All Dictyostelium discoideum strains used in this study derived from the axenic strain AX2-214 (considered as wild type), and were cultivated in Petri dishes in HL5 medium (Formedium) at 22C. Dictyostelium cells expressing GFP-cofilin or mCherry-cofilin, or the mutant strain lacking Aip1 (Konzok et al., 1999) and expressing GFP-cofilin were described earlier (Ishikawa-Ankerhold et al., 2017).

For all experiments, Dictyostelium cells were cultivated in HL5 medium at 22°C. For the specific treatments, the medium was replaced by HL5 medium supplemented with the respective agent.

Sodium azide (NaN ₃ ): 10 mM in HL5 medium. To test the stability of rods when the stress stimulus is removed, cells treated with 10 mM sodium azide for 60 min were washed and incubated in HL5 medium for recovery.

2, 4-Dinitrophenol (DNP): 200 µM in HL5 medium.

Hyperosmotic shock (SOR): 200 and 400 mM D-sorbitol (Sigma-Aldrich) in HL5 medium.

Sorbic acid (SA) and propionic acid (PA): the respective pH values of the HL5 medium were adjusted with stock solutions of 14 mM sorbic acid or 134 mM propionic acid.

2-deoxy-glucose (2-DG): HL5 medium without glucose (Formedium HLB0102) supplemented with 10 g/L 2-deoxy-D-glucose (Roth)

Cytochalasin D (CytD) and Latrunculin A (LatA) (both Sigma-Aldrich): final concentrations of 1, 2, 4, 6, 8, 10, and 20 µM, or 1, 5, and 10 μM, respectively, in HL5 medium were tested. Dictyostelium cells expressing GFP-cofilin were plated on 12-mm cover glasses placed in 24-well plates and treated with cytochalasin D or latrunculin A for 60 min before induction of rod assembly with 10 mM sodium azide for 60 min. Then, cells were fixed with methanol at −20°C for 15 min and stained for actin.

The experiments were repeated at least 3 times for each treatment. For quantification, cells with cytoplasmic rods were counted from at least 10 different fields of view (image size of 512 × 512 µm).

For quantification of cells with cytoplasmic rods, Zen blue software was used (Carl Zeiss Microscopy).

For cytosolic pH measurements, the ratiometric fluorescent protein pHluorin2 (Mahon 2011) was cloned via HindIII and ClaI into a pDEX-based Dictyostelium vector enabling expression under control of an actin-15 promoter and conferring resistance to blasticidin (Müller-Taubenberger, 2006). The pEntry-mCherry-pHluorin2 vector (GFP-RFPpHluorin; plasmid Y872) was a kind gift of Prof. Simon Alberti (Dresden), and was used as template for amplification in combination with the primers 5′-gtg aag ctt aaa atg cca act ttg tac aaa aaa gca gg-3′ and 5′-gtg atc gat tta ttt gta tag ttc atc cat gcc atg tg-3′.

AX2 cells were transfected by electroporation using a gene pulser (Bio-Rad Laboratories) at 0.8–0.9 kV and 3 µF in 4-mm cuvettes. Transformants were selected and cultivated in the presence of 10 µg/ml blasticidin-S (ICN Biomedicals Inc.).

Cells were plated on round 12-mm glass cover glasses coated with poly-lysine (Merck) and placed in 24-well plates, and after 1 h were subjected to the specified treatment. Thereafter, cells were fixed in methanol at −20°C for 15 min and labeled as described (Ishikawa-Ankerhold et al., 2017). After fixation, cells were washed three times with PBS, and then incubated in blocking buffer (PBS plus 2% bovine serum albumin) for 1 h at room temperature (RT). After blocking, the cells were washed three times with PBS and immunolabeled with the primary antibodies. Secondary antibodies were added for 1 h at RT. Then, samples were washed and mounted using Gelvatol or Dako. Primary antibodies used were specific for: actin (mAb Act1) (Simpson et al., 1984), Aip1 (mAb 246-153-2 and 246-404-2) (Konzok et al., 1999), coronin (mAb 176-3-6) (de Hostos et al., 1991), cofilin (pAb 496) (Ishikawa-Ankerhold et al., 2017), and isotype IgG (Molecular Probes). Secondary antibodies used in the study were Cy3, Alexa Fluor-488, -563, or -594 goat anti-mouse or anti-rabbit IgG (Molecular Probes). For some experiments, Atto 550-phalloidin (Molecular Probes, Sigma-Aldrich), and DAPI (1:1,000) were added together with the secondary antibodies.

A confocal LSM 880 microscope (Carl Zeiss Microscopy) equipped with an Airyscan module and Plan-Apo 20x/NA 0.8 and Plan-Apo 63x/NA 1.46 oil immersion objectives, or a confocal LSM 510 microscope (Carl Zeiss Microscopy) equipped with a 63x Neofluar 1.4 oil immersion objective were used for image acquisition.

For live-cell microscopy, Dictyostelium cells expressing GFP-cofilin were plated into an open chamber or into 35-mm glass-bottom dishes (ibidi, µ-Dish 35 mm, high glass bottom), and incubated and recorded in medium or in medium adapted to the respective treatment conditions.

FRAP (fluorescence recovery after photobleaching) measurements were performed as described previously for nuclear rods (Ishikawa-Ankerhold et al., 2017). In short, living cells with cytoplasmic GFP-cofilin rods were photobleached using a LSM 510 Meta laser scanning confocal microscope with a 488-nm laser line at 100% intensity for 20 interactions as described in the detailed protocol (Müller-Taubenberger and Ishikawa-Ankerhold, 2013).

For the actin-cofilin rod disassembly experiments, the GFP-cofilin cells were plated on coverslips coated with poly-lysine, and treated with 10 mM sodium azide for 1 h at RT to induce the assembly of cytoplasmic actin-cofilin rods. After rod induction, the cells were washed to remove the chemical stimulus and incubated with medium for recovery. After 0, 5, 10, 15, 20 and 30 min under medium conditions, the cells were fixed with cold methanol (−20°C) for 15 min, washed and mounted for image acquisition.

For ratiometric pH measurements, Dictyostelium cells expressing GFP-mRFP-pHluorin2 were seeded at 1 × 10⁴ per ml into 35-mm glass-bottomed dishes (ibidi, µ-Dish 35 mm, high glass bottom), before imaging. The images were acquired using a LSM 880 confocal microscope with an Airyscan module (Carl Zeiss Microscopy). All imaging parameters for excitation, detection and all software settings were kept identical for the different sample acquisitions [laser intensity 488 nm (2%) and 561 nm (2%), image size (212 × 212 µm), Z-stack (30 µm), objective Plan-Apochromat ×20/0.8, pixel time (0.77 s), frame time (943.72 s)].

The images were analyzed with the ZEN software using the plugin (Histo), where a defined ROI of 4 µm² was applied to the selected cells to obtain the mean fluorescence intensity from each channel (green and red). The ratios of green/red channels were plotted as median with interquartile range to represent the pH fluorescence changes inside the cells. pHluorin2 allowed us to monitor decreases of the intracellular pH after acidification of the extracellular medium, which was identified by an increase in the ratio of pH-sensitive GFP versus pH-insensitive mCherry fluorescence. In measurements where we tested the application of HL5 medium adjusted to pH values below 5.5 (pH 5.25 and 5.0), we did not detect a linear increase of the GFP fluorescence anymore, and therefore show only the measurements for pH 5.75 and 5.5.

Statistical analyses were calculated and graphically plotted with GraphPad Prism 9 (GraphPad Software, La Jolla, CA, United States). Data are means ± standard deviation of the mean (SD). Statistical analyses were conducted as described in the figure legends. Findings were considered statistically significant at p < 0.05.

Earlier work conducted in our laboratory and data from other studies suggested that cytoplasmic actin-cofilin rods are induced by treatments that cause depletion of ATP (Ashworth et al., 2003; Bernstein and Bamburg, 2003; Bernstein et al., 2006; Huang et al., 2008). The current investigations aimed to identify similarities and differences between nuclear and cytoplasmic rods in Dictyostelium cells, and to analyze the formation of cytoplasmic rods under different stress conditions by live-cell imaging.

In first experiments, ATP-depletion was provoked experimentally by addition of sodium azide (Ishikawa-Ankerhold et al., 2017), which acts by uncoupling the respiratory chain. Sodium azide induces cytoplasmic actin-cofilin rods both in wild-type as well as in GFP-cofilin expressing cells to a very similar extent (Supplementary Figure S1), and in order to follow formation of actin-cofilin rods by live-cell imaging, most experiments were performed with GFP-cofilin expressing cells. Addition of sodium azide to the cells causes within minutes a rapid shape change and reorganization of the actin cytoskeleton (Figure 1A and Supplementary Video S1). First actin-cofilin rods become visible within 10 min of sodium azide treatment, and after 1 h, all cells are rounded up, and rods are detected in about 80% of cells. Image reconstruction using 3D rendering confirmed the cytoplasmic localization of actin-cofilin rods (Figure 1B and Supplementary Video S2).

Our earlier studies on nuclear rods showed that these contain in addition to actin and cofilin a distinct set of other proteins comprising actin-interacting protein 1 (Aip1), coronin (CorA), filactin (Fia), and the 34-kDa actin-bundling protein B (AbpB), and we found a finely tuned spatial-temporal pattern of protein recruitment during formation of rods inside the nuclei (Ishikawa-Ankerhold et al., 2017). For cytoplasmic rods, we confirmed actin and cofilin as the main constituents (Figure 1C, Supplementary Figure S1C, and Supplementary Video S3), and by immunofluorescence labeling we identified in addition Aip1 and coronin (Figure 1C), but did not detect, in contrast to nuclear rods, a sequential recruitment.

Like for nuclear rods, the compaction of cytoplasmic rods requires Aip1, a protein that induces the disassembly of actin filaments in conjunction with cofilin (Nadkarni and Brieher, 2014). Induction of cytoplasmic rods by sodium azide in a mutant lacking Aip1 (Konzok et al., 1999) revealed that rods are less compacted (Figures 1D,E), a result suggesting that the turnover of actin is a critical determinant for unimpeded rod formation.

This assumption is further supported by the application of drugs that interfere with actin polymerization. We have used cytochalasin D (CytD) that by binding to the high affinity growing ends prevents the elongation of actin nuclei and filaments (Casella et al., 1981), and latrunculin A (LatA), a drug that binds actin monomers and thus prevents polymerization of actin (Morton et al., 2000). The formation of cytoplasmic rods was increasingly inhibited with rising concentrations of CytD (Supplementary Figure S2A). Similarly, LatA inhibited the emergence of rods in a dose-dependent manner (Supplementary Figure S2B). These results indicate that under conditions of low ATP, the turnover of actin and the availability of actin monomers are crucial and limiting for the formation of cytoplasmic rods. In contrast to intranuclear rods, cytoplasmic actin-cofilin rods are not stainable with phalloidin (Figure 1B and Supplementary Video S2), which indicates that these assemblies are structurally different to cortical actin and intranuclear rods, a finding that has been described already earlier (Nishida et al., 1987; Ishikawa-Ankerhold et al., 2017).

In order to address the change in the dynamics of cytoplasmic actin-cofilin rods over time (i.e., newly assembled versus “aged” rods), we analyzed the exchange between the assembled and diffuse pool of GFP-cofilin by FRAP (fluorescence recovery after photobleaching). For this, GFP-labeled rods were bleached using high laser power (100%), and fluorescence recovery was monitored over time (Figures 2A–C and Supplementary Video S4). The mobile fraction represented by the fluorescence recovery of GFP-cofilin rods after photobleaching was about 20%, meaning that cytoplasmic rods are rather stable structures with low protein exchange.

The formation of actin-cofilin rods is rapidly reversible when the stress stimulus is removed (Figures 2D,E). After induction of cytoplasmic rods by addition of 10 mM sodium azide, the percentage of cells with cytoplasmic rods decreased rapidly when the sodium azide was washed out and replaced by standard HL5 medium. 5 min after replacement of sodium azide by medium, cytoplasmic rods started to disassemble, recognizable as shortening cofilin bundles. In the subsequent recovery phase, the bundles were shortened further and appeared as punctuated small cofilin aggregates, which continued to disintegrate.

Another drug that causes energy depletion is 2,4-dinitrophenol (DNP). DNP acts as protonophore and causes uncoupling of respiratory chain oxidative phosphorylation thereby inhibiting ATP production. Previous work by Gerisch and colleagues (Jungbluth et al., 1994) showed that addition of 50 µM DNP causes the reversible compartmentalization of the actin cytoskeleton. However, the available microscopical techniques were limited at that time, and did not allow a detailed resolution of cytoskeletal structures.

We have visualized the formation of cytoplasmic actin-cofilin rods in GFP-cofilin expressing cells after addition of DNP (Figure 3A). Live-cell microscopy revealed a rapid reorganization of the actin cytoskeleton after addition of DNP, and rods became visible within 10–15 min (Supplementary Figure S3A). After 2 h with DNP, 50% of cells showed cytoplasmic rods. Rod formation was fully reversible after an incubation time of 3 h by replacing the DNP medium with standard medium similar to the recovery after sodium azide treatment.

Next, we tested hyperosmotic stress as another condition that we assumed to potentially induce cytoplasmic actin-cofilin rods. Hyperosmotic stress was induced by addition of 200 and 400 mM sorbitol (Figure 3B). It was reported earlier that in consequence of hyperosmotic conditions, Dictyostelium cells rearrange their cytoskeleton. This transition was postulated to constitute the major osmoprotective mechanism in Dictyostelium accompanied by a rapid acidification of the cytosol (Ott et al., 2000; Pintsch et al., 2001), but the cytoskeletal changes were not investigated in greater detail. In addition, it was found that vesicle mobility and membrane flow were downregulated under hyperosmotic conditions (Pintsch et al., 2001) a notion supported in studies of fluid- to solid-like state transitions more recently (Munder et al., 2016). We found that hyperosmotic conditions caused a rapid formation of cytoplasmic actin-cofilin rods (Figure 3B). After 2 h of incubation with 200 mM sorbitol, about one third of the cells showed cytoplasmic rods, with 400 mM sorbitol about 50% (Figure 3C).

Since hyperosmotic conditions were described to be accompanied by a rapid drop in intracellular pH, we next set out to determine changes in the intracellular pH of Dictyostelium cells. Previous studies using ³¹P-NMR (nuclear magnetic resonance) showed that the intracellular pH of growth-phase Dictyostelium cells is around pH 7.3 (Kay et al., 1986). To directly monitor changes of the intracellular pH, we initially tested established intracellular pH indicators used for pH determination in other eukaryotic cell lines. Under our experimental conditions, BCECF (Invitrogen) and pHrodo (Life technologies) were not usable because they were rapidly sequestered into the endosomal compartment of Dictyostelium cells. We therefore generated a vector for expression of the ratiometric pH sensor pHluorin2 (Mahon, 2011), and examined pHluorin2 expressing Dictyostelium cells in live-cell measurements in order to monitor changes in the intracellular pH after reduction of the pH of the extracellular medium to 5.75 and 5.5. While DNP had no effect on the intracellular pH of Dictyostelium cells, application of 400 mM sorbitol caused a reduction of the cytoplasmic pH (Figure 3D) consistent with earlier results (Pintsch et al., 2001).

In order to investigate whether an acidification of the cytosol itself could trigger the formation of cytoplasmic actin-cofilin rods, we systematically examined the formation of rods by increasingly lowering the pH of the extracellular medium by the addition of weak acids. Under standard laboratory conditions, axenically growing Dictyostelium cells are cultivated in medium with a pH of 6.4. By addition of weak membrane-permeable acids, sorbic or propionic acid, we reduced the pH of the growth medium (5.75, 5.5, 5.25, 5.0) and followed the formation of cytoplasmic actin-cofilin rods either in fixed cells (Figure 4A) or by live-cell microscopy (Figure 4B). All pH conditions tested provoked the formation of cytoplasmic actin-cofilin rods. The rods varied from bar-shaped at low extracellular pH (pH 5.75 and 5.5) to more needle-like condensates at very low extracellular pH conditions (pH 5.25 and 5.0) (Figures 4A,B and Supplementary Figure S3B). With sorbic acid buffered medium, the percentage of cells with rods was highest with the slightest pH-shift (pH 5.75), and then decreased at lower values (Figure 4C). With propionic acid buffered medium, about 60% of cells formed cytoplasmic rods at pH 5.25, but rods were also detected at a considerable extent at the other pH values tested (Figure 4D). Ratiometric measurements of cells in sorbic and propionic acid containing medium confirmed the reduction of the intracellular pH compared to control cells (Figure 4E).

The life cycle of Dictyostelium cells is characterized by two main phases, the growth stage, in which the cells multiply by mitotic divisions, and a developmental stage, in which the cells aggregate to form a multicellular slug and finally fruiting bodies. The transition from growth to development is induced by starvation.

In the course of experiments where we tested energy depletion, we cultivated Dictyostelium cells in glucose-free medium containing 2-deoxyglucose (2-DG). In contrast to mammalian cells, application of 2-DG did not result in the same effect as ATP depletion with sodium azide or DNP. Instead, we observed that after 5–6 h a low percentage of cells (<5%) were characterized by the appearance of rods. After 18 h, about 20% of cells showed cytoplasmic rods (Figures 5A,C). Ratiometric measurements of pHluorin-cells after application of 2-DG medium revealed a significant drop in the intracellular pH after 18 h (Figure 5D). After 24 h, rods were not detectable anymore, but most cells were characterized by an elongated shape typical for Dictyostelium cells of the aggregation stage (Figure 5B).