Regulation of Lineage Commitment Distinct from Positive Selection

See allHide authors and affiliations

Science  05 Nov 1999:
Vol. 286, Issue 5442, pp. 1149-1153
DOI: 10.1126/science.286.5442.1149


Developing αβ T cells diverge into the CD4 and CD8 lineages as they mature in the thymus. It is unclear whether lineage commitment is mechanistically distinct from the process that selects for the survival of T cells with useful T cell receptor (TCR) specificities (positive selection). In HD mice, which lack mature CD4+ T cells, major histocompatibility complex (MHC) class II–restricted T cells are redirected to the CD8 lineage independent of MHC class I expression. However, neither TCR-mediated signaling nor positive selection is impaired. Thus, the HD mutation provides genetic evidence that lineage commitment may be mechanistically distinct from positive selection.

Developing αβ thymocytes go through three major phenotypic stages, first expressing neither CD4 nor CD8 (double negative; DN), then expressing both (double positive; DP), and finally expressing only one or the other (CD4+8 or CD48+) (SP). At the DP stage thymocytes are selected to undergo the alternative outcomes of negative selection, positive selection, or death by neglect depending on the interaction of the αβTCR complex with intrathymic major histocompatibility complex ligands (1,2). Coincident with positive selection, thymocytes undergo lineage commitment, a process that ensures the correlation of TCR specificity toward class I or II MHC with the cell's functional phenotype as a CD8+ killer or a CD4+ helper T cell. Various mechanisms have been proposed to explain how this correlation is achieved (3–6). At the molecular level CD4 and CD8 coreceptors (7, 8) as well as the Ras-MAPK (9) and Notch (10) pathways are likely to play a role. The mutant HD mouse (11) is deficient in generation of peripheral CD4+ T cells because of a specific defect in thymic development not affecting antigen presentation or CD4 function, distinct from other spontaneous and induced mouse mutants with similar phenotypes (12). Although the HD defect was shown to be intrinsic to the hematopoietic lineage (11), it is unclear whether it maps to thymocytes or to bone marrow–derived nonthymocytes, which could cause aberrant negative selection of class II–restricted thymocytes.

To test this, we cotransferred bone marrow fromHD −/− and HD +/+ mice into the same RAG −/− recipients (13). If another cell type were acting in trans to prevent thymocytes from maturing to the CD4 lineage, then in a mixed chimera thymocytes of both HD −/− andHD +/+ origins should behave similarly. On the other hand, if the defect were intrinsic to developing thymocytes,HD −/− and HD +/+thymocytes should behave distinctly. To distinguish between thymocytes of HD −/− and HD +/+origin, we obtained HD +/+ bone marrow from B6.SJL congenic mice that carry the CD45.1 allele on an otherwise C57BL/6 background, in contrast toHD −/− mice that carry the CD45.2allele. Thymocytes of different origins behaved autonomously—that is,HD −/− (CD45.2 +) cells matured exclusively to the CD8 lineage, whereasHD +/+ (CD45.1 +) cells gave rise to CD4+ and CD8+ SP T cells in normal proportions (Fig. 1). In addition, thymocytes ofHD −/− origin give rise to a population of CD4+8lo cells, an intermediate stage that in normal mice includes precursors of both SP CD4+ and CD8+ mature T cell subsets (5, 14) and that is increased in HD −/− mice (11). In agreement with the thymic subset composition, peripheral T cell populations of HD +/+ andHD −/− origin consist, respectively, of both SP CD4+ and CD8+ cells or only SP CD8+cells. In addition, HD −/− peripheral T cells include a minor population of CD4+8+ cells, which is a typical feature of HD −/− mice (11). Thus, abnormal development ofHD −/− thymocytes is not mediated by another hematopoietic cell type but is inherent to developing thymocytes. It can be further deduced that the defective HD product is unlikely to be a soluble factor.

Figure 1

Coreconstitution ofRAG −/− hosts withHD −/− and HD +/+ bone marrow. (Upper) Thymocytes and lymph node (LN) cells from an irradiated RAG −/− mouse reconstituted with bone marrow from both HD −/−(CD45.2+) and HD +/+(CD45.1+) mice were stained with antibodies specific for CD4 and CD8, as well as CD45.1 or CD45.2, and CD69 or HSA. Samples were gated on the basis of CD45.1/CD45.2 expression, to distinguish cells ofHD −/− and HD +/+ origin, as indicated. CD4:CD8 expression profiles are also shown for total cell samples. (Lower) HSA and CD69 staining profiles of gated SP CD4+ and SP CD8+ thymocytes. Mature cells bear the HSAloCD69lo phenotype. Note the absence of HSAlo and CD69lo cells in theHD −/− SP CD4+ subset.

Thymocyte development depends on signaling via the TCR, coreceptors, and associated tyrosine kinases p56lck and ZAP-70. Initially, p56lck phosphorylates components of the TCR complex, thereby leading to the recruitment of additional signaling factors including ZAP-70, which in turn is phosphorylated by p56lck. We have examined these early signaling events inHD −/− mice by an in vitro immune complex kinase assay (15, 16).HD −/− and HD +/+thymocytes exhibit equivalent phosphorylation of CD3ζ, CD3ɛ, ZAP-70, p56lck, and the exogenous substrate enolase, showing that p56lck is activated normally and that ZAP-70 associates with the TCR complex and is phosphorylated normally (Fig. 2A). In agreement, immunoblot analysis of TCR complexes from unstimulated thymocytes or from TCRβ/CD4 antibody (anti-TCRβ/CD4) cross-linked HD −/−thymocytes show normal constitutive phosphorylation of CD3ζ (p21 isoform) and the normal appearance of higher molecular weight phosphorylated isoforms of CD3ζ and ZAP-70 after cross-linking (Fig. 2B) (17). To measure the function of TCR-mediated signaling in mature T cells, we assayed the proliferative response of HD −/− splenocytes to cross-linking with plate-bound anti-CD3ɛ or anti-TCRβ (18). The response ofHD −/− splenocytes, although less than that of wild-type (WT) animals, is equivalent to that ofI-A β b −/− mice, which similarly lack peripheral CD4+ T cells (Fig. 2C). Consistent with normal TCR expression and signaling inHD −/− mice, breeding experiments showed that the HD defect is unlinked to genes encoding proximal components of the TCR-mediated signaling cascade—that is CD3δ, TCRαβ, and p56lck—and DNA sequencing showed that the ZAP-70 mRNA fromHD −/− mice is normal (19).

Figure 2

Assays of T cell function inHD −/− mice. (A) In vitro kinase activity of anti-CD3ζ immunoprecipitates from anti-TCRβ/CD4 stimulated (lanes +) or unstimulated (lanes −) total thymocytes ofHD −/− or WT mice. (B) Anti-phosphotyrosine immunoblot analysis of CD3ζ isoforms fromHD −/− or WT thymocytes stimulated (lanes +) or unstimulated (lanes −) with anti-TCRβ/CD4. (C) Proliferation of total splenocytes from HD −/−,I-A β b −/−, and WT mice in response to anti-CD3ɛ or anti-TCRβ treatment. Background proliferation in response to phosphate-buffered saline alone has been subtracted.

The absence of mature SP CD4+ T cells inHD −/− mice indicates that class II–restricted thymocytes undergo an abnormal developmental fate—that is, either their development is blocked or they undergo alternative development, specifically, to the CD8 lineage. The latter possibility is suggested by the increased representation of SP CD8+ thymocytes inHD −/− relative to WT mice (11). To evaluate these two possibilities, we crossed the AND TCR transgene onto the HD −/− background, thereby limiting the repertoire of developing thymocytes to a single class II–restricted specificity (20, 21). HD −/− mice bear the H-2b haplotype and express the I-Ab product, which mediates efficient positive selection of AND+ thymocytes (22). To exclude the possibility of rearrangement and expression of endogenous TCR products, we did these experiments on a RAG −/−background. A comparison of peripheral T cells fromAND + HD −/− RAG −/− H-2b mice withAND + HD +/− RAG −/− H-2b littermates revealed different subset distributions (Fig. 3). Although,AND + HD +/− mice show the expected predominance of SP CD4+ T cells,AND + HD −/− mice give rise exclusively to SP CD8+ T cells. Analysis of thymocytes from AND + HD −/− RAG −/− mice directly shows the occurrence of redirection of class II–restricted AND +thymocytes to the CD8 lineage. Thus, AND + HD −/− mice contain 10 times more SP CD8+ thymocytes than their AND + HD +/− littermates. As usual, inHD −/− mice, a substantial population of CD4+8lo intermediate cells is also observed (Fig. 3). In normal mice, thymocytes that are undergoing positive selection show high surface expression of the activation marker CD69 (23), whereas the most mature SP thymocytes that are ready to exit to the periphery exhibit low levels of CD69 and heat stable antigen (HSA) but high levels of CD62L (24). To compare thymocyte maturation routes in AND + HD +/− versus HD −/−mice, we determined the CD4/CD8 expression profiles of cells contained within the CD69hi and CD62Lhi HSAlopopulations. In AND + HD +/− mice both CD69hi and CD62Lhi HSAlo populations consist primarily of CD4+8 thymocytes (Fig. 3). In contrast, inAND + HD −/− mice, the CD69hi population consists almost entirely of CD4+8lo cells, whereas the CD62LhiHSAlo population is composed predominantly of CD48+ cells. Thus class II–restricted CD48+ cells in AND + HD −/− mice are mature and probably pass through the intermediate CD4+8lo stage, consistent with the proposed maturation route of class I–restricted CD48+ thymocytes in normal mice (5,14).

Figure 3

T cell development inHD −/− and HD +/− mice carrying the class II–restricted AND TCR transgene. Thymocytes and lymph node (LN) cells from AND + HD −/− RAG −/− andAND + HD +/− RAG −/− mice were stained with antibodies specific for CD4, CD8, CD69, HSA, and CD62L and analyzed by flow cytometry. Histograms represent total cell samples or specific gated subsets, as indicated. CD69+ thymocytes are thought to be undergoing positive selection, and CD62LhiHSAlo thymocytes are considered fully mature. Note the absence of fully mature SP CD4+ and SP CD8+subsets in AND + HD −/− RAG −/− and AND + HD +/− RAG −/− mice, respectively.

Some class II–restricted TCRs can support inefficient differentiation to the CD8 lineage in normal mice or efficient maturation in CD4 −/− mice (6, 25). In both cases, alternative development requires MHC class I expression by thymic epithelial cells. To test whether there is a similar requirement for class I expression in the redirection of class II–restricted thymocytes in HD −/− mice, we crossed the HD mutation onto a β2 M −/− background in which class I expression is blocked (21, 26). Doubly deficient HD −/−β2 M −/− mice arising from this cross were identified phenotypically by the simultaneous absence of MHC class I expression and peripheral CD4+ T cells. These mice still generate SP CD8+thymocytes, which are mature based on expression of αβTCR, CD69, HSA, and CD62L (Fig. 4) (27). In agreement, peripheral T cells consist exclusively of CD8+ cells. To confirm this result, we generated radiation chimeras by transferring hematopoietic cells ofHD −/− origin into HD +/+β2 M −/− hosts. Because positively selected thymocytes upregulate class I expression (28), HD −/− thymocytes undergoing positive selection in these chimeras can be readily distinguished from residual host-derived thymocytes on the basis of their expression of the MHC class I product H-2Kb. Consistent with our previous result, HD −/− precursors give rise only to mature cells of the CD8 lineage as well as to CD4+8lo intermediate cells (29).

Figure 4

T cell development in HD −/−mice in the absence of MHC class I expression. Thymocytes and lymph node (LN) cells from HD −/−β2 M + I-A β b +,HD +β2 M −/−,HD −/−β2 M −/−,HD + I-A β b −/− andHD −/− I-A β b −/− mice were stained with antibodies specific for CD4 and CD8 and analyzed by flow cytometry. Note the presence of substantial numbers of SP CD8+ thymocytes and mature T cells inHD −/−β2 M −/− mice.

To similarly test the dependence of thymic development inHD −/− mice on MHC class II expression, we transferred HD −/− bone marrow into irradiatedI-A β b −/− RAG −/− hosts. SP CD8+ thymocytes and peripheral T cells are still generated in these chimeras, indicating that MHC class I–dependent maturation to the CD8 lineage is unaffected by the HD defect (Fig. 4). However, the CD4+8lo subset that is so abundant inHD −/− mice is substantially reduced, confirming that most of these cells in HD −/−mice are class II restricted.

Finally, to assess the MHC requirements for development of a particular class II–restricted TCR in theHD −/− background, we transferredAND + HD −/− RAG −/− bone marrow into both β2 M −/− RAG −/− H-2b andI-A β b −/− RAG −/− H-2b recipients. Consistent with our preceding results, SP CD8+ thymocytes of the mature CD69loCD62LhiHSAlophenotype and peripheral SP CD8+ T cells are still generated in the absence of MHC class I, as are intermediate CD4+8lo thymocytes expressing the CD69hi CD62LloHSAhi phenotype (Fig. 5). In contrast, in the MHC class II–deficient background neither intermediate nor fully mature subsets are detected, nor are there any peripheral T cells [except for a minor population of peripheral CD48 T cells, also found inAND + I-A β b −/− HD + mice and thus unrelated to the HD phenotype (29); similar DN cells arise in other TCR transgenics in the absence of positive selection (30)]. Thus maturation ofAND + CD8+ T cells in theHD −/− background depends on expression of a selecting class II ligand and is independent of MHC class I.

Figure 5

MHC dependence of AND TCR-mediated development inHD −/− mice. Thymocytes and lymph node cells (PBLs) from irradiated RAG −/−β2 M −/− (top row) or RAG −/− I-A β b −/− (bottom row) recipients reconstituted 5 weeks earlier with bone marrow from AND + HD −/− RAG −/− mice were stained with antibodies specific for CD4 and CD8 as well as CD69, CD62L, and HSA. Some samples were gated on the basis of CD69, HSA, and CD62L expression, as in Fig. 3, to distinguish cells that have undergone positive selection (CD69+) or completed maturation (HSAloCD62Lhi). These subsets were absent in the case of chimeras generated in RAG −/− I-A β b −/− hosts.

Redirection and complete maturation of class II–restricted thymocytes to the CD8 lineage favors the instructive model and is inconsistent with the stochastic/selective model of lineage commitment, because the latter predicts that thymocytes that adopt the incorrect lineage would be incapable of surviving the subsequent selection step. Of further mechanistic importance is the fact that redirection inHD −/− mice is independent of class I MHC. In contrast, when such redirection occurs in HD +mice it is strictly dependent on class I MHC, suggesting that CD8 engagement is normally required for CD8 lineage commitment (6,25). The abolition of this requirement inHD −/− mice suggests that a CD8 lineage-determining pathway is constitutively activated. It has been proposed that lineage commitment is regulated by TCR signal strength, so that stronger and weaker signals give rise to CD4+ and CD8+ T cells, respectively (6). It seems unlikely, however, that redirection of class II–restricted thymocytes in HD −/− mice could be due to a weakened TCR signal, because all other TCR-mediated processes examined inHD −/− mice, in particular positive selection, appear to be normal. Instead these results argue that lineage-specifying signals are qualitatively distinct from the TCR-mediated signals that support positive selection.

  • * To whom correspondence should be addressed. E-mail: dj_kappes{at}


Stay Connected to Science

Navigate This Article