Targeting Integrin and Integrin Signaling in Treating Thrombosis

Abstract

The critical roles of integrins in thrombosis have enabled the successful development and clinical use of the first generation of integrin antagonists as represented by abciximab (Reopro), eptifibatide (Integrilin), and tirofiban (Aggrastat). These integrin α_IIb β₃ antagonists are potent anti-thrombotics, but also have significant side effects. In particular, their induction of ligand-induced integrin conformational changes is associated with thrombocytopenia. Increased bleeding risk prevents integrin antagonists from being used at higher doses and in patients at risk for bleeding. To address the ligand-induced conformational changes caused by current integrin antagonists, compounds that minimally induce conformational changes in integrin α_IIb β₃ have been developed. Recent studies on the mechanisms of integrin signaling suggest that selectively targeting integrin outside-in signaling mechanisms allows for potent inhibition of thrombosis while maintaining hemostasis in animal models.

Introduction

Integrins, a family of cell adhesion receptors, play important roles in cell adhesion, spreading, retraction, migration, anchorage-dependent survival and proliferation. Integrins exist as an α:β heterodimeric complex of transmembrane proteins. In blood platelets, the most abundant integrin is integrin α_IIb β₃. Integrin α_IIb β₃ binds to fibrinogen through the HHLGGAKQAGV sequence in the C-terminus of the fibrinogen γ chain and RGD sequences in the α chain. RGD-like sequences are also present in several other integrin-binding adhesive proteins including vitronectin, fibronectin and von Willebrand factor. In addition, platelets express integrins α_V β₃, α₂ β₁, α₆ β₁, and α₅ β₁, among which α₅ β₁ and α_V β₃ also recognize the RGD sequence. Integrin α₂ β₁ and α₆ β₁ bind to collagen and laminin¹. By binding to adhesive proteins, the integrins mediate platelet adhesion to injured vascular wall and platelet aggregation, which is important for the maintenance of hemostasis, preventing excessive bleeding. The importance of integrin α_IIb β₃ in hemostasis is exemplified in patients suffering from Glanzmann’s thrombasthenia, in which genetic deficiencies in integrin α_IIb β₃ causes bleeding diathesis². Integrin α_IIb β₃ is critical for arterial thrombosis³, which is evident by the protective effects seen in experimental models of thrombosis using either pharmacologic inhibition or genetic deletion/mutation of integrin α_IIb β₃^{4, 5}; and by the clinical efficacy of α_IIb β₃ antagonists^6–8. However, despite successful clinical use of integrin antagonists as potent anti-thrombotics, their use is primarily limited to patients undergoing percutaneous coronary intervention, mainly due to significant bleeding risk. In fact, increased bleeding risks are a major problem shared by all currently available anti-thrombotic drugs. In this review, we briefly discuss the major problems associated with the currently used integrin antagonists, and new advances in developing the next generation of integrin antagonists.

Current α_IIb β₃ Integrin Antagonists

The three current FDA-approved platelet integrin antagonists are designed to block the ligand binding function of integrin α_IIb β₃. Among these drugs, abciximab (Reopro) is a ~48 kilodalton mouse/human chimeric antibody fragment that binds to an epitope near the ligand binding site of β₃^{4, 9–12}; eptifibatide (Integrilin), is a 832 dalton synthetic disulfide-linked cyclic heptapeptide ligand-mimetic, containing an integrin binding sequence, KGD, based on a snake venom peptide, barbourin^{9, 12–14}; tirofiban (Aggrastat) is a 495 dalton synthetic compound, engineered to mimic RGD sequence^{9, 12, 14–16}. Both eptifibatide and tirofiban are integrin ligand mimetics, which interact with the ligand-binding site of integrin α_IIb β₃¹². Tirofiban appears to be specific for α_IIb β₃. Eptifibatide inhibits α_IIb β₃ and α_V β₃, and abciximab inhibits α_IIb β₃, α_V β₃ and α_M β₂^{12, 17, 18}. All three integrin antagonists are administered intravenously. Orally active integrin antagonists were also developed. However, clinical trials of oral integrin antagonists suggested increased mortality instead of beneficial effects^{19, 20}.

The current integrin antagonists have each demonstrated clear therapeutic benefits in high-risk patients undergoing percutaneous coronary intervention (PCI), as indicated by significant reductions in death and reoccurrence of myocardial infarction^{6, 7, 9, 14}. There have also been clinical trials studying the effect of integrin antagonist treatment on patients suffering from acute ischemic stroke. Although, these trials so far have been mainly designed for the purpose of determining safety, and thus the therapeutic efficacy in stroke patients is yet to be conclusively established. In these trials, α_IIb β₃ antagonist treatment alone showed no beneficial impact on mortality or debilitating stroke-related outcomes^{21, 22}, but increased the incidence of symptomatic or fatal intracranial hemorrhage^{21, 23}, with the exception of a trial of tirofiban²⁴. In the tirofiban trial, no significant difference in hemorrhage was found between placebo and tirofiban groups, although the placebo group had significantly more patients also treated with aspirin, which may influence the outcome. Some clinical trials tested a combination of fibrinolytic therapy, using recombinant tissue plasminogen activator (r-tPA), and integrin antagonists, and suggested that integrin α_IIb β₃ antagonists may have a beneficial effect by reducing adverse outcome due to stroke^24–26; although, there is increased risk of hemorrhage, especially with abciximab²⁵. In other clinical trials, fibrinolytic therapy, a reduced dose of r-tPA (<0.6 mcg/kg), together with eptifibatide-treatment shows similar bleeding profiles as the normal dose of r-tPA (0.9 mcg/kg) alone^26–28. Treatment of patients with reduced r-tPA doses in combination with an integrin antagonist implicate the investigators’ consideration of potential hemorrhagic risk of the combination therapy.

The benefit of current integrin antagonists over other anti-platelet agents for general antithrombotic therapy is their rapid onset of action, potency, and low inter-patient variability^{7, 9, 14}. By contrast, there is significant interpatient variability in response to aspirin (irreversible COX-1 inhibitor) or clopidogrel (P2Y12 inhibitor), mainly due to drug resistance^{7, 12, 16}. However, the potent effects of current integrin antagonists are associated with increased bleeding risk^29–31, which can be potentially life-threatening. Bleeding risk limits the use and dose of integrin antagonists, and thus also limits their effectiveness^{11, 31}. Abciximab, tirofiban and eptifibatide all cause thrombocytopenia, which may be associated with conformational changes of integrins following the binding of these drugs³².

New Inhibitors That Minimally Induce Conformational Changes

Integrin structure and conformations

Both α and β chains of the α_IIb β₃ complex contain a long extracellular region, a single-pass transmembrane region, and a short cytoplasmic tail. The amino terminal region of the α and β chains interact to form what is known as the ‘head’, which contains the ligand binding pocket where a conserved structural motif, known as the metal ion-dependent adhesion site (MIDAS) is critical³³. In α_IIb β₃ the MIDAS is on the β₃ -subunit and thought to stabilize ligand-binding by coordinating a metal ion with the aspartic acid on RGD-containing ligands³⁴. Some other integrin α-subunits contain an additional domain called the interactive domain (I-domain), which also contain a MIDAS¹². Below the head region are two long ‘legs’: in α_IIb, two calf domains and a thigh domain constitute a leg; whereas in β₃, four integrin epidermal growth factor-like domains, two hybrid domains, and a plexin semaphorin integrin domain form the other leg.

Integrin molecules undergo conformational changes upon receptor activation and ligand-binding^34–36. Integrin α_IIb β₃ is kept in a resting (low-affinity) state in normal circulation, preventing undesirable thrombus formation. This state is maintained by interactions between the α and β chains within the transmembrane and membrane proximal cytoplasmic domains which constrain the ectodomain³⁷. The resting state has been suggested to correspond to the ‘bent’ conformation as revealed in crystal structure and electron microscopy (EM)^{33, 38, 39,40}. Integrin activation induces the separation of α and β transmembrane and cytoplasmic domains and “un-bending” of the ectodomain, resulting in an ‘extended’ active conformation^41,42. The extended conformation with a ‘closed’ configuration, wherein the β₃ head and hybrid domain form an acute angle, represents an active intermediate affinity state, which is recognized by the RGD sequence or HHLGGAKQAGV sequence in fibrinogen. Binding of ligand recognition sequences induces further conformational changes, resulting in an 'open' head piece conformation, which is the high affinity state^{35, 38, 39, 41–44} (Fig. 1A–C). Between these major conformational states, six different intermediate states have also been suggested, based on crystal structures of the ectodomain of α_IIb β₃⁴³. EM studies using intact α_IIb β₃ in a nanodisc suggest different pictures of ‘bent’ and ‘extended’ conformations^40,45. Different from models obtained from crystal structures of integrin ectodomains, electron microscopy analyses of intact α_IIb β₃ show that the resting integrin headpiece points away from the membrane and that the intermediate extended integrin conformation contains crossed legs^{40, 45}. The differences in models of resting and activated integrin structure are possibly due to the contribution of the transmembrane/cytoplasmic domains to integrin conformation³⁹. The ligand-induced conformational changes are physiologically important because they: (1) expose new epitopes and binding sites on integrins (ligand-induced binding sites, LIBS)^{36, 46}; (2) enable the initial interaction of ‘resting’ integrins with the exposed RGD-like sequence in certain ligands (such as immobilized fibrinogen) to transform integrins into a high affinity form (ligand-induced integrin activation), bypassing the need for inside-out signaling⁴⁷; and (3) are important for integrin clustering and “outside-in” signaling^{44, 48}.

Figure 1. Conformational states during integrin activation and ligand binding.

(A) The resting (low-affinity) state is maintained by interactions between the α and β chains within the transmembrane and membrane proximal cytoplasmic domains which constrain the ectodomain. (B) The activated (intermediate affinity) state has an extended conformation with a ’closed’ configuration. (C) Binding of ligand recognition sequences induces further conformational changes, resulting in an ’open’ high affinity state. TM, transmembrane domain. Curved black arrow indicates the swing-out motion of the β3 subunit hybrid domain upon full integrin activation.

Integrin antagonists that minimally affect integrin conformation

Tirofiban and eptifibatide are RGD mimetics, and thus cause “ligand-induced conformational changes”^{49, 50}, resulting in exposure of LIBS and ligand-induced integrin activation⁵¹, although these monomeric RGD-like peptides or compounds in general do not appear to directly induce integrin outside-in signaling^{47, 49, 52}. The conformational changes induced by ligand mimetic antagonists are thought to be important for the adverse effect of thrombocytopenia³². Abciximab also induces LIBS and thrombocytopenia⁵³⁵⁴. The ability of these antagonists to induce an active conformation of integrins carries the risk of possible pro-thrombotic effects after antagonist dissociation^50,20. There were some reports of such antagonist-induced prothrombotic effects^{50, 52, 55}. Recently, new small molecule integrin antagonists have been developed that exhibit increased specificity and potency without exposing β₃ LIBS epitopes^{56, 57}. RUC-1 and its more potent derivative RUC-2, inhibit the ligand binding function of integrins, platelet aggregation and in vivo thrombus formation, and importantly they do not induce integrin activation^{56, 57}. RUC-1 interacts with α_IIb whereas RUC-2 appears to interact with β₃ Mg²⁺ coordinating sites. Interestingly, unlike RUC-1 and current integrin antagonists, RUC-2 competes with Mg²⁺ for binding to the β₃ -subunit, and its inhibitory effects are attenuated by adding exogenous Mg^{2+ 57}.

Targeting Integrin Signaling

Inside-out signaling

Platelets circulating in blood vessels are normally in a resting state and become activated and adherent only when exposed to the site of vascular injury or platelet agonists. Platelet agonists elicit platelet activation signals via various receptor-mediated intracellular signaling pathways⁵⁸. These intracellular signals converge to transform α_IIb β₃ from a ‘resting’ state to an ‘activated’ state^{1, 58}. This process is called inside-out signaling (Fig. 2). A key requirement for integrin inside-out signaling is the induction of the binding of talin to the membrane proximal half of the β₃ cytoplasmic domain which includes an important NPXY motif. Talin-binding induces unclasping of the transmembrane and cytoplasmic domains of α_IIb and β₃, and thus integrin activation^59–62. This talin-dependent integrin activation is facilitated by kindlin, which interacts with the C-terminal region of the β₃ cytoplasmic domain^{62, 63}. It is conceivable that disruption of talin/kindlin binding to integrin α_IIb β₃ or disruption of the signal responsible for the induction of talin/kindlin binding would also inhibit integrin activation and thus thrombus formation, as evidenced by talin1 gene deletion or mutations^{63, 64}.

Figure 2. Inside-out and Outside-in signaling of the platelet integrin α_IIb β₃.

For a detailed description of individual agonist-induced platelet activation pathways, please see ref⁵⁸. PLC, phospholipase C; PI3K, phosphoinositide 3-kinase; GPCRs, G-protein coupled receptors; IP3, inositol trisphosphate; DAG, diacylglycerol; PKC, protein kinase C; Syk, spleen tyrosine kinase; SFK, src family kinase, CalDAG-GEF1, calcium and diacylglycerol-regulated guanine nucleotide exchange factor 1; RIAM, Rap1-GTP interacting adapter molecule; G13, guanosine nucleotide-binding protein alpha subunit 13; GEF, Guanine nucleotide exchange factor; GAP, GTPase-activating protein; ITAM, immunoreceptor tyrosine-based activation motif; TXA₂, thromboxane A₂.

Inhibitors of inside-out signaling

Current platelet inhibitors including ADP receptor antagonists (e.g. clopidogrel), cyclooxygenase inhibitors (e.g. aspirin), and thrombin receptor inhibitors (e.g. vorapaxar), primarily exert their effects by inhibiting early receptor signaling pathways that initiate inside-out signaling and integrin activation¹⁶. Pharmacological inhibition of inside-out signaling was demonstrated with cell-permeable peptides containing talin binding sequences^{65, 66}. A cell permeable peptide corresponding to α_IIb residues 1000–1008 important in talin binding and β₃ interaction, were also used to inhibit integrin activation⁶⁷. Because inhibition of inside-out signaling results in the loss of the activation of the ligand binding function of integrins, it is expected that the inhibitors of inside-out signaling should show characteristics similar to that of integrin antagonists, which inhibit both thrombosis and hemostasis. Indeed, talin1 deletion or mutational disruption of talin-binding site (β₃ L746A) protected mice from thrombosis, but they still displayed impaired hemostasis as shown by prolonged tail bleeding times in these mice^63,64. However, one report suggests that partial inhibition of talin binding to the integrin β₃ NPXY motif caused defective thrombus formation, with only minor bleeding side effect⁶⁸. It remains to be investigated whether partial inhibition of α_IIb β₃ activation may also result in less potent anti-thrombotic effects or whether finding the right balance between potent anti-thrombotic effects and hemorrhagic side effects may allow anti-thrombotic therapy with proper control of bleeding risk.

Outside-in signaling

Ligand binding to integrins not only mediates platelet adhesion and primary aggregation but also induces signal transduction into cells that triggers the activation of vast intracellular signaling networks and cytoskeleton reorganization^{58, 69}. This process is known as outside-in signaling. Outside-in signaling leads to a series of cellular responses including platelet spreading, stable adhesion, granule secretion and clot retraction, which greatly amplify platelet aggregation and thrombus size^{69, 70} (Fig. 2).

Several protein tyrosine kinases have been shown to be important in outside-in signaling including focal adhesion kinase, ILK, and Syk, Src family kinases (SFK)^{58, 69}. In particular, integrin β₃ -bound c-Src⁷¹, is now recognized as a key early signaling molecule. Following integrin ligation, c-Src has been shown to phosphorylate two NXXY motifs in the β₃ tail^{58, 69}; and induce activation of the phosphoinositide 3-kinase (PI3K) pathway⁷², inhibition of RhoA⁷³, and activation of the Syk-ITAM pathway⁷⁴. Src-dependent transient inhibition of RhoA and activation of PI3K is necessary for platelet spreading on integrin ligands⁷². The PI3K pathway and the Syk-ITAM pathway stimulate granule secretion^{72, 74}. Tyrosine phosphorylation in β₃ may also help recruit phosphotyrosine-binding proteins, such as SHC and myosin heavy chain^{58, 69}. Phosphorylation at Y⁷⁴⁷ also regulates talin binding and thus the direction and dynamics of integrin signaling⁶⁴, and phosphorylation at Y⁷⁵⁹ protects β₃ from calpain cleavage⁷⁵. These events are important for controlling the switch between platelet spreading and retraction⁷³. Interestingly, the role of c-Src requires its interaction with the β₃ cytoplasmic domain. Deletion of the c-Src binding RGT sequence in the C-terminus of β₃ abolished the ability of c-Src to mediate cell spreading even when constitutively active c-Src was expressed⁷⁶. Thus, it appears that targeting c-Src binding to integrins may selectively inhibit integrin outside-in signaling. This notion is supported by a study using β₃ -RGT-deleted integrin-expressing mice, which are defective in platelet responses associated with outside-in signaling, and protected against arterial thrombosis, but display only a mild defect in inside-out signaling⁷⁶.

The most proximal signaling mediator of outside-in signaling identified so far is Gα13. Gα13 directly interacts with an ExE motif in the cytoplasmic domain of integrin β-subunits, and this binding is required for c-Src activation and Src-dependent outside-in signaling⁷⁷. Gα13 binding to β₃ occurs only during early phase outside-in signaling. The Gα13 binds to an ExE motif located near talin binding sites of β₃. However, the ExE motif is not required for talin binding, and Gα13-binding is not involved in integrin activation⁶⁶. Thus, suppression of Gα13 expression or disruption of the Gα13-binding site in β₃ selectively inhibits the early phase of outside-in signaling responsible for stabilization and amplification of a thrombus, but does not affect inside-out signaling nor the ligand binding function of integrins.

Selective inhibitors of outside-in signaling

Recent conceptual advances in integrin outside-in signaling reveal the potential in developing selective inhibitors of integrin outside-in signaling as new anti-thrombotic drugs. A major advantage for targeting outside-in signaling is that inhibition of outside-in signaling should not affect primary platelet adhesion and aggregation, which is critical for hemostasis, but should limit the size of a thrombus to prevent vessel occlusion (Fig. 3). Our laboratory has recently demonstrated the potential of such an approach with a myristoylated ExE motif peptide that selectively inhibits Gα13-integrin interaction⁶⁶. This inhibitor selectively inhibits outside-in signaling, platelet spreading and the second wave of platelet aggregation without affecting primary platelet aggregation. Importantly, this inhibitor potently inhibits occlusive thrombosis in mouse models in vivo without affecting bleeding time, unlike eptifibatide, which dramatically prolongs bleeding time⁶⁶. Thus, selective inhibitors of outside-in signaling as a new antithrombotic strategy have the potential to selectively inhibit arterial thrombosis without causing excessive hemorrhage. Although, it is still important to consider and investigate potential off-target effects caused by selective targeting of Gα13.

Figure 3. A schematic showing the anti-thrombotic effect of selective inhibitors of outside-in signaling without causing hemorrhage.

(Reprinted from Nature, 503: 131-135, 2013).

Since SFK is a required signal downstream of Gα13 in outside-in signaling, inhibitors of SFK could potentially be effective inhibitors. However, SFK play multiple roles in platelets and other cells. For example, SFK is important in the ITAM pathway and GPIb-IX signaling pathways⁵⁸, and thus is important in inside-out signaling, which limits the value of SFK inhibitors as selective outside-in signaling inhibitors. However, blocking the interaction between SFK and integrins may selectively inhibit outside-in signaling. A myristoylated peptide inhibitor derived from the c-Src-binding sequence of β₃, abolished platelet spreading without affecting ADP-induced fibrinogen binding⁷⁸. However, disruption of the β₃ c-Src- binding site in mice provided protection from thrombosis, but also mildly affected hemostasis⁷⁶.

Conclusions

All current integrin antagonists function by blocking the binding of integrin ligands^{9, 12, 34}. These inhibitors induce conformational changes in integrins, which are associated with thrombocytopenia and possibly other adverse effects. New inhibitors with minimal conformational effects may potentially help resolve this issue. A major problem associated with the current integrin antagonists is that at doses where they exhibit high potency they also increase the risk of hemorrhage. Emerging evidence suggests that selective inhibition of outside-in signaling has the potential to have potent antithrombotic effects without causing bleeding.