The ability for the professional APC to cross-present Ag to MHC class I from parenchymal cells is essential for priming as well as tolerance of CD8+ T cells against intracellular Ags. Since cross-presentations of non-cell-associated free Ags are inefficient, the roles of molecular chaperones or heat shock proteins (HSPs) in chaperoning Ags to APCs have been postulated. We herein genetically addressed this hypothesis using mice that were defective of heat shock factor 1 (Hsf1), a major transcription factor for HSPs. Hsf1−/− mice have a decreased expression of several HSPs including HSP90 and HSP70. Using multiple Ag systems, we demonstrated that cross-priming of Ag-specific CD8+ T cells was inefficient when Ag expression was restricted to Hsf1−/− non-APCs. Our study provides the first genetic evidence for the roles of Hsf1 in regulating cross-presentation of MHC class I-associated Ags.
Although the Ag cross-presentation (XP)3pathway has been shown to be essential for the priming of CD8+ T cells against viruses (1, 2, 3), tumors (4), histoincompatible Ags (5), and self-Ags (6), the underlying mechanism by which cell-associated Ag is cross-presented to MHC class I of APCs remains unresolved. In particular, it is unclear why cell-associated Ag is much more efficient than free Ag in gaining access to the XP pathway (7). One possible reason is the existence of abundant heat shock proteins (HSPs) in the cells that can chaperone antigenic peptides into the Ag XP pathway (8). HSP-peptide complexes, formed either naturally in the cells (9) or reconstituted in vitro (10), can interact with APCs, leading to the XP of HSP-chaperoned peptides to MHC class I of APCs. However, it remains unclear whether HSPs are contributing to Ag XP in vivo. We reasoned that if the “HSP-centric model” is physiologically relevant, the efficiency of Ag XP must be dependent on the integrity of the intracellular homeostasis of HSPs. We have addressed this prediction genetically using mice that were defective of heat shock factor 1 (Hsf1), a major transcription factor for HSPs. We found that Ags from Hsf1−/− cells were poorly cross-presented to the MHC class I pathway both in vivo and in vitro.
Materials and Methods
Hsf1−/− mice and TCR-transgenic mice for Kb-SIINFEKL (OT-I) were bred and maintained at the University of Connecticut Health Center (Farmington, CT) according to the established guidelines. All other mice were obtained from The Jackson Laboratory (Bar Harbor, ME).
Cells and reagents
ADK14NP, a replication incompetent recombinant adenovirus expressing a full-length nucleoprotein (NP) of PR8 influenza virus, was generously provided by J. Yewdell (National Institute of Allergy and Infectious Diseases, Bethesda, MD) (3). Abs against HSPs were obtained from Stressgen (Victoria, British Columbia, Canada) and Neomarker (Montreal, Ontario, Canada). Peptides were synthesized by GeneMed Synthesis (South San Francisco, CA). Mouse embryonic fibroblast (MEF) was derived using a standard method and immobilized with a retroviral vector expressing a large T Ag of SV40 (a kind gift from T. Blankenstein, Berlin, Germany) (11).
Protein extraction and Western blot
ELISPOT and in vivo XP
Hsf1+/− or Hsf1−/− mice were immunized intradermally with 2 × 108 PFU of ADK14NP or with NP147–155 peptides plus CFA. Mice were sacrificed 14 days later. The frequency of Ag-specific CD8+ T cells in the spleen were measured by ELISPOT as published elsewhere (14). The labeling of OT-I cells with CFSE, adoptive transfer and in vivo XP assay, was performed as described previously (15).
In vitro XP assay
Bone marrow dendritic cells (DCs) were derived from mice in GM-CSF (16). Large T-specific T cell lines were kindly provided by J. Kovalchin and P. Srivastava (University of Connecticut Health Center). The T cells were weekly restimulated in vitro with irradiated SVB6 and used for the in vitro XP assay 12 days after restimulation. The XP efficiency was indexed by IFN-γ release after coculturing of 1 × 104 MEFs, 2 × 104 DCs, and 2 × 104 T cells in a U-bottom 96-well plate for 24 h. Alternatively, B3Z cells (17) were incubated with DCs and OVA-loaded MEFs, followed by ELISPOT to measure IL-2 production as an index for the efficiency of XP of Kb-SIINFEKL.
Student’s t test was used for statistical analysis. Values of p < 0.05 were considered to represent statistically significant differences.
Effect of Hsf1 ablation on HSP expression and immune response
Previous studies have shown that Hsf1 is essential for the stress induction of inducible HSPs including HSP70 and HSP25 (18, 19, 20). To determine whether there were alterations of steady-state levels of HSPs in Hsf1−/− cells, we systemically compared the expressions of several major HSPs that have been implicated earlier in chaperoning peptides for XP. We found that both wild-type (WT) and Hsf1−/− cells expressed a similar level of HSC70 (the constitutive member of HSP70) and an endoplasmic reticulum (ER) HSP gp96 (Fig. 1,A). However, the basal level of HSP90 was significantly decreased in Hsf1−/− MEFs, bone marrow-derived DCs (BMDCs), liver, and skin compared with corresponding tissues from WT mice (Fig. 1). Inducible HSP70 was present in the WT but not Hsf1−/− MEFs. These results confirmed that Hsf1 is essential not only for the induction of inducible HSPs, but also for the constitutive expression of HSP90 and HSP70.
To determine whether Hsf1 ablation results in global impairment in the immune system, we performed a phenotypical analysis of the hemopoietic system. We found that the absolute number and distributions of both lymphoid (CD3, CD4, CD8, B220, Dx5) and myeloid compartments (CD11b, CD11c, Gr-1) in the spleen, thymus, and bone marrow were comparable between WT and Hsf1−/− mice (Fig. 2,A and data not shown). No differences in the expressions of MHC class I and II were observed between WT and Hsf1−/− cells (data not shown). Serum Ig level was normal in the absence of Hsf1 (Fig. 2,B). Moreover, WT and Hsf1−/− T cells proliferated to a similar extent in response to Con A stimulation, as assessed by CFSE dilution (Fig. 2 C). This data, along with our previous studies that DCs differentiated and matured normally, based on the expression of CD86, CD40, and MHC class II and their ability to stimulate allogeneic naive CD4+ T cells in the MLR in the absence of Hsf1 (13, 17), demonstrated that Hsf1 is not essential for the development and basal functions of the immune system.
Cross-priming of CD8+ T cells against Hsf1−/− cell-associated Ag
To study the role of Hsf1 in the cross-priming of CD8+ T cells in vivo, we immunized Hsf1−/− mice and their littermates with ADK14NP, a recombinant adenovirus expressing NP under the control of a keratinocyte-specific promoter. Previous studies have shown that Kd-restricted NP147–155-specific CD8+ T cells can be cross-primed efficiently in WT BALB/c mice, even when the NP expression is directed to keratinocytes only (3). Indeed, NP147–155-specific CD8+ T cells could be primed readily in Hsf1+/− mice, as measured by an IFN-γ ELISPOT assay 14 days after infection (Fig. 3,A, left panel). By contrast, skin infection of Hsf1−/− mice with ADK14NP led to a much more reduced frequency of NP147–155-specific CD8+ T cells. To rule out the possibility that such a defect was due to a low NP-specific CD8+ T cell precursor frequency in Hsf1−/− mice, we also immunized mice with NP147–155 peptide emulsified in CFA. The priming of T cells in this case was not dependent on Ag XP. We found that the direct priming of CD8+ T cells in WT and Hsf1−/− mice was equally efficient (Fig. 3 A, right panel). The fact that IFN-γ-producing cells were normal in response to peptide vaccination also argues against the possibility that Hsf1 ablation impairs IFN-γ production.
To determine whether the defect of Hsf1−/− mice in cross-priming CD8+ T cells resides at the level of Ag-expressing cells, we developed a cross-priming system using WT or Hsf1−/− MEFs. We immortalized MEFs by the stable transduction with SV40 large T Ag (MEF-LT). The expression of large T Ag (LT) by WT or Hsf1−/− MEFs was comparable by Western blot analysis (Fig. 3,B, inset). However, immunization of CB6F1 (H-2d/b) mice with irradiated Hsf1−/− MEFs (H-2d) resulted in inefficient cross-priming of LT-specific H-2b-restricted CD8+ T cells in vivo as measured by an IFN-γ ELISPOT assay against the H-2b+ SV40-transformed fibroblast SVB6 (Fig. 3,B). To study the cross priming of LT-specific T cells in vitro, we cocultured MEF-LT with WT C57BL/6 BMDCs and H-2b-restricted LT-specific T cells. Twenty-four hours later, the ability of MEFs (H-2d) to channel LT to the XP pathway of DCs (H-2b) to stimulate T cells was indexed by the release of IFN-γ. We found that LT Ag from WT MEFs was cross-presented efficiently to stimulate T cells (Fig. 3,C). By contrast, Hsf1−/− MEFs were inefficient in mediating the XP of LT-derived peptides. The response was Ag specific, since without LT, MEFs did not induce significant IFN-γ release in this assay. Interactions between MEF and DCs were essential, since coculturing MEFs with either T cell or DC alone did not induce IFN-γ release (Fig. 3 C).
To determine whether the defect in the XP can be extended to nonviral systems, we studied the XP of the chicken OVA-derived SIINFEKL peptide to CD8+ cells. We used a well-described OT-I mouse model that transgenically expressed TCR for Kb-SIINFEKL (6). CFSE-labeled CD8+ OT-I cells (purified by magnetic beads conjugated with anti-CD8 Ab; the purity was >95% in all experiments based on staining with Vα1-specific mAb) were adoptively transferred into naive C57BL/6 mice (H-2b), followed by immunization with WT or Hsf1−/− MEFs (H-2d) osmotically loaded with OVA. Three days later, CFSE intensity was analyzed by flow cytometry as a reflection of OT-I proliferation in vivo. The loading efficiency of OVA to WT and Hsf1−/− MEFs was identical (50 ng OVA/5 × 105 cells, data not shown). However, immunization with WT MEFs drove OT-I proliferations more efficiently than immunization with the OVA-loaded Hsf1−/− MEF. Similar differences were observed when OVA-loaded WT or Hsf1−/− DCs were used as the source of Ags (Fig. 3,D). The defect of Hsf1−/− cells in channeling OVA to the XP pathway could be overcome by increasing the amount of cells for immunization, indicating that the defect is quantitative rather than qualitative. To corroborate our in vivo data of the OVA system, we incubated DCs with OVA-loaded WT or Hsf1−/− MEFs in vitro along with B3Z hybridoma that is specific for Kb-SIINFEKL complexes. Once again, XP of OVA from Hsf1−/− cells was inefficient, as reflected by limited production of IL-2 (Fig. 3 E).
We have studied the molecular basis for the Ag XP from the angle of Ag-expressing cells. We postulate that there must exist accessory molecules in every cell type that assist the Ag XP, since cell-free Ag is clearly not cross-presented efficiently (7). We discovered that Ags from Hsf1−/− cells were unable to gain effective access to the XP pathway to stimulate T cell proliferation (Fig. 3,D) as well as to trigger IFN-γ and IL-2 release (Fig. 3). The difference between WT and Hsf1−/− cells withstood in vitro and in vivo, regardless of what Ags were tested (OVA, NP, and LT).
The precise mechanisms for the inefficient Ag XP from Hsf1−/− cells are under active investigation. Our data clearly indicate that the presence of Ag alone does not dictate the efficiency of Ag XP (Fig. 3), which is in agreement with the suggestion that other non-Ag factors must play a role in facilitating Ag XP (7). Our study demonstrated that these putative factors are regulated by Hsf1, the major transcription factor for HSPs. The finding that XP efficiency correlates with the level of cytosolic HSP90 and HSP70 is intriguing for two reasons. HSP90 is known to play important roles for the direct Ag presentation to MHC class I (21). In addition, HSP90 forms superchaperone complexes with HSP70 and Hsf1 in the cytosol to regulate protein folding (22, 23, 24). Further studies are necessary to address whether cytosolic HSPs are uniquely involved in Ag XP.
Our findings do not prove that the HSP-peptide complex is the preferred mode of Ag in XP to MHC class I. In contrast, our study also does not contradict, with the recent surge of “the protein-centric model of Ag XP,” that protein Ags alone, rather than peptides or the HSP-peptide complex, are the more relevant forms of XP Ag (25, 26, 27). Our data call for the reconciliation of the “HSP-centric” and “protein-centric” models. We suggest that the complex of HSPs with protein Ags (HSP-protein complex) rather than the HSP-peptide complex or free peptide/protein Ags is the key substrate for the Ag XP physiologically. Such an hypothesis is in line with the normal function of HSPs; i.e., it is the nascent polypeptide chains, not the peptides, that HSPs keenly care about and interact with (28, 29). The first and foremost roles of HSPs are to promote protein folding. The complex of HSPs with peptides is the by-product of this interaction during protein catabolism. Most likely, the efficient protein catabolism machinery leaves free peptides and HSP-peptide complexes in a minute quantity in the cell at all times. Our new model is consistent with several well-established phenomena including: 1) the XP of cell-associated Ag is more efficient than free Ag (7); 2) inhibition of protein catabolism enhances XP (26); and 3) free peptides, introduced into cells through minigene expression and other means, are not efficiently cross-presented (26, 30). Finally, our model would predict that HSPs in the cytosol are more important than HSPs in other subcellular localizations in facilitating Ag XP, which explains why ER chaperones are dispensable for Ag XP (30, 31) and why the XP of Hsf1−/− cell-associated Ag is reduced despite normal levels of ER chaperones such as gp96 (Fig. 1).
We wish to emphasize that the Hsf1-regulated genes were not restricted to HSPs. It is also possible that Hsf1-controlled HSPs are regulating the fusion of phagosomes with the ER, a process that has been implicated in the Ag XP (1, 32, 33, 34). Thus, more studies in Hsf1−/− mice should be fruitful for uncovering other molecules and mechanisms that are important in the Ag XP. In addition, the Hsf1−/− mouse should be a useful model for studying the roles of Ag XP in a variety of fundamental immunological processes, such as immune surveillance, autoimmunity/tolerance, and immunity against intracellular microorganisms.
We thank Bei Liu, Yi Yang, and Jie Dai (all from Z.L.’s laboratory) for technical help and stimulating discussions throughout the course of the study. We are grateful to members from the Center for Immunotherapy of Cancer and Infectious Diseases and the Immunology Graduate Program from the University of Connecticut School of Medicine for helpful discussions.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in part by National Institutes of Health Grants CA90337 and CA 100191 (to Z.L.).
Abbreviations used in this paper: XP, cross-presentation; DC, dendritic cells; ER, endoplasmic reticulum; Hsf1, heat shock factor 1; HSP, heat shock protein; NP, nucleoprotein; MEF, mouse embryonic fibroblast; WT, wild type; BMDC, bone marrow-derived DC; LT, large T Ag.