Environmentally induced autoimmune diseases: potential mechanisms.

Environmental and other xenobiotic agents can cause autoimmunity. Examples include drug-induced lupus, toxic oil syndrome, and contaminated l-tryptophan ingestion. Numerous mechanisms, based on (italic)in vitro(/italic) evidence and animal models, have been proposed to explain how xenobiotics induce or accelerate autoimmunity. The majority of these can be divided into three general categories. The first is those inhibiting the processes involved in establishing tolerance by deletion. Inhibiting deletion can result in the release of newly generated autoreactive cells into the periphery. The second mechanism is the modification of gene expression in the cells participating in the immune response, permitting lymphocytes to respond to signals normally insufficient to initiate a response or allowing the antigen-presenting cells to abnormally stimulate a response. Abnormal gene expression can thus disrupt tolerance maintained by suppression or anergy, permitting activation of autoreactive cells. The third is the modification of self-molecules such that they are recognized by the immune system as foreign. Examples illustrating these concepts are presented, and related mechanisms that have the potential to similarly affect the immune system are noted. Some mechanisms appear to be common to a variety of agents, and different mechanisms appear to produce similar diseases. However, evidence that any of these mechanisms are actually responsible for xenobiotic-induced human autoimmune disease is still largely lacking, and the potential for numerous and as yet unidentified mechanisms also exists.

ONE of the major problems facing tumour immunologists studying cellmediated reactions results from the disparity in the results with different in vitro cytotoxicity assays. Such disparity could be caused by variations in methodological parameters, which include the nature of the target cells and the length of the incubation period in vitro with the effector cells. Both the original visual microcytotoxicity assay (MA) (Takasugi & Klein, 1970) and several derived radioisotopic methods (Jagarlamoody et al., 1971;Cohen et al., 1972;Hashimoto & Sudo, 1971;Perlmann & Holm, 1969;Hashimoto et al., 1969;Bean et al., 1973) require long incubation periods and target cells in monolayers. Most of them reveal simultaneously cytostatic and cytolytic activities due to different effector cell populations, including T and non-T cells as it has been shown, for example, in the murine sarcoma virus (MSV) system (Lamon et al., 1973;Plata et al., 1974;Owen & Seeger, 1973). The cytostatic phenomenon mediated by non-T cells being, at least for the main part, non-specific (Owen & Seeger, 1973;Senik et al., 1974) it is often difficult to assess the specificity of the reactions measured by such methods. More antigen-specific reactions are generally obtained using short-term assays with ascitic lymphoma cells or tumour cells cultivated in suspension as targets. Under these conditions, very clear results can be obtained in the MSV system, regardless of the isotopic marker e.g., 51Cr , 125IUdR (Oldham & Herberman, 1973) or 3H-proline (Shiku et al., 1975;Oldham et al., 1977). These tests appear especially valuable in revealing the activity of cytolytic T lymphocytes (CTL). They are, however, much less efficient with target cells in monolayers. Relatively few in vitro-maintained tumour cells being available in suspension, only a limited number of tumour antigens can, therefore, be studied under optimal conditions. The availability of a specific cytotoxic assay using target cells in monolayers but revealing only cytolytic reactions would, therefore, constitute a major advance in tumour immunology.
The proline assay (PA) was initially proposed (Bean et al., 1973) because: (1) 3H-proline is retained by the target cells longer than 51Cr or 3H-thymidine; (2) it is less toxic than 125IUdR or [3H]-TdR; (3) when released by destroyed target cells the label is not re-utilized, because cold proline is present in large excess in the medium; (4) the results are not noticeably affected by proliferation or cytostasis of tumour cells during the incubation, so that only cytolysis is measured. This method appeared, therefore, as a candidate to replace other tests in the study of target cells in monolayers. It has notably been applied in the study of chemically induced sarcomas (Shiku et al., 1975) and the MSV tumour (Weiland & Mussgay, 1976). Here we report results of experiments using the latter system, aimed at determining the nature of effector cells, the level of H-2 restriction of target-cell cytolysis and the antigenic specificities involved. The results show on the one hand that the PA clearly reveals CTL-mediated reactions but, on the other hand, that the results obtained are not identical with those found using the Crrelease test (CRT). The major discrepancies concern the frequent detection of non-Tcell-mediated cytolysis, the lower precision in the quantification of the reactions, the less clear H-2 restriction of the CTL activity, and the unexpected lack of activity of some effector-cell populations. This suggests that some caution is necessary in the interpretation of antitumour reactions measured using PA, further emphasizing the difficulty in comparing antitumour cell-mediated reactions in different assays.
Viruses.-Tumours were induced in vivo, either by MSV-Moloney isolate, (maintained in vivo by regular acellular passages in newborn B6) or by the Friend leukaemia virus (FLV) (maintained in vivo in adult BALB/c). To infect cultured cells in vitro, the same agents or in vitro-produced viruses were used. The Moloney leukemia virus (MLV) was harvested from the supernatant of virusinfected non-Fv.1-restricted 3T3.FL lines, initially derived from NIH Swiss embryos. The Gross leukaemia virus (GLV) was similarly obtained from an in vitro-infected SCI line derived from wild mice, the Rauscher leukaemia virus (RLV) from a BALB/c 3T3 line in vitro transformed by a RLV pseudotype of MSV.
Cell lines.-Their main characteristics are summarized in Table I. Immune lymphocytes.-Anti-MSV immune lymphocytes were harvested from the spleens of adult mice inoculated 10-15 days before with 0-2 ml of a 10-1 dilution of the virus. Anti-FLV immune lymphocytes were obtained similarly from the spleen of adult B6, 10-20 days after an 0-1 ml i.p. inoculation of 1/5 diluted FLV. Spleen-cell pellets were incubated 20 sec in distilled water to eliminate red blood cells and the normal osmolarity was then adjusted by adding hypertonic NaCl. The cells were washed in medium and their concentration adjusted to the test density. Normal spleen cells from the same inbred strain of mice were used as controls.
Effector-cell purification.-(1) T cell elimination. Non-T cells were purified by eliminating Thy-1-2 cells from the whole-spleen cell suspensions using AKR anti-Thy-1-2 serum and rabbit complement. The preparation and specificity of the anti-Thy-1-2 serum, and the technical conditions used have been described previously (Leclerc et al., 1973).
The 3H-proline assay (PA).-PA was performed using a slightly modified version of the original method of Bean et al., (1973). Target-cell monolayers at 80% confluence in T-30 flasks were washed twice with minimum Eagle's medium (MEM) lacking non-essential amino acids (including proline) and incubated overnight in MEM lacking non-essential amino acids plus 15% FCS at 370C in a 5% C02 atmosphere, in the presence of 100 ,tCi 3H-proline (L-proline 3H-5, sp. act.: 22 Ci/mmol, CEA, Gif sur Yvette, France). The following day the culture was washed twice with complete MEM containing 15% FCS and 2% non-essential amino acids and incubated for 20-30 min at 37°0. The cells were detached with 0.05% trypsin, centrifuged 10 min at 800g, suspended in 1 ml MEM containing 10% FCS and 1% non-essential amino acids (test medium), and the cell concentration adjusted to 105 cells/ml. Ten ,ul containing 1000 target cells were distributed with a microlitre syringe into the wells of microtest II plates prefilled with 0-1 ml of test medium. Effector cells were added -4 h later in 0-1 ml of test medium, the lymphoid target cell (L/T) ratios varying from 25:1 to 300:1. Six to 8 replicates were used for each L/T ratio.
After 24-48 h incubation at 37°C under 5% C02, the plates were inverted, shaken slightly to remove medium, then submerged X3 in prewarmed PBS containing 10% FCS, and wiped with blotting paper. The cells remaining alive were harvested by adding 250 ,ul of 0.05% trypsin to each well, and transferred to scintillation vials. The arithmetic mean of 6-8 wells was calculated in order to estimate percentage relative inhibition of tumour cells after incubation with normal or immune lymphocytes, this inhibition being expressed as follows: inhibition= 100 x (mean ct/min after incubation i with normal lymphocytes J (mean ct/min after incubationi with immune lymphocytes J mean ct/min after incubation with normal lymphocytes The activity of the effector-cell suspension was also expressed in lytic units per 106 effector cells (LU) as previously reported (Plata et al., 1975): one LU represents the number of effector cells necessary to decrease by 50% the radioactivity of 1000 target cells. All statistical analyses were performed using Student's t test. The levels of significance were expressed as usual: NS=not significant, *=0.05>P>0.01, **=0.01>P>0 001, *** =P<0.001.
The 51Cr-release test (CRT) was performed as previously described  Transforming  14.
4z Eq co a) using as target cells the B6 Molon induced MBL2 lymphoma cells. Blocking of the cytolysis by antiser target cells in 10 ,ul of medium c( 10% FCS and 1% non-essential am were incubated for 2 h at 37°C in 5' the presence of 0 1 ml of normal inact immune serum diluted 1: 2 or mediu The sera were then removed using E pipette, and 3 x 105 effector cells were added to each well. After 24 h activity was calculated by compe cytolysis of target cells in the pr normal or immune serum.

Levels and specificity of the a' reactions detected in PA
The spleen cells of MSV re harvested at the beginning of rejection some 12-16 days aft inoculation, were always effectiv against MSV tumour cells, whe L/T ratios were used (Table I: decreasing ratios, lower levels of i activity were detected. When ali nti-MS V the same effector and target-cell preparations were tested in CRT, no activity was found (Table III). Nevertheless, the same effector-cell preparations were always highly efficient in CRT against MBL2 lymphoma cells in suspension. The maximum cytolytic activity detected against MSV tumour cells in monolayer (PA) or against the antigenically related MBL2 cells in suspension (ORT) were in the same range, but when different L/T ratios were used in both tests it appeared that, in terms of LU/106 effector cells, CRT was more sensitive than PA. However, the latter method was able to show cytolysis of MSV tumour cells, which CRT failed to detect.
A good level of cytolysis was also found gressors, in PA when normal mouse embryonic tum.our fibroblasts (MEF) infected in vitro with re in PA different transforming or non-transforming type C viruses, were used as target cells n 300: 1 (Table IV). MLV, FLV or RLV-infected L). With cells were regularly lysed, whereas normal cytolytic MEF or GLV-infected MEF were uniquots of affected. From these results it can be suggested that: (a) an "FMR-like" antigen f MSV_ could be involved in PA as in classical s detected CRT (Gomard et al., 1978) and (b)    1 See footnotes t, + and § of Table II. cytes and MSV tumour cells resulted in a strong secondary cytolytic activity against MBL2 target cells (results not given).

Nature of the effector cells in PA
The cytolytic activity being specifically abrogated by anti-Thy-1-2 and complement treatment of the effector cells, it appeared dependent on the presence of T lymphocytes (Table VI). Macrophages were apparently not concerned, since carbonyl iron and magnet treatments did not significantly decrease attacker-cell efficiency. The role of a non-phagocytic but plastic-adherent cell cannot be ruled out, since the activity of the whole-spleen cell suspension was clearly decreased by plastic adherence (Table VI) and still more by plastic adherence plus carbonyl iron treatment. It must be noted, however, that after such treatments the activity of the treated cells was always greater if measured in a 48-h rather than in a 24-h assay. This suggested that relatively time-consuming and aggressive manipulations could have non-specifically altered the effector cells, which then need more than 24 h to restore their normal functions. This hypothesis was reinforced by the fact that passing through nylon-wool columns, which takes only a relatively short time, did not decrease the cytolytic activity of the spleen-cell suspensions. We concluded, therefore, that the CTL were, at least for the most part, the effector cells of the anti-MSV reaction measured in PA. The role of cytolytic T lymphocytes (CTL) would be further supported by the existence of an H-2 restriction of cytolytic activity, this property being one of the major characteristics of CTL in the MSV (Gomard et al., 1976) as well as in many other systems (Doherty et al., 1976;Dennert 1976;Forman 1976;Shearer et al., 1976). The experiments reported in Table VII showed that such an H-2 restriction can be found in PA, allogeneic MSV tumour cells being lysed significantly but at a 2-3-times lower level than syngeneic targets. However, we have never found in PA, the very strong H-2 restriction which is regularly detected in CRT with lymphoma target cells (Gomard et al., 1978).
The involvement of H-2 normal antigens in the effector-target-cell interaction is clearly confirmed by the observation that preincubation of H-2b tumour cells with anti-H-2b antibodies specifically abrogated their sensitivity to syngeneic anti-MSV effector cells in PA (Table VIII) as previously shown in the CRT (Gomard et al., 1977).

Detection of natural killer (NK) cells in PA
The cytolytic activities of non-immune spleen cells were measured in PA by comparing cytolysis in the presence of normal lymphoid cells and in medium alone. The results in Table IX show that in PA normal murine spleen cells had a strong killer activity for: (a) normal mouse fibroblasts; (b) type C virus-infected murine cells whether transformed or untransformed; (c) normal xenogeneic cells. This NK-cell activity was not H-2 restricted, and did not depend on a viral antigen, since it was also found with nonvirus-infected cells. It was not dependent on tumour antigen(s) since normal cells were also affected. It appeared, therefore, as mainly non-specific and predominantly determined by the general sensitivity of the target cells to immune cytolysis, a   I See footnotes t, + and § of Table II. clear parallelism existing, for example, between the sensitivity to NK cells and the sensitivity to related anti-H-2 lymphocytes. Table X shows that the effector cells were non-phagocytic and non-T. It is important to emphasize that aliquots of the same effector-cell populations were 63.8*** 2250±293 1432±324 36-4*** 4 9 NS 1842±474 1810±344 1-7 NS always devoid of cytolytic activities when tested in CRT against ascitic tumour cells (results not given).
specific cell-mediated cytolytic reactions against MSV tumour cells. The cytolysis induced by immune lymphocytes is predominantly, if not exclusively, due to T cells. The possible involvement of a minor population of plastic-adherent cells does not change this conclusion, since its activity was abrogated by an anti-Thy 1-2 and complement treatment, suggesting that it too may be a T-cell subpopulation.
The cytolytic activity appears very specific and probably directed against the same "FMR-like" antigen which is recognized by anti-MSV CTL in CRT (Gomard et al., 1978). Despite the use of tumour target cells in monolayers and of relatively long in vitro incubations, PA gives much more specific results than MA. The fact that PA does not measure cytostatic phenomena is probably determinant (Seeger et al., 1974). Moreover, according to the recent results of Brooks et al. (1978) we may suppose that the use of labelled amino acids in place of radioactive nucleotides explains its better specificity than that of other isotopic MA. PA allows one to detect the cytolysis of MSV tumour cells themselves, whereas CRT fails to do so, or provides only very weak and hardly reproducible results. This advantage of PA contrasts with its lower sensitivity than classical CRT using lymphoma cells as target. It is probably related to the longer in vitro incubations which are possible with PA, allowing lysis of relatively insensitive tumour cells. Such incubations are hardly possible in CRT due to the high level of spontaneous marker elution, except when specially selected tumour cells are used. Moreover, the cellular lesion which is necessary to allow the detachment of altered cells from the plastic could be an earlier step in cell death than the release of 5lCr-labelled large molecules. PA appears, therefore, as a useful test in tumour immunology when tumour-cell populations growing in monolayer have to be used. Nevertheless, it must be emphasized that in the MSV system, PA and CRT, which both reveal a CTL-mediated reaction directed against an "FMR-like" antigen, do not provide identical results. At least 3 major differences have been detected in our experiments: (1) A natural killer activity was regularly found in PA but not in CRT. It has been detected, however, by others with the latter method (Bean et al., 1973;Oldham et al., 1977). The discrepancy is probably related to the variable sensitivity of tumour cells to NK-cell-mediated cytolysis, with an especially high sensitivity of in vitro tumour cells (Sendo et al., 1975). The fact that PA measures the cells unsticking could account for its specially high ability to reveal natural killing since it is well established that non-T-cells are frequently responsible for such phenomena, independent of target-cell cytolysis (Golstein, 1970). The important point is that the NK-cell-mediated reactions could mask, in PA experiments much more than in CRT, weak CTL-mediated cytolysis.
(2) Anti FLV-effector cells were much less efficient in PA than anti-MSV, whereas both kinds of CTL behave alike in CRT. The reasons for this surprising phenomenon is unclear, and the role of different cellsurface antigens appears unlikely, as discussed above. An explanation may perhaps be found in the high degree of adhesiveness of anti-FLV-CTL. It is known that these cells can be retained by nylon-wool columns  and we observed that they were much nmore adherent to target-cell monolayers than anti-MSV-CTL. Whatever its origin, this phenomenon could alter the CTL-tumour-cell interaction and it could be responsible for false-negative reactions suggesting an incorrect pattern of antigen specificity.
(3) The H-2 restriction of CTL reactivity was far weaker in PA than in CRT. This could be related to quantitation problems, since when strongly efficient CTL are tested the H-2 restriction can only be established by quantitative experiments using several L/T ratios (Gomard et al., 1978). Such experiments are hardly possible in PA since the cytotoxic activity falls very rapidly with decreasing L/T ratios. The NK-cell activity is also increased by prolonged in vitro incubations. This NK activity, which increases the background cytolysis, can in turn mask some of the specific cytolysis, especially at low L/T ratios when syngeneic but not allogeneic effector cells should be efficient at a relatively weak level. PA, therefore, appears a priori to be a bad method for revealing H-2 restriction phenomena, and this may explain the weakness of the H-2 restriction of viral mammary-tumour cell cytolysis previously reported (Stutman, 1977). It is remarkable, in view of the above considerations, that it was nevertheless possible to establish that H-2 antigens were involved in PA reactivity, as in the CRT, as demonstrated by the strong blocking activity of anti-H-2 antibodies.
In conclusion, PA can be a useful method of testing CTL-mediated antitumour or anti-viral immune reactions, but several peculiarities of the test should be borne in mind in order to avoid misinterpretation of the results. Once again, it appears that the technical parameters are especially determinant in the detection of cellular anti-tumour reactions, as recently emphasized by other investiga-tions in different systems (Chou-Chik Ting et al., 1977a, b;Oldham et al., 1977;Brooks et al., 1978).