Mlck1a is expressed in zebrafish thrombocytes and an essential component for thrombus formation

Abstract

Background

We have used the advantages of the zebrafish model system to demonstrate which of the vertebrate Myosin Light Chain Kinase (MLCK) genes is expressed in thrombocytes and important for thrombus formation.

Methods and Results

Here we report that Mlck1a is an essential component for thrombus formation. Phylogenetic data revealed four zebrafish orthologous for three human MLCK genes. To investigate expression of the zebrafish mlck genes in thrombocytes we, compared GFP-tagged platelets with other cells by microarray analysis, and showed that mlck1a expression was 4.5 fold enriched in platelets. Furthermore, mlck1a mRNA and mRNA for the platelet-specific cd41 co-localized in thrombi. Expression of other mlck subtypes was lower in GFP-tagged platelets (mlck1b; 0.77 fold enriched) and absent in thrombi (mlck1b, -2, -3). To investigate the role of Mlck1a in thrombus formation, we knocked down mlck1a using two morpholinos. This resulted in impaired morphology changes of platelets adhering on fibrinogen. In a thrombosis model, in which thrombocytes adhere to the vessel wall damaged by laser irradiation, thrombus formation was slowed down in mlck1a deficient embryos.

Conclusion

We conclude that Mlck1a is the subtype of MLCK that contributes to platelet shape change and thrombus formation.

Introduction

Platelets contribute to arterial thrombosis by formation of a platelet plug, release of platelet activating and vasoconstrictive components and generation of a pro-coagulant surface. Following adhesion to the damaged vessel wall, platelets undergo a profound change in morphology. Two signaling pathways leading to platelet shape change have been described[1]. First, there is the Ca²⁺-independent route, initiated by thrombin, thromboxane A₂ and lysophosphatidic acid that triggers myosin phosphorylation through Rho-kinase (ROCK) and inhibition of myosin phosphatase. Second, there is a Ca²⁺ dependent route initiated by ADP, thrombin and collagen that activates Ca²⁺/calmodulin-dependent myosin light chain kinase (MLCK). MLCK induces phosphorylation of myosin light chain resulting in cytoskeletal rearrangements, folding of membrane surfaces and the contractile wave that centralizes the secretory granules[2, 3]. Myosin light chain phosphorylation starts before shape change and can be detected in the aggregometer. Once started, the extent of shape change correlates strongly with the phosphorylation of myosin light chain[4, 5] and with the association of myosin to actin[6, 7].

Mammals express three genes for MLCK[8]. MLCK1 (also named smooth muscle MLCK or MYLK1) is ubiquitously expressed in various tissues. MLCK2 (or skeletal muscle MLCK or MYLK2) is expressed in skeletal and cardiac muscle tissue. MLCK3 (or cardiac MLCK or MYLK3) is expressed in cardiac muscle tissue and may play a role in cardiogenesis[9].

MLCK phosphorylates the regulatory light chain of myosin on Ser 19 at the N-terminus[8]. The only myosin isoform expressed in platelets is myosin IIa[10]. Phosphorylation of myosin IIa leads to assembly of filaments, which mediate interaction with actin, forming a contractile unit similar to actomyosin in smooth muscle cells[11]. Contractile force is generated by the movement of myosin along actin, a process that requires phosphorylation of the myosin regulatory light chain[12]. Phosphorylation of the myosin heavy chain inhibits filament formation[13]. Mutations in the myosin IIa gene (MYH9) cause thrombocytopenia with large platelets and a mild bleeding tendency. These platelets aggregate normally but fail to undergo shape change[14-17]. A similar but more severe phenotype is seen in mice with a megakaryocyte-restricted MYH9 disruption. This leads to absent shape change and clot retraction and a dramatic increase in the bleeding time[18]. Platelets adhering to fibrinogen form normal lamellopodia but do not make stress fibers. In a carotid artery thrombosis model, coverage of the injured area is incomplete and thrombus stability is impaired leading to strong embolization.

The zebrafish is an attractive model to study platelet proteins involved in thrombus formation because of transparency of the embryos, labeling of thrombocytes with a GFP tag under the control of the cd41 promoter and accessibility of gene interference by morpholinos[19, 20]. The zebrafish has a coagulation cascade that is very similar to that of humans. Their platelets, although possessing a nucleus, express the fibrinogen receptor αIIbβ3 and the von Willebrand Factor receptor Glycoprotein Ib and aggregate upon stimulation with collagen, ADP and von Willebrand Factor /ristocetin[21-23]. In the present study we made use of these advantages to identify the MLCKs involved in platelet shape change and thrombus formation. We found zebrafish orthologues for all three human MLCK genes and show exclusive expression of mlck1a in zebrafish thrombocytes. Knock down of mlck1a by reverse genetics greatly impaired spreading of thrombocytes on a fibrinogen surface and leads to retarded vessel occlusion in an in vivo thrombosis model.

Materials and Methods

Zebrafish husbandry and lines

Embryos were obtained by mass matings of adult TL or Tg(cd41:GFP) fish and raised at 28°C. The transgenic line Tg(cd41:GFP)[19] was a kind gift from R. Handin (Boston, MA). In this line the sequence for green fluorescent protein (GFP) is fused to the promotor elements of the platelet specific gene CD41, to generate fluorescent thrombocytes.

A thrombosis model in zebrafish

Zebrafish larvae 3-5 days post-fertilization (dpf) were anaesthetized in MS222 solution (Sigma, St. Louis, MO) and immobilized in 1% low-melting agarose (Invitrogen, UK) on a microscope slide. A thrombus was induced by delivering a pulsed laser light pumped through coumarin 440 dye (445 nm) (MicroPoint Laser System, Photonic Instruments Inc., St. Charles, Illinois) at 7 pulses per second for 5 seconds through a 10× objective on a Zeiss Axioscope microscope (Carl Zeiss Light Microscopy, Göttingen, Germany). Laser damage was induced in the endothelial layer of the posterior cardinal vein and the dorsal aorta, as indicated, at the position of somite 5 posterior to the cloaca. Thrombus formation was recorded with a Hamamatsu ORCA-ER C4742-80 digital camera (Hamamatsu Photonics, Herrsching am Ammersee, Germany). Time to occlusion (TTO) was defined as the time between the start of laser irradiation and complete occlusion of blood flow.

Morpholino injections

Morpholinos (MOs) were obtained from Gene Tools (http://www.gene-tools.com) and diluted in water containing 0.2% phenol red. One cell stage embryos were injected (maximum volume of 2 nL) as described[20]. Embryos were injected with prothrombin or mlck1a specific morpholinos (8 ng/embryo each). Control embryos were injected with a p53 morpholino (8 ng/embryo each). Morpholino sequences were:

• MO_prothrombin: 5′-GTTTGGCTCCCATCCTTGAGAGTGA-3′; MO1_mlck1a:

• 5′-TATGCAAGTGTTCATACTCACCGAC-3′; MO2_mlck1a: 5′-TGATATACTCACGTGCCCAGTCGG-3′;

• MO_p53: 5′-GCGCCATTGCTTTGCAAGAATTG-3′

Microarray gene expression profiling

Three thousand cd41–GFP⁺ (3 dpf) cells were sorted by fluorescence-activated cell sorter (FACS) analysis, using single cell lysates from 3-4 day old embryos. Propidium iodide (PI; Sigma) was added as a marker (1 μg/mL) to indicate dead cells and debris. Cell sorting was performed based on PI exclusion, forward scatter, side scatter and GFP fluorescence using a FACS Vantage Flow Cytometer (Becton Dickinson, San Jose, CA). All cell populations were sorted twice to optimize cell purity. These were compared with 3000 cd41-GFP⁻ cells (a mixture of other cells of the embryos). Total RNA was purified and analyzed using Affymetrix zebrafish gene chips, as described[24].

Whole-Mount in situ hybridization

In situ hybridizations were performed essentially as described[25]. Embryos were mounted in glycerol and documented with a Zeiss Axioplan mounted with a Leica DFC 480 Camera.

Spreading of cd41-GFP positive cells on fibrinogen coated glass

For analysis of static adhesion, glass slides (Menzel Gläser 18×18 mm) were cleaned overnight with bromic acid and rinsed with distilled water. Slides were coated with 100 μl of 100 μg/ml fibrinogen (Kordia, Leiden, the Netherlands) for 60 minutes at 22 °C. Coverslips were blocked with 1% BSA in PBS (30 minutes, 22 °C) and washed with Hepes/Tyrode buffer. cd41-GFP positive larvae (4 dpf) were anesthetized by MS222 solution and placed on the fibrinogen-coated slide in a drop of Hepes/Tyrode solution. Blood was withdrawn by an incision through the inflow tract of the heart. After withdrawal of the blood, the embryos were removed from the cover slip. The coverslip was covered with a second coverslip (fibrinogen-free) and incubated (30 minutes, 22 °C). cd41-GFP⁺ cells were scored for morphologic features. Different cell morphologies were: (1) round cells, absence of initiation of cell shape change (2) cells that were fully spread on fibrinogen. The number of round cells was measured.

Statistical analysis

Results are expressed as means ± SEM with number of experiments. Statistical comparisons of two groups were by students t-test, using Graphpad Prism 4.0 software. Differences were considered significant at p<0.05.

Results

A thrombosis model in zebrafish

We used a thrombosis model in zebrafish and confirmed the contribution of the coagulation system by a knock down approach of the prothrombin gene. Tg(cd41:GFP) larvae were irradiated and the accumulation of GFP⁺ thrombocytes was recorded. Approximately five seconds after inducing endothelial injury, the first thrombocytes adhered to the vessel wall and formed a thrombus. The thrombus rapidly grew in size and after about 45 seconds had occluded the vessel and blood flow came to a stand still (figure 1A-G). To investigate the contribution of the coagulation system, we knocked down prothrombin expression by introducing a morpholino targeted against the translation start site (MO_prothrombin). Three days after fertilization laser damage was induced in the posterior cardinal vein and the time to occlusion (TTO) was measured. Knockdown of thrombin expression increased the TTO from 34 ± 6 to 132 ± 12 s (n = 28 embryos per group) (figure 1H). These data confirm that the coagulation mechanism contributes to thrombus formation in a zebrafish, similarly as in humans.

Figure 1. Laser induced thrombosis in Tg(cd41:GFP) zebrafish larvae (5 dpf) and the effect of the knockdown of prothrombin.

(A) Endothelial cells in the pericardinal vein were damaged by laser irradiation. Approximately five seconds after endothelial injury the first blood cells adhered to the vessel wall and started forming a thrombus (Ai), which grew rapidly within the next seconds (Aii).

(B) To keep variation between fish as low as possible the same point (5 somites posterior to the anal fin) was taken as a target for laser irradiation of the endothelium.

(C-G) Induction of a thrombus in Tg(cd41:GFP) resulted in fluorescent cells adhering to the injured vessel wall. These cells were identified as zebrafish thrombocytes. The arrowhead indicates the location of laser injury. Panels D-G display different intervals after laser injury.

(H) The average time to occlusion after knockdown of prothrombin by an ATG-morpholino (MO_prothrombin) compared to uninjected controls (UIC). A significant difference is indicated by *.

Expression of mlck genes in thrombocytes

We identified four zebrafish orthologues for the three human MLCK genes (supplemental Figure 1). To assess which of these zebrafish mlck genes is expressed in zebrafish thrombocytes, FACS analyses was performed on a single cell suspension of Tg(cd41:GFP) larvae three days post fertilization. A CD41⁺ cell population was detected in this transgenic line, which represented about 0.18 % of total cell number (figure 2Ai). The cd41-GFP⁺ population was enriched by two rounds of sorting to a purity of 83.8%. Then, approximately 4000 cells GFP⁺ cells (figure 2Aii) were compared with 4000 GFP⁻ cells (figure 2Aiii) by micro-array analysis. Expression of the thrombocyte marker cd41 was 8.7 fold higher in GFP⁺ cells than in GFP⁻ cells, confirming the specificity of this marker. Two probe sets were present for mlck1a on the micro-array chip. Both showed 4.5 times higher expression of mlck1a. In contrast, for mlck1b this number was 0.77 indicating lower expression levels of this gene in thrombocytes (figure 2B). Unfortunately, the array did not contain probe sets for mlck2 and mlck3 genes. Together, these data indicate that zebrafish thrombocytes express mlck1a and not mlck1b.

Figure 2. Expression of mlck genes in thrombocytes.

(A) A single cells suspension of about thousand 3 day old Tg(cd41:GFP) zebrafish embryos was sorted on GFP expression by FACS. Of the total cell population (Ai) 0.18% of the cells showed positive GFP expression (Aii). Four thousand GFP⁺ cells (purity of 83.8% after two rounds of sorting) were sorted from these embryos and compared to 4000 GFP⁻ cells (Aiii) by microarray analysis.

(B) Micro-array analysis of RNA of four thousand GFP⁺ cells compared to four thousand GFP⁻ cells. Cd41 was used as a positive control. The data show 4.5 times higher expression of mlck1a (no probes were available for mlck2 and mlck3).

Expression of mlck subtypes in a zebrafish thrombus

To confirm differences in expression of mlck1a and mlck1b and search further for expression of mlck2 and -3, we performed whole-mount in situ hybridization in zebrafish laser-irradiated at the posterior cardinal vein. Embryos were fixed and incubated with probes specific for mlck1a, -1b, -2 and -3. A probe for cd41 was used as a positive control. In line with the array data, mlck1a was expressed at the site of thrombus formation (figure 3A). Individual cells were visible in the posterior cardinal vein at the site of thrombus formation and intersomitic vessels. Staining for mlck1a strongly overlapped staining for cd41, confirming expression in thrombocytes. Probes for mlck1b, -2 and -3 showed no staining of blood cells in the posterior cardinal vein both on whole-mount and cross-sections (figure 3B). The expression domains of the zebrafish mlck genes in non laser-irradiated whole-mount embryos are shown in supplemental figure 2. These findings confirm that mlck1a is the major mlck subtype (and possibly the only one) expressed in zebrafish thrombocytes.

Figure 3. In situ hybridization with mlck specific probes after induction of a thrombus in the pericardinal vein.

(A) After induction of a thrombus, embryos were fixed and wholemount in situ hybridization was performed with probes specific for the thrombocyte marker cd41 and mlck1a. The probe for cd41 was used as a positive control and showed staining of single cells on the site of thrombus formation. Sections of these embryos showed staining at the location of the pericardinal vein. The probe for mlck1a clearly displayed a similar staining pattern as the cd41 probe on wholemount and sectioned embryos.

(B) Probes for mlck1b, mlck2, and mlck3 showed no staining at the site of thrombus formation.

cd41-GFP⁺ cells with knocked-down mlck1a show a decreased spreading on fibrinogen

To clarify the role of mlck1a in thrombocyte function and thrombus formation, we knocked down gene expression by designing two splice site morpholinos upstream of the protein kinase domain. Morpholino 1 (MO1_mlck1a) blocked the splicedonor site of exon 6 and morpholino 2 (MO2_mlck1a) that of exon 9 (figure 4A). Injection with MO_p53 served as a control. Injection of 8 ng of morpholinos did not change the wild-type appearance of the embryo (data not shown).

Figure 4. Knockdown of mlck1a attenuates spreading of cd41-GFP⁺ cells on fibrinogen.

(A) Target sites of the 2 splice site morpholinos (MO1_mlck1a and MO2_mlck1a) for mlck1a. Both were designed to target an exon-intron boundary upstream of the 5′ end of the protein kinase domain.

(B) Expression by RT-PCR of mlck1a in the embryos injected with a p53 targeting morpholino showed a normal sized band. Injection of MO1 resulted in two bands, a wildtype band and a shorter band. Injection of MO2 resulted in three bands: a wildtype band and two shorter bands.

(C-D) Fluorescent thrombocytes on fibrinogen coated slides at room temperature. Images of fluorescent channel of the different morphologic features of spreading cd41-GFP⁺ cells on fibrinogen. After 30 minutes, mainly two different cells morphologies were seen: (C) round cells, absence of initiation of cells shape change (D) cells that were fully spread on fibrinogen.

(E) Number of thrombocytes that preserved their round morphology in knockdowns of mlck1a by MO1_mlck1a and MO2_mlck1a compared to the control situation (MO_p53).

The efficiency of the morpholinos was determined by RT-PCR with mlck1a specific primers of cDNA extracted from embryos of the same experiment and compared with MO_p53 injected embryos (figure 4B). Injection of MO1_mlck1a induced the appearance of a second band of smaller size. Knockdown of mlck1a with MO2_mlck1a induced two extra bands. These bands were cut out of the gel and cloned. For MO1_mlck1a, 3 out of 17 clones showed a wild type sequence (supplemental figure 3A); 14/17 showed a 36 nucleotide deletion. The protein sequence of this deletion showed a high level of orthology with human, mouse and rat (supplemental figure 3B). For MO2_mlck1a, 4/18 colonies showed a normal boundary for exon 8 - 9, while in 14/18 colonies either exon 9 (6/18) or both exon 9 and 10 (8/18) were spliced out (supplemental figure 3C). Both splice forms resulted in a frameshift and premature termination of the transcript. These data show that both MOs severely reduce the level of functional protein.

The impact on thrombocyte function was measured in a static adhesion assay. Thrombocytes were isolated and spread on a fibrinogen-coated surface. cd41-GFP⁺ thrombocytes were scored for changes in morphology, as defined (figure 4C-D). In the control, the MO_p53-treated zebrafish, the number of thrombocytes that completely preserved the resting, round morphology was 12.0 ± 3.3% of CD41-GFP⁺ cells. This number increased to 32.5 ± 2.1% and 31.4 ± 4.4% (n = 3 experiments; p < 0.05) in thrombocytes from zebrafish treated with MO1_mlck1a and MO2_mlck1a respectively. This data indicates that the thrombocytes’ ability to undergo a morphology change upon contact with fibrinogen depends on mlck1a.

Knockdown of mlck1a retards thrombus formation

In the MO_p53 injected embryos, laser damage of the dorsal aorta endothelium induced thrombus formation and complete occlusion of the vessel after 31 ± 2 seconds (n = 19; figure 5). Knockdown of mlck1a induced a 70% fold increase in occlusion time in MO1_mlck1a injected embryos and a 120% fold increase in MO2_mlck1a injected embryos (p<0.005). These data show that in zebrafish, platelet Mlck1a is important for formation of a thrombus. The observations that the human megakaryocytic CHRF-288-11 cell-line expresses mlck1a but not mlck2 and -3 and that the MLCK inhibitor ML-7 inhibits collagen-induced aggregation by human platelets suggest that the role for Mlck1a established in zebrafish might be similar in human individuals (supplemental figure 4).

Figure 5. Knockdown of mlck1a increases the time to occlusion.

Time to occlusion was measured in knockdown of mlck1a by two (MO1_mlck1a and MO2_mlck1a) morpolinos and in controls (MO_p53). Time to occlusion was defined as the time between the start of laser irradiation and complete arrest of blood flow.

Discussion

This study addressed the question which MLCK gene contributes to platelet shape change and thrombus formation, using the zebrafish as a tool for rapid gene inactivation. Transient knockdown of mlck1a by a reverse genetic approach decreased the ability of thrombocytes to spread on fibrinogen and impaired thrombus formation in an in vivo thrombosis model. These data indicate that mlck1a is the subtype that drives the phosphorylation of myosin light chain and thereby the change in platelet morphology. We found no evidence to indicate that mlck1b and mlck2 and -3 were not involved. Activation of Mlck1a is a late step in a pathway that links receptor activation with phosphorylation of myosin IIa. A signaling protein upstream of Mlck1a is calmodulin. In human platelets stimulated with the glycoprotein VI ligand convulxin, inhibition of calmodulin by W-7 caused a 50% fall in shape change[26]. Inhibition of myosin by blebbistatin strongly decreased aggregation/secretion, impaired the formation of stress fibers on a collagen surface and increased thrombus embolization in vitro[27]. Thus, signaling though Mlck1a is vital for platelet morphology changes and thrombus formation. Activation of the Mlck1a pathway appears a general property of platelet activating agents with some stimulating through Gq and phospholipase Cβ, e.g. thrombin and ADP, and others through phospholipase Cγ, e.g. collagen[1]. Toth-Zsamboki[28] et al, showed that platelet shape change can also be induced by P2X1- mediated Ca²⁺ influx and Ca²⁺/CaM-dependant initiation of myosin IIa phosphorylation. MLCK activity is enhanced by P2X1 mediated ERK2 activation which leads to amplified platelet secretion. Aggregation was completely inhibited by the MLCK inhibitor ML-7. cAMP-dependent protein kinase phosphorylates MLCK, which interferes with binding of calmodulin and causes overall inhibition[29]. In our study MLCK-depleted thrombocytes showed impaired spreading on fibrinogen. We have demonstrated that inhibition of Mlck1a results in an impaired thrombus formation in vivo. It is possible that this impaired thrombus formation is due to unstable thrombi and increased embolization, which results in increased occlusion times. A parallel route that controls the phosphorylation state of myosin light chain is through Rho-kinase (ROCK) and inhibition of myosin light chain phosphatase. Inhibition of ROCK by Y27632 hardly affected myosin light chain phosphorylation and shape change after stimulation with ADP or collagen related peptide[1] and inhibition of ROCK had a minor effect on static adhesion to different surfaces[30-32]. Mice treated with Y27632 were unable to form stable thrombi and displayed fast embolization in an in vivo thrombosis model[27]. Activation of this pathway is restricted to thromboxane A₂, thrombin and lysophosphatidic acid whose receptors in addition to Gq also activate G12/13 and signaling to ROCK[1].

Mammals express three genes for MLCK. MLCK1 (also named smooth muscle MLCK or MYLK1) expresses three transcripts due to alternate promoters[8]. The first transcript codes for the short isoform (130 kDa) and contains a kinase domain, three immunoglobin domains, a fibronectin domain and an actin binding domain and is ubiquitously expressed in adult tissue with the highest concentration in smooth muscle cells[33]. The second transcript codes for the long isoform (also referred to as 210-kDa MLCK), contains an extra six immunoglobin domains and two actin-binding domains in addition to the sequence of the short isoform and is expressed in embryonic smooth muscle cells and non-muscle cells[34]^, [35, 36]. A third transcript results in the expression of the 17 kDa protein telokin and only contains the C-terminal Ig domain[37]. Telokin plays a role in Ca²⁺ desensitization of smooth muscle force by cyclic nucleotides[38]. Mice with blocked expression of the three MLCK1 transcripts developed to full size, but died within 1-5 hours after birth[39].

Mammalian platelets expresses MLCK1, which is activated by Ca²⁺/calmodulin, and MLCK2 also contains a Ca²⁺/calmodulin binding site[8]. The actin-binding sequence in 130 kDa MLCK1 was found necessary for high affinity binding[40]. The function of the other domains in the different MLCK’s is unclear and is remains uncertain why certain cells express a specific MLCK gene. We found that the zebrafish transcribes four genes for Mlck (supplemental figure 2). This difference in number compared to other vertebrates is caused by the expression of two genes for mlck1, mlck1a and mlck1b in zebrafish. Although the zebrafish genome has orthologues for the mlck2 and mlck3 genes, their thrombocytes only express mlck1a.

A recent study in neutrophils suggests that the function of MLCK1 might extend beyond the regulation of myosin light chain kinase. In a lung injury model, mice lacking the long, 210 kDa MLCK1 isoform showed impaired attachment of neutrophils to the endothelium and further migration[41]. The defect was caused by impaired activation of β₂ integrins, revealing a role of this MLCK1 isoform in integrin regulation in normal mouse neutrophils. This effect was surprisingly independent of myosin II activity. The myosin light chain could be phosphorylated in MLCK1 depleted neutrophils and inhibition of MLCK and myosin II led to contradictory effects on neutrophil adhesion and actin polymerization. Platelets express several β1-integrins (α2β1, α5β1 and α6β1) and β3 integrins (αIIbβ3 and αVβ3), but no β2- integrins. Future studies are needed to demonstrate a similar effect of MLCK signaling to integrins in platelets.