Reactive Oxygen Species Regulate Protrusion Efficiency by Controlling Actin Dynamics

Productive protrusions allowing motile cells to sense and migrate toward a chemotactic gradient of reactive oxygen species (ROS) require a tight control of the actin cytoskeleton. However, the mechanisms of how ROS affect cell protrusion and actin dynamics are not well elucidated yet. We show here that ROS induce the formation of a persistent protrusion. In migrating epithelial cells, protrusion of the leading edge requires the precise regulation of the lamellipodium and lamella F-actin networks. Using fluorescent speckle microscopy, we showed that, upon ROS stimulation, the F-actin retrograde flow is enhanced in the lamellipodium. This event coincides with an increase of cofilin activity, free barbed ends formation, Arp2/3 recruitment, and ERK activity at the cell edge. In addition, we observed an acceleration of the F-actin flow in the lamella of ROS-stimulated cells, which correlates with an enhancement of the cell contractility. Thus, this study demonstrates that ROS modulate both the lamellipodium and the lamella networks to control protrusion efficiency.


Introduction
Reactive oxygen species (ROS), including superoxide anion (O 22 ), hydrogen peroxide (H 2 O 2 ), and singlet oxygen, are generated as byproducts of biological reactions catalyzed by oxidative enzymes such as xanthine oxidase, cytochrome P-450, mitochondrial oxidases and NADPH oxidases [1,2,3,4]. ROS have been identified as major contributors to biological damage in organisms, acting to irreversibly alter or destroy the function of target molecules or membranes [5]. However, an appropriate or normal production of ROS has been demonstrated to function as an important signaling component and to play a major role in the host defense against bacterial and fungal pathogens [6].
Interestingly, ROS have emerged as important regulators of cell motility. Indeed, ROS derived from NADPH oxidase control colon adenocarcinoma cell migration [7,8] and studies in vascular smooth muscle cells (VSMC) and in endothelial cells have shown that growth factors-stimulated migration required ROS generation [9,10]. Conversely, the inhibition of ROS production by a NADPH oxidase inhibitor diphenylene iodonium (DPI) or by deletion of Nox1, a member of the NADPH oxidase family, has been reported to reduce the speed and directionality of migrating cells [8,11]. Recent works in zebrafish revealed the existence of a tissue-scale gradient of H 2 O 2 induced by wounding and required for leukocyte migration toward the wound, highlighting the role of H 2 O 2 as an important signal to direct cell motility [12].
Migrating cells respond to chemoattractant gradients by adopting a polarized morphology, with their leading edge oriented in the direction of the gradient. Leading edge protrusion is thought to initiate migration and set the direction of movement. Cells with weak polarity lose the ability to protrude in a single direction, resulting in random trajectories and reduced migration speed. Protrusion of the membrane is tightly coupled to actin polymerization, both through the elongation of pre-existing filaments and by nucleation of new filaments. Substantial evidence has shown that the Arp2/3 complex mediates new actin nucleation and branching observed in lamellipodia [13], although other data have implicated the action of formins [14] and filamin [15]. Other major regulators of leading edge actin dynamics are cofilin and ADF [16]. Cofilin/ADF act both to increase the number of free barbed ends available for actin elongation via their severing function, and to enhance the turnover of existing filaments necessary for actin remodeling via their depolymerizing function.
Previous works in epithelial cells defined two dynamically, molecularly, and functionally distinct F-actin networks at the cell leading edge: the lamellipodium and the lamella [17]. The lamellipodium is defined by a fast treadmilling F-actin network [18,19] characterized by a thin band of high F-actin polymerization at the leading edge, adjacent to a similarly thin band of strong disassembly. In contrast, the lamella is dominated by random patterns of alternating F-actin assembly and disassembly puntae and is characterized by a slow actin retrograde flow. Important molecular differences between the two F-actin networks have been revealed by pharmacological and biochemical analysis. The lamellipodium is enriched in Arp2/3 and cofilin, both responsible for the fast actin treadmilling [19]. The lamella, however, is composed of myosin II and tropomyosin that regulate the contractile machinery responsible for the F-actin retrograde flow in this region. Tropomyosin suppresses the interaction of cofilin with F-actin and blocks Arp2/3 nucleation [20]. Recent studies have shown that microinjection of skeletal muscle tropomyosin selectively depleted the lamellipodium from the cell protrusion [21]. Surprisingly, cells without a lamellipodium were still highly motile, thus suggesting that the lamellipodium plays more a role in directional sensing and persistence.
Interestingly, ROS can influence F-actin dynamics both directly and indirectly. On the one hand, direct oxidation of actin by ROS has been shown to affect polymerization. Treatment of actin in vitro with high concentrations of H 2 O 2 decreased the polymerization and elongation rates of actin filaments [22]. Other studies have shown that migrating cells produced ROS at the membrane ruffles which in turn increased actin polymerization [11,23]. On the other hand, cytoskeletal rearrangements could result from indirect regulation of actin dynamics by ROS, translated through redox-sensitive enzymes. As such, the oxidation of the lowmolecular weight protein tyrosine phosphatase gives rise to a decrease in RhoA activity required for Rac-induced formation of membrane ruffles [24]. In addition, the ROS-dependent activation of the cofilin phosphatase slingshot-cofilin pathway via H 2 O 2mediated oxidation of 14-3-3zeta was shown to stimulate the formation of cofilin-actin rods [25].
Despite these findings, the precise effect of ROS on actin dynamics in vivo is still not clearly defined. The study presented here aimed at deciphering the molecular mechanisms controlling F-actin dynamics in migrating cells in response to H 2 O 2 stimulation, which is critical to understand how ROS promote cell motility.

H 2 O 2 Regulates Cell Migration and Protrusion Dynamics in PtK1 Cells
For our studies, we used mammalian PtK1 cells, derived from marsupial kidney epithelium, that have been shown to be ideal for imaging studies as they polarize in a large and thin cell front [17,18,21,26,27,28]. We developed a chemotaxis assay to evaluate the effects of H 2 O 2 on PtK1 cell migration. To this end, cells were exposed to a continuous flow of H 2 O or H 2 O 2 -containing solution from the tip of a microinjection needle for 45 min ( Figure 1A-flow indicated in red). Cells present in the same imaging field but not in contact with the flow were used as control ( Figure 1A-no flow). Analysis of individual cell tracks revealed that cells exposed to H 2 O 2 displayed similar total path lengths, but increased net path lengths and directionality compared with control cells (Figure 1B-D). Moreover, cells exposed to H 2 O 2 displayed similar migration rates compared to control cells ( Figure 1E). These results indicate that PtK1 cells are more directional in presence of H 2 O 2 and are consistent with data obtained in zebrafish in which H 2 O 2 acts as an immediate signal that attracts leukocytes to a wound site [12].
During the chemotaxis assays, we observed that cells exposed to H 2 O 2 seemed to protrude more than control cells. Thus, we investigated how H 2 O 2 treatment affected protrusion dynamics of PtK1 cells. First, we wanted to confirm that H 2 O 2 was incorporated within the cells following the treatment. To this end, PtK1 cells were incubated with the H 2 O 2 -specific probe PY1-AM (Peroxy-Yellow 1 Acetoxymethyl-ester) [29,30], and treated with 500 mM of H 2 O 2 for 20 min. Cells were visualized by epifluorescence microscopy before and after H 2 O 2 addition ( Figure 1F). We observed an increase in PY1-AM fluorescence intensity after stimulation with H 2 O 2 which confirmed that H 2 O 2 was present in these cells. Although the probe uptake was not homogenous in the entire cell island ( Figure 1F), PY1-AM fluorescence intensity ratio after vs. before H 2 O 2 addition was 1.6960.36 at the front of the cell island and 1.1360.15 at the back, thus indicating that H 2 O 2 was incorporated in the cells at the edge of PtK1 islands before the cells located inside the islands ( Figure 1F, right panel).
We next analyzed the protrusion behavior of cells in response to H 2 O 2 using phase-contrast live-cell microscopy. PtK1 cells were starved overnight and then incubated either with control media ( Figure 1G, top row) or with media containing 500 mM H 2 O 2 ( Figure 1G, center row) for 45 min (Videos S1, S2). Analysis of protrusion dynamics by kymographs ( Figure 1G, right panels) revealed that cells treated with H 2 O 2 exhibited a wider and more persistent protrusion ( Figure 1H-I) compared to control cells. Analysis of protrusion velocity demonstrated that control cells or cells treated with H 2 O 2 protruded faster than they retracted and had similar protrusion and retraction velocities ( Figure 1J). Finally, we analyzed whether the effects of H 2 O 2 on cell protrusion can be abrogated by sodium pyruvate, a well-characterized H 2 O 2 scavenger [31]. Our data indicated that the protrusion parameters tested in cell treated with sodium pyruvate alone were not significantly different from control cells ( Figure 1H-J). However, the addition of sodium pyruvate in cells treated with H 2 O 2 induced a strong decrease in the protrusion width and persistence compared to H 2 O 2 -treated cells ( Figure 1G, bottom row, Figure 1H-I, Video S3), no effect was detected on protrusion and retraction velocities compared to control or H 2 O 2 -treated cells ( Figure 1J). Combined, these results indicate that H 2 O 2 treatment induces the formation of a persistent protrusion in PtK1 cells.

H 2 O 2 Affects Actin Dynamics in PtK1 Cells
Protrusive activity is driven by a tight regulation of the actin cytoskeleton at the front of the cells. To determine the effects of H 2 O 2 treatment on actin dynamics at the leading edge of PtK1 cells, we used quantitative fluorescent speckle microscopy (qFSM) [32] (Figure 2). Cells were injected with X-rhodamine-labelled of motility parameters of PtK1 cells exposed or not to a flow of H 2 O (control) or H 2 O 2 , including total path length (B), the total distance traversed by cells over time; net path length (C), the net distance that the cells traversed from the first to the last frame; directionality (D), the ratio of net to total path length; and cell velocity (E). The data result from n $38 cells analyzed for each condition. Error bars represent s.e.m. ***, p,0.001 compared to control (flow) and H 2 O 2 (no flow). (F) PtK1 cells were incubated with the H 2 O 2 -sensitive probe PY1-AM (5 mM for 30 min) and then treated with H 2 O 2 (500 mM for 20 min). One representative image is shown before and after H 2 O 2 stimulation. Red lines highlight the leading edge of the cells. The scale bar is 30 mm. In the right panel, fluorescence intensity of PY1-AM was measured in cells located at the front and the back of PtK1 islands before and after H 2 O 2 addition. Fluorescence intensity was averaged from 15 cells. Error bars represent s.e.m. **, p,0.01 compared to -H 2 O 2 (front) and + H 2 O 2 (back). (G) After starvation, PtK1 cells were incubated for 15 min in media only and 45 min with control media (+H 2 O) or media containing 500 mM H 2 O 2 alone or in combination with 5 mM sodium pyruvate, a ROS scavenger. In this case, cells were pretreated with ROS scavenger for 30 min before the experiment. Phase-contrast images were taken from movies of each condition. Images of control (top row), H 2 O 2 -(center row) and actin, then starved for 3 h, and FSM time-lapse series were acquired at 10 sec intervals for 10 min (Figure 2A and Videos S4, S5). Kymograph analysis of F-actin velocity indicated that, in control cells, F-actin underwent a slow retrograde flow in the lamella (0.6 mm/min, Figure 2B, top panel, and Figure 2E) and a faster retrograde flow in the lamellipodium (1.1 mm/min, Figure 2B, top panel, and Figure 2D), as previously observed in other studies [17,18,21,26]. In comparison to control cells, cells treated with H 2 O 2 displayed a fast F-actin retrograde flow not only at the leading edge but also throughout the entire protrusion ( Figure 2B, bottom panel). Quantification of the F-actin flow rates using kymographs indicated that cells treated with H 2 O 2 had an increase in the flow speed at the leading edge by ,1.6 fold (1.75 mm/min) compared to control cells (1.1 mm/min) ( Figure 2D), while at 5 mm from the leading edge, the retrograde flow was increased by ,1.8 fold (1.1 mm/min) compared to unstimulated control cells (0.6 mm/min) ( Figure 2E). These data were supported by spatially resolved maps of F-actin flow speed which showed an increase in the flow speed ( Figure 2C) and a widening of the region of fast F-actin flow at the leading edge of H 2 O 2 -treated cells compared with untreated cells ( Figure 2C). Together, these results indicate that H 2 O 2 regulates actin dynamics in PtK1 cells and induces the formation of a wide region of fast actin retrograde flow.

H 2 O 2 Induces Cofilin Activation and Increases the Formation of Polymerization-competent Free Barbed Ends at the Cell Edge in PtK1
To investigate whether the F-actin dynamics phenotype observed in H 2 O 2 -treated cells resulted from a regulation of the lamellipodium and/or the lamella, we next examined the distribution of the molecular markers of both regions. We first examined the localization of endogenous inactive phosphorylated cofilin (P-cofilin), a lamellipodium marker, in cells unstimulated or treated with 500 mM of H 2 O 2 for various periods of time. In control cells (unstimulated), P-cofilin localized in diffuse punctae throughout the cell edge and within the cell body ( Figure 3A). Upon H 2 O 2 stimulation, we observed a gradual decrease of Pcofilin from the leading edge throughout the protrusion ( Figure 3A). Quantification of P-cofilin relative to F-actin fluorescence intensities from the leading edge toward the cell center indicated a ,1.9 fold decrease in P-cofilin by 15 min that reached a 3 to 4 fold decrease by 30-60 min at the cell edge, as compared to unstimulated cells ( Figure 3B-D and Figures S1-S2A). Quantification of F-actin fluorescence intensity revealed a slight increase of F-actin at the leading edge of cells treated with H 2 O 2 for 15-30 min ( Figure 3C). We next analyzed by western blot the P-cofilin and cofilin expression levels in total cell lysates from cells treated or not with H 2 O 2 . As shown in Figure 3E, the level of P-cofilin was slightly decreased after 30 min of H 2 O 2 treatment, followed by a significant decreased at 45-60 min upon stimulation, as previously observed [25]. The level of cofilin was similar in all conditions ( Figure 3E). Quantification of the ratio Pcofilin/cofilin revealed a decrease of 70% and 95% in cofilin phosphorylation after 45 and 60 min of H 2 O 2 stimulation, respectively, compared to untreated cells ( Figure 3F). We next examined the effect of sodium pyruvate on the level of P-cofilin upon H 2 O 2 stimulation. As shown in Figure 3E-F, the presence of ROS scavenger strongly abrogated the cofilin dephosphorylation observed upon H 2 O 2 stimulation in PtK1 cells, thus confirming the specific effect of H 2 O 2 on cofilin activation. We previously observed that H 2 O 2 was incorporated in the cells located at the edge of PtK1 islands before the cells present inside the islands ( Figure 1F). This suggests that cofilin was activated first in the cells located at the edge of cell islands, which were analyzed by immunofluorescence. This could explain the cofilin activation delay observed between immunofluorescence and western blot, in which the whole population of cells was analyzed. Taken together, these data indicate that cofilin is activated upon H 2 O 2 stimulation in PtK1 cells and that this activation occurs at the cell edge starting at 15 min after treatment.
Since cofilin is thought to supply polymerization-competent free barbed filament ends to the lamellipodium [33], we further analyzed the localization and density of free barbed ends upon H 2 O 2 treatment in PtK1 cells. In starved control cells (unstimulated), few free barbed ends were observed at the cell edge ( Figure 4A). In contrast, the concentration of free barbed ends was slightly increased after 15 min of H 2 O 2 treatment and dramatically increased after 30-60 min in a narrow rim along the leading edge ( Figure 4A). Quantification of the fluorescence intensity of barbed ends to F-actin ratio confirmed that free barbed ends were increased at the cell edge by ,1.8 fold after 15 min and by ,3 fold after 45 and 60 min of treatment ( Figure 4B-D and Figure  S2B). These data were consistent with the increase of cofilin activity we observed previously ( Figure 3E-F). As described above, quantification of the fluorescence intensity of F-actin slightly increased after H 2 O 2 stimulation ( Figure 4C). Free barbed filament ends localized in a 1.3 mm region along the leading edge ( Figure 4B-D). The width of the barbed ends area was not significantly modified upon H 2 O 2 treatment. To further support these results, we analyzed the spatial organization of F-actin assembly-disassembly rates in PtK1 cells treated or not with H 2 O 2 . The lamellipodium is typically defined by a narrow band of polymerization adjacent to a narrow band of depolymerization [17]. Actin turnover maps, obtained from FSM time-lapse series, revealed that control cells had a mixed region of both polymerization (bright red punctae) and depolymerization (bright green punctae) events along the cell edge ( Figure 4E). In starved cells, the lamellipodium is so thin ( Figure 2B, top panel) that the integration of all the frames abolished the distinction between local assembly/ disassembly bands. In contrast, H 2 O 2 -treated cells exhibited a wider band of strong polymerization along the leading edge juxtaposed to a similar band of rapid F-actin depolymerization ( Figure 4E, bright red and bright green, respectively), indicative of a fast treadmilling lamellipodium. Altogether, our data suggest that H 2 O 2 treatment induces the formation of polymerization-competent free barbed ends and increases F-actin turnover at the edge of PtK1 cells.

H 2 O 2 Induces Arp2/3 Recruitment and ERK Activation at the Leading Edge of PtK1 Cells
Several studies suggest a synergy between the severing activity of cofilin and the branching activity of Arp2/3 to regulate actin dynamics in the lamellipodium [34,35,36]. The Arp2/3 complex drives leading edge protrusion by nucleating actin monomers to assemble a dense meshwork of short branched filaments [37]. Thus, we analyzed the distribution of the Arp2/3 complex in PtK1 cells stimulated with H 2 O 2 for up to 60 min. In unstimulated control cells, p34-Arc, a subunit of the Arp2/3 complex, was detected throughout the cell ( Figure 5A). We observed a dramatic increase of p34-Arc recruitment to the cell edge after H 2 O 2 treatment ( Figure 5A). Quantification of the fluorescence intensity from the leading edge toward the cell center confirmed that p34-Arc increased in density at the leading edge of H 2 O 2 -treated cells as soon as 15 min after stimulation, and with the highest concentration at 30 min (,4 fold above control) ( Figure 5B-D and Figure S2C). Thus, H 2 O 2 treatment induced the recruitment of the Arp2/3 complex at the leading edge of PtK1 cells.
Interestingly, a recent study has shown a link between the regulation of Arp2/3 and ERK (extracellular signal-regulated kinase) signaling. Indeed, the activity of ERK specifically regulates protrusion initiation through the phosphorylation of the WAVE2 regulatory complex, which is required for its functional binding and activation of Arp2/3 [38]. Thus, we analyzed by western blot the activity of ERK in PtK1 cells treated with 500 mM of H 2 O 2 for 0 to 60 min. Of note, the western blot with pERK antibody ( Figure 6A) showed two bands at 42 and 44 kDa. This antibody recognized both pERK1 (top band, 44 kDa) and pERK2 (bottom band, 42 kDa). Upon H 2 O 2 stimulation, pERK level increased at 15 min compared to control cells and by 30 min the level of pERK decreased to the basal level of control cells ( Figure 6A). The ERK protein level did not change during the time of stimulation ( Figure 6A). Quantification of the ratio pERK/ERK indicated a notable ,1.35 fold increase in ERK phosphorylation after 15 min of H 2 O 2 stimulation while no significant activation was detected at 30-60 min compared to untreated cells ( Figure 6B). Conversely, addition of ROS scavenger to the culture media abrogated the ERK activation observed in the presence of H 2 O 2 alone ( Figure 6A-B). We next examined the localization of active phosphorylated ERK in starved PtK1 cells stimulated with H 2 O 2 . In unstimulated condition, pERK was distributed throughout the cell ( Figure 6C). After 15 min of treatment, pERK signal was increased not only at the cell edge but also throughout the entire protrusion and at 30-60 min, pERK signal decreased within the cell body but was still detected in punctae at the cell edge ( Figure 6C). Quantification of the pERK/F-actin fluorescence intensity ratio confirmed a significant increase of 1.5 fold in pERK localization at the edge of H 2 O 2 -treated cells compared to control cells ( Figure 6D-F and Figure S2D). Of note, we observed that other MAPK, such as p38 and JNK, were activated during H 2 O 2 stimulation ( Figure S3F-G). However, since ERK signaling has been involved in the regulation of protrusion dynamics, we decided to further investigate its role in response to H 2 O 2 . To this end, we treated the cells with a MEK inhibitor, UO126. We analyzed by western blot the pERK and ERK expression levels in total cell lysates from cells treated with UO126 and stimulated  with H 2 O 2 . As shown in Figure S3A, the level of pERK was completely decreased after 15 min of H 2 O 2 treatment and a slight increase was observed at 60 min. The level of ERK was similar in all conditions ( Figure S3A). We next analyzed the protrusion dynamics as described above. We observed that inhibition of ERK in  Figure S3E and Figure S2H). Analysis of F-actin fluorescence intensity revealed no effect on F-

H 2 O 2 Regulates the Distribution of Lamella Markers and the Contractile Machinery in PtK1 Cells
Using FSM, we showed that H 2 O 2 treatment dramatically increased the F-actin retrograde flow from the leading edge to further inside the protrusion (Figure 2). This phenotype could result from (i) a widening of the lamellipodium, (ii) an increase in F-actin retrograde flow in the lamella, or (iii) an increase in both the lamellipodium and the lamella. To differentiate between these  We analyzed the localization of myosin IIA and tropomyosin, which are exclusively associated with the lamella and involved in the regulation of the contractile machinery [17,21]. In control cells (0 min H 2 O 2 ), myosin IIA was excluded from the lamellipodium and appeared in a gradient of punctae within the lamella ( Figure 7A). Upon H 2 O 2 treatment, myosin IIA was depleted farther from the cell edge at 30-60 min after stimulation ( Figure 7A). These observations were confirmed by quantification of the fluorescence intensity of myosin IIA ( Figure 7B), measured from the cell edge into the cell center. The ratio myosin IIA/Factin ( Figure 7C-D and figure S2E) indicated that in control cells, myosin IIA was localized at ,0.8 mm from the cell edge, while after 30 min of treatment, myosin IIA was depleted in a ,1.8 mm area from the leading edge and decreased by 1.4 fold in intensity compared to control cells. At 45 and 60 min after addition of H 2 O 2 , myosin IIA was reduced in a 3 to 3.5 mm region from the cell edge, respectively. These data indicate that H 2 O 2 displaced myosin IIA from the cell edge. We next analyzed by immunofluorescence the localization of high-molecular weight isoforms of tropomyosin, which regulate myosin II interaction with F-actin as characterized in skeletal muscle [39]. In control cells or cells treated with H 2 O 2 for 15 min, we observed that tropomyosin localized throughout the cell ( Figure 7E). After 30-60 min of stimulation, tropomyosin was drastically reduced from the cell edge ( Figure 7E). Quantification of the fluorescence intensity ratio of tropomyosin/F-actin from the leading edge toward the cell center confirmed our observation and indicated that at 30-45 and 60 min of stimulation, tropomyosin was reduced by ,1.8 to 2.7 fold at the leading edge of PtK1 cells compared to untreated cells and was depleted in the first micrometers adjacent to the cell edge ( Figure 7F-H and Figure S2F). Thus, upon H 2 O 2 treatment, myosin IIA and tropomyosin were depleted much farther from the cell edge compared to unstimulated cells, suggesting a widening of the lamellipodium.
To test this hypothesis, we next treated the cells with blebbistatin, a nonmuscle myosin II ATPase inhibitor [40], and analyzed by qFSM the F-actin dynamics in the protrusion. These results indicate that, in H 2 O 2 stimulated cells, the region of fast F-actin flow further away from the leading edge is myosin IIsensitive and thus correspond to an increase in the lamella F-actin dynamics.
To confirm these data, we measured the width of the lamellipodium in all the conditions. The results shown in Figure 8D  Since the F-actin flow rate was enhanced in the lamella during H 2 O 2 stimulation, we next determined whether this was due to an increase of the cell contractility in this region. Therefore, we examined by immunofluorescence the localization of serine 19phosphorylated myosin II regulatory light chain (pMLC), an indicator of myosin II activity [41]. In control cells and in cells stimulated for 15 min with H 2 O 2 , pMLC was distributed throughout the cell. At 30-60 min, we observed a net decrease of pMLC localization at the cell leading edge (Figure 8E), confirming the expansion of the lamellipodium upon H 2 O 2 stimulation. As mentioned above, the fluorescence intensity of Factin was slightly enhanced during the time of stimulation compared to control cells ( Figure 8G). Interestingly, we observed an increase of pMLC in the lamella of H 2 O 2 -treated cells. These observations were confirmed by quantification of the pMLC/Factin fluorescence intensity ratio indicating a ,1.3 fold increase at 5 mm from the leading edge in cells stimulated for 15-30 and 45 min compared to control cells ( Figure 8F-H and Figure S2G). These results indicate that H 2 O 2 treatment increases the F-actin retrograde flow in the lamella by activating the contractile machinery of the cells.

Discussion
Productive protrusions allowing motile cells to sense and migrate toward a chemotactic gradient require a tight control of the actin cytoskeleton. Over the past years, substantial evidences indicated times. The scale bar is 10 mm. (D and E) Fluorescence intensity of pERK (D) and F-actin (E) in cells treated with H 2 O 2 500 mM, measured from the cell edge (0 mm) into the cell center (10 mm). (F) pERK/F-actin fluorescence intensity ratio in cells treated with H 2 O 2 500 mM, measured from the cell edge (0 mm) into the cell center (10 mm). In (D)-(F), the data shown represent one experiment and are averaged from at least 13 cells for each condition. The experiment was repeated three times with similar results ( Figure S2D). Protrusion width (G), persistence of protrusion (H) and protrusion/retraction velocities (I) in starved Ptk1 cells incubated for 45 min with control media (+DMSO) or media containing 500 mM H 2 O 2 alone or in combination with UO126, a MEK inhibitor. In (G)-(I), the data shown result from the analysis of at least 23  have been accumulated involving ROS in the regulation of the actin network reorganization [11,42,43,44]. However, the precise mechanisms of how ROS affect cell protrusion are not well elucidated yet. Detailed analysis of the spatio-temporal regulation of actin dynamics relies on the use of state-of-the art microscopy techniques, such as fluorescent speckle microscopy. One limitation of this technique is the requirement for the cell lines to form flat protrusions during the migration process, excluding neutrophils. To study the effects of ROS on cell migration and dynamics of the cell leading edge, we thus used PtK1 cells in which F-actin organization, kinetics, and kinematics have been extensively characterized. To validate the choice of PtK1 cells, we examined their response to an external source of ROS. Consistent with the studies discussed above, we showed that the presence of H 2 O 2 modulated the migration of PtK1 cells by increasing their directionality. In addition, ROS stimulation in these cells induced the formation of a persistent protrusion.
Signals that promote cell motility are often correlated with increased cofilin activity, while those inhibiting motility enhance cofilin phosphorylation [16,45]. Cofilin is a crucial mediator of actin cytoskeletal dynamics: it severs F-actin filaments, generating substrates for Arp2/3-mediated branching activity, and promotes F-actin depolymerization, leading to the recycling of actin monomers at the leading edge for further polymerization [16,19,36,45,46,47,48,49]. Cofilin has been shown to control the directionality of motile cell in a nonmetastatic tumor cell line (MTC) and in fibroblasts [50,51]. Furthermore, recent studies reported that ROS induced the activation of cofilin (i.e. dephosphorylation) in HeLa or VSMC [25,52,53] leading to an increase of cell migration. In contrast, chemotaxis was decreased in oxidatively stressed human T cells where oxidized cofilin was dephosphorylated but did not mediate F-actin depolymerization [54]. In our study, we observed a significant increase of cofilin activation during H 2 O 2 stimulation that correlates with enhanced Arp2/3 localization at the cell edge, increased free barbed ends formation, and faster actin retrograde flow in the lamellipodium, as previously described [26]. Conversely, inhibition of ROS has been shown to block actin assembly and reduce actin retrograde flow in growth cones [55].
The precise spatio-temporal regulation of cofilin activity, as well as its local concentration within the cell leading edge are critical for cell migration [56,57]. Studies in vitro revealed that active cofilin at low concentration leads to severed actin filaments; at higher concentration, active cofilin binds cooperatively to actin filaments without severing them; finally a very high concentration of active cofilin induces de novo nucleation that results in actin filament assembly [56]. In vivo, this can be complicated as levels of active cofilin can vary from cell-to-cell and depend on stimuli. H 2 O 2 treatment in PtK1 cells induced an increase of active cofilin that correlates with faster F-actin turnover at the cell edge and results in a persistent protrusion. Interestingly, overexpression of active cofilin S3A in PtK1 cells was shown to decrease protrusion efficiency even though F-actin treadmilling was enhanced [26]. Altogether, these results may be explained by the relative difference in the concentration of local active cofilin in cells. Upon H 2 O 2 stimulation, our results suggest that the cells have achieved a balance of cofilin activities leading to well structured actin networks and efficient chemotactic cells. Overexpression of active cofilin, however, induced a decoupling of lamellipodium/ lamella actin networks and affected the protrusion efficiency.
Other signaling pathways downstream of ROS may also be involved in the regulation of actin dynamics. Among them, the ERK-MAPK pathway has been implicated as a signaling cascade by which cells react to H 2 O 2 [58,59,60]. Importantly, recent studies have discovered that active ERK at the cell edge is critical for cell migration [38]. ERK controls the productive advancement of the cell edge by activating WAVE2 regulatory complex, required for the activation of Arp2/3 complex and subsequent actin polymerization. Our study reports an increase of active ERK at the leading edge after H 2 O 2 stimulation and this activation is involved in the formation of a persistent protrusion. In addition, we showed that ERK activity contributes to the recruitment of Arp2/3 at the cell edge which may then works in synergy with cofilin [34,35,36] to increase the F-actin retrograde flow in the lamellipodium of H 2 O 2 -treated cells. This finding is particularly interesting as a recent work in zebrafish larvae has demonstrated that H 2 O 2 induces direct oxidation-mediated activation of Lyn, a member of Src-family kinases, which in turn activates ERK at the cell edge of neutrophils [61]. This event appears to be critical to mediate leukocyte migration toward wounds in response to H 2 O 2 . Since the activity of p38 and JNK, two MAPK family members involved in the regulation of actin dynamics [62,63], is increased upon H 2 O 2 treatment in PtK1 cells ( Figure S3F-G), it would be interesting to further investigate their contribution in protrusion dynamics.
Cell motility requires the coordination of the lamellipodium and lamella networks. Ponti et al. suggested that the lamella is necessary for the cells to migrate whereas the lamellipodium could serve as an exploratory function [17]. Interestingly, in addition to the effect of H 2 O 2 on the lamellipodium, we also observed an increase of the actin retrograde flow in the lamella that coincides with enhanced myosin contractility. Importantly, several regulators of the contractile apparatus are ROS sensitive, which might explain the increase of MLC phosphorylation observed in the protrusion of H 2 O 2 -treated cells. The Rho/ROCK (Rho-associated protein kinase) pathway is activated by ROS in both aorta and VSMC, thus promoting MLC phosphorylation by inhibiting the regulatory subunit of MLC phosphatase [64,65]. In addition, Pak, another major regulator of myosin contractility that can directly phosphorylate MLC [66], is activated by ROS [67]. The increased actin retrograde flow in the lamella observed upon H 2 O 2 stimulation could thus result from Pak activation.
Our current study indicates that H 2 O 2 increases the protrusion efficiency by modulating both the lamellipodium and lamella.  Interestingly, we have recently shown that Paks regulate cell migration by modulating F-actin retrograde flow in both networks [27]. Further investigation to identify the direct targets of H 2 O 2 controlling the molecular mechanisms that regulate the two actin networks will provide a better understanding of the role of ROS in cell migration.

Chemotaxis Assay
Using a microinjection system (Eppendorf), PtK1 cells were exposed or not to a constant flow of H 2 O (control) or 1.5 mM H 2 O 2 mixed with rhodamine dextran (0.5 mg/ml). Phase-contrast time-lapse image series were acquired at 10 s intervals for 45 min using a 10X/0.25 phase objective lens (Olympus) on an inverted microscope (Olympus IX70) equipped with a Cool SNAP HQ camera. Images of rhodamine dextran were acquired every 2 min to localize the flow and determine which cells are exposed to the flow (Flow). Cells present in an area not exposed to the flow (No flow) were considered as controls. Movements of individual cells were tracked using MetaMorph software to calculate velocity and motility parameters. To localize and quantify the relative number of actin filament free barbed ends, live cells were permeabilized with 0.25 mg/ml saponin in the presence of 0.5 mM X-rhodamine actin and fixed as previously described [69]. Epifluorescence images of fixed cells were acquired on an inverted microscope (Eclipse TE 2000-U, Nikon) equipped with an electronically controlled shutter, filter wheels, and a 14-bit cooled CCD camera (Cool SNAP HQ, Photometrics) controlled by MetaMorph software (Universal Imaging Corp.) with a 60X/1.4 NA Plan Apo DIC objective lens (Nikon).

Immunofluorescence Analysis
Quantification of the fluorescence of myosin IIA, pMLC, tropomyosin, p34-Arc, P-cofilin, free barbed ends, pERK and Factin as a function of the distance from the cell edge was obtained with custom software written in Matlab (MathWorks). Bands of constant distance to the cell edge were constructed, and individual fluorescence intensities were accumulated and averaged in each band to produce graphs representing fluorescence intensity versus distance from the leading edge (see Figure S1 for details).

Fluorescent Speckle Microscopy
PtK1 cells were plated on coverslips three days prior to experiments and X-rhodamine-conjugated actin, prepared as described previously [68], was injected into cells at 1 mg/ml. Cells were starved for ,3 h after injection and incubated for ,30 min in control-or H 2 O 2 500 mM-containing medium in which oxyrase (Oxyrase Inc., Mansfield, OH) has been added to inhibit photobleaching. For myosin IIA inhibition, cells were incubated in culture medium containing 50 mM blebbistatin (Calbiochem, Philadelphia, PA) or blebbistatin + H 2 O 2 500 mM for 30 min. Actin fluorescent speckle microscopy time-lapse series were acquired at 10 s intervals for 10 min using a 60X(X1.5)/ 1.49 NA Plan Apo objective lens (Nikon) on a spinning disk confocal microscope system equipped with a Cool SNAP HQ camera.

Image Analysis and Quantification
F-actin flow rates at the leading edge of the cells were measured by kymograph analysis as previously described [70]. Five randomly placed lines normal to the cell edge were used to construct kymographs for each cell. FSM time-lapse image series were analyzed using the fsmCenter software package written in phalloidin staining (red) in starved PtK1 cells treated with H 2 O 2 500 mM for the indicated times. The scale bar is 10 mm. Red lines highlight the leading edge of the cells. (F and G) Fluorescence intensity of pMLC (F) and F-actin (G) in cells treated with 500 mM H 2 O 2 , measured from the cell edge (0 mm) into the cell center (10 mm). (H) pMLC/F-actin fluorescence intensity ratio in cells treated with 500 mM H 2 O 2 , measured from the cell edge (0 mm) into the cell center (10 mm). In (F-H), the data shown represent one experiment and are averaged from at least 13 cells for each condition. The experiment was repeated three times with similar results ( Figure S2G). doi:10.1371/journal.pone.0041342.g008 Matlab (MathWorks) which allows the visualization of F-actin turnover maps [18,71] and F-actin flow maps. To determine the width of the lamellipodium, we used the kymographs and measured the distance of the region of fast F-actin flow at the leading edge which corresponds to the lamellipodium.

H 2 O 2 Detection in Living Cells
The presence of ROS in the cells was monitored with PY1-AM (Peroxy-Yellow 1 Acetoxymethyl-ester), a selective fluorescent indicator for H 2 O 2 which has been provided by C. J. Chang (University of California, Berkeley). PtK1 cells plated on glass bottom dishes were incubated with 5 mM PY1-AM in Hank's balanced salt solution, HBSS (Invitrogen, Carlsbad, CA) for 30 min at 37uC, washed once with HBSS and treated with 500 mM H 2 O 2 for 20 min at 37uC. Then, PY1-AM fluorescence was observed with an inverted microscope (Eclipse TE 2000-U, Nikon) using a 40X/1.4 NA Plan Apo Ph3 objective lens (Nikon).

Statistical Analysis
Statistical analyses presented were determined using two-tailed Student's t test. Figure S1 Method of immunofluorescence analysis. Quantification of the fluorescence of P-cofilin, free barbed ends, p34-Arc, pERK, myosin IIA, tropomyosin, pMLC and F-actin as a function of the distance from the leading edge was obtained with custom software written in Matlab (MathWorks) as explained below with the example of P-cofilin (H 2 O 2 -0 min). (A) Individual fluorescence intensities of P-cofilin (left panel) and F-actin (right panel) were measured from the cell edge (0 mm) into the cell center (10 mm) for 18 cells. (B) These fluorescence intensities were then averaged and are presented with the corresponding standard deviation. (C) The ratio between the averaged P-cofilin and Factin intensities was calculated and plotted against distance from the leading edge. This curve corresponds to the fluorescence intensity ratio P-cofilin/F-actin after 0 min of H 2 O 2 treatment (D, left panel, black curve) for the first experiment which is shown in Figure 3D. The same process has been used for the other stimulation times (D, left panel, colored curves) and for two other independent experiments (D, center and right panels).