Imrnunology Today, voL 7, No. 11, 1986
How do T cellsmediateautoimmune thyroiditis? Susceptibility to experimental autoimmune thyroiditis (EAT) is linked to the mouse H-2 complex 7. Initiation of the condition by the thyroid antigen, thyroglobulin (Tg), requires T cells from susceptible mice2. These T cells are autoreactive in that they recognize Tg and proliferate in vitro in response to it 3. The 1980s have brought forth confirmatory data from several laboratories on T-cell autoreactivity to mouse Tg both in vivo4'5 and in vitro ~13, and its correlation with EAT susceptibility. T cells from patients with Hashimoto's thyroiditis (HT), may proliferate in response to stimulation in vitro with human Tg14-16. The capacity of murine T cells to expand and differentiate in vitro 17'18and to serve as effector cells causing thyroiditis in vivo/,8, 71,12is strongly indicative of their pathogenic role. We are now in a position to discuss not only if T cells mediate autoimmune thyroiditis, but how they might contribute to the pathogenic process. At least two antigen-activated T-cell subsets seem to participate in initiating and perpetuating thyroid destruction. In this article, Yi-chi Kong and her colleagues present a synopsis of the supporting data from the mouse model and parallel findings from HT patients. As a rule the helper T cells (Th) are driven by Tg to cooperate with B cells in production of autoantibodies, but no definite role in pathogenesis has been shown for Tg autoantibodies in HT 19 •20 or rat EAT21 . Four additional reasons why Th cells are not included as participants in thyroid destruction are deduced from the mouse model. (1) Comparable titers of Tg autoantibodies occur after immunization with Tg in complete Freund's adjuvant (CFA) in both EAT-susceptible and -resistant strains. Hence, MHC-linked susceptibility is differentiated on the basis of the extent of thyroid infiltration ~. (2) The immunoglobulin classes of Tg autoantibodies are indistinguishable between susceptible and resistant mice 22. (3) All susceptible mice, given repeated injections of syngeneic Tg without adjuvant, produce high titers of autoantibodies, but only up to 50% develop thyroid lesions 4. Thus, as seen in the human, high levels of antibodies, presumably of similar quality, are insufficient to initiate disease. (4) Activated immune T cells adoptively transfer thyroiditis in the absence of significant antibody production8,11,12 What cells other than Th are implicated in lesion development? Using I-I-2 recombinant strains of the same BI0 background, susceptibility to EAT, encoded by the putative Ir-Tg gene(s), has been mapped to the I-A subregion 23. Since only T cells from immunized susceptible, but not resistant, mice proliferate in response to mouse Tg stimulation in vitro 6'10'11, certain mouse Tg epitopes presented on la ÷ cells must be promoting the expansion of T cells other than Th. These epitopes include the ones shared with human Tg 1°. The proliferative responses are blocked by monoclonal antibodies (mAb) to I-A subregion products 1°'24 and L3T4 (Ref. 25), and are abrogated by eliminating Thy-1+, Lyt I + cells 6'I°.
Departmentof lrnrnunologyand Microbiology, WayneState University Schoolof Medicine, Detroit, M148201 USA; and Istituto Scientificodi Medicina Interna, Universityof Genoa, 16132Genoa,Italy ~) 1986, ElsevierSciencePublisllers BV., Amsterdam 0167 4919/86/502.00
Yi-chi M. Kong,MarcelloBagnascoand G. Walter Canonica These characteristics have gained importance because expansion in vitro markedly enhances the detectability of effector T cells for EAT7'81°-12'17'18. In addition, DR antigen is expressed in the thyrocytes of human patients with autoimmune thyroid disease 26,27 and similar inhibition of the patients' T-cell proliferation by mAb to DR antigens occurs 15. At present, two assays have been developed to measure effector T cells in murine EAT. One is to test their capacity to produce thyroiditis in normal recipients. Whereas the direct transfer of spleen cells or lymph node cells from immunized mice does not transfer the disease 12'28 the injection of T cells restimulated and expanded with Tg in vitro for several days to weeks results in severe thyroiditis in most recipient mice 8,~'~2. These adoptive transfer studies have permitted the following observations which are difficult to make from actively immunized individuals. (1)The initial effector T cells retained at the target organ are Tg-specific, and must lodge in sufficient numbers to begin the pathogenic process. Antigen-specific retention seems a reasonable premise; ten times the number of purified protein derivative-activated cells from mice immunized with CFA 12 or Tg-CFA (Simon and Kong, unpublished) do not transfer EAT. Although the number of localized Tg-specific effector T cells required to mediate severe EAT is not known, dose-response experiments show that more cells are required after restimulation with concanavalin A (Con A) than with Tg 12,28, whereas Tg-enriched T-cell lines require fewer cells 8. (2) Severity of thyroiditis increases with time after transfer for up to 21 days 7'8'11'12, suggesting continued expansion/differentiation of transferred cells. The chronic process also implicates two kinds of recruitment of host cell participation, the nonspecific kind by newly released lymphokines and the specific kind by Tg from disrupted follicles. (3) The autoreactive T-cell repertoire includes inducer T cells (Ti) recognizing shared epitopes on Tg from heterologous species 8 ' 12 , 73 . For example, although human Tg is inferior to mouse Tg as an immunogen for EAT induction, it can serve as an activator of effector T cells in vitro, whether the cells are derived from mouse or human Tg-immunized mice 12. In contrast, such autoreactive T cells cannot be detected in genetically resistant mice. Heterologous bovine and porcine Tg can also expand the mouse Tg-primed effector population in vitro for adoptive transfer of EAT29. These findings raise the interesting possibility that conserved regions of Tg are thyroiditogenic, in addition to mouse Tg-specific (nonconserved) epitopes. The other assay developed for effector T cells in EAT monitors the cytotoxic potential of lymph node cells from mouse Tg-immunized mice for thyroid target monolayers ~7. Only after stimulation in vitro with Tg for 5-6 days is cytotoxicity detectable. The cytotoxic T cells
337
Frostru
3311
(To) are Lyt 2 ÷ and recognize Tg on thyroid target cells in the context of class I (H-2 D,K) antigens. Their in-vitro generation and differentiation require the presence of antigen-activated Thy-1 ÷, Lyt 1÷ cells. In addition to mouse Tg 17, shared epitopes on human Tg 12, Con A 24, as well as thyroid cells24 also generate functional Lyt 1÷ T cells for the development of To. We reason that, during the culture period of 5-6 days, the Lyt 1÷ cells proliferate and serve as amplifier T cells (Ta), abetting the differentiation of Tc from their precursors, probably by providing lymphokines such as interleukin-2 (IL-2). These findings put forth evidence for the first time that Tc may play a major role in the pathogenesis of organ-specific autoimmune disease ~7. The existence and characteristics of Tc have recently been corroborated by others using mouse lymphocytes primed in vitro with thyroid epithelial cells (TEC) (Ref. 18). The relative contribution of lc to thyroid damage in vivo requires further study. Clearly, after culture for 3 days, the effector T-cell populations mediating the transfer of EAT contain precursors of Tc~7 which can undergo further differentiation in the host. Host T cells and Tc precursors can be recruited by the activated cells to perpetuate thyroid destruction, as mentioned above. Indeed, Lyt 2 + cells constitute a small percentage of Tg-enriched T-cell lines which transfer EAT8. We have examined the kinetics of T-cell subset accumulation in the thyroids of immunized mice developing EAT3°. There is a shift with time in the Lyt 1÷: Lyt 2 ÷ ratio from about 7:1 to 2:1. The early predominance of Lyt 1÷ cells is followed by a relative increase in Lyt 2 ÷ cells. Whereas the specificities and functions of these T-cell subsets are unknown, more than likely they contain the two antigenactivated T-cell subsets under discussion. The Lyt 2 + cells could include Tg-specific, class I antigen-restricted Tc. The Lyt 1+ cells could include Tg-specific, class II antigenrestricted T~ and perhaps T cells with functions resembling delayed type hypersensitivity (TD-like). The presence of TD in immunized mice has been detected by intradermal testing but it does not correlate with lesion development 9. Thus, although Thy-1 ÷ cells in the thyroid decline with time from about 50% to 28%, as more non T-non B cells (macrophages?) infiltrate to the thyroid 3°, it is not known if TD-like cells play a role by producing lymphokines. Three lines of studies in HT patients lend support to the role of T cells in pathogenesis when viewed in concert with evidence from the mouse model. First, DR antigens are expressed by TEC in diseased glands26,27; such TEC can present antigen to T-cell clones 3~. DR/la expression by normal TEC is inducible with interferon-3, in both humans 32 and mice33, and cultured normal TEC can prime T cells in vitro in mice 7'18. As discussed above, adoptive transfer studies show that activated T cells migrate to normal mouse thyroids in an antigen-specific manner to initiate the chronic pathogenic process. These activated effector T cells may be producing interferon--/, thereby inducing DR/la expression in situ and promoting additional stimulation of Ti in the host. The second line of studies on the in-vitro response of peripheral blood (PB) T cells to human Tg also infers a pathogenic role of T cells in HT. These T cells belong to a subset of Ti/T h bearing the 5/9 surface, marker34, and their proliferation is inhibited by mAb to DR antigens ~s. A second mAb B9.4 identifies a 5/9-T8 + subpopulation which has been shown to contain both Tc and their
Immunology Today, voL 7, No. 17, 1986
precursors 3s. Whereas the CD8 + cells alone do not proliferate in response to Tg, co-culture with CD5/9 + T cells results in a marked increase in CD8 ÷ cells 16. Interestingly, although PB T cells from Graves' disease (GD) patients also proliferate in response to Tg, the Tginfluenced CD8 + proliferation is observed only with HT T cells. As yet, we do not know the specificity of the CD8 ÷ cells in PB; T-cell clones displaying lectin-dependent cytotoxicity have been derived from this T-cell subset 36'37. It appears that, as in murine EAT, the 5•9 + T cells serve as Tg-specific T a in HT, providing the necessary signal for the expansion~differentiation of CD8 + cells with cytolytic capacity. The third line of studies supporting a T-cell role comes from examining the thyroid infiltrate for the distribution of T-cell subsets. Using different mAb to compare thyroidal T cells between HT and GD patients, others have reported either no differences in the inducer~helper to cytotoxic/suppressor ratios or higher ratios in HT thyroids 27. In contrast, we have observed that, in addition to 5/9 + (Ti/]-h) cells being present, about 50% of total T cells in HT thyroids are CD8 + (Ref.16 and 38), compared to only 19% in GD thyroids. Similarly high percentages of CD8 ÷ cells in HT thyroids have recently been confirmed 37. The difference in the proportion of thyroidal CD8 ÷ cells between HT and GD may reflect differences in pathogenic mechanisms, with antibodies playing a prominent role in GD. The high percentage of CD8 ÷ cells in HT thyroids is reminiscent of the increase of Lyt 2 ÷ cells (relative to Lyt 1÷ cells) in the thyroids of mice developing EAT3°. In both species, investigations on T-cell functions and antigen specificities have only just begun. As with PB T cells, high cloning efficiency has been observed with thyroidal T cells in HT; more than 60-70% of the clones exhibit lectin-dependent cytotoxicity and several of them also have natural killer activity36'37. The significance of these findings in relationship to thyroid destruction is unknown at present. Nevertheless, we (Bagnasco et al., unpublished) have recently found a higher frequency (65%) of interferon-3'producing cytolytic T-cell clones derived from HT thyroids, compared to those from PB (about 30%). These observations provide further support for the abovementioned-role of interferon-3, on TEC. The concept of clonal balance between suppressor T cells (Ts) and Ti/Th has previously been advanced to explain why genetically susceptible individuals are kept free of autoimmune disease, despite the constant presence of autoantigens 39. In murine EAT, the existence of Ts which can be activated with Tg has been documented 5. To override Ts influences, Tg can be administered with or without adjuvant with ensuing thyroiditis
How do T cellsmediateautoimmune thyroiditis? Susceptibility to experimental autoimmune thyroiditis (EAT) is linked to the mouse H-2 complex 7. Initiation of the condition by the thyroid antigen, thyroglobulin (Tg), requires T cells from susceptible mice2. These T cells are autoreactive in that they recognize Tg and proliferate in vitro in response to it 3. The 1980s have brought forth confirmatory data from several laboratories on T-cell autoreactivity to mouse Tg both in vivo4'5 and in vitro ~13, and its correlation with EAT susceptibility. T cells from patients with Hashimoto's thyroiditis (HT), may proliferate in response to stimulation in vitro with human Tg14-16. The capacity of murine T cells to expand and differentiate in vitro 17'18and to serve as effector cells causing thyroiditis in vivo/,8, 71,12is strongly indicative of their pathogenic role. We are now in a position to discuss not only if T cells mediate autoimmune thyroiditis, but how they might contribute to the pathogenic process. At least two antigen-activated T-cell subsets seem to participate in initiating and perpetuating thyroid destruction. In this article, Yi-chi Kong and her colleagues present a synopsis of the supporting data from the mouse model and parallel findings from HT patients. As a rule the helper T cells (Th) are driven by Tg to cooperate with B cells in production of autoantibodies, but no definite role in pathogenesis has been shown for Tg autoantibodies in HT 19 •20 or rat EAT21 . Four additional reasons why Th cells are not included as participants in thyroid destruction are deduced from the mouse model. (1) Comparable titers of Tg autoantibodies occur after immunization with Tg in complete Freund's adjuvant (CFA) in both EAT-susceptible and -resistant strains. Hence, MHC-linked susceptibility is differentiated on the basis of the extent of thyroid infiltration ~. (2) The immunoglobulin classes of Tg autoantibodies are indistinguishable between susceptible and resistant mice 22. (3) All susceptible mice, given repeated injections of syngeneic Tg without adjuvant, produce high titers of autoantibodies, but only up to 50% develop thyroid lesions 4. Thus, as seen in the human, high levels of antibodies, presumably of similar quality, are insufficient to initiate disease. (4) Activated immune T cells adoptively transfer thyroiditis in the absence of significant antibody production8,11,12 What cells other than Th are implicated in lesion development? Using I-I-2 recombinant strains of the same BI0 background, susceptibility to EAT, encoded by the putative Ir-Tg gene(s), has been mapped to the I-A subregion 23. Since only T cells from immunized susceptible, but not resistant, mice proliferate in response to mouse Tg stimulation in vitro 6'10'11, certain mouse Tg epitopes presented on la ÷ cells must be promoting the expansion of T cells other than Th. These epitopes include the ones shared with human Tg 1°. The proliferative responses are blocked by monoclonal antibodies (mAb) to I-A subregion products 1°'24 and L3T4 (Ref. 25), and are abrogated by eliminating Thy-1+, Lyt I + cells 6'I°.
Departmentof lrnrnunologyand Microbiology, WayneState University Schoolof Medicine, Detroit, M148201 USA; and Istituto Scientificodi Medicina Interna, Universityof Genoa, 16132Genoa,Italy ~) 1986, ElsevierSciencePublisllers BV., Amsterdam 0167 4919/86/502.00
Yi-chi M. Kong,MarcelloBagnascoand G. Walter Canonica These characteristics have gained importance because expansion in vitro markedly enhances the detectability of effector T cells for EAT7'81°-12'17'18. In addition, DR antigen is expressed in the thyrocytes of human patients with autoimmune thyroid disease 26,27 and similar inhibition of the patients' T-cell proliferation by mAb to DR antigens occurs 15. At present, two assays have been developed to measure effector T cells in murine EAT. One is to test their capacity to produce thyroiditis in normal recipients. Whereas the direct transfer of spleen cells or lymph node cells from immunized mice does not transfer the disease 12'28 the injection of T cells restimulated and expanded with Tg in vitro for several days to weeks results in severe thyroiditis in most recipient mice 8,~'~2. These adoptive transfer studies have permitted the following observations which are difficult to make from actively immunized individuals. (1)The initial effector T cells retained at the target organ are Tg-specific, and must lodge in sufficient numbers to begin the pathogenic process. Antigen-specific retention seems a reasonable premise; ten times the number of purified protein derivative-activated cells from mice immunized with CFA 12 or Tg-CFA (Simon and Kong, unpublished) do not transfer EAT. Although the number of localized Tg-specific effector T cells required to mediate severe EAT is not known, dose-response experiments show that more cells are required after restimulation with concanavalin A (Con A) than with Tg 12,28, whereas Tg-enriched T-cell lines require fewer cells 8. (2) Severity of thyroiditis increases with time after transfer for up to 21 days 7'8'11'12, suggesting continued expansion/differentiation of transferred cells. The chronic process also implicates two kinds of recruitment of host cell participation, the nonspecific kind by newly released lymphokines and the specific kind by Tg from disrupted follicles. (3) The autoreactive T-cell repertoire includes inducer T cells (Ti) recognizing shared epitopes on Tg from heterologous species 8 ' 12 , 73 . For example, although human Tg is inferior to mouse Tg as an immunogen for EAT induction, it can serve as an activator of effector T cells in vitro, whether the cells are derived from mouse or human Tg-immunized mice 12. In contrast, such autoreactive T cells cannot be detected in genetically resistant mice. Heterologous bovine and porcine Tg can also expand the mouse Tg-primed effector population in vitro for adoptive transfer of EAT29. These findings raise the interesting possibility that conserved regions of Tg are thyroiditogenic, in addition to mouse Tg-specific (nonconserved) epitopes. The other assay developed for effector T cells in EAT monitors the cytotoxic potential of lymph node cells from mouse Tg-immunized mice for thyroid target monolayers ~7. Only after stimulation in vitro with Tg for 5-6 days is cytotoxicity detectable. The cytotoxic T cells
337
Frostru
3311
(To) are Lyt 2 ÷ and recognize Tg on thyroid target cells in the context of class I (H-2 D,K) antigens. Their in-vitro generation and differentiation require the presence of antigen-activated Thy-1 ÷, Lyt 1÷ cells. In addition to mouse Tg 17, shared epitopes on human Tg 12, Con A 24, as well as thyroid cells24 also generate functional Lyt 1÷ T cells for the development of To. We reason that, during the culture period of 5-6 days, the Lyt 1÷ cells proliferate and serve as amplifier T cells (Ta), abetting the differentiation of Tc from their precursors, probably by providing lymphokines such as interleukin-2 (IL-2). These findings put forth evidence for the first time that Tc may play a major role in the pathogenesis of organ-specific autoimmune disease ~7. The existence and characteristics of Tc have recently been corroborated by others using mouse lymphocytes primed in vitro with thyroid epithelial cells (TEC) (Ref. 18). The relative contribution of lc to thyroid damage in vivo requires further study. Clearly, after culture for 3 days, the effector T-cell populations mediating the transfer of EAT contain precursors of Tc~7 which can undergo further differentiation in the host. Host T cells and Tc precursors can be recruited by the activated cells to perpetuate thyroid destruction, as mentioned above. Indeed, Lyt 2 + cells constitute a small percentage of Tg-enriched T-cell lines which transfer EAT8. We have examined the kinetics of T-cell subset accumulation in the thyroids of immunized mice developing EAT3°. There is a shift with time in the Lyt 1÷: Lyt 2 ÷ ratio from about 7:1 to 2:1. The early predominance of Lyt 1÷ cells is followed by a relative increase in Lyt 2 ÷ cells. Whereas the specificities and functions of these T-cell subsets are unknown, more than likely they contain the two antigenactivated T-cell subsets under discussion. The Lyt 2 + cells could include Tg-specific, class I antigen-restricted Tc. The Lyt 1+ cells could include Tg-specific, class II antigenrestricted T~ and perhaps T cells with functions resembling delayed type hypersensitivity (TD-like). The presence of TD in immunized mice has been detected by intradermal testing but it does not correlate with lesion development 9. Thus, although Thy-1 ÷ cells in the thyroid decline with time from about 50% to 28%, as more non T-non B cells (macrophages?) infiltrate to the thyroid 3°, it is not known if TD-like cells play a role by producing lymphokines. Three lines of studies in HT patients lend support to the role of T cells in pathogenesis when viewed in concert with evidence from the mouse model. First, DR antigens are expressed by TEC in diseased glands26,27; such TEC can present antigen to T-cell clones 3~. DR/la expression by normal TEC is inducible with interferon-3, in both humans 32 and mice33, and cultured normal TEC can prime T cells in vitro in mice 7'18. As discussed above, adoptive transfer studies show that activated T cells migrate to normal mouse thyroids in an antigen-specific manner to initiate the chronic pathogenic process. These activated effector T cells may be producing interferon--/, thereby inducing DR/la expression in situ and promoting additional stimulation of Ti in the host. The second line of studies on the in-vitro response of peripheral blood (PB) T cells to human Tg also infers a pathogenic role of T cells in HT. These T cells belong to a subset of Ti/T h bearing the 5/9 surface, marker34, and their proliferation is inhibited by mAb to DR antigens ~s. A second mAb B9.4 identifies a 5/9-T8 + subpopulation which has been shown to contain both Tc and their
Immunology Today, voL 7, No. 17, 1986
precursors 3s. Whereas the CD8 + cells alone do not proliferate in response to Tg, co-culture with CD5/9 + T cells results in a marked increase in CD8 ÷ cells 16. Interestingly, although PB T cells from Graves' disease (GD) patients also proliferate in response to Tg, the Tginfluenced CD8 + proliferation is observed only with HT T cells. As yet, we do not know the specificity of the CD8 ÷ cells in PB; T-cell clones displaying lectin-dependent cytotoxicity have been derived from this T-cell subset 36'37. It appears that, as in murine EAT, the 5•9 + T cells serve as Tg-specific T a in HT, providing the necessary signal for the expansion~differentiation of CD8 + cells with cytolytic capacity. The third line of studies supporting a T-cell role comes from examining the thyroid infiltrate for the distribution of T-cell subsets. Using different mAb to compare thyroidal T cells between HT and GD patients, others have reported either no differences in the inducer~helper to cytotoxic/suppressor ratios or higher ratios in HT thyroids 27. In contrast, we have observed that, in addition to 5/9 + (Ti/]-h) cells being present, about 50% of total T cells in HT thyroids are CD8 + (Ref.16 and 38), compared to only 19% in GD thyroids. Similarly high percentages of CD8 ÷ cells in HT thyroids have recently been confirmed 37. The difference in the proportion of thyroidal CD8 ÷ cells between HT and GD may reflect differences in pathogenic mechanisms, with antibodies playing a prominent role in GD. The high percentage of CD8 ÷ cells in HT thyroids is reminiscent of the increase of Lyt 2 ÷ cells (relative to Lyt 1÷ cells) in the thyroids of mice developing EAT3°. In both species, investigations on T-cell functions and antigen specificities have only just begun. As with PB T cells, high cloning efficiency has been observed with thyroidal T cells in HT; more than 60-70% of the clones exhibit lectin-dependent cytotoxicity and several of them also have natural killer activity36'37. The significance of these findings in relationship to thyroid destruction is unknown at present. Nevertheless, we (Bagnasco et al., unpublished) have recently found a higher frequency (65%) of interferon-3'producing cytolytic T-cell clones derived from HT thyroids, compared to those from PB (about 30%). These observations provide further support for the abovementioned-role of interferon-3, on TEC. The concept of clonal balance between suppressor T cells (Ts) and Ti/Th has previously been advanced to explain why genetically susceptible individuals are kept free of autoimmune disease, despite the constant presence of autoantigens 39. In murine EAT, the existence of Ts which can be activated with Tg has been documented 5. To override Ts influences, Tg can be administered with or without adjuvant with ensuing thyroiditis
