The organic layers were combined, dried, aiid evapd in vacuo. The product was recrystd from a CsHs-hexane solvent mixture to give 174.9 g (727,), mp 117-118". CHCI,.
Methyl 2-Phenoxathiinyl Ketone 10-Oxide (9).--A solution of m-chloroperbenzoic acid (5.3 g, 0.026 mol) in 25 ml of CH2C12 was added to 5.0 g (0.026 mol) of 7 dissolved in 100 ml of CH2CIS. This was maintained at -70' for 30 min. The reaction mixture was washed with aq XaHCO,, and the solvent was evapd in vucuo. Tlc of the residue indicated a mixture of sulfoxide and sulfone. The components were separated by chromatography over silica gel. The sulfone was eluted with 4: 1 c&~-EtO.4c, and t,he desired sulfoxide with EtOAc alone. The column afforded 3.3 g (49.37,) of the siilfoxide, mp 128-130". Methyl 2-Phenoxathiinyl Ketone 10,lO-Dioxide (lo).-To compd 7 (24.2 g, 0.1 mol) in 250 ml of glacial A4cOH at, 60' was added, rapidly, with stirring, 307, H202 (24 ml, 0.24 mol). This was heated at 80" for 4 hr. Upon cooling, the product crystallized from iicOH. hfter filtration and drying, 24.5 g (89.5%) of the product was obtained, mp 168-170". 2-(l-Methoxyethyl)phenoxathiin (20).-111 a flame-dried apparatiis inider X2,50% XaH i n oil suspension (1.0 g, 0.0204 mol) was added t o 24 ml of PhlIe. -4solii of 1 (5.0 g, 0.0804 mol) in 100 ml of P h N e was added. The reaction mixlure was refluxed for 2 hr and then cooled to room temp. Me1 (3.0 g, 0.0204 mol)
was added dropwise with stirring. The reaction mixture was washed with H20 to remove the XaI formed, dried, filtered, and evapd in z'acuo. The residue proved to be t,wo-spot material by tlc. The residue was dissolved in CsHs-EtOilc and passed over 300 g of silica gel. Short-path distillat,ion of the eluate from the column yielded 2.9 g (54.6Ci) of 2-(l-methoxyethyl)phelioxathiin. 2-(Methylcarbamate-l-ethyl)phenoxathiin(21 ).-.4 mixtiire of 1 (6.8 g, 0.028 mol), K2C03 (0.83 g, 0.006 mol), CHsC12 (20 ml), 1 drop of H20, 1 drop of MeOH, and bleSCO (1.87 g, 0.033 mol) was refluxed overnight. The solvent x a s removed in vacuo, and the crude prodirct was re td from lIed3O-hesaiie togive5.1 g (60.8yc),mp 124-126". 2- (1-Acetoxyethyl)phenoxathiin (23).--By the met hod of Fritz and Schenkl3 a soln of 2 .1/ A c 2 0 in EtOAc was made. To 5.0 g (0.0204 mol) of 1 was added 33 ml of 2 d i Ac20 at, 5'. This was stirred for 10 min. The sollL was washed with satd ?;aHCO1, dried, filtered, and evapd in ~'aczio. The crude prodllct was short-path distd t o give 3.8 g of prodiict (9'7.6%).
Acknowledgmeat. --IYc \\.ish t o thwiik 111.. C ' . liedmarl for statist,ical a t i w l j of hiolvgical data. (13) J.
13;.
S.Fritz and G . H. Scllenk. Anal. Chem.. 31, 1808 (19591.
Structure-Activity Correlations for the Central. Nervous System Depressant 2-Imidazolidinones ERICJ. LIEN,'&J I E H D I HUSSAIN,''] AND
:\IICHAEL
1'.
Cr0LI)EPi"
School of I'harmucy, L'niacrsity of Southern California, Cnivwsit!j Park, I,os .tngd(,s, CaliJoi~nic~!/0007
Received September 10, 19/39 The CNS depressant activities of a series of substituted 2-imidazolidinoues are found to be highly dependeiil the log of the octanol-water partition coefficient'. Chlorpromazine fits the same regression lines. Exceptions to this dependence on the lipohydrophilic character are discussed in terms of steric factors.
oil
The dependence of biological activity in a set of congeneric drugs on lipohydrophilic character has been shown in many types of drug action.2 I n particular, a study by Hansch and his coworkersza pointed out that variation in the CNS depression activity of several t'ypes of hypnotics depends parabolically upon the relative lipohydrophilic character as represented by the 1-octanol-water partition coefficient (log P ) . The hypnotics reported included barbiturates, thiobarbiturates, tertiary alcohols, carbamates, substituted amides, and N,N'-diacylureas. In all of these compounds a value of 2 was found for log P for maximum activity (log Po). Similar log Po values for this wide range of structurally dissimilar compounds are taken as a strong indication that these compounds have the same ratedotermining step i n producing CNS depressioti. This step may lie in the peiietratiori of the inter- and/or intracellular membranes or one of the prior barriers such as the blood-brain barrier. As an extension of this type of study, interesting arid ample biological data published by Wright, et u Z . , ~ were analyzed by regression analysis. (1) (a) To whom requests for reprints should he addressed; (b) R . &I. and J. L. Converse Fund Fellow; (c) NSF-C'RP participant. summer. 1969, GP-5829. ( 2 ) (a) C. Hansch. A . H. Steward, S. M. .indemon, and L). Bentlei., J . M e d . Chem., 11, 1 (1968); (b) E. J. Lien, C. Hansch, and S. M . Anderson. i h i d . , 11, 430 (1968); (e) E. J. Lien, J . Agr. Food Chem.. 17, 1265 (1969): ((1) J . I.ien. Sympusium o n Firnpitoxic .kction. 61st .\nniial hleetinr of .hiierican l'li).tupatliulugical Sucietj-, .illgust 19, I Q t i Y , Spokane, \Vasliingtun. (3) It-. 13. K r i y h t . Jr., H. ,J. Brabander, R. Hardy, and .i. C . Osterbery. J. M e d . Chem.. 9 , 852 (1966). 13;.
Methods r '
l h e set of corigeiiers analyzed was a series of substituted 2-imidazolidinones (ethyleneurea derivatives). Chlorproma,zine was included for comparison (see Titble I a,nd Figure 1). To place the drugs on %: conimoti basis for quantitation the biological activity, giveii originally in milligrams per kilogram, n:as converted into log l / c where c is the effective concentration in moles per kilogram. Three different measurements of C S S depressiori (1) reduction of motor were used by Wright, et activity of mice by 50% as measured b;; an actophotomet'er (MDDjo),( 2 ) inhibition of 507; of the test group of mice to walk from the midpoint of a horizontal steel rod to H platform at the ends (HWD,,), and (3) inabilitj. of 50% of tht, test mice t,o rerriaiti o t i w 60" iricliried screeii (ISDja). All injectioiis wercl itit'raperitoneal. The log P values were calculated from the experimental value of 1-(2-dimethylaminoethy1)-3- (m-methoxyphenyl)-2-imidazolidinone and the TT constants reported by Hansch arid his ~ o w o r l i e r s(see ~ , ~Table 11). The Hammett's u constant, a measure of the electronic effect of a substituent, was given adequate analysis, arid the steric effect mas not considered quantitatively. u values mere iiicluded only for the substituents on the phenyl ring. Since t,hP alkyl sub(1) T . I'ujitn, J. Iwasa. and C. Hansch, J . .lmer. C/~t,m.Soc.. 8 6 , 5373 ( I 964).
1
3.0
1 1.0
1.5
2 ,?
3.0
B
,
2.0
2.0
n o t included In eq. ld. L.0
3.0 log
E
log
I1
4.0
3.0
5.0
6.0
6.0
Tasr,l3 I11 Equation
SUMMARY O F k h a . l l l O N S CORREI, lT1NG CNS DEPRESSANT A m i w r Y WIl'H LOG Test n r S
+
log I'o
NO.
?vIDL),a
21
0.58
0.49
4 83
la
+
3IL)L)dJ
19
0.77
0.36
4.30
lb
lIL)Dorj
17
0.84
0.27
AlDD60
15
0.91
0.19
4.16 (3.63-5,96)" 4.17 (3.75-5.11)"
RWIl5a 1:w1>,0 ISl>),o
21 18 17
0.83 0.91 0.92
0.23 0 19 0.14
lug l/c = -0.096 (log P)' 0 933 log P 3 001 log l/c = -0 152 (log P)2 1.307 log P 2 547 log l / c - 0 163 (log P ) z 1 354 log P 2 584 log l / c = -0 163 (log P ) z 1 359 log P 2 638 log l/c = 0.354 log P 2.81:: log 1, c = 0 403 log P 2 662 log 1 'c = 0 331 log P 2 5.52 a 9 5coconfidence interval.
+ + + +
P
+ +
+ + +
Results 'l'lie equations obtained from the regression are sunimwrized in Table 111, where TZ is the number of data points used, I' is the correlation coefficient, and s is the standard deviation. The dependence of the CNS depressant activity on log P is also presented in Figure 1. AIore scattering in the log l / c us. log P plot for the _11DD5,test is observed than in the other two tests. Kevertheless, the (log P ) z term in equations IC and I d is significant a t the 99.5 percentile level. The ideal log P value for the maximum activity (log Po) is about 4.2. In equations IC, Id, and 2b data points with deviations higher than or close to 2s were not included. For equations Id, 2b, and 3, more than 81% ( I ' ~> 0.81) of the variance in the data can be accounted for by these equations. For compounds with u constants available, addition of this term does not result in improved correlations.
Discussion Yariations in log P ( P = octanol-water partition coefficient) account for the relative biological activities very well. The increasing order of activity in both RWDj0 and ISDjo tests of the molecules having various substituents parallels the increasing log P value (see Figure l ) , while in the AIDDjo test a parabolic dependence on log P emerges. This overall agreement is gratifying considering the wide variations resulting from biological testing and the approximations used in calculating most of the log P values. As C S S depressants, it may be assumed that these imidazolidinones would fit into the same pattern as those reported by Hansch and his coworkers.2a Indeed, the urea group of the imidazolidinones is similar to that of the barbiturates and that of the diacylureas. The reahon why linear equations rather than parabolic equations are obtained for the RWD,, and ISDj0 tests may be due to the fact that the log P of the unprotonated forms are used, while a t physiological p H the amino group will be protonated more than 99%. The protonated species will have log P several log units lower than that of the uncharged form. Another possibility is that these may be different type of sedative-hypnotics than those having log Po about 2 . The two groups would have different ratelimiting steps and probably different mechanisms of action.
IC
Id 2x 2b 0
The fact that addition of u term does not improve the correlations significantly may be due to insulation of the electronic effect of the ring substituent by the saturated carbon atoms. In other words the pka values of the compounds examined do not vary significantly. Among the three tests examined, measurement of decreased motor activity (AIDDjo) using an actophotometer appears to be the most sensitive (log l / c = 3.8 to 5.6). Ataxia and paralysis tests (RJVDjo and ISDjo) require higher effective dosages (log l / c = 2.95 to 4.72), but the results give somewhat better correlations (see Figure 1 and eq 2a-3). The linear equations obtained for the RWDSOand ISD,, tests suggest that the log Po for the maximum activity has not yet been reached and that the points examined lie on the initial linear portion of the parabola, while in the MDDjo test it can be seen that the curve begins to level off before chlorpromazine is reached (log P = 5.35). It is interesting to note that the slope and the intercepts of equations 2a, 2b, and 3 are very close. While lipohydrophilic character accounts for most of the variance in the data, there are three major observations not explainable in terms of the partition coefficient. These are the systematically lower than expected activities \Then 12 is increased to three (see Figure l ) , when the phenyl is para rather than Ineta substituted3 and when the cyclic urea ring is unsaturated.6 Although no quantitative data were reported on the para-substituted compounds, and therefore no mathematical analysis of the steric factors is possible, the lessened activities of both the para-substituted and the ? I = 3 compounds may be explained in the following terms. It has been hypothesized that drugs of this type may bind to lipoprotein membranes through dipole-dipole interaction with the protein portion of the membranes, and hydrophobic interactions of the rest of the molecule with the lipoid portion of the These interactions may then change membrane conformation and alter permeability and excitability. In accordance with this, Sprites and Guth have suggested that chlorpromazine acts by decreasing mitochondrial membrane permeability.8 Since (6) W. U. Wright, Jr., 1I. J. Brabander, and R . 11. IIardy. Jr.. J . ATIed. Chem., 9 , 868 (1966). (7) (a) 4 . Shulman and G. Laycock. Eur. J . Pharmncol., 1, 295 (1967); (b) E. J. Lien and W. I).Iiumler, J. .Wed. Chem., 11, 214 (1968). ( 8 ) hl. A. Sprites and P. S. Guth, Nature, 190, 274 (1966).
Stereochemical Studies on Medicinal Agents. \ l l I . l Absolute Stereochemistries of Isomethadol Isoniers"."
The a h J i i i e .-;terecrche~iiisti.ies o f optically acdve CY- itlid ij-isornefhadol ciia~stet~eouiei.~ have beell dediiced 1)). the asymmetric iiiductioii procedure of Prelog. The dissoriatiori c*oiist.aiitsof the title compounds and isomet,hadoiie suggest the absence of intramolecular +X-H. .O i i i ay medium. 11. d a h indicate that, bot,h di:tst,ereomers are int,ernally bonded in CC1, and that the CY isomer is more st'rongly I T bonded. Results from iimr .studies support, the ir dat.a and are consistent with coiiformat8ioiislla and 12a i i i CDCl orily oneisomer (3S,5S)possesses high pot,ency suggests that, tht. proper combiiiatioii of configura The possible influences of conformat,ional isomerism on analgetic act,ivities are dimissed. I t is proposed iha1 the pot,ency differences hetweeii eiiantiomers are due primarily t,o the ohht8i.iirdlive role of t,he 5- or & M e groii~)* of ( +bisomethadone and ( +j-met.hadone.
\\ 1i:ive~rrportrd' previouhlj on t h r absolute stereoc l i ~ n i i ~ t r iofw CY- arid 0-nwthadol enantiomers and have off'cwd an explanation for the inversion of receptor 1terroiPlectivitv at the C-6 n~ymnietriccenter of these diastertwmeric analgetics. A s a continuation of our iriterc.st in this area, ~ t - ehave investigated3 a group of (L
