Functional CD8؉ but not CD4؉ responses develop independent of thymic epithelial MHC

Marianne M. Martinic*†‡, Maries F. van den Broek*, Thomas Ru¨ licke§¶, Christoph Huberʈ, Bernhard Odermatt**, Walter Reith††, Edit Horvath*, Raphael Zellweger*, Katja Fink*, Mike Recher*, Bruno Eschli*, Hans Hengartner*, and Rolf M. Zinkernagel*‡

*Institute of Experimental Immunology, §Institute of Laboratory Animal Science, and **Institute of Clinical Pathology, University Hospital Zurich, CH-8091 Zurich, Switzerland; ʈDepartment of Immunology, The Scripps Research Institute, La Jolla, CA 92037; and ††Department of Pathology and Immunology, University of Geneva Medical School, CH-1211 Geneva 4, Switzerland

Contributed by Rolf M. Zinkernagel, August 7, 2006 The role of nonthymic epithelial (non-TE) MHC in T cell repertoire shown by the presence of high titers of CD4ϩ T cell-dependent selection remains controversial. To analyze the relative roles of VSV-specific neutralizing comparable to those present thymic epithelial (TE) and non-TE MHC in T cell repertoire selection, in control mice. However, concerns emerged that T cells restricted we have generated tetraparental aggregation chimeras (B6- to nude MHC in these chimeras might have had a survival advan- nude7BALB͞c and B67BALB͞c-nude) harboring T and B cells from tage in the periphery over TE MHC-restricted T cells. It is well both parents, whereas TE cells originated exclusively from the known that after leaving the mature T cells are able to non-nude donor. These chimeras mounted protective virus-specific survive only in the periphery when their TCR remain in continuous TE and non-TE MHC-restricted T cell responses. To further evaluate interaction with TCR-restricting MHC molecules (20–22). In the whether non-TE MHC alone was sufficient to generate a functional periphery, mature T cells will primarily be in contact with their T cell repertoire, we generated tetraparental aggregation chimeras surrounding T and B cells, which in the RagϪ/Ϫb7nuded chimeras ؊ ؊ lacking MHC class II (B6-nude7MHCII / ) or both MHC molecules exclusively express nude MHC. Even if the nude MHC-restricted T ؊ ؊ ؊ ؊ (B6-nude7MHCI / II / ) on TE cells, but not on cells of B6-nude cells after leaving the thymus were initially a minor population, in origin. Chimeras with MHC-deficient TE cells mounted functional the periphery these T cells might have a survival advantage over the ؉ ؉ virus-specific CD8 but not CD4 T cell responses. Thus, maturation initially more numerous TE MHC-restricted T cells. ؉ of functional CD4 T cell responses required MHC class II on Therefore, to ensure that neither TE MHC- nor nude MHC- ؉ thymic epithelium, whereas CD8 T cells matured in the absence of restricted T cells have a survival advantage in the periphery, we TE MHC. have now generated B6-nude7BALB͞c and B67BALB͞c-nude tetraparental aggregation chimeras. In these chimeras, TE cells are T cell selection ͉ tetraparental aggregation chimeras exclusively of BALB͞c or B6 origin, respectively. T and B lympho- cytes and professional -presenting cells (APC), however,

he thymus is the organ responsible for the differentiation of originate from both donors. We infected this set of chimeras and IMMUNOLOGY Tprogenitor T cells into mature T cells expressing self-MHC- control mice with LCMV and analyzed MHC restriction and restricted and self-tolerant T cell receptors (TCR) (reviewed in refs. function of both CD4ϩ and CD8ϩ T cells. Our results also show that 1 and 2). It is well established that thymic epithelial (TE) cells play T cells from B6-nude7BALB͞c and B67BALB͞c-nude chimeras a very important role in T cell repertoire selection (ref. 3; reviewed were restricted to TE and non-TE MHC. CD4ϩ and CD8ϩ T cells in refs. 4 and 5). However, the role for non-TE cells in T cell were both functional, and chimeras showed protective immune repertoire selection is uncertain. Most of the studies investigating responses comparable to those from control mice. non-TE MHC-associated T cell repertoire selection came to the As already mentioned above, TE cells are involved in the conclusion that non-TE MHC is not efficient in selecting a func- selection of a functional T cell repertoire. However, it is unclear tional T cell repertoire (ref. 6; reviewed in refs. 4 and 7). The few whether the presence of MHC molecules on these TE cells is studies showing involvement of non-TE MHC in T cell repertoire required for T cell repertoire selection or whether MHC molecules selection were not accepted for several reasons: because non-TE expressed by non-TE cells are sufficient to select a functional T cell MHC-restricted T cell responses were weak and required several repertoire. Previous studies with bone marrow chimeras had shown rounds of restimulation, because the used were not defined that non-TE MHC inefficiently selects a CD8ϩ T cell repertoire (12, at the peptide level [(Tyr,Glu)-poly(DL-Ala)-poly(Lys)-polymer], 23). Furthermore, experiments using bone marrow chimeras with or because the results were solely from in vitro experiments (8–17). hosts harboring MHC class II-deficient TE cells showed that CD4ϩ Although studies with nude mice reconstituted with a fully alloge- T cells were not positively selected (refs. 24 and 25; reviewed in ref. neic thymus graft showed almost exclusive restriction to nude MHC 26). In contrast, in C2ta pIVϪ/Ϫ mice, which lack MHC class II alone, these studies were also not accepted because of suspected rescue of nude thymic rudiment (18). More recently, however, the availability of new tools and mice has allowed the controversial issue Author contributions: M.M.M. and M.F.v.d.B. contributed equally to this work; M.M.M. and of the role of non-TE versus TE MHC in T cell repertoire selection R.M.Z. designed research; M.M.M., M.F.v.d.B., T.R., C.H., B.O., E.H., R.Z., K.F., M.R., and B.E. performed research; W.R. contributed new reagents͞analytic tools; M.M.M., M.F.v.d.B., to be readdressed. We had generated tetraparental aggregation H.H., and R.M.Z. analyzed data; and M.M.M., M.F.v.d.B., and R.M.Z. wrote the paper. chimeras from thymus-deficient nude mice and T and - The authors declare no conflict of interest. deficient RagϪ/Ϫ mice (19). In these RagϪ/Ϫb7nuded tetraparental Ϫ/Ϫ b Abbreviations: LCMV, lymphocytic choriomeningitis virus; TE, thymic epithelial; TCR, T cell aggregation chimeras TE cells were exclusively of Rag H-2 ; VSV, vesicular stomatitis virus; APC, antigen-presenting cell; VSV-IND, VSV-Indiana. d origin; however, T and B cells were exclusively of nude H-2 origin. †Present address: Division of Developmental Immunology, La Jolla Institute for Allergy and After infection of these chimeras with lymphocytic choriomenin- Immunology, 9420 Athena Circle, La Jolla, CA 92037. gitis virus (LCMV) or vesicular stomatitis virus (VSV), the MHC ‡To whom correspondence may be addressed. E-mail: [email protected] or rolf.zinkernagel@ restriction and functionality of T cells were analyzed. Protective usz.ch. ϩ LCMV-specific CD8 T cell responses restricted to TE and non-TE ¶Present address: Institute of Laboratory Animal Science, University of Veterinary Medicine ϩ MHC were detected in all chimeras tested. In addition, CD4 T Vienna, A-1210 Vienna, Austria. cells restricted to non-TE MHC were present and functional as © 2006 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0606707103 PNAS ͉ September 26, 2006 ͉ vol. 103 ͉ no. 39 ͉ 14435–14440 Downloaded by guest on October 1, 2021 expression on TE cells but not on professional APC and which have CD8ϩ T cells from chimeric but not from control mice showed two MHC class II-sufficient non-TE cells in the thymus, low levels of distinct populations, one binding to gp33͞Db and the other binding CD4ϩ T cell-dependent NP-CGG-specific IgG were detected (27). to np118͞Ld tetramer (Fig. 1A), indicating that chimeric CD8ϩ T We analyzed whether, in a thymus with MHC-deficient TE cells but cells were restricted to both TE and non-TE MHC. Similar results MHC-sufficient non-TE cells present from birth, the selected were obtained by analyzing splenic CD8ϩ T cells (data not shown). CD4ϩ and CD8ϩ T cells were able to mount functional virus- In parallel, splenocytes from day-8 immune chimeric and specific T cell responses. To analyze these questions, we generated control mice were restimulated for6hwithLCMV-gp33–41 ϩ a second set of tetraparental aggregation chimeras and used C2ta (gp33) or LCMV-np118–126 (np118) peptide, and CD8 T cells pIVϩ/Ϫ and C2ta pIVϪ/Ϫ mice as control mice. We first generated were analyzed for IFN-␥ secretion (Fig. 1B). Again, CD8ϩ T cells B6-nude7MHCIϪ/ϪIIϪ/Ϫ tetraparental aggregation chimeras. In from chimeric but not from control mice produced IFN-␥ after these chimeras, TE cells are exclusively of MHCIϪ/ϪIIϪ/Ϫ origin restimulation with either peptide, indicating that both TE and and therefore lack MHC class I and II molecules. All other cells non-TE MHC-restricted CD8ϩ chimeric T cells were functional including professional APC originate from both donors, but only (Fig. 1B). In addition, chimeric and control splenocytes from those from nude origin will express MHC molecules. Next we day-13 immune mice were restimulated for 5 days with LCMV generated B6-nude7MHCIIϪ/Ϫ tetraparental aggregation chime- peptide-labeled irradiated splenocytes and tested for their ability ras. In these chimeras, TE cells originate exclusively from to lyse LCMV peptide-labeled target cells in a 5-h 51Cr-release MHCIIϪ/Ϫ mice and, therefore, lack MHC class II but not MHC assay (Fig. 1C). Splenocytes from chimeras lysed peptide-labeled class I molecules. Again, all other cells originate from both donors; target cells of both haplotypes, confirming the TE and non-TE therefore, professional APC will express MHC class II molecules MHC-restricted specificity of CD8ϩ chimeric T cells in an only when they are derived from nude origin. We infected these additional functional assay (Fig. 1C). Finally, in all chimeras chimeras and control mice with LCMV and VSV and analyzed the tested LCMV was cleared by day 13 from blood, spleen, kidney, function of both CD4ϩ and CD8ϩ T cells. CD8ϩ T cells from liver, and lung as efficiently as in control mice (data not shown). B6-nude7MHCIϪ/ϪIIϪ/Ϫ chimeras but not from MHCIϪ/ϪIIϪ/Ϫ In conclusion, even when from both parents were controls showed virus-specific immune responses. Virus-specific present, chimeric CD8ϩ T cells were restricted to TE and non-TE CD4ϩ T cells, however, could not be detected in mice expressing MHC, and, furthermore, they were functional and protective. MHC class II-deficient TE cells. However, it is noteworthy to mention that, in all chimeras tested, T cell responses restricted to TE MHC were stronger than the Results ones restricted to non-TE MHC (Fig. 1 A–C). This outcome is Functional CD8؉ T Cell Responses in B6-Nude7BALB͞c and B67BALB͞ probably because of the fact that TE MHC-restricted T cells c-Nude Tetraparental Aggregation Chimeras Are Restricted to TE and were more numerous because of their survival advantage in Non-TE MHC. We started with the generation of B6-nude7BALB͞c the thymus due to higher cell numbers expressing TE than and B67BALB͞c-nude tetraparental aggregation chimeras. In non-TE MHC. these chimeras TE cells were exclusively of BALB͞c or B6 origin, respectively; however, all other cells, including lymphocytes and Chimeric CD4؉ T Cells Are Restricted to TE and Non-TE MHC and professional APC, were derived from both donors. We infected Provide Help to B Cells. We then analyzed MHC restriction and both chimeras and control mice i.v. with 200 pfu of LCMV strain functionality of CD4ϩ T cell responses. For this purpose, chimeras WE (LCMV-WE). Seven days after infection, blood cells from and control mice were infected i.v. with 200 pfu of LCMV-WE. B6-nude7BALB͞c and B67BALB͞c-nude chimeras were ana- Because LCMV peptides binding to H-2IAd MHC class II mole- lyzed for chimerism by staining T cells, B cells, and macrophages for cules have not yet been identified, we restimulated chimeric and b d the expression of B6 H-2 and BALB͞cH-2 MHC class I mole- control splenocytes 8 days after infection with the LCMV-gp64–80 cules (Fig. 4A, which is published as supporting information on the peptide (gp64) presented on H-2IAb MHC class II molecules. Six PNAS web site). All cell types analyzed always showed two distinct hours after restimulation, CD4ϩ T cells were analyzed for IFN-␥ populations, one expressing H-2b and the other expressing H-2d secretion (Fig. 1D). As expected, B6 and B67BALB͞c-nude but MHC class I molecules, confirming that indeed all lymphocytes, not BALB͞c CD4ϩ T cells secreted IFN-␥ after restimulation with macrophages (Fig. 4A), and dendritic cells (data not shown) gp64 (Fig. 1D). Interestingly, B6-nude7BALB͞c CD4ϩ T cells also originated from both donors. Similar results were obtained by secreted IFN-␥ after restimulation with gp64, which is solely analyzing chimeras before LCMV infection and analyzing different presented by non-TE MHC, showing that, similar to CD8ϩ T cells, tissues such as thymus, spleen, and mesenteric and inguinal lymph chimeric CD4ϩ T cells were also restricted to TE and non-TE MHC nodes (data not shown). These data indicated that, in the periphery (Fig. 1D). However, as was already shown for CD8ϩ T cells, CD4ϩ of these chimeras, as opposed to the previous RagϪ/Ϫb7nuded T cells showed stronger TE than non-TE MHC-restricted responses chimeras, neither non-TE MHC-restricted nor TE MHC-restricted (Fig. 1D). T cells had a survival advantage. To confirm that TE cells in It has previously been reported that LCMV-specific IgG is strictly B6-nude7BALB͞c and B67BALB͞c-nude chimeras were exclu- dependent on cognate CD4ϩ T cell help (28, 29). We therefore sively of BALB͞c or B6 origin, respectively, we stained thymic examined LCMV-GP- and LCMV-NP-specific IgG titers 20 days sections for expression of cytokeratin (universal marker for epi- after i.v. infection with 200 pfu of LCMV-WE. Chimeric and thelial cells) and B6 H-2IAb or BALB͞cH-2IAd MHC class II control mice showed similar concentrations of LCMV-GP-specific molecules using fluorescence microscopy (Fig. 4B). In B67BALB͞ (Ͼ1:30,000 titer dilution) and LCMV-NP-specific (Ͼ1:600 titer c-nude chimeras, B6 H-2IAb but not BALB͞c H-2IAd MHC class dilution) IgG, confirming that CD4ϩ T cells from LCMV-immune II staining colocalized with cytokeratin staining, which also coin- chimeras were able to provide cognate help to B cells (data not cided with the TE network, confirming that TE cells were exclu- shown). sively of B6 origin (Fig. 4B, orange staining). H-2IAd MHC class In conclusion, even when lymphocytes from both parents were IIϩcytokeratinϪ cells were also present (Fig. 4B, bright green-yellow present, non-TE MHC-restricted CD4ϩ T cells were positively staining), confirming the existence of professional APC of BALB͞ selected, and these CD4ϩ T cells were functional. c-nude origin in the thymus; corresponding results were obtained with B6-nude7BALB͞c chimeras (Fig. 4B). Mice Lacking MHC Expression on TE Cells Still Have Functional CD8؉ To assess MHC restriction and the function of CD8ϩ T cell T Cell Responses. To analyze whether non-TE MHC alone was responses in these chimeras, 8 days after infection blood cells were sufficient in selecting a functional T cell repertoire or whether and stained with anti-CD8 and LCMV-specific tetramers (Fig. 1A). which MHC molecules were needed on TE cells for proper T cell

14436 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0606707103 Martinic et al. Downloaded by guest on October 1, 2021 Fig. 1. CD8ϩ and CD4ϩ T cells from chimeras are restricted to TE- and non-TE MHC and are functional. Chimeras and control mice were infected i.v. with 200 pfu of LCMV-WE. (A) Eight days after infection, blood cells were gated on CD8ϩ T cells and stained for LCMV-specific gp33͞Db and np118͞Ld tetramer binding. Numbers in the lower right and upper left quadrants represent the percentage of CD8ϩ T cells binding to LCMV- specific gp33͞Db and np118͞Ld tetramer, respectively. Data are representative of four independent experiments and show one representative mouse per group (total, n ϭ 5–6 mice per group). (B) Eight days after infection, splenocytes were restimulated in vitro for 6 h with LCMV-gp33–41 peptide (gp33) (Left) or LCMV-np118–126 peptide (np118) (Right) followed by surface staining for CD8 and intracellular staining for IFN-␥. Numbers in dot plots indicate the percentage of CD8ϩ T cells secreting IFN-␥. Data are representative of four independent experiments and show one representative mouse per group (total, n ϭ 5–6 mice per group). (C) Thirteen days after infec- tion, splenocytes were restimulated in vitro for 5 days with LCMV peptide-labeled spleen cells and assayed in a 5-h 51Cr- release assay for lysis of peptide-loaded EL-4 (H-2b) and P815 (H-2d) target cells. Curves represent individual mice. Data are representative of three independent experiments containing two to four mice per group. gp33, gp(33–41); np118, np(118– 126). (D) Eight days after infection, splenocytes were restim- ulated in vitro for 6 h with LCMV-gp64–80 peptide (gp64) followed by surface staining for CD4 and intracellular staining for IFN-␥. Numbers beside circles indicate the percentage of CD4ϩ T cells secreting IFN-␥. Data are representative of four independent experiments and show one representative mouse per group (total, n ϭ 4–5 mice per group). IMMUNOLOGY

repertoire selection, we generated tetraparental aggregation MHCIϪ/ϪIIϪ/Ϫ mice (Fig. 2B Bottom) produced IFN-␥ after re- chimeras from thymus-deficient B6-nude mice and MHC class I stimulation with either peptide, indicating that CD8ϩ T cells from and class II double-deficient mice. In the resulting B6- chimeras but not from MHCIϪ/ϪIIϪ/Ϫ mice were functional. Fur- nude7MHCIϪ/ϪIIϪ/Ϫ tetraparental aggregation chimeras, TE thermore, the percentage of total CD8ϩ T cells in chimeras and in cells and all other cells of MHCIϪ/ϪIIϪ/Ϫ origin were MHC- MHCIϪ/ϪIIϪ/Ϫ mice compared with B6 mice was reduced by only deficient, whereas cells of B6-nude origin were MHC-sufficient. 1.6 times in the chimeras, whereas it was reduced up to 25 times in Immunofluorescence analysis of thymic sections confirmed the MHCIϪ/ϪIIϪ/Ϫ mice (Fig. 2B). absence of MHC molecules on TE cells and the presence of MHC Thirteen days after infection, chimeric and control splenocytes molecules on non-TE cells of B6-nude origin (Fig. 5, which is were restimulated for 5 days with LCMV peptide-labeled irradiated published as supporting information on the PNAS web site). To splenocytes and tested for their ability to lyse LCMV peptide- analyze whether selection on non-TE MHC alone was sufficient to labeled target cells in a 5-h 51Cr-release assay (Fig. 2C). Only generate a repertoire that can give rise to a functional CD8ϩ T cell chimeric and B6, but not MHCIϪ/ϪIIϪ/Ϫ, splenocytes lysed peptide- response, chimeric and control mice were infected with 200 pfu of labeled target cells (Fig. 2C). Finally, 13 days after infection, blood LCMV-WE i.v. Eight days after infection, blood and CD8ϩ T cells of LCMV-infected mice was analyzed for viral clearance. Two of were stained for binding to the different LCMV-specific tetramers, three chimeric mice were able to clear LCMV from blood as gp33͞Db and np396͞Db tetramers (Fig. 2A). CD8ϩ T cells from efficiently as control mice, whereas all MHCIϪ/ϪIIϪ/Ϫ control mice chimeras and B6 mice always showed binding to either gp33͞Db were unable to clear the virus at any time points tested (data not or np396͞Db tetramers (Fig. 2A Top and Middle). However, shown). The third chimera still had detectable virus in the blood MHCIϪ/ϪIIϪ/Ϫ mice practically were devoid of CD8ϩ T cells, and (40,000 pfu͞ml), but less than any of the MHCIϪ/ϪIIϪ/Ϫ control the few remaining CD8ϩ T cells did not bind to any of the tetramers mice (200,000 Ϯ 72,629 pfu͞ml). (Fig. 2A Bottom). Similar results were obtained after analysis of In conclusion, selection and survival of a functional virus-specific splenic CD8ϩ T cells (data not shown). These results indicated that CD8ϩ T cell repertoire did not require MHC expression on TE mice lacking MHC expression on TE cells but expressing MHC cells; MHC expression on non-TE cells alone was sufficient. molecules on cells other than TE cells both in the thymus and in the periphery showed selection and survival of non-TE MHC-restricted Mice Without MHC Class II-Positive TE Cells Lack Virus-Specific CD4؉ CD8ϩ T cells able to bind LCMV-specific tetramers. T Cell Responses. We next analyzed whether a functional CD4ϩ T Next, day-8 LCMV-immune spleen cells from chimeras and cell response could be selected in mice lacking MHC class II control mice were restimulated for6hwithLCMV-gp33–41 peptide expression on TE cells but not on professional APC. For this ϩ Ϫ/Ϫ (gp33) or LCMV-np396–404 peptide (np396), and CD8 T cells were purpose we generated B6-nude7MHCII tetraparental chime- analyzed for IFN-␥ secretion (Fig. 2B). Again, CD8ϩ T cells from ras. These chimeras lacked MHC class II expression on TE cells and B6 mice (Fig. 2B Top) and chimeras (Fig. 2B Middle) but not from on professional APC of MHCIIϪ/Ϫ origin but not on professional

Martinic et al. PNAS ͉ September 26, 2006 ͉ vol. 103 ͉ no. 39 ͉ 14437 Downloaded by guest on October 1, 2021 Fig. 2. Functional and virus-specific CD8ϩ T cell responses in B6- nude7MHCIϪ/ϪIIϪ/Ϫ chimeras. B6-nude7MHCIϪ/ϪIIϪ/Ϫ chimeras and control mice were infected i.v. with 200 pfu of LCMV-WE. (A) Eight days after infec- ϩ Fig. 3. Mice lacking MHC class II expression on TE cells lack functional tion, blood cells were gated on CD8 T cells and stained for LCMV-specific virus-specific CD4ϩ T cell responses. (A) Chimeras and control mice were ͞ b ͞ b gp33 D and np396 D tetramer binding. Numbers in the lower right and infected i.v. with 200 pfu of LCMV-WE. Eight days after infection, splenocytes upper left quadrants represent the percentage of CD8ϩ T cells binding to were restimulated in vitro for 6 h with LCMV-gp64–80 peptide followed by ͞ b ͞ b LCMV-specific gp33 D and np396 D tetramer, respectively. Data are of one surface staining for CD4 and intracellular staining for IFN-␥.(Top) Mice with ϭ representative mouse per group (total, n 3–4 mice per group). (B) Eight days MHC class II-sufficient TE cells (MHC IIϩ TE cells) and MHC class II-sufficient after infection, splenocytes were restimulated in vitro for 6 h with LCMV- professional APC (MHC IIϩ prof. APC). (Middle) Mice with MHC class II-deficient gp33–41 peptide (gp33, Left) or LCMV-np396–404 peptide (np396, Right) fol- TE cells (MHC IIϪ TE cells) and MHC IIϩ professional APC. (Bottom) Mice with lowed by surface staining for CD8 and intracellular staining for IFN-␥. Numbers Ϫ Ϫ ϩ MHC II TE cells and MHC II professional APC. Numbers in bold indicate the in bold indicate the percentage of CD8 T cells secreting IFN-␥, and numbers ϩ ␥ ϩ percentage of CD4 T cells secreting IFN- , and numbers in italics indicate the in italics indicate the percentage of total CD8 T cells. Data are of one percentage of total CD4ϩ T cells. Data are of one representative mouse per ϭ representative mouse per group (total, n 3–4 mice per group). (C) Thirteen group (total, n ϭ 2–4 mice per group). (B) CD4ϩ T cell-dependent VSV- days after infection, splenocytes were restimulated in vitro for 5 days with neutralizing IgG titers were monitored at different time points after i.v. 51 LCMV peptide-labeled spleen cells and assayed in a 5-h Cr-release assay for infection with 2 ϫ 108 pfu of UV-inactivated VSV-IND. Data represent mean Ϯ b lysis of peptide-loaded EL-4 (H-2 ) target cells. Curves represent individual SEM of three individual mice per group and are representative of one ϭ mice. Data are of one representative mouse per group (total, n 2–3 mice per experiment. group). gp33, gp(33–41); np396, np(396–404).

ϩ ␥ APC of B6-nude origin (Fig. 6, which is published as supporting CD4 T cells from these mice showed secretion of IFN- after restimulation with gp64 (Fig. 3A Top). In contrast, all mice with information on the PNAS web site, and data not shown). We also ϩ analyzed C2ta pIVϪ/Ϫ mice, which lack MHC class II expression on MHC class II-deficient TE cells had a low percentage of total CD4 Ϫ/Ϫ TE cells and on cells of nonhematopoietic origin after IFN-␥ T cells, and, with the exception of the C2ta pIV mice, none of ␥ ϩ stimulation, whereas professional APC are MHC class II-sufficient them showed IFN- -secreting CD4 T cell responses (Fig. 3A et al Middle and Bottom). Analysis of later time points after LCMV (27) (Fig. 6). Waldburger . (27) had previously shown that ϩ CD4ϩ T cell responses against NP-CGG were greatly reduced in infection also yielded an absence of IFN-␥-secreting CD4 T cell C2ta pIVϪ/Ϫ mice. We were interested to analyze the virus-specific responses (data not shown). To investigate whether the relatively ϩ CD4ϩ T cell responses in B6-nude7MHCIϪ/ϪIIϪ/Ϫ, B6- small population of CD4 T cells could provide sufficient cognate nude7MHCIIϪ/Ϫ, C2ta pIVϪ/Ϫ, and control mice after infection help to B cells to enable them to produce virus-neutralizing Ϫ Ϫ Ϫ Ϫ with 200 pfu of LCMV-WE. Eight days after infection, splenocytes antibodies, we infected B6-nude7MHCI / II / , B6- 7 Ϫ/Ϫ Ϫ/Ϫ ϫ 8 were restimulated for6hwiththeLCMV-gp64–80 peptide, and nude MHCII , C2ta pIV , and control mice i.v. with 2 10 CD4ϩ T cells were assayed for IFN-␥ secretion (Fig. 3A). Mice with pfu of UV-inactivated VSV-Indiana (VSV-IND). It has been MHC class II-sufficient TE cells (B6 and C2ta pIVϩ/Ϫ mice) had shown previously that VSV-neutralizing IgG responses are strictly a normal percentage of total CD4ϩ T cells, and LCMV-specific dependent on CD4ϩ T cell help (30). Sera from VSV-infected mice

14438 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0606707103 Martinic et al. Downloaded by guest on October 1, 2021 were analyzed at different time points for VSV-neutralizing IgG tionally skew the outcome of the experiment, were circumvented by (Fig. 3B). Only mice with MHC class II-sufficient TE cells (B6 and the present approach. C2ta pIVϩ/Ϫ mice) were able to mount high titers of VSV- The detection of functional virus-specific CD8ϩ but not CD4ϩ T neutralizing IgG (Fig. 3B). All mice with MHC class II-deficient TE cell responses in mice with MHC-deficient TE cells was surprising. cells, however, were unable to mount any VSV-neutralizing IgG One possible explanation is that the total number of MHC class I responses (Fig. 3B). The same was true for LCMV-specific CD4ϩ and II molecules has to be above a certain threshold to ensure T cell-dependent IgG responses; none of the mice with MHC class selection and survival of CD8ϩ and CD4ϩ T cells, respectively. II-deficient TE cells showed any LCMV-specific IgG titer (data not Whereas TE cells are MHC-deficient in B6-nude7MHCIϪ/ϪIIϪ/Ϫ shown). Taken together, these results showed that selection of a chimeras, all cells originating from nude donor express MHC class functional CD4ϩ T cell repertoire required the presence of MHC I molecules. Therefore, continuous interaction with MHC class I class II-sufficient TE cells. molecules is guaranteed both in the thymus and in secondary lymphoid organs. In this situation, CD8ϩ T cells with intermediate- Discussion to-high-avidity TCR seem to become positively selected and survive The generation of tetraparental aggregation chimeras, B6- subsequently in the periphery. As already shown by previous nude7BALB͞c and B67BALB͞c-nude, has enabled us to studies, CD8ϩ T cells with low-avidity TCR probably will not be analyze the role of non-TE and TE MHC molecules in T cell able to survive in these mice because of low total numbers of repertoire selection in a model where lymphocytes originated TCR-restricting MHC class I molecules (23, 31). In contrast, in mice from both donors but TE cells were derived exclusively from the with MHC class II-deficient TE cells, the total number of MHC non-nude donor. After LCMV infection, CD8ϩ and CD4ϩ T class II molecules in the thymus, not on TE but on professional cells in these chimeras mounted functional and protective T cell APC, is probably not sufficient to ensure selection and survival of responses comparable to control mice, and T cells were restricted CD4ϩ T cells. to TE and non-TE MHC. However, non-TE MHC-restricted T Because thymocytes are in constant interaction with each other, cell responses were somewhat lower than TE MHC-restricted one can speculate as to whether expression of MHC class II responses. In our previously reported RagϪ/Ϫb7nuded and molecules on these cells would increase the total MHC class II Scidd7nudeb chimeras, primary CD8ϩ T cell responses re- numbers above the threshold level needed for CD4ϩ T cell selection stricted to non-TE MHC could always be measured in a direct and survival. In two recent studies this question was elegantly ex vivo cytotoxicity assay, whereas in B6-nude7BALB͞c and answered by the generation of transgenic mice expressing the B67BALB͞c-nude chimeras only Ϸ25% of the mice showed human class II transactivator molecule (CIITA) under the control primary CD8ϩ T cell responses restricted to non-TE MHC (data of the CD4 promoter or the proximal promoter of p56lck. The not shown); however, all responded well after secondary in vitro resulting CIITATg mice expressed MHC class II molecules on stimulation. This finding may reflect quantitative and͞or qual- immature thymocytes and mature T cells (32, 33). The authors itative differences favoring survival and͞or proliferation of TE generated irradiation bone marrow chimeras by injecting bone MHC-restricted T cells in the thymus, correlating with the marrow cells from CIITATg mice into irradiated MHC class II- smaller number of cells expressing non-TE MHC compared with deficient recipients and showed that successful CD4ϩ T cell selec- the ones expressing TE MHC. In the periphery, both TE and tion occurred in the absence of MHC class II molecules on TE cells non-TE MHC-restricted T cells should have a comparable (32, 33). Whereas murine thymocytes do not express MHC class II IMMUNOLOGY survival-favoring environment because the proportion of lym- molecules, human, guinea pig, rat, and sheep immature thymocytes phocytes, APC, and all other somatic cells originating from and͞or peripheral T cells express MHC class II molecules (32, 33) each donor was about equivalent. In our earlier series of and could therefore participate in CD4ϩ T cell selection. RagϪ/Ϫb7nuded tetraparental aggregation chimeras, non-TE Thus, the presence of any MHC class II (independent of haplo- MHC-restricted T cells may have had a survival advantage, type or maybe even species) on TE cells seems to be necessary and because in the thymus as well as secondary lymphoid organs sufficient for the development of a functional CD4ϩ T cell reper- lymphocytes were exclusively of nude origin, and therefore the toire, whereas MHC class II on non-TE cells alone was not total number of cells expressing non-TE MHC was higher than sufficient, at least not in mice. This concept is supported by the number of cells expressing TE MHC. experiments in which mice received a xenogeneic porcine thymus The question of whether non-TE MHC alone is sufficient to graft. These mice showed the presence of CD4ϩ T cells restricted efficiently select a functional T cell repertoire had been studied to porcine MHC class II even in the absence of porcine MHC class previously with bone marrow chimeras and TCR transgenic mice. II in the periphery. The cells could survive in the periphery because Most of these studies came to the conclusion that non-TE MHC of the presence of murine MHC class II IAb, which shows highly only inefficiently selected a T cell repertoire (12, 23). For example, conserved sequence, with porcine MHC class II DQd Bix and Raulet (12) injected MHC class I-sufficient bone marrow probably substituting therefore the need for porcine MHC class II into irradiated MHC class I-deficient recipients and showed that in the periphery (34). These results are compatible with an attrac- CD8ϩ T cell numbers were reduced. Importantly, CD8ϩ T cells tive generalization proposed here: Maturing T cells need to contact exhibited allospecific T cell responses only when primed and MHC but not a particular allelic form of MHC. MHC class I contact boosted several times, whereas usually these responses are very on TE or non-TE cells is sufficient for maturing CD8ϩ T cells; strong in the absence of priming and do not require extensive however, maturing CD4ϩ T cells require MHC class II contact on boosting. In our B6-nude7MHCIϪ/ϪIIϪ/Ϫ chimeras, however, TE cells. where TE cells are devoid of MHC molecules but non-TE cells are These findings suggest that positive selection reflects more the MHC-sufficient, CD8ϩ T cell numbers were reduced only 1.6 times. reaching of a quantitatively critical number of contacts between the After LCMV infection, chimeras showed functional virus-specific T TCR and its specific antigen presented on MHC class I or II cell responses. It is important to remember that the chimeras (allele-independent) than of antigen-independent but MHC allele- described in this study, in comparison to bone marrow chimeras, dependent contacts (1, 2, 35–38). were not lethally irradiated. Therefore, MHC-sufficient non-TE cells in our chimeras were present in the thymus from the beginning, Materials and Methods probably giving the developing thymocytes restricted to non-TE Mice. Tetraparental aggregation chimeras were generated at the MHC a survival advantage, which is unlikely to occur in bone Institute of Laboratory Animal Science of the University of marrow chimeras. In addition, irradiation damage and subclinical Zurich (Zurich, Switzerland), as previously described (19). graft-versus-host or host-versus-graft reactions, which may addi- C57BL͞6 (H-2b) (B6) and BALB͞c (H-2d) mice were purchased

Martinic et al. PNAS ͉ September 26, 2006 ͉ vol. 103 ͉ no. 39 ͉ 14439 Downloaded by guest on October 1, 2021 from Harlan (Horst, The Netherlands). C57BL͞6-nude (B6-nude, M) in DMSO were diluted 1͞1,000 and 1͞100,000 in medium for H-2b) and BALB͞c-nude (H-2d) mice were from Biological Re- intracellular stains and for restimulations, respectively. search Laboratories (Fu¨llinsdorf, Switzerland). MHCIϪ/ϪIIϪ/Ϫ and MHCIIϪ/Ϫ (H-2b) mice were from Taconic Europe (Lille Skensved, Intracellular Cytokine Staining for IFN-␥. Approximately 1 ϫ 106 Denmark). C2ta pIVϩ/Ϫ and C2ta pIVϪ/Ϫ mice (H-2b) (27) were splenocytes were incubated for6hat37°Cinatotalvolumeof200 ␮ ␮ obtained from Walter Reith (University of Geneva Medical School, l of medium with 2.5 M relevant peptide, with medium alone, or with 50 ng͞ml phorbol 12-myristate 13-acetate plus 500 ng͞ml Geneva, Switzerland). ␮ ͞ ͞ Mice were bred and maintained in specific pathogen-free con- ionomycin, in the presence of 5 g ml brefeldin A and 50 units ml recombinant mouse IL-2. After incubation, cells were surface- ditions. Experiments were done according to institutional guide- stained with anti-H-2Dd-FITC (clone 34-2-12) and anti-CD8b.2- lines and Swiss federal and cantonal laws on animal protection. phycoerythrin (clone 53-5.8). Cells were then fixed with 2% para- formaldehyde and permeabilized with permeabilization buffer 51 Cell Lines, Cr-Release Assay, VSV-IND Neutralization Assay, and (flow cytometry buffer containing 0.1% saponin) followed by b d Virus. EL-4 (H-2 ) and P815 (H-2 ) cells were obtained from the intracellular staining with anti-mouse IFN-␥-allophycocyanin American Type Culture Collection (Manassas, VA). The 51Cr- (clone XMG1.2). release assay after restimulation in vitro, the VSV-IND neutraliza- tion assay, LCMV-WE, and VSV-IND have been previously de- Immunofluorescence. Immunofluorescence histology of TE cells has scribed (39–42). been previously described (19). Images were recorded with an Olympus BX61 fluorescence microscope (Olympus, Melville, NY) Antibodies, LCMV Tetramers, and Peptides. All antibodies used for and processed by using AnalySIS Software (Soft Imaging System, flow cytometry were obtained from BD Biosciences Pharmingen Mu¨nster, Germany). (San Jose, CA). Samples were acquired on a FACSCalibur and analyzed by using FlowJo software (Version 6.3.1; Tree Star, Statistical Analysis. Statistical analysis was performed with Prism 4 Ashland, OR). Double tetramer stains with phycoerythrin- and software (GraphPad, San Diego, CA). allophycocyanin-labeled tetramers have been described (19). The LCMV-specific peptides [gp33 (gp , KAVYNFATC), gp64 We thank Dr. E. Ling for critical reading of the manuscript, K. Rappold 33–41 and K. Osei-Boadum for animal husbandry, A. Nowotny for technical (gp64–80, GPDIYKGVYQFKSVEFD), np396 (np396–404, assistance with histological analysis, and N. Wey for technical help and FQPQNGQFI), and np118 (np118–126, RPQASGVYM)] were syn- guidance on fluorescence microscopy. This work was supported by the thesized by Neosystems (Strasbourg, France). Peptide stocks (10Ϫ2 Swiss National Science Foundation and the Kanton of Zurich.

1. Sebzda E, Mariathasan S, Ohteki T, Jones R, Bachmann MF, Ohashi PS (1999) 23. Zerrahn J, Volkmann A, Coles MC, Held W, Lemonnier FA, Raulet DH (1999) Annu Rev Immunol 17:829–874. Proc Natl Acad Sci USA 96:11470–11475. 2. Starr TK, Jameson SC, Hogquist KA (2003) Annu Rev Immunol 21:139–176. 24. Markowitz JS, Auchincloss H, Jr, Grusby MJ, Glimcher LH (1993) Proc Natl 3. Miller J (1961) Lancet 2:748–749. Acad Sci USA 90:2779–2783. 4. Moller, GE (1978) Immunol Rev 42:3–270. 25. Grusby MJ, Auchincloss H, Jr, Lee R, Johnson RS, Spencer JP, Zijlstra M, 5. von Boehmer H, Aifantis I, Gounari F, Azogui O, Haughn L, Apostolou I, Jaenisch R, Papaioannou VE, Glimcher LH (1993) Proc Natl Acad Sci USA Jaeckel E, Grassi F, Klein L (2003) Immunol Rev 191:62–78. 90:3913–3917. 6. Ron Y, Lo D, Sprent J (1986) J Immunol 137:1764–1771. 26. Grusby MJ, Glimcher LH (1995) Annu Rev Immunol 13:417–435. 7. von Boehmer H (1990) Annu Rev Immunol 8:531–556. 27. Waldburger JM, Suter T, Fontana A, Acha-Orbea H, Reith W (2001) J Exp Med 8. Matzinger P, Mirkwood G (1978) J Exp Med 148:84–92. 194:393–406. 9. Doherty PC, Bennink JC (1979) J Exp Med 149:150–157. 28. Ahmed R, Butler LD, Bhatti L (1988) J Virol 62:2102–2106. 10. Longo DL, Schwartz RH (1980) Nature 287:44–46. 29. Ochsenbein AF, Pinschewer DD, Odermatt B, Ciurea A, Hengartner H, 11. Wagner H, Rollinghoff M, Rodt H, Thierfelder S (1980) Eur J Immunol Zinkernagel RM (2000) J Immunol 164:6296–6302. 10:521–525. 30. Leist TP, Cobbold SP, Waldmann H, Aguet M, Zinkernagel RM (1987) 12. Bix M, Raulet D (1992) Nature 359:330–333. J Immunol 138:2278–2281. 13. Hugo P, Kappler JW, McCormack JE, Marrack P (1993) Proc Natl Acad Sci 31. Martinic MM, Rocha B, McCoy KD, Hengartner H, Zinkernagel RM (2004) USA 90:10335–10339. Eur J Immunol 34:1041–1049. 14. Pawlowski T, Elliott JD, Loh DY, Staerz UD (1993) Nature 364:642–645. 32. Li W, Kim MG, Gourley TS, McCarthy BP, Sant’angelo DB, Chang CH (2005) 15. Vukmanovic S, Grandea AG, III, Faas SJ, Knowles BB, Bevan MJ (1992) Immunity 23:375–386. Nature 359:729–732. 33. Choi EY, Jung KC, Park HJ, Chung DH, Song JS, Yang SD, Simpson E, Park 16. Takahama Y, Suzuki H, Katz KS, Grusby MJ, Singer A (1994) Nature SH (2005) Immunity 23:387–396. 371:67–70. 34. Rodriguez-Barbosa JI, Zhao Y, Zhao G, Ezquerra A, Sykes M (2002) 17. Cibotti R, Punt JA, Dash KS, Sharrow SO, Singer A (1997) Immunity J Immunol 169:6697–6710. 6:245–255. 35. Matzinger P (1993) Immunol Rev 135:81–117. 18. Zinkernagel RM, Althage A, Waterfield E, Kindred B, Welsh RM, Callahan 36. Singer A (1988) J Immunol 140:2481–2483. G, Pincetl P (1980) J Exp Med 151:376–399. 37. Katz DH (1980) Adv Immunol 29:137–207. 19. Martinic MM, Rulicke T, Althage A, Odermatt B, Hochli M, Lamarre A, 38. Cosgrove D, Chan SH, Waltzinger C, Benoist C, Mathis D (1992) Int Immunol Dumrese T, Speiser DE, Kyburz D, Hengartner H, Zinkernagel RM (2003) 4:707–710. Proc Natl Acad Sci USA 100:1861–1866. 39. Speiser DE, Stubi U, Zinkernagel RM (1992) Nature 355:170–172. 20. Kirberg J, Berns A, von Boehmer H (1997) J Exp Med 186:1269–1275. 40. Bachmann MF, Kundig TM, Kalberer CP, Hengartner H, Zinkernagel RM 21. Goldrath AW, Bevan MJ (1999) Nature 402:255–262. (1993) J Virol 67:3917–3922. 22. Tanchot C, Lemonnier FA, Perarnau B, Freitas AA, Rocha B (1997) Science 41. Lehmann-Grube F (1971) Virol Monogr 10:1–173. 276:2057–2062. 42. McLaren LC, Holland JJ, Syverton JT (1959) J Exp Med 109:475–485.

14440 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0606707103 Martinic et al. Downloaded by guest on October 1, 2021