Escherichia coli treated for 1 h with 100 microM rac-3,4-dihydroxybutyl 1-phosphonate (DBP), a glycerol-3-phosphate analog, die when sorted at 5 degrees C, whereas the viability of untreated cells is relatively unaffected. This observation formed the basis of a selection procedure that was used to isolate mutants that are partially resistant to DBP. One such mutant, strain 6204, is constitutive for DBP transport, exhibits a particularly high degree of cold resistance, has the same doubling time as the parent, and is similar to the parent strain in terms of incorporation of DBP into the lipid fraction. Glycerol-3-phosphate and phosphatidylglycerol phosphate synthetases obtained from strain 6204 and its parent were identical in terms of DBP recognition. The parent strain is killed when incubated in the presence of a combination of 70 microM rac-DBP and 0.25% deoxycholate, whereas strain 6204 continues to grow, albeit more slowly, in the presence of this combination. Strain 6204 can be distinguished from the parent strain on agar plates (low phosphate minimal medium with glucuronate as the sole carbon source) containing 15 microM rac-DBP. The insertion of Tn10 near the 6204 mutation has facilitated genetic manipulations. All phenotypic effects attributed to strain 6204 appear to be due to a single mutation. Genetic analysis indicates that Tn10, inserted near the gene responsible for DBP resistance, maps in the vicinity of 27 min. Three-factor crosses reveal a gene order of hemA-Dbpr-Tn10(zch)-trp. The only gene for phosphoglyceride metabolism known to map in this region is the gene associated with cardiolipin synthetase, cls. Genetic results suggest that the mutation responsible for DBP resistance maps in or very near cls. Analysis of the lipids isolated from untreated strain 6204 (and from each of the transductants prepared by P1 vir-mediated transfer of DBP resistance of wild-type strains) reveals that cardiolipin synthesis is defective. These results strongly suggest that the mutation responsible for DBP resistance has its primary effect on cardiolipin synthesis. To further test this hypothesis, strains with an authentic cls mutation were constructed and examined for resistance to DBP. These strains had growth properties that were identical with those of strain 6204. Wild-type strains and mutants defective in cardiolipin synthesis were treated with DBP and 20 mM magnesium or calcium chloride. Simultaneous treatment of either cell type with DBP and divalent cation not only failed to stimulate growth but, quite the contrary, had a marked synergistic growth inhibitory effect.