like partheniol, was a perhydroazulene and a new member of the class of guaianolides. The dehydrogenation dynamically more stable isomer. This conclusion is supported by the partial conversion of I1 t o X on treatment with potassium carbonate in xylene. (22) In basic solution, parthenin consumed one mole and noLparthenone two moles of periodic acid. We ascribe this to the existence of the equilibrium
0-C-&C-C-
HOH
O=C-CH-C-C-
I
I
I
1
OH OH
260.1
1-01. 8-1 TABLE I N.M.R. PEAKS OF PARTHENIN
Compounds
I
H I
Hs
7.55d( 6)
ti. 18d(6)
5.08d(7)
DERIVATIVES' CHI
Cs-Me
Go-Me
5.59d(3) 6,2F)d(3)
1.28
l.lld(8)
1.11 0.83 1.35
1 . 0 8 , l .11,1.21 (comb. int. 6 protons) 1 . 0 8 d ( i .5) 1.80 2.03
1.30
1.21d(7), 1.24d(6)
4.66df5)
I1 Iv
5'
HE
7,97d(6j
6.09d(6)
5. %it
2 . 90d 3,0413
5.5Obr 4.4Sd( 6)
5.59d(3)
C r M e
6.29