Induction of heat shock proteins in B-cell exosomes

Exosomes are a discrete population of nano-vesicles (40-90 nm in diameter) secreted by a variety of cells including B-lymphocytes, T-lymphocytes, dendritic cells, mast cells, epithelial cells, reticulocytes and others (Blanchard et al., 2002; Johnstone and Ahn, 1990; Raposo et al., 1996; Skokos et al., 2001; van Niel et al., 2001; Zitvogel et al., 1998). Currently, there is much discussion about the physiological functions of exosomes, with mounting evidence supporting a role in modulating cellular immunity (Andre et al., 2002; Wolfers et al., 2001; Zitvogel et al., 1998).

Proteomic methods are uncovering the molecular complexity of exosomes purified from different cells or body fluids, and together such analyses are useful for highlighting important cell-type specific differences between exosomes of different sources. They also identify commonly expressed exosome-proteins, such as MHC molecules, tetraspanins, adhesion molecules, cytosolic enzymes and certain stress proteins including hsc70 and hsp90 (Bard et al., 2004; Hegmans et al., 2004; Mears et al., 2004; Pisitkun et al., 2004; Thery et al., 2001; Wubbolts et al., 2003). However, these molecular analyses have essentially been limited to those exosomes produced under steady-state conditions, representing therefore a constitutive exosomal phenotype. It is not yet clear whether the composition – and maybe also the function(s) – of exosomes are dynamic and modified by external microenvironmental factors. So far, only two ways have been described how the extracellular milieu can influence the composition and/or quantity of exosomes that are secreted. First, the inhibition of dendritic cell exosome secretion in response to factors that trigger phenotypic maturation of the dendritic cell (such as LPS) (Kleijmeer et al., 2001; Zitvogel et al., 1998). Second, the activation of T cells (through ligation of the CD3-TCR complex) triggers an elevated exosome secretion with an altered exosomal composition such as higher levels of exosomally-expressed MHC molecules, tetraspanins and adhesion molecules (Blanchard et al., 2002).

Here, we studied whether conditions of cellular stress (specifically heat stress), can modify the quantity and/or composition of exosomes produced by B-lymphocytes. Our data show for the first time a dynamic and highly specific alteration in exosome composition in response to heat stress, resulting in an increased expression of selected heat shock proteins (hsps) (hsp27, hsc70, hsp70 and hsp90) in proportion to the degree of stress. Exosomes purified from either steady-state cells or heat stressed cells were not, however, able to trigger the phenotypic maturation of monocyte-derived dendritic cells. Furthermore, our analysis demonstrates that these stress proteins are encapsulated within the exosome lumen and are not present on the exosome surface.

Overall, our findings indicate that alterations in exotype-phenotype are a component of the cellular response to stress. The physiological importance of these alterations is not yet clear, but exosomes might enable the elimination of hsp-protein complexes from the cell in a manner that does not directly lead to the activation of dendritic cells.

Epstein Barr virus (EBV)-immortalised human B-lymphoblastoid cell lines (B-LCL) were prepared as previously described (Doyle et al., 1994) from healthy donors, designated B-LCL(PC) and B-LCL(LL). An additional EBV-transformed cell line termed IB4 (Hurley et al., 1991) was a gift from Martin Rowe (Infection and Immunity, Wales College of Medicine, Cardiff University, UK). The Jurkat cell line was obtained from the MRC Co-operative (Cell Culture Laboratory, Wales College of Medicine, Cardiff University). Cells were seeded at 2-3 million cells/ml in serum free AIM-V medium, subjected to heat stress (0-3 hours, 42°C in a humidified incubator with 5% CO₂) and cultured for a total of 24 hours before exosomes were isolation. All culture reagents were purchased from Gibco/Life technologies.

Twenty-four hours after heat stress, B-cell lines were washed in PBS, fixed and permeabilised (Beckman Coulter Fix/Perm Kit), and stained with a panel of anti-hsp antibodies (10 μg/ml per 10⁶ cells in 100 μl). Expression levels were analysed by flow-cytometry (FACScan, BD). The mean fluorescence intensities (MFI) of four cell lines were measured. In addition, cell lysates (plus 2% NP-40 with protease inhibitors) were prepared as described (Clayton et al., 2003), and analysed for hsps by western blot. Densitometric analysis of blots was performed using Quantity One Software (BioRad).

Exosomes produced by B cells or Jurkat cells during a 24-hour culture period, were isolated from conditioned culture medium by differential centrifugation and flotation on 30% sucrose-D₂O as described (Andre et al., 2002; Lamparski et al., 2002). This method has been shown to effectively separate exosomes from non-exosomal membrane-vesicles or membranous fragments and other contaminants. Exosome-pellets were resuspended in 100 μl of PBS and aliquots stored at –70°C. Exosome protein was determined by BCA assay (Pierce). Exosomes derived from untreated B cells are referred to as `control-exosomes', whereas those from heat-treated cells are referred to as `heat-shocked exosomes'.

Equal volumes of SDS-sample buffer were added to cell lysates or purified exosomes (1-5 μg protein/well), samples were boiled at 98°C for 10 min and separated on 10% acrylamide gels. After transfer to PVDF membranes and blocking overnight (3% milk and 0.05% Tween-20 in PBS), primary antibody (0.5-2 μg/ml) was added for 1 hour, followed by washing and the application of secondary HRP-conjugated antibody (1:26,000 dilution, DAKO). Detection of bands was performed with the ECL⁺ system (Amersham/Pharmacia).

Exosomes derived from control or heat-stressed cells were incubated with aldehyde-sulphate latex beads (surfactant free, white, 4 μm, Interfacial Dynamics), as described elsewhere (Lamparski et al., 2002). Exosome-bead complexes were stained with antibodies against known exosome surface markers including CD81 (Pharmingen) and MHC class II (DAKO) as positive controls, or with a panel of hsp-specific antibodies. Specific staining was compared to isotype-matched control antibodies by flow-cytometry.

Equal quantities of exosomes from control cells or cells heat-shocked for 3 hours were incubated with a panel of hsp-specific antibodies (1 μg antibody/sample) in a final volume of 100 μl for 3 hours at room temperature. A pan-MHC-class-II antibody was used as a positive control (Dako). Goat-anti-mouse-coated beads (Dynal) at 10×10⁶ beads/sample were added. After 2 hours, beads were pelleted (using a magnet) and washed 6 times to remove unbound material. Beads were incubated in SDS-sample buffer and analysed by western blot for the presence of MHC class I.

Peripheral blood mononuclear (PBM) cells were isolated by histopaque centrifugation (Sigma), and added to 24-well plates (5 million cells/ml/well) in AIM-v for 2 hours. Adhered cells, mainly monocytes, were incubated in AIM-v that contained IL-4 and granulocyte/macrophage-colony stimulating factor (GM-CSF) at 1000 U/ml. Lymphocytes were distinguished from non-lymphocytes, based on a dot-plot of their forward scatter (FSC) and side scatter (SSC) characteristics on a flow cytometer. By day 6, 90% of FSC/SSC-gated cells were MHC class II high, CD14^– by flow-cytometry.

MDDCs (day 6) were harvested and seeded into fresh 48-well plates at a density of 5×10⁵ cells/ml. Exosomes from control or cells heat-shocked for 3 hours were added (at various concentrations of up to 100 μg/10⁵ MDDCs). As a positive control, lipopolysaccatide (LPS, Sigma) was added at 0.1 or 10 μg/ml; as a negative control MDDCs were left untreated. After 48 hours, cells were harvested, fixed (Becton Dickinson Fix/Perm Kit) and stained with FITC-conjugated antibodies against CD83 or isotype-matched control. Cell-free supernatants were analysed for IL-6, IL-10 and IL-12 by ELISA (R&D Duoset Kit).

Mouse anti-human hsp27 (IgG2a, clone F-4), hsp60 (IgG3, clone H-1), hsp70 (IgG2a, clone w27), hsc70 (IgG2a, clone b-6), hsp90 (IgG2a, clone F-8) and LAMP-2 (IgG1, clone H4B4) were from Santa Cruz Biotechnology. Rat anti-human gp96 (IgG2a, clone 9G10) was from Stressgen Biotechnologies. Anti-human CD83-FITC and isotype-matched control were from Immunotech. Antibodies specific for EBV LMP-1 were a gift from Martin Rowe (clone CS1, 2, 3, and 4). Mouse anti-human MHC class I (IgG2a, clone HC10) was obtained from the MRC Co-operative (Cell Culture Laboratory, Wales College of Medicine, Cardiff University, UK). Goat anti-mouse IgG-HRP or rabbit anti rat-IgG-HRP was from DAKO.

The constitutive expression of hsp90 and/or hsc70 (hsp73) in exosomes derived from antigen-presenting cells (Thery et al., 1999; Wubbolts et al., 2003), reticulocytes (Geminard et al., 2001) and epithelial cells (van Niel et al., 2001) has been reported previously. However, potential alterations in the quantity and/or repertoire of exosomal hsps following stress have not been addressed to date.

We investigated whether cellular stress leads to altered hsp expression in exosomes and induced hsps by elevated temperature. B-cell lines were exposed to a temperature of 42°C for varying times (up to 3 hours) to generate different levels of non-lethal heat stress. After returning the cells to 37°C for 24 hours, we analysed changes in the levels of hsps, specifically of hsp27, hsp70, hsc70, hsp60 and hsp90, therefore covering most hsp families (Srivastava, 2002). Fig. 1a shows the expression levels of these hsps in normal cells, with high constitutive levels of hsc70 and hsp90, and significantly less (around 7- to 20-fold) expression of hsp27, hsp70 and hsp60. Following only 30 minutes of heat stress, a significant increase of hsp70 expression occurred (Fig. 1b) that rose to approximately sevenfold above its constitutive level after 3 hours. Some increase in cellular hsp27, hsp60 and hsp90 was evident, but this was not statistically significant. In all B-cell lines tested, cellular hsc70 was reduced by approximately 30% compared with its constitutive levels (Fig. 1b).

Exosomes secreted during the 24-hour incubation period post heat-stress, were analysed by western blotting for hsp expression, with each lane containing the same quantity of purified exosome protein (Fig. 2a). A time-dependent (duration of heat stress) elevation in exosomal hsp27, hsp70, hsc70 and hsp90 was evident. However, not all exosome-associated proteins were increased, levels of MHC class I, the EBV-encoded latent membrane protein-1 (LMP-1) and the lysosome associated membrane protein LAMP-2 were unaltered (Fig. 2a). Hsp60 or gp96 were not detected in these samples (not shown), which agrees with previously published reports of exosomes from other cellular sources (Skokos et al., 2003; Wolfers et al., 2001). To demonstrate the selective absence of hsp60 and gp96 from exosomes even under conditions of stress, we compared cell lysates to exosomes by western blot. Although bands for hsp60 and gp96 were seen in lanes containing cell lysates, no bands were evident in lanes containing exosomes. By contrast, the same samples showed a time-dependent elevation in hsp70, evident in both lysates and exosomes, and again MHC class I remained at constant levels (Fig. 2b). Although our flow-cytometric analysis of B cells revealed a reduction in hsc70 levels due to heat stress (Fig. 1b), western blots of exosomes showed a dramatic elevation in hsc70 (Fig. 2a). To confirm these findings, hsc70 levels in cell lysates and exosomes were compared by western blot (Fig. 2c), showing a stress-mediated reduction of cellular hsc70 and an increase in exosomal hsc70. The same samples demonstrated elevated expression for hsp70 in both lysates and exosomes (Fig. 2c). This suggests that a proportion of hsc70 is released from the cell in the form of exosomes after heat stress. Also examined was TSG101, a protein involved in the production of multivesicular bodies (Katzmann et al., 2003) that ultimately give rise to exosomes. TSG101 levels were slightly elevated in stressed exosomes, suggesting that the cellular machinery for exosome assembly can also be activated by cell stress (Fig. 2d). To investigate whether induction of hsps in exosomes is a phenomenon unique to EBV-immortalised B cells, we repeated these studies in cells of a EBV-negative human T-cell line (Jurkat). Exposure of these cells to heat stress, resulted in elevated levels of exosomal hsp90 and hsp70 (Fig. 2d), although the constitutive levels of exosomal hsps was much lower in Jurkat exosomes.

In conclusion, our findings demonstrate that the exosomes secreted during a response to heat stress have a composition that is indeed different from that of steady-state exosomes and, although broadly reflective of changes that occur within the cell, exosomes express a unique and inducible repertoire of selected stress proteins.

We next investigated whether heat stress also altered the quantity of exosomes produced by B cells and Jurkat cells. Exosomes were purified from supernatants of the same number of control cells or cells that had been heat-stressed for 3 hours. The total protein of each exosome pellet was measured (by BCA assay) for each preparation, and total protein of heat-stressed cells was compared with that produced by non-stressed cells (Fig. 3). Multiple preparations of exosomes from three B-cell lines and from Jurkat cells showed only small elevations (less than 1.3-fold) in the number of exosomes produced, with an overall mean ratio of 3-hour-stress-exosomes to control exosomes of 1.245±0.07 to 1 (mean±s.e.m., n=20).

We investigated the location of exosome-associated hsps, present on the outer surface or within the vesicle lumen, because this could have a direct bearing on the function of exosome-associated hsps.

Control and heat-shock exosomes were prepared once, and this stock preparation was used in all the following experiments (Fig. 4). Initial analysis was by western blot where elevated levels of exosomal hsp70 and hsc70 in heat-shock exosomes was confirmed (Fig. 4a). Exosomes were subsequently coupled to aldehyde-sulphate latex beads (Lamparski et al., 2002), and the expression of exosome-surface molecules was analysed by flow-cytometry of the exosome-bead complexes (Fig. 4b). Positive staining for the exosome surface markers CD81 and MHC class I was demonstrated by flow-cytometry, however there was no positive staining for hsp27, hsp70, hsc70 or hsp90. Furthermore, this analysis demonstrated that there was no difference between control and heat-shock exosomes, indicating that under conditions of stress, the exosome-surface does not become hsp-positive. Beads coated with exosomes underwent several washing steps during antibody staining, which might have resulted in the loss of surface-associated hsp. To confirm that our negative flow-cytometric findings were not due to washing away surface hsp, exosome-beads that had been subjected to washes, but had not been labelled with antibodies were boiled in SDS-sample buffer at 98°C for 10 minutes; the solubilised total exosome proteins were then analysed by western blot (Fig. 4c). There, elevated levels of hsp70 and hsc70 were found in the samples of heat-shock exosomes but not of control exosomes, indicating that exosomal hsps had not been lost during exosome-bead washing steps, and that the detection of elevated hsps levels absolutely required the disruption of the exosome structure (Fig. 4c).

To further support these data, a more sensitive method was used to detect hsps on the exosome surface. Intact exosomes were precipitated out of solution with antibodies against exosome-surface molecules, and precipitates were analysed by western blot for the presence of MHC class I. Control or 3 hour heat-shock exosomes precipitated equally well with antibodies against MHC class II (Fig. 4d). By contrast, irrelevant isotype-matched antibody was ineffective at exosome precipitation and precipitations with antibodies against hsp27, hsp70, hsc70 or hsp90 were equally unsuccessful at precipitating exosomes. There was no indication that heat stress was a stimulus for locating hsps to the exosome surface, because no difference was seen between control and heat-shock exosomes (Fig. 4d). Together, our data point to a lumenal and not a surface location for hsps in B-cell-derived exosomes.

One possible mechanism by which stress proteins can modulate immunological responses is through activating dendritic cell maturation. This phenomenon is usually associated with the binding of soluble hsps to cell surface signalling receptors such as CD91 (Basu et al., 2001; Singh Jasuja et al., 2000). We examined whether exosomes enriched in hsps are capable of modulating the phenotype of immature dendritic cells towards maturation, by measuring alterations in cell-surface maturation markers (e.g. CD83) and alterations in cytokine secretion.

MDDCs were cultured in GM-CSF and IL-4 from adherent peripheral blood cells, and were at day 6 morphologically and phenotypically still immature. Treatment of day 6 MDDCs with 10 μg/ml LPS, a potent maturation agent, resulted in marked elevation in CD83 protein levels after 48 hours. By contrast, treating MDDCs with control exosomes or heat-shock exosomes, even at doses as high as 100 μg exosomes per 10⁵ dendritic cells, had no effect on CD83 protein levels (Fig. 5a). The culture medium of the above MDDCs was also examined for possible exosome-mediated changes in cytokine secretion. There was no significant difference between control and heat-shock-exosome-treated MDDCs in the levels of IL-6, IL-10 and IL-12 (assessed by ELISA; Fig. 5b), and the levels of TNFα, IL-1β, IL-10, IL-12, IL-8 and IL-6 at day 3 or day 6 post stimulation (measured by cytometric bead array, data not shown). These data indicate that stress-derived exosomes, like control-exosomes, do not modify the status of maturation of dendritic-cells.

Here, we have investigated whether heat stress can affect the molecular composition of nanometer-sized vesicles (exosomes) that are produced by cultured B-cell lines. Our data demonstrates that exposure of B cells to elevated temperature (42°C up to 3 hours) results not only in alterations in expression pattern of cellular hsps, but also results in hitherto unknown changes in the phenotype of secreted exosomes. Specifically, heat stress drives the selective enrichment of certain hsps (hsp27, hsc70, hsp70, hsp90) within exosomes, while others, such as hsp60, remain excluded from exosomes. Other housekeeping exosome proteins, such as MHC molecules and LAMP-2, are unchanged following heat stress. Heat stress also mediates the increased quantity of exosome secretion (approximately 1.2- to 1.3-fold). The functional significance of the release of hsps into the extracellular milieu in the form of exosomes remains unknown.

Stress proteins act to stabilise misfolded proteins, and prevent the formation of protein aggregates under conditions of heat stress. Their presence in the extracellular environment has been proposed to be an indicator of cellular injury, with important roles in inflammation and immunity. Stress proteins act essentially as soluble factors that interact with various cell-surface receptors to drive inflammatory processes (Lamb et al., 1989; Wang et al., 2002) including the maturation of dendritic cells (Basu et al., 2000; Zeng et al., 2003). Extracellular hsps might also act as molecular chaperones, by forming associations with antigenic polypeptides and thus efficiently delivering exogenous antigens to the endogenous antigen-processing-pathway of dendritic cells (Binder et al., 2001; Manjili et al., 2002; Srivastava and Udono, 1994; Zeng et al., 2003).

Exosomes are also thought to offer an efficient mode of transferring antigens, e.g. from cancer cells to dendritic cells, and the notion that certain stress proteins that are expressed by exosomes of diverse sources might play a role in exosome-immunity has attracted much interest. For example, exosomes derived from ascitic fluid of cancer patients were hsp-positive and prove an effective form of antigen for dendritic-cell cross-presentation to CD8⁺ T cells in vitro (Andre et al., 2002). Similarly, exosomes isolated from several murine tumour cell lines expressed high levels of hsp90 and hsc70, and were immunologically effective as vehicles for delivering antigen to dendritic cells, yet failed to trigger dendritic cell maturation (Wolfers et al., 2001). Recent in vivo studies have shown that co-administration of exosomes with adjuvants (such as CpG) is required to generate optimal immune responses (Chaput et al., 2003). Exosomes may therefore be viewed as an antigen source of insufficient inherent-adjuvant activity to drive alterations in dendritic cell maturation. To date, the only exception of this is a report by Skokos et al., which describes that exosomes of mast-cell origin can alter the dendritic-cell phenotype towards maturation in a CD91-dependent manner (Skokos et al., 2003); however, exosomes from other cellular sources, including B cells, cannot stimulate dendritic-cell maturation. Mast-cell exosomes, which exhibit a different hsp repertoire from B-cell exosomes, might have unique properties such as the ability to express hsp60, to mediate these effects (Skokos et al., 2003). So far however, all these studies have involved steady-state exosomes, and have not addressed the importance of stress-mediated induction of hsp.

The data presented here demonstrate that exosomes from heat-stressed B cells, which express elevated levels of stress proteins, have no effect on the maturation status of MDDCs. There was no difference in the expression of CD83 (or other markers including CCR7 and CCR5, data not shown) in dendritic-cells following exposure to high levels of control or heat stressed exosomes, and there was no effect on cytokine production (IL-6, IL-10, IL-12 or others) by dendritic cells. Our explanation for this surprising lack of stimulatory function lies in the precise location of exosomal hsps. Direct analysis of the exosome surface by flow-cytometry and immuno-precipitation experiments has demonstrated that the hsps lie encapsulated within the exosome lumen even after heat stress (Fig. 4), a stimulus that can often redistribute hsps to, for example cellular surfaces (Feng et al., 2002; Feng et al., 2003). Exosomal-hsps are therefore not directly available for binding to dendritic-cell-surface receptors such as CD91 or other receptors that trigger phenotypic dendritic-cell changes (Basu et al., 2001; Delneste et al., 2002; Ohashi et al., 2000; Wang et al., 2001).

When searching for a possible function for exosome-borne stress proteins, one could imagine that extracellular or intracellular processing of exosomes are necessary to liberate the encapsulated hsps. Intracellular processing, which probably involves exosome-adhesion molecules (Clayton et al., 2004) that bind to the dendritic-cell surface before endocytosis and processing (Morelli et al., 2004), is certainly compatible with earlier studies documenting exosomes as antigen-delivery-vehicles to dendritic cells (Altieri et al., 2004; Andre et al., 2002; Wolfers et al., 2001). The relative potency of stress-derived exosomes in such cross-presentation mechanisms is the subject of ongoing studies in our laboratory.

This work was funded by the Leukaemia Research Fund, Cancer Research Wales, and European Commission Grant QLRT-2001-00093. We are grateful to Fabrice Andre and Laurence Zitvogel, Institute Gustave Roussy, Paris, for their assistance in exosome purification.