- Email: [email protected]
Quantitative structure-activity relationships and mixture toxicity of organic chemicals in Photobacterium phosphoreum: The microtox test
ECOTOXICOLOCY
AND
ENVIRONMENTAL
SAFETY
9, 17-25
(1985)
Quantitative Structure-Activity Relationships of Organic Chemicals in Photobacterium The Microtox Test FRANS
JOOPEIERMENS, Department
BUSSER,
and Mixture Toxicity phosphoreum:
PETERLEEUWANGH,'
of Veterinary Pharmacology. Pharmacy and Toxicology, Biltstraat 172, 3572 BP Utrecht, The Netherlands Received
January
ANDAALTMUSCH University
of Utrecht.
3. 1984
Quantitative structure-activity relationships were calculated for the inhibition of bioluminescence 01‘ Photobacterium phosphoreum by 22 nonreactive organic chemicals. The inhibition was measured using the Microtox test and correlated with the partition coefficient between noctanol and water (Pm,), molar refractivity (MR), and molar volume (MW/d). At log P,, < 1 and > 3, deviations from linearity were observed. Introduction of MR and MW/d improved the quallity of the relationships. The influences of MR or MW/d may be related with an interaction of the tested chemicals to the enzyme system which produces the light emission. The sensitivity of the Microtox test to the 22 tested compounds is comparable to a 1Qday acute mortality test with guppies for chemicals with log P, < 4. The inhibition of bioluminescence by a mixture of the tested compounds was slightly less than was expected in case of concentration addition. The Microtox test can give a good estimate of the total aspecific “minimum toxicity” of polluted waters. When rather lipophilic compounds or pollutants with more specific modes of action are present, this test will underestimate the toxicity to other aqUatiC life. 0 1985 Academic Press, Inc.
INTRODUCTION In aquatic toxicology quantitative structure-activity relationships (QSAR) have been applied mostly to the study of acute lethal toxicity of chemicals (Veith and Konasewich, 1975; Kiinemann, 1981a; Konemann and Musch, 1981). The acute lethal toxicity (14-day LC& of a group of 50 nonionized nonreactive organic chemicals was correlated successfully with partition coefficients between n-octanol and water (P,,,) by Konemann (1981a). Similar studies with the same group of chemicals Iwere carried out by Slooff et al. (1983) with 14 aquatic species and by Hermens et al. (1984a) in toxicity tests with Daphnia magna. The acute lethal toxicity of these chemicals is thought to be related with membrane perturbation, comparable with the action of the volatile anesthetics (Kiinemann, 198 la). We are interested in the question of whether such good correlations also occur in studies with other aquatic species and with toxicity criteria on the sublethal level. In this study such a sublethal toxicity test, the Microtox test, was used. This Microtox test, developed by Beckman Instruments Inc., is a very rapid bioassay based on the measurement of the inhibition of bioluminescence of Phofohacterium phosphoreum. EC& values (concentrations at 50% inhibition of bioluminescence) of 22 chemicals from the group of nonreactive organic chemicals were correlated with Pwt and molar refractivity (MR). PO,, represents, among other things, the influence r Present address: Institute
for Pesticide
Research,
Marijkeweg 17
22, 6709 PG Wageningen, 0147-6513/85
The Netherlands. $3.00
Copyright B 1985 by Academic Press. Inc All rights of reproduction tn any form reserved.
18
HERMENS ET AL.
has been used of hydrophobicity on membrane permeation. Molar refractivity successfully in enzyme-binding studies (Hansch and Leo, 1979). MR, as a measure of polarizability of molecules, is related to London dispersion forces and is usually experimentally obtained with the Lorenz-Lorentz equation (Hansch and Leo, 1979) MR=--
n2 - 1 MW n2+2 d ’
(1)
where n = index of refraction, d = density, and MW = molecular weight. Within a group of chemicals a high quality QSAR may indicate a similar mode of action. Konemann (198 1b) and Hermens et al. ( 1984a,b) demonstrated that the toxicity of mixtures of similarly acting chemicals can be predicted by concentration addition. We also determined the effect of a mixture of the tested chemicals on the inhibition of bioluminescence. The result of this experiment was evaluated with the Mixture Toxicity Index (MTI) as proposed by Konemann ( 198 1b). METHODS The Microtox test (Beckman, Model 2055) and calculation of EC& values (concentrations which reduce bacterial luminescence by 50%) were performed according to the procedures described in the Beckman Instruments Manual (1980). Five concentrations of the tested chemical (ratio between concentrations was a factor 2) were incubated with the bacteria for 15 min at 15°C. Pm values were calculated using the “hydrophobic fragmental constant” method of Rekker (1977); MR values were calculated by adding fragment values as tabulated by Hansch and Leo (1979). QSAR were calculated with a computer program based on the method of least squares. Statistical comparison of the quality of the QSAR was based on comparison of the standard deviations with an F test as described by Hansch (1973, pp. 150-158). RESULTS
AND
DISCUSSION
I. QSAR STUDIES The EC& values, together with the physicochemical properties of the tested compounds are summarized in Table 1. QSAR of all 22 compounds are presented in Table 2. Introduction of a (log P)= term (Eq. 3) does not improve the quality of the QSAR, while introduction of MR (Eq. 4) results in a significantly better correlation (P < 0.01). It is surprising that introduction of (log P)= does not result in a better QSAR, since in Fig. 1 a deviation from linearity is observed at high log P values. At low log P values, however, a nonlinear behavior is also observed. Probably these two phenomena, opposite in direction, cancel out in Eq. 3. Therefore we calculated QSAR in two log P ranges in which only one deviation from linearity is observed. Tables 3 and 5 summarize the QSAR in the log P ranges of - 1.3 to 3 and 1 to 5, respectively. Influence of (log POCJ2 1. Log P range -1.3 to 3. The positive (log P)= term in Eqs. 7 and 9 represent the nonlinear dependence of log (l/E&) from log P for the hydrophilic compounds
QUANTITATIVE
STRUCTURE-ACTIVITY
TABLE
1
ECso VALUES AND PHYSICOCHEMICALPROPERTIES
No. 1
2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18 19 20 21 22
Compound Diethyleneglycol hdethanol Acetone n-Propanol n-Butanol Diethylether Pentanol-3 Dichloromethane 1,2-dichloroethane n-Hexanol Benzene Trichloroethene n-Heptanol Toluene I, I,1 -Trichloroethane hdonochlorobenzene n-Octanol o-Xylene 1,3-Dichlorobenzene 3,4-Dichlorotoluene 1,2.3-Trichlorobenzene 1,2,3,4-Tetrachlorobenzene Lindane
19
RELATIONSHIPS
OFCOMPOUNDS
1%
1%
ECso”
P mb
MR’
5.44 6.12 5.56 5.16 4.58 4.88 4.23 4.53 4.05 2.82 3.31 3.16 1.93 2.29 1.78 2.12 1.56 1.94 1.35 0.94 1.14 0.94’ 1.58’
-1.30 -0.79 -0.30 0.27 0.80 0.88 1.21 1.51 1.76 1.86 2.13 2.20 2.39 2.59 2.75 2.81 2.92 3.09 3.53 3.98 4.20 4.94 3.53/
25.3 8.20 15.9 17.5 22.2 22.2 26.8 16.4 21.0 31.3 26.4 25.5 35.9 31.0 26.0 31.4 40.5 35.6 36.4 41.0 41.4 46.4
’ Experimentally determined concentrations at 50% inhibition
MWldd 94.8 40.5 73.6 74.8 91.5 103.8 108.2 64.0 80.1 125.6 88.9 89.7 141.4 106.3 99.6 101.8 157.5 120.6 114.1 128.2
of bioluminescence of Photobacterium
phosphoreum (rmol/liter). b P,, calcu!lated according to Rekker ( 1977). ’ MR calculated with fragment values as tabulated by Hansch and Leo (1979). d MW/d (molecular weight/density) calculated with data taken from Weast (197 I). ’ Extrapolated ECSa; no concentrations with more than 50% inhibition could be tested. ‘From Hetmens and Leeuwangh (1982).
with low log P. Equations 7 and 9 are significantly better (P -C 0.01) than the equations without a (log P)* term (Eqs. 6 and 8). Deviations from linearity at low log P values can be related with the diffusion of small hydrophilic molecules through aquous membrane pores (Seydel and Schaper, 1982). It is assumed that water soluble molecules with diameters lower than 4 X 10P” m, such as methanol, ethanol, and urea, are able to pass freely through the aquous membrane channels (O’Flaherty, 1981, p. 89). In a study of permeability through human and dog red cell membranes of a seriesof straight chain amides by Sha’afi et ~1. (197 1) a deviation from the linear relation between permeability and chain length was found for those amides with shorter chain lengths than propionamide. The PO,,of propionamide. calculated according to Rekker (1977), is -0.74. The significant improvements of the QSAR with a positive (log P)2 term in Eqs. 7 and 9 can be related to the diffusion of compounds (for instance diethyleneglycol, methanol, and acetone) through aquous membrane pores because the diameter of these straight chain hydophilic moleculesis smaller than 4 X lo-” m. A complication is that MR of diethyleneglycol, the most hydrophilic compound, is relatively high.
20
HERMENS
ET
TABLE QSAR
AND
2
CORRELATIONSBETWEENPHYSICCXHEMICAL PROPERTIES OFALL
COMP~UNDS(TABLE
1,No.
l-22)
Equation (n = 22)”
Eq. no.
2 3 4 5
AL.
log ( 1/EC,,) = 0.995 log P - 5.14 0.011 (log P)2 + 0.957 log P - 5.13 0.667 log P + 0.0636 MR - 6.30 MR = 5.15 log P + 18.2
rb
SC
0.952 0.952
0.53 0.54 0.42 5.2
0.971
0.852
’ ECSo= concentration at 50% inhibition of luminescence by P. phosphoreum; P = partition coefficient between n-octanol and water; MR = molar refractivity. b r = correlation coefficient calculated by rz = 1 - (SS,/S$) in which SS, is sum of squares of deviations about regression and SS2 is equal to the sum of squares of deviations about mean value of log ECso. c s = standard error of estimate.
Omitting this compound in the QSAR results, however, in the same significant improvements by introduction of a (log P)2 term (Eqs. 15 and 17 from Table 4). 2. Log P range I to 5. Introduction of a negative (log P)2 term (Eqs. 20 and 22, Table 5) results in significantly better correlations (P < 0.01). The coefficients of the (log P)2 term in these equations are opposite in sign to those in Tables 3 and 4, which is due to the opposite deviations from linearity. Nonlinear relationships between biological activity and hydrophobicity at high PKt are often found (Hansch, 197 1) and may be related with a reduced rate of uptake for the higher lipophilic compounds. The optimum value for hydrophobicity (log PO) in Eq. 20 is 4.6. This optimum hydrophobicity depends on, among other things, the lipid content of the membranes. The value of 4.6 for the inhibition of bioluminescence by P. phosphoreum (a gramnegative bacterium) is very near to the log PO of about 4.4 for other data with gramnegative bacteria, as summarized by Hansch and Glave (197 1).
;;~.,/,*~,“, ,,, -1
0
12
3
4
5
log %ct
FIG. 1. Relation between concentrations at 50% inhibition of bioluminescence and hydrophobicity compounds from Table I (0, oxygen compounds: 0, compounds which do not contain an oxygen X. lindane).
for atom:
QUANTITATIVE
STRUCTURE-ACTIVITY
TABLE QSAR AND CORRELATIONS IN log Eq. no.
BETWEEN
P,, RANGE
PHYSICOCHEMICAL TO 3.0 (TABLE
MR = 0.270 MW/d
PROPERTIES 1, No. l-17)
OF COMPOUNDS
sn
(n = 17)”
log ( I/EGO) = 1.06 log P - 5.21 0.273 (log P)’ + 0.537 log 0.734 log P + 0.0745 MR 0.189 (log P)2 + 0.475 log 0.228 (log P)2 + 0.467 log 0.286 (log P)’ + 0.644 log MR = 4.32 log P + 18.9
6 7 8 9 10 11 12 13
3
-1.3
Equation
21
RELATIONSHIPS
P P P P
0.929 0.972 0.97 1 0.987 0.988 0.986 0.706 0.949
- 5.45 6.61 + 0.0511 MR - 6.34 + 0.0123
MW/d
- 6.38
+ 0.593 Z - 5.99
- 1.2
u See footnotes to Table 2: MW/d = quotient molecular weight and density; for oxygen compounds and 0.0 for chemicals which do not contain an oxygen
I = indicator atom.
0.56 0.37 0.38 0.26 0.25 0.27 5.8 2.6 variable,
1 .O
The QSAR with log P alone (Eqs. 2, 6, 14, 19, 25, and 30) have rather high coefficients. This was also found by Hansch (1973, p. 131) in studies of the inhibition of bioluminescence by a series of alcohols. Hansch concluded that inhibition of bacterial luminescence appearsto involve a different kind of membrane processthan inhibition of bacterial growth, becauseQSAR with growth as parameter have slopesof about 0.7. Influence
of MR, MW/d
The improvement of the QSAR with MR (significant at at least P < 0.05) in Eqs. 8, 9, 16, 17, 2 1, and 22 is related to the relatively higher toxicity and molar refractivity of the oxygen compounds in comparison with those compounds which do not contain an oxygen atom (seealso Figs. 1 and 2). MR brings these two groups together in one equation. Becausethe correlations between log( l/E&) and log P,,, for these two groups are practically parallel (Fig. 1) with an indicator variable, these two groups can be brought together in one equation (Eqs. 11 and 23). Therefore it is interesting to see if differences in polar&ability also influence the activity within the two groups. QSAR with these two groups are given in Tables 6 and 7. Including TABLE QSAR AND CORRELATIONS IN log Eq. no.
14 15 16 17 18 L?See footnotes
BETWEEN
P,, RANGE Equation
-0.8
4
PHYSICOCHEMICAL TO 3.0 (TABLE
PROPERTIES 1, No. 2-17)
(n = 16)”
log ( l/&o) = I.22 log P - 5.54 0.219 (log P)* + 0.702 log P - 5.52 0.861 log P + 0.0595 MR - 6.46 0.209 (log P)* + 0.378 log P + 0.0567 MR - 6.40 MR = 6.11 log P + 15.4 to Table
2.
OF COMPOUNDS
ra
sa
0.954 0.971 0.97 1 0.986 0.844
0.45 0.37 0.38 0.27 4.6
HERMENS
22
ET AL.
TABLE
5
QSAR AND CORRELATIONS BETWEEN PHYSICOCHEMICAL PROPERTIES OF COMPOUNDS IN log PXt RANGE 1.O TO 5.0 (TABLE 1, No. 7-22) Eq. no.
Equation
ra
(n = 16)”
sa
log ( ~/EGO) =
19 20 21 22 23 24
log P - 5.22 (log P)' + 2.98 log P - 7.80 log P + 0.0622MR - 6.10 (log P)* + 2.49 log P + 0.0621MR - 8.55 (log P)* + 3.27 log P + 0.750 Z - 8.56 MR = 6.82 log P + 13.4 1.03 -0.322 0.582 -0.311 -0.351
0.899 0.954 0.930 0.979 0.988 0.854
0.53 0.38 0.46 0.27 0.20 4.4
‘See footnotes to Tables 2 and 3.
MR in the equations with the oxygen compounds
result in less significant (P < 0.1) improvements of the correlations (Eqs. 27 and 28) in comparison with those in Tables 3-5. Within the group of compounds which do not contain an oxygen atom, the influence of MR is not obvious. The standard deviation of Eq. 32 is the same as in Eq. 30, while including MR in the QSAR with (log P)* and log P results in a significantly lower standard deviation at P < 0.05 (Eqs. 3 1 and 33). Also, the coefficient of MR in Eq. 33 is opposite in sign to all the other equations with MR. Maybe this phenomenon is related with the colinearity between log P and MR. In comparison with the other calculated QSAR, the correlation between log P and MR for this group is very high (compare Eq. 34 with Eqs. 12, 18, 24, and 29). MR usually is higher colinear with molar volume (Hansch et al. 1977). Replacing, for instance, MR in Eq. 9 by the quotient molecular weight/density (MW/d), as a measure of molar volume, in fact results in a QSAR of similar quality (Eq. 10). As expected, the correlation between MR and MW/d for these compounds is rather high (Eq. 13). The influence of molar refractivity (MR), molar volume (MW/d), or the indicator variable (I) can be explained in several ways: The influence of molar volume on membrane permeation in our experiments is unlikely. Negative influences of molecular weight and molar volume on membrane MR
50. 0 40.
. 0
. .
30.
0
-
0
0
.*
lo
0
00
.
. 20.
00
0
.
-1
0
1
2
3
4
5
log
pact
FIG. 2. Relation between MR and log P, for compounds from Table 1. (0, oxygen compounds; 0, compounds which do not contain an oxygen atom).
QUANTITATIVE
STRUCTURE-ACTIVITY
TABLE
23
RELATIONSHIPS
6
QSAR AND CORRELATIONS BETWEEN PHYSICOCHEMICAL PROPERTIES OFOXYGEN COMPOUNDS (TABLE1, No. l-7, 10, 13, 17) Eq. no.
Equation (n = 10)’
ra
sn
25 26 27 28 29
log ( 1/EC,,) = 1.09 log P - 5.09 0.265 (log P)* + 0.664 log P - 5.37 0.609 log P + 0.0827 MR - 6.75 0.165 (log P)’ + 0.544 log P + 0.0483 MR - 6.23 MR = 5.82 log P + 20.0
0.938 0.982 0.98 1 0.990 0.823
0.58 0.34 0.35 0.27 5.8
‘See footnotes to Table 2
permeability have been observed by Lien (1975) and Sha’afi et al. (197 l), while in the calcula.ted QSAR the influence of molar volume is positive. According to Hansch and Leo (1979) positive coefficients with MR suggesta binding action via dispersion forces, possibly in the more polar regions of the receptor molecule. A4R is used successfully in correlation studies involving enzyme binding (Hansch and Leo, 1979). The light-emitting reaction in bioluminescence is catalyzed lby the enzyme luciferase (Hastings and Nealson, 1977). Therefore the influence of MR may be related with the inhibition of this enzyme or of the reduced Flavin mononucleotide (FMNH,)-donating system. The higher activity of the oxygen compounds and therefore the influence of the indicator variable can be explained with the possibility of these compounds to form hydriogen bonds with the receptor molecule. According to Franks and Lieb (1982) and Sandorfy (1978) an aspecific effect, such as anethesia, may also depend on a polar and molar volume factor as well as on hydrophobicity. It cannot be indicated which explanation is the right one or if other mechanisms are involved. This indistinctness is partly caused by the high colinearity between the paramleters for the tested compounds. If more data on the toxicity of other compounds in the Microtox test become available, probably more unambiguous conclusions can be drawn. It is clear however that in this test hydrophobicity is not the only parameter which influences the activity of the compounds. TABLE QSAR A:VD CORRELATIONS WHICH
Do
NOT
CONTAIN
BETWEEN AN OXYGEN
PHYSICOCHEMICAL ATOM (TABLE
PROPERTIES 1,
OF COMPOUNDS
No. 8. 9, 1 I. 12. 14- 16. 18-22)
Equation (n = 12)”
Eq. no. 30 31 32 33 34
7
ra
s*
0.924 0.99 1 0.930 0.995 0.973
0.49 0.18 0.49 0.14 2.2
1% ( 1/EGO) =
1.08 log P - 5.49 -0.411 (log P)* + 3.69 log P - 9.21 0.559 log P + 0.062 MR - 5.91 -0.469 (log P)’ + 4.55 log P - 0.059 MR - 9.34 MR = 8.39 log P + 6.7
a See footnotes to Table 2.
24
HERMENS ET AL.
II. MIXTURE
TOXICITY STUDY AND AN EVALUATION OF THE MICROTOX
TEST
Exposure of the bacteria to an equitoxic mixture (concentrations in identical fractions of the ECSo) of 21 of the tested compounds (without diethyleneglycol) resulted in a C (c/EC&, at 50% inhibition of bioluminescence, of 2.0. This corresponds with a Mixture Toxicity Index of 0.77 + 0.08 (calculated according to Kijnemann (198 1b) and assuming a standard error in log ECsO of 0.10). A MT1 of 1.0 corresponds with concentration addition and is expected for mixtures of chemicals with “simple similar actions” (Kiinemann, 198 lb). The high quality QSAR, which could be calculated, suggest that the chemicals act similarly. Slight deviations from concentration addition were observed earlier in other sublethal toxicity studies with D. magna (Hermens et al. 1984a). The sensitivity of the Microtox test to the tested compounds is comparable to a 1Cday acute mortality test with guppies, as can be concluded by comparing the QSAR intercepts. The intercept value in Eq. 2, which amounts to 5.14, does not deviate much from the intercept of 4.87 of a QSAR with guppies calculated with the same group of chemicals by Kijnemann (198 1a). The Microtox test, however, is not sensitive to very lipophilic compounds. A deviation from the linear relationship log ECso vs log P,,,t occurs at log Pact of about 3.0 (Fig. l), while in the 1Cday test with guppies no deviation from linearity was observed at log Pmt < 5.7 (Kiinemann, 1981a). As suggested by Kanemann, this QSAR probably predict a “minimum toxicity” of hydrophobic nonionized organic chemicals. This minimum action can be masked by more specific modes of action. Lindane, for instance, is about 300 times more toxic in the 1Cday LC 50 experiment with guppies (Hermens and Leeuwangh, 1982). In the Microtox test, however, only the aspecific minimum toxicity of lindane is observed (see Fig. 1). It is very likely that, with other compounds which act more specifically in toxicity tests with higher organisms, only the aspecific “minimum toxicity” will appear in the Microtox test. Further, because only a slight deviation from concentration addition was found in the mixture toxicity study, this Microtox test can give a good estimate of the total aspecific “minimum toxicity” of, for instance, an effluent. When such an effluent contains very lipophilic compounds or pollutants with more specific modes of action, the Microtox test will underestimate the toxicity to other aquatic life. ACKNOWLEDGMENTS This work was supported by the Ministry of Housing, Physical Planning, and Environment (The Netherlands). We thank Professor Dr. H. van Genderen and Dr. H. KBnemann for their stimulating discussion and comments on the manuscript.
REFERENCES BECKMAN Interim FRANKS,
INSTRUMENTS INC. (1980). Operating Instructions, Microtox (Model 2055) Toxicity Manual 110679B-9-80. N. P., AND LIEB, W. R. (1982). Molecular mechanisms of general anaesthesia. Nature
Analyzer. (London)
300,487-493. C., AND ANDERSON, S. (1967). The structure-activity relationship in barbiturates and in similarity to that in other narcotics. J. Med. Chem. 10, 745-753. HANSCH, C. (197 1). Quantitative structure-activity relationships in drug design. In Drug Design (E. J. Ari&ns, ed.), Vol. I, pp. 271-342. Academic Press, New York. HANSCH, C., AND GLAVE, W. R. (197 1). Structure-activity relationships in membrane perturbing agents. Mol. Pharmacol. 7. 337-354. HANSCH,
QUANTITATIVE
STRUCTURE-ACTIVITY
RELATIONSHIPS
25
HANSCH, C. (1973). Quantitative approaches to pharmacological structure-activity relationships. In Structure-Activity Relationships (C. J. Cavallito, ed.), Vol. I, Pergamon Press, Oxford/New York. HANSCH, C., SMITH, R. N., ROCKOFF, A., CALEF, D. F., Jow, P. Y. C., AND FUKUNAGA, J. Y. (1977). Structure-activity relationships in papain and bromelain hgand interactions. Arch. Biochem. Biophys. 183, 383-392.
HANSCH, C.. AND LEO, A. J. (1979). Sobstituent Constants for Correlation Analysis in Chemistry and Biology. Wiley, New York. HASTINGS, J. W., AND NEALSON, K. H. (1977). Bacterial luminescence. Annu. Rev. Microbial. 31, 549595.
HERMENS, J., AND LEEUWANGH, P. (1982). Joint toxicity of mixtures of 8 and 24 chemicals to the guppy (Poecilia reticulata). Ecotoxicol. Environ. SaJ 6. 302-3 10. HERMENS, J.. CANTON, H., JANSSEN. P.. AND DE JONG, R. (1984a). Quantitative structure-activity relationships and mixture toxicity studies of chemicals with anaesthetic potency: Acute lethal and sublethal toxicity to Daphnia magna. Aquat. Toxicol. 5, 143-154. HERMENS. J., LEEUWANGH, P., AND MUS~H, A. (1984b). Quantitative structure-activity relationships and mixture toxicity studies of chloro- and alkylanilines at acute lethal toxicity level to the guppy (Poecilia reticulata). Ecoio-yicoi. Environ. Suf 8. 388-394. KONEMANN, H. (198 la). Quantitative structure-activity relationships in fish toxicity studies. 1. Relationship for 50 industrial pollutants. To,yicology 19. 209-22 I. KONEMANN, H., (198 lb). Fish toxicity tests with mixtures of more than two chemicals: A proposal for a quantitative approach and experimental results. Toxicology 19, 229-238. KBNEMANN, H., AND MUSCH, A. (1981). Q uantitative structure-activity relationships in fish toxicity studies. 2. ‘The influence of pH on the QSAR of chlorophenols. Toxicology 19, 223-228. LIEN, E. J. (1975). Structure-absorption-distribution relationships: Significance for drug design. In Drug Design (E. J. Ariens. ed.), Vol. 5, pp. 81-131. Academic Press, New York. O’FLAHERTY. E. J. (198 1). To.xicants und Drugs: Kinetics and Dynamzcs. Wiley, New York. REKKER, R. IF.(1977). The Hydrophobic Fragmental Constant. Elsevier, Amsterdam. SANDORFY,C. (1978). Intermolecular interactions and anaesthesia. Anesthesiology 48, 357-359. SEYDEL, J. K.. AND SCHAPER. K. J. (1982). Quantitative structure-pharmacokinetic relationships and drug design. Pharmucol. Ther. 15. I3 I- 182. SHA’AFI. R. I.. GARY-B• BO. C. M., AND SOLOMON. A. K. (1971). Permeability of red cell membranes to small hydrophylic and lipophilic solutes. J. Gen. P!z~~siol.58, 238-258. SLOOFF, W.. CANTON, J. H.. AND HERMENS. J. (1983). Comparison of the susceptibility of 22 fresh water species to I5 chemical compounds. I. (Sub)acute toxicity tests. Aquat. To.~icol. 4, 113-128. VEITH, G. D.. AND KONASEWICH, D. E. (eds.) (I 975). Structure-activity Correlations in Studies of Tosicity and Bioconcentration with Aquutic Species. Great Lakes Advisory Board, Windsor, Ontario. WEAST. R. C. (ed.) (197 1). Handhook of Chernistr.v und Phy.sic.s.CRC Press. Cleveland.
AND
ENVIRONMENTAL
SAFETY
9, 17-25
(1985)
Quantitative Structure-Activity Relationships of Organic Chemicals in Photobacterium The Microtox Test FRANS
JOOPEIERMENS, Department
BUSSER,
and Mixture Toxicity phosphoreum:
PETERLEEUWANGH,'
of Veterinary Pharmacology. Pharmacy and Toxicology, Biltstraat 172, 3572 BP Utrecht, The Netherlands Received
January
ANDAALTMUSCH University
of Utrecht.
3. 1984
Quantitative structure-activity relationships were calculated for the inhibition of bioluminescence 01‘ Photobacterium phosphoreum by 22 nonreactive organic chemicals. The inhibition was measured using the Microtox test and correlated with the partition coefficient between noctanol and water (Pm,), molar refractivity (MR), and molar volume (MW/d). At log P,, < 1 and > 3, deviations from linearity were observed. Introduction of MR and MW/d improved the quallity of the relationships. The influences of MR or MW/d may be related with an interaction of the tested chemicals to the enzyme system which produces the light emission. The sensitivity of the Microtox test to the 22 tested compounds is comparable to a 1Qday acute mortality test with guppies for chemicals with log P, < 4. The inhibition of bioluminescence by a mixture of the tested compounds was slightly less than was expected in case of concentration addition. The Microtox test can give a good estimate of the total aspecific “minimum toxicity” of polluted waters. When rather lipophilic compounds or pollutants with more specific modes of action are present, this test will underestimate the toxicity to other aqUatiC life. 0 1985 Academic Press, Inc.
INTRODUCTION In aquatic toxicology quantitative structure-activity relationships (QSAR) have been applied mostly to the study of acute lethal toxicity of chemicals (Veith and Konasewich, 1975; Kiinemann, 1981a; Konemann and Musch, 1981). The acute lethal toxicity (14-day LC& of a group of 50 nonionized nonreactive organic chemicals was correlated successfully with partition coefficients between n-octanol and water (P,,,) by Konemann (1981a). Similar studies with the same group of chemicals Iwere carried out by Slooff et al. (1983) with 14 aquatic species and by Hermens et al. (1984a) in toxicity tests with Daphnia magna. The acute lethal toxicity of these chemicals is thought to be related with membrane perturbation, comparable with the action of the volatile anesthetics (Kiinemann, 198 la). We are interested in the question of whether such good correlations also occur in studies with other aquatic species and with toxicity criteria on the sublethal level. In this study such a sublethal toxicity test, the Microtox test, was used. This Microtox test, developed by Beckman Instruments Inc., is a very rapid bioassay based on the measurement of the inhibition of bioluminescence of Phofohacterium phosphoreum. EC& values (concentrations at 50% inhibition of bioluminescence) of 22 chemicals from the group of nonreactive organic chemicals were correlated with Pwt and molar refractivity (MR). PO,, represents, among other things, the influence r Present address: Institute
for Pesticide
Research,
Marijkeweg 17
22, 6709 PG Wageningen, 0147-6513/85
The Netherlands. $3.00
Copyright B 1985 by Academic Press. Inc All rights of reproduction tn any form reserved.
18
HERMENS ET AL.
has been used of hydrophobicity on membrane permeation. Molar refractivity successfully in enzyme-binding studies (Hansch and Leo, 1979). MR, as a measure of polarizability of molecules, is related to London dispersion forces and is usually experimentally obtained with the Lorenz-Lorentz equation (Hansch and Leo, 1979) MR=--
n2 - 1 MW n2+2 d ’
(1)
where n = index of refraction, d = density, and MW = molecular weight. Within a group of chemicals a high quality QSAR may indicate a similar mode of action. Konemann (198 1b) and Hermens et al. ( 1984a,b) demonstrated that the toxicity of mixtures of similarly acting chemicals can be predicted by concentration addition. We also determined the effect of a mixture of the tested chemicals on the inhibition of bioluminescence. The result of this experiment was evaluated with the Mixture Toxicity Index (MTI) as proposed by Konemann ( 198 1b). METHODS The Microtox test (Beckman, Model 2055) and calculation of EC& values (concentrations which reduce bacterial luminescence by 50%) were performed according to the procedures described in the Beckman Instruments Manual (1980). Five concentrations of the tested chemical (ratio between concentrations was a factor 2) were incubated with the bacteria for 15 min at 15°C. Pm values were calculated using the “hydrophobic fragmental constant” method of Rekker (1977); MR values were calculated by adding fragment values as tabulated by Hansch and Leo (1979). QSAR were calculated with a computer program based on the method of least squares. Statistical comparison of the quality of the QSAR was based on comparison of the standard deviations with an F test as described by Hansch (1973, pp. 150-158). RESULTS
AND
DISCUSSION
I. QSAR STUDIES The EC& values, together with the physicochemical properties of the tested compounds are summarized in Table 1. QSAR of all 22 compounds are presented in Table 2. Introduction of a (log P)= term (Eq. 3) does not improve the quality of the QSAR, while introduction of MR (Eq. 4) results in a significantly better correlation (P < 0.01). It is surprising that introduction of (log P)= does not result in a better QSAR, since in Fig. 1 a deviation from linearity is observed at high log P values. At low log P values, however, a nonlinear behavior is also observed. Probably these two phenomena, opposite in direction, cancel out in Eq. 3. Therefore we calculated QSAR in two log P ranges in which only one deviation from linearity is observed. Tables 3 and 5 summarize the QSAR in the log P ranges of - 1.3 to 3 and 1 to 5, respectively. Influence of (log POCJ2 1. Log P range -1.3 to 3. The positive (log P)= term in Eqs. 7 and 9 represent the nonlinear dependence of log (l/E&) from log P for the hydrophilic compounds
QUANTITATIVE
STRUCTURE-ACTIVITY
TABLE
1
ECso VALUES AND PHYSICOCHEMICALPROPERTIES
No. 1
2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18 19 20 21 22
Compound Diethyleneglycol hdethanol Acetone n-Propanol n-Butanol Diethylether Pentanol-3 Dichloromethane 1,2-dichloroethane n-Hexanol Benzene Trichloroethene n-Heptanol Toluene I, I,1 -Trichloroethane hdonochlorobenzene n-Octanol o-Xylene 1,3-Dichlorobenzene 3,4-Dichlorotoluene 1,2.3-Trichlorobenzene 1,2,3,4-Tetrachlorobenzene Lindane
19
RELATIONSHIPS
OFCOMPOUNDS
1%
1%
ECso”
P mb
MR’
5.44 6.12 5.56 5.16 4.58 4.88 4.23 4.53 4.05 2.82 3.31 3.16 1.93 2.29 1.78 2.12 1.56 1.94 1.35 0.94 1.14 0.94’ 1.58’
-1.30 -0.79 -0.30 0.27 0.80 0.88 1.21 1.51 1.76 1.86 2.13 2.20 2.39 2.59 2.75 2.81 2.92 3.09 3.53 3.98 4.20 4.94 3.53/
25.3 8.20 15.9 17.5 22.2 22.2 26.8 16.4 21.0 31.3 26.4 25.5 35.9 31.0 26.0 31.4 40.5 35.6 36.4 41.0 41.4 46.4
’ Experimentally determined concentrations at 50% inhibition
MWldd 94.8 40.5 73.6 74.8 91.5 103.8 108.2 64.0 80.1 125.6 88.9 89.7 141.4 106.3 99.6 101.8 157.5 120.6 114.1 128.2
of bioluminescence of Photobacterium
phosphoreum (rmol/liter). b P,, calcu!lated according to Rekker ( 1977). ’ MR calculated with fragment values as tabulated by Hansch and Leo (1979). d MW/d (molecular weight/density) calculated with data taken from Weast (197 I). ’ Extrapolated ECSa; no concentrations with more than 50% inhibition could be tested. ‘From Hetmens and Leeuwangh (1982).
with low log P. Equations 7 and 9 are significantly better (P -C 0.01) than the equations without a (log P)* term (Eqs. 6 and 8). Deviations from linearity at low log P values can be related with the diffusion of small hydrophilic molecules through aquous membrane pores (Seydel and Schaper, 1982). It is assumed that water soluble molecules with diameters lower than 4 X 10P” m, such as methanol, ethanol, and urea, are able to pass freely through the aquous membrane channels (O’Flaherty, 1981, p. 89). In a study of permeability through human and dog red cell membranes of a seriesof straight chain amides by Sha’afi et ~1. (197 1) a deviation from the linear relation between permeability and chain length was found for those amides with shorter chain lengths than propionamide. The PO,,of propionamide. calculated according to Rekker (1977), is -0.74. The significant improvements of the QSAR with a positive (log P)2 term in Eqs. 7 and 9 can be related to the diffusion of compounds (for instance diethyleneglycol, methanol, and acetone) through aquous membrane pores because the diameter of these straight chain hydophilic moleculesis smaller than 4 X lo-” m. A complication is that MR of diethyleneglycol, the most hydrophilic compound, is relatively high.
20
HERMENS
ET
TABLE QSAR
AND
2
CORRELATIONSBETWEENPHYSICCXHEMICAL PROPERTIES OFALL
COMP~UNDS(TABLE
1,No.
l-22)
Equation (n = 22)”
Eq. no.
2 3 4 5
AL.
log ( 1/EC,,) = 0.995 log P - 5.14 0.011 (log P)2 + 0.957 log P - 5.13 0.667 log P + 0.0636 MR - 6.30 MR = 5.15 log P + 18.2
rb
SC
0.952 0.952
0.53 0.54 0.42 5.2
0.971
0.852
’ ECSo= concentration at 50% inhibition of luminescence by P. phosphoreum; P = partition coefficient between n-octanol and water; MR = molar refractivity. b r = correlation coefficient calculated by rz = 1 - (SS,/S$) in which SS, is sum of squares of deviations about regression and SS2 is equal to the sum of squares of deviations about mean value of log ECso. c s = standard error of estimate.
Omitting this compound in the QSAR results, however, in the same significant improvements by introduction of a (log P)2 term (Eqs. 15 and 17 from Table 4). 2. Log P range I to 5. Introduction of a negative (log P)2 term (Eqs. 20 and 22, Table 5) results in significantly better correlations (P < 0.01). The coefficients of the (log P)2 term in these equations are opposite in sign to those in Tables 3 and 4, which is due to the opposite deviations from linearity. Nonlinear relationships between biological activity and hydrophobicity at high PKt are often found (Hansch, 197 1) and may be related with a reduced rate of uptake for the higher lipophilic compounds. The optimum value for hydrophobicity (log PO) in Eq. 20 is 4.6. This optimum hydrophobicity depends on, among other things, the lipid content of the membranes. The value of 4.6 for the inhibition of bioluminescence by P. phosphoreum (a gramnegative bacterium) is very near to the log PO of about 4.4 for other data with gramnegative bacteria, as summarized by Hansch and Glave (197 1).
;;~.,/,*~,“, ,,, -1
0
12
3
4
5
log %ct
FIG. 1. Relation between concentrations at 50% inhibition of bioluminescence and hydrophobicity compounds from Table I (0, oxygen compounds: 0, compounds which do not contain an oxygen X. lindane).
for atom:
QUANTITATIVE
STRUCTURE-ACTIVITY
TABLE QSAR AND CORRELATIONS IN log Eq. no.
BETWEEN
P,, RANGE
PHYSICOCHEMICAL TO 3.0 (TABLE
MR = 0.270 MW/d
PROPERTIES 1, No. l-17)
OF COMPOUNDS
sn
(n = 17)”
log ( I/EGO) = 1.06 log P - 5.21 0.273 (log P)’ + 0.537 log 0.734 log P + 0.0745 MR 0.189 (log P)2 + 0.475 log 0.228 (log P)2 + 0.467 log 0.286 (log P)’ + 0.644 log MR = 4.32 log P + 18.9
6 7 8 9 10 11 12 13
3
-1.3
Equation
21
RELATIONSHIPS
P P P P
0.929 0.972 0.97 1 0.987 0.988 0.986 0.706 0.949
- 5.45 6.61 + 0.0511 MR - 6.34 + 0.0123
MW/d
- 6.38
+ 0.593 Z - 5.99
- 1.2
u See footnotes to Table 2: MW/d = quotient molecular weight and density; for oxygen compounds and 0.0 for chemicals which do not contain an oxygen
I = indicator atom.
0.56 0.37 0.38 0.26 0.25 0.27 5.8 2.6 variable,
1 .O
The QSAR with log P alone (Eqs. 2, 6, 14, 19, 25, and 30) have rather high coefficients. This was also found by Hansch (1973, p. 131) in studies of the inhibition of bioluminescence by a series of alcohols. Hansch concluded that inhibition of bacterial luminescence appearsto involve a different kind of membrane processthan inhibition of bacterial growth, becauseQSAR with growth as parameter have slopesof about 0.7. Influence
of MR, MW/d
The improvement of the QSAR with MR (significant at at least P < 0.05) in Eqs. 8, 9, 16, 17, 2 1, and 22 is related to the relatively higher toxicity and molar refractivity of the oxygen compounds in comparison with those compounds which do not contain an oxygen atom (seealso Figs. 1 and 2). MR brings these two groups together in one equation. Becausethe correlations between log( l/E&) and log P,,, for these two groups are practically parallel (Fig. 1) with an indicator variable, these two groups can be brought together in one equation (Eqs. 11 and 23). Therefore it is interesting to see if differences in polar&ability also influence the activity within the two groups. QSAR with these two groups are given in Tables 6 and 7. Including TABLE QSAR AND CORRELATIONS IN log Eq. no.
14 15 16 17 18 L?See footnotes
BETWEEN
P,, RANGE Equation
-0.8
4
PHYSICOCHEMICAL TO 3.0 (TABLE
PROPERTIES 1, No. 2-17)
(n = 16)”
log ( l/&o) = I.22 log P - 5.54 0.219 (log P)* + 0.702 log P - 5.52 0.861 log P + 0.0595 MR - 6.46 0.209 (log P)* + 0.378 log P + 0.0567 MR - 6.40 MR = 6.11 log P + 15.4 to Table
2.
OF COMPOUNDS
ra
sa
0.954 0.971 0.97 1 0.986 0.844
0.45 0.37 0.38 0.27 4.6
HERMENS
22
ET AL.
TABLE
5
QSAR AND CORRELATIONS BETWEEN PHYSICOCHEMICAL PROPERTIES OF COMPOUNDS IN log PXt RANGE 1.O TO 5.0 (TABLE 1, No. 7-22) Eq. no.
Equation
ra
(n = 16)”
sa
log ( ~/EGO) =
19 20 21 22 23 24
log P - 5.22 (log P)' + 2.98 log P - 7.80 log P + 0.0622MR - 6.10 (log P)* + 2.49 log P + 0.0621MR - 8.55 (log P)* + 3.27 log P + 0.750 Z - 8.56 MR = 6.82 log P + 13.4 1.03 -0.322 0.582 -0.311 -0.351
0.899 0.954 0.930 0.979 0.988 0.854
0.53 0.38 0.46 0.27 0.20 4.4
‘See footnotes to Tables 2 and 3.
MR in the equations with the oxygen compounds
result in less significant (P < 0.1) improvements of the correlations (Eqs. 27 and 28) in comparison with those in Tables 3-5. Within the group of compounds which do not contain an oxygen atom, the influence of MR is not obvious. The standard deviation of Eq. 32 is the same as in Eq. 30, while including MR in the QSAR with (log P)* and log P results in a significantly lower standard deviation at P < 0.05 (Eqs. 3 1 and 33). Also, the coefficient of MR in Eq. 33 is opposite in sign to all the other equations with MR. Maybe this phenomenon is related with the colinearity between log P and MR. In comparison with the other calculated QSAR, the correlation between log P and MR for this group is very high (compare Eq. 34 with Eqs. 12, 18, 24, and 29). MR usually is higher colinear with molar volume (Hansch et al. 1977). Replacing, for instance, MR in Eq. 9 by the quotient molecular weight/density (MW/d), as a measure of molar volume, in fact results in a QSAR of similar quality (Eq. 10). As expected, the correlation between MR and MW/d for these compounds is rather high (Eq. 13). The influence of molar refractivity (MR), molar volume (MW/d), or the indicator variable (I) can be explained in several ways: The influence of molar volume on membrane permeation in our experiments is unlikely. Negative influences of molecular weight and molar volume on membrane MR
50. 0 40.
. 0
. .
30.
0
-
0
0
.*
lo
0
00
.
. 20.
00
0
.
-1
0
1
2
3
4
5
log
pact
FIG. 2. Relation between MR and log P, for compounds from Table 1. (0, oxygen compounds; 0, compounds which do not contain an oxygen atom).
QUANTITATIVE
STRUCTURE-ACTIVITY
TABLE
23
RELATIONSHIPS
6
QSAR AND CORRELATIONS BETWEEN PHYSICOCHEMICAL PROPERTIES OFOXYGEN COMPOUNDS (TABLE1, No. l-7, 10, 13, 17) Eq. no.
Equation (n = 10)’
ra
sn
25 26 27 28 29
log ( 1/EC,,) = 1.09 log P - 5.09 0.265 (log P)* + 0.664 log P - 5.37 0.609 log P + 0.0827 MR - 6.75 0.165 (log P)’ + 0.544 log P + 0.0483 MR - 6.23 MR = 5.82 log P + 20.0
0.938 0.982 0.98 1 0.990 0.823
0.58 0.34 0.35 0.27 5.8
‘See footnotes to Table 2
permeability have been observed by Lien (1975) and Sha’afi et al. (197 l), while in the calcula.ted QSAR the influence of molar volume is positive. According to Hansch and Leo (1979) positive coefficients with MR suggesta binding action via dispersion forces, possibly in the more polar regions of the receptor molecule. A4R is used successfully in correlation studies involving enzyme binding (Hansch and Leo, 1979). The light-emitting reaction in bioluminescence is catalyzed lby the enzyme luciferase (Hastings and Nealson, 1977). Therefore the influence of MR may be related with the inhibition of this enzyme or of the reduced Flavin mononucleotide (FMNH,)-donating system. The higher activity of the oxygen compounds and therefore the influence of the indicator variable can be explained with the possibility of these compounds to form hydriogen bonds with the receptor molecule. According to Franks and Lieb (1982) and Sandorfy (1978) an aspecific effect, such as anethesia, may also depend on a polar and molar volume factor as well as on hydrophobicity. It cannot be indicated which explanation is the right one or if other mechanisms are involved. This indistinctness is partly caused by the high colinearity between the paramleters for the tested compounds. If more data on the toxicity of other compounds in the Microtox test become available, probably more unambiguous conclusions can be drawn. It is clear however that in this test hydrophobicity is not the only parameter which influences the activity of the compounds. TABLE QSAR A:VD CORRELATIONS WHICH
Do
NOT
CONTAIN
BETWEEN AN OXYGEN
PHYSICOCHEMICAL ATOM (TABLE
PROPERTIES 1,
OF COMPOUNDS
No. 8. 9, 1 I. 12. 14- 16. 18-22)
Equation (n = 12)”
Eq. no. 30 31 32 33 34
7
ra
s*
0.924 0.99 1 0.930 0.995 0.973
0.49 0.18 0.49 0.14 2.2
1% ( 1/EGO) =
1.08 log P - 5.49 -0.411 (log P)* + 3.69 log P - 9.21 0.559 log P + 0.062 MR - 5.91 -0.469 (log P)’ + 4.55 log P - 0.059 MR - 9.34 MR = 8.39 log P + 6.7
a See footnotes to Table 2.
24
HERMENS ET AL.
II. MIXTURE
TOXICITY STUDY AND AN EVALUATION OF THE MICROTOX
TEST
Exposure of the bacteria to an equitoxic mixture (concentrations in identical fractions of the ECSo) of 21 of the tested compounds (without diethyleneglycol) resulted in a C (c/EC&, at 50% inhibition of bioluminescence, of 2.0. This corresponds with a Mixture Toxicity Index of 0.77 + 0.08 (calculated according to Kijnemann (198 1b) and assuming a standard error in log ECsO of 0.10). A MT1 of 1.0 corresponds with concentration addition and is expected for mixtures of chemicals with “simple similar actions” (Kiinemann, 198 lb). The high quality QSAR, which could be calculated, suggest that the chemicals act similarly. Slight deviations from concentration addition were observed earlier in other sublethal toxicity studies with D. magna (Hermens et al. 1984a). The sensitivity of the Microtox test to the tested compounds is comparable to a 1Cday acute mortality test with guppies, as can be concluded by comparing the QSAR intercepts. The intercept value in Eq. 2, which amounts to 5.14, does not deviate much from the intercept of 4.87 of a QSAR with guppies calculated with the same group of chemicals by Kijnemann (198 1a). The Microtox test, however, is not sensitive to very lipophilic compounds. A deviation from the linear relationship log ECso vs log P,,,t occurs at log Pact of about 3.0 (Fig. l), while in the 1Cday test with guppies no deviation from linearity was observed at log Pmt < 5.7 (Kiinemann, 1981a). As suggested by Kanemann, this QSAR probably predict a “minimum toxicity” of hydrophobic nonionized organic chemicals. This minimum action can be masked by more specific modes of action. Lindane, for instance, is about 300 times more toxic in the 1Cday LC 50 experiment with guppies (Hermens and Leeuwangh, 1982). In the Microtox test, however, only the aspecific minimum toxicity of lindane is observed (see Fig. 1). It is very likely that, with other compounds which act more specifically in toxicity tests with higher organisms, only the aspecific “minimum toxicity” will appear in the Microtox test. Further, because only a slight deviation from concentration addition was found in the mixture toxicity study, this Microtox test can give a good estimate of the total aspecific “minimum toxicity” of, for instance, an effluent. When such an effluent contains very lipophilic compounds or pollutants with more specific modes of action, the Microtox test will underestimate the toxicity to other aquatic life. ACKNOWLEDGMENTS This work was supported by the Ministry of Housing, Physical Planning, and Environment (The Netherlands). We thank Professor Dr. H. van Genderen and Dr. H. KBnemann for their stimulating discussion and comments on the manuscript.
REFERENCES BECKMAN Interim FRANKS,
INSTRUMENTS INC. (1980). Operating Instructions, Microtox (Model 2055) Toxicity Manual 110679B-9-80. N. P., AND LIEB, W. R. (1982). Molecular mechanisms of general anaesthesia. Nature
Analyzer. (London)
300,487-493. C., AND ANDERSON, S. (1967). The structure-activity relationship in barbiturates and in similarity to that in other narcotics. J. Med. Chem. 10, 745-753. HANSCH, C. (197 1). Quantitative structure-activity relationships in drug design. In Drug Design (E. J. Ari&ns, ed.), Vol. I, pp. 271-342. Academic Press, New York. HANSCH, C., AND GLAVE, W. R. (197 1). Structure-activity relationships in membrane perturbing agents. Mol. Pharmacol. 7. 337-354. HANSCH,
QUANTITATIVE
STRUCTURE-ACTIVITY
RELATIONSHIPS
25
HANSCH, C. (1973). Quantitative approaches to pharmacological structure-activity relationships. In Structure-Activity Relationships (C. J. Cavallito, ed.), Vol. I, Pergamon Press, Oxford/New York. HANSCH, C., SMITH, R. N., ROCKOFF, A., CALEF, D. F., Jow, P. Y. C., AND FUKUNAGA, J. Y. (1977). Structure-activity relationships in papain and bromelain hgand interactions. Arch. Biochem. Biophys. 183, 383-392.
HANSCH, C.. AND LEO, A. J. (1979). Sobstituent Constants for Correlation Analysis in Chemistry and Biology. Wiley, New York. HASTINGS, J. W., AND NEALSON, K. H. (1977). Bacterial luminescence. Annu. Rev. Microbial. 31, 549595.
HERMENS, J., AND LEEUWANGH, P. (1982). Joint toxicity of mixtures of 8 and 24 chemicals to the guppy (Poecilia reticulata). Ecotoxicol. Environ. SaJ 6. 302-3 10. HERMENS, J.. CANTON, H., JANSSEN. P.. AND DE JONG, R. (1984a). Quantitative structure-activity relationships and mixture toxicity studies of chemicals with anaesthetic potency: Acute lethal and sublethal toxicity to Daphnia magna. Aquat. Toxicol. 5, 143-154. HERMENS. J., LEEUWANGH, P., AND MUS~H, A. (1984b). Quantitative structure-activity relationships and mixture toxicity studies of chloro- and alkylanilines at acute lethal toxicity level to the guppy (Poecilia reticulata). Ecoio-yicoi. Environ. Suf 8. 388-394. KONEMANN, H. (198 la). Quantitative structure-activity relationships in fish toxicity studies. 1. Relationship for 50 industrial pollutants. To,yicology 19. 209-22 I. KONEMANN, H., (198 lb). Fish toxicity tests with mixtures of more than two chemicals: A proposal for a quantitative approach and experimental results. Toxicology 19, 229-238. KBNEMANN, H., AND MUSCH, A. (1981). Q uantitative structure-activity relationships in fish toxicity studies. 2. ‘The influence of pH on the QSAR of chlorophenols. Toxicology 19, 223-228. LIEN, E. J. (1975). Structure-absorption-distribution relationships: Significance for drug design. In Drug Design (E. J. Ariens. ed.), Vol. 5, pp. 81-131. Academic Press, New York. O’FLAHERTY. E. J. (198 1). To.xicants und Drugs: Kinetics and Dynamzcs. Wiley, New York. REKKER, R. IF.(1977). The Hydrophobic Fragmental Constant. Elsevier, Amsterdam. SANDORFY,C. (1978). Intermolecular interactions and anaesthesia. Anesthesiology 48, 357-359. SEYDEL, J. K.. AND SCHAPER. K. J. (1982). Quantitative structure-pharmacokinetic relationships and drug design. Pharmucol. Ther. 15. I3 I- 182. SHA’AFI. R. I.. GARY-B• BO. C. M., AND SOLOMON. A. K. (1971). Permeability of red cell membranes to small hydrophylic and lipophilic solutes. J. Gen. P!z~~siol.58, 238-258. SLOOFF, W.. CANTON, J. H.. AND HERMENS. J. (1983). Comparison of the susceptibility of 22 fresh water species to I5 chemical compounds. I. (Sub)acute toxicity tests. Aquat. To.~icol. 4, 113-128. VEITH, G. D.. AND KONASEWICH, D. E. (eds.) (I 975). Structure-activity Correlations in Studies of Tosicity and Bioconcentration with Aquutic Species. Great Lakes Advisory Board, Windsor, Ontario. WEAST. R. C. (ed.) (197 1). Handhook of Chernistr.v und Phy.sic.s.CRC Press. Cleveland.