Heat shock proteins (HSP) Hsp70 and gp96 prime class I-restricted cytotoxic T cells against Ags present in the cells from which they were isolated. The immunization capacity of HSPs is believed to rely on their ability to bind antigenic peptides. In this study, we employed the well-established OVA and β-galactosidase (β-gal) antigenic model systems. We show that in vitro long-term established OVA and β-gal-specific CTL clones release TNF-α and IFN-γ when incubated with Ag-negative Hsp70 and gp96. In the absence of antigenic peptides, HSP-mediated secretion of TNF-α and IFN-γ requires cell contact of the APC with the T cell but is not MHC-I restricted. Moreover, Hsp70 molecules purified from Ag-negative tissue, e.g., negative for antigenic peptide, are able to activate T cells in vivo, leading to significant higher frequencies in OVA-specific CD8+ T cells. In unprimed animals, these T cells lyse OVA-transfected cell lines and produce TNF-α and IFN-γ after Ag stimulus. Taken together our data show that, besides the well-established HSP/peptide-specific CTL induction and activation, a second mechanism exists by which Hsp70 and gp96 molecules activate T cells in vivo and in vitro.

Heat shock proteins (HSP)3 from tumor tissue are potent vaccines against tumors from which the HSPs have been purified 1 . Srivastava and coworkers 2, 3 have recently shown that gp96 and Hsp70 molecules purified from chemically induced tumors caused complete tumor regressions in in vivo tumor challenge protocols. Because HSPs from tumorigenic and normal tissue do not differ in their primary structure, peptides associated with gp96 and Hsp70 seem to account for the immunogenic capacities of HSP molecules 4 . Stripping of the HSP bound peptides abrogates the immunogenicity of the HSP preparation.

CD8+ T cells play a major role in HSP-mediated tumor regression, whereas immunization with tumor cells also requires CD4+ T cells 5 . The paradigm of HSP immunization has also been extended to other antigenic model systems: gp96 preparations isolated from cells expressing a given set of minor H Ags or expressing a transfected β-galactosidase (β-gal) protein 6 can be used to immunize and elicit CTL responses against the minor Ags or β-gal-derived peptides, respectively. Confirmation of these results has been obtained in an independent antigenic system in which vesicular stomatitis virus (VSV)-infected cells were the source for the purified gp96 molecules 7 . gp96 molecules isolated from VSV-infected cells contained the major H2-Kb binding epitope from the VSV nucleoprotein VSV52–59 (RGYVYQGL). This peptide associates with gp96 in VSV-infected cells regardless of the MHC haplotype of the cell providing a biochemical explanation for the “cross-priming” 8 phenomenon that can be observed after immunization with gp96 preparations.

All three described model systems (minor H, β-gal, and the VSV-Ags) investigating the function of HSP-mediated CTL induction depend on the presence of HSP-associated peptides. Minor H Ag-negative or uninfected cell lines, as a source of HSP, do not work effectively in these immunization protocols.

In this report, we analyzed a potential capacity of HSP molecules to activate T cells that are not restricted to HSP-bound peptides. We used the well-known OVA-antigenic system and purified Hsp70 and gp96 from the OVA-transfected cell line E.G7 9 . E.G7 cells process and present the H2-Kb binding epitope OVA257–264 (SIINFEKL) 10, 11 that can be detected in in vitro chromium release assays with the help of the H2-Kb/SIINFEKL-specific CTL clone 4G3 12 . In contrast to the described systems that were dependent on HSP-associated antigenic peptides, immunization with HSPs derived from OVA-negative tissue leads to the appearance of H2-Kb/SIINFEKL-specific CTL lines.

Eight- to 10-week-old BALB/c and C57BL/6 mice were obtained from Charles River WIGA (Sulzfeld, Germany) and maintained in the animal facilities at the Bernhard-Nocht-Institute. A-20 is a H-2d B-cell line, EL4 is a H-2b T cell lymphoma, and E.G7 cells are EL4 cells transfected with the chicken/OVA gene 9 . P13.1 cells (H-2d) are P815 cells transfected with the β-gal gene 6 . 4G3 represents a SIINFEKL/H-2b, 0805B a TPHPARIGL/H-2d-specific CTL clone. The synthetic SIINFEKL peptide was a kind gift of Prof. H. U. Weltzien (Max-Planck-Institute for Immunobiology, Freiburg, Germany). The β-gal peptide and CTL clone 0805B were provided by Dr. H. G. Rammensee (Department of Immunology, Institute of Cell Biology, Eberhard-Karls-University, Tübingen, Germany). All cells were cultured in RPMI 1640 medium supplemented with 5% FCS, 2β-ME, and l-glutamine. In the case of E.G7 250 μg/ml and in the case of P13.1 cells, 1 mg/ml of G418 was added as a selection antibiotic. CTL clones were supplemented with 10% FCS and IL-2 (100 U/ml). The CTL line 4G3 was restimulated weekly with irradiated E.G7 cells (7 × 105/ml). In the case of β-gal-specific 0805B cells, irradiated syngenic spleen cells (106/ml) and 1 μM β-gal peptide were incubated in 96-well round-bottom culture dishes (Greiner, Nurtingen, Germany). To induce peritoneal macrophages, mice were injected i.p. with 500 μl pristane (Sigma, Deisenhofen, Germany). Peritoneal exudate cells (PEC) were harvested 5–6 days later by rinsing the peritoneum with ice cold medium. Only freshly prepared PEC were used in the experiments described. FACS analysis of isolated PEC showed a >90% staining for the macrophage surface marker Mac-1. Abs to gp96 (anti-grp94, SPA-850) and to Hsp70 (SPA-820) were obtained from Stressgen Biotechnologies (Victoria, Canada). Quantification of TNF-α and IFN-γ was performed as described previously 13 . For IFN-γ ELISAs, we used XMG 1.2 (0.5 μg/ml) and RA-6A2 (2 μg/ml) mAbs, and for TNF-α-specific ELISAs, we used the Ab pairs G281-2626 (2 μg/ml) and MP6-XT3 (1 μg/ml) (PharMingen, Hamburg, Germany).

gp96 was purified as described with minor modifications 14 . Briefly, P13.1 and E.G7 cells were grown in roller bottles to generate a 20-ml cell pellet. Mouse liver-derived gp96 was purified from 60 g and Hsp70 from 20-g cell pellets. The pellets were homogenized in four volumes of hypotonic buffer (10 mM NaHCO3 and 0.5 mM PMSF (pH 7.0)) and centrifuged at 100,000 × g for 90 min. For gp96 purification the 50–70% ammoniumsulfate precipitate was solubilized in Con A binding buffer (125 mM NaCl, 20 mM Tris-HCl, 1 mM MgCl2, and 1 mM CaCl2 (pH 7.2)) and was applied to a Con A/Sepharose column followed by extensive washing. Con A-bound material was eluted with 10% α-methyl-mannoside. The eluate buffer was exchanged to DEAE-binding buffer (5 mM sodium phosphate and 300 mM NaCl (pH 7.0)), employing PD10 gel filtration units (Pharmacia, Freiburg, Germany). This partially purified gp96 material was applied to a DEAE-Sepharose column (Pharmacia) and washed and bound material was eluted with DEAE-binding buffer containing 700 mM NaCl.

Hsp70 from P13.1, E.G7, and liver cell pellets was purified from the 100,000 × g supernatant by first exchanging the lysis buffer to ADP-binding buffer (20 mM Tris-Acetate, 2 mM MgCl2, 15 mM 2-ME, and 20 mM NaCl (pH 7.0)). The material was applied to an ADP-agarose column (Sigma). The column was washed subsequentlywith ADP-binding buffer containing 500 mM NaCl and 20 mM NaCl. The ADP-bound material was eluted with ADP-binding buffer containing 3 mM ADP. The elution buffer was exchanged to Hsp70-DEAE-binding buffer (20 mM sodium phosphate and 20 mM NaCl (pH 7.0)) again using PD10 columns. The eluate was applied to a DEAE-column and washed. Hsp70 coeluted at a concentration of 150 mM NaCl in DEAE-binding buffer. Fractions of purified gp96 and Hsp70 material were tested in SDS-PAGE and Western blot analysis using mAb specific for Hsp70 and gp96. Fractions containing Hsp70 and gp96 as the major proteins were used for additional experiments. Protein concentration was determined employing Coomassi Plus Protein Reagent (Pierce, Rockford, IL).

Mice were immunized i.p. with 30 μg of gp96 purified from liver, P13.1 cells, or E.G7 cells, 2 × 107 irradiated P13.1 or E.G7 cells in 500 μl PBS, or with PBS only. Mice were sacrificed on day 10. For Hsp70 immunization, mice were injected twice weekly s.c. with 15 μg of Hsp70, 2 × 107 irradiated P13.1, or E.G7 cells in 100 μl PBS, and the mice were sacrificed 7 days after the second immunization. For bulk culture, splenocytes (5 × 106/ml) were restimulated with 1 μM of OVA or β-gal peptide in upright 10-ml culture flasks. CTL activity was tested in a standard chromium release assay five days after restimulation. For frequency analysis, splenocytes (106/ml) were cultured in 96-well round-bottom culture dishes with 1 μM of β-gal peptide or irradiated E.G7 cells (5 × 105/ml). Fifty splenocyte lines were tested for Ag-specific lysis on day 6 after restimulation. E.G7- and P13.1-specific lysis was calculated by subtracting the specific lysis of Ag-negative target cells (EL4, P815) from the specific lysis of the Ag-expressing target cell (E.G7, P13.1) for each point.

Indicated types of APC or freshly prepared PEC (104/well), CTL clones (105/well) and indicated amounts of HSP, LPS, or peptide were cocultured in 96-well round-bottom culture dishes for 30 h. A total of 50 μl of the supernatant was tested for IFN-γ and TNF-α in a standard ELISA. In trans-well experiments, 0.2-μm anopore membrane transwell strips (Nunc, Naperville, IL) were employed to block direct cell contact between CTL and APC within one well.

HSP preparations are able to immunize against tumors and to induce CTLs in vivo as described by coworkers of Srivastava 13 and other groups. Moreover, HSP are highly efficient in chaperoning associated peptides into the MHC-I related Ag presentation pathway, as shown in in vitro re-presentation assays 3 . Because in some of our T cell assays (M.B. and A.v.B. unpublished data) the addition of Ag-negative HSP preparations resulted in the induction of lymphokines and proliferation of T cells, we started to analyze a possible T cell-activating potential of HSP that functions irrespective of the HSP-associated peptides.

To this end, we purified Hsp70 and gp96 molecules from the OVA-transfected cell line E.G7 and control tissues (Fig. 1). We have recently shown 15 that Hsp70 and gp96 molecules purified from E.G7 cells are associated with the major H2-Kb binding epitope OVA257–264. As shown in Fig. 2, the H2-Kb/OVA257–264-specific CTL clone 4G3 releases substantial amounts of TNF-α when E.G7-purified Hsp70 (Fig. 2,A) or gp96 (Fig. 2,B) molecules were added to the cultures. However, the same amount of TNF-α could be detected in cultures when Ag-negative HSP, e.g., liver or RMA-derived Hsp70/gp96, was titrated into the cell culture system. To exclude the possibility that the response to Ag-negative HSPs is a characteristic of the 4G3 clone, we repeated the experiment with a β-gal-specific CTL clone, 0805B. Again, we observed no Ag-specific induction of IFN-γ after the addition of P13.1-derived gp96 (Fig. 3, A and B) and Hsp70 (Fig. 3 C), because Hsp70 and gp96 preparations purified from liver tissue showed a comparable effect. Although gp96 was purified by Con A-containing columns, a contamination with Con A can be largely ruled out, because the stimulating activity in our gp96 preparations could only be observed in DEAE fractions in which gp96 was present. Moreover, recombinant His-tagged gp96 purified in the absence of Con A showed comparable results (S. Moré, M. B., and A.v.B., unpublished data).

FIGURE 1.

Purification of Hsp70 and gp96. A, SDS-PAGE followed by silver-staining of purified Hsp70 (left panel) and gp96 (right panel) material from liver tissue. A 10-kDa ladder protein standard is indicated on the right in the silver staining. B, Western blot analysis of the same Hsp70 and gp96 preparations as shown in A with the anti-Hsp70 mAb SPA-820 and the anti-gp96 mAb SPA-850. Numbers on the right indicate relative molecular masses in kDa. Proteins with low molecular masses in the gp96 Western blot probably represent gp96 degradation products.

FIGURE 1.

Purification of Hsp70 and gp96. A, SDS-PAGE followed by silver-staining of purified Hsp70 (left panel) and gp96 (right panel) material from liver tissue. A 10-kDa ladder protein standard is indicated on the right in the silver staining. B, Western blot analysis of the same Hsp70 and gp96 preparations as shown in A with the anti-Hsp70 mAb SPA-820 and the anti-gp96 mAb SPA-850. Numbers on the right indicate relative molecular masses in kDa. Proteins with low molecular masses in the gp96 Western blot probably represent gp96 degradation products.

Close modal
FIGURE 2.

Unspecific stimulation of 4G3 cells with Ag-positive and Ag-negative HSP. Shown is the TNF-α release (y-axis) after the addition of titrated amounts of gp96 (A) and Hsp70 (B) to cultures containing 5 × 105/ml 4G3 cells alone (○), pristane induced H-2b-positive macrophages (2.5 × 105/ml, □), or APC and 4G3 cells together (▪). Shown is a typical result of four independent experiments.

FIGURE 2.

Unspecific stimulation of 4G3 cells with Ag-positive and Ag-negative HSP. Shown is the TNF-α release (y-axis) after the addition of titrated amounts of gp96 (A) and Hsp70 (B) to cultures containing 5 × 105/ml 4G3 cells alone (○), pristane induced H-2b-positive macrophages (2.5 × 105/ml, □), or APC and 4G3 cells together (▪). Shown is a typical result of four independent experiments.

Close modal
FIGURE 3.

OVA257–264 and β-gal-specific CTL are stimulated irrespective of the HSP source. A total of 2.5 × 105/H-2d-positive pristane-induced BALB/c macrophages were incubated alone (□) or together with 5 × 105 OVA257–264-specific 4G3 cells (▪) or β-gal-specific 0805B cells (▴) for 24 h with titrated amounts of gp96 purified from liver (A) or P13.1 cells (B). Shown is the induced IFN-γ in the supernatant (y-axis). C, IFN-γ was determined in the supernatants of cultures as described for Fig. 3 A. Hsp70 (10 μg/ml) purified from liver tissue with the help of ADP/agarose columns (ADP) or ATP/agarose columns (ATP) was compared with P13.1-derived Hsp70. Shown is a typical result of three independent experiments.

FIGURE 3.

OVA257–264 and β-gal-specific CTL are stimulated irrespective of the HSP source. A total of 2.5 × 105/H-2d-positive pristane-induced BALB/c macrophages were incubated alone (□) or together with 5 × 105 OVA257–264-specific 4G3 cells (▪) or β-gal-specific 0805B cells (▴) for 24 h with titrated amounts of gp96 purified from liver (A) or P13.1 cells (B). Shown is the induced IFN-γ in the supernatant (y-axis). C, IFN-γ was determined in the supernatants of cultures as described for Fig. 3 A. Hsp70 (10 μg/ml) purified from liver tissue with the help of ADP/agarose columns (ADP) or ATP/agarose columns (ATP) was compared with P13.1-derived Hsp70. Shown is a typical result of three independent experiments.

Close modal

Although the gp96 preparation derived from P13.1 cells (Ag-positive) induced slightly higher amounts of IFN-γ when compared with liver-derived gp96 (Fig. 3,B), the inclusion and effective stimulation of the OVA257–264-specific CTL clone 4G3 in the same experiment revealed that this induction was independent of gp96-associated peptides and reflected a higher stimulating capacity of that particular gp96 preparation. Interestingly, in this set of experiments we used H-2d-positive macrophages induced in BALB/c mice as accessory cells. Because the 4G3 clone is H-2Kb restricted, the induction of IFN-γ was triggered by the “wrong” MHC-I haplotype. This result is not due to an alloreactive reaction of 4G3, because coculture of H-2d-positive PEC and 4G3 in the absence of HSP did not induce IFN-γ or TNF-α. Interestingly, Hsp70 purified from liver stimulated the CTL clones to release IFN-γ when either ATP-agarose or ADP-agarose was applied for purification (Fig. 3 C). Because ATP-agarose in contrast to ADP-agarose was described 13 to release endogenous Hsp70-associated peptides, these results indicate that Hsp70 molecules, irrespective of the complexed peptide-ligands, stimulate 4G3 and 0805B cells. However, the described experimental setup does not reveal the cytokine source, either T cells or macrophages being responsible for the secretion of TNF-α or IFN-γ, respectively. However, we think that the CTL clones in our system are the cytokine-producing cells, because experiments in which the CTL clones, following an incubation of 2 h with gp96, were separated from the adherent macrophages, resulted in TNF-α and IFN-γ release only in the separated T cell fraction and not in the adherent macrophages (data not shown). Thus, a short HSP-mediated stimulus is sufficient to induce cytokine secretion in 4G3 and 0805B T cells.

Another typical read-out system to determine the specific activation of CTL clones is the chromium release assay, which in contrast to the quantification of cytokines, lasts only 4 h. Therefore, we repeated the experiments shown in Fig. 3 with 51Cr-labeled macrophages. As shown in Fig. 4, incubation of gp96 derived from P13.1 and liver cells together with CTL clones resulted in killing 51Cr-labeled macrophages. Again, the CTL-mediated lysis was not haplotype-restricted, e.g., 0805B cells (β-gal/H2-Ld-restricted) lysed to a significant degree C57Bl/6 macrophages (H-2b).

FIGURE 4.

51Cr release assay with gp96-modified accessory cells. A total of 2 × 105/ml 4G3 cells (▪) or 0805B cells (▴) were incubated with 2 × 104/ml 51CR-labeled PEC (H-2b) in the presence of the indicated amount of liver or P13.1-derived gp96. As a control served a SIINFEKL/TPHPARIGL peptide mixture (Ag) or PEC (H-2b, ▵) incubated without CTL clones in a 4-h standard chromium release assay. Specific lysis (triplicate values) is shown on the y-axis. Lysis of unmodified PEC was <15% and is subtracted from each value. Comparable results were obtained with H2d-positive PEC (data not shown).

FIGURE 4.

51Cr release assay with gp96-modified accessory cells. A total of 2 × 105/ml 4G3 cells (▪) or 0805B cells (▴) were incubated with 2 × 104/ml 51CR-labeled PEC (H-2b) in the presence of the indicated amount of liver or P13.1-derived gp96. As a control served a SIINFEKL/TPHPARIGL peptide mixture (Ag) or PEC (H-2b, ▵) incubated without CTL clones in a 4-h standard chromium release assay. Specific lysis (triplicate values) is shown on the y-axis. Lysis of unmodified PEC was <15% and is subtracted from each value. Comparable results were obtained with H2d-positive PEC (data not shown).

Close modal

To correlate the T cell-stimulating activity to HSP molecules we tested individual fractions eluting from the DEAE column in a T cell assay. Although Hsp70 elutes from the DEAE column as the final purification step as the most dominant protein (compare Fig. 1), some Hsp70 preparations contained an additional protein band corresponding to 40 kDa. The identity of this protein is unclear, however, Hsp70 is known to interact with different proteins such as Hsp40 or Hip 16 , respectively, with both proteins displaying a molecular mass of ∼40 kDa in SDS-PAGE. Fig. 5 shows that the induction of IFN-γ in 4G3 cells (Fig. 5,C) correlates with the elution profile of Hsp70 molecules visualized in silver staining (Fig. 5,A) and Western blot analysis (Fig. 5,B) and not, for example, with the copurifying protein (Fig. 5,A, lane 2) that showed a different migration shift (smaller than Hsp70) and eluted before Hsp70 molecules from the DEAE column (Fig. 5 C, lane 2).

FIGURE 5.

T cell-stimulating activity coelutes with Hsp70. Purified Hsp70 preparations derived from liver tissue were collected as 2-ml fractions. Aliquots of the individual fractions were either stained with silver (A) or blotted with anti Hsp70-specific mAbs (B) and used in a T cell assay (C) as described for Fig. 2. Centricon-50, enriched fractions (lanes 1–7, 50 μl each), were added to cultures containing 4G3 cells (1 × 106/ml), □), pristane-induced PEC (H-2b, 5 × 105/ml, ▵) or 4G3 cells and PEC (▪). Numbers on the right indicate molecular mass markers in kDa. The marker in A is a 10-kDa protein marker (Life Technologies, Eggenstein, Germany).

FIGURE 5.

T cell-stimulating activity coelutes with Hsp70. Purified Hsp70 preparations derived from liver tissue were collected as 2-ml fractions. Aliquots of the individual fractions were either stained with silver (A) or blotted with anti Hsp70-specific mAbs (B) and used in a T cell assay (C) as described for Fig. 2. Centricon-50, enriched fractions (lanes 1–7, 50 μl each), were added to cultures containing 4G3 cells (1 × 106/ml), □), pristane-induced PEC (H-2b, 5 × 105/ml, ▵) or 4G3 cells and PEC (▪). Numbers on the right indicate molecular mass markers in kDa. The marker in A is a 10-kDa protein marker (Life Technologies, Eggenstein, Germany).

Close modal

Although strictly dependent on the presence of accessory cells (PEC), our previous results suggested that the stimulation of OVA and β-gal-specific CTL clones with Hsp70 and gp96 molecules was not MHC restricted. To confirm this finding, we repeated the experiments with a broadened set of accessory cells. We incubated Hsp70 and gp96 molecules together with 4G3 (Fig. 6,A) or 0805B (Fig. 6,B) cells, respectively, and MHC-I matched and mismatched macrophages. As shown in Fig. 4, HSP-mediated TNF-α secretion does not depend on the expression of the “correct” TCR-restricting MHC-I molecules on the accessory cells. H-2d and H-2b molecules induced TNF-α equally well after conincubation with gp96 and Hsp70 molecules. However, this HSP-mediated TNF-α release was dependent on the type of APC. MHC-I expressing tumor cell lines (RMA) as well as B cell lines (A20) or a long-term cultivated macrophage cell line (J774, data not shown) were not able to induce TNF-α. Thus, stimulation of CTL clones with HSPs obviously requires certain properties of macrophages that are not present on long-term cell lines but on freshly ex vivo cultured macrophages.

FIGURE 6.

HSP-mediated stimulation is not MHC-I restricted. A total of 5 × 105/ml 4G3 cells (A) or 0805B cells (B) were incubated with the indicated type of APC as described for Fig. 2. TNF-α in the supernatant (y-axis) was determined after the addition of 10 μg/ml gp96 or Hsp70 molecules (liver, ATP purified; 10 μg/ml). Shown is a typical result of three independent experiments.

FIGURE 6.

HSP-mediated stimulation is not MHC-I restricted. A total of 5 × 105/ml 4G3 cells (A) or 0805B cells (B) were incubated with the indicated type of APC as described for Fig. 2. TNF-α in the supernatant (y-axis) was determined after the addition of 10 μg/ml gp96 or Hsp70 molecules (liver, ATP purified; 10 μg/ml). Shown is a typical result of three independent experiments.

Close modal

The release of TNF-α and IFN-γ induced by both HSP preparations, Hsp70 and gp96, is not dependent on the source of the purified HSP preparation, e.g., TNF-α and IFN-γ production could be measured after incubation with both Ag-positive and Ag-negative HSP preparations. To exclude the possibility that the observed TNF-α in cultures containing macrophages is due to contaminations with endotoxins in the process of HSP purification, we compared our HSP preparations to LPS at concentrations that induced comparable TNF-α secretion. We performed two series of additional experiments. In the first set, we boiled LPS and the HSP preparations prior to the incubation with CTL clones and macrophages (data not shown). In the second set of experiments, we incubated CTL clones and macrophages together and thus allowed cell contact, whereas in a parallel experiment we prevented direct cell contact with the help of trans-well stripes (Fig. 7). Either boiling or the prevention of direct cell contact completely abolished the TNF-α secretion by coincubation with HSPs, whereas the TNF-α inducing effect of LPS in both cases is only mildly (by boiling) or not at all (in the trans-wells) affected. These experiments largely rule out: 1) that contaminating endotoxins and 2) that a cell contact-independent cytokine cross-talk are responsible for the cytokine inducing effects.

FIGURE 7.

HSP-mediated activation of T cells requires cell contact. 4G3 cells and H-2b-positive pristane-induced macrophages were incubated together (PEC and CTL) or were separated by trans-well stripes (PEC//CTL) in 96-well round-bottom culuture plates. After the addition of 30 μg/ml gp96 (A) or 10 μg/ml LPS (B), TNF-α in the supernatant of the cultures was determined as described in Materials and Methods. C, As an internal control for the integrity of the trans-well stripes served OVA257–264-pulsed macrophages. Shown is the mean of triplicate values representative for two independent experiments.

FIGURE 7.

HSP-mediated activation of T cells requires cell contact. 4G3 cells and H-2b-positive pristane-induced macrophages were incubated together (PEC and CTL) or were separated by trans-well stripes (PEC//CTL) in 96-well round-bottom culuture plates. After the addition of 30 μg/ml gp96 (A) or 10 μg/ml LPS (B), TNF-α in the supernatant of the cultures was determined as described in Materials and Methods. C, As an internal control for the integrity of the trans-well stripes served OVA257–264-pulsed macrophages. Shown is the mean of triplicate values representative for two independent experiments.

Close modal

Next, we analyzed the Ag-independent stimulating activity of Hsp70 molecules also in vivo. Two batches of Hsp70 were prepared; one was derived from E.G7 cells (OVA+) and the second was derived from normal mouse liver tissue. In a first set of experiments, we immunized four groups of C57BL/6 mice: the first group were control mice immunized with PBS (negative control group), the second group received 2 × 107 irradiated E.G7 cells (OVA+, positive control group), the third group was immunized i.p. with E.G7-derived Hsp70, and the fourth group with liver-derived Hsp70. After 7 days, mice were sacrificed, and spleen cells were cultured in the presence of antigenic OVA peptide as described in Materials and Methods. After 6 days of culture, induced OVA257–264-specific CTLs were tested in in vitro cytotoxic chromium release assays. As shown in Fig. 8, immunization with E.G7 cells resulted in a significantly higher frequency of OVA-specific CTL when compared with PBS. However, priming of OVA-specific CTL with Hsp70 worked efficiently with both employed Hsp70 preparations; that is, also the OVA-Ag-negative liver-derived Hsp70 was able to induce OVA-specific CTL that were able to lyse E.G7 cells (Fig. 8) and to secrete TNF-α and IFN-γ after coincubation with E.G7 cells (data not shown). The induced CTL clones were OVA257–264/H2-Kb-specific, because killing could be blocked by preincubation with H2-Kb- (but not with H2-Db) specific mAb (data not shown).

FIGURE 8.

Increased in vivo induction of OVA-specific CTL is independent of the Hsp70 source. Three mice per group were immunized with PBS, 2 × 107, irradiated-E.G7 cells, liver-derived Hsp70, or E.G7-derived Hsp70. Mice were sacrificed on day 7 after immunization as described in Materials and Methods. Shown is the specific lysis of E.G7 cells (y-axis), e.g., unspecific lysis of EL4 cells is subtracted for each value of 50 individual short time cell lines (○) measured in a standard 51Cr release assay. Background lysis of EL4 cells varied in all four groups from 0 to 25%. Similar results were obtained with CTL derived from bulk cultures (data not shown). Shown is a typical result of four independent experiments.

FIGURE 8.

Increased in vivo induction of OVA-specific CTL is independent of the Hsp70 source. Three mice per group were immunized with PBS, 2 × 107, irradiated-E.G7 cells, liver-derived Hsp70, or E.G7-derived Hsp70. Mice were sacrificed on day 7 after immunization as described in Materials and Methods. Shown is the specific lysis of E.G7 cells (y-axis), e.g., unspecific lysis of EL4 cells is subtracted for each value of 50 individual short time cell lines (○) measured in a standard 51Cr release assay. Background lysis of EL4 cells varied in all four groups from 0 to 25%. Similar results were obtained with CTL derived from bulk cultures (data not shown). Shown is a typical result of four independent experiments.

Close modal

Similar results were obtained with gp96 preparations derived from E.G7 and liver tissue. In contrast to Hsp70, gp96 rather weakly induced OVA-specific CTL in the liver/gp96 and the E.G7/gp96 groups (data not shown). Thus, the presence of OVA peptides did not increase the HSP-mediated in vivo priming of OVA-specific CTL. However, in agreement with earlier reports, we were able to show that gp96 immunization in the β-gal system resulted in an induction of β-gal-specific CTL after the immunization with P13.1-derived gp96, whereas gp96 from liver tissue did not induce a specific CTL response at all (Ref. 6 and data not shown).

Our results suggest that besides the well-established HSP/peptide-specific priming, a second mechanism may exist that is able to activate Ag-specific T cells. Depending on the Ag, this result manifests itself as activation of T cell clones in vitro (OVA/β-gal) or additionally as enhanced in vivo priming of CTL (OVA). Immune responses to HSP that have been detected after exposure to a broad spectrum of infectious agents, especially cellular responses directed against mycobacterial Hsp70 molecules, are profound 17 . It seems that the immune system is routinely stimulated to respond to Hsp70, and such stimulation may cause an expansion of Hsp70-reactive cells 18 . Although surface expression of Hsp70 molecules on some APCs and tumor tissue has been reported 19 , Hsp70 is normally located in the cytoplasm of cells, meaning that the immune system is rarely confronted with Hsp70/peptide complexes. Thus, one way to explain our findings is that Hsp70 and gp96 molecules themselves, irrespective of bound ligands, are capable of inducing activation signals in the T cell.

On the other hand, HSP-associated peptides seem to reflect the whole or at least a great part of the interior world of a given cell type, enabling the induction of a broad spectrum of Ag-specific CTL clones. Members of the Hsp70 family bind preferentially linear polypeptide fragments, and therefore represent central components of the protein folding machinery in the cell 20, 21 . In this context, it could be shown that gp96, an endoplasmic reticulum resident HSP, is able to induce crosspriming after immunization with gp96 preparations 22 . One could imagine that even Hsp70 preparations derived from OVA-negative tissue are associated with cross-reactive self-peptides or other ligands that lead to activation of OVA-specific T cells. In our in vitro experiments, we can exclude the possibility that cross-reactive “non-OVA” peptides are responsible for the observed stimulation of the H-2Kb/OVA257–264-specific CTL clone 4G3, because stripping of liver gp96 and Hsp70-associated peptides did not sensitize this clone on RMA target cells (Fig. 3 C and Refs. 13 and 15).

Immunization with irradiated cells worked very efficiently in our in vivo experiments. A tempting working hypothesis would be that these irradiated APCs do not only display antigenic peptides in context of MHC-I molecules on their cell surfaces but also release immunogenic HSP/peptide complexes after cellular breakdown. This release of HSP molecules could reflect a danger signal 23 , attracting many different T cells and other members of the immune system to start a broad primary immune response. However, replacing ADP-purified Hsp70 preparations (e.g., still complexed with endogenous peptides) with ATP-purified HSP molecules (e.g, peptide-negative) using a method first described by Tamura et al. 13 resulted in comparable results in vivo and in vitro arguing against a central role for peptides at least for the OVA-system (data not shown and Fig. 3 C).

Most of the experiments published so far that concentrated on the induction of CTL or the protection from tumors employed gp96 preparations. However, recent reports also started to analyze the function of in vitro reconstituted Hsp70/peptide complexes 24 or used Hsp70 fusion proteins 25 . Interestingly, in vitro complexation of Hsp70 molecules with OVA-peptides led to the induction of OVA-specific CTLs. Because E.G7-purified Hsp70 (despite being associated with OVA257–26415) did not specifically generate an OVA-specific CTL response, one could argue that in vitro reconstituted Hsp70, due to the high number of associated synthetic peptides, has a higher immunizing capacity. Our in vivo data generally show that Hsp70 has a broader, and at least for OVA, unspecific stimulating capacity, whereas gp96 administration induced a specific amount of Ag-specific CTLs (in case of β-gal) or displayed no induction at all (in the case of E.G7-derived gp96). It may be worthwhile to see whether differences in the immunization capacities of Hsp70 and gp96 molecules can be further proven, which may help us to better understand the powerful immunological potential of HSP.

The mechanism that enables Hsp70 and gp96 molecules to trigger T cells in vitro is not understood. Because minimal amounts of HSPs are suffcient to induce protection in tumor-challenging protocols, putative Hsp70 and gp96 receptors on the cell surface that introduce the uptaken HSP/peptide complexes directly into the MHC-I pathway have been postulated. However, so far there is no evidence for the existence of such receptors. Obviously the choice of the correct accessory cells is very important for detecting HSP-induced re-presentation of HSP/peptide complexes. To this end all long-term cultured cell lines used as accessory cells did not work in our assays; only freshly isolated macrophages responded to the coincubation with HSP molecules. Although pristane-induced PEC contain a few contaminating T cells (<5%), these T cells do not seem to be responsible for the observed effects, because cultures of PEC and HSP without additional T cell clones did not contain significant amounts of TNF-α and IFN-γ. It may well be that only these ex vivo purified macrophages express surface structures that are able to interact with HSP complexes, whereas in vitro cultured cells have lost such expression markers. The induction of TNF-α and IFN-γ is not dependent on the expressed MHC-I haplotype. Ongoing studies are concentrating on the identification of molecules that mediate or at least enhance the HSP-induced stimulation of T cells.

We thank Dr. P. K. Srivastava for helpful discussions and for sharing unpublished data and Dr. B. Bröker for critically reading the manuscript.

1

This work was supported by the Volkswagenstiftung.

3

Abbreviations used in this paper: HSP, heat shock protein; β-gal, β-galactosidase; VSV, vesicular stomatitis virus; PEC, peritoneal exudate cell(s).

1
Blachere, N. E., H. Udono, S. Janetzki, Z. Li, M. Heike, P. K. Srivastava.
1993
. Heat shock protein vaccines against cancer.
J. Immunother.
14
:
352
2
Udono, H., P. K. Srivastava.
1993
. Heat shock protein 70-associated peptides elicit specific cancer immunity.
J. Exp. Med.
178
:
1391
3
Suto, R., P. K. Srivastava.
1997
. A mechanism for the specific immunogenicity of heat shock protein-chaperoned peptides.
Science
269
:
1585
4
Srivastava, P. K., H. Udono, N. E. Blachere, Z. Li.
1994
. Heat shock proteins transfer peptides during antigen processing and CTL priming.
Immunogenetics
39
:
93
5
Udono, H., D. L. Levey, P. K. Srivastava.
1994
. Cellular requirements for tumor-specific immunity elicited by heat shock proteins: tumor rejection antigen gp96 primes CD8+ T cells in vivo.
Proc. Natl. Acad. Sci. USA
91
:
3077
6
Arnold, D., C. Wahl, S. Faath, H. G. Rammensee, H. Schild.
1997
. Influences of transporter associated with antigen processing (TAP) on the repertoire of peptides associated with the endoplasmatic reticulum-resident stress protein gp96.
J. Exp. Med.
186
:
461
7
Nieland, T. J. F., M. C. A. Tan, M. Monnee-van Muijen, F. Koning, A. M. Kruisbeek, G. M. van Bleek.
1996
. Isolation of an immunodominant viral peptide that is endogenously bound to the stress protein GP96/GRP94.
Proc. Natl. Acad. Sci. USA
93
:
6135
8
Matzinger, P., M. J. Bevan.
1977
. Induction of H-2-restricted cytotoxic T cells: in vivo induction has the appearance of being unrestricted.
J. Exp. Med.
33
:
92
9
Moore, M. W., F. R. Carbone, M. J. Bevan.
1988
. Introduction of soluble protein into class I pathway of antigen processing and presentation.
Cell
54
:
777
10
Niedermann, G., S. Butz, H. G. Ihlenfeldt, R. Grimm, M. Lucciari, H. Hoschützky, G. Jung, B. Maier, K. Eichmann.
1995
. Contribution of proteasome-mediated proteolysis to the hierarchy of epitopes presented by major histocompatibility complex class I molecules.
Immunity
2
:
289
11
Dick, L. R., C. Aldrich, S. C. Jameson, C. R. Moomaw, B. C. Pramanik, C. K. Doyle, G. N. DeMartino, M. J. Bevan, J. M. Forman, C. A. Slaughter.
1994
. Proteolytic processing of ovalbumin and β-galactosidase by the proteasome to yield antigenic peptides.
J. Immunol.
152
:
3884
12
Walden, P. R., H. N. Eisen.
1990
. Cognate peptides induce self-destruction of CD8+ cytolytic T lymphocytes.
Proc. Natl. Acad. Sci. USA
87
:
9015
13
Tamura, Y., P. Peng, K. Liu, M. Daou, P. K. Srivastava.
1997
. Immunotherapy of tumors with autologous tumor-derived heat shock protein preparations.
Science
278
:
117
14
Udono, H., P. K. Srivastava.
1994
. Comparison of tumor-specific immunogenicities of stress-induced proteins gp96, hsp90, and hsp70.
J. Immunol.
152
:
5398
15
Breloer, M., B. Fleischer, A. von Bonin.
1998
. Isolation of processed, H-2Kb binding ovalbumin-derived peptides associated with the stress proteins HSP70 and gp96.
Eur. J. Immunol.
28
:
1016
16
Höhfeld, J., S. Jentsch.
1997
. GrpE-like regulation of the hsc70 chaperone by the anti-apoptotic protein BAG-1.
EMBO J.
16
:
6209
17
Selkirk, M. E., D. A. Denham, F. Partano, R. M. Maizels.
1989
. Heat shock cognate 70 is a prominent immunogen in Brugian filariasis.
J. Immunol.
143
:
299
18
Kaufmann, S. H., U. Vath, J. E. Thole, J. D. Van Embden, F. Emmrich.
1987
. Enumeration of T cells reactive with mycobacterium tuberculosis organisms and specific for the recombinant mycobacterial 64-kDa protein.
Eur. J. Immunol.
17
:
351
19
Li, Z., P. K. Srivastava.
1994
. A critical contemplation on the role of heat shock proteins in transfer of antigenic peptides during antigen presentation.
Behring Inst. Mitt.
94
:
37
20
Hartl, F. U..
1996
. Molecular chaperones in cellular protein folding.
Nature
381
:
571
21
Fourie, A. M., J. F. Sambrooks, M. J. H. Gething.
1994
. Common and divergent peptide binding specificities of hsp70 molecular chaperones.
J. Biol. Chem.
269
:
30470
22
Arnold, D., S. Faath, H. G. Rammensee, H. Schild.
1995
. Cross-priming of minor histocompatibility antigen-specific cytotoxic T cells upon immunization with the heat shock protein gp96.
J. Exp. Med.
182
:
885
23
Matzinger, P..
1994
. Tolerance, danger and the extended family.
Annu. Rev. Immunol.
12
:
991
24
Blachere, N. E., Z. Li, R. Y. Chandawarkar, R. Suto, N. S. Jaikaria, S. Basu, H. Udono, P. K. Srivastava.
1997
. Heat shock protein-peptide complexes, reconstituted in vitro, elicit peptide-specific cytotoxic T lymphocyte response and tumor immunity.
J. Exp. Med.
186
:
1315
25
Suzue, K., X. Zhou, H. N. Eisen, R. A. Young.
1997
. Heat shock fusion proteins as vehicles for antigen delivery into the major histocompatibility complex class I presentation pathway.
Proc. Natl. Acad. Sci. USA
94
:
13146