Previous Next

Serological Definition of Cell Surface Antigens of Mouse Leukemia

Lloyd J. Old

Elizabeth Stockert
Memorial Sloan-Kettering Cancer Center
New York, New York

The discipline of tumor immunology, like so many other biological fields of contemporary importance, owes its origin to the innovators whom we honor at this workshop. Without their development and characterization of inbred strains, it would be difficult to imagine how the principles that underlie modern immunological approaches to cancer could have been established. The demonstration of tumor-specific antigens in chemical and virus-induced tumors, the development of methods to analyze cellular and humoral immunity to tumors, and the finding that tumor immunity could be bolstered by microbial agents, such as BCG, represent the legacy from work with the inbred mouse that has engendered the vast interest in applying similar approaches to human cancer.

To the cancer immunologist, cell surface antigens are of paramount importance, for it is novel antigens incorporated into the tumor cell surface that can serve as transplantation antigens and provoke immunological rejection. Although such transplantation techniques provided the first evidence for tumor-specific surface antigens, serological techniques have far more potential for analyzing the surface composition of tumor cells. With advances in methods to determine cellular immune reactions in vitro, these can be expected to take their place with serological techniques in the analysis of cell surface antigens of tumors. Present knowledge about these antigens, however, has come almost exclusively from the application of serological techniques, and of these, the cytotoxic test has proved most successful in the identification and analysis of surface antigens, some of which would have been impossible to discover by techniques involving graft rejection.

The cytotoxic test, originally developed by Gorer and O'Gorman to detect antibody to H-2 products, is a simple technique in which cells are lysed by antibody in the presence of a suitable source of complement ( 1). The test has evolved through a series of modifications and, in its present form, is an analytical method of unrivaled specificity and sensitivity for the detection of cell surface antigens. Because leukemia cells can be easily obtained in free cell suspension and are exquisitely sensitive to cytotoxic antibody, they have been a favorite object of study. The great majority of mouse leukemias are T cells of thymic origin, and this has allowed the serologist to study the leukemia cell side by side with its normal counterpart, the thymocyte. As a consequence of these advantages, more is known about the surface antigens of normal an malignant T cells than about any other cell population of the mouse ( 2, 3, 4).

Four general classes of cell surface antigens have been distinguished on mouse leukemias: 1) conventional H-2 alloantigens, that are present on virtually all cells of the adult mouse; 2) differentiation alloantigens, e.g., Thy-1 and the Lyt series, and several products of the H-2-Ir and Q complex whose presence signify selective gene activation in cells undergoing T cell differentiation; 3) MuLV-related antigens, e.g., GCSA, GIX, G(RADA1), that owe their origin to genetic information of endogenous murine leukemia viruses; and 4) TL (thymus-leukemia) antigens that occur as differentiation alloantigens restricted to normal thymocytes in TL+ strains of mice and as leukemia-specific antigens in mouse strains not normally expressing TL antigens (TL- strains).

Before discussing the MuLV and TL systems in more detail, some mention should be made of the immunization procedures that were devised to produce antibody to these cell surface antigens ( Table 1). To define non-H-2 specificities, the challenge was to develop methods of immunization or testing that would eliminate the contribution of H-2 antibodies. This has been accomplished in three principal ways. The first can be illustrated by the approach that led to the detection of TL and GCSA antigens and involves immunization with H-2 incompatible leukemia cells and testing the resulting antiserum on syngeneic leukemia cells ( 5, 6). The rationale behind this immunization was that the allogeneic and syngeneic leukemia cells share a common antigen. In the case of TL, C57BL mice were immunized with A strain spontaneous leukemia and the antiserum was tested on a C57BL x-ray-induced leukemia, thus eliminating reactions due to H-2 and other alloantibodies. The second method involves donor-recipient combinations that are H-2 compatible and this was first used to produce antibody to θ, or as it is now known, the Thy-1 antigen ( 7). Syngeneic immunization, although theoretically the method of choice to produce antibody with leukemia specificity, has failed to uncover leukemia-specific antigens in spontaneous, x-ray or chemically induced mouse leukemias. Two antigenic systems, however, were defined with antisera raised by syngeneic immunization; the FMR complex of antigens associated with leukemias induced by Friend, Moloney, and Rausher leukemia viruses ( 8, 9) and the MuLV-related GIX antigen, originally identified with sera produced in W/Fu rats immunized with syngeneic MuLV-induced leukemia cells ( 10). The ease with which antibody can be raised to these antigens is undoubtedly related to the strong immunogenicity of FMR mouse leukemias and MuLV-induced rat leukemias in syngeneic hosts. These leukemias can undergo regression after initial growth, in contrast to the invariable growth of transplants of spontaneous leukemias or leukemias induced by physical or chemical agents in their strain of origin.

In addition to antisera prepared by deliberate immunization, naturally occurring antibodies in normal mouse serum are becoming increasingly valuable as reagents to define the spectrum of cell surface antigens specified by murine leukemia viruses ( Table 2). With the exception of antibody to GCSA, which has not yet been found in the serum of normal mice, antibody to the other MuLV-related specificities is found with characteristic frequency in different mouse strains. Thus far, F1 hybrids have been the best source of these antibodies, and this most probably has to do with heterozygosity at Ir loci which broadens the range of antigens that can be recognized.

The list of MuLV-related specificities recognized by mouse antibody is rapidly growing and is coming to parallel the remarkable diversity of MuLV recognizable by virological and biochemical methods. The use of the letter G to designate cell surface antigens related to naturally occurring MuLV was intended to honor Ludwig Gross, the discoverer of this class of oncornaviruses. Until a more definite nomenclature can be devised for these antigens, it has been proposed that new MuLV-related cell surface specificities of the G class be named after the prototype normal or leukemic cell used in defining the antigen, and this has been done in the case of G(RADA1), G(ERLD), and G(AKSL2). Because the GCSA and GIX designations are widely used, it would seem unwise to rename them according to this new convention at the present time.

Each of the five MuLV-related surface antigens can be distinguished on the basis of their strain distribution, tissue distribution, and appearance in leukemias and other tumors of strains that normally do not express these antigens. Their specification by MuLV is shown in two general ways: relation to MuLV structural components, either by absorption studies with purified MuLV proteins or by immunochemical characterization of the cell surface molecule, and antigen induction following MuLV infection of permissive cells. In this way, GCSA has been shown to be a glycosylated polyprotein related to p15 and p30 viral core proteins, and GIX, G(RADA1) and G(ERLD) have been shown to be type-specific determinants of gp70 molecules. In assays of antigen induction by MuLV, GCSA appears to be a general marker for MuLV infection, with GCSA expression being induced by ecotropic, xenotropic, and amphotropic MuLV. In the case of the gp70-related antigens, GIX and G(RADA1) induction is a common property of ecotropic MuLV, particularly those with N-tropism, and G(ERLD) is closely related to xenotropic MuLV in induction assays. The most recently defined MuLV-related specificity, G(AKLS2), is closely related to xenotropic MuLV in induction assays. The most recently defined MuLV-related specificity, G(AKLS2), is related to the dualtropic MuLV, MCF 247, a virus thought to arise by recombination between ecotropic and xenotropic MuLV ( 19), and recognition of the G(AKLS2) system provides a way to examine the natural history of this virus in the mouse.

Of the various MuLV-related cell surface specificities, the GIX trait has been the subject of the most comprehensive immunogenetic analysis ( Table 3). In normal mice, GIX behaves like an alloantigen, being present in some mouse strains and not in others. It also has the characteristics of a differentiation antigen, being present in some tissues and absent in others, and this feature is most evident in strains such as 129, where thymocytes are the only GIX+ lymphoid cells. Thymocytes from different GIX+ strains show characteristic quantitative differences in GIX expression and, as absorption capacity follows a ratio of 3:2:1, the three phenotypes have been termed GIX3, GIX2, and GIX1. Segregation data are consistent with a two-gene specification of the GIX trait, and these two genes have been designated Gv-1 and Gv-2. Despite considerable effort, the chromosomal loci of these genes remain unknown ( 4, 20). The relation of GIX to MuLV has been shown in a variety of ways: 1) in vivo infection of GIX- mouse and rat strains with MuLV leads to GIX appearance, 2) in vitro infection of permissive cells with certain ecotropic and MCF MuLV leads to GIX appearance, and 3) the GIX determinant on the surface of thymocytes and MuLV-infected cells resides on a molecule with antigenic and biochemical characteristics of MuLV-gp70. Peptide maps of the GIX-gp70 on normal thymocytes suggest a relationship to xenotropic MuLV ( 21), even though MuLV-induction assays have shown a close association of the GIX trait with ecotropic MuLV ( 12). This would suggest that thymocyte GIX represents the product of an ancestral gene coding for a recombinant gp70 molecule that has become fixed in the species, with the GIX determinant originally contributed by ecotropic MuLV and the remainder of the gp70 molecule by xenotropic MuLV. The fact that the MCF 247 MuLV, a virus with both ecotropic and xenotropic properties, codes for GIX is consistent with this idea.

From studies of leukemic mice, we know that all mice have the genetic information specifying GIX, even though they may not express GIX during normal life as a consequence of differentiation signals or MuLV activation. The basis for this statement is the observation that GIX+ leukemias occur in GIX- strains of mice as well as in GIX+ strains, and this may occur in the absence of MuLV replication. This activation of normally silent genetic information as a consequence of leukemogenesis was first recognized during our analysis of another class of cell surface molecules, the TL system of antigens ( 5). TL antigens and the GIX antigen have many features in common, with the notable exception that there is presently no evidence linking MuLV to the TL system.

In normal mice, TL is found exclusively on thymocytes, no other normal tissue expressing the antigen ( Table 4). Mouse strains can be typed TL+ or TL- on the basis of the presence or absence of TL antigen on thymocytes and, like GIX, antigen-negative mice have no alternative antigens specified by an alternate allele. TL is inherited as a Mendelian dominant trait, and linkage studies have placed the TL locus, designated Tla, on chromosome 17 < 2 units from the D end of the H-2 complex. The key feature of the TL system in regard to malignancy is the anomalous occurrence of TL+ leukemias in strains with the TL- phenotype ( Table 5). This can best be seen by comparing the TL phenotype of normal thymocytes and of leukemia cells in mice with different Tla haplotypes. The three Tla haplotypes determine three TL phenotypes in normal mice. A strain mice with the Tlaa haplotype express three specificities of TL, TL.1, TL.2, and TL.3, on their normal thymocytes. C57BL/6 mice, the prototype strain with the Tlab haplotype, express no TL antigens on normal thymocytes. BALB/c mice with the Tlac haplotype express TL.2 alone on normal thymocytes. In mice with the Tlaa haplotype, TL+ leukemias resemble normal thymocytes and no evidence for appearance of anomalous TL components has been found. Leukemias arising in mice with the Tlab or Tlac haplotype may be either TL- or TL+, and TL+ leukemias invariably express anomalous TL components; TL.1, TL.2 and TL.4 in the case of C57BL leukemias and TL.1 or TL.1 and TL.4 in the case of BALB/c leukemias.

The anomalous appearance of TL antigens has been attributed to the universal presence of structural genes for TL antigens in the mouse. According to this view, regulatory genes determine whether TL is expressed on normal thymocytes. Thus, segregation for the Tl trait in normal mice in based on expression vs. nonexpression alleles at the regulatory locus rather than presence vs. absence of TL structural genes. Leukemogenesis leads to an alteration in this regulatory mechanism in TL- mice, with consequent activation or derepression of TL genetic information and the appearance of TL antigens on the surface of leukemia cells. Antigen systems with the three distinct features of TL -- namely, genetic linkage to the major histocompatibility complex, restriction to normal thymocytes, and anomalous appearance on leukemias -- have not been found in any other species, but it would be surprising if the mouse were unique in this regard.

With the range of surface markers that have been identified on T cells of the mouse, the surface phenotype of normal thymocytes and leukemias of thymic origin is becoming well characterized. A comparison of the surface antigens of three leukemias that have been extensively studied with the corresponding normal thymocyte population is shown in Table 6.

The thymic origin of the AKR spontaneous leukemia AKSL2 is shown by the Thy-1 marker, despite the fact that no Lyt antigens could be detected on this leukemia. As is true of most spontaneous leukemias of AKR mice, the leukemia cells express no anomalous TL components. The five MuLV-related specificities found on the leukemia cells are also present on normal thymocytes. Thus, with regard to TL and MuLV-related antigens, the surface phenotype of this AKR leukemia does not differ from normal thymocytes, at least in a qualitative sense.

The A strain x-ray-induced leukemia RADA1 also lacks Lyt antigens, but its T cell origin is indicated by Thy-1 and by the three TL specificities (TL.1, 2, 3) that appear on RADA1 as differentiation alloantigens. The other surface feature that distinguishes RADA1 from normal A strain thymocytes is the anomalous appearance of the MuLV-related specificity G(Rada1) on the leukemia cell.

The surface phenotype of the C57BL x-ray-induced leukemia ERLD illustrates the phenomenon of anomalous TL appearance. In contrast to the TL- phenotype of normal C57BL thymocytes, leukemia ERLD has a TL.1, 2, 4 phenotype. With regard to other surface antigens, ERLD and C57BL thymocytes are identical.

The recognition that cell surface antigens such as TL and G(RADA1) may be leukemia specific in some strains, yet be normally expressed as differentiation alloantigens or MuLV-related antigens in other strains, has been an important contribution of basic immunogenetics to tumor immunology. As yet, no transformation-specific surface antigen restricted to leukemia cells of the mouse has been found. TL.4 comes closest to fulfilling this characteristic, having never been found on any normal cell of any mouse strain. Recent work, however, has shown that anomalous TL components, including Tl.4, are expressed early in the preleukemic phase of x-ray leukemogenesis, prior to the emergence of fully autonomous cells, and should therefore be considered markers for preleukemic changes rather than as transformation-specific traits ( 22).

Table 7 summarizes current knowledge about the categories of surface antigens of mouse leukemia cells ( 4). The term "derepression antigens" has been used to distinguish antigens appearing on the surface of tumor cells that are coded for by normally silent genetic information. The TL and MuLV-related antigens appearing on leukemias of TL- and MuLV- strains would be prime examples of derepression antigens. Tumor antigens coded for by genes active only in embryonic or fetal life would also belong in this category, but, despite considerable interest in the possibility of such tumor antigens, their existence remains to be proven. Another category of cell surface antigens of particular interest to the tumor immunologist is the individually distinct or unique tumor-specific antigen, first demonstrated by transplantation techniques in chemically induced sarcomas of the mouse ( 23, 24, 25) and now known to be present on several other tumor types of mouse and rat. These antigens are characterized by a remarkable polymorphism, no two tumors, even if induced in the same mouse, sharing identical antigens. The origin of these unique antigens has been the subject of considerable speculation, with both genetic and epigenetic theories having been advanced, but their nature remains as obscure now as it was when they were discovered. With the recognition that MuLV exists in a far more polymorphic state than originally envisioned, the possibility that these unique antigens owe their origin to MuLV genes must be reconsidered, especially since recombinational event between different classes of MuLV and between MuLV and host genes, which appears likely as a source of MuLV variation, would be expected to give rise to an enormous repertoire of new antigens. Further understanding awaits serological and biochemical definition of these antigens, and advances in both these directions have recently been made with a methylcholanthrene sarcoma of BALB/c mice ( 26, 27). As yet, little effort has gone into the serological detection of unique antigens on leukemia cells. Their presence could be easily obscured by cross-reacting systems such as Tl or MuLV-related antigens. With the background of information we now have concerning the surface phenotype of leukemia cells, planned immunizations devised to detect new systems of surface antigens, including the individually distinct type, should make the next phase of immunogenetic analysis a most revealing one.


1. Gorer, P.A., and O'Gorman, P. (1956). Transplant. Bull. 3: 142.

2. Old, L.J., and Boyse, E.A. (1973). Harvey Lect. 67: 273.
See also PubMed.

3. Boyse, E.A., and Old, L.J. (1978) Harvey Lect. 71: 23.
See also PubMed.

4. Old, L.J., and Stockert, E. (1977). Ann. Rev. Genet. 11: 127.

5. Old, L.J., Boyse, E.A., and Stockert, E. (1963). J. Natl. Cancer Inst. 31: 977.

6. Old, L.J., Boyse, E.A., and Stockert, E. (1965). Cancer Res. 25: 813.
See also PubMed.

7. Reif, A.E., and Allen, J.M.V. (1964). J. Exp. Med. 120: 413.
See also MGI.

8. Old, L.J., Boyse, E.A., and Sockert, E. (1964). Nature 201: 777.

9. Klein, E., and Klein, G. (1964). J. Natl. Cancer Inst. 32: 547.
See also PubMed.

10. Stockert, E., Old, L.J., and Boyse, E.A. (1971). J. Exp. Med. 133: 1334.
See also MGI.

11. Snyder, H.W., Jr., Stockert, E., and Fleissner, E. (1977). J. Virol. 23: 302.
See also PubMed.

12. O'Donnell, P.V., and Stockert, E. (1976). J. Virol. 20: 545.
See also PubMed.

13. Obata, Y., Ikeda, H., Stockert, E., and Boyse, E.A. (1975). J. Exp. Med. 141: 188.
See also PubMed.

14. Obata, Y., Stockert, E., Boyse, E.A., Tung, J-S., and Litman, G.W. (1976). J. Exp. Med. 144: 533.
See also PubMed.

15. Tung, J-S., Vitetta, E.S., Fleissner, E., and Boyse, E.A. (1975). J. Exp. Med. 141: 198.
See also PubMed.

16. Obata, Y., Stockert, E., O'Donnell, P.V., Okubo, S., Snyder, H.W., Jr., and Old, L.J. (1978). J. Exp. Med. 147: 1089.
See also PubMed.

17. Obata, Y., Stockert, E., DeLeo, A.B., O'Donnell, P.V., Snyder, H.W., Jr., and Old, L.J. Manuscript in preparation.

18. Stockert, E., DeLeo, A.B., O'Donnell, P.V., Obata, Y., and Old, L.J. Manuscript in preparation.

19. Hartley, J.W., Wolford, N.K., Old, L.J., and Rowe, W.P. (1977). Proc. Natl. Acad. Sci. 74: 789.
See also MGI.

20. Stockert, E., Boyse, E.A., Sato, H., and Itakura, K. (1976). Proc. Natl. Acad. Sci. 73: 2077.
See also MGI.

21. Tung, J-S., O'Donnell, P.V., Fleissner, E., and Boyse, E.A. (1978). J. Exp. Med. 147: 1280.
See also PubMed.

22. Stockert, R., and Old, L.J. (1977). J. Exp. Med. 146: 271.

23. Gross, L. (1943). Cancer Res. 3: 326.

24. Foley, E.J. (1953). Cancer Res. 13: 835.
See also PubMed.

25. Prehn, R.T., and Main, J.M. (1957). J. Natl. Cancer Inst. 18: 769.
See also PubMed.

26. DeLeo, A.B., Shiku, H., Takahashi, T., John, M., and Old, L.J. (1977). J. Exp. Med. 146: 720.
See also PubMed.

27. Natori, T., Law, L.W., and Appella, E. (1977). Cancer Res. 37: 3406.
See also PubMed.

Previous Next