Aquaporins in platelet function

Abstract

Structurally, aquaporins (AQPs) are small channel proteins with monomers of ~ 30 kDa that are assembled as tetramers to form pores on cell membranes. Aquaporins mediate the conduction of water but at times also small solutes including glycerol across cell membranes and along osmotic gradients. Thirteen isoforms of AQPs have been reported in mammalian cells, and several of these are likely expressed in platelets. Osmotic swelling mediated by AQP1 sustains the calcium entry required for platelet phosphatidylserine exposure and microvesiculation, through calcium permeable stretch-activated or mechanosensitive cation channels. Notably, deletion of AQP1 diminishes platelet procoagulant membrane dynamics in vitro and arterial thrombosis in vivo, independent of platelet granule secretion and without affecting hemostasis. Water entry into platelets promotes procoagulant activity, and AQPs may also be critical for the initiation and progression of venous thrombosis. Platelet AQPs may therefore represent valuable targets for future development of a new class of antithrombotics, namely, anti-procoagulant antithrombotics, that are mechanistically distinct from current antithrombotics. However, the structure of AQPs does not make for easy targeting of these channels, hence they remain elusive drug targets. Nevertheless, thrombosis data in animal models provide compelling reasons to continue the pursuit of AQP-targeted antithrombotics. In this review, we discuss the role of aquaporins in platelet secretion, aggregation and procoagulation, the challenge of drugging AQPs, and the prospects of targeting AQPs for arterial and venous antithrombosis.

KEYWORDS:

• Aquaporin
• Platelets
• Thrombosis
• Procoagulant Membrane Dynamics
• Haemostasis

Introduction

The aquaporins (AQPs) are a family of channel proteins that form pores on biological membranes and mediate the conduction of water but at times also allow small solutes such as glycerol across cell membranes and along osmotic gradients [1-5]. Structurally, AQPs are small proteins with monomers of ~ 30 kDa that are assembled as tetramers to form pores on cell membranes (Figure 1) [1,5]. There are 13 types of AQPs in mammalian cells, and of these it is the aquaglyceroporin subset which also transports small and polar non-electrolytes (glycerol, urea, hydrogen peroxide and carbon-dioxide) [8–10]. Aquaporin-1 (AQP1, Figure 1) is a channel protein and the first member of the AQP family to be identified [2,11]. Aquaporin 1 forms passive membrane channel that, in human and mouse, regulates water and small solute permeation across lipid bilayers and has been shown to regulate cell volume and membrane dynamics in other cell types [12,13]. The distribution of aquaporins in mammalian tissue is broad; their expression has been reported in epithelial and related cell types involved in fluid transport in several organs, but also in adipocytes where there is no obvious role in fluid transport [8]. AQP1 expression has been reported in renal tubules, micro-vessels, choroid plexus, ciliary epithelium, corneal endothelium, pain-processing C-fibers, vascular endothelium, tumor vessels, red blood cells and platelets [8,14,15]. Reviews of data from aquaporin knockout mice and from human with loss-of-function mutations in aquaporins suggest that modulation of water channels may have broad clinical indications, such as in the treatment of intracranial hypertension and edema, epilepsy and brain swelling, tumor angiogenesis and proliferation, obesity, nephrogenic diabetes insipidus, glaucoma and epidermal hydration [5,8,16,17].

Figure 1. Structure of aquaporins. (a) A top view of the extracellular face of an aquaporin 1 (AQP1) homotetramer, with monomers labeled 1–4, based on the X‑ray structure of bovine AQP1 (Protein Data Bank (PDB) code: 1J4N). The tetrameric structure was modeled by the interactive PDBePISA (proteins, interfaces, structures and assemblies) tool. (b) A schematic of AQP membrane topography. Both the amino and carboxyl terminals of AQP1 are located in the cytoplasm [6,7]. (c)Structure of the bovine AQP1 monomeric unit, which shows key helical domains (labeled M1–M8) and connecting linkers (labeled a–e). (d)A view into the extracellular vestibule of bovine AQP1. The constriction region (in green) is made up of aromatic and arginine residues (known as the ar/R constriction; residues Phe58, His182 and Arg197); extracellular Asn‑Pro‑Ala (NPA) residues (Asn194, Pro195 and Ala196) are shown in orange; backbone α‑carbonyl hydrogen‑bond acceptors (Ile192, Cys191, Gly190 and Gly189) are shown in violet; and hydrophobic side chains comprising nonpolar amphipathic surface are shown in black. Figure and description are from Verkman et al. (2014) [8]. License Number 4916061052663

Expression and Subcellular Localization of AQPs in Platelets

Aquaporins have been identified in most living organisms, indicating the participation of this family of proteins in diverse biological processes [5]. Aquaporin-1 is expressed in all vascular endothelial cells outside the brain [18,19], and a few cells types co-express isoforms of aquaporins, for example, human erythrocytes express AQP1 [6,20] and AQP3 [21,22]. Data about the expression of aquaporins in platelets are sparse, and the authenticity of available data is limited by the lack of molecular evidence [23,24]. Genome-wide RNA-seq analysis indicates expression of AQP1, 3, 4, 9 and 11 RNA in human platelets [25]. Using quantitative real-time PCR analysis, Wright et al [26]. detected AQP10, encoding an aqua-glyceroporin, and low levels of AQP1 in platelet lysates [26]. Misztal et al [27]. used polyclonal antibodies in immunodetection to suggest the presence of AQP isoforms 0–7 and 9–12 in human platelet plasma membranes and in the dense and alpha granules [27]. Notably, the authors localized AQP11 and AQP12 to the platelet secretory granules [27]. Additionally, Goubau et al [28]. used immunostaining to show colocalization of AQP7 with CD63, a marker for platelet dense granules [28]. Using platelets isolated from AQP1-null mice, Agbani et al [14]. validated the specificities of commercial antibodies and confirmed the expression of AQP1 in human and mouse platelets by immunocytochemistry and western blot experiments, and established the subcellular localization of AQP1 to internal membrane systems likely aligning to the open canalicular system [14]. The authors failed to localize significant AQP1 expression to particular platelet granule populations and did not observe significant AQP1 translocation after platelet stimulation, suggesting that the gene is predominantly expressed on internal membranes in human and mouse platelets [14]. A similar approach, of genetic deletion of aquaporin isoforms, is now needed to establish the expression and localization of other aquaporin isoforms in platelets.

Regulation of Aquaporins

It seems likely that isoforms of aquaporins are localized to both the platelet plasma membrane and platelet granule membranes [14,27,28]; but how are platelet aquaporins regulated in these locations? In other cell types, a major regulatory mechanism for aquaporins is translocation from an intracellular location to the plasma membrane. For example, AQP2 is shuttled from sub-apical vesicles in the kidney to the apical plasma membrane upon activation of the collecting duct cells with vasopressin [29]. Similarly, AQP1 has been shown to interact with moesin, which provides coordinated linkage of the cytoskeleton with the plasma membrane, and AQP1 also interacts with the lysosomal trafficking regulator–interacting protein, LIP [30]. While there is transcript evidence for the expression of these proteins in platelets [25], data on their interactions with platelet aquaporins is lacking. Similarly, it is not known whether AQPs interact with known regulators of granule secretion, including Munc13-4, VAMP8, Stx11, PKCα, PKCδ and Rab27a/b. Also, it is not clear whether phosphorylation events regulate aquaporins activity in platelets. In other cell types, protein kinase C (PKC) promotes AQP1 translocation, regulating cell volume and dynamic membrane changes [15]. There is also evidence of PKA-dependent phosphorylation of AQP1 [31], but these or similar events are yet to be demonstrated in platelets. Therefore, the mechanisms by which platelets mobilize aquaporins for secretion and procoagulation functions remain unclear and research in this area is needed.

Role of Aquaporins in Platelet Secretion and Aggregation

Vesicular swelling events that mediate content expulsion in several secretory cells is regulated by aquaporins [32,33]. However, such a role in platelets remains to be established. Goubau et al [28]., reported on the relevance of G264V AQP7 mutation to platelet dense granule secretion. The prevalence of G264V AQP7 mutation has been reported to be 3.75 and 8% in Japanese and Caucasian populations, respectively [34,35]. Hematologic analysis of G264V AQP7 mutation showed normal platelet count, but a mildly increased mean platelet volume relative to controls; AQP7 G264V platelets were enlarged, more rounded with centrally localized granules and a pronounced open canalicular system. It remain unclear how these changes may affect features of platelet procoagulant membrane dynamics, such as ballooning. A subclinical ATP secretion defect was also reported in response to epinephrine stimulation of AQP7 G264V platelets, and the aggregation defect was similar to that in patients with storage pool disease [28,36]. Notably, the mutation of AQP7 G264V did not precipitate any clinical bleeding problems. Moreover, the authors reported normal aggregation and secretion in response to ristocetin, arachidonic acid, U46619, ADP, and Horm collagen [28]. It is therefore unlikely that AQP7 plays a major role in platelet granular secretion or aggregation. Agbani et al [14]. likewise investigated the possibility that AQP1 might be a major regulator of platelet secretion and aggregation. Constitutive deletion of AQP1 did not affect hematological parameters nor the expression of surface glycoproteins, or the number and distribution of alpha and dense granules. Also, dense and alpha granule secretion, integrin α_IIbβ₃ activation (inside-out signaling), platelet lamellipodial and filopodial spreading over fibrinogen (outside-in signaling), as well as platelet aggregation remained normal after collagen stimulation [14]. Clearly, AQP1 is not a major regulator of these responses.

There is some published pharmacology-based evidence for the involvement of aquaporins in human platelet secretion, which reports significant reduction in secretion from dense and alpha granules, as well as lysosomes, in a collagen dose-dependent manner [27]. The role of specific AQP isoforms in these secretory responses cannot be deciphered from these studies. Importantly, these observations are not consistent with the molecular evidence for AQPs role in platelet secretion, and one explanation for this may be in the dilemma of finding a selective inhibitor of platelet AQPs. Studies now show that heavy metal and sulfhydryl-reactive small molecules, previously reported to inhibit AQPs, have no effects on the water permeability of these channels [37]. Furthermore, these compounds lack selectivity and specificity and have cytotoxic effects [8,28,38–40]. Therefore, convincing evidence of the role of AQP isoforms in platelet secretion and aggregation responses will require the use of genetic modifications or knockout models.

Role of Aquaporins in Platelet Procoagulation

Given the highly complex nature of the platelet procoagulation process and its contribution to thrombosis, separate AQP subtypes may drive distinct platelet mechanisms, and the differential contributions of AQPs to procoagulation is well orchestrated. It is also possible that different AQP subtypes are simultaneously involved in a given procoagulation mechanism. A series of studies by Agbani et al. [14,41–45] revealed the spatiotemporal dynamics and drivers of the morphological transformation that platelets undergo during thrombosis and hemostasis. The mechanism of platelet procoagulant remodeling was shown to involve a coordinated fluid entry system of Na⁺, Cl⁻ and water influx, together with a regulated disruption of the platelet microtubule network [45]. Also, elements of platelet procoagulant membrane dynamics (PMD), such as membrane ballooning, phosphatidylserine (PS) externalization and microvesiculation, can independently and in concert, amplify the clotting process [41,43–46]. Furthermore, platelet responses to exposed sub-endothelial collagen have been shown to lead to the formation of distinct phenotypes which play different roles in hemostasis and thrombosis [45,47,48]. Using 4-dimensional high-resolution and dynamic imaging approaches, subpopulations of human platelets were shown, after collagen stimulation, to transform into a highly procoagulant phenotype characterized by PS-exposing ballooned membranes and extensive microvesiculate contact surfaces, in a process described as procoagulant-spreading [45]. At least one isoform of AQPs (AQP1) has been shown to mediate three pivotal elements of platelet PMD, namely PS exposure, procoagulant-spreading and microvesiculation [14]. Platelet exposure to acetazolamide, which has been shown to block AQP1 [38,49] and AQP4 [50,51], leads to a markedly diminished procoagulant response in terms of ballooning, procoagulant-spread and in vivo arterial thrombosis [41,45]. However, as with several putative small-molecule inhibitors of AQPs [8,27,37],^38,39 [40,52], there remains the question of whether acetazolamide is selective for AQP1, in addition to the conundrum of delineating which of the AQP subtypes mediates its antithrombotic response. As with secretion and aggregation, the most convincing evidence for the role of AQPs in platelet procoagulation has come from AQP knockout mice and from humans with loss-of-function mutations in AQPs.

The effect of homozygous AQP7 G264V mutation on platelet procoagulation and thrombin generation is not known. However, AQP1 appears to play a major role in the fluid entry mechanism and remodeling required for the platelet procoagulant function [41,43–45]. Using AQP1 knockout mice generated by targeted gene disruption [53], Agbani et al [14]. showed that loss of AQP1 did not affect normal hemostatic response or procoagulant membrane ballooning [14], but resulted in a marked reduction in platelet procoagulant-spreading, microvesiculation and in vivo thrombus formation [14]. The role of AQP1 in these events was dependent on the spatiotemporal dynamics of calcium entry required for the conversion of quiescent platelets into procoagulant phenotype [45,54,55]. Compared to red blood cells, a broader expression of AQPs in platelets is possible [25,56]; however, the significant phenotypic changes associated with ablation of AQP1 alone would suggest that while functional redundancy in AQPs is plausible for the formation of ballooned platelet membrane [14], such mechanism is unlikely in the regulation platelet procoagulant-spreading and microvesiculation after collagen stimulation.

AQP1 mediated procoagulant membrane dynamics in collagen-stimulated platelets by enabling the sustained rise in cytosolic calcium required for phosphatidylserine exposure and microvesiculation, through its facilitation of calcium entry via mechanosensitive cation (MSC) channels TRPC6 and Piezo1 (Figure 2). Previous studies in mammalian cells have shown that membrane stretching due to osmotic swelling can induce transient calcium entry [65] through calcium permeable stretch-activated or MSC channels, several of which are expressed in platelets [62,66–68]. MSC channels can bind to and functionally interact with several isoforms of AQPs [69–71], and a similar coordination between AQP1 and MSC channels is pivotal for the platelet procoagulant response to collagen. Likely, AQP1 and MSC channels are proximally located in platelets as indicated by colocalization and fluorescence resonance energy transfer activity [14]; and this further supports the possibility that these structurally dissimilar channels form a functional symbiotic unit that mediates volume regulation and swelling-induced calcium entry [14]. A similar symbiotic mechanism underlies the functional significance of AQP4–TRPV4 interactions in the Xenopus laevis expression system [70]. Additional studies are needed to understand how a rise in calcium, induced by AQP1-regulated swelling, may be translated to regulate the procoagulant response. AQP1 was shown to directly interact with transmembrane protein 40 (TMEM40) which may act as calcium-dependent lipid scramblase or calcium-activated chloride channel, either of which will regulate procoagulant responses in a calcium-dependent manner [72,73].

Figure 2. Water entry via aquaporin-1 mediates a stretch-induced amplification of calcium entry and the full procoagulant response of collagen-activated adherent platelets. Upon contact with sub-endothelial collagen, subpopulation of quiescent platelets (a) signal via glycoprotein receptor VI and integrin α2β1 to cause a rise in cytosolic calcium and the formation of non-procoaguant, conventional spread, non-balloon platelets (CSNB; C) [45,57]. Other subpopulations may undergo the add-on activation of nonspecific cation channels as well as Ca²⁺ activated chloride channels (CACC), resulting in an initial salt entry which is then followed by the influx of water along its concentration gradient. AQP1 facilitate rapid cell swelling leading to stretching of the plasma membrane which activates the opening of mechanosensitive cation (MSC) channels. Of these, TRPC6 of the family of transient receptor potential cation (TRPC) channels [58–61] and Piezo1 [62,63] channels are likely candidates, which open to allow additional influx of extracellular calcium, sustaining the rise in cytosolic calcium (b–d) which is critical for activation of the lipid ‘scramblase’ leading to PS externalization [41] (d,e, d–h). The internal hydrostatic pressure initiates membrane swelling at regions of high calpain activity [16,41,64]. In combination with external osmotic pressure, this leads to full-scale irreversible membrane ballooning (d–g, d,e). Ballooning is temporally correlated to the formation of the expansive procoagulant surface, which subsequently breaks up by multiple coalescence events to form PS^+ve microvesicles [43,44,45] (f–h). Both ballooned non-spread (BNS; E) and ballooned procoagulant-spread platelets (BAPS; H) increase the procoagulant area mediating the acceleration of coagulation at wound sites [43–45]. Data from this study indicate AQP1 has a specific role in the membrane swelling events that control procoagulant-spreading, but not ballooning. Upon ablation of AQP1, platelet swelling kinetics are slowed and unable or only weakly activate MSC channels for enhanced Ca²⁺ entry. This results in limited Ca²⁺ influx and procoagulant-spreading, thus favoring the formation of the BNS platelet phenotype (d,e) and limited thrombosis (Figure and Legend are unedited and from Agbani et al. 2018 [14]; https://creativecommons.org/licenses/by/4.0/)

The Challenge of Drugging Aquaporins

Identifying a selective and clinically effective modulator of aquaporins function in platelets is challenging. Assays commonly used to estimate aquaporins activity, such as oocyte swelling rates and cytoplasmic calcein fluorescence signal, have now been shown to be affected by non-aquaporin factors [8,74]. The structure of AQPs does not make for easy targeting of these channels either. Aquaporins are simple passive pores, but the small size and narrow pore diameter of the functional aquaporin monomer not only restricts access of small molecules but also size-limits the options for drug targeting. Additionally, since aquaporins do not allow the passage of charged molecules, electrostatic interactions of potential inhibitors are limited to those between hydrogen bond donors and acceptors [8]. Furthermore, the lack of complex gating or transport mechanisms in aquaporins also limits the pharmacology options for potential small molecule inhibitors. Remarkably, water permeation through aquaporins channels has remained largely unaffected by mutations in the extracellular and cytoplasmic domains of aquaporins; hence it has been suggested that the binding of an inhibitor will need to occur deep in the narrow aquaporin pore to effectively stop water transport through these channels (Figure 1) [8]. Alternatively, the development of allosteric modulation of the channel, at sites distant from the channel, may be another possible approach. Taken together, to identify a small molecule inhibitor that meets these requirements is demanding. Lastly, given that there are 13 homologous human aquaporin isoforms, several of which have broad tissue expression and play vital roles in maintaining hemostasis, the blockade of aquaporins by the currently available pharmacological agents which lack specificity and selectivity is not a rational approach to modulate platelet function. So, to date, platelet aquaporins remain valuable but elusive antithrombotic drug targets [8]. These notwithstanding, data from in vitro platelet procoagulant responses and in vivo thrombosis in mice provide compelling reasons to continue the pursuit of platelet aquaporin targeted antithrombotics [14]. Approaches to identify an authentic aquaporin inhibitor should utilize the many high-resolution X-ray crystal structures of aquaporins as the basis for computationally-based screening and for the large-scale functional screenings of drugs in current clinical use as well as of new compounds or lead molecules of diverse chemistry.

Targeting Aquaporins for Arterial and Venous Antithrombosis

Data from clinical use of antithrombotics show that over 25% of patients on antiplatelet drugs will experience treatment failure and suffer an ischemic event [75]. Attempts to improve clinical effectiveness by combining antiplatelet drugs, for example P2Y₁₂ blockers and thromboxanes inhibitor in single treatment regimens have precipitated increased bleeding as a side effect [76–78]. Clearly, this demonstrates the need for new drugs that are antithrombotics via mechanisms distinct from agents presently used in the clinics. Based on the molecular evidence that has established a role for AQP1 in the platelet procoagulant response and in vivo arterial thrombosis [14], targeting platelet AQP1 holds promise for the discovery of this much needed new class of antithrombotic, since AQP1 deletion essentially spares platelet secretory functions and does not perturb hemostasis [14]. Although the role for platelets in arterial thrombosis is better established, it is clear that they are present in the venous thrombus and play an important role [79–81]. Recent data suggest that platelets also play a critical role in the early development of venous thrombosis, providing a localized procoagulant site adjacent to the endothelium, anchored by the vWF secreted from the endothelium [79–81]. Because water entry into platelets promotes procoagulant activity [14], it is possible that platelet aquaporins are critical for the initiation and progression of venous thrombosis. Aquaporins may therefore represent attractive targets for future development of a new class of antithrombotics in the treatment of both arterial and venous thrombosis.

Conflict-of-interest disclosure

The authors declare no competing financial interests.

Contribution

E.O.A. and A.W.P. wrote the review.

Funding and Acknowledgement

British Heart Foundation grant BHF: PG/16/102/32647Canadian Hemophilia Society (CHS) 26/03/2019