LIM Kinase Inhibitor Pyr1 Reduces the Growth and Metastatic Load of Breast Cancers Free

LIM kinases (LIMK) regulate the architecture of the actin cytoskeleton by phosphorylation and inactivation of actin depolymerization factors of the ADF/cofilin family (1). Independently of this effect on actin microfilament dynamics, LIMKs regulate microtubule dynamics (2–4), but whether this regulation occurs through a direct binding of LIMK to microtubules (5) or through phosphorylation of an associated protein (6) is still under debate. When LIMKs are inhibited, microtubules are stabilized and actin microfilaments are severed and disorganized (2, 4).

Thus, LIMKs function as central network hubs coordinating several cellular- and tissue-level responses by regulating both actin microfilament and microtubule assembly (7). In pathophysiologic conditions, pharmacologic inhibition of LIMK could have antitumor and antimetastatic effects, given the involvement of the actin and the microtubule cytoskeleton in cell division and in cell motility. The LIMK family of serine/threonine kinases includes two highly related members, LIMK1 and LIMK2 (1). LIMK activity is mainly regulated by the Rho-GTPases (RhoA, Rac, and Cdc42) through their downstream kinases ROCK, PAK1, PAK4, and MRCK (1). The activation of the Rho-GTPases and their effectors, including LIMK, have been reported as playing important roles in tumor development and progression (8–13). Expression of LIMK or cofilin phosphorylation is elevated in malignant melanoma (14), glioma (15), prostate (4, 16, 17), and breast tumors (18, 19). In breast cancer models, activation of LIMK is the last step of an integrin-linked machinery of cytoskeletal regulation that enables tumor initiation and metastatic colonization (20). Thus, LIMK are enzymes whose activity is elevated in cancers compared with normal tissue. Consequently, their inhibition could selectively target tumors and offer a large therapeutic window.

Chemotherapy is a component of the treatment of invasive breast cancers. One of the most important classes of chemotherapy agents is the taxanes, which bind and stabilize microtubules. Taxane resistance, however, limits treatment options and creates a major challenge for clinicians. Taxane's general antimitotic and microtubule-stabilizing actions also result in severe side effects, such as myelosuppression or neurotoxicity. Rather than chemotherapeutic agents interacting directly with the microtubule network, the use of drugs that target microtubule regulators, such as LIMK, is hence an attractive alternative therapeutic strategy.

LIMKs are considered as emerging targets for cancer therapy (21), and an increasing number of inhibitors is reported in the literature (2, 22–28). Among these inhibitors, few, if any, fulfill the three criteria that are important for in vivo experiments, that is, high selectivity, complete characterization of the effects on both actin and microtubule dynamics, and knowledge of toxicity on animals.

We have previously identified and characterized a highly selective LIMK inhibitor, Pyr1 (2). Although ATP competitive, Pyr1 inhibits only LIMK out of 110 kinases tested. When applied on cells, Pyr1 stabilizes microtubules, induces a cell-cycle arrest at the S-G₂–M phases, and, through inhibition of cofilin phosphorylation, blocks actin microfilament dynamics. We have also shown that Pyr1 was active in vitro on paclitaxel-sensitive and -resistant cancerous cell lines and displayed a therapeutic activity in an in vivo murine model of leukemia L1210, while being well tolerated (2).

Because of the selectivity of this cell-permeable inhibitor and its good tolerance in vivo, the aims of this study were (i) to investigate the effect of LIMK inhibition on breast cancer development and (ii) to test the hypothesis that LIMK inhibition is efficient in paclitaxel-resistant cancers. The effects of Pyr1 on paclitaxel-resistant breast cancers have been analyzed thoroughly, both in vitro and in vivo on xenografted models of primary tumor growth and on metastasis. In response to Pyr1 treatment, intratumoral cell movement and tumor cell morphology have been monitored using intravital imaging. Our results show that Pyr1 displays an antitumoral activity. Intravital microscopy revealed morphologic changes of the tumor cells and perturbation of their motile behavior within the tumor when treated with Pyr1. Finally, although Pyr1 did not prevent metastases, it led to an important reduction of the metastatic load. These results indicate that LIMK inhibitors might represent both a pharmacologic alternative to the treatment of taxane-resistant primary tumors and potent agents to reduce the growth of metastases.

Murine mammary adenocarcinoma cells TS/A-pGL3 (20, 29) and human cells MDA-MB-231 and MCF7 (ATCC), routinely tested and authenticated by the ATCC, were cultured as recommended. MDA-MB-231 cells overexpressing the transcription factor ZNF217 (MDA-MB-231-ZNF217rvLuc2) were grown as described previously (30). For intravital microscopy experiments, MDA-MB-231 cells were modified in J. van Rheenen's team (for details, see Supplementary Information). Mouse embryonic fibroblast (MEF) cells were a generous gift from Dr. Richard Hynes (Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA). The expression of the constitutive active form of the Src kinase (SrcY527F) was performed as described in the Supplementary Information.

The antibodies used were from Cell Signaling Technology, that is, cofilin (ref. 5175), phospho-cofilin (ser3; ref. 3313), and β-actin (ref. 4967).

Immunofluorescence analysis of the modification of cellular microtubule dynamics using nocodazole was realized as described previously (2).

Cell viability was analyzed using MTT assay as described previously (2).

Cells (5 × 10⁴) were plated on top of a layer of Matrigel in Transwell chambers (Biocoat, BD Biosciences). After 24-hour incubation with 25 μmol/L Pyr1 or 0.25% DMSO, nuclei of cells that reached the bottom of the transwell were stained with Hoechst. Cell invasion was quantified by counting the number of invading cells using ImageJ software (NIH, Bethesda, MD).

FRAP analysis was performed as described previously (31) on MEF SrcY527F cells transfected with GFP-actin and treated or not with 25 μmol/L Pyr1 just before bleaching. Changes of fluorescent intensity within the bleached area were measured over 2 minutes, and the characteristic time of recovery was quantified using the ZEN software from Zeiss.

Cells were seeded in culture inserts (ibidi, 80206). Two days later, inserts were removed and 25 μmol/L Pyr1 or 0.25% DMSO was added in the medium. Recovery of the wound was recorded during 12 hours using video microscopy. Velocity, total displacement, and persistence were calculated using the MTrackJ plugin (http://www.imagescience.org/meijering/software/mtrackj/manual/) from ImageJ software (NIH, Bethesda, MD).

Spheroids were derived from MDA-MB-231 and MDA-MB-231-ZNF217rvLuc2 cells using the Matrigel-on-top culture as described by Shibue and colleagues (32). After 5 days of culture, spheroids were treated with 25 μmol/L Pyr1 or 0.25% DMSO. Filopodia number and length were quantified using an inverted microscope. A total of 25 spheroids from three different experiments were analyzed in each group.

All animal studies were conducted in accordance with European Union guidelines and approved by the regional ethics committee. The animals were examined daily for mortality and morbidity. Weight was monitored twice a week, and behavior was carefully examined every day from the beginning of treatments (grooming, postures, spontaneous movement in the cage, and touch response)

Populations of 5 × 10⁵ TS/A-pGL3 cells, stably transfected with luciferase, were suspended in PBS and injected into the mammary fat pad of 30 NMRI nude mice. Seven days later, mice were randomized in three equal groups and drugs [Pyr1, synthesized by C-H Nguyen, Institut Curie, France, 10 mg/kg, PTX (Sigma, T1912, 10 mg/kg), or vehicle (36% PEG 400, 10% DMSO, and 54% NaCl 0.9%)] were injected intraperitoneally daily. Tumor growth was monitored by bioluminescence twice a week. Before each bioluminescence imaging (IVIS Kinetic; Caliper), anesthetized mice received an intraperitoneal injection of luciferin (Promega).

Populations of 10 × 10⁶ MDA-MB-231 cells were suspended in PBS/Matrigel v/v (BD Biosciences) and injected subcutaneously into the right flank of 30 NMRI nude mice. When tumors were palpable, that is, 21 days after cell injection, mice were randomized in 3 equal groups and drugs were injected intraperitoneally daily, as described above. Tumor growth was monitored 3 times a week with a sliding caliper.

Populations of 5 × 10⁵ MDA-MB-231 Dendra2 cells were suspended in PBS and injected into the mammary fat pad of 7 female NSG. The first group (4 mice) was injected with 10 mg/kg Pyr1 and the second (3 mice) with vehicle [36% polyethylene glycol (PEG) 400, 10% DMSO, and 54% NaCl 0.9%]. Treatments were injected intraperitoneally daily and started between 30 and 45 days after cell injection, that is, when tumors were palpable and window implanted. Imaging sessions started 2 days after mammary window implantation. Each session lasted 2 hours with one stack every 15 minutes. Each mouse was imaged twice a week during 2 weeks. Mice were anesthetized and intravital imaging was achieved as described previously (33). Details of the methods used for tumoral cells migration tracking and for quantification of fluorescent signal in MDA-MB-231 Dendra2 tumors are presented in the Supplementary Information.

Populations of 2.5 × 10⁵ MDA-MB-231-ZNF217rvLuc2 cells stably transfected with luciferase were suspended in PBS and injected in the left ventricle of 40 NMRI nude mice. The quality of cell implantation was checked immediately after injection by bioluminescence, and only mice with validated implantation were included in the experiment. Treatments started 3 days before cell injection. The first group of 15 mice was injected intraperitoneally daily with 10 mg/kg Pyr1 and the second group of 9 mice was injected with vehicle (36% PEG 400, 10% DMSO, and 54% NaCl 0.9%). Metastatic colonization was monitored by bioluminescence as described above.

MDA-MB-231 Dendra2 tumors were fixed overnight at 4°C in periodate-lysine-paraformaldehyde (PLP) buffer (4% PFA 2.5 mL, NaIO₄ 0.0212 g, l-lysine 3.75 mL, P-buffer pH 7.4 3.75 mL). The fixed tissues were then washed twice with PLP buffer and placed for 6 hours in 30% sucrose at 4°C. Tumors were embedded in OCT tissue-freezing medium (Jung). Tumor sections (14 μm) were incubated with Ki67 antibody (Abcam, 66155) overnight and then with A647-conjugated secondary antibody. Proliferation was evaluated as the number of Ki67-positive cells per field ± SEM. A total of 6 fields were examined and counted for each tumor in each group.

TUNEL labeling was performed using In Situ Cell Death Detection Kit (Invitrogen, C10247). Tumor sections (14 μm) were counterstained with Hoechst. The apoptotic index corresponded to the number of TUNEL-positive cells per field. A total of 6 fields were examined and counted for each tumor in each group.

Statistical analyses were performed using t test, except for tumor growth and cell migration in vivo experiments, for which a Mann–Whitney test was used. Results with P < 0.05 were considered to be statistically significant.

As the expression level of LIMK1 and its activity have been reported to be increased in invasive breast cancer cells (34, 35), we first characterized the different cell lines used in this study, that is, TS/A-pGL3, MDA-MB-231, and MDA-MB-231-ZNF217rvLuc2, regarding their level of expression of LIMK1. Using Western blotting, we found that the endogenous level of LIMK1 was at least 50% higher in these invasive cell lines, as compared with MCF-7 cells, a noninvasive cell line (Supplementary Fig. S1A and S1B; refs. 34, 36).

These cell lines have been described to show resistance to PTX (37–39). We confirmed that PTX had almost no effect on TS/A-pGL3 cell viability and reduced MDA-MB-231 and MDA-MB-231-ZNF217rvLuc2 cell viability by only 30% to 40% over 24 to 48 hours, relative to vehicle control (Supplementary Fig. S2).

We then tested if the LIMK inhibitor Pyr1 was active on these cell lines by measuring its effect on cofilin phosphorylation and on microtubule dynamics. Cofilin phosphorylation was quantified by Western blotting, and slowing down of microtubule dynamics was assessed by evaluation of the resistance of the microtubule network to nocodazole-induced depolymerization. Nocodazole binds free tubulin and prevents its incorporation into microtubules, inducing microtubule depolymerization. Stabilized microtubules, with slow dynamics, have reduced exchanges with the free tubulin pool and are thus less sensitive to nocodazole-induced depolymerization (2).

Although showing some variation between cell lines, Pyr1 consistently inhibited cofilin phosphorylation in a dose-dependent manner (Fig. 1A). Moreover, Pyr1 protected the microtubule network from nocodazole-induced depolymerization (Fig. 1B), indicating that microtubules were stabilized.

We then analyzed the toxicity of Pyr1 on these cell lines. Toxicity profiles were obtained by determining population cell viability in response to a 48-hour incubation with Pyr1 (0–25 μmol/L). As shown in Fig. 1C, Pyr1 has a significant effect on the proliferation of these cell lines, reducing their viability by about 90%. The GI₅₀ (50% of growth inhibition) of Pyr1 was 1.4 μmol/L for TS/A-pGL3, 3.9 μmol/L for MDA-MB-231, and 7 μmol/L for MDA-MB-231-ZNF217rvLuc2.

Thus, in vitro, Pyr1 affects the proliferation of invasive cells that exhibit resistance to PTX.

As Pyr1 affects the in vitro proliferation of mammary and breast cancer cell lines, we further analyzed its effect on tumors xenografted to mice and compared it with the PTX effect, administrated at a therapeutic dose (40).

Although Pyr1 did not induce detectable adverse effects (Supplementary Fig. S3), PTX-treated mice were motionless, displayed swollen abdomen, and lost weight after 2 weeks of treatment. This led to the arrest of the experiment for ethical issues. Consistent with the in vitro cell viability results, PTX was unable to stop the growth of TS/A-pGL3 tumors, whereas Pyr1 stopped their growth (Fig. 2A and B).

Pyr1 contains an ester moiety, which could be subjected to hydrolysis in vivo. We thus analyzed the intravenous concentration of Pyr1 and of its 9-OH metabolite M1, lacking the ester moiety, after a single intraperitoneal injection of 10 mg/kg of Pyr1. We found that 25 minutes after the initial injection, Pyr1 was undetectable in the blood, whereas its metabolite M1 was present. M1 concentration decreased progressively to reach its basal level within 2 hours (Supplementary Fig. S4). This is in line with the high distribution volume, which indicates that compounds exit quickly from the plasmatic compartment. As we have shown that M1 is also able to inhibit LIMK, both in vitro and in cells (compound 3; ref. 2), we hypothesize that the observed effect results mainly from the combined action of Pyr1 and its 9-OH metabolite. For the sake of clarity, the generic term Pyr1 will be further used to refer to Pyr1 and its metabolite.

Tumors were excised at the end of the experiment, and several markers related to microtubule and actin regulation (Supplementary Table S1) were quantified using a reverse-phase protein array (RPPA). Among these markers, the level of phospho-cofilin in tumors was not significantly different between Pyr1-treated mice and vehicle-treated mice. The levels of acetylated- and detyrosinated tubulin, which are indirect markers of microtubule stabilization (2), were found significantly increased in Pyr1-treated mice, as compared with the vehicle-treated mice. Such an increase was not detected in the tumors of PTX-treated mice, which could be correlated with the absence of a PTX effect on tumor growth (Supplementary Fig. S5). These results strongly suggest that the Pyr1 effect in tumors involves microtubule stabilization.

Next, the effect of Pyr1 and PTX on subcutaneous MDA-MB-231 xenografts was also compared. When the tumors reached a palpable size (200–300 mm³), Pyr1 and PTX were injected daily. As for TS/A-pGL3 experiments, the endpoint of the experiment was governed by PTX-induced side effects. Pyr1, as well as PTX, induced a statistically significant decrease (40%–50%) of the tumor volume (Fig. 2C).

Tumors were excised at the end of the experiment and cut in two pieces. One piece was used to analyze the structure of the tumors. Hematoxylin and eosin staining of the tumor sections showed that the cellular density of Pyr1- and PTX-treated tumors was greatly reduced, confirming the antitumor effect of these compounds (Supplementary Fig. S6).

The phosphorylated-cofilin and the detyrosinated tubulin contents were analyzed by Western blotting, using the other piece, and quantified. Whereas no statistical difference was observed when comparing the level of phosphorylated-cofilin in vehicle- and Pyr1- or PTX-treated tumors, a consistent enhancement of the level of detyrosinated-tubulin was observed in Pyr1- and PTX-treated tumors (Supplementary Fig. S7). Again, this suggests that the antitumor effect of the PTX and Pyr1 involved a stabilization of microtubules.

Taken together, these results indicate that Pyr1 has a potent antitumor effect on primary mammary tumors in breast cancer models, even on PTX-resistant tumors, such as TS/A-pGL3.

It has been shown that interfering with LIMK function either by using RNAi (16) or by overexpression of a dominant negative form of LIMK1 in metastatic breast cancer cells (34) or by pharmacologic LIMK inhibition (41) results in reduced cell invasiveness.

Moreover, as LIMK expression is correlated with the aggressiveness of cancer cells (13, 19, 20) we decided to investigate whether LIMK inhibition by Pyr1 impacted TS/A-pGL3, MDA-MB-231, and MDA-MB-231-ZNF217rvLuc2 invasiveness in vitro.

We examined the Pyr1 effect on the motility of cells using a wound-healing assay. We observed that Pyr1 significantly reduced the motility of the three cell lines (Fig. 3A). After tracking the cells individually, we found that Pyr1 reduced the speed of cell movement by approximately 75%. The total displacement was reduced at least by 50%. Pyr1 also reduced the directionality of the cell movement (Fig. 3B).

We then analyzed the Pyr1 effect on invasive migration of these cells through Matrigel in Transwell chambers. Cells were seeded in the insert of the chambers and after 24 hours, the nuclei of invasive cells were counted. As shown in Fig. 3C, Pyr1 exerted a strong inhibitory effect on the invasion of the three cell lines.

Invasion involves specialized finger-like actin structures, called invadosomes in cancer cells, which can be induced by the expression of a constitutively active form of Src (42). These structures self-assemble into round metastructures known as rosettes or rings (31). We investigated the effect of LIMK inhibition on actin dynamics in invadosomes using MEF cells expressing a constitutively active form of Src (Src Y527F) and Lifeact-RFP.

We first observed that cells treated for 2 hours with 25 μmol/L of Pyr1 showed disorganized invadosomes. Instead of the normal ring structure, the actin cytoskeleton was often reorganized into actin spots, indicating that actin dynamics was perturbed (Fig. 3D). FRAP analysis of the few remaining invadosomes after Pyr1 treatment allowed measuring the net flux of GFP-actin into these structures. We found that the time of recovery doubled in invadosomes of Pyr1-treated cells compared with control cells, indicating that actin dynamics was strongly slowed down (Fig. 3E; Supplementary Movie S1). This result suggests that the cyclic phosphorylation/dephosphorylation of cofilin is central for actin dynamics in invadosomes (43) and that complete blockade of cofilin phosphorylation through LIMK inhibition leads to unbalanced actin dynamics.

We conclude from these experiments that, in vitro, Pyr1 is able to slow down cell motility and to suppress invasion.

We further explored the consequences of LIMK inhibition in the processes that allow the survival and proliferation of cancer cells after their settling in the parenchyma of distant tissues. It has been shown that the ability of breast cancer cell lines to settle in a foreign tissue is determined by their capacity to extend abundant actin-rich protrusions morphologically resembling filopodia, called filopodium-like protrusions (FLP), when cultured in three dimensions (36). The activation of the ILK/β-parvin/cofilin pathway leads to the activation of LIMK to govern FLP lifetime (20). Blocking this pathway through the production of constitutively active cofilin has been shown to impair FLPs formation (20). We wondered whether pharmacologic inhibition of LIMK could also impact on FLP formation. MDA-MB-231 and MDA-MB-231-ZNF217rvLuc2 cells were thus propagated in 3-dimensional cultures using the “Matrigel-on-top” method (36), in which cells are plated above a layer of 100% Matrigel and then covered with culture medium containing 2% Matrigel. In such conditions, these cells grow as spheroids. We found that about half of MDA-MB-231 and MDA-MB-231-ZNF217rvLuc2 spheroids extend FLPs. After a 2-hour treatment with 25μmol/L of Pyr1, the percentage of spheroids with FLPs was reduced to 39% for MDA-MB-231 and to 33% for MDA-MB-231-ZNF217rvLuc2 cells (Fig. 4A and B). We measured the length of the remaining FLPs and found that they were reduced by 37% for both MDA-MB-231 and MDA-MB-231-ZNF217rvLuc2 cells (Fig. 4C). Thus, pharmacologic inhibition of LIMK has a profound effect on FLP abundance and length.

As the velocity properties of cell motility in vivo are often different from those observed on two dimensional substrates (44, 45), we then investigated the Pyr1 effect on tumor cell motility in vivo, using intravital imaging. MDA-MB-231 cells expressing the Dendra2 fluorescent protein were implanted into murine mammary fat pads. When the tumors were palpable, that is, 30 to 45 days after implantation, mammary imaging windows were surgically implanted. Mice were then injected daily with 10 mg/kg of Pyr1. As shown in Fig. 5A and in Supplementary Movies S2 and S3, acquired images indicated that cells in the tumors of mice treated with Pyr1 for at least 8 days were less packed than cells in the tumors of the vehicle-treated mice. To quantify this effect, we measured the area covered by fluorescent cells. We found that there were statistically less fluorescent cells in tumors of mice treated with Pyr1 (Fig. 5B). These results are consistent with the reduced number of tumor cells observed in the above-described experiment on subcutaneous MDA-MB-231 xenografts treated with Pyr1 (Supplementary Fig. S6). Indeed, cell proliferation, as assessed by Ki67 staining and quantification, was significantly decreased in tumors of Pyr1-treated mice (Fig. 5C), whereas the number of apoptotic cells, detected with TUNEL staining, was significantly increased in these tumors (Fig. 5D). These observations confirmed the antitumor effect of Pyr1.

Even though the measurement of cell velocity within the tumor did not show any significant difference upon Pyr1 treatment (Fig. 6A), cell morphology was affected. Whereas 90% of cells in vehicle-treated mice are elongated, 60% of cells displayed a rounded morphology in Pyr1-treated mice (Figs. 5A and 6B). The migratory properties of elongated and rounded cells were separately analyzed. Pyr1 treatment induced a significant decrease of both the velocity and the distance covered by elongated cells, as compared with vehicle-treated cells (Fig. 6C and D). In contrast, the velocity of rounded cells almost doubled. The total distance covered by rounded cells was three times higher and their persistence was lower upon Pyr1 treatment, as compared with elongated cells (Fig. 6C–F).

Altogether, our results indicate that in vivo Pyr1 treatment increases the velocity of cell movements in the tumor.

If Pyr1 treatment increases the velocity of malignant cells, it is expected that the migration of the cells and the establishment of distant metastases will be enhanced. Dendra2 protein expression in cells allowed the postmortem evaluation of the colonization of distant organs by tumor cells at the end of the intravital imaging experiment. Quantification of the fluorescent cells in lung cryosections indicated that Pyr1 treatment did not affect the number of metastases, contrary to expectation (Fig. 7A and B). The size of the metastases was however significantly reduced by Pyr1 treatment, as the number of cells per metastasis in Pyr1-treated mice was decreased by 90% (Fig. 7A and C), confirming the strong effect of Pyr1 on cell proliferation.

To further explore the impact of Pyr1 on in vivo metastatic colonization, the aggressive MDA-MB-231-ZNF217rvLuc2 cells (30) were injected directly into the blood stream of Pyr1-treated mice and control mice. Bioluminescence imaging demonstrated that although Pyr1 did not affect the number of metastases, it had a clear impact on the global metastatic load (Fig. 7D–F). Interestingly, this effect was long lasting, as the global metastatic load did not rise steeply even 12 days after the end of the treatment (Fig. 7E). We conclude from this experiment that Pyr1 treatment had no effect on metastasis establishment but induced a strong reduction of metastases growth.

In the Rho pathway, LIMK are the most distal kinases that directly control microtubule and actin dynamics. This position in the signaling network makes them attractive targets for pharmacologic inhibition in a therapeutic perspective. In this study, we have explored the efficacy of the LIMK inhibitor Pyr1 on breast cancers and compared the effects of Pyr1 with those of PTX. We found that Pyr1 was effective on PTX-resistant tumors, leading to the reduction of tumor size, with no detectable adverse side effects on mice. Assessing the mechanism of action of a new therapeutic agent is important (46). In contrast to our in vitro results, we could not detect a statistically significant decrease of the level of cofilin phosphorylation in Pyr1-treated tumors. Several explanations may account for these differences. First, cofilin phosphorylation was measured on whole-tumor extracts and putative intratumoral differences may have been overlooked. Alternatively, as the behavior of cells, including the migration properties, in vivo is often different from that observed on 2-dimensional substrates (44, 45), it is possible that the differences in cofilin phosphorylation and cell migration velocities are due to the 2D in vitro and the 3D in vivo assay systems. In contrast, we found that the amount of detyrosinated tubulin was systematically enhanced in Pyr1-treated tumors, but such an enhancement that occurs upon a null background of detyrosinated tubulin may be easier to detect than a decrease of a normally expressed marker, such as phospho-cofilin.

Although we do not exclude the possibility that Pyr1 exerts its antitumor effect through a modification of actin microfilament dynamics, our data show that its effect strongly correlates with the stabilization of the microtubule network.

The assembly and disassembly of actin in cellular structures, such as lamellipodia and filopodia, is not only regulated by LIMK but has been postulated to be an integral component of LIMK-regulated cell invasion (20, 21). We have analyzed Pyr1 effects, both in vitro and in vivo, on cell motility and invasion. We found that although Pyr1 inhibits in vitro cell motility, its effect is more complex in vivo. First, tumors treated with Pyr1 were heterogeneous, comprising two tumor cell populations: cells with a rounded morphology and cells with an elongated morphology. Second, motility differed according to the cell shape: upon Pyr1 treatment, the motility of elongated cells decreased, while the motility of rounded cells increased. This heterogeneity likely reflects the variations of intratumoral Pyr1 concentrations, which generate two different phenotypic outcomes that differ by their cell motility and cell shape features (47, 48).

The disparity between in vitro and in vivo observations, regarding the effect of Pyr1 on cell motility could have other explanations. Besides differences in drug dosing, differences in the duration of in vivo treatment and of in vitro assay could be at the origin of this disparity.

The change of cell shape and motility parameters observed in Pyr1-treated tumors could also result from modifications of the biomechanical properties of interstitial tissue (49). More generally, a global effect of Pyr1 on the tumor microenvironment could perturb the fine balance between chemical and mechanical signals produced by the different cell types and thus modify the shape and the motility phenotype of tumor cells (50).

Overall, although in vitro Pyr1 has both an antiproliferative and antimigration effect, in vivo experiments indicate that the antiproliferative effects of the drug are stronger than the migration effects.

Although Pyr1 did not inhibit in vivo tumor cell migration and metastasis seeding, we consistently observed that the size of the metastases remained small, regardless whether the cells migrated from the primary tumor or whether they were directly injected into the circulation. Our observations indicate that pharmacologic inhibition of LIMK impairs the proliferation of cancer cells at their new sites of implantation. These findings are consistent with the results obtained by Shibue and colleagues, following experimental implantation of breast cancer cells bearing a constitutively active cofilin mutant that mimics the inhibition of LIMK (20).

Taken together, our data indicate that LIMK inhibitors, such as Pyr1, could represent potent agents to decrease the growth of both primary tumors and their metastasis. Moreover, they are a possible pharmacologic alternative to overcome the tumor resistances frequently observed when patients are treated with taxanes.

R. Prudent is the Chief Operating Officer at and has ownership interest in Cellipse. L. Lafanechère has ownership interest in Cellipse SAS. No potential conflicts of interest were disclosed by the other authors.

Conception and design: C. Prunier, J. Vollaire, O. Destaing, R. Prudent, C. Albiges-Rizo, L. Lafanechère

Development of methodology: C. Prunier, V. Josserand, J. Vollaire, P.A. Cohen, J.-L. Coll, J. van Rheenen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Prunier, V. Josserand, J. Vollaire, E. Beerling, C. Petropoulos, O. Destaing, C. Montemagno, A. Hurbin, L. de Koning

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Prunier, V. Josserand, J. Vollaire, C. Petropoulos, O. Destaing, C. Montemagno, A. Hurbin, R. Kapur, P.A. Cohen, C. Albiges-Rizo, L. Lafanechère

Writing, review, and/or revision of the manuscript: C. Prunier, V. Josserand, A. Hurbin, R. Prudent, L. de Koning, R. Kapur, P.A. Cohen, C. Albiges-Rizo, J.-L. Coll, M. Billaud, L. Lafanechère

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V. Josserand, J. Vollaire, E. Beerling

Study supervision: A. Hurbin, J.-L. Coll, J. van Rheenen, L. Lafanechère

The authors thank C.H. Nguyen for Pyr1 synthesis, C. Lecerf, F. Bard, and A. Barbet for performing the RPPA experiments, M. Barreto from SimplicityBio for statistical analysis, P. Poullet et S. Liva for RPPA data representation, A. Zomer and M. Alieva for technical help for intravital microscopy, K. Sadoul for critical reading of the manuscript.

This work is supported by the CNRS, the INSERM, and by grants from the Fondation de France and from Association Espoir (to L. Lafanechère). It was supported by ANR Project Nanoluc ANR-11-BSV5-0018, France Life Imaging, French program "Investissement d'avenir," grant "Infrastructure d'avenir en biologie", ANR-11-INBS-0006 (J.-L. Coll and A. Hurbin). The RPPA platform is supported by the Cancéropôle Ile-de-France. C. Albiges-Rizo, C. Petropoulos, and O. Destaing were supported by LNCC. C. Prunier was a PhD fellowship from the French Ministry of Research and from the French Foundation ARC and the recipient of a subvention of Région Rhône-Alpes and of an EOLE grant.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.