INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Immunoglobulin (Ig) and T-cell receptor (TCR) genes are assembled from separate variable (V), diversity (D), or joining (J) gene segments in a process known as V(D)J recombination (4, 44, 55). Normally, V(D)J recombination is restricted to, and required for, early B and T lymphocyte development. It depends on a unique activity, called the V(D)J recombinase, which targets recombination signal sequences (RSSs; consisting of a heptamer, a spacer of 12 or 23 bp, and a nonamer) flanking the rearranging sides of all V, D, and J segments. Rearrangement events primarily involve pairs of segments with RSSs of asymmetrical spacer length (12/23 rule).

The functional core of the V(D)J recombinase consists of the lymphoid-restricted RAG-1 and RAG-2 gene products which recognize, pair off, and cleave the RSSs from two rearranging segments, a step also involving architectural proteins from the high-mobility-group family (22, 61). These cleavages consist of DNA double-strand breaks (DSBs) introduced precisely at the junction between the heptamer and adjacent coding sequences, yielding two distinct products: the hairpin-sealed coding ends (CEs) and the phosphorylated, blunt-ended signal ends (SEs). Eventually, the cleaved products are assembled together, a process which requires, in addition to the RAG factors, components of the general DNA DSB repair machinery (26, 54). The two CEs are assembled to form a coding joint (CJ) on the rearranged chromosome following hairpin opening and possible deletion and/or addition of nucleotides at the free extremities. The SEs are fused without further processing, yielding a signal joint (SJ) usually contained within a circular piece of extrachromosomal DNA.

V(D)J recombination is strictly controlled with respect to the lymphoid cell lineage, stage of cell differentiation, and allele usage. For example, there is a strong bias for TCR gene rearrangement to occur only in T cells (and for Ig gene rearrangement to occur only in B cells). Also, during thymic cell development, TCR rearrangement initiates in CD4 CD8 double-negative (DN) cells prior to TCR rearrangement which occurs during the transition from DN to CD4⁺ CD8⁺ double-positive (DP) cells (37). Moreover, during TCR rearrangement, D-to-J recombination usually precedes complete V-to-DJ recombination (in DN CD44⁺ CD25⁺ and DN CD44 CD25⁺ thymocytes, respectively). Finally, V-to-DJ joining is subject to allelic exclusion, a process whereby productive rearrangement (e.g., those encoding a TCR  chain) on one allele precludes further V recombination. It has been hypothesized that these levels of control reflect the ability of the individual loci and/or segments to serve as substrates for the recombinase, a concept known as V(D)J recombinational accessibility (63, 67). Recent evidence has revealed that recombinational accessibility, as tested by specific DSB formation, is a regulated property of lymphocyte chromatin (69). However, the molecular mechanism(s) and structural basis underlying these phenomena are not yet understood. Proposed activating factors include transcription across the unrearranged (germ line) region, the down-modulation of CpG methylation, and/or other epigenetic changes in chromatin (e.g., level of histone acetylation) affecting nucleosome positioning (13, 20, 54).

In studies using transgenic and knockout mouse techniques, it has been demonstrated that transcriptional cis-regulatory elements affect the recombination potential of adjacent gene segments (29, 63, 67). One striking example has come from studies on the TCR gene locus and the associated enhancer. This large locus (~500 kb in the mouse) (Fig. 1A) carries two homologous regions, each containing one D segment (D1 or D2), six functional J segments (J1.1 to 6 or J2.1 to 6; one additional pseudo-J is present in each J cluster), and one constant (C) region gene (C1 or C2). Most of the V genes are located 5' of the D-J clusters, except for one (V14) oriented in the opposite transcriptional direction 3' of C2. A single transcriptional enhancer (E), contained within a 560 bp HpaI-NcoI DNA fragment (E560), has been found 5.9 kb downstream of the C2 exons (38, 50). E560 activates V(D)J recombination within a transgenic substrate in early developing thymocytes from both embryonic and adult mice (12, 56). Moreover, deletion of the endogenous E560 fragment by using gene targeting in embryonic stem cells and production of mutant mice results in a drastic inhibition of cis rearrangement of the mutated TCR alleles (5, 6), indicating that the fragment contains regulatory sequences that are absolutely required for V(D)J recombination to proceed normally at that locus. Finally, detailed analyses of D and J SE intermediates as well as D-to-J CJ and SJ products in thymocytes from mice homozygous for the targeted deletion (E/ mice) lead to the surprising finding that loss of accessibility alone may not fully account for the defect in recombination, suggesting an unsuspected role for E during the recombination reaction (30).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Structural organization of E and truncated E variants used in this study. (A) Location within the TCR locus and structure of the 560-bp HpaI-NcoI DNA fragment containing E (not drawn to scale). The TCR V, D, and J segments as well as the C exons are represented by shaded boxes, and E is represented by an oval. 5' and 3' indicate transcriptional orientations of these various elements within the TCR locus, with the exception of the single V14 gene located downstream of E, which lies in the opposite direction. An enlargement of the 560-bp enhancer (E560) shows the several nuclear factor binding sites (E1 to E7) previously identified by EMSA within this fragment (71). (B) Partial restriction endonuclease map of the microinjected inserts and structure of the E variants. The TCR V, D, and J segments are represented by open boxes, their flanking RSSs are represented by shaded (23-bp spacer) or open (12- or 13-bp spacer) triangles, the IgH Cµ exons are represented by shaded boxes, the cosmid sequences are represented by hatched boxes, and E is represented by an oval (these various elements are not drawn to scale). Restriction endonuclease sites are indicated as B (BamHI), Bg (BglII), Hp (HpaI), Nc (NcoI), and R (EcoRI). The structural organization of each E variant is shown, including the 80- and 30-bp sequences which, in some variants, flank the 5' and/or 3' sides of the E1 and E4 motifs, respectively (delineation of E motifs according to Takeda et al. [71]). Locations of the V14-hybridizing probe used in Southern analysis of BglII-restricted genomic DNA from the transgenic mice and of the 5.5-kb BglII-hybridizing fragment from the unrearranged substrate are indicated.

Studies of mouse and human E, using in vitro footprinting analysis and electrophoretic mobility shift assays (EMSA), have identified multiple nuclear factor binding sites along this element which are conserved between the two species, supporting their functional importance (25, 58, 71). Several such motifs, including GATA- and helix-loop-helix (HLH)-binding E-box motifs, a cyclic AMP response element-like sequence, and Ets and core-binding factor (CBF) sites, are shared by other lymphoid gene enhancers and have been shown to be required for E activity (25, 31, 42, 59, 70, 71). Accordingly, transcription factors known to bind discrete enhancer motifs were also found to activate reporter genes placed under the control of related E sequences, including notably GATA-3 (47) and Ets as well as CBF (36, 59, 70, 78). In addition, screening of cDNA expression libraries with oligonucleotide probes spanning E or related sequences contributed to the identification of novel factors with homology to POU domain proteins (51) or to CACCC box binding proteins (73). Finally, signal transduction pathways involving raf and ras were proposed to play a role in E-mediated transcriptional activation (79). Despite all of these studies however, E structural and functional organization remains less well understood compared to regulatory elements from other lymphoid genes such as the Ig heavy-chain (IgH) intronic enhancer or the TCR enhancer (23, 53). Notably, in contrast to these other elements, a minimal core E sequence has not yet been unequivocally defined (25, 58, 71). Moreover, the functional importance of individual motifs and their associated factors to act nonredundantly as positive or negative elements is still unclear; such motifs include GATA and Ets (31, 47, 59, 70).

In this study, we used transgenic mice to better characterize the function of the E element in vivo, focusing especially on its role in the activation of germ line transcription and V(D)J recombination. Notably, we have defined a core TCR enhancer for recombination which is comprised of two known nuclear factor binding motifs as well as additional, previously unidentified sequences which we show are absolutely required to support cis rearrangement of adjacent TCR gene segments.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of recombination substrates. All recombination substrates were derived from the construct VDJCµB*R*. VDJCµB*R* is identical to the VDJCµ construct (19) except that it lacks the BamHI and EcoRI restriction sites located within exon 2 of the Cµ gene and 5.3 kb further downstream, respectively. Both sites were eliminated following BamHI and EcoRI partial endonucleolytic cleavages, fill-in of the extremities, and religation. This allowed direct cloning of the several E variants, using the adjacent BamHI and EcoRI sites located 130 bp 3' of the J1.2 gene segment in VDJCµB*R*. The E variants were produced by PCR amplification, using a 4.95-kb E-containing genomic fragment subcloned into pGEM-4Z (Promega France, Charbonnières, France) as a template (6), as well as appropriate forward and reverse primers flanked by BamHI and EcoRI sites, respectively. With position 1 assigned to the first nucleotide in the HpaI site from the 560bp HpaI-NcoI E-containing fragment (38), the various truncated variants of E used in this study were as follows: E200 (+124 to +324), E180 (+124 to +304), E169 (+155 to +324), E149 (+155 to +304), E98 (+226 to +324), E72 (+252 to +324), and E3' (+300 to +560). Each E fragment inserted into the recombination substrate was verified by nucleotide sequencing.

Production of transgenic mice. Linearization, purification, microinjection of the individual DNA inserts into fertilized (C57BL/6 × CBA/J)F₂ oocytes, identification of the transgenic founders, and production and maintenance of transgenic mouse lines were performed as described previously (12, 18, 19). The characteristics of the transgenic animals in lines 1003 (carrying an enhancerless recombination substrate) and E2 (carrying the E560-driven substrate, hereafter referred to as the E560 line) have been reported previously (12, 19). Nontransgenic littermates in the E560 line were used as the source of wild-type (WT) genomic DNA and total RNA. All mice were housed in a specific-pathogen-free animal facility in accordance with institutional guidelines. Mice were sacrificed for analysis between 4 and 6 weeks of age.

Southern blot analyses. Preparation of genomic DNA, restriction enzyme digests, agarose gel electrophoresis, DNA blotting onto nylon membranes (Hybond-N⁺; Amersham, Les Ulis, France), preparation and [-³²P]dCTP labeling of the V14 probe, and hybridization procedures were performed as described previously (12, 19). Images were generated from hybridized filters by use of a phosphorimager (BAS 1000; Fuji, Raytest France S.A.R.L., Courbevoie, France) and quantitated with MacBAS software.

DNA and RNA PCR assays. Genomic DNA and total RNA were simultaneously extracted from transgenic cell populations (~5 × 10⁵ cells), using TRIzol (GIBCO BRL, Cergy Pontoise, France) as recommended by the manufacturer. RNA samples were treated with RNase-free DNase I (Pharmacia, Orsay, France) and were converted to cDNA by reverse transcription (RT), using the SuperScript II reverse transcriptase (GIBCO BRL). Analysis of transgenic specific V(D)J rearrangements and J germ line transcripts by, respectively, DNA long-range PCR (LR-PCR and RNA RT)-PCRs were performed as described previously (12, 18). LR-PCRs were performed for 25 cycles of 1 min at 94°C, 30 s at 56°C, and 2 min at 72°C; RT-PCRs were performed for 28 cycles of 30 s at 94°C, 30 s at 58°C, and 30 s at 72°C. Oligonucleotide primers for amplification of fragments encompassing substrate D-to-J and V-to-DJ rearrangements and substrate J1-IgH Cµ exons, as well as fragments encompassing -actin, CD14, and C2 exons, were as described previously (12, 30). Sequences of forward and reverse primers for amplification of fragments encompassing endogenous D1-to-J1.1/J1.2 rearrangements were 5-CTGGTGGTTTCTTCCAGCCCT-3 and 5-CCTTCCTCTGATTACCAGAAC-3'. PCR amplification of endogenous V-to-J2 rearrangements was performed as described by Schlissel and Baltimore (62). After amplification, PCR products were electrophoresed through 1% agarose-0.5% NuSieve gels, transferred to nylon membranes (Hybond-N⁺; Amersham), and hybridized with [-³²P]ATP-labeled locus-specific oligonucleotide probes internal to the corresponding primers. Images were generated and quantitated as described above.

Antibodies and fluorescence-activated cell sorting (FACS) purification of lymphocytes. Fluorescein isothiocyanate (FITC)- and phycoerythrin-conjugated monoclonal antibodies (MAbs) against CD25 (7D4), CD44 (Pgp-1), CD3 (145-2C11), and B220 (RA3-6B2) were purchased from PharMingen (San Diego, Calif.). Lymphocyte preparation and cell staining with saturating levels of MAbs were performed as described previously (10, 18). Cell sorting of peripheral T and B cells was accomplished with anti-CD3 and anti-B220 MAbs. For cell sorting of DN thymic cell populations, an initial depletion of CD4⁺ and/or CD8⁺ cells was performed with anti-CD4- and anti-CD8-containing supernatants and rabbit complement, as described previously (33), before staining with anti-CD44 and anti-CD25 MAbs. Cell sorting was performed with a FACStar Plus (Becton Dickinson, Mountain View, Calif.).

Fluorescence in situ hybridization (FISH) analysis. Metaphase spreads were prepared from transgenic splenic lymphocytes stimulated with concanavalin A (ConA) at 37°C for 72 h and cultured with 5-bromodeoxyuridine (60 µg/ml) added for the final 6 h of culture to ensure chromosomal R banding. Chromosome spreads were hybridized with a biotinylated 10-kb EcoRI fragment containing the Cµ exons (19), using standard protocols (49, 57). Biotinylation of the Cµ probe was carried out by nick translation using biotin-16-dUTP (Boehringer Mannheim, Meylan, France) as recommended by the manufacturer. Before hybridization, the biotinylated probe was annealed for 45 min at 37°C with a 200-fold excess of murine Cot-1 DNA (GIBCO-BRL) in order to compete with nonspecific repetitive sequences; 200 ng of the probe (at 10 µg/ml in the hybridization solution) was used per slide. The hybridized probe was detected with FITC-conjugated avidin (Vector Laboratories, Burlingame, Calif.). Chromosomes were counterstained with propidium iodide diluted in antifade solution (pH 11.0) as described previously (43). For each chromosomal spread, a total of 30 metaphase cells were analyzed. Over 85% of the cells showed, in addition to the two hybridization signals detectable at the distal end of chromosomes 12 (corresponding to the endogenous IgH loci), a specific signal corresponding to the transgenic site.

In vivo footprinting analyses. The in vivo genomic footprint analysis was performed by the dimethyl sulfate (DMS)/ligation-mediated PCR (LM-PCR) technique (52) essentially as described by Algarté et al. (1). Briefly, thymocytes from 4- to 6-week-old WT or RAG-1-deficient (RAG-1/) mice (68) were harvested and subjected to DMS treatment, and the DNA was extracted. In parallel, DNA from cells of the same thymuses was first extracted, deproteinized, and then subjected to DMS treatment (naked DNA). The resulting DMS-treated DNA samples were cleaved by pipiridine and subjected to LM-PCR. Specific oligonucleotide primers, encompassing the E sequence, were used to carry out a primer extension reaction for both top and bottom strands. The resulting double-stranded DNA was ligated to the LM-PCR unidirectional linker consisting of two strands of dissimilar length: LM-PCR.1 (5'-GGGGTGACCC GGGAGATCTGAATTC-3') and LM-PCR.2 (5'-CTAGACTTAAG-3'). The linker-ligated DNA was subsequently subjected to PCR using E-specific oligonucleotide primers internal to that used for the primer extension step and primer LM-PCR.1 and a final two cycles of PCR with a third [-³²P]ATP-labeled E-specific oligonucleotide primer. The resulting labeled PCR products were resolved on 5% Hydrolink Long Ranger sequencing gels (TEBU, Le Perray en Yvelines, France) and exposed to X-ray film. The primers used for the primer extension, PCR, and labeling steps were, respectively, as follows: bottom strand, E1.1 (5'-GACCGATTCCATCAAAGAG-3'), E1.2 (5'-GCAACTGAAGAGATGCATTCCTGGG-3') and E1.3 (5'-GAGATGCATTCCTGGGACTTTTCGGTTCC-3'); top strand, E2.1 (5'-TTAGAGACCCTCCTCTTGG-3'), E2.2 (5'-GGTGATAGCTAGAGGCTGAGGTAGA-3'), and E2.3 (5'-AGAGGCTGAGGTAGAAAGGGCTGCATGAG-3').

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Experimental design. The E560 mouse DNA fragment (E560 [Fig. 1]) has been shown by in vitro footprinting and EMSA to include seven nuclear factor-binding motifs, designated E1 to E7 (71). The aim of this study was to define which sequences within E560 are involved in targeting TCR variable segments for cis recombination and expression in vivo. Thus, we used a well-characterized transgenic minilocus system in which substrate rearrangements are strictly dependent on the presence of a transcriptional enhancer (19, 66, 74). Briefly, the minilocus (Fig. 1B) is comprised of the unrearranged V14, D1, J1.1, and J1.2 gene segments (and flanking RSSs) in the same transcriptional orientation such that substrate D-to-J and V-to-(D)J rearrangements result in deletion of the corresponding intervening sequences; within these transgenes, the former rearrangements can occur in both T and B cells, whereas the latter are highly restricted to T cells. The TCR region is linked to a downstream DNA fragment containing the IgH Cµ gene, thus conferring a hybrid structure to the minilocus that facilitates molecular analyses of the integrated transgenes. Truncated versions of E were inserted between the TCR and IgH regions in the minilocus, a location in which the E 560 fragment was previously shown to activate cis recombination and transcription (12, 56). In preliminary experiments using this transgenic approach, we found that a 200-bp fragment (E200) comprised of sequences upstream of the E5 motif (including the E1 to E4 motifs and 80 and 30 bp of 5' and 3' flanking sequences, respectively [Fig. 1B]) induces transgene rearrangements, whereas a downstream fragment containing the E5, E6, and E7 motifs (E3') is essentially inactive (G. Bouvier and P. Ferrier, unpublished data). The present study further dissects the recombinational activity of the 5' region of E, using several truncated versions of E200 as schematized at the bottom of Fig. 1B.

Recombination activity of the transgenic substrates. The linearized recombination substrates were used to produce transgenic mice, utilizing standard protocols. The resulting transgenic founders were backcrossed onto WT (C57BL/6 × CBA/J)F₁ mice to derive transgenic lines. At least three independent transgenic lines shown to contain from 1 to 18 copies of the intact transgene were established for each construct (Table 1). Substrate rearrangements in the individual lines were first analyzed by Southern blot assay using BglII-restricted genomic DNA from transgenic thymuses and a V14-specific hybridization probe to allow for the identification of site-specific rearrangements within the integrated transgenes, as previously described (19) (see the legends to Fig. 1 and 2). DNA from a nonrearranging tissue (kidney) was used as a negative control. In the case of the E98 substrate, two additional founders (E98 D and E98 E) were sacrificed and similarly analyzed. Overall, these analyses demonstrated substantial differences in the efficiency of V(D)J recombination among the various transgenes, as judged by the presence or absence of thymus-specific hybridizing bands, the sizes of which correspond to predicted rearranged fragments within the transgenic substrates (Fig. 2; results from all transgenic lines are reported in Table 1). Specifically, substrate D-to-J rearrangement was clearly detected in thymus DNA carrying the E200 construct (in three of four lines), the E169 construct (in all four lines), and the E98 construct (in three of five lines) but not in thymus DNA carrying the E180, E149, or E72 construct (three lines in the first two cases; four lines in the latter). When absent from thymus DNA, substrate D-to-J rearrangement was also not detected in DNA from other lymphoid tissues such as the lymph nodes, spleen, and bone marrow (data not shown). Generally, substrate V-to-(D)J rearrangement was barely visible by this assay, including in mice exhibiting substrate D-to-J joints (Fig. 2). This may be due to the fact that V-to-(D)J joining in T-lineage cells is frequently less efficient within this recombination transgene (12, 18, 19, 56), coupled to the related sizes of the unrearranged and V-to-(D)J rearranged hybridizing fragments (Fig. 2, legend; also see the results from PCR analyses reported below).

View larger version (78K): [in this window] [in a new window]   FIG. 2.   Southern blot analysis of the microinjected constructs in transgenic mouse tissues. Ten micrograms of genomic DNA from thymus (Th) and kidney (Kd) of 4-week-old transgenic mice in the indicated lines and from the thymus of a WT animal was digested with BglII and assayed by Southern blot analysis for hybridization to the [-32P]dCTP-labeled V14 probe. In the unrearranged substrate, the probe labels a 5.5-kb BglII fragment. Because of deletion of the intervening sequences following V(D)J recombination within the transgene, the probe is expected to label additional fragments of predicted sizes depending on the type of rearrangement (19). The positions of the BglII fragments containing the endogenous V14 gene (Edg., 3.7 kb), the unrearranged substrate (NRg., 5.5 kb), and the D-to-J (DJ) and V-to-DJ (VDJ) rearrangements are indicated. The predicted sizes of both the DJ and VDJ fragments may vary depending on the length of the inserted E variant (in the range of 128 bp), with the largest V14-hybridizing DJ (9.74-kb) and VDJ (5.74-kb) fragments being seen in the E 200 thymus. Arrowheads indicate substrate rearrangements of lesser abundance which most likely correspond to other types of recombinase-mediated joints, as described previously (19). The asterisk indicates a 6.1-kb hybridizing fragment present in both thymus and kidney DNA from transgenic mouse E180 B, which most likely corresponds to a truncated copy of the transgene.

View larger version (73K): [in this window] [in a new window]   FIG. 3.   LR-PCR analysis of the microinjected constructs in transgenic mouse tissues. Fifty nanograms of genomic DNA from thymus (lanes 5 to 13) and kidney (lane 4) of 4-week-old transgenic mice in the indicated lines and from a WT thymus (lane 14) was tested by PCR using substrate-specific oligonucleotide primers (scheme at the top) and a J1.2-specific oligonucleotide probe, as described previously (12). The transgenic mouse in line 1003 carries six copies of the enhancerless minilocus (19). Top panel, hybridization images of DNA LR-PCR-amplified products using a forward primer located upstream of the D1 segment (primer 1) and a reverse primer located on the 5' side of the IgH region (primer 3) in the transgenic construct to allow the simultaneous amplification of substrate DNA fragments containing either unrearranged (NRg.) or rearranged D and J segments (DJ). Middle panel, hybridization images of DNA LR-PCR-amplified products using a forward primer located in the V14 gene (primer 2) and the reverse primer 3 to allow the amplification of substrate fragments containing rearranged V and J segments (VDJ); the faint band visible in the rearranged samples corresponds to DNA fragments carrying a V-to-D partial rearrangement, as described previously (12). Bottom panel, amplification products of the nonrearranging CD14 gene, used here to control for sample loading. Lanes 1 to 3 show a titration of transgenic thymus DNA in the rearranging line E200:6t28. Input DNA was kept constant (50 ng) by using decreasing amounts of thymus DNA and increasing amounts of kidney DNA. Lane 1, undiluted thymus; lanes 2 and 3, 1-to-10 and 1-to-100 dilutions, respectively. Quantification of the levels of substrate rearrangement per transgene copy in the individual lines was performed by phosphorimager scanning of the hybridizing images and correction for the amount of loaded DNA and the copy number per diploid genome. Phosphorimager analysis of the CD14 hybridization indicated that sample loading varied no more than 2.5-fold.

Transcription activity of the transgenic substrates. A host of studies, including transgenic analyses of various forms of recombination substrates, have shown that transcription of germ line Ig and TCR gene loci and/or segments tightly correlates with their activation for V(D)J recombination during lymphoid cell development (29, 67). However, a few examples which support a possible dissociation between transcriptional and recombinational activities have been reported (2, 18, 35, 40, 56). Using a previously described RT-PCR assay (12), we analyzed germ line transcription through the unrearranged J segments in the various integrated transgenes. High levels of germ line J transcripts were found in thymus RNA from the rearranging E200, E169, and E98 as well as nonrearranging E180 and E149 transgenic mice (Fig. 4A and Table 1). Among these lines, levels of J transcripts in thymus RNA varied no more than threefold according to quantification analysis (Fig. 4 and data not shown). However, when accounting for the number of nonrearranged copies of the transgene in the different lines (calculated following densitometric scanning of Southern blots of thymus DNA similar to those in Fig. 2), we found that levels of J transcripts from the nonrearranging lines were generally the lowest. For example, J transcription in line E180:A was calculated to represent ~20% of that found in the E200:6t28 line, whereas levels of transcription in the other two rearranging mouse lines shown in Fig. 4A were somewhat higher (see figure legend for details). Thus, the difference between the levels of germ line transcription between line E180:A and the rearranging lines is relatively modest. A somewhat different picture was found in assays using thymus RNA from the nonrearranging E72 transgenic animals. Although J transcripts could generally be detected, they were produced at significantly lower levels, exhibiting variations between transgenic animals in the same litter, including some that had no detectable J transcription (Fig. 4B, Table 1, and data not shown). Compared to serial dilutions of thymus cDNA from a nonrearranging E180:A transgenic mouse, we estimated the level of J transcripts per transgene copy to be decreased 10- to 25-fold, depending on the E72 line studied (i.e., 0.8 to 2% of J transcription in the E200:6t28 line [Table 1]). These data indicate that all of the truncated E variants tested in this study except E72 are efficient in stimulating J transcription within the integrated transgene.

View larger version (37K): [in this window] [in a new window]   FIG. 4.   Germ line transcription of the transgenic substrates. An RT-PCR assay (schematized at the top) was used to specifically detect germ line J transcripts from the transgenic substrates (12), starting with 10 ng of total RNA from transgenic thymus (Th) and kidney (Kd) (A and B, lanes 1 to 7 and 4 to 10, respectively) or with total RNA from FACS-purified DN CD44+ CD25+ (44+25+) and CD44 CD25+ (4425+) thymocytes (C) in the indicated lines. In panel A, lanes 1 shows a transgenic thymus cDNA in the rearranging line E200:6t28. In panel B, lanes 1 to 3 show a titration of transgenic thymus cDNA in the nonrearranging line E180:A. Input cDNA was kept constant. Lane 1, thymus cDNA undiluted; lanes 2 and 3, thymus cDNA diluted 1 to 10 and 1 to 50 in kidney cDNA. Analysis of -actin transcripts was performed in parallel to control for sample loading. Because primers which crossed intron-exon boundaries were used, contaminating DNA could not account for the hybridizing amplified products. Quantifications were carried out as described for Fig. 3. For the lines shown in panel A, for example, the percentages of J transcripts calculated for the individual lines, after correction for differences in sample loading and the amount of unrearranged substrate, were as follows: E200:6t28, 100%; E180:A, 20%; E169:B, 118%; and E98:C, 145%.

Effect of transgene integration on enhancer activity. The integration of transgenes into the mouse germ line following pronuclear injection of DNA usually results in an array of tandemly repeated, head-to-tail copies at a single chromosomal site. Depending on the site of chromosomal integration, side effects which generally have a negative effect on transgene expression can result (48). For example, it has been suggested that silencing of transgene expression could be a consequence of the spreading of centromeric heterochromatin within the nearby integrated transgenic array, a mechanism analogous to position effect variegation (PEV) in drosophila (7, 16, 21). Among the several E reporter transgenes which we found to rearrange and/or be expressed in a majority of independent transgenic lines, there were a few in which both transgene recombination and transcription could not be detected (e.g., Table 1, lines E200:4t26, E98:A, and E98:B). Since the copy number was stable and there was no evidence of structural alteration of the transgenes within these lines (R. K. Tripathi, M.-G. Mattei, and P. Ferrier, unpublished data), it seemed possible that the transgenic arrays had been silenced for expression following integration within or close to chromosomal centromeres. To test this hypothesis, we performed FISH analysis on spread chromosomes from ConA-stimulated splenocytes in transgenic animals of the nonrearranging E200:4t26 and E98:B lines, using as a probe the IgH Cµ DNA fragment (~10 kb) present on the 3' side of the recombination substrates. The rearranging lines E200:6t28 and E98:C were similarly analyzed. Under these conditions, the probe is expected to hybridize to the endogenous IgH loci, thus labeling the telomeric region of the two homologous chromosomes 12, as well as to the transgenic array, leading to an additional spot located on an unpredictable chromosome. In agreement with our hypothesis, we found that the transgenic arrays in lines E200:4t26 and E98:B are located within the centromeric region of, respectively, a large and a medium-sized chromosome (Fig. 5, left panels). Significantly, in both cases, a signal (corresponding presumably to the transgenic substrates) is associated with heterochromatin visible as condensed areas in interphase nuclei (Fig. 5, right panels). In marked contrast, the transgenic signal in line E200:6t28 is located in the middle part of a medium-sized chromosome, distant from any possible effect of centromeric heterochromatin (bottom panels); a similar image of chromosomal midarm integration was observed in transgenic splenocytes from line E98:C (data not shown). As expected, in all cases, fluorescent signals are visible at the distal end of both chromosomes 12, corresponding to the endogenous IgH loci.

View larger version (64K): [in this window] [in a new window]   FIG. 5.   Localization of IgH Cµ sequences by chromosomal FISH analysis to metaphase chromosomes (left panels) and interphase nuclei (right panels) from ConA-stimulated splenocytes of hemizygous transgenic mice. The top and middle panels are from the nonrearranging lines E200:4t26 and E98:B; the bottom panels are from the rearranging line E200:6t28. The biotinylated IgH Cµ probe was revealed with avidin-FITC. Arrows indicate locations of the endogenous IgH Cµ locus at the distal end of chromosomes 12; arrowheads indicate locations of transgenic areas in the various lines. Chromosomes and nuclei were counterstained with propidium iodide.

E98 confers lineage and temporal specificity to transgene recombination. The minimal TCR enhancer fragment that was found to activate transcription and recombination within the transgenic substrates integrated at sites distant from centromeric heterochromatin was E98. To test whether this element does so in a lineage- and development stage-restricted manner, we analyzed D-to-J rearrangement by LR-PCR using genomic DNA prepared from sorted populations of T and B peripheral lymphocytes or DN CD44⁺ CD25⁺ and DN CD44 CD25⁺ thymocytes from transgenic mice of line E98:C. Genomic DNA from the corresponding cells purified from E560 transgenic mice was similarly tested. Previously, we found that transgene rearrangement in this line is T-cell specific and occurs early during T-cell ontogeny and differentiation (12). Significantly, fragments containing substrate D-to-J rearrangements were found to predominate in purified T compared to B cells from the E98:C mouse, similar to the recombination profiles observed with a transgenic mouse from line E560 (Fig. 6A, top panel). Additional LR-PCR assays to analyze fragments containing endogenous D-to-J and V-to-J rearrangements confirmed the T- and B-lineage origin of the sorted cells (Fig. 6A, middle panels). Moreover, D-to-J recombination products were detected in purified DN CD25⁺ thymocytes from both E98 and E560 mice, again with profiles related to those observed for endogenous TCR gene rearrangement (Fig. 6B). Together with our finding of germ line J transcripts in DN thymocytes from the E98:C mice (see above), these data strongly suggest that E98 represents a true TCR core enhancer for recombination, as it appears to confer appropriate cell lineage and temporal specificity.

View larger version (45K): [in this window] [in a new window]   FIG. 6.   (A) Lineage and (B) developmental stage specificity of transgene rearrangement in line E98:C. Lymphoid cells were purified by cell sorting from individual transgenic mice of lines E98:C and E560, as indicated. The sorted cells were >98.5% pure as determined by FACS reanalysis. Genomic DNA prepared from the indicated purified cell populations was analyzed by LR-PCR to detect transgene D-to-J rearrangements (A and B, top panels), endogenous D-to-J rearrangements (A and B, second panels from the top), and endogenous V-to-J rearrangements (A, third panel from the top). Amplification of fragments containing endogenous C2 sequences was carried out to control for sample loading (A and B, bottom panel). Unrearranged and rearranged TCR-hybridizing fragments are labeled as in Fig. 3A. VJ, VJ-containing fragment. The sorted cells analyzed were peripheral T and B cells (T and B, respectively) (A) and DN CD44+ CD25+ and CD44 CD25+ thymocytes (44+25+ and 4425+, respectively) (B). In panel B, DNA from thymocytes (Th) and peripheral T cells (T) of a transgenic mouse in line E98:C were used as controls.

Developing thymocytes show nuclear factor occupancy within E98. The E98 core enhancer defined in this study contains the previously identified E3 and E4 nuclear factor binding sites as well as additional sequences 3' of E4. E3 contains a typical E-box motif, while E4 has been shown to be comprised of a composite Ets-CBF binding site (36, 70, 71, 78). Moreover, examination of the E4 3' flanking sequences by using the TRANSFAC database of transcription factor binding sites (76, 77) revealed that this region also contains an E-box motif (Fig. 7). To test whether the E98 element is bound by nuclear factors in vivo, we analyzed enhancer occupancy within the endogenous TCR locus by genomic footprinting using LM-PCR (52). This technique can reveal nuclear factor binding in chromosomal DNA following comparison of LM-PCR products between DNA which is treated in vivo with the membrane-permeable DNA-methylating agent DMS and that treated in vitro after extraction (thus generating the complete pattern of G residues across the given region). This is visualized as either protected G's or hypersensitive G's and A's for the in vivo versus in vitro DMS-treated DNA. The LM-PCR strategy was adapted to analyze the 250-nucleotide (nt) region spanning from 5' of E1 to 3' of E7. In addition to WT thymocytes, we analyzed thymocytes from RAG/ mice. Whereas the former are comprised of mostly DP cells carrying a productively rearranged TCR locus, the latter are made up of DN cells arrested in development at a stage where TCR gene rearrangement normally takes place (e.g., CD25⁺) (37). The endogenous locus was chosen for this analysis instead of the transgenic loci because of technical difficulties encountered in specifically amplifying the E region from the flanking A-T-rich IgH sequences in the integrated substrates (R. K. Tripathi, D. Payet, S. Spicuglia, W. M. Hempel, and P. Ferrier, unpublished data).

View larger version (38K): [in this window] [in a new window]   FIG. 7.   Analysis of in vivo occupancy of E in primary thymocytes. Thymocyte DNA from a (WT) mouse and from 10 RAG/ mice was methylated with DMS either as naked DNA in vitro (N) or as chromosomal DNA in intact cells in vivo (C). Methylated DNA samples were treated with piperidine and subjected to LM-PCR using E-specific primers. The top and bottom panels show footprints for the top and bottom strands, respectively. The corresponding sequence is shown below. Protected guanines are indicated by filled circles, while hypersensitive adenines are indicated by open circles to the right of the panels and above the top strand and below the bottom strand of the sequence shown. Positions of known protein binding sites within the analyzed E sequences, as well as the location of the element defined in this study 3' of the E4 motif, are indicated. Only the sequences showing in vivo occupancy within the 250-nt E region studied are shown.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Previous genetic analyses demonstrate a crucial role of E in regulating TCR gene recombination and, in addition, suggest a possible dual function for the control of both chromatin structure and V(D)J recombination at the TCR gene locus (5, 6, 12, 30, 56). These functions are most likely mediated by the combined action of nuclear factors bound to E in developing T cells. Provocatively, several of the identified E-binding factors also interact with Ig and other TCR enhancers (39, 42), which were similarly demonstrated to affect V(D)J recombination at their own loci (29, 67). Despite these findings, however, gene-targeting studies have so far failed to reveal an indisputable role for the relevant factors in regulating V(D)J recombinational accessibility (14). This may be due to either a redundancy between members of a given transcription factor family or a pleiotropic role of the inactivated factor in cell differentiation and/or survival. The definition of minimal cis-regulatory elements required to modulate the recombination of linked gene segments may, in addition to providing molecular clues as to the mechanism(s) involved, constitute an alternative approach to address the issue of which trans-acting factors are critical for this process.

We have shown that a 98-bp domain from the 560-bp E-containing fragment, comprised of two well-defined nuclear factor binding motifs (E3 and E4) and a 30-bp flanking sequence 3' of E4, is sufficient to promote germ line transcription and V(D)J recombinase-mediated rearrangements within an enhancer-dependent recombination substrate integrated into the mouse genome in several transgenic mouse lines. An overlapping 72-bp region that lacks the upstream E3 motif was found to be uniformly inactive when tested in this system, whereas longer E subfragments containing additional E1 and E2 motifs but lacking most of the sequences 3' of E4 (e.g., E180 and E149; both subfragments contain only 10 bp 3' of E4) were relatively efficient in activating germ line transcription but not recombination. These data indicate that the E98 domain represents a TCR core enhancer for V(D)J recombination and point toward a novel E sequence that plays a critical role in this process. The implications of these results, in terms of the molecular mechanisms and nuclear factors involved, are discussed below.

The V(D)J recombination-enhancing E subfragments identified in this study, including the E98 core, promoted rearrangement and germ line transcription from the integrated transgenes in several independent transgenic mouse lines, in conformity with the correlation between the two activities generally observed at antigen receptor gene loci (20, 67). In a few instances, however, we found that some E subfragments (e.g., E200 and E98) were not active after transgene integration within or close to a chromosomal centromere. Most likely, the subfragments could not antagonize the repressive effect of centromeric heterochromatin structures. In a study investigating the function of two enhancer elements, the 5' HS2 enhancer from the human -globin locus and the metallothionein enhancer, it was found that instead of regulating the level of expression of a linked reporter cassette, both enhancers act in cis to protect constructs from repression by flanking chromatin and to suppress PEV (72). Strikingly, however, the strength of this effect was shown to depend on the site of integration, indicating that as in PEV, chromatin varies in its ability to repress activity in a given gene unit. These data support a model in which enhancers (and their bound factors) directly interact with repressive chromatin structures to permit and stabilize expression rather than to increase the rate of transcription. While E activity of the subfragments is suppressed by centromeric heterochromatin, the possibility remains that it can overcome the repressive effects of chromatin regions characteristic of highly regulated loci, such as the TCR locus, as opposed to those characteristic of housekeeping genes, somewhat analogous to the situation just described. We anticipate a similar mode of action for other recombination enhancers; already, the Ig µ enhancer core has been shown to establish factor access in chromatin independent of transcriptional stimulation (34).

Unexpectedly, early-programmed J transcription (e.g., in DN CD25⁺ thymocytes) was detected within transgenes carrying the rearrangement-defective E180 subfragment, suggesting that this element could also confer chromatin accessibility to the transgenic templates in a significant proportion of cells undergoing TCR gene rearrangement. How then could the V(D)J recombinase apparently ignore the homologous gene segments in the particular transgenes? A trivial explanation would be that significant differences in accessibility still exist between rearranging and nonrearranging loci, despite the modest difference in germ line transcription, and/or that a higher threshold of chromatin access is required to permit recombination as opposed to germ line transcription. Alternatively, germ line transcription in the E180-containing transgenes may be qualitatively inadequate, for example, not being initiated at the promoter upstream of D1, an element which has been shown by gene targeting to be required for rearrangement of this gene segment (75). Finally, in line with our recent report of unresolved DSBs at RSSs from enhancer-deleted TCR alleles, it is also possible that specific nucleoprotein complexes organized at E regulatory elements may affect the V(D)J recombination process at a step beyond that of chromatin access (30). Of note, two potential functions of E, increasing chromatin accessibility and affecting directly the course of the recombination reaction, may not be mutually exclusive. If this were the case, it would provide an explanation for the apparent paradox of J transcription in the absence of recombination in the E180 transgenic mouse lines. Current analyses of recombination intermediates and germ line transcripts in lymphoid cells from transgenic mouse models, including some established on a RAG/ background, may help us to distinguish among these possibilities. In any case, our data add to the increasing experimental evidence concerning transgenic and endogenous TCR loci, suggesting that transcription per se is not sufficient to permit rearrangement of the given gene segments (5, 18, 45, 56).

It is believed that all lymphoid cells are derived from a common lymphoid cell progenitor (65). As they differentiate, the developing lymphocytes progressively lose their multilineage potential. In the rearranging E98:C line, transgene D-to-J rearrangements were found to predominate in T as opposed to B lymphocytes. Accordingly, these products were readily observed in DN CD44⁺ CD25⁺ thymocytes, a subpopulation which contains (and ) T-cell precursors but can no longer give rise to B cells (65). Overall, transgene rearrangement profiles within the E98:C line (as well as within the E560 line [Fig. 6 and reference 12]) mimic those at the endogenous TCR locus (11, 24, 46). Coupled to the fact that the reporter transgenes in the E98:C and E560 lines are located on different chromosomes (Tripathi, Mattei, and Ferrier, unpublished data) thus presumably within independent genomic contexts, these data strongly suggest that the regulatory elements within the E core act positively in regulating cis recombination, independently of the action of adjacent motifs outside the core domain. Interestingly, also in a study using a transgenic approach, it was found that rearrangements within substrates carrying DNA fragments containing the full-length TCR enhancer (E) are appropriately timed (e.g., with respect to those at the endogenous TCR-J locus) and strictly limited to the T lineage during thymocyte differentiation (12, 41). In contrast, those under the control of a shorter core E fragment initiate slightly earlier during ontogeny and occur in both and T cells (60). Because some nuclear factor binding motifs are shared by the E and E cores, it is tempting to speculate, based on our results as well as those of Roberts et al. (60), that nucleoprotein complexes comprised of the corresponding factors could enhance cis recombination in early T cells, this activity at the TCR-J locus being delayed by the combinatory effect of factors bound outside the E core. Contrary to our results, however, equivalent levels of D-to-J junctions in both T and B cells within E560-containing transgenes have been reported (56). We note that among the four types of transgenic mice analyzed, two were of high copy number (25 and 60 copies), a fact which could lead to the dysregulation of cell lineage specificity in V(D)J recombination (9). One way to reconcile these data would be that E560 has the capacity to enhance cis recombination in both T and B cells, but in the latter case less efficiently and only occasionally, an assumption which would parallel the differential transcriptional efficiency of related fragments following transfection into lymphoid T- and B-cell lines (25, 71).

The E98 subfragment identified here as a core enhancer of V(D)J recombination exhibits several interesting characteristics. Extensive (90%) homology is found between the mouse and human sequences over the entire fragment, including the 3' flanking sequences shown to be strictly required for the activity, whereas overall homology is less conserved (<70%) outside this particular domain (58; Tripathi, Payet, et al., unpublished data). In addition, within this subfragment, the E3 and E4 sites that have been shown to interact with nuclear factors from early developing thymocytes following EMSA and in vitro footprinting analyses (71; Tripathi, Payet, et al., unpublished data) as well as in vivo genomic footprinting (this study) overlap with, respectively, a bHLH-binding E-box motif and the Ets-CBF composite motif, both of which are commonly found in antigen receptor gene enhancers (17, 23, 59). Notably, a direct role for bHLH factors on the regulation of V(D)J recombination in T lymphocytes is supported by several studies (3, 64), including, in the case of the TCR locus, the finding that enforced expression of Id3 (which inhibits many bHLH transcription factors) in cultured human thymocytes prevents the appearance of cells with D-to-J rearrangements (28). Interestingly, another typical E box is also present within the most 3' 20-bp flanking sequences of E98 (bottom of Fig. 7). Although we have not been able to definitively confirm binding at this site by using conventional DNA-protein interaction assays, we have obtained evidence suggesting that when present, the 3' sequences modify binding at the E4 site, leading to the formation of a distinct nucleoprotein complex(es) (Tripathi, Payet, et al., unpublished data). One attractive possibility would be that the E box within the 3' sequence, along with another site(s) (E3 and/or E4), contributes to a bipartite DNA motif that nucleates a multimeric complex in developing T cells, as recently described in a study of nucleoprotein structures involving the LIM-only protein Lmo2 (27). Alternatively, the 3' sequence may help to stabilize the interaction of cofactors which bind to the minimal enhancer. Intriguingly, within the TCR enhancer has been found a sequence which, in the absence of demonstrable factor binding, is required for the formation of a ternary complex involving a coactivator (ALY) and a bound factor (8). Further analyses in our lab will focus on a deeper characterization of the molecular complexes identified at E98 and their putative role in regulating V(D)J cis recombination.

Indirectly, our results suggest that the E motifs outside of E98 may play a role during early T-cell differentiation distinct from activating recombination and/or at a latter stage of T-cell life, for example, during T-cell activation. In particular, this may be the case for the motif in E1 which was proposed to represent the main GATA-3 factor binding site within E (31). Indeed, expression of GATA-3 has been found to be down-modulated during periods of TCR gene recombination in thymocytes (32), making it unlikely that this factor could contribute directly to regulation of the TCR rearrangement process. More intriguing, however, is the function of the second Ets- and CBF-interacting element within E6 which, like that in E4, is heavily occupied in early thymocytes (Fig. 7). These elements have been shown to be responsible for the inducibility of the TCR enhancer in response to phorbol ester treatment, which mimics the signals for a range of cellular changes associated with T-cell differentiation and activation (58, 59). It is, therefore, possible that each of the two motifs plays a unique role in enhancer function, the E4 site being more closely involved with modulating recombinational accessibility. It is equally possible, however, that these two elements act cooperatively to ensure the chromatin-opening activity of E, the E6 site being, for example, required for maximal efficiency as proposed for other types of cis-regulatory elements (21). These issues are currently being addressed by inserting E subfragments at the endogenous TCR locus.