Molecular chaperones in the brain endothelial barrier: neurotoxicity or neuroprotection?

Abstract

Brain microvascular endothelial cells (BMECs) interact with astrocytes and pericytes to form the blood–brain barrier (BBB). Their compromised function alters the BBB integrity, which is associated with early events in the pathogenesis of cancer, neurodegenerative diseases, and epilepsy. Interestingly, these conditions also induce the expression of heat shock proteins (HSPs). Here we review the contribution of major HSP families to BMEC and BBB function. Although investigators mainly report protective effects of HSPs in brain, contrasted results were obtained in BMEC, which depend both on the HSP and on its location, intra- or extracellular. The therapeutic potential of HSPs must be scrupulously analyzed before targeting them in patients to reduce the progression of brain lesions and improve neurologic outcomes in the long term.—Thuringer, D., Garrido, C. Molecular chaperones in the brain endothelial barrier: neurotoxicity or neuroprotection?

In response to various exogenous stresses, brain cells activate endogenous stress responses. The most general answer may be heat shock response (HSR). HSR is triggered by a wide variety of stress, including oxidative and metabolic stress, ischemia and reperfusion, gliomas, and neurodegenerative diseases. HSR is characterized by an almost exclusive synthesis of heat shock protein (HSP) families. HSPs are chaperones or cochaperones whose expression is intended to attenuate the accumulation of denatured cytoplasmic proteins known to form in response to the stressors listed above (1–4). The main families of chaperones are classified according to their MW: Hsp40 (DNAJA, B), Hsp60 (HSPD1), Hsp70 (HSPA1A, B), Hsp90α, β (HSP90AA1, HSP90B1), Hsp110 (HSPH1), and small HSPs to which belong Hsp27 (HSPB1) and αB-crystalline (HSPB5) (5, 6). Most chaperones require the help of cochaperones to increase protein association rates, ATPase activity, and nucleotide exchange. Hsp60, Hsp70, Hsp90, and Hsp110 each use ATP binding and hydrolysis to induce binding, folding and release cycles of a protein (Fig. 1A). The small HSPs, which are ATP-independent, work in refolding reactions by forming large oligomers capable of storing proteins in an aggregation independent state. Most HSPs are up-regulated under the same stress conditions but they perform different functions.

Figure 1.

The HSP chaperone system at the BBB level. A) Collaboration between HSPs in protein folding. Under nonstressful conditions, HSPs bind to newly synthesized polypeptides and facilitate their folding (folded protein), whereas under stressful conditions they recognize denatured proteins and prevent their premature degradation. 1) Hsp40 transiently associates with an altered or misfolded protein to deliver it to Hsp70. 2) Hsp40 binds to Hsp70 and promotes its ATPase activity, thereby generating ADP-linked Hsp70, which stably interacts with the misfolded protein. After nucleotide exchange (replacement of ADP with ATP), the protein is released from Hsp70 and leaves the chaperone system as a correctly folded protein. 3) During stress, Hsp70, in its ADP binding form, encompasses denatured protein. 4) The renaturation of the protein by Hsp70 requires the participation of cochaperones such as Hsp110 or Hsp90. In all cases, the replacement of ADP with ATP is necessary to release the correctly folded protein from these HSP complexes. 5) Some proteins require a subsequent folding step by a group of chaperones having a common structure, chaperonins: for example, Hsp60/Hsp10 present in mitochondria. 6) Under conditions where the system is saturated by a massive influx of proteins to replicate, a transient storage of these is carried out by oligomeric structures formed by Hsp27. These reservoirs of folding intermediates would then direct the proteins to the different protein steps of folding machinery. B) Interactions between astrocytes, pericytes, BMECs, and NSCs/NPCs ensure adequate cerebral blood flow. BMECs are tightly connected through transmembrane proteins present in the tight junctions (occludin), gap junctions (connexin, Cx), and adherens junctions (cadherin). These junction structures interact with cytosolic adaptor proteins that link junctions to the actin cytoskeleton, leading dynamic structures that respond to the local microenvironment. In stressful conditions, HSPs are overexpressed in and released by all cells in the brain. HSPs modulate the expression and/or activity of MMPs involved in basal lamina degradation; for example, MMP-9, which contributes to the disruption of BBB causing cerebral hemorrhage, is decreased by Hsp70 but increased by Hsp90α. Extracellular HSPs bind to TLRs and stimulate specific signaling pathways in the different cell types. In BMEC, these events result in actin polymerization (actin-F), actomyosin contraction (tension), junctional remodeling, and disruption of intercellular contacts, leading to BBB permeability increase. They also signal to adult niches of NSCs/NPCs to increase proliferation and differentiation in healthy or diseased cells. These processes allow transendothelial migration of leukocytes and can be induced by leukocyte secretions themselves. For example, Hsp70 released by M2 monocytes disrupts interendothelial communication by decreasing endothelial expression of Cx43 (67).

Defining the role of HSPs in the normal and pathologic brain is complicated by the diversity of both cell types and modes of cell-to-cell communication (7). For instance, overexpression of Hsp70 is generally associated with increased neuronal resistance to cerebral ischemia and better recovery from it (8). Endothelial cells, in particular, generate a stress response with induction of Hsp70 after cerebral ischemia (9). However, this Hsp70 response coincides with the death of endothelial cells and pericytes (10). More recently, extracellular vesicles (EVs) released from endothelial cells have been reported to contain Hsp70 and directly protect neurons from ischemia/reperfusion injury by activating their growth, migration, and invasion and by inhibiting their apoptosis (11). EVs carry not only HSPs but also small noncoding microRNAs (miRs) potentially related to the expression of hsp genes in the recipient cell (12–15). This review will attempt to answer the following questions: Can we define a specific role of each Hsp in the blood–brain barrier (BBB)? Are molecular chaperones toxic or protective?

HSP EXPRESSION AT THE BBB LEVEL

The BBB is a continuous monolayer of brain microvascular endothelial cells (BMECs) and is sheathed with pericytes and perivascular astrocytes end-feet (Fig. 1B). In addition to pericytes, undifferentiated or differentiated neural stem cells (NSCs)/neural progenitor cells (NPCs) also contribute to the induction and maintenance of the BBB (16). Transdifferentiation of NSCs/NPCs into BMEC has been described, and conversely, BMECs may generate NSCs/NPCs (17). The microvascular network is very close to each neuron (∼20 µm of a microvessel) promoting the transmission of BMEC secretory products to neurons and vice versa (18).

During hyperthermia (higher than 40°C), the victims suffer from cognitive deficits and dementia, associated with BBB rupture, microhemorrhage, cerebral edema, and cell death (19) as also observed in Alzheimer’s and Parkinson’s diseases, schizophrenia, or ischemia-associated damage (20, 21). Very few reports have focused on BMEC’s HSR. Exposure of BMEC cultures to thermal shock results in both tight junction disassembly and overexpression of Hsp70, Hsp90, and Hsp110 (22). Recovery to 37°C induces overexpression of Hsp47 in the first hours. This latter expression was concomitant with the reassembly of tight junctions between BMEC in vitro. Such an increase was also reported in gliomas, during tumor angiogenesis in vivo, where the newly formed vessels have morphologic and functional abnormalities (leaks and anastomoses) related to the disassembly of tight junctions between BMECs. By expressing Hsp47, pericytes promote angiogenesis by producing the basal lamina of microvessels but do not induce re-expression of tight junction between BMEC and thus the establishment of the BBB (23, 24). Interestingly, sequential expression of HSPs has also been reported in primary vascular endothelial cell cultures exposed to microgravity as observed aboard an orbital station (25). The adaptive endothelial response resulted in early up-regulation of Hsp70, which was progressively replaced by the increase of other stress proteins, including Hsp27. The early increase in Hsp70 appeared crucial in this adaptation process because its inhibition (by small interfering RNA) induced cell death. Thus, the sequential expression of HSPs seems first to quickly adapt the cells to stress conditions, then to establish a new homeostatic status to preserve the viability and function of cells. It should be noted that a mild hyperthermia (38.5°C) increased NSCs/NPCs proliferation and neuronal differentiation through an initial and brief increase in Hsp70 followed by the more sustained increase in Hsp27, whereas no significant effect was reported on Hsp90 expression (26). In contrast, a mild hypothermia after a heat stress decreases Hsp70 expression in NPCs, as reported during the differentiation of NSCs/NPCs in brain cells (27).

Overexpression of Hsp27 in BMECs preserves the integrity of BBB by suppressing aberrant actin polymerization, stress fiber formation, and translocation of junctional proteins (28). Structural aberrations in BMECs have been described during ischemic stroke and would be responsible for the early rupture of BBB, occurring within 30–60 min following reperfusion after focal ischemia (29). It is worth noting that Hsp70 was overexpressed in a 300-μm-wide area surrounding the BBB disturbance zone. Matrix metalloproteinases (MMPs) involved in basal lamina degradation can lead to cerebral hemorrhage. Hsp70 expression decreases MMP expression and proteolytic activity in a traumatic brain injury model (30). As a result, cerebral hemorrhage decreases and neurologic function improves. Conversely, Hsp70 deficiency results in increased lesions, brain hemorrhages, and MMP activity. Studies on brain tissue and serum in patients with ischemic stroke demonstrate that MMP9 contributes to the disruption of BBB (31). Hsp90, the isoform α, is overexpressed and correlated with MMP9 levels in patients, and its inhibition by 17-dimethylaminoethylamino-17-demethoxygeldanamycin strongly attenuates the BBB disruption (32). The pharmacological inhibition of Hsp90 resulted in up-regulation of Hsp70 and Hsp40, which could control the expression of several synaptic proteins but could also drive the misfolded protein to proteasome degradation (33–36). Another chaperone involved is Hsp110, which interacts with Hsp70 and increases its ATPase activity (37, 38). Hsp110 is known to suppress the aggregation and folding of proteins and to protect them from the adverse effects of various stresses. Nevertheless, its physiologic function seems more complex. The absence of hsp110 gene in mice is associated with hyperphosphorylation of tau microtubule protein and accumulation of β-amyloid (Aβ) plaques, as described in Alzheimer’s disease (AD). The activity of Hsp110 is also present at the synaptic level and is recognized for its protection against the aggregation of α-synuclein in Parkinson’s disease (37). Thus, Hsp110 is critical in various neurodegenerative diseases called tauopathies (39). Unlike Hsp110, Hsp90 facilitates the aberrant aggregation of tau, and increases tau toxicity and Aβ metabolism (40). In addition, by modifying the structure of eNOS, Hsp90 affects the functions of eNOS in particular by changing its affinity to cofactors such as Ca²⁺ and l-arginine, and, subsequently, NO generation capacity (41). This results in phosphorylation of tau (p-tau) in neuronal tissue (42). Uncoupled eNOS in BMECs produce an excess of superoxides that react with NO produced by microglia or astrocytes to form peroxynitrite, in turn increasing the accumulation of p-tau in astrocytes (43). These Hsp90 mechanisms link endothelial dysfunction to the pathogenesis of AD. Hsp90 inhibitors have proven their efficiency in reducing tau pathology (40). However, these Hsp-modulating chemicals may affect cellular targets in cells other than the chaperones themselves, such as the many Hsp90 client proteins, for example. Some work reveals intriguing roles of Hsp90 in transcription regulation (44), but its contribution in tauopathies has proven to go beyond individual transcription factors (45).

When expressed on the endothelial cell surface, HSPs constitute possible autoantigens in the vessel wall. Such is the case of Hsp60. This chaperone participates in the folding of mitochondrial proteins and facilitates the proteolytic degradation of denatured proteins. Hsp60 is also detected on the membrane surface of stressed endothelial cells, and the antibody against Hsp60 induces endothelial cytotoxicity especially in cerebrovascular disease (46). Mutations in the Hsp60 gene have been associated with neurodegenerative disorders (47). An abnormal level of Hsp60 expression and subcellular localization was also detected in inflammatory diseases and gliomas (48). Moreover, extracellular Hsp60 increased secretion of proinflammatory factors by microglia, causing neuronal death (49–52). Concerning Hsp70, it has been reported to be overexpressed in primary glioblastoma (GBM) cells, whereas neighboring endothelial cells, isolated from GBM tissues, did not (53). More recently, Hsp90 has been reported to mediate the fusion of multivesicular bodies (MVBs) with the plasma membrane, leading to release of EVs (54). This activity requires an amphipathic α-helix that becomes exposed when the protein dimer is open. The membrane-binding amphipathic helix is a common motif encountered in various peptides and proteins such as Hsp90 (55, 56). Hence, by controlling the open vs. closed state of the Hsp90 dimer, cells can regulate MVB fate: open Hsp90 promotes the vesicular release, whereas closed Hsp90 blocks the process (54).

MODULATION OF BBB FUNCTION BY EXTRACELLULAR HSP

HSPs are released outside the cells where they display a function different from that of intracellular chaperone (Fig. 2A). Extracellular HSPs could be alert stress signals for other cells to prevent the spread of insult and promote repair (57, 58). Plasma membrane translocation, lipid vesicle–associated release, and passive postdeath release of necrotic cells have been proposed to explain the presence of HSPs in the extracellular medium (59). Extracellular HSPs take several forms: membrane-bound or not, associated with client protein or not (60). The diversity of their forms is probably associated with various interactions with cells, both in terms of target cell types and activated signaling pathways (59, 61, 62).

Figure 2.

Expression of HSPs outside the cell. A) Mechanisms for the cell release and uptake of HSPs. 1) Conventional exocytosis of Hsp (the stress-induced form) or heat-shock cognate (Hsc; the constitutively synthesized form) in the extracellular space, so that they are present in the extracellular fluid. 2) An unknown mechanism allows Hsp/c to concentrate in the small vesicles (exosome). 3) Exocytosis of the MVB (exosomes within a larger vesicle) resulting from autophagy. 4).Conventional apocrine secretion, in which small vesicles (ectosome) containing cytoplasmic constituents bud off the cell surface. The released vesicles (ectosome and exosome) can then interact with recipient cells (BMEC) in 2 hypothetical ways: either to fuse with the plasma membrane and release its contents or to be phagocytosed, forming another MVB. Soluble extracellular Hsp interacts with plasma membrane receptors or is internalized into the cytoplasm of the recipient cell by endocytosis, where it is free to interact with other cytoplasmic components. B) Hypothetical HSP expression model at the endothelial cell surface that constitute possible autoantigens in the vessel wall. Membrane associations and lipid interactions are reported for small HSPs, Hsp60, Hsp70, and Hsp90. 1) Excess HSPs are routed to MVBs where HSPs are linked to the inner leaflet of the vesicular membrane. The mechanisms of HSPs supply of the endolysosomal system are not known, but would involve mechanisms of autophagy and direct membrane crossing. 2) Fusion of MVB with the plasma membrane and release of exosomes containing HSPs. 3) HSPs are exposed to the cell surface (131). 4) By a conventional apocrine secretion, HSPs are found on the outer leaflet of ectosomes. Thus, HSPs could be expressed on the surface of the plasma membrane and ectosomes (104).

Receptor-mediated signaling by free extracellular HSP has been controversial because of the fact that recombinant preparations of HSP were contaminated with bacterial products (i.e., LPS, flagellins, lipoproteins, and DNA). These contaminants can activate cells via pattern recognition receptors, including TLRs, which result in activation of NF-κB and high production of various chemokines (63), similar to the response observed with HSP (64). Attention has been given to possible endotoxins and, more specifically, to LPS contamination, given the implications of TLR2 and/or TLR4 as suspected HSP receptors. The debate over possible endotoxin contamination is due to the ability of HSP to sequester LPS and protect it from the methods used for its detection or inactivation (i.e., proteinase or boiling). Nevertheless, the recent use of recombinant HSPs produced by mammalian cells has allowed the emergence of the notion that surface receptors are indeed capable of recognizing extracellular HSPs or even fragments of HSPs. The first receptor recognized is CD91 (or LDL receptor–related protein), which binds Hsp60, Hsp70, and gp96 (61, 65). Despite earlier controversy, TLR2 and TLR4 are still considered candidate receptors for Hsp60, Hsp70, Hsp90α, and gp96 (66–71). TLR3, which specifically recognizes double-strand RNA derived from virus and dead cells, is stimulated by Hsp27 (72, 73). These studies ascribed new functions to HSPs. For example, extracellular Hsp90α stimulates wound healing and metastasis of GBM cells (66). Extracellular Hsp27 also influences the behavior of surrounding cells (74). At the microvascular level, this extracellular chaperone promotes tumor angiogenesis by increasing VEGF in endothelial cells (72). Extracellular HSPs have been reported as pro- or anti-inflammatory (1). Surface-bound Hsp70 specifically activates NK cells, whereas Hsp70 when released into the extracellular medium exerts immunoregulatory effects, including up-regulation of adhesion molecules and release of cytokines and chemokines [a process called “chaperokine” activity of HSP (61, 75, 76)]. Hsp70, exogenously added or released by circulating monocytes, blocks the gap-junctional coupling between endothelial cells, reducing the transendothelial migration of monocytes (67). It is well known that the immune system contributes to neuronal cell death because of microglial activation, leukocyte recruitment, and cytokine secretion (77). The involvement of extracellular HSPs could be seen as an attempt to correct the inflammatory condition. Furthermore, extracellular Hsp70 interacts with Alzheimer’s Aβ peptides, arresting their oligomerization in fibers and plaques that are neurotoxic (36, 78–80). Thus, the final outcome seems to depend on the microenvironment.

HSPs are transferred from cell to cell via cell-derived EVs. EVs are released from all types of brain cells, healthy or diseased (81), and are bidirectionally transported through the BBB (7, 82–85). Depending on their subcellular origin, EVs are classified into 2 main categories: ectosomes (also called microparticles) that are released after budding from the plasma membrane as well as exosomes, which are produced inside the MVBs and released after fusion of the MVBs with the plasma membrane (86, 87). It should be noted that apoptotic bodies (∼800–5000 nm diameter) that are released from apoptotic cells exhibit ectosome characteristics but are rarely described in intracellular communication because they would be rapidly destroyed by phagocytes (87). Little is known about EVs derived from BMECs (12, 85, 88–98). Secreted by BMECs, EVs are heterogeneous in size and include ectosomes (∼150–1000 nm diameter) and exosomes (∼20–150 nm diameter) (99, 100). Their protein profiles reveal that ectosomes and exosomes share about one-third of their proteins (88). The remaining proteins are specific to the EV type (e.g., 544 proteins in ectosomes and 209 proteins in exosomes). HSPs are present in both types of EVs. Constitutively expressed heat shock members [heat-shock cognate (Hsc)] are present in almost all intracellular compartments. It is therefore not surprising to find them in both types of EVs. Inducible HSP forms (Hsp) are, however, described as a result of cellular stress so that their presence in EVs may reflect the state of stress of the secretory cell. It is easy to imagine that HSPs contribute to the correct folding and assembly of the polypeptide of newly synthesized proteins and are expected to play cytoprotective roles when transferred to the target cell. It has been reported that EVs isolated from BMEC under normal or ischemic conditions have diametrically opposite effects (101). Injected in a mouse model of transient cerebral ischemia, the normal EVs decreased the permeability of BBB by overexpressing tight junction proteins in BMEC, and improved local cerebral blood flow, reducing the volume of the infarcted area (101). In contrast, ischemic EVs favored the propagation of the insult. What would be the role of HSPs in this network of organelles? In the tumor context, Hsp60, Hsp70, and Hsp90 are secreted by healthy and cancerous cells via EVs but have opposite effects, suppressors or stimulators, on tumor cell growth (13, 102–104). Hsp27, αB-crystalline, and Hsp20 are also exported by exosomes with neuroprotective effects (105).

HSPs are also present in vesicular membranes (Fig. 2B). Their membrane location is difficult to interpret (106). For instance, Hsp60 is expressed on the surface of exosomes secreted by tumor cells and its depletion in in vivo models of GBM leads to the suppression of intracranial tumors (104). What are the interactive partners of HSPs at the membrane level? Rab GTP hydrolase [i.e., Rab1 involved in endoplasmic reticulum (ER)-to-Golgi and Rab3A transport involved in synaptic vesicle fusion] require Hsp90. This observation has led the authors to propose that Hsp90 could serve as a general regulator for the recycling of Rab GTP hydrolase in the exocytic and endocytic trafficking pathways involved in cell signaling and proliferation (107). The vesicle trafficking between Golgi stacks is dependent on the ATPase function of Hsp90 and can be inhibited by Hsp90-specific drugs (108). More recently, Hsp90 has been reported to release exosomes (54).

HSP OVEREXPRESSION: NEUROTOXIC OR NEUROPROTECTIVE?

Microvascular endothelial responses occurring after acute neurologic insults usually reduce pathogenic processes in the brain (28, 109, 110), although BMEC could also synthesize neurotoxic proteins, such as thrombin in AD (111). Are they related to specific HSP expression? Very few studies have analyzed the contribution of HSPs to cerebral endothelial functions. Nevertheless, we know that the suppression of some HSPs exacerbates neuronal cell death or, conversely, improves recovery (Table 1).

TABLE 1.

HSPs in CNS diseases

Type	Function	Biologic activity	Pathologic	Preventive	AUC (au)	Reference
Hsp27	Antiaggregation	Sustained angiogenesis (TLR3) VEGF release	GBM growth			72, 147
	Antioxidative	Anti-inflammatory effect		Cortical spreading depression		148, 149
	Antiapoptotic	BBB integrity (anti-actin polymerization)		I/R-induced neurovascular injury		28
	Thermotolerance	Resistance to chemotherapeutics	GBM growth		0.871	150, 151
		Prevent tau accumulation		Neurodegeneration (AD, PD, MS)		152, 153
		Neurite outgrowth, NSC/NPC differentiation		Neurodegeneration (AD, PD, MS)		26, 153
Hsp40	Protein folding	Hsp40/Hsp70 complex	Meningioma			154
	Cochaperone of	Release of neurodegenerative proteins	AD, PD			155, 156
	Hsp70 promoting
	its ATPase activity
Hsp60	Mitochondrial	Neuroinflammation (TLR4)	GBM, epilepsy, MS, ASD			3, 157, 158
	Protein folding	Proliferation (mTOR pathway)	GBM			158
Hsp70	Protein folding and	Neuro-inflammation (TLR4)	ASD		0.987	159, 160
	Membrane transport		Meningioma, GBM		0.779	161
	Antiapoptotic	BBB disruption	ICH			162
	Immunomodulatory	Transendothelial cell migration	ALS		0.826	163
		Neuroinflammation (TLR2/TLR4)	Ischemic stroke			68
		Invasion, endocytosis	AD, glioma			131
		Anti-inflammatory and neuroprotective		AD, PD, HKD		112, 115, 123
HSP90α	Folding of many	Cell proliferation	GBM			164
	regulatory proteins	Migration (TLR4)	Cell invasion, metastasis			66, 165
	Interaction with	Membrane deformation		Hyperthermia		166
	signaling pathways	Fusion of MVBs with plasma membrane	Propagation (exosome)			54
	Exosomal secretion
HSP110	Protein	Reduce injury at the impact site		Traumatic brain injury		167
	Disaggregation	Suppress cancer cell apoptosis	GBM growth			169
	Stress tolerance	Prevent tau accumulation		Neurodegeneration (AD, PD, MS)		168
		Increase levels of BDNF		Depression		170

List of HSPs examined and their main chaperone functions, their biologic activity, including those interacting with TLRs, their involvement in CNS diseases (pathologic or preventive), and the AUC determined from biopsies fluid (plasma, cerebrospinal fluid, or both) that reflects the reliability of the HSP assay as biomarkers for early diagnosis in patients (3, 28, 54, 66, 68, 131, 147–170). ALS, amyotrophic lateral sclerosis; ASD, autism spectrum disorders; au., arbitrary units with values from 0 to 1; BDNF, brain-derived neurotrophic factor; HKD, Huntington’s and Kennedy’s diseases; ICH, intracerebral hemorrhage; IR, ischemia-reperfusion; MS, multiple sclerosis; PD, Parkinson’s disease.

The protective effects of brain HSPs have been reported in various models of brain diseases (2, 28, 112–114). For example, overexpression of Hsp70 appears to be anti-inflammatory and to play a neuroprotective role in neurodegenerative diseases (such as AD and Parkinson’s disease) (112, 115). Its beneficial effects may be due to both its chaperone role and its ability to protect neuronal cells against various types of toxic factors (116). Brain ischemia can be modeled in vitro by exposing cell cultures to oxygen and glucose deprivation or excitotoxin exposure (117). Cultured hippocampal cortical neurons from transgenic mice overexpressing Hsp70 were protected from excitotoxin exposure and oxygen and glucose deprivation (118). Isolated glial cultures of this same mouse strain were also resistant. Conversely, the down-regulation of Hsp70 with an antisense oligonucleotide aggravated the damage following thermal shock. Although Hsp70 transgenic mice are protected against myocardial ischemia, this is not always the case for cerebral ischemia (119). The reasons for these discrepancies may be due to differences in transgene expression in different organs, or to the varying susceptibilities of brain regions. Some groups have studied the permanent arterial occlusion in various strains of Hsp70 transgenic mice with contradictory results. These different results may be due to transgenic animal limitations, such as alterations in the development of other biochemical systems caused by overexpression of transgenes. The extent of Hsp70 expression could also be different between transgenic strains because different promoters have been used (120–122). After occlusion of the cerebral artery, overexpression of Hsp70 improves the survival of the striatal neurons only during the first 2 h following the aggression. Nevertheless, these observations confirm that HSPs protect neuronal cells by their chaperone function, possibly by preventing protein aggregation. This is also the case in polyglutamine diseases (Huntington’s and Kennedy’s disease, certain spinocerebellar ataxic disorders) (123). This heterogeneous group of neurodegenerative diseases is characterized by a cellular accumulation of glutamine aggregates, resulting in cell death (124). Overexpression of Hsp70 in spinocerebellar ataxia models has been reported to significantly reduce protein aggregation and cytotoxicity as well as to improve behavioral test scores (125).

However, overexpression of Hsp70 is not protective in all cases. If overexpression of Hsp70 protects hippocampal and cortical neurons against shock or thermal ischemia, it does not protect against the toxicity of glutamate or 3-nitropropionic acid (126) or apoptotic stimuli (127). Glial cultures derived from Hsp70-overexpressing transgenic mouse brains are resistant to hydrogen peroxide damage but not to other stresses (128). Interestingly, hippocampal neurons are more resistant to stress than cortical neurons. Thus, Hsp70 protects against certain types of brain lesions but not all, and its protective effects may be related to the nature and severity of the attack. HSPs can be pathogenic factors (3). Associated with cancer, there are “chaperonopathies by collaborationism” in which the involved chaperone improves the survival and growth of tumor cells by inhibiting apoptosis and the anti-tumor immune response or by promoting neoangiogenesis. There is evidence of a positive correlation between HSP levels and the progression of gliomas, suggesting the use of anti-HSP antibodies in anti-tumor vaccines. Nevertheless, studies involving a large number of patients are needed to clearly define the relationship between HSPs and the aggressiveness of tumors. The inhibition of Hsp90 offers a dual therapeutic approach in aging and neurodegenerative diseases (129): on the one hand, it increases the protein hyperphosphorylation and their subsequent aggregation by reducing the activity of aberrant proteins; on the other hand, it can protect against the toxicity of proteins via the induction of Hsp70 and other chaperones.

Some HSPs are used as biomarkers of brain-related diseases [see area under the curve (AUC) in Table 1], including early diagnosis of brain tumors (88, 130, 131). Cotransported with neurotoxic or misfolded proteins in EVs, these chaperones contribute to the spread of their neurotoxicity and to the pathogenesis of chronic neurodegenerative diseases (2, 14). There are concepts about the role of chaperones in the organelle network and their regulation by miRs (8). Indeed, EVs also contain small, noncoding miRs, which regulate many aspects of BMEC biology, such as angiogenesis, inflammation, and BBB function, with clinical consequences (132–138). Interestingly, endothelial EVs contain neprilysin, a zinc-dependent metalloprotease, also called Calla for Common Acute Lymphoblastic Leukemia Antigen (139). This enzyme cleaves peptide hormones, such as enkephalins, substance P, and neurotensin, and also degrades Aβ peptide (140). This emphasizes the therapeutic potential of endothelial EVs in BBB.

CONCLUDING REMARKS

The cerebral microenvironment includes many types of resident or infiltrated cells. Resident cells have received the most attention so far. Nevertheless, the influence of their microenvironment in the progression of cerebral pathologies remains poorly understood. An imbalance in the cellular production of HSPs contributes to various neuropathies. Whether the chaperone system is directly or indirectly involved in these pathogenic processes must be determined. There is sufficient evidence to encourage the development of therapeutic strategies targeting HSPs, either to block their activity if they promote the progression of the disease, or to improve their performance when they are protective (6, 116, 141). However, the design of this type of drug is hampered by the fact that HSPs are actually more than cytosolic chaperones and should be considered as important regulators of signaling pathways. Their membrane localization and the interactions they maintain with lipids are associated with the facilitation of endocytosis or the release of exosomes and antiapoptotic mechanisms in cancers, for example (81). Thus, the membrane-HSP mechanistic interactions must be decrypted to reduce the toxicity of drugs. Future studies should also take into account the neurogenic niches closely linked to BMECs. In the adult brain, NSCs/NPCs contribute, directly or as a source of proangiogenic factors, to the re-endothelialization of the BBB during brain injury (17, 142). The metabolic activity and function of these neural cells are tightly regulated by many systemic and local factors, such as as HSPs, and can generate cancerous endothelial cells or metastasis, spreading glioma (17, 143). Given the importance of the sequential expression of HSPs in stressed BMECs and in the differentiation of NSCs/NPCs, it will be essential to show which HSPs contribute to functional neurons, astrocytes, or oligodendrocytes, capable of repairing lesions in the adult brain or of inducing cerebral angiogenesis (144). The answer to these questions will be beneficial for the implementation of new therapeutic concepts based on biomaterials and the use of HSPs in the treatment of cerebrovascular diseases, brain trauma, neurodegeneration, brain tumors, or neuroinfection (15, 145, 146).

Glossary

Aβ

β amyloid

AD

Alzheimer’s disease

BBB

blood–brain barrier

BMEC

brain microvascular endothelial cell

ER

endoplasmic reticulum

EV

extracellular vesicle

GBM

glioblastoma

HSP

heat shock protein

HSR

heat shock response

miR

microRNA

MMP

matrix metalloproteinase

MVB

multivesicular body

NPC

neural progenitor cell

NSC

neural stem cell

AUTHOR CONTRIBUTIONS

D. Thuringer conceived, drafted, and revised the manuscript and figures; and C. Garrido revised the manuscript.