RAG1 and RAG2 are the lymphocyte-specific components of the V(D)J recombinase. In vitro analyses of RAG function have relied on soluble, highly truncated “core” RAG proteins. To identify potential functions for noncore regions and assess functionality of core RAG1 in vivo, we generated core RAG1 knockin (RAG1c/c) mice. Significant B and T cell numbers are generated in RAG1c/c mice, showing that core RAG1, despite missing ∼40% of the RAG1 sequence, retains significant in vivo function. However, lymphocyte development and the overall level of V(D)J recombination are impaired at the progenitor stage in RAG1c/c mice. Correspondingly, there are reduced numbers of peripheral RAG1c/c B and T lymphocytes. Whereas normal B lymphocytes undergo rearrangement of both JH loci, substantial levels of germline JH loci persist in mature B cells of RAG1c/c mice, demonstrating that DJH rearrangement on both IgH alleles is not required for developmental progression to the stage of VH to DJH recombination. Whereas VH to DJH rearrangements occur, albeit at reduced levels, on the nonselected alleles of RAG1c/c B cells that have undergone D to JH rearrangements, we do not detect VH to DH rearrangements in RAG1c/c B cells that retain germline JH alleles. We discuss the potential implications of these findings for noncore RAG1 functions and for the ordered assembly of VH, DH, and JH segments.
Ig and TCR variable region genes are assembled during early lymphocyte development from component variable (V), diversity (D), and joining (J) gene segments. V(D)J recombination is initiated by the RAG1 and RAG2 proteins via introduction of DNA double strand break (DSBs) between the V, D, and J coding segments and flanking recombination signal (RS) sequences. RAG1 and RAG2 are necessary in vivo and sufficient in vitro for initiation of V(D)J recombination (for review see reference 1). After DNA cleavage, RAG-cleaved coding and RS ends are joined by ubiquitously expressed, nonhomologous end-joining (NHEJ) proteins (for review see reference 2).
V(D)J recombination is tightly regulated during lymphocyte development within the context of lymphocyte lineage specificity, developmental stage specificity, and feedback regulation of allelic exclusion (for reviews see references 2, 3). Furthermore, the developmental progression of lymphocytes requires the productive assembly and expression of antigen receptor genes. In developing B lymphocytes, IgH genes are assembled before IgL genes; whereas in developing αβ T cells, TCRβ genes are assembled before TCRα genes (for review see reference 3). Both IgH and TCRβ genes are assembled via an ordered process in which D to J rearrangements proceed to completion on both alleles, followed by V segment rearrangement to a preexisting DJ complex (4, 5). Productive rearrangement and expression of IgH μ chains in ckit+/B220int/CD43+/CD25−/CD19+ progenitor (pro–)B cells and TCRβ chains in CD4−CD8− (double negative [DN]) pro–T cells induces cellular expansion and differentiation, respectively, to the corresponding ckit−/B220+/CD43lo/CD25−/CD19+ pre–B cell stage and the CD4+/CD8+ (double positive [DP]) thymocyte stage (6, 7). Functional rearrangement and expression of IgL or TCRα chains as part of surface IgM or αβ TCR signals precursor B or T lymphocytes to develop into mature peripheral lymphocytes (for reviews see references 7, 8).
Biochemical activities of the RAG1 and RAG2 proteins have been extensively characterized in vitro using purified proteins and defined DNA substrates (for review see reference 9). Full-length RAG proteins are largely insoluble when overproduced; consequently, in vitro studies have used truncated “core” RAG proteins. Core RAG1 (aa 384–1,008 of 1,040 residues) and core RAG2 (aa 1–383 of 527 residues) comprise the minimal regions of RAG1 and RAG2 necessary for recombination of extrachromosomal substrates in nonlymphoid cells (10–13). Whereas core RAGs mediate the basic V(D)J recombination reaction in vitro, the activities of these truncated proteins are not identical to those of full-length RAGs. In this context, noncore RAG regions influence both V(D)J recombination efficiency and products formed when assayed in cell lines (14–19).
Noncore RAG regions are evolutionarily conserved, suggesting these regions may serve important functions (11). Studies of A-MuLV–transformed pre–B cells have implicated noncore RAG regions in accessory and/or regulatory functions in chromosomal V(D)J recombination (18, 19). Thus, NH2-terminal elements of RAG1 are required for complete DH to JH rearrangement in A-MuLV transformants (19), whereas the COOH terminus of RAG2 may be more important for VH to DJ H rearrangement than for DH to JH rearrangement (18). In this regard, mice expressing core RAG2 in place of endogenous RAG2 (which we will refer to here as RAG2c/c) display impaired lymphocyte development at the pro–B and pro–T cell stages, accompanied by decreased levels of V to DJ rearrangement (20, 21). Inactivating mutations in RAG1 or RAG2 lead to a complete block in lymphocyte development in mice (22, 23) and are a cause of human T−B− SCID (24, 25). However, missense mutations in RAG1 or RAG2 that result in reduced, and possibly altered, V(D)J recombinase activity lead to Omenn Syndrome (OS), a disease characterized by lack of B cells and a reduced, oligoclonal T cell repertoire (24, 26). In addition, frameshift mutations in noncore RAG1 regions can result in NH2-terminal RAG1 truncations that lead to OS-like immunodeficiencies (27, 28). To address the role of noncore RAG1 regions in vivo, we have generated mice containing specific replacement of the full-length endogenous RAG1 gene with a gene encoding the mouse core RAG1 protein.
Materials And Methods
Antibodies and Flow Cytometry.
Single cell suspensions from lymphoid tissues were stained with antibodies conjugated to FITC, PE, and cytochrome C (CyC) by standard procedures. The following antibody conjugates (BD Biosciences and Southern Biotechnology Associates, Inc.) were used: CyC anti-CD44, PE anti-CD25, CyC anti-CD4, FITC anti-CD8, FITC anti-CD43, FITC anti-GR1, FITC anti-TCRγ,δ, CyC and FITC anti-B220, and PE anti-IgM. A total of eight RAGc/c mice, six RAG+/c, and six RAG+/+ mice between three and five wk of age were analyzed. pro– and pre–B cell populations were sorted based on CD43 and B220 expression after gating out all cells staining with IgM. Cell sorting was performed on independent BM samples from five RAG1c/c, one RAG1+/c, and four RAG1+/+ littermates.
Generation and Analysis of B Cell Hybridomas.
Single cell splenocyte suspensions from RAG1+/+, RAG1+/c, and RAG1c/c mice were cultured with anti-CD40 (1 μg/ml) and IL-4 (10 ng/ml) for 4 d and activated B cells fused to the NS-1 fusion partner (ATCC TIB-18). Hybridomas were screened for isotype secretion by sandwich ELISA using isotype-specific antibodies from SBA. Genomic DNA from each Ig-secreting hybridoma was assayed for Ig rearrangements by Southern blotting. JH rearrangement status and the number of alleles was determined via a JH-specific probe on StuI- or EcoRI-digested DNA (29). VH to DJH rearrangements were detected with a probe from 5′ of DFL16; VH to DJH rearrangements will result in deletion of this fragment (hybridomas nos. 10 and 12); whereas nonrearranged alleles or D to JH rearranged alleles will retain this band. DFL16 to JH joins will result in a band of unique size. Nonrearranged JH alleles were confirmed by Southern blot analyses using a probe from 3′ of DQ52 (30). D to JH rearrangements result in deletion of this fragment; thus, alleles that have not undergone DH to JH rearrangements will retain this germline band. Cell lines with only one detectable IgH allele originating from B cells were not further analyzed.
Purification of B and T Cell Fractions.
Single cell suspensions of splenocytes from RAG1+/+, RAG1+/c, and RAG1c/c were cultured with anti-CD40 and IL4 for B cell activation or ConA and IL-2 for T cell activation. Activated B cells from day 2 cultures were purified using biotinylated anti-CD19–specific antibodies in conjunction with streptavidin-conjugated MACS microbeads and MACS Separation Columns (Miltenyi Biotech). T cells cultured with ConA + IL-2 for 6 d were purified by Lympholyte-M sedimentation (Cedarlane). Purity of separated B and T cells was analyzed by flow cytometry and found to be at least 90%. DP T cell DNA was isolated by staining thymocytes with FITC anti-CD4 and PE anti-CD8 antibodies and sorting on a MO-Flo (Becton Dickinson). DN3 thymocyte DNA was isolated by first depleting CD4+ thymocytes using anti-CD4-conjugated MACS microbeads and staining the unbound cell fraction with CYC anti-CD44, PE anti-CD25, and a combination of FITC anti-CD4, -CD8, -B220, -Mac1, -GR1, and -TCRγ,δ antibodies and sorting on live-gated FITC−, CYC−, and PE+ cells.
PCR Analysis of IgH and TCR Rearrangements.
Genomic DNA from sorted pro– and pre–B cell populations and purified CD4+CD8+ DP or CD4−CD8−CD44−CD25+ DN3 T cells was analyzed for IgH and TCRβ rearrangements by PCR amplification (31, 32). LMPCR for SE breaks were performed as described previously (33). Oligos and PCR primers are listed in Table S1 . Briefly, PCR reactions contained 90, 30, 10, and 3.3 ng or 250, 50, 10 and 2 ng of genomic DNA for IgH and TCRβ amplifications, respectively, 20 pmol of each primer, 0.2 mM dNTPs, 50 mM KCl, 10 mM Tris-HCl (pH 8.0), 2.5 mM MgCl2, and 0.5 U QIAGEN Taq polymerase. Amplification conditions were as follows: 94°C for 45 s, 60°C for 1 min, 72°C for 2.5 min for 30 cycles. Amount of input genomic DNA was normalized by PCR amplification of a coding exon of the ATM gene. PCR products were analyzed by agarose gel electrophoresis, transferred to Zetaprobe membrane, and probed with nested oligonucleotide probes to detect rearrangements. PCR analyses were performed at least three times on genomic DNA samples from sorted B cells and purified T cells, respectively, isolated from at least three RAG1c/c mice and either RAG1+/c or RAG1+/+ littermates.
Generation of Core RAG1 Mice.
The pMS127 plasmid containing the core RAG1 coding sequence and three copies of a human c-Myc tag at the COOH-terminal end was provided by M. Sadofsky (Albert Einstein College of Medicine) (11). The core RAG1 coding sequence contains a single nucleotide change (A to G) that results in a methionine to valine substitution at position 484. A similar form, with an additional 9-residue Histidine COOH-terminal tag, has been used extensively in vitro to define the biochemical and V(D)J recombination activities intrinsic to the RAG recombinase (34–36) 2.1 kb of homologous sequence immediately 5′ of RAG1 was cloned along with core RAG1 coding sequence into the SalI site of pLNTK (37). A 3.3-kb MluI to BamH1 fragment was cloned into the XhoI site to complete the core RAG1 replacement vector. TC1 embryonic stem cells (provided by P. Leder, Harvard Medical School) were transfected with the targeting vector as described previously (38). A positively selected recombinant was identified by Southern blotting using the 5′ and 3′ targeting probes (Fig. S1). The LoxP flanked neomycin resistance (Neor) gene was deleted by transiently transfecting the ES cell clone with a plasmid-based Cre recombinase expression construct (37). Subclones of this ES cell line were injected into C57Bl/6 blastocysts to generate chimeric mice. RAG1+/c mice were intercrossed to produce 129sv RAG1c/c mice.
The Generation of Mice that Express the Core RAG1 Protein.
To evaluate in vivo functions of noncore RAG1 regions, we used gene targeting to replace the endogenous RAG1 gene with a recombinant gene encoding the mouse core RAG1 protein (aa 384–1008) (Fig. S1). We first generated targeted ES clones with endogenous RAG1 replaced on a single allele by the synthetic core RAG1 gene and a 3′loxP-PGK-Neor cassette (Fig. S1, A and B). We then used transient expression of Cre recombinase to delete the PGK-Neor cassette (Fig. S1 B). Chimeric mice from Cre-deleted ES cell clones were bred to 129sv mice to obtain mice carrying the core RAG1 gene in their germline. Heterozygous core RAG1 (RAG1+/c) mice were intercrossed to produce RAG1c/c mice and RAG1+/c and RAG1+/+ littermate controls (Fig. S1 C). The core RAG1 protein was expressed at elevated levels compared with the full-length protein, as predicted by previous studies (13, 15), and was found mostly in nuclear extracts (Fig. S1 D). RAG1c/c mice were housed in a pathogen-free environment and survived into adulthood without overt defects.
Core RAG1 Mice Exhibit Impaired Lymphocyte Development.
We found reduced thymic cellularity of RAG1c/c mice compared with RAG1+/c and RAG1+/+ control mice (average thymocyte count for RAG1c/c thymuses = 29% ± 11% of controls; Fig. 1 A). FACS® analyses of RAG1c/c thymocytes demonstrated an almost normal distribution of CD4+/CD8+ (DP) and CD4+/CD8− or CD4−/CD8+ (single positive) populations (Fig. 1 A). However, RAG1c/c mice accumulated CD4−/CD8− (DN) pro–T cells with a strikingly high percentage of cells amassing at the CD25+CD44− (DN3) stage compared with RAG1+/c and RAG1+/+ control mice (Fig. 1 B). TCRβ rearrangements are completed within the DN3 stage, with productive TCRβ expression inducing cellular proliferation, differentiation to the CD25−CD44− DN4 stage, and cessation of further Vβ to DJβ rearrangement (allelic exclusion) (for review see reference 39). Thus, the accumulation of RAG1c/c DN3 thymocytes is consistent with diminished capacity of core RAG1 to efficiently mediate complete VβDJβ rearrangements.
TCRα chain expression is not required for the DN to DP transition (for reviews see references 7, 39). However, we found an increased ratio of TCRβlo to TCRβint DP thymocytes in RAG1c/c mice compared with controls (Fig. 1 D). This is consistent with delayed TCRα rearrangement, since DP thymocytes from mice that lack either the TCRα enhancer or TCRα genes likewise express low levels of TCRβ at their cell surface in the absence of TCRα chains (40, 41). We did not detect TCRβ− DP thymocytes in RAG1c/c mice, confirming that delayed development due to decreased V(D)J recombination efficiency does not allow progression to the DP stage in the absence of TCRβ gene expression (Fig. 1 D). The absolute numbers of splenic αβ T cells in RAG1c/c mice were reduced ∼50% from that of RAG1+/c and RAG1+/+ controls (Fig. 1 C). We conclude that overall T cell development in RAG1c/c mice is impaired with respect to the pro– to pre–T cell transition, accompanied by a modest reduction in overall T cell numbers.
We detected a statistically significant reduction in the percentage and absolute number of CD43loB220+ pre–B cells in RAG1c/c mice relative to RAG1+/c and RAG1+/+ control mice (average pre–B cell count RAG1c/c BM = 6.5 ± 2.9 × 105 versus 10.3 ± 1.3 × 105 in control littermates, P = .020 by paired t test; Fig. 2 A) (6). Correspondingly, the absolute splenic B cell numbers in RAG1c/c mice were ∼50% of those in control mice (Fig. 2 C). On the other hand, absolute BM CD43+B220int pro–B cell numbers were relatively normal in RAG1c/c mice (average pro–B cell count RAG1c/c BM = 5.2 ± 1.7 × 105 versus 5.5 ± 1.5 × 105 in control littermates); and the percentages of Hardy fractions D, E, and F in the BM of RAG1c/c and littermate controls were also comparable (Fig. 2, A and B) (6). Therefore, B cell development in RAG1c/c mice shows a modest impairment in the pro– to pre–B cell transition and a corresponding reduction in overall B cell numbers.
TCRβ and TCRα Gene Rearrangements in RAG1c/c Mice.
Since RAG1c/c mice exhibit impaired T cell development, we assayed for potential alterations in TCRβ rearrangement. We employed PCR primers located 5′ of Dβ1 or Dβ2 and 3′ of Jβ1 or Jβ2 to amplify Dβ1 to Jβ1 and Dβ2 to Jβ2 rearrangements from purified CD44−CD25+ DNIII or DP thymocytes from RAG1c/c and control mice (32). The levels of PCR products representing most Dβ1 to Jβ1 rearrangements were reduced in DNA isolated from DN3 T cells of RAG1c/c mice compared with controls (Fig. 3 A). Similarly, Dβ2 to Jβ2 levels were reduced and were accompanied by increased levels of germline Dβ2/Jβ2 bands (Fig. 3 A). In contrast, Dβ1 to Jβ2 rearrangement levels were comparable between RAG1c/c and control mice (Fig. 3 A). We found a similar reduction in Dβ1 to Jβ1 and Dβ2 to Jβ2 rearrangements in DNA isolated from sorted RAG1c/c DP thymocytes, but Dβ1 to Jβ2 rearrangements again were not markedly changed (Fig. S2 A). There was also a diminished level of 3′ Dβ broken RS ends in DN3 thymocyte DNA from RAG1c/c mice, consistent with reduced RS cleavage by core RAG1 (Fig. 3 C).
We employed Vβ8- and Vβ10-specific primers in conjunction with 3′ Jβ1 or 3′ Jβ2 primers to determine whether Vβ rearrangement is altered in RAG1c/c versus WT mice. Vβ8 and Vβ10 rearrangements to either Jβ1 or Jβ2 were dramatically reduced in RAG1c/c DN3 thymocytes (Fig. 3 B). Likewise, 5′Dβ RS ends were also reduced in RAG1c/c DN3 thymocytes, consistent with reduced RAG-mediated cleavage during Vβ to DJβ rearrangement (Fig. 3 C). Conversely, there was little or no reduction in Vβ8 or Vβ10 to Jβ2 rearrangements in RAG1c/c DP thymoyctes (Fig. S2, A and B), as expected given selection for at least one (productive) Vβ(D)Jβ rearrangement in each DP cell. Therefore, the ability of core RAG1 to mediate TCRβ rearrangements in vivo appears compromised for both the Dβ to Jβ and Vβ to DJβ steps. In WT mice, TCRα genes rearrange on both alleles during the DP T cell stage; thus, all mature T cells will have deleted the sequences located between the 3′ most Vα and 5′ most Jα segments (42). Southern blot analysis using a probe from between Vα and Jα gene segments (probe 8) (40) demonstrated that both TCRα alleles rearrange to completion in mature T cells of RAG1c/c and WT mice (Fig. 3 D). Although this finding does not rule out a modest reduction in TCRα rearrangement in RAG1c/c DP cells, it shows that overall TCRα gene rearrangements are not substantially altered.
Rearrangements of IgH and Igκ in RAG1c/c Mice.
We used a PCR approach to analyze DH to JH and VH to DJH rearrangements in sorted CD43+B220int pro–B and CD43loB220+ pre–B cells of RAG1c/c and control mice. Normal B cells usually rearrange DH to JH on both IgH alleles (4, 6). Moreover, all pre–B cells, based on selection for IgH μ chain expression, will have a productive VHDJH rearrangement on one of their two JH alleles (43). Approximately 40% of normal mature B cells have a nonproductive VHDJH rearrangement on their second nonselected allele, with the remaining ∼60% having a DJH rearrangement on the second nonselected allele (4). Therefore, the overall level of VH to DJH rearrangements can be reduced by, at most, ∼30% in pre–B or mature B cell populations. Consequently, PCR methods cannot accurately quantify the effects on VH to DJH rearrangement in these populations. However, PCR can be used to assay for overall reduction in the level of DH to JH recombination, which would be reflected by increased levels of germline alleles.
We employed PCR primers located 5′ of DHQ52 and 3′ of the JH4 gene segment to amplify DHQ52 to JH rearrangements. Assembled DHQ52JH complexes can be excised from the chromosome upon subsequent rearrangement of 5′ DH segments with 3′ JH gene segments (44, 45). Therefore, the use of a 3′ primer downstream of JH4 permits specific amplification of DHQ52 to JH rearrangements retained within the IgH locus on alleles that lack VH to DJH rearrangements. Joining of DHQ52 to JH segments, especially JH4, was reduced in both RAG1c/c pro–B and pre–B cells (Fig. 4 A). Consistent with reduced DJH joining, we observed increased levels of germline JH alleles, most notably in the pre–B cell population (Fig. 4 A). We conclude that RAG1c/c pro–B and pre–B cells have reduced levels of DHQ52 to JH rearrangements. To examine other DH segments, we employed a degenerate 5′ DH RS primer (31) to amplify most DH to JH rearrangements. The level of DH to JH1 and JH4 rearrangements also was reduced in both RAG1c/c pro– and pre–B cells, but DH rearrangements to JH2 and JH3 were present at control levels (Fig. 4 A). Notably, the level of 3′ DFL16 RS ends did not appear to be markedly reduced in RAG1c/c pro–B cells (Fig. 4 D).
We also analyzed rearrangement of three frequently used VH gene segment families (VH7183, VHQ52, and VH558) in RAG1c/c B lineage cells. We employed a 3′ JH4 primer along with 5′ primers specific to these VH families to amplify family-specific VHDJH rearrangements (31). Rearrangement of all three VH families were reduced in RAG1c/c pro–B cells (Fig. 4 B). However, in RAG1c/c pre–B cells, rearrangements of these same VH families were only modestly reduced (Fig. 4 B). As outlined above, this is not surprising because all pre–B cells will have at least one productive VHDJH rearrangement. Overall utilization of tested VH families in pro– or pre–B cells was not dramatically altered (Fig. 4 B), indicating that, at least at a gross level, usage of endogenous VH families by core RAG1 is not obviously influenced by RS sequence [RSS] or VH chromosomal location. We also detected a reduction in the level of 5′DFL16 and 5′DQ52 RS ends in the DNA of pro–B cells, indicating a reduced capacity of core RAG1 to mediate VH to DJH RS cleavage compared with full-length RAG1 (Fig. 4 D). Therefore, RAG1c/c pro–B cells have reduced overall levels of VHDJH rearrangements but relative utilization of different VH families is not markedly altered.
To analyze Igκ rearrangements, we performed PCR analyses on DNA isolated from sorted pro– and pre–B cells using a Vκ-specific primer in conjunction with a primer located downstream of Jκ4. We detected equivalent levels of Vκ to Jκ rearrangements in pre–B cells of RAG1c/c and control mice, indicating that the noncore regions of RAG1 are not required for accumulation of normal levels of Vκ to Jκ rearrangements (Fig. 4 C). We also found an equivalent level of 5′ Jκ1 RS ends in pre–B cells of RAG1c/c and control mice, consistent with the finding that Vκ to Jκ rearrangements are not substantially diminished (Fig. 4 D).
Mature B Cells with Germline JH Alleles in RAG1c/c Mice.
Normally, DH to JH rearrangement occurs on both IgH alleles in developing pro–B cells before the onset of VH to DJH rearrangement (6). However, our PCR analyses suggested possible retention of germline JH alleles in RAG1c/c pre–B cells. To confirm this, we performed Southern blot analyses on genomic DNA from purified CD19+ splenic B cells of RAG1c/c and control mice using a JH-specific probe. Whereas RAG1+/+ B cells had, as expected, little or no unrearranged JH loci, RAG1c/c B cells contained a substantial level of the germline-sized JH fragment (Fig. 5 and not depicted). To more precisely quantify the reduction in JH rearrangements on the nonproductive allele in RAG1c/c B cells, we analyzed JH rearrangements in B cell hybridomas. Previous analyses have shown that a low percentage of hybridomas from WT mice retain germline JH alleles (4, 30), despite the fact that essentially no germline JH locus is detectable in purified splenic B cells (4, 46). These findings suggested that WT hybridomas with germline JH alleles most likely result from tripartite fusions involving non–B lineage cells. Consistent with prior studies (30), Southern blot analyses revealed that most control hybridomas had rearrangements of both JH alleles; although a few retained a germline-sized band (Table I). In dramatic contrast, 59% of the RAG1c/c B cell hybridomas contained a germline JH allele (Fig. 5, A and C; Table I). This very high level of retained germline JH alleles in RAG1c/c B cell hybridomas confirms the findings with purified CD19+ B cells (Fig. 5 and not depicted). Therefore, expression of core RAG1 in the absence of WT RAG1 results in a developmentally altered pattern of IgH locus rearrangement in which VH to DJH rearrangements frequently occur in developing RAG1c/c B lineage cells that have not yet completed DH to JH rearrangement on both alleles.
To determine the level of VHDJH versus DJH rearrangements on the rearranged nonselected JH allele of RAG1c/c B lineage cells, we also quantified the number of DJH and VHDJH rearrangements in RAG1c/c and control B cell hybridomas. Southern blotting with a probe that hybridizes between the VH and DH gene segments demonstrated that 26 out of 42 (62%) and 12 out of 42 (28%) of the second alleles contained, respectively, DJH and VHDJH rearrangements in control B cell hybridomas (Table I). In contrast, whereas the majority of RAG1c/c B cell hybridomas contained one allele in germline configuration, the remaining 23 out of 66 (35%) and 4 out of 66 (6%) of the nonproductive alleles contained, respectively, DJH and VHDJH rearrangements (Fig. 5 and Table I). Therefore, there is a significant reduction in the percentage of RAG1c/c B cells containing VH to DJH rearrangements on the nonproductive allele.
All RAG1c/c B cell hybridomas with germline JH loci also contained hybridizing sequences between the most JH-proximal VH gene segment and the most 5′ DH gene segment (Fig. 5 C and Table I) (30), indicating lack of direct VH to DH rearrangements. In addition, we were unable to amplify direct VH to DH rearrangements in WT or RAG1c/c pro– or pre–B cell populations by PCR using sets of primers designed to amplify rearrangements involving the three VH gene segment families VH558, VH7183, and VHQ52 and either DFL16 or DQ52 gene segments (unpublished data). Similarly, using an upstream VH81X primer in conjunction with the DHR primer that anneals to the 3′ RS sequence of DSP and DFL gene segment families (31), we were unable to amplify direct VH81X to DH rearrangements (unpublished data). Thus, despite the notable reduction in DH to JH rearrangement, none of the RAG1c/c B lymphocytes appeared to have undergone direct VH to DH rearrangements on the nonselected second allele (Fig. 5 and Table I). Therefore, RAG1c/c mice exhibit a marked impairment in both DH to JH and VH to DJH rearrangement.
RAG1c/c Mice Exhibit an Increased Level of DH to JH Inversional Hybrid Joints.
We used a PCR strategy to compare the levels of normal and hybrid inversional DH to JH joints (DJHinv) between RAG1c/c and control mice (Fig. S3). Previous studies have shown that DJHinv joints occur in WT mice at levels ∼3 logs lower than normal DH to JH joints (47). Primary amplifications of DJHinv joints in DNA isolated from pro– and pre–B cells of both WT and RAG1c/c mice revealed an approximate three to fivefold increase in the levels of DJHinv joints detected in the latter (Fig. S3). To confirm that the amplified products observed were in fact the products of DJHinv rearrangements, we cloned and sequenced the junctions of 14 unique WT and 35 unique RAG1c/c DJHinv joints (Fig. S4). Consistent with previous studies (47), we estimate 13 of the 14 DJHinv joints cloned from WT mice and 31 out of 35 from RAG1c/c mice were hybrid joints (HJ), with the remainder likely being DJHinv coding joints (Fig. S4). These HJs form when the 23-bp RS previously associated with a JH gene segment becomes juxtaposed to a DH-coding gene segment originally associated with a 12-bp RS, albeit in an inverted orientation (Fig. S3) (47). The junctions of HJs isolated from both WT and RAG1c/c mice frequently displayed a loss in DH coding sequence, and all but four of the RAG1c/c clones had P or N nucleotide additions (Fig. S4).
Altered V(D)J Recombination in RAG1c/c Mice.
We demonstrate that lymphocyte development and V(D)J recombination is modestly reduced in mice that express the core RAG1 protein in place of the WT RAG1 protein. The occurrence of significant levels of lymphocyte development in RAG1c/c mice is in accord with findings that the RAG1 core protein maintains substantial levels of cleavage activity in vitro. Thus, nearly 40% of the highly conserved RAG1 protein is not required to access RSSs and initiate V(D)J recombination in the context of chromosomal DNA. In addition, noncore regions of RAG-1 are not absolutely necessary for recruitment of factors required to complete the V(D)J recombination reaction in vivo. Moreover, the absolute numbers of prolymphocytes are not as substantially reduced in RAG1c/c mice, as is found in NHEJ-deficient mice, suggesting that DNA DSBs initiated by the core RAG1 protein are properly repaired (48, 49). However, we cannot rule out a modest diminution in joining activities. Overall, RAG1c/c mice clearly exhibit an impairment in both B and T cell development at the pro to prelymphocyte transition, the same stage at which RAG1- (and RAG2-) deficient mice exhibit a complete developmental block (22, 23).
In normal mice, D to J rearrangements take place on both TCRβ or IgH alleles before V to DJ rearrangement (4–6). However, RAG1c/c mice develop mature lymphocytes with germline JH loci on their nonproductive alleles, indicating that the noncore regions of RAG1 are required for fully efficient D to J rearrangement. Retention of germline JH loci occurs despite the fact that DH to JH rearrangement initiates at an earlier developmental stage than VH to DJH rearrangement (6). Expression of core RAG1 in the absence of WT protein also results in a reduced frequency of VHDJH rearrangements on nonselected alleles that contain DJH rearrangements. Therefore, reduced frequency of both DJH and VHDJH rearrangements on the nonselected alleles in RAG1c/c mice likely reflects a decrease in overall V(D)J recombination efficiency. The decreased levels of 3′ Dβ RS ends, which are associated with D to J rearrangements, along with the decreased levels of 5′ DH and 5′ Dβ RS ends, which are associated with V to DJ rearrangements, further supports this conclusion. Of note, previous studies showed that Igκ and TCRα rearrangements were not reduced in RAG2c/c mice (20, 21) which was argued to reflect a correlation between recombination efficiency and specific RS sequence motifs (21). In this regard, we also find that Vκ to Jκ rearrangements are comparable between RAG1c/c and control mice and that the TCRα locus rearranges on both alleles of RAG1c/c T cells.
Although the above findings clearly demonstrate that overall V(D)J recombination is affected in RAG1c/c mice at the IgH and TCRβ loci, we cannot unequivocally rule out the possibility of more subtle effects of truncated core RAG1 on D to J versus V to DJ rearrangements. Notably, expression of core RAG1 in place of full-length RAG1 resulted in the same or even increased production of DJHinv hybrid joints suggesting that noncore regions of RAG1 may preferentially favor this form of hybrid D to J join. In this regard, it is possible that the noncore region of RAG1 may contribute to the proper assembly or stability of synaptic and/or postcleavage complexes, thereby increasing overall recombination efficiency while leading to a decreased frequency of aberrant recombination products. This would be consistent with previous studies showing an increased frequency of hybrid joins and transposition events by core RAGs (17, 50, 51). However, we note that we cannot rule out the possibility that the steady-state increase in DJHinv hybrid joins results from a decreased level of secondary DJH joins that would delete such rearrangements in normal preB cells.
VH to DJH and Vβ to DJβ rearrangements were dramatically reduced in developing lymphocytes of mice that express core RAG2 in place of full-length RAG2 (20, 21). One study concluded that D to J recombination at the IgH and TCRβ loci was unaffected in core RAG2-expressing mice and showed that the reduction in V to DJ rearrangement correlated with decreased RAG cleavage activity at the 5′ D and V RSSs (21). A separate study found a similar effect on V to DJ rearrangements but clearly demonstrated through direct hybridoma analyses that DH to JH rearrangement was also impaired in core RAG2-expressing mice (20). However, VHDJH rearrangements on nonselected alleles were not detected in core RAG2-expressing B cell hybridomas, suggesting that the homozygous core RAG2 mutation affects VH to DJH rearrangement more substantially than DH to JH rearrangement (20, 21). In contrast, we have demonstrated VHDJH rearrangements do occur on nonselected alleles of RAG1c/c B cell hybridomas and find that reduction in D to J and V to DJ rearrangements are accompanied by reduced cleavage at both 5′ and most 3′ D RSSs.
Implications of Altered V(D)JH Rearrangement Patterns in RAG1c/c B Lineage Cells.
Ordered assembly of IgH and TCRβ genes appears to be mediated through developmental stage-specific D versus V segment accessibility (4, 52). Our detection of mature B cells with nonrearranged IgH alleles demonstrates that the expression of core RAG1 in place of full-length RAG1 allows developing B cells to progress to a stage in which VH to DJH rearrangement can take place in the absence of DJH rearrangements on both alleles. Thus, these studies indicate that there is no mechanistic requirement for DH to JH rearrangement on both alleles for progression to a developmental stage in which VH gene segments become accessible. However, we did not detect direct VH to DH joins in the absence of DJH rearrangements in the RAG1c/c B cells. Although absence of direct VH to DH joins might be explained by recombinational efficiency, it would also be consistent with a mechanistic requirement for a DH to JH deletional rearrangement before a subsequent VH to DJH rearrangement on the same allele. If so, VH to DJH rearrangement may require deletion of sequences between DHQ52 and the JH1 gene segment and/or be specifically targeted only to an assembled DJH complex versus a nonrearranged upstream germline DH segment. Thus, rearrangement may allow establishment of VH gene accessibility, for example, by activating the 5′ DH RSS via the germline DH promoter, or may facilitate the capture or synapsis of VH-associated RS target sequences. In the latter context, it is notable that deletion of the 3′ Dβ RSS allowed direct Vβ to Dβ joining in the absence of Dβ to Jβ joining (53), suggesting that the 3′ Dβ RSS may be responsible for maintaining ordered rearrangement, perhaps by providing a high affinity RAG binding site.
Putative Functions of Noncore RAG1 Regions.
The NH2 terminus of RAG1 contains a zinc finger ring domain involved in homodimerization (54) and cysteine-containing elements with suggested multimerization functions (19). The ring finger domain has E3 ubiquitin ligase activity, although a physiological function has yet to be linked to this activity (55). In addition, three basic regions within the NH2 terminus of RAG1 associate with SRP1 (56), a protein known to interact with nuclear localization sequences and to be involved in facilitating nuclear transport (57). Although mutations in these basic regions affect binding to SRP1, nuclear localization is not dramatically altered because there are additional nuclear localization sequences within the RAG1 core region (56). In this regard, we show that core RAG1 is expressed at somewhat higher levels than WT RAG1, with the majority of protein being located in the nucleus. Elevated expression of core RAG1 compared with full-length RAG1 probably reflects the deletion of sequences regulating its degradation (13, 16, 19).
The observation that both DJH and VHDJH rearrangements are affected in RAG1c/c mice suggests the core RAG1 protein simply may be inefficient at mediating V(D)J recombination. Likewise, our PCR analyses indicate that the relative efficiencies of certain Vβ to DJβ rearrangements in RAG1c/c DP T cells is very similar to that of core RAG2-expressing T cells (21). In this context, the modest reduction in peripheral lymphocyte numbers probably reflects a selection and expansion of the few B and T cells that productively rearrange their antigen receptor genes. Although the nature of the V(D)J recombination defect is unclear, one could speculate that multimerization motifs within the noncore regions of RAG1 could be important for synapsis between cleavage complexes and/or other proteins associated with distant RS/coding gene segments, with early disassembly of the cleavage complex occurring in their absence. E3 ubiquitin ligases have been linked to protein degradation, cell cycle control, and DNA repair (58) and that such an activity is contained within the NH2-terminal region of RAG1 raises some intriguing possibilities. For instance, it was shown recently that RAG2 is targeted for degradation by ubiquitination (59). Ubiquitination mediated by RAG1 could ensure that a RAG-associated complex is disassembled and removed from coding gene segments to allow for efficient ligation by NHEJ proteins. Another possibility would be that RAG1-dependent ubiquitination of histones flanking RAG-initiated DSBs may alter chromatin structure to facilitate subsequent repair (55, 60).
RAG Mutations and Immune Deficiencies.
Mutations in both RAG1 and RAG2 can result in partial recombination activity that in humans can lead to T−B− SCID and OS. One group of frameshift mutations identified in OS and OS-like patients leads to alternative ATG usage resulting in NH2-terminally truncated RAG1 proteins (27, 28). In vitro experiments demonstrated that, like the core RAG1 protein, proteins encoded by these frameshift mutant alleles remain partially functional for V(D)J recombination; however, in vivo they lead to dramatically reduced lymphocyte numbers and to immune deficiency (27, 28). Although the NH2-terminal truncation of the murine core RAG1 protein does not fully recapitulate the effects of the frameshift mutant alleles detected in these OS patients, RAG1c/c mice do exhibit some similar phenotypes. For example, germline Dβ-Jβ sequences have been isolated from the T cells of OS patients, demonstrating incomplete TCRβ rearrangement (61). The targeted core RAG1 replacement results in the deletion of a larger stretch of NH2-terminal sequences than the human frameshift mutations. Yet RAG1c/c mice have only slightly reduced numbers of B and T lymphocytes compared with OS patients that have variable T cell numbers but virtually undetectable B cells (26, 62). Many human RAG1 frameshift mutations result in decreased protein expression and cellular mislocalization (27), but core RAG1 in RAG1+/c mice is expressed at levels higher than that of full-length RAG1 protein and is expressed predominantly in the nucleus. Therefore, development of OS in patients with NH2-terminal frameshift mutations may not simply be due to the absence of the noncore RAG1 regions but may also be influenced by reduction in the overall amount or localization of the truncated RAG1 protein. The generation of mice harboring mutations in RAG1 that directly mimic those found in human OS patients should provide insights into the relationship between RAG activity and the development of specific immune deficiencies.
We thank Drs. Martin Gellert and Margie Oettinger for advice and suggestions.
C.H. Bassing and C. Zhu were Associates of the Howard Hughes Medical Institute. J. Sekiguchi is a Special Fellow of the Leukemia and Lymphoma Society. M.J. Sadofsky is a Fellow of the Leukemia and Lymphoma Society. F.W. Alt is an investigator of the Howard Hughes Medical Institute. This work was supported by National Institutes of Health grants AI35714 and AI20047 to F.W. Alt.
The online version of this article contains supplemental material.
Abbreviations used in this paper: CyC, cytochrome C; DN, double negative; DP, double positive; DSB, double strand break; NHEJ, nonhomologous end-joining; OS, Omenn Syndrome; pro, progenitor; RS, recombination signal; RSS, RS sequence.