V(D)J Recombination in Mature B Cells: A Mechanism for Altering Antibody Responses

See allHide authors and affiliations

Science  10 Oct 1997:
Vol. 278, Issue 5336, pp. 298-301
DOI: 10.1126/science.278.5336.298


The clonal selection theory states that B lymphocytes producing high-affinity immunoglobulins are selected from a pool of cells undergoing antibody gene mutation. Somatic hypermutation is a well-documented mechanism for achieving diversification of immune responses in mature B cells. Antibody genes were also found to be modified in such cells in germinal centers by recombination of the variable (V), diversity (D), and joining (J) segments. The ability to alter immunoglobulin expression by V(D)J recombination in the selective environment of the germinal center may be an additional mechanism for inactivation or diversification of immune responses.

One of the key features of adaptive immunity is the ability to respond to an enormous number of different antigens. To account for this diversity the clonal selection theory proposes selective expansion of cells with antibody gene mutations during specific immune responses (1). Antibody diversity is initially generated in the bone marrow by a lymphocyte-specific V(D)J recombinase that assembles antigen receptor genes (2). Each B cell has only one antibody receptor, a phenomenon known as allelic exclusion (3). Exclusion is essential for clonal selection and is established by a feedback mechanism from the membrane-bound B cell receptor (BCR) that extinguishes the expression of the recombinase-activating genes (RAG1 and RAG2) (4). Self-reactive BCRs generated during this random gene recombination process can be eliminated in the bone marrow by continued recombination (receptor editing) or by deletion (5). B cells leaving the bone marrow are hypothesized to have fixed receptors that can only be altered by somatic hypermutation, a process that would maintain allelic exclusion (6). Thus, somatic hypermutation is thought to be the mechanism of antibody gene mutation predicted in the clonal selection theory (1).

RAG1 and RAG2 are the recombination signal sequence (RSS)–specific endonucleases that activate V(D)J recombination (7). In addition to being found in developing lymphocytes, RAGs are transcribed in germinal centers (8), which are the foci of hypermutation (9), switch recombination (10), and B cell clonal expansion in response to antigen (11). The finding that RAGs are expressed in germinal centers suggested that they might mediate antigen receptor diversification in mature B cells that are responding to antigenic stimulation. However, expression of the RAG genes does not necessarily translate into immune receptor gene recombination:RAG1 is transcribed in the brain, with no known function to date (12), and RAG1 and RAG2coexpression does not result in antibody gene assembly in T cells or T cell receptor gene recombination in B cells.

To determine whether RAGs induced in mature B cells were active, we assayed for de novo, RAG-specific DNA double-stranded breaks by the ligation-mediated polymerase chain reaction (LM-PCR) (13,14). There are few potential RSS targets for the RAGs in mature B cells because they have extensive V(D)J gene rearrangements on both alleles of heavy and light chain loci. To increase the sensitivity of the assay, we used mice with targeted VB1-8DJH2 and V3-83Jκ2 replacement alleles [μi/+κi/+ mice (where the superscript i stands for inactivated); Fig.1A] (15, 16). The rearranged receptors in these mice allelically exclude the endogenous heavy and light chain alleles, which should therefore remain unrecombined (15, 16) and available for cleavage by RAGs (17).

Figure 1

Jκ and JH signal breaks in B cells stimulated with LPS and IL-4. (A) Diagrammatic representation of the κ (V3-83) and μ (VB1-8) Ig replacement loci. (B) LM-PCR used to detect signal breaks was performed on high molecular weight DNA (27) isolated from cultured cells at the indicated times, according to (13). The assay detects Jκ2 and JH2-JH3 signal breaks as indicated. Briefly, linker was added to 3 to 5 μg of DNA and ligated for 24 to 36 hours. PCR products were run on a 1.8% agarose gel, Southern (DNA) blotted on nylon filters, and hybridized with appropriate kinase-labeled oligonucleotide probes. CD14 is a DNA loading control. PCR primers, cycling conditions, and probe sequences were as previously published (13). Symbols: V, variable region coding exons; D, diversity segments; J, joining segments; C, constant region coding exons; VJκ, rearranged V3-83; VDJH, rearranged VB1-8; d0, day zero; d2, day 2; μ κ, μi/+κi/+ mice; WT, wild-type mice; BM, bone marrow cells from WT mice.

RAG genes were induced in spleen cells from μi/+κi/+ mice and wild-type controls by stimulation with bacterial lipopolysaccharide (LPS) and interleukin-4 (IL-4) (8). As a control, B cells were stimulated with LPS alone, which does not induce the RAGs but does produce B cell activation and proliferation (8). JH and Jκ signal breaks were either absent or difficult to detect in unstimulated cells, whereas both JH and Jκ signal breaks were found in the μi/+κi/+ B cells cultured in LPS and IL-4 for 2 days (Fig. 1B). In contrast, μi/+κi/+ B cells that had been cultured in LPS alone had no signal breaks. Wild-type B cells responded to LPS and IL-4 stimulation somewhat similarly to μi/+κi/+ B cells, though the amount of JH and Jκ signal breaking induced was only 10 to 20% of that found in the μi/+κi/+ mice, as measured by phosphorimager and dilution analysis (Fig. 1B). The low frequency of signal breaks found in the wild type could be explained simply on the basis of reduced availability of RSSs.

In the bone marrow, after RSS-directed, RAG-mediated DNA cleavage, joining reactions lead to antigen receptor gene assembly. We used PCR to determine whether VDJH and VJκ joints accumulated on the allelically excluded alleles in mature μi/+κi/+ B cells. The V3-83Jκ replacement (15) deleted the Jκ1 segment (Fig. 1A); therefore, any joining that uses the Jκ1 segment must take place on the wild-type allele. Few VJκ1 joints were found in wild-type bone marrow and mature B cell controls, and VJκ joints were virtually undetectable in unstimulated μi/+κi/+ B cells (Fig.2A). In contrast, VJκ1 joints were abundant in μi/+κi/+ B cells that had been cultured for 48 hours with LPS and IL-4, but not with LPS alone (Fig. 2A). Heavy chain gene rearrangements were assessed in the same samples with the use of VJ558L primers (18). The VB1-8DJH replacement uses JH2 and deletes JH1, JH3, JH4, and JH5 (Fig. 1A); thus, any joining that involves any of those four JH segments must be on the wild-type allele (16). VJ558LDJH1 joints were found in the wild-type B cells but were virtually undetectable in μi/+κi/+ B cells (Fig. 2B). The near absence of these joints in the μi/+κi/+ B cells was expected because endogenous VDJH rearrangements are inhibited by the VB1-8DJH replacement (16). In vitro stimulation with LPS and IL-4, but not with LPS alone, resulted in significant accumulation of VJ558LDJH1 joints in the μi/+κi/+ B cells (Fig. 2B). These joints are suggestive of new V(D)J rearrangements at the previously excluded wild-type allele.

Figure 2

VJκ and VDJH rearrangements in B cells stimulated with LPS and IL-4. (A) PCR amplification of VJκ1 rearrangements with a Vκ degenerate primer, which anneals to 80% of all mouse Vκ genes, and a primer for Jκ1 (28). PCR products were analyzed as in Fig. 1B. Primer sequences, PCR conditions, and probes were as described (28). (B) PCR amplification of VJ558L-JH1 rearrangements with primers for the VJ558L family of VH genes and for JH2 (18). PCR products were visualized with an Apa I–Xba I mouse J region probe (18) that detects JH1 and JH2. JH2 amplification was uniform in all samples. Symbols: VJκ1, amplified VJκ1 rearrangements; VDJH1, amplified DNA from the VJ558L V gene family; IVS, amplified DNA from intervening sequences (loading control); other symbols as in Fig. 1.

To determine whether B cells stimulated with LPS and IL-4 change the specific immunoglobulin M (IgM) they express, we stained μi/+κi/+ B cells and controls with the Ac146, an antibody that recognizes the idiotype produced by the VB1-8DJH– V3-83Jκ replacement combination (19). As expected, 90% of the unstimulated B220hi μi/+κi/+ lymphocytes were positive for both Ac146 and membrane IgM expression (Ac146+ mIgM+), and only 1% of the cells were Ac146 negative and mIgM positive (Ac146mIgM+) (Fig. 3). After 24 hours in culture with LPS and IL-4 the number of μi/+κi/+ B cells that were idiotype negative had increased from 1 to 10%, and by 48 hours this number had increased to 53% (Fig. 3). In contrast, during the same period of time the percentage of B cells that expressed the idiotype had decreased by half (90 to 45%) (Fig. 3).

Figure 3

Fluorescence-activated cell sorting (FACS) analysis of B cells stimulated with LPS and IL-4. Uninduced cells (d0) or cells cultured with LPS and IL-4 for 1 or 2 days (d1 and d2) were stained with biotinylated Ac146 antibody (15, 19) and with fluoroscein isothiocyanate–labeled antibody to mouse IgM (anti-mouse IgM) (Pharmingen). Lymphocytes were gated on B220hi cells. The numbers in each quadrant represent percentages of gated lymphocytes. All data acquisition and analysis were done on a FACScan with CellQuest software (Becton Dickinson). Symbols: μ κ Ku, μi/+κi/+Ku80−/− mice; Ac146, Ac146 antibody staining; anti-IgM, anti-mouse IgM staining; other symbols as in previous figures.

To verify that the observed conversion was specifically due to the induction of the V(D)J recombinase system, we bred the μi/+κi/+ mice to be homozygous for the Ku80 deletion (μi/+κi/+Ku80−/−). Ku80−/− animals cannot complete V(D)J recombination and therefore have no mature B cells (20). However, breeding of the μi/+κi/+ mice to the Ku80−/−background reconstitutes the mature B cell compartment (21). If the observed increase in Ac146mIgM+ cells was due to V(D)J recombination in mature B cells, stimulation of splenocytes from μi/+κi/+Ku80−/− mice with LPS and IL-4 should not lead to an increase in idiotype-negative cells. The percentage of μi/+κi/+ Ku80−/− B cells that were Ac146mIgM+ was unaltered after 48 hours of culture Fig. 3 (22). Thus, when RAG1 and RAG2 are induced in mature B cells they can cut DNA at RSSs, and the new breaks are associated with accumulation of otherwise allelically excluded gene rearrangements.

Continued RAG expression in developing B cells in the bone marrow can result in receptor editing by deletional replacement of a prerearranged Ig gene (5). To determine whether a prerearranged Ig gene could also be replaced in mature B cells, we bred mice to be homozygous for the V3-83Jκreplacement and heterozygous for the VB1-8DJHreplacement (μi/+κi/i) (15, 16). Because the Jκ3, Jκ4, and Jκ5 alleles remain in the germline configuration, further VJκrecombination in μi/+κi/i B cells could only be accomplished by looping out the Vκ3-83-Jκ2 replacement. Although VJκ4 rearrangements were difficult to detect in maure μi/+κi/i B cells, Jκ4 signal breaks and VJκ4 recombination products were abundant in μi/+κi/i B cells after they were cultured for 2 days in LPS and IL-4 (Fig. 4A). The relative increase in VJκ4 rearrangements was by a factor of 10 to 20, as measured by semiquantitative dilution PCR and phosphorimaging. The limited cell division and cell death in the first 48 hours of culture with LPS and IL-4 made it likely that these VJκ4 joints represent new rearrangements that “edit” out the preexisting receptors (22).

Figure 4

Molecular substitution of the V3-83Jκ2 replacement allele by VJκ4 and signal breaks in vivo. (A) Top, PCR amplification of VJκ4 rearrangements (29). Middle, induction of DNA breaks in splenic μi/iκi/i B cells cultured with LPS and IL-4. LM-PCR was performed as described in Fig. 1B. PCR primers, cycling conditions, and probe sequences were as previously published (13). Bottom, amplification of the CD14 locus is the DNA loading control. (B) Jκ signal breaks induced in GL-7+ B cells in vivo. T cell–depleted, activated (A), and resting (R) B cells were used to make high molecular weight DNA for LM-PCR (see Fig. 1B). CD14 is the DNA loading control. Symbols: μ κ/κ, μi/+κi/i homozygous kappa-inactivated mice; VJκ4, VJκ4 rearrangements; Jκ, Jκ signal breaks; other symbols as in previous figures.

To determine whether RAGs expressed in germinal centers in vivo were active, we examined B cells from μi/+κi/+ mice immunized with trinitrophenol coupled to keyhole limpet hemocyanin (TNP-KLH) for signal breaks (13, 14). Germinal center B cells were enriched by density centrifugation as measured by GL-7 staining. Jκ breaks were found 7 days after immunization in the activated but not in the resting B cell fractions (Fig. 4B), and the breaks were associated with a loss of Ac146 staining. This time course of break induction correlates with the reported expression of RAG1 in germinal centers 5 days after immunization (8, 23). Thus, antigen-driven activation of the RAGs in germinal center B cells was associated with de novo Jκ breaks and perhaps the expression of previously excluded membrane Igs.

Does V(D)J recombination really occur in mature B cells in more physiologic situations? We would predict that light chain editing is not uncommon, because V-Jκ rearrangements could occur on any rearranged allele that still contains Jκ5 or more 5′ Jκ genes. In contrast, heavy chain gene editing may be a less frequent event. Any heavy chain allele that has a functional or nonfunctional VH gene would not be available for standard V(D)J recombination because all DH-associated RSSs would have been deleted. Nevertheless, heavy chain gene replacement using internal cryptic RSSs in VDJH genes has been documented and would remain possible (24).

Active gene rearrangement in mature B cells that express membrane μ suggests that allelic exclusion may be violated in germinal centers and conflicts with the generally accepted idea that membrane μ regulates allelic exclusion (4, 25). However, three mechanisms are likely to reduce the number of “allelically included” B cells that can emerge from the selective environment of the germinal centers: (i) somatic mutation and Ig gene inactivation; (ii) elimination by selection of mixed- or low-affinity receptors (26); and (iii) incompatibility between heavy and light chain pairs. Support for the idea that most replacements would be selected against comes from the immunochemical observation that RAG1 is abundant in clusters of apoptotic B cells in the germinal centers (23).

In conclusion, affinity maturation, which is thought to be driven by somatic mutation, might also involve V(D)J recombination and receptor editing. Edited receptors would usually comprise the original heavy chain with an altered light chain. Because a disproportionately large part of the antigen-binding pocket is frequently contributed by the heavy chain, changes in the light chain might be expected to yield antibodies with various affinities for the original immunogen.

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


Stay Connected to Science

Navigate This Article