Characterization of Chromatin Structure and Enhancer Elements for Murine Recombination Activating Gene-2¹

During the development of B and T lymphocytes, Ig and TCR variable region genes are assembled from germline V, (D), J gene segments by a site-specific recombination reaction known as V(D)J recombination (1). The V(D)J recombination is initiated by the products of recombination activating gene (RAG)⁵-1 and RAG-2 (2, 3). They act together to recognize recombination signal sequences that flank V, D, J gene segments and introduce the dsDNA breaks between the recombination signal sequences and the gene segments (4). The expression of RAG-1 and RAG-2 is restricted to lymphoid cells undergoing V(D)J gene rearrangement (5, 6, 7, 8). In the absence of either the RAG-1 or RAG-2 gene products, the development of mature lymphocytes is completely abrogated, and hence results in severe combined immune deficiency in mouse and human (9, 10, 11). In humans, missense mutations in either the RAG-1 or RAG-2 result in Omenn syndrome, which is characterized by a few T lymphocyte repertoires (12).

The transcription of RAG is regulated at different levels. Fuller et al. (13, 14) have reported, and we have also reported, that the DNase I hypersensitive (HS) site was identified in the promoter region of mouse and human RAG-1 only in RAG-expressing lymphocytes, indicating that RAG-expression is regulated at the chromatin level. At the promoter level, human RAG-1 and RAG-2 promoter regions did not show the lymphocyte-specific expression of the reporter gene in vitro (14, 15, 16, 17). On the contrary, it was shown that the mouse RAG-2 promoter conferred lymphoid specificity in the reporter gene assay and that the distinct transcriptional factors regulate lineage-specific activation of the RAG-2 promoter in B- and T-lineage cells (18, 19, 20, 21). Recently it was shown that cis-elements other than the RAG-2 promoter region were also involved in driving the lymphoid- and differentiation stage-specific RAG-2-expression in vivo (22, 23). These indicated the existence of the cis-regulatory element, such as enhancer, in the 5′-upstream region of mouse RAG-2.

In this study, we analyzed the DNase I HS sites of 25-kb DNA sequences adjacent to the murine RAG-2 locus and identified three lymphoid-specific DNase I HS sites, one of which corresponded to the promoter region and specifically associated with RAG-2 expression. We further identified that two enhancer elements in this 25-kb DNA both corresponded to HS sites. We found that one of the enhancer elements (distal enhancer) exhibited enhancer activity only in lymphoid cell lines in transient transfection assays. To test whether the enhancer functions in vivo, we generated and analyzed the transgenic mice carrying enhanced green fluorescence protein (EGFP) reporter gene connected to the distal enhancer element. We show that this enhancer was active in thymus and bone marrow. We further demonstrate that EGFP was expressed only in the CD4⁻CD8⁻ subpopulation in the thymus and B220⁺IgM⁻ cells in the bone marrow. Our data suggest that chromatin structure and the cis-regulatory element may regulate the tissue- and stage-specific expression of murine RAG-2. We also demonstrate that C/EBP may bind to this element and function as one of the activator(s) or coactivator(s) for the enhancer activity.

18.81 (pre-B cell, RAG-2⁺), BAL-17 (B cell, RAG-2⁺), WEHI231 (B cell, RAG-2⁻), LSB11-1 (T cell, RAG-2⁺) (24), LSB11-14 (T cell, RAG-2⁺) (24), EL4 (T cell, RAG-2⁻), 110TC (T cell, RAG-2⁻) (24), WEHI3 (myeloid, RAG-2⁻), L (fibroblast, RAG-2⁻), and NIH3T3 (fibroblast, RAG-2⁻) were grown in RPMI 1640 medium supplemented with 10% FCS, 100 U/ml penicillin, and 0.1 mg/ml streptomycin at 37°C in 5% CO₂.

DNase I HS sites were analyzed as previously described (14). Briefly, 1.5 × 10⁷ nuclei were treated with 0–2 U/μl of DNase I (Takara, Kyoto, Japan) at 25°C for 3 min. The reaction was stopped by addition of 3 ml of proteinase K solution (10 mM Tris-HCl, pH 8.0, 0.1 M EDTA, 0.5% SDS, and 0.1 mg/ml proteinase K). Then the genomic DNA was extracted, digested by appropriate restriction enzymes, electrophoresed on 0.7% agarose gel, blotted onto GeneScreen Plus membrane (NEN, Boston, MA), and hybridized with a ³²P-labeled mouse RAG-2 fragment.

For the preparation of the EGFP construct with the minimal mouse RAG-2 promoter (pR2P-EGFP), the −86/+147 region of the murine RAG-2 promoter (18) was inserted into the SalI and BamHI sites of the pEGFP-1 (Clontech Laboratories, Palo Alto, CA). To analyze the 5′-upstream region of murine RAG-2, a λDNA clone containing the 5′-upstream region of murine RAG-2 was cloned by screening the λFIX II murine genomic DNA library (Stratagene, La Jolla, CA). An 8-kb HindIII-HindIII fragment (Fig. 1,A) was cut out and subcloned into pBluescript II SK⁺ (Stratagene). The D, E, F, G, H, and P fragments (Fig. 2,A), and the D1, D2, D3, P1, P2, and P5 fragments (Fig. 3) were excised from an 8-kb HindIII-HindIII fragment by restriction enzymes, and subcloned into 5′-upstream of the RAG-2 promoter in the pR2P-EGFP. The P4 fragment was prepared by PCR using oligonucleotides 1 and 2 as primers. Amplified fragments were inserted into the pR2P-EGFP. The murine TCRβ 3′ enhancer (25) was subcloned into the SalI and BamHI sites of the pR2P-EGFP. The HSV thymidine kinase (TK) promoter (GenBank accession no., M15234) was subcloned into the KpnI and SmaI sites of pEGFP-1 (pTKP-EGFP). pD3-TKP-EGFP was prepared by inserting the D3 fragment into the pTKP-EGFP. For constructing pSVen-TKP-EGFP, the SV40 enhancer (SVen) was excised from the PicaGene Enhancer Vector 2 (Nippon Gene, Tokyo, Japan) and inserted into pTKP-EGFP. The D3 fragment with mutation for the C/EBP binding site (Mut-C/EBP) was prepared by PCR using the cloned genomic DNA as a template and oligonucleotides 3 and 4, as well as oligonucleotides 5 and 6 as primers. All the constructs were verified by restriction enzyme digestion or sequencing analysis.

To introduce EGFP constructs into lymphoid cells, 2.5 μg of EGFP reporter gene and 2.5 μg of human CD23 expression vector (pMECD23) (26) were transfected into 2.5 × 10⁶ cells by the DEAE-dextran method as described previously (18). For transfection of EGFP constructs into fibroblasts, 15 μg of EGFP reporter gene and 15 μg of pMECD23 were transfected into the cells using the calcium phosphate method as previously described (18). Cells were harvested 22–24 h after transfection, and stained with biotinylated anti-human CD23 mAb (18.6) (26) followed by PE-conjugated streptavidin (BD PharMingen, San Diego, CA). The relative expression of EGFP and CD23 was analyzed by flow cytometry using a FACSCalibur (BD Biosciences, San Jose, CA). Data were analyzed with CellQuest software (BD Biosciences). Where indicated, the results were analyzed using the Student’s t test with a p value of <0.05 considered significant.

The EGFP reporter gene constructs, pR2P-EGFP and pD3-R2P-EGFP, were digested with XhoI and AflII, and the DNA fragments containing the EGFP reporter gene were digested with the murine RAG-2 promoter/SV40 poly(A) signal, or with the D3 fragment, were purified by a Prep-A-Gene DNA Purification system (Bio-Rad, Hercules, CA). The transgenic mice were generated by Japan SLC (Hamamatsu, Japan). The mice were analyzed for the presence of the transgene by PCR using oligonucleotides 7 and 8, or 9 and 10 for examining D3-R2P-EGFP, and oligonucleotides 8 and 11 to examine R2P-EGFP. The copy number of the integrated DNA was determined by Southern blotting. To analyze the expression of the transgene, total RNA was extracted from various tissues and treated with RNase-free DNase I to avoid the contamination of genomic DNA as a template. The transcripts were detected by RT-PCR using oligonucleotides 8 and 9 and by Southern blotting using the EGFP gene as a probe as described previously (27).

For analyzing EGFP expression in the thymocyte subpopulation, thymocytes were prepared from the transgenic mice (line 108) and stained with FITC-conjugated anti-CD8 Ab and PE-conjugated anti-CD4 Ab. CD4⁺CD8⁺ and CD4⁺CD8⁻ thymocytes were sorted twice using Epics-Elite (Beckman Coulter, Fullerton, CA). CD4⁻CD8⁻ thymocytes were prepared by deleting CD4⁺ or CD8⁺ thymocytes using HO2.2 anti-CD8 Ab and RL172 anti-CD4 Ab, and complement, as described previously (24, 28). For analyzing EGFP expression in subpopulations of the bone marrow, bone marrow cells were incubated with biotin-conjugated anti-B220 Ab, followed by streptavidin-conjugated magnetic microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) and FITC-conjugated anti-IgM Ab. B220⁺IgM⁺and B220⁺IgM⁻ bone-marrow cells were sorted using autoMACS (Miltenyi Biotec) and Epics-Elite. The purity of each population was analyzed by FACSCalibur.

Nuclear extracts were prepared according to the method described by Schreiber et al. (29). EMSA was performed by incubating nuclear extracts with radio-labeled oligonucleotides and then subjecting them to electrophoresis as previously described (18). The probes of the distal enhancer were prepared by PCR using the primer pairs: oligonucleotides 7 and 14 for probe 1 (−382/−282), 15 and 16 for probe 2 (−281/−198), 17 and 18 for probe 3 (−197/−90), or 19 and 6 for probe 4 (−89/−1). Consensus binding sequences of lymphoid transcription factor (Lyf)-1, acute myeloid leukemia (AML)-1, GATA, E-box, Nkx2.5, NF-κB, C/EBP, Oct-1, and c-Myb were prepared by annealing oligonucleotides of 20 and 21, 22 and 23, 24 and 25, 26 and 27, 28 and 29, 30 and 31, 32 and 33, 34 and 35, or 36 and 37, respectively.

The primers for PCR were as follows: 1, 5′-AAGCTTTCGCATCCTTTTCTACTTCCGTTG-3′; 2, 5′-GAATTCGGGGATAGGCATGGCTCTTATGA-3′; 3, 5′-AGCTTGCTAGAACAAGAT-T-3′; 4, 5′-GTTGGCTTTAGACTAGTCTAATACACAA-3′; 5, 5′-TTGTGTATTAGACTAGTCTAAAGCCAAC-3′; 6, 5′-ATCACGCTCCACAGGCATTG-3′; 7, 5′-AAGCTTAAAGGCATAAAAATGGCTGT-3′; 8, 5′-CGTCGCCGTCCAGCTCGACCAG-3′; 9, 5′-CAAGCTGACCCTGAGTTCA-3′; 10, 5′-TCGATGTTGTGGCGGATCTT-3′; 11, 5′-TTCTGTCTCCCTCAACCATC-3′; 12, 5′-ATGGACGAGCTGTACAAGTA-3′; 13, 5′-ACAAACCACAACTAGAATGC-A-3′; 14, 5′-ACTTACCATTATGTCCTATGCT-3′; 15, 5′-AAAGGCATAAAAATGGCTGT-3′; 16, 5′-AGAGTGGTCTGTATTTTGTTTCCT-3′; 17, 5′-GCTCTGGCTATCACTGTCAC-3′; 18, 5′-AGGACTTCTGGGAGGCGA-3′; 19, 5′-GTTGTGTATTATTAAGTAAT-3′; 20, 5′-GATCCATTTTGGGAGAAAGT-3′; 21, 5′-ACTTTCTCCCAAAATGGATC-3′; 22, 5′-AATTCGAGTATTGTGGTTAATACG-3′; 23, 5′-CGTATTAACCACAATACTCGAATT-3′; 24, 5′-CACTTGATAACAGAAAGTGATACTCT-3′; 25, 5′-AGAGTTATCACTTTCTGTTATCAAGTG-3′; 26, 5′-CTTTTTACCAGGTGGTCTCTA-3′; 27, 5′-TAGAGACCACCTGGTAAAAAG-3′; 28, 5′-CACTGCCCAGTCAAGTGTTCG-3′; 29, 5′-CGAACACTTGACTGGGCAGTG-3′; 30, 5′-GTAGGGGACTTTCCGAGCTCG-3′; 31, 5′-CGAGCTCGGAAAGTCCCCTAC-3′; 32 AND 33, 5′-TGCAGATTGCGCAATCTGCA-3′; 34, 5′-ATGAATATGCAAATCAGGTGA-3′; 35, 5′-TCACCTGATTTGCATATTCAT-3′; 36, 5′-TACAGGCATAACGGTTCCGTAGTGA-3′; 37, 5′-TCACTACGGAACCGTTATGCCTGTA-3′.

DNase I HS sites are often associated with cis-acting elements, such as enhancers, silencers, or promoters (30, 31, 32, 33, 34, 35). To identify the cis-acting elements of murine RAG-2, we examined the DNase I HS sites within a 25-kb region encompassing the 5′-upstream and 3′-downstream regions of murine RAG-2 (Fig. 1,A). Three DNase I HS sites (HS1, HS2, and HS3) were identified in a RAG-2-expressing pre-B cell line (18.81). HS1, HS2, and HS3 were located at −8.1, −2.6, and −0.2 kb relative to the transcriptional initiation site. These DNase I HS sites were not detected in a myeloid cell line (WEHI3) and a fibroblast cell line (L) (Fig. 1, B and C), suggesting that these HS sites are lymphoid-specific. To confirm this, we examined the DNase I HS sites in T cell lines as well as other B cell lines and nonlymphoid cell lines. As shown in Table I, HS1, HS2, and HS3 existed in both B (18.81 and BAL17) and T cell lines (LSB11-1, LSB11-14) which expressed RAG-2, but not in myeloid (WEHI3) and fibroblast cell lines (L and NIH3T3). HS1 was found not only in RAG-2-expressing cell lines, but also in RAG-2-nonexpressing lymphoid cell lines (WEHI231 and EL4). HS2 was not clearly detected in the RAG-2-nonexpressing EL4 T cell line or the WEHI231 B cell line. HS3, which corresponds to the RAG-2 promoter region, was present only in the cells expressing RAG-2. These DNase I HS sites were also identified using StuI digestion (data not shown).

The presence of the lymphoid-specific DNase I HS sites in the 8-kb 5′-upstream region (Fig. 1,A) suggested that this region might be of importance in the lymphoid-specific transcriptional regulation of RAG-2. Therefore, we investigated whether DNA elements in this region could affect the mouse RAG-2 minimal promoter by using the EGFP reporter gene (Fig. 2). Constructs containing consecutive or overlapping fragments of the murine RAG-2 5′-upstream region extending from −300 bp to ∼−8.3 kb were prepared and assayed for EGFP expression in the transiently transfected 18.81 cells. Two discrete fragments (D and P) were shown to augment EGFP expression ∼6- to 8-fold compared with the control EGFP reporter gene (Fig. 2,C). The fragment connected in the reverse orientation (DR and PR) to the RAG-2 promoter retained its abilities to stimulate the transcription, indicating that the elements functioned as enhancers (Fig. 2,C). D and P fragments coincided with the DNase I HS sites shown in Fig. 1 (HS1 and HS2, respectively).

To identify the core regulatory elements within P and D fragments, a series of deletion mutants of each fragment was constructed into the EGFP reporter genes (Fig. 3). The EGFP constructs were transfected into 18.81 cells, and the relative EGFP expression was determined. As shown in Fig. 3,A, deletion of the 5′ portion of the D fragment to the EcoRV (D1) or PstI (D2) site completely abolished the enhancer activity. The 380-bp HindIII-EcoRV fragment (D3) showed the full enhancer activity, suggesting that the element necessary for maximal enhancer activity is present in D3. Of note is that this 380-bp core enhancer segment coincided with the HS1 site. The fragment D3 constructed in the reverse orientation (D3R) exhibited the corresponding enhancer activity compared with that in forward orientation (D3) (Fig. 3,A), showing that the enhancing activity of D3 is orientation-independent. By a similar analysis of series of deletion mutants, the core enhancer element in the P fragment was shown to reside within P4 (a 170-bp fragment) (Fig. 3 B). It functioned in an orientation-independent manner and coincided with the HS2 site. Taken together, two enhancer elements, D3 and P4, were identified in the 5′-upstream region of murine RAG-2 and both of them werelocated at the lymphocyte-specific DNase I HS sites.

To determine the cell specificity of the distal enhancer element (D3), EGFP reporter constructs containing the RAG-2 promoter with or without D3 (D3-R2P-EGFP and R2P-EGFP) were transiently transfected into 18.81 (pre-B), BAL17 (B), and EL4 (T) cells, and enhancer activity was examined. The EGFP reporter construct containing a TCRβ enhancer (TCRβen-R2P-EGFP) was used as a positive control. As shown in Fig. 4,A, D3 increased the mouse RAG-2 promoter activity ∼5- to 9-fold in both B (18.81, BAL17) and T cell lines (EL-4). To test the enhancer activity in nonlymphoid cells, the EGFP reporter constructs containing the TK promoter with or without D3 (D3-TKP-EGFP and TKP-EGFP) were transfected into the lymphoid cell line (18.81) and nonlymphoid cell lines (L and NIH3T3), and the relative enhancer activity was assessed. The EGFP reporter construct containing the SV40 enhancer (SVen-TKP-EGFP) was used as a positive control. We used the TK promoter and the SV40 enhancer in the EGFP reporter constructs because the RAG-2 promoter and TCRβ enhancer did not function in nonlymphoid cells. As shown in Fig. 4 B, the enhancer activity of D3 was detected in 18.81 cells, but not in either L or NIH3T3 cells. These results indicate that D3 functions in a lymphoid-specific manner.

To verify the results obtained in vitro, we examined the function of the distal enhancer in vivo by generating transgenic mice that carried the EGFP reporter construct containing D3 driven by the murine RAG-2 minimal promoter (D3-R2P-EGFP). The construct without D3 was used to generate the control mice (R2P-EGFP) (Fig. 5,A). The total RNA from thymus, bone marrow, spleen, liver, kidney, and brain of the 11 independent transgenic lines were examined for the expression of the transgene by RT-PCR. Fig. 5,B (upper panel) shows a representative result of a transgenic mouse (D3-R2P-EGFP, line 108). D3 directed the expression of the EGFP reporter gene in thymus and bone marrow but not in the spleen, liver, kidney, and brain. In the control transgenic mice (R2P-EGFP), no EGFP expression was detected (Fig. 5,B, lower panel). The data of all of the 11 transgenic lines was shown in Table II. Six of 11 lines (108, 208, 301, 314, 401, and 404) expressed EGFP only in thymus and bone marrow, but not in other organs. These data indicate that D3 enhanced mouse RAG-2 promoter activity in vivo specifically in primary lymphoid organs. Two of the transgenic lines (213 and 313) expressed EGFP in all of the tissues examined, probably due to the integration of the element into the vicinity of other cis-acting elements as described by Ellmeier et al. (36). Interestingly enough, three lines (119, 215, and 302) expressed EGFP not only in primary lymphoid organs, but also in brain, suggesting that D3 is active in some situations in the brain.

To further determine which subpopulations of the thymus or the bone marrow in the transgenic mice express EGFP, various thymocyte subpopulations, CD4⁻CD8⁻ (DN), CD4⁺CD8⁺ (DP), and CD4⁺CD8⁻ (SP), or bone marrow subpopulations, B220⁺IgM⁻ and B220⁺IgM⁺, were prepared from a transgenic line (108) and their EGFP expression was analyzed. As shown in Fig. 5 C, EGFP was expressed in the DN thymocytes, but not in the DP or SP thymocytes in the thymus, while it was expressed in B220⁺IgM⁻, but not in B220⁺IgM⁺, cells in bone marrow.

To delineate the transcription factors regulating the D3 distal enhancer element, putative binding sites for transcription factors were searched using the Genome Net Database (Bioinformatics Center, Institute for Chemical Research, Kyoto University, Kyoto, Japan). Fig. 6,A shows the binding sites for several transcription factors that are thought to function in lymphoid cells such as Lyf-1, AML-1, GATA, E-box, NF-κB, Nkx2.5, C/EBP, Oct-1, or c-Myb. To try to identify the cell type-specific transcriptional factors that bind to the D3 fragment, an EMSA was performed using probe 1 (−382/−282), probe 2 (−281/−198), probe 3 (−197/−90), and probe 4 (−89/−1) of the D3 region with nuclear extracts from lymphoid 18.81 and nonlymphoid NIH3T3 cells (Fig. 6,B). To confirm the specific complex-formation, a 200-fold excess of unlabeled probe was added as a competitor. As shown in Fig. 6,B, nine lymphoid-specific complexes (C1-C3 with probe 1; C4 with probe 2; C5 with probe 3; C6-C9 with probe 4) were detected. To determine the binding sites responsible for the complex formation, we then performed a competition assay. We added the excess amount of the oligonucleotides that contained consensus binding sites for each of the possible lymphoid-specific transcription factors in the D3 enhancer, Lyf-1, AML-1, GATAs, E-box, NF-κB, Nkx2.5, C/EBP, Oct-1, and c-Myb (shown in Fig. 6,A), and examined their ability to inhibit the complex formation. Only the addition of the excess oligonucleotide containing the consensus binding site for C/EBP inhibited the complex formation between nuclear extract 18.81 and probe 4 (Fig. 7). This inhibition was specific because the −88/−61-D3 enhancer sequence containing the mutated binding site for C/EBP did not inhibit the complex formation (Fig. 7). Addition of oligonucleotides containing consensus binding sites for the transcription factors other than C/EBP did not inhibit the complex formation (data not shown). These results show that C/EBP may bind to the D3 enhancer.

To delineate the function of the binding site for C/EBP for D3 enhancer activity, we constructed the D3 enhancer containing the mutated binding site for C/EBP that was linked to R2P-EGFP (Mut-C/EBP). We then transfected it into 18.81 cells, and compared its enhancer activity with that of the wild-type D3-R2P-EGFP (shown in Fig. 4). We found the enhancer activity of Mut-C/EBP decreased significantly compared with that of the wild-type enhancer (p < 0.01), although the inhibition was not so remarkable (Fig. 8). These data suggest that C/EBP binds to the D3 enhancer and functions as one of transcription factors or coactivators responsible for the enhancer activity.

We and others have shown that the tissue- and stage-specific expression of RAG-1 and RAG-2 is regulated by alteration of the chromatin structure of the promoter region (13, 14) and/or transcriptional factors that bind the promoter region (18, 19, 20, 21). In the present study, we characterized the chromatin structure of the 25-kb DNA segment adjacent to murine RAG-2, and identified three lymphocyte-specific HS sites (Fig. 1). Subsequently, we determined the enhancer element(s) in the 5′-upstream region of RAG-2 using EGFP reporter assays. Two enhancer elements (D3 and P4) were identified that corresponded to the lymphocyte-specific HS sites (HS1 located at −8.1 kb and HS2 at −2.3 kb 5′ upstream of mouse RAG-2) (Figs. 2 and 3). One of the enhancer elements, the distal enhancer element (D3), was shown to be specifically active in lymphoid cell lines in in vitro transient assays (Fig. 4) and active in the primary lymphoid organs (i.e., thymus and bone marrow) in the D3 minigene transgenic mice (Fig. 5). Our data indicate that the tissue- and stage-specific expression of murine RAG-2 may be regulated not only by the promoter region, but also by chromatin remodeling and other cis-regulatory element(s).

Recently, Yu et al. (22) have reported that approximately the 10-kb 5′-upstream region of murine RAG-2 is sufficient for the expression of the GFP reporter gene in B lineage cells as well as in a subset of thymocytes in vivo. Monroe et al. (23) have also demonstrated that the 9-kb 5′-upstream region of murine RAG-2 is enough to rescue B cell- and T cell-development using the RAG-2^−/− blastocyst complementation assay. Their results indicated that there are different regulatory elements in the −2- to −7-kb and the −7- to −9-kb regions, respectively. The proximal element contributed to B cell lineage expression, while the distal element functioned to regulate both B and T cell lineage expression. These studies suggest that at least two lymphoid-specific regulatory elements, such as enhancer(s), may exist in this 5′-upstream region of murine RAG-2. With this respect, we identified P4 and D3 elements locating at −2.3- and −8.1-kb 5′ upstream of RAG-2, respectively, the results corresponding to the putative elements proposed by Monroe et al. (23).

The D3 enhancer was functional in vivo. We generated transgenic mice carrying a minigene containing D3, RAG-2 promoter, and EGFP and analyzed expression of EGFP in various tissues. We showed that 6 of 11 lines expressed EGFP only in the thymus and bone marrow, but not in spleen, liver, kidney and brain, indicating that D3 enhanced mouse RAG-2 promoter activity in vivo specifically in primary lymphoid organs. We further investigated the expression of EGFP in subsets of thymocytes and bone marrow cells in the transgenic mice. We revealed that D3 was active in CD4⁻CD8⁻ (DN), but not in CD4⁺CD8⁺ (DP) or CD4⁺CD8⁻ (SP), thymocytes in the thymus, and was also active in B220⁺IgM⁻, but not in B220⁺IgM⁺, cells in the bone marrow (Fig. 5 C). These results support the previous observation by Yu et al. (22) which found that the 10-kb 5′ upstream region of RAG-2 functions for the expression of the GFP reporter gene in CD4⁻CD8⁻ thymocytes, as well as B lineage cells, in vivo.

An interesting finding that emerged during analysis of the transgenic mice is that the transgene was active in the brain in 3 lines of 11 transgenic founders (Table II). Is it an artifact? Because most of the transgenic mice expressed the transgene only in primary lymphoid organs but not in the brain, D3 was not sufficient for the expression of the transgene in the brain. However, those three transgenic lines expressed the transgene in the brain as well as primary lymphoid organs, but not in other organs. This observation indicates that D3 may be able to augment the transcription in the brain in some situations, such as being located near some cis-regulatory elements or being in the presence of an additional cis-element in RAG locus. With regard to this, Chun et al. (37) have shown that RAG-1, but not RAG-2, is expressed in the nervous system. Yu et al. (22) have also shown that the 5′-upstream region of murine RAG-2 can regulate the expression of mouse RAG-1. Thus, the D3 enhancer could be involved in the expression of murine RAG-1 in the brain.

Although EGFP transcripts were clearly detected by RT-PCR in thymocytes and bone marrow cells in the D3-R2P-EGFP transgenic mice (Fig. 5), we could not detect EGFP proteins by either flow cytometry or Western blotting. There are two possibilities to explain this result. First, it is due to the copy numbers of transgenes in the mice. In fact, the copy numbers in the transgenic lines did not exceed 10 copies (Table II). When we established and analyzed stable D3-R2P-EGFP transfectants of EL-4 cells whose EGFP-proteins were detected by flow cytometry, they contained >50 copies of EGFP-construct (data not shown). Another possibility, which we think is more likely, is that our D3 transgenic mice are missing cis-regulatory element(s) such as locus control region (LCR). LCR, which locates near, or sometimes apart from, enhancer element(s), was found to be required for high level, copy number-dependent gene activation in vivo (38, 39). In this respect, Yu et al. (22) generated transgenic mice carrying a bacterial artificial chromosome DNA that spanned a long range of the 5′-upstream to 3′-downstream region of the mouse RAG locus, where the EGFP reporter gene was inserted. It was shown that EGFP proteins were detected by flow cytometry in the thymocytes and bone marrow cells of these transgenic mice. Furthermore, generation of EGFP-knockin mice of the RAG-1 or RAG-2 locus that carry a single copy of EGFP showed that only one copy of EGFP is enough to induce EGFP proteins in the transgenic mice (7, 40). These results strongly suggest that other cis-element(s) or LCR are necessary for the D3 enhancer element to induce higher levels of EGFP expression. This element(s) should be determined.

Transcripts of EGFP were detected in bone marrow and thymus, but not detected in spleen, showing that the enhancer activity of distal element is correlated with RAG expression in immature lymphoid cells. The result suggests that the cell-specific transcription of RAG-2 is also regulated by lymphoid-specific trans-acting factors. With this regard, a database search revealed the putative binding sites of transcription factors in the 380-bp core distal enhancer element (Fig. 6,A). Using EMSA, we demonstrated that C/EBP, one of the transcriptional factors that controls the transcription of genes involved in a broad range of physiological processes, as well as lymphoid cell development (41, 42), may bind to the D3 enhancer (Figs. 6,B and 7). We also showed that C/EBP may function as one of the transcription factors or coactivators responsible for the enhancer activity (Fig. 8). It is necessary to determine other transcription factors and/or associated molecules that are responsible for the full D3 enhancer activity. Identification of these factors may elucidate the transcriptional regulation by the distal enhancer element.

We thank Sanae Hirota and Taro Kitagawa for their technical assistance, and Kaoru Hata for her secretarial work. We are also grateful for the kind supply of the cell lines 18.81 and BAL-17 by Nobuo Sakaguchi and the cell line WEHI3 by Hitoshi Kikutani.