The role of cyclic gmp in regulating myosin during chemotaxis of Dictyostelium: evidence from a mutant lacking the normal cyclic gmp response to cyclic amp

During chemotaxis of the cellular slime mould Dic -tyostelium discoideum, binding of the chemoattractant cyclic AMP to the cell surface receptors induces a number of intracellular responses, including an increase in the inos-itol phosphates (Europe-Finner and Newell, 1987, 1989; Van Haastert et al., 1989), cyclic GMP (Mato et al., 1977; Wurster et al., 1977), cyclic AMP (Shaffer, 1975), influx of calcium (Bumann et al., 1984; Abe et al., 1988), phos-phorylation of myosin II (Berlot et al., 1985), and association of actin and myosin II with the cytoskeleton (McRob-bie and Newell, 1983; Hall et al., 1988; Liu and Newell, 1988).

Mutants that are defective in specific parts of the signal transduction pathway are very useful in elucidating the roles of various components of the signal transduction system. Streamer F mutants are defective in the structural gene for cyclic GMP-specific phosphodiesterase and show a pro-longed cyclic AMP-induced cyclic GMP accumulation that correlates with a prolonged period of elongation during cell chemotactic movement (Ross and Newell, 1981; Van Haastert, et al., 1982; Coukell and Cameron, 1986). A study of the cytoskeleton in these mutants revealed that the asso-ciation of the myosin II heavy chain (MHC) with the Triton X-100-insoluble cytoskeleton was also prolonged (Liu and Newell, 1988). Further investigation showed that the phos-phorylation of MHC, normally peaking at about 30 seconds after cyclic AMP stimulation in the wild-type strains, was delayed in these mutants (Liu and Newell, 1991). These studies with the streamer mutants have suggested that cyclic GMP plays an important role in the regulation of myosin in Dictyostelium.

Studies on mutants lacking normal myosin II (myosin null mutants) have indicated that MHC is not essential for chemotaxis (although it is required for other cellular func-tions such as cytokinesis; Knecht and Loomis, 1987; De Lozanne and Spudich, 1987; Wessels et al., 1988). How-ever, these mutants have a more rounded shape and are less efficient at chemotactic movement, suggesting that a role of MHC is in the regulation of cell polarity and hence effi-cient directional movement needed for chemotaxis. A ques-tion that could be asked is: if cyclic GMP regulates the function of MHC, what would be the effect on myosin of a mutant that was defective in its cyclic GMP response? Would such a mutant show some similarity to the MHC null mutant such as having a more rounded shape and dif-ficulty in cell division? Would the events downstream of cyclic GMP such as MHC phosphorylation and MHC asso-ciation with the cytoskeleton be greatly altered due to the lack of cyclic GMP response?

In this report, we present results that answer some of these questions. We have studied mutant KI-10, which lacks the normal cyclic GMP accumulation in response to stim-ulation with cyclic AMP. It is a mutant selected by its inability to move chemotactically towards folate and cyclic AMP, its phenotype being due to a dominant mutation on linkage group I (Kuwayama et al., 1993). It shows a normal cyclic AMP-induced InsP₃ response but cannot be stimu-lated to produce more than the basal level of cyclic GMP, although in vitro tests show that its guanylate cyclase activity and the activity of cyclic GMP phosphodiesterase are essentially normal. These data imply that mutant KI-10 has a defect between the cell surface cyclic AMP receptors and guanylate cyclase that prevents the normal cyclic AMP stimulation of this enzyme.

Dictyostelium discoideum parental strain XP55 (Ross and Newell, 1981) and mutant KI-10 derived from XP55 (Kuwayama et al., 1993) were grown in association with Klebsiella aerogenes, strain OXF1, on SM nutrient agar (Sussman, 1966). The harvesting and ³²P-labelling of amoebae were carried out according to the method of Liu and Newell (1991). Unless stated elsewhere, all the cells were subjected to 8 hours development before they were stimu-lated with cyclic AMP.

The cyclic GMP assays were carried out according to the method of Van Haastert et al. (1981) using a radioimmunoassay kit sup-plied by Amersham International PLC.

Quantification of whole cell protein was carried out using the method of Peterson (1977), a modification of the method of Lowry et al. (1951). The isolation of Triton X-100-insoluble cytoskeletons, the purification of monoclonal anti-MHC antibody, immuno-precipitation of MHC, electrophoresis and densitometry were per-formed as described by Liu and Newell (1991).

In order to assess the effect of the mutation in KI-10 on the cytoskeletal actin response, developing cells were stim-ulated with 100 nM cyclic AMP and the actin present in the cytoskeleton was assayed. It was found that this response was seen in the mutant with similar timing to that seen for the parental strain published previously (Liu and Newell, 1988), (Fig. 3). At the dose of 100 nM cyclic AMP used in these experiments, however, the response seen at 5 seconds was significantly lower than the parental controls (the prestimulus values also being significantly lower as indicated in the legend to Fig. 3). The dose-response curve of cyclic AMP-dependent actin association with the cytoskeleton in the parental strain XP55 gave a similar pat-tern (Fig. 4) to that in wild-type strain NC4 as previously reported by McRobbie and Newell (1983), in which the level of the cytoskeletal actin was induced to the maximum by a dose of 10⁻⁷ M cyclic AMP, then slightly decreased with a further increment in cyclic AMP.

In the mutant KI-10, a similar dose-response curve was seen but the dose of cyclic AMP for maximal response was shifted from 10⁻⁷ M to 10⁻⁶ M (Fig. 4). Whether such a shift is due to the altered association of MHC with the cytoskeleton (which may facilitate the association of actin with the cytoskeleton by actin-myosin interaction under normal conditions) is unclear.

In contrast to the actin response, the cytoskeletal MHC response to 100 nM cyclic AMP is abolished in this mutant (Fig. 5) compared to that of XP55 (Liu and Newell, 1988). The dose-response curve of cyclic AMP-dependent MHC association with the cytoskeleton in the strain XP55 is shown in Fig. 6. The maximal response in the cytoskeletal MHC was induced by 10⁻⁷ M cyclic AMP and was main-tained with a further increase in cyclic AMP dose. In the mutant there was no significant response even up to mM cyclic AMP.

Results from the study of Streamer F mutants of Dic -tyostelium discoideum suggested that cyclic GMP may regulate MHC association with the cytoskeleton by inhibiting its phosphorylation (Liu and Newell, 1991). KI-10 mutant provides a means of testing this hypothesis. After stimu-lating ³²P-labelled developing cells of KI-10 with 200 nM cyclic AMP, myosin II was immunoprecipitated using an anti-MHC monoclonal antibody and analyzed by autoradiography. The results clearly demonstrated that the response in MHC phosphorylation seen in XP55 was abol-ished (Fig. 7).

During experiments with gel electrophoresis it was noticed that more MHC was isolated from cells of KI-10 than from the same number of XP55 cells. It was also observed by microscopic observation that amoebae of mutant KI-10 were somewhat larger than those of XP55 (although for amoebae such observations are subjective and not easily quantified). However, assay of whole cell protein in these two strains indicated that mutant KI-10 did indeed contain about 40% more protein than the parental strain XP55 (Fig. 8). When equal amounts of whole cell protein from the mutant and parental strains were compared (rather than the same number of cells), equal amounts of MHC were immunoprecipitated from both strains, indicating that myosin content was increased proportionally in the KI-10 mutant.

One way in which size could be increased in the mutant would be via an inhibitory effect on cell doubling. When cell growth was measured, the mutant was found to have a markedly longer doubling time compared to XP55 (Table 1).

The data shown here provide further evidence in support of the hypothesis that cyclic GMP regulates the association of MHC with the cytoskeleton and that this regulation involves MHC phosphorylation. In the stmF mutant, a pro-longed cyclic GMP accumulation in response to cyclic AMP is followed by prolonged inhibition of MHC phos-phorylation, prolonged cytoskeletal MHC association and prolonged cell elongation. In the KI-10 mutant, which shows no cyclic AMP-stimulatable cyclic GMP accumula-tion, the cytoskeletal MHC accumulation and MHC phos-phorylation responses were absent, supporting the proposed link between these events. The actin response, however, was still present in the mutant although maximal only at higher stimulus concentrations compared to the parental strain, showing that the lack of cyclic GMP does not greatly affect this aspect of chemotaxis. The reason for the shift in the dose-response curve to higher cyclic AMP concentra-tions is not understood but may be an indirect effect of the changes in myosin distribution in the cell.

The study of a single mutant such as KI-10 leaves open the possibility of two independent mutations being coinci-dentally present in the mutant that, together, lead to the observed effects. This possibility has recently been rendered unlikely by results obtained with another independently derived mutant (SA219) in a different genetic background. This mutant was found to be similar to KI-10 in having lost its cyclic GMP response to cyclic AMP. Like KI-10, it was also found to show an actin response that was similar to wild type and, like KI-10, it lacked the cytoskeletal MHC response and phosphorylation of myosin II in response to cyclic AMP (Liu and Newell, unpublished data).

An interesting feature of mutant KI-10 is that it has a higher protein content per cell than the parental strain XP55. It has been shown by studying MHC null strains formed by homologous recombination or antisense mRNA (De Lozanne and Spudich, 1987; Knecht and Loomis, 1987; Manstein et al., 1989) that MHC is necessary for cell divi-sion. These MHC null cells could not undergo normal cytokinesis and formed large syncytia in axenic suspension culture. Such cells could, however, divide inefficiently on solid substrata by parts of the syncytia moving apart. In the case of mutant KI-10, although MHC is present, its regu-lation by cyclic GMP is defective. If cyclic GMP is a more general regulator of myosin besides its effects on myosin during chemotaxis, the amoebae may be less efficient in regulating the formation of the contractile ring during cell division. As a consequence the amoebae would take longer to complete cell division, and (because protein synthesis would continue) would be larger than the wild-type cells, producing the observed phenotype. It is of interest in this respect that another mutant (KI-8) that is defective in cyclic GMP formation (and which lacks any measurable basal unstimulated level of this nucleotide) also shows this enlarged size and slower rate of cell doubling (H. Kuwayama, S. Ishida and P.J.M. Van Haastert, unpub-lished). The effect on cell division is not sufficiently severe in KI-10 and KI-8, however, to produce multinucleate cells as seen with the myosin-null strains.

We thank Julian Gross and Peter van Haastert for helpful dis-cussions, Ken Johnson for drawing the figures and the SERC for financial support.