Insight into the divergent role of TRAIL in non‐neoplastic neurological diseases

Abstract Tumour necrosis factor–related apoptosis‐inducing ligand (TRAIL) is a member of the tumour necrosis factor (TNF) superfamily which mainly induces apoptosis of tumour cells and transformed cell lines with no systemic toxicity, whereas they share high sequence homology with TNF and CD95L. These unique effects of TRAIL have made it an important molecule in oncology research. However, the research on TRAIL‐related antineoplastic agents has lagged behind and has been limited by the extensive drug resistance in cancer cells. Given the several findings showing that TRAIL is involved in immune regulation and other pleiotropic biological effects in non‐malignant cells, TRAIL and its receptors have attracted widespread attention from researchers. In the central nervous system (CNS), TRAIL is highly correlated with malignant tumours such as glioma and other non‐neoplastic disorders such as acute brain injury, CNS infection and neurodegenerative disease. Many clinical and animal studies have revealed the dual roles of TRAIL in which it causes damage by inducing cell apoptosis, and confers protection by enhancing both pro‐ and non‐apoptosis effects in different neurological disorders and at different sites or stages. Its pro‐apoptotic effect produces a pro‐survival effect that cannot be underestimated. This review extensively covers in vitro and in vivo experiments and clinical studies investigating TRAIL. It also provides a summary of the current knowledge on the TRAIL signalling pathway and its involvement in pathogenesis, diagnosis and therapeutics of CNS disorders as a basis for future research.

cell lineages or transformed target cells, 3 and leaving normal cells unaffected. 4 The ability to selectively induce apoptosis in tumour cells without affecting healthy cells makes TRAIL a vital module in the field of cancer treatment, 3 including wide utilization in the treatment of glioma in the CNS. 5,6 However, it has recently emerged that many cancers are becoming TRAIL-resistant. This can be attributed to several TRAIL limitations which include poor agonistic activity and stability of recombinant soluble TRAIL. Cancer cells are also capable of exploiting endogenous TRAIL/TRAIL-R system to their advantage.
Many cancers have been TRAIL-resistant, and this has made scientists explore more strategies and nanotechnological advancements to enhance the apoptosis promoting effect of trail on cancer cells. 7,8 In addition to its expression in a wide range of normal tissues, TRAIL messenger RNA is expressed in human natural killer cells, B cells, monocytes and dendritic cells following cytokine stimulation. 9,10 This means that the regulation of TRAIL-mediated cell death is more complex than simply interacting with five distinct receptors identified and characterized previously. 11 Therefore, in addition to anticancer effects, the pleiotropic influence of TRAIL has also been observed in various pathophysiological processes involving multiple systems, such as autoimmune diseases such as rheumatoid arthritis and systemic lupus erythematosus, cardiovascular diseases such as acute myocardial infarction and atherosclerosis. [12][13][14][15] Among CNS diseases, inflammation to a certain extent acts as an important link in the complex pathogenesis regulated by TRAIL. This mediates neuron damage after acute injury, promotes the formation of amyloid β plaque in Alzheimer's disease and plays a specific protective role in experimental autoimmune encephalomyelitis. Based on such a complex and ambiguous background, this paper provides an extensive review of in vitro and in vivo experimental and clinical studies, and then provide a summary of the current knowledge on the TRAIL signalling pathway and its involvement in pathogenesis, diagnosis and therapeutics during CNS disorders. The information provided here is expected to form a basis for future studies.

| TR AIL S I G NALLING SYS TEM AND REL ATED REG UL ATI ON DEMON S TR ATED IN ON COLOG IC AL RE S E ARCHE S
In humans, two TRAIL receptors contain a functional cytoplasmic death domain and thus can transduce the signals to induce apoptosis. They are known as membrane-bound death receptor 4 (DR4/ TRAIL-R1) and death receptor 5 (DR5/TRAIL-R2). These two receptors are similar in structure (58% identity) and are highly distributed in cells and tissues including peripheral blood lymphocytes (PBLs), spleen and thymus. They mediate apoptosis by associating with Fasassociated death domain protein (FADD). [16][17][18] An additional two receptors, namely glycosyl-phosphatidylinositol (GP1)-anchored DcR1 (TRAIL-R3) without an intracellular domain, and DcR2 (TRAIL-R4) containing a truncated death domain serve as decoy receptors.
TRAIL can bind to DcR1 and DcR2 by their cysteine-rich extracellular domain, but this combination cannot induce apoptosis because it lacks a functional death domain (DD). This suggests that the apoptotic capacity of TRAIL can be resisted by competitively binging to decoy receptors. [19][20][21] Another unique soluble decoy receptor named osteoprotegerin (OPG), which is capable of inhibiting osteoclastogenesis in bone remodelling by interacting with OPG ligand (previously described as receptor activator of NF-κB ligand (RANKL), 22 also combines with TRAIL and impairs TRAIL-induced apoptosis. 23 As described above, TRAIL facilitates pro-apoptotic and non-apoptotic effects by binding to death or decoy receptors. The apoptosis course can be triggered by either intrinsic or extrinsic pathways initiated from the formation of a homotrimer of DR4 or DR5. 24 This homotrimer subsequently recruits FADD and pro-caspase 8/10 to form the death-inducing signalling complex (DISC). 25 Cellular FADD-like IL-1β-converting enzyme inhibitory protein (c-FLIP) can also be recruited to DISC as an inhibitor of caspase cascade via heterodimerizing with pro-caspase 8. 26 In this multi-protein complex of DISC, pro-caspase 8/10 is cleaved and activated autocatalytically, producing a vigorous proteinase caspase 8/10 which can cleave multiple downstream proteins such as pro-caspases 3, 6 and 7, and BH3 interacting domain death agonist (Bid). 27,28 Recently, according to the ratio of XIAP to caspase 3 and the DISC's capacity to cleave pro-caspase 3, cells were classified into type I and type II. 29,30 On the one hand, type I cells activate caspase 8/10 which are sufficient or more than enough to cleave pro-caspase 3 and then trigger apoptosis directly, and this pathway is known as an extrinsic apoptotic pathway or receptor-dependent pathway 30-32 ( Figure 1A1). On the other hand, type II cells activate caspase 8/10 converting Bid into truncated Bid (tBid) which later interacts with Bak and Bax proteins on the membrane of mitochondria and leads to the release of cytochrome c and SMAC/Diablo from the unstable mitochondria outer membrane. Cytochrome c couples with pro-caspase 9 and apoptotic protease-activating factor-1 (APAF-1) and assembles into apoptosome complex, which sequentially activates caspase 9 and executioner caspase 3. The SMAC/Diablo also promotes apoptosis as it serves as the negative effect on XIAP, which is a direct inhibitor of final effective apparatus, caspase 3/9. The effect of SMAC/Diablo might be crucial in case of insufficient caspase 8. This pathway is known as an intrinsic apoptotic or mitochondrion-dependent pathway 30-32 ( Figure 1A2).
Apart from classic pathways, TRAIL can also participate in the non-apoptotic pathway in certain situations (high concentrations of TRAIL). This non-canonical signalling can lead to cell survival, proliferation and migration via transcription of several genes at the stimulation of kinase signallings, such as nuclear factor kappalight-chain-enhancer of activated B cells (NF-κB), mitogen-activated protein kinases (P38/MAPK), c-Jun N-terminal kinase (JNK), phosphatidylinositide 3-kinases (PI3K/Akt) and extracellular regulated kinase (ERK). 33,34 For instance, NF-κB, a familiar transcription factor, could up-regulate the transcription of genes such as c-FLIP, Bcl-XL and XIAP. These genes encode intracellular cytokines that are capable of blocking specific sites during the apoptotic processes as explained earlier, thus inhibits apoptosis. 35 The P38/MAPK as a multifunctional kinase could regulate inflammation, cell proliferation and differentiation as well as apoptosis. 36 The ERK protects the integrity of the endothelial cell and promotes proliferation. 37 The PI3K-independent activation of protein kinase B/Akt plays a crucial role not only in cell survival triggered by growth factors, extracellular matrix and other stimuli but also in activation of NF-kB pathway 38 ( Figure 1B).
TRAIL selectively driven apoptosis of tumour cells has been extensively studied in the field of cancer treatment as it was found in the mid-1990s. However, one of the biggest barriers to its effective clinical application lies in the drug resistance against TRAIL-based therapeutics. Among various reasons including decline in of death receptors, activation of oncogenes and silence of tumour suppressor genes, and disorder of anti-apoptotic proteins, the dark side of TRAIL-mediated signalling which also known as non-canonical survival signalling pathways has achieved more attention than TRAIL-induced apoptotic signalling. 39,40 Additionally, according to the cellular context, TRAIL signalling towards death or survival may be affected by factors such as variation between the expression of death receptors and decoy receptors, shuttling of death receptors F I G U R E 1 The TRAIL signalling pathway. A, Apoptotic signalling pathway. The binding of soluble TRAIL to the homotrimer of DR4/5, the DD on receptors, the recruited FADD and pro-caspase 8 form the DISC complex, which triggers autocatalytic activation of caspase 8 in DISC. Subsequently, in (A1) extrinsic pathway, caspase 8 activates caspase 3/6/7 directly; in (A2) intrinsic pathway, caspase 8 cleaves Bid into tBid which later interact with Bax/Bac on the mitochondrial membrane leading to lysis of mitochondria. The Cyt-C released from mitochondria cleaves pro-caspase 9 into caspase 9 in apoptosome complex, which in turn activates caspase 3/6/7. The effector caspase 3/6/7 activated by both pathways induces cell apoptosis. B, Non-apoptotic signalling pathway. Adaptor proteins including TRADD, TRAF2, RIP1 and IKK-γ are recruited to form the secondary signalling complex through a serious complex processes involving several molecules such as NF-κB, P38/MAPK, JNK, PI3K/Akt and ERK. These pathways induce cell survival, proliferation and migration among other processes. APAF-1, apoptotic protease-activating factor-1; Bcl-2, Bcl-XL, Mcl-1, Bac, Bax all belong to Bcl-2 family, B cell leukaemia 2 family; Bid, BH3 domain-containing protein; c-FLIP, cellular FADD-like IL-1β-converting enzyme inhibitory protein; DISC, death-inducing signalling complex; DR4/5, death receptor 4/5; ERK, extracellular regulated kinase; FADD, Fas-associated death domain; IKK-γ, inhibitor of κB (IκB) kinase-γ; JNK, c-Jun N-terminal kinase; MAPK, mitogen-activated protein kinases; NF-κB, nuclear factor kappa-light-chain-enhancer of activated B cells; PI3K, phosphatidylinositide 3-kinases; RIP1, receptor-interacting kinase 1; tBid, truncated Bid; TRADD, TNFR1-associated death domain; TRAF2, TNF receptor-associated factor 2; TRAIL, TNF-related apoptosis-inducing ligand; TRAIL-R, TNF-related apoptosis-inducing ligand receptor; XIAP, X-linked inhibitor of apoptosis protein and regulation of non-coding RNA. 39,40 In a large amount of cancer treatment studies, these factors were reported to confer TRAILbased therapeutics potential adverse effect such as proliferation, migration and metastasis of cancer cells. For example, studies on breast cancer metastasis reported that overexpressed DR5 was able to improve the expression of C-X-C chemokine receptor type 4 (CXCR4) on the surface of MDA-MB-231 BCa cells, whose migratory potential towards stromal cell-derived factor 1 (SDF1) was subsequently enhanced. 41 Interestingly, a natural agent a-mangostin was found to significantly promote shuttling of nuclear accumulation of DR5 to cell surface to break the suppressed condition of TRAILdriven apoptosis. 39 Moreover, a-mangostin can also positively regulate TRAIL-mediated apoptosis by significantly enhancing the expression of DR5 through down-regulating miR-133b. 39 Actually, TRAIL pathways were demonstrated to be regulated by abundant non-coding RNAs such as long non-coding RNA (lncRNA) and mi-croRNA (miRNA), in which miRNA was most widely studied. 42 Noncoding RNAs possess a bright prospect in the field of TRAIL-based tumour therapy and drug resistance relief because most of which positively or negatively regulate the components of TRAIL/TRAIL-R system. 39,42 Unfortunately, there are very limited researches about non-cording RNA on CNS diseases up to now. The study of TRAIL in CNS diseases still stays in the classical apoptotic pathway, and even the dark side has not gain enough attention, let alone non-coding RNAs as the regulators. It reminds us that many valuable and novel experiences can be drawn from the field of tumour researches for the study of TRAIL in CNS.

| The role of TRAIL in ischaemic stroke and haemorrhagic stroke
Cerebral stroke, one of the fatal and disabling diseases worldwide, which cause brain tissue damage because of the sudden rupture of blood vessels or the inability of blood flow because of blockage of blood vessels in the brain, includes haemorrhagic stroke and ischaemic stroke. 43 There are certain research foundations of TRAIL on ischaemic stroke, whereas no literature about haemorrhagic stroke and TRAIL can be retrieved in PubMed. Thus, in this section, we mainly discuss the progress of TRAIL/TRAIL-R system in ischaemic stroke, and our unpublished data which showed the relationship between TRAIL and haemorrhagic stroke will be discussed in the discussion section. Accounting for the majority of stroke events, ischaemic stroke shows various sudden clinical symptoms such as lacunar-like syndromes, aphasia or neglect, ataxic syndromes, visual field defects and associated neurobehavioural syndromes which mainly based on occlusion of different blood vessels. 44 Ischaemic stroke causes tissue impairment not only through direct damage such as oxygen and glucose deprivation or increased intracranial pressure but also through secondary inflammatory response followed by cell death. After the onset of ischaemic stroke, the hypoxia-ischaemia area suffers from fatal cerebral swelling, and later releases tremendous amounts of inflammatory mediators and damage-associated molecular patterns (DAMPs). 45 This is usually followed by prominent infiltration of immune cells. 45 The TRAIL protein is not found in normal brain tissue, 46 but TRAIL is showed to participate in the neuronal apoptosis through a ceramide-mediated c-Jun signalling pathway after brain injury. 47,48 Recruitment of leucocytes by crossing the compromised blood-brain barrier (BBB) and activation of local microglia seem to play a crucial role in the initiation of neuronal apoptosis. [49][50][51] Several pre-clinical studies indicate that TRAIL is mainly expressed in active microglia and macrophages, whereas DR5 is predominantly expressed in neurons after transient global cerebral ischaemia. 52 Meanwhile, the competitive inhibitor-soluble DR5 can reduce the cerebral pathological harm by competitively binding to free TRAIL. 52 Similar changes in TRAIL and its receptor are also found in hypoxia-ischaemia (HI) of immature rat models. 50,53 In these studies, however, DcR1 was overexpressed in the cerebral cortex. 50 Accounting for approximately 85% of all stroke cases, 54 ischaemic cerebral stroke has attracted much attention. However, many agents that are effective in pre-clinical studies have not suitable for clinical application. 55 Because of this, some researchers have provided more information on an endogenous neuroprotective mechanism, for example ischaemic pre-conditioning (IPC), 56 which causes an effect by regulation of intrinsic TRAIL and its receptors rather than extrinsic agents. The IPC conferred tolerance to ischaemic neurons when rats were subjected to sublethal brain ischaemia in 30 mins of tMCAO study as reported by Cantarella et al. 49 The TRAIL and its death receptors were down-regulated, whereas decoy receptors were up-regulated after IPC, and this led to the suppression of intrinsic and extrinsic apoptotic pathways. Recently, Xu et al reported that remote limb pre-conditioning (RPC) ameliorates brain damage after ischaemic cerebral stroke with similar regulation of TRAIL. 54 As an oxygen-free radical scavenger, edaravone provides neuronal protection against hypoxic-ischaemic brain damage by suppressing TRAIL and active caspase 3 protein. 57 Intriguingly, previous studies have demonstrated that the condition of oxidative stress characterized by excessive accumulation of reactive oxygen species mediates the overexpression of DR5 through key downstream transcription factors such as C/EBP-homologous protein (CHOP). 58 On the basis of this theory, various modulators, inducers and sensitizer of ROS were researched to enhance the sensitivity of TRAIL in a lot kinds of cancer cells, which is a promising research target towards TRAIL-related tumour drug resistance. 59,60 Conversely, in most cases, the primary task of the TRAIL-based therapeutics for CNS diseases is inhibiting the apoptotic pathway. Therefore, exploring an effective ROS inhibitor may be a novel and promising target compared with the traditional direct inhibition on TRAIL or TA B L E 1 A list of clinical cohort researches of sTRAIL or TRAIL mRNA as well as other relative factors as biomarkers in CNS nonneoplastic diseases its death receptors. Utilization of edaravone on hypoxic-ischaemic brain damage filled this research blank and created a precedent.
Coalition pre-conditions mentioned above with the application of TRAIL signalling pathway inhibitors may provide a comprehensive approach to ischaemic cerebral stroke (ICS) treatment.
The TRAIL and its receptors have been recognized as biomarkers in various diseases (Table 1)  Moreover, it has been discovered that TRAIL is one of the various overexpressed inflammatory mediators after spinal cord injury and the expression is induced not only by intraspinal injection of quisqualic acid (QUIS) but also by clip compression injury persisting for 30 minutes. The pathophysiological characteristic of QUIS is similar to traumatic SCI, and the peak of TRAIL mRNA expression levels was observed at the 1-to 2-week time-point following QUIS injury. 71 A relatively early expression of TRAIL was observed in another study.

| The role of TRAIL in traumatic brain and spinal cord injuries
High-intensity TRAIL protein was inclined to colocalize with oxygenase-1 (HO-1) in motor neurons after 16 hour post-operation. 72 Cantarella et al were the first to highlight the involvement of TRAIL in the inflammatory response and cellular apoptosis after SCI. In agreement with several findings concluded from SCI models, the processes mentioned above could be halted by the TRAILneutralizing antibody, therefore reduce apoptotic cells and promote neurological recovery. There is a synergizing function between glucocorticoid-induced TNF receptor superfamily-related ligand (GITRL) relative pathway and TRAIL pathway in the process of SCIrelated apoptosis. 73

| The role of TRAIL in central nervous system infection
The

| The role of TRAIL in Alzheimer's disease
Alzheimer's disease (AD) is a progressive neurodegenerative disorder resulting from complicated interactions among multiple factors that progressively lead to a deficiency in neurons and the decline of cognitive functions. 86 With the aetiology still remaining unknown, dementia is clinically characterized by memory disorder, aphasia, loss of use, loss of recognition, impairment of visuospatial skills, executive dysfunction and personality and behaviour changes. With increasing life expectancy, AD is advance irresistibly prevalent all over the world.
In present America, more than 5 million residents were affected by AD and related financial burden reached $230 billion. These two numbers were excepted to reach 13.8 million patients and $1.1 trillion by 2050. 86 Against the backdrop of increasing ageing population worldwide, multi-filed and interdisciplinary theories and approaches have been applied in clinical research with regard to pathogenesis, early biomarkers and novel therapeutic interventions. 87,88 Several studies have confirmed that amyloid-β plaques and neurofibrillary tangles are the two main pathological changes during AD. Nevertheless, recently, third core pathogenesis known as microglial-related inflammation which contributes to driving and exacerbating AD pathology has been accepted as another central mechanism. 86,88,89 Before the involvement of TRAIL in β-amyloid protein (βAP), induced neurotoxic effects were demonstrated by in vitro SH-SY5Y neuroblastoma models. 90 Plenty of evidence has proved that the c-Jun N-terminal kinase (JNK) pathway, one of the TRAILinduced non-apoptotic signalling pathways, contributes to βAP-induced death of cortical neurons by induction of apoptosis-related gene transcription. 90,91 Moreover, TRAIL is highly overexpressed in AD patients in vivo, whereas its expression has not been detected in the brains of the age-matched non-demented patients. And TRAIL protein tends to be detected in AD-affected neuron-rich cerebral cortex, mainly in the proximity of amyloid plaques. 92   the TRAIL had the potential to become the prognostic marker of treatment response to IFN-β in MS. This proposal was supported by the positive correlation between pre-treatment assessment of TRAIL expression and prognosis after IFN-β treatment. 110 Contrarily, IFN-β enhances TRAIL's inhibitory effect on T cell activation as well as its expression in vivo and in vitro. 111

| CON CLUS I ON AND PER S PEC TIVE
TRAIL, a signalling protein expressed during specific pathological conditions, plays a multifaceted role in various non-neoplastic neurological diseases mentioned above ( Figure 2). Actually, by mediating cell apoptosis, TRAIL mainly causes tissue damage ac-

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
No data, models or code was generated or used during the study.