Abstract
Although heat shock proteins are thought to function primarily as intracellular chaperones, the release and potential extracellular functions of heat shock proteins have been the focus of an increasing number of studies. Our particular interest is HspB1 (Hsp25/27) and as astrocytes are an in vivo source of HspB1 it is a reasonable possibility they could release HspB1 in response to local stresses. Using primary cultures of rat cortical astrocytes, we investigated the extracellular release of HspB1 with exposure to amyloid-β (Aβ). In order to assess potential mechanisms of release, we cotreated the cells with compounds that can modulate protein secretion including Brefeldin A, Methyl β-cyclodextrin, and MAP kinase inhibitors. Exposure to Aβ (0.1, 1.0, 2.0 μM) for 24–48 h resulted in a selective release of HspB1 that was insensitive to BFA treatment; none of the other inhibitors had any detectable influence. Protease protection assays indicated that some of the released HspB1 was associated with a membrane bound fraction, and analysis of exosomal preparations indicated the presence of HspB1 in exosomes. Finally, immunoprecipitation experiments demonstrated that the extracellular HspB1 was able to interact with extracellular Aβ. In summary, Aβ can stimulate release of HspB1 from astrocytes, this release is insensitive to Golgi or lipid raft disruption, and HspB1 can be found either free in the medium or associated with exosomes. This release suggests that there is a potential for extracellular HspB1 to be able to bind and sequester extracellular Aβ.
INTRODUCTION
Heat shock proteins (HSPs) are a family of chaperone proteins that can be upregulated in response to various cellular or environmental stressors. The particular HSP family member induced may differ among different cells and in response to differing stresses. These proteins can act as cellular chaperones promoting appropriate protein folding and removing misfolded or aggregated proteins [1–4]. We have been investigating the small heat shock protein, HspB1 (Hsp25/27), which unlike some of the other HSPs, functions in an ATP-independent manner and does not participate in protein folding per se [5, 6]. HspB1 can play a protective role in neurons but its effects may differ from those of Hsp70 and other HSPs (reviewed in [7–10]). HspB1 can act as a chaperone enabling the sequestration of misfolded proteins, but is also important in regulating cytoskeletal elements [11–14]. HspB1 has been noted to be increased in AD brains along with accumulation of HSPs in plaques, neurofibrillary tangles, and Lewy bodies [3, 15]. We have previously reported that HspB1 can protect cortical neurons from the deleterious effects of Aβ exposure in vitro [16] pointing to a possible protective mechanism. In a recent study, we explored the potential effects of the HspB1 on amyloid-β precursor protein (AβPP) processing and distribution within HEK293 stable cell lines expressing either AβPPwt or AβPPsw. Expression of HspB1 was observed to alter AβPP expression and processing in these cell lines, and furthermore, the presence of HspB1 decreased the amount of amyloid-β (Aβ)42 released by the cell lines [17].
Despite the lack of endogenous neuronal expression, a number of studies have shown that exogenous expression of HspB1 can provide a protective influence in a variety of disease-related models including ischemia, stroke, amyotrophic lateral sclerosis, and Huntington’s disease [18–24]. Although it was necessary in our experiments, and those cited above, to express exogenous HspB1, there are potentially other ways in which endogenous HspB1 might protect neurons in vivo. One possibility is that glial cells could release HspB1 that can be then taken up by adjacent neurons [25], or alternatively act to sequester Aβ. We have previously observed that HspB1 can be released into the medium of cultured cells [17] and HspB1 is also found in the cerebrospinal fluid and serum in vivo [26–29].
Our hypothesis is that astrocytes are able to release HspB1 in response to a local stimulus (for example, local accumulation or release of Aβ) that would then be available to provide a protective effect by either sequestering amyloid or being available to be taken up by other cells, such as neurons. Here we report that treatment of primary astrocytes with Aβ results in the release of HspB1, and that this release appears to occur via a non-classical method of secretion.
METHODS
Cell culture
Dissection and dissociation
Monolayer cultures of astrocytes were prepared from P1-P2 rat brain cortex according to established protocols [30, 31]. All animal usage was approved by the Institutional Animal Care Committee(IACC protocol KM-14-10). Briefly, the brain was removed and placed in ice cold Hanks Balanced Salt Solution (HBSS, Invitrogen/Gibco) containing 1% Penicillin/Streptomycin, and 0.2% HEPES (Invitrogen/Gibco), and the cortex was dissected from the brain. The hippocampus, meninges, and blood vessels were then peeled away from the cortex, and cortical tissue was enzymatically dissociated. The tissue was centrifuged, resuspended in Dulbecco’s modified Eagle Medium (DMEM, Gibco) with 10% FCS, and 1% Pen/Strep/glutamine and subjected to trituration. The cell suspension was centrifuged at 1300 RPM for 5 min and the resulting cell pellet was suspended in 10 ml of medium for plating in T-75 flasks. Cells were cultured for 5–7 days and confluent cultures were subsequently shaken overnight at 37°C on a platform rotary shaker (150–170 rpm) to remove microglia, oligodendrocytes, and neurons [32]; the medium was then replaced with fresh medium and the flasks returned to the incubator for 24 h to allow cells to recover. The remaining cells were then removed from the flasks with 0.025% trypsin and replated in DMEM-FCS in T-25 flasks, 6-well plates, or onto collagen-coated 16-well glass slides. The medium was changed every 3 days and also 24 h prior to any experimental manipulations. In some cases, the medium was changed to a low serum formulation (DMEM with 1% exosome-free FCS). These cultures were >95% astrocytes as assessed by immunocytochemistry (ICC) with anti-GFAP. Secondary passage astrocytes were employed for all experimental procedures.
Culture treatments
Prior to experimental treatments, medium was changed to a low-serum (1% exosome-free FCS) medium. Cultures were treated with Aβ at varyingconcentrations (0.1, 1, or 10 μM) or times of exposure (24 or 48 h); scrambled peptide was employed as a control. To assess for release of proteins into the medium, culture medium was collected at 24 and 48 h after Aβ addition. Medium was centrifuged to remove any cellular debris and then concentrated using Amicon concentrators (10 KD cutoff).
For inhibitor studies, the inhibitors were added 1 h prior to Aβ treatment and cells were cultured for a further 24–48 h prior to cell or conditioned medium sampling. Brefeldin A (BFA, Calbiochem; 10 μM) blocks protein export from the endoplasmic reticulum (ER) and disrupts the Golgi apparatus, and blocks the classical protein secretion mechanism [33–35]. Methyl β-cyclodextrin (MBC, Calbiochem; 10 μM) depletes membrane cholesterol and disrupts lipid rafts [33, 35]. Cycloheximide (CHX, 1 μM) blocks de novo protein synthesis and was employed to assess whether release requires de novo protein synthesis [33]. To test involvement of protein kinases, we employed inhibitors of p38 MAPK (SB20315, Calbiochem; 10 μM) and of p42/44 MAPK (U0126, Calbiochem; 10 μM) [36, 37]. Vehicle (DMSO) controls were included in all experiments.
Protein and conditioned media collection
Cell lysates and conditioned medium were collected 24 and 48 h post-treatment. Conditioned medium was collected on ice with protease inhibitor cocktail tablet (Roche Diagnostics, Laval, QC) added immediately upon collection. Media was centrifuged at 14,000 g for 5 min at 4°C to discard any cell debris. Cell lysates were collected by adding 1 ml ice cold TBS with 200 mM sodium vanadate and scraping cells off the plate with a rubber policeman. Cells were pelleted for 5 min at 4,000 g at 4°C and resuspended in ice-cold protein lysis buffer (1% NP40, 10% glycerol, O-β thioglucopyranoside, protease inhibitor tablet, 200 mM sodium vanadate, sodium fluoride and magnesium chloride in TBS) and stored at – 80°C until analysis.
Western blot analysis
Western blot analysis was performed with samples of total cellular lysate and conditioned medium. Protein concentrations were determined using a BSA protein assay kit (Pierce Chemicals, Rockford, IL). Laemmli sample buffer (10% SDS, glycerol, 1M Tris pH 6.8, dH20, 0.01% Bromophenol blue) containing fresh β-mercaptoethanol (BME) was added to 50 μg of cellular lysate or 200 μg of conditioned media protein, boiled and separated on an pre-cast 4–20% gradient Tris-glycine gel using the X-Cell Surelock System (Invitrogen). Separated protein was then transferred to a nitrocellulose membrane and after transfer, blots were stained with Ponceau Red to assess equivalency of protein loading. Blots were washed with TBS-T (1M Tris base, 2.5M NaCl, 50% Tween) to remove Ponceau Red and blocked with either 3% milk or BSA depending on the primary antibody dilution conditions for 1 h to prevent non-specific binding. Once blocked, blots were incubated with antibodies to one or more of the following proteins overnight at 4°C on a shaking platform: HspB1 (SPA-801, Enzo Life Sciences); clusterin/ApoJ (Santa Cruz SC8354); actin (Sigma-Aldrich A2066); GAPDH (Abcam ab9485); Integrin a6 (DHB P2C62C4); Hsc70/Hsp70 (SMC104A, Enzo Life Sciences); TSG101 (Santa Cruz SC7964). Following washing, signal was detected with horseradish peroxidase labeled secondary antibodies (1:5000 – 1:10000 in 3% milk) and Super Signal West Pico chemiluminescence substrate (ECL; Thermo Scientific, Rockford, IL) for 5 min and developed using films. Densitometry analysis was performed using ImageJ software and images prepared with Adobe Photoshop graphics software.
Immunoprecipitation (IP)
Conditioned medium samples were used for IP with either anti-HspB1 or anti-Aβ (clone 6E10, Covance). 10–15 ml of medium was concentrated 2–3 fold using 3 KD Amicon centrifuge filters, protein concentration was determined and 200 μg of protein was used for IP experimentation. Either anti-HspB1or 6E10 was added to the medium samples and incubated for 1 h with rotation followed by addition of 20 μl of magnetic protein A/G beads overnight to immunoprecipitate any Aβ and HspB1 complexes that formed. Samples were exposed to a magnet to separate the immunoprecipitates (magnetic A/G beads, antibody and any protein complexes attached) from supernatant. Protein concentration of supernatant was determined and 30 μg was added to 5X Laemmli sample buffer with fresh dithiothreitol (DTT). The IP sample was resuspended with 40 μl of 2X Laemmli sample buffer with fresh DTT and both supernatant and IP samples were electrophoresed as per our western blot protocol [17]. Blots were probed with 6E10 and anti HspB1 (either rabbit (SPA-801) or goat (SC polyclonal) antibodies).
Proteinase protection assay
Conditioned medium (CM) was collected and aliquots (50–100 μg protein) were treated with Proteinase K. Briefly, 120 μl of CM was treated with 4 μl of PK (100, 10 or 1 mg/ml in 50 mM Tris-HCl pH8, 10 mM CaCl2) and incubated on ice or at 25°C for 2 h. The reaction was quenched by addition of loading buffer, samples boiled and electrophoresed. Subsequent blots were probed with anti-HspB1, Integrin α6, or clusterin/ApoJ.
Exosome isolation
Exosomes were isolated from conditioned medium via a standard ultracentrifugation protocol [38]. As noted above, astrocytes were cultured in a low-serum medium containing 1% exosome-free FBS, and treated for 24 h with either Aβ or vehicle control (DMSO). The CM was then collected and cleared of cellular debris by two rounds of low speed centrifugation (2000 g, 10 min, and 10,000 g for 30 min). The supernatant was then centrifuged at 100,000 g for 3 h; the resulting pellets were washed (x 2) with PBS and re-centrifuged at 100,000 g for 1 h [38, 39]. Pellets were resuspended in 50–150 μl of PBS and either analysed immediately or stored at –80°C until use. For western blot analyses, 50–200 μg protein were electrophoresed and blots probed with anti-HspB1, TSG101, Hsc/Hsp70,clusterin/ApoJ.
Electron microscopy
5–10 μl of exosomes was mixed with an equivalent volume of 1% LMP agarose, fixed with Karnovsky fixative for 24 h and stored in 0.1M Na cacodylate buffer until processed. The specimens were osmicated, dehydrated using graded alcohol and acetone followed by infiltration with EPON resin, embedded in molds, and polymerized overnight at 70°C. Thin sections were cut using Reichert ultra-cut S at 85 nm using a Diatome diamond knife, mounted on 300 mesh copper grids and dried for 30 min. Grids were then stained with uranyl acetate followed by lead citrate stain. Grids were then examined in a JOEL 1200EX electron microscope and images captured using a SIA – L3 C digital camera; the camera images were calibrated using a carbon line grating.
Immunocytochemistry
Astrocytes were plated on collagen-coated 16-well Lab-Tek® chamber slides (Lab-Tek®). Cells were fixed in 4% paraformaldehyde in phosphate buffered saline (PBS) for 15 min, washed with PBS and permeabilized with 0.1% Triton X and blocked with 5% donkey serum for 1 h. Primary antibodies included GFAP (Chemicon MAB360), HspB1 (SPA-801, (Assay Designs), Golgi 58K (Abcam ab9845). Cells were incubated in primary antibodies overnight (20 h) at 4°C, washed in PBS and incubated with secondary antibodies for 1 h in the dark. Secondary antibodies were: DylightTM 488-conjugated AffiniPure donkey anti-mouse IgG; Dylight 649-conjugate AffiniPure donkey anti-rabbit (1:250, Jackson Laboratories, West Grove, PA). Cells were rinsed with TBS-½T (Tris-buffered saline with 0.25% Tween) and in some cases, were stained with DAPI in TBS for 5 min. Cells were again washed with TBS-½T and cover-slipped using polyvinyl alcohol mounting medium with DABCO® (Sigma–Aldrich). Images were routinely acquired in three channels (488, 549, 647) using confocal scanning microscopy with sequential Z-stage scanning (Olympus Fluoview 1000 microscope).
Statistical analysis
Statistical analysis was performed in GraphPad Prism 6.0 (GraphPad Software Inc., La Jolla, CA). Figures are shown with mean values±SEM with significance determined by either one-way ANOVA testing followed by Tukey post-hoc tests or t-tests to compare two groups, for example±Aβ. Significance was determined at p < 0.05 unless otherwise stated.
RESULTS
Astrocytes express and release HspB1
Astrocytes isolated from neonatal rat cortex were cultured and expression of HspB1 was assessed by western blotting and ICC. Astrocytes robustly express HspB1 (see Fig. 1B; see also Supplementary Figure 1) and we were interested in determining whether HspB1 would be released into the culture medium with exposure to Aβ, and if so whether that stimulus was selective. We exposed cells to Aβ (10 μM), scrambled control peptide, and the vehicle (DMSO), and collected the medium and the cells 24 h after exposure. As shown in Fig. 1A, the CM contained HspB1 with increasing amounts observed with Aβ exposure (compared to the no treatment control condition), with quantitation of the extracellular HspB1 under control, vehicle, scrambled peptide, and Aβ (10 μM) treatments displayed graphically. Expression of HspB1 in total cell lysates does not appear to be influenced by any of the treatments (Fig. 1B).
To further investigate the effects of Aβ, we carried out experiments testing release of HspB1 with 0.1, 1.0, and 2.0 μM Aβ for 24 and 48 h. The medium and cells were collected at either 24 or 48 h of treatment and resulting western blots probed for expression of HspB1. As shown in Fig. 2A, HspB1 in the medium accumulates with longer exposure; the blots were also probed with anti-Aβ (6E10) to show the amount of peptide detectable in the medium. Based upon the results of these experiments, we chose 1.0 μM Aβ as the concentration to use in subsequent experiments.
Mechanism of HspB1 release
To gain insight into whether this release was passive (perhaps due to cell damage) or via a secretory pathway, we carried out a series of experiments initially using heat shock as the stress stimulus and treatments with several different inhibitors of protein synthesis or transport within the cell. Heat stress has been routinely used to stimulate extracellular release of several Hsps in tumor cell lines as well as in primary cells [35, 40–43]. Figure 3 shows that HS resulted in an increase in the amount of HspB1 released into the medium, as well as a modest increase in cellular expression (averaged over three separate experiments). In order to determine whether this was regulated in any way, we exposed the cells to several treatments that have been shown to alter protein secretion. Astrocytes were treated with the chemicals for 1 h prior to the stress stimulus (control cells were similarly treated but not exposed to the heat stress) and then the medium and cells were collected 24 h later. Although CHX (inhibitor of protein synthesis) attenuated the HS-induced increase in cellular expression as expected, there was a significant increase in extracellular HspB1, which was likely due to cellular damage since the cultures did not appear to be particularly healthy with the CHX treatment. Treatment with BFA (which blocks protein export from the ER and disrupts Golgi function and is considered an inhibitor of classical protein secretion [34]), had no significant effect in the non-heat stressed cells, in particular there was no decrease in release. While there was an increase compared to control, this was not significantly different from the HS condition. This result is similar to that reported for Hsp70 and Hsp90, as well as other non-classically secreted proteins such as IL-6 or FGF [33, 44].
Release of HspB1 with Aβ and inhibitors
We next examined release of HspB1 in astrocytes treated with 1 μM Aβ and the various agents. We employed BFA to block the classical secretion pathway and MBC to disrupt lipid rafts [33, 45] (Fig. 4). Prior reports had suggested that activation of protein kinases by heat stress were involved in regulating the release of Hsp70 from astrocytes [46], so we employed an inhibitor of MAPK (U0126) as well as a p38MAPK inhibitor (SB20358); the latter is of particular interest since p38 phosphorylates HspB1 and is involved in its interaction with actins, which could potentially be involved in regulating release [14]. Both EDTA and MgCl2 have been shown to influence release of other leaderless proteins by chelation of calcium and ATP [33, 47]. Figure 4A presents a representativewestern blot for vehicle control samples treated with the various compounds, while Fig. 4B shows a representative blot for the Aβ samples with the same compounds. In this experimental series, we also assessed release of clusterin (a known secretory protein, secretion of which should be blocked by BFA) to compare with the HspB1, and integrin α6 (which is an integral membrane protein often found on the surface of released microvesicles).
The blots show quite clearly that although clusterin release is blocked by BFA, that of HspB1 is enhanced both in the vehicle control and the Aβ-treated samples; note that these are the same samples probed sequentially. Although the reason for the increase with BFA is unclear, ICC assessment of cells under each of these different conditions did not show any obvious cell death that could result in enhanced passive release (see Supplementary Figure 1).
Figure 4C presents the graphical analysis of HspB1 release with the various treatments. Aβ results in increased release in all samples compared to the vehicle control samples. Cotreatment with BFA results in increased release in both the vehicle and Aβ-treated samples, although none of the other inhibitor cotreatments results in increased release over the Aβ treatment. Treatment with EDTA resulted in cellular detachment from the substrate and thus was not continued. We also tested the influence of KCl (to induce depolarization) and glibenclamide (an ABC transporter inhibitor), although neither of these had a significant effect on the detection of extracellular HspB1. None of the treatments had any apparent effect on cellular levels of HspB1 or clusterin (data not shown).
Extracellularly released HspB1 is resistant to proteinase K treatment
The results so far indicated that HspB1 did not appear to be released via a classical secretion pathway. Some non-classically secreted proteins are packaged into vesicles, although other possibilities include passive release (perhaps related to membrane disruption), lysosomal secretion, blebbing, or exosomal release. One way to assess whether the protein is free in solution or associated with a membrane-bound structure is to carry out a protease protection assay (to test for vesicle-independent release [47]). We tested this possibility by treating aliquots of the CM with proteinase K (PK) and assessing protein degradation by western blotting. As shown in Fig. 5, treatment of the CM with 1 μg/ml of PK completely abolishes the signal for integrin α6 (used as a marker for an integral membrane protein associated with the outer surface of membrane vesicles). The signal for clusterin is somewhat diminished at 1 μg/ml, but abolished with 10 μg/ml of PK. In contrast, degradation of the HspB1 requires the highest PK concentration. This is suggestive of at least some of the released HspB1 being protected by inclusion in a membrane bound structure.
Presence of HspB1 in exosomes
To determine whether HspB1 might be being released in vesicles, we then isolated exosomes from CM from vehicle- and Aβ-treated astrocytes. Presence of exosomes in the preparations was confirmed by electron microscopy (Fig. 6A, B), as well as by the presence of TSG101 (an exosomal marker, Fig. 6E) in western blots of exosomal preparations. In Fig. 6C-E, a western blot probed sequentially for HspB1 and TSG101 is presented. Here the exosomal fraction (exo), the concentrated conditioned medium before exosomal fractionation (CM) and the supernatant from the exosomal pellet (Sup) have been run on the same blot. The top panel (6C) is a short exposure showing HspB1 in the CM and in the exosomal supernatant; faint bands can be observed in the exosomal fraction lanes (arrow). With a longer exposure, the presence of HspB1 in the exosomal fraction is more obvious (note that this HspB1 antibody tends to show a doublet for HspB1 particularly in CM samples). TSG101 is also clearly detectable in the exosomal fraction although because of the amount of BSA and other medium components in the concentrated medium, detection of TSG101 in the medium fractions is precluded. The ponceau red stained blot image is shown in Fig. 6F. These results suggest that HspB1 can be detected in exosomes, but HspB1 free in the medium is likely more abundant, since the amount in the supernatant after exosome isolation is not depleted (Fig. 6C).
Extracellular HspB1 interacts with Aβ
We then tested whether extracellular HspB1 could interact with Aβ present in the medium. We have previously reported that HspB1 can interact directly with Aβ, based upon IP experiments [17]. Here CM samples were subjected to IP with either anti-HspB1or anti-Aβ (6E10). Representative western blots of four separate IP experiments are presented in Fig. 7. InFigure 7A, HspB1 has been subjected to IP and the blots probed sequentially with anti-Aβ (6E10) and then anti-HspB1 (after stripping). The top panel shows that Aβ is co-precipitated with HspB1, while the bottom panel shows the blot probed with anti-HspB1. The bottom band appears to be the specific HspB1 band that can be detected just below the light chain IgG, based on the positive control of r-HspB1 (last lane). 7B shows a corresponding IP of Aβ (with 6E10) probed with anti-HspB1 first, followed by anti-Aβ; in this case, there is little detectable HspB1 in the IP samples.
Our data thus show that astrocytes can release HspB1 extracellularly, that this release is increased by stress and by exposure to Aβ in particular, occurs via a non-classical mechanism that involves some exosomal release. Furthermore, the released HspB1 appears to be able to interact with the Aβ present in the medium.
DISCUSSION
Astrocytes robustly express HspB1, which can be upregulated by heat stress and released into the extracellular milieu. Our results show that Aβ can elicit increased HspB1 release compared to the vehicle control treatment. The release in both the vehicle and Aβ-treated conditions was not inhibited by BFAtreatment.
Although HSPs are thought to function primarily as intracellular chaperones, the release and potential extracellular functions of HSPs have been the focus of an increasing number of studies (reviewed in [44, 48–52].
Most secreted proteins possess an N-terminal signal peptide that directs their sorting to the ER and subsequently through the ER-Golgi compartment for release via conventional ER-Golgi secretory pathway [34, 53, 54, 34, 53, 54]. There are, however, a large number of proteins that have been shown to exit cells via pathways independent of the conventional secretory pathway. Generally these proteins lack the signal peptide and their release is not blocked by BFA [53, 54]. This unconventional release is regulated to some degree and often induced by stress, and many of the proteins (both cytosolic and nuclear) secreted in by non-conventional pathways have roles in inflammation, tissue repair and angiogenesis. Several different categories of release have been described including direct translocation of proteins through the plasma membrane to the extracellular compartment and release via lysosome, exosome or by membrane blebbing and vesicle shedding [53–56].
Heat shock proteins lack the classic N-terminal leader sequence normally associated with the classical secretion pathway, and reported mechanisms underlying their release into the extracellular environment tend to be dependent on cell type and context [44, 49–51]. HspB1 has been reported to be released by a variety of cell types including glial tumor cells (exosomes) [41], vascular endothelial cells (soluble) [57], B cells (exosomes) [40], macrophages (lysosome-like vesicles) [27], neuroblastoma cells [58], HEK293 cells [17]. HspB1 release from endothelial cells was noted as being soluble, since the secreted HspB1 interacted with soluble VEGF to regulate angiogenesis; interestingly phosphorylation of HspB1 inhibited its release in these experiments [57]. Secretion of HspB1 from macrophages was regulated by estrogen and intracellular colocalization of HspB1 and LAMP in lysosomal vesicles was observed [27]. We have previously reported that overexpression of HspB1 in HEK293 results in HspB1 release into the culture medium although in that study we did not investigate release mechanisms [17]. The acrylamide-induced increase in extracellular Hsps in neuroblastoma cells may have been a result of passive release following increased cellular toxicity [58].
In our prior studies, we have shown that HspB1 promotes survival in PC12 cells [36], primary peripheral [59] and central neurons [16]. HspB1 also plays a role in axonal initiation and extension and branching of primary neuron axons [16, 61]. HspB1 is generally found localized in the cytosol in a diffuse or granular appearance, while in migrating cells and growth cones it is found along the leading edge of lamellopodia associated with actin [14, 60]. Heat stress in PC12 cells results in the redistribution of HspB1 from the cytosol to the cytoskeletal fraction, particularly increasing its association with actin, but also causes membrane blebbing, with blebs displaying localization of HspB1 and actin [14]. In the current study, we detected sporadic cells with blebs following treatment with Aβ and BFA, although given the sporadic nature of this occurrence we do not think it likely that it could fully account for the increased HspB1 release we observed. It is, however, possible that the BFA treatment resulted in membrane leakiness, although this would not explain the reduction in the release of clusterin.
How does Aβ stimulate HspB1 release? Our results suggest that Aβ selectively increases HspB1 release, but how this comes about is not clear. Aβ can bind to phospholipids in the cell membrane and to a variety of cellular receptors including p75, nicotinic AChRs, glutamate receptors [62, 63]. Aβ can also form stable membrane pores and channels with resulting dysregulation of Ca flux and which could promote protein release across the membrane or via vesicular release. [64]. In our experiments, treatment of the cells with KCl (to induce membrane depolarization) had little effect on HspB1 release.
The extracellular function of HspB1 is not entirely clear. Like other Hsps, HspB1 has been suggested to play a role in immunomodulation [49], with a number of studies reporting an anti-inflammatory action by increasing production of anti-inflammatory cytokines by monocytes and macrophages [27, 65]. Lee and colleagues have recently reported that soluble HspB1 inhibits the function of VEGF by a direct interaction with VEGF which can result in decreased angiogenesis and tumor metastasis; they also suggest that VEGF itself can inhibit HspB1 release and thus regulate angiogenesis [57]. Cellular receptors that have been reported to bind HspB1 include the scavenger receptors and toll-like receptors. In preliminary studies we have exposed cortical neurons and astrocytes to recombinant HspB1. We did not see any detectable influence on neuron survival nor internalization of HspB1 over the course of these short-term experiments (10 min-6 h). In the astrocyte cultures exposed to rHspB1 (endotoxin-free), we observed activation of signaling pathways, in particular MAPK and Akt; however, we also noted that the act of changing the medium results in pathway activation although this was enhanced when rHspB1 was also provided. Further study is required to determine what cellular receptors HspB1 might bind to, what signaling pathways are activated, and whether there is any influence on cellular survival or local inflammatory responses.
There have been numerous studies reporting upregulation of glial HspB1 in response to various stimuli, including heat, excitotoxicity, and ischemia both in vitro and in vivo [16, 66–71]. Increased expression of HspB1 is reported in several neurodegenerative disorders (e.g., [72–76]), however, there have been no reports of extracellular release of HspB1 from glial cells under these conditions. HspB1 promotes neuronal survival in response to various stresses, and it could be acting intracellularly (to chaperone the cytoskeleton or protein aggregates [14, 37, 61, 77], or alternatively it could be released into the extracellular space where it could potentially be sequestering amyloid [17, 78–80]. Overexpression of HspB1 in AβPPswe/PS1dE9 transgenic mice resulted in decreased appearance of amyloid plaques, as well as attenuating the behavioral deficits associated with this mouse model [81].
A number of studies have reported that small Hsps including HspB1 can influence Aβ aggregation and toxicity as well as sequester toxic oligomers [77, 80]. HspB1 has been localized to plaques in AD brain samples [15] as well as in transgenic mouse models of AD [77]. In the latter study, HspB1 was not only localized in plaques in AD mouse model brains, but HspB1 added to culture medium was shown to sequester toxic oligomers of Aβ and attenuate neuronal death. Interestingly, the authors questioned how an intracellular chaperone could act on externally added Aβ, and comment that this problem could be solved if the glial HspB1 were externalized [77]. Our results provide evidence that HspB1 can indeed be released from glial cells, and relevantly, in response to extracellularly added Aβ.
In summary, our results show that relatively low concentrations of Aβ can stimulate release of HspB1 from astrocytes, via a non-classical secretion mechanism. HspB1 can be found either free in the medium or associated with exosomes. Further, HspB1 and Aβ in the medium can interact, in the sense that IP of either HspB1 or Aβ coprecipitates the other component.
