Role of platelets in neuroinflammation: a wide-angle perspective

Platelets have long been implicated, or at least been suspected, in the etiology of a variety of neuropathologies, most obviously including ischemic stroke but also others such as multiple sclerosis (MS). In recent decades, a series of discoveries have been made which place those conjectures on a sound rational footing. Broadly speaking, the essence of these findings is that platelets possess an unexpectedly large variety of receptors and secretory products, additional to those serving their classical role in hemostasis and thrombosis, which are active in inflammation, immunity, and tissue repair. This versatility is remarkable in view of their very small size and lack of cell nuclei. Indeed, in the early days they were regarded as nothing more than cellular debris. These recent advances, together with the fact that platelets are often the first cells to arrive at sites of vascular injury, suggest the hypothesis that they may be central players in neurodegenerative diseases.

As the title states, this review provides a wide-angle perspective on platelets as mediators of inflammation and immunity, with emphasis on neurological implications. Therefore, it is not possible to treat each topic in the depth it deserves. Most of the topics are large and specialized fields in themselves with their own wealth of literature. However, the references supplied will lead the interested reader to more comprehensive accounts. Some good reviews of platelets in inflammation are available [1] but the present review is more wide-ranging and exhibits the relevance to neurology specifically at every opportunity.

Platelets, properly termed thrombocytes, were traditionally considered to function exclusively in hemostasis and thrombosis, a role for which they are superbly adapted. Platelets are produced as fragments of megakaryocytes and, according to the convincing arguments of Martin [2], this fragmentation occurs during passage through the lungs. Like erythrocytes, they lack nuclei but unlike erythrocytes they do possess mitochondria. They are about 1/4 the diameter of erythrocytes and about 1/24 as numerous, but they preferentially circulate along the vessel wall [3], positioning them to respond immediately to vascular injury.

Equally important is the ability of activated platelets, and microparticles from them which are released concomitantly, to catalyze the coagulation cascade. This is accomplished by an activation-dependent membrane inversion or "flip-flop" by which normally in-facing phospholipids (PL) become exposed to plasma [6]. These PLs are mainly anionic, such as phosphatidylserine (PS), and render the membrane procoagulant by promoting assembly of the vitamin K-dependent coagulation factors into their active complexes (prothrombinase and factor X-ase), to generate the thrombin that converts fibrinogen to the fibrin plug. The role of vitamin K in this process is to serve as a cofactor for the post-translational addition of an extra carboxy group to glutamic acid residues (Gla domains), enabling them to bind calcium, much as citrate does, thereby anchoring the coagulation factors to suitable PL membrane surfaces (such as PS) via calcium bridging.

Since hemostasis and thrombosis is not the focus of this article, readers wishing more depth are referred to standard textbooks of hematology [7–10] or more specialized books [11].

It had become clear by 1995 that platelets also play major roles in inflammation and immunity [12–15]. This concept was amply confirmed and extended by subsequent findings [16], beginning with the discovery around the same time of the first chemokine receptor on platelets (referenced below). The following paragraphs briefly review the major classes of platelet-derived factors which are active in inflammatory states, with emphasis on implications for neurological disorders.

The matrix metalloproteinases (MMPs) are a group of about 40 homologous proteases which are secreted to the extra-cellular matrix (ECM) in an inactive form. As a group these enzymes depend on a metal ion for activity, generally zinc. A subgroup of these enzymes, the membrane-type (MT-MMP), remains at the plasma membrane and can activate the secreted forms of the enzyme. Other closely related enzymes are those of the ADAM's (A Distintegrin And Metalloprotease) and the ADAMTS's (ADAM with ThromboSpondin domain) [17, 18]. Many are also known by their earlier common names. They have varying degrees of substrate specificity towards many proteins in the vicinity of the ECM including collagen, fibronectin, gelatin, laminin and other stromal components ("stromelysins"). Accordingly, their activities must be finely controlled and regulated by the aforementioned activators, specific inhibitors known as TIMP's (Tissue Inhibitor of MMP), and by the circulating plasma protein, alpha-2-macroglobulin. MMPs are important in many aspects of human development and tissue remodeling, and play significant roles in pathogenesis of neuroinflammatory disorders.

MMPs have been generally recognized as major participants in disruption of the blood-brain barrier (BBB) in MS [19] and there is persuasive evidence that they play a direct causal role [20]. This evidence consists of close association of rising levels, particularly of MMP-9, prior to onset of exacerbations in humans and in the animal model of MS known as experimental autoimmune encephalomyelitis (EAE) [21–23]. Relative reduction of TIMP's was also seen. In EAE model, disease severity is sharply attenuated by inhibition of MMP or by gene knockout. Similar results were observed with an alternative model of MS, demyelinating canine distemper virus infection [24]. In human immunodeficiency virus (HIV) infection, MMP-9 in particular was implicated in disruption of the BBB [25]. MMPs have also been implicated in post-ischemic brain injury [26]. MMP-9 may be particularly noxious [27] but MMP-2 is also discussed as a biomarker [28], as are others for the case of MS [20]. The mechanism usually proposed is that infiltrating leukocytes employ MMPs to disintegrate the basement membranes of cerebral endothelial cells to enter the CNS.

Accordingly, inhibitors of MMPs are under active investigation for treatment of MS and some other neuropathologies [29]. At least one action of tetracycline derivatives such as minocycline and doxycycline is inhibition of MMPs [30–32]. Indeed, interferon-beta1a (IFN-beta) is reported to have such action [33, 34]. On the other hand, several groups caution against prolonged and non-specific inhibition of MMPs for therapy because of their equally prominent role in repair, recovery and normal CNS health [35–37].

How might platelets be placed in these events? First, platelets have been shown to express MMPs -1, -2, -3, -9; ADAM-10, -17; all four TIMP's except TIMP-3; MT1-MMP (also known as MMP-14); and ADAMTS-13 [38, 39]. The cited review [38] is up-to-date but limits discussion to the role of MMPs in platelet function and does not explore the likely effects of MMPs secreted during platelet activation on bystander cells, analogous to the manner in which neutrophils or glial cells can injure neurons in the vicinity [40]. Second, platelets are often the first responders to sites of injury or inflammation. Third, it is known that platelets can interact with leukocytes to form circulating complexes, first demonstrated by Rinder et al [41, 42], extended by our laboratory to platelet microparticles (PMP) [43], and now widely employed as an assay of activated or inflammatory states, e.g. [44–47]. It has been shown that formation of platelet-leukocyte complexes is associated with the expression and activation of MMPs -1, -2, -3 and -9 [48]. It is not clear in that paper if the source of MMP is platelets or leukocytes (it is likely both) but inhibitors of MMP in the media reduced formation of the complexes while active MMP promoted them.

Fourth, is the effect of platelet-derived microparticles (PMP). Although little work of a specifically neurological nature has been done in this area, cancer researchers have demonstrated that PMP strongly promote the invasive potential of prostate and breast cancer cells in a manner dependent on MMP expression [49–51], a process resembling leukocyte infiltration to the CNS. PMP express most of the platelet membrane proteins, therefore likely including MT-1 MMP, which was notable in the study [51]. The work by Jy et al of our laboratory demonstrated that PMP can directly activate and bind to neutrophils [43]; however, MMP activities were not measured. We have recently presented evidence that cell-derived microparticles, especially PMP, may play an important role in neurodegenerative diseases [52, 53]. Although studies of platelet activation in MS are rare in the recent literature, the data of Cananzi et al from patients in remission clearly demonstrate chronic platelet activation in MS [54] (their figure two), confirming earlier reports, which they cite, and which we discuss in the closing section of this paper.

Chemokines are central to inflammation chiefly by signaling leukocyte migration/infiltration and differentiation, but also by other actions [55]. About 50 are known and half as many receptors. They are low molecular weight proteins (7-12 kDa) and are known both by acronyms for their common names and by systematic nomenclature, the latter consisting of two main groups, CC and CXC, and a few exceptions. These symbols refer to the amino acids (aa), Cys-Cys or Cys-X-Cys where X is any other aa. Suffix L for "ligand" or R for "receptor" is applied. The alpha chemokines are the CXCL's and beta chemokines are the CCL's. In general, the CCL's bind only to CCR's, while CXCL's bind only to CXCR's. Beyond that, the receptors are promiscuous to varying degrees, suggesting complex and subtle signaling depending on relative concentrations and affinities of the ligands present.

Meanwhile, recent advances have created more complexity, not simplification. It should be noted that our knowledge of platelet chemokines and receptors for them is based largely on murine studies and tissue culture with limited cell types. For example, human umbilical vein endothelial cells (HUVEC) are widely used because they are easy to obtain and grow, but they often exhibit quite different responses compared to brain micovascular EC, as we have observed in microparticle studies [65] and pointed out for in vitro studies of antiphospholipid antibodies [66]. Endothelial cell (EC) activation is not a simple yes/no effect but a multi-pathway phenomenon having both immediate effects (within seconds of stimulation, such as CD62E cell surface expression) and slower effects (24-48 hr) that depend on genetic upregulation of the pathways activated by, for example, exposure to TNF-α [67, 68].

The cytokines constitute a lengthier list of signaling molecules, notably including the interleukins, and overlaps somewhat with the chemokines. Some helpful summary tables (such as distributed by R&D Systems) include both groups, and a number of members are commonly listed in both families, e.g. interleukin 8 (IL-8) is now CXCL8, and RANTES, once considered a cytokine, is now assigned as CCL5. Originally, a sharp distinction was drawn between factors that attracted leukocytes (chemokines) and factors with other effects (cytokines, e.g. the interleukins) but that distinction is increasingly blurry in view of the pleiotropic actions of so many of these substances.

As the first cells to arrive on the scene of vascular injury, it makes sense that platelets would be involved with tissue repair as well as plugging leaks. Accordingly, some of the most informative reviews of platelet cytokines are oriented to the role of platelets in wound healing [93–95]. One may suspect important roles in neurological tissue repair as well. The 2008 review by Nurden's group [94] includes dozens of factors in addition to the chemokines listed above. Although no attempt will be made to list them here, the platelet "growth factors" include EGF, TGF-β1, -β2, PDGF, HGF, FGFb (FGF-2), and a series of pro- and anti-angiogenic factors, e.g., VEGF-A, -C. Platelets also secrete the antimicrobial peptides known as thrombocidins. Of special interest in neurology is the presence of semaphorin 3A (a.k.a. collapsin-1), another of many "CNS-specific" agents found in platelets. Also listed in [94] are the cytokines TRAIL, LIGHT, SDF-1α, HMGB-1, etc., and the interleukins 1L-8 and IL-6sR in platelets. Their review misses a few, notably IL-1 from platelets [96, 97], which is secreted on platelet microparticles. Further discussion of these platelet-derived agents (which are difficult to neatly classify) is beyond the scope of this review, our purpose being to bring to wider attention the astonishing variety of bioactive platelet-derived agents. The following few examples may not be widely appreciated.

The discovery of toll-like receptors (TLRs) on platelets in 2004 was another completely unexpected development. Other immune functions of platelets had been earlier noted, including generation of killer-like reactive oxygen species (ROS) [13], phagocytic activity, secretable antimicrobials (thrombocidins), interactions with leukocytes and endothelial cells by direct contact or secretory signaling, and as the principal source of CD40L (CD154) [100, 110].

First discovered in the fruit fly, about a dozen TLR's are now known and each is tuned to recognize a distinct class of pathogen-associated molecular patterns (PAMP's). The PAMP's which are recognized include bacterial components such as flagellin, lipopolysaccharide (LPS), lipoproteins, peptidoglycans, certain regions of DNA, and so on. The true TLR's are transmembrane surface proteins, but proteins with similar PAMP-recognition functions for virions occur in the cytoplasm [112, 113]. The membrane TLR's, some of which function as heterodimers (e.g. TLR2 plus TLR1 or TLR6), upon engagement with ligand, pass their signals through a series of accessory transducers to the cell nucleus where antimicrobial genes are activated to yield products such as IL-1 and TNF-α. Since platelets lack nuclei, however, they seemed irrelevant to studies of the tissue distribution of TLR's.

To our knowledge, it was Shiraki et al who discovered the first TLR's in platelets, TLR1 and TLR6, by mRNA, Western blotting, and flow cytometry [114]. Furthermore, their expression was upregulated in response to IFN-γ. The next year, Cognasse et al reported also TLR2, TLR4, and TLR9 on platelets [115], and observed that expression levels increased two-fold upon platelet activation. More recently, the latter authors have followed up with further insights on platelet TLR's, and propose a major and unique role of platelets in bridging innate to adaptive immunity on this basis [116, 117].

These findings add a new dimension to studies of TLR's in neuroinflammatory conditions. For example, the study by Chearwae and Bright of TLR4 and TLR9 in a model of MS documented upregulation of these receptors in T cells after induction of EAE, and favorable reduction along with amelioration of symptoms with prostaglandin and curcumin treatment [118]. However, now that platelets are known to also possess these receptors, it appears that their participation in such processes deserves consideration in future experiments of that kind.

Peroxisome proliferator-activated receptors (PPAR's), of which three are known, are ligand-activated nuclear transcription factors of the hormone receptor superfamily. They are widely disseminated and appear to function chiefly in regulating metabolism, notably of fats, and this fact has elevated the fibrate drugs (agonists of PPAR) to a level of importance second only to the statins for prevention and treatment of coronary artery disease. However, like the statins, they seem to have "pleiotropic" effects, prominently including anti-inflammatory actions now under investigation for treatment of MS [119–121]. Indeed, the role of PPAR's in CNS disorders is of much current interest [122, 123], with promise of applications in Parkinson's and Alzheimer's diseases among others [124, 125]. Yang et al has reported on PPARα regulation of immunity and the EAE model of MS [126]. PPAR's appear to be an important target of the NSAID's [127]. Interestingly, it appears that lysophophatidic acid (LPA) is a natural activator of PPAR [128], and that agonists of endocannabinoid receptors also stimulate PPAR in a model of MS [129]. Importantly for MS, PPAR appears to control inflammation induced by CD4+ T cell infiltration, at least in vitro [130].

Since platelets lack nuclei they are not expected to have PPAR's, but platelets are full of surprises. Both PPARβ/δ and PPARγ were found expressed in platelets [131]. Furthermore, the same authors have demonstrated that PPARγ is released from activated platelets on microparticles (PMP), which is taken up by promonocytic cell line in tissue culture; and that agonists of PPARγ induced platelet release as judged by secretion of sCD40L and thromboxane A₂ (TxA₂) [131]. Although the full implications of these late findings remains to be seen, they are paralleled by other developments in PPAR research, such as the recognition that active PPAR is not necessarily confined to the surface of the nucleus. The therapeutic potential of agonists of PPAR, such as rosiglitazone (Avandia™) and pioglitazone (Actos™) for PPARγ, needs further study but it has been shown that they inhibit platelet release of CD40L [131].

Recognition that certain lipids perform critical signaling functions began in the 1970's with the observation that human semen caused contraction of uterine muscle strips, leading to identification of the prostaglandins, so called for that reason. Related families of active lipids, such as the leukotrienes and thromboxanes, subsequently came to light and are collectively known as the eicosanoids, for the 18-carbon arachidonic acid precursor in the cell membrane. From the outset, platelets were seen as a major source of these agents. Indeed, a common measure of platelet activation today is the circulating level of thromboxane B₂ (TXB₂), the stable breakdown product of thromboxane A₂ (TXA₂, half-life 30 sec's). That early work culminated in understanding the mechanism of aspirin and led to development of the other COX inhibitors and NSAID's, most of which inhibit platelet activation. Since that work is well-known, we shall limit discussion to brief review of some of the more interesting recent developments relevant to the link between platelets and neuro-inflammatory diseases. Several recent reviews are available on the general subject of bioactive lipids [132] and with emphasis on the neuronal nucleus [133].

In the last decade it has come to attention that several of the coagulation proteins are inflammatory or anti-inflammatory, and these are relevant to neurodegenerative diseases because they are implicated in MS (see later). They relate to platelets since the two systems go hand-in-hand: Activated coagulation stimulates platelets (e.g., thrombin) and activated platelets amplify coagulation.

The complement (C) system is a humoral arm of the innate immune system and can attack self-cells when marked by autoantibodies or due to defects in proteins that protect against C. It is a complicated system of circulating proteins, analogous to the coagulation system in that its activation involves a series or cascade of proteolytic conversions of zymogens to their active forms, resulting in several ultimate products, notably, the C5b-9 membrane attack complex which kills by punching holes in the target cell. The C system has several links to the coagulation system [215, 216]. For example, protein S, a cofactor of the protein C system, is carried in circulation largely bound to C4 binding protein (C4bp), a complement component. It is well known that C is central to the pathology of many autoimmune and inflammatory disorders.

Platelets possess C receptors CR2, CR3, CR4, C1q, C1-inhibitor, and factors D, and H. Others are listed in some sources but are disputed as they may be acquired from plasma. Platelets are capable of deploying the lethal C5b-9 complex [217]. In addition, like other blood cells, platelets contain a set of membrane proteins which specifically protect them against autologous C-mediated attack, i.e. against self-injury by C [218]. These are CD55 (decay accelerating factor, DAF), CD59 (membrane inhibitor of reactive lysis, MIRL), and homologous restriction factor (HRF) [219, 220]. Morgan mentions others less well defined. Recently, another such protective protein, Crry, was identified in murine platelets and erythrocytes [221, 222], later extended by those authors to work on leukocytes [223], but Crry appears to be absent in humans. Defects in these proteins can result in pathology, e.g. paroxysmal nocturnal hemoglobinuria (PNH). We investigated CD59 on platelets from PNH patients and found levels ≈10% of normal, but findings on altered sensitivity to C-mediated lysis were inconclusive [224].

A recent finding of significance in neurology is that platelet-bound complement fragment, C4d, is a highly specific biomarker for systemic lupus erythematosus (SLE) and neuropsychiatric lupus [225]. Those authors propose this as a biomarker of cerebrovascular inflammation generally. Moreover, it correlated closely with positive lupus anticoagulant (LA) (p < 0.0001), and less well with positive anti-cardiolipin (aCL) (p = 0.035) [225]. Work by the same authors demonstrated that platelet-bound C4d was associated with ischemic stroke [226]. We have drawn attention to evidence for C-mediated injury in the neuropsychiatric aspect of anti-phospholipid syndrome, which can resemble MS, as referenced [66]. According to Roach et al, C5a signaling in macrophages is synergistic with PAF and with LPA [227].

Despite the presence of the above mentioned protective proteins, platelets are very sensitive to C-mediated attack. We have witnessed the serum-dependent fragmentation of platelets into microparticles when opsonized with an anti-platelet IgM [228]. However, platelets are not entirely defenseless against C, as it has been shown that attack complexes are selectively shed from the platelet membrane on platelet microparticles (PMP), allowing recovery of the parent cell [229, 230]. Butikofer et al has shown that the microparticles released from erythrocytes are selectively enriched in proteins that protect against autologous C, at the expense of these proteins in the parent cell, and in their discussion cite evidence that same is true of platelets [231]. This implies that microparticle shedding sensitizes the remnant cell to C-mediated injury.

The list of active agents in platelets has been expanding in recent years by application of new technologies: proteomics, lipidomics, and mRNA transcript analysis. Hundreds of proteins were identified -- and many others were not identified -- in the supernatant of activated platelets [232–234]. Some 578 proteins were identified just in platelet microparticles [235]. Several proteomic studies of the whole platelet membrane were recently reviewed [236], with a special section devoted to proteins of the lipid rafts. (Lipid rafts are regions of the membrane which resist detergent solubilization and play critical roles in platelet physiology [237, 238]). Raft regions are selectively shed with microparticles, e.g. [239].

The platelet "transcriptome" is also growing. Although platelets lack nuclei, thousands of transcripts have been identified in platelets, including coding for enzymes, interleukin receptors, etc., many of which were previously unknown in platelets, e.g. PEAR-1 [240–242]. Some of the same leaders in the field have sought to extend proteomic studies to include platelet-specific genes [243]. A recent paper claims to show that nuclear factor NFkB is not only present in platelet but active [244], which seems impossible since NFkB is a nuclear transcription factor, as discussed in an editorial [245]. The still-nascent field of lipidomics, which applies the same methods as proteomics (mass spectrometry) but to lipids, promises great advances in sorting out the myriad of bioactive lipids. Potentially important new platelet proteins continue to come to light, such as the septins, reviewed in relation to neurodegenerative disorders [246], and the vanilloid receptor (TRPV1) for noxious stimuli such as capsaicin [247] but also for several catechol amine metabolites such as homovanillic acid [248]. We shall cite the significance of some of this work to MS presently.

In 1985, Stahl pointed out a number of parallels between platelets and neurons, with focus on the storage and secretion of serotonin (5-HT), and the similar sensitivities to many CNS-active drugs [249]. Reed et al devoted a section of their review of mechanisms of platelet secretion to a comparison with vesicle trafficking in neurons [250], echoing work by Lemons et al [251]. Steidl et al in a study of CD34+ hematopoietic progenitor cells identified numerous ion channels, neuromediators, and other proteins previously assumed to be restricted to the CNS [252]. The reason for abundant acetylcholinesterase activity of erythrocytes (but not platelets [253]) remains a mystery attracting much interest [254–257]. The above-mentioned finding of endocannabinoid receptors, semaphorin A, monoamine oxidase (MAO), etc., on platelets is more of the same.

Serotonin (5HT) does not itself cross the BBB but some of its metabolites or precursors do, resulting in a degree of correlation between peripheral and CNS levels of 5HT [258]. Platelets are the main source of circulating 5HT but arteries and veins can act as reservoirs [259]. Of the several receptors for 5HT, platelets possess 5HT(1), as well as a serotonin transporter (SERT). Among the more surprising recent discoveries is the role of 5HT in regulating bone density [260].

Of direct relevance to inflammation is the demonstration that 5HT enhanced monocyte stimulation of CD4+ T cells and cytokine production following LPS exposure [261]. That paper also reports increased plasma 5HT in Alzheimer's disease (AlzD), which correlated with disability index [261]. Ciz et al reported inhibition of the oxidative burst of phagocytes by 5HT, via action on the 5HT receptor and on myeloperoxidase activity [262].

Treatment of MS with selective serotonin reuptake inhibitors (SSRI's) was reportedly very promising [263]. Plasma levels of 5HT are governed mainly by the number of SERT's on the platelet, with more or less of them appearing at the membrane surface depending on external 5HT concentration [264]. Accordingly, the number of SERT's per platelet is sensitive to SSRI's. Platelet SERT density correlated closely with P-selectin expression, a marker of platelet activation and cell-cell interaction, consistent with the concept of 5HT as a weak agonist [265]. However, Galan et al concluded that, contrary to tradition, 5HT is not a weak agonist of platelets, but instead sensitizes them or potentiates their responsiveness [266].

Abdellah et al observed significant differences in the action of SSRI on platelets, depending on polymorphisms of the gene for SERT (SLC6A4) and its promoter (5-HTTLPR) [267]. In an EAE model, it was found that knock-out of SERT caused increased plasma 5HT and attenuated symptoms, with effects most pronounced in female animals [268]. Interestingly, a strong gender effect was seen also with the influence of cannabinoids (in habitual cannabis smokers with cognitive impairment) on 5HT uptake in plasma [269]. In MS, levels of 5HT in CSF appear to correlate well with progression of disease [270]. Norepinephrine, at plasma levels seen in stress disorders, desensitizes the 5HT(1) receptor by uncoupling it from SERT via G proteins, leading the authors to conclude that NE can modulate 5HT responses and the action of SSRI's [271].

It was recently shown that 5HT can become covalently bound to a number of proteins, a process termed seritonylation, including to small GTPase's, with roles at the vessel wall [272] and in platelet activation [273]. The latter authors report that low levels of plasma 5HT markedly prolong the bleeding time. In summary, these brief notes from a large literature suggest that 5HT from platelets, or acting on platelets, may be a significant factor in neuroinflammatory conditions, especially in a restricted microenvironment such as at the BBB.

Although the focus of this review has been on MS, many of the papers cited refer also to other neuroinflammatory disorders. Here we take the example of Alzheimer's disease (AlzD). In 1998, we reported chronic platelet activation in AlzD [274] but that work was largely ignored in the blinding light of the β-amyloid hypothesis. Several recent developments are now forcing a "radical rethink" of the disease. (i) Platelet activation in AlzD has been confirmed [275]. Levels of homocysteine, considered a marker of hypercoagulable state, also impact on AlzD [276]. (ii) Two recent large genetic studies uncovered, in addition to APOE, three other AlzD-related genes, notably clusterin and CR1, both signatures of complement involvement [277, 278]. (iii) Mounting evidence indicates that AlzD is inflammatory in nature and etiology [279], and that it can be controlled by exercise [280] and possibly diet [281]. (iv) Consistent with an inflammatory vascular etiology is the presence of complement fragment C1q in the brain lesions of a mouse model [282] and of the coagulation protein, fibrinogen, in human AlzD plaque [283]. (v) The hypothesis that age-related cognitive decline generally is mainly a vascular condition is supported by finding that retinal microvascular abnormalities predict decline [284]. (vi) To the extent that β-amyloid is truly causative, it should be noted that platelets are the principal source of amyloid precursor protein (APP) [285, 286]. However, the relation of plasma levels of β-amyloid forms to AlzD has been controversial, and ratios of some of the forms are associated as much with vascular dementia as with AlzD [287]. In this connection, it is relevant to note that platelets appear to be responsible for post-surgical cognitive impairment [288].

Finally, (vii) recent work linking all of the above into one coherent hypothesis was brought to our attention during peer review of this paper: Platelets are now strongly implicated in the overexpression of the enzyme which liberates the offending peptide, amyloid-β (Aβ), from APP, known as BACE (β-site APP Cleaving Enzyme), or generically, "β-secretase". BACE was first identified as the long-sought β-secretase in 1999 [289] but those authors did not assay platelets in their study of its tissue distribution. In 2004, a direct relation was found between early-stage AlzD and platelet activity of BACE, as well as ADAM10, leading those authors to propose this assay as a diagnostic aid [290]. (ADAM10, also on platelets, has similar activity [291]). This was confirmed in 2005 [292] and again in 2008 [293]. BACE inhibitors to protect against progression of AlzD are in active development, but Hu et al advise caution because of their finding that BACE inhibition also delays (but does not totally abolish) remyelination by blocking cleavage of neuroregulins [294]. Interestingly, it was recently shown that platelet membrane cholesterol content modulates the activity of β-secretase, possibly explaining reported relations between cholesterol, dietary lipids, and AlzD [295].

MS is an immune-mediated demyelinating disease of the CNS and can be regarded as a model neuroinflammatory condition. Like most other autoimmune disorders, the etiology and pathophysiology of MS remains uncertain but most agree that a combination of genetic and environmental factors are required to initiate an immune reaction against CNS antigens. For example, Gong et al [296] has proposed a 5HT deficiency at high latitudes as a predisposing factor in the epidemiology.

The possible involvement of platelets in MS was first studied by Putnam in 1935, who considered a role for venule thrombosis in CNS demyelination [297]. In the 1950's-60's, at least ten studies appeared on the relation of platelets to CNS demyelination, several of which reported augmented platelet adhesiveness in MS, which correlated with disease activity [298–301]. More recently, a number of observations of platelet abnormalities in MS patients have appeared [302–304], and others cited below. We became interested when colleague W. Sheremata encountered MS patients with severe immune thrombocytopenic purpura (ITP) [305], leading to our report of increased platelet microparticles and platelet activation marker CD62P (P-Selectin) in MS [306]. Thus, chronic platelet activation in MS may now be regarded as well established, including by the report of Cananzi et al cited earlier [54]. Epidemiological studies have found a prevalence of ITP-like thrombocytopenia in MS patients about 25-fold higher than in the general population [304, 307]. As earlier mentioned, mild thrombocytopenia, which could often be overlooked as insignificant, could be a signature of PAF activity.

It may be objected that a modest degree of platelet activation in MS is simply a consequence of general inflammation. However, the finding of platelet-specific GP IIb/IIIa in lesions of MS patients [191] comes close to "Smoking-Gun" evidence of platelet involvement. Relatedly, several elements of coagulation have been detected in the lesions including fibrin, tissue factor (TF), and protein C, suggestive of a procoagulant state [191, 203].

For these and other reasons, MS has been described as a vascular disease [308]. The adhesion molecule, PECAM-1, may be particularly important in this regard. It was reported in 1999 that levels of serum soluble PECAM-1 (sPECAM-1) are significantly elevated in patients with active, gadolinium-enhancing lesions [309]. In 2001, we reported that PECAM-1-positive endothelial microparticles (EMP) are elevated in MS patients only during relapses, and correlated well with gadolinium-enhancing lesions [310]. In 2005, another study documented increased sPECAM-1 during acute relapse and in remission compared with progression [311]. It is not clear if the microparticle-bound form is functionally distinct from the soluble form but we have discussed examples of "soluble" biomarkers which are in fact membrane particle-bound [52] (Part D). These and other findings suggest that PECAM-1, whether soluble or MP-bound, may be at least an indicator of BBB disruption in MS and a biomarker of disease activity, and probably a key participant. The PECAM-1 gene is located on chromosome 17 in the region 17q23 [312] and, given that the region 17q22 was proposed to be an MS-susceptibility factor, PECAM-1 seemed to be a good candidate gene. However, at least two studies failed to find convincing support for that hypothesis [313, 314].

Our findings, together with evidence reviewed in prior sections, suggest a broader hypothesis, namely, that platelet interaction with leukocytes at the endothelium of the BBB is responsible for the release of PECAM-1 to the circulation, and associated infiltration of leukocytes. Platelets are capable of directly activating both lymphocytes and dendritic cells [103, 315]. In addition, it is tempting to postulate involvement of platelet activating factor (PAF) in view of its potency at disrupting endothelial junctions, and likely signature of thrombocytopenia (see Section 9). Elevation of PAF in the CSF and plasma of RRMS patients was reported, and the authors concluded that PAF is likely responsible for the early disruption of the BBB in MS [193]. Moreover, PAF receptors are up-regulated in MS lesions [191].

Differences in these parameters between RRMS and progressive MS have been noted in several reports, hinting at distinctive pathophysiologic pathways. Humm et al observed such differences in responses to prednisone [316]; and PAF activity was higher in RRMS than in secondary progressive MS [193]; and stronger differences were observed in sP-Selectin and other markers in RRMS compared to secondary progressive [311].

As earlier mentioned (Section 9), PAF is inactivated by PAF-AH but no association of the inactivating mutation of the PAF-AH gene with RRMS or progressive MS was found [189]. Nevertheless, the PAF-AH activities in MS were significantly lower than in healthy controls [189]. Such a decrease of PAF-AH activity may in part be responsible for the reported increase of PAF in MS plasma and CSF, and therefore could contribute to the inflammation and vascular permeability changes seen in the CNS of MS. On the other hand, a small Japanese study did find a significant association between MS susceptibility and a PAFR polymorphism causing a modest but significant reduction of PAF-dependent signaling [190].

We also found that platelet-associated IgM (but not IgG) is increased in MS patients [317]. We feel that this is of potential importance to understanding the pathophysiology of MS. However, present knowledge of antiphospholipid antibodies is too fragmentary to offer much insight on the significance of this finding, as discussed [66].

In summary, there are numerous mechanisms by which platelets could substantially contribute to the pathophysiology of MS. We do not pretend to have any specific hypothesis, nor do we propose that some bizarre platelet abnormality actually causes MS (although that is not impossible). Rather, the purpose of this review is to call attention to the neglected platelet and its potential to modulate inflammatory processes.

This review indicates that platelets could be pivotally involved with neurodegenerative and autoimmune conditions. At present, the potential role of platelets in such disorders has been neglected, although well appreciated in the cardiovascular and cancer research fields. The majority of studies in tissue culture designed to elucidate the pathophysiology of neurodegenerative diseases such as MS have investigated interactions of leukocyte subsets and endothelial cells, but it is likely that these interactions could be significantly modulated in the presence of platelets from patients vs. healthy controls.

The question may arise, exactly what kind of hypothesis could link platelets to such a wide variety of neurological conditions? Our view is that platelets are partners with leukocytes and other immunological effectors (complement, TLR's, PPAR's, resolvins, PAF, CD40/CD40L, etc.), amplifying or otherwise modulating those effectors in ways distinctive for each condition, with actions likely to be most prominent at the BBB.

A number of topics and many references were cut from this review because of excessive length. They include the role of platelets in several viral infections, consideration of other possibly relevant receptors (e.g., vanilloids, vasopressin), new work dissecting prothrombotic from inflammatory pathways, and further details on the topics covered. However, it is hoped that enough has been said to inspire new respect for the humble little platelet.

If this review succeeds in raising awareness of the potential roles of platelets in neurodegenerative and neuroinflammatory conditions, the labor of assembling it will have been amply rewarded.