Coordinate Regulation of RAG1 and RAG2 by Cell Type-Specific DNA Elements 5' of RAG2

See allHide authors and affiliations

Science  13 Aug 1999:
Vol. 285, Issue 5430, pp. 1080-1084
DOI: 10.1126/science.285.5430.1080


RAG1 and RAG2 are essential for V(D)J recombination and lymphocyte development. These genes are thought to encode a transposase derived from a mobile genetic element that was inserted into the vertebrate genome 450 million years ago. The regulation of RAG1 and RAG2 was investigated in vivo with bacterial artificial chromosome (BAC) transgenes containing a fluorescent indicator. Coordinate expression of RAG1 andRAG2 in B and T cells was found to be regulated by distinct genetic elements found on the 5′ side of the RAG2 gene. This observation suggests a mechanism by which asymmetrically disposed cis DNA elements could influence the expression of the primordial transposon and thereby capture RAGs for vertebrate evolution.

Vertebrates assemble immunoglobulins (Igs) and T cell receptors (TCRs) by a site-specific DNA recombination reaction known as V(D)J recombination (1). V(D)J recombination occurs only in lymphocytes and is catalyzed by the protein products of the recombinase-activating genes RAG1 and RAG2(2). RAGs initiate V(D)J recombination by recognizing recombination signal sequences that flank Ig and TCR variable, diversity, and joining gene segments (3). RAG-catalyzed DNA cleavage proceeds by a mechanism similar to bacterial Tn10 transposition (4). Like Tn10, RAGs nick double-stranded (ds) DNA (3) and then cleave the opposite strand by trans-esterification (5). The hairpin products of trans-esterification can then be opened by RAGs or Tn10 in a second DNA-nicking reaction (6, 7). Ubiquitous DNA ds-break repair factors, including Ku70 and 80, DNA-dependent protein kinase, XRCC4, and ligase IV, are required to repair the RAG-induced breaks and for normal lymphocyte development (8).

Coordinate expression of RAG1 and RAG2 is restricted to lymphocytes and is essential for lymphocyte development (2, 9). In the absence of either RAG1 or RAG2, B and T cell development is completely abrogated (9). Conversely, aberrant RAG expression results in abnormal lymphocyte development (10). Despite the central importance of regulated RAG expression in the development, evolution, and function of the immune system, little is known about the molecular mechanisms that control RAG transcription.

To examine the mechanisms that regulate RAG2 expression in vivo, we produced transgenic mice that carry bacterial artificial chromosomes (BACs) modified by homologous recombination to encode a green fluorescent protein (GFP) reporter instead of RAG2 (11,12). Three overlapping BACs (NG, HG, and MG) that cover the RAG locus were obtained by screening libraries with oligonucleotides complementary to RAG1 (Fig. 1). The HG BAC was further modified by homologous recombination to create additional BAC constructs (Fig. 1).

Figure 1

Diagram of BAC transgenes and summary of indicator expression in T cells and B cells. RAG2-GFP (12) and RAG1-YFP expression was detected by flow cytometry. To produce R1Δ, HGΔ, HGΔ2, and HGΔ3, HG was further modified by homologous recombination (26). HYG and HYGΔ were produced by inserting EYFP (Clontech) at theRAG1 start codon of HG and HGΔ, respectively. HYG and HYGΔ were further modified to produce INSΔ and R2Δ, respectively (26). Symbols: ∗, RAG1-YFP or RAG2-GFP (or both) present at levels comparable to NG in DN but not in DP thymocytes (Fig. 2B); ^, RAG1-YFP present at low levels in developing B cells (See Fig. 4A) and DN thymocytes (21), but not in DP thymocytes.

We first investigated whether GFP-containing BACs could direct proper transcriptional regulation of RAG2 in lymphocytes by analyzing mice transgenic for the NG BAC, as the RAG coding regions were located centrally within this genomic insert (Fig. 1) (12). Bone marrow B cells were developmentally staged by cell surface markers and analyzed by flow cytometry (13) (Fig. 2A). Green fluorescence was detected in bone marrow fraction A1, which contains the most immature committed B cell precursors, and fluorescence increased in the pre-pro-B cell subpopulation, where V(D)J recombination is first detected (fraction A2) (13). RAG2-GFP expression peaked in pro-B cells that undergo immunoglobulin heavy chain D-J and V-DJ recombination (fractions B and C) and decreased in large cycling pre-B cells (fraction C′) (13, 14). Small pre-B cells undergoing light chain gene recombination also had high green fluorescence (fraction D) that disappeared in mature recirculating CD43B220high B cells (fraction F). This temporal pattern of RAG2-GFP expression matched reported endogenous RAG2 expression (13,14), and coordinate expression of RAG2-GFP mRNA and endogenous RAG2 mRNA was confirmed (12, 15). Thus, NG BAC contained all cis elements necessary for the proper expression of RAG2 mRNA in B cells.

Figure 2

GFP expression in the bone marrow, thymus, and spleen of NG transgenic mice. (A) Cytofluorographic analysis of GFP expression in developing B cells of NG transgenic mice. Histograms show GFP expression on gated fractions (13). (B) Confocal micrograph showing GFP expression in the thymus from an NG transgenic mouse. Thymic cortex (CTX) and medulla (MDL) are indicated. (C) Cytofluorographic analysis of GFP expression in T cells from thymus and spleen of NG transgenic mice. Histograms show GFP fluorescence in CD4CD8 DN and CD4+CD8+ DP thymocytes and in CD4+CD8 and CD4CD8+SP splenocytes. Thymocytes and spleen cells were stained with phycoerythrin (PE) antibody to CD8 and biotin antibody to CD4 (Pharmingen; 53-6.7 and RM4-4, respectively) developed with streptavidin red-670 (Gibco-BRL). (D) Histogram shows GFP expression in CD25+ or CD25 DN thymocytes. APC-GK1.5 (antibody to CD4), APC-53-6.7 (antibody to CD8a), and BI-PC61 (antibody to CD25) followed by TR-Avidin and PI to label dead cells were used to visualize thymocyte DN subsets.

Consistent with previous in situ work on endogenous RAGexpression in the thymus, GFP expression was most prominent in the cortex, where immature CD4+CD8+ double-positive (DP) T cells undergo V(D)J recombination (Fig. 2B) (16). In contrast, the thymic medulla, which contains more mature T cells, was only weakly fluorescent.

Thymocytes were then analyzed by flow cytometry. Among CD4CD8 double-negative (DN) thymocytes, γ, δ, and β chain rearrangements are detected in the CD25+ subpopulation. Consistent with this, RAG2-GFP was expressed in the CD25+ subset of DN thymocytes (Fig. 2, C and D). DP thymocytes were fluorescent, but mature CD4+CD8 or CD4CD8+single-positive (SP) T cells in the spleen were not (Fig. 2C) (15). Thus, the NG BAC contained all cis elements necessary for regulated transcription of RAG2 in T cells as well as in B cells (17).

Two additional BAC constructs, HG and R1Δ (Fig. 1), also directed proper RAG2 transcription in T and B cells (Fig. 3A); that is, the pattern of expression was the same as in NG BAC transgenic mice. As judged by the overlap between the NG, HG, and R1Δ BACs, region I is not required for regulated RAG2 expression in T or B cells (Fig. 1). Thus, the 100-kb genomic overlap between NG and R1Δ is sufficient to direct regulated expression of RAG2 in both T and B cells (Fig. 1) (17).

Figure 3

RAG2-GFP expression in mice transgenic for modified BACs. (A) (Left) Histograms show GFP expression in bone marrow cells electronically gated on the basis of PE antibody to B220 and biotin antibody to CD43 staining (Pharmingen; RA3-6B2 and S7, respectively, developed with streptavidin red- 670). (Right) Histograms show GFP expression in T cells from thymus and spleen gated as indicated. WT, wild type; DN, CD4CD8 double negative; DP, CD4+CD8+ double positive. (B) RAG2-GFP expression in HGΔ DN thymocytes. Histograms show GFP expression in electronically gated CD25+ or CD25 DN thymocytes. DP and SP thymocytes were depleted with CD4 and CD8 magnetic beads (Dynal) and stained with PE antibody to CD25, APC antibody to CD4, and APC antibody to CD8 (Pharmingen; PC61, RM4-4, and 53-6.7, respectively).

The MG BAC has only 10 kb of genomic sequence 5′ ofRAG2 that extends into region II, and the HGΔ BAC has only the overlap between MG and HG BACs (Fig. 1), yet mice transgenic for either BAC had normal RAG2-GFP expression in developing B cells and DN thymocytes (Fig. 3A). In contrast, RAG2-GFP expression in DP thymocytes was reduced in both sets of transgenic mice (Fig. 3A).

To narrow the cis region necessary for RAG2 expression in DP thymocytes, we created the HGΔ2 and HGΔ3 BACs, which contain genomic sequences extending an additional 5.5 kb and 19.5 kb, respectively, into region II as compared with HGΔ (Fig. 1). HGΔ2 and HGΔ3 transgenic mice resembled HGΔ transgenic mice in that GFP fluorescence was found in developing B cells and DN thymocytes, but not in DP thymocytes (Fig. 1). Thus, the DNA sequences required to direct fully regulated expression of RAG2 in DP thymocytes are found in a 55-kb interval between the ends of BACs HGΔ3 and NG in region II (Fig. 1).

To determine whether RAG2 expression in DN thymocytes in HGΔ transgenic mice was properly regulated as in B cells, we subfractionated these cells according to CD25 expression. As in NG mice, CD25+ DN thymocytes from HGΔ mice expressed RAG2-GFP and CD25 cells did not (Figs. 2D and 3B). Thus,RAG2 was properly regulated in DN thymocytes in HGΔ transgenic mice, and expression of RAG2 in DN and DP thymocytes can be dissociated. We conclude that the 23 kb of sequence found in the overlap between R1Δ and HGΔ BACs contains sufficient information to direct expression of R2-GFP in a developmentally appropriate fashion in B cells and DN T cells but not in DP T cells.

A pattern of greatly reduced RAG2-GFP expression in DP thymocytes coupled with relatively normal GFP levels in DN thymocytes and B cells was observed in 18 of 19 independent founders transgenic for the MG, HGΔ, HGΔ2, or HGΔ3 BACs (Fig. 1). This abnormal pattern ofRAG2 expression suggests the existence of proximal elements, which contribute to, but cannot fully direct, T cell expression ofRAG2. Consistent with this, the human and mouseRAG1 and RAG2 promoters have basal transcriptional activity (18). In particular, in vitro experiments indicate that the murine RAG2 promoter has lymphoid specificity and is regulated differently in T and B cells (18). We conclude that a promoter distal element or elements in region II likely interact with more proximal elements to confer complete transcriptional regulation of RAG2 (19).

To examine the regulation of RAG1 in vivo and to determine whether RAG1 and RAG2 are coordinately expressed, we created the HYG BAC by inserting a yellow fluorescence protein (YFP) indicator into the RAG1 gene of the HG BAC (Fig. 1). Four lines of transgenic mice that carry HYG were analyzed; two of the four lines expressed both genes and two lines expressed neither (Fig. 1). In both expressing lines, B and T cells displayed a broader distribution of yellow than green fluorescence (Fig. 4A). Nevertheless, RAG1-YFP and RAG2-GFP expression were coordinate, and their temporal regulation was indistinguishable from that shown when RAG2 was assayed alone in HG transgenic mice (Figs. 1 and 4A) (20). Deletion of the intergenic sequence between RAG1 and RAG2did not alter the pattern of expression of either RAG1-YFP or RAG2-GFP (INSΔ, Fig. 1). Thus, the intergenic sequence was not essential for regulation of either RAG1 or RAG2, and the HYG BAC carries sufficient information to direct the expression ofRAG1 and RAG2 in B cells and T cells in a developmentally regulated and coordinate fashion.

Figure 4

Expression of RAG1-YFP and RAG2-GFP in mice transgenic for modified BACs. (A) Dot plots show YFP and GFP coexpression in bone marrow lymphocytes and thymocytes electronically gated on the basis of forward and side scatter. (B) Two models of transcriptional activation at the RAG locus. E, proposed cis regulatory element or elements.

To explore the cis requirements for RAG1 expression in T cells, we generated the HYGΔ BAC by inserting YFP into the start codon of RAG1 in HGΔ (Figs. 1 and 4A). HYGΔ directed coordinate expression of RAG1-YFP and RAG2-GFP in developing B cells (Fig. 4A) and DN thymocytes (Fig. 1), and, again, there was a broader distribution of YFP fluorescence than GFP fluorescence. In contrast, neither YFP nor GFP was expressed in DP thymocytes in the eight transgenic HYGΔ founders (Fig. 1). Thus, the 105-kb genomic interval in region II between the ends of HYGΔ BAC and the HYG BAC is essential for regulated expression of RAG1 in DP thymocytes, but not in DN thymocytes or B cells (Fig. 1). Expression ofRAG1 in DP thymocytes, therefore, requires DNA elements found on the RAG2 side of the locus.

To determine whether the cis elements required for regulated expression of RAG1 in B cells and DN T cells are on the RAG1or RAG2 side of the RAG locus, we removed sequences in region II from the HYGΔ BAC to create the R2Δ BAC (Figs. 1 and 4A). In five of the seven R2Δ founders, low yellow fluorescence was detected in pro- and pre-B cells and in DN thymocytes (Figs. 1 and 4A) (21), indicating that elements on the RAG1 side of the locus contribute to regulation of RAG1 in these cells (18) but that these elements are not sufficient for high expression of RAG1. In contrast to B cells and DN thymocytes, RAG1-YFP was not detectable in DP thymocytes in R2Δ transgenic mice (Fig. 1) (21). We conclude that bothRAG1 and RAG2 are in part regulated by cis elements located 5′ of RAG2 in region II.

RAG1 and RAG2 are unusual genes. They have remained closely linked and convergently transcribed throughout vertebrate evolution and have similarities to transposases (2, 22), including in vitro transposase activity (23). This led to the proposal that RAG1 andRAG2 entered the vertebrate genome as a mobile genetic element (2) and that transposase target sequences were inserted into a primordial receptor gene to create the first split antigen receptor gene (23, 24). The finding that RAG1 and RAG2 are coordinately regulated by asymmetrically disposed elements supports this hypothesis and suggests that the original integration may have occurred at a site where elements on one side of the transposon were able to influence the expression of both RAG genes coordinately. Lymphoid specificity may have been added after primordial transposition.

How can a set of elements on the RAG2 side of the RAG locus control transcription of both RAG1 and RAG2? One possibility is that elements on the RAG2 side of the locus alternately loop between the RAG1 and RAG2promoters by a switching mechanism similar to that of the human β globin locus (25). Such a switching model would require that RAGs remained linked but would not explain why they remained in a convergent transcriptional orientation (Fig. 4B). A second possibility is that the regulatory elements in region II loop to interact with both RAG1 and RAG2promoters simultaneously (Fig. 4B). DNA looping to coordinate transcription would require that RAG1 and RAG2remain closely linked in the genome, would allow for separate transcriptional control in T and B cells, and might favor conservation of the original convergent orientation of RAG1 andRAG2.

  • * To whom correspondence should be addressed. E-mail: nussen{at}


Stay Connected to Science

Navigate This Article