Training-set of 19 compounds - 11 haloalkanes and 8 haloalkenes - was
selected from a group of 58 halogenated aliphatic hydrocarbons using s
tatistical experimental design. Strictly defined method was used for p
reparation of water solutions and toxicity measurements of volatile an
d poorly water-soluble halogenated aliphatic substances. The acute tox
icity expressed as the effective concentrations (EC50) was determined
for the compounds in the training set using the Microtox test. The qua
ntitative structure-activity relationships (QSAR) models for haloaliph
atic compounds were constructed using the Projection ot latent structu
res method. Size of the molecules was the most important parameter for
toxicity of saturated haloaliphatic compounds. This characteristics c
an be related to accumulation of the haloalkanes in biological membran
es or binding to biomacromolecules. Toxicity of 2-chlorobutane was sig
nificantly higher than expected from its size. This compound, as the o
nly representative of beta-substituted chloroderivatives in the data s
et, has probably different mode of action than terminally substituted
compounds. Three unsaturated compounds - cis-1,2-dichloroethylene, tra
ns-1,4-dichlorobutene, and cis-1,4-dichlorobutene - displayed similar
mode of action to that observed for haloalkanes, while another two hal
oalkenes 3-chloropropylene, and 2,3-dibromopropylene displayed differe
nt - reactive - type of toxicity. The steric parameters had to be comp
lemented by four electronic descriptors for explanation of their high
toxicity. (C) 1998 Elsevier Science Ltd.