Unraveling a Revealing Paradox : Why Major Histocompatibility Complex I–signaled Thymocytes “Paradoxically” Appear as CD4⁺8^lo Transitional Cells During Positive Selection of CD8⁺ T Cells

Immature CD4⁺8⁺ (double positive [DP]^*) thymocytes are signaled in the thymus to differentiate into mature single positive T cells by TCR engagement of intrathymic MHC–peptide complexes (1). TCR engagement of intrathymic peptide–MHC-II complexes results in differentiation of DP thymocytes into mature CD4⁺ T cells whereas TCR engagement of intrathymic peptide–MHC-I complexes results in differentiation of DP thymocytes into mature CD8⁺ T cells (2, 3). The mechanism by which the MHC class specificity of surface TCR complexes determines lineage choice remains an unresolved problem in developmental immunology. It is thought that lineage choice is a consequence of the coreceptor molecules that individual TCRs use to engage intrathymic ligands, as TCR engagement of peptide–MHC-II complexes generally requires coengagement of surface CD4 coreceptor molecules whereas TCR engagement of peptide–MHC-I complexes generally requires coengagement of surface CD8 coreceptor molecules (4, 5). However, the actual mechanism by which TCR and coreceptor coengagements influence lineage choice remains a matter of intense controversy.

Classical models of lineage commitment are of two general types: instructional and stochastic (6–10). Instructional models of lineage commitment propose that TCR plus CD4 coengagements on DP thymocytes terminate CD8 gene expression and dictate commitment to the CD4⁺ T cell lineage by transducing signals that differ either qualitatively or quantitatively (in strength or duration) from signals transduced by TCR plus CD8 coengagements that terminate CD4 gene expression and dictate commitment to the CD8⁺ T cell lineage (11–15). Stochastic models of lineage commitment propose that coengagement of TCR with either coreceptor molecule signals DP thymocytes to randomly terminate expression of one coreceptor gene, with only thymocytes still expressing matching TCR and coreceptor molecules responsive to subsequent survival signals (16, 17). In both models, lineage commitment is postulated to occur in DP thymocytes and result in silencing of the opposite coreceptor gene.

As a result, classical models of lineage commitment predict that DP thymocytes that have committed to the CD4⁺ T cell lineage have terminated CD8 gene expression and therefore will lose CD8 surface protein expression and appear as CD4⁺8^lo transitional cells. Reciprocally, signaled DP thymocytes that have committed to the CD8⁺ T cell lineage have terminated CD4 gene expression and will lose CD4 surface protein expression and appear as CD4^lo8⁺ cells. However, experimental observations are discordant with these predictions as assessment of CD4⁺8^lo and CD4^lo8⁺ transitional populations revealed that a significant proportion of CD8 lineage cells initially appear as CD4⁺8^lo transitional cells (18–22). The appearance of CD8-committed thymocytes as CD4⁺8^lo transitional cells is paradoxical on two levels, as: (a) CD8-committed thymocytes are thought to have terminated CD4 gene expression, yet these cells display high levels of CD4 proteins on their cell surface, and (b) CD8-committed thymocytes are thought to express CD8 coreceptor genes, yet these cells display low levels of CD8 proteins on their cell surface. Although the paradoxical loss of CD8 surface proteins on MHC-I–signaled thymocytes had not been predicted, it could be reconciled with classical lineage commitment models by hypothesizing that it was due to a direct effect of MHC-I signals on CD8 surface proteins that resulted in reduced CD8 expression on CD8-committed thymocytes (9, 22, 23). This classical view of CD8⁺ T cell differentiation is depicted in Fig. 1 A.

In contrast, the kinetic signaling model of lineage commitment proposes that signaled DP thymocytes terminate CD8 gene expression and transcriptionally convert into CD4⁺8⁻ intermediate thymocytes that still retain the potential to differentiate into either CD4⁺ or CD8⁺ T cells (24–26). So kinetic signaling postulates that lineage commitment occurs after DP thymocytes have converted into CD4⁺8⁻ intermediate thymocytes. According to the kinetic signaling model, lineage direction is determined in CD4⁺8⁻ intermediate thymocytes simply by whether TCR-mediated positive selection signals persist or cease. If TCR signaling persists, intermediate CD4⁺8⁻ thymocytes differentiate into CD4⁺ T cells, but if TCR signaling ceases, intermediate CD4⁺8⁻ thymocytes differentiate into CD8⁺ T cells. As a result, a critical step in the differentiation of intermediate CD4⁺8⁻ thymocytes into CD8⁺ T cells is the silencing of CD4 and reinitiation of CD8 gene expression, referred to as “coreceptor reversal.” Consequently, the kinetic signaling model predicts that MHC-I–signaled thymocytes initially appear as CD4⁺8^lo cells because they initially terminate CD8 gene expression before undergoing coreceptor reversal to transcriptionally become CD4⁻8⁺ thymocytes. The kinetic signaling view of CD8⁺ T cell differentiation in the thymus is depicted in Fig. 1 B.

Determining the basis for reduced CD8 surface protein expression on MHC-I–signaled thymocytes during CD8⁺ T cell differentiation tests predictions made by different concepts of how lineage commitment occurs. The classical concept that lineage commitment occurs in DP thymocytes and results in irreversible silencing of the opposite coreceptor gene predicts that low CD8 surface protein expression on MHC-I–signaled CD4⁺8^lo thymocytes is due to direct effects on CD8 surface protein expression, not CD8 gene expression. In contrast, the kinetic signaling model predicts that low CD8 surface protein expression on MHC-I–signaled CD4⁺8^lo thymocytes is due to down-regulation of CD8 gene expression. Consequently, this study has attempted to determine the actual basis for reduced CD8 surface protein expression on MHC-I–signaled CD4⁺8^lo thymocytes. These results demonstrate that reduced CD8 surface expression on MHC-I–signaled CD4⁺8^lo thymocytes is not due to internalization nor slow reexpression of CD8 surface proteins, but is due to down-regulated CD8 gene expression. In addition, this study demonstrates that down-regulation of CD8 gene expression does not imply commitment to the CD4 lineage, as many intermediate CD4⁺8⁻ thymocytes retain the ability to differentiate into CD8⁺ T cells. These results support the kinetic signaling model but challenge more classical concepts of lineage commitment.

Mice deficient in MHC-II (MHC-II^o; reference 27), CD8α (CD8α^o; reference 5), or CD8β (CD8β^o; reference 28) were used at 8–12 wk of age and maintained in our own animal facility. MHC-II^o mice were bred with CD8α^o mice to generate II°CD8α^o mice, and were bred with CD8β^o mice to generate II°CD8β^o mice.

CD8α transgenic mice constructed with human CD2 (hCD2)-based CD8α transgenes encoding either full-length CD8α or tailless CD8α′ molecules have been described (29, 30). Transgenic and endogenous CD8α proteins expressed different allelic forms of CD8α with transgenic CD8α proteins being CD8α.1 and endogenous CD8α proteins being CD8α.2. The CD8α transgenic mouse lines expressed transgenic and endogenous CD8α proteins at comparable levels. The CD8α.1 transgene was introduced by mating into MHC-II^o, II^oCD8α^o, and II^oCD8β^o mice to generate MHC-II^o mice that either expressed both endogenous and transgenic CD8α proteins or expressed only transgenic CD8α proteins.

Line 30 transgenic mice expressing a transgene encoding hCD2 cDNA under the control of the CD4 locus control region (including the CD4 silencer element) were provided by D. Littman (Skirball Institute of Biomolecular Medicine, New York, NY; 31). Surface expression of hCD2 protein on thymocytes and T cells in Line 30 mice serves as a faithful reporter of CD4 promoter activity and strictly parallels CD4 gene expression (31). Line 30 mice were mated with MHC-II^o mice to generate MHC-II^o Line 30 mice.

Fluorochrome-conjugated antibodies with the following specificities were used for direct immunofluorescence: pan CD8α (53-6.7; BD Biosciences), CD8α.1 (116-113.1; reference 32), CD8α.2 (2.43; reference 33), CD4 (GK1.5; BD Biosciences), TCRβ (H57-597; BD Biosciences), and hCD2 (G11; Caltag). Recombinant murine IL-7 (R&D Systems) was added to cultures where indicated.

Cells were harvested, stained with fluorochrome-conjugated antibodies, and analyzed on a multi-laser FACSVantage™ SE (Becton Dickinson). Dead cells were excluded by forward light scatter gating and propidium iodide staining. Analysis was performed using software developed at the National Institutes of Health.

Where indicated, thymocytes were pronase treated to remove preexisting surface CD4 and CD8 proteins (22). 5 × 10⁶/ml cells were treated with 0.01% pronase (Calbiochem) for 15 min at 37°C and then cultured at 37°C overnight during which time they reexpressed the coreceptor molecules they were actively transcribing (22). For quantitative analysis of surface coreceptor reexpression, CD4 and CD8 fluorescence on thymocytes at various times after pronase stripping was quantitated in total fluorescence units and expressed as a percentage of CD4 and CD8 fluorescence on DP thymocytes after full coreceptor reexpression at 24 h, which was set equal to 100% and was comparable to CD4 and CD8 fluorescence on untreated DP thymocytes.

Purified thymocyte subpopulations were placed in suspension cultures as previously described (24, 26). Where indicated, recombinant murine IL-7 was added at 6 ng/ml final concentration.

To determine the mechanism by which CD8 surface protein expression is reduced on MHC-I–signaled thymocytes, we introduced a CD8α.1 transgene into MHC-II^o mice so that MHC-I–signaled thymocytes would express two different CD8α molecules, i.e., endogenously encoded CD8α.2 and transgenically encoded CD8α.1 molecules (29, 30). Endogenously encoded CD8α.2 proteins and transgene-encoded CD8α.1 proteins were expressed on the cell surface at similar levels and were identical except for the single amino acid change in the extracellular domain that is responsible for their allelism. However, the genes encoding endogenous and transgenic CD8α proteins were quite different from one another. The endogenous CD8α.2 gene utilizes CD8 promoter and CD8 regulatory elements, whereas the CD8α.1 transgene utilizes heterologous hCD2 promoter/enhancer elements (34). Thus, if MHC-I signaling initially reduces CD8 surface expression on DP thymocytes by directly affecting CD8 surface proteins, both endogenous CD8α.2 and transgenic CD8α.1 protein levels will be equally reduced as the two CD8 proteins are essentially identical. However, if MHC-I signaling initially reduces CD8 surface expression on DP thymocytes by targeting CD8 regulatory elements and down-regulating CD8 gene expression, surface expression of endogenously encoded CD8α.2 proteins will be reduced but surface expression of transgenically encoded CD8α.1 proteins will not be reduced.

Immature thymocytes that have been stimulated by intrathymic ligands to undergo positive selection up-regulate TCR surface expression to become TCR^hi cells. Gating on TCR^hi cells in MHC-II^o mice and assessing surface expression of endogenously encoded CD4 and CD8β molecules revealed two major populations of MHC-I–signaled thymocytes in both normal and CD8α transgenic mice: a transitional thymocyte population that is CD4⁺ CD8β^lo and a mature population that is CD4⁻ CD8β^hi (Fig. 2). Previous studies have documented that transitional CD4⁺ CD8^lo cells are the progeny of MHC-I–signaled DP thymocytes that then differentiate into CD8⁺ T cells (20–22). Importantly, it can be seen that surface expression of endogenous CD8α molecules paralleled that of surface CD8β molecules on both wild-type and CD8α transgenic mice, in that endogenous CD8α surface expression was low on CD8β^lo cells and high on CD8β^hi cells (Fig. 2). In sharp contrast, surface expression of transgenic CD8α molecules on CD8α transgenic thymocytes remained high on both CD8β^lo and CD8β^hi cells (Fig. 2). Thus, although intrathymic MHC-I signals in DP thymocytes reduced surface expression of endogenous CD8α molecules, MHC-I signals did not alter surface expression of transgenic CD8α molecules on the same cells. Because endogenous and transgenic CD8α proteins are essentially identical, these results indicate that intrathymic MHC-I signals do not reduce CD8α surface protein expression by inducing CD8α protein internalization.

Although transgenic CD8α molecules are essentially identical to full-length endogenous CD8α molecules, thymocytes also express a splice variant of endogenous CD8α (referred to as CD8α′) lacking a cytosolic tail (35). Consequently, we considered the unlikely possibility that intrathymic MHC-I signals might have reduced surface expression of endogenous, but not transgenic, CD8α molecules on CD8β^lo cells by selectively internalizing CD8 complexes containing a CD8α′ chain. We introduced either the full-length CD8α or tailless CD8α′ transgene into II^oCD8α^o mice that lack endogenous CD8α expression so that all surface CD8αβ complexes would contain only CD8α transgenic molecules (Fig. 3). Gating on TCR^hi cells and assessing surface expression of endogenously encoded CD4 and CD8β molecules revealed the same two major populations of MHC-I–signaled thymocytes in both CD8α and CD8α′ transgenic mice: a transitional thymocyte population that was CD4⁺8β^lo and a mature population that was CD4⁻8β^hi. Importantly, surface expression of either full-length or tailless CD8α transgenic molecules was only slightly reduced on CD8β^lo thymocytes relative to CD8β^hi thymocytes (Fig. 3). These results exclude the unlikely possibility that intrathymic MHC-I signals reduce surface CD8 protein expression by selectively internalizing CD8α′ complexes.

Because surface CD8αβ heterodimers bind to MHC-I molecules with greater avidity than do CD8αα homodimers (30), it was conceivable that intrathymic MHC-I engagements might result in selective internalization of CD8αβ complexes but not CD8αα complexes. Consequently, we considered that intrathymic MHC-I signals might have failed to reduce surface expression of transgenic CD8α molecules because transgenic CD8α molecules were expressed as CD8αα homodimers whereas endogenous CD8α molecules were expressed as CD8αβ heterodimers. To assess this possibility, we introduced the CD8α transgene into II^oCD8β^o mice so that the only CD8 complexes thymocytes could express would be CD8αα complexes composed of endogenous and/or transgenic CD8α molecules (Fig. 4). Gating on TCR^hi cells and assessing surface expression of endogenously encoded CD4 and CD8α.2 molecules again revealed two major populations of MHC-I–signaled thymocytes: a transitional thymocyte population that was CD4⁺ CD8α.2^lo and a mature population that was CD4⁻ CD8α.2^hi (Fig. 4). Even though all surface CD8 complexes on these thymocytes could only be CD8αα homodimers, surface expression of endogenous CD8α.2 molecules was obviously reduced on CD8α.2^lo thymocytes, whereas surface expression of transgenic CD8α.1 molecules was not reduced on the very same CD8α.2^lo cells (Fig. 4). The detection of TCR^hi transitional CD4⁺ CD8α.2^lo thymocytes in II^oCD8β^o mice directly excludes the possibility that intrathymic MHC-I signals fail to reduce surface expression of CD8αα homodimers.

These results demonstrate that MHC-I–signaled CD4⁺8^lo thymocytes display reduced surface expression of all endogenously encoded CD8 proteins (CD8α, CD8α′, and CD8β), but not any transgene-encoded CD8 proteins (CD8α, CD8α′). Because endogenously encoded and transgene-encoded CD8α proteins are essentially identical, these results demonstrate that CD8 protein internalization cannot be the mechanism by which intrathymic MHC-I signals reduce CD8 surface protein expression and convert positively selected DP thymocytes into transitional CD4⁺8^lo cells.

Having excluded CD8 protein internalization as the basis for reduced CD8 surface protein expression on MHC-I–signaled thymocytes, we next considered whether CD8 protein externalization might be the explanation. It has been proposed that TCR-signaled DP thymocytes undergo a complex set of changes in coreceptor surface protein expression that occur independently of, and before, changes in coreceptor gene expression (9, 23). In this proposal, TCR-signaled DP thymocytes initially remove both CD4 and CD8 proteins from the cell surface to phenotypically convert into CD4^lo8^lo thymocytes that then phenotypically become CD4⁺8^lo cells because CD4 proteins are more rapidly reexpressed on the cell surface than CD8 proteins. Thus, transgenic CD8α proteins might also be more rapidly reexpressed on the cell surface than endogenous CD8α proteins, explaining why CD4⁺8^lo cells expressed low surface levels of endogenous CD8α proteins but high surface levels of transgenic CD8α proteins (Figs. 2–4). Consequently, we compared relative reexpression rates of transgenic and endogenous CD8 surface proteins on thymocytes after surface coreceptor proteins had been stripped away by treatment with the extracellular protease, pronase. We did this in two different ways. We used allele-specific anti-CD8α mAbs to determine reexpression rates of endogenous CD8α.2 and transgenic CD8α.1 proteins on DP thymocytes expressing both CD8α proteins (Fig. 5 A), and we used a pan-CD8α mAb to determine reexpression rates of endogenous CD8α.2 and transgenic CD8α.1 proteins on DP thymocytes expressing only one or the other CD8α protein (Fig. 5 B). In both experimental situations, CD4 proteins were reexpressed on the cell surface at a faster rate than CD8 proteins and could be detected on the cell surface first (Fig. 5, A and B), as originally described (23). However, endogenous and transgenic CD8α proteins were reexpressed on the cell surface at identical rates, regardless of how this experiment was performed (Fig. 5, A and B). Consequently, for differential reexpression rates of coreceptor proteins to be the explanation for MHC-I–signaled CD4⁺8^lo thymocytes, CD4⁺8^lo thymocytes would have to be low for expression of both endogenous and transgenic CD8α proteins as their reexpression rates are identical. But, despite identical reexpression rates, MHC-I–signaled CD4⁺8^lo thymocytes only exhibited low surface levels of endogenously encoded CD8α proteins but exhibited high surface levels of transgenically encoded CD8α proteins (Figs. 2–4). Thus, the phenotypic appearance of MHC-I–signaled thymocytes as CD4⁺8^lo cells is not due to differential rates of coreceptor protein externalization.

We conclude that the appearance of MHC-I–signaled thymocytes as CD4⁺8^lo cells is neither due to rapid internalization nor slow externalization of CD8 surface proteins. Rather, the difference in expression of endogenous and transgenic CD8α molecules on MHC-I–signaled CD4⁺8^lo thymocytes must reflect the impact of TCR signals on endogenous CD8 regulatory elements that are not present in the hCD2-based transgene and result in the selective down-regulation of endogenous CD8 gene expression.

If MHC-I signals down-regulate CD8 gene expression in positively selected DP thymocytes thereby causing them to become transitional CD4⁺8^lo cells before their terminal differentiation into CD8⁺ T cells, MHC-I–signaled CD4⁺8^lo thymocyte populations should contain two distinct subpopulations: one subpopulation consisting of signaled (i.e., TCR^hi) thymocytes that are transcriptionally CD4⁺8⁻, and the other subpopulation consisting of signaled TCR^hi cells that are the immediate progeny of CD4⁺8⁻ cells that have just undergone coreceptor reversal and become transcriptionally CD4⁻8⁺. To experimentally identify such subpopulations among MHC-I–signaled CD4⁺8^lo thymocytes, we subjected sorted CD4⁺8^lo thymocytes to the coreceptor reexpression assay in which they were treated with extracellular pronase to remove preexisting CD4 and CD8 surface coreceptor proteins and cultured overnight to allow surface reexpression of the coreceptor molecules actively being transcribed (Fig. 6; reference 22). Note that we have previously documented that the coreceptor reexpression assay reveals the coreceptor genes that individual thymocytes are actively transcribing and the coreceptor mRNAs that they contain (22, 24). Applying the coreceptor reexpression assay, we found that CD4⁺8⁻ and CD4⁻8⁺ TCR^hi subpopulations were both present among electronically sorted CD4⁺8^lo thymocytes from MHC-II^o mice (Fig. 6, populations a and c). Also unavoidably included in the CD4⁺8^lo sorting gate were unsignaled (i.e., TCR^lo) DP thymocytes that reexpressed both CD4 and CD8 (Fig. 6, population b), and unsignaled TCR^lo DP thymocytes that were presumably in the early stages of apoptosis and so reexpressed neither CD4 nor CD8 (Fig. 6). Importantly, both of these latter subpopulations were TCR^lo, indicating that they had not been signaled in vivo (Fig. 6). Thus, this experiment confirms that MHC-I–signaled TCR^hi thymocytes that phenotypically appear as CD4⁺8^lo cells actually consist of two TCR^hi subpopulations that, as revealed by the coreceptor reexpression assay, are CD4⁺8⁻ and CD4⁻8⁺.

Consequently, we asked if in vivo MHC-I–signaled TCR^hi CD4⁺8^lo thymocytes that reexpressed only CD4 coreceptor proteins and therefore were CD4⁺8⁻ for coreceptor gene expression still retained the potential to differentiate into CD4⁻8⁺ cells. To address this question, we used MHC-II^o mice containing a reporter transgene (referred to as Line 30; reference 31) that is composed of hCD2 cDNA under the control of CD4 enhancer and silencer elements and therefore faithfully reports in vivo CD4 transcription by surface expression of hCD2 reporter protein. We obtained in vivo–signaled CD4⁺8^lo thymocytes from such MHC-II^o Line 30 mice by electronic cell sorting (Fig. 7, a and b), identified the subpopulation that was transcriptionally CD4⁺8⁻ in the coreceptor reexpression assay (Fig. 7, d), and purified cells that were CD4⁺8⁻ hCD2⁺ by further electronic sorting (Fig. 7, c). The purified cells were then placed into short-term cultures with IL-7, a cytokine present in the normal thymus, to maintain cell viability (24, 26). Unlike sorted CD4⁺8⁺ cells that remained DP in culture (Fig. 7, g and i), most CD4⁺8⁻ hCD2⁺ cells underwent a dramatic change, terminating CD4 gene transcription (as indicated by absent CD4 and absent hCD2 reexpression) and reinitiating CD8 gene expression (as revealed by CD8 reexpression; Fig. 7, f and h). Thus, most in vivo MHC-I–signaled CD4⁺8⁻ thymocytes underwent coreceptor reversal in vitro, converting into CD4⁻8⁺ cells. A minority of CD4⁺8⁻ hCD2⁺ cells did not change coreceptor gene expression in culture (Fig. 7, f and h).

We conclude that intrathymic MHC-I–signaled DP thymocytes phenotypically convert into CD4⁺8^lo cells because they have transiently terminated CD8 gene expression to become CD4⁺8⁻ cells. We also conclude that many such CD4⁺8⁻ cells remain lineage uncommitted as they retain the potential to reverse coreceptor gene expression and to differentiate into CD4⁻8⁺ cells.

Most models of lineage commitment incorporate the concept that lineage commitment occurs in DP thymocytes and results in permanent silencing of the opposite coreceptor gene (9, 10). From this perspective the paradoxical appearance of MHC-I–signaled DP thymocytes as CD4⁺8^lo intermediates would be due to removal or redistribution of CD8 surface proteins, and not due to down-regulation of CD8 gene expression. However, this study documents that removal or redistribution of CD8 surface proteins is not the basis for conversion of MHC-I–signaled DP thymocytes into CD4⁺8^lo cells, but rather that it is due to down-regulation of CD8 gene expression. This study also demonstrates that MHC-I–signaled thymocytes that have down-regulated CD8 gene expression to become CD4⁺8⁻ cells still retain the potential to differentiate into CD8⁺ T cells. Thus, initial cessation of CD8 gene expression does not imply commitment to the CD4⁺ T cell lineage, as it occurs in MHC-I–signaled thymocytes during differentiation into CD8⁺ T cells.

That many MHC-I–signaled DP thymocytes phenotypically convert into CD4⁺8^lo cells during differentiation into CD8SP T cells was a surprising observation made independently by two different laboratories (21, 22) that has since been confirmed by others (18, 19, 23, 36). Importantly, MHC-I–signaled CD4⁺8^lo thymocytes are not dead-end cells, but are cells on their way to differentiating into CD8SP T cells (18, 21, 22). Nevertheless, the molecular basis for this observation has been uncertain. Two explanations have been proposed that can reconcile the appearance of MHC-I–signaled CD4⁺8^lo thymocytes with the concept that lineage commitment occurs in DP thymocytes and results in permanent silencing of the opposite coreceptor gene. The first explanation posits that MHC-I–specific TCR interactions signal DP thymocytes to internalize surface CD8 coreceptor proteins even as they induce DP thymocytes to terminate CD4 gene expression (22). The second explanation is more complex and posits that MHC-I–specific TCR interactions signal DP thymocytes to remove both CD4 and CD8 coreceptor proteins from the cell surface that are then reexpressed at different rates, with the effect that MHC-I–signaled DP thymocytes transiently appear as CD4⁺8^lo cells though they continue to actively express both coreceptor genes (9, 23). Both explanations have now been excluded by this study.

This study used an hCD2-based CD8α.1 transgene that is expressed on thymocytes at similar levels to that of endogenous CD8α.2 proteins. Endogenous and transgenic CD8α proteins are essentially identical, but the genes encoding them are controlled by entirely different regulatory elements, as expression of endogenous CD8α genes is controlled by endogenous CD8 promoter/enhancer elements (37–42) whereas expression of the CD8α transgene is controlled by heterologous hCD2 promoter/enhancer elements (29, 34). This study found that MHC-I–signaled CD4⁺8^lo thymocytes expressed low surface levels of endogenously encoded CD8α proteins but expressed high surface levels of transgenically encoded CD8α proteins, an observation that is not consistent with a direct effect of MHC-I signals on CD8α surface protein expression. Nevertheless, to document that internalization of CD8 surface complexes was not the mechanism by which MHC-I–signaled thymocytes appeared in vivo as CD4⁺8^lo cells, we evaluated and excluded the possibility that MHC-I signals internalized only specific subsets of CD8 surface protein complexes, such as those containing only tailless CD8α′ proteins, those consisting only of CD8αβ heterodimeric complexes, or only of CD8αα homodimeric complexes. And to document that slow externalization of CD8 protein complexes was not the mechanism by which MHC-I–signaled thymocytes appeared in vivo as CD4⁺8^lo cells, we evaluated and excluded the possibility that transgenic CD8α proteins, like CD4 proteins, might be reexpressed on the cell surface at a faster rate than endogenous CD8α proteins (23). Consequently, regardless of whether MHC-I signals stimulate the removal and/or reexpression of surface coreceptor proteins, this study demonstrates that such events are not the mechanism by which MHC-I–signaled thymocytes convert in vivo into CD4⁺8^lo thymocytes. Rather, the appearance of MHC-I–signaled thymocytes as CD4⁺8^lo cells must result from transient down-regulation of endogenous CD8 gene expression.

Supporting this conclusion, we found that in vivo MHC-I–signaled CD4⁺8^lo thymocytes consisted of two subpopulations of TCR^hi cells with different patterns of coreceptor gene expression: one subpopulation that had indeed down-regulated CD8 gene expression to become CD4⁺8⁻, and the other subpopulation whose coreceptor gene expression was CD4⁻8⁺. That the transitional CD4⁺8^lo population contained TCR^hi thymocytes with a coreceptor gene expression pattern of CD4⁺8⁻ was concordant with our conclusion that MHC-I–signaled thymocytes phenotypically appeared as CD4⁺8^lo cells because of down-regulated CD8 gene expression. Indeed, this study demonstrates that despite having ceased CD8 gene expression, such CD4⁺8⁻ cells could undergo coreceptor reversal into CD4⁻8⁺ cells in the presence of IL-7, a cytokine that is present within the normal thymus (43). As a result, TCR^hi cells that are transcriptionally CD4⁻8⁺ and in the process of differentiating into mature CD8⁺ T cells are also present within the transitional CD4⁺8^lo thymocyte population because they are the immediate progeny of cells that were transcriptionally CD4⁺8⁻ before undergoing coreceptor reversal. Thus, cessation of CD8 gene expression in in vivo–signaled thymocytes does not necessarily imply CD4 lineage commitment, as such cells may undergo coreceptor reversal and differentiate into CD8⁺ T cells (25).

It seems to us that the present results cannot be easily reconciled with the concept that lineage commitment occurs in DP thymocytes and results in permanent silencing of the opposite coreceptor gene. Cessation of coreceptor gene expression was originally equated with commitment to the opposite coreceptor lineage (6, 16). For example, the original instructional model postulated that MHC-specific TCR plus coreceptor engagements dictated which coreceptor gene was silenced, whereas stochastic models postulated that the selection of which coreceptor gene to silence occurred randomly. Updated versions of the original instruction model no longer equate cessation of coreceptor gene expression with lineage commitment, but postulate that the strength or duration of the TCR signal in DP thymocytes commits the cell to one coreceptor lineage and that lineage choice is revealed by its subsequent silencing of the opposite coreceptor gene (11, 14, 15, 44). Importantly, cessation of coreceptor gene expression is still considered to reflect an individual cell's “commitment” to the opposite coreceptor lineage. In contrast, the present results are consistent with the kinetic signaling perspective that intermediate CD4⁺8⁻ thymocytes are lineage-uncommitted cells in which lineage commitment then occurs. In the kinetic signaling model, transient cessation of CD8 gene expression occurs during differentiation of MHC-I–signaled DP thymocytes and results in cessation of CD8-dependent MHC-I–specific signaling which, in turn, promotes coreceptor reversal and differentiation into CD8SP T cells (24, 25). However, not all MHC-I–signaled thymocytes go through a stage where they appear as CD4⁺8^lo cells. For example, DP thymocytes that express TCR with low apparent affinity for intrathymic MHC-I ligands appear to differentiate directly into CD8SP T cells without ever appearing as CD4⁺8^lo cells (36, 45). It is possible that such MHC-I–signaled cells do transiently down-regulate CD8 gene expression as proposed by the kinetic signaling model, but their TCR signals are of such short duration as a result of low ligand affinity that these cells undergo coreceptor reversal and reinitiate CD8 gene expression before surface CD8 protein levels can detectably decline (25). Whether or not signals transduced by low affinity MHC-I–specific TCR in fact induce transient down-regulation of CD8 gene expression before undergoing lineage commitment remains to be determined.

This study demonstrates that the appearance of MHC-I–signaled CD4⁺8^lo thymocytes is due to down-regulation of CD8 gene expression rather than to effects on CD8 surface proteins, but it is possible that MHC-I–specific TCR signals may affect CD8 mRNA stability as well as CD8 transcription. Our present observation that MHC-I–specific TCR signals cause DP thymocytes to discontinue expression of all endogenously encoded CD8 genes (CD8α and CD8β) without affecting expression of CD8 transgenes indicates that TCR signals target regulatory elements present within the endogenous CD8 gene locus that are absent from the hCD2-based transgene. Indeed, in vitro observations from this laboratory have previously demonstrated that antibody-mediated TCR engagement of purified DP thymocytes selectively terminates endogenous CD8 gene transcription and modestly destabilizes both CD4 and CD8 coreceptor mRNAs (46, 47). We think that similar events are likely occurring in in vivo–signaled DP thymocytes.

Although our understanding of CD4 and CD8 gene transcription is still far from complete, current knowledge of how CD8 and CD4 genes are transcriptionally regulated provides some clues as to how TCR signaling in DP thymocytes might selectively result in transient cessation of CD8 gene expression and conversion of signaled DP thymocytes into CD4⁺8⁻ cells. To oversimplify the situation in developing thymocytes, lineage-specific expression of CD8 and CD4 coreceptor genes appears to be regulated by fundamentally opposite mechanisms, as CD8 gene expression is specifically up-regulated by activation of stage-specific CD8 enhancer elements (37–42) and CD4 gene expression is specifically down-regulated by activation of a CD4 silencer element (31, 48, 49). As a result, the “basal” or “default” state of coreceptor gene expression in developing thymocytes might be considered to be CD4⁺8⁻, as this represents a transcriptional state in which neither CD8 enhancer elements nor CD4 silencer elements are activated. However, enhancer usage within the CD8 gene locus is known to shift during differentiation of immature DP thymocytes into mature CD8⁺ T cells, as “immature” CD8 enhancer elements drive CD8 gene expression at the immature DP thymocyte stage of differentiation and “mature” CD8 enhancer elements drive CD8 gene expression at the mature CD8⁺ T cell stage of differentiation. Thus, TCR signals can induce the basal transcriptional state in DP thymocytes by simply terminating activation of immature CD8 enhancer elements. It is interesting to further speculate (26) that TCR signals in developing thymocytes might block activation of all lineage-specific regulatory elements (i.e., CD8 enhancer and CD4 silencer elements) with the result that persistent TCR signaling would maintain CD4 gene expression “on” and CD8 gene expression “off” (i.e., CD4⁺8⁻), resulting in thymocyte differentiation into CD4⁺ T cells as proposed by the kinetic signaling model.

In conclusion, this study demonstrates that the appearance of MHC-I–signaled thymocytes as CD4⁺8^lo cells results from transient cessation of CD8 gene expression. Indeed, this study demonstrates that initial cessation of CD8 gene expression does not imply commitment to the CD4⁺ T cell lineage, as many in vivo MHC-I–signaled CD4⁺8⁻ cells are able to undergo coreceptor reversal and differentiate into CD8⁺ T cells. Thus, these observations support the kinetic signaling model and challenge the concept that lineage commitment occurs in DP thymocytes and results in permanent silencing of the opposite coreceptor gene.

We thank Drs. Jon Ashwell and B.J. Fowlkes for their critical readings of the manuscript.