The genomic arrangement of T cell receptor variable genes is a determinant of the developmental rearrangement pattern

Abstract

Developmentally regulated V(D)J recombination profoundly influences immune repertoires, but the underlying mechanisms are poorly understood. In the endogenous T cell receptor Cγ1 cluster, the 3′ Vγ3 gene (closest to Jγ1) rearranges preferentially in the fetal period whereas rearrangement of the 5′ Vγ2 gene predominates in the adult. Reversing the positions of the Vγ2 and Vγ3 genes in a genomic transgene resulted in decreased rearrangement of the now 5′ Vγ3 gene in the fetal thymus and increased rearrangement of the now 3′ Vγ2 gene. The reversed rearrangement pattern was not accompanied by significant changes in chromatin accessibility of the relocated Vγ genes. The results support a model in which the 3′ location is the key determinant of rearrangement in the fetus, after which there is a promoter-dependent inactivation of Vγ3 rearrangement in favor of Vγ2 rearrangement.

Specific V genes in T cell receptor (TCR) γ (1), TCRδ (2), and IgH (3, 4) loci rearrange preferentially at different stages of ontogeny. The mechanisms that differentially regulate recombination of various V, D, and J gene segments are poorly understood. One model is that local transcriptional regulatory elements control the accessibility of corresponding recombination signal sequences (RSS) to the recombinase, thereby regulating the lineage and stage specificity of V(D)J recombination (5–9).

Another factor that may contribute to developmental regulation of gene rearrangement is the gene segment location on the chromosome. In IgH, TCRγ, and TCRδ loci, V genes that rearrange early in ontogeny are typically relatively proximal to the D or J gene segments (3, 6, 10–13). However, the evidence to date that rearrangement frequency is determined by the V gene location is purely correlative.

TCR Vγ gene rearrangements are subject to a particularly striking degree of developmental regulation. The TCR Cγ1 cluster in mice consists of four Vγ gene segments in the order (5′ to 3′) Vγ5, -2, -4, and -3, all of which recombine primarily with a single downstream J gene segment, Jγ1. In early fetal thymocytes [embryonic day (E) 15], nearly all Jγ1 rearrangements consist of Vγ3 and to a lesser extent Vγ4 rearrangements (11, 12). In contrast, Vγ2 and Vγ5 rearrangements dominate in the adult thymus (6, 11).

The ontogenic pattern of Vγ gene rearrangement is important in establishing the unique distribution of specific subsets of γδ T cells in different tissues. Early fetal γδ cells harboring Vγ3 rearrangements migrate to the skin where they represent essentially all of the resident T cells, called dendritic epidermal T cells (DECs) (14). On the other hand, Vγ2⁺ and Vγ5⁺ γδ T cells represent the majority of γδ T cells in the lymph nodes, spleen, and blood.

Several lines of evidence demonstrate that developmentally regulated Vγ gene segment recombination is an intrinsic genetically programmed process. The most direct experiment used γB, a 39-kb genomic transgene containing most of the unrearranged TCR Cγ1 cluster, in which each of the Vγ genes (Vγ2, Vγ4, and Vγ3) contained a frame-shift mutation that prevents functional expression. The Vγ genes in the γB transgene rearranged at the appropriate stage in thymic development despite the fact that the rearrangements could not directly influence the fate of cells in which they occurred (9). In like fashion, a targeted TCRδ gene deletion, which abrogates the formation of TCRγδ receptors, did not disrupt the normal developmental pattern of Vγ gene rearrangement (15). Earlier studies demonstrated that nonproductive Vγ gene rearrangements exhibit a similar rearrangement pattern as productively rearranged alleles in the same cells (16). The pattern of Vγ gene germ-line transcription correlates with the rearrangement pattern, suggesting a possible link between rearrangement and transcriptional regulatory cues in this system (6). Furthermore, the decreased rearrangement of the Vγ3 gene in the later stages of development correlated well with progressive local deacetylation (17).

We previously reported that Vγ gene promoter elements play a key role in regulating the developmental stage specificity of Vγ gene rearrangement (9). Starting with the γB transgene backbone, fragments containing the promoter regions of the Vγ2 and Vγ3 genes were reciprocally swapped to generate the γB-Pr-Sw transgene (previously called the γSw transgene). The promoter exchange reversed the pattern of rearrangement of these Vγ genes in the adult thymus, suggesting that the Vγ gene usage pattern in the adult thymus is controlled by promoter sequences.

Surprisingly, the promoter swap did not alter the pattern of Vγ transgene rearrangements in fetal thymocytes or in Vγ3⁺ DEC cells, a population that is derived from early fetal Vγ3⁺ thymocytes. Apparently, a distinct form of regulation is responsible for selective Vγ gene rearrangement in early fetal thymocytes. Here, we demonstrate that the fetal, but not the adult, pattern of rearrangement is determined by the location of Vγ genes in the gene cluster. The results represent direct evidence that V gene location is a determinant of the rearrangement pattern.

Methods

Generation of Transgenic Mice. The γB and γB-Pr-SW (formerly called γSW) transgenic mice have been described (9). The γB-Gn-Sw transgene is identical to the γB transgene except that a 2.3-kb Spel-EcoR1 fragment containing the entire Vγ2 gene segment was exchanged with a 1.6-kb HindIII-EcoRI fragment containing the entire Vγ3 gene segment. The transgene inserts were injected into fertilized (C57BL/6 × CBA/J)F₂ eggs. Transgenic founders were identified by PCR of genomic DNA (9) and backcrossed serially to CBA/J mice (National Cancer Institute, Bethesda). The mice used in the experiments had been back-crossed two to four times. Transgene copy number was determined by quantitative Southern blotting of tail genomic DNA.

PCR Primers and Linkers. L2, L4, L3, J1, V2-3′a, V3-3′a, V3-3′b, 5′tubulin, and 3′tubulin primers have been described (6, 18). Primers PSVγ2 and PSVγ3 have been also described (9) as have the BW linker and BW1 primer (19). The sequence of VG2I primer is: GCTTCGTCTTCTTCCTCCAAGG.

Semiquantitative PCR and RT-PCR. CD4^-CD8^- thymocytes were prepared by magnetic sorting (20). DECs were prepared as described (21). Genomic DNA and total RNA were prepared as described (9, 22). Semiquantitative analysis of rearrangement of the Vγ2 and Vγ3 genes was performed as described (9). The ratios of Vγ2 (Tg)/Vγ3 (Tg) transgene rearrangement levels displayed below each data set in Fig. 2 were based on the previous determination of a ratio of 1/8 in the γB (3) sample, by comparison with cell lines (9). The ratios in the other DNA samples were calculated by comparing the observed level of rearrangements based on the titrations shown to that in the γB (3) sample, normalizing based on the tubulin PCRs.

Fig. 2.

Rearrangements in DECs of γB-Gn-Sw and γB transgenic mice. DECs from all three lines of γB-Gn-Sw transgenic mice were compared with γB (3) by semiquantitative PCR (9). Transgene copy number is indicated in brackets. Three-fold serial dilutions of DEC DNA samples were PCR amplified with the following primers (see Fig. 1): L2/J1 (for Vγ2), L3/J1 (for Vγ3), and 5′tubulin/3′tubulin (for Tub). PCR products were digested with NruI (for Vγ2) or EcoRI (for Vγ3) to distinguish endogenous (E) from transgene (Tg) rearrangements. Semiquantitative PCR of tubulin was used to normalize the DNA samples. Ratios of the levels of Vγ2 to Vγ3 rearrangements of the transgene are displayed below each data set (see Methods).

The semiquantitative RT-PCR assay for quantitation of germline Vγ transcripts was modified from a previously described method (6, 20). Comparable amounts of total cellular RNA were reverse transcribed with the V2-3′a primer (for Vγ2) or the V3-3′a primer (for Vγ3) by using SuperScript II RNase H^- reverse transcriptase (GIBCO/BRL). Reverse transcription products were serially diluted and subjected to semiquantitative PCR in the presence of [α-³²P]dCTP. For amplifying Vγ2 sequences, we used the L2 or PSVγ2 primers in conjunction with the VG2I primer. For Vγ3, we used the L3 or PSVγ3 primers in conjunction with the V3-3′b primer. The PCR products were digested with NruI (for Vγ2) or EcoRI (for Vγ3) and run on a 5% polyacrylamide gel, which was dried and exposed to an x-ray film and a PhosphorImager (Molecular Dynamics). A semiquantitative RT-PCR for tubulin was used to normalize the samples.

Detecting RSS Breaks Flanking Vγ2 and Vγ3 by Linker-Mediated PCR (LM-PCR). RSS breaks flanking Vγ2 and Vγ3 were measured by LM-PCR, as described, with minor modifications (19). Briefly, thymic genomic DNA was ligated with a BW linker, serially diluted, and subjected to radioactive semiquantitative PCR with the BW1 and Vγ2-3′a primers to detect Vγ2 breaks, or with the BW1 and Vγ3-3′b primers to detect Vγ3 breaks. Semiquantitative PCR for tubulin was used to normalize the samples.

Restriction Enzyme Accessibility Assay. The assay was performed as described (23). Briefly, nuclei were prepared from E15 total thymocytes or adult CD4^-CD8^- thymocytes by Nonidet P-40 lysis and resuspended in suitable buffers supplemented with restriction enzymes (Asp-718 for Vγ2; EcoRI for Vγ3) for 1-hr digestion. Genomic DNA was purified from the digested nuclei and subjected to LM-PCR to detect restriction enzyme cleavage.

Results

A Transgenic Model to Test Role of V Gene Location in Developmentally Regulated V(D)J Recombination. The γB transgene consists of 39 kb of contiguous genomic DNA containing Vγ2, Vγ4, Vγ3, Jγ1, and Cγ1 in their normal configuration (24). We previously used the identical backbone in the construction of γB-Pr-Sw, the transgene used to demonstrate that promoter sequences regulate V gene usage in adult thymocytes (9) (Fig. 1). Starting with γB, a 2.3-kb SpeI-EcoRI fragment containing Vγ2 was reciprocally swapped with a 1.6-kb HindIII-EcoRI fragment containing Vγ3, to generate the γB-Gn-Sw construct (Fig. 1). The 5′ endpoints of the exchanged fragments were identical to those used for the γB-Pr-Sw. Therefore, the exchanged segments in the new γBGn-Sw construct contain the identical promoter regions swapped in the promoter swap construct γB-Pr-Sw and, in addition, the entire coding regions and 450–660 bp of DNA downstream of each RSS. In this way, the three transgenes can be compared directly to distinguish the roles of the promoter sequences and gene location in V-J recombination. Each of the Vγ genes (Vγ2, Vγ4, and Vγ3) in all three constructs contained in the coding region a restriction enzyme linker (NruI, XhoI, and EcoRI, respectively) that allows the transgene and its transcripts to be readily distinguished from the corresponding endogenous γ genes and transcripts (9, 25). Furthermore, because the linkers introduce frameshift mutations that render the mutant proteins nonfunctional, the transgene is expected to confer no selective advantage or disadvantage to cells that express it.

Fig. 1.

Comparison of γB, γB-Gn-Sw, and γB-Pr-Sw transgenes. The 5′ DNase hypersensitive site A (HSA) and the 3′ Cγ1 enhancer, 3′ E_Cγ1 (E) are indicated. All transgenes have the same backbone. Differences are depicted in the expanded segments. The boxes above the line are the coding (V and leader) sequences. Boxes below the line are segments subjected to exchange. Vγ2 sequences are in black and Vγ3 are hatched. Asterisks indicate restriction enzyme sites introduced into the transgene. Restriction enzymes are as follows: S, SpeI; H, HindIII; R1, EcoRI; and RV, EcoRV. Arrows indicate positions and orientation of primers: 1, L2; 2, PSVγ2; 3, VG2I; 4, V2-3′a; 5, L3; 6, PSVγ3; 7, V3-3′b; 8, V3-3′a; and 9, J1. The labeled line segments indicate probes used in Southern blots.

Four γB-Gn-Sw transgenic founders were generated and crossed with CBA mice. Three of the founders transmitted the transgene to their offspring, and one founder, a mosaic, did not transmit the transgene.

V Gene Location Determines Vγ Rearrangement Pattern in DECs. Vγ gene rearrangements in DECs, which are derived from early fetal thymocytes, were analyzed by semiquantitative PCR with a Vγ-specific primer and a common Jγ1 primer (9). These primers amplify both transgene and endogenous rearrangements, which can be distinguished by electrophoresis after cleavage with restriction enzymes specific for the transgene (NruI for Vγ2 and EcoRI for Vγ3). Compared with a typical γB transgenic line, DECs from all three γB-Gn-Sw transgenic lines exhibited substantially elevated levels of transgene Vγ2 rearrangements and substantially diminished levels of transgene Vγ3 rearrangements (Fig. 2). Normalized to transgene copy number, Vγ2 transgene rearrangements in γB-Gn-Sw transgenics were elevated 5- to 10-fold, and Vγ3 rearrangements were depressed 5- to 10-fold, in comparison with the mean values from several γB transgenics. The average ratio of transgene Vγ3 to Vγ2 rearrangements in DEC cells is ≈8:1 in γB, 12:1 in γB-Pr-Sw, and 6:1 in the endogenous locus (9). The corresponding ratio for the γBGn-Sw transgenics is ≈1:8, nearly a perfect reciprocal of the normal pattern. The mosaic γB-Gn-Sw transgenic founder exhibited a similar reversal in the pattern of rearranged Vγ3/Vγ2 genes except that the absolute Vγ gene rearrangement levels were lower (data not shown). The reversed pattern of Vγ rearrangement in DEC cells of the γB-Gn-Sw transgene was also confirmed by Southern blot analysis (Fig. 6, which is published as supporting information on the PNAS web site). These results demonstrate that swapping the location of Vγ2 and Vγ3 genes completely reversed the rearrangement pattern of Vγ2/Vγ3 genes in DECs. The consistency of this effect in different lines indicates that it is independent of transgene integration site.

V Gene Location Determines Vγ Rearrangement Pattern in Fetal but Not Adult Thymocytes. Transgene rearrangements in E14 and E15 fetal thymi were also assessed. Compared with the γB transgenics, transgene Vγ2 rearrangements in γB-Gn-Sw transgenics were elevated by an average of 4- to 5-fold whereas transgene Vγ3 rearrangements were diminished by an average of 15- to 20-fold (Fig. 3A and Table 1). Consequently, the calculated ratio of rearranged Vγ2/Vγ3 for the γB-Gn-Sw transgene (6:1) is the near inverse of the corresponding ratio for the γB transgene (1:10) or the endogenous TCRγ genes (1:4) (9). The overall level of Vγ2 plus Vγ3 gene rearrangements per transgene copy in the γB-Gn-Sw mice was only modestly lower (2- to 3-fold) than in γB mice in the E15 fetal thymocytes (Table 1). A larger difference was observed in E14 fetal thymocytes, but this is at a stage where rearrangement is first initiated and the values for all of the transgenics were very low. These data suggest that, in the fetal thymus, the relative rearrangement levels of individual Vγ genes are controlled by the locations of the V gene segments in the TCR Cγ1 cluster, with little or no influence from the upstream promoter regions, the V coding regions, or the immediate downstream regions of the Vγ genes including the RSSs.

Fig. 3.

Semiquantitative PCR of Vγ2 and Vγ3 rearrangements in early fetal and adult thymocytes of transgenic mice. (A) Comparison of Vγ rearrangement in E14 thymus and E15 thymocytes. (B) Comparison of Vγ rearrangement in adult CD4^-CD8^- thymocytes. The Vγ2 standard is hybridoma DN2.3, which contains two copies of rearranged Vγ2. The Vγ3 standard is the WRD.34 cell line, which contains two copies of rearranged Vγ3. See Fig. 2 legend for other details. Similar data for other lines are summarized in Table 1.

Table 1. Rearrangement of Vγ2 and Vγ3 genes in thymocytes of the transgenic mice.

		E14 fetus			E15 fetus			Adult
Line	Copy No.	V2*	V3*	V2/V3†	V2	V3	V2/V3	V2	V3	V2/V3
γB-Gn-Sw1	20	0.00032	0.00012	2.7	0.0069	0.0012	5.8	0.130	0.032	4.1
γB-Gn-Sw2	28	0.00019	0.00002	11.6	0.0060	0.0013	4.9	0.069	0.032	3.0
γB-Gn-Sw3	17	ND	ND	ND	0.0073	0.0012	6.2	0.081	0.020	3.9
γB-Gn-Sw mean		0.00026	0.00007	7.1	0.0067	0.0012	5.6	0.093	0.028	3.7
γB-Pr-Sw mean‡		0.00003	0.00150	0.015	0.0018	0.0680	0.040	0.042	0.220	0.19
γB mean‡		0.00005	0.00325	0.024	0.0016	0.0190	0.095	0.120	0.049	2.47
End‡	2	0.00120	0.00330	0.364	0.0247	0.0930	0.266	0.380	0.015	25.33

ND, not done.

^*V2 and V3 refer to rearrangements of Vγ2 and Vγ3 per gene copy from PCR data

^†V2/V3 is ratio of rearranged Vγ2 to rearranged Vγ3 per gene copy

^‡Data for End (endogenous TCRγ), γB, and γB-Pr-Sw transgenes are from Baker (9)

In adult thymocytes, Vγ2 rearrangements were more abundant than Vγ3 rearrangements in both γB-Gn-Sw and γB transgenic thymocytes (Fig. 3B, Table 1, and Fig. 7, which is published as supporting information on the PNAS web site). The average ratio of Vγ2/Vγ3 rearrangements in the three γBGn-Sw transgenic lines was 3.7, comparable to 5.0 in the γB transgenic mice (9), suggesting that Vγ gene location has no or little effect on Vγ gene rearrangement at the adult stage. Similar results were obtained in an analysis of adult splenic γδ T cells (data not shown).

Extent of RSS Breaks Correlates with Location-Dependent Vγ Gene Rearrangement Pattern. To assess the possible steps of V-J recombination impacted by Vγ gene location, we examined whether V gene location determined the extent of RAG-mediated DNA breaks at the RSSs flanking the Vγ genes in the various transgenes. Breaks were assayed in thymocyte DNA by LM-PCR amplification of signal ends (19). By performing a limited number of PCR cycles, the experiment relied on the high number of transgene copies to distinguish breaks of transgene origin from those of endogenous origin. Under these conditions, no or little PCR products were detected in nontransgenic DNA from E15 fetal thymocytes (Fig. 4A). Significantly, in E15 fetal thymocytes, the γB-Gn-Sw transgene contained >5-fold more Vγ2 breaks and >5-fold fewer Vγ3 breaks than the control γB transgene, as determined by phosporimaging and taking into account the difference in transgene copy number (Fig. 4A Left). These changes correlated well with the rearrangement pattern and suggested that the location of Vγ genes in the transgene determined the RAG cleavage pattern. In adult thymocytes, Vγ2 breaks were predominant, and Vγ3 breaks were rare in both γB and γB-Gn-Sw transgenes, also consistent with the rearrangement pattern (Fig. 4A Right). In the γB-Pr-Sw transgene, in contrast, Vγ2 breaks in adult thymocytes were diminished and Vγ3 breaks were sharply increased, consistent with the rearrangement pattern of this transgene.

Fig. 4.

RSS breaks and germ-line transcription of Vγ2 and Vγ3 genes in the transgenes. (A) Comparison of RSS breaks flanking Vγ2 and Vγ3 in the transgenes. Samples were serially diluted 5-fold (for E15 fetal samples). Semiquantitative PCR for tubulin was used as a loading control. (B) Vγ2 and Vγ3 germ-line transcripts emanating from transgenes in fetal and adult thymocytes. Shown is semiquantitative RT-PCR analysis of germ-line transcripts in total RNAs from adult CD4^-CD8^- thymocytes and E14 fetal thymus of γB and γB-Gn-Sw transgenic mice. Primers for reverse transcription (RT) included oligo(dT) (for the tubulin control), V2-3′a (for Vγ2), or V3-3′a (for Vγ3) (see Fig. 1). RT products were then 3-fold serially diluted and subjected to the semiquantitative PCR. Either 5′tubulin/3′tubulin primer set (for Tub) or L2/VG2I primer set (for Vγ2), or PSVγ3/V3-3′b primer set (for Vγ3) was used for PCR of corresponding RT products. Vγ2 and Vγ3 PCR products were digested with Nru1 (for Vγ2) or EcoRI (for Vγ3) before gel fractionation. RNA samples without RT were also directly subjected to the semiquantitative PCR to confirm the absence of DNA contamination in the samples. (C) Germ-line transcripts of γB and γB-Pr-Sw transgenes in adult CD4^-CD8^- thymocytes. The PSVγ2/VG2I primer set was used for Vγ2 PCR. All others were as in B.

Germ-Line Transcription of Vγ2/Vγ3 Genes in γB and γB-Gn-Sw Transgenes. Germ-line transcription of endogenous Vγ genes has been shown to correlate with the pattern of gene rearrangement (6). Vγ gene germ-line transcripts directed by the different transgenes were assayed by reverse transcription with primers 3′ of the unrearranged Vγ3 or Vγ2 genes but within the exchanged segments in the γB-Gn-Sw transgene. Serially diluted cDNA was amplified with a 5′ primer in the Vγ coding region, and a downstream 3′ primer internal to the primer was used for reverse transcription. The PCR products were cleaved with transgene-specific restriction enzymes to distinguish transgene-encoded and endogenous germ-line transcripts.

In E14 fetal thymocytes, transgene-derived germ-line Vγ2 transcripts were slightly more abundant in the γB-Gn-Sw sample than in the γB sample, and Vγ3 germ-line transcripts slightly less abundant (Fig. 4B Left). When data from different lines were normalized for transgene copy number and averaged, these differences were small, averaging 2.4-fold for Vγ2 and 1.5-fold for Vγ3. Although the increase in Vγ2 germ-line transcription did not grossly differ in magnitude from the increase in the level of Vγ2 rearrangement (4- to 5-fold), the minor decrease in Vγ3 germ-line transcription contrasted with the much larger decrease in Vγ3 rearrangements (15- to 20-fold) and RSS breaks evident with the γB-Gn-Sw transgene at the fetal stage. Thus, the reduced rearrangement of Vγ3 was not accompanied by a corresponding decrease in germ-line transcription.

There was little difference in germ-line transcription in adult thymocytes between the γB and γB-Gn-Sw transgenes, as expected based on their similar rearrangement pattern (Fig. 4B Right). As a comparison, we examined germ-line transcription of the γB-Pr-Sw transgene, which exhibits a reciprocal rearrangement pattern in adult thymocytes compared with γB. The analysis revealed that Vγ3 germ-line transcripts per transgene copy were increased by an average of 5-fold compared with γB whereas Vγ2 transcripts were decreased by an average of 3-fold, consistent with the expected effects of promoter regions (Fig. 4C). The magnitude of these differences in germ-line transcription correlated well with the changes in rearrangement observed with this transgene (9). We conclude that, whereas the promoter-dependent adult pattern of transgene rearrangement correlates well with changes in germ-line transcription, the location-dependent fetal pattern of transgene rearrangement does not.

Restriction Enzyme Accessibility of Vγ Genes in γB and γB-Gn-Sw Transgenics. The changed rearrangement pattern as a result of relocalization of the Vγ genes in the transgene could be due to changes in the “accessibility” of the local chromatin to the recombinase. Vγ gene accessibility was examined by assaying the susceptibility of Vγ genes in thymocyte nuclei to restriction enzyme cleavage, by using LM-PCR. Preliminary analysis showed that accessibility of the Vγ2 and Vγ3 genes to restriction enzymes was substantially lower in irrelevant perfused adult liver cells than in adult thymocytes (data not shown). In E15 fetal thymocytes, an Asp-718 site in the Vγ2 coding region (Fig. 5A) was slightly less accessible in γB-Gn-Sw transgenics, despite the increase in rearrangement (Fig. 5B). The PCR products were essentially all cleaved by NruI, specific for a site unique to the transgenes (Fig. 5B). Also in E15 fetal thymocytes, a transgene-specific EcoRI site in the coding region of the Vγ3 gene near the RSS (Fig. 5A) was equally accessible, or only slightly less accessible (depending on the experiment) in the γB-Gn-Sw transgene compared with the γB transgene (Fig. 5C). A minor reduction in accessibility of Vγ3 gene in the γB-Gn-Sw transgene was observed at another restriction enzyme site, ApaLI (data not shown). An analysis of adult thymocyte nuclei revealed a pattern very similar to that of fetal thymocytes: the two transgenes showed similar levels of accessibility of the Asp-718 site in Vγ2, and the γB-Gn-Sw transgene exhibited a slight reduction in accessibility of the EcoRI or ApaLI sites in the Vγ3 gene (Fig. 5 D and E and data not shown). Thus, the strongly altered pattern of rearrangement and RSS breakage due to relocation of the Vγ genes in fetal thymocytes was not accompanied by similar alterations in Vγ gene accessibility. These findings suggest that V gene location determines the rearrangement pattern predominantly by a mechanism independent of accessibility.

Fig. 5.

Restriction enzyme accessibility of Vγ2 and Vγ3 genes in the transgenes. (A) Restriction enzyme sites in Vγ2 and Vγ3 genes used in the accessibility assays. The NruI and EcoRI sites indicated with asterisks are transgenespecific sites. Vertical arrows indicate the restriction sites probed in these analyses. Nuclei of E15 fetal thymocytes (B and C) or adult CD4^-CD8^- thymocytes (D and E) were treated with restriction enzymes to detect the accessibility of the Vγ2 (Asp-718) or Vγ3(EcoRI) genes. After DNA extraction, cleavage was assayed by radioactive LM-PCR. (B and D) The accessibility of the Asp-718 site in the Vγ2 gene. Samples were serially diluted 5-fold. Semiquantitative PCR for the tubulin gene was used as a loading control. A portion of each PCR product was digested with NruI before gel fractionation to confirm its origin from the transgene. (C and E) The accessibility of the EcoRI site in Vγ3 gene was determined. In this assay, the DNA was digested with a second restriction enzyme, ApaL1 (located 5′ of the EcoR1 site) after DNA extraction but before LM-PCR. Therefore, the ApaL1 PCR product corresponds to DNA that was uncleaved by EcoRI and serves as an internal reference. In C, two experiments are shown for comparison.

Discussion

The results reported here demonstrate that the locations of Vγ2 and Vγ3 genes in a TCRγ transgene determine the pattern of gene rearrangement in fetal thymocytes and DECs. The results represent an experimental demonstration that the relative location of V gene segments is a direct determinant of preferential V gene recombination. It is possible that a similar principle contributes to developmentally regulated V gene segment rearrangement in the IgH, TCRδ, and possibly other receptor gene families. Previous studies have shown that RSSs (26) and promoters (9) can differentially regulate gene segment rearrangement during development, but the present results argue that these elements do not play a primary role in determining the fetal/DEC pattern of rearrangement.

V gene location did not affect the rearrangement pattern in the adult thymus. Therefore, Vγ gene rearrangement at the adult stage is largely dependent on promoters, as previously suggested (9). The separate regulation of the fetal and adult patterns of rearrangement was also suggested by the finding that E2A deficiency reverses the adult but not the fetal pattern of rearrangement (27). Taken together, the results indicate that the “developmental switch” in Vγ gene usage is imposed by at least two mechanisms, one sensitive to gene location and the other dependent on differential V-promoter activity.

In comparing the transgenes, the RSS of the most proximal Vγ gene was more frequently broken in chromatin of fetal thymocytes, suggesting that location of the gene determines its susceptibility to RAG-mediated cleavage. It was notable, however, that accessibility and germ-line transcription of Vγ2 and Vγ3 genes in fetal thymocytes were not substantially altered between the γB and γB-Gn-Sw transgenics. These results suggest that both Vγ2 and Vγ3 genes are substantially accessible in early fetal thymocytes of both transgenes, consistent with a report that Vγ2 and Vγ3 in the endogenous locus exhibited similar levels of histone hyperacetylation at the fetal stage (17). Although both genes are apparently accessible at the early fetal stage, the downstream gene undergoes preferential RSS breakage and recombination.

One explanation of the location dependence of rearrangement in the fetal stage is that the locus is separated into discrete domains under independent regulation, and the swap of the Vγ genes places each into the domain of the other. For example, it has been proposed that D_H-J_H joining activates the rearrangement of a domain containing 3′ V_H genes in the IgH locus, but that signaling by IL-7 or the v-abl tyrosine kinase is necessary for rearrangement of V_H genes in upstream domains (28, 29). If Vγ genes are organized in domains, however, their regulation is not well correlated with accessibility or germ-line transcription at the fetal stage.

Another possibility is that proximity to Jγ1 itself may determine the Vγ gene rearrangement frequency at the fetal stage. Proximity to Jγ1 may increase the likelihood of random interactions between the two participating RSSs, or there may be a processive component to the recombination process. Preferential pairing of the most proximal Vγ RSS with the Jγ1 RSS could also account for increased levels of breaks observed at the RSS of the proximal Vγ gene because cleavage occurs preferentially at paired RSSs.

It is also possible that the key determinant of rearrangement at the early fetal stage is proximity of the Vγ gene to a downstream transcriptional regulatory element, reminiscent of a study showing that proximity of globin genes to the β globin locus control region determines the extent of globin gene transcription early in erythropoiesis (30). We disfavor this explanation for our results because we observed that Vγ gene location is not a primary determinant of germ-line transcription levels or accessibility, both properties that would arguably be associated with enhancer activity. Furthermore, we showed that the fetal pattern of rearrangement was normal on a chromosome from which the obvious candidate element in the downstream region, 3′ E_Cγ1, was deleted (20). It remains possible that proximity to another, undefined, downstream element determines the fetal rearrangement pattern, but by a mechanism other than regulating locus accessibility.

At the adult stage, Vγ3 rearrangement is repressed, and Vγ2 rearrangement predominates. A strong reduction in Vγ3 germline transcription and histone hyperacetylation accompanies this switch (6, 17). When the Vγ2 and Vγ3 promoters were swapped, Vγ2 rearrangement decreased and Vγ3 rearrangements increased (9). As shown here, the changes in rearrangement were mirrored by comparable changes in germ-line transcription. Thus, promoter-dependent repression of Vγ3 rearrangement is likely accompanied by reduced accessibility and transcription of Vγ3 and may also involve increased accessibility of Vγ2. The picture that begins to emerge is one where the two genes are both relatively accessible in the early stages, with an advantage to the more proximal gene (normally Vγ3), followed by the late stage where the Vγ3 gene is repressed and/or Vγ2 is activated, leading to a strong preference for Vγ2 rearrangement.