Dynamics of intracellular cGMP during chemotaxis in Dictyostelium cells

Cyclic guanosine 3′,5′-monophosphate (cGMP) is a ubiquitous second messenger involved in various physiological functions, such as phototransduction in the eye and smooth muscle relaxation. cGMP mainly regulates enzyme and channel activity through cGMP-dependent kinases or cGMP-binding proteins (Beavo and Brunton, 2002; Feil et al., 2022; Hofmann, 2020; Lucas et al., 2000; Tsai and Kass, 2009).

cGMP is also an important second messenger in Dictyostelium cells, a model organism for studying cell migration, chemotaxis and development. Vegetative Dictyostelium cells show chemotaxis toward folic acid. After starvation for several hours, they begin to aggregate by chemotaxis to form a multicellular structure. In the aggregation stage, cyclic adenosine 3′,5′-monophosphate (cAMP) secreted from cells is used as a chemoattractant, which acts as a diffusible chemical guidance cue for aggregation. cAMP secreted from the aggregation center induces neighboring cells to secrete cAMP, which causes outward relay and propagation of cAMP signals as concentric or spiral traveling waves (Devreotes, 1989; Tomchik and Devreotes, 1981). The cells move faster inward with an elongated cell shape during the rise of the wave, and slow down or stop moving during the fall of the wave (Varnum et al., 1985). Dark-field microscopy clearly shows propagating waves with different light-scattering patterns owing to the differences in the cell shapes between the fast-migrating and stopping phases (Alcantara and Monk, 1974; Rietdorf et al., 1998).

The binding of cAMP to the trimeric G-protein-coupled membrane receptor leads to the activation of intracellular signaling pathways, involving adenylyl cyclase, guanylyl cyclase (GC), and Ca²⁺ channels (Chung et al., 2001; van Haastert and Kuwayama, 1997; Yumura et al., 1996), and pathways mediated by phosphoinositide 3-kinase, phospholipase A2 and TorC2 (van Haastert et al., 2007; Yan and Jin, 2012). When aggregation-competent cells are stimulated with external cAMP, the intracellular cGMP (cGMPi) levels transiently increase, with a peak at 10 s, and return to basal levels 30 s after stimulation (Mato et al., 1977a). Additionally, intracellular cAMP (cAMPi) and Ca²⁺ (Cai²⁺) levels also transiently increase upon stimulation (Bumann et al., 1984; Gerisch and Wick, 1975; Wick et al., 1978; Yumura et al., 1996). cGMPi concentration is strictly regulated via production, by two types of GC, soluble GC (sGCA) and membrane-bound GC (GCA), and degradation, mainly by cGMP-stimulated phosphodiesterase PdeD (also known as GbpA) in a similar way to what occurs in other organisms (Ross and Newell, 1981; Veltman et al., 2005). After the activation of the trimeric G-protein RasG, a small G-protein directly activates sGC to produce cGMP (Van Haastert et al., 2021).

The chemically mutagenized cell line KI8, which exhibits only a subtle activity of both GCA and sGCA, does not show chemotaxis (Kuwayama et al., 1993). GCA/sGCA double-null cells aggregate slowly and show a strongly reduced chemotactic index (Bosgraaf et al., 2002a). In contrast, the streamer F mutants, which are also generated by chemical mutagenesis, show a prolonged increase in cGMPi levels and an extended fast-migrating phase during chemotaxis (Ross and Newell, 1981). PdeD-null cells, which have high levels of cGMPi, show a highly polarized morphology and chemotaxis index (Bosgraaf et al., 2002a). Mutant cells deficient in the cGMP-binding protein GbpC also show a strongly reduced chemotactic index and delayed cell aggregation. Thus, the cGMP pathway is considered to play an important role in chemotaxis. Recently, a good review on the history of cGMP research in Dictyostelium has been published (Van Haastert et al., 2021).

When Dictyostelium cells are stimulated with chemoattractants, such as folic acid and cAMP, cytosolic actin and myosin II are transiently translocated to the cell cortex (Yumura and Fukui, 1985; Yumura, 1993, 1994). Biochemical analysis has also shown that actin and myosin II levels increased in Triton X-100 insoluble cytoskeletons after stimulation. Mutant cells deficient in GC have a diminished increase in myosin II levels, and mutant cells deficient in cGMP-stimulated phosphodiesterases show a much longer association of myosin II with the cortex (Bosgraaf et al., 2002a; Liu and Newell, 1988, 1991). A high concentration of external cGMP, a small fraction of which enters the cells, can induce similar dynamics of actin and myosin II to external folic acid as a chemoattractant in vegetative cells (Yumura, 1994). GbpC null cells do not show translocation of myosin II to the cell cortex upon stimulation (Bosgraaf et al., 2002a). A recent research project using caged cGMP directly showed that an increase in cGMPi levels induces the translocation of myosin II to the cortex (Pfannes et al., 2013). However, the precise spatiotemporal regulatory mechanism of cGMPi and its relevance to myosin II regulation remain unknown.

In this study, we aimed to visualize the dynamics of cGMPi during chemotaxis in living Dictyostelium cells. To this end, Green cGull was used as a genetically encoded fluorescent probe for cGMPi (Matsuda et al., 2017). Using this probe, we successfully observed cGMPi propagating in waves and found that the phases of cGMPi, cAMPi and Cai²⁺ waves were synchronized. These waves preceded the phases of the oscillations of actomyosin and cell velocity. In addition, inducing an acute increase in cGMPi concentration using caged cGMP caused a transient shuttling of myosin II between the cytosol and cell cortex, suggesting a direct link between cGMP signaling and myosin II dynamics.

When cAMP, a chemoattractant for Dictyostelium, is externally applied to aggregation-competent cells, the level of intracellular cGMP (cGMPi) begins to increase at 0.8 s after cAMP application with a peak time of 10 s and returns to the resting level at 20–30 s (Bosgraaf et al., 2002b; Mato et al., 1977a; Valkema and Van Haastert, 1992; Wurster et al., 1977). However, given that these measurements of cGMPi were based on lysed cells and isotope dilution and immunoassays, its spatiotemporal dynamics were unknown.

To visualize the dynamics of cGMPi in live Dictyostelium cells, we used Green cGull, a cGMP fluorescent probe composed of the cGMP-binding domain of cGMP-specific mouse phosphodiesterase 5α inserted into the yellow fluorescent protein Citrine. Upon binding to cGMP, conformational changes in its cGMP-binding domain increase the fluorescence intensity (Matsuda et al., 2017). We created cells expressing codon-optimized Green cGull for Dictyostelium (D-Green cGull) under the control of actin15 promoter (Fig. 1A). The cells stably expressing this probe showed no defects in cell migration or developmental progression (Tanaka et al., 2019).

First, we tested the characteristics of the probe in vitro. Fig. 1B shows the calibration curves depending on the concentration of exogenous cGMP and cAMP obtained by measuring the fluorescence intensities of the lysate of aggregation-competent cells expressing D-Green cGull. This probe had an increased fluorescence level depending on the increase in cGMP levels, but not cAMP levels. The range for the measurement by Green cGull was between 10⁻⁷ and 10⁻³ M cGMP. Next, we measured the fluorescence intensities of a suspension of living cells over time after cAMP exposure (Fig. 1C). To prevent cAMP relay between the cells, 5 mM caffeine was added. Fluorescence began to increase 6.1±0.7 s after cAMP application, peaked at 24.1±2.3 s, and returned to the resting level at 98.2±10.3 s (mean±s.d., n=11). The responses persisted for a longer period compared to what was found in the previous studies mentioned above. Fig. 1D shows the correlation between the applied cAMP concentration and the fluorescence intensity of D-Green cGull at the peak. Half-maximal cGMPi responses occurred between 10⁻⁹ and 10⁻⁸ M cAMP. This concentration corresponds closely to that which causes half-maximal occupancy of cAMP receptors (Janssens and Van Haastert, 1987).

In Dictyostelium cells, cGMP is synthesized by two types of GC, membrane-bound (GCA) and soluble (sGCA) (Roelofs et al., 2001). The latter predominantly synthesizes cGMP. GCA/sGCA double-knockout cells show significantly reduced cGMP production (Roelofs and Van Haastert, 2002). As expected, the mutants expressing D-Green cGull did not show any increase in fluorescence upon cAMP stimulation (blue in Fig. 1D). Therefore, D-Green cGull is a reliable probe for the measurement of cGMPi levels.

Next, we observed individual cells expressing D-Green cGull under confocal fluorescence microscopy. The fluorescence of D-Green cGull was evenly distributed in the cells, suggesting that this probe exists diffusely in the cytosol (Fig. 1E). Fig. 1F shows the typical time course of fluorescence intensity when stimulated with cAMP, related to Fig. 1E. Fluorescence transiently increased, similar to what was seen in the suspended cells, as measured by the fluorometer. We did not observe any local increase in a specific subcellular region, even with high magnification and rapid image acquisition.

Fig. 2A,B shows the typical time course of fluorescence images of aggregating cells expressing D-Green cGull. Fluorescence did not show any fluctuations before aggregation began. Thereafter, the cells became highly mobile and showed chemotaxis toward the aggregation center by forming cell streams. Fluorescence waves of D-Green cGull propagated in the distal direction from the aggregation center (Fig. 2A; Movie 1). In the late aggregation stage, cells turned around the center, and fluorescence propagated outward in a spiral or concentric pattern, turning around the center in the opposite direction (Fig. 2B; Movie 2). It is plausible that these waves represent the cell responses following the propagation of extracellular cAMP signaling waves. Fig. 2C shows a typical time course of the fluorescence intensity in a small area (containing 10–20 cells), indicating that the responses of cGMPi are propagated in a wave-like manner.

Previous reports have visualized the oscillatory propagations in cAMPi and Cai²⁺ (Hashimura et al., 2019; Horikawa et al., 2010). To compare the dynamics of cGMPi with those of these second messengers, cells expressing Flamindo2 (cAMPi probe) or D-GCaMP6s (Cai²⁺ probe) were observed. Fig. 2D,E (also see Movies 3 and 4) shows the typical time course of the fluorescence intensity of individual probes, confirming that the responses of cAMPi and Cai²⁺ are propagated in a wave-like manner. Because the fluorescence of Flamindo2 decreased at high cAMP concentrations, the plot was inverted.

cGMP has been reported to regulate the actin cytoskeleton as well as myosin II (Liu and Newell, 1988; Veltman and Van Haastert, 2006). It has been reported that actin shows propagating waves in the aggregation stream (Vicker and Grutsch, 2008). Although the propagation of myosin II has not been observed in the aggregation stream, phosphorylation of myosin II and its association with the cytoskeleton in response to the external cAMP have been reported (Berlot et al., 1985; McRobbie and Newell, 1983). We examined the dynamics of actin and myosin II in cells expressing GFP–lifeact, an F-actin marker, and GFP–myosin II under confocal microscopy. The propagation became much clearer upon frame subtraction, showing only the difference in each frame (Movies 5 and 6). Fig. 2F,G shows typical time courses of the relative fluorescence intensities of individual probes. The fluorescence intensities of actin and myosin II also propagated in a wave-like manner. Their increases reflect the assembly of actin and myosin II filaments at the cortex as shown below.

Fig. 2H shows a summary of the periodic durations of individual probes, indicating that the responses of these signals and cytoskeletal proteins are propagated with a similar periodicity (221.8±37.3 s, mean±s.d., n>15 each).

In the propagating wave of excreted cAMP from the aggregation center, cells respond to rising gradients of cAMP with stimulated motility, and to falling gradients with depressed motility (Varnum et al., 1985). It would be expected that the second messengers and cytoskeletal proteins showing propagating properties are correlated with cell migration.

We examined the correlation between the migration velocity of individual cells and the waves of signals and cytoskeletal proteins. In the fluorescence measurements in Fig. 2C, we did not provide any information about the individual cells because the cells crossed within the area of fixed position. Thus, the fluorescence intensities and cell velocities were simultaneously measured over time by tracking single cells. Fig. 3A,C,E,G,I shows the correlation between cell velocity (blue) and the fluorescence intensities of five probes (cGMP, cAMP, Ca²⁺, F-actin and myosin II; red). These are synchronized in almost the same phase, although Cai²⁺ increased more steeply and cell velocity remained high even after Cai²⁺ returned to basal levels. Fig. 3B,D,F,H,J shows results of cross-correlation of the curves shown in Fig. 3A,C,E,G,I, respectively. Fig. 3K shows phase differences between the periodic waves of individual probes and cell velocities, based on the cross-correlation analysis (the latter is set to 0). The peaks of F-actin and myosin II waves were almost consistent with those of cell velocities, indicating that F-actin and myosin II increase in a synchronized manner with cell velocity. On the other hand, the peaks of cGMPi, cAMPi, and Cai²⁺ waves (−14.3±5.3 s, −24.3±5.3 s and −22.9±4.9 s, mean±s.d., n>7 each) preceded those of F-actin, myosin II, and cell velocities, indicating that they might be upstream signals for the chemotactic movements.

F-actin is mainly localized in the cell cortex, including pseudopods. Fig. 3L shows fluorescence images of cells expressing GFP–lifeact at the peak and trough of cell velocity under confocal fluorescence microscopy (optical thickness, 1 µm). Fig. 3N shows fluorescence intensities in the cell cortex at the peak and trough of the cell velocity, indicating that F-actin levels increase in the cell cortex at the peak (n=15 for each). Myosin II is mainly localized in the posterior cortex. Fig. 3M shows fluorescence images of cells expressing GFP–myosin II at the peak and trough of cell velocity. Fig. 3O shows fluorescence intensities in the cell cortex at the peak and trough of the cell velocity, indicating that myosin II levels increase in the cell cortex at the peak (n=15 for each).

A previous study using a photoactivatable caged version of cGMP (DMACM-caged 8-Br-cGMP) showed that myosin II is transiently translocated to the cell cortex after uncaging by UV illumination (Pfannes et al., 2013). However, it is unclear whether UV illumination actually increases cGMPi levels, because cGMPi was not measured.

We applied permeable caged cGMP to the cells expressing D-Green cGull. After 1 h of incubation, a UV laser (405 nm) was locally illuminated (2-µm diameter circle) in the cells to release cGMP. Fig. 4A shows the typical time course of the fluorescence intensity of D-Green cGull with repeated uncaging. At every UV illumination (arrows), the cGMPi level increased immediately after uncaging, but after reaching the maximum level of 1.1 (relative value against the resting level), it autonomously decreased to the resting level.

Next, we applied the caged cGMP to the cells expressing GFP–myosin II, then the cells were UV irradiated repeatedly. Fig. 4B shows the typical time course of the fluorescence intensity of GFP–myosin II in the cell cortex with repeated uncaging (arrows). Myosin II levels increased immediately after uncaging, but the increase was very subtle (up to 1.05). Because the relative fluorescence level of cGMPi was 1.5–1.6 when stimulated with external cAMP, the level of uncaged cGMP might be not enough to cause more myosin II to translocate to the cortex.

Given that cGMP is rapidly degraded by PdeD (Bader et al., 2007; Meima et al., 2002), we used a PdeD knockout mutant expressing D-Green cGull. Fig. 4C shows a typical time course of the fluorescence intensity of D-Green cGull after uncaging (arrows). The fluorescence in the mutant was much higher than that in wild-type cells, and the duration was prolonged. Fig. 4E shows typical time courses of fluorescence images upon uncaging (arrow), indicating that the locally uncaged cGMP spreads over the cytosol.

Next, PdeD-null cells expressing GFP–myosin II were incubated with caged cGMP and then UV was applied to the cells. Fig. 4D shows the time course of the fluorescence intensity of GFP–myosin II after uncaging (arrow). Fluorescence transiently decreased at 5–10 s and then increased with a peak at ∼60 s, which is a similar response to the changes in the amount of myosin II in the cytoskeleton in wild-type cells upon cAMP stimulation. Fig. 4F shows a typical time course of total internal reflection fluorescence (TIRF) images of PdeD-null cells expressing GFP–myosin II upon UV illumination. TIRF microscopy visualizes ∼100 nm in depth from the surface of the coverslip, which corresponds to the thickness of the cell cortex, including the cell membrane and underlying cortical actin layer. TIRF microscopy enables observation of individual actin and myosin II filaments (Bretschneider et al., 2004; Itoh and Yumura, 2007; Yumura et al., 2008). After UV illumination, myosin filaments rapidly disappeared from the cortex (5–10 s) and then accumulated, with a peak time of 60 s. Although myosin II has been suggested to associate with the actin cortex in a filamentous form upon cAMP stimulation (Bosgraaf et al., 2002a; Egelhoff et al., 1993; Liu and Newell, 1991; Yumura and Fukui, 1985), the present study visualized its association for the first time.

Together, we conclude that the increase of cGMPi directly induces similar dynamics of myosin II to those seen upon chemotactic stimulation.

In the present study, we visualized the dynamics of cGMPi in live Dictyostelium cells during chemotaxis using the fluorescent probe D-Green cGull. When the cells were stimulated with external cAMP, the concentration of cGMPi transiently increased in the cytosol. During natural aggregation, the response of cGMPi propagated in a wave-like manner from the aggregation center, indicating that the cells respond to the propagation of extracellular cAMP signaling waves. GC-null cells did not show any fluctuation in fluorescence, and PdeD-null cells showed enhanced dynamics. Therefore, D-Green cGull is a reliable indicator of cGMPi. In the future, the absolute concentration of cGMPi should be examined using fluorescence ratiometry combined with a cGMP-insensitive fluorophore.

The time course profiles of cAMP-stimulated cGMPi dynamics obtained in previous studies were significantly different from those obtained in the present study. In previous reports, the cGMPi level was found to increase at 0.8 s after cAMP application with a peak time of 10 s and return to the resting level at 20–30 s (Bosgraaf et al., 2002a; Mato et al., 1977b; Valkema and Van Haastert, 1992). In contrast, in the present study, the cGMPi level began to increase ∼6 s after cAMP application, peaked at 24 s, and returned to the resting level at 98 s. This discrepancy might be due to the measurement methods. In the previous study, cGMP level was measured by radioimmunoassay of the cell lysate. If the results found from the previous biochemical methods are correct, the PDE-based cGMP probe might have a lower affinity for cGMP, resulting in a slow response. Otherwise, the dissociation of cGMP from the probe might be slow. The response of the probe depends on the on-rate and off-rate of cGMP binding to the probe in relation to how fast cGMP levels change. The kinetics of binding of cGMP to the probe needs to be examined in the future.

Previous studies have reported that chemotactically activated sGC localizes at the leading edge of the cell in a cAMP gradient (Veltman et al., 2005), indicating that the cGMPi level locally increases within a cell. The gradient of cGMPi from the anterior to the posterior end of the cell might preferentially induce myosin II translocation toward the posterior cortex, thereby inducing cell polarity. Incidentally, myosin II-null cells show less polarity, extending multiple pseudopods. In addition, sGCA-null and GbpC-null cells also show low polarity (Bosgraaf et al., 2005). However, we did not find any local increase in cGMPi levels in a specific subcellular region, even with high magnification and rapid image acquisition, although it was visualized in the experiments using caged cGMP. Also, local uncaging at the anterior or posterior regions of cells did not affect the cell migration or cell polarity in the present study. Mutant cells in sGC that lose localization at the leading edge exhibit normal cGMP production and myosin II localization (Veltman and Van Haastert, 2006), which further suggests that the gradient of cGMPi is not required for chemotaxis. However, we cannot rule out the possibility that cGMPi locally increases in the vicinity of the cell membrane. A recent report visualized the characteristic concentration changes in cAMPi in the vicinity of the membrane in human cultured cells by using membrane-bound probes (Anton et al., 2022).

We compared cGMPi oscillations with those of other second messengers, such as cAMPi and Cai²⁺. These oscillations were synchronized almost in phase with each other. However, for a more detailed analysis, their timing needs to be compared upon double expression of probes with different colors in the future. Previous reports have indicated that the dynamics of cAMPi coordinate cell migration, but disappear in the late developmental stage, although cell migration propagates in waves (Hashimura et al., 2019). The disappearance of cAMPi dynamics in the late developmental stage has been disputed (Singer et al., 2019). Moreover, we did not find any changes in cAMPi levels at the vegetative stage, suggesting that the involvement of cAMPi signaling is limited to later than aggregation stage.

Cai²⁺ signaling might not be required for Dictyostelium chemotaxis. Mutant cells deficient in the inositol 1,4, 5-trisphosphate receptor-like protein, in which Ca²⁺ entry in response to chemoattractants is abolished, still exhibit normal chemotaxis and cell velocity (Traynor et al., 2000). However, Ca²⁺ ionophores can induce similar actin and myosin II dynamics to those seen upon cAMP stimulation, and BAPTA-AM, a chelator of Cai²⁺ suppresses them (Yumura, 1993). Ca²⁺ reduces the activity of GC (Valkema and Van Haastert, 1992), and external Ca²⁺ reduces myosin II dynamics in permeabilized cells (Van Haastert et al., 2021). When the cell membrane is wounded, the influx of Ca²⁺ through the wounded pore induces the dissociation of myosin II from the wounded cell cortex (Tanvir et al., 2021). Therefore, the role of Ca²⁺ in chemotaxis requires further investigation.

cGMPi might contribute to chemotaxis. Loss of both GCA and sGCA severely impairs chemotaxis and normal cell migration (Roelofs and Van Haastert, 2002). The observation that the oscillations of cGMP_i coincide well with those of cell velocity also proves that cGMPi contributes to chemotaxis. However, we did not observe any fluctuations in cGMPi in non-chemotactic or dividing cells. Therefore, it does not appear to be involved in cell migration without any cues, or cell division, suggesting that there might be another unknown signal involved in these processes.

The assembly of myosin II into bipolar filaments is essential for the generation of a force, as they interact with actin filaments in a manner similar to that seen in muscle contraction. The assembly of monomeric myosin II into filaments is regulated by the phosphorylation of myosin heavy chains (MHCs), which are regulated by MHC-specific kinases (MHCKs) and phosphatases (de la Roche and Côté, 2001). MHCKA, one of the six MHCKs in Dictyostelium, localizes to the anterior region, whereas MHCKB and MHCKC localize to the posterior region of migrating cells (Nagasaki et al., 2002; Yumura et al., 2005). However, signals upstream of MHCKs remain unclear. cGMP is a candidate upstream signal for the initiation of myosin II dynamics, because a jump in cGMPi using caged cGMP induced myosin II to translocate to the cortex. Moreover, cGMP is upstream of myosin II light chain (MLC) kinase (Silveira et al., 1998). MLC phosphorylation enhances the motor activity of myosin II (Griffith et al., 1987). Both filament assembly and enhancement of motor activity of myosin II may facilitate the exertion of force by interacting with the cortical actin, thus driving cell migration.

Although myosin II level increased very slightly in wild-type cells upon uncaging of the caged cGMP, myosin II in PdeD-null cells showed similar large responses to those in wild-type cells upon cAMP stimulation, indicating that a certain threshold increase in cGMPi is required for the initiation of the myosin II dynamics. It is known that phosphorylation of MHC, which mediates the translocation of myosin II from the cortex to the cytosol, is required because mutant myosin II that mimics the dephosphorylated form, over-assembles in the cortex (Egelhoff et al., 1993; Yumura and Uyeda, 1997). Because myosin II showed a biphasic response (the first decrease and the second increase in the cortex) upon uncaging of cGMP, and cGMPi showed a monophasic response, it is not plausible that the changes in the concentration of cGMPi directly regulate the individual (decrease and increase) myosin II responses. A recent report described that sGC and cGMP-binding protein (GbpC) regulate actin assembly, which might indirectly regulate the dynamics of myosin II (Tanabe et al., 2018). Further clarification of the signaling pathway linking the cGMP and myosin II dynamics is needed in the future.

D. discoideum (AX2) and mutant cells were cultured in plastic dishes at 22°C in HL5 medium (1.3% bacteriological peptone, 0.75% yeast extract, 85.5 mM D-glucose, 3.5 mM Na₂HPO₄·12H₂O, and 3.5 mM KH₂PO₄, pH 6.3), as previously described (Yumura, 2001). The cells were transformed with extrachromosomal vectors for the expression of D-Green cGull (Tanaka et al., 2019), Flamindo2 (Tanaka et al., 2019), D-GCaMP6s (Pervin et al., 2018; Talukder et al., 2020; Tanaka et al., 2019), GFP–lifeact (Yumura et al., 2014) and GFP–myosin II (Yumura, 2001) via electroporation or laserporation, as previously described (Yumura et al., 1995; Yumura, 2016). Transformed cells were selected in HL5 medium supplemented with 10 µg/ml G418 (Wako, Japan). To obtain cells in the aggregation stage, HL5 medium was replaced with BSS (3 mM CaCl₂, 10 mM KCl, 10 mM NaCl, and 3 mM MES, pH 6.3), and the cells were incubated for 8–9 h. Otherwise, after cells were incubated in BSS at 22°C, for 1 h, they were kept at 10.5°C for 12–14 h. When they were returned to 22°C, they began to aggregate.

The gene encoding Green cGull (Matsuda et al., 2017) was synthesized by Integrated DNA Technologies (Coralville, IA, USA) after optimizing the code for Dictyostelium. The gene was inserted between the XbaI and BamHI sites of the pDYU1G vector (Kondo and Yumura, 2020) for stable expression in Dictyostelium under the control of an actin15 promoter.

Cells (10⁷) expressing D-Green cGull were lysed in 3 ml of a lysis solution [100 mM NaCl, 10 mM EGTA, 4 mM NaF, 1 mM dithiothreitol (DTT), 1% Triton X-100, 10 mM PIPES pH 7.5] supplemented with a protease inhibitor cocktail (Sigma-Aldrich, Japan) on ice and centrifuged at 12,000 g for 10 min. After the indicated concentrations of cGMP or cAMP were added to the supernatant (0.3 ml), the fluorescence intensities were measured using a fluorometer (FL-2500; Hitachi High-Tech, Tokyo, Japan). The excitation and emission wavelengths were 488 and 522 nm, respectively. DTT was included in the lysis buffer to prevent the degradation of cGMP (Bader et al., 2007). We did not observe any decrease of fluorescence during recording.

Live cells were pre-incubated with 5 mM caffeine (Wako, Japan) to prevent endogenous cAMP oscillations. Immediately after the addition of the indicated concentration of cAMP to the cell suspension, the fluorescence intensities were measured over time using a fluorometer by gently mixing the solution with a vortex mixer or bubbles of oxygen.

Aggregation-competent cells were settled in a glass-bottom chamber and slightly compressed using an agarose block (1.5%, dissolved in BSS, 1 mm thick) to improve the cell imaging (Fukui et al., 1986; Yumura et al., 1984). Under these conditions, cells aggregated normally by forming aggregation streams. Fluorescence images of cells expressing fluorescent proteins were observed under a confocal microscope (LSM510 or LSM710; Zeiss, Oberkochen, Germany) equipped with 20×, 40× or 63× objectives and an argon laser (with standard filter settings for GFP). The optical thickness was set at 1 or 2.5 µm. The exposure time was 100 ms with an interval of 2–10 s. TIRF microscopy (based on IX71 microscope; Olympus) was conducted as previously described (Yumura et al., 2013).

First, (7-dimethylaminocoumarin-4-yl)methyl-8-bromoguanosine-3′, 5′-monophosphate (DMACM-caged 8-Br-cGMP) was dissolved in DMSO to prepare a 10 mM stock solution. Cells expressing D-Green cGull or GFP–myosin II were incubated with caged cGMP at a final concentration of 10 µM for 1 h in the dark. We also assessed the use of a much longer incubation time of up to 24 h, but the results were not affected. Fluorescence images were acquired with a 63× objective, using confocal or TIRF microscopy. To uncage the caged cGMP, a laser (405 nm) was applied to a circular area of 2 µm diameter in the cells.

The acquired images were analyzed using the ImageJ software (http://rsbweb.nih.gov/ij/). At low magnification, the fluorescence intensities in a small rectangular area with 10–15 cells were measured over time. For single cells, after obtaining the cell outline at high magnification, the mean fluorescence intensity and cell velocity were determined. Cell velocities were determined using the manual tracking plugin in ImageJ software. To quantify the F-actin and myosin II levels in the cell cortex, the integrated fluorescence intensities in a 1 µm-thick outline including the cell membrane were calculated as described previously (Tanaka et al., 2017). The periodic durations of the waves of the individual probes were estimated using the power spectrum based on Fourier transformation. Cross-correlation analysis was performed as described previously (Hashimura et al., 2019). Graphs were created using GraphPad Prism version 7 software (GraphPad Inc., USA).

Statistical analyses were performed using the GraphPad Prism version 7 software. Data were analyzed using one-way ANOVA with Tukey's multiple comparison test, and are presented as the mean±s.d.

We are grateful to the members of the Dictyostelium research community, the Dicty Stock Center, and the NBRP Nenkin for providing several of the mutants that were used in this study. We would like to thank Dr H. Hashimura for his technical advice. We thank Dr T. Kitanishi-Yumura for her critical reading of the manuscript and helpful comments.