Steering cell migration by alternating blebs and actin-rich protrusions
- 3.7k Downloads
High directional persistence is often assumed to enhance the efficiency of chemotactic migration. Yet, cells in vivo usually display meandering trajectories with relatively low directional persistence, and the control and function of directional persistence during cell migration in three-dimensional environments are poorly understood.
Here, we use mesendoderm progenitors migrating during zebrafish gastrulation as a model system to investigate the control of directional persistence during migration in vivo. We show that progenitor cells alternate persistent run phases with tumble phases that result in cell reorientation. Runs are characterized by the formation of directed actin-rich protrusions and tumbles by enhanced blebbing. Increasing the proportion of actin-rich protrusions or blebs leads to longer or shorter run phases, respectively. Importantly, both reducing and increasing run phases result in larger spatial dispersion of the cells, indicative of reduced migration precision. A physical model quantitatively recapitulating the migratory behavior of mesendoderm progenitors indicates that the ratio of tumbling to run times, and thus the specific degree of directional persistence of migration, are critical for optimizing migration precision.
Together, our experiments and model provide mechanistic insight into the control of migration directionality for cells moving in three-dimensional environments that combine different protrusion types, whereby the proportion of blebs to actin-rich protrusions determines the directional persistence and precision of movement by regulating the ratio of tumbling to run times.
KeywordsCell migration Protrusion orientation Directionality Persistence Bleb Actin-rich protrusion Run and tumble
Alignment index (a measure of the local persistence)
Maternal zygotic oep
Green fluorescent protein
Automated protrusion analyzer
Polar order parameter
Standard error of the mean
Constitutively active version of Ezrin
Efficient directed migration is assumed to rely on high directional persistence [1, 2, 3]. Indeed, in a stable chemotactic gradient, straight trajectories allow to reach the target in a minimal time. In contrast, lower directional persistence has been associated with poorly directed migration such as in the absence of chemotactic cues or in shallow chemotactic gradients [2, 3]. For instance, the persistence of fibroblasts and dendritic cells has been shown to decrease in presence of a uniform concentration of chemoattractant when compared to migration of the same cells in a chemotactic gradient . Yet, cells undergoing directed migration in vivo often display trajectories with frequent direction changes and low persistence compared to directed migration in culture [5, 6, 7]. Such trajectories have been described as biased random walks or as series of runs and tumbles, i.e., alternating phases with high and low directional persistence [8, 9, 10, 11]. In zebrafish primordial germ cells, whose chemotactic migration during development can be described as a succession of run and tumbles, low persistence and frequent direction changes associated with tumbling have been proposed to fine-tune the migration of these cells, as they progress to intermediate targets during development [9, 12]. However, the cellular mechanisms controlling directional persistence during animal cell migration in vivo are poorly understood, and the functional importance of a proper control of this parameter remains elusive.
Here, we investigate the cellular control and function of directional persistence during cell migration in vivo. We use zebrafish early mesendoderm progenitor cells, which, during early gastrulation, predominantly migrate as single cells and display frequent direction changes . We have previously shown that mesendoderm progenitors can form different protrusion types, including blebs and actin-polymerization driven ones, and that enhancing the formation of blebs decreases migration directional persistence . Therefore, we reasoned that mesendoderm progenitors represented a good model for investigating migration directionality in vivo.
We first show, using an unbiased trajectory analysis algorithm, that lateral progenitors migrating towards the forming body axis alternate run and tumbling phases. We then employ a transplantation assay to investigate how protrusion formation relates to migration directionality during single cell migration of progenitor cells. Using custom-made cell segmentation and protrusion detection software, we show that run phases correlate with the formation of directed actin-rich protrusions, while enhanced blebbing is observed during tumbles. Changing the proportion of blebs to actin-rich protrusions changes the ratio of tumbling to run times. Strikingly, we observe that both decreasing and increasing the ratio of tumbling to run times increase cell dispersion during migration, indicative of reduced migration precision. A theoretical model quantitatively recapitulating the characteristics of progenitor cell migration indicates that an optimal tumbling-to-run ratio enhances migration precision in a changing environment. Together, our experiments and model suggest that the precision of mesendoderm progenitor cell migration depends on the ratio of tumbling to run times, and that this ratio is controlled by adjusting the proportion of blebs to actin-rich protrusions formed by these cells.
Zebrafish lateral mesendoderm progenitors display run-and-tumbling during directed migration
Even though lateral progenitors display mostly single cell migration in early gastrulation , they still transiently interact with neighboring mesendoderm progenitors, which could influence their trajectories. To investigate the migration of these cells in an in vivo environment while avoiding any influence of transient contacts with neighboring cells, we transplanted single mesendoderm cells, into the lateral side of maternal zygotic oep (MZoep) mutant embryos, which lack mesendoderm progenitors . Transplanted cells display directed migration between the yolk and the overlying ectoderm towards the dorsal side of the embryo, as their wt counterparts, but do not have neighboring cells to interact with . Thus, they represent a good model system for the study of single cell migration in a complex in vivo environment. We acquired trajectories of mesendoderm progenitors injected with a fluorescent histone transplanted into MZoep hosts and applied the same automated analysis as described above to their trajectories. We found that, similarly to progenitors transplanted into wt hosts, the cells displayed multi-modal trajectories that can be described as successions of run and tumble phases (Fig. 1f–h). Similar to progenitors migrating in wt hosts, the average ratio of tumbling to run times was 0.68 ± 0.38 (mean ± SD, n = 23 trajectories), instantaneous cell speed was approximately 1.8 times higher during run phases compared to tumble phases (Fig. 1e), and tumbles resulted in a significant direction change, with an average angle between successive runs of 68 ± 37 degrees (mean ± SD, n = 23 trajectories).
Taken together, our analysis indicates that zebrafish mesendoderm progenitors alternate phases of directed migration (runs) and reorientation events (tumbles) during directed migration in vivo.
Protrusion formation during run and tumbling phases
To analyze the orientation of each protrusion type with respect to the direction of cell migration, we developed a new software package for three-dimensional (3D) cell and protrusion segmentation and automated detection and identification of individual protrusions (Automated Protrusion Analyzer (APA), Fig. 2a–c and Additional file 4: Figure S2). Protrusion identification and classification is based on detection of changes in cell surface curvature and morphological differences between protrusion types. APA identifies two types of protrusions: blebs and actin-rich protrusions (Fig. 2b). Actin-rich protrusions are distinguished from blebs by the presence of actin (labeled with Lifeact) in all phases of their expansion (Additional file 3: Movie 1), and by a higher curvature than blebs (Additional file 1: Supplementary Methods). Using APA, we could monitor the center of mass of the cells and each protrusion formed, as well as the intensity of actin in actin-rich protrusions during 3D migration (Fig. 2b, c). As lamellipodia size and actin content have been shown to correlate with migration speed , we analyzed the angle distribution of actin-rich protrusions weighted with the total intensity of the Lifeact signal in the protrusion. Thus, this weighted distribution mostly reflects the orientation of larger actin-rich protrusions. The overall orientation of a specific protrusion type was quantified using the polar order parameter (POP). The POP magnitude indicates how sharply focused the protrusion angle distribution is (Additional file 1: Supplementary Methods).
We then used these automated analysis tools to relate protrusion formation to mesendoderm progenitors’ run-and-tumbling behavior. Run-and-tumbling was evident in 11 out of 17 two-photon high-resolution timelapses (Fig. 2d); in the remaining timelapses, cells displayed directed motion only, likely because the shorter (10–30 min long) high-resolution movies necessary for protrusion analysis are sometimes too short to capture the tumbling behavior. Analysis of the timelapses where run-and-tumbling could be quantified showed that, during run phases, mesendoderm cells formed actin-rich protrusions in the direction of migration (Additional file 5: Movie 2, Fig. 2d–f) and poorly oriented blebs, as evidenced by the clear difference in POP between the two protrusion types (POP = 0.444 ± 0.151 for actin-rich protrusions vs. 0.187 ± 0.197 for blebs in run phases, mean ± standard error of the mean (SEM), Fig. 2f). In contrast, tumble phases were associated with the formation of an increased number of randomly oriented blebs (Fig. 2e) and a decrease in the focus of actin-rich protrusion formation (POP = 0.158 ± 0.132 for actin-rich protrusions formed during tumble phases, mean ± SEM, Additional file 5: Movie 2, Fig. 2f). In about 15 % of the tumble events, less blebbing was observed and a change in direction was achieved by the formation of a new leading edge actin-rich protrusion (corresponding to the two cells labeled as blue data points in Fig. 2e, Additional file 6: Movie 3). Taken together, our observations suggest that actin-rich protrusions may drive directed migration of mesendoderm progenitors whereas blebs mainly contribute to cell re-orientation.
Modulating the proportion of blebs to actin-rich protrusions changes the ratio of tumbling to run times without affecting protrusion orientation
Modulating the ratio of tumbling to run times affects migration precision
To test whether the ratio of tumbling to run times observed in mesendoderm progenitors might indeed optimize migration precision, we developed a stochastic model of cells migrating towards a target moving at constant speed. We represented the moving cells by active Brownian particles randomly switching between run and tumble phases (Fig. 5b, Additional file 1: Supplementary Methods, Additional file 10: Figure S4 and Additional file 11: Figure S5). During run phases cells perform directed active Brownian motion with stochastic speed and a direction fluctuating around a mean value oriented towards the target with a detection error. During tumble phases cells are randomly moving without any preferred direction. We constrained the model parameters by comparing characteristic observables of motion obtained from simulated tracks (analyzed with the same procedure as applied to the experimental data) to experimental measurements. Specifically, several parameters describing cell velocity, as well as run and tumble durations were compared between simulations and experiments. A parameter search yielded a set of parameters very accurately accounting for measured experimental values in control mesendodermal cells (Additional file 1: Table S2 and Additional file 1: Supplementary Methods for details). We found that, with this selected set of parameters, the combined 2D distribution of alignment and cell speed, and the probability distribution of cell speeds during run and tumble phases were well captured by the simulations without further fitting (Fig. 5c, compare to Fig. 1g, and Fig. 5d, e). These observations indicate that the numerical model accurately captures the aspects of cell migration relevant to the observed progenitor trajectories.
Using the estimated parameters, we then systematically varied the run time of the model cells and assessed the precision of cell migration by computing the distance to target and the dispersion of the cell population at the end of the experiment (te = 1.5 h). We found that the distance to target decreased as a function of the run time, indicating that longer runs are more favorable for overall cell velocity. Strikingly, cell dispersion showed a clear minimum around the mean run time measured for control mesendoderm progenitors. This prediction is consistent with the increased cell dispersion measured for CAEzrin and ezrin-MO cells (Fig. 5a), which display run times longer and shorter than control cells, respectively. Taken together, our experiments and model thus indicate that the ratio of tumbling to run times is a critical factor controlling the precision of cell migration in vivo.
Low directional persistence is often thought to be a consequence of a shallow chemotactic gradient resulting in the formation of unfocused protrusions [1, 3]. Here, we show that the directional persistence of zebrafish mesendoderm progenitors migrating in vivo does not depend on the directional focus of protrusion formation, but rather is determined by the ratio of persistent run phases to tumble phases associated with cell reorientation. Interestingly, progenitor cells appear to control the ratio of tumbling to run times by adjusting the proportion of blebs to actin-rich protrusions formed during migration. Blebs have previously been implicated in mediating directed migration of primordial germ cells during zebrafish embryogenesis , and of a number of cancer lines in culture and in vivo [23, 24]. In zebrafish primordial germ cells, bleb growth appears to expand the cell body forward, and subsequent anchoring of the bleb neck to the substrate by adhesive contacts to surrounding cells is thought to drive cell migration . Our finding that blebs in mesendoderm progenitor cells are predominantly associated with tumbling reorientation events suggests that, in these cells, blebs are primarily used for exploring the environment, whereas actin-rich protrusions drive directed migration during run phases. Specifically, undirected bleb formation, as observed during tumble phases, induces displacement of the cell towards random directions and might thus provide a stochastic way of exploring the environment. This difference in bleb function between primordial germ cells and mesendoderm cells may be due to the fact that mesendoderm progenitors form directed actin-rich protrusions, whereas primordial germ cell migration appears to rely exclusively on blebs .
The run and tumbling behavior of control mesendoderm progenitors appears highly similar for cells in wt and in MZoep hosts. Furthermore, our experiments indicate that the ratio of run and tumbling can be modulated in single transplanted cells by tuning the amount of Ezrin activity. To account for experimental variability between different embryos, cells with increased or decreased Ezrin activity were always co-transplanted with control cells in the same MZoep embryo (internal controls) (see also ). These observations indicate that run and tumbling is largely a cell autonomous behavior. Nonetheless, it remains to be investigated whether extracellular factors, such as the distribution, organization and nature of extracellular matrix or the proximity to the chemotactic signal followed by the cells, influence run and/or tumbling in zebrafish mesendoderm progenitors.
Run-and-tumbling is a common feature of bacterial chemotaxis, where it is a strategy for efficient gradient sensing , but has also been observed in a variety of eukaryotic motile cells, including primordial germ cells , chlamydomonas , and mammary epithelial cells . Bacteria are too small to accurately measure a chemoattractant gradient without moving, and use temporal comparisons instead, leading to a biased random walk with longer run phases in the direction of the chemotactic gradient. Animal cells are large enough to polarize in a gradient without motion  and thus alternating run and tumbling phases during migration is likely to serve a different function than in bacterial chemotaxis. It has been speculated that tumble-associated direction changes might increase the precision of chemotactic cell migration in animal cells [12, 21]. Our observation that changing the ratio of tumbling to run times impairs the focus of cell migration provides direct experimental evidence supporting this hypothesis. Indeed, both increasing and decreasing the tumbling to run ratio by modulating the bleb-to-actin-rich protrusion ratio led to impaired cell migration precision (Fig. 5a). Distinct molecular pathways regulate the formation of blebs and actin-rich protrusions [23, 30], suggesting that the ratio between the two protrusion types could be readily tuned. Such a sub-specialization of protrusion function would allow cells to easily modulate the frequency of re-orientation events during migration in complex and changing environments. Our theoretical model, which recapitulated key features of mesendoderm progenitor migration, predicts that an optimal tumble to run ratio enhances migration precision. Indeed, too long runs increase cell dispersion by overly amplifying initial errors in migration direction, whereas too short runs increases cell dispersion because frequent direction changes enhance heterogeneity in direction between cells. Furthermore, it is possible that alternating run and tumbles enhances the robustness of migration to noise in, for example, lamellipodium orientation .
Our experiments and model indicate that mesendoderm progenitors may be operating close to an optimum tumbling to run ratio for precise migration in the in vivo context of the developing zebrafish embryo. Taken together, our data suggest that, by adjusting the proportion of blebs to actin-rich protrusions, mesendoderm cells modulate the ratio of run to tumbling times, and thereby control the precision of their migration. A number of cell types have been reported to combine blebs and actin-rich protrusions during migration [32, 33, 34, 35]. Future studies will have to investigate whether blebs and actin-rich protrusions also have distinct functions in these cells types.
Embryo staging and maintenance
MRNA, morpholino, and dye injection
mRNA was synthesized as previously described . For single cell transplantation, wt TL embryos were injected with 50 pg of Lifeact-GFP , 3.25 ng of Dextran Alexa Fluor-595 (D22913, Invitrogen), and 100 pg of cyc alone (control) or together with 4 ng of ezrin-UTR-MO , to generate ezrin-MO cells or 150 pg of CAEzrin mRNA (T564D of Danio rerio’s gene as in ) to generate CAEzrin cells.
For tracking of cell nuclei in low magnification transplantation experiments, wt donor embryos were injected with 100 pg of cyc together with Alexa Fluor-488 conjugated histone H1 (H13188, Invitrogen) (control), or 100 pg of histoneH2Azf::mcherry plus 150 pg of CAEzrin mRNA (CAEzrin cells). MZoep host embryos were injected with Dextran Alexa Fluor-647 (D22914, Invitrogen) (see also ).
Transplantation experiments, cell imaging, and bleb size measurements
For transplantation experiments, wt and experimental TL donors and MZoep dharma::GFP host embryos were dechorionated with Pronase (2 mg/mL in E2) and transferred onto an agarose plate with E3 medium. Two to three cells were taken from control and experimental donor embryos at dome stage (4.5 hpf) and co-transplanted into the emerging lateral mesendoderm of a host embryo labeled with Dextran Alexa Fluor-647 at 50 % epiboly (5.5 hpf).
For low magnification experiments, time-lapse images were obtained with an upright Leica SP5 confocal microscope equipped with a 20× water immersion lens, using 488-nm Argon, DPSS 561 nm, and 633-nm HeNe laser lines. Frames were captured at 90 s intervals for 3 h (~5.5–8.5 hpf). The temperature was kept constant in all videos (28 °C).
For big magnification transplantation experiments, images were obtained with a Zeiss 710 two-photon microscope equipped with a 63×/1.2 objective, using 910 nm wavelength of the Chamaleon laser. Frames were captured at 10–25 s intervals for 10–30 min, between 6 and 8 hpf.
For bleb size measurements, the projected area of each bleb at its maximal extension was measured using ImageJ and normalized to the projected area of the whole cell.
For cell dispersion measurements, pictures were taken with a dissecting microscope (Olympus SZX 12) equipped with a QImaging Micropublisher 5.0 camera approximately 2 hours post-transplantation.
For single transplanted cells in low magnification movies nuclei tracking in three dimensions (x, y, and z) was performed with Imaris 7.3.0 software. The instantaneous and net speeds, as well as directional persistence (ratio of the net displacement to the distance actually traveled by the cells), were extracted from the tracks.
Analysis of the directions of protrusion formation in combination with cell tracking in higher magnification movies was performed using the APA software, described in Additional file 1: Supplementary Methods.
t tests were performed after the data were confirmed to have normal distribution and equal variance; otherwise, Mann–Whitney U tests were applied. P values were computed in R. For low magnification cell transplantation experiments and variance of cell position (used to assess cell dispersion), one-sided t test was used, which compared experimental data points to an equal sized group of 1. We also computed the P values with ttest2 from Matlab, which compared experimental data points with a random distribution of numbers around one with the same standard deviation as our data. ttest2 yielded similar results and conclusions.
To numerically describe the angular distribution of protrusions, we used the polar order parameter (POP), as explained in detail in Additional file 1: Supplementary Methods. We consider two POP values to be significantly different when their SEMs do not overlap.
Definition of run-and-tumbling phases
For longer trajectories (Figs. 1 and 4e, f), a timeframe of 1.5 min was used as it maximized the amount of embryos we could image simultaneously without a change in the run-to-tumble behavior or in the instantaneous speed. Run-and-tumbling phases were automatically extracted using and unbiased procedure described in Additional file 1: Supplementary Methods [39, 40]. For the analysis of short cell trajectories (timeframe ~10 s, Figs. 2 d–f, 3e, 4 h), “runs” were defined as phases where the trajectory does not deviate by more than 45 degrees from the direction at the beginning of the run or if a change in direction larger than 45 degrees persists for less than 5 timeframes. “Tumbles” were defined as phases where a change of direction higher than 45 degrees occurs and persists for longer than 5 timeframes.
Measurements of cell dispersion
Cell dispersion was assessed using cell position variance, as measured by adding the variances in x and y of the positions of control and experimental cells approximately 2 hours after they were co-transplanted at the same location in a host embryo at 50 % epiboly. Only embryos with at least three control cells and three experimental cells were considered. The ratio has been normalized to transplanted control cells in the same embryo (internal controls) to account for experimental variability between individual transplantation experiments.
We thank K. Lee, C. Norden, A. Webb, and the members of the Paluch lab for comments on the manuscript. We are grateful to P. Rørth and Peter Dieterich for discussions, S. Ares, Y. Arboleda-Estudillo and S. Schneider for technical help, M. Biro for help with programming, and the BIOTEC/MPI-CBG and IST zebrafish and imaging facilities for help and advice at various stages of this project.
This work was supported by the Max Planck Society, the Medical Research Council UK (core funding to the MRC LMCB), and by grants from the Polish Ministry of Science and Higher Education (454/N-MPG/2009/0) to EKP, the Deutsche Forschungsgemeinschaft (HE 3231/6-1 and PA 1590/1-1) to CPH and EKP, a A*Star JCO career development award (12302FG010) to WY and a Damon Runyon fellowship award to ADM (DRG 2157-12). This work was also supported by the Francis Crick Institute which receives its core funding from Cancer Research UK (FC001317), the UK Medical Research Council (FC001317), and the Wellcome Trust (FC001317) to GS.
ADM, PR, CPH, GS, and EP designed the research and wrote the manuscript. ADM performed most of the experiments and the analysis of protrusions; MB helped with analysis and KY helped with experiments. PR performed the run and tumble data analysis and simulations. PR and GS developed the theoretical model. WY developed APA. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Protrusion formation in a single wt transplanted mesendodermal cell. Time-lapse of a wt mesendoderm progenitor (left) and the same cell segmented using APA (right). See also Fig. 1b. Plasma membrane (GPI-RFP) is red; actin cortex (Lifeact-GFP) is green. Scale bar = 10 μm. Time in minutes:seconds. (MP4 21029 kb)
Run and tumbling during migration of a control mesendoderm progenitor. Example time-lapse of a transplanted mesendoderm control progenitor alternating phases of directed migration (run) and phases when the cell temporarily stalls, preceding reorientation (tumbles). Plasma membrane (GPI-RFP) is red; actin cortex (Lifeact-GFP) is green. Scale bar = 10 μm. Time in minutes:seconds. (MP4 14122 kb)
Cell reorientation by formation of a new actin-rich protrusion. Example time-lapse of a transplanted mesendoderm control progenitor changing migration direction by forming a new leading edge actin-rich protrusion. Plasma membrane (GPI-RFP) is red; actin cortex (Lifeact-GFP) is green. Scale bar = 10 μm. Time in minutes:seconds. (MP4 11663 kb)
Bleb formation is enhanced in ezrin morphant single transplanted mesendodermal cells. Example time-lapse of an ezrin-MO injected transplanted mesendoderm progenitor. Plasma membrane (GPI-RFP) is red; actin cortex (Lifeact-GFP) is green. Scale bar = 10 μm. Time in minutes:seconds. (MP4 30553 kb)
Formation of actin-rich protrusions is enhanced in CAEzrin-expressing single transplanted mesendodermal cells. Example time-lapse of a transplanted mesendoderm progenitor expressing CAEzrin. Plasma membrane (GPI-RFP) is red; actin cortex (Lifeact-GFP) is green. Scale bar = 10 μm. Time in minutes:seconds. (MP4 43410 kb)
- 36.Westerfield M. The zebrafish book: a guide for the laboratory use of zebrafish (Danio rerio). 4th ed. Eugene: University of Oregon Press; 2000.Google Scholar
- 41.Gardiner C. Stochastic Methods. Berlin: Springer; 2009.Google Scholar
- 45.Hansen N. The CMA Evolution Strategy: A Comparing Review. In: Lozano JA, Larraaga P, Inza I, Bengoetxea E, editors. Towards a New Evolutionary Computation. Number 192 in Studies in Fuzziness and Soft Computing. Heidelberg: Springer; 2006. p. 75–102.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.