Proteasome proteolysis supports stimulated platelet function and thrombosis

Abstract

Objective

Proteasome inhibitors are in use to treat hematologic cancers, but also reduce thrombosis. Whether the proteasome participates in platelet activation or function is opaque since little is known of the proteasome in these terminally differentiated cells.

Approach and Results

Platelets displayed all three primary proteasome protease activities, which MG132 and bortezomib (Velcade^®) inhibited. Proteasome substrates are marked by ubiquitin, and platelets contained a functional ubiquitination system that modified the proteome by mono- and poly-ubiquitination. Systemic MG132 strongly suppressed formation of occlusive, platelet-rich thrombi in FeCl₃-damaged carotid arteries. Transfusion of platelets treated ex vivo with MG132 and washed prior to transfusion into thrombocytopenic mice also reduced carotid artery thrombosis. Proteasome inhibition reduced platelet aggregation by low thrombin concentrations and ristocetin-stimulated agglutination through the GPIb-IX-V complex. This receptor was not appropriately internalized after proteasome inhibition in stimulated platelets, and spreading and clot retraction after MG132 exposure also were decreased. The effects of proteasome inhibitors were not confined to a single receptor as MG132 suppressed thrombin-, ADP-, and LPS-stimulated microparticle shedding. Proteasome inhibition increased ubiquitin decoration of cytoplasmic proteins, including the cytoskeletal proteins Filamin A and Talin-1. Mass spectrometry revealed a single MG132-sensitive tryptic cleavage after R₁₇₄₅ in an extended Filamin A loop, which would separate its actin-binding domain from its carboxy terminal GPIbα binding domain.

Conclusions

Platelets contain a ubiquitin/proteasome system that marks cytoskeletal proteins for proteolytic modification to promote productive platelet-platelet and platelet-wall interactions.

Introduction

Proteasome inhibitors are widely used for the treatment of hematologic cancers, specifically relapsed/refractory multiple myeloma and mantle cell lymphoma,^1–3 yet little is known of their impact on platelet function and hemostasis. Potentially, these agents may also target platelets because a cyclic course of the proteasome inhibitor bortezomib (Velcade^®) induces significant cyclic thrombocytopenia,⁴ and proteasome inhibition hastens platelet death and turnover.⁵ Bortezomib therapy also associates with reduced thrombosis.^{6, 7} Beyond this, there is little evidence of proteasome action on the platelet proteome, their activation, or on thrombosis.

Platelets are recruited to the vascular wall at high shear through the GPIb-IX-V receptor complex,⁸ where the GPIbα subunit binds von Willebrand factor (vWF)⁹ while its C-terminal cytoplasmic domain interacts with the actin binding protein Filamin A with high affinity.¹⁰ This interaction tethers the receptor complex to the platelet cytoskeleton, maintaining the cytoskeletal architecture of resting platelets and those adhering in vessels at high shear rates.¹¹ The GPIbα subunit also contains a high affinity binding site for thrombin that contributes to platelet activation when thrombin concentrations are low.¹² These interactions are essential for thrombosis.¹³ Thrombotic platelet deposition is modeled in mice by FeCl₃ induced injury of carotid arteries,^{14, 15} resulting in rapid platelet adhesion and formation of an occlusive platelet-rich thrombus at the site of injury.

Nucleated cells remove proteins from their proteome by proteasome-catalyzed proteolysis. This multimeric complex consists of a 20S catalytic core of non-catalytic α-subunits and three distinct β-subunits that hydrolyze peptide bonds of unfolded protein substrates by β1 caspase-like hydrolysis, β2 tryptic hydrolysis, and β5 chymotryptic cleavage.^16–19 The proteasome is capped by a 19S regulatory subunit that conducts substrate recognition, deubiquitination, unfolding, and protein translocation into the 20S core.^{20, 21} Substrates for proteasome hydrolysis are recognized by the 19S core through the covalent conjugation of monomeric or polymeric chains of the ~8 kDa ubiquitin to the targeted protein.^22–24

Newly released reticulated platelets express a richer proteome than older, dense platelets.²⁵ In addition, platelets contain several components of the 20S proteasome core^26–29 and possess at least the chymotryptic activity of the proteasome.³⁰

Here we show the expression of a functional ubiquitin/proteasome system in platelets. By investigating the impact of proteasome inhibition on a well established mouse model of thrombosis and on a range of ex vivo activities, we conclusively demonstrate that the platelet proteasome contributes to cellular activation and function. Therapeutic proteasome inhibition in platelets produces a hypothrombotic state, but also might augment anti-platelet therapy.

Materials and Methods

Materials and Methods are available in the online-only Data Supplement.

Results

The platelet proteasome aids occlusive thrombosis

Human platelets expressed all three primary proteolytic activities of the proteasome, effectively hydrolyzing peptides through chymotryptic, tryptic and caspase-like cleavage (Fig. 1A). The luminescent product was the same for each luminogenic proteasome substrate, so the caspase-like activity was approximately four times more efficient than either the tryptic or chymotryptic activity. The trimeric leucinyl proteasome inhibitor MG132 at 30 μM completely blocked platelet chymotryptic and caspase-like activities, and reduced the tryptic activity by half. MG132 at 30 μM is just optimally inhibitory (Supplemental Fig. I). The structurally unrelated specific boron-based proteasome inhibitor, bortezomib, inhibited both chymotryptic and caspase-like activities, but not the tryptic activity (not shown). Proteasome activity was retained in murine platelets that hydrolyzed these three substrates at equivalent ratios (Supplementary Fig. I).

Figure 1. Platelets express a functional proteasome that contributes to occlusive thrombosis.

(A) Platelets contain a functional proteasome. Proteasome proteolytic activities were individually assayed using luminogenic substrates for indicated catalytic activities in lysates of untreated and MG132-treated platelets (n=3; ^**p≤ 0.01). (B) Fluorescent platelet accretion after FeCl₃ damage to carotid arteries is reduced by MG132. Video frames at 1 min intervals of thrombus formation in FeCl₃-damaged carotid arteries (n=5). (C) MG132 lengthens the time to vascular occlusion. Time to cessation of blood flow in mice treated as in the preceding panel determined by cessation of platelet movement (n=8 experimental, 6 control; ^***p < 0.001). (D) ex vivo MG132 pretreatment of platelets prolongs occlusion time after transfusion into irradiated thrombocytopenic mice. Platelets from donor mice were first pretreated for 30 min with MG132 and were washed before transfusion into normal BL6 mice and FeCl3-induced thrombosis of carotid arteries initiated 15 min later (n=4 experimental, 3 control; ^*p < 0.05).

We determined whether proteasome inhibition affected thrombosis by systemically injecting MG132 into BL6 mice to achieve an estimated initial circulating concentration of 30 μM along with rhodamine dye to fluorescently label circulating platelets. We initiated occlusive thrombosis 15 min later in a surgically exposed external carotid artery by a brief ectopic application of 7.5% FeCl₃.^{14, 15} This oxidative insult to the vascular wall resulted in the deposition of fluorescently labeled platelets along the damaged vessel wall that increased over several minutes (Fig. 1B). Typically, complete occlusion of the vessel occurred by 10 min after FeCl₃ treatment, but in animals previously injected with MG132 occlusion was significantly delayed to 25 min (Fig. 1C).

The delay in thrombosis after systemically administering MG132 need not solely reflect participation of platelet proteasomes in thrombosis. We therefore isolated and washed platelets from wild-type BL6 mice and treated these isolated cells with 30 μM MG132 or buffer, washed the cells by centrifugation and transfused these cells into mice previously rendered thrombocytopenic by gamma irradiation. We induced carotid artery thrombosis with ectopic FeCl₃ as before to find control platelets were fully functional and occluded the carotid artery by 10 min (Fig. 1D). We again found occlusive thrombosis was delayed in mice reconstituted with platelets treated with MG132 ex vivo. The platelet proteasome thus participates in arterial thrombosis.

Platelets contain a stimulatable ubquitination system

Modification of platelet function by MG132 suggests platelets may designate proteins as substrates for proteolysis by ubiquitination, as in nucleated cells. We determined whether platelets contained E1, E2, E3 enzymes that sequentially conjugate ubiquitin to target proteins. In order to ensure that we assessed only platelet proteins, we also isolated highly purified, negatively selected platelets that were essentially free of nucleated cells (less than 1 monocyte in 10⁹ platelets³¹). Western blotting showed that these platelets contained the E1 activating enzyme UBE1, E2 and E3 ligases, and Rpn2 and the 11S α-subunit of the 19S regulatory complex (Fig. 2A).

Figure 2. Platelets contain a functional ubiquitination system.

(A) Platelets contain components of ubiquitin-proteasome system. Immunoblots of platelet E1 activating enzyme, E2 and E3 conjugating enzymes, and proteasome regulatory cap proteins (n=3). (B) Platelet E1 is functional. Platelet lysates substituted for recombinant E1 and together with recombinant E2, E3, ubiquitin and ATP ubiquitinated recombinant p53 (n=3). (C) Agonist stimulation increases platelet proteome ubiquitination. Lysates of platelets stimulated with buffer, thrombin (thr) or ADP were immunoblotted with FK2 antibody against mono-and polyubiquitinated proteins (n=3). (D) MG132 increases proteome ubiquitination. Lysates of platelets, treated or not with MG132 were stimulated, or not, before immunoblotting with FK2 (n=3). (E) MG132 increases poly-ubiquitination of platelet proteins (n=3). Western blotting of platelet lysates with FK1 antibody (n=3).

We used in-vitro ubiquitination of exogenous p53, since platelets lack this transcription factor, to determine whether platelet UBE1 was functional. Incubation of recombinant p53 with recombinant E1 and appropriate recombinant E2 and E3 enzymes along with ubiquitin and ATP resulted in robust p53 ubiquitination with formation of numerous slowly migrating adducts (Fig. 2B). Substitution of a platelet lysate for recombinant E1 also promoted in-vitro ubiquitination of p53, although the higher molecular weight ladder was less prominent than produced by recombinant E1 (Fig. 2B).

We next determined whether endogenous platelet proteins were modified by ubiquitin. Western blotting with FK2 antibody that recognizes both mono- and poly-ubiquitin chains showed that the proteome of quiescent cells contained multiple ubiquitinated proteins (Fig. 2C). The intensity of ubiquitination increased after thrombin activation, although the pattern and position of the adducted bands was unaltered compared to control cells. ADP stimulation also increased ubiquitination of platelet proteins, but this primarily reflected an increase in modification of very slowly migrating proteins (Fig. 2C).

We investigated whether inhibiting the platelet proteasome altered the amount of ubiquitinated proteins. Pretreatment with MG132 increased both the abundance and the levels of ubiquitinated proteins as detected with the anti-ubiquitin antibody (Fig. 2D). In fact, MG132 was significantly more effective than thrombin or ADP stimulation in enhancing decoration of the proteome with ubiquitin, and the combination of agonist stimulation and MG132 was not different from the effect of MG132 alone (Fig. 2D). We observed a similar increase in ubiquitinated proteins after treatment with bortezomib (not shown). Probing the platelet proteome with FK1 antibody that recognizes only poly-ubiquitinated proteins revealed less abundant modification in unstimulated cells (Fig. 2E). This was increased by agonist stimulation and was again greatly enhanced by MG132 treatment. Comparison of FK2 (Fig. 2D) and FK1 (Fig. 2E) immunoblots showed distinct patterns of mono-and poly-ubiquitination of the platelet proteome.

Filamin A is ubiquitinated, and truncated by the proteasome

Filamin A links the GPIb-IX-V complex to actin filaments of the cytoskeleton to modify cytoskeletal shape.³² Filamin A was present in the soluble fraction of quiescent platelets as a 225 kD fragment of the 280 kDa native protein (Fig. 3A). MG132 inhibition of the proteasome decreased the amount of this smaller band and increased the amount of intact Filamin A. Capture of ubiquitinated platelet proteins by a sushi domain column followed by immunoblotting using anti-Filamin A antibody showed that the fragment of Filamin A in quiescent cells was constitutively modified with ubiquitin (Fig. 3B). This approach also showed that MG132 increased the amount of Filamin A ubiquitination, and increased the apparent size of the ubiquitinated Filamin A (Fig. 3B). That Filamin A was ubiquitinated and that a larger, more extensively ubiquitinated protein accumulated after MG132 treatment was confirmed by the converse experiment where Filamin A was immunoprecipitated from platelet lysates and then probed for ubiquitin (Fig. 3C) using FK2 antibody.

Figure 3. MG132 protects cytoskeletal protein cleavage.

(A) Filamin A is ubiquitinated and its length increased by MG132. Western blot of Filamin A in the soluble fraction of platelets before and after MG132 exposure (n=3). (B) Ubiquitination and size of Filamin A are increased by MG132. Western blot of Filamin A in eluates of ubiquitinated platelet proteins captured with sushi columns (n=3). (C) Filamin A ubiquitination and size are increased by MG132. FK2 western blot of mono- and poly-ubiquitinated proteins immuoprecipitated with anti-Filamin A. (D) Talin-1 is ubiquitinated and its size increased by MG132. Western blot of Talin-1 before and after MG132 exposure (n=3). (E) Ubiquitination and size of Talin-1 are increased by MG132. Talin-1 in ubiquitinated platelet proteins captured by sushi domain chromatography (n=3). (F) Talin-1 ubiquitination and size are increased by MG132. FK2 western blot of mono- and poly-ubiquitinated proteins immuoprecipitated by anti-Talin-1 (n=3). (G) MG132 increases high molecular weight proteins. Coomassie stained gel of resolved platelet cytoplasmic proteins. (H) Mass spectrometer determination of the mz 126/127 ratio of TMT-labeled peptides along the Filamin A sequence (n=2). (I) Pictogram of Filamin A structure.

The cytoskeletal protein Talin-1 participates in cell spreading, and similar experimental approaches of western blotting (Fig. 3D), sushi domain capture of ubiquitin adducts (Fig. 3E), and immunoblotting of captured Talin-1 for mono-and poly-ubiquitin chains (Fig. 3F) showed that Talin-1 was present in the cytosol in a rapidly migrating, ubiquitinated form. Again, MG132 treatment increased its apparent size and ubiquitin content.

Coomassie blue staining of soluble proteins showed MG132 treatment did not alter overall platelet protein mass or composition, but there were two abundant exceptions (Fig 3G). Two new, slowly migrating bands appeared in the soluble fraction of platelets treated with MG132. Mass spectrometry showed Filamin A to be present in the new band 1, and Talin-1 in bands 1 and 2 of the resolved soluble proteins (data not shown), corresponding to the altered mobility detected by western blotting.

We used Tandem Mass spectrometry Tags (TMT)—chemically identical tags differentially substituted with heavy atoms that enable quantitative multiplexed analysis—to quantitatively compare the amount of Filamin A peptides in tryptic digests of control and MG132 treated platelets. Untreated (band 3) and MG132 treated (band 1) platelets were separately digested with trypsin, their primary amines exhaustively modified with isobaric TMT tags (m/z 126 and m/z 127, respectively) before combined analysis by mass spectrometry. The peptides confirmed both bands contained Filamin A, and the tags showed the abundance of Filamin A tryptic peptides in band 1 and band 3 uniformly gave a m/z 126/127 ratio of 1.9. This ratio plunged to 0.08 after the tryptic peptide containing residues 1718–1745 (Fig. 3H), demonstrating increased abundance of the full length protein after MG132 treatment. Chymotryptic digestion of band 3 isolated from control cells generated a distinct peptide map and selected reaction monitoring identified a new Filamin A peptide with S₁₇₄₆ as its amino terminus (not shown). This identifies the R₁₇₄₅ - S₁₇₄₆ bond as the site of cleavage in control cells, which maps into the unfolded hinge 1 region of Filamin A (Fig. 3I). Notably, this cleavage is tryptic-like, and is distinct from the previously determined calpain cut site.³³

Proteasome inhibition reduces cytoskeleton-dependent functions

Filamin A is a critical regulator of cytoskeletal structure and function, so we next tested the hypothesis that proteasomal inhibition alters platelet functions regulated by cytoskeletal dynamics, including GPIb function, microparticle generation, and clot retraction.

First, we imaged cytoskeletal dependent spreading by Total Internal Reflection Microscopy (TIRF) that detects only fluorophore closely opposed (≤ 200 nm) to a glass matrix. Calcein-labeled platelets adhered over time after activation by thrombin with extension of filopodia followed by lamellipodia (video supplement) that resulted in adherent, spread cells (Fig. 4A). In contrast, while MG132-treated cells adhered and spread after thrombin stimulation, they did so with less frequency, rapidity, and ultimately were less splayed (Fig. 4B).

Figure 4. MG132 suppresses stimulated spreading, microparticle shedding, and clot retraction.

(A) MG132 disturbs thrombin stimulated adhesion and spreading. Interaction of control or thrombin (thr, 0.025 U) stimulated platelets with a glass substrate were imaged by Total Internal Reflection Microscopy after 5 min (n=3) (scale bar = 10 μM). (B) Platelet area after MG132 treatment. Platelet area in Panel A was quantified by Imagepro plus software (n=3). ^***p≤ 0.001 (C) Microparticle shedding by platelets stimulated with 0.2 U thrombin with or without MG132 (n=3). ^**p≤ 0.01 (D) Bortezomib (n=3). ^*p≤ 0.05 (E) Microparticle shedding by platelets stimulated with lipopolysaccharide (LPS, n=3). ^*p≤ 0.05 (F) MG132 represses clot retraction. Ratio of the image surface area of images of thrombin-induced clots over time (n=3). ^**p≤ 0.01

Stimulated platelets release pro-thrombotic microparticles from their surface that depends on cytoskeletal rearrangement after stimulation.³⁴ Thrombin induced a ~6 fold increase in microparticle shedding (Fig. 4C and 4D), which was significantly reduced by pre-treating the platelets with either MG132 (Fig. 4C) or bortezomib (Fig. 4D). MG132 also blocked microparticle release from ADP (not shown) or LPS-stimulated platelets (Fig. 4E). The majority of these particles express phosphatidylserine that promote thrombosis through tissue factor pro-coagulant activity (Supplemental Fig. II).

Cytoskeletal rearrangement retracts newly formed thrombi, promoting wound repair.^{35, 36} Stimulation of platelets with thrombin induced rapid formation of a thrombus that then consolidated over time (Fig. 4F). MG132 interfered with this process, ultimately decreasing retraction by 60%. MG132 does not act by reducing surface CD36 or GPVI, nor does MG132 modulate intracellular Ca⁺⁺ concentrations (Supplemental Fig. III). MG132 does, however, prevent thrombin-induced loss of surface GP1bα (Supplemental Fig. III).

Proteasome inhibition selectively reduces aggregation stimulated by low concentrations of thrombin

Thrombin stimulates platelets by cleaving surface PAR1 (protease-activated receptor 1) to create a self stimulatory terminal SFLLRN peptide,³⁷ but at low concentrations thrombin activates platelets through the high affinity GPIb-IX-V receptor.¹² Both MG132 and bortezomib reduced homotypic platelet aggregation at low concentrations of thrombin (Fig. 5A).

Figure 5. MG132 reduces aggregation in response to low dose thrombin, and prevents GP1bα downregulation.

(A) MG132 and bortezomib suppress platelet aggregation induced through the high affinity thrombin receptor, GP1b-IX-V. Aggregation induced by 0.025 U thrombin (thr) was reduced by MG132 and bortezomib (n=3). (B) MG132 failed to reduce aggregation stimulated by the PAR1 (protease-activated receptor 1) agonist peptide SFLLRN (n=3). (C) Neither MG132 nor bortezomib block aggregation from thrombin activation of PAR1. Platelet aggregation by 0.1 U thrombin with or without SZ2 antibody (n=3). (D) GP1bα or PAR1 inhibition alone does not induce aggregation (n=3). (E) MG132 repressed stimulated GPIbα down regulation. Flow cytometry of surface CD42b (GPIbα) expression on platelets treated or not with MG132 and stimulated or not with 0.025 U thrombin (n=3). (F) Bortezomib repressed surface GPIbα down regulation. Flow cytometry of GPIbα expression on control and thrombin stimulated platelets (n=3). (G) MG132 suppressed ristocetin induced agglutination. Quiescent platelets pretreated or not with MG132 were agglutinated with ristocetin (n=3). (H) Bortezomib suppressed ristocetin induced agglutination. Quiescent platelets pretreated or not with bortezomib were agglutinated with ristocetin (n=3).

The effect of proteasome inhibitors was on the high affinity GPIb-IX-V complex because the SZ2 antibody against GPIbα (CD42b) fully blocked aggregation at a low concentration of thrombin (Fig. 5A). The PAR1 agonist SFLLRN induced platelet aggregation that was completely blocked by a PAR1 specific antagonist (RWJ56110, Fig. 5B). However, neither MG132 (Fig. 5B) nor bortezomib (not shown), suppressed aggregation induced by SFLLRN. Aggregation induced by a low, submaximal amount of SFLLRN also was unimpeded by MG132 (Supplementary Fig. IV). Thus, the inhibition observed at low thrombin concentration by proteasome inhibitors is independent of PAR1. MG132 also failed to affect aggregation induced by AYPGFK stimulation of PAR 4 (Supplementary Fig. IV). Furthermore, neither MG132 nor bortezomib inhibited platelet aggregation stimulated by a higher thrombin concentration that acts through PAR1 (Fig. 5C). The modifiers SZ2, MG132, RWJ56110 or bortezomib (not shown) by themselves did not induce aggregation (Fig. 5D).

Proteasome protease was required by the high affinity thrombin receptor because epoxomicin inhibited platelet aggregation by low thrombin concentrations, but not high thrombin concentrations that stimulate PAR1 (Supplementary Fig. V). Epoxomicin is a fungal metabolite, structurally unrelated to either MG132 or bortezomib, that specifically inhibits proteasome protolysis without inhibiting calpain, trypsin, chymotrypsin or cathepsin B at concentrations 50 times higher than we used to block platelet aggregation.³⁸

The GPIb-IX-V complex is displayed on the surface of quiescent cells, and is down regulated after thrombin stimulation.³⁹ We found both MG132 (Fig. 5E) and bortezomib (Fig. 5F) stabilized the GPIb-IX-V complex on the surface of stimulated cells, and actually enhanced surface GPIbα (CD42b) expression. Ristocetin increases the affinity of GPIbα for vWF, agglutinating unactivated cells, and pre-treatment of quiescent platelets with MG132 (Fig. 5G) or bortezomib (Fig. 5H) reduced ristocetin and botrocetin (not shown) induced platelet agglutination. Thus despite the increased surface expression of the GPIb-IX-V complex after proteasome inhibition, the complex is less able to interact with vWF than in control platelets.

Discussion

Multiple myeloma is associated with an increased incidence of venous^40–42 and arterial⁴³ thrombotic disease, and bortezomib therapy suppresses these thrombotic states.⁷ Platelets display proteasome chymotryptic activity that is greatly stimulated by soluble agonists.³⁰ Additionally, bortezomib represses ADP-induced aggregation²⁷ and platelets isolated from patients receiving bortezomib are hyporesponsive to other stimuli.^{6, 7} A second proteasome inhibitor, PSI, suppresses thrombosis in hypertensive animals.⁴⁴

We extend these observations by showing that platelets contain an intact and functional ubiquitin/proteasome system that participates in agonist stimulated responses, especially those aided by cytoskeletal rearrangement. We found that inhibition of the proteasome’s proteolytic activity reduced thrombosis at sites of oxidative damaged to murine carotid arterial walls. Since MG132 was introduced into the circulation in this experiment, protection need not have been a direct effect of MG132 on platelet function, but the target of MG132 in platelets is in fact the proteasome. We treated purified platelets with MG132 ex vivo, washed them, and then transfused these cells into mice rendered thrombocytopenic by prior gamma irradiation. This platelet-specific exposure to MG132 interfered with the in vivo function of platelets because MG132 treated platelets were deficient in their occlusion of a FeCl₃-damaged carotid artery. Remarkably, this ex vivo exposure exactly mimicked systemic exposure to MG132. This establishes a role for the platelet proteasome in thrombosis.

Platelets contained a functional ubiquitin system that modified cellular proteins to mark them as proteasome substrates. The proteome of quiescent platelets contained numerous ubiquitin-protein conjugates whose adduction was increased upon stimulation, consistent with the prior observation that collagen activation stimulates ubiquitination of platelet Syk kinase⁴⁵ through the E3 ligase Cbl-b.⁴⁶ Two of the proteins decorated with ubiquitin in quiescent platelets were Filamin A and Talin-1. Agonist stimulation modestly increased the amount of ubiquitin esterified in the platelet proteome, but blockade of the proteasome was far more effective than agonist stimulation for this, with the result that heavily ubiquitinated Filamin A, and Talin-1 accumulated in the cytoplasm of cells with diminished proteasome proteolytic activity.

Filamin A is organized into an actin-binding domain (ABD), 24 tightly compacted FERM immunoglobulin-like repeats connected with short linking sequences, and two unstructured hinge regions. Cytoplasmic Filamin A migrated more quickly than the intact protein during gel electrophoresis, suggesting it primarily was a fragment. Quantitative comparison of tryptic peptides using tandem mass spectrometry tags showed that MG132 protected Filamin A from cleavage to the smaller 225 kDa fragment. Identification of Filamin A peptides present only after MG132 treatment and identification of the new amino terminal peptide after chymotryptic digestion by multiple reaction monitoring showed cleavage of the Filamin A protein occurred between R1745 and S1746. This is in a hinge region that is not compacted into FERM domains. Cleavage at this site is distinct from the previously determined calpain cut site,³³ and shows the protease was tryptic since it occurred after an arginine residue. The proteasome subunits functionally interact with one another⁴⁷ to degrade proteins to amino acids and small peptides, but the proteasome also processes large proteins to functional fragments, as shown by NF-κB p105 proteolysis to the p50 transcription factor.⁴⁸ We now identify the longer unstructured Filamin A loop as another site of limited proteasome proteolysis, but our data is associative so we cannot ascribe this proteolysis as causal in MG132 inhibition of platelet adhesion or spreading.

Platelets pretreated with MG132 or bortezomib aggregated less in response to low concentrations of thrombin. Adhesion, spreading, microparticle shedding, and the ability to generate the force to retract formed clots was also reduced by proteasome inhibition. Cytoskeletal interactions aid microparticle shedding from stimulated platelets, and proteolysis by activated calpain^{49, 50} and caspase-3⁵¹ promote microparticle shedding through cytoskeletal proteolysis.⁵⁰ The reduction of microparticle shedding is additionally revealing because MG132 suppressed shedding after stimulation by LPS activation of TLR4, thrombin activation of PAR1, and ADP activation of P2Y12. So, MG132 interference with stimulated platelet function is not restricted to GPIb-IX-V activation by thrombin.

We establish that platelets, like nucleated cells, express a functional ubiquitin/proteasome system that enables them to ubiquitinate their proteome. This decoration increases upon stimulation, and modulates an array of responses from several receptors that all engage the cytoskeleton. The cytoskeletal proteins Filamin A and Talin-1 are targets of ubiquitination, and proteasome mediated proteolysis. The ubiquitin/proteasome system of platelets affects their response to thrombotic stimuli, and proteasome inhibition effectively delayed arterial thrombosis. Bortezomib is already in clinical practice for multiple myeloma and mantle cell lymphoma so elucidation of its anti-platelet activity defines a new salutary function of these inhibitors.

Significance.

Proteasome inhibitors are widely used to treat hematologic cancers, yet little is known of their impact on platelet function and hemostasis. We find platelets have a complete ubiquitin/proteasomal system that participates in cytoskeletal-dependent platelet processes. Proteasome inhibition inhibited platelet aggregation to low dose thrombin, and suppressed occlusive thrombus formation in FeCl₃-damaged carotid arteries. Therapeutic proteasome inhibition may reduce coagulation and augment anti-platelet therapy.

Abbreviations

vWF

von Willebrand factor

TMT

Tandem Mass spectrometry Tags

TIRF

Total Internal Reflection Microscopy