Spontaneous transitions between amoeboid and keratocyte-like modes of migration

The motility of adherent eukaryotic cells is driven by the dynamics of the actin cytoskeleton. Despite the common force-generating actin machinery, different cell types often show diverse modes of locomotion that differ in their shape dynamics, speed, and persistence of motion. Recently, experiments in Dictyostelium discoideum have revealed that different motility modes can be induced in this model organism, depending on genetic modifications, developmental conditions, and synthetic changes of intracellular signaling. Here, we report experimental evidence that in a mutated D. discoideum cell line with increased Ras activity, switches between two distinct migratory modes, the amoeboid and fan-shaped type of locomotion, can even spontaneously occur within the same cell. We observed and characterized repeated and reversible switchings between the two modes of locomotion, suggesting that they are distinct behavioral traits that coexist within the same cell. We adapted an established phenomenological motility model that combines a reaction-diffusion system for the intracellular dynamics with a dynamic phase field to account for our experimental findings.

. Computation of geometric and dynamic quantities of a motile cell and their representation as kymographs. (A) Cell track with underlying spatio-temporal coordinate system shown for exemplary contours and virtual markers: θ = 0 (black filled circle) and θ ∈ { π 2 , π, 3π 2 } (empty circle). (B) A quantity of interest (the example here is the local motion) can be displayed as a two-dimensional kymograph representation, with the time coordinate on the x-axis and the normalized arc length coordinate θ ∈ [0, 2π) on the y-axis (colored lines representing single cell contours). (C) Local dispersion showing regions of virtual marker thinning (expansions, red), virtual marker clustering (retractions, blue), and neutral regions (green). (E) Distances from virtual markers on the contour to the center of mass of the cell (so-call contour shape, color-coded in purple). (D, F) Corresponding kymographs showing local dispersion and distances to the center of mass as a function of time and space.
10 μm 10 μm A B Figure S2. Initial conditions for the model simulations. All simulations, which have been taken into account for the analysis in this article, started with either 25% (A) or 75% (B) of the cell area covered by high values of the effective force generating component c. After starting the simulations, the cell shape and the intracellular distribution of c randomly evolved according to the stochastic model equations, so that correlations with the initial cell shape and c distribution were rapidly lost, see also the corresponding movie.

Small protrusions at the leading edge of fan-shaped cells
At the leading edge of persistently moving fan-shaped cells, we repeatedly observed the emergence of small, short-lived protrusions. In Fig. S4A, the formation and decay of such a protrusion is illustrated in a sequence of snapshots (a)-(e) from the fluorescence recording of a stable fan-shaped cell. The entire event extends over a short time window of about 36 s, which is typical for the life-time of these structures. Due to their small size, their formation does not affect the overall value of the relative wave area Q, see Fig. S4B, where Q remains around 0.6 for the entire measurement time. However, in the local dispersion kymograph of the cell contour in panel C, protrusion formation can be clearly detected as a strongly localized, extending region (red), with a simultaneous increased retraction along the rest of the contour (blue), see the time period between (a) and (e), as well as the time period shortly after t = 200 s.
The growing protrusion is accompanied by a decrease of the fluorescence intensity in its vicinity. In particular, the front of the actin wave that pushes the cell forward is directly affected by the growing protrusion, such that the leading wave segment is disrupted, indicating a depletion of F-actin, which causes a breakup of the wave segment at the position of protrusion formation, see Fig. S4A at time frames (b)-(d). We thus conclude that the growing protrusion competes with the wave for the common pool of actin, consuming some of it in its surrounding in order to push the plasma membrane outwards. When the protrusion has decayed, the disrupted actin segment at the wave front heals and recovers its initial fluorescence intensity. This can be also seen in the fluorescence intensity kymograph in Fig. S4D, taken along the red dashed line displayed in the first snapshot of panel A.
Small protrusions at the leading edge of fan-shaped cells have been noted earlier  and can be also induced by chemotactic cues (Ecke and Gerisch, 2019). However, based on the limited amount of available data, their precise nature remains elusive. They resemble small, short-lived pseudopods that are commonly observed in D. discoideum cells when moving in an amoeboid fashion (Swanson and Taylor, 1982;Rubino et al., 1984;Wessels et al., 1988) but it is also conceivable that their formation is driven by a blebbing mechanism (Tyson et al., 2014).  Table S1.  Table S1.  Table S1.

Analysis of the distance from the contour to the center of the cell
Our model was not specifically tuned to reproduce geometrical features of the experimentally observed cell shapes. Thus, the distance from the center of the cell to the contour provides an additional opportunity to compare experimental observations and simulation results, i.e., an independent quality check for the performance of our model.
In order to interpret the data, let us first consider what we expect for the different cases. During amoeboid migration without any external cue, pseudopod activity is randomly distributed all around the cell contour. Hence, on a time scale much longer than the typical pseudopod lifetime, the distance from the center to the contour of an amoeboid cell will evolve randomly and will, when averaged over time, approach half of the cell diameter for each point on the contour (a typical cell diameter is about 10µm). The mean distance to center for amoeboid contours will thus converge to a flat profile at a distance of around 5µm, while the histogram of the center to contour distances for all individual contour positions will be symmetrically distributed around this value. This is indeed observed both in the experimental (Fig. S9) and in the numerical data (Fig. S10). When considering time intervals, during which the cell moves in an amoeboid fashion (highlighted in yellow in the kymographs (B) in Figs. S9 and S10), we see in the experimental as well as in the simulation data no distinct coherent patterns. Figures S9 (C) and S10 (C) depict the distance from the center to the contour for all individual contours (grey lines) in the highlighted time interval, and the black line is the average over all these contours, displaying an almost constant value of around 5µm. Figures S9 (D) and S10 (D) show the histograms for all contour positions during the highlighted time interval, which almost symmetrically center around the mean value.
In the case of fan-shaped motion, the cell adopts a kidney-like shape, elongated perpendicular to the direction of motion. Thus, the front and rear parts of the cell border are closer to the center compared the left and right parts. This shape is stably maintained while the cell is moving. Hence, we expect to see a characteristic pattern in the distances of the contours from the center for these cells. This is indeed observed during time intervals, when cells move in the fan-shaped mode, see the time interval highlighted in yellow in the kymographs (G) in Figs. S9 and S10. Here, the kidney shape is reflected by an alternating pattern of dark and bright stripes. This is also visible in the plots of the individual contours from the highlighted time interval (grey lines) and in their average (black line) in Figs. S9 (H) and S10 (H). As the kidney shape persists over time, the histograms of the distances taken over the individual contours display a characteristic asymmetric distribution with three local maxima, which is, however, more clearly seen in the numerical than in the experimental data, see Figs. S9 (I) and S10 (I). Video 7: Video of a fluorescent D. discoideum DdB NF1 KO cell, corresponding to Figure S5. Breakdown of the actin wave and the associated switching from fan-shaped migration to amoeboid motion. The frame rate is 4 seconds. F-actin is labeled with Lifeact-GFP. Total time 360 seconds.