The production of distinct sets of T cell receptor (TCR) γδ+ T cells occurs in an ordered fashion in thymic development. The Vγ3 and Vγ4 genes, located downstream in the TCRγ Cγ1 gene cluster, are expressed by the earliest waves of developing TCRγδ+ T cells in the fetal thymus, destined for intraepithelial locations. Upstream Vγ2 and Vγ5 genes are expressed in later waves in the adult and constitute most TCRγδ+ T cells in secondary lymphoid tissue. This developmental pattern is caused in part by a preference for rearrangements of the downstream Vγ3 and Vγ4 genes in the early fetal stage, which switches to a preference for rearrangements of the upstream Vγ2 and Vγ5 gene rearrangements in the adult. Our gene targeting studies show that the downstream genes rearrange preferentially in the early fetal thymus because of their downstream location, independent of promoter or recombination signal sequences and unrelated to the extent of germline transcription. Remarkably, gene deletion studies show that the downstream genes competitively inhibit upstream rearrangements at the fetal stage. These data provide a mechanism for specialization of the fetal thymus for the production of T cells expressing specific genes.

It has been estimated that TCRγδ+ T cells exhibit substantially greater potential sequence diversity than TCRαβ or primary Ig (1). Much of this diversity is caused by a great potential variability in CDR3 sequences in the TCR δ chain, but there is also diversity in the number of V, J, and (in the case of δ) D gene segments. The γ locus consists of three functional Jγ-Cγ genes and up to seven genes, with four of the genes clustered together upstream of Jγ1-Cγ1 (Fig. 1 B) (24).

Despite all of the potential diversity, the expressed diversity of γδ TCRs differs dramatically depending on the location of TCRγδ+ T cells in the adult. Sessile populations of intraepithelial TCRγδ+ T cells are found in the epidermis and the vaginal and tongue epithelium (5). The epidermal TCRγδ+ T cells, called dendritic epidermal T cells (DETCs), all express Vγ3 and Vδ1, whereas TCRγδ+ T cells in the vaginal and tongue epithelium all express Vγ4 and Vδ1 (6). There is essentially no TCRγδ diversity in these locations, as all of the cells express the same and Vδ, and the corresponding chains essentially lack N regions or other types of junctional diversity (7). Accordingly, TCRγδ+ T cells in these locations have been called “invariant γδ T cells.”

TCRγδ receptors in other locations show much more diversity. Many intraepithelial γδ T cells resident in the intestinal epithelium use Vγ5 associated with various Vδ chains, but these chains, unlike the invariant intraepithelial cells, show much more junctional diversity (8). In secondary lymphoid organs, γδ T cells using Vγ2, Vγ5, Vγ1.1, and Vγ1.2 are predominant, associated with numerous chains (2). Again, these chains show substantial junctional diversity. The striking differences between the repertoires of invariant γδ T cells and γδ T cells in the secondary lymphoid organs suggest different functions for these cells. Invariant γδ T cells such as DETCs are thought to recognize predictable self-ligands, which are up-regulated by cell stress such as that associated with wounding or tumorigenesis (911). Some of the diverse TCRγδ+ T cells in secondary lymphoid organs also recognize self-ligands (12), but others may play a more conventional immunological role by recognizing variable antigens associated with pathogens, consistent with the much greater receptor diversity observed in these populations.

A key question is how such different repertoires of TCRγδ+ T cells can be produced during development of the immune system. A striking finding was that invariant Vγ3+ and Vγ4+ TCRγδ+ T cells are only produced in the early fetal thymus (between embryonic days 13 and 17 [E13–17]), whereas the thymus switches to producing only the variable Vγ2+, Vγ5+, Vγ1.1+, and Vγ1.2+ populations after E16 (1315) (for review see reference 16). This specialization is determined by the stage of development of both precursor T cells and thymic stroma (5, 17). The switch has been shown to consist of both programmed events that determine the types of rearrangements that occur at different stages of ontogeny, and a cellular selection process that operates at least in the fetal period to select cells of defined specificity for export to the appropriate peripheral location, such as the skin (for review see reference 18).

Some of the relevant mechanisms are exemplified by the Cγ1 cluster, where the Vγ3 and Vγ4 genes expressed by invariant TCRγδ+ T cells are located in the downstream position, closer to Jγ1, whereas the Vγ5 and Vγ2 genes expressed by many variable TCRγδ+ T cells are located upstream (Fig. 1 A). The gene rearrangement pattern changes in ontogeny so that the Vγ3 and Vγ4 genes rearrange preferentially in the early fetal period, whereas the upstream Vγ2 and Vγ5 genes rearrange preferentially in the adult stage (13). Studies with mutated TCRγ transgenes or TCRδ knockout mice show that the stage-specific rearrangement pattern is independent of cellular selection processes (16, 19, 20). Rearrangements that occur in the fetal period usually lack N regions (as a result of the absence of terminal deoxynucleotidyl transferase [TdT]) and exhibit little or no junctional diversity (21, 22), whereas the adult-stage rearrangements contain abundant N regions and junctional diversity. On top of these programmed events in development, there is considerable evidence for a selection process that acts in the fetal thymus to select Vγ3+ cells that express appropriate Vδ1 chains, and to equip these cells with homing receptors necessary for export from the thymus and emigration to the epidermis (23, 24).

Previous studies of V(D)J rearrangement suggest several possible mechanisms for differential V gene rearrangements. Early studies of VH genes (25, 26), later extended to and gene families (13, 27), showed that rearrangements of downstream V genes are often favored in the fetal period, raising the possibility that the downstream location is inherently favorable for rearrangement at this stage. Later studies showed a correlation between germline transcription of rearranging gene segments and subsequent rearrangements (28, 29) (for review see reference 30). Studies with isolated nuclei showed that the local “accessibility” of rearranging gene segments to exogenously added RAG recombinase is also correlated with subsequent rearrangement (31). The relative roles of transcription versus simple accessibility have proven challenging to dissect, but a recent report provided direct evidence that germline transcription is causally related to rearrangement in the TCRα locus (32). Transcription, accessibility, and gene rearrangement have also been linked in some cases to local histone acetylation and/or lack of methylation of gene segments (33, 34). Finally, the specific sequence of a recombination signal sequence (RSS) can restrict its capacity to recombine with other RSS, despite compatibility with the 12/23 joining rule (35). Any or all of these mechanisms may work together or separately to impose the observed pattern of rearrangements.

In previous experimental analyses of the TCRγ locus, we have focused on the upstream Vγ2 gene as a representative adult-stage gene, and the downstream Vγ3 gene as a representative fetal-stage gene. In adult stage thymocytes, the upstream Vγ2 gene rearranges ∼25 times more frequently than the downstream Vγ3 gene (20, 36). We implicated the promoter sequences in this regulation by showing that the preference for Vγ2 over Vγ3 rearrangements in adult thymocytes, as well as the germline transcription pattern, was reversed when the promoter regions of the Vγ3 and Vγ2 genes were swapped in a transgenic backbone consisting of unrearranged Vγ2-4, Jγ1, and Cγ1 gene segments (20, 36). Other studies showed that poor rearrangement of Vγ3 in adult thymocytes correlates with low germline transcription and low levels of histone acetylation (29, 34) of the Vγ3 gene as compared with the Vγ2 gene. Furthermore, inhibitors of histone deacetylase were shown to enhance Vγ3 rearrangements in thymic organ cultures reconstituted with adult bone marrow stem cells (34). These data indicated that rearrangement of Vγ4/Vγ3 genes is strongly repressed in the adult stage by a mechanism that depends at least in part on the V gene promoter segments.

Two lines of evidence suggest that the pattern of rearrangement is regulated by a distinct mechanism in the early fetal thymus. First, although the downstream Vγ3 gene rearranges approximately four times more frequently than the upstream Vγ2 gene in fetal thymocytes (20, 36), both genes were germline transcribed at a relatively high level and showed similarly high levels of histone acetylation (34, 36). Second, swapping the Vγ3 and Vγ2 promoter regions in a transgenic construct did not reverse the fetal pattern of rearrangement, as it did the adult pattern (20). In light of these and additional data with transgenic rearrangement constructs, we proposed that Vγ2 and Vγ3 genes are similarly accessible and transcriptionally active in fetal thymocytes, and that the preference for rearrangement of Vγ3 arises because of an inherent preference for rearrangement of the gene that lies closest to Jγ1 (18, 36). In addition, we suggested that the preference is established at least in part because of a competition in which the rearrangement of J-proximal genes inhibits the rearrangement of more distal genes. In the adult stage, repression of the Vγ3 and Vγ4 genes is proposed to prevent rearrangement of these segments, thus removing the competitive barrier to Vγ2 rearrangement (Fig. 1 B). In this study, we have used a gene targeting approach to directly test two aspects of this proposal in the context of the endogenous TCRγ locus. The results provide definitive support for the contentions that the downstream genes enjoy an inherent preference for rearrangement, and that rearrangement of the downstream genes competitively inhibits rearrangement of the upstream Vγ2 gene segment. Furthermore, both of these effects occur independently of changes in germline transcription, suggesting that the position of the gene by itself determines the rearrangement pattern in the fetal thymus.

Results

Gene-targeted mice to test the role of gene location in programmed V gene rearrangement

To address the role of V gene location in V gene rearrangement in the endogenous TCR Cγ1 locus, we used gene targeting to create a locus with two Vγ2 genes, one in the normal upstream position and the second at the downstream position normally occupied by the Vγ3 gene (Fig. 1 A). This arrangement created the opportunity to examine the rearrangement of essentially identical genes located in distinct locations in the genome. To create this locus, an endogenous gene segment containing the downstream Vγ4-Vγ3 gene pair was replaced by homologous recombination with an in vitro–assembled Vγ4-Vγ2 pair flanked by loxP sites and a neo cassette, in which the 2.3-kb fragment containing Vγ2 included the Vγ2 promoter region, RSS, and flanking sequences (Fig. 1 A; and Fig. S1). The introduced Vγ2 copy contained a silent single bp substitution that destroyed a KpnI site in the middle of the Vγ2 gene, providing the basis for assays to distinguish this sequence from the wild-type upstream copy when analyzing genomic DNA or mRNA. The loxP sites flanking Vγ4-Vγ2 provided a separate option to delete these genes in the germline, as described below. After deleting the neo cassette by transfection of embryonic stem (ES) cells with a Cre expression plasmid (Fig. 1 and Fig. S1), we achieved germline transmission of the knock-in (KI) allele, named 242KI. Homozygous 242KI/242KI mice and +/+ littermates were generated by intercrossing (Fig. S1). Because the ES cells were from the 129/SvJ strain, the founder mice were crossed to 129/J mice to minimize genetic diversity and segregation.

Preferential rearrangement of the downstream (J-proximal) Vγ2 gene in the fetal thymus of 242KI/242KI mice

The extent of gene rearrangement in fetal (E14 or E15) or adult thymocytes was determined by semiquantitative PCR or by Southern blotting. The relative levels of rearrangement of the two Vγ2 genes were determined by digesting the PCR products with KpnI (Fig. 2 A, left), or by performing Southern blots with DNA digested with restriction enzymes that allowed the two rearrangements to be distinguished by size (Fig. 2 B). In E14 fetal thymocytes, the downstream Vγ2 gene segment rearranged four to five times more frequently than the upstream Vγ2 gene. In E15 fetal thymocytes, the difference decreased to twofold (Fig. 2, A and B). Accordingly, the total level of Vγ2 rearrangements (the sum of rearrangements of the upstream and downstream Vγ2 genes) was elevated approximately fourfold in the 242KI E14 fetal thymocytes compared with wild-type fetal thymocytes (Fig. 2 C). The substituted Vγ2 gene in the downstream position did not detectably alter the extent of Vγ4 rearrangements (unpublished data).

It remained possible that these estimates of relative Vγ gene rearrangement levels were distorted by cellular selection that influenced the abundance of cells with different rearrangements. To address this possibility, we crossed the 242KI mice to TCRδ−/− mice, thus preventing formation of a γδ TCR that could influence cellular selection. Analysis of 242KI/TCRδ−/− fetal thymocyte DNA by PCR showed that the downstream Vγ2 rearrangements were favored over upstream Vγ2 rearrangements to a similar extent as in 242KI/TCRδ+/+ mice (Fig. 2 A, right).

The four- to fivefold preference for rearrangement of the downstream Vγ2 gene in 242KI/242KI mice is comparable to the approximately fourfold preference for rearrangement of the downstream Vγ3 gene observed in wild-type fetal thymocytes (20, 36). Therefore, the location of the Vγ2 gene segment in the V gene array determines its preference for rearrangement in the fetal period independently of the RSS or the upstream promoter regions, which were identical in the two Vγ2 genes under comparison.

Deletion of Vγ4 and Vγ3 results in increased rearrangement of Vγ2 at the fetal stage

Preferential rearrangement of the J-proximal gene at the fetal stage might reflect a competition between genes for rearrangement to Jγ1, in which the downstream gene holds an advantage. If this were the case, deletion of the downstream genes should result in an increased frequency of rearrangements of the upstream gene. Deletion of the downstream genes was possible because the Vγ4-Vγ2 gene pair in the targeted locus was flanked by loxP sites. Among the Cre-transduced ES cell colonies, we identified some in which this entire segment, as well as the neo cassette, were deleted by Cre-mediated recombination (Fig. 1 and Fig. S1). The deletion reduced the distance between Vγ2 and Jγ1 by only a modest 20% compared with the wild-type locus. The knockout allele was named Δ43 to signify the fact that this targeted allele corresponds to a deletion of the Vγ4 and Vγ3 genes relative to the wild-type locus. After generating chimeras with the recombinant ES cells and achieving germline transmission of the Δ43 allele by crossing to 129/J mice, homozygous Δ43/Δ43 mice were generated by intercrossing.

The amount of Vγ2 gene rearrangement to Jγ1 in Δ43/Δ43 mice was compared with that in wild-type (+/+) littermates on the 129 background. As shown by semiquantitative PCR, the deletion of Vγ3 and Vγ4 resulted in an approximately fivefold increase in Vγ2 rearrangement in E14 fetal thymocytes (based on three experiments; Fig. 3 A and not depicted). Thus, deletion of Vγ3 and Vγ4 resulted in increased levels of Vγ2 rearrangement in E14 fetal thymocytes to about the level of Vγ3 rearrangement observed in wild-type mice. Comparable PCR analysis (not depicted) and genomic Southern blotting with a Vγ2 probe (Fig. 3 B and not depicted) revealed that Vγ2 rearrangements were still elevated in Δ43/Δ43 mice on E15, though to a lesser extent. A separate examination of rearrangements of Vγ5, which is also upstream of Vγ3/Vγ4, showed an approximately threefold increase in Δ43/Δ43 mice compared with wild-type mice in E14 fetal thymocytes (Fig. S2). These data indicated that rearrangements of the proximal Vγ3 and Vγ4 genes in the wild-type locus normally competitively inhibit rearrangements of the more distal Vγ2 and Vγ5 genes.

Location-dependent rearrangement patterns in the fetal thymus are not associated with changes in germline transcription

Because germline transcription has been causally associated with V(D)J rearrangements, we assessed whether the rearrangement patterns we observed correlated with the abundance of germline transcripts. To quantify germline transcripts, dilutions of cDNA generated by RT with random hexamer primers were PCR amplified with an upstream primer corresponding to the Vγ2 coding region and a downstream primer corresponding to a sequence just 3′ of the Vγ2 gene RSS. No products were observed when RT was omitted from the reaction, confirming that the products corresponded to transcripts as opposed to contaminating genomic DNA (Fig. 4 A). For analysis of germline transcripts in thymocytes from 242KI/242KI mice, the PCR products were digested with KpnI to discriminate whether they originated from the wild-type upstream gene (cleaved) or the knocked-in downstream gene (uncleaved). Based on this analysis, germline transcripts of the downstream Vγ2 gene were, if anything, slightly less abundant than upstream Vγ2 germline transcripts in E14 fetal thymocytes (ratio = 0.83), despite the much increased level of rearrangement of the downstream gene at this time point.

Two caveats should be noted in interpreting the germline transcription data. First, a correction should be applied to these data, because preferential rearrangement of the downstream gene copy is predicted to decrease the availability of unrearranged downstream gene copies, compared with upstream copies, for transcription. However, it is likely that this correction is negligible in light of previous data showing that <1% of total gene copies are rearranged in E14 thymocytes (20, 36). The second caveat is that it is conceivable that the altered nucleotide sequence of the knocked-in downstream copy of Vγ2 resulted in altered stability or processing of the corresponding germline transcripts, resulting in an underestimate of the level of germline transcription. We think it unlikely, however, that the single bp change introduced near the middle of the Vγ2 exon would cause such a substantial change in transcript stability or processing. Based on these considerations, the data support the conclusion that the higher level of rearrangement of the 3′ Vγ2 gene in fetal thymocytes occurs in the absence of increases in germline transcription.

We also used a restriction enzyme accessibility assay to determine whether the two copies of the Vγ2 gene were differentially accessible in the nuclei of E15 fetal thymocytes from 242KI/242KI mice. A ScaI restriction enzyme site ∼380 bp upstream of the Vγ2 ATG codon was nearly equally accessible to restriction enzyme in the two copies of Vγ2 in 242KI/242KI mice in the nuclei of E15 fetal thymocytes (Fig. 4 B). Therefore, the different levels of rearrangement did not correlate with the accessibility of this site to restriction enzyme.

To assess whether the increased rearrangement of Vγ2 in fetal thymocytes from Δ43/Δ43 mice could be attributed to greater germline transcription, we quantified Vγ2 germline transcripts in Δ43/Δ43 versus +/+ E14 and E15 fetal thymocytes. In two experiments, fetal thymocytes from the two strains contained essentially identical levels of Vγ2 germline transcripts (Fig. 5 and not depicted). Thus, the substantial increase in Vγ2 rearrangements that accompanied deletion of Vγ4 and Vγ3 is not accompanied by a corresponding increase in Vγ2 germline transcription.

Influence of V gene location on the rearrangement and accessibility of genes at the adult stage

In adult-stage wild-type thymocytes, Vγ3 and Vγ4 rearrangements are normally strongly depressed compared with the fetal stage. We therefore tested whether replacing Vγ3 with Vγ2 in 242KI/242KI mice prevents the repression of downstream rearrangements in adult thymocytes. The abundance of upstream Vγ2 rearrangements was higher in adult than fetal thymocytes of 242KI mice, just as it was in wild-type mice (Fig. 2 B). However, in contrast to the situation in fetal thymocytes, the majority (∼70%) of Vγ2 rearrangements in adult 242KI/242KI thymocytes involved the upstream Vγ2 gene (Fig. 2, A and B), representing an ∼10-fold decrease in downstream Vγ2 rearrangements relative to upstream Vγ2 rearrangements between the fetal and adult stages. Although there was a shift toward upstream Vγ2 rearrangements in adult 242KI thymocytes, the overall preference for rearranging the upstream gene was smaller in 242KI mice (approximately 2-fold) than it was in adult wild-type thymocytes (which exhibit a 25-fold preference for Vγ2 over Vγ3 rearrangements).

We also determined whether the decreased proportion of downstream Vγ2 rearrangements at the adult stage was paralleled by a decrease in the proportion of germline transcripts emanating from the downstream gene. The analysis showed a reduction of approximately threefold in the proportion of germline transcripts emanating from the downstream gene (Fig. 4 A). These data must be adjusted based on the relative number of copies of each unrearranged Vγ2 gene in the population being examined. A separate PhosphorImager analysis (see Materials and methods) of three Southern blots, examining band intensity in thymocytes as compared with liver DNA, showed that the ratio (downstream/upstream) of unrearranged Vγ2 genes in adult thymocytes was ∼0.74 (unpublished data). Hence, the abundance of Vγ2 germline transcripts emanating from the downstream Vγ2 gene was reduced by ∼2.2-fold after normalization for the number of gene templates in the population. The decreased proportion of rearrangements and germline transcripts of the downstream Vγ2 gene segment in adult thymocytes suggests that sequences in the vicinity of the downstream gene repress both germline transcription and rearrangement at the adult stage. Because the downstream Vγ2 gene segment is identical to the upstream segment, the relevant repressive sequences are presumably not in the downstream Vγ2 gene segment itself, but more likely in neighboring sequences. Hence, these data are consistent with the existence of sequences in the downstream region in addition to those previously attributed to the Vγ3 promoter (which is absent from the 242KI chromosome) (20), which locally repress rearrangements in the adult thymus.

In adult thymocytes, rearrangements of the upstream Vγ2 gene were similarly abundant in wild-type mice as in either 242KI/242KI (Fig. 2 B) or Δ43/Δ43 (Fig. 3 B) mice. Furthermore, in adult thymocytes, Vγ2 germline transcripts were similarly abundant in wild-type, 242KI/242KI (Fig. 4 A, legend), and Δ43/Δ43 (Fig. 5) thymocytes. Hence, the alterations at the downstream position in these two chromosomes did not alter the efficiency of rearrangements or germline transcription at the upstream position in adult thymocytes.

Discussion

Previous studies (20, 36) demonstrated that the downstream Vγ3 gene is preferentially rearranged in the early fetal thymus compared with Vγ2 by a factor of approximately fourfold. Other studies show that the fraction of productive rearrangements is inherently lower for Vγ2 than for rearrangements of other V genes, because Vγ2 uniquely contains an in-frame stop codon near its 3′ end, which must be removed (presumably by endonuclease action) during the recombination process (16). The relative paucity of Vγ2 rearrangements at the fetal stage, coupled with the rarity of productive Vγ2 rearrangements, are critical in ensuring that nearly all TCRγδ+ fetal thymocytes express Vγ3 (or Vγ4) and not Vγ2. The absence of TdT at the fetal stage also plays a role in the process of Vγ3 gene rearrangement. Without TdT and the N nucleotides it adds, the frequency of rearrangements that occur at the site of a dinucleotide homology shared by the Vγ3 and Jγ1 gene segments is increased, thus ensuring that most productive rearrangements carry precisely the same (canonical) Vγ-Jγ junctional sequence (22). These mechanisms, coupled with a selection process for Vγ3+ T cells that operates in the fetal thymus (24), conspire to convert the early fetal thymus into an organ specialized in part for the production of Vγ3+ DETCs.

Our data using gene-targeted chromosomes address a key part of this coordinated process by probing the mechanisms that lead to preferential Vγ3 gene rearrangement in the fetal thymus. The results have implications for other Ig/TCR genes as well, because preferential rearrangement of J-proximal V genes is also observed in early stage development of B cells and T cells in the case of the IgH and TCRδ gene families. The results of this study provide compelling in vivo evidence that the location of a gene is a critical determinant of its rearrangement pattern in the fetal thymus and provide direct evidence that the downstream gene exerts competitive inhibition of upstream V gene rearrangement at the fetal stage.

Previous studies provided evidence that at the fetal stage, the unrearranged Vγ2 and Vγ3 gene segments are acetylated on histone residues to a similar extent, and are germline-transcribed similarly as well (34, 36). These similarities suggested that neither accessibility nor transcriptional activity of the genes was a likely explanation for the preferential rearrangement of the downstream genes, raising the possibility that gene location by itself plays a role in rearrangement at the fetal stage. This interpretation was further suggested by our previous study performed with transgenes consisting of unrearranged Vγ/Jγ/Cγ genes, in which the positions of the Vγ3 and Vγ2 gene segments were swapped within the transgene (36). The present study, however, is a definitive analysis, as it was performed by targeting the endogenous locus, and is therefore free of certain artifacts that can confound interpretation of transgenic mice, such as the existence of multiple tandem copies of the transgene that create the opportunity for complex rearrangements to take place and that may override natural regulatory processes, as well as the potential absence of flanking regulatory sequences that may influence the rearrangement process. Equally important, the current study is superior as it examines two copies of the same complete gene segment (Vγ2) located at the upstream and downstream positions. Hence, differences in rearrangement cannot be attributed to differences in the RSS or promoters of the genes, or other features associated with the coding and immediately flanking DNA, all of which were included in the inserted downstream Vγ2 copy.

The results showed that the downstream gene enjoyed an approximately fivefold advantage in rearrangement frequency in the early fetal thymus. The upstream and downstream genes were germline transcribed to a similar extent, suggesting that the advantage did not result from greater accessibility or transcription-related enhancement of rearrangement. We propose, instead, that the advantage of the downstream gene arises from a greater likelihood of encountering the Jγ1 RSS, either because of its closer proximity or because of a processive aspect to the rearrangement process that favors the first encountered gene. We cannot rule out the possibility that the advantage arises from closer proximity of the downstream gene to a downstream regulatory element. However, we previously showed that rearrangement is not significantly altered in the fetal thymus of mice with targeted deletions of either or both of the two enhancer-like elements in the locus, including the downstream enhancer E3′Cγ1 (37).

Based on these findings, we hypothesized that the fetal pattern of rearrangement is determined by a competition between the downstream and upstream gene segments, with an advantage accruing to the downstream gene owing to its closer proximity to Jγ1 (Fig. 1 B). The design of our gene-targeted allele enabled us to test for the first time a key prediction of this hypothesis by determining whether deletion of the intervening Vγ4 and Vγ3 genes altered the putative competition so as to promote rearrangement of the upstream Vγ2 gene. The results demonstrated that deletion of Vγ4 and Vγ3 resulted in enhanced rearrangements of the upstream Vγ2 gene segment in the fetal thymus to an extent commensurate with the amount of rearrangement that normally occurs at the downstream gene. These findings provide direct support for the notion that genes are normally in competition in the fetal stage, and that J-proximal rearrangements occur at the expense of J-distal rearrangements. The small reduction in the length of the segment of DNA connecting Vγ2 and Jγ1 resulting from the deletion, ∼20%, appears unlikely to account for an approximately fivefold increase in rearrangement in E14 fetal thymocytes.

As discussed in the Introduction, the pattern of rearrangement in the adult thymus is imposed by a distinct mechanism that is regulated by promoter–specific sequences and leads to repression of Vγ3 and Vγ4 rearrangements that is dependent on local histone deacetylation. Considering all these observations, we proposed that the Vγ3 promoter contains sequences that locally repress Vγ3 and Vγ4 rearrangements in the adult, abolishing competition from downstream genes and enhancing upstream gene rearrangements (18). The notion that Vγ3-associated sequences repress rearrangement in the adult thymus predicts that replacing Vγ3 and its promoter with corresponding Vγ2 sequences, as in the 242KI mice, should result in less effective silencing of downstream rearrangements in adult thymocytes. The data are consistent with this interpretation, because the bias in favor of rearrangement of the upstream Vγ2 gene in adult 242KI/242KI mice (2-fold) was small in comparison to the bias in favor of Vγ2 over Vγ3 rearrangements in adult wild-type mice (25–50-fold). On the other hand, although the overall level of rearrangement increased substantially between the fetal and adult stages, presumably because of enhanced overall activity of the locus, the relative abundance of downstream Vγ2 rearrangements in adult thymocytes of 242KI/242KI mice declined by ∼10-fold compared with the situation in fetal thymocytes. A plausible explanation for these data is that the 242KI allele retains residual repressive sequences active at the adult stage. One possibility is that the Vγ4 gene, which remains in the 242KI allele, contains these putative residual repressive sequences.

The results reported in this paper support a significant role of gene position and V gene competition in controlling the abundance of specific V gene rearrangements. These features must therefore be considered in conjunction with transcriptional activity, accessibility, differences in RSS sequences, and other factors as important regulators of V(D)J recombination. All of these factors may be exploited in the evolution of mechanisms to create various specific repertoires in the adaptive immune system. This is illustrated by the dynamic behavior of the TCRγ locus during development, in which the positional influence, gene competition and other factors work in tandem to enable the early fetal thymus to specialize in the production of invariant TCRγδ+ T cells destined for epithelial locations, whereas transcriptional repression and/or loss of accessibility enable the adult thymus to switch its efforts toward the production of Vγ2+ and other TCRγδ+ subsets found abundantly in secondary lymphoid organs in adult mice.

Materials And Methods

Generation of 242KI and Δ43 knockout mice.

A single gene targeting event was used to generate 242KI and Δ43 knockout mice. The targeting vector (Fig. 1 A and Fig. S1) was based on the pKSTKNeoLoxP plasmid (a gift from R. Rickert, Burnham Institute for Medical Research, La Jolla, CA). It included a thymidine kinase cassette at the 5′ end, a 5′ homology region corresponding to a 3.2-kb SnaBI-BglII fragment upstream of the Vγ4 gene, and a 3′ homology region corresponding to a 3.2-kb EcoRI-BamHI fragment downstream of the Vγ3 gene. Between the arms, we inserted a PGK-neo cassette and a Vγ4-Vγ2 gene pair comprising a 1.7-kb BglII-HindIII Vγ4 fragment ligated to a 2.5-kb SpeI-EcoRI Vγ2 fragment. The latter fragment was the same fragment used in an earlier transgenic study (20, 36), and it included the Vγ2 promoter, leader exon, intron, coding segment, RSS, and 3′ flanking DNA. In the present study, the inserted Vγ2 fragment contained a silent 1-bp mutation that destroyed the unique KpnI site located in the coding region of this gene (ggtacc→ggtatc) to enable discrimination from the upstream wild-type gene. The PGK-neo cassette was flanked by loxP sites, and an additional loxP site was located downstream of Vγ2 at the 3′ end of the inserted segments (Fig. 1 A). The locations of the three loxP sites enabled us to delete the PGK-neo cassette or the entire inserted sequence, depending on whether the 5′ loxP site recombined with the central loxP site or the 3′ loxP site.

The targeting construct was transfected into J1 ES cells of the 129/SvJae strain (38), followed by selection with G418 and ganciclovir. Targeted ES cell clones were identified by Southern blotting with a 5′ flanking probe (Fig. S1 A). Correct targeting at the 3′ end was confirmed by PCR (Fig. S1 B). The targeted ES cell clones were transiently transfected with a Cre-expressing plasmid (39) and screened by Southern blotting for clones that had deleted only the PGK-neo cassette (242KI), or had deleted both the PGK-neo cassette and the Vγ4-Vγ2 fragment (Δ43; Fig. S1 D). The recombinant ES cell clones were separately injected into C57BL/6 blastocysts, and chimeric mice were crossed with 129/J and C57BL/6J mice. Heterozygous pups from the 129 cross were intercrossed to generate homozygous 242KI/242KI or Δ43/Δ43 mice on the 129 background. Heterozygous pups from the C57BL/6J cross were serially backcrossed to C57BL/6J mice six times before intercrossing. Blotting genomic DNA with Vγ3, Vγ2, and other probes confirmed germline transmission of the targeted alleles (Fig. S1 E and not depicted). Animal studies were approved and overseen by the University of California, Berkeley Animal Care and Use Committee.

Cell, DNA, and RNA preparations.

Total and CD4CD8 (double-negative) thymocytes were prepared as previously described (37). DNA and total cellular RNA were prepared as previously described (20).

Primers.

Primers L2, J1, PSVγ2, VG2I, V2-3′a, 5′Jγ1 5′-2, Cγ1 3′-3, 5′ tubulin, and 3′ tubulin have been previously described (18, 20, 29, 34, 36). Primer PSVγ2′ is as follows: tcgaaagctttaggagtgtaacaatacac.

Semiquantitative PCR and RT-PCR analysis.

The semiquantitative PCR assays were performed essentially as previously described (20). In brief, thymic genomic DNA was subjected to PCR in the presence of α-[32P]deoxycytidine triphosphate (dCTP) with the primer set L2/J1 to amplify Vγ2 gene rearrangements. To distinguish whether the radiolabeled PCR products originate from rearrangements of the endogenous upstream Vγ2 gene versus the knocked-in downstream Vγ2 gene in 242KI mice, they were digested with Kpn1 before electrophoresis on a polyacrylamide gel and analysis (PhosphorImager; Molecular Dynamics).

Semiquantitative RT-PCR performed as previously described (18, 36), with minor modifications, was used to quantify Vγ2 or Jγ1 germline transcripts. RNA samples from E14 fetal thymocytes or adult CD4CD8 thymocytes were treated with DNase to eliminate any genomic DNA contamination and reverse transcribed with Superscript II RNase H RT using random hexamer primers. The RT products were serially diluted (by the factors indicated in the figures) and subjected to PCR in the presence of α-[32P]dCTP with primer sets PSVγ2/V2-3′a (for analysis of Vγ2 transcripts in 242KI mice), PSVγ2′/Vγ2-3′a (for analysis of Vγ2 transcripts in Δ43 mice), or 5′Jγ1 5′-2/Cγ1 3′-3 (for Jγ1). In the case of samples from 242KI mice, the RT-PCR products (307 bp) were digested with Kpn1 to distinguish the origin of the products. RT-PCR products corresponding to the mutated 3′ Vγ2 copy cannot be cleaved by KpnI, whereas products corresponding to the endogenous 5′ Vγ2 gene are cleaved into smaller fragments (274 + 33 bp). Parallel semiquantitative RT-PCRs of β-tubulin transcripts were used to normalize the samples. In all cases, RNA samples without RT were subjected to PCR in parallel to assure that there was no genomic DNA contamination.

Southern blot analysis of Vγ2 gene rearrangements.

Thymic genomic DNA samples were digested with NcoI/BamHI and analyzed by Southern blotting using a Vγ2-specific probe, as previously described (20).

Restriction enzyme accessibility assay.

The assay was performed to assess restriction enzyme accessibility of the two copies of Vγ2 genes in 242KI mice. In brief, nuclei were prepared from E15 fetal thymocytes and digested with ScaI for 1 h. After the ScaI digestion, DNA was prepared from the nuclei and digested with NcoI. The DNA was analyzed by Southern blotting using a Vγ2 probe.

Online supplemental material.

Fig. S1 depicts and confirms the gene targeting procedure used to generate 242KI and Δ43 mice. Fig. S2 shows changes in the rearrangement of Vγ5 in Δ43 mice.

Acknowledgments

We thank Hector Nolla for expert assistance in cell sorting and Robert Rickert for the gift of the pKSTKNeoLoxP plasmid. We thank Timothy Nice, Michael Whang, and Sangho Lee for comments on the manuscript, and Yana Natanzon and Jennifer Beck for technical support.

This work was supported by a grant from the National Institutes of Health to D.H. Raulet (R01AI31650).

The authors have no conflicting financial interests.

References

References
1
Davis, M.M., and P.J. Bjorkman.
1988
. T cell antigen genes and T cell recognition.
Nature.
334
:
395
–402.
2
Raulet, D.H.
1989
. The structure, function, and molecular genetics of the gamma/delta T cell receptor.
Annu. Rev. Immunol.
7
:
175
–207.
3
Haas, W., P. Pereira, and S. Tonegawa.
1993
. Gamma/delta cells.
Annu. Rev. Immunol.
11
:
637
–685.
4
Hayday, A.C.
2000
. [gamma][delta] cells: a right time and a right place for a conserved third way of protection.
Annu. Rev. Immunol.
18
:
975
–1026.
5
Allison, J.P., and W.L. Havran.
1991
. The immunobiology of T cells with invariant γδ antigen receptors.
Annu. Rev. Immunol.
9
:
679
–705.
6
Lafaille, J.J., A. DeCloux, M. Bonneville, Y. Takagaki, and S. Tonegawa.
1989
. Junctional sequences of T cell receptor γδ genes: implications for γδ T cell lineages and for novel intermediate of V-(D)-J joining.
Cell.
59
:
859
–870.
7
Asarnow, D.M., W.A. Kuziel, M. Bonyhadi, R.E. Tigelaar, P.W. Tucker, and J.P. Allison.
1988
. Limited diversity of γδ antigen receptor genes of Thy-1+ dendritic epidermal cells.
Cell.
55
:
837
–847.
8
Bonneville, M., C.A. Janeway Jr., K. Ito, W. Haser, I. Ishida, N. Nakanishi, and S. Tonegawa.
1988
. Intestinal intraepithelial lymphocytes are a distinct set of γδ T cells.
Nature.
336
:
479
–481.
9
Havran, W.L., Y.H. Chien, and J.P. Allison.
1991
. Recognition of self antigens by skin-derived T cells with invariant gamma delta antigen receptors.
Science.
252
:
1430
–1432.
10
Girardi, M., D.E. Oppenheim, C.R. Steele, J.M. Lewis, E. Glusac, R. Filler, P. Hobby, B. Sutton, R.E. Tigelaar, and A.C. Hayday.
2001
. Regulation of cutaneous malignancy by gammadelta T cells.
Science.
294
:
605
–609.
11
Jameson, J., K. Ugarte, N. Chen, P. Yachi, E. Fuchs, R. Boismenu, and W.L. Havran.
2002
. A role for skin gammadelta T cells in wound repair.
Science.
296
:
747
–749.
12
Crowley, M.P., A.M. Fahrer, N. Baumgarth, J. Hampl, I. Gutgemann, L. Teyton, and Y. Chien.
2000
. A population of murine gammadelta T cells that recognize an inducible MHC class Ib molecule.
Science.
287
:
314
–316.
13
Garman, R.D., P.J. Doherty, and D.H. Raulet.
1986
. Diversity, rearrangement and expression of murine T cell gamma genes.
Cell.
45
:
733
–742.
14
Havran, W.L., and J.P. Allison.
1988
. Developmentally ordered appearance of thymocytes expressing different T cell antigen receptors.
Nature.
335
:
443
–445.
15
Havran, W.L., and J.P. Allison.
1990
. Thy-1+ dendritic epidermal cells of adult mice arise from fetal thymic precursors.
Nature.
344
:
68
–70.
16
Raulet, D.H., D.M. Spencer, Y.-H. Hsiang, J.P. Goldman, M. Bix, N.-S. Liao, M. Zijlstra, R. Jaenisch, and I. Correa.
1991
. Control of gammadelta T cell development.
Immunol. Rev.
120
:
185
–204.
17
Ikuta, K., T. Kina, I. MacNeil, N. Uchida, B. Peault, Y.-h. Chien, and I.L. Weissman.
1990
. A developmental switch in thymic lymphocyte maturation potential occurs at the level of hematopoietic stem cells.
Cell.
62
:
863
–874.
18
Xiong, N., and D.H. Raulet.
2007
. Development and selection of gammadelta T cells.
Immunol. Rev.
215
:
15
–31.
19
Itohara, S., P. Mombaerts, J. Lafaille, J. Iacomini, A. Nelson, A.R. Clarke, M.L. Hooper, A. Farr, and S. Tonegawa.
1993
. T cell receptor δ gene mutant mice: independent generation of αβ T cells and programmed rearrangements of γδ TCR genes.
Cell.
72
:
337
–348.
20
Baker, J.E., D. Cado, and D.H. Raulet.
1998
. Developmentally programmed rearrangement of T cell receptor Vgamma genes is controlled by sequences immediately upstream of the Vgamma genes.
Immunity.
9
:
159
–168.
21
Asarnow, D.M., D. Cado, and D.H. Raulet.
1993
. Selection is not required to produce invariant T cell receptor gamma-gene junctional sequences.
Nature.
362
:
158
–160.
22
Zhang, Y., D. Cado, D.M. Asarnow, T. Komori, F.W. Alt, D.H. Raulet, and J.P. Allison.
1995
. The role of short homology repeats and TdT in generation of the invariant gamma delta antigen receptor repertoire in the fetal thymus.
Immunity.
3
:
439
–447.
23
Mallick-Wood, C.A., J.M. Lewis, L.I. Richie, M.J. Owen, R.E. Tigelaar, and A.C. Hayday.
1998
. Conservation of T cell receptor conformation in epidermal γδ cells with disrupted primary Vγ gene usage.
Science.
279
:
1729
–1733.
24
Xiong, N., C. Kang, and D.H. Raulet.
2004
. Positive selection of dendritic epidermal gammadelta T cell precursors in the fetal thymus determines expression of skin-homing receptors.
Immunity.
21
:
121
–131.
25
Yancopoulos, G.D., S.V. Desiderio, M. Paskind, J.F. Kearney, D. Baltimore, and F.W. Alt.
1984
. Preferential utilization of the most JH-proximal VH gene segments in pre-B-cell lines.
Nature.
311
:
727
–733.
26
Yancopoulos, G.D., B.A. Malynn, and F.W. Alt.
1988
. Developmentally regulated and strain-specific expression of murine VH gene families.
J. Exp. Med.
168
:
417
–435.
27
Chiei, Y.-H., M. Iwashima, D.A. Wettstein, K.B. Kaplan, J.F. Elliott, W. Born, and M.M. Davis.
1987
. T cell receptor δ gene rearrangements in early thymocytes.
Nature.
330
:
722
–727.
28
Yancopoulos, G.D., and F.W. Alt.
1985
. Developmentally controlled and tissue-specific expression of unrearranged VH gene segments.
Cell.
40
:
271
–281.
29
Goldman, J.P., D.M. Spencer, and D.H. Raulet.
1993
. Ordered rearrangement of variable region genes of the T cell receptor gamma locus correlates with transcription of the unrearranged genes.
J. Exp. Med.
177
:
729
–739.
30
Sleckman, B.P., J.R. Gorman, and F.W. Alt.
1996
. Accessibility control of antigen-receptor variable-region gene assembly: role of cis-acting elements.
Annu. Rev. Immunol.
14
:
459
–481.
31
Stanhope-Baker, P., K.M. Hudson, A.L. Shaffer, A. Constantinescu, and M.S. Schlissel.
1996
. Cell type-specific chromatin structure determines the targeting of V(D)J recombinase activity in vitro.
Cell.
85
:
887
–897.
32
Abarrategui, I., and M.S. Krangel.
2006
. Regulation of T cell receptor-alpha gene recombination by transcription.
Nat. Immunol.
7
:
1109
–1115.
33
Durum, S.K., S. Candèias, H. Nakajima, W.J. Leonard, A.M. Baird, L.J. Berg, and K. Muegge.
1998
. Interleukin 7 receptor control of T cell receptor γ gene rearrangement: role of receptor-associated chains and locus accessibility.
J. Exp. Med.
188
:
2233
–2241.
34
Agata, Y., T. Katakai, S.K. Ye, M. Sugai, H. Gonda, T. Honjo, K. Ikuta, and A. Shimizu.
2001
. Histone acetylation determines the developmentally regulated accessibility for T cell receptor γ gene recombination.
J. Exp. Med.
193
:
873
–879.
35
Bassing, C.H., F.W. Alt, M.M. Hughes, M. D'Auteuil, T.D. Wehrly, B.B. Woodman, F. Gärtner, J.M. White, L. Davidson, and B.P. Sleckman.
2000
. Recombination signal sequences restrict chromosomal V(D)J recombination beyond the 12/23 rule.
Nature.
405
:
583
–586.
36
Xiong, N., J.E. Baker, C. Kang, and D.H. Raulet.
2004
. The genomic arrangement of T cell receptor variable genes is a determinant of the developmental rearrangement pattern.
Proc. Natl. Acad. Sci. USA.
101
:
260
–265.
37
Xiong, N., C.H. Kang, and D.H. Raulet.
2002
. Redundant and unique roles of two enhancer elements in the TCR gamma locus in gene regulation and gammadelta T cell development.
Immunity.
16
:
453
–463.
38
Li, E., T.H. Bestor, and R. Jaenisch.
1992
. Targeted mutation of the DNA methyltransferase gene results in embryonic lethality.
Cell.
69
:
915
–926.
39
Torres, R., and R. Kuhn.
1997
. Laboratory Protocols for Conditional Gene Targeting. Oxford University Press, New York. 184 pp.

Abbreviations used: DETC, dendritic epidermal T cell; E, embryonic day; ES, embryonic stem; KI, knock-in; RSS, recombination signal sequence(s); TdT, terminal deoxynucleotidyl transferase.

N. Xiong's present address is Dept. of Veterinary and Biomedical Sciences, The Pennsylvania State University, University Park, PA 16802.

Supplementary data