1

Pharmacological inhibition of LIM Kinase stabilizes microtubules and inhibits neoplastic growth

Renaud Prudent1*, Emilie Vassal-Stermann2,3*, Chi-Hung Nguyen4,5,6, Catherine Pillet2, Anne

Martinez1, Chloé Prunier1, Caroline Barette2, Emmanuelle Soleilhac2, Odile Filhol8, Anne

Beghin9, Glaucio Valdameri10, Stéphane Honoré7, Samia Aci-Sèche 2,11, David Grierson4,5,12,

Juliana Antonipillai13, Rong Li13, Attilio Di Pietro10, Charles Dumontet14, Diane Braguer7, Jean-

Claude Florent4,5, Stefan Knapp15, Ora Bernard 13,16 and Laurence Lafanechère 1,2

1 Institut Albert Bonniot, CRI INSERM/UJF U823, Team 3 “Polarity, Development and Cancer”,

Rond-point de la Chantourne, 38706 La Tronche Cedex, France

2 CEA, DSV, iRTSV/CMBA, 17 rue des Martyrs, Grenoble F-38054, France

3 current address: Laboratoire d'Enzymologie Moléculaire, Institut de Biologie Structurale,

UMR 5075 CNRS-CEA-Université Joseph Fourier, 41, rue Jules Horowitz, 38027 Grenoble

Cedex 1, France

4 Institut Curie, Centre de Recherche, 26 rue d’Ulm, Paris F-75248, France

5 CNRS, UMR 176, 26 rue d’Ulm, Paris F-75248, France

6 Institut Curie, Centre de Recherche, Bâtiment 112, Université Paris-Sud, Orsay F-91405,

France

7 INSERM UMR 911, Centre de Recherche en Oncologie biologique et Oncopharmacologie,

Aix-Marseille Université, Marseille, 13005, France.

8 INSERM, U1036, CEA, iRTSV/BCI, Grenoble, F-38054, France

9 Centre Commun de Quantimétrie, Faculté de Médecine Rockefeller, 8 avenue Rockefeller

69008 Lyon, France

2

10 Equipe Labellisée Ligue 2009, Institut de Biologie et Chimie des Protéines FR 3302, BM2SI

UMR 5086 CNRS/Université Lyon 1, 7 passage du Vercors, 69367 Lyon, France

11 current address: Centre de Biophysique Moléculaire, rue Charles Sadron , 45071 Orléans

cedex 2, France

12 current address: UBC Faculty of Pharmaceutical Sciences #2146 East Mall, Vancouver, BC,

Canada V6T 1Z3

13 St Vincent’s Institute of Medical Research, 9 Princes St Fitsroy, Victoria 3065, Australia

14 INSERM 590, Faculté Rockefeller, 8 Avenue Rockefeller, 69008 Lyon, France and Université

Lyon 1, ISPB, Lyon, F-69003, France

15 Oxford University, Department of Clinical Pharmacology, Old Road Campus Building,

Oxford OX37DQ, UK

16 The University of Melbourne, Department of Medicine, St Vincent’s Hospital, Fitzroy,

Victoria 3065, Australia

Grant Support This work was supported by the CNRS, the CEA, the Institut Curie, the

INSERM, and by grants from the Association pour la Recherche sur le Cancer, the GRAVIT

consortium and from the French National Agency for Research (grant ANR- 2010-EMMA-013-

01). E.V-S. was a fellow of the Région Rhône-Alpes. R.P. is a fellow of Fondation de France.

Contact

LAFANECHERE Laurence; PhD; Institut Albert Bonniot, CRI INSERM/UJF U823,

Team 3 « Polarity, Development and Cancer”

Rond-point de la Chantourne

38706 La Tronche Cedex, France

3

Phone: + (33) 476 54 95 71

e-mail: laurence.lafanechere@ujf-grenoble.fr

Additional Footnotes

* Renaud Prudent and Emilie Vassal-Stermann have contributed equally to this work

4

Abstract

The emergence of tumor resistance to conventional microtubule-targeting drugs restricts their

clinical use. Using a cell-based assay that recognizes microtubule polymerization status to screen

for chemicals that interact with regulators of microtubule dynamics, we identified Pyr1, a cell

permeable inhibitor of LIM Kinase, which is the enzyme that phosphorylates and inactivates the

actin depolymerizing factor cofilin. Pyr1 reversibly stabilized microtubules, blocked actin

microfilament dynamics, and inhibited cell motility in vitro and showed anticancer properties in

vivo, in the absence of major side effects. Pyr1 inhibition of LIM Kinase caused a microtubule

stabilizing effect, which was independent of any direct effects on the actin cytoskeleton. In

addition, Pyr1 retained its activity in multidrug resistant cancer cells that were resistant to

conventional microtubule targeting agents. Our findings suggest that LIM Kinase functions as a

signaling node that controls both actin and microtubule dynamics. LIM Kinase may therefore

represent a targetable enzyme for cancer treatment.

5

Introduction

Microtubules are filaments composed of α-tubulin and β-tubulin heterodimers. They are highly

dynamic polymers and their dynamics is tightly regulated by a balance between activities of

microtubules-stabilizing and microtubule destabilizing proteins (1-3).

Moreover, the interaction of the microtubule growing tips with some proteins such as CLIP 170

results in microtubule stabilization (4, 5). The binding of all of these proteins to tubulin is tightly

regulated, by phosphorylation/dephosphorylation processes (6, 7) or, directly, by tubulin

modifications such as the tubulin tyrosination cycle (8, 9).

A detailed knowledge of the different players and how they interact both spatially and temporally

to regulate microtubule functions is, however, still lacking.

Because of microtubules key role in mitosis, microtubule-targeting agents are powerful

anticancer drugs. Tubulin is now considered as one of the most highly validated cancer target (10,

11). These drugs have, however, some limitations due to side effects, principally

myelosuppression and neurotoxicity. Moreover, many cancers are, or become, resistant to these

drugs (12). This is often the result of multidrug resistance caused by overexpression of ATP-

binding cassette transporters (13). Several strategies have been proposed for the development of

more effective and less toxic anticancer drugs. One of them is to identify drugs targeting proteins

that regulate microtubule functions, and that would lead to mitotic arrest and/or apoptosis.

The aim of this study was to identify chemical compounds able to interact with regulators of

microtubule dynamics. We therefore performed a screening using a cell-based assay which

probes the microtubule polymerization status (14). We identified a compound (Pyr1) that slows

down microtubule dynamics. Pyr1 is toxic for cancerous cell-lines, including drug-resistant cell

6

lines. Pilot studies on mice with xenografted tumors indicate that Pyr1 prevented tumor growth

and was well tolerated. Pyr1 also affected the actin cytoskeleton and inhibited cell motility. We

have identified LIM kinases (LIMKs) as the main targets of Pyr1. The activity of LIMK1/2 is

regulated mainly by the Rho-dependent kinases ROCK1 (15) and ROCK2 (16) and the p21

activated kinases PAK1 (17) and PAK4 (18) and inactivated mainly by the Slingshot

phosphatases (19). Phosphorylation and inhibition of the actin depolymerizing protein cofilin by

LIMKs is the last step of a cascade that regulates actin polymerization. Here we demonstrate that

Pyr1 inhibits cofilin phosphorylation. We established that the microtubule stabilizing effect of

Pyr1 results from its inhibitory effect on LIMKs and is independent of its effect on the actin

cytoskeleton. Thus LIMKs inhibition by Pyr1 explains the observed phenotypes on both actin and

tubulin.

Material and Methods

Screen for chemical modulators of microtubule dynamics

The screen was performed as described in Vassal et al. (14). For a summarized description see

the Supplementary methods.

Chemical reagents, recombinant and purified proteins, plasmids, antibodies and cell lines:

All reagents used are indicated in the Supplementary methods.

Immunofluorescence microscopy

To visualize microtubules, cells were seeded either on glass coverslips or in microplates

(ViewPlate-96, PerkinElmer Inc., USA). They were permeabilized in warm OPT buffer (80 mM

Pipes, 1 mM EGTA 1 mM MgCl2, 0.5% triton X-100, and 10% glycerol, pH 6.8) and fixed for 6

min in -20°C methanol. To visualize F-actin, cells were washed in warm PBS and fixed with 4%

PFA in PBS at 37°C, followed by cell permeabilization with 0.1% Triton X-100 in PBS. After

7

incubation with primary antibodies, cells were incubated with Cy3 (Jackson ImmunoResearch

Laboratories), or Alexa 488 (Invitrogen) secondary antibodies. To visualize F-actin, Alexa 488-

phalloidin (Invitrogen) was included with the secondary antibody.

Images were captured either with a charge-coupled device camera (CoolSNAP ES; Roper

Scientific) in a straight microscope (Nikon Eclipse 90i) controlled by Nikon software (Universal

Imaging Corp.) using a 40X, 60X or 100X Plan-Neofluar oil objectives, with an IN Cell Analyzer

1000 automated microscope (GE-Healthcare) using a 40X objective, controlled by IN Cell

software allowing automated cell imaging or with an Orca R2 N/B camera (Hamamatsu) in a

straight microscope (Zeiss AxioImager Z1) controlled by Axiovision software (Carl Zeiss) using

a 63X oil objective. When necessary, cells were observed with a confocal microscope (Leica).

Immunofluorescence analysis of growing microtubules + ends

HeLa cells were incubated for 2 hours with the tested compound, fixed for 6 min in -20°C

methanol and processed for immunofluorescence using anti-EB1 specific antibody.

Tubulin polymerization assay

Microtubule polymerization assay was adapted from (8). Briefly, microtubule assembly was

carried out in a half area 96-well black plate (Greiner, #675090) using a microplate reader

FLUOstar OPTIMA (BMG Labtechnologies). Wells were charged with either microtubule

proteins or pure tubulin (final concentration of 25 µM and 30 µM, respectively) in MEM buffer

with 10 µM DAPI and variable concentrations of compounds. Following 10 minutes incubation,

assembly was started by injection of GTP and MgCl2 to a final concentration of 1 mM and 5 mM

respectively, in 100 µL. The excitation and emission wavelengths were set at 360 and 450 nm,

respectively, and the fluorescence of microtubule-bound DAPI was monitored as a function of

time at 37°C. Fluorescence signal at time 0 was subtracted from each of the subsequent

fluorescence readings. Each compound was assayed in triplicate.

8

Actin-pyrene polymerization assay

Mg-ATP-monomeric actin (2 μM, 10% pyrene labeled) in MME buffer containing 0.2 mM ATP

was polymerized at room temperature after addition of one-tenth volume of 10X-buffer (500 mM

KCl, 20 mM MgCl2, 2 mM EGTA, 100 mM imidazole, pH 7.5). Experiments were conducted in

the presence or absence of different compounds, as indicated in the figure legend. The

polymerization was followed by changes in pyrene fluorescence using a Xenius SAFAS

fluorimeter (Safas SA, Monaco).

Tumor xenograft experiments

12 B6D2F1 female mice were injected subcutaneously in the right flank with 2.106 L1210 cells.

Mice were then randomly separated into two groups of 6 mice each. In one group, Pyr1 (10

mg/Kg/day in PEG400) was injected intraperitoneally, on the left flank, daily for 10 days. In the

other group, PEG400 alone was injected, daily for 10 days. Each day the mice were observed for

survival. They were weighted twice a week. The experiment was stopped 20 days after the last

vehicle treated mice death. The ethics committee of Lyon1 University validated this experimental

protocol.

All the other experimental procedures are described in the supplementary methods section

Results

Hits selection process

The screening assay is based on the properties of the tubulin enzymes involved in the tubulin

tyrosination cycle, tubulin tyrosine ligase and tubulin carboxypeptidase, Because of their

substrate properties, dynamic microtubules, are composed of tyrosinated tubulin (Tyr-tubulin)

whereas stabilized microtubules are composed of detyrosinated tubulin (Detyr-tubulin) (20). Thus

depolymerization or stabilization of the microtubule network can be distinguished by double-

9

immunofluorescence using specific Tyr- and Detyr-tubulin antibodies (14). We selected 16

compounds that enhance the Detyr-tubulin signal by 20%. After checking the compounds

structure, only 5 chemically non-reactive compounds that generate stable Detyr-microtubules in

cells were selected for further characterization (Supplementary Table 1).

It is well established that sodium azide can stabilize microtubules, thus generating Detyr-tubulin

(21). To eliminate compounds acting via similar mechanisms, we analyzed compounds effects on

mitochondria and discarded compounds targeting mitochondria as did sodium azide

(Supplementary Table 1). We also eliminated compounds with no analogues, leaving one

remaining compound, 9-Benzoyloxy-5,11-dimethyl-2H,6H-pyrido[4,3-b]carbazol-1-one, (Pyr1,

Fig. 1a). Incubation of HeLa cells with 25 µM of Pyr1 resulted in Detyr-microtubules generation

(Fig. 1b).

Pyr1 suppresses microtubule dynamic instability without directly targeting tubulin

To check if the enrichment of Detyr-microtubules results from an effect on the tubulin

tyrosination enzymes or from an effect on microtubules dynamics, we first studied the resistance

of the microtubule network to nocodazole-induced depolymerization. Nocodazole binds free

tubulin and prevents its incorporation into microtubules, inducing microtubule depolymerization.

Microtubules with slow dynamics have reduced exchanges with the free tubulin pool and are thus

less sensitive to nocodazole-induced depolymerization (22). We found that a pretreatment of cells

with Pyr1 protect the microtubule network from nocodazole-induced depolymerization (Fig. 2a).

We next examined the distribution of the plus-end tracking protein (+TIP) EB1 (End Binding

protein 1) that specifically probes growing microtubule plus ends. Cells treated with DMSO, 5

µM paclitaxel, or 10 and 25 µM Pyr1 were stained with an anti-EB1 specific antibody (Fig. 2b).

We observed a decrease of the number of EB1 comets per cell in paclitaxel-treated cells (92%

reduction of the number of EB1 comets) as well as in cells treated with Pyr1 (84% reduction of

10

the number of EB1 comets for 25 µM Pyr1), together with a size reduction of remaining EB1

comets (45% and 27 % reduction of the comets surface for 5 µM paclitaxel and 25µM Pyr1,

respectively) suggesting a strong reduction of microtubule growth dynamics (Supplementary

Fig.1).

We then analyzed the effect of Pyr1 on microtubule dynamic instability parameters, using time-

lapse fluorescence microscopy on GFP-EB3 transfected cells (23) (Supplementary Movie 1). As

expected, Pyr1 suppressed the microtubule growth rate, strongly reduced the microtubule growth

length as indicated by the dose-dependent increase of the distance-based catastrophe frequency,

and increased time spent in pause. Such effects have been previously described for microtubule

stabilizing agents (24) (Supplementary Table 2).Taken together, these experiments clearly show

that Pyr1 suppresses microtubule dynamics instability.

Close analysis of the morphology of the mitotic spindles revealed that the spindles of cells treated

with the compound were abnormal, with disorganized asters. We found that 84.2 % of the mitotic

spindles were normal in DMSO-treated cells and 12.8 % abnormal, whereas 9.7 % of the mitotic

spindles were normal in Pyr1-treated cells and 82.9 % were abnormal. These defects were clearly

different from the multiple asters observed in cells treated with paclitaxel. Spindles depolymerize

in nocodazole- treated cells (Fig. 2c).

As agents interfering with microtubule dynamics lead to a G2-M cell cycle arrest, we analyzed

Pyr1 effect on the cell cycle, by flow cytometry. Although Pyr1 effect was not as strong as that of

nocodazole or paclitaxel, we found that it induced a cell cycle arrest at the S-G2/M phases (Fig.

2d).

To test whether Pyr1 directly interacts with tubulin, we assayed it in an in vitro tubulin assembly.

Paclitaxel is a well-characterized microtubule-stabilizing agent that can induce in vitro tubulin

assembly at low tubulin concentrations that would not induce spontaneous tubulin assembly. We

11

found that Pyr1 could not induce in vitro tubulin assembly, neither when assayed on tubulin in

the presence of associated protein (microtubule proteins, Fig. 2e) nor on pure tubulin (Fig. 2f).

This experiment strongly suggests that Pyr1 does not interact directly with tubulin.

Pyr1 overcomes drug resistant (MDR) cell phenotype

Because Pyr1 was able to stabilize microtubules, with a mechanism of action different to that of

taxanes, we tested if it could be of therapeutic interest. We first examined its toxicity on several

cancerous cell lines, including drug-resistant cell lines overexpressing the glycoprotein P or

transporters such as ABCG2 and MRP1 that renders cells insensitive to current chemotherapies.

Pyr1 reduced the viability of these cell lines in a dose-dependent manner, with a GI50 (50% of

growth inhibition) in the micromolar range (Table 1).

Interestingly, Pyr1 toxicity is the same for the drug-sensitive human cell lines, and for their

multidrug-resistant counterparts, indicating that it is not substrate of the glycoprotein P or of

ABCG2 and MRP1 transporters.

Pyr1 is well tolerated in mice and shows anticancer activity

The core structure of Pyr1 is similar to that of ellipticine, an alkaloid with proven antineoplastic

activity, through DNA intercalation and inhibition of topoisomerase II. The ellipticine derivative

N2-methyl-9-hydroxyellipticium has been used in the treatment of breast cancer, but it has been

retrieved due to its toxicity. Previous studies have shown that Pyr1 is a weak DNA intercalator

and has no effect on topoisomerase II (25).

To determine whether Pyr1 could serve as chemotherapeutic agent we examined its toxicity in

mice. Mice received daily intraperitoneal injections of Pyr1 (from 15 to 60 mg/Kg), during 7 days.

Transient signs of discomfort related to administration of the vehicle, were observed in all

animals (Supplementary Table 3). The weights of Pyr1-treated animals and vehicle-treated

animals were not significantly different (Supplementary Table 4). Hematological parameters,

12

were normal after Pyr1 treatment. The only small changes observed for the highest doses were

not considered as having significant toxicity (Supplementary Table 5). After treatment period, a

full post-mortem examination was performed. Yellow foci on the liver, for mice that received

high-dose of Pyr1, were observed which correlated with acute to sub-acute peritoneal

inflammation seen in the hepatic capsule (data not shown) but without signs of severe hepatic

damage. Thus, under these experimental conditions, no severe adverse effects related to Pyr1

administration were observed up to the dose of 30 mg/kg/day, indicating that Pyr1 is well

tolerated.

We then conducted a pilot study using an in vivo model of leukemia L1210-bearing mice of the

B6D2F1strain. The injected dose of Pyr1 was chosen to be 3 times lower than the dose (30

mg/kg/day) for which the first adverse signs were detected. Vehicle treated mice were all dead at

day 70, whereas all the Pyr1 treated mice survived (Fig. 3) and were still alive 20 days later,

when the experiment was stopped. Thus, Pyr1 induced a complete survival gain (p < 0.002) with

no apparent toxicity (no loss of weight, data not shown).

These experiments indicate that Pyr1 has a good therapeutic efficacy and is well tolerated.

Pyr1 affects the organization of F-actin but does not directly target it

We then tried to identify Pyr1 cellular targets. As the microtubule and actin microfilament

networks are highly interconnected (26), we investigated the effect of Pyr1 on the actin

cytoskeleton. Cells treated with Pyr1 were stained for filamentous actin (F-actin) using

fluorescein-phalloidin. Pyr1 affected the organization of the actin microfilaments, which was not

observed when cells were exposed to paclitaxel (Fig. 4a).

Because of the central role of actin dynamics in cell motility, we examined the effect of Pyr1 on

MCF10A cells motility, using time-lapse video microscopy. We found that Pyr1 rapidly and

completely blocked cell motility, even at the lowest concentration assayed (10 µM) (Fig. 4b).

13

Pyr1 was not able to directly affect actin assembly in vitro, and could not modify actin assembly

kinetics, contrary to phalloidin, which can stabilize F-actin in vitro and to latrunculin B which

inhibits actin assembly (Fig. 4c). These results indicate that Pyr1 does not directly interact with

actin.

We have also tested the possibility that the effect of Pyr1 on F-actin influences microtubule

dynamics which could indirectly result in the stabilization of microtubules. We thus pre-treated

cells with cytochalasin D, which depolymerizes F-actin (27), and examined Pyr1 effect on

microtubule dynamics (Supplementary Fig. 2). We observed that even when the microfilaments

are completely depolymerized, Pyr1 still induces the formation of Detyr-microtubules, indicating

that the microtubule network is stabilized. This result strongly suggests that Pyr1-induced

microtubule stabilization is independent of the inhibitor effect on the actin network.

Taken together, these findings indicate that Pyr1 targets regulators of actin as well as tubulin.

Pyr1 selectively inhibits LIM Kinases activity

Previous screening identified a class of Pyr1 structurally related compounds that often targets

protein kinases (25, 28). We therefore tested the ability of Pyr1 to inhibit the activity of a panel of

66 protein kinases known to be involved in the regulation of the cytoskeleton. Pyr1 inhibited only

the activities of a member of NIMA-related kinase family (NEK11), Mixed Lineage Kinase 1

(MLK1) and LIM kinase 1 (LIMK1) (Fig. 5a and Supplementary Table 6). The highest

inhibition was observed for LIMK1, which showed only 4% residual in vitro kinase activity. A

well-established substrate of LIMK1 is cofilin, a protein that regulates actin dynamics. Using

cofilin as substrate, we found that Pyr1 inhibited the activity of recombinant LIMK1 and LIMK2

(Fig. 5b and 5c) in vitro with an IC50 of 50 and 75 nM, respectively (Supplementary Fig. 3a).

The inhibitor behaved as an ATP competitor (Supplementary Fig. 3b).

14

As selectivity is a major issue for ATP-competitive kinase inhibitors we extended the evaluation

of the compound on the activity of 45 additional different kinases, using a thermal stability shift

assay (29). We did not find significant interaction of the identified inhibitor with any of the

screened kinases except LIMK1, which showed a Tm shift of 8.16oC, consisted with the high

potency of this inhibitor (Supplementary Table 7).

Another substrate of LIMK1 is p25 (30). We investigated the ability of Pyr1 to inhibit the in vitro

phosphorylation of bacterially expressed and purified human GST-p25 by recombinant LIMK1.

Although the phosphorylation of GST-p25 was not efficient in control conditions, we found that

Pyr1 inhibits it (Fig. 5d).

To ascertain that Pyr1 can also inhibit LIMK activity in cells, we examined the level of phospho-

cofilin (P-cofilin) in cells treated for two hours with different concentrations of Pyr1. The level of

P-cofilin was greatly reduced with increasing amounts of Pyr1, while the protein levels of cofilin

were not affected. A complete inhibition of cofilin phosphorylation was observed after incubation

of cells with 25 µM of Pyr1 (Fig. 5e). This concentration was chosen to determine the time

course of Pyr1 effect in cells. Complete inhibition of cofilin phosphorylation was observed 20

minutes after addition of Pyr1 (Supplementary Fig. 3c).

We also investigated whether the inhibition of LIMK1 activity by Pyr1 was reversible in vitro

and in cells. To achieve this, the kinase activity of a LIMK1-Pyr1 mix was assayed before and

after size-exclusion chromatography, allowing the potential separation of the kinase from

unbound inhibitor. After gel-filtration, a significant increase in LIMK activity was observed

(Supplementary Fig. 3d). Western blot of protein lysates prepared from cells first treated with

Pyr1, and then re-incubated in fresh medium without Pyr1, showed that cofilin phosphorylation

was again detectable 30 min after Pyr1 washout and reached basal levels after two hours

(Supplementary Fig. 3e). These results clearly show that Pyr1 binding to LIMK is reversible.

15

Similarly, the suppression of microtubule dynamic instability by Pyr1 observed in HeLa cells was

also found reversible (Supplementary Table 8).

We also checked if Pyr1 inhibits NEK11 and MLK1 activities. In vitro, Pyr1 has weak activity on

both NEK11 and MLK1 (Supplementary Fig. 4ab). Pyr1 effect was also checked in cells.

NEK11 is known to destabilize Cdc25A (31) and a downstream target of MLK1 is Jun Kinase.

Looking on Cdc25 expression and on the phosphorylation status of Jun Kinase, respectively, we

found that Pyr1 has no effect on the activity of neither NEK11 nor MLK1 (Supplementary Fig.

4cd).

LIMK inhibition causes microtubule stabilization

The identification of LIMK1/2 as targets of Pyr1 strongly suggested that the inhibition of cofilin

phosphorylation is the mechanism that underlies the actin network reorganization and the loss of

cell motility of cells treated with Pyr1. The link between LIMK inhibition and the observed

microtubule stabilization is, however, less obvious. To investigate this issue we compared the

potency of several Pyr1 structural analogues to inhibit LIMK1 activity in vitro (Supplementary

Table 9), to modify actin microfilaments dynamics, as assessed by the resistance of Pyr1-induced

actin structures to latrunculin depolymerization, to induce the generation of Detyr-microtubules

(Supplementary Fig. 5a) and to inhibit cofilin phosphorylation in cells (Supplementary Fig.

5b). We found a good correlation between the levels of LIMK1 inhibition by the tested

compounds and the stabilization of both the actin structures and the microtubules. These

experiments also allowed the determination of Pyr1 structure-activity relationships

(Supplementary Table 9). In another experiment, we tested if (N-[5-[2,6-Dichloro-phenyl]-5-

difluoromethyl-2H-pyrazol-3-yl]thiazol-2-yl)-isobutyramide, a recently described LIMK

inhibitor (32-34), also called LIMKi (Fig. 6a), was able to stabilize microtubules. We

16

demonstrated that LIMKi was indeed able to increase the Detyr-microtubules cell content (Fig.

6b) and to stabilize them to a nocodazole induced depolymerization (data not shown). These

results indicate that a LIMK inhibitor structurally different from Pyr1 is also able to induce a

stabilization of the microtubule network.

We then tested if the overexpression of LIMK1 could counteract the microtubule stabilizing

effect of Pyr1. Cells overexpressing LIMK1 (Fig. 6c) were treated with increasing concentrations

of Pyr1 and microtubule stability was then probed with nocodazole, as described above. We

found that, at all the concentrations assayed, the microtubule network of the LIMK1 expressing

cells treated with Pyr1 was less stabilized, as compared with control cells (Fig. 6d).

Finally, we analyzed the effect of LIMKs invalidation on microtubule stabilization, as assessed

by the generation of Detyr-microtubules. We invalidated LIMK1 and LIMK2 into MCF7 cells

and found that LIMKs invalidation not only decreases cofilin phosphorylation, but also increases

Detyr-tubulin content (Supplementary Fig. 6). We found that LIMKs expression, especially that

of LIMK2, was more difficult to silence into Hela cells. We succeeded, however, using shRNA

lentivirus, to diminish LIMK1 expression in such cells. In these conditions, we found a large

amount of cells with Detyr-microtubules, indicating that microtubules are stabilized

(Supplementary Fig.6).

Taken together, our results demonstrate that specific inhibition of LIMK activity by the small

molecule inhibitor Pyr1 is responsible of its effects on the cytoskeletal proteins tubulin and actin.

Discussion

Our strategy based on a screening of a chemical library led to the discovery of a cell-permeable

inhibitor of LIMK1/2, enzymes known to phosphorylate and inactivate cofilin, an actin-

depolymerizing factor. As the readout of the assay used for the library screening was the

immunofluorescence detection of a tubulin modification indicative of the cellular microtubule

17

stability status, it was somewhat surprising that this approach led to the identification of an

inhibitor of a known actin regulator. However, such a whole cell-based assay probes for

inhibitors of entire pathways, allowing the cell biology to dictate the best targets (35). Using

cytochalasin D, an actin-depolymerizing agent, we demonstrated that the effect of Pyr1 on

microtubules was independent of its induced actin cytoskeleton reorganization. Thus, this

screening approach allowed the selection of a chemical compound able to target LIMKs, a shared

signaling regulation node of the actin and the microtubule cytoskeletons.

Because of Pyr1 ability to suppress microtubules dynamic instability, we tested if it has

antimitotic and cytotoxic properties on cancer cell lines. We found that this was the case.

Moreover, Pyr1 had the same toxicity for the drug-sensitive human cell lines and for their

multidrug-resistant counterparts. This suggests that Pyr1 may be used as an alternative or in

addition to standard chemotherapy, in drug resistant tumors. Our pilot study conducted on tumors

xenografts in mice indicated that Pyr1 was indeed able prevent tumor growth at doses that are

well tolerated by the animals. Moreover, because it is active at doses much lower than the doses

where the first signs of toxicity appear, Pyr1 has a convenient therapeutic window.

Although we found a close correlation between LIMK inhibition, actin disorganization and

microtubule stability, the mechanism by which Pyr1 induces microtubule stabilization is not

completely understood. It was demonstrated that phosphorylation of p25 by LIMK1 inhibits its

ability to polymerize microtubules, resulting in reduced levels of stable microtubules (30).

Therefore a Pyr1-induced dephosphorylation of p25 could be responsible of the observed

stabilization of microtubules. We cannot, however, rule out the possibility that Pyr1-induced

microtubule stabilization results from an effect of Pyr1 on other targets.

18

The mitotic spindles of cells treated with Pyr1 show a disorganization of astral microtubules. This

phenotype is similar to that obtained by Kaji et al. after LIMK1 knockdown (36), further

supporting our conclusion that the LIMKs are the main cellular targets of Pyr1.

Pyr1 induced rapid and almost complete cofilin dephosphorylation in vitro and in cells. This

indicates that the basal activity of cofilin phosphatases (slingshot and chronophin) must be high

in the cells tested and that Pyr1 may also targets closely related cofilin kinases, i.e. TESK.

Pyr1 acts as an ATP competitive inhibitor of LIMK. A major concern in the use of kinase

inhibitors that are ATP antagonist is their target specificity, since the ATP-binding motif is

present all kinases as well as in other proteins. We therefore tested Pyr1 ability to inhibit the in

vitro activity of 110 protein kinases and although it is possible that Pyr1 targets other kinases

such as TESK, than the ones already assayed, it shows, however, a high selectivity. The kinases

assayed include some vital kinases such as the insulin receptor kinase (IR). The absence of Pyr1

effect on these kinases could contribute to the observed limited animal toxicity of the compound.

It is anticipated that inhibition of cofilin phosphorylation that results in activation of cofilin

would lead to actin severing and depolymerization of actin filaments. After two hours of

treatment with Pyr1, we observed a complete disorganization of the actin microfilament network

with decreased level of F-actin. This phenotype was not previously observed when LIMK1 and

LIMK2 were knocked down by specific siRNA (32, 34). A possible explanation of this disparity

is that there are major differences between the mechanism of action of siRNA and

pharmacological inhibitors. Indeed, additionally to differences in kinetics, siRNA experiments

reduce protein levels, whereas a small molecule only inhibits protein enzymatic activity.

Moreover, LIMKs contain protein-protein interaction domains. They, thus, may have a variety of

yet uncharacterized functions, in addition to their enzymatic activity. These other functions might

be involved in the observed reorganization of the actin network.

19

One major advantage of small molecules, particularly when their effect is reversible, as is the

case for Pyr1, is that they allow high temporal and spatial control. Their effects can be observed

in real time by live cell imaging. To date only two cell-permeable inhibitors of LIMK have been

described with limited experimental data regarding their specificity (32-34, 37, 38). Moreover,

LIMK is the last node in the signaling pathway that regulates the actin and the microtubule

cytoskeletons. Thus, Pyr1, with its high specificity, is a new reagent that can be added to the

biologist toolbox for dissecting cytoskeleton dependent mechanisms and to investigate the effect

of acute and chronic inhibition of LIMKs, on different cells and organisms.

We have shown that LIMKs inhibition totally blocks in vitro cell migration. It has been recently

shown, using siRNA or a small molecule inhibitor, that the activity of LIMK1/2 is required for

invasive path generation by tumor and tumor-associated stromal cells (34). In contrast to our

results, the authors found that LIMK inhibition did not affect cell motility. This disparity can be

explained by the differences in the cell lines studied. The activity of cofilin kinases and cofilin

phosphatases may vary among different cells. Yet, small changes in the cofilin phosphorylation

state may have a great impact on the cell invasiveness (34). Differences in the affinity for LIMKs

or in the selectivity of the two inhibitors could also explain their different effects on cell motility.

Nevertheless, the cofilin pathway (39) including its upstream effectors ROCK, PAK and LIMK

(40-43) was proposed as a validated target for the treatment of cancer metastasis. Hence, in

addition to its use in basic research and possibly its therapeutic use as an alternative antimitotic

compound in the treatment of paclitaxel resistant tumors, Pyr1 has the potential to be used for the

inhibition of cancer metastasis.

Acknowledgments

The authors thank Drs Laurent Blanchoin, Raja Boujemaa-Paterski , Claude Cochet, Yasmina

Saoudi, Manuel Théry, Jérémie Gaillard, Florence Mahuteau, Marie-Paule Teulade-Fichou for

20

their advices and help. Marianne Bombled and Doriane Poloni for technical assistance. The

authors also thank the ChemAxon company (http://www.chemaxon.com).

21

References

1. Andersen SS. Spindle assembly and the art of regulating microtubule dynamics by MAPs and Stathmin/Op18. Trends Cell Biol. 2000;10:261-7. 2. Niethammer P, Kronja I, Kandels-Lewis S, Rybina S, Bastiaens P, Karsenti E. Discrete states of a protein interaction network govern interphase and mitotic microtubule dynamics. PLoS Biol. 2007;5:e29. 3. Lieuvin A, Labbe JC, Doree M, Job D. Intrinsic microtubule stability in interphase cells. J Cell Biol. 1994;124:985-96. 4. Akhmanova A, Hoogenraad CC. Microtubule plus-end-tracking proteins: mechanisms and functions. Curr Opin Cell Biol. 2005;17:47-54. 5. Small JV, Kaverina I. Microtubules meet substrate adhesions to arrange cell polarity. Curr Opin Cell Biol. 2003;15:40-7. 6. Cassimeris L, Spittle C. Regulation of microtubule-associated proteins. Int Rev Cytol. 2001;210:163-226. 7. Johnson GV, Stoothoff WH. Tau phosphorylation in neuronal cell function and dysfunction. J Cell Sci. 2004;117:5721-9. 8. Peris L, Thery M, Faure J, Saoudi Y, Lafanechere L, Chilton JK, et al. Tubulin tyrosination is a major factor affecting the recruitment of CAP-Gly proteins at microtubule plus ends. J Cell Biol. 2006;174:839-49. 9. Peris L, Wagenbach M, Lafanechere L, Brocard J, Moore AT, Kozielski F, et al. Motor-dependent microtubule disassembly driven by tubulin tyrosination. J Cell Biol. 2009;185:1159-66. 10. Zhou J, Giannakakou P. Targeting microtubules for cancer chemotherapy. Curr Med Chem Anticancer Agents. 2005;5:65-71. 11. Dumontet C, Jordan MA. Microtubule-binding agents: a dynamic field of cancer therapeutics. Nat Rev Drug Discov. 2010;9:790-803. 12. Rowinsky EK, Donehower RC. Antimicrotubule agents, In Cancer Chemotherapy and Biotherapy. Philadelphia: Lippincott-Raven; 1996. 13. Haimeur A, Conseil G, Deeley RG, Cole SP. The MRP-related and BCRP/ABCG2 multidrug resistance proteins: biology, substrate specificity and regulation. Curr Drug Metab. 2004;5:21-53. 14. Vassal E, Barette C, Fonrose X, Dupont R, Sans-Soleilhac E, Lafanechere L. Miniaturization and validation of a sensitive multiparametric cell-based assay for the concomitant detection of microtubule-destabilizing and microtubule-stabilizing agents. J Biomol Screen. 2006;11:377-89. 15. Ohashi K, Nagata K, Maekawa M, Ishizaki T, Narumiya S, Mizuno K. Rho-associated kinase ROCK activates LIM-kinase 1 by phosphorylation at threonine 508 within the activation loop. J Biol Chem. 2000;275:3577-82. 16. Sumi T, Matsumoto K, Nakamura T. Specific activation of LIM kinase 2 via phosphorylation of threonine 505 by ROCK, a Rho-dependent protein kinase. J Biol Chem. 2001;276:670-6. 17. Edwards DC, Sanders LC, Bokoch GM, Gill GN. Activation of LIM-kinase by Pak1 couples Rac/Cdc42 GTPase signalling to actin cytoskeletal dynamics. Nat Cell Biol. 1999;1:253-9. 18. Soosairajah J, Maiti S, Wiggan O, Sarmiere P, Moussi N, Sarcevic B, et al. Interplay between components of a novel LIM kinase-slingshot phosphatase complex regulates cofilin. Embo J. 2005;24:473-86. 19. Edwards DC, Gill GN. Structural features of LIM kinase that control effects on the actin cytoskeleton. J Biol Chem. 1999;274:11352-61. 20. Lafanechere L, Job D. The third tubulin pool. Neurochem Res. 2000;25:11-8. 21. De Brabander M, Geuens G, Nuydens R, Willebrords R, De Mey J. Microtubule stability and assembly in living cells: the influence of metabolic inhibitors, taxol and pH. Cold Spring Harb Symp Quant Biol. 1982;46 Pt 1:227-40. 22. Saoudi Y, Fotedar R, Abrieu A, Doree M, Wehland J, Margolis RL, et al. Stepwise reconstitution of interphase microtubule dynamics in permeabilized cells and comparison to dynamic mechanisms in intact cells. J Cell Biol. 1998;142:1519-32. 23. Honore S, Braguer D. Investigating microtubule dynamic instability using microtubule-targeting agents. Methods Mol Biol. 2011;777:245-60.

22

24. Kamath K, Jordan MA. Suppression of microtubule dynamics by epothilone B is associated with mitotic arrest. Cancer Res. 2003;63:6026-31. 25. Prudent R, Moucadel V, Nguyen CH, Barette C, Schmidt F, Florent JC, et al. Antitumor Activity of Pyridocarbazole and Benzopyridoindole Derivatives that Inhibit Protein Kinase CK2. Cancer Res. 2010;70:9865-74. 26. Rodriguez OC, Schaefer AW, Mandato CA, Forscher P, Bement WM, Waterman-Storer CM. Conserved microtubule-actin interactions in cell movement and morphogenesis. Nat Cell Biol. 2003;5:599-609. 27. Kueh HY, Charras GT, Mitchison TJ, Brieher WM. Actin disassembly by cofilin, coronin, and Aip1 occurs in bursts and is inhibited by barbed-end cappers. J Cell Biol. 2008;182:341-53. 28. Hoang TM, Favier B, Valette A, Barette C, Nguyen CH, Lafanechere L, et al. Benzo[e]pyridoindoles, novel inhibitors of the aurora kinases. Cell Cycle. 2009;8:765-72. 29. Fedorov O, Marsden B, Pogacic V, Rellos P, Muller S, Bullock AN, et al. A systematic interaction map of validated kinase inhibitors with Ser/Thr kinases. Proc Natl Acad Sci U S A. 2007;104:20523-8. 30. Acevedo K, Li R, Soo P, Suryadinata R, Sarcevic B, Valova VA, et al. The phosphorylation of p25/TPPP by LIM kinase 1 inhibits its ability to assemble microtubules. Exp Cell Res. 2007;313:4091-106. 31. Melixetian M, Klein DK, Sorensen CS, Helin K. NEK11 regulates CDC25A degradation and the IR-induced G2/M checkpoint. Nat Cell Biol. 2009;11:1247-53. 32. Scott RW, Hooper S, Crighton D, Li A, Konig I, Munro J, et al. LIM kinases are required for invasive path generation by tumor and tumor-associated stromal cells. J Cell Biol.191:169-85. 33. Ross-Macdonald P, de Silva H, Guo Q, Xiao H, Hung CY, Penhallow B, et al. Identification of a nonkinase target mediating cytotoxicity of novel kinase inhibitors. Mol Cancer Ther. 2008;7:3490-8. 34. Scott RW, Hooper S, Crighton D, Li A, Konig I, Munro J, et al. LIM kinases are required for invasive path generation by tumor and tumor-associated stromal cells. J Cell Biol. 2010;191:169-85. 35. Peterson JR, Mitchison TJ. Small molecules, big impact: a history of chemical inhibitors and the cytoskeleton. Chem Biol. 2002;9:1275-85. 36. Kaji N, Muramoto A, Mizuno K. LIM kinase-mediated cofilin phosphorylation during mitosis is required for precise spindle positioning. J Biol Chem. 2008;283:4983-92. 37. Harrison BA, Whitlock NA, Voronkov MV, Almstead ZY, Gu KJ, Mabon R, et al. Novel class of LIM-kinase 2 inhibitors for the treatment of ocular hypertension and associated glaucoma. J Med Chem. 2009;52:6515-8. 38. Manetti F. LIM kinases are attractive targets with many macromolecular partners and only a few small molecule regulators. Med Res Rev. 2011. 39. Wang W, Eddy R, Condeelis J. The cofilin pathway in breast cancer invasion and metastasis. Nat Rev Cancer. 2007;7:429-40. 40. Eswaran J, Soundararajan M, Knapp S. Targeting group II PAKs in cancer and metastasis. Cancer Metastasis Rev. 2009;28:209-17. 41. Ding Y, Milosavljevic T, Alahari SK. Nischarin inhibits LIM kinase to regulate cofilin phosphorylation and cell invasion. Mol Cell Biol. 2008;28:3742-56. 42. Horita Y, Ohashi K, Mukai M, Inoue M, Mizuno K. Suppression of the invasive capacity of rat ascites hepatoma cells by knockdown of Slingshot or LIM kinase. J Biol Chem. 2008;283:6013-21. 43. Scott RW, Olson MF. LIM kinases: function, regulation and association with human disease. J Mol Med. 2007;85:555-68.

23

Figure legends

Figure 1: Effect of Pyr1 on cellular Tyr- and Detyr-microtubule content.

(a) Chemical structure of Pyr1

(b) Cells treated with Pyr1 have increased Detyr-microtubules content.

HeLa cells were treated for 2 hours with 25 µM Pyr1, or with 0.25% DMSO alone (DMSO,

control) as indicated. Left: Tyr-tubulin staining ; Right : Detyr-tubulin staining. Bar = 10 µm.

Figure 2: Effect of Pyr1 on the stabilization of cellular microtubules in cells and in vitro and

its effect on the cell cycle.

(a) Immunofluorescence analysis of cell microtubules and their resistance to nocodazole-induced

depolymerization. HeLa cells were incubated for 2 hours with DMSO, 5 µM Paclitaxel or 25 µM

Pyr1. 10 μM nocodazole was then added for 30 minutes. Cells were then stained for total tubulin

(red). Nuclei were stained with Hoechst. (b) Immunofluorescence analysis of EB1 comets

HeLa cells were incubated for 2 hours with DMSO, 5 µM Paclitaxel, or 10 and 25 µM Pyr1 as

indicated and then stained for EB1. (c) Comparison of nocodazole, paclitaxel and Pyr1 effects on

the mitotic spindles. HeLa cells were incubated for 2 hours with DMSO, 10 µM nocodazole, 5

µM paclitaxel or 25 µM Pyr1, as indicated and then stained for total tubulin. Representative

metaphase spindles are shown. (d) Pyr1 induces a cell cycle arrest at S-G2/M. HeLa cells were

cultured for 26 hours, with 0.25% DMSO or 10 µM of nocodazole, paclitaxel or Pyr1 as

indicated. The cell cycle parameters were analyzed by flow cytometry. Histograms represent the

percentage of the total cell population in each cell cycle phase: sub G1 (yellow), G1 (blue), S

(green), G2/M (red), > 4N (purple). (e) Microtubule proteins and (f) pure in vitro tubulin

polymerization assay. Tubulin (0.5mg/mL for microtubule proteins and 1mg/mL for pure tubulin)

was allowed to polymerize at 37°C in the presence of 1 µM paclitaxel (red), 25 µM Pyr1 (green)

24

or equivalent amount of DMSO (0.25%, orange). Results are presented as mean ± standard error

of the mean (SEM) of three independent experiments. Bars = 10 µm.

Figure 3: Effects of Pyr1 on tumor growth in mice

Survival curve of L1210 syngenic leukemia-bearing mice treated with Pyr1 or with the vehicle

alone, as indicated. Both groups significantly differ (p < 0.002) at day 70 (Mann-Whitney U test).

Figure 4: Effects of Pyr1 on the actin cytoskeleton and cell motility

(a) Immunofluorescence analysis of Pyr1 effects on F-actin. HeLa cells were incubated for 2

hours with DMSO, 5 µM Paclitaxel or 25 µM Pyr1, as indicated. Cells were then stained with

phalloidin (green) and Hoechst (blue). Bar = 20 µm. (b) Analysis by time-lapse microscopy of

the random migration of MCF10A cells treated with DMSO (control) or 10 and 25 µM of Pyr1,

as indicated. Tracks of individual cells, obtained from 3 representative stacks, revealed that Pyr1

completely abolished cell motility. (c) Actin-pyrene polymerization assays with 2 μM actin (10%

pyrene-labelled): Control (blue), with 10 µM Pyr1 (orange), with 2 µM phalloidin (green) and

with 1 µM latrunculin B (red).

Figure 5: LIM Kinase is a target of Pyr1

(a) Inhibitory effect of Pyr1 on the kinase activity of a panel of 66 kinases including LIMK1. The

data are expressed as the percent of the activity determined in the absence of inhibitor. (b, c) Pyr1

inhibits the phosphorylation of GST-cofilin by GST-LIMK1 (b) or GST-LIMK2 (c). Increasing

concentrations of Pyr1 were incubated with kinase assay mix. Phosphorylated proteins were then

separated on SDS/PAGE and analyzed by autoradiography (upper panel) and Coomassie Blue

staining (lower panel). GST-LIMK1 autophosphorylation is also abrogated in presence of Pyr1.

(d) Inhibition of LIMK1-mediated GST-p25 phosphorylation by Pyr1. Increasing concentrations

of Pyr1 were incubated with kinase assay mix. Phosphorylated proteins were then separated by

SDS/PAGE and analyzed by autoradiography (upper panel) and Coomassie Blue staining (lower

25

panel). (e) Effect of Pyr1 on cofilin phosphorylation in cells. HeLa cells were treated for 2 hours

with increasing concentrations of Pyr1, as indicated. Cells lysates were separated on SDS-PAGE

and transferred for immunoblotting with anti-Phosphocofilin (ser3) or cofilin as indicated.

Figure 6: LIMK1 inhibition induces microtubule stabilization

(a) Structure of (N-[5-[2,6-Dichloro-phenyl]-5-difluoromethyl-2H-pyrazol-3-yl]thiazol-2-yl)-

isobutyramide, also called LIMKi.(b) Comparison of the effect of Pyr1 and of LIMKi on Detyr-

microtubules formation. HeLa cells were treated for 2 hours with DMSO, 5 µM paclitaxel, 25

µM Pyr1 or LIMKi, as indicated. Cells were stained for Detyr-microtubules. (c) Analysis of

LIMK1 content in the different cell extracts. Lysates of unstransfected cells, and of cells

transfected with the control plasmid pEFrFlagPGKpuropAv18 or with

pEFrFlagLIMK1PGKpuropAv18 plasmid overexpressing full-length LIMK1 were separated on

SDS-PAGE and transferred for immunoblotting with anti-LIMK1 antibody. (d) LIMK1

overexpression can reverse Pyr1-induced microtubule stabilization. HeLa cells were transfected

with the control plasmid pEFrFlagPGKpuropAv18 or with pEFrFlagLIMK1PGKpuropAv18

plasmid overexpressing full-length LIMK1. Cells were treated for 2 hours with different amounts

of Pyr1, as indicated. 10 μM nocodazole was then added for 30 minutes. Cells were then stained

for total tubulin. Bars = 10 µm.

26

Table 1: Effect of Pyr1 on the viability of different cell lines

Cell Line Type Concentration of Pyr1 that

inhibits 50% of the cell growth

HeLa Human cervical adenocarcinoma 1 μM

786-O Human renal

adenocarcinoma 55 μM

NCI H460 Human lung carcinoma 6 μM

MCF-7 Human breast

adenocarcinoma 62 μM

MDA-MB-231

Human breast carcinoma 25µM

MES-SA Human uterine sarcoma 26 μM

MES-SA/DX5 Multi drug-resistant cell line derived from the MES-SA

27 μM

HEK-293 Cell line derived from human

embryonic kidney 3 µM

HEK-293-ABCG2 HEK-293 stably transfected with the ABCG2 transporter

2 µM

BHK-21 Cell line derived from baby

hamster kidney 1 μM

BHK-21-MRP1 BHK-21 stably transfected

with MRP1 (multidrug resistance protein 1)

1 μM

27