5-hydroxy-2(5H)-furanone - American Chemical Society

M. F., Oishi, S., Dahlem, A. M., Beasley, V. R., and Carmichael,. W. W. (1988) Analysis and purification of toxic peptides from cyanobacteria by rever...
0 downloads 0 Views 794KB Size
Chem. Res. Toxicol. 1991,4, 540-545

540

method for purification of toxic peptides produced by cyanobacteria. Toxicon 26, 433-439. (28) Harada, K.-I., Matsuura, K., Suzuki, M., Oka, H., Watanabe, M. F., Oishi, S., Dahlem, A. M., Beasley, V. R., and Carmichael, W. W. (1988) Analysis and purification of toxic peptides from cyanobacteria by reversed-phase high-performance liquid chromatography. J. Chromatogr. 448, 275-283. (29) Krishnamurthy, T., Szafraniec, L., Hunt, D. F., Shabanowitz, J., Yates, J. R., Hauer, C. R., Carmichael, W. W., Skulberg, O., Codd, G. A., and Missler, S. (1989) Structural characterization of toxic cyclic peptides from blue-green algae by tandem mass spectrometry. Proc. Natl. Acad. Sci. U.S.A. 86, 770-774. (30) Harada, K.-I., Matsuura, K., Suzuki, M., Watanabe, M. F., Oishi, S., Dahlem, A. M., Beasley, V. R., and Carmichael, W. W. (1990) Isolation and characterization of the minor components associated with mirocystins LR and RR in the cyanobacterium (blue-green algae). Toxicon 28, 55-64. (31) Botes, D. P., Wessels, P. L., Kruger, H., Runnegar, M. T. C., Santikarn, S., Smith, R. J., Barna, J. C. J., and Williams, D. H. (1985) Structural studies on cyanoginosins-LR, -YR,-YA and -YM, peptide toxins from Microcystis aeruginosa. J. Chem. Soc., Perkin Trans. 1, 2747-2748. (32) Harada, K.-I., Ogawa, K., Matsuura, K., Murata, H., Suzuki, M., Watanabe, M. F., Itezono, Y., and Nakayama, N. (1990)

Structural determination of geometrical isomers of microcystins LR and RR from cyanobacteria by two-dimensional NMR spectroscopic techniques. Chem. Res. Toxicol. 3, 473-481. (33) Kaiser, F. E., Gehlrke, C. W., Zumwalt, R. W., and Kuo, K. C. (1974) Amino acid analysis: Hydrolysis, ion-exchange cleanup, derivatization, and quantitation by gas-liquid chromatography. J. Chromatogr. 94, 113-133. (34) Leimer, K. R., Rice, R. H., and Gehrke, C. W. (1977) Complete mass spectra N-trifluoroacetyl-n-butyl esters of amino acids. J. Chromatogr. 141, 121-144. (35) Gelpi, E., Koenig, W. A,, Gilbert, J., and Oro, J. (1969) Combined gas chromatography-mass spectrometry of amino acid derivatives. J. Chromatogr. Sci. 7 , 604-613. (36) Evans, W. C., and Walker, N. (1947) The synthesis of a-amino-y-@-hydroxypheny1)butyric acid, a homolohe of tyrosine. J. Chem. SOC.,1571-1573. (37) Namikoshi, M., Rinehart, K. L., Sakai, R., Sivonen, K., and Carmichael, W. W. (1990) Structures of three new cyclic heptapeptide hepatotoxins produced by the cyanobacterium (blue-green alga) Nostoc sp. strain 152. J. Org. Chem. 55, 6135-6139. (38) Al-Layl, K. J., Poon, G. K., and Codd, G. A. (1988) Isolation and purification of peptide and alkaloid toxins from Anabaena flos-aquae using high-performance thin-layer chromatography. J. Microbiol. Methods 7 , 251-258.

Structure-Activity Relationships of Bacterial Mutagens Related to 3-Chloro-4-(dichioromethyl)-5-hydroxy-2(5H)-furanone: An Emphasis on the Effect of Stepwise Removal of Chlorine from the Dichloromethyl Group Robert T. LaLonde,* Gary P. Cook, Hannu Perakyla, and Lin Bu Department of Chemistry, College of Environmental Science and Forestry, State University of New York, Syracuse, New York 13210 Received March 29, 1991 T h e Salmonella typhimurium (TA100)mutagenicities of six structural analogues of 3chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone (MX) were determined and compared. These were also compared to previously determined mutagenicities for another four analogues. This study was conducted for the primary purpose of ascertaining the effect of C-6 chlorineby-hydrogen replacement on mutagenicity. The compounds assayed were 3-chloro-4-(chloromethyl)-5-hydroxy-2 (5H)-furanone (3), 3-chloro-4-(chloromethyl)-2(5H)-furanone (4), 3chloro-4-methyl-5-hydroxy-2(5H)-furanone(7), 3-chloro-4-methyl-2(5H)-furanone(a), 4methyl-5-hydroxy-2(5H)-furanone(9), and 4-methyl-2(5H)-furanone (10). Compounds 3,4, and 7 were mutagenic whereas 8-10 were not. All six compounds were stable under assay conditions. Mutagenicity data for the three active compounds were combined with data of another four active compounds studied previously to obtain an expanded data set. Mutagenicities of the seven compounds were compared, pairwise, in 21 comparisons and then by multiple regression analysis. On the average, chlorine-by-hydrogen replacement of a single chlorine located a t a chloromethyl group (C-6) had a markedly greater effect in reducing mutagenicity than a similar replacement at C-3 or a hydroxyl-by-hydrogen replacement a t C-5. The chlorine-by-hydrogen replacement at C-6of compound 3 resulted in the greatest mutagenicity reduction of any single replacement and amounted to a 103-fold diminished mutagenicity.

Introduction This paper concerns further structure-activity investigations of chlorinated ligno-humic products possessing the 2(5H)-furanone skeleton. Among this family of compounds is 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone (MX) (1, Figure 1). MX is the most potent member of this family of direct-acting mutagens. It is a product of the chlorine bleaching stage of softwood pulping (I), results from the chlorination of humic waters (2,3),and is observed in numerous samples of chlorine-disinfected

drinking water (4-8), where its contribution to total mutagenicity ranges from 3 to 33% of the water concentrate (5). The molar mutagenicity of MX in the Ames Salmonella typhimurium (TA100) assay ranges from approximately 1000 to IOOOO rev/nmol (7,9,IO).1 The gradual, stepwise reduction of Mx mutagenicity has been effected in a series of MX analogues through a sysAbbreviations of molar mutagenicity units: rev/nmol = revertants per nanomole; rev/fimol = revertants per micromole.

0893-228x/91/2704-0540$02.50/00 1991 American Chemical Society

Mutagenicity of Chlorinated 2(5Z€)-Furanones

confirmation of previously determined mutagenicities. We believed that if mutagenicities for compounds 3, 4, and 7-10 were studied together, comparison of structure-activity data across the larger series of 4-methyl-2(5H)furanones would be justified. This larger series of compounds would start with MX, extend through the compounds with a lower degree of chlorine content, and end with the chlorine-free compounds 9 and 10.

ClQ

ClQ

HO x

Chem. Res. Toxicol., Vol. 4,No. 5, 1991 541

o x o

cQ l HOAgAO

clQ

5

7 HQ

HO A O X O

Experimental Procedures

HAoAO

"Q 6

HQ

8

HxOxo

10

9

HOxoxO

11

clQ

HOxoxO

HxOxo

CQ '

12

HAOAO

Figure 1. Structures of MX (1) and ita 4-methyl-2(5H)-furanone analogues.

tematic, successive replacement of hydroxyl group and chlorine atoms by hydrogen (11). The compounds studied included structures 1-6, Figure 1. The magnitude of mutagenicity reduction for a given type of replacement could range from approximately 10- to 100-fold. This indicated an interactive effect between the chlorine atom and hydroxyl group substituents in some combinations. All compounds studied possessed a t least one chlorine atom at C-6. The mutagenicity of a single C-6 chlorine-free compound also was compared although the mutagenicity of this compound had been determined at an earlier date (10)rather than in the same series of assays that included the compounds whose mutagenicities were being compared. Therefore, we wished a more direct comparison of mutagenicities of compounds possessing the C-6 C1 atom with those lacking this atom. Mutagenicities would be determined for all compounds in the same series of assays. The newly assayed compounds reported here include two monochlorinated compounds, 7 and 8, and two non-chloro compounds, 9 and 10. All four compounds lack a C-6 chlorine atom. Also included in this series of assays were two dichloro compounds, 3 and 4. Both possess a single C-6 chlorine atom and were included in our recently reported studies of the more highly chlorinated 4-methyl2(5H)-furanones (11, 12). Compounds 3 and 4 were included in the present study to measure the decrease in mutagenicity resulting from the total removal of chlorine from C-6. Moreover, their renewed study would allow a

Chemicals. Assayed samples were analyzed for purity by 'H NMR and capillary gas chromatography (GC). These samples were purified by flash chromatography until the 'H NMR and GC showed that the sample was homogeneously one substance with a purity exceeding 99%. Samples of 3 and 4 had been prepared previously (13)but were rechromatographed and analyzed for purity by the usual method, prior to assay. The common starting point for the preparation of compounds 7-10 was 5-morpholino-4-methyl-2(5H)-furanone, prepared by the method of Bourguignon and Wermuth (14). Hydrolysis and ethanolysis of the latter compound, also by the methods of Bourguignon and Wermuth (14),produced compound 9 and its corresponding ethoxy analogue, respectively. Properties of 9 were as follows: mp 50-51 OC; 'H NMR (300 MHz) 6 2.11 (d, 1.5 Hz, 3 H, C-6 H), 5.87 (m, 1 H, C-3 H), 6.03 (8, 1 H, C-5 H), 6.25 (br s, 1H, OH); N M R 6 172.8 (C-2),166.7 (C-4),118.1 (C-3), 100.1 (C-5), 13.3 (C-6). The known compound 10 (15) was prepared by reducing 9 in THF solution with NaBH4. Over a period of 0.5 h, 322 mg of NaBH4 (8.8 mmol) in 30 mL of THF was injected, in a dropwise manner, into a cooled solution of 1.0 g of 9 (8.8 mmol) in 30 mL of THF. After stirring the resulting mixture at 25 OC for an additional 2.5 h, 3 drops of concentrated HC1 was added and stirred for 0.5 h. Then, 6 N HCl was added to pH 2 and the volume doubled with water. The water-THF mixture was extracted once with 75 mL of ether and then 3 times with 25-mL portions of ether. The combined extracts were dried over Na&304, concentrated, dried again with NazS04,dissolved in dichloromethane, filtered, and concentrated once more. The GC-monitored flash chromatography of the resulting slightly yellow oil on silica gel gave 0.5 g of 10 (15) (58%): IR (CHC13)1770 (s), 1740 (s), 1630 (m), 1300 (m), 1140 (s), 1040 (s), 990 (m), 910 (m), 880 (m),840 cm-I (m);GC (120OC, Nzat 1.1mL/min) 4 min, 'H NMR (300 MHz) 6 2.16 (narrow m, 3 H, C-6 H), 4.79 (narrow m, 2 H, C-5 H), 5.84 (9, J = 1.7 Hz, 1H, C-3 H); I3C NMR 6 173.7 (C-2), 166.4 (C-4), 115.2 (C-3), 73.4 (C-5), 13.3 ((2-6). Compound 7 was prepared from 5-ethoxy-2(5H)-furanone. Three grams of the latter compound (21.1 nmol) was dissolved in 40 mL of CC14,in which 300 mg of FeC13was suspended. Dry chlorine was passed through the stirred mixture for 4 h. The mixture was purged with N2,washed with water, and dried. The CCl, was removed by a rotary evaporator. Flash chromatography of the liquid residue on silica gel, eluting with hexanes-EtOAc (1O:l) gave 470 mg of 3-chloro-5-ethoxy-2(5H)-furanone. This product was refluxed for 20 min with 5 mL of concentrated aqueous HCl, cooled, and extracted with EtOAc. The solution was dried and concentrated on the rotary evaporator. The residue was flash chromatographed with hexanes-EtOAc (53) on silica gel to obtain 150 mg of 7 (10) (4.8% overall): mp 71-72 "C; IR (CH2ClZ)1790 (s), 1650 (m), 1470 (m), 1290 (m), 1180 (s), 1140 (m), 970 (s), 810 (m), 790 cm-I (m); 'H NMR (300 MHz) 6 2.09 (narrow m, 3 H, C-6 H), 5.78 (br s, 1 H, OH), 6.05 (br s, 1H, C-5 H);13CNMR 6 167.4 (C-2), 156.6 (C-41, 122.2 (C-3), 98.6 (C-5), 11.7 (C-6). Compound 8 was prepared by reduction of 7 with NaEiH, in methanol according to the procedure of LaLonde et al. for the reduction of 2,3-dichloro-5-hydroxy-2(5Zf)-furanone(16). In this manner, 200 mg of 7 (1.35 mmol) and 80 mg of NaBHl in 12 mL of MeOH produced 140 mg of liquid 8 (78%): GC (initially 170 "C, finally 250 OC at 30 OC/min, N2 at 1.1mL/min) 3 min; IR (CH,Cl,) 1775 (s), 1748 (s), 1690 (s), 1640 (m), 1430 (s), 1400 (s), 1250 (s), 1010 (s), 980 (m), 880 cm-' ( 8 ) ; 'H NMR (300 MHz) 6 2.12 ( ~ ,H, 3 C-6 H), 4.76 ( 8 , 2 H, C-5 H);"C NMR 6 168.5 (C-2), 156.0 (C-4), 119.7 (C-3),72.0 (C-5), 12.4 (C-6);GCMS m / z 132 (C11, M), 117 (M - CH,), 103 (C11, M - [CHO]),97 (M - CI), 87

LaLonde et al.

542 Chem. Res. Toxicol., Vol. 4, No. 5, 1991

Table I. S. typbimurium (TA100) Mutagenicities and a Summary of Statistical Results from the Linear Regression Analyses of Three Assays of Compounds 3,4, and 7, Three MX Analogues Derived from 4-Methyl-2(5H)-furanone spontaneous molar mean molar compd assay no. revertantsa nb slope r2 mutagenicity (I@ mutagenicity (A@ 3 1 177 9 2.80" 0.966 512 9 2.30 0.985 421 2 165 3 172 9 2.20 0.975 402 445 4 1 177 7 23.3O 0.969 3.90 2 165 6 18.5 0.906 3.10 3 172 7 19.6 0.930 3.28 3.42 7 1 177 8 3.27O 0.943 0.486 8 2.95 0.937 0.438 2 165 3 172 6 3.34 0.962 0.495 0.473

Revertants/plate. Number of doses including the zero dose. Revertants/nmol,obtained by multiplying the slope (in units of revertants/ng) by the molecular weight of the compound (in ng/nmol), or by multiplying the slope (in units of revertants/pg) by the molecular weight (in pglpmol) and then dividing by 1OOO. In units of revertants/ng. e In units of revertants/pg. (M - CO,H), 75 (base peak, Cll, M - [CHO + CO]), 73,67 (M - [HCl + CHO]). Anal. Calcd for C6H5C102:C, 45.31; H, 3.80;

C1, 26.75. Found: C, 45.16; H, 3.70; C1, 26.71. Spectra and Chromatography. 'H NMR (300 MHz) and I3C NMR (75.45 MHz) were determined in CDC13solution on a Brucker AMX 300 spectrometer. Chemical shift values (6) are relative to TMS (6 = 0.00 ppm). W spectra of compounds 7-10 were recorded for the stability studies on a Kontron UVIKON860 spectrometer. Solutions of compounds 7-10 were prepared and the study of their stabilities was conducted as previously described (12). IR spectra were recorded by a Perkin-Elmer 1310 spectrometer on samples in solution as indicated in the immediately preceding subsection. Designations s, m, and w refer to strong, medium, and weak, respectively. GCMS were obtained from a Finnigan 4021 mass spectrometer as E1 spectra at 70 eV. Analytical gas chromatography (GC) was performed on a Varian 3300 gas chromatograph equipped with a 30 m x 0.75 mm SPB-5 capillary column. Mutagenesis Assay. The histidine-requiring (his-) S. typhimurium tester strain TAlOO supplied by Dr. Bruce Ames (Universityof California, Berkeley) was used in the standard plate incorporation mutagenesis assays performed according to Maron and Ames (17)for 72-h periods. The mutagenic activities of all compounds, added in freshly prepared DMSO solutions to the top agar, were determined without activation by rat liver homogenate fraction S9. Three plates per dose level were prepared, and mutagenic responses were determined simultaneously for three compounds in three assays, numbered 1-3. Additionally, prior assays had been carried out with compounds to establish toxic dose levels in relation to the lower nontoxic dose-response levels. Values for mutagenicity as rev/@ or rev/ng2 were obtained as the positive linear regression slopes from the linear portion of the dose-response plots. The spontaneous TAlOO mutants, observed from the DMSO controls, were taken as the zero-dose points. Each assay included positive DMSO, crystal violet, ampicillin, and sodium azide controls. Bacterial controls were conducted in quintuplicate. Statistical Treatments. The two sets of mutagenicity measurementsresulting from three current assays of compounds 3 and 4, three previous assays of compound 3, and four previous assays of compound 4 were compared by t test as a difference between two group means using a pooled estimate of the variance. Similarly,the t test was applied to the means of two separate sets of data obtained earlier (11, 12) for compounds 1 and 2. Determinations stemming from these analyses are given under Results. The sample data for the multiple regression analysis consisted of a set of 38 assays of seven mutagenic compounds with various sets of combinations for the substituents represented by W-2, as detailed under Results. The log of each of the 38 mutagenicity values was regressed by the least-squares method on the four predictor variables W-2 to obtain their coefficients. The t test, with 33 degrees of freedom and the standard errors for each coefficient, was employed to test the null hypothesis that there was no difference between any two coefficient values.

* Abbreviations of mutagenicity in units of weight: revlpg = revertants per microgram; rev/ng = revertants per nanogram.

Results The two dichloro compounds 3 and 4 and the monochloro compound 7 proved to be mutagenic. The compounds lacking chlorine, namely, 9 and 10, and the monochloro 8 were nonmutagenic, either because the revertant count at all nontoxic dose levels was indistinguishable from spontaneous, as was the case for compounds 9 and 10, or because the response failed to meet the criteria for mutagenicity of Prival and Dunkel (I@, as was the case for compound 8. The dose-response plots for compounds 3,4, and 7 were obtained from three different assays, in which all three compounds were tested simultaneously. Because the mutagenic doses spanned a range of several powers of 10, it was necessary to multiply the dose of the most active compound, namely, 3, by a factor of 100 in order to display all three sets of plots together. Thus, the three displays given in Figure 2 are produced primarily to exemplify the results of assays 1-3 in a manner that readily assists visual assessment of response linearity and revertant standard deviations. Slopes in units of rev/ng or rev/pgl were obtained from the dose-response plots using linear regression of the data of each plot. Multiplication of these slopes by the molecular weight in units of ng/nmol or pg/pmol gave the molar mutagenicities (M)in units of rev/nmol or rev/pmol, respectively. From values of M, the mean molar mutagenicities (M) were obtained for each of the three mutagenic compounds. Values of the least-squares slopes, M, M ,and relevant statistical indicators are shown in Table I. The M values obtained for compounds 3 and 4 differed insignificantly from those obtained for the same two compounds in the previous study (11). Similarly, the two sets of M values for compounds 1 and 2 from the two previous studies (11,12) differed insignificantly from one another. Mutagenicity reduction as a result of different compound assay stabilities was addressed and dismissed for compounds 3 and 4, whose chemical half-life values were estimated experimentally to be 44 and 2.9 h, respectively (11). The same experiments applied to the monochloro compounds 7 and 8 and the nonchlorinated compounds, 9 and 10, demonstrated that none of these compounds underwent change in a 25-h period under assay conditions. Having concluded that each compound's M values r e sulted from its different intrinsic mutagenicity, the M values of compounds 3,4, and 7 were compared by ratios (Table 11). Table I1 includes the comparisons of not only compounds 3,4, and 7 but also the four additional, more highly chlorinated mutagenic compounds studied earlier (11,12). The comparison of 3 to 7 revealed that the C-6 C1-by-H replacement in 3 resulted in a mutagenicity re-

Chem. Res. Toxicol., Vol. 4, No. 5, 1991 543

Mutagenicity of Chlorinated 2(5H)-Furanones

Table 11. Comparisons of Mean Molar Mutagenicity (i@) by Ratios Resulting from Chlorine Atom and Hydroxyl Group Replacements and Shifts replacements and shifts" compounds mutagenicity ratio m/n to C1 by H compared, H byC1 OH b y H H by OH reduction ratio, nearest m to nb c-3 C-6 c-3 c-5 c-5 m/n multiple of 10 1 to 6 12900 104 1 to 4 922 103 1 to 2 21.5 10' 1020 103 1 to 5 1 to 3 7.71 10' 8350 104 1 to 7 600 103 2 to 6 43.0 10' 2 to 4 47.5 2 to 5 10' 10-1 2 to 3 0.359 + 102 2 to 7 389 1670 103 3 to 6 + 120 102 3 to 4 132 3 to 5 102 10s 3 to 7 1080 14 4 to 6 10' l.llC 4 to 5 10' 4 to 7 9.06 12.6 10' 5 to 6 10' 8.19 5 to 7 0.64gd 6 to 7

+

+ + + + ++ + + + + ++

+

+ +

+ + + +

~

+ + +

+ +

+

+

+ + +

+ +

+ +

+

+

" The indicated replacement or shift results in the conversion of compound m to n. Compound m is given first, n second. There is no significant difference between mutagenicity means of the two compounds compared according to t test. dThe mutagenicity means of the two compared compounds is significantly different at the 95% confidence level according to the t test ( t = 3.886, DF = 5, P = 0.00578). 2800 2600 2400

2200 2000

- c-5 OH

-2

- C-6 CI

184

1800 -

a

\1

4.28

0

F i g u r e 3. A summary comparison of_4-methyl-2(5H)-furanone mean molar mutagenicity values (M) in units of rev/nmol.' Compounds 1-12 are arranged to illustrate the effect of the stepwise removal of C-3 and C-6 C1 and C-5 OH on S. typhimurium (TA100) mutagenicity starting from the most highly chlorinated and hydroxylated compound, MX (1). When no M v_alueis given, no results could be obtained for the compound. M values given for compounds 3 and 4 were obtained- by combining results from this and a previous study (11). The M values given for compounds 5 and 6 were obtained from the same previous study while M for compounds 1 and 2 were obtained from combining values from two previous studies (11, 12).

1600

F 1400 0 1200

4

>

; 1000 800 600 400

200

o

20

40

60

80 100 120 140 160 '80

D o s e in m c g x f a c t o r

F i g u r e 2. Doseresponse plots for assay 1 for compounds 3 (a), 4 ( O ) ,and 7 (A). Standard deviations (SD) are represented by the vertical error bars, which are concealed when S D is less than the height of a symbol. The horizontal-axis units are pg (mcg). Doses of compounds 4 and 7 were a t the microgram level, and the scaling factor is 1. Doses of compounds 3 were a t the nanogram level, but were converted to micrograms and multiplied by a scaling factor of 100 for uniformity of display and ease of comparison in this figure.

duction of 1000-fold. Comparison of 3 to 4 showed that the (2-5 OH-by-H replacement in the same compound produced a mutagenicity reduction of about 100-fold. Accordingly, the double group replacement indicated in the comparison of 4 to 7 resulted in a 10-fold reduction. To assess generally the influence the several replacements were having,the combined total of 38 mutagenicities from this and previous studies (11,12) were regressed on four predictor variables, W - 2 , representing four substit-

uents attached to the I-methyl-2(5H)-furanone skeleton (Table III). A value of 1or 0 was assigned to the predictor variable when W , X , or Y was represented by a chlorine or hydrogen atom, respectively. Similarly, a value of 1or 0 was assigned when 2 was represented by a hydroxyl group or hydrogen, respectively. Values assigned to X and Y were restricted to the following. If X was 0, then Y could be 1or 0; if X was 1, then Y must be 1. The relation of predictor variables to dependent variables is given in the model represented by eq 1,which was transformed to the working model, eq 2. M = Ae(aW+@X+yY+6Z+c) (1)

+

log M = constant + CYW+ /3X + rY + 6.2 e (2) Regression of log M on variables W, X,Y, and 2 yielded the regression coefficients given in Table 111, which also includes the pertinent statistical indicators for this regression. The coefficient y for the Y variable is significantly greater than any of the other three coefficients. The difference between CY and 6 is insignificant, where= the

LaLonde et al.

544 Chem. Res. Toxicol., Vol. 4, No. 5, 1991

c1

+OH-

C1

- H2O d

f--

coo -

-t H2O

-OH13 14 Figure 4. The base-promoted opening of the 2(5H)-furanone ring and subsequent migration of the carbon-carbon double bond in 11

..

11.

Table 111. Multiple Linear Regression: Log M on Predictor Variables E', X , Y,and 2,Representing Four Substituents Attached to 4-Methyl-2(5H)-furanone

Z

variable

W X Y

Z const

LOAO probabilitv