CD40-Independent Engagement of Mammalian hsp70 by Antigen-Presenting Cells¹

CD40 plays several significant roles in the functioning of APCs. CD40-CD40L interactions between CD4⁺ T cells and professional APCs condition the APC such that they are equipped for activation of CD8⁺ T cells (1, 2, 3). The signals generated through CD40 receptors allow for the maturation and subsequent alteration of the phenotype of the APC (4). In addition the interaction of CD40 and CD40L plays a crucial role in T cell-dependent Ab responses as evidenced by the hyperIgM syndrome, characterized by a genetic mutation in the CD40L gene (5).

Heat shock proteins (HSPs)³ bind to and stimulate dendritic cells (DCs) such that they acquire a mature phenotype; this is characterized by the up-regulation of several costimulatory molecules including B7-2, CD40, and MHC class II and secretion of cytokines IL-1, IL-12, TNF-α, and GM-CSF (reviewed in Ref. 6). In addition, peptides chaperoned by murine hsc73 (hsp70) are acquired by the DC and presented on MHC molecules to T cells. The engagement of DC by HSPs occurs through cell surface receptors. Recently, CD40 on APCs was shown to interact with mycobacterial but not mammalian hsp70 and this interaction led to signaling followed by secretion of cytokines and chemokines by the APCs (7). More recently, CD40 has been implicated in representation of hsp70-chaperoned peptides by APCs and has been suggested to be a receptor for mammalian hsp70 (8). Even more recently, Ohashi and colleagues (9) reported that immunization with hsp70 and an LCMV-derived antigenic peptide can break tolerance to the antigenic peptide expressed as a self Ag in transgenic mice and that the tolerance-breaking activity of hsp70 is not seen in CD40^−/− mice. Bacterial LPS acts in a manner similar to hsp70 but its action is independent of CD40.

In this study, we examine the role of CD40 in hsp70-APC interaction through a series of independent parameters including binding of hsp70 to immature and mature DCs, stimulation of translocation of NF-κB by hsp70-APC interaction, elaboration of cytokines in response to hsp70-APC interaction, representation of hsp70-chaperoned peptides and elicitation of Ag-specific cytotoxic T lymphocyte by immunization with hsp70-peptide complexes. These results point unambiguously to a CD40-independent means of interaction of mammalian hsp70 with APCs.

CD40-deficient mice on the H-2^b (B6.129P2-Tnfrsf5^tm1Kik) and H-2^d (CNCr.129P2-Tnfrsf5^tm1Kik) background were obtained from The Jackson Laboratory. CD40-deficient mice were designated CD40^−/− compared with their respective wild-types C57BL/6 and BALB/c (CD40^+/+), also from The Jackson Laboratory.

For DCs, bone marrow cells were harvested and cultured for 6 days in 200 ng/ml GM-CSF. Cells obtained at the end of 6 days were described as immature DCs. For maturation, immature DCs were cultured for an additional day in 100 ng of LPS. These cells were called mature DCs and had increased expression of CD40, B7 and MHC class II (data not shown). RAW264.7 and RAW309Cr.1 macrophage cell lines were obtained from American Type Culture Collection and maintained in the specified medium.

Hsp70 was purified from CT26 or EG7 cell pellets or from murine liver tissue as previously published with few modifications (10). In brief, tumor cells were lysed by dounce homogenization in 30 mM NaHCO₃ (pH 7.1) with protease inhibitors. Livers were homogenized in a blender in PBS with protease inhibitors. Lysate from tumor cells or liver was centrifuged at 100,000 × g. The supernatant was applied to an ADP-agarose column in buffer D (20 mM Tris-acetate, 20 mM NaCl, 15 mM 2-ME, 3 mM MgCl₂, 0.5 mM PMSF (pH 7.5)). Columns were washed with buffer D plus 0.5 M NaCl, eluted with 3 mM ADP in buffer D, buffer exchanged to 20 mM sodium phosphate and applied to a DEAE column. Protein was eluted with 150 mM buffer as a single protein as determined by SDS-PAGE and immunoblotting. Purification of hsp70 by this method retains the chaperoned peptide bound to hsp70 as shown previously (10) and also in this study. Hsp70 derived from liver was complexed to peptide as previously described (11). Peptides used were AH1–19 (RVTYHSPSYVYHQFERRAK) and OVA-20 (SGLEQLESIINFEKLTEWTS) with the presented epitope underlined and were synthesized at Genemed Synthesis.

To generate T cells, mice were intradermally immunized with 10 μg of hsp70 per mouse in a total volume of 100 μl. Two immunizations were given 1 wk apart. One week after the last immunization, spleens were harvested and cultured with peptide-pulsed irradiated cells. During the second week of culture, ConA supernatant was added and cultures were tested for cytotoxicity by ⁵¹Cr release assays 5 days later. Peptides used in the spleen cell stimulations and in the Cr release assay were the OVA8 (SIINFEKL, H-2^b system) and AH1 (SPSYVYHQF, H-2^d system).

Recombinant mouse receptor associated protein (RAP) was synthesized at Antigenics. Soluble CD40 and anti-CD40 mAb was purchased from R&D Systems. Mouse albumin, rat IgG, and LPS were purchased from Sigma-Aldrich. Rabbit p38 Ab was purchased from Cell Signaling Technology.

Mouse cytometric bead array kits for inflammation were purchased from BD Biosciences. ELISA kits for IFN-γ, IL-12p70, and IL-12p40 detection were purchased from Pierce. FITC conjugation kits were purchased from Sigma-Aldrich. All kits were used precisely according to the manufacturer’s recommendation. Nuclear extraction and NF-κB ELISA kits were purchased from ActiveMotif and used according to the manufacturer’s recommendations.

Murine hsp70 was purified from mouse tissues or cell lines (see Materials and Methods) and labeled with FITC. The hsp70 was analyzed by SDS-PAGE and immunoblotting before and after labeling with FITC (Fig. 1,A, inset). Hsp70-FITC preparations (10 μg/ml) were incubated with paraformaldehyde-fixed bone marrow-derived immature DCs from C57BL/6 mice. After removing excess unbound protein through repeated washing, cells were analyzed by flow cytometry. Hsp70-FITC was observed to stain the DCs extensively (Fig. 1,A). The binding studies were conducted in the presence of various CD40-competitor molecules (anti-CD40 Ab, CD40L, soluble CD40 (sCD40)) using subsaturating quantities of hsp70-FITC to increase the sensitivity of inhibition. Anti-CD40 Ab even at a 100-fold molar excess failed to interfere with the intensity of staining of DCs with hsp70-FITC (Fig. 1) and behaved identically to an isotype control Ab and mouse serum albumin. CD40L and excess soluble sCD40 also did not prevent labeling of the DCs with hsp70-FITC at 50-fold molar excess. In contrast, RAP, a ligand for the previously described hsp70 receptor CD91 (11), abolished completely the staining of DCs with hsp70-FITC at 50-fold molar excess. These experiments were performed in parallel with immature DCs isolated from CD40^−/− mice (H-2^b) with identical results: the intensity of staining of immature DCs by hsp70-FITC was comparable between wild-type and CD40^−/− mice and anti-CD40 Abs, CD40L, and sCD40 failed to interfere with staining of DCs with hsp70-FITC. RAP inhibited the staining of DCs with hsp70 in both strains.

The binding and competition studies were repeated with paraformaldehyde-fixed, LPS-matured DCs from CD40^+/+ and CD40^−/− mice (Fig. 1,B). As expected, mature DCs expressed higher levels of CD40 (data not shown), although hsp70-FITC staining of mature and immature DCs was comparable (see y-axes of Fig. 1, A and B). The competition data with mature DCs were identical to those with immature DCs.

As additional controls, DCs labeled with hsp70-FITC were competed with unlabeled hsp70 at a 50-fold molar excess concentration, leading to competition of >95% (Fig. 1,B, left). Soluble CD40 used in Fig. 1, A and B was tested and observed to inhibit the staining of DCs with PE-labeled anti-CD40 Ab (Fig. 1 B, right). RAP did not inhibit the staining of DCs by anti-CD40 Abs.

Studies similar to those shown in Fig. 1, A and B were conducted with primary cultures of peritoneal exudates cells (PEC) of CD40^+/+ and CD40^−/− mice and essentially identical results were obtained. None of the three CD40-binding/inhibiting reagents had any influence on staining of macrophage with hsp70-FITC, while RAP did (data not shown).

Two macrophage cell lines RAW264.7 and RAW309Cr.1 have previously been reported to be disparate in their ability to bind several HSPs including hsp70 (11). RAW264.7 and RAW309Cr.1 were stained with Abs to CD40 or CD91 (Fig. 2,A). Although both cell types had identical expression of CD40, only RAW264.7 expressed CD91. There was no detectable expression of CD91 on RAW309Cr.1 as shown previously (11) (Fig. 2,A). We now tested the ability of RAW264.7 and RAW309Cr.1 to bind hsp70. The two cells were fixed and stained with hsp70-FITC. RAW264.7 but not RAW309Cr.1 cells were noted to stain abundantly (Fig. 2,B). These data are summarized in Fig. 2,C and highlight the lack of correlation of hsp70 binding with CD40 expression. We studied the internalization of hsp70 by RAW264.7 cells in relation to CD40 on the cell surface. When cells were incubated at 4°C with hsp70, CD40 (red) and hsp70 (green) remained predominantly on the cells surface with only marginal internalization (Fig 2,D, left). When the temperature was raised to 37°C, cells endocytosed hsp70 as demonstrated by significant intracellular green staining of cells. However, unlike CD91 (12) CD40, which has not been demonstrated to be an endocytosing receptor, (13) remains on the cells’ surface (Fig 2 D, right).

The release of cytokines by HSP-stimulated DCs occurs via the NF-κB pathway (14). Initially, we established some parameters of translocation of NF-κB with CD40^+/+ or CD40^−/− DCs pulsed with LPS (Fig. 3,A). DCs from both sets of mice were pulsed with LPS (1eu/ml) for 20 min. Cells were lysed and nuclear extracts prepared. Equal quantities of nuclear extract were tested for the presence of NF-κB by ELISA. NF-κB appeared in nuclei in response to incubation with LPS (squares) within 20 min (Fig. 3,A and see Ref. 14). To test the specificity of the assay, it was conducted in presence of competing NF-κB binding oligonucleotides (that inhibited NF-κB) or mutated, noncompeting sequences that do not bind NF-κB. The NF-κB-competing oligonucleotides extinguished the signal obtained in the absence of any oligonucleotide while the noncompteing oligonucleotides did not (Fig. 3,A). In another experiment, DCs from wild-type or CD40^−/− mice were pulsed with 100 μg/ml hsp70 over several time points between 0 and 20 min. After the indicated amount of time, cells were lysed and nuclear extracts prepared and tested for the presence of NF-κB. The kinetics of NF-κB activation was identical between wild-type and CD40^−/− DCs (Fig. 3,B). The NF-κB activation obtained with hsp70 pulsing was competed out with NF-κB-binding oligonucleotides, but not mutated sequences that do not bind NF-κB (Fig. 3 C).

One functional outcome of the engagement of hsp70 with its receptor on APCs and the activation of NF-κB is the release of cytokines. Murine and mycobacterial hsp70 (TB) have been shown to stimulate APCs to release TNF-α, IL-12, and IL-1β, and to up-regulate costimulatory molecules. We tested whether release of cytokines by DCs stimulated with murine hsp70 was CD40-dependent. DCs were pulsed with hsp70 in increasing doses for 20 h. The supernatants were collected and tested for TNF-α and IL-12. Release of TNF-α (Fig. 3,D) or IL-12 (Fig. 3,E) by CD40^−/− DCs was observed to be comparable to DCs from wild-type mice. No differences were observed in the dose dependence or in the release of IL-4 and IL-5 (data not shown). p38 has been suggested as a signaling molecule that is phosphorylated when APCs are pulsed with hsp70 (15). p38 was analyzed in wild-type and CD40^−/− DCs that had been pulsed for 10 min with hsp70, by immunoblotting. No differences were observed in the quantity of p38 between the two types of DCs (Fig. 3 E, inset). Gp96 was immunoblotted as a loading control.

LPS (1eu/ml) pulsed DCs also released TNF-α (Fig. 3,D, inset). The LPS-mediated induction of cytokine release was, as expected, sensitive to inclusion of polymyxin B in the assay. Concerns have been raised in the literature regarding the possibility that the innate immunological activities of recombinant HSP preparations derive from contaminating LPS and are not inherent to HSPs (16, 17). We have always shared these concerns (14) and, as a consequence, do not use recombinant HSPs in our studies (10, 11, 14). Despite that precaution, we have always used a panel of tests to rule out contaminating LPS as a source of the innate immunological activities reported by us. In the present instance, the role of contaminating LPS was ruled out by the following observations: 1) hsp70 was purified from murine tissues and its LPS content, if any, was below the limit of detection (0.1eu/ml) and 2) inclusion of polymyxin B in the assay did not diminish the amount of cytokine released when stimulation was done with hsp70 but did inhibit the same parameter significantly when stimulation was done with LPS (Fig. 3 D, inset).

Similar studies were performed with CD11b⁺ PECs instead of DCs from CD40^+/+ and CD40^−/− mice and with identical results: hsp70-elicited release of cytokines was similar in PECs from wild-type or CD40^−/− mice (data not shown).

The receptor-dependent uptake of hsp70-peptide complex leads to the processing and presentation of the chaperoned peptides by MHC molecules of the APCs to T cells (11). This phenomenon of representation is detected by the following assay (18). HSP peptide complexes are incubated with receptor-bearing APCs and T cells specific to the peptides chaperoned by hsp70. Activity of T cells is quantified by release of IFN-γ, as a measure of representation. The peptide complexed to hsp70 is usually a 19 or 20mer containing the 8/9mer sequence ultimately presented by MHC I to T cells. Representation assays were conducted with hsp70 complexed to AH1–19 (hsp70-AH1–19), APCs from wild-type or CD40^−/− mice and cloned T cells specific for AH1 peptide of gp70 of murine leukemia virus (19) presented on L^d. IFN-γ secretion was measured as an index of T cells stimulation. DCs from wild-type and CD40^−/− mice were able to take up hsp70 (40 μg/ml) and present the AH1 peptide to T cells with equal efficiency (Fig. 4). Eliminating either one of the components of the representation system (i.e., T cells, DCs, or hsp70) abrogates representation. To ensure that DC numbers, and thus receptor molecules, were limiting, DC numbers were titrated down from the optimal number of 10,000. At all numbers of DCs used there were no differences between representation with wild-type and CD40^−/− DCs, and no re-presentation was observed with less than 1000 DCs (data not shown). Hsp70 complexed to an OVA-20 peptide was not able to stimulate IFN-γ release demonstrating the specificity of the system. We extended these observations by including various CD40 inhibitors at 50-fold molar excess concentrations into the representation assay (Fig. 4). Soluble CD40, CD40L, and anti-CD40 mAbs, added as competitors, which should inhibit hsp70 from binding cell surface CD40, all failed to inhibit representation as did the negative control protein serum albumin. In contrast, RAP, a CD91 binding protein, inhibited representation (Fig. 4) (20). As seen from the data, identical results were obtained with DCs from wild-type and CD40^−/− mice.

Groups of BALB/c (H-2^d) or C57BL/6 (H-2^b) mice were immunized twice at weekly intervals with 10 μg of hsp70 derived from CT26 (AH1 expressing) or E.G7 (AH1 nonexpressing, OVA expressing) cells. CD40^−/− mice obtained in both backgrounds (see Materials and Methods) were immunized with each of the two hsp70 preparations. Control mice were immunized with PBS. Spleen cells of immunized mice were harvested 1 wk after the last immunization and were cultured in vitro, and priming was monitored by the ability of cultured cells to lyse ⁵¹Cr labeled targets in a peptide-dependent manner (Fig. 5). In the OVA system, C57BL/6 mice immunized with E.G7-derived hsp70 generated titratable and K^b/SIINFEKL-specific T cell responses independent of their CD40 phenotype (Fig. 5,A). CD40^+/+ and CD40^−/− mice showed identical responses. Specificity was demonstrated by the lack of K^b/SIINFEKL-specific responses in mice immunized with CT26-derived hsp70 or PBS. In the AH1 system, mice immunized with CT26-derived hsp70 generated T cells that lysed AH1 pulsed cells (Fig. 5 B). The expression of CD40 in BALB/c mice did not affect the ability to prime immune responses. Mice immunized with PBS or E.G7-derived hsp70 did not mount an L^d/AH1-specific response.

We have examined in this study in vitro and in vivo the role of CD40 in interaction of murine hsp70 with a number of murine APCs (immature and mature primary cultures of DCs, primary cultures of PECs and two macrophage lines), using several independent read-outs (binding, translocation of NFκB, cytokine release, representation of hsp70-chaperoned peptides). A number of independent tools including CD40^−/− mice, soluble CD40 (sCD40), CD40L, and anti-CD40 Abs have been used. The various consequences of hsp70-APC interaction were compared between CD40^+/+ and CD40^−/− mice and were found to be identical in kinetics and magnitude. As CD40^−/− mice have been generated by targeting the CD40 gene for homologous recombination, the CD40^−/− mice retain some CD40 sequences. The lack of functional differences between the mice may therefore be attributable to these retained sequences. To address this issue, the various assays have been conducted in presence of a number of additional reagents such as CD40L, anti-CD40 Ab, and sCD40. All the experimental conditions tested in all possible combinations indicate unambiguously that all known effects of mammalian hsp70 on APCs can be mediated in a CD40-independent manner.

These results should be considered in light of previous data where CD40 was suggested to be a receptor for mammalian hsp70. Becker et al. (8) made that claim on basis of the following observations. First, they observed association of recombinant GST-tagged CD40 with murine hsp70 in cell lysates. This association was enhanced in the presence of excess ADP and peptide and inhibited by the hsp70 cochaperone Hip. Second, murine hsp70 could bind APCs better after the APCs were stimulated with LPS. As LPS-stimulated APCs express higher levels of a number of molecules including CD40, the authors argued that their data are consistent with a role for CD40 as an hsp70 receptor. Third, APCs take up peptides with CD40 and that such uptake is hsp70-facilitated. The second and third observations are consistent with hsp70-CD40 interaction but do not constitute evidence to that interaction. The first observation claims to show hsp70-CD40 interaction and requires careful scrutiny in light of the data shown in Fig. 1. Finally in this regard, the claim by Becker et al. that CD40 is a representation receptor for hsp70-peptide complexes is inconsistent with the observations that CD40 is not an endocytosing receptor, but purely a signaling receptor (13). The data shown in the present study, as well as the examination of the data presented by Becker et al. combined with the known nature of the CD40 molecule argue overwhelmingly against the idea that CD40 is a receptor for mammalian hsp70.

In several antigenic systems, depletion of CD4 T cells has no impact on the ability to prime CD8 T cells following hsp70 immunization (21, 22, 23), an observation that is remarkably similar to mice immunized with another HSP, gp96 (24). Because CD40-ligand is provided largely by CD4 T cells, these data would lend support to our conclusions in this study, a lack of a necessary interaction between hsp70 and CD40. As mentioned in the Introduction, the idea of the hsp70-CD40 interaction may appear to draw indirect support from the work of Millar et al. (9) who reported recently that immunization with recombinant hsp70 and an LCMV-derived antigenic peptide can break tolerance to the antigenic peptide expressed as a self Ag in transgenic mice and that the tolerance-breaking activity of hsp70 is not seen in CD40^−/− mice. LPS can also act precisely in the same capacity as hsp70 but this effect of LPS is not abrogated in CD40^−/− mice. The requirement of CD40 for the activities of hsp70 suggest that CD40 is, most likely, involved in some broader role than as a receptor for hsp70, and indeed, Millar et al. do not suggest CD40 to be a receptor for hsp70.

Interestingly, CD40 was reported as a signaling receptor for mycobacterial hsp70 by Lehner and colleagues (7). These authors detected binding of mycobacterial hsp70 with murine CD40 through direct binding studies and through inhibition by anti-CD40 Abs of chemokine secretion by macrophage upon stimulation with mycobacterial hsp70. They also specifically showed that CD40 binds mycobacterial but not murine hsp70. This work draws support from a recent study (25) where it was observed that CD40^−/− mice succumb to Mycobacterium tuberclosis infection, while CD40L^−/− mice are M. tuberculosis-resistant. The observation suggested the existence of an alternative ligand for CD40 and the authors suggest hsp70 to be this ligand. In studies in vitro, recombinant M. tuberclosis hsp70 was demonstrated to cause the release of IL-12 from DCs similar to that seen during infection; this response was absent in CD40^−/− mice, encouraging the authors to suggest an involvement of CD40 as a receptor for hsp70. Collectively, the studies of Wang et al. (7) and Lazarevic et al. (25) are consistent with each other and with the fact that CD40 is a signaling receptor.

These studies raise the very interesting possibility that the mammalian CD40 can distinguish between mammalian and mycobacterial hsp70 despite the extensive homologies between them. This would make the phylogenetic variations among HSPs to be candidates for patterns associated with microbial pathogens and CD40 to be a candidate for a patterns associated with microbial pathogens receptor.

The authors have no financial conflict of interest.