- Email: [email protected]
Association of Peptides with Heat Shock Protein gp96 Occurs in Vivo and Not after Cell Lysis
Biochemical and Biophysical Research Communications 262, 813– 818 (1999) Article ID bbrc.1999.1306, available online at http://www.idealibrary.com on
Association of Peptides with Heat Shock Protein gp96 Occurs in Vivo and Not after Cell Lysis Antoine Me´noret, Ping Peng, and Pramod K. Srivastava 1 Center for Immunotherapy of Cancer and Infectious Diseases, University of Connecticut School of Medicine, Farmington, Connecticut 06030
Received August 6, 1999
Immunization of mice and rats with gp96 preparations isolated from syngeneic cancers has been shown to elicit protective immunity to a number of cancers. The specific immunogenicity of gp96 preparations derives from the antigenic peptides chaperoned by the gp96 molecule and not from gp96 molecules per se. Studies reported here demonstrate that the association of peptides with gp96 occurs in vivo and is not a procedural artifact which occurs in vitro after cell lysis. This demonstration has a bearing on the proposed functional role of HSP peptide association in antigen processing and presentation by MHC I molecules. © 1999 Academic Press Key Words: cancer; immunity; antigen presentation.
Previous studies from our laboratory have shown that heat shock protein (HSP) preparations derived from cancer cells elicit cancer-specific protective immunity (see 1). The immunogenicity of HSP preparations has been shown to derive not from HSPs per se, but from antigenic peptides associated with them. Removal of HSP-bound peptides has been shown to abrogate the immunogenicity of HSP preparation (2); conversely, reconstitution of HSP-peptide complexes in vitro has been shown to generate immunogenic moieties (3). The association of HSPs with peptides has been proposed to reflect a function in vivo. The HSPs have been said to constitute a relay line that chaperones antigenic peptides from the point of their generation in the cytosol to their loading on MHC I molecules in the endoplasmic reticulum (ER) (4). A basic premise of this relay-line hypothesis is that HSP-peptide association occurs in vivo and is not an artifact, even if a useful artifact, which occurs in vitro, after lysis of the cells for biochemical purification of the HSPs. Studies reported here address this question and show that association of 1 To whom correspondence should be addressed. Fax: 860-6794365. E-mail: [email protected]. Abbreviation used: HSP, heat shock protein.
peptides with the ER-resident HSP gp96 does not occur in vitro. MATERIALS AND METHODS Mice and antibody. BALB/cJ mice were obtained from the Jackson Laboratory (Bar Harbor, ME) and were maintained in the mouse facilities at Fordham University. Rat monoclonal antibody SPA-850 (clone 9G10, specific of gp96) was purchased from Neomarker (Fremont, CA). Metabolic cells labeling. MethA ascites cells were cultured in RPMI without methionine, and with 10% dialysed FCS for 12–15 hours, followed by a change of medium supplemented with 40 mCi/ml trans 35S methionine (ICN Biomedicals, Costa Mesa, CA) for 16 hours. Dissociation of peptides from 100,000g supernatant. MethA cells metabolically radiollabelled for 24 hours with 35S methionine were lysed in hypotonic buffer (30 mM sodium bicarbonate pH 7). To avoid degradation of proteins into peptides, cells were lysed in the presence of 14 protease inhibitors: phenylmethylsulfonyl fluoride (PMSF) 2 mM, ethylenediaminetetreacedic acid (EDTA) 1 mM, ethylene glycolbis(b-aminoethyl ether)N,N,N9,N9-tetraacetic acid (EGTA) 1 mM, (all obtained from Sigma, St. Louis, MO), and Antipain 20 mg/ml, Bestatin 5 mg/ml, Chemostatin 20 mg/ml, E64 20 mg/ml, Leupeptine 1 mg/ml, Pepstatine 1 mg/ml, Pefabloc 40 mg/ml, and Apoprotein 10 mg/ml (all obtained from Boehringer Mannheim, Indianapolis, IN). The cell lysate was centrifuged at 100,000g for 90 minutes. Peptides associated with proteins contained in the 100,000 supernatant were extracted for 1 hour in ATP containing buffer (6 mM ATP, 10 mM MgCl 2, 14 mM 2Me, pH 7) at room temperature as described (2). Peptides were separated by centrifugation through Centriprep-3. Peptides still bound to protein after ATP treatment were acid eluted by sonication in 0.1% trifluoroacetic acid (TFA). Bulk peptides eluted by ATP and TFA were fractionated by reverse phase column (BioCad20 microanalytical HPLC Poros RH2 column, Perseptive Biosystem, Cambridge, MA), equilibrated with 0.1% TFA in H2O and eluted by acetonitrile. During a linear gradient of 0% to 30% of acetonitrile 70 fractions (1 ml/fraction) were collected and counted for b radioactivity. SDS-PAGE and protein blotting. Proteins were resolved on SDSPAGE, subjected to electrophoresis, blotted to nitrocellulose, and probed with the anti-gp96 antibody, as described (5). Purification of gp96. Gp96 was purified as described (6) but adapted to an automated serial chromatographic technique. Briefly, cells (5 3 10 8) were washed three times in PBS, lysed by dounce homogenization in hypotonic buffer (30 mM sodium bicarbonate pH 7, 0.5 mM phenylmethylsulfonyl fluoride) and centrifuged at
813
0006-291X/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
Vol. 262, No. 3, 1999
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
100,000g for 90 minutes. The 100,000g supernatant was brought to 50% ammonium sulfate saturation. The supernatant from this step was brought to 70% ammonium sulfate saturation and centrifuged at 4500g for 30 min. The pellet was suspended in ConA-PBS (5 mM phosphate buffer, 2 mM MgCl 2, 0.3 M NaCl) and applied on a BioCad20 HPLC installed with ConA-sepharose (Pharmacia, Piscataway, NJ) in tandem with an anion exchange column Poros HQ/M (Perseptive Biosystems, Cambridge, MA). ConA-bound material was eluted with 5 mM phosphate buffer containing 2 mM MgCl 2, 0.3 mM NaCl and 6% a-methylmannopyranoside and transferred automatically on Poros HQ/M column. Fractions were collected and checked for purity of gp96 by SDS-PAGE. Tumor rejection assay. Mice (6 – 8 weeks old, female) were immunized with gp96 or PBS twice at weekly interval and challenged by intradermal injection of live tumor cells 1 week after the last immunization. Kinetics of tumor growth was monitored in individual mice.
RESULTS A two-part strategy was employed in order to determine if the gp96-bound peptides associated with gp96 in vivo or if the association occurred post-lysis, i.e. in vitro. In the first part, metabolically labeled cells were lysed and the lysates were analysed for presence of free peptides. It was observed that while the amount of free peptides in aqueous lysates was quite limited, a significantly larger variety and quantity of peptides could be shown to be bound to proteins and could be eluted from them. In the second part, a wide variety of peptides of diverse sequences and lengths were added to whole cells before the cells were lysed. Cell lysates were prepared and gp96 preparations obtained from such lysates. As the added peptides were labeled, they could be monitored; it was shown that no exogenously added peptides associated with gp96. Aqueous cell lysates contain relatively few free peptides. We reasoned that if gp96 molecules associated with peptides after cell lysis, cells would have to contain a large repertoire of free peptides which would then be available for binding to free gp96 molecules. As 100,000g supernatants of aqueous cell lysates are the starting material for purification of gp96, the 100,000g supernatant of metabolically radiolabeled MethA cells were analysed for the presence of free peptides. As degradation of proteins by endogenous proteases can generate a new set of peptides leading to an overestimation of the quantity of free peptides in the lysates, a number of protease inhibitors of different specificities (14 inhibitors, as described under Materials and Methods) were included in the lysis buffer. The 100,000g superanate was filtered through a micro-filtration device as described under Materials and Methods and the low molecular weight (,3 kDa) material isolated and separated on a C18 reverse phase column (Fig. 1). Except for a single peak eluting at fraction 20, no free peptides were detected in significant quantity. In parallel, an equivalent quantity of 100,000g supernatant was treated with 6 mM ATP, as described under Materials and Methods and molecules of ,3 kDa size
eluted and resolved on a C18 column. It was observed (Fig. 1A), that in contrast to the eluate from un-treated 100,000g supernatant, eluates from the ATP-treated 100,000g supernatant contain a wide range of significant quantities of peptides, which elute across the entire spectrum of the acetonitrile gradient. This observation indicates that although the peptides are present in the cell lysate, they are not in freely diffusing, soluble state, and at least a proportion of them are associated in an ATP-sensitive linkage with other proteins. A number of HSPs are known to be associated with peptides in an ATP-sensitive manner and our data are consistent with the idea that the peptides are associated with hsp70 or similar chaperones. To further dissociate peptides still complexed with the ATP-treated proteins, the ATP-treated, high molecular weight material was treated for 10 minutes to 0.1% TFA (final concentration) at room temperature, and low molecular weight material isolated and resolved as before, on the C18 column. This analysis revealed a new repertoire of peptides. Thus, fractions 5 to 10 of the TFAtreated sample contain peptides which are not detected in the untreated sample and fractions 29 –32 and fractions 36 –39 contain peptides which are not detected in the untreated sample nor in the ATP treated sample. Based on the total quantity of radiolabeled peptides eluted by ATP and 0.1% TFA, more than 70% of peptides are found associated with molecules larger than 3 kDa. In order to rule out the possibility that the peptides eluted by ATP and TFA treatments were generated by proteolytic activities or other forms of degradation resulting from such treatments (rather than elution of pre-bound peptides), the proteins retained on centricon-3 at the end of the experiment were analysed using gel electrophoresis and immunobloting analysis. Autoradiography revealed identical protein profiles for untreated and ATP/TFA-treated samples (Fig. 1B). The physical integrity of ATP/TFA-treated proteins was further demonstrated for a soluble chaperone, hsp70, which was detected as a single band before and after treatment, indicating that no degradation products were generated by such treatments (Fig. 1B). Gp96 does not associate with peptides during purification. If the peptide-association of gp96 occurs after cell lysis and in vitro, it should be possible to add labeled peptides to cell lysates and to monitor their association with gp96. In order for this experiment to be adequately stringent, it should be done not with a given peptide, but with the widest possible array of peptides and the peptides should be available to gp96 at the first in vitro moment. In order to obtain a collection of peptides with such an array of peptides, lysates of metabolically labeled cells were treated for 24 hours with trypsin (1 mg/ml) and chymotrypsin (1 mg/ml) at 37°C. Insoluble material was removed by centrifugation and low-molecular-
814
Vol. 262, No. 3, 1999
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 1. (A) Chromatographic profiles of peptides eluted from untreated, ATP-treated, and TFA-treated soluble cellular proteins. Metabolically labeled MethA cells were lysed in an aqueous buffer and a 100,000g supernatant was obtained. Equal quantities of the supernatant were either untreated or treated with 6 mM ATP or 0.1% TFA and 6 mM ATP, as described under Materials and Methods. The extracts were centrifuged through Centriprep-3 and the eluates were applied to a C18 reverse phase column in 0.1% TFA and eluted by an acetonitrile gradient. Fractions were counted and plotted. Chromatographic profiles of peptides eluted from untreated (- - -), ATP-treated (—), and TFA-treated (– – –) extracts are shown. (B) The untreated and ATP- and TFA-treated 100,000g supernates were analysed by SDS-PAGE and autoradiography to determine if the treatments had resulted in degradation of proteins themselves. This was further tested specifically for hsp70, which was detected by immunoblotting before and after treatments.
weight peptides were separated by centrifugation through Centriprep-3. Analysis of peptides thus generated, by tricine-PAGE showed them to be a heterogeneous collection of sizes (data not shown). The labeled peptides were mixed with 5 3 10 8 intact
MethA cells, which were used to purify gp96. The peptides were therefore available to gp96 at the first moment of lysis and thereon. The radioactive peptides and gp96 were monitored at each step of purification (Fig. 2). It was observed that most of the
815
Vol. 262, No. 3, 1999
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 2. Purification of gp96 in presence of exogenous radiolabeled peptides does not result in association of gp96 with exogenous peptides. Radiolabeled peptides were added to whole MethA cells and were followed during each step of purification of gp96. MethA cells (5 3 10 8) were lysed in the presence of the radiolabeled peptides. The peptide were followed by counting while gp96 molecules were followed by SDS-PAGE, silver staining, and immunoblotting using a monoclonal antibody specific for gp96.
radioactivity segregated into gp96-negative fractions during the first steps of purification, and no measurable radiolabeled peptides could be detected in the Concanavalin A-bound fraction which is highly enriched for gp96. The Concanavalin A-bound fraction was applied to the final purification step, the anion exchange HPLC column and the fractions were analysed for protein content (by SDS-PAGE), for presence of gp96 (by immunoblotting) and for the presence of labeled peptides (by counting). It was observed that the fractions 11–14, where gp96 elutes, were completely free of radioactivity. Autoradiography of SDS-PAGE was performed to determine if exogenous radiolabeled peptides formed complexes with proteins during the purification. The gp96 bands were found to be devoid of radioactivity (data not shown), further confirming that the exogenously added peptides did not bind gp96 during purification. Immunization of naive BALB/c mice with the gp96 preparations thus obtained was observed to confer protection against a challenge with live MethA cells (Fig. 3). Indicate that although no exogenous peptides bound to gp96 during purification, the gp96 preparations retained their immunogenic activity due to the in vivo-bound pool of peptides.
DISCUSSION The studies reported here show that (i) most peptides present in cells are not free, but associated with proteins, (ii) the association is non-covalent and can be disrupted by ATP or TFA, and (iii) exogenously added peptides do not bind to gp96 in vivo fortuitously under the conditions used for purification of gp96. These observations have a number of implications. First, they indicate that the tumor-protective ability of HSPpeptide complexes derived from tumors does not result from an artifactual association of peptides with HSPs such as gp96, hsp70 and hsp90. Secondly, these observations are consonant with a key element of the relayline model of antigen processing, wherein it was proposed that the peptides, after their generation in the cytosol, do not simply diffuse to the TAP and other molecules on their way to be loaded on MHC I molecules in the lumen of the ER, but are actively chaperoned by HSPs (4). The element of the relay-line model supported by the present observations is that cells appear to contain relatively small quantities of freely diffusing peptides; a much larger proportion of peptides of a wider variety are associated with molecules from which they may be dissociated by treatment with
816
Vol. 262, No. 3, 1999
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 3. Immunogenicity of MethA-derived gp96. The gp96 preparation in Fig. 2 was used in tumor rejection assays. Groups of 5 mice each were immunized with PBS (A) or 10 mg gp96 derived from MethA (B) as described under Materials and Methods and challenged with 100,000 MethA cells. Each line represents the kinetics of tumor growth in one mouse.
ATP or TFA. The aqueous extraction methods used here eliminate the MHC I molecules as possible sources of the dissociable peptides and are consistent with hsp70 and gp96 molecules as being the peptidebinding chaperones. The idea that peptides destined for binding to MHC I molecules do not diffuse in solution is also supported by the general poor solubility in aqueous buffer of most of the synthetic 8 mer peptides known to bind MHC-I or longer synthetic peptides which include MHC-I epitope (7, and unpublished). Chaperoning of peptides by other proteins is expected to minimize the chances of fortuitous association of peptides with irrelevant proteins, permit substantially higher local concentrations of peptides than those obtainable by diffusion and thus maximize the chances of productive association with relevant proteins, and finally, reduce the possibility of proteolytic destruction of peptides. In addition to the extensive studies on tumors, chaperoning of peptides by the gp96 molecules has now been demonstrated in case of viral antigens (8), minor histocompatibility antigens (9) and model antigens (9). In case of viral antigens, Nieland et al. (1996) observed recently that gp96 preparation from vesicular stomatis virus (VSV)-infected cells contain VSV-specific Kbbinding peptides, regardless of whether or not the preparation is derived from cells of the b haplotype. These results are consistent with our earlier prediction that due to their monomorphism, gp96 molecules in any cell type would be found associated not only with the MHC I-binding peptides specific for cells of that haplotype but for all haplotypes (4). This prediction had been experimentally confirmed by showing that
gp96 preparations derived from VSV-infected cells of the b or d haplotypes can prime a VSV-specific b-restricted response in b haplotype mice (10). Similar results were obtained by Arnold et al. in case of a model cytosolic antigen and minor histocompatibility antigens (9). These results indicate that peptides dissociated from gp96 or other HSPs represent a source of precursors of antigenic peptides, destined to be presented by MHC-I molecules. Collectively, these data and previous studies (11) indicate that gp96 acts as a classical chaperone protein, which associates with peptides in vivo. In contrast to the hsp70-peptide interaction, the mechanism of gp96-peptide interaction is not clearly understood yet, but is considered to be conformationally dependent and energetically regulated, resulting in a long-lived interaction of gp96 with peptides. ACKNOWLEDGMENTS This work was supported by NIH Grants CA64394 and CA44786 and a research agreement with Antigenics, Inc. We gratefully acknowledge Nathalie Blache`re, Daniel Levey, and Rajiv Chandawarkar for their collective criticisms and helpful discussions. Antoine Me´noret was supported partially by the French Institut National de la Sante´ et de la Recherche Medical (INSERM).
REFERENCES
817
1. Srivastava, P. K. (1993) Advances in Cancer Research 62, 153– 177. 2. Udono, H., and Srivastava, P. K. (1993) J. Exp. Med. 178, 1391– 1396. 3. Blachere, N. E., Li, Z., Chandawarkar, R. Y., Basu, S., Jaikaria,
Vol. 262, No. 3, 1999
4. 5. 6. 7.
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
N. S., Udono, H., and Srivastava, P. K. J. (1997) J. Exp Med. 183, 1315–1322. Srivastava, P. K., Udono, H., Blachere, N. E., and Li, Z. (1994) Immunogenetics 39, 93–98. Me´noret, A., Meflah, K., and Le Pendu, J. (1994) Int. J. Cancer 56, 400 – 405. Srivastava, P. K. (1997) Immunological Methods Manual, p. 737, Academic Press, San Diego, CA. Dick, T. P., Ruppert, T., Groettrup, M., Kloetzel, P. M., Kuehn,
8.
9. 10. 11.
818
L., Koszinowski, U. H., Stevanoviv, S., Schild, H., and Rammensee, H-G. (1996) Cell 86, 253–262. Nieland, T. J. F., Agnes, A., Tan, M. C., Monnee-Van Muijen, M., Koning, F., Kruisbeek, A. M., and Van Bleek (1995) Proc. Natl. Acad. Sci. USA 93, 6135– 6139. Arnold, D., Faath, S., Rammensee, H-G. R., and Schild, H. J. (1995) J. Exp. Med. 182, 885– 889. Suto, R., and Srivastava, P. K. (1995) Science 269, 1585–1588. Li, Z., and Srivastava, P. K. (1995) EMBO J. 12, 3143–3151.
Association of Peptides with Heat Shock Protein gp96 Occurs in Vivo and Not after Cell Lysis Antoine Me´noret, Ping Peng, and Pramod K. Srivastava 1 Center for Immunotherapy of Cancer and Infectious Diseases, University of Connecticut School of Medicine, Farmington, Connecticut 06030
Received August 6, 1999
Immunization of mice and rats with gp96 preparations isolated from syngeneic cancers has been shown to elicit protective immunity to a number of cancers. The specific immunogenicity of gp96 preparations derives from the antigenic peptides chaperoned by the gp96 molecule and not from gp96 molecules per se. Studies reported here demonstrate that the association of peptides with gp96 occurs in vivo and is not a procedural artifact which occurs in vitro after cell lysis. This demonstration has a bearing on the proposed functional role of HSP peptide association in antigen processing and presentation by MHC I molecules. © 1999 Academic Press Key Words: cancer; immunity; antigen presentation.
Previous studies from our laboratory have shown that heat shock protein (HSP) preparations derived from cancer cells elicit cancer-specific protective immunity (see 1). The immunogenicity of HSP preparations has been shown to derive not from HSPs per se, but from antigenic peptides associated with them. Removal of HSP-bound peptides has been shown to abrogate the immunogenicity of HSP preparation (2); conversely, reconstitution of HSP-peptide complexes in vitro has been shown to generate immunogenic moieties (3). The association of HSPs with peptides has been proposed to reflect a function in vivo. The HSPs have been said to constitute a relay line that chaperones antigenic peptides from the point of their generation in the cytosol to their loading on MHC I molecules in the endoplasmic reticulum (ER) (4). A basic premise of this relay-line hypothesis is that HSP-peptide association occurs in vivo and is not an artifact, even if a useful artifact, which occurs in vitro, after lysis of the cells for biochemical purification of the HSPs. Studies reported here address this question and show that association of 1 To whom correspondence should be addressed. Fax: 860-6794365. E-mail: [email protected]. Abbreviation used: HSP, heat shock protein.
peptides with the ER-resident HSP gp96 does not occur in vitro. MATERIALS AND METHODS Mice and antibody. BALB/cJ mice were obtained from the Jackson Laboratory (Bar Harbor, ME) and were maintained in the mouse facilities at Fordham University. Rat monoclonal antibody SPA-850 (clone 9G10, specific of gp96) was purchased from Neomarker (Fremont, CA). Metabolic cells labeling. MethA ascites cells were cultured in RPMI without methionine, and with 10% dialysed FCS for 12–15 hours, followed by a change of medium supplemented with 40 mCi/ml trans 35S methionine (ICN Biomedicals, Costa Mesa, CA) for 16 hours. Dissociation of peptides from 100,000g supernatant. MethA cells metabolically radiollabelled for 24 hours with 35S methionine were lysed in hypotonic buffer (30 mM sodium bicarbonate pH 7). To avoid degradation of proteins into peptides, cells were lysed in the presence of 14 protease inhibitors: phenylmethylsulfonyl fluoride (PMSF) 2 mM, ethylenediaminetetreacedic acid (EDTA) 1 mM, ethylene glycolbis(b-aminoethyl ether)N,N,N9,N9-tetraacetic acid (EGTA) 1 mM, (all obtained from Sigma, St. Louis, MO), and Antipain 20 mg/ml, Bestatin 5 mg/ml, Chemostatin 20 mg/ml, E64 20 mg/ml, Leupeptine 1 mg/ml, Pepstatine 1 mg/ml, Pefabloc 40 mg/ml, and Apoprotein 10 mg/ml (all obtained from Boehringer Mannheim, Indianapolis, IN). The cell lysate was centrifuged at 100,000g for 90 minutes. Peptides associated with proteins contained in the 100,000 supernatant were extracted for 1 hour in ATP containing buffer (6 mM ATP, 10 mM MgCl 2, 14 mM 2Me, pH 7) at room temperature as described (2). Peptides were separated by centrifugation through Centriprep-3. Peptides still bound to protein after ATP treatment were acid eluted by sonication in 0.1% trifluoroacetic acid (TFA). Bulk peptides eluted by ATP and TFA were fractionated by reverse phase column (BioCad20 microanalytical HPLC Poros RH2 column, Perseptive Biosystem, Cambridge, MA), equilibrated with 0.1% TFA in H2O and eluted by acetonitrile. During a linear gradient of 0% to 30% of acetonitrile 70 fractions (1 ml/fraction) were collected and counted for b radioactivity. SDS-PAGE and protein blotting. Proteins were resolved on SDSPAGE, subjected to electrophoresis, blotted to nitrocellulose, and probed with the anti-gp96 antibody, as described (5). Purification of gp96. Gp96 was purified as described (6) but adapted to an automated serial chromatographic technique. Briefly, cells (5 3 10 8) were washed three times in PBS, lysed by dounce homogenization in hypotonic buffer (30 mM sodium bicarbonate pH 7, 0.5 mM phenylmethylsulfonyl fluoride) and centrifuged at
813
0006-291X/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
Vol. 262, No. 3, 1999
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
100,000g for 90 minutes. The 100,000g supernatant was brought to 50% ammonium sulfate saturation. The supernatant from this step was brought to 70% ammonium sulfate saturation and centrifuged at 4500g for 30 min. The pellet was suspended in ConA-PBS (5 mM phosphate buffer, 2 mM MgCl 2, 0.3 M NaCl) and applied on a BioCad20 HPLC installed with ConA-sepharose (Pharmacia, Piscataway, NJ) in tandem with an anion exchange column Poros HQ/M (Perseptive Biosystems, Cambridge, MA). ConA-bound material was eluted with 5 mM phosphate buffer containing 2 mM MgCl 2, 0.3 mM NaCl and 6% a-methylmannopyranoside and transferred automatically on Poros HQ/M column. Fractions were collected and checked for purity of gp96 by SDS-PAGE. Tumor rejection assay. Mice (6 – 8 weeks old, female) were immunized with gp96 or PBS twice at weekly interval and challenged by intradermal injection of live tumor cells 1 week after the last immunization. Kinetics of tumor growth was monitored in individual mice.
RESULTS A two-part strategy was employed in order to determine if the gp96-bound peptides associated with gp96 in vivo or if the association occurred post-lysis, i.e. in vitro. In the first part, metabolically labeled cells were lysed and the lysates were analysed for presence of free peptides. It was observed that while the amount of free peptides in aqueous lysates was quite limited, a significantly larger variety and quantity of peptides could be shown to be bound to proteins and could be eluted from them. In the second part, a wide variety of peptides of diverse sequences and lengths were added to whole cells before the cells were lysed. Cell lysates were prepared and gp96 preparations obtained from such lysates. As the added peptides were labeled, they could be monitored; it was shown that no exogenously added peptides associated with gp96. Aqueous cell lysates contain relatively few free peptides. We reasoned that if gp96 molecules associated with peptides after cell lysis, cells would have to contain a large repertoire of free peptides which would then be available for binding to free gp96 molecules. As 100,000g supernatants of aqueous cell lysates are the starting material for purification of gp96, the 100,000g supernatant of metabolically radiolabeled MethA cells were analysed for the presence of free peptides. As degradation of proteins by endogenous proteases can generate a new set of peptides leading to an overestimation of the quantity of free peptides in the lysates, a number of protease inhibitors of different specificities (14 inhibitors, as described under Materials and Methods) were included in the lysis buffer. The 100,000g superanate was filtered through a micro-filtration device as described under Materials and Methods and the low molecular weight (,3 kDa) material isolated and separated on a C18 reverse phase column (Fig. 1). Except for a single peak eluting at fraction 20, no free peptides were detected in significant quantity. In parallel, an equivalent quantity of 100,000g supernatant was treated with 6 mM ATP, as described under Materials and Methods and molecules of ,3 kDa size
eluted and resolved on a C18 column. It was observed (Fig. 1A), that in contrast to the eluate from un-treated 100,000g supernatant, eluates from the ATP-treated 100,000g supernatant contain a wide range of significant quantities of peptides, which elute across the entire spectrum of the acetonitrile gradient. This observation indicates that although the peptides are present in the cell lysate, they are not in freely diffusing, soluble state, and at least a proportion of them are associated in an ATP-sensitive linkage with other proteins. A number of HSPs are known to be associated with peptides in an ATP-sensitive manner and our data are consistent with the idea that the peptides are associated with hsp70 or similar chaperones. To further dissociate peptides still complexed with the ATP-treated proteins, the ATP-treated, high molecular weight material was treated for 10 minutes to 0.1% TFA (final concentration) at room temperature, and low molecular weight material isolated and resolved as before, on the C18 column. This analysis revealed a new repertoire of peptides. Thus, fractions 5 to 10 of the TFAtreated sample contain peptides which are not detected in the untreated sample and fractions 29 –32 and fractions 36 –39 contain peptides which are not detected in the untreated sample nor in the ATP treated sample. Based on the total quantity of radiolabeled peptides eluted by ATP and 0.1% TFA, more than 70% of peptides are found associated with molecules larger than 3 kDa. In order to rule out the possibility that the peptides eluted by ATP and TFA treatments were generated by proteolytic activities or other forms of degradation resulting from such treatments (rather than elution of pre-bound peptides), the proteins retained on centricon-3 at the end of the experiment were analysed using gel electrophoresis and immunobloting analysis. Autoradiography revealed identical protein profiles for untreated and ATP/TFA-treated samples (Fig. 1B). The physical integrity of ATP/TFA-treated proteins was further demonstrated for a soluble chaperone, hsp70, which was detected as a single band before and after treatment, indicating that no degradation products were generated by such treatments (Fig. 1B). Gp96 does not associate with peptides during purification. If the peptide-association of gp96 occurs after cell lysis and in vitro, it should be possible to add labeled peptides to cell lysates and to monitor their association with gp96. In order for this experiment to be adequately stringent, it should be done not with a given peptide, but with the widest possible array of peptides and the peptides should be available to gp96 at the first in vitro moment. In order to obtain a collection of peptides with such an array of peptides, lysates of metabolically labeled cells were treated for 24 hours with trypsin (1 mg/ml) and chymotrypsin (1 mg/ml) at 37°C. Insoluble material was removed by centrifugation and low-molecular-
814
Vol. 262, No. 3, 1999
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 1. (A) Chromatographic profiles of peptides eluted from untreated, ATP-treated, and TFA-treated soluble cellular proteins. Metabolically labeled MethA cells were lysed in an aqueous buffer and a 100,000g supernatant was obtained. Equal quantities of the supernatant were either untreated or treated with 6 mM ATP or 0.1% TFA and 6 mM ATP, as described under Materials and Methods. The extracts were centrifuged through Centriprep-3 and the eluates were applied to a C18 reverse phase column in 0.1% TFA and eluted by an acetonitrile gradient. Fractions were counted and plotted. Chromatographic profiles of peptides eluted from untreated (- - -), ATP-treated (—), and TFA-treated (– – –) extracts are shown. (B) The untreated and ATP- and TFA-treated 100,000g supernates were analysed by SDS-PAGE and autoradiography to determine if the treatments had resulted in degradation of proteins themselves. This was further tested specifically for hsp70, which was detected by immunoblotting before and after treatments.
weight peptides were separated by centrifugation through Centriprep-3. Analysis of peptides thus generated, by tricine-PAGE showed them to be a heterogeneous collection of sizes (data not shown). The labeled peptides were mixed with 5 3 10 8 intact
MethA cells, which were used to purify gp96. The peptides were therefore available to gp96 at the first moment of lysis and thereon. The radioactive peptides and gp96 were monitored at each step of purification (Fig. 2). It was observed that most of the
815
Vol. 262, No. 3, 1999
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 2. Purification of gp96 in presence of exogenous radiolabeled peptides does not result in association of gp96 with exogenous peptides. Radiolabeled peptides were added to whole MethA cells and were followed during each step of purification of gp96. MethA cells (5 3 10 8) were lysed in the presence of the radiolabeled peptides. The peptide were followed by counting while gp96 molecules were followed by SDS-PAGE, silver staining, and immunoblotting using a monoclonal antibody specific for gp96.
radioactivity segregated into gp96-negative fractions during the first steps of purification, and no measurable radiolabeled peptides could be detected in the Concanavalin A-bound fraction which is highly enriched for gp96. The Concanavalin A-bound fraction was applied to the final purification step, the anion exchange HPLC column and the fractions were analysed for protein content (by SDS-PAGE), for presence of gp96 (by immunoblotting) and for the presence of labeled peptides (by counting). It was observed that the fractions 11–14, where gp96 elutes, were completely free of radioactivity. Autoradiography of SDS-PAGE was performed to determine if exogenous radiolabeled peptides formed complexes with proteins during the purification. The gp96 bands were found to be devoid of radioactivity (data not shown), further confirming that the exogenously added peptides did not bind gp96 during purification. Immunization of naive BALB/c mice with the gp96 preparations thus obtained was observed to confer protection against a challenge with live MethA cells (Fig. 3). Indicate that although no exogenous peptides bound to gp96 during purification, the gp96 preparations retained their immunogenic activity due to the in vivo-bound pool of peptides.
DISCUSSION The studies reported here show that (i) most peptides present in cells are not free, but associated with proteins, (ii) the association is non-covalent and can be disrupted by ATP or TFA, and (iii) exogenously added peptides do not bind to gp96 in vivo fortuitously under the conditions used for purification of gp96. These observations have a number of implications. First, they indicate that the tumor-protective ability of HSPpeptide complexes derived from tumors does not result from an artifactual association of peptides with HSPs such as gp96, hsp70 and hsp90. Secondly, these observations are consonant with a key element of the relayline model of antigen processing, wherein it was proposed that the peptides, after their generation in the cytosol, do not simply diffuse to the TAP and other molecules on their way to be loaded on MHC I molecules in the lumen of the ER, but are actively chaperoned by HSPs (4). The element of the relay-line model supported by the present observations is that cells appear to contain relatively small quantities of freely diffusing peptides; a much larger proportion of peptides of a wider variety are associated with molecules from which they may be dissociated by treatment with
816
Vol. 262, No. 3, 1999
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 3. Immunogenicity of MethA-derived gp96. The gp96 preparation in Fig. 2 was used in tumor rejection assays. Groups of 5 mice each were immunized with PBS (A) or 10 mg gp96 derived from MethA (B) as described under Materials and Methods and challenged with 100,000 MethA cells. Each line represents the kinetics of tumor growth in one mouse.
ATP or TFA. The aqueous extraction methods used here eliminate the MHC I molecules as possible sources of the dissociable peptides and are consistent with hsp70 and gp96 molecules as being the peptidebinding chaperones. The idea that peptides destined for binding to MHC I molecules do not diffuse in solution is also supported by the general poor solubility in aqueous buffer of most of the synthetic 8 mer peptides known to bind MHC-I or longer synthetic peptides which include MHC-I epitope (7, and unpublished). Chaperoning of peptides by other proteins is expected to minimize the chances of fortuitous association of peptides with irrelevant proteins, permit substantially higher local concentrations of peptides than those obtainable by diffusion and thus maximize the chances of productive association with relevant proteins, and finally, reduce the possibility of proteolytic destruction of peptides. In addition to the extensive studies on tumors, chaperoning of peptides by the gp96 molecules has now been demonstrated in case of viral antigens (8), minor histocompatibility antigens (9) and model antigens (9). In case of viral antigens, Nieland et al. (1996) observed recently that gp96 preparation from vesicular stomatis virus (VSV)-infected cells contain VSV-specific Kbbinding peptides, regardless of whether or not the preparation is derived from cells of the b haplotype. These results are consistent with our earlier prediction that due to their monomorphism, gp96 molecules in any cell type would be found associated not only with the MHC I-binding peptides specific for cells of that haplotype but for all haplotypes (4). This prediction had been experimentally confirmed by showing that
gp96 preparations derived from VSV-infected cells of the b or d haplotypes can prime a VSV-specific b-restricted response in b haplotype mice (10). Similar results were obtained by Arnold et al. in case of a model cytosolic antigen and minor histocompatibility antigens (9). These results indicate that peptides dissociated from gp96 or other HSPs represent a source of precursors of antigenic peptides, destined to be presented by MHC-I molecules. Collectively, these data and previous studies (11) indicate that gp96 acts as a classical chaperone protein, which associates with peptides in vivo. In contrast to the hsp70-peptide interaction, the mechanism of gp96-peptide interaction is not clearly understood yet, but is considered to be conformationally dependent and energetically regulated, resulting in a long-lived interaction of gp96 with peptides. ACKNOWLEDGMENTS This work was supported by NIH Grants CA64394 and CA44786 and a research agreement with Antigenics, Inc. We gratefully acknowledge Nathalie Blache`re, Daniel Levey, and Rajiv Chandawarkar for their collective criticisms and helpful discussions. Antoine Me´noret was supported partially by the French Institut National de la Sante´ et de la Recherche Medical (INSERM).
REFERENCES
817
1. Srivastava, P. K. (1993) Advances in Cancer Research 62, 153– 177. 2. Udono, H., and Srivastava, P. K. (1993) J. Exp. Med. 178, 1391– 1396. 3. Blachere, N. E., Li, Z., Chandawarkar, R. Y., Basu, S., Jaikaria,
Vol. 262, No. 3, 1999
4. 5. 6. 7.
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
N. S., Udono, H., and Srivastava, P. K. J. (1997) J. Exp Med. 183, 1315–1322. Srivastava, P. K., Udono, H., Blachere, N. E., and Li, Z. (1994) Immunogenetics 39, 93–98. Me´noret, A., Meflah, K., and Le Pendu, J. (1994) Int. J. Cancer 56, 400 – 405. Srivastava, P. K. (1997) Immunological Methods Manual, p. 737, Academic Press, San Diego, CA. Dick, T. P., Ruppert, T., Groettrup, M., Kloetzel, P. M., Kuehn,
8.
9. 10. 11.
818
L., Koszinowski, U. H., Stevanoviv, S., Schild, H., and Rammensee, H-G. (1996) Cell 86, 253–262. Nieland, T. J. F., Agnes, A., Tan, M. C., Monnee-Van Muijen, M., Koning, F., Kruisbeek, A. M., and Van Bleek (1995) Proc. Natl. Acad. Sci. USA 93, 6135– 6139. Arnold, D., Faath, S., Rammensee, H-G. R., and Schild, H. J. (1995) J. Exp. Med. 182, 885– 889. Suto, R., and Srivastava, P. K. (1995) Science 269, 1585–1588. Li, Z., and Srivastava, P. K. (1995) EMBO J. 12, 3143–3151.