Immunology Today, voL 7, No. 5, 1986
'Immune surveillance'without immunogenicity The hypothesis of immune surveillance against cancer is based on two premises; (1) that transformed and normal cells generally have different antigenic qualities, and (2) that the immune system responds to the antigenically modified cells in essentially the same way as it responds to invasive microorgam isms. Both premises have been questioned. Here, Zvi Grossman and Ronald Herberman suggest that lymphoid cells not only mediate immune responses but also assist in regulating the differentiation of a variety of normal cells. They do so by recognizing self rather than foreign antigens. By forcing and steering the turnover of tissue cells, lymphoid cells prevent the accumulation of small irregular phenotypic and karyotypic changes in the tissue. Tumor escape from surveillance may therefore be described as escape from regulatory differentiation pressures. The analysis of populations has reached a high level of conceptual sophistication in the study of genetics and ecology. In contrast, cell-population dynamics is usually given surprisingly little weight in interpreting observations in immunology and cancer biology. The tendency to seek all the answers inside (or on) the cell surface has recently increased with the advance of molecular biology. The Ehrlich-Thomas-Burnet theory of immunological surveillance of cancer ~ grew out of conceptions about cancer which are still largely accepted today: (1) cancer is caused by discrete change, or changes, in the cell genome; and (2) a series of additional mutations, in the broad sense of the word, account for the progressive evolution of the tumor phenotypes- these mutations are due to the development of genetic instability in the transformed cells s,6. The theory also contained the following conjectures: (1) transformation events occur frequently and yield cells with abnormally coded proteins; and (2) these proteins are recognized in most cases as foreign by the immune cells, and the immune system destroys the antigenically modified cells - cancer arises when the immune system is depressed or when the transformed cells lack antigenicity 3,4. The observation that immunosuppression increases the frequency and spread of certain tumors (reviewed in Ref. 7) was considered to corroborate this prediction. However, the limited range of tumors which arise under immune suppression indicates that the spontaneous transformation events may be less frequent than was thought and that genetic predisposition and other, different mechanisms may be controlling tumorigenesis 8'9. The cells proposed to carry out surveillance now include T and B cells and also natural killer (NK) cells and macrophages. Are spontaneously transformed cells antigenic? It has been difficult to demonstrate directly that most spon-
ITeI-AvivUniversity,TelAviv, Israel 1,2 Pittsburgh Cancer Institute and Departments of'Medicine and Mathematics, University of Pittsburgh, Pittsburgh, PA 15213-2592, USA. ~) 1986, ElsevierScience Publishers B.V., Amsterdam
Zvi Grossman1 and Ronald B. Herberman2 taneous tumors are immunogenic. It was proposed that subtl e changes in the pattern of activation of 'banal genes '1°'11 might be responsible for tumorigenicity. Except for quantitative differences, phenotypes of leukemic cells are 'normal gene products expressed appropriately in accord with the cells' maturation and proliferation status '12. In modern terms, proto-oncogenes are normal genes that may sometimes cause cancer simply by being expressed at a level different from that in normal cells. Even if certain transformed cells are antigenically different from their normal antecedents, or acquire antigenic qualities later, it is questionable whether the immune response to neoplasia could be regarded as equivalent to homograft rejection, as was originally postulated 2. Some aspects of transplantation phenomena and of the development of immunologic tolerance are particularly pertinent. The term 'sneakingthrough' describes a reproducible growth pattern that was observed when different numbers of potentially immunogenic tumor cells were transplanted: a small number grew progressively, a medium number was rejected, and a large number again broke through 13'14. This phenomenon was reproduced and analysed with the help of a simple mathematical model is which has been a paradigm for the analysis of tumor escape mechanisms16 18. The analyses suggest that the strength of a response depends not on the magnitude of the antigenic load but rather on the associated gradients of change of antigen levels. The typical rise and fall in the number of effector cells in response to a sudden challenge is characteristic of a transient phenomenon. In contrast, persistent or gradually changing antigenic stimulation selects low-affinity lymphocyte populations that cannot easily be depleted by terminal differentiation or blocked by specific factors. The sneaking-through phenomenon was linked to the induction of 'low-zone tolerance' by repeated stimulation with small antigen concentrations ~6-18. Thus, it may be quite possible for a potentially immunogenic tumor to develop in the presence of an intact immune system. Instead of mounting a destructive response, the immune system may adapt, through a process of selection, to the tumor growth. Similarly, positive selection of low-affinity proliferating lymphocytes in the vicinity of a parasite population may provide protection of the parasite from more effective clones m. These considerations complicate questions about the antigenicity and immunogenicity of physiological tumors and require an examination of alternative explanations for the observed association between immunosuppression and tumor promotion. In the absence of foreign determinants this association might not be causal but rather the suppressive agents might be carcinogenic as such, inducing genetic transformations in the target cells. This is the only explanation consistent with both a tack of immunogenicity and the notion that genetic instability has developed in the cells. On the other hand an approach to cancer development
Immunology Today, vol. 7, No. 5, 1985
which emphasizes the interplay between cellular characteristics and multicellular organization 2°-22 can offer a different explanation. We propose that, in tissues with a rapid cellular turnover, most of the actively dividing cells are forced to be transitory. (i.e., they are in the process of differentiating to another state). The 'force' is associated with the organization of the tissue, is dynamic and allows for the variability that is required physiologically in the proliferative activity and growth characteristics of the cells. At the same time, it prevents the prolonged accumulation of small phenotypic and karyotypic changes in the progeny of transitory cells. (The susceptibility of these cells to the cumulative effect of such changes is demonstrated by the phenotypic plasticity2°'23-38 and genetic instabilitys'6'2° of cultured cells and established tumors.) The minority of self-renewing (stem) cells which remain in the tissue for a prolonged period is protected from the transformation process by its small number, its ,primitiveness,21,22, its slow rate of division39, 'protective niches '4°'41, and by other mechanisms 42. We suggest that lymphoid cells have a regulatory function, forcing and steering the turnover of tissue cells. This function goes beyond the classical role of lymphoid cells as mediators of immune responses against foreign invaders. We also suggest that impairment of this regulatory function is itself part of the tumorigenic process. This hypothesis is further developed below in two stages: the first part outlines a broad scenario for the dynamics of tumor progression, following a recently proposed model of leukemogenesis; the second part describes a role for lymphoid cells in this scenario. Tumor progression Many tissues, especially those with a rapid turnover, are renewed by means of stem cells. The relative rates of self-renewal and differentiation of normal mitotic cells are regulated by microenvironmental and feedback interactions among cells. Self-renewal capacity is a measure of cells' resistance to differentiation pressures; normally, it declines as the cells undergo a series of phenotypic changes in their patterns of gene expression - a process called maturation - heading towards the stage of terminal differentiation. The differentiation pressures in the tissue change with the size and composition of the tissue, providing a form of feedback control. At steady state, the feedback differentiation pressure is adjusted so that the most primitive cells are replicating and differentiating at the same rate, while the more differentiated cells are forced to be transitory, namely to have a higher probability of maturation than of replication. The rule that cells become more responsive to differentiation signals as they mature makes maturation an autocatalytic process. This, together with the feedback assumption, ensures the stability of the maturation hierarchy21'22, namely of the normal phenotypic profile. We postulate, however, that cell replication is also potentially autocatalytic: the force that drives replication tends to endow the cell with a slowly increasing capacity for self-renewal or, equivalently, decreasing inducibility to differentiation. For the most primitive cells, the two opposing forces are balanced endowing them with 'stemness'. For more differentiated cells, the autocatalytic circuit of differentiation is normally dominant.
rostrumExtensive evidence supports the concept of phenotypic adaptability, namely the capacity of somatic cells to change their patterns of gene expression, in a heritable manner, in response to changes in their microenvironment. Tissue interactions determine the developmental fate of cells in the embryo and are required in adult life to maintain the identity of cells2°. There are many examples of unusual phenotypic switches in tumors and in cultured cells that appear to result not from mutations or genetic aberrations but from inheritable epigenetic effects 23 28. A role for DNA methylation in stabilizing epigenetically induced changes of gene expression has been proposed 29'3°. The assumption that cellular growth characteristics are subject to adaptive changes in both directions, and not only to down-regulation of the selfrenewal capacity with differentiation, is indirectly sup ported by several observations21'22. External influences can change the self-renewal capacity of hemopoietic cells31'32. DNA methylation may have a role in regulating blast cell self-renewal 33. Striking heterogeneities exist in the capacity of hemopoietic cells belonging to the same 'compartment' to form colonies 34. Cultured cell lines and clones often undergo gradual loss of differentiative capacities and of responsiveness to various inducers 2°'3s'36. Some growth factors can regulate the expression and affinity of their own receptors 37'38. To describe cancer progression as a consequence solely of a sequence of intra-cellular events is not compatible with an approach based on the assumption that a network of cell-cell interactions is involved in the regulation of cell division and differentiation. If in somatic cells possessing extensive division capacity inheritable changes occur which affect the cells' growth and differentiation characteristics, the tissue composition is bound to change. This in turn necessarily leads to additional cellular changes, and s o o n 20 22 Certain perturbations of the tissue can weaken the differentiation feedback loop, leading to a self-driven cascade of interrelated changes in the growth characteristics of the individual cells and in the microenvironment, and eventually to malignancy. As transitory cells are progressively relieved from external maturation pressures, they not only increase their self-renewal activity but become less inducible to maturation. Such a transformation toward cellular 'immortality' may be initially largely independent of DNA aberrations, although the aberrations could eventually contribute to it. The cells may eventually specialize in continuous division, instead of other functions of differentiated cells; they adapt to express a set of genes associated with division at the expense of other genes, including presumably those genes responsible for maintaining the karyotypic integrity of the cell. Cell crowding, with the associated distortion of the population balance in the tissue, may represent such a perturbation, which can destabilize the normal cell organization 21,22. A mathematical model 2t of leukemia progression supports the following scenario. A genomic, heritable event in an early hemopoietic cell reduces the responsiveness of the clone to feedback. Consequently, the clone expands to maintain the balance between self-renewal and maturation of stem cells; global expansion, encompassing the whole marrow, is associated with chronic myelocytic leukemia (CML), whereas local, microscopic expansion is thought to occur in the preleukemic stage of acute leukemia. Cells then crowd
rostruMr. together and there is the beginning of selection in favor of blast cells and against (nondividing) mature cells and (slowly dividing) primitive cells. If crowding exceeds a certain threshold, blast cells adapt to the distorted microenvironmental conditions by increasing their selfrenewal capacity. As a result, the balance among mature and immature cell populations is distorted even further. This process is one of positive feedback - a snowballlike process of slipping control - leading eventually to blast cell dominance. Alternative scenarios have also been proposed 21'22. It may be noted that destabilization occurs not because of the accumulation of phenotypic changes in any particular cell and its progeny. The effect of previous changes is imprinted on the collective memory of the system through the modification of the tissue composition. The role of the initiating heritable event is only to cause and maintain a state of cell crowding for a sufficient length of time.
Tissue organization, lymphoid cells and surveillance
The leukemogenesis model described above assumed that cells in the mature compartment positively affect the differentiation of earlier hemopoietic cells21'22 In general, the feedback relationships in the tissue may include interacting cell populations which are not necessarily derived from a common stem cell. The cell populations which stimulate and regulate each other are normally linked in a stable state of 'functional heterogeneity'. A major perturbation, e.g. DNA rearrangement in a clonogenic cell, cytotoxic drug or some kind of prolonged external stimulation, may produce a modified set of constraints, incompatible with the maintenance of functional heterogeneity, and malignancy may follow through a process of selection and adaptation. We conjecture that lymphoid cells are an important component of the regulatory cell population which generates maturation pressures in some tissues - in particular in the hemopoietic tissues. Since maintenance of these pressures, we propose, is essential for the cell population balance in the tissue and for the stability of its phenotypic profile, our conjecture defines a new mode of immune surveillance. Thus, 'tumor escape from surveillance' may represent a reversal of the feedback circuit of differentiation pressures, involving suppression of some regulatory lymphoid populations as part of the change in the composition of the tissue. It may be sufficient for this suppression to be only local and limited in time, to allow for the acquisition of the malignant, differentiation-resistant phenotype. It has become clear that T cells are a direct source of factors affecting a wide range of lymphoid and hemopoietic cells (reviewed in Ref. 43). Although generally associated by immunologists with antigen presentation, la antigens may play another role in the regulation of hemopoiesis4 3 : they may be involved in la-restricted communication with regulatory T cells. The recognition that the levels of histocompatibility antigens on progenitor cells are influenced by external factors may indicate that MHC products have such a role in hemopoiesis. In addition to T cells, large granular lymphocytes may function not only as natural killer (NK) cells but also as 'natural helper' cells44'4s, and so may be involved in similar regulatory functions. Moreover, growing evidence ir~dicates the existence of reciprocal communication between the immune system and several other cell systems. For instance, the immune and neuroendocrine systems appear to
Immunology Today, voL 7, No. 5, 1986
share a common set of hormones and hormone receptors46. In general, lymphokines and MHC antigens are obvious potential channels of communication. Golub has already expressed the idea that the generation of effector B and T cells may be only part of the manifestations of the immune response 47. He pointed out that the consequences of the interactions of lymphocytes with self-antigens are poorly understood. In one theory of the immune system, latent proliferative responses of lymphocyte precursors, in particular to selfantigens, are seen as a major factor in the generation of memory and tolerance, in determining the balance between help and suppression and in the development and maintenance of self-tolerance and MHC-restriction 16'17. Golub speculated about 'a major function of the immune system as a regulator of normal, cellular function'. Prehn and Lapp~ also criticized the 'wide-spread perception that an immune system is basically a defense reaction', and have consistently advocated the idea that immune interactions with a tumor may often be stimulatory48. It is likely that lymphocytes are more central to the regulation of lymphoid and hemopoietic tissues than to that of other cell systems. This is consistent with the observation that immunosuppression is more often associated with lymphomas and leukemias than with other cancers. One prediction of this new concept of surveillance is that, even when disruption of the immune system is not an apparent cause of the disease, observable changes in the distribution and phenotype of T cells and NK cells may accompany the tumorigenic process. When a preleukemic state exists, or when there are predictable switches in the state of malignancy over time, there may be an opportunity to identify such changes and correlate them with the progression of the tumor. Such changes could be effectively studied by a combination of multivariate flow cytometry and functional assays. One example is the development of secondary acute leukemia in patients with multiple myeloma. Local, subtle changes in lymphoid cell distribution may be more significant during the early stage of the transformation process than global hypoplasia, which may explain in part the lack of consistent association of carcinogenesis with immune deficiency. The approach we discuss here underscores the need to understand the functional heterogeneity within a normal tissue in order to learn how to restore it or prevent its disruption. To probe experimentally into the relationships between transformable normal cells and lymphoid cells it would be necessary to disturb normal homeostatic relations in the tissue. The general methodology required has been discussed by Rubin 2°. The methods employed should not involve the use of carcinogens. Selective depletion/modulation of lymphoid subsets or of lymphokines for prolonged periods of time in limited locations might lead to detectable changes, both phenotypic and functional, in the cells subject to the modified environment. Potential changes in responsiveness to growthrelated factors, in expression of la antigens and in recognition by lymphoid cells are of particular interest, since they may indicate slipping control. Such changes could then be studied in relation to those occurring under more comprehensive perturbations, including carcinogenic ones, or to changes which can be induced in culture. Clonal analysis, and the ability to reverse the changes by restoring normal conditions in the tissue or
Immunology Today, voL 7, No. 5, 1986
by manipulations in culture, could be used to assess their adaptive nature (versus selection of rare mutations in the population). The alternative concept of anti-tumor surveillance w e propose is not conceived as an 'immune response' in the usual sense. However, as Mitchison has recently observed, referring to lymphocytes, 'most of the elements of any complex control system look inwards at one another'49; here it may be recognized that the lymphoid populations and the cells under their surveillance constitute components of one such system. Note added in proof
Dr Marc Lapp~ has drawn our attention to a paper entitled Possible significance of immune recognition of preneop!astic and neoplastic cell surfaces, which he wrote in 1972 (National Cancer Institute Monogr. 35, 49-55). This paper suggested a primary role for the immune system in non-neoplastic surveillance as "a modulator of cell-surface configuration" and a regulator of "the morphologic expression of 'normalcy'". It may be noted that we prefer to view the immune system as a part of the normal, self-sustaining 'functional heterogeneity', rather than a 'regulator of normalcy'. Although the two approaches differ in several respects, and although we have just become aware of Lappc~'s article, we are pleased to acknowledge that our hypothesis can be viewed as an extension and specification of his idea and thus to reiterate the fact that many of the basic concepts which we develop have historical origins. References
1 Ehrlich, P. (1957)in The Collected Papers of Paul Ehrlich, Vol. II (Himmelweit, F. ed.) pp. 550-562, Pergamon Press, London 2 Thomas, L. (1959) in Cellular and HurnoralAspects of the Hypersensitive States (Lawrence, H.S. ed.) pp. 529-532, Hoeber, New York 3 Burnet, F.M. (1964) Br. Meal. Bull. 20, 154-158 4 Burnet, F.M. (1970)Prog. Exp. TurnorRes. 12, 1-27 5 Foulds, L. (1969) Neoplastic Development, Academic Press, New York 6 Nowell, P.C. (1976) Science 194, 23-28 7 Herberman, R.B. (1983) in Advances in Host Mechanisms, Vol. 2 (Gallin, J. and Fauci, A.S. eds) pp. 241-272, Raven Press, New York 8 Klein, G. (1980) in Immunology 80, Progress in Immunology /V(Fougereau, M. and Dausset, J. eds) pp. 680-687, Academic Press, New York 9 Gorelik, E. (1983)Adv. Can. Res. 39, 71-120 10 Mintz, B. and Fleischman, R.A. (1981)Adv. Can. Res. 34, 211-278 11 Greaves, M.F. (1979) in Tumor Markers (Boelsma, E. and Rumke, P. eds) pp. 201-211, Elsevier, Amsterdam 12 Greaves, M.F. and Janossy, G. (1978) Biochim. Biophys. Acta 516, 193-230 13 Old, L.J., Boyse, E.A., Clark, D.A. etaL (1962)Ann. N.Y. Acad. Sci. 101,80-106
14 Bonmassar, E., Menconi, E., Goldin, A. etal. (1974) J. Natl.
Cancer Inst. 53, 4 7 5 4 7 9 15 Grossman, Z. and Berke, G. (1980) J. Theor. Biol. 83,
267-296 16 Grossman, Z. (1982) Eur. J. IrnrnunoL 12, 747-756 17 Grossman, Z. (1984)lrnrnunol. Rev. 79, 119 138 18 De Boer, R.J. and Hogeweg, P. (1985) J. Theor. Biol. 113, 719 19 Grossman, Z., Greenblatt, C. and Cohen, I.R.J. Theor. Biol. (in press) 20 Rubin, H. (1985) CancerRes. 45, 2935-2942 21 Grossman, Z. Int. J. Math. Modeling (in press) 22 Grossman, Z. (1986) EMBOJ. 671-677 23 Schirrmacher, V. (1980) Irnrnunobiology 157, 89-98 24 Kerbel, R.S., Frost, P., Liteplo, R. etal. (1984) J. Cell. Physiol. (Suppl.) 3, 87-97 25 Frost, P. and Kerbel, R.S. (1984) CancerMetast. Rev. 2, 375-378 26 Frost, P., Liteplo, R.G., Donaghue, T.P. etaL (1984) J. Exp. Med. 159, 1491-1501 27 Farber, E. (1984) Biochirn. Biophys. Acta 738, 171-180 28 Smith, L.J. and McCulloch, E.A. (1984) Blood63, 1324-1330 29 Razin, A. and Cedar, H. (1984) Int. Rev. Cytol. 92, 159-185 30 Cedar, H. (1985) in DNA Methylation: Biochernistryand BiologicalSignificance (Rzin, A., Cedar, H. and Riggs, D. eds) pp. 147-163, Springer-Verlag, New York 31 Spooncer, E., Boettiger, D. and Dexter, T.M. (1984) Nature (London) 310, 228-230 32 Chang, L.J-A. and McCulloch, E.A. (1981) Blood 57, 361-367 33 Motoji, J., Joang, T., Tritchler, D. et al. (1985) Blood 65, 894-901 34 McCulloch, E.A., Smith, L.J. and Minden, M.D. (1982) Cancer Surv. 1,279-298 35 Brooks, C.G., Urdal, D.L. and Henney, C.S. (1983)IrnmunoL Rev. 72, 43 72 36 Shortman, K., Wilson, A. and Scollay, R. (1984)J. Irnrnunol. 132, 584-593 37 Smith, K.A. and Cantrell, D.A. (1985)Proc. NatlAcad. Sci. LISA 82,864-868 38 Reem, G.H. and Yeh, N.E. (1985)J. Irnrnunol. 134, 953-958 3g Lajtha, L.G. (1979)Differentiation 14, 23 34 40 Schofield, R. (1978) Blood Cells 4, 7-25 41 Grossman, Z. Leuk. Res. (in press) 42 Cairns, J. (1975) Nature (London) 255, 197-200 43 Schrader, J.W. (1983) CRCCrit. Rev. Irnrnunol. 4, 197-277 44 Grossman, Z. and Herberman, R.B. (1982) in NK Cells and Other Natural Effector Cells (Herberman, R.B. ed.) pp. 229238, Academic Press, New York 45 Grossman, Z. and Herberman, R.B. (1986) CancerRes. (in press) 46 Blalock, J.E. in The RoleofLeukocytes in Heat Defense (Oppenheim, J.J. and Jacobs, D. eds) Alan R. Liss, Inc., New York (in press) 47 Golub, E.S. (1981) Ce1127, 417~418 48 Prehn, R.T. and LappS, M.A. (1971) Transplant. Rev. 7, 26-54 49 M itchison, N.A. (1985) Nature (London) 316, 676