γδ Lineage Commitment

Immunity, Vol. 4, 37–45, January, 1996, Copyright 1996 by Cell Press

T Cell Receptor d Gene Rearrangement and T Early a (TEA) Expression in Immature ab LineageThymocytes: Implications for ab/gd Lineage Commitment Anne Wilson,* Jean-Pierre de Villartay,† and H. Robson MacDonald* *Ludwig Institute for Cancer Research Lausanne Branch University of Lausanne Epalinges Switzerland † Institut National de la Sante ´ et de la Recherche Me´dicale U429 Hoˆpital Necker Paris France

Summary Mature T cells comprise two mutually exclusive lineages expressing heterodimeric ab or gd antigen receptors. During development, b, g, and d genes rearrange before a, and mature gd cells arise in the thymus prior to ab cells. The mechanism underlying commitment of immature T cells to the ab or gd lineage is controversial. Since the d locus is located within the a locus, rearrangement of a genes leads to deletion of d. We have examined the rearrangement status of the d locus immediately prior to a rearrangement. We find that many thymic precursors of ab cells undergo VDJd rearrangements. Furthermore, the same cells frequently coexpress sterile T early a (TEA) transcripts originating 39 of Cd and 59 of the most upstream Ja, thus implying that individual ab lineage cells undergo sequential VDJd and VJa rearrangements. Finally, VDJd rearrangements in immature ab cells appear to be random, supporting models in which ab lineage commitment is determined independently of the rearrangement status at the TCR d locus. Introduction The generation of T lineage cells in the thymus from a precommitted bone marrow–derived precursor cell follows a complex developmental program of sequential gene expression and T cell receptor (TCR) gene rearrangements (reviewed by Fowlkes and Pardoll, 1989; Rothenberg, 1992). This leads to the expression at the surface of one of two types of TCR composed of a pair of somatically rearranged a and b or g and d heterodimers in conjunction with the CD3 signaling complex (reviewed by Ashwell and Klausner, 1990; Borst et al., 1993). In the thymus, TCR g, d, and b genes become transcriptionally active prior to TCR a, and during ontogeny, T cells expressing gd TCRs appear in the fetal thymus several days before cells expressing ab TCRs (Pardoll et al., 1987). In addition, the location of the TCR d locus within the TCR a locus (between the Va and Ja gene segments) ensures that on any given chromosome the entire TCR d region is deleted upon TCR a rearrangement (Chien et al., 1987). These findings imply the existence of a highly efficient regulatory mechanism

by which the TCR a and TCR d genes are independently rearranged and expressed, and by which commitment to either the ab or gd lineage may be effected. Such a mechanism should, in particular, preclude TCR a rearrangement in gd lineage cells. Strong evidence for the independence of ab and gd lineages has been obtained in mutant mice in which TCR constant (C) region genes have been rendered nonfunctional by homologous recombination. In TCR a 2/ 2 or TCR b2/2 mice (Mombaerts et al., 1992; Philpott et al., 1992), gd lineage cells develop normally, while mature ab lineage cells are completely absent. Conversely, in TCR d 2/ 2 mice, ab lineage cells develop normally, while gd lineage cells are absent (Itaharo et al., 1993). These experiments formally demonstrate that ab and gd lineage cells can develop independently. However, they do not directly address the mechanism of lineage commitment during normal T cell development. Several models of ab/gd lineage commitment have been proposed. The successive rearrangement model proposes that initial rearrangements occur in all immature T cells at the TCR g and TCR d loci and, if productive, commit the cells to the gd lineage. However, if either of these rearrangements are nonproductive the cell goes on to attempt rearrangements at the TCR b and TCR a loci, thereby committing it to the ab lineage (Pardoll et al., 1987). A second model is based on the existence of specific silencer elements (Winoto, 1991; Haas and Tonegawa, 1992). According to this hypothesis, ab and gd T lineages arise independently from precommitted precursors, and lineage commitment is assured by the activation of specific silencers operating on either TCR g (for ab lineage) or TCR a (for gd lineage) genes. While the TCR g silencer acts at the transcriptional level (Ishida et al., 1990), the TCR a silencer (Winoto and Baltimore, 1989b) may restrict accessibility of DNA to the recombinase machinery, as initially proposed by Yancopoulos and Alt (1986). A third model (de Villartay and Cohen, 1990) postulates that a programmed deletion event occurs at the TCR d locus in cells committed to the ab lineage. In the human thymus, a dRec element upstream of the TCR d locus frequently recombines to cJa, the most 59 of the Ja segments, thereby deleting the Dd, Jd, and Cd regions (de Villartay et al., 1988). Three murine homologs of dRec have been isolated. However, these elements rearrange less frequently to cJa. Instead, they are joined either to Dd2 (which may then be joined to cJa), or less often to Jd1 (Arden, 1992). In either case, the formation of a functional d chain would be impossible. A further consequence of these dRec recombinations is that the Va region located upstream of dRec is brought into proximity to the Ja locus, thereby potentially enhancing the efficiency of subsequent Va–Ja rearrangement. In attempting to distinguish between these models, it is noteworthy that very little information is available regarding the status of the TCR a/d locus at distinct developmental stages in the normal thymus. Such information would be relevant, since each of the three lineage

Immunity 38

commitment models makes distinct predictions as to the configuration of the TCR d locus prior to TCR a rearrangement in ab lineage cells. According to the successive rearrangement model, nonproductive VDJd rearrangements should be overrepresented prior to VJa rearrangements in ab lineage cells. In contrast, the silencer model would be permissive for random VDJd rearrangements in immature ab cells because the protein product of a TCR d rearrangement could never be expressed in the absence of TCR g. Finally, the programmed deletion model would not predict the occurrence of frequent VDJd rearrangements in immature ab cells, since most such rearrangements would delete dRec elements postulated to be involved in the deletional event. In view of these considerations, we decided to examine the extent and nature of TCR a/d transcripts in developing thymocytes committed to the ab lineage. In the adult thymus, CD4lo stem cells (the earliest identifiable intrathymic precursors) progress along a developmental pathway via a series of discrete CD42CD82 doublenegative (DN) subsets principally characterized by the differential expression of the surface glycoproteins CD24 (HSA), CD44 (Pgp1), and CD25 (interleukin-2Ra [IL-2Ra]), prior to the expression of the coreceptor molecules CD4 and CD8 (Wilson et al., 1988; Pearse et al., 1989; Egerton et al., 1990; Nikolic-Zugic, 1991; Scollay, 1991; Shortman, 1992; Godfrey and Zlotnik, 1993; Kruisbeek, 1993). Upon intrathymic transfer, the CD4lo precursor and each of the DN subsets can give rise to cells of both ab and gd lineages (Wu et al., 1991a, 1991b; Shortman et al., 1991). In addition, some cells from the latest DN subset in the developmental pathway can differentiate into gd expressing cells in vitro (Petrie et al., 1992). Based on these results, it appears that the precursors of gd cells do not progress beyond the late DN stage consistent with the fact that mature thymic gd cells have a DN phenotype. In contrast, the CD41CD81 double-positive (DP) thymocyte subset is generally accepted to be committed to the ab lineage and undergoes subsequent selection and maturation to mature CD41 or CD81 single-positive (SP) TCR ab cells. Previous studies have demonstrated the existence of a rapidly cycling intermediate between the late DN and the DP subsets (Paterson and Williams, 1987; MacDonald et al., 1988; Shortman et al., 1988). These cells resemble the late DN in that they are CD3lo, HSAhi, CD252, and CD442, but as they express either CD4 or CD8 in the absence of a mature TCR they are called immature single positives (ISP). Since the transition from ISP to DP is rapid and occurs in the apparent absence of stimuli or growth factors in simple culture as well as in vivo (Guidos et al., 1989; Petrie et al., 1990, 1992; Wilson et al., 1989), ISP are assumed to be part of the ab lineage. We show here that ISP thymocytes express no detectable TCR a transcripts but include a high proportion (approximately 50%) of cells expressing full-length (VDJC) TCR d mRNA. Moreover, most TCR d expressing ISP thymocytes coexpress sterile transcripts originating from the TEA region located 39 of Cd and 59 of the most upstream Ja. Since TEA transcription has been postulated to be involved in the opening of the TCR a

Figure 1. TCR Gene Expression in Subsets of Immature and Mature T Cells Thymocyte subsets and lymph node T cells were prepared as described in Experimental Procedures. Both CD251 and CD252 DN subsets were sorted to be CD3lo HSA1CD44lo. Total RNA (4 mg) was loaded into each lane. The same blot was sequentially hybridized with the indicated probes. Exposure times were as follows: 48 hr for Ca and Cd; 18 hr for Cb; 24 hr for Cg; and 6 hr for b-actin. TCR b transcripts in SP cells, which are expressed at much lower levels than in DP thymocytes (Maguire et al., 1990), could readily be detected in longer exposures (data not shown).

locus for subsequent rearrangement (de Villartay et al., 1987; de Villartay and Cohen, 1990; Shimizu et al., 1993), our data strongly suggest that individual thymocytes of the ab lineage can undergo sequential VDJd and VJa rearrangements at the TCR a/d locus. The implications of these findings for models of ab/gd lineage commitment will be discussed. Results TCR mRNA Expression by Immature Thymocyte Subsets To define more precisely the stage of T cell development where ab and gd lineages may diverge, we first purified subsets of immature and mature thymocytes from adult mouse thymus by FACS sorting in sufficient quantities to perform Northern blot analysis with TCR constant region probes. As shown in Figure 1, transcripts of TCR

Expression of TCRd and TEA in Immature ab T Cells 39

b, g, and d genes were strongly expressed at the CD251 DN stage. As expected, both short and longer transcripts (corresponding to [D]J and V[D]J rearrangements, respectively) were observed for each of these TCR loci at this early stage. In subsequent CD25 2 DN, ISP, and DP stages, TCR b transcripts remained at high levels, whereas TCR g transcripts rapidly decreased, consistent with the hypothesis that mature gd cells are derived from DN precursors, whereas ab lineage cells progress through to the ISP and DP stages. TCR a transcripts were undetectable at early stages of development but were first strongly expressed in the DP thymocyte subset. These TCR a transcripts were exclusively full length (corresponding to VJa rearrangements). Our data are inconsistent with earlier reports, suggesting that TCR a transcripts are already present at moderate levels in late DN stages (Pearse et al., 1989; Nikolic-Zugic and Moore, 1989); however, it is possible that low levels of mature (surface ab-expressing) DN cells contaminated their purified preparations. Interestingly, ISP thymocytes that are known to be a transitional population between DN and DP cells, both in vivo and in vitro (MacDonald et al., 1988; Guidos et al., 1989; Wilson et al., 1989; Petrie et al., 1990; Hugo and Petrie, 1992), expressed high levels of TCR d transcripts but no detectable TCR a or TCR g transcripts, confirming previous in situ hybridization studies with TCR constant region probes (Held et al., 1990; Wilson et al., 1994). Although TCR g mRNA was not detected in ISP thymocytes by Northern blot, frequent Vg–Jg rearrangements were seen in ISP DNA by polymerase chain reaction (PCR) (data not shown). Significantly, full-length (VDJC) TCR d transcripts were expressed at high levels in the ISP thymocyte subset, and these transcripts disappeared at the DP stage concomitant with the appearance of VJCa transcripts. These Northern blot data are consistent with the possibility that successive VDJd and VJa rearrangements can occur in immature cells of the ab lineage. TEA Transcripts Are Selectively Expressed in the ISP Subset TEA is a genetic region located 39 of the Cd gene and 59 of cJa, the most upstream Ja element (de Villartay et al., 1987). It has been postulated that transcription of TEA is an obligatory early event in the opening of the TCR a locus for subsequent VJa rearrangement (reviewed by de Villartay and Cohen, 1990). To test this hypothesis, we therefore examined the expression of TEA by Northern blot in thymus subsets. Although TEA was rather weakly expressed, transcripts were only detectable in the ISP subset (Figure 2). This distribution of TEA expression provides the strongest evidence to date that TEA transcription occurs immediately prior to VJa rearrangement (and concomitant Ca expression), which can only be detected at the subsequent DP stage (see Figure 1). To characterize further TEA and TCR d expression in ISP thymocytes at the single cell level, FACS-sorted ISP cells were prepared as shown in Figure 3, and fused directly with the BW5147a2b 2 thymoma (Born et al., 1988). This thymoma has two rearranged TCR a alleles and thus has deleted both the TCR d and TEA loci on

Figure 2. Expression of TEA in Thymus Subsets Subsets and controls are as in Figure 1. Exposure times were 2 weeks for TEA and overnight for b-actin.

both chromosomes, thereby facilitating analysis of expression of these genes in the hybrids. Of 37 ISP hybrids produced, the vast majority (30 of 37) expressed TEA transcripts (Figure 4; data not shown). These TEAexpressing ISP hybridomas included a high proportion (about 50%) with one or two fully rearranged TCR d genes, whereas the remainder were mostly characterized by the presence of short DJCd transcripts. In contrast, no TEA transcripts were detectable in the BW5147 fusion partner (Figure 4) or in a panel of 20 control T–T hybridomas derived from FACS-sorted TCR gd1 thymocytes (data not shown). These latter data confirm earlier studies indicating that TEA is not transcribed in mature gd T cells (reviewed by de Villartay and Cohen, 1990). Furthermore, they formally exclude the possibility that TEA expression is induced in the ISP hybrids by transactivating factors produced by BW5147. Collectively, these data are compatible with the hypothesis that ISP thymocytes represent transitional ab lineage cells that undergo sequential rearrangements at the TCR a/d locus.

Figure 3. FACS Sorting of ISP Thymocytes CD42CD252 thymocytes were stained with anti-CD8–Red613, antiHSA–PE, and anti-CD3–fluorescein isothiocyanate and the CD81CD3lo HSAhi (ISP) cells sorted. The sorting gate is indicated and on reanalysis after sorting the cells were >98% pure.

Immunity 40

Figure 4. TCR d and TEA Expression by ISP Thymocyte Hybridomas Total cell RNA was prepared and 5 mg loaded into each lane. Hybridization was performed sequentially with the TEA probe, the Cd probe, and then the b-actin probe. Exposure times were as follows: 5 days for TEA; overnight for Cd; 6 hr for b-actin.

TCR d Transcripts in Mature T Cells from TCR ab Transgenic Mice In normal mature ab T cells, the TCR d locus is usually deleted on both chromosomes as a consequence of TCR a rearrangements (Malissen et al., 1992). However, in certain TCR ab transgenic mice, the early expression of the transgenic TCR during thymus development prevents endogenous TCR a rearrangement in positively selected cells, presumably owing to down-regulation of the recombinase-activating genes RAG1 and RAG2 upon TCR engagement at the DP stage (Borgulya et al., 1992; Bra¨ndle et al., 1992). If our hypotheses were correct, one might predict that the TCR d locus in positively selected transgenic ab cells would remain frozen in the rearranged configuration normally present in ISP thymocytes just prior to TCR a rearrangement. We therefore examined TCR d transcripts in mature lymph node T cells of ab transgenic mice and their nontransgenic littermates. The transgenic TCR expressed by these mice is positively selected to the CD8 lineage in the presence of the MHC class I molecule H-2Db (Pircher et al., 1989). Furthermore, mature TCR transgenic T cells do not rearrange endogenous TCR a genes, since several CD81 clones derived from these mice did not delete the Cd locus as determined by Southern blot (Ohashi et al., 1990). As shown in Figure 5, lymph node cells from H-2 b TCR ab transgenic mice expressed high levels of TCR d transcripts, which were undetectable in littermate controls. These TCR d transcripts originated from the positively selected CD8 subset, since in situ hybridization analysis of sorted cells (Table 1) revealed high levels of Cd expressing CD81 cells but negligable values in the CD4 subset, consistent with the fact that mature CD41 cells in these transgenic mice have rearranged endogenous TCR a genes (Crompton et al., 1992). Importantly, fully rearranged (VDJCd) and short (DJCd) TCR d transcripts in the mature ab transgenic cells were present at the same ratio as in ISP thymocytes from normal mice (cf Figure 5 and Figure 1), as would be expected if the TCR d locus in the transgenic cells were frozen at the ISP stage.

Sequence Analysis of VDJd Rearrangements among ISP Thymocytes As mentioned earlier, the status of VDJd rearrangements in immature ab lineage cells has important implications for the mechanism of ab/gd lineage commitment. We therefore PCR-amplified reverse-transcribed RNA isolated from the 22 ISP hybridomas that expressed fulllength VDJCd transcripts using a panel of Vd primers. In addition, Vd–Cd PCR products were amplified from cDNA prepared from freshly sorted ISP thymocytes (see Figure 3) and cloned into bacteria. A total of 77 Vd–Cd PCR products (representing 6 of the 8 known Vd families) were successfully sequenced. As summarized in Table 2, the overall frequency of productive (in-frame) TCR d rearrangements (29%) was close to the theoretical value (1/3) that would be expected for random VDJd joining. Discussion Control of Rearrangement of the TCR a/d Locus The data presented in this report provide strong evidence that sequential VDJd and VJa rearrangements at the TCR a/d locus can occur in individual cells of the ab lineage during thymus ontogeny. This conclusion is supported by the observed coexpression of VDJCd and sterile TEA transcripts in normal ISP thymocytes and in a large fraction (approximately 50%) of T cell hybridomas derived from ISP cells, a subset that is known to contain the immediate precursors of ab DP thymocytes (Guidos et al., 1989; Wilson et al., 1989). TEA transcription has been postulated to be necessary for (or associated with) the opening of the Ja locus for subsequent VJa rearrangement (de Villartay and Cohen, 1990). However, although TEA transcripts have been observed in the fetal thymus (Shimizu et al., 1993), no direct evidence for their involvement in VJa rearrangement has been reported. In this context, our finding that TEA transcripts are temporally regulated during thymus development such that expression is only detected in the ISP subset provides compelling evidence that TEA may play a role in the opening of the Ja cluster for subsequent recombination.

Expression of TCRd and TEA in Immature ab T Cells 41

Table 1. Presence of Cd-Expressing Mature CD8SP Cells in TCR ab Transgenic Micea Sorted subset

Source of cells

Percent positive cells b Cd

Ca

Normal thymus

ISP CD4SP CD8SP

48 6 3 161 362

662 65 6 3 74 6 4

Transgenic thymus

CD4SP CD8SP

361 56 6 7

75 6 5 73 6 5

Normal lymph node

CD4SP CD8SP CD4SP CD8SP

0 262 361 44 6 5

78 78 78 75

Transgenic lymph node

6 6 6 6

4 2 2 3

a

In situ hybridization was performed on sorted thymocytes or lymph node cells from TCR ab transgenic mice or littermate controls using Cd and Ca probes as described (Held et al., 1990; Wilson et al., 1994). b For each slide, 500–1000 individual sorted cells were analyzed. Data are expressed as mean 6 SD (4 slides per group).

Figure 5. Expression of TCR d Transcripts in Mature ab T Cells from TCR Transgenic Mice Total RNA from lymph node cells of ab transgenic or littermate control mice was blotted and sequentially probed with Cd and Ca. The longer TCR a transcript originates from the transgenic cDNA construct (Pircher et al., 1989) Ethidium bromide staining is indicated as a measure of RNA loading.

Gene-targeting experiments at the TEA locus will be required to test formally this hypothesis. Previous attempts to document sequential TCR d and TCR a rearrangements at the TCR a/d locus directly have led to conflicting results. In one study, extrachromosomal circular thymocyte DNA produced as a result of VJa joining was analyzed and found to contain TCR d rearrangements (Takeshita et al., 1989). However, in another study using similar techniques, only germline TCR d genes were detected in excision circles (Winoto and Baltimore, 1989a). Although the basis of this discrepancy is not clear, it is likely that prior VDJd rearrangement is not an absolute prerequisite for VJa joining. Indeed, only 50% of the ISP hybridomas analyzed here expressed VDJd rearrangements that could theoretically be present in VJa excision circles. The occurrence of frequent VDJd rearrangements in

cells of the ab lineage is consistent with several recent studies of TCR a/d gene regulation in transgenic mice. In particular, Lauzurica and Krangel (1994a) showed that the human TCR d enhancer (located within the Jd–Cd intron) was able to activate VDJ rearrangement of a transgenic human TCR d gene minilocus. Importantly, the extent of rearrangement of the transgene was equivalent in ab and gd lineage cells, suggesting that there is no lineage-specific control of rearrangement at the TCR d locus. In contrast, the same minigene construct bearing the human TCR a enhancer (located 39 of Ca) only underwent VDJ rearrangement in ab lineage cells (Lauzurica and Krangel, 1994b). Similar results were obtained by Capone et al. (1993), who found that VDJ rearrangement of a transgenic TCR b minilocus construct under the control of the mouse TCR a enhancer was restricted to ab lineage cells. Collectively, these latter studies suggest that the activity of enhancer elements located in the TCR a locus may restrict VJa rearrangements to cells of the ab lineage. However, they in no way exclude the possibility that these VJa rearrangements are preceded by VDJd rearrangements, mediated (presumably) by the presence of an active TCR d enhancer. The relative timing of VDJd and VJa rearrangements in ab lineage cells has not been addressed in the present Table 2. Sequence Analysis of VDJd Rearrangements in ISP Thymocytes Family

Total

Productive

(percent)

Vd2 Vd3 Vd4 Vd5 Vd6 Vd7

9 8 14 11 30 5

1 4 3 2 11 1

(11) (50) (21) (18) (37) (20)

All Vd

77

22

(29)

Vd–Cd PCR products originating from ISP hybridomas (total of 22) or randomly cloned ISP thymocyte cDNA (total of 55) were sequenced. The frequency of productive (in-frame) rearrangements is indicated for each Vd family analyzed.

Immunity 42

study. Although we have chosen to focus on ISP thymocytes because of their precise developmental stage and unambiguous lineage commitment, it is likely that the VDJCd transcripts that are present in these cells represent TCR d rearrangements that have occurred earlier. Indeed, VDJCd transcripts are already detectable in early (CD251) DN thymocytes and persist throughout later DN and ISP stages (this study). Furthermore, in transgenic mice bearing a human TCR d minigene locus under the control of the TCR d enhancer, VDJ rearrangements (which are not lineage specific) occur early and appear to be completed by the late DN stage (Lauzurica and Krangel, 1994b). Thus, it seems probable that VDJd rearrangements occur simultaneously in early DN cells of both the ab and gd lineages. In contrast, the TCR a enhancer is only able to direct VDJ rearrangements of a transgenic minigene locus in ab cells, and this rearrangement occurs much later at the ISP stage (Capone et al., 1993). Accordingly, we must postulate that ab lineage early DN cells with a rearranged VDJd gene would retain this configuration at the TCR a/d locus (perhaps through several cell divisions) until activation of the TCR a enhancer at the ISP (or early DP) stage resulted in a subsequent VJa rearrangement. In the context of such a model, transcriptional downregulation of the recombinase-activating genes RAG1 and RAG2 occurs selectively in rapidly cycling late DN and ISP thymocytes (Wilson et al., 1994), and the RAG2 protein has been shown to be rapidly degraded (due to phosphorylation) in cycling cells (Lin and Desiderio, 1993). Both mechanisms may favor the stable maintainance of the TCR a/d locus in ab lineage cells between these two putative waves of rearrangement. Diaz et al. (1994) have recently proposed an integrated model for the regulation of rearrangement at the TCR a/d locus. Using transgenic mice, these authors obtained evidence for the existence of a TCR a/d locus control region (LCR) located 39 of Ca and the TCR a enhancer. According to their model, this LCR would be instrumental in opening the TCR a/d locus for subsequent rearrangement and transcription of both the TCR a and TCR d genes. In cells of the gd lineage, activity of the LCR would be restricted to the TCR d locus, since the TCR a locus would be turned off by interactions with a specific silencer mechanism (see below). In contrast, the LCR could potentially interact with either the TCR d or a locus in cells of the ab lineage, where the TCR a silencer would be inactive. Although the authors favor a variant of their model in which interactions of the LCR with the TCR a locus would occur preferentially in ab cells (due to its relative proximity), our data would suggest that such a constraint is unnecessary.

VDJd Rearrangements and Lineage Commitment As outlined in the Introduction, three models to explain T cell commitment to the ab or gd lineages have been postulated. In the successive rearrangement model (Pardoll et al., 1987), all T cell precursors initially attempt g and d rearrangements and ab lineage cells are rescued from the pool of cells that have nonproductive rearrangements at either the g or d locus. Alternatively,

putative silencers at the g (Ishida et al., 1990) or a (Winoto and Baltimore, 1989b) loci may assure commitment of T cells to the ab or gd lineage, respectively, independently of the rearrangement status of their TCR genes. Finally, the programmed deletion model (de Villartay and Cohen, 1990) holds that TCR d locus deletion is a prerequisite for commitment to the ab lineage. Whereas the first two models would be compatible with the frequent occurrence of VDJd rearrangements that we observe in the precursors of ab T cells, the latter hypothesis (at least in its strictest form) is not supported by our data. In attempting to distinguish between the successive rearrangement and silencer models of ab/gd lineage commitment, it is necessary to evaluate the quality of VDJd rearrangements in immature ab cells. Thus, the former model predicts a selection against productive (in-frame) VDJd rearrangements in ab precursors (since either g or d must be out-of-frame to allow a cell to undergo subsequent a rearrangement), whereas the latter predicts random VDJd rearrangements in these cells (since it imposes no constraints on productive rearrangement at either g or d locus). In the simplest situation (ie., where an individual precursor cell attempts rearrangements at the g and d locus on both chromosomes), the successive rearrangement model would predict a frequency of approximately 20% in-frame VDJd rearrangements in ab precursors that are derived exclusively from failed gd cells (for calculations see Dudley et al., 1995), whereas the silencer model would anticipate 33% (ie., random) in-frame VDJd rearrangements in ab cells. In this respect, our collective data (29% productive VDJd rearrangements in ISP thymocytes) are more readily compatible with the silencer model. Consistent with this interpretation, we observed that TCR g transcripts were rapidly down-regulated in ab lineage cells subsequent to the CD251 DN stage. However, it remains possible that the TCR g transcriptional silencing mechanism is only activated in ab cells following lineage commitment, where it functions to maintain (rather than initiate) the lineage decision (Sim et al., 1995). Finally, it should be noted that two very recent reports (published while this manuscript was under review) have provided evidence that the frequency of in-frame Vd4 or Vd5 rearrangements in immature or mature ab lineage cells is around 20%, thus supporting the successive rearrangement model (Dudley et al., 1995; Livak et al., 1995). Although the reason(s) for this apparent discrepancy with our data are not clear, it is noteworthy that both Dudley et al. (1995) and Livak et al. (1995) analyzed VDJd rearrangements in genomic DNA of unknown origin isolated from ab lineage cells, whereas we examined only expressed VDJd sequences. Thus, it cannot be excluded that a preferential stability of in-frame transcripts may have biased our estimate of the frequency of productive VDJCd rearrangements.

Experimental Procedures

Isolation of Thymus Subsets C57BL/6 female mice were purchased from HARLAN OLAC (Bicester, United Kingdom) and used at 4–6 weeks of age for the preparation of all thymocyte subsets, and at 12 weeks of age for the isolation

Expression of TCRd and TEA in Immature ab T Cells 43

of peritoneal macrophages. Transgenic mice on a C57BL/6 background expressing an ab TCR specific for lymphocytic choriomeningitis virus in the context of H-2Db have been described (Pircher et al., 1989). Thymus subsets were prepared as described previously (Wilson et al., 1992, 1994) by a combination of complement-mediated cytotoxicity, magnetic bead depletion, and three color FACS sorting on a FACStar Plus (Becton Dickinson, Mountain View, California). Most of the antibodies used for both the depletions and for FACS sorting have been described elsewhere (Wilson et al., 1994) and both culture supernatants and fluorescein isothiocyanate or biotin conjugates were prepared in this laboratory. Additional phycoerythrin (PE), and Red613 antibody conjugates were as follows; HSA–PE (Pharmingen, San Diego, California); anti-TCRd–PE (Caltag, San Francisco, California); and CD8–Red613 (GIBCO BRL, Gainsberger, Maryland). Analysis was performed using the LYSYS ll program on either a FACScan or FACStar Plus. Generation of Hybridomas Hybridomas of ISP thymocytes were made by fusing the TCRa2b 2 variant of the BW5147 thymoma (Born et al., 1988) with FACS-sorted CD42CD81CD3lo thymocytes using a standard T cell hybrid fusion protocol. Thymocytes were depleted of CD41 and CD251 cells as described above, and after removal of dead cells on a Ficoll gradient the CD81CD3lo HSAhi cells sorted and reanalyzed for purity (Figure 3). As this subset of thymocytes has been shown previously to be mostly in cycle (MacDonald et al., 1988; Shortman et al., 1988), fusion was performed immediately after sorting in the absence of any stimuli. Immediately postfusion, the cells were placed in bulk culture, HAT medium was added after 48 hr, and after a further 24 hr the cells were harvested and the viable cells were cloned directly onto irradiated (5000 rads) macrophage monolayers (40,000 per well) in 96-well microtiter plates at 1 cell per well using the ACDU deposition unit of the FACStar Plus sorter (Becton Dickinson, Mountain View, California). No selection for CD8 surface expression by the hybrids was done, as on fusion with the BW5147 line CD8 expression is lost due to an unknown mechanism. After cloning and expansion (and prior to RNA preparation) these ISP hybridomas were screened for the surface expression of CD3 and all found to be low to negative. A second series of hybridomas was prepared in the same way from concanavalin A (2 mg/ml) and IL-2 (200 U/ml) activated gd1 thymocytes (MacDonald et al., 1990). For cloning, these cells were positively selected by staining with anti-TCRd–PE and screened during expansion for surface expression of TCR gd. The fusion efficiency was around 10-fold less for the ISP thymocytes compared with that of the activated thymic gd cells. Northern Blot Analysis and In Situ Hybridization In situ hybridization using 35S-labeled Ca or Cd RNA probes was performed on sorted thymocyte or lymph node populations as described in detail elsewhere (Held et al., 1990; Wilson et al., 1994). For Northern blot analysis, total cell RNA was prepared by CsCl2 gradient centrifugation and separated on formaldehyde agarose minigels as previously described (Wilson et al., 1994). Control RNA samples were the BW5147a2b 2 fusion partner, an ab CTL clone, activated thymic gd1 cells, and the P815 mastocytoma line. The RNA probes used for Northern blot were Cd, Ca, Cb2, Cg, and b-actin (Wilson et al., 1994) and TEA (Shimizu et al., 1993). These were all labeled with [a-32P]UTP by RNA transcription and the hybridization performed as previously described (Wilson et al., 1994). PCR and Cloning of Vd–Cd Products cDNA was prepared using AMV-reverse transcriptase (Boehringer Mannheim, Mannheim, Federal Republic of Germany) and oligo dT15–18 (Pharmacia Biotech Europe, Brussels, Belgium), using RNA from those hybridomas with TCR d transcripts corresponding to fulllength (VDJC) mRNA or from freshly isolated ISP thymocytes. For the hybridomas, a screening PCR was performed using a panel of Vd–Cd primers (listed below), to determine which Vd gene(s) were used by each and the resulting products purified for sequencing using the Geneclean kit (United States Biochemical, Cleveland, Ohio). PCR sense primers (59to 39) were as follows: Vd1, (GGAATTCA GAAGGCAACAATGAAAG); Vd2, (GTTCCCTGCAGATCCAAGCC); Vd3, (TTCCTGGCTATTGCCTCTGAC); Vd4, (CCGCTTCTGTGTGAAC

TTCC); Vd5, (CAGATCCTTGCAGTTCATCC); Vd6, (TCAAGTCCATCA GCCTTGTC); Vd7, (GAAAGCTTCAGTGCAAGAGTC); Vd8, (GCTACA GCACCCTGCACATC); nomenclature as in Raulet (1989). For each, an antisense Cd primer (TTGATGGCAATGGTCTTGGC) was used, giving products of around 500 bp. A separate PCR of the VaBWB/ Ca from the BW5147 fusion partner was included as a positive control for each of the cDNA preparations tested. The cDNA from freshly isolated ISP thymocytes was PCR amplified and purified in the same manner. Amplified DNA from each Vd family (except Vd1, which was not amplified from these cells) was cloned into the pCR-ScriptTMSK(1) cloning vector using the pCRScriptTMSK(1) cloning kit (Stratagene, La Jolla, California). Plasmid DNA was purified from positive bacterial colonies using the QIAprep spin plasmid kit (QIAGEN Gmbh, Hilden, Federal Republic of Germany). Sequencing of the VDJd junctions in the PCR products amplified from either the ISP hybridomas or cloned directly from ISP thymocytes was performed on both strands by the dideoxy-termination method (Casanova, 1993), using sequencing grade [a-35S]dATP (New England Nuclear, Du Pont de Nemours). Sequencing primers were as follows: an internal antisense TCR Cd primer close to the JCd junction (CAACATTTGTTCCATTTTTC); the same Vd sense primer as used for the PCR; and a Jd1 primer (TCCACAGTCACTTGGGTTCC). The products were analyzed on 8M urea, 6% acrylamide sequencing gels (Biorad, Hercules, California), fixed, dried, and exposed to x-ray film at room temperature for 2–21 days depending on signal intensity. Acknowledgments The authors wish to thank H. Pircher for the TCR transgenic mice, P. Zaech and C. Knabenhans for excellent FACS sorting, D. Reed for help in PCR cloning, A. Zoppi for help with the manuscript, and the Swiss Institute for Experimental Cancer Research (Epalinges, Switzerland) oligonucleotide synthesis laboratory for synthesis of PCR and sequencing primers. Received February 2, 1995; revised December 2, 1995. References Arden, B. (1992). Diversity of novel recombining elements suggests developmentally programmed expression of the T cell receptor a/d locus. Eur. J. Immunol. 22, 1287–1291. Ashwell, J.D., and Klausner, R.D. (1990). Genetic and mutational analysis of the T-cell antigen receptor. Annu. Rev. Immunol. 8, 139–167. Borgulya, P., Kishi, M., Uematsu, Y., and von Boehmer, H. (1992). Exclusion and inclusion of a and b T cell receptor alleles. Cell 69, 529–537. Born, W., White, J., O’Brien, R., and Kubo, R. (1988). Development of T cell receptor expression: studies using T cell hybridomas. Res. Immunol. 7, 279–291. Borst, J., Brouns, G.S., de Vries, E., Verschuren, M.C.M., Mason, D.Y., and van Dongen, J.J.M. (1993). Antigen receptors on T and B lymphocytes: parallels in organization and function. Immunol. Rev. 132, 49–84. Bra¨ ndle, D., Mu¨ller, C., Ru¨licke, T., Hengartner, H., and Pircher, H. (1992). Engagement of the T-cell receptor during positive selection in the thymus down-regulates RAG-1 expression. Proc. Natl. Acad. Sci. USA 89, 9529–9533. Capone, M., Watrin, F., Fernex, C., Horvat, B., Krippl, B., Wu, L., Scollay, R., and Ferrier, P. (1993). TCRb and TCRa gene enhancers confer tissue- and stage-specifity on V(D)J recombination events. EMBO J. 12, 4335–4346. Casanova, J.-L. (1993). Sequencing double-stranded linear DNA with sequenase and (a-35S)dATP. Meth. Mol. Biol. 23, 191–197. Chien, Y., Iwashima, M., Kaplan, K.B., Elliot, J.F., and Davis, M.M. (1987). A new T-cell receptor gene located within the a locus and expressed early in T cell differentiation. Nature 327, 677–682.

Immunity 44

Crompton, T., Pircher, H., and MacDonald, H.R. (1992). CD4182 thymocytes bearing MHC class I–restricted T cell receptors: evidence for homeostatic control of early stages of CD4/CD8 lineage development. J. Exp. Med. 176, 903–907. Diaz, P., Cado, D., and Winoto, A. (1994). A locus control region in the T cell receptor a/d locus. Immunity 1, 207–217. Dudley, E.C., Girardi, M.C., Owen, M.J., and Hayday, A.C. (1995). ab and gd T cells can share a late common precursor. Curr. Biol. 5, 659–669. Egerton, M., Shortman, K., and Scollay, R. (1990). The kinetics of immature murine thymocyte development in vivo. Int. Immunol. 2, 501–507. Fowlkes, B.J., and Pardoll, D.M. (1989). Molecular and cellular events in T cell development. Adv. Immunol. 44, 207–264. Godfrey, D.I., and Zlotnik, A. (1993). Control points in early T-cell development. Immunol. Today 14, 547–553. Guidos, C.J., Weissman, I.L., and Adkins, B. (1989). Intrathymic maturation of murine T lymphocytes from CD81 precursors. Proc. Natl. Acad. Sci. USA 86, 7542–7546. Haas, W., and Tonegawa, S. (1992). Development and selection of gd T cells. Curr. Opin. Immunol. 4, 147–155. Held, W., Mueller, C., and MacDonald, H.R. (1990). Expression of T cell receptor genes in the thymus: localization of transcripts in situ and comparison of mature and immature subsets. Eur. J. Immunol. 20, 2133–2136. Hugo, P., and Petrie, H.T. (1992). Multiple routes for late intrathymic precursors to generate CD41CD81 thymocytes. Adv. Mol. Cell. Biol. 5, 37–53. Ishida, I., Verbeek, S., Bonneville, M., Itaharo, S., Berns, A., and Tonegawa, S. (1990). T cell receptor gd and g transgenic mice suggest a role of a g gene silencer in the generation of ab T cells. Proc. Natl. Acad. Sci. USA 87, 3067–3071.

Nikolic-Zugic, J. (1991). Phenotypic and functional stages in the intrathymic development of ab T cells. Immunol. Today 12, 65–70. Nikolic-Zugic, J., and Moore, M.W. (1989). T cell receptor expression on immature thymocytes with in vivo and in vitro precursor potential. Eur. J. Immunol. 19, 1957–1960. Ohashi, P.S., Wallace, V.A., Broughton, H., Ohashi, C.T., Ferrick, D.A., Jost, V., Mak, T.W., Hengartner, H., and Pircher, H. (1990). Specific deletion of the J–Cd locus in murine a/b T cell clones and studies using transgenic mice. Eur. J. Immunol. 20, 517–522. Pardoll, D.M., Fowlkes, B.J., Bluestone, J.A., Kruisbeek, A., Maloy, W.L., Coligan, J.E., and Schwartz, R.H. (1987). Differential expression of two distinct T cell receptors during thymocyte development. Nature 326, 79–81. Paterson, D.J., and Williams, A.F. (1987). An intermediate cell in thymocyte differentiation that expresses CD8 but not CD4 antigen. J. Exp. Med. 166, 1603–1606. Pearse, M., Egerton, M., Wilson, A., Shortman, K., and Scollay, R. (1989). An early thymocyte development sequence marked by transient expression of the IL-2 receptor. Proc. Natl. Acad. Sci. USA 86, 1614–1618. Petrie, H.T., Hugo, P., Scollay, R., and Shortman, K. (1990). Lineage relationships and developmental kinetics of immature thymocytes: CD3, CD4, and CD8 acquisition in vivo and in vitro. J. Exp. Med. 172, 1583–1588. Petrie, H.T., Scollay, R., and Shortman, K. (1992). Commitment to the T cell receptor-ab or -gd lineages can occur just prior to the onset of CD4 and CD8 expression among immature thymocytes. Eur. J. Immunol. 22, 2185–2188. Philpott, K.L., Viney, J.L., Kay, G., Rastan, S., Gardiner, E.M., Chae, S., Hayday, A.C., and Owen, M.J. (1992). Lymphoid development in mice congenitally lacking T cell receptor ab-expressing cells. Science 256, 1448–1452. Pircher, H., Bu¨rki, K., Long, R., Hengartner, H., and Zinkernagel, R.M. (1989). Tolerance induction in double specific T cell receptor transgenic mice varies with antigen. Nature 346, 559–561.

Itaharo, S., Mombaerts, P., Lafaille, J., Iacomini, J., Nelson, A., Clarke, A.R., Hooper, M.L., Farr, A., and Tonegawa, S. (1993). T cell receptor d gene mutant mice: independent generation of ab T cells and programmed rearrangements of gd TCR genes. Cell 72, 337–348.

Raulet, D.H. (1989). The structure, function, and molecular genetics of the g/d T cell receptor. Annu. Rev. Immunol. 7, 175–208.

Kruisbeek, A.M. (1993). Development of ab T cells. Curr. Opin. Immunol. 5, 227–234.

Rothenberg, E.V. (1992). The development of functionally responsive T cells. Adv. Immunol. 51, 85–214.

Lauzurica, P., and Krangel, M.S. (1994a). Enhancer-dependent and independent steps in the rearrangement of a human T cell receptor d transgene. J. Exp. Med. 179, 43–55.

Scollay, R. (1991). T-cell subset relationships in thymocyte development. Curr. Opin. Immunol. 3, 204–209.

Lauzurica, P., and Krangel, M.S. (1994b). Temporal and lineagespecific control of T cell receptor a/d gene rearrangements by T cell receptor a and d enhancers. J. Exp. Med. 179, 1913–1921. Lin, W.-C., and Desiderio, S. (1993). Regulation of V(D)J recombination activator protein RAG-2 by phosphorylation. Science 260, 953–959. Livak, F., Petrie, H.T., Crispe, I.N., and Schatz, D.G. (1995). In-frame TCR d gene rearrangements play a critical role in the ab/gd T cell lineage decision. Immunity 2, 617–627. MacDonald, H.R., Budd, R.C., and Howe, R.C. (1988). A CD32 subset of CD42CD81 thymocytes: a rapidly cycling intermediate in the generation of CD41CD81 cells. Eur. J. Immunol. 18, 519–523. MacDonald, H.R., Schreyer, M., Howe, R.C., and Bron, C. (1990). Selective expression of CD8a (Ly-2) subunit on activated thymic gd cells. Eur. J. Immunol. 20, 927–930. Maguire, J.E., McCarthy, S.A., Singer, A., and Singer, D.S. (1990). Inverse correlation between steady-state RNA and cell surface T cell receptor levels. FASEB J. 4, 3131–3134. Malissen, M., Trucy, J., Jouvin-Marche, E., Cazanave, P.-A., Scollay, R., and Malissen, B. (1992). Regulation of TCR a and b allellic exclusion during T-cell development. Immunol. Today 13, 315–322. Mombaerts, P., Clarke, A.R., Rudnicki, M.A., Iacomini, J., Itohara, S., Lafaille, J.J., Wang, L., Ichikawa, Y., Jaenisch, R., Hooper, M.L., and Tonegawa, S. (1992). Mutations in T-cell antigen receptor genes a and b block thymocyte development at different stages. Nature 360, 225–231.

Shimizu, T., Takeshita, S., Muto, M., Kubo, E., Sado, T., and Yamagishi, H. (1993). Mouse germline transcript of TCR a joining region and temporal expression in ontogeny. Int. Immunol. 5, 155–160. Shortman, K. (1992). Cellular aspects of early T cell development. Curr. Opin. Immunol. 4, 140–146. Shortman, K., Wilson, A., Egerton, M., Pearse, M., and Scollay, R. (1988). Immature CD42CD81 murine thymocytes. Cell. Immunol. 113, 462–479. Shortman, K., Wu. L., Kelly, K., and Scollay, R. (1991). The beginning and the end of gd T cell development in the thymus. Curr. Topics. Microbiol. Immunol. 173, 71–80. Sim, G.-K., Olsson, C., and Augustin, A. (1995). Commitment and maintainance of the ab and gd T cell lineages. J. Immunol. 154, 5821–5831. Takeshita, S., Toda, M., and Yamagishi, H. (1989). Excision products of the T cell receptor gene support a progressive rearrangement model of the a/d locus. EMBO J. 8, 3261–3270. de Villartay, J.-P., and Cohen, D.I. (1990). Gene regulation within the TCR-a/d locus by specific deletion of the TCR-d cluster. Res. Immunol. 141, 618–623. de Villartay, J.-P., Lewis, D., Hockett, R.D., Waldmann, T.A., Korsmeyer, S.J., and Cohen, D.I. (1987). Deletional rearrangement in the human T-cell receptor a-chain locus. Proc. Natl. Acad. Sci. 84, 8608–8612. de Villartay, J.-P., Hockett, R.D., Coran, D., Korsmeyer, S.J., and Cohen, D.I. (1988). Deletion of the human T-cell receptor d-gene by a site-specific recombination. Nature 335, 170–174.

Expression of TCRd and TEA in Immature ab T Cells 45

Wilson, A., D’Amico, A., Ewing, T., Scollay, R., and Shortman, K. (1988). Subpopulations of early thymocytes: a cross-correlation flow-cytometric analysis of adult mouse Ly-22L3T42 (CD82CD42) thymocytes using eight different markers. J. Immunol. 140, 1461– 1469. Wilson, A., Petrie, H.T., Scollay, R., and Shortman, K. (1989). The acquisition of CD4 and CD8 during the differentiation of early thymocytes in short-term culture. Int. Immunol. 1, 605–612. Wilson, A., Pircher, H., Ohashi, P., and MacDonald, H.R. (1992). Analysis of immature (CD4-CD8-) thymic subsets in T-cell receptor ab transgenic mice. Dev. Immunol. 2, 85–94. Wilson, A., Held, W., and MacDonald, H.R. (1994). Two waves of recombinase gene expression in developing thymocytes. J. Exp. Med. 179, 1355–1360. Winoto, A. (1991). Regulation of the early stages of T-cell development. Curr. Opin. Immunol. 3, 199–203. Winoto, A., and Baltimore, D. (1989a). Separate lineages of T cells expressing the ab and gd receptors. Nature 338, 430–432. Winoto, A., and Baltimore, D. (1989b). ab lineage-specific expression of the a T cell receptor gene by nearby silencers. Cell 59, 649–655. Wu, L., Antica, M., Johnson, G.R., Scollay, R., and Shortman, K. (1991a). Developmental potential of the earliest precursor cells from the adult mouse thymus. J. Exp. Med. 174, 1617–1627. Wu, L., Scollay, R., Egerton, M., Pearse, M., Spangrude, G.J., and Shortman, K. (1991b). CD4 expressed on earliest T-lineage precursor cells in the adult murine thymus. Nature 349, 71–73. Yancopoulos, G.D., and Alt, F.W. (1986). Regulation of the assembly and expression of variable-region genes. Annu. Rev. Immunol. 4, 339–368.