Presently, the exact role of neuronal autophagy is still unclear when neuronal autophagy is induced by various intrinsic and extrinsic insults including ischemia/reperfusion (I/R), nutrient deficiencies, neurotoxins excitotoxic stimuli, and inflammation [13, 27, 28]. On the one hand, under these stressful conditions, mild to moderate induction of autophagy can maintain neuronal homeostasis, clear the aggregated protein and damaged mitochondria, preserve the energy balance, and alleviate the endoplasmic reticulum stress, referred to as adaptive autophagy [29]. On the other hand, excessive increases in autophagic activity might lead to enhanced degradation of essential cellular components and unrestrained induction of autophagy may be a causative factor for cell death, although the duration and extent of autophagy for cell death induction remain to be explored in depth [30, 31]. This unchecked autophagy is termed maladaptive autophagy.

Maladaptive autophagy in neurons is often triggered upon ischemic stroke and causes neuronal death [32, 33]. Many researches demonstrate that inhibiting neuronal autophagy progress significantly reduces the brain infarct volume in rat middle cerebral artery occlusion (MCAO) models. In addition, the blockade of neuronal autophagic cell death makes for the neuroprotective action of many compounds, including amlodipine, atorvastatin, lithium, N-acetyl-serotonin, and so on [34].

On the contrary, there are also many studies showing the association between autophagy and neuroprotection. Inhibition of autophagy by 3-methyladenine (3-MA), a specific inhibitor of autophagic/lysosomal protein degradation, reversed the neuroprotection of tunicamycin and thapsigargin, and enhance the I/R-induced release of cytochrome c and the downstream activation of apoptosis in neurons in oxygen glucose deprivation (OGD) and MCAO model [35].

Furthermore, many researches have supported that autophagy flux plays an important role in determining the neuroprotection or destruction aspect of autophagy. Autophagic responses and autophagy flux can be impaired in beclin 1 heterozygous mice compared to wild mice, which significantly lowered neuronal survivability and degraded neurological recovery score. Then, using rapamycin to enhance autophagy and autophagy flux on both the wild type and beclin 1 heterozygous mice. However, beclin 1 heterozygous mice did not produce any change in autophagy flux and neuronal survivability compared to untreated mice. Hence, disruption of autophagy and autophagy flux pathways is detrimental to the brain, and increasing autophagy is neuroprotective when autophagy flux is undisturbed [36, 37].

Compared to numerous studies around macroautophagy and stroke, the effect and mechanism of microautophagy and CMA in stroke are unknown, which is limited by immature technology and unidentified receptors. However, a growing body of evidence shows that CMA and microautophagy are crucial for neuronal proteostasis and the development of neurodegenerative diseases [38, 39].

Glial cells, including astrocyte, microglial, and oligodendrocyte, are the most abundant and widely distributed cells in the central nervous system (CNS), and activated glial cell autophagy has been well-proved to play critical roles in the regulation of neurological functions under stroke circumstances [40, 41].

Accumulating reports have proven that among all the cell types that can be targeted, astrocytes are undoubtedly the most promising [42], which may regulate synaptic transmission [43]. And, synaptic transmission is closely associated with neuron plasticity, playing an important part in the prognosis of patients with ischemic stroke. Meanwhile, astrocyte autophagy is common and can be divided into basal autophagy and induced autophagy under physiological and stressful conditions [44]. In the early times of ischemic stroke, activated autophagy is detrimental to astrocytes. Ischemic astrocytes in brain tissue can observe the accumulation of autophagy-like vacuoles containing electron-dense material. OGD could induce astrocyte autophagy and remarkably elevate the expression of early autophagy markers beclin 1 and light chain (LC) 3-II/LC3-I ratio. Inhibition of autophagy with 3-MA and bafilomycin AI (Baf-AI) mildly but significantly attenuated the death of damaged astrocytes under ischemic insult [45]. Said another report [46], the circular RNA HECTD1 (circHECTD1) significantly increased in ischemic brain tissue, and the knockdown of circHECTD1 restrained astrocyte autophagy and ameliorated neuronal damage. From another aspect, autophagy reportedly has a protective effect on astrocytes after ischemia and hypoxia. In a recent study, Kasprowska et al. [47] devoted to investigating whether autophagy activation is beneficial or harmful to astrocytes. After short-term OGD exposure (1, 4 h), compared with control cells, 3-MA-treated cells showed a higher level of cleaved caspase-3, showing the promotion of cell apoptosis. Nevertheless, exposed to OGD such 8 h or 24 h, astrocytes autophagy treated by 3-MA were significantly downregulated, thereby causing time-dependent inhibition of cleaved caspase-8 and tumor necrosis factor-α (TNF-α) related extrinsic apoptotic pathway and intrinsic apoptotic pathway (cleaved caspase-9) [47].

In the context of injury ischemia, studies on microglia and oligodendrocytes are limited. Microglial, the first immune defense of the CNS, is equivalent to a peripheral macrophage, involved in mediating neuroinflammation after ischemic stroke [48]. Yang et al. [49] took focal cerebral ischemia model by permanent MCAO in mice to mainly observe microglial autophagy and inflammation response in vivo. When using 3-MA (autophagy inhibitor), in addition to reduced microglial autophagy and inflammation, they also found significant reductions in infarct size, edema formation, and neurological deficits. Besides, Xia et al. [50] found an interesting phenomenon that the intensity of microglial autophagy was dependent on the time of ischemia. Activated microglia show a wide range of different functional states, mainly including pro-inflammatory (M1) and anti-inflammatory (M2) states. Moreover, inhibition of autophagic flux activated the NF-κB pathway, which is the main regulator of M1 microglial phenotype, and decreased the activity of cyclic adenosine monophosphate (AMP) response element-binding protein (CREB) participating in the polarization of M2 macrophage to modulate microglial phenotype between M1 and M2 phases.

The interaction of microglial and astrocytes jointly participates in the formation of glial scar which is the barrier for axonal regrowth and functional recovery in the late times of ischemic stroke [51], in which astrocytes are activated [52]. Wu et al. [53] reported that fused in sarcoma induced excessive autophagy and activated astrocytes which was a new target for improving the outcome after stroke. It is speculated that activated astrocytes autophagy exists which increase after stroke, nonetheless, researches on pathological change in stroke are more prevailing. If the hypothesis is true, autophagy facilitates axonal regrowth and attenuates glial scar. Transforming growth factor-beta inhibited autophagy pathway to induce the secretion of chondroitin sulfate proteoglycans of astrocytes which is responsible for glial scar [54].

Oligodendrocytes are the only cells capable of forming myelin in the CNS, which are abundant in both gray and white matter of the brain and spinal cord, especially white matter. In many clinical hypoxic-ischemic insults, white matter is the priority to be damaged. Similarly, making up the white matter myelin sheath-oligodendrocytes are also vulnerable to hypoxic-ischemic insults, so oligodendrocytes have more possibility to be explored [55]. Currently, the link between oligodendrocyte autophagy and neurodegenerative disease or spinal cord injury-related demyelination has been clarified. However, the role of oligodendrocyte autophagy in cerebral ischemia is under investigation, which is more sensitive and vulnerable to ischemic insults [56]. And hopefully, more knowledge of different aspects of oligodendrocyte autophagy will be shown in the future.

Since then, accumulating evidence has proven that autophagy regulation plays a dramatically vital role in neurodegenerative diseases and ischemic stroke. Meanwhile, endothelial cells, the major component of the blood-brain barrier (BBB), modulate themselves to maintain homeostasis by regulating the vascular tone and permeability [57]. And, ischemic stroke in the early stage is often accompanied by the disruption of BBB, and the inflammatory factors enter the brain parenchyma through BBB, which are doomed to cause irreversible damage to neurons. Above all, endothelial autophagy has the clinical potential to be a therapeutic target of ischemic stroke.

Anatomically, BBB is composed of self-fusion of brain microvascular endothelial cells (BMECs) with tight junctions (TJs) and the interaction of other components, including astrocytes, pericytes, and perivascular microglial [58]. In the pathological process of ischemic stroke, the interruption of blood flow in the brain leads to the inevitable formation of BMECs in the brain microenvironment [59]. Finally, TJ protein levels are downregulated, and BBB loses its protective function. Claudin 5 is specially involved in TJ proteins and participates in forming the backbone of TJ chain [60]. Under damaged circumstances, endothelial cells autophagy is activated to alleviate the hypoxia-induced BBB injury via regulating the redistribution and degradation of claudin 5 [6]. Induced by I/R or OGD/re-oxygenation (OGD/R), long noncoding RNAs (lncRNAs) metastasis associated lung adenocarcinoma transcript 1 (MALAT1) is one of the highly upregulated endothelial lncRNA and acts as a protective role in BMECs against cerebral ischemic insults. Similar to circRNA, lncRNA MALAT1 served as an endogenous sponge with miR-26b and downregulated the expression of miR-26b, which inhibited BMEC autophagy and survival [61].

Here, this review will give more emphasis on specific autophagy, given the significance for neuropathies, cancer, and heart disease. In ischemia, mitochondria loss the ability to produce adenosine triphosphate for energy due to the lack of nutrients and oxygen, leading to adenosine triphosphate-dependent Ca²⁺ channel injury and Ca²⁺ load of cells and mitochondria [62]. Under the circumstances, mitochondrial autophagy, termed mitophagy, is emerging to clear out dysfunction or excessive mitochondria producing reactive oxygen species and inducing cell death to control the quality and quantity of mitochondria [63]. Mitophagy, a selective catabolism mechanism [64], modulates dynamic the process of mitochondria involved with fission and fusion and completes mitochondrial turnover in stroke which has the potential to achieve pre-diagnosis and prognosis of stroke and realizes clinical value [15]. Besides, owing to the disturbance of Ca²⁺ balance, the endoplasmic reticulum stress is triggered which refers to the homeostasis disorder and dysfunction of the endoplasmic reticulum and has tight relation with autophagy [56]. Autophagy inhibitor 3-MA used in vivo and in vitro, models tended to aggravate the endoplasmic reticulum stress and increased the level of caspase-12 and caspase-3 that induced apoptosis [65]. And the mechanism of some CNS protectants is inhibiting the endoplasmic reticulum stress level and enhancing autophagy to perform neuroprotective function [66]. Peroxisomes are selectively eliminated through autophagy in yeast mediated by PpAtg30, a Pichia pastoris protein. It involves the formation of pexophagy intermediates and mediates the selection of peroxisomes, which are delivered to the autophagy machinery for pexophagy [67]. Mature ribosomes are also selectively degraded via autophagy with the help of ubiquitin protease [68]. Compared with yeast, this specialized autophagy in mammalian systems is under-studied.

In different stages of stroke including hyperacute, acute, subacute, and chronic phase, activated autophagy plays different roles during the development of stroke. It has demonstrated that induction of adenosine 3’,5’-monophosphate (cyclic AMP)-activated protein kinase-dependent autophagy is involved in the ischemic preconditioning-inducing neuroprotection [69]. What’s more, in the acute stage, activation of autophagy with ataxia-telangiectasia mutated (ATM)/cell cycle checkpoint kinase 2 (CHK2)/beclin 1 axis can relieve oxidative stress to protect neurons against injury [70]. Hence, several studies suggest that activation of autophagy at the early or before stage of ischemic stroke is neuroprotective [14]. During the acute and subacute stages, activated autophagy tends to play a detrimental role, enlarging the ischemia area and promoting cell apoptosis. At 24 h after ischemia, Nogo-A/Nogo-66 receptor 1, an autophagy inhibitor, significantly reduced ipsilateral thalamus neuronal loss and gliosis [71]. In the chronic stage, autophagy is activated excessively, which is possible to bring secondary thalamic damage to influence functional recovery. The role of autophagy in different disease stages of stroke has summarized in Figure 1.

Based on reports, some of the circRNAs promote cell autophagy. Han et al. [46] observe that circHECTD1 expressed highly and increased dramatically in the transient MCAO group, coincident with the circRNA microarray and heat map. Served as an endogenous sponge, upregulated circHECTD1 can interact with miR-142, whose knockdown inhibited astrocyte activation induced by OGD-R in vivo and in vitro. And through the forecast, the TIPARP has a conserved miR-142 binding site within its 3’-untranslated region (UTR) in most species, which is the downstream target of the circHECTD1/miR-142 axis and promotes astrocyte activation proved by experiments. Meanwhile, the circHECTD1/miR-142 axis promoted astrocyte autophagy via downstream TIPARP, leading to activated astrocyte enhancement. In conclusion, circHECTD1 acted as a sponge of miR-142, and upregulated TIPARP expression, activating astrocytes via enhancing autophagy [46]. Besides, another report said, upregulated circular RNA FoxO3 (circFoxO3) activates BMECs autophagy. Mechanistic target of rapamycin complex 1 (mTORC1), a complex formed with mTOR, Raptor, and mammalian lethal with SEC13 protein 8 (mLST8) is involved in the negative regulation of autophagy. circFoxO3 indirectly regulates mTORC1 in two different ways. One way is that circFoxO3 competitively sequestered mTOR to inhibit mTORC1 activity, promoting autophagy activation. Under normal physiological conditions, E2F transcription factor 1 (E2F1), a transcriptional regulator of mTORC1, is obligated to enter the nucleus and bind to DNA for activation of 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3 (PFKFB3) transcription, facilitating the translocation of mTORC1 to lysosomes with the assistance of PFKFB3. However, the interaction between circFoxO3 and E2F1 blocked the pathway of E2F1 entering the nucleus, preventing its transcriptional regulation. Through the above two ways, mTORC1 activity is suppressed, promoting autophagy activation [79].

Moreover, the insight of Chen et al. [75] is focused on the role of the ischemic-preconditioned astrocyte-derived exosomes (IPAS-Exos) against ischemic stroke. They found that IPAS-Exos played a protective role in neuroprotection both in vitro and in vivo, with a decrease in the ratio of LC3-II/LC3-I and beclin 1 (two autophagy-related proteins) and an increase of p62. All the data clarified IPAS-Exos inhibited ischemia-induced neuronal death by regulating autophagy. How did IPAS-Exos regulate autophagy to protect neuronal cells? Through circRNA microarray, they found that circSHOC2 was highly expressed in IPAS-Exos and worked as a neuron protector. Mechanically, circSHOC2 acted as a molecular sponge for miR-7670-3p in neurons, inhibiting miR-7670-3p activity, thereby increasing the expression of SIRT1. In recent years, the downstream target molecule SIRT1 has been found to induce autophagy by regulating the tuberous sclerosis complex subunit 2 (TSC2)/mTOR/S6 kinase 1 (S6K1) signaling pathway, indicating the anti-autophagic role of circSHOC2 [75].

In the mice and cells with the I/R model, the expression levels of circ_016719 and mitogen-activated protein kinase 6 (Map2k6) were significantly increased and miR-29c was decreased. What’s more, the knockdown of circ_016719 expression reversed the increased expression of autophagy proteins, showing that circ_016719 regulates autophagy in ischemia by circ_016719/miR-29c/Map2k6 signaling pathway [93]. Additionally, Xu et al. [94] showed that circular RNA Akap7 (circAkap7) played a protective role in transient MCAO in mice and knockdown of ATG12 reversed the therapeutic effects of circAkap7, showing that circAkap7 attenuates I/R-induced cellular injury by promoting ATG12-mediated autophagy.

circRNAs not only influence disease development by promoting autophagy but also regulate by inhibiting autophagy [93]. In MCAO rats, circ_0025984 acts as a molecular sponge and absorbs miR-134-3p in astrocytes, and decreases the expression of ten-eleven translocation 1 (TET1). After translocating into the nucleus, TET1 can bind to the promoter of 150-kDa oxygen-regulated protein (ORP150) and converts 5-methylcytosine (5mC) into 5-hydroxymethylcytosine (5hmC), leading to DNA demethylation and increased expression of ORP150, which eventually activates autophagy mediated by ATG7. Thereby, circ_0025984 protects astrocytes from ischemia-induced autophagy by targeting the miR-143-3p/TET1 pathway and might inhibit cerebral injury induced by ischemic stroke [95].

A summary of the functions of different autophagy-related circRNAs in ischemic stroke has been represented in Figure 2.