Similarly to spindle MT arrays, F-actin filaments are often arranged in anti-parallel contractile fibers or in more complicated contractile meshworks, for example near the leading edge of migrating cells. Cell migration is powered by a coordinated cycle of leading-edge protrusion in the direction of migration, substrate adhesion of the protrusion, generation of tension on new focal adhesions (FAs) to advance the cell body, and de-adhesion of the trailing cell rear. F-actin is required for each step of the cycle. Spatio-temporally coordinated regulation of the interaction of F-actin with specific binding proteins and myosin motors is required for the actin cytoskeleton to perform such diverse mechanical functions. Because of a lack of appropriate image analysis methods, the dynamics of contractile F-actin structures has been understudied. We, therefore, proceeded and tested the applicability of our IFTA to the investigation of contractile arrangements followed by rapid F-actin meshwork re-arrangements in migrating epithelial cells and similarly highly dynamic neuronal growth cones.
Anti-parallel F-actin flow is immediately expected in migrating cells. Retrograde flow of the dendritic F-actin network in the lamellipodia meets the anterograde flow of cell body F-actin in the so-called “convergence zone”, resulting in anti-parallel F-actin movements and/or F-actin depolymerization. To test whether overlapping contractile F-actin flow patterns exist in normal migrating cells, we re-analyzed previously published FSM datasets of PtK1 cells. Previous analysis of F-actin dynamics (3) identified the presence of two anti-parallel F-actin flow fields: retrograde flow in the lamella and lamellipodia, and forward F-actin movement in the cell body (25). Disappearance of flow vectors in the so-called “convergence zone”, where the two anti-parallel flow fields meet, was interpreted as F-actin disassembly, and previous tracking methods have not provided evidence of any anti-parallel (contractile) F-actin movement in the “convergence zone”. However, visual inspection of these FSM image sequences shows that the method fails to correctly track direction and speed of overlapping F-actin flows (yellow vectors on the inset at the upper right corner of Fig. 2A). In contrast, our IFTA tracker found a considerable overlap of the two interdigitating networks and specifically emerging anti-parallel flows in intracellular regions previously thought to be zones of F-actin depolymerization (side-by-side comparison is presented in Fig. 2A and the inset with yellow vectors at the upper right corner of the figure). In the area marked with a red circle, similarly to (25), we measured a dense anterograde flow of about 2µm/min. In the area thought to be convergent with no flux, we measured two overlapped networks, an anterograde flow (in red vectors) and a retrograde flow (in yellow vectors). It is of significant difficulty to detect any zone of absolute flow convergence in this dataset, i.e., an area of no speckle flux and diffusion of free dimers, as suggested by the measurement depicted by the short speed vectors (in yellow color on the inset) of length close to zero in Fig. 2A (inset taken from reference 25). Thus, although F-actin depolymerization certainly occurs in the so-called “convergence zone” and previously measured depolymerization rates are likely largely overestimated. The reason for this error is that anti-parallel flow fields cancel each other out when regional vector filtering is performed, during the iterative data processing, in previous tracking methods (25). Remarkably, our measurements showed that the two, a retrograde and an anterograde, interlaced speckle flow fields are extending way beyond the so-called “convergence zone”. Our method shows that the two anti-parallel actin speckle flows are clearly interdigitating and fluxing in opposite directions in the areas of overlap, thus suggesting these areas are significantly contractile.
3.2.2. Anti-parallel actin flows in the “convergence zone” at the base of neuronal growth cones
In neuronal growth cones, the F-actin network is organized similarly to the front of migrating cells and we next asked whether IFTA can dissect the arrangement of contractile F-actin networks in migrating neurons. In growth cones, two anti-parallel F-actin networks are thought to merge at the base (26) but visual inspection of published data shows contraction at the narrow entry point at the neck of the growth cone (Video 3, supplemental data). Thus, a “convergence zone” was thought to exist near the axonal shaft as previous analyzes ((27), see Fig. 2B) have not been able to measure the velocities of the F-actin flows at the growth cone neck. Using IFTA, we were able to identify clearly present anterograde and retrograde flows which were interdigitated. We measured heterogeneous flux rates with a plethora of actin speckle speeds, in the range of 1 to 3 µm/min (Fig. 2C). The anterograde flow passing through the neck of the growth cone was oriented straight toward the center of the leading edge exhibiting high level of speed heterogeneity (Fig. 2B, yellow arrows pointing to the left with a distribution shown in yellow on Fig. 2C).
Both results have a very intuitive interpretation as, at the cone, the F-actin polymer meshwork translocates from a narrow zone of high density of parallel speckle motion to a wider zone within the growth cone, hence the vectors are mostly parallel, yet speckles move at high variety of different speeds. Oppositely, the retrograde flow showed narrower slower on average speed distribution (i.e., the high-speed speckles were suppressed) but with a high level of heterogeneity in terms of angular orientation, as the F-actin translocation followed the outer shape of the cell (Fig. 2B, red arrows pointing to the right with a distribution shown in red on Fig. 2C). These results are also intuitive in terms of geometrical interpretation as polymerization rates (i.e., speckle motion speeds) equalize at the entrance of the neck of the cone and, at the same point of entry, the wider network becomes narrower, hence the higher angular homogeneity of the F-actin network entering the neck.
Besides F-actin networks in growth cones in control cells, we analyzed cells after addition of 70 µM blebbistatin. Blebbistatin is a small molecule inhibitor which inhibits the activity of myosin II. After treatment, we measured slower F-actin turnover rates and a reduction of the anti-parallel contractile motion in perturbed cells (data not shown). Again, this result seemed very intuitive given the effects myosin II filaments exert on the actin organization. This provided us with confidence that IFTA can truthfully detect shifts in velocity, both for changes in the speeds and spatial orientation, for a complex organization of flows.
3.2.4. Contractile multi-directional actin flows in the “convergence zone” in the area between the lamella and lamellipodia of epithelial cells with over-expressed tropomyosin-2
Tropomyosins are long coiled-coil dimers that bind along F-actin filaments. In muscle cells, tropomyosins modulate myosin binding and activity (28, 29). In non-muscle cells, increased amounts of tropomyosin alter lamella and lamellipodia dynamics (30), but the underlying molecular mechanisms and, to what extent different tropomyosin isoforms differently affect F-actin dynamics, are not understood. Anti-parallel F-actin flow is myosin-mediated, and it is reasonable to assume that tropomyosins could regulate myosin-mediated anti-parallel F-actin contractions. Indeed, excess tropomyosin in migrating cells has previously been shown to render the F-actin network hyper-contractile (30). We, therefore, used our novel tracking method to test how different tropomyosin isoforms modulate F-actin dynamics. Concretely, we over-expressed two tropomyosin isoforms, one from the α-gene (tropomyosin-2 (Tm2)) and one from the δ-gene (tropomyosin-4 (Tm4)) in PtK1 cells. Visually, we observe highly dynamic, re-organizing F-actin networks. We first applied a tracking method used in (30) for actin speckle flow measurements to successfully detect and decouple the effects of oppositely organized F-actin meshworks (31). It failed to measure the speeds in the area of overlap and displayed speed vectors of, close to, zero (data not shown) as described in section 3.2.2. above. Also, in both cases of over-expression of the two tropomyosin isoforms texture-based tracking (5) yields no interpenetrating speckle fluxes (see the insets with yellow vectors in Fig. 3A and Fig. 3C). We then used IFTA to quantify the effects of tropomyosin over-expression on F-actin dynamics. First, using IFTA, we measured that Tm2 overexpression increases the speed of F-actin flux by about 30%, with speeds ranging from 1.9um/min to 2.5um/min, while preserving the organization of F-actin in retrograde and anterograde flows interdigitating at the so-called “convergence zone”. Exceptionally, we identified examples of four overlapping flows in cells overexpressing Tm2 (Fig. 3B, see the speeds of four vectors flows in the color histogram) which seemed to be coupled in orthogonal pairs (Video 4, supplemental data). None of the four speckle flow fields was a result of stage drift. We established that there are four significant flows by using expectation-maximization based automated selection of the number of vector cluster directions (2, 32). Two of the flows exhibit perpendicular to each other anterograde type of motion (in magenta and yellow in Fig. 3A and Fig. 3B) with faster flow speeds and higher level of heterogeneity and standard deviation of the speed distributions. Oppositely, the two retrograde flows (in red and green in Fig. 3A and Fig. 3B) have slower and narrower speed distributions. Those two flows (the one represented with red arrows on Fig. 3A in particular) seem to stem from overall underlying polymer translocation rather than active F-actin meshwork activity. Their speed distributions have lower standard deviation, the reason for which might be that the whole structure is moving, and with that translocation, all speckles move with very similar speeds. This observation, together with the high edge activity, compared with wild type (control) cells, suggests that Tm2 plays an essential role in the rapid remodeling of F-actin filaments at the edge, yet preserving the shape of the speed distributions. Tm2 acts as an F-actin filament stabilizer and the long F-actin filaments decorated with Tm2 are stable and, therefore, can lead to the formation of protrusions in multiple directions.
Figure 3A (scale bar equals 5µm). Tm2 over-expression increases anti-parallel left-right actin flux and leads also to anti-parallel up-down translocation, i.e., a four-directional actin flow. We identified examples of four overlapping flows in cells over-expressing Tm2 (Fig. 3B shows the corresponding four-color histogram) which seem to be coupled in pairs (see also Video 4, supplemental data). Two of these flow fields exhibit perpendicular to each other anterograde type of motion (in magenta and yellow vectors) with faster flow speeds and higher level of heterogeneity and standard deviation of the speed distribution (Fig. 3B). Oppositely, the two retrograde flows (in red and green vectors) have slower and narrower distributions; those two flows (the one represented with red arrows in particular) seem to stem from an overall underlying polymer translocation rather than an active F-actin meshwork activity as the speed distributions have lower standard deviation, the reason for which might be that the whole structure is moving and, thus, all associated speckles move with very similar speeds. This observation, together with the high edge activity, compared with wild type cells, suggests that Tm2 plays a role in the fast remodeling of F-actin filaments at the edge, yet preserving the shape of the speed distributions. Tm2, thus, works in this scenario like an F-actin filament stabilizer and long F-actin filaments decorated with Tm2 are stable and, therefore, can lead to the formation of protrusions in a different direction. In comparison to IFTA, texture-based tracking (5) yielded no interpenetrating speckle fluxes, but rather displayed vectors of a retrograde flow at the cell edge and a zone of no flux in the lamella, both of which are incorrect (see the inset with yellow vectors on Fig. 3A).
Figure 3C (scale bar equals 5µm). Tm4 over-expression makes actin flux more heterogeneous. In the case of Tm4 over-expression, actin speckles often exhibit two peaks in the distribution. The two “waves” of speckles resemble a distribution in which, first, there is a distribution similar to the wild type speeds and then, there is a second faster peak centered around the average F-actin flux speed for data with over-expressed Tm2. Hence, many of the Tm4 over-expression datasets demonstrated an incorporation of both a wild type behavior and a Tm2 over-expression behavior. We also observed that while the retrograde or anterograde speeds of the F-actin flows were increased with the over-expression of Tm2, the over-expression of Tm4 created visibly different flow patterns with multi-directional flows where a sub-population behaves like wild type, while another one moves faster and in a distinctly heterogeneous way. Tm4 is a weaker actin binder and therefore its over-expression can stabilize nascent F-actin filaments. In the histogram shown on Fig. 3D (in three colors – green, red, and yellow), one can observe the very narrow peak formed by a flow moving parallel to the cell edge (red arrows and red bins). Remarkably, with values ranging 2.17–2.18µm/min, there is a clear indication of a highly organized motion with extremely narrow speed and angular variation and standard variation. In comparison to IFTA, texture-based tracking (5) yields no interpenetrating speckle fluxes, but rather displayed vectors of a retrograde flow at the cell edge and a zone of no flux in the lamella, both of which are incorrect (see the inset with yellow vectors on Fig. 3B).
Overall, in contrast to Tm2 over-expression, the retrograde flow (in green color on Fig. 3C and Fig. 3D) is faster than the anterograde, ranging from values lower than the anterograde to significantly faster. The over-expression of the Tm4 isoform, the more transient actin binder, however, has lesser effects on the organization of the retrograde F-actin meshwork, i.e., a smaller number of organized retrograde flows are present, yet all of them exhibit an increased heterogeneity of the distribution of speckle flux speeds suggesting that Tm2 over-expression fragments at a higher degree the F-actin flows (for instance, resulting in four overall actin flows, from the wild type two flows). Another difference between the two isoforms is that after an over-expression of Tm2, the anterograde flows are faster than the retrograde, while, oppositely, the over-expression of Tm4 leads to a slower anterograde speckle motion.
See Table 1 in Materials and Methods for further information regarding the conditions and perturbations.
3.2.5. Contractile multi-directional actin flows in the “convergence zone” in the area between the lamella and lamellipodia of epithelial cells with over-expressed tropomyosin-4
We also studied the effects of over-expressing Tm4, another isoform. In contrast to Tm2 over-expression, the retrograde flow is on overall faster than the anterograde, ranging from values lower than the anterograde to significantly faster in cells over-expressing Tm4. The retrograde or anterograde speeds of actin speckles for Tm2 over-expression data are normally distributed in a unimodal normal distribution which is faster than wild type data. Oppositely, in the case of Tm4 over-expression, actin speckles often exhibit two peaks in a bi-modal distribution. The two “waves” of speckle vectors resemble a distribution in which there is a first, slower, mode with a distribution similar to the wild type speeds and then there is an additional second, faster, peak centered around the average F-actin flux speed for data with over-expressed Tm2. Hence, many of the Tm4 over-expression datasets demonstrated actin dynamics of embedded both wild type behavior and Tm2 over-expression behavior. We also observed that even when the retrograde or anterograde speeds of the F-actin flows were increased (like in the case of over-expression of Tm2), Tm4 often created distinctly different flow patterns of three-directional flows, some of which have sub-populations behaving like in wild type cells. In cells over-expressing Tm4, we often observed drastic directional changes in the flow organization within the, up to, 10 minutes duration of the acquired movies; in contact, such changes never occurred in cells over-expressing Tm2. Tm4 is the weaker actin binder (33) and therefore its over-expression can stabilize nascent F-actin filaments, which might explain these interesting differences.
Remarkably, in the histogram shown in Fig. 3D (three color histogram distribution), one can observe the very narrow peak formed by a flow moving parallel to the cell edge (red arrows on Fig. 3B); with values ranging 2.17–2.18µm/min, a clear indication of highly organized motion with extremely narrow speed and angular variation. We did not observe any stage drift neither any movement of the cell body, therefore this measurement was not a result of an artifact, but rather represents a very interesting result demonstrating the ability of Tm4 to induce a very coordinated contraction parallel to the cell edge. This is measurement, which could serve as a quantitative readout in functional drug testing in the context of bone remodeling. Osteoclast activity is reduced with lowering the levels of Tm4, leading to the inhibition of bone resorption. Putative anti-aging compounds could be tested ex vivo for their ability to maintain a heterogeneous actin flows perpendicular to the cell edge, while inducing coordinated, fast polymer translocation in parallel to the cell edge. This quantitative assay could be used to evaluate compounds to treat and prevent diseases of the brain ranging from stroke to dementia.
In conclusion, the analysis of Tm2/4 data demonstrates that IFTA can detect differential effects of tropomyosin isoforms on contractility. Firstly, we concluded that over-expression of the Tm2 isoform, which is a stronger binder, causes a massive re-organization of the F-actin flows, resulting in flows in multiple directions with varying, narrower (for the retrograde flows) and wider (for the anterograde flows), speed distributions. Drugs to improve heart muscle and other muscle activity can be tested ex vivo based on this quantitative assay. Secondly, the over-expression of the Tm4 isoform, the weaker actin binder, which is a more transient binder, has lesser effects on the organization, i.e., causes a smaller number of organized flows, yet they seemingly exhibit an increased heterogeneity of the distributions of speckle flux speeds. Drugs to improve bone repair and brain function can be tested ex vivo based on this quantitative assay.
The ability of IFTA to detect subtle differential effects in the organization of F-actin meshworks in areas of overlap demonstrates the power of this quantitative method. Overall, our analysis shows that, besides the two distinct F-actin networks driving the protrusion at the leading edge (31), in migrating cells there are also multiple, distinct contractile F-actin networks in the so-called “convergence zone” at the interface between the lamella and lamellipodia. Our work demonstrates the complexity of the mechanisms regulating the area of the cell where the cell edge ends and the cell body begins, and strongly suggests the existence of putative drug targets.