A Novel Element Upstream of the Vγ2 Gene in the Murine T Cell Receptor γ Locus Cooperates with the 3′ Enhancer to Act as a Locus Control Region

The rearranged (rr)V2H⁺E⁺ transgene was previously described as the γ transgene containing the EcoRI–SalI fragment of the G8 TCR-γ gene 22,25. The rrV2H⁺E⁻ transgene was identical except that the 2.8-kb KpnI–SalI fragment containing 3′E_Cγ1 was removed from the 3′ end. The rrV2H⁻E⁺ transgene lacked the 1.5-kb EcoRI–NcoI fragment containing HsA at the 5′ end. The rrV2H⁻E⁻ transgene lacked both of these fragments. The γD(H⁺E⁺) transgene was assembled from BALB/c DNA derived from phage clones. It included the 5-kb EcoRI fragment containing Vγ2 and HsA and a 15.5-kb MboI fragment containing Jγ1, Cγ1, and 3′E_Cγ1 that extended from 4.8 kb upstream of Jγ1 to 4 kb downstream of Cγ1. Compared with germline DNA, the transgene lacked 18.5 kb of DNA between Vγ2 and Jγ1, including the Vγ3 and Vγ4 genes. The Vγ2 gene in the γD transgenics contained an XhoI linker at the ClaI site in the coding region that disrupted the reading frame and allowed discrimination between transgenes and endogenous genes. The γD(H⁺E⁻) transgene was identical to γD(H⁺E⁺) except it lacked the 2.6-kb KpnI–MboI 3′ fragment containing 3′E_Cγ1. The γD(H⁻E⁺) transgene lacked the 3.5-kb EcoRI–XbaI 5′ fragment containing HsA. The γD(H⁻E⁻) transgene lacked the 2.6-kb 3′ fragment containing 3′E_Cγ1 and a 1.5-kb EcoRI–NcoI fragment containing HsA. The transgene constructs, free of vector DNA, were injected into fertilized (C57BL/6 × CBA/J)F₂ eggs. Transgenic founders were either analyzed directly or were backcrossed repeatedly to B6 mice (rrV2 lines) or CBA/J mice (γD lines; purchased from the National Cancer Institute, Bethesda, MD) to generate transgenic lines. Mice were bred and maintained in specific pathogen-free facilities at the University of California at Berkeley.

The DNase I hypersensitive assays were performed on thymocytes and LPS blasts as described 26 except that the cells were lysed in a saponin solution 27. The quantities of DNase I (Type IV; Sigma Chemical Co.) per tube were as follows: 0.2, 0.4, 0.8, 1.6, 3.2, 6.4, 12.8, 25, and 50 μg. The control tube contained water. The DNase I hypersensitive assays on liver were performed as previously described 28. LPS blasts were made by incubating spleen cells, from which CD4⁺ and CD8⁺ cells had been depleted by complement lysis, with 40 μg/ml LPS (Salmonella typhosa; Difco Labs., Inc.) at a concentration of 2 × 10⁶ cells/ml for 3 d. More than 90% of the resulting cells stained positive for the B cell marker B220.

Peripheral T cells were prepared from a mixture of spleen and lymph node cells by passing the cells over nylon wool columns. To purify α/β and γ/δ T cells, the isolated peripheral T cells were combined with thymocytes, and the mixture was partially depleted of CD4 and CD8 cells by complement lysis followed by cell sorting with an Epics Elite flow cytometer (Coulter Immunology) using anti-γδ (GL3–FITC) and anti-αβ (H57.597–biotin) antibodies. Fetal thymus was timed by designating the day of the plug as day 0. The whole fetal thymus, including the capsule, was used to isolate RNA. CD4⁻CD8⁻ double-negative (DN) thymocytes were prepared by complement lysis of whole thymocytes with anti-CD4 (RL172) and anti-CD8 (3.168.8 or AD415) antibodies and a mixture of guinea pig complement (GIBCO BRL) and rabbit complement followed by isolation of live cells on a Ficoll gradient. CD4⁺CD8⁺ thymocytes (double-positives [DPs]) and CD4⁺CD8⁻ and CD8⁺CD4⁻ thymocytes (single-positives [SPs]) were sorted on an Epics Elite flow cytometer (Coulter Immunology) using anti-CD4 and anti-CD8 antibodies. The enriched DN populations of the γD lines employed for the analysis of transcription in different developmental stages (see Fig. 4) were not sorted and hence were only ∼50% pure. For semiquantitative transcription analysis of γD lines, DN, DP, and SP thymocytes were sorted to >99% purity using anti-CD4 and anti-CD8 antibodies.

Total RNA was prepared by the single step method using water-saturated phenol as described 29. 20 μg of tRNA was added as a carrier. Genomic DNA was prepared from defined numbers of cells as described 29. Lambda DNA (2.5 μg; New England Biolabs, Inc.) was added as carrier.

RNase protection assays 30 were performed with a riboprobe generated using T7 RNA polymerase and a linearized pKS Bluescript™ vector (Stratagene Inc.) construct containing the KpnI–BsrI fragment (273 bp) spanning the V–J junction of the G8 γ gene. The control riboprobe specific for γ actin mRNA 31 was generated using SP6 RNA polymerase. Densitometric analysis was performed using a PhosphorImager^® (Molecular Dynamics).

Serial threefold dilutions of DNA were prepared in the presence of 50 μg/ml bacteriophage lambda DNA (New England Biolabs, Inc.), and PCR reactions were performed as described 11 with the L2 and J1 primers 32. The transgene contained an XhoI linker in the V2 coding sequence; thus, digestion of the products with XhoI distinguished the endogenous product from the transgene product. The sample dilutions were compared with a standard curve prepared with DNA from the DN2.3 hybridoma 33, which contains two Vγ2 rearranged genes and four tubulin genes. A β tubulin PCR was used to normalize the samples. The bands were visualized by autoradiography, and their intensities were measured on a PhosphorImager^®. Reverse transcriptase (RT)-PCR was employed for a parallel analysis of transcript levels in the γD transgenic lines. The procedure was done as described using either oligo-dT or J1 primer for reverse transcription and the L2 and J1 primers (or tubulin primers) for PCR 32, with or without RT, except that 1 μCi α-[³²P]dCTP was added during the PCR amplification step, and 28 cycles of amplification were performed. The PCR products were digested and analyzed as described for the genomic PCR.

Anti–Vγ2 TCR (UC3-10A6) and anti–δ TCR (GL3) were purified and conjugated with biotin and FITC, respectively. Anti-CD8α–Tricolor was purchased from Caltag Labs., and anti-CD4–Red 613 was from GIBCO BRL. Unseparated thymocytes from adult mice were stained with all four antibodies in the first step and streptavidin–PE (Molecular Probes, Inc.) in the second step. Gated TCR-γ/δ⁺ CD4⁻CD8⁻ thymocytes were examined for Vγ2 expression on an Epics XL-MCL flow cytometer (Coulter Immunology).

Initially, we compared the in vivo activity of two γ transgene constructs consisting of Vγ2, 4, and 3 gene segments upstream of the Jγ1-Cγ1 genes, all in their germline configurations. The two constructs were identical except that one lacked a 2.8-kb 3′ fragment that contains 3′E_Cγ1 (Fig. 1 A shows a map of the germline Cγ1 locus). We found that several independent transgenic lines of each type consistently underwent rearrangement of Vγ2 to Jγ1 in thymocytes and that both constructs were efficiently transcribed (data not shown; see below for a similar analysis). These data indicated that 3′E_Cγ1 is not absolutely required for either transgene rearrangement or expression and raised the possibility that the constructs contained a second cis-acting enhancer-like element. A clue to the site of such an element came from a previous study of a transgene construct (19L5) containing a rearranged Vγ2–Jγ1Cγ1 gene lacking 3′E_Cγ1 that was not expressed in vivo (reference 20 and Raulet, D., unpublished data). Compared with the constructs above, 19L5 lacked a 1.5-kb segment of DNA upstream of Vγ2 on its 5′ end. These data raised the possibility that a relevant enhancer element might lie on this 1.5-kb DNA segment 5′ of Vγ2.

Examination of the region upstream of Vγ2 demonstrated a clear DNase I hypersensitive site, denoted HsA, in adult thymocytes (Fig. 1 B). HsA mapped to the region 3 kb upstream of Vγ2, corresponding to the 5′ region that was absent from the 19L5 construct compared with the rearrangement constructs described above. The site was not DNase hypersensitive in B lymphocytes (LPS blasts, 90% B cells) or liver cells (Fig. 1 B).

In a parallel analysis, the 3′E_Cγ1 region was weakly hypersensitive in normal adult thymocytes (data not shown). The DNase hypersensitivity of 3′E_Cγ1 (designated HsE) was more clearly demonstrated in a transgenic line with 15 copies of an integrated TCR-γ transgene, called γB, that consists of 40 kb of contiguous germline DNA from the γ locus (Fig. 1 C; reference 11). HsE was also hypersensitive to DNase I in B cells of the transgenic mice but was not hypersensitive in liver cells. Several other hypersensitive sites, most of them weak, were also detected in the transgene, but these have not been corroborated in nontransgenic cells (data not shown).

To systematically investigate the transcription-enhancing activities of HsA and 3′E_Cγ1 in vivo, we compared four transgene constructs containing a prerearranged Vγ2–Jγ1Cγ1 gene (Fig. 2 A). The rrV2H⁺E⁺ construct containing both HSA and 3′E_Cγ1 was previously described 25. The rrV2H⁺E⁻ construct contained HsA but lacked the 2.8-kb 3′ fragment containing 3′E_Cγ1, rrV2H⁻E⁺ contained 3′E_Cγ1 but lacked a 1.5-kb 5′ fragment containing HsA, and rrV2H⁻E⁻ lacked both the 5′ and 3′ fragments. Founders were either killed and analyzed directly or bred to generate transgenic lines. Transgene copy numbers were determined by Southern blot analysis. In the cases where founder mice were analyzed directly, we determined transgene copy number in the cells being examined to minimize the effects of the transgene mosaicism that sometimes occurs in founder animals.

A quantitative RNase protection assay was used to measure Vγ2 transcripts in RNA from peripheral T cells and thymocytes from the transgenic mice. The riboprobe spanned the unique V–J junctional region of the transgene, allowing the specific detection of transgene-encoded transcripts as full length protected products. γ Transcript levels were normalized by inclusion of a control γ actin probe in each reaction. The results demonstrated that the transgene was efficiently transcribed in peripheral T cells from all three rrV2H⁺E⁺ lines and from all six rrV2H⁺E⁻ lines (Fig. 2 B). Similarly, transgene transcription was detected in thymocytes from all of these transgenic lines (Fig. 2 B). Transcription was T cell–specific, as no transcripts could be detected in B cells or kidney cells from several representative transgenic lines (data not shown). In contrast to the transgenes containing HsA, transgene transcription could not be detected in either peripheral T cells or thymocytes of the five rrV2H⁻E⁻ transgenic lines. Thus, the fragment containing HsA consistently enhances T cell–specific γ gene expression in chromatin templates in the absence of 3′E_Cγ1.

In contrast to the consistent expression of transgenes containing HsA, sporadic expression was observed in the case of the rrV2H⁻E⁺ transgene, which contained 3′E_Cγ1 but not HsA. Of the 14 lines tested, expression was detected in 5 or 6 lines in peripheral T cells and in 7 or 8 lines in thymocytes (Fig. 2 B). No expression of the transgene was detected in B cells or kidney cells from several representative transgenic lines (data not shown). Sporadic expression of the rrV2H⁻E⁺ transgene was dependent on the 3′ fragment containing 3′E_Cγ1 because, as already discussed, the rrV2H⁻E⁻ transgene lacking this fragment was not expressed in five independent lines. Hence, 3′E_Cγ1 can enhance transcription in chromatin templates but is subject to transgene position effects.

To allow quantitative comparisons, transgene expression levels in peripheral T cells were plotted against transgene copy number. One unit of transcripts was defined as the level of transcripts directed by an endogenous rearranged Vγ2 gene. This value was determined by parallel analysis of the DN2.3 γ/δ cell line, which contains two rearranged Vγ2 alleles (Fig. 2 B). In transgenic lines that contained HsA (rrV2H⁺E⁺ and rrV2H⁺E⁻), the graphs revealed a roughly proportional relationship between the number of integrated transgene copies and the levels of transgene expression (Fig. 2 C). Furthermore, the slope of the graphs was ∼1, indicating that the level of transcripts per transgene copy was roughly the same as the level directed by an endogenous Vγ2 gene. Even an rrV2H⁺E⁺ transgenic line with only two transgene copies exhibited a similar level of transgene expression per copy as the endogenous gene (Fig. 2 D). In contrast, the rrV2H⁻E⁺ transgene, which lacked HsA but contained 3′E_Cγ1, was transcribed at detectable levels in less than half of the lines (Fig. 2C and Fig. D). The lines where transcripts were detectable were all high-copy lines. No transcripts were detected in the lines harboring the rrV2H⁻E⁻ construct. These results demonstrated that the transgenes that contained HsA exhibited position-independent, roughly copy number–dependent transcription of the transgene in peripheral T cells. In contrast, the transgene that contained 3′E_Cγ1 but not HsA exhibited severe position effects.

Transgene expression in γ/δ cells was investigated by determining transcript levels in sorted γ/δ cells in one transgenic line of each type and by assessing the effect of the transgene on the percentage of Vγ2⁺ cells among thymic γ/δ cells in several lines. Abundant transgene transcripts were present in γ/δ T cells from the three lines examined, representing an rrV2H⁺E⁺ line, a high-expressing rrV2H⁻E⁺ line, and an rrV2H⁺E⁻ line (Fig. 3 A). This result was also confirmed by RT-PCR assay using purified peripheral γ/δ T cells (data not shown). Similar levels of transcript were found in sorted α/β T cells from the three lines. The expression of the transgene in α/β T cells is probably due to the absence from the transgene of a transcriptional silencer that inhibits expression of endogenous γ genes in α/β T cells 24,34 (see Discussion).

Flow cytometry was employed to determine the percentage of Vγ2⁺ thymic γ/δ cells in transgenic lines of each type (Fig. 3 B). We chose lines that had the most similar transgene copy numbers to minimize the effect of gene dosage. In nontransgenic mice, ∼35–50% of thymic γ/δ cells expressed Vγ2. The percentage was unaffected in two rrV2H⁻E⁻ lines (41–48%) but was elevated to 80–94% in the two rrV2H⁺E⁺ transgenic lines. The percentage was also elevated in two rrV2H⁺E⁻ lines (∼70%) and two rrV2H⁻E⁺ lines that exhibited high levels of transgene transcripts in the thymus (∼80%). In contrast, two rrV2H⁻E⁺ lines that were expressed poorly at the mRNA level also showed no enhancement in the percentage of Vγ2⁺ cells (30–37%). Thus, transgene expression at the mRNA level in bulk populations correlated with Vγ2 surface expression in γ/δ cells. Furthermore, the position effects exhibited by the rrV2H⁻E⁺ transgene in bulk populations were recapitulated in the analysis of γ/δ cells.

Normalization of the transcript levels determined by RNase protection to transgene copy number demonstrated that the rrV2H⁺E⁻ transgene was expressed at higher levels in peripheral T cells than in thymocytes in all six transgenic lines, by an average of 4.5-fold (Fig. 2 D). In contrast, the rrV2H⁻E⁺ transgene was expressed at lower levels in peripheral T cells than in thymocytes in all the lines where expression could be detected, by an average of fivefold. In the rrV2H⁺E⁺ lines, the transcript levels in thymocytes were similar to the levels in peripheral T cells, with one low-copy line exhibiting marginally higher (twofold) expression in thymocytes. These data suggested that HsA and 3′E_Cγ1 are differentially regulated in peripheral T cells and thymocytes.

To clarify the developmental activity of the two elements, we examined representative lines for transgene expression during thymocyte ontogeny and in subsets of adult thymocytes (Fig. 4). The transgene with both elements, rrV2H⁺E⁺, was expressed well in fetal thymocyte populations from day 14–18 of gestation. Similarly, two rrV2H⁻E⁺ lines that exhibited transgene expression in adult thymocytes also exhibited substantial transgene expression in fetal thymocytes. In contrast, the rrV2H⁺E⁻ transgene was expressed very poorly in fetal thymocytes in both lines tested. We conclude that the HsA element displays poor enhancing activity in fetal thymocytes, whereas the 3′E_Cγ1 element, when not subject to position effects, evinces relatively strong activity in fetal thymocytes.

Transgene expression levels were also determined in adult immature CD4⁻CD8⁻ (DN) thymocytes, immature CD4⁺CD8⁺ (DP) thymocytes, and a mixture of the relatively mature SP CD4⁺CD8⁻ and CD4⁻CD8⁺ thymocytes (Fig. 4). The rrV2H⁺E⁺ transgene was expressed well in all of these cell populations. Consistent with the ontogeny data, all four rrV2H⁺E⁻ transgenics tested exhibited poor expression in immature DN and DP thymocytes but strong expression in SP thymocytes and peripheral T cells. In contrast, in two lines where the rrV2H⁻E⁺ transgene was expressed well in unseparated thymocytes, expression was relatively strong in the DN, DP, and SP populations but weak in peripheral T cells (Fig. 4). These results suggest that 3′E_Cγ1, when not subject to negative position effects, functions well as an enhancer in immature thymocytes. In contrast, HsA by itself does not enhance transcription in immature thymocytes. As expected, little or no transgene expression was observed in DN thymocytes from two rrV2H⁻E⁺ lines in which the transgene was expressed poorly in unseparated thymocytes and from the one rrV2H⁻E⁻ line tested (data not shown).

Four new transgenic recombination substrates were prepared to examine the role of HsA and 3′E_Cγ1 in γ gene recombination (Fig. 5 A). The γD construct consisted of a 5-kb genomic fragment containing HsA and Vγ2 attached to a 15.5-kb genomic fragment containing Jγ1, Cγ1, and 3′E_Cγ1; γD(H⁺E⁻) was identical to γD except it lacked 2.6 kb of DNA containing 3′E_Cγ1; γD(H⁻E⁺) contained the 3′E_Cγ1 fragment but lacked 3.5 kb of DNA containing HsA; and γD(H⁻E⁻) lacked the 3′E_Cγ1 fragment as well as a 1.5-kb fragment of DNA encompassing HsA. In all of the transgenes, the Vγ2 gene contained a frameshift mutation to prevent expression of a functional protein. Rearrangement and expression of the transgenes was determined by semiquantitative PCR or RT-PCR, respectively, in thymocyte populations that had been enriched in DN cells (∼50% DN thymocytes).

Vγ2–Jγ1 transgene rearrangements were easily detected in the single γD transgenic line that was examined (Fig. 5b and Fig. c). For comparative purposes, a separate analysis showed that the level of transgene rearrangement was one half to one third that of endogenous Vγ2 gene rearrangement levels after normalizing for gene copy. Approximately similar levels of rearrangement were detected in five of the six γD(H⁺E⁻) transgenic lines, which lacked 3′E_Cγ1; one line exhibited low levels of rearrangement. These findings corroborated the initial data in which 3′E_Cγ1 was not necessary to support γ gene rearrangement in transgenic substrates. The role of HsA in stimulating recombination was suggested by the results with the γD(H⁻E⁻) transgene, which was identical to γD(H⁺E⁻) except that it lacked the HsA fragment. Rearrangement was undetectable in three of these transgenic lines and reduced by a factor of three to five in the remaining three lines. Hence, although low levels of rearrangement occurred in some transgenic lines in the absence of both HsA and 3′E_Cγ1, the fragment containing HsA stimulated high levels of rearrangement.

Rearrangement of the γD(H⁻E⁺) transgene was approximately normal in one transgenic line, undetectable in two lines, and reduced severalfold in a fourth line (Fig. 5b and Fig. c). This pattern of rearrangement, indicating clear position effects, cannot be clearly distinguished from the pattern observed in the γD(H⁻E⁻) lines. Therefore, it is unclear from these data whether the 3′E_Cγ1 element plays a discrete role in stimulating γ gene rearrangement (see Discussion).

Transcripts of the rearranged genes were detected by RT-PCR in each of the γD, γD(H⁺E⁻), and γD(H⁻E⁺) lines where recombination was detected (Fig. 5 B), and the relative levels were roughly correlated with the extent of rearrangement. In contrast, no such transcripts were detected in the γD(H⁻E⁻) lines that exhibited low levels of rearrangement, supporting the earlier conclusion that transcription of the rearranged genes requires HsA and/or 3′E_Cγ1. To confirm that the developmental pattern of transgene transcription in the γD lines paralleled that of the rrV2 lines, sorted DN, DP, and SP thymocytes from γD(H⁺E⁺) and γD(H⁺E⁻) lines were assayed for transcripts of the rearranged transgene by semiquantitative RT-PCR (Fig. 5 D). The results demonstrate that transgene expression was high in each population from the γD(H⁺E⁺) line but was lower in DNs, undetectable in DPs, and high in SPs from the γD(H⁺E⁻) line, consistent with the results from the rrV2 lines. The weak signal in the DNs of the γD(H⁺E⁻) line is likely derived from the 5–10% of γ/δ T cells present in this population, as the transgene transcripts were hardly detectable in the CD3⁻CD4⁻CD8⁻ population of this line (data not shown).

With the use of multiple restriction enzyme digests, we localized the DNase I hypersensitive site associated with HsA to a 462-bp PstI–NcoI fragment (data not shown). Although we have not proven that this small fragment contains the functional site defined by the transgenic studies, other studies have shown a colocalization of cis-acting functional sites and DNase I hypersensitive sites 35. The sequence of this fragment revealed several consensus sites for known transcription factors, including sites for ebox proteins, myb, gata 3, lef/tcf, stat, and gaga factors (Fig. 6).

The data indicate that 3′E_Cγ1 functions as an enhancer in vivo. In terms of enhancing transcription, 3′E_Cγ1 seems to play a more important role than HsA in immature thymocytes. DNase I hypersensitivity of 3′E_Cγ1 in thymocytes was easily detected in transgene templates. The hypersensitivity in B cells may be due to the absence from the transgene of “silencer elements” present in the endogenous locus, although we emphasize that the transgenes were not expressed in B cells. The endogenous 3′E_Cγ1 site was clearly hypersensitive in a dendritic epidermal γ/δ T cell line (Goldman, J., and D.H. Raulet, unpublished data) but was difficult to detect in thymocytes, perhaps because endogenous γ gene expression is silenced in most thymocytes. Overall, the 3′E_Cγ1 element has the properties of a typical non-LCR enhancer element in that it is active in transient transfection assays, exhibits DNase I hypersensitivity, and enhances transcription in vivo but is subject to transgene position effects.

The HsA element is a T cell–specific enhancer-like element that promotes transcription of rearranged γ genes in mature T cells. In addition, HsA stimulates recombination of transgenic γ rearrangement substrates. HsA was hypersensitive to DNase I in thymocytes but not in B cells or liver cells.

Although HsA exhibited clear enhancer activity in mature T cells when integrated as a multicopy transgene, it was devoid of enhancer activity in transient transfection assays in the PEER and Jurkat cell lines, in which 3′E_Cγ1 was active (data not shown). Both of these cell lines are unlikely to represent immature cells where HsA is nonfunctional, because the Jurkat line, at least, appears to be relatively mature based on its capacity to produce cytokines after TCR cross-linking. It is possible that HsA only functions with a homologous (γ gene) promoter element or only in the context of chromatin. Other instances have been reported where an element enhanced transcription when integrated in chromatin but not in transient transfection assays 18,36,37.

HsA, when combined with 3′E_Cγ1 as in the rrV2H⁺E⁺ transgene, confers efficient transgene expression in cells that normally express γ genes, including DN thymocytes and purified γ/δ cells, but does not drive expression in non-T cells. The enhanced percentage of Vγ2⁺ thymic γ/δ cells with the various lines of transgenic mice provides further evidence that the transgenes are indeed expressed in γ/δ cells. Significantly, expression of the rrV2H⁺E⁺ transgene was independent of transgene position effects, and the level of transgene expression was proportional to the number of transgene copies. Thus, the combination of HsA and 3′E_Cγ1 exhibits several characteristics of LCRs. We have not demonstrated that these elements are effective in single transgene copies, as none of the relevant lines contained just a single copy. However, an rrV2H⁺E⁺ line with two transgene copies exhibited high levels of transgene expression.

Although the rrV2H⁺E⁺ transgene was regulated appropriately in most respects, it was inappropriately expressed in α/β lineage T cells, unlike endogenous γ genes of this type. Previous studies provided evidence that the absence of expression of endogenous γ genes in α/β lineage cells is due to an associated “silencer” element 24. Although the silencer has not been subsequently defined or localized in detail, we have recently shown that transgenes containing an additional 10 kb of flanking 3′ DNA compared with the rrV2H⁺E⁺ transgene, when present at low copy number, are strongly downregulated in α/β but not γ/δ T cells (reference 34 and Kang, J., and D.H. Raulet, unpublished data). These data are consistent with the conclusion that transgene expression in α/β T cells observed here is due to a lack of cell type–specific repressive elements.

Transgenes containing only HsA or 3′E_Cγ1 were clearly expressed inappropriately. The transgene containing 3′E_Cγ1 but not HsA exhibited severe position effects and was not expressed in a copy number–dependent fashion, suggesting that 3′E_Cγ1 by itself is insufficient to open the chromatin. The transgene containing HsA but not 3′E_Cγ1 was poorly expressed in some cells in which endogenous γ genes are expressed well, such as DN thymocytes and fetal thymocytes. Nevertheless, HsA by itself did stimulate transcription in peripheral T cells in every line tested and was expressed in a roughly copy number–dependent fashion in peripheral T cells. The effects of HsA suggest that it may isolate linked genes from the inhibitory effects of neighboring chromatin. The putative chromatin-opening activity of HsA is probably operative even in cells where HsA alone functioned poorly as an enhancer. In total or DN thymocytes, HsA without 3′E_Cγ1 was expressed poorly, and 3′E_Cγ1 without HsA was subject to position effects; together, the elements supported high-level position-independent expression in both populations (Fig. 2 and Fig. 4). Thus, HsA may relieve position effects in immature thymocytes, cooperating with 3′E_Cγ1 to yield maximal levels of expression. Consistent with the role of HsA as a chromatin-opening element, we found that the 3′E_Cγ1 site was not DNase I hypersensitive in an rrV2H⁻E⁺ transgenic line that expressed the transgene poorly but was hypersensitive in an rrV2H⁺E⁺ transgenic line (data not shown).

Other LCRs have been shown to involve cooperative elements that enhance transcription and exhibit chromatin-opening activity 12,13,16,17,37,38,39,40. Hence, chromatin-opening elements may be at least partially separable from classical enhancers in several LCRs, including the TCR-γ LCR.

In the absence of 3′E_Cγ1, HsA drove transcription in mature SP thymocytes and peripheral T cells but not immature thymocytes. This was true of all four lines tested. In contrast, 3′E_Cγ1 by itself, when not subject to position effects, drove expression in DN, DP, and SP thymocytes but did so less well in peripheral T cells. Correspondingly, 3′E_Cγ1 functioned relatively well in fetal thymocytes. It will be of considerable interest to address the developmental roles of these two elements in vivo, where different sets of Vγ genes are used in the fetal and adult stages. Other instances have been reported of lymphocyte receptor genes with multiple, developmentally regulated enhancer elements. For example, both the CD4 and CD8 loci contain elements that seem to function differently in mature versus immature T cells 41,42,43,44,45,46,47.

In addition to promoting transcription of γ genes, our results with the γD series of recombination substrates indicate a role for HsA in supporting rearrangement of γ genes, even in the absence of 3′E_Cγ1. These results are of interest given the fact that HsA is a poor enhancer in immature thymocytes, the population in which rearrangement presumably takes place. It will be of interest to assess in future studies whether rearrangement promoted by HsA in the absence of 3′E_Cγ1 primarily involves the chromatin-opening activity of HsA or is associated with prior transcription of unrearranged Vγ genes. For technical reasons, we have been unable to address whether such germline transcription occurs from the transgenes.

The γD(H⁻E⁻) transgene underwent weak and sporadic rearrangement, as did the γD(H⁻E⁺) transgene. These data alone were therefore insufficient to assess the role of 3′E_Cγ1 in stimulating γ gene rearrangement. The finding that transgene rearrangement occurs to a limited extent in some lines lacking both elements is surprising, as enhancer elements are usually required for V(D)J recombination. It is possible that the transgene integrated into especially open chromatin in these lines. Alternatively, it remains possible that elements in the transgenes other than 3′E_Cγ1 and HsA participate in stimulating γ gene rearrangement. However, it is notable that no transcription of the rearranged transgenes was detected, confirming the importance of 3′E_Cγ1 and HsA in transcription.

All of the enhancer elements identified to date in antigen receptor loci are located either downstream of the constant regions or within the J–C introns. The location of HsA between Vγ2 and Vγ5 is therefore a novel scenario for antigen receptor genes. A ramification of the inter-V region location of HsA is that rearrangements of Vγ5 to Jγ1 will delete the element. Therefore, HsA must be unnecessary for supporting transcription of rearranged Vγ5 genes. One possibility is that HsA is only necessary to initially open the chromatin surrounding the Vγ genes and that subsequent maintenance of an open configuration is controlled by other elements and/or factors. Alternatively, there may exist additional elements upstream of the Vγ5 gene that support chromatin opening in the relevant cells. As the Vγ5 gene is unusual in that it is believed to undergo rearrangement preferentially in intestinal epithelial lymphocytes rather than thymocytes 48, it would not be surprising if the gene was regulated differently than the other Vγ genes. Finally, it is possible that the endogenous γ locus is sufficiently open in the absence of HsA for at least some cells to efficiently transcribe γ genes. It will be of interest to explore these possibilities by deleting HsA and/or 3′E_Cγ1 at their endogenous locations.

We thank Astar Winoto and Ben Ortiz for comments on the manuscript, Ann Lazar and Chern-sing Goh for technical assistance, and Peter Schow for assistance with flow cytometry.

This work was supported by National Institutes of Health grant RO1-AI31650 to D.H. Raulet.