Oleanolic acid stimulation of cell migration involves a biphasic signaling mechanism

Cell migration is a critical process for wound healing, a physiological phenomenon needed for proper skin restoration after injury. Wound healing can be compromised under pathological conditions. Natural bioactive terpenoids have shown promising therapeutic properties in wound healing. Oleanolic acid (OA), a triterpenoid, enhances in vitro and in vivo cell migration. However, the underlying signaling mechanisms and pathways triggered by OA are poorly understood. We have previously shown that OA activates epidermal growth factor receptor (EGFR) and downstream effectors such as mitogen-activated protein (MAP) kinase cascade and c-Jun N-terminal kinase (JNK), leading to c-Jun transcription factor phosphorylation, all of which are involved in migration. We performed protein expression or migration front protein subcellular localization assays, which showed that OA induces c-Jun activation and its nuclear translocation, which precisely overlaps at wound-edge cells. Furthermore, c-Jun phosphorylation was independent of EGFR activation. Additionally, OA promoted actin cytoskeleton and focal adhesion (FA) dynamization. In fact, OA induced the recruitment of regulator proteins to FAs to dynamize these structures during migration. Moreover, OA changed paxillin distribution and activated focal adhesion kinase (FAK) at focal adhesions (FAs). The molecular implications of these observations are discussed.


Results
Oleanolic acid over-expresses c-Jun and promotes its activation at the wound edge during migration. C-Jun is overexpressed in response to OA stimulation 12 in non-malignant mink lung epithelial cells, Mv1Lu. This cell line is recognized as a good epithelial model for the study of cell motility due to its ability to migrate and also to stop proliferation when the cells reach confluence [38][39][40] . Mv1Lu cells were stimulated with OA during a wound healing scratch assay. The evolution of the migration front was monitored and the expression of c-Jun was studied by immunostaining. The advantage of this approach is to directly study proteins involved in OA enhanced migration, focusing on its topographic and subcellular localization along the wound edge. The presence of OA produced a neat overexpression of c-Jun at the wound edge, which was significantly high after 6 h ( Fig. 1a; Supplemental Fig. 1a).
Also, upon OA treatment (6 h), levels of phospho-c-Jun, regarded as phospho-Ser 63, were significantly high at the wound edge ( Fig. 1b; Supplemental Fig. 1b). Remarkably, c-Jun and p-c-Jun overexpression occurred at the nucleus of actively migrating cells. Although c-Jun and p-c-Jun levels were enhanced at the migration edge of Mv1Lu, there was a clear decrease in the c-Jun protein expression and its active form as cells were further away from the wound edge. The visualization of actin, by using phalloidin staining (Supplemental Fig. 2), revealed critical changes in the intensity of the cytoskeleton, which were very evident after OA stimulation.
All these data suggest that the effect of OA is especially powerful at wound-edge cells, activating and overexpressing c-Jun protein there, which correlates well with the improvement of cell migration exhibited by OA 12 . Activation of different OA targets exhibits different timing. Previously, it has been shown that OA produces an activation of EGFR in MDA-MB-231 cells that is independent of protein translation 12 . MDA-MB-231 cells are a good model for cell migration [41][42][43] and exhibit a good expression of EGFR 44 . Indeed, EGFR inhibitor (EGFRi), PD153035, was capable of completely reducing OA-stimulated migration of MDA-MB-231 ( Fig. 2a,b) 12 . It is important to measure the effect of OA at the protein signaling level. By using sub-confluent cells, we can interpret what the effect of OA might be on the wound edge 12,45 . To gain further knowledge of the activation of EGFR by OA and its involvement in the activation of previously described protein targets 12 , we conducted an OA-induction and monitored it at different times. Immediately after stimulation with OA (1 h), the activation by phosphorylation of JNK1/2 and c-Jun was detected together with the total level increase of both proteins (Fig. 2c,d). Strikingly, when looking at EGFR, we could detect its OA activation at three hours, but no sign of activation was seen at earlier time points. So, compared with c-Jun or JNK1/2, activation of EGFR showed a delayed stimulation by OA (Fig. 2c,d). Moreover, c-Jun phosphorylation experienced a further increase at three hours, which may suggest a double response and different phosphorylation dynamics for both proteins, c-Jun and EGFR. Coherently with this, the activation of JNK kinase seemed to be also biphasic, showing an early activation and a later (6 h) higher activation.
Activation of EGFR has been linked to phosphorylation of different residues at the Focal Adhesion Kinase (FAK), which leads to the Tyr 925 phosphorylation 35 . Tyr 925 phosphorylation of FAK is a key event on the management of cell migration 46 . Interestingly, OA produced the phosphorylation of FAK with a time dynamic parallel to the phosphorylation of EGFR (Fig. 2c,d).
Altogether, these results suggest that OA stimulates different targets to activate several key elements related to cell migration showing a clear time-uncoupling for different proteins.
The use of specific inhibitors against EGFR, MEK and JNK confirms an EGFR independent OA-induced pathway in MDA-MB-231 cells. The study of OA-activation of different proteins showed time uncoupling between c-Jun phosphorylation and EGFR activation. To assess whether there was an effective independence of both phenomena, cells were stimulated with OA in the presence of a specific inhibitor for EGFR, PD153035 (EGFRi). Although EGFRi importantly reduced OA-stimulated phosphorylation of EGFR, a minute effect was perceived on c-Jun phosphorylation or JNK1/2 phosphorylation. However, the presence of JNK1/2 inhibitor SP600125 (JNKi) caused a drastic negative effect on c-Jun phosphorylation (Fig. 4a,b). When Figure 1. Oleanolic acid induces the expression of c-Jun and promotes its phosphorylation at the edge of wound-scratched Mv1Lu cells. Confluent Mv1Lu cells were scratched and allowed to migrate for the indicated times. Cells were treated with 5 µM OA or DMSO equivalent volume (control). Cells were immunostained with specific antibodies against c-Jun (a) and its active phosphorylated form, p-c-Jun (b). Co-staining with phalloidin and Hoechst-33258 was used to show actin cytoskeleton and nuclei, respectively. Images of c-Jun/p-c-Jun fluorescence were converted into pseudo-color with ImageJ software to show the intensity of c-Jun staining. Color rainbow scale represents fluorescence intensity for either c-Jun or phospho-c-Jun. Actin fibers (F-actin): red. Nuclei: blue. Images were obtained with a confocal microscope. This experiment was repeated at least three times. Representative images are shown. Scale bar indicates 50 µm.  (Fig. 4a,b). All these data suggest a complex and intricate signaling network that is interconnected to activate several proteins in response to OA. Nevertheless, a clear independence of c-Jun phosphorylation could be seen in the activation of EGFR phosphorylation upon OA stimulation, while FAK activation was dependent on EGFR activity. We wanted to assess the effect of the different inhibitors previously used on the activation of the same proteins upon OA stimulation at Mv1Lu cells. In this scenario, the stimulation with OA produced a late activation of ERK that was inhibited with EGFRi or MEKi but not with JNKi. The activation of c-Jun, however, was not prevented by EGFRi (Supplemental Fig. 3a,b). The activation (phosphorylation) of c-Jun after OA stimulation overlapped with migrating cells at the wound-healing scratch assay front (Fig. 5a,b). Strikingly, both JNKi and MEKi prevented the phosphorylation of c-Jun. On the other hand, the presence of EGFRi enabled the activation Asterisks indicate statistically significant differences between the selected conditions according to a one-way ANOVA statistical analysis: (*p < 0.05, **p < 0.005, ***p < 0.001 and ****p < 0.0001). (c) Total protein extracts from serum-starved sub-confluent MDA-MB-231 cells treated with 10 µM OA, EGF or DMSO equivalent volume as vehicle control. Different proteins were assayed at the indicated times (h) targeting phospho-FAK (Tyr 925), phospho-EGFR (Tyr 1068), phospho-ERK1/2 (Thr 202/Tyr 204), phospho-JNK1/2 (Thr 183/Tyr 185), phospho-c-Jun (Ser 63). Additionally, total protein expression was assayed: EGFR, FAK, ERK1/2, JNK1/2 and c-Jun. β-Actin was used as a loading control. A representative experiment is shown. (d) Plots with intensity values of each protein assayed by Western-blot, by gathering the data of three independent experiments. Intensity values were quantified and collected by ImageJ software. Asterisks indicate statistically significant differences between the selected conditions according to a one-way ANOVA statistical analysis: (*p < 0.05, **p < 0.005, ***p < 0.001 and ****p < 0.0001). www.nature.com/scientificreports/ of c-Jun further supporting the fact that c-Jun activation by OA was not dependent on EGFR activation. All these data reinforced the notion that the activation of c-Jun by OA in different cell lines has, at least, two distinct non-related mechanisms, one of which does not require EGFR activation.
OA promotes changes in actin fibers and paxillin distribution. In cell migration, actin protein is essential for building a suitable dynamic architecture that allows cells to migrate 23,47,48 . As shown before, actin varies in response to OA in the migrating cells (see Supplemental Fig. 2). Actin interacts, among others, with paxillin, which is an adaptor protein, playing a role in cytoskeleton anchoring to focal adhesion (FA) and its remodeling during cell migration 24,27,37,49 . One of the consequences of OA stimulation was the phosphorylation of paxillin at Ser 178 (Supplemental Fig. 4a,b). JNK regulates cell migration through the phosphorylation of paxillin at Ser 178 50 , and indeed JNK is activated in response to OA (see Fig. 2c,d). Both phosphorylation of JNK and phosphorylation of paxillin were coincidental in time, thus suggesting a correlation between both phenomena and the activation by OA.
To further understand the effect of OA on cell migration, we immunostained paxillin in wound-healing scratched Mv1Lu cells. When immediately-scratched cells were observed, paxillin was distributed as compacted focal adhesions (Fig. 6). As soon as 6 h after stimulation with OA, paxillin staining pattern changed to display smaller FA and denser distribution. This was especially noticeable at places where the cells were forming Asterisks indicate statistically significant differences between the selected conditions according to a one-way ANOVA statistical analysis: (*p < 0.05, **p < 0.005, ***p < 0.001 and ****p < 0.0001). (c) Total protein extracts from serum-starved sub-confluent Mv1Lu cells treated with 5 µM OA, EGF or DMSO equivalent volume as vehicle control. Different protein phosphorylation was assayed at the indicated times (h) targeting phospho-EGFR (Tyr 1068), phospho-ERK1/2 (Thr 202/Tyr 204), phospho-JNK1/2 (Thr 183/Tyr 185) and phospho-c-Jun (Ser 63). Additionally, total protein expression was assayed: ERK1/2, JNK1/2 and c-Jun. β-Actin was used as a loading control. A representative experiment is shown. (d) Plots with intensity values of each protein assayed by Western-blot, by gathering the data of three independent experiments. Asterisks indicate statistically significant differences between the selected conditions according to a one-way ANOVA statistical analysis: (*p < 0.05, **p < 0.005, ***p < 0.001 and ****p < 0.0001). , became more apparent after 12 h after OA treatment. This correlated well with the fact that OA enhances cell migration 12 . All these observations suggest that OA produces high dynamization of cytoskeleton and FA compatible with a higher migration rate. In a previous publication, we showed that EGFRi, JNKi or MEKi had an inhibitory effect on OA-stimulated cell migration 12 . To further comprehend the consequences of using these inhibitors, we studied the subcellular actin architecture of the cells upon OA stimulation in the presence of these inhibitors. Both OA and EGF produced clear changes in cell architecture losing actin stress fibers, which were still evident with JNKi and MEKi but not with EGFRi (Supplemental Fig. 6a). Concerning FAs, the use of MEKi and EGFRi decreased their density and increased their size, although the effect of EGFRi was milder size-wise (Supplemental Fig. 6b,c). In contrast, in response to OA, neither the size nor the density of FA seemed to be affected by the presence of JNKi.

OA treatment produces FAK activation and promotes focal adhesion remodeling in scratched
Mv1Lu cells. Analyses of various FAK mutants have revealed that FAK targeting to FA is important in the regulation of its activity 51 . Previously, we showed that OA produced an activation of FAK at residue Tyr 925 (see Figs. 2,4) in MDA-MB-231 cells. Western blot results in Mv1Lu cells confirmed that OA caused a strong FAK phosphorylation on Tyr 925, similarly to EGF, but at a different timing (Fig. 7a). Thus, we decided to study the OA effect on phospho-Tyr 925 FAK subcellular localization at wound-edge migrating cells. As a consequence of wounding, in control cells, we observed a p-FAK sharp and pointed pattern located at FA 6 h after wound scratch  (Ser 63). Additionally, total protein expression was assayed: EGFR, FAK, JNK1/2 and c-Jun. β-Actin was used as a loading control. A representative experiment is shown. (b) Plots with intensity values of each protein assayed by Western-blot, by gathering the data of three independent experiments. Intensity values were quantified and collected by ImageJ software. Asterisks indicate statistically significant differences between the selected conditions according to a one-way ANOVA statistical analysis: (*p < 0.05, **p < 0.005, ***p < 0.001 and ****p < 0.0001). . Pearson´s correlation showed significant values between paxillin and p-FAK (Tyr 925) upon OA treatment that were not noticeable immediately after wounded cells or control (DMSO-treated) cells (Fig. 7d). OA-like results were obtained upon EGF treatment. Similarly, in MDA-MB-231, we observed that OA produced the phosphorylation of the FAK that was overlapping with the focal structures revealed by paxillin staining. Pearson's correlation of the staining of both proteins confirmed that phospho-Tyr 925 activated FAK was localized at the focal structures upon OA stimulation (Supplemental Fig. 8a-c).

Scientific Reports
Therefore, all these results suggest that OA treatment induces an activation of FAK that occurs mainly at the FAs.

Effect of inhibitors on the OA activation of FAK.
In culture cells, the treatment with OA produced the phosphorylation of FAK at Tyr 925. The use of different inhibitors (EGFRi and MEKi) proved the dependence of this phosphorylation on EGFR and MAPK signaling pathways (see Fig. 4). In fact, the size and the density of FA were affected by EGFRi and MEKi (see Supplemental Fig. 6). To investigate further, we tested the effect of the aforementioned inhibitors on the OA activation of FAK at the wound-edge migrating cells and studied its colocalization with paxillin. Again, the stimulation with OA caused a general phosphorylation at Tyr 925 of FAK, which was evident due to a higher increase in the signal at the FA (Fig. 8a) and its evident colocalization with paxillin (Supplemental Fig. 9). Coherently, this effect was suppressed by the presence of EGFRi and MEKi. However, the presence of JNKi did not affect the activation of FAK measured as localization of phospho-Tyr 925 at the Color rainbow scale represents fluorescence intensity for phospho-c-Jun. Actin fibers (F-actin): red. Nuclei: blue. Images were obtained with a confocal microscope. This experiment was repeated at least three times. Representative images are shown. Scale bar indicates 10 µm. (b) Plot represents the data obtained by p-c-Jun intensity at cell nuclei. In every condition, each point on the plot represents p-c-Jun intensity at the nucleus of one cell, quantified by ImageJ software. With the collected data of p-c-Jun intensity, a one-way ANOVA statistical analysis was performed (*p < 0.05, **p < 0.005, ***p < 0.001 and ****p < 0.0001).  All these data suggest a clear OA-stimulated FAK-dependent remodeling of FA in which EGFR and MEK, but not JNK, are involved.

Discussion
In this paper, we have disclosed more molecular details of OA mode of action on epithelial cells to produce migration and we have studied in depth the effect of OA on critical components of the cell migration machinery. Regarding OA effect on cell migration, it has been reported to inhibit migration on a laryngeal squamous carcinoma cell line (Hep-2) at 30 µM OA concentration in serum-free medium with an additional negative effect on proliferation 11 . Also, on a hepatocellular carcinoma (HepG2) cell line, concentrations of 25 and 50 µM 10 or 30 µM in the same cell line 9 have a clear migration inhibitory action. Finally, in HeLa cells, 50 µM of OA has a negative effect on migration and induces apoptosis 8 . However, we previously published that there is a narrow interval of OA concentrations in serum-free medium, 5 µM and 10 µM for Mv1Lu or MDA-MB-231, respectively, that stimulates migration 12 . Above these concentrations, OA inhibits migration in these cell lines 12 . Similarly, the same range of OA concentration for migration stimulation was reported on mouse fibroblasts, 20 µM in serum-free medium, with a migration inhibitory effect above that concentration 6 . This apparent discrepancy could be attributed to the use of different concentrations of OA, although further research should be conducted to possibly stablish differences between the cell lines that may explain the divergence in the results.
Previously, we showed that OA activates c-Jun by phosphorylation in Mv1Lu and MDA-MB-231 cells 12 . C-Jun is a key transcription factor involved in cell migration 52,53 . Active c-Jun plays an important role in the signaling of different growth factors, indeed, conditional mutations of this transcription factor at the epidermis are known to delay wound closure 54 . Our study confirmed that phosphorylation of c-Jun at the wound scratch edge is crucial for the induction of migration of epithelial cells by OA. Remarkably, OA-stimulated c-Jun overexpression and activation only occurred at the nucleus of actively migrating cells, which is consistent with the c-Jun function to promote migration 52,53 upon OA stimulation. EGF stimulation of cells produces a very extensive c-Jun overexpression and phosphorylation 45,55 . In contrast, the effect of OA was specific to the wound edge of the scratch assay, thus being more precise and avoiding the stimulation of cell migration beyond necessary.
The blockage of upstream molecular signaling by selectively inhibiting MAPKs (MEK1-ERK1/2) or JNK pathways partially inhibited OA-induced migration in Mv1Lu or MDA-MB-231 cells 12 . Moreover, inhibition of Figure 6. OA promotes changes in FAs and actin cytoskeleton. Confluent Mv1Lu cells were scratched and allowed to migrate for 6 and 12 h. Cells treated with 5 µM OA or DMSO equivalent volume (vehicle control) were immunostained with specific antibodies against paxillin (green). Co-staining with phalloidin and Hoechst-33258 was used to show actin cytoskeleton (magenta) and nuclei (blue), respectively. This experiment was repeated at least three times. Representative images are shown. Scale bar indicates 10 µm. Quantification of the density of FA as FA number per filopodia area. Quantification of FA size (average size) at the filopodia area. One-way ANOVA statistical analysis was performed (*p < 0.05, **p < 0.005, ***p < 0.001 and ****p < 0.0001).  12 . Interestingly enough, the use of EGFRi did not completely suppress phosphorylation of c-Jun at the wound edge. Moreover, OA stimulation of c-Jun phosphorylation measured by western blot was not suppressed by the use of EGFRi. In a previous paper, we focused on the activation of EGFR by OA as the main event inducing migration in both Mv1Lu and MDA-MB-231 cells 12 . However, the evidence presented in this paper suggests a very early stimulation of c-Jun phosphorylation that is independent of the activation of EGFR in both Mv1Lu and MDA-MB-231 cells. Additional pathways converge into c-Jun activation 56,57 ; nevertheless, all the c-Jun activation we saw, independent of or dependent on EGFR was mediated by JNK1/2. JNK has been shown to be required for Drosophila dorsal closure 15,16 . Certainly, the use of JNKi completely blocked the phosphorylation of c-Jun. In any case, further research must be carried out in order to reveal the possible pathway that specifically produces the activation of c-Jun, independently of EGFR activation. www.nature.com/scientificreports/ C-Jun, the founding member of the AP-1 family, contributes to its own transcriptional upregulation when it is activated by phosphorylation, producing a self-regulating positive feed-back 56 . OA phosphorylation/protein total level increment of c-Jun and phosphorylation of EGFR showed different time dynamics, with EGFR having a later activation compared to c-Jun activation, which suggests a complex or multiple step mechanism for OA activation of EGFR. In this context, transactivation of EGFR is a phenomenon that activates this receptor in the absence of canonical EGFR external stimuli 58 . EGFR transactivation is triggered by G-protein-coupled receptors (GPCRs) and is now recognized as a key mechanism that couples distinct signal transduction pathways to diverse cellular stimuli 58 . It is plausible that EGFR stimulation by OA could be due to receptor transactivation. Indeed, a clue comes from the fact that OA has been identified as a weak agonist for some GPCR receptors 59 , which in turn would activate EGFR 60 and could also have a direct stimulatory effect on the MAPK signaling pathway 61 . Besides, as shown by Bernabé et al. 12 , the stimulation of EGFR phosphorylation by OA is independent of protein synthesis. Despite all of this, activation of EGFR is clearly a key point in the process of migration induced by OA, because the inhibition of the receptor by PD153035 completely suppresses OA stimulation of migration 12 .
In Mv1Lu cells, the OA-induced phosphorylation of ERK experienced a delay in comparison to c-Jun phosphorylation, and it is in synchrony with the activation of EGFR. This fact suggests that MAPK signaling pathway mainly depends on the OA-activation of EGFR. Unfortunately, in MDA-MB-231, the high overexpression of EGFR 44 , causes a high level of ERK phosphorylation making it impossible to clearly distinguish ERK activation in response to OA.
So far, our results show that, in Mv1Lu and MDA-MB-231 cell lines, JNK and c-Jun OA-activation are independent of EGFR. On the other hand, our results for ERK phosphorylation support an EGFR delayed activation mechanism independent of the mechanism early activating JNK/c-Jun. In any case, MAP kinases seem to be involved in both mechanisms, early and late, because PD98059 (MEK) inhibitor blocks the activation of both EGFR and JNK/c-Jun. This way of behaving would suggest a possible feedback mechanism between the MAPK pathway and the activation mechanism that implicates both EGFR and putative GPCR. This crosstalk mechanism between both receptors has been associated with the activation of Src 61 , but it is worth exploring whether there could be a similar crosstalk for other kinases such as ERK1/2. In the wounding of corneal cells, the activation of www.nature.com/scientificreports/ EGFR is mediated by a shedding mechanism related to GPCR receptors 62 . Strikingly, in this system, the inhibition of ERK1/2 phosphorylation by the use of PD98059 caused a halt in the activation of EGFR, thus suggesting that, in addition to functioning as an EGFR downstream effector, ERK1/2 also mediates EGFR transactivation to a variety of stimuli 62 . Similarly, in both MDA-MB-231 and Mv1Lu cells, the use of PD98059 upon OA stimulation caused a strong inhibitory effect on EGFR activation and c-Jun phosphorylation. This suggests that, upon stimulation with OA, a similar MEK-involved crosstalk mechanism could be at work for the transactivation of EGFR. In fact, in Mv1Lu cells, a minute stimulation of ERK could be seen 1 h after OA stimulation (see Fig. 3). At in vitro wound assays, a close look at the wound migrating edge showed morphological variations in migrating cells at the wound edge compared to cells placed in the rear. These morphological differences consist in the development of cytoplasmic protrusions such as lamellipodia and ruffles 19 . Indeed, the presence of OA increased the number of lamellipodia and ruffles at the cells at the edge of the wound. In our case, the stimulation with OA also modified the intensity and definition of actin microfilaments there. In fact, actin depolymerization and debranching happens during migration facilitating the dynamic remodeling of the actin network and the cyclic extension and retraction of lamellipodia 20 . So, the observed changes were suggestive of an active migratory status of the cells 19,20 . Cell migration, a very well-coordinated process, requires the continuous formation of focal complexes (FC) that mature into FAs 19,20 . Paxillin is a fundamental component for FA functioning because it is able to integrate multiple signaling inputs 24 . By providing a structural scaffold, paxillin facilitates the synchronized binding of different protein components, particularly signaling molecules 63 . Our results showed a differential organization of FA, revealed by paxillin staining, in response to OA. Certainly, OA caused an increased density but a decreased size of the foci. The phosphorylation of paxillin Ser 178 by JNK produced FA increased dynamics and cell motility 28 . OA stimulation produced a time phosphorylation on this residue that correlated well with JNK activation. However, in response to OA, neither the size nor the density of FA were affected by the presence of this inhibitor (JNKi), which suggests that there might be other means of focal adhesion arrangements in response to OA, independently of JNK activation. Further research is needed to stablish the particular contribution of paxillin Ser 178 phosphorylation to the migration induced by OA.
Coordination and integration of growth-factor signaling and FA remodeling is facilitated by the close physical proximity of key signaling molecules. This is the case for EGFR and FAK 64 . FAK Tyr 925 can be phosphorylated as a consequence of cell integrin assembling or growth factor stimulation 31,36 . Interestingly, OA treatment produced the phosphorylation of EGFR and thus of FAK with a very similar timing outcome. Indeed, FAK activation was completely suppressed by EGFRi. Additionally, FAK Tyr 925 phosphorylation was also suppressed by MEKi, indicating the direct involvement of EGFR via MEK in the activation of FAK in response to OA. So, regarding FA remodeling and in contrast to JNKi, the presence of EGFRi or MEKi produced bigger and consolidated FAs, thus supporting previous findings indicating the need of an EGFR/MEK active signaling for a successful OAdriven migration 12 .
The use of different functional mutants of FAK has revealed that FAK localization to FAs is important for the regulation of FA dynamization 51 . Hence, increased FAK localization to FAs raises its turnover at the protrusion front 34 . Our results show that active FAK co-localized with paxillin in response to OA, thus indicating that an active remodeling of FAs was taking place in response to the triterpenoid. Active FAK localization at FAs was dependent on the activation of EGFR because it was inhibited by EGFRi. Also, the participation of active MEK was required as the use of MEKi also prevented Tyr 925 phosphorylated FAK localization at FAs. These data confirmed an OA responsive axis EGFR/MEK/ERK/FAK as a motor for FA dynamization 49 . Additionally, mobilization of phospho-Tyr 925 FAK to the cytosol indicates an intense remodeling of FAs. Certainly, active migration requires a continuous remodeling of FAs 27 . The absence of inhibitory activity on the FA localization of active FAK in the presence of JNK reinforces the idea of an independent but complementary promigratory event. Such promigratory event, although not required for FA managing, may be involved in other aspects of cell migration. So, even though active FAK mobilization to FAs did not seem to be affected by JNKi, OA-stimulated cell migration is partially inhibited 12 .
In this paper, we have further dissected the molecular mechanisms underlying the effect of OA on cell migration. Our results showed that OA induces different and interconnected signaling pathways. Given the powerful effect of OA on cell-migration, the need for additional understanding of those mechanisms seems relevant and very pertinent. Finally, these findings could provide an interesting mode of action for OA that is worth investigating in other related molecules.

Materials and methods
Preparation of oleanolic acid. Oleanolic acid (purity > 97%) (Merck, Darmstadt, Germany) was dis- www.nature.com/scientificreports/ medium was changed to FBS-free medium for 24 h. At the initial time (T0), a cross-shaped scratch was made on the cell monolayer using a sterile p-200 µl pipette tip. After replacing FBS-free culture medium to wash out released cells, OA (indicated concentration at each experiment) or an equivalent volume of DMSO was added. Additionally, EGFR pharmacological inhibitor PD153035 (EGFRi) 2.5 µM, was added to OA sample. After a 24-h incubation period, cells were fixed with 4% formaldehyde (Applichem GmbH, Darmstadt, Germany) in PBS (Biowest, Nuaillé, France) for 10 min. Finally, well plates were washed twice with PBS. Pictures were taken at 10× magnification using an optical microscope equipped with a digital camera (Motic Optic AE31, Motic Spain, Barcelona, Spain). To quantify cell migration, the areas of the gaps in the wounds were measured by ImageJ software. The initial cell-free surface was subtracted from the endpoint cell-free surface and plotted in a graph 66 .
Cell-front migration assay, subcellular localization assay. Mv1Lu or MDA-MB-231 cells were grown until they reached confluence on round glass-coverslips using appropriate medium with 10% FBS. At that point, cells were washed and the medium was replaced by FBS-free medium for 24 h. After this, the established epithelium was scratched using a razor blade, making an artificial wound with enough space to allow cells to migrate. The new wound was set as time 0 of the experiment and then OA, DMSO or 10 ng/ml epidermal growth factor (EGF, Sigma-Aldrich, St Louis, MO, USA) was added to the medium. When using specific inhibitors, cells were treated 30 min before OA stimulation with the following pharmacological inhibitors: 2.5 µM PD153035 (Epidermal Growth Factor inhibitor, EGFRi) 68  When a wider view of the migration front was required, especially in c-Jun staining (indicated in figures), four linked fields were acquired by the "Tile scan" ZEN tool. Subsequently, tile scans fluorescent signals were converted to a linear mode and covered with a full data range using the Rainbow look up table (Rainbow LUT) in ImageJ software. In order to quantify c-Jun and phospho-c-Jun levels in immunofluorescence pictures, the image was analyzed and quantified by ImageJ software. For this purpose, pictures in 8-bit three-channel format (Red, Green, Blue, RGB) were divided into three separate color channels (three pictures). By using blue channel picture (Hoechst staining), regions of interest (ROIs) were stablished to define each nucleus, creating as many ROIs masks as nuclei in the image. Then, by overlapping these masks onto the corresponding green channel picture (c-Jun/p-c-Jun staining), we calculated the green intensity value of each nucleus (ROI). The quantified signal of each nucleus was used as a replicate to obtain c-Jun/p-c-Jun intensity data in each of the conditions performed. On the other hand, "Z stack" ZEN feature was used when deep cytoskeleton structure observance was required, taking a proper number of pictures along the Z axis. Focal adhesion (FA) quantifications were performed as described in Horzum et al. 67 by using CLAHE and Log3D macros for ImageJ. Briefly, focal adhesions were quantified from paxillin stained acquired pictures. We used three different images for each condition. Specifically, cell filopodia were selected as regions of interest (ROIs) and the resulting areas (containing FAs) were considered for further analysis. A number of five filopodia were taken into account from each picture. Then, the number of FAs were calculated in each filopodia by using the previously mentioned macros. The obtained number was divided by the total filopodia area to determine FA density. In parallel, the size of each FA was measured using the macros mentioned above. Finally, FA average size was calculated for each treatment. For colocalization determination of paxillin and p-FAK, ZEN interface software was used. The analysis was performed by dividing two pictures (replicates) of each experimental condition into three equal regions of interest (ROIs) including front and back of the edge. Then, overlapped pixels between channel 1 signal (paxillin) and channel 2 signal (p-FAK) were marked for a Pearson's correlation coefficient determination included in ZEN software (ZEISS, Jena, Germany). www.nature.com/scientificreports/ USA). For protein quantification, western pictures in 8-bit format were processed in ImageJ software. Subsequently, in all western-blot pictures, a lane was stablished for each of the samples. In each lane, the band was selected according to the appropriate size (kDa) of the protein of interest. In general, and for the phosphorylated version of each interest protein, each band's intensity peak was plotted and the area under the plot was measured using "Wand (tracing) tool" of ImageJ to obtain the intensity value. To normalize the data, obtained intensity values were referred to obtained intensity values of either the unphosphorylated form of the protein or a loading control protein such as β-actin if the unphosphorylated form was undetectable (non-available antibody for detecting the unphosphorylated form). Statistical analysis. All the collected data were analyzed using Graph Pad Prism 7 software. In every analysis, classic statistical parameters were calculated and statistical tests were performed with a 95% confidence interval, consequently, p values lower than 0.05 were considered to be statistically significant. At the figure legends, the asterisks denote statistically significant difference between treatments (*p < 0.05, **p < 0.005, ***p < 0.001 and ****p < 0.0001). Data of intensity values, collected from Western-blots, were analyzed by a one-way ANOVA test, comparing the mean of each condition with the mean of every other condition. Then, a Tukey's multiple comparisons test was performed. Data of intensity values obtained from c-Jun and p-c-Jun nuclei quantifications were analyzed by a two-way ANOVA test. Specifically, a Tukey's multiple comparisons test was performed, comparing the mean of each condition with the mean of every other condition; however, each comparison was made within each picture sector. In this way, same picture sectors were compared (for example, sector 1, S1) between different conditions. On the other hand, to see differences between sectors from the same condition a one-way ANOVA was performed, as described elsewhere. Data of colocalization Pearson's coefficient in the immunocytochemistry assays were analyzed with a one-way ANOVA test, comparing the mean of each condition with the mean of every other condition. Then, a Tukey's multiple comparisons test was performed.

Data availability
In order to clarify Western-blot data, we displayed cropped blots of each Figure