Inhibition of platelet function by targeting the platelet cytoskeleton via the LIMK-signaling pathway

Actin is highly abundant in platelets, and platelet function is dependent on actin structures. Actin filaments are dynamic structures involved in many cellular processes including platelet shape changes and adhesion. The actin cytoskeleton is tightly regulated by actin-binding proteins, which include the members of the actin depolymerising factor (ADF)/cofilin family. LIM kinase (LIMK) and slingshot phosphatase (SSH-1L) regulate actin dynamics by controlling the binding affinity of ADF/cofilin towards actin. We hypothesised that inhibition of LIMK activity may prevent the changes in platelet shape during their activation and therefore their function by controlling the dynamics of Factin. Therefore, inhibition of LIMK activity may represent an attractive new strategy to control and inhibit platelet function; particularly the formation of stable platelet aggregates and thus stable thrombi.


Introduction:
Changes in platelet shape and functions are regulated by cytoskeletal rearrangements involving dynamic actin polymerisation and depolymerisation (1,2). Platelets circulate in the blood as discs in an inactive form with a distinct organisation of microtubules (MTs) and actin filaments (F-actin). Actin comprises ~20-30% of the total platelet proteins (3). During activation, platelets undergo dramatic changes including a rapid shape change from a disc to a flat sphere; formation of pseudopods; disassembly of the peripheral microtubules ring; migration of the dense bodies towards the centre of the cell; secretion from its granules; conformational change and cytoskeletal anchorage of the integrin receptor αIIbβ3 (GPIIb/IIIa, CD41/61) and finally aggregation. All of these changes require alterations in the membrane skeleton as well as actin and tubulin remodelling (2).
The changes in the actin cytoskeleton are important for platelet activity in many cardiovascular functions. The dynamic process of the actin cytoskeleton is regulated by actin regulatory proteins including members of the actin depolymerising factors ADF/Cofilin. The activity of cofilin is regulated by the members of the LIM kinase family, LIMK1 & 2, that phosphorylate cofilin on serine 3 resulting in its inactivation (4,5). LIMK1 and LIMK2 are serine protein kinases that are activated by phosphorylation of threonine 508/505, respectively, by the effectors protein kinases belonging to the Rho GTPase family that includes Rho kinase (ROCK) and p21 activating kinases (PAK 1 and 4) (6)(7)(8). Activated LIMKs phosphorylate and inactivate their substrate cofilin resulting in the accumulation of actin filaments within the cells (4,5). On the other hand, the cofilin phosphatase slingshot (SSH1L) (also known as LIMK1 phosphatase) reactivates cofilin by dephosphorylation of both cofilin and LIMK1 (9,10). LIMK1 and SSH-1L play a major role in actin dynamics via the regulation of ADF/cofilin. Figure 1 represents a schematic diagram of actin filament formation via the downstream of ROCK/LIMK1 signal transduction pathway and the cofactors involved in the process.
The finding by Pandey et al. that LIMK1 activity is important for thrombininduced change of platelet shape and aggregation suggests that LIMK1 may be a new therapeutic target to inhibit platelet aggregation (11). Furthermore, these studies also demonstrated that ROCK activated LIMK1 during platelet shape change and aggregation. Interestingly, although LIMK1 was activated, no change in the level of P-cofilin was observed during shape change, while cofilin de-phosphorylation was observed during platelets aggregation (11).
Other studies have demonstrated a role for thrombin in ROCK and LIMK1 activation. Addition of thrombin to human vein umbilical endothelial cells (HUVECs) resulted in ROCK activation followed by LIMK1 activation, increased P-cofilin and F-actin levels (12). Interestingly Aslan et al. demonstrated that also PAK, the other positive regulator of LIMK, regulates platelet shape change and aggregation (13).
The studies conducted by Estevez

Materials and Methods: Blood collection and platelet isolation
Platelet rich plasma (PRP) from human blood collected by venipuncture from healthy volunteers taking no medications and anti-coagulants was prepared as previously described. Washed platelets prepared from citrated PRP (1 ml) were passed through a pre-washed sepharose CL-2B column (Sigma) and eluted by addition of 1 ml of modified Tyrode's buffer (150 mM NaCl, 2.5 mM KCl, 12 mM NaHCO 3 , 2 mM MgCl 2 , 2 mM CaCl 2 , 1mg/ml BSA, 1mg/ml dextrose; pH 7.4).
After elution, platelets were diluted with modified Tyrode's buffer either 1:50 for flow cytometry analysis or 1:5 for staining and adhesion assay purposes. For western blotting, PRP was acidified to pH 6.5 by addition of ACD buffer (1.32% w/v sodium citrate, 0.48% w/v citric acid, and 1.47% w/v dextrose) followed by addition of 10 mM Theophylline, to inhibit platelet activation during centrifugation. Acidified PRP was centrifuged at 720g for 10 minutes at 22°C, and the platelet pellet was washed with 5 mL of PBS containing 1 mM of Theophylline. After centrifugation, platelets' pellet was resuspended in PBS containing Ca 2+ /Mg 2+ .

Static adhesion and immunostaining assays
Human platelets purified through Sepharose CL-2B column were added onto fibrinogen-coated cover slips blocked with 1% BSA for 1 hour at RT. Platelets, pre-incubated with the ROCK inhibitor (Y-27632), or LIMK inhibitor (22j) for 1 hour at RT, were allowed to adhere in the presence of 2 M ADP at 37°C for 30 minutes. Coverslips were washed twice with Tyrode's buffer and fixed with 1x Cellfix for 30 minutes followed by two washes with Tyrode's buffer before mounting with Vectashield and visualized by microscopy (Nikon A1R confocal).
Washed mouse platelets (1x10 7 /L in PBS) were added onto fibrinogen-coated cover slips as described above and permeabilized with 0.1% Triton X-100 for 30 minutes at RT followed by staining with Alexa 488 phalloidin for 30 minutes at RT. Platelets were visualized using a Nikon A1R confocal microscope (Japan) Apo x40 water 1.1 n.a objective. Stress fibres (F-actin) were visualized under 2D mode N-SIM structured illumination microscopy using a Nikon Ti inverted microscope (Japan) with an Apo x100 oil 1.49n.a. objective and EMCCD camera (Andor Technology Ixon 3). The images were reconstructed using the NIS ELEMENTS software 4.10.

Flow cytometry
Platelet rich plasma (PRP) isolated by gradient centrifugation from healthy human blood was incubated with and without ROCK inhibitor (Y27632, 20-300

Platelet aggregometry
PRP aliquots of human blood pre-incubated for 1 hour with the ROCK or LIMK inhibitors were subjected to platelet aggregometry using an aggregometer (AggRam TM System) in the presence of 1 μM ADP.

Clot retraction assay
PRP was incubated with 22j or solvent for 60 minutes at RT. Clot formation was initiated in glass tubes by the addition of 1 U/mL thrombin in the presence of 10 mM CaCl 2 . The retraction was measured after 1 hour. The extent of clot retraction was expressed as the volume of serum extruded from the clot as a percentage of the total reaction volume.

In vitro kinase assay
The in vitro kinase assay was performed as previously described. Briefly, target proteins were purified and then subjected to an in vitro kinase assay for 30 minutes at 37C in the presence of 5 Ci [ 32 P] -ATP.
Phosphatase assays were carried out at 30C for 2 hours using 15 L of 32 P-GST-cofilin as a substrate for SSH-1L in a phosphatase buffer [30 mM Tris-HCl (pH 7.4), 30 mM KCl, 1 mM EDTA, 0.1 mg/ml BSA and protease inhibitors].
After centrifugation at 12,000g for 30 seconds, the radioactivity released by SSH-1L from 32 P-GST-Cofilin was determined by  counter (PerkinElmer). The beads containing GST-32 P-cofilin were subjected to western blotting. 32 P-GSTcofilin was detected by a Phosphorimager and quantified using Image Quant probe software. The level of total GST-cofilin protein was determined by immunoblotting and quantified using the above software.

Tail bleeding time
Mice were anaesthetized with ketamine (100 mg/kg), and xylazine (5 mg/kg) by intraperitoneal injection and the tip of their tail (5 mm) was cut and immediately immersed into saline, pre-warmed at 37  C. Bleeding time was monitored and recorded as the time needed for the cessation of the visible blood stream, for at least 1 minute. Maximum bleeding time was defined as 20 minutes and mice bleeding for more than 20 minutes were sacrificed. For drug administration, the right side jugular vein was bluntly exposed and catheterized.

Carotid artery thrombosis
Ferric chloride-induced injury in the carotid artery of mice was used as previously described (22). Mice were anaesthetized with ketamine (100 mg/kg) and xylazine (5 mg/kg) by intraperitoneal injection. Anaesthetized mice were placed under a dissecting microscope in a supine position, and a midline incision of the skin extending from the mandible to the sternal notch was made.
The fascia was bluntly dissected up to the left common carotid artery.
Thrombosis was induced by applying a piece of filter paper (1×3 mm, GB003, Schleicher & Schuell) saturated with 10% ferric chloride (Sigma) under the left common carotid artery for 3 minutes. A piece of plastic sheet (3×6 mm) was laid under the filter paper to prevent the absorption of ferric chloride by surrounding tissue. After removing the filter paper, the artery was rinsed with saline, and a nano-Doppler flow probe (0.5 VB or 0.5 PBS, Transonic) was positioned over the artery. The blood flow was measured by a flow meter (T106 Transonic) and recorded with a PowerLab data acquisition unit (AD Instruments). Thrombotic occlusion was defined by a drop in blood flow to 0.03 mL/min for at least 20 seconds, and reopening was defined by an increase in blood flow to 0.04 mL/min for at least 20 seconds. Maximum occlusion time was defined as 60 minutes and mice that had not occluded after 60 minutes were sacrificed.

Drug administration
The right jugular vein was bluntly exposed and catheterized. All injections were conducted via this catheter. 100 mg/kg 22j or an equal volume of solvent was injected at 1 minute before the start of the ferric chloride injury. For reopening experiments, 500 U/g BW Urokinase and 100 mg/kg BW 22j or an equal volume of solvent were injected immediately after blood flow ceased.

Statistical analysis
A Mann Whitney t-test (mean ± standard deviation are presented) was conducted on the occlusion experiments if time measurements were in the predefined range. If they were not within this range, a Gehan-Breslow-Wilcoxon test (median is presented) was conducted. A Gehan-Breslow-Wilcoxon test was also conducted on all bleeding time experiments. A Chi-square test was conducted on the reopening experiment. Differences were considered to be significant at p < 0.05. Other results are shown as means ± SEM from ≥ 4 independent experiments and from various donors. Groups were compared by t-test and a value of p < 0.05 was considered significant.

Expression of the cytoskeleton regulators in platelets
It was previously demonstrated that LIMK1 protein is expressed in platelets (11). Using rat anti-LIMK1 monoclonal antibodies (mAb) (20), we have confirmed LIMK1 expression in platelets by western blotting (Figures 1B).
Furthermore, we have shown that adult human platelets also express a number of other molecules that play roles in downstream of LIMK signaling pathway including cofilin, phospho-cofilin (P-cofilin), PKD and slingshot-1L (SSH1L) proteins ( Figure 1B). As these proteins are involved in the regulation of actin polymerization, it is highly suggestive that they are important for the regulation of platelet function as previously suggested for LIMK1 and cofilin (11).

LIMK plays an important role in platelet cofilin phosphorylation
Since cofilin is a major substrate of LIMK, the level of P-cofilin correlates with LIMK activity. We therefore tested the effects of LIMK and ROCK inhibition on cofilin phosphorylation after platelet incubation with the LIMK inhibitors 22j (19), BMS3 (17) (16) and Pyr1 (18) as well as the ROCK inhibitor Y-27632 (11) for 5 minutes (Figure 2). Incubation of platelets with these inhibitors resulted in a significant reduction in the level of P-cofilin, with 22j having the greatest effect on the levels of both P-cofilin and P-LIMK ( Figure 2C). As the high concentration of 22j ( Figure 2B) may also inhibit ROCK activity (19), we tested the effect of 10 and 100 M 22j on the phosphorylation of the ROCK substrate, myosin light chain (MLC) using an anti-phospho-MLC (P-MLC) antibody.
However, high concentrations of 22j did not affect MLC phosphorylation whereas it significantly reduced P-cofilin levels ( Figure 2D), suggesting that the inhibition of LIMK activity by 22j is specific under these conditions.

SSH-1L is responsible in part for the down regulation of P-cofilin levels after LIMK1 inhibition
We have demonstrated that inhibition of ROCK and LIMK1 activity by pretreatment of platelets with the ROCK inhibitors as well as LIMK inhibitors for 5 minutes results in dramatic reduction in the level of P-cofilin, suggesting the involvement of a cofilin phosphatase in this process. We therefore compared the activity of SSH-1L, an established cofilin and LIMK1 phosphatase, in lysates prepared from untreated platelets and platelets treated with the ROCK or LIMK inhibitors. First, we confirmed that indeed the level of P-cofilin is reduced after treatment with these inhibitors and that SSH-1L is expressed in all samples ( Figure 3A & C). SSH-1L activity was measured by the amount of that the amount of 32 P released from 32 P-cofilin by SSH1L immunoprecipitated from the ROCK or LIMK inhibitor-treated platelets is greater than the amount released by SSH-1L purified from control platelets ( Figure 3B & D).
The ROCK/LIMK1 signaling pathway regulates platelet activity.
To confirm the previous findings that the ROCK signaling pathway is important for the regulation of platelet function, we measured the activity of ADPstimulated platelets in the presence of different amounts of the ROCK inhibitor Y27632 using P-selectin expression, a marker for platelet activation. P-selectin expression induced by 2 M ADP was gradually decreased with increasing amounts of Y27632 (0-300 M) ( Figure 4A). The inhibition of ROCK activity with Y27632 partially inhibits P-selectin expression.
Another marker of platelet activation is the conformational change of the platelet integrin GPIIb/IIIa (23) (PAC-1). This receptor is constitutively expressed on platelets, however, upon platelet activation, it undergoes a conformational change that can be detected with a specific single-chain antibody. To further support our findings that ROCK activity is involved in platelet function, GPIIb/IIIa expression was measured by flow cytometry using His-tagged activation specific single-chain antibody (21). While ADP-stimulated human platelets result in GPIIb/IIIa activation, the addition of Y27632 decreases this level of activation by approximately 2 folds ( Figure 4B).

LIMK is involved in human platelet activity and aggregation
To confirm that the role of ROCK in platelet activity is via the regulation of the LIMK downstream signaling pathway, we investigated the changes in P- To further confirm our findings that the LIMK signaling pathway is important for platelet function, an aggregation assay was performed in the presence and absence of the specific LIMK inhibitors 22j (5 M), BMS3 (5 M), Pyr1 (5 M) and the ROCK inhibitor Y-27632 (10 M). Incubation with all these LIMK inhibitors led to a significant reduction in platelet aggregation ( Figure 4E).

Platelet adhesion is mediated by ROCK and LIMK
To test whether the ROCK/LIMK signaling pathway plays a role in integrinmediated platelet adhesion, platelets pre-incubated with 20 M Y-27632 or 3 M 22j were activated with 2 µM ADP and allowed to spread on fibrinogencoated cover slips. Incubation with Y-27632 and 22j reduced the number of platelets that adhered to fibrinogen ( Figure 5A and B). These results suggest that platelet adhesion to fibrinogen occurs via activation of ROCK signaling cascade mainly via LIMK.

ROCK/LIMK control the actin cytoskeleton during platelet spreading
To investigate if LIMK is involved in regulating the tensile strength of the actin cytoskeleton, a clot retraction assay was performed with PRP in the presence of 22j. A significant reduction in platelet clot retraction was observed with increasing concentrations of the LIMK inhibitor 22j ( Figure 5C). To confirm our findings that platelet actin polymerization is regulated by LIMK, F-actin was visualized using fluorescence microscopy. Platelets treated with 20 M Y-27632 or 3 M 22j displayed reduced adhesion ( Figure 5A & 6A), reduced Factin content ( Figure 6B) and reduced spread surface area ( Figure 6C). Stress fibre formation in platelets treated with either 22j or solvent was resolved with structured illumination microscopy ( Figure 6D). Platelets treated with 22j showed an accumulation of chopped or truncated F-actin fragments at various lengths. However, we observed only a small trend towards reduction of the Factin/G-actin ratio for both Pyr1 and 22j treated platelets compared to control platelets ( Figure 6E and F).

Effects of LIMK inhibition on tail bleeding and thrombus occlusion time in mice
To further explore the involvement of LIMK in platelet function and its potential therapeutic implication, we performed tail-bleeding time assays with C57Bl/6 WT mice treated with 22j (100 mg/kg) or Y-27632 (30 mg/kg). A significantly prolonged bleeding time was observed in WT mice treated with 22j and Y-27632 ( Figure 7A). Bleeding in the control mice stopped within 1-3 minutes (p< 0.01) while that of the Y-27623 and the 22j treated mice stopped after 15 and 20 minutes, respectively ( Figure 7A). In addition, the carotid artery occlusion time following topical exposure to 10% ferric chloride in control mice was 662.0 ± 85.19 seconds, while the occlusion time of 22j administered mice was significantly increased to 998.6 ± 298.6 (p< 0.01) ( Figure 7B). Similarly, the occlusion time of the Y-27632 administered mice was significantly increased from its PBS control 534 ± 59.14 to 1412 ± 284.2 seconds (p< 0.02). Our in vitro studies of pharmacological LIMK inhibition corroborate our in vivo experiments demonstrating that LIMK inhibition impairs platelet function resulting in a prolongation of bleeding and occlusion times.

22j is a potential facilitator of thrombolysis
To investigate whether 22j can facilitate thrombolysis, we used ferric chloride to induce a thrombotic occlusion of the carotid artery in wild type mice followed by an injection of 500 U/g body weight (BW) urokinase, which is a clinically used fibrinolytic drug, and monitored the time taken for the vessel to reopen.
When 22j (100 mg/kg) was injected simultaneously with urokinase, 50% of the mice showed reopening of the vessel compared to only 14% when injected with urokinase alone (n=14 each, p<0.05). This indicates that 22j may be a potential drug candidate for facilitating thrombolysis in combination with a fibrinolytic drug.

Discussion:
Anti-platelet therapy is highly beneficial for patients with cardiovascular diseases such as myocardial infarction and stroke. However, bleeding complications with current drug therapies can prevent their use in many patients that would otherwise benefit from them. The aim of this study is to understand the basic mechanisms involved in the regulation of the platelet cytoskeleton in order to identify new targets and novel strategies for antiplatelet therapy, which preferentially target thrombus stabilisation.
Actin and tubulin comprise the majority of the platelet proteins and changes in their polymerization state are essential for platelet shape change, granule secretion and aggregation. Inhibition of actin polymerization has been shown to reduce the stability of platelet aggregates (24). It is therefore likely that These authors also demonstrated the involvement of LIMK1 in thrombinstimulated platelets (11). In agreement with that study, we confirmed the expression of LIMK1 and cofilin in resting human platelets ( Figure 1B).
We have further investigated the involvement of LIMK in the regulation of platelet function by using LIMK specific inhibitors. We observed reduced levels of LIMK and cofilin phosphorylation in platelets treated with the specific LIMK inhibitors BMS3 (17), Pyr1 (18) and 22j (19). We also noticed a rapid drop in P-cofilin levels with increasing amounts of 22j (Figure 2 B). We have also observed that inhibition of LIMK activity by 22j or the ROCK inhibitor Y27632 control platelet adhesion as well as F-actin levels ( Figure 5 & 6). Furthermore, inhibition of PAK activity by a PAK specific inhibitor was shown to control platelet spreading and F-actin levels (25). PAK and ROCK are two well-established upstream regulators of LIMK. Here we show that the inhibition of LIMK activity by the LIMK specific inhibitors inhibits platelet aggregation ( Figure   4E), indicating that LIMK plays an important role in platelet function through control of F-actin levels, as previously demonstrated for thrombin-activated platelets by Pandey et al. (11).
During platelet activation by increasing concentrations of ADP, there was no change in the level of cofilin phosphorylation. However, when platelets were pre-incubated with the ROCK or LIMK inhibitors (Figure 2) cofilin phosphorylation was dramatically reduced. As mentioned above, the dramatic change in p-cofilin levels cannot be explained only by the reduction in LIMK1 activity. We therefore propose that the increased activity of cofilin phosphatase is responsible for the dephosphorylation of cofilin within 5 minutes inhibition of ROCK, or LIMK. As SSH1L is expressed in platelets, it is reasonable to suggest that it may play a role in the regulation of p-cofilin levels in platelets. Indeed, the experiments presented in Figure 3 indicate that SSH1L activity is increased after incubation with the ROCK inhibitor suggesting that inhibition of this enzyme results in SSH1L activation. We and others have previously demonstrated that SSH1L activity is regulated by phosphorylation (10). Eiseler et al. (26) have shown that phosphorylation of SSH1L by protein kinase D1 (PKD1) at Serine 978 generates a 14-3-3 binding site. SSH1L binding to 14-3-3 inhibits its interaction with F-actin. As SSH1L activity is greatly enhanced by its binding to F-actin, inhibition of this association results in down regulation of SSH1L activity (26). PKD1 is activated by the Rho-ROCK pathway, suggesting that inhibition of ROCK may result in PKD1 inhibition and upregulation of SSH1L activity leading to cofilin dephosphorylation after ROCK inhibition. Indeed, we have shown that PKD is expressed in platelets ( Figure   1B). We have previously demonstrated that LIMK1 interacts with SSH1L resulting in its dephosphorylation (10). It is possible that binding of LIMK1 to its inhibitor releases the interaction between these two enzymes favouring the interaction between SSH-1L and cofilin resulting in increased SSH-1L activity  (14). In contrast, our studies with the LIMK inhibitor treatments did show significant bleeding prolongations. Interestingly, relatively high levels of P-cofilin protein in LIMK1 KO platelets were found by Estevez et al. (14) suggesting that either their LIMK1 KO model is not a complete knockout, and the kinase domain may expressed as previously suggested by Bernard (28), or LIMK2 is expressed in platelets, and that phosphorylates cofilin. However, previously Pandey et al. demonstrated that LIMK2 is not expressed in platelets (11). Till to date, LIMK1 and LIMK2 are the only known kinases that phosphorylate cofilin. We have been previously demonstrated the complete inhibition of phospho cofilin by knocking down of both LIMK 1, and LIMK2 with siRNA in human cell lines including breast cancer cells (16). The LIMK inhibitors that we used in this project, 22j, pyr1 and BMS3, inhibit both LIMK1 and LIMK2 and therefore, we can conclude that both LIMK1 and LIMK2 have a significant role in platelet function.
Therapeutic targeting of the platelet cytoskeleton is an attractive novel approach for anti-platelet therapy. Thrombus destabilization in the clinical setting of thrombolysis, for example in patients with myocardial infarction, is of particular interest. Whereas fibrinolytic drugs can dissolve the extracellular fibrin mesh, a LIMK inhibitor could destabilize the intracellular cytoskeleton.
Indeed, we have shown that the LIMK inhibitor 22j in combination with urokinase was more efficacious than urokinase alone in restoring blood flow in the carotid artery following vessel injury and thrombotic occlusion. This opens up a new strategy in pharmacological thrombolysis that warrants to be tested further.  Purified platelets were stimulated for 1 minute with ADP (0, 2, 20 M) or preincubated with 22j for 5 minutes prior to incubation with ADP. Platelet lysates were immunoblotted with anti-P-cofilin Abs followed by probing the stripped membrane with rabbit anti-cofilin Abs. The numbers below the blots represent the ratio between the levels of P-cofilin to cofilin. (E) The LIMK inhibitor 22j inhibits LIMK activity but not that of ROCK. Immunoblots of lysates prepared from platelets incubated for 10 minutes with 22j or DMSO were probed for phospho-MLC (P-MLC), P-cofilin and LIMK1. GAPDH was used as loading control.