OPN is a non-classical extracellular matrix glycoprotein referred to as a matricellular protein. It is different from a classical extracellular matrix protein and has the following common characteristics: it undergoes transient upregulation of expression at a specific stage of development or under a pathological condition, can be present as a soluble protein, induces cell motility, and can interact with a variety of biologically active substances such as growth factors, chemokines and proteases to modulate the function of the bioactive substances and to act as a reservoir of the bioactive substances. OPN gene disruption produces no apparent phenotype, although OPN knock-out mice manifest diverse phenotypes responsive to numerous insults [17, 18]. In contrast, a classical extracellular matrix protein is constitutively expressed, present as a structural component and provides scaffolds for the stability of cell–cell adhesions [17]. Transcription of OPN is accelerated by transforming growth factor-$\beta$, platelet-derived, epidermal or basic fibroblast growth factors, cytokines (interleukin-1$\alpha$, interleukin-2, or tumor necrosis factor-$\alpha$) and endothelin [18]. OPN is known to have 16 signal peptides in the N-terminus, in addition to Ser/Thr phosphorylation sites and a Gly-Arg-Gly-Asp-Ser motif [19].

The gene of OPNs is located in a small integrin-binding ligand N-linked glycoprotein cluster on chromosome 4 (4q13) in humans [20]. OPN genes contain 7 exons, and 6 of the 7 exons are translated into total length of OPNs, which are also called OPN-a [21]. In humans, alternative translation and splicing develop four splicing variants: OPNs-b lacking exon 5, OPNs-c lacking exon 4, OPNs-4 lacking exons 4 and 5, and OPNs-5 containing one extra exon by retaining a portion of intron 3 [22]. OPN can be proteolytically cleaved by thrombin and matrix metalloproteases (MMPs) [22]. Upon thrombin cleavage at the Arg168-Ser169 site, OPN transforms into two types of OPN fragment, one an N-terminal fragment (OPN-N) and the other a C-terminal fragment (OPN-C) [14]. OPN-Ns contain several cell adhesive motifs, which are highly conserved, such as Arg-Gly-Asp (RGD) sequences interacting with integrin receptors including $\alpha$v$\beta$1, $\alpha$v$\beta$3, $\alpha$v$\beta$5, $\alpha$5$\beta$1 and $\alpha$8$\beta$1; and cryptic Ser-Val-Val-Tyr-Gly-Leu-Arg (SVVYGLR)-containing motif sequences binding to $\alpha$4$\beta$1, $\alpha$4$\beta$7 and $\alpha$9$\beta$1 integrin receptors: the latter ones are exposed only after the cleavage by thrombin [23, 24]. OPN-C binds to CD44 variants [14]. MMPs-2, -3, -7, -9 and -12 also cleave OPN at various sites, contributing to functional changes of OPN [22]. Additionally, OPNs are subject to a variety of post-translational modifications such as transglutamination, glycosylation, sulfation and phosphorylation, which also potentially control OPN function [25].

OPN exists in two distinct isoforms, one secreted, the other intracellular: the former isoforms exert their functions through activating their receptors on cell surface, while the latter ones bind to myeloid differentiation primary-response protein 88 (MyD88), which is the down-stream of Toll-like receptors, to exert their effects [26]. The coupling of secreted OPN and a specific receptor induces a signal transduction pathway downstream of OPN to secrete cytokine and chemokine, and to cause cellular apoptosis, differentiation, adhesion and migration, and is physiologically or pathologically involved in various processes such as homeostasis, angiogenesis and immune responses in a large number of tissues [11, 27]. Another complexity is that the biological role of OPN is highly diverse, ranging from beneficial to harmful, and is often seemingly contradictory in accordance with the biological scenarios on OPN’s upregulation [28]. For example, even in the same organ, OPN acts detrimentally for degenerative or demyelinating disorders including Alzheimer’s disease, Parkinson’s disease and multiple sclerosis, while acts favorably against acute brain injuries after ischemic and hemorrhagic strokes, although OPN is upregulated in all the situations [29].

OPN exists in a body fluid and as with other matricellular proteins is available as biomarkers to monitor the progression or regression of tumors and cardiovascular, inflammatory and fibrotic diseases [30]. In patients with aneurysmal SAH, plasma OPN-a levels increased from an acute stage to peak at four to six days post-SAH [31]. In SAH patients with worse admission clinical grades and poor three month outcomes, plasma OPN-a levels were significantly higher at least from day 1 to day 12, while higher OPN-a levels were revealed in patients with delayed-onset non-iatrogenic cerebral infarct at days 1–3 and 10–12, DCI at days 4–9 and chronic shunt-dependent hydrocephalus at days 10–12 [16, 31]. OPN-a levels in the peripheral blood at days 1–3 after SAH may reflect the severity of EBIs [30], which have been shown to influence the development of delayed cerebral infarction by experimental and clinical studies [30, 32]. Cerebrospinal fluid (CSF) concentrations of OPNs (unknown subtypes) were much higher compared with plasma ones at days one, four and eight in aneurysmal SAH patients, indicating that OPN is primarily produced in the central nervous system after SAH [33]. It was also suggested that an EBI-unrelated cause of DCI was present, because DCI was accompanied with higher plasma levels of OPNs-a not at days 1–3 but 4–9 [31].

It has been reported that OPNs are upregulated in injured brain tissues and cerebral arteries in a delayed manner after experimentally induced SAH and induce intrinsic defense mechanisms on EBIs, cerebral vasospasm and therefore DCI [11, 34]. If the findings in experimental studies can be applied to clinical practice, plasma OPN-a may show a delayed increase by reflecting the severity of preceding tissue injuries associated with EBI, cerebral vasospasm and DCI and/or the extent of the resultant inflammatory reactions [17, 31]. OPN that is upregulated in injured tissues may exert healing effects, but intrinsically induced OPN levels may be insufficient to recover the injury immediately. However, in the case that exogenous OPNs are administered, OPNs may exert neuroprotective effects [11, 34]. Nevertheless, it should also be remembered that when OPN remains highly induced from days 4–6 to days 10–12 post-SAH, chronic hydrocephalus requiring CSF shunting surgery may develop, possibly via an undue OPN’s healing effect that promotes fibroses in the subarachnoid and other CSF spaces [16]. Despite this, plasma OPN-a levels could be used as a biomarker for EBI, cerebral vasospasm, DCI, chronic hydrocephalus requiring shunt surgery or poor outcomes on the basis of measurement time points [16, 30]. Recently, machine learning analyses have been developed and constructed early prediction models for the occurrence of DCI, angiographic cerebrovascular spasm, and delayed cerebral infarct using admission clinical factors and plasma levels of three types of matricellular proteins (OPN-a, periostin, and galectin-3) at post-SAH days one to three in a clinical setting [35]. Periostin and galectin-3 have been implicated in EBI and DCI or delayed cerebral infarction after SAH [36, 37, 38, 39, 40, 41, 42, 43, 44]. However, a clinical study demonstrated that plasma OPN-a levels at days one to three were also one of the three most important features in all of the three prediction models with high accuracy and sensitivity [35].

OPN plays diverse roles in neuroinflammation, blood-brain barrier (BBB) disruption and cellular apoptosis and has a positive effect on chemotaxis, proliferation, differentiation or survival of a variety of cells such as smooth muscle cells in the cerebral artery, neural progenitor or stem cells and neuroblasts in the central nervous system by interacting with integrin and CD44 receptors [14]. Collectively, OPNs are believed to be neuroprotective during an early phase of SAH and represent potential therapeutic targets against post-SAH cerebral tissue injuries. However, clinical trials relevant to OPNs have never been conducted. Recombinant OPN (r-OPN) or OPN peptide has been administered intrathecally or intranasally to animal models of SAH, but the administration route limits the clinical applications of OPNs [14]. Orally administrated Ephedra sinica (a Chinese herb extract) and preconditioning with hyperbaric oxygen conferred protection on acute brain injuries by enhancing endogenous OPN’s signaling [14, 45], offering an alternative possibility of the therapeutic application. Also, computer-aided drug design methods have advanced to minimize ligands that should be screened in bioassays, saving the cost, time, and efforts required for developing a new drug [46]. Such technologies will help develop new OPN-relevant drugs in future studies.

EBI is defined in clinical settings as post-ictus cerebral damage that occurs within 72 hours following rupture of an intracranial aneurysm and before the development of DCI [47]. In experimental SAH studies, endovascular perforation models in mice or rats have shown an acute metabolic change similar to clinical conditions and have been established as the most suitable models for studies of EBI after SAH [5, 48, 49]. Cerebral tissue injuries that occur under ischemic or hemorrhagic stroke are characterized by imbalances of metabolism and energy within cerebral tissue cells and the structural damages are followed by a decrease in cerebral blood flow and the disruption of cellular membranes [50]. This leads to mitochondrial impairment, inflammatory or oxidative reactions, ionic gradient breakdown, glutamate-mediated excitotoxicity, stress signaling and finally to cell deaths [50]. After SAH, decreasing CSF glucose-to-lactate ratios were reported to be accompanied by unfavorable outcomes [51], but the role of extracellular lactate is still controversial if the lactate supports basal metabolism in neurons or contributes to neurotransmission activity per se [52, 53]. Recently, it has been revealed that neurons and glia orchestrate a metabolic reaction to sustain energy demands and the redox balance of neuronal activities or cause post-stroke brain injuries such as aquaporin 4-medated brain edema and neuronal apoptosis, which may be reduced by hypothermia-induced suppression of metabolic activity [50, 53, 54, 55, 56]. However, in the models, EBI has been studied in the same time frame of 72 hours as the time course of cerebrovascular spasm occurrence: therefore, it is impossible to differentiate brain injuries by EBI from those caused by cerebral vasospasm or DCI [5, 48, 49]. Nevertheless, it has been revealed that EBI involves various pathophysiological processes including a surge of intracranial pressure, subsequent transient occurrence of global cerebral ischemia, mechanical brain injuries by associated intracerebral hemorrhage and acute hydrocephalus, and subarachnoid spread of blood-derived products [6, 57]. These events are followed by excitotoxicity, inflammatory and free radical reactions, microcirculatory disturbance, increased permeability of BBB, cortical spreading depolarization or depression, and others, finally leading to neuronal apoptosis [6, 57]. Glial cells are considered to contribute to post-stroke brain injuries or pathophysiology in an acute phase [58, 59], as well as intimately and actively control neuronal activities and synaptic neurotransmissions in brain function [60, 61]. These phenomena interact each other: for example, BBB disruption causes further infiltration of toxic blood components and inflammatory cells into brain tissue and furthermore aggravates inflammatory reactions, BBB permeability, brain edema and brain injury [62, 63, 64]. Although OPNs are pleiotropic glycoproteins, OPNs have been always protective for disruption of BBB and apoptosis of neurons, which are important constituents of EBI following experimentally induced SAH [11].

Post-SAH ischemia by elevated intracranial pressure, mechanical injury on cerebral tissues and extravasation of blood components as well as their degradation products disrupt BBB associated with activation of multiple inflammatory and other signaling pathways that are independent or interconnected, exacerbating EBI [62, 65, 66]. Although post-SAH BBB disruption is caused by many molecules and pathways and has crucial roles in EBI [62], BBB disruption induces only transient damage and recovery mechanisms may ensue [67]. In endovascular perforation SAH models in rats, endogenous OPN induction occurred in reactive astrocytes and brain capillary endothelial cells and peaked at 3 days after SAH, at which BBB disruption spontaneously recovered [68]. Upregulated OPN increased expression of angiopoietin-1, in addition to mitogen-activated protein kinase (MAPK) phosphatase-1 that is an endogenous MAPK inhibitor, to inactivate MAPKs such as extracellular signal-regulated kinase (ERK) 1/2, c-Jun N-terminal kinase and p38, and suppressed expressions of vascular endothelial growth factor-A through RGD-dependent integrins [68]. These bioactive molecules are all implicated in progression or regression mechanisms of BBB injuries, acting simultaneously or at different stages through the course of BBB disruption after SAH [68]. Vascular endothelial growth factor-A potently induces the disruption of BBB, while angiopoietin-1 blocks effects of vascular endothelial growth factor-A, possibly by regulating MAPK activation [69]. The blockage of endogenous OPN expression prevented the pathways and the recovery of BBB disruption [70]. Endogenous induction of OPNs is considered to serve as an important intrinsic mechanism for BBB protection or repair after SAH (Fig. 1).

Fig. 1.

Possible pathways of osteopontin (OPN) to suppress the disruption of blood-brain barrier (BBB) after subarachnoid hemorrhage (SAH). OPN induces angiopoietin-1 and mitogen-activated protein kinase (MAPK) phosphatase-1 (MKP-1) through Arg-Gly-Asp (RGD)-dependent integrins, then inactivates MAPKs to downregulate matrix metalloprotease (MMP)-9 as well as vascular endothelial growth factor-A. OPN also inhibits nuclear factor-kappaB (NF-$\kappa$B) signaling, and downregulates MMP-9. Additionally, OPN activates CD44 and P-glucoprotein glycosylation signals. All these pathways contribute to the inhibition of BBB disruption after SAH. IL-1$\beta$, interleukin-1$\beta$; TIMP-1, tissue inhibitor of MMP-1.

OPNs have also been experimentally examined as to the possibility as therapeutic agents. In endovascular perforation SAH models in rats, r-OPN administration prevented BBB disruption, brain edema formation as well as body weight loss and improved neurological function after SAH by deactivating a nuclear factor-kappaB (NF-$\kappa$B)-mediated signaling pathway, resulting in MMP-9 downregulation, tissue inhibitor of MMP-1 maintenance and the resultant preservation of substrates for MMP-9 such as brain capillary basal lamina protein laminin and the tight junction protein zona occludens (ZO)-1, both of which are important elements of BBB [28]. There is much evidence that MMP-9 degrades extracellular matrix proteins of cerebral microvessels including basal lamina proteins consisting of laminin, collagen IV and fibronectin, as well as ZO-1 belonging to endothelial tight junction-related proteins, loss of which causes disruption or increased permeability of BBB [14]. However, post-SAH increases in active interleukin-1$\beta$, which activates NF-$\kappa$B signaling, were not suppressed by r-OPN, suggesting that r-OPN blocked intracellular signaling upstream of NF-$\kappa$B via RGD-dependent integrins [70] (Fig. 1). In the same rat models, intranasal vitamin D3 reportedly upregulated expression of endogenous OPN isomers (OPNs-a and OPNs-c, but not OPNs-b) in astrocytic and brain capillary endothelial cells [71]. Induced OPN activated CD44 and P-glucoprotein glycosylation signals in the capillary endothelium and attenuated the disruption of BBB and cerebral edema [71]. The inhibition of complement C3 by an extract of Ephedra sinica also alleviated the disruption of BBB and cerebral edema associated with improved neurological functions, possibly via the upregulation of sonic hedgehog and OPN signaling and the consequent reduction of MMP-9 expression [45].

There are other possible mechanisms of OPN protection of the BBB, which have not been investigated in SAH yet. For example, administration of exogenous OPNs has reduced oxidative stress that activates NF-$\kappa$B; as a result, OPNs downregulate MMP-9 expressions and upregulate a tissue inhibitor of MMP-1, leading to the maintenance of the integrity of BBB [14, 28]. Furthermore, exogenous OPNs are known to promote focal adhesion kinase (FAK) phosphorylation and subsequent activation of phosphatidylinositol 3-kinase (PI3K), inducing Ras-related C3 botulinum toxin substrate 1 (Rac-1) to preserve the integrity of BBB [14]. Additionally, endogenous OPNs have been reported to induce polarization of reactive astrocytes that is pivotal to the complete coverage of neovessels by astrocytic end-feet and the BBB integrity [72].

EBI finally leads to neuronal death or apoptosis, which has been demonstrated experimentally and clinically [73, 74]. The PI3K–Akt pathway is related to an antiapoptotic mechanism in neurons after experimentally induced SAH [75] and the activation of the pathway has been demonstrated to exert an antiapoptotic role [76]. Activation of FAK, a cytoplasmic tyrosine kinase, which is triggered via integrin or CD44 receptors, stimulates the PI3K–Akt signaling pathways [14, 77]. Phosphorylated Akt can then suppress proapoptotic proteins such as B-cell lymphoma 2 (Bcl-2)-associated X protein (Bax) and Bcl-2-associated agonist of cell death (BAD) and upregulate antiapoptotic proteins including Bcl-2 [14]. ERK1/2 is another classic downstream protein of FAK [78, 79].

Accumulating studies have demonstrated that OPNs exert direct antiapoptotic action through integrin and CD44 receptors [14]. However, the underlying mechanism for OPNs to inhibit caspase-3 activation remains unclear, although cleaved caspase-3 may be a common downstream protein to apoptosis after SAH [14]. Administration of r-OPN was reported to attenuate neuronal apoptosis in the cerebral cortex and brain edema and improved neurological status in endovascular perforation SAH models in rats, possibly through the activation of the FAK–PI3K–Akt signaling pathway that inhibits capase-3 cleavage [77]. In the same model, although SAH upregulated endogenous OPNs and autophagy (ATG)-related proteins (Beclin 1, ATG5, and microtubule-associated protein light chain 3 [LC3] II to I ratio), r-OPN administration further caused an increase in expressions of ATG-related proteins [79]. r-OPN inhibited apoptosis of neurons by ATG activation and the regulation of ATG-apoptosis interactions, accompanied by increased expressions of an antiapoptotic protein Bcl-2 and decreased expressions of proapoptotic proteins (cleaved caspase-3 and Bax) [79]. The same group also reported that FAK–ERK1/2 signaling may be involved in r-OPN-enhanced ATG and reduced neuronal apoptosis in post-SAH EBI [78] (Fig. 2).

Fig. 2.

Possible pathways of osteopontin (OPN) that protect against neuronal apoptosis after subarachnoid hemorrhage. OPN activates focal adhesion kinase (FAK) through Arg-Gly-Asp (RGD)-dependent integrins, subsequently activating the phosphatidylinositol 3-kinase (PI3K)-Akt signaling, which in turn inactivates proapoptotic caspase-3 and B-cell lymphoma 2 (Bcl-2)-associated X protein (Bax). OPN-mediated FAK activation also activates extracellular signal-regulated kinase (ERK)1/2 and then upregulates autophagy (ATG)-related proteins to inhibit neuronal apoptosis. LC3, microtubule-associated protein light chain 3.

In a rat intracerebral hemorrhage model, intracerebroventricularly injected r-OPN increased phosphorylated Akt expression, which phosphorylated and subsequently inactivated pro-apoptotic glycogen synthase kinase 3 beta (GSK-3$\beta$), suppressing Bax-to-Bcl-2 ratios and caspase-3 activation that result in a decrease in cerebral edema and cell deaths [80]. Similar mechanisms may be implicated in r-OPN-induced antiapoptotic effects after SAH.