Development of heat shock protein as a vaccine vehicle:

The need for more effective immunological prophylaxis (vaccines) and therapies for cancer and infectious diseases caused by intracellular pathogens has spurred intense investigation of immunogens and immunization strategies that elicit effective CD8+ CTL. Our work focuses on recombinant heat shock protein 70 (hsp70) fusion proteins: simple, defined and highly purified vaccines. We have found that immunization with antigens genetically fused to hsp70 elicits strong and long-lived humoral and cellular immune responses in the absence of adjuvant (Suzue et al., 1996). Hsp70 fusion proteins elicit antigen-specific CTL responses and protect mice from tumor challenge (Suzue et al., 1997). We have now shown the ability to elicit CTL resides in a 200-amino acid domain of hsp70 and is independent of CD4 (Huang et al., 2000).

What are heat shock proteins? 

Heat shock proteins (hsps) are highly conserved, ubiquitous and abundant proteins, essential for cellular viability (reviewed in Ang et al., 1991). Prokaryotic and eukaryotic hsp60 and hsp70 vary by only 50% at the nucleotide level (Lindquist and Craig, 1988; Jindal et al., 1989; Zeilstra-Ryalls et al., 1991). Expression of most hsps is induced by heat and other stresses including hypoxia, nutrient deprivation, oxygen radicals, metaboic disruption, viral infection (Morimoto and Milarski, 1990; Young, 1990; Young and Elliot, 1989), phagocytosis (Polla, 1991) and transformation. Under conditions of stress hsps constitute as much as 15% of prokaryotic cellular proteins, while stress increases expression of eukaryotic hsps much more modestly. Most hsps function as chaperones: hsps participate in folding, assembly and dissasembly of protein complexes and may also assist in translocation of proteins from one compartment to another (Ang et al., 1991; Gething and Sambrook, 1992; Rothman, 1989). In fact, accumulation of unfolded or misfolded proteins is a form of stress that induces expression of hsps. Hsps 60, 70 and 90 are generally found in the cytosol and mitochondria. A more distantly related family of chaperones, including gp96 and calreticulin, are located in the endoplasmic reticulum.

Hsp70 

Heat shock protein 70 (hsp70) is a 70 kD molecular chaperone possessing coupled ATPase and peptide-binding functions. The amino-terminal 44 kD of hsp70, mixed alpha-helices and beta-pleated sheets, binds and hydrolyzes ATP, driving the folding action of hsp70 (Flaherty et al., 1990). The carboxy-terminal 27 kD domain is a loose beta barrel which binds 6-amino acid hydrophobic protein motifs, the object of folding activity (Zhu et al., 1996). There is a large family of hsp70 proteins; family members differ in subcellular localization and degree of induction by stress.

Hsps and immunity

Hsps also appear to have a variety of important and surprising roles in immunity. Two hsp70 genes are located within the human major histocompatibility complex class III region, between the complement and tumor necrosis factor genes (Sargent et al., 1989). An hsp90 gene is linked to a minor histocompatibility locus near murine H-2 (Romano et al., 1989). Hsp 60 and 70 are major targets of immune responses to a wide variety of pathogens including bacteria, fungi, helminth (worms) and protozoan parasites (Young and Elliot, 1989; Kaufmann, 1990; Young et al., 1990; Young, 1990).

Hsps are targets of anti-pathogen immune responses.

Pathogens	Disease	Hsp
Bacteria
Bordetella pertussis	pertussis	hsp70, 60
Borrelia burgdorferi	Lyme disease	hsp60
Brucella abortus	brucellosis	hsp60
Chlamydia trachomatis	blinding trachoma	hsp70, 60
Coxiella burnetti	Q fever	hsp60
Heliobacter pylori	gastritis	hsp60, 10
Legionella pneumophila	Legionnaire's disease	hsp60
Mycobacterium leprae	leprosy	hsp70, 60, smaller
Mycobacterium tuberculosis	tuberculosis	hsp70, 60, smaller
Treponema pallidum	syphillis	hsp60
Fungi
Aspergillus fumigatus	aspergillosis	hsp60
Candida albicans	candidasis	hsp90
Histoplasma capsulatum	histoplasmosis	hsp60, 70
Helminths
Brugia malayi	lymphatic filariasis	hsp70
Onchocera volvus	ocular filariasis	hsp70
Schistosoma mansoni	schistosomiasis	hsp90, 70, smaller
Schistosoma japonicum	schistosomiasis	hsp70
Protozoa
Leishmania braziliensis	leshmaniasis	hsp70
Leishmania donovani	visceral leshmaniasis	hsp90, 70
Plasmodium falciparum	malaria	hsp70
Trypanosoma cruzi	Chagas' disease	hsp70
(from Suzue and Young, 1996)

Up to 20% of CD4+ T cells responding to mycobacterial infection are specific for hsp60 (Emmrich et al., 1986; Kaufmann et al., 1987). In fact, immunization with a variety of pathogen hsps induces strong immune responses and provides protection against disease caused by these pathogens (reviewed by Suzue and Young, 1996). The strength of immune responses to pathogenic hsps may be a result of the abundance and induction of these proteins. In addition, hsps also appear to be fairly immunogenic; multiple B cell and T cell epitopes are found on mycobacterial hsp60 and 70, as well as other hsps (Mehra et al., 1986; Lamb et al., 1987; Thole et al., 1988; van Schooten et al., 1988; Richman et al., 1989; Mattei et al., 1989; Munk et al., 1990; Requena et al., 1993; Adams et al., 1994; Offtung et al., 1994). It has been proposed that strong memory responses are elicited and maintained early in life by self hsps or hsps present in normal gut and skin (microbial) flora. This hsp-specific immunological memory might form a first line of adaptive defense against microbial invaders and might account for the strength of hsp-specific immune responses to pathogens (Murray and Young, 1992).

Immunization with hsps from pathogens protects animals from disease.

Pathogen	Hsp	Animal Model
Bacteria
Heliobacter pylori	hsp60, 10	mouse
Legionella pneumophila	hsp60	guinea pig
Mycobacterium leprae	hsp60, 10	mouse
Mycobacterium tuberculosis	hsp70	mouse, guinea pig
Mycobacterium tuberculosis	hsp60	mouse
Yersinia enterocolitica	hsp60	mouse
Fungi
Candida albicans	hsp90	mouse
Histoplasma capsulatum	hsp60, hsp70	mouse
(from Suzue and Young, 1996)

 

Are hsps involved in autoimmune responses? 

The strength of immune responses to these pathogen hsps makes the apparent tolerance of healthy individuals to their own (self) hsps all the more remarkable. While the hsps of pathogens and mammals are very similar (~50% identical at the amino-acid level) most people don’t develop dangerous autoimmune responses to self hsps. In fact, healthy people do possess T cells which recognize these self hsps, but suffer no ill effects (Rees et al., 1988; Lamb et al., 1989; Munk et al., 1989). The presence of such self-reactive cells in healthy people suggests that these cells are highly regulated (von Boehmer, 1988; Schwartz, 1989). However, there is also some evidence from animal models of human autoimmune diseases, such as arthritis, that suggests that T cells which recognize mycobacterial hsp60 and may sometimes cross-react with self hsps and cause disease (Holoshitz et al., 1983 and 1984; Shinnick et al., 1988; van Eden et al., 1988; Young et al., 1988).

Heat shock proteins are being used as vaccine vehicles.

There is now substantial evidence that native heat shock proteins (hsps) isolated from tumors can be used as adjuvant-free anti-tumor vaccines in animal models; hsp70 and the distantly related chaperones gp96 and calreticulin share this vaccine activity (Udono et al., 1993; Udono et al., 1994; Suto et al., 1995; Blanchere et al., 1997; Tamura et al., 1997; Nair et al., 1999). In addition, chemical conjugation (Lussow et al., 1991; Barrios et al., 1992; Perraut et al., 1993) or genetic fusion (Suzue et al., 1996 and 1997; Rico et al., 1998) of antigens to mycobacterial hsp70s creates potent and customized immunogens that can elicit MHC class I-restricted, CD8+ cytotoxic T cell responses sufficient to mediate rejection of tumors expressing the fusion partner (Suzue et al., 1997). While most of this work has been done in mice, hsp conjugate vaccines have elicited immune responses in monkeys (Perraut et al., 1993) and hsp fusion vaccines are being tested in humans.

Hsp fusion protein vaccines do not require adjuvants.

We have found that a recombinant protein consisting of the HIV-1 p24 antigen fused to the amino-terminus of mycobacterial hsp70 elicits both humoral and cellular immune responses in mice (Suzue et al., 1996). Antibody responses were stronger and more persistent (up to 68 weeks) when p24 was fused to hsp70. Splenocytes from mice immunized with the fusion protein proliferated and produced IFN-γ, IL-2 and IL-5 in response to in vitro stimulation with p24. Physical linkage of p24 and hsp70 was required for increased immunity. Importantly, hsp70 fusion proteins induced these immune responses without adjuvants. Most vaccines require adjuvants to provoke effective and protective immune responses. However, most adjuvants used in research cause powerful and unpleasant side effects in humans; thus only alum, a very weak adjuvant, is used in human vaccines. There is currently a great deal of interest in developing vaccines, like hsp70 fusion proteins, that generate powerful immune responses without the use of adjuvants. Our early work suggested that hsp70 functions not as an adjuvant, but as an exceptionally powerful carrier, capable of eliciting both T cell and B cell responses (Suzue et al., 1996).

Hsp fusion proteins elicit CTL responses in the absence of adjuvant.

We then fused ovalbumin (OVA), a well characterized T cell antigen, to the amino-terminus of mycobacterial hsp70 in order to further characterize T cell response to hsp70 fusion proteins. Mice were immunized twice with 120 pmoles of OVA.TBhsp70 fusion protein in saline. Mice immunized with OVA.TBhsp70, but not irrelavant OVA.p24 fusions or OVA alone, produced OVA-specific cytotoxic T cells (CTL) which lysed cells expressing the immunodominant OVA MHC class I epitope, SIINFEKL. In other words, hsp70 fusion proteins cross-prime CTL well; hsp fusions appear to gain access the MHC class I processing and presentation pathway in a non-classical manner. This was unexpected as immunization with soluble proteins, especially in the absence of adjuvant, rarely elicit CTL responses. The CTL response to OVA.TBhsp70 was restricted to CD8+ T cells. Physical linkage of hsp70 and antigen was, again, required for induction of antigen specific CTL responses. Immunization with OVA.TBhsp70 fusion proteins, in saline, provided partial protection from tumor challenge with OVA-expressing melanoma.

How do hsp fusion proteins elicit CTL responses?

The means by which soluble hsp70 fusion proteins stimulate CD8 cytotoxic T cell (CTL) responses are unknown. Possible mechanisms include: 1) strong hsp-specific CD4+ helper cell responses that enhance what might otherwise be a minimal response to the soluble proteins (Barrios et al., 1992; Suzue et al., 1996; Horwitz et al., 1998; Könen-Waisman et al., 1999); and 2) the chaperone function of hsps delivers the fusion protein to intracellular compartments of antigen-presenting cells for processing into short peptides and loading onto MHC class I (Young et al., 1990; Schild et al., 1999). We have recently demonstrated that hsp70 fusion proteins can elicit CD8+ CTL in the absence of CD4+ T lymphocytes and that this function resides in a 200-amino acid segment of TBhsp70, indicating that chaperone activity is not required (Huang et al., 2000). The findings of this recent work are discussed below.

The ability of hsp70 fusion proteins to elicit CTL is CD4-independent.

While there is evidence that the hsp moiety of mycobacterial hsp fusion proteins acts as an effective carrier in the classic sense, enhancing B cell responses to chemically conjugated pneumococcal polysaccharides (Könen-Waisman et al., 1999) and malarial polypeptide (Barrios et al., 1992), carriers are not known to stimulate CTL production. We thought it more likely that hsp70 fusion proteins provide hsp70-specific cognate CD4+ T cell help to OVA-specific CD8+ CTL by activating shared professional antigen presenting cells (APCs) as demonstrated recently (Bennett et al., 1997 and 1998; Ridge et al., 1998; Schoenberger et al., 1998).

We expressed two recombinant proteins in E. coli: OVA and OVA fused to the amino-terminus of M. tuberculosis hsp70 (OVA.TBhsp70). We tested the cognate help hypothesis using CD4 knockout mice (CD4-/-). Wild-type C57BL/6, CD4-/-, and β2-microglobulin-/- mice were each immunized twice with 120 picomoles of OVA or OVA.TBhsp70 fusion protein in saline. As expected, immunization of wild-type mice with OVA.TBhsp70, but not OVA alone, generated CTL specific for the immunodominant CTL epitope of OVA (SIINFEKL). Unexpectedly, CTL were also induced when the CD4-/- mice were immunized with OVA.TBhsp70, demonstrating that hsp70 fusion proteins' ability to enhance immune responses does not require CD4 . β2-microglobulin-/- mice, which have very few CD8+ T cells, did not develop OVA-specific CTL after immunization with OVA.TBhsp70 or with OVA alone. The similarity of CD8+ CTL responses to OVA.TBhsp70 in CD4-/- and wild-type mice suggests that hsp70 fusion proteins are relatively potent CD8+ CTL immunogens.

A similar result, showing that CD4 is not required for the CD8+ CTL response elicited by another mycobacterial heat shock fusion protein (hsp65 fused to a polypeptide containing an epitope for 2C-specific CD8+ T cells), has also been obtained using CD4-/- mice (Cho et al. personal communication). In addition, the ability of a non-homologous hsp, gp96, to elicit tumor rejection requires CD4+ T cells at tumor challenge, but not during priming with tumor-derived gp96 (Udono et al., 1994).

Both mycobacterial and murine (self) hsp70 enhance immune responses.

It has been proposed that the immunostimulatory effects of mycobacterial hsp fusion proteins may be due to the bacterial origin of the hsp moiety (Schild et al., 1999). We therefore made OVA.hsp70 fusion proteins with the murine homologue of TBhsp70 (Hunt and Calderwood, 1990), here referred to as mhsp70. Immunization of wild-type C57BL/6 mice with OVA.mhsp70, but not OVA, elicited CTL responses equivalent to those generated by the mycobacterial hsp70 fusion protein. The response to OVA.mhsp70 was also independent of CD4. Since a CD4+ T cell response to self (murine) hsp70 is unlikely, the effectiveness of the murine hsp70 fusion protein is in accord with the more direct evidence for CD4-independence obtained using CD4-/- mice (see above).

A 200-amino acid domain of hsp70 is sufficient to elicit CTL.

To determine whether the chaperone activity of hsp70 is required to elicit CTL, we divided TBhsp70 into four linear segments and fused OVA to the amino-terminus of each segment, creating OVA.TBhsp70s I-IV. Each segment corresponds to a distinct structural domain of hsp70 as described by Flaherty et al. (1990) and Zhu et al. (1996). The amino-terminal ATP-binding domain was divided into its two structural lobes: I (aa 1-166) and II (aa 160-362). The carboxy-terminal peptide-binding domain was divided into a b -sandwich domain, III (aa 360-517), and an a -helical domain, IV (aa 510-625).

C57BL/6 mice were immunized with 120 pmoles of OVA, OVA.TBhsp70, and OVA fused to segments I, II, III or IV. Immunization with intact OVA.TBhsp70 and OVA fused to segment II lysed cells expressing ovalbumin. In contrast, immunization with OVA and OVA fused to segments I, III, and IV did not elicit CTL. Levels of cytolysis induced by immunization with OVA.TBhsp70 and OVA fused to segment II were indistinguishable. These results show that segment II, half of the ATP-binding domain of TBhsp70 (aa 161-362), is sufficient to induce anti-OVA CTL response in the absence of adjuvant. Since it is highly unlikely that segment II possesses chaperone activity we conclude that the ability of the fusion proteins to elicit CD8+ T cell does not depend on the hsp moieties' chaperone properties. Interestingly, the sequence of TBhsp70 segment II does not appear to be related to proteins outside of the hsp70 family.

So how do hsp70 fusion proteins enhance immunity?

How then do hsp fusion proteins act as CD8+ CTL immunogens? Our data would support a model in which hsp70 bypasses the need for CD4+ help by directly or indirectly activating or affecting the maturation state of antigen presenting cells (APCs) such as dendritic cells in a manner similar to some viruses (Ruedl et al., 1999). According to this model, hsp70 fusion proteins may activate a few CD8+ T cells to release immunostimulatory cytokines in draining lymph nodes. These cytokines may, in turn, provide the help required to up-regulate expression of costimulatory molecules on APCs in the lymph node, leading to further CD8+ T cell activation (Ruedl et al., 1999). Recent studies demonstrate that exposure of macrophages to bacterial and human hsp60 (Chen et al., 1999; Kol et al., 1990), murine hsp70 and gp96 (Suto et al., 1995; Breloer et al., 1999) increases expression of adhesion molecules and cytokines. We are currently examining expression of costimulatory molecules, adhesion molecules, and cytokines by APCs after exposure to fusion proteins made with full-length hsp70 or segment II.

Conclusions:

Whatever the underlying mechanism, the ability of hsp70 fusion proteins to elicit CTL responses in the absence of CD4+ cells suggests that hsp70 may be a useful vehicle for the development of prophylaxis and therapy of HIV-1 and its opportunistic infections. Infection by HIV and its simian cousin SIV can lead to a substantial reduction in CD4+ T cells, thereby crippling the host's immune response to HIV and other pathogens. This loss of CD4+ cells is thought to impair the development and maintenance of CD8+ CTL responses (Kalams et al., 1999). Recent studies conclude that strong HIV-specific CTL responses are required to keep HIV-1 infection in check and to destroy HIV-infected cells (Harrer et al., 1996; Yang et al., 1996; Yang et al., 1997; Matano et al., 1998; Wagner et al., 1998). It will thus be of interest to determine whether hsp70 fusion constructs can elicit anti-SIV CTL responses in SIV-infected macaques having low CD4+ T cell counts, and if similar effects are observed in HIV-infected humans.

Young Lab Home Page