Immune surveillance and protection from cancer

Cancer appears to have engaged the minds of immunologists almost since the beginning of immunology itself. Ehrlich, who opined on all things immunological, believed that the immune system could protect the host from cancer.

Burnet and Thomas refined that idea into the immune surveillance hypothesis of cancer. The hypothesis lived in limbo for several decades, failing to thrive and failing to die, until very recently when the work of Old, Schreiber, and their colleagues demonstrated that mice with a compromised immune status were more prone than immunocompetent mice to develop an array of cancers.

The immune surveillance hypothesis is often regarded as the intellectual underpinning of cancer immunology. Although the hypothesis itself has contributed little to our attempts to treat cancer through immunological means, it has profound implications for understanding the functions of the immune system.

Tumor immunity in the primary host

Entirely unrelated to these two lines of enquiry, the study of cancer immunity saw a revival at the hands of those who were transplanting chemically induced tumors into the many inbred mice that began to be available in the 1950s. These investigators ‘showed “highly successful” immunization against the “transplantable tumors” and expressed great hopes about cancer vaccination’.

In hindsight, these successful immunizations were simply a result of allogeneic differences between tumors and the host strain of mouse, a theme that played a seminal role in definition of the MHC, but had no relevance for cancer immunity.

Nonetheless, amidst the barrage of experiments where MHC-mismatched tumors were transplanted into mice, were the experiments of Ludwik Gross, and later those of Prehn and Main, and of George and Eva Klein, who showed that, even when MHC-matched tumors were used to immunize mice, protection against subsequent tumor growth could be achieved (Fig. 22.1). These studies led to two principles, which have informed much of cancer immunology since and are discussed below.

Fig. 22.1 Immunogenicity of chemically induced tumors

(1) Mice immunized with tumor cells are protected against subsequent tumor growth. (2) The primary animal that develops the tumor is also immune to subsequent challenges with the same tumor.

Cancers elicit protective immunity in the primary and syngeneic host

Mice and rats of a given haplotype can be immunized with irradiated cancer cells that arose in animals of the same haplotype. When they are challenged with live cancer cells, they are able to resist the tumor challenge. The following further observations and deductions have been derived from these results.

• The immunogenicity of tumors has provided the foundation stone for the idea of tumor-specific antigens. If one could immunize, then antigens must exist.

• Tumor immunity depends on many factors. The degree of tumor immunity depends upon the type of cancer and the method of its induction, or lack of induction. UV-induced cancers are highly immunogenic, methyl-cholanthrene-induced tumors less so, and spontaneous tumors even less so. Nonetheless, immunogenicity of tumors has been demonstrated in all model systems tested.

• The primary animal develops immunity to subsequent tumor challenge. Furthermore it is also immune to subsequent challenges with the same tumor (see Fig. 22.1).

• Protective immunity is only seen in prophylactic immunization and not therapeutic immunization – once a mouse has been implanted with a tumor, immunization with irradiated cancer cells (derived from the growing tumor) does nothing to mitigate tumor growth (Fig. 22.2). Exploration of differences between prophylaxis and therapy has led to fundamental insights into tumor immunity.

Fig. 22.2 Difference between prophylaxis against future tumors and therapy of pre-existing tumors

Protective immunity is only seen in prophylactic immunization and not therapeutic immunization – once a mouse has been implanted with a tumor, immunization with irradiated cancer cells (derived from the growing tumor) does nothing to mitigate tumor growth.

It is obviously not possible to test immunogenicity of tumors in humans by the transplantation-challenge experimental paradigm. There is no other reliable method of determining immunogenicity of tumors. It is therefore impossible to comment on the immunogenicity of human tumors. Much of the work in human cancer immunity has been done with melanomas leading to suggestions that, among human tumors, melanomas are particularly immunogenic. This is an erroneous belief – melanomas are simply the easiest human tumors to culture in vitro and hence the easiest to study.

Spontaneous tumors are also antigenically distinct

Cross-reactivity among tumors does occur occasionally. Typically, such cross-reactive immunity has been observed to be significantly weaker than the individually specific antigenicity. Efforts at characterization of such cross-reactive tumor-protective antigens have not made much headway, except in the case of virally induced tumors. In contrast, there has been considerable success in identification of individually distinct antigens.

In addition to these experimental studies, several clinical observations point to the existence of tumor-protective immunity in humans. These include the increased relative risk of cancers in patients who are immunosuppressed because they are kidney transplant recipients (Fig. 22.3) or for a variety of other reasons (Fig. 22.4).

Fig. 22.3 Relative risk of tumors in immunosuppressed kidney transplant

In all forms of immunodeficiency the relative risk of developing tumors in which viruses are known to play a role is greatly increased. This is the case for all those listed except cancer of the lung. The relative risks vary in different studies according to the duration of follow-up and the presence of co-factors such as sunlight for skin cancer.

Fig. 22.4 Tumor viruses and immunodeficiency

Skin cancer is the most common form of tumor in absolute numbers in organ transplant recipients. In other forms of immunodeficiency, tumors of the immune system dominate. Most normal adults carry both Epstein–Barr virus (EBV) and many papilloma viruses throughout life with no ill effects because they have antiviral immunity.

Q. The observation that immunosuppressed individuals are more susceptible to tumors can be explained in at least two different ways. Suggest what these explanations might be

A. The individual may be unable to control virus infections that lead to cancer (oncogenic viruses), or they may be unable to survey and control primary tumors that arise by mutation. The explanations are not mutually exclusive.

Characterization of tumor antigens

The two broad approaches to the identification of tumor antigens are shown in Figure 22.5 and discussed individually below. Not surprisingly, the approaches have yielded results that are not fully concordant. These differences have helped highlight a fascinating interplay between immunity and tolerance to tumors, discussed below.

Fig. 22.5 Comprehensive list of major types of tumor antigen

A comprehensive list of the major types of tumor antigen.

Tumor-specific antigens defined by immunization all belong to the family of HSPs

When tumors were biochemically fractionated and individual protein fractions tested for their ability to elicit protective tumor immunity, a number of tumor-protective antigens were identified in diverse tumor models, such as mouse sarcomas, melanomas, colon and lung carcinomas, and rat hepatomas.

Interestingly, regardless of the tumor models used, all antigens were found to belong to the family of proteins known as the heat-shock proteins (HSPs), which:

• could elicit protective immunity;

• were of the HSP90 (gp96 and HSP90), HSP70 (HSP110 and HSP/c70), calreticulin, and HSP170 (also known as grp170) families.

HSPs must be isolated directly from tumors to be immunologically active

Two aspects of HSP-elicited tumor immunity are notable.

• First, HSPs are present in normal tissues as well as tumors, and normal tissue-derived HSPs do not elicit rejection. They must be isolated from tumors to be immunologically active.

• Second, HSPs elicit immunity specifically against the individual tumors from which they are isolated.

These two observations suggested that HSPs in tumors differ from those in normal tissues and that HSPs in each tumor differ from the same molecules in other tumors.

This conundrum was resolved by the demonstration that the HSP molecules chaperone peptides in a peptide-binding pocket, much as the MHC molecules do, although the structural details of the pockets in HSP and MHC differ (Fig. 22.6). The specificity of immunogenicity derives from the peptides rather than the HSP itself – dissociation of HSP-associated peptides from HSPs abrogate the tumor rejection activity.

Fig. 22.6 Unique immunogenicity of HSP-peptide complexes

HSP molecules chaperone peptides in a peptide-binding pocket. The specificity of immunogenicity derives from the peptides rather than the HSP itself – dissociation of HSP-associated peptides from HSPs abrogated the tumor rejection activity.

HSPs can chaperone many different peptides hence the HSP-chaperoned peptides contain among them any tumor-specific antigenic epitopes present in the tumor cell or the antigenic fingerprint of the tumor from which the HSPs are isolated.

HSPs can chaperone many different types of antigen

HSP-chaperoned peptide pools contain cytotoxic T lymphocyte (CTL) epitopes and CTL epitope precursors for any antigens that the cell expresses (Fig. 22.w1). This evidence comes from mouse and human tumors, normal tissues, and virus-infected cells.

Fig. 22.w1 CTL epitopes isolated from HSP-peptide complexes

HSP-chaperoned peptide pools contain CTL epitopes and CTL epitope precursors for any antigens that the cell expresses.

(Redrawn from Srivastava P. Nat Rev Immunol 2002;2:185–194. Copyright 2002, Nature Reviews Immunology.)

HSP molecules bind to APCs and target peptides with high efficiency

The HSP molecule itself plays at least two crucial roles other than chaperoning peptides:

• HSPs bind antigen-presenting cells (APCs) such as macrophages and dendritic cells (DCs) through HSP receptors such as CD91 and thus target the peptides chaperoned by them into the APCs with high efficiency;

• further, the HSP-chaperoned peptides, once introduced into the APC, follow the endogenous as well as the exogenous pathway of antigen presentation and are processed and re-presented by the MHC I and MHC II molecules of the APCs (Fig. 22.7).

Fig. 22.7 Mechanism of specific immunogenicity of HSP-peptide complexes

Once introduced into the APC, the HSP-chaperoned peptides follow the endogenous class I as well as the exogenous class II pathway of antigen presentation and are processed and represented by the MHC class I and MHC II molecules of the APCs. HSP ligation induces production of chemokines and cytokines, and upregulation of CD40 and CD80/85. (MCP-1, monocyte chemotactic protein-1; MIP-1α, macrophage inflammatory protein-1α.)

(Redrawn from Srivastava P. Nat Rev Immunol 2002;2:185–194. Copyright 2002, Nature Reviews Immunology.)

Q. How can an APC present internalized antigen via MHC class I molecules?

A. It can do this via the mechanism of cross-presentation.

It is by this mechanism that immunization with tumor-derived HSPs elicits a CD8 as well as CD4 response against the tumors. In addition, the HSP molecules stimulate the APCs to mediate maturation of DCs and secretion of an array of cytokines that provide the innate milieu for the adaptive response.