Relationship between major histocompatibility antigens and disease

Relationship between major histocompatibility antigens and disease Possible associations with human arenavirus diseases MICHAEL B. A. OLDSTONE 1 Histocompatibility antigens, virus infections, and disease are discussed relative to, avenues of research in humans with arenavirus infections. The data implicating a relationship between histocompatibility complexes in man and animals and diseases of the central nervous system are reviewed. Histocompatibility antigens may share common antigenic determinants with viruses, act as receptor sites for attachment of viruses, and be altered by viruses. In addition, genes regulating immune responses to a variety of natural and synthetic antigens are linked, in many species, to the major histocompatibility complex. Since injury associated with virus infections may be largely due to the activity of the immune system, study of immune response genes may provide insight into understanding resistance to disease. Further, histoincompatibility reactions can activate latent viruses with resultant disease. An intriguing yet unsolved puzzle in microbiology and medicine is the identification of factors deter- mining host resistance or susceptibility to infec- tious disorders. Although the variation in different hosts' ability to resist infectious diseases is well established, the precise mechanisms have not been clearly defined. The roles played by the immune system, interferon, the infecting agent, and a variety of additional specific and nonspecific factors are difficult to analyse for any given infection. For example, following measles virus infection, why does one child develop the usual illness with rash, mal- aise, and fever, another meningoencephalitis, a third post-infectious encephalitis, and a fourth subacute sclerosing panencephalitis-a latent slow virus infec- tion? Such variability in host responses to patho- genic viruses is the rule and not the exception, perhaps best typified by the range of clinical findings that follow poliomyelitis virus infection, from para- lysis to an inapparent, subclinical reaction. A variety of mechanisms may govern resistance or susceptibility. Some occur at the cell-virus level, yet others derive from the host's immune response. In the cell-virus system, susceptibility is most likely to be related to the presence of suitable receptors for 1 Department of Immunopathology, Scripps Clinic and Research Foundation, La Jolla, CA 92037, USA. virus on the cell's plasma membrane, although sometimes viruses attach to and penetrate into cells without subsequent viral multiplication. In these instances, intracellular factors that prevent uncoat- ing, allow degradation of virus, or disallow mem- brane assembly, may be present. With other viruses, susceptibility or resistance depends in a large mea- sure on the host's immunological capability. Viruses as self-replicating agents provide a supply of macro- molecular antigens and usually elicit a host immune response. The related tissue injury is usually caused when the host reacts against the virus or virus- altered cell surfaces. In acute virus infections, an imbalance of the immune response and virus replica- tion occurs, which either removes the virus and produces immunity or causes the host's death. In chronic virus infections, both ongoing virus replica- tion and a continuous host anti-viral immune re- sponse occur. For several reasons attention has focused on the histocompatibility complex, virus infections, and resultant disease. First, the histocompatibility anti- gen complex is located on the surfaces of cells. Thus, these antigens may react with the virus, or immune reagents combining with viral agents at the cell's surface. Histocompatible antigens may act as recep- tors for invading viruses or be structurally altered during viral infection. Theoretically, viruses that bud 3374 - 479 - BULL. WORLD HEALTH ORGAN., Vol. 52, 1975 M. B. A. OLDSTONE Table 1. Possible relationships between histocompa- tibility antigens and virus infections 1. Association of Histocompatibility Products with Disease 1. Histocompatibility antigens and viruses may share antigenic determinants. 2. Histocompatibility antigens may act as specific receptor sites for attachment of viruses. 3. Histocompatibility antigens may be altered by viruses. 4. Association of histocompatibility complex to immune response genes: Ir gene controls the ability of host to respond to virus antigens. 11. Association of Histoincompatible Reactions with Disease 1. Histoincompatible reactions may activate latent viruses. from the surfaces of cells may incorporate histocom- patibility antigens in their outer coat and, if altered and immunogenic, act to immunize the host against self. In addition, it has become clear that the regula- tion of immune responses to many antigens is under the control of genes that map near the major histocompatibility complex (1). Finally, there has been an increasing number of reports associating histocompatibility types with specific diseases of man and experimental animals (reviewed 1, 2). This report will focus on two major subjects (Table 1). These are, first, the association of histo- compatibility products or markers with diseases and, second, the association of histoincompatibility reac- tions with activation of latent viruses and disease. RELATIONSHIP BETWEEN HISTOCOMPATIBILITY ANTIGENS AND VIRUS INFECTIONS Histocompatibility antigens and viruses may share antigenic determinants In theory, when viruses bud from plasma mem- branes, part of the host's antigenic material may be assimilated in their outer membranes. At present there is no evidence to indicate that arenaviruses share antigenic determinants with the histocompati- bility complex. Using fluorescein isothiocyanate-con- jugated monospecific antiserum to lymphocytic cho- riomeningitis (LCM) virus in concert with rhoda- mine-conjugated antiserum to H-2 determinants, we stained surfaces of cells infected with LCM virus. Our preliminary results indicate that virus-specific cell surface antigens and histocompatibility antigens do not share common determinants. More compel- ling evidence that LCM virus does not share a common determinant with host histocompatibility antigens comes from observations that after non- lethal LCM virus infections in utero, at birth, or in adulthood, mice mount specific antiviral immune responses (3, 4). Such responses are unlikely to occur if virus and histocompatibility antigens have com- mon determinants. Histocompatibility antigens may act as specific recep- tor sites for virus attachment Recent reports associating HL-A antigens with multiple sclerosis and paralytic poliomyelitis have raised the possibility that histocompatibility antigens may act as specific receptor sites for the attachment of viruses. Morris & Pietsch studied 111 patients who developed paralytic poliomyelitis before the introduction of the Salk vaccine and compared them with 395 healthy blood donors (5). They found significant correlations between the occurrence of paralytic poliomyelitis and the presence of the HL- A 3 or HL-A 7 haplotype (p <0.02, p <0.002, respec- tively). Both the general incidence of multiple scle- rosis and the number of patients with multiple sclerosis and high titres of antibodies to measles virus correlated with HL-A 3 and HL-A 7 (6). Together these reports suggest that HL-A 3 and HL- A 7 could provide favourable receptor sites for the virus on the cell surface, resulting in infection of and damage to central nervous system cells (5). In both paralytic poliomyelitis and multiple sclerosis the association with HL-A seems to be related not to the initiation of viral infection but rather to an end stage of target injury. Several workers are actively studying whether virus-cell interactions are associated with HL-A as regards viral adsorption, penetration, expression, or production. Recent work from my laboratory done in collaboration with Luc Perrin and Barry Joseph has shown that individual 8866 human lymphoid cells that bear HL-A 3 and HL-A 7 determinants express more measles virus antigens at the cell surface and are more readily lysed by specific antimeasles virus antibody and complement than WIL-2 human lymphoid cells that do not bear HL-A 3 and HL-A 7 determinants (Table 2). Studies of measles virus infection in other human lymphoid cells bearing HL- A 3 and HL-A 7 markers will determine whether these findings are restricted to the two cells under study or represent a general phenomenon. Since the LD 7-a determinant was recently found to be more closely linked to disease than HL-A 7, these experi- 480 MAJOR HISTOCOMPATIBILITY ANTIGENS AND DISEASE Table 2. Measles virus and cells with various HL-A marl Parameters 24 h measured WIL-2 8866 V % cells lysed 24 43 No. antibody molecules bound per cell b 8x106 7x106 7 a WIL-2 HL-A 1,2,5; 8866 HL b 5 ,g of 1251 Fab specific anti ments should be repeate LD 7-a marker. Similar virus are needed. It will whether or not there is a c type and susceptibility or l If such a correlation is foi appropriately selected ce] expression, production, an ated injury may provide vw standing of how these vi infections. Several years ago we nc immune lysis in lymphoid LCM virus were inoculated, there was no significant kers a difference in the quantities of infectious virus pro- duced in vitro by individual SWR/J and C3H/St48 h 72 h cultures [SWR/J (H-2q) 9 embryos: mean ± SE: 4.1 VIL-2 8866 WIL-2 8866 ±0.5 x 105 PFU/ml of culture fluid compared with C3H/St (H-2k) 7 embryos: 2.8 ±0.5 x 105 PFU/ml of 49 91 45 99 culture fluid, P = 0.1]. Studies of infectious virus and viral antigen made in H-2q and H-2k mice over a one-cycle growth stage have not yet been done. Ixlo6 1.2x107 4x106 1.7X107 Nevertheless, while we found no significant differ- ences in the production of infectious virus 7 days after culture, we did observe that these same H-2q ibody added to 0.5 x 106 cells, embryos were significantly more susceptible to injury associated with LCM virus infection than were H-2k embryos. Table 3 shows that measuring the release d with cells bearing the of double-stranded DNA from individual embryo studies with poliomyelitis cultures as an indication of cellular injury, we found also be useful to know that all 9 SWR/J embryo cultures released signifi- ,orrelation between HL-A cantly more DNA as compared to sham-infected resistance to arenaviruses. cultures (SWR/J uninfected: mean ±SE: 1.5±0.4 ng and, careful studies using of DNA released/ml of culture fluid; SWR/J LCM[ls for viral absorption, infected: 9.2±0.8 ng/ml released, P > 0.001). In Ld immunologically medi- contrast, there was no significant release of DNA aluable clues to an under- from C3H/St mice infected with LCM virus over ruses function in human that seen with C3H/St sham-infected mice (C3H/St uninfected: 1.3±0.6 ng/ml, C3H/St LCM virus )ted that H-2q mice were infected: 1.8±0.8 ng/ml released). more susceptible to the manifestations of acute and chronic diseases associated with LCM virus infection than were H-2k mice (3, 7). Over the last 5 years, it has been possible to reproduce these observations, although the magnitude of difference initially ob- served has lessened. At present we find routinely that weanling or adult H-2q mice (SWR/J, C3H.q/SN and DBA/1J) die sooner and with a more intense inflammatory response after inoculation with a lethal dose of LCM virus than do H-2k mice (C3H/HeJ, C3H/St, C57 Br/cdJ). In general, adult H-2q mice die 6 to 9 days after a lethal dose ofLCM virus, whereas age-matched and sex-matched H-2k mice die between the 8th and 14th day. When the titres of various dilutions of virus are determined, there is usually a corresponding 2-3 log difference in susceptibility to virus from the 6th to the 9th day. At the 14th day after inoculation the difference in LD50 titres is in most instances no longer significant. Our inability to repeat the larger and longer-lasting quan- titative differences seen earlier most likely reflects changes in the new virus pools used or changes in the mice themselves (8). Raymond Welsh and I used individual mouse embryos and plaque formation assay to show that 7 days after uniform doses of Histocompatibility antigens may be altered by virus Ability of a virus to alter histocompatibility anti- gens was reported by Hecht & Summers (9). Study- ing the effects of vesicular stomatitis virus infection on L (H-2k) cells these investigators found that over 70% of the H-2 alloantigenic activity was lost upon infection. In contrast, infection with encephalomyo- carditis virus, a picornavirus, showed no loss of H-2 activity. Recently, Zinkernagel & Doherty reported that the in vitro killing of LCM-virus-infected targets through the mediation of T cells is restricted to the histocompatibility complex (10). These workers showed that sensitized T cells taken from mice inoculated 6-8 days previously with LCM virus killed syngeneic target cells infected with LCM virus significantly more effectively than they killed allo- geneic infected target cells. Using hybrids and vari- ous recombinants, these investigators extended their findings to show that T-cell-mediated cytotoxicity was linked closely to the H-2 complex and mapped, not in the Ir region, but either at the D or K ends (11).. Subsequently these workers and other col- leagues in Canberra have shown T-cell-mediated 481 M. B. A. OLDSTONE Table 3. Release of double-stranded DNA in association with LCM virus infection of individual H-2q and H-2k embryo cultures a nanograms of DNA released per ml of tissue culture fluid Embryo No. SWR/J (H-2q) C3H/St (H-2k) noninfected infected PFU noninfected infected PFU 1 0.2 7.2 5 x 105 0.0 0.2 4 x 105 2 1.4 7.3 6 x 105 0.0 0.2 1 x 105 3 0.3 7.4 4x 105 0.7 0.8 ND 4 1.0 7.0 4 x 105 1.7 5.9 2 x105 5 3.0 13.4 2 x 10 5 0.7 0.7 3 x 105 6 4.4 12.9 5 x 105 2.2 1.5 2 x 105 7 1.2 10.1 1 x 105 5.0 5.0 3 x105 8 1.0 9.8 4 x 105 9 1.1 7.4 4 x 105 mean 4 SD 1.5 ± 0.4 9.2 i 0.8 1.3 + 0.6 1.8 ± 0.8 P >0.001 NS a Individual embryos were cultured and on the second passage either sham treated or infected with LCM virus, Armstrong strain at a multiplicity of infection of 0.7. Supernatants were collected 7 days later, freed of any cell debris, and analysed for release of double-stranded DNA in tissue culture fluid by use of inhibition assay or for infectious virus by plaque assay. For comparison, a lytic virus, such as vesicular stomatitis virus, can release on the average 98 ng of DNA from similar cultures. P values determined by t test. Despite the increased susceptibility of SWR/J cells to LCM virus as compared with C3H/St cells, the amounts of virus produced by these cultures were not significantly different (P = 0.1, see text). PFU: plaque forming units per ml, ND: not done, NS: not significant. killing and H-2 restriction to another virus, ecto- melia, and to a bacteria, lysteria monocytogenes. In addition, Shearer, studying T-cell-mediated cyto- toxicity in trinitrophenol-(TNP)-modified target cells, showed that killing occurred only with TNP- modified targets and a syngeneic system (12). T cell cytotoxicity did not occur with TNP-modified allo- geneic targets. Thus, in at least 4 systems, T cell killing is restricted to the presence of a given histo- compatibility locus. Zinkernagel & Doherty have suggested that T cell killing is associated with altera- tion of the cell surface proteins by chemical or viral infection and that these proteins map in the region of the major histocompatibility complex. These find- ings should be extended and the active determinants (H-2 and viral) defined, quantitated, and mapped. Experiments to determine whether these observa- tions pertain to infections in humans are under active evaluation in several laboratories. It would be germane to know the relationship between arena- viral antigens on surface membranes and HL-A antigens and to establish whether arenavirus infec- tions alter a patient's HL-A determinants. Association of histocompatibility complex to immune response genes Reports of a variety of histocompatibility-linked, specific immune response (Ir) genes in several spe- cies (1) provoked investigations to determine whether anti-tissue and anti-viral immune responses might be under a similar genetic control. Basic work on histocompatibility antigens and Ir genes has emphasized three major points. First, histocompati- bility-linked Ir genes occur with native as well as synthetic antigens. Second, the dose effect of immu- nogen and H-2-linked responses correlate. Third, responses vary with immunogen and with strain, so a high responder to one antigenic challenge may be a low responder to a second or different immunogen. Working with the model of allergic encephalomye- litis (reviewed 2, 13), Williams and his group showed that immune responses in Lewis rats to encephalo- genic basic protein were significantly greater than those seen in BN rats. Similarly, Lewis rats devel- oped both histological and clinical disease while BN rats did not (13). Genetic backcrossing experiments 482 MAJOR HISTOCOMPATIBILITY ANTIGENS AND DISEASE indicated that there was a relationship between histocompatibility genotype and the occurrence of encephalomyelitis. Similar results were obtained independently by Gasser and his associates in the same model (14). Others investigating autoimmune thyroiditis in mice and chickens have shown a re- lationship of histocompatibility antigens to thyroid- itis (15). The degree of tissue injury accompanying infec- tion with LCM virus depends upon the dose of inoculated virus, the immune response mounted by the host against the virus, the interaction of struc- tural or coated viral antigens with products of the host's immune response, and the consequences of this interaction. Adult mice inoculated intercere- brally with a large dose of virus suffer a fatal acute disseminated necrotizing inflammatory disease. Immunosuppression induced by any of several meth- ods prevents acute disease in adult mice despite adequate viral replication. Reconstitution of im- mune responses results in disseminated necrotizing inflammatory disease. Experiments using H-2q, H- 2k mice, their hybrids, and the cross between the hybrid and both parental strains indicated that mice with an H-2q allele were significantly more suscep- tible to acute LCM virus infection than mice that excluded all H-2q alleles (7). In most of our experi- ments the hybrid was more susceptible than either parent. There was some variability among suscep- tible groups suggesting that other host factors were also participating. From these observations we infer- red that immune responses to LCM virus are in part genetically controlled and that the responsible genes map close to the major histocompatibility locus. These results are currently being re-evaluated using purified LCM virus, various recombinant mouse strains, and quantitative assessments of immune responses to the virus. ASSOCIATION OF HISTOINCOMPATIBLE REACTIONS WITH DISEASE Histoincompatible reactions may activate latent viruses In addition to the possible relationships between histocompatible antigen products and susceptibility to disease, other data strongly indicate that histo- incompatible reactions may lead to activation of latent viruses and disease. This work has centered around the study of cytomegalovirus and oncorna- virus in animal model systems and may have practi- cal consequences for the understanding of certain human disorders. Cytomegalovirus, a DNA virus, which causes a significant percentage of human disease, can persist in a latent state. This virus is activated in pregnant women, patients 1 to 2 months after receiving a transplanted kidney, and patients receiving multiple blood transfusions. Such frequent occurrences of cytomegalovirus infection suggest that this virus might be activated by means of immunologic reaction to foreign antigens. Lars Olding, Fred Jensen and I investigated this possibility by first establishing a model of latent cytomegalovirus infection in mice, second, assaying various tissues for viral expression or genome and, third, attempting to activate virus by in vitro allo- geneic reactions (16). Injection of several strains of *mice with murine cytomegalovirus in utero or at birth produced a model of latent cytomegalovirus infection. Hybridization studies done in collabora- tion with David Kingsbury on tissues from these mice clearly demonstrated the viral genome in less than 1 copy per cell, in cells from the spleen and salivary gland, but no detectable genome in cells from kidney, liver, brain, or thymus. We were unable to recover infectious virus from spleens of latently infected mice when these cells were co- cultured with syngeneic lymphocytes or fibroblasts. In contrast, co-culture with allogeneic cells led to the activation and recovery of virus in repeated experi- ments (16). After co-culture with allogeneic cells we almost always recovered virus from spleen cells but never from kidney, thymus, brain, or liver cells. Using a variety of immunological techniques, we determined that cytomegalovirus was harboured in B lymphocytes derived from the bone marrow and not in the T cell population derived from the thymus. Other experiments using disrupted cells and various mitogens showed that the virus resided in a latent stage in B lymphocytes and that active cell prolifera- tion was associated with virus activation. These observations suggest that similar activation results from histoincompatible reactions in vivo, and if the parallel from mice to man is complete, that the mechanism of cytomegalovirus activation in human disease is similar. Recent experiments in our labora- tory using both skin grafting and graft-versus-host reactions indicates that cytomegalovirus may be activated in vivo by allogeneic effects. Allogeneic activation of murine leukaemia virus in vivo and in vitro has been reported (17). In this model, as contrasted to the cytomegalovirus infection, oncor- navirus appears to be latent in T and not in B cells. 483 484 M. B. A. OLDSTONE Tests using allogeneic interactions may prove of value for the detection and isolation of other viruses. CONCLUSIONS In conclusion, diverse work in laboratories using different model systems of tissue injury and disease suggests possible relationships between disease and histocompatibility products on one hand and histo- incompatible reactions and activation of virus and disease on the other hand. Though much of this work is as yet incomplete, the approaches described here may well be of value in understanding related human diseases. ACKNOWLEDGEMENTS This is publication number 886 from the Department of Immunopathology, Scripps Clinic and Research Foundation, La Jolla, California. This research was supported by US Public Health Service grants AI-09484, AI-07007 and NS-12428; the Violet June Kertell Memorial Grant for research on Multiple Sclerosis from the National Multiple Sclerosis Society; and a grant from The National Foundation, No. 1-364. The author acknowledges the scientific collaboration of Drs Welsh, Olding, Joseph, Perrin, and Jensen in various parts of this work. R1SUMIm RELATION ENTRE LES PRINCIPAUX ANTIGENES D HISTOCOMPATIBILITE' ET LA MALADIE. ASSOCIATION POSSIBLE AVEC DES MALADIES HUMAINES A ARENAVIRUS Les notions d'antigenes d'histocompatibilite, d'infec- tions virales et de maladie sont examin6es sous l'angle des voies de la recherche sur les infections a arenavirus chez 1'homme. Les donnees faisant intervenir une rela- tion, che l'homme et les animaux, entre des complexes d'histocompatibilite et des maladies du systeme nerveux central sont passees en revue. Les antigenes d'histocom- patibilite ont peut-etre avec ceux des virus des d6ter- minants communs, ils peuvent agir comme sites recep- teurs pour la fixation des virus et etre modifies par ces derniers. En outre, les genes regissant les reponses immu- nitaires a des antigenes naturels et synth6tiques tres divers sont lies, chez bien des especes, au principal complexe d'histocompatibilite. Les lesions associ6es aux infections virales pouvant etre en grande partie dues a l'activit6 dy syst6me immunitaire, 1'etude des genes regissant la reponse immunitaire peut accroitre les connaissances rela- tives A la resistance A la maladie. En outre des reactions d'histo-incompatibilite sont susceptibles d'activer des virus latents et de provoquer ainsi la maladie. REFERENCES 1. BENACERRAF, B. & MCDEVITT, H. Histocompatibil- ity-linked immune response genes. Science, 175: 273- 279 (1972). 2. VLADUTIU, A. & ROSE, N. HL-A antigens: Associa- tion with disease. Immunogenetics, 1: 305-328 (1974). 3. OLDSTONE, M. & DIXON, F. Pathogenesis of chronic disease associated with persistent lymphocytic choriomeningitis viral infection. I. Relationship of antibody production to disease in neonatally infected mice. J. exp. Med., 129: 483-505 (1969). 4. OLDSTONE, M. & DiXON, F. Persistent lymphocytic choriomeningitis viral infection. III. Virus-anti viral antibody complexes and associated disease following transplacental infection. J. Immunol., 105: 829-837 (1970). 5. MoRRIs, P. & PIETSCH, M. A possible association between paralytic poliomyelitis and multiple sclero- sis. Lancet, 2: 847-848 (1973). 6. JERSILD, C. ET AL. Association between HL-A anti- gens and measles antibody in multiple sclerosis. Lancet, 1: 151-152 (1973). 7. OLDSTONE, M. ET AL. Histocompatibility-linked gen- etic control of disease susceptibility: Murine lympho- cytic choriomeningitis virus infection. J. exp. Med., 137: 1201-1212 (1973). 8. OLDSTONE, M. & DIXON, F. Changes in susceptibility of C3H/HeJ mice to LCM virus infection. J. Im- munol., 111: 1613-1615 (1973). 9. HECHT, T. & SUMMERS, D. Effect of vesicular stomati- tis virus infection on the histocompatibility antigen of L cells. J. Virol., 10: 578-585 (1972). 10. ZINKERNAGEL, R. & DOHERTY, P. Restriction of in vitro T cell-mediated cytotoxicity in lymphocytic choriomeningitis within a syngeneic or semi- allogeneic system. Nature (Lond.), 248: 701-702 (1974). MAJOR HISTOCOMPATIBILITY ANTIGENS AND DISEASE 485 11. BLANDEN, R. ET AL. Genes required for cytotoxicity against virus-infected target cells in K and D regions of H-2 complex. Nature (Lond.), 254: 269-270 (1975). 12. SHEARER, G. Cell-mediated cytotoxicity to trinitro- phenyl-modified syngeneic lymphocytes. Eur. J. Immunol., 4: 527-533 (1974). 13. WILLiAMS, R. & MooRE, M. Linkage of susceptibility to experimental allergic encephalomyelitis to the major histocompatibility locus in the rat. J. exp. Med., 138: 775-783 (1973). 14. GASSER, D. ET AL. Genetic control of susceptibility to experimental allergic encephalomyelitis in rats. Science, 181: 872-873 (1973). 15. VLADUT[U, A. & ROSE, N. Autoimmune murine thyroiditis. Relation to histocompatibility (H-2) type. Science, 174: 1137-1139 (1971). 16. OLDING, L. ET AL. Pathogenesis of cytomegalovirus infection. I. Activation of virus from bone marrow derived (B) lymphocytes by in vitro allogenic reaction. J. exp. Med., 141: 561-572 (1975). 17. HIRSCH, M. ET AL. Activation of leukemia viruses by graft versus host and mixed lymphocyte reactions in vitro. Proc. Nat. Acad. Sci. (Wash.), 69: 1069-1072 (1972). DISCUSSION LEHMANN-GRUBE: As Oldstone and Dixon have shown in earlier work, there are marked differences between inbred mouse strains as regards virus titres of carrier mice and as regards susceptibility to the lethal action of the virus. We were not able to confirm the differences between mouse strains, but our studies did not include the SWR/J strain. In titration experiments, we have now included C3H-He/J, SWR/J, and ordinary outbred albino mice, all of them matched according to age and sex. No differences at all were found in the LCM strain WE virus titres in these three mouse strains, which should have been on the extremes of the previously published range of inbred mouse strains as far as their susceptibility to the virus is concerned. In a second experiment, using LCM strain CA 1371, there was again no difference in susceptibility. I have no explanation for the discrepancies between our two laboratories. All I can say is, we have done numerous titrations of a similar kind and have never obtained evidence that differences exist between mouse strains as regards susceptibility to LCM virus. OLDSTONE: In the late 1960s we found sharp differences between strains. It concerns us greatly that in the middle 1970s we can no longer demonstrate these differences. I am unable to account for this, but it may be partly a host problem, since we can no longer produce carriers with C3H mice; when we inject them, they die. The only thing that we now find consistently is that the SWR/J mouse certainly dies more quickly and has a much greater inflammatory response than the C3H mouse, though at the end of 14-21 days, the virus titres are not significantly different. We have also found that our hybrid mice are distinctly more susceptible than either parent strain. MIMs: If I have understood you correctly, you are refer- ring to intracerebral adult susceptibility. In our Australian laboratories some years ago, we showed that the suscep- tibility of adult C57 black mice to intracerebral infection was basically different from that of the other mouse strains we used. We found the answer in terms of a dif- ference in the intensity of the cell-mediated immunity response in these two strains of mice. I am happy to accept that there are differences, although I have nothing to suggest as the explanation. Inevitably, discrepant results will prove to be partly due to virus strain, mouse strain, and technical differences. The strong genetic determinants of susceptibility in immune responses must undoubtedly make a difference, and I often wonder whether the C57 mice we use in our laboratories are exactly the same as the C57 mice used in somebody else's laboratory. WOODRUFF: Dr Oldstone's reference to immune complex disease in trypanosomiasis as well as in virus infections is of some interest in relation to work that we have been doing. We were interested particularly in the anaemia that occurs in African trypanosomiasis. Having found that this has a strong haemolytic component, we looked for a cause of the haemolysis and were able to show that there is complement fixed on the cell mem- brane of the red cells in these patients. It then occurred to us that perhaps the red cell membrane is no different from other membrane within the patient. Referring then to experimental models, we were able to demonstrate, using fluorescein-conjugated antimouse complement serum, that in mice infected with trypanosomiasis there is strong immunofluorescence in the glomeruli. WALKER: Dr Oldstone's last slide made mention of tissue injury in LCM infection. We have investigated this with the electron microscope, but have not been able to demonstrate, in the ependyma, choroid epithelium, or meninges, any sites of pathology. We did find infil- tration of mononuclear cells, probably T cells, which come out of capillaries, go through the adventitia, and through the choroid epithelium into the cerebrospinal fluid. What type of tissue injury did you find, Dr Old- stone ? M. B. A. OLDSTONE OLDSTONE: This was a finding reported by Theofilopolous & Dixon. I imagine that it probably referred to cellular infiltrate in the liver. MIMs: The previously point is quite an important one because, although cytotoxicity in vitro can be readily demonstrated, I am not sure that there is much evidence of the actual destruction of cells by cytotoxicity in the meninges of the adult infected mouse. As Doherty has shown, the major pathogenic event is inflammatory infil- tration, with the production of oedema in the subependy- mal and submeningeal areas detectable by injecting protein labelled with Evans blue. This is the nature of the immunopathological event; it is not cell destruction, but inflammatory infiltration and oedema in the limited space-the rigid box-of the skull. WALKER: We have also done experiments using a protein tracer, horseradish peroxidase, with a molecular weight of 40 000, administered by perfusion at the time of death in LCM-infected mice and we found that the tight junctions were completely intact. In these experiments we used the Armstrong strain. This is admittedly a dif- ferent strain from that used by Doherty, but there was no evidence of oedema, nor was there evidence of leakage of the horseradish peroxidase molecule. OLDSTONE: The choroid plexus leakages I referred to were leakages that were described in the persistently infected mouse and not in the acutely infected mouse. MONJAN: I can confirm the observations on lack of choroid plexus damage. In cerebellar disease in the rat, where there is massive destruction of cerebellar tissue, careful electron microscopic examination of the choroid plexus shows it to be completely normal except for infiltrates. NATHANSON: It seems fairly clear that the classical con- vulsive event represents disordered function of the nervous system. I think it is also fairly clear that we do not know exactly what the steps are that connect the infiltrates that everybody talks about with the final convulsion. OLDSTONE: Moreover, in order to build up immune com- plexes, the rate of degradation of those complexes must be exceeded by the rate of deposition. This only takes place over a prolonged period of time. HOTCHIN: While discussing what causes convulsive death, it is well to bear in mind that as early as 2 days, certainly 3 days, after inoculation of LCM you can precipitate typical convulsive death by giving Escherichia coli endo- toxin. We have not seen any pathological histology, but presumably there is cerebral oedema. 486