Heavy Chain Revision in MRL Mice: A Potential Mechanism for the Development of Autoreactive B Cell Precursors¹

Immunoglobulin gene rearrangement is a complex and ordered process that originates with recombination of D to J_H, then V_H to D-J_H genes at the heavy chain locus. Once a functional heavy chain is created, a similar process drives rearrangement at the light chain locus, eventually resulting in production of Abs capable of responding to a diverse number of foreign Ags.

In the autoimmune disease systemic lupus erythematosus, B cells produce Abs reactive to self Ags such as DNA or chromatin (1, 2, 3, 4, 5). The B cells producing these Abs are deleted, anergized, or edited in normal individuals (6, 7, 8, 9). Although somatic mutation and V gene usage are partly accountable for the specificity of these autoreactive Abs, unconventional Ig gene rearrangements at the heavy chain locus such as D-D fusions may also be responsible (10, 11, 12). D-D fusions result from the joining of two heavy chain D genes and create a drastic change in the amino acid sequence of the heavy chain complementarity-determining region 3 (CDR3).³ The presence of positively charged amino acids such as arginine increase the affinity for Ab binding to DNA or DNA complexed to nuclear proteins such as histones (4, 10, 13). Although D-D fusions are uncommon in the normal Ab repertoire, these unusual rearrangements have frequently been observed in Abs with antinuclear specificities (12, 14).

Previous results in our laboratory demonstrated that these unusual rearrangements (D-D fusions and D inversions) occur more frequently in the Ab repertoire of newborn autoimmune-prone MRL/MpJ-+/+ (MRL) mice when compared with C3H/HeJ (C3H) normal controls (15). In addition, the autoimmune strain used more frequently upstream D genes and the most D-distal J_H genes. In comparison, the nonautoimmune C3H mice tended to use the most 3′ D gene, DQ52, and the most D-proximal J_H gene, J_H 1. This suggests that MRL mice may have undergone secondary gene rearrangements that delete evidence of a primary rearrangement. Thus, the MRL strain may be prone to generate secondary gene rearrangements at the heavy chain locus that are more likely to include atypical junctions (15).

In this study, we wanted to determine whether similar rearrangement patterns and atypical junctions are also present in the MRL adult pre-B cell repertoire. Thus, we analyzed the heavy chain gene rearrangements in B220⁺IgM⁻ cells in both MRL and C3H mice. Again, the MRL strain demonstrated an increased frequency of unconventional Ig heavy chain rearrangements when compared with C3H mice. However, the pattern of D and J_H use was different in adult pre-B cells compared with newborns. Therefore, we propose a model of secondary gene rearrangements at the heavy chain locus in MRL mice, which explains the differences in gene usage between the newborn and adult libraries and the frequent occurrence of atypical rearrangements in MRL mice.

Male and female animals from the lupus-prone MRL/MpJ⁺⁺ (MRL) and the nonautoimmune C3H/HeJ (C3H) strains were obtained from The Jackson Laboratory (Bar Harbor, ME; Ref. 16). C3H was chosen as a control for MRL because the MRL strain is partly derived from C3H and both strains share the same Igh j allotype. The animals were maintained in our facility and sacrificed at 3 mo of age.

Bone marrow cells were obtained as previously described (17, 18). Briefly, a single-cell suspension was obtained by flushing femurs from five mice with ice-cold staining media (deficient RPMI 1640 medium without l-glutamine or phenol red (Cellgro, Herndon, VA) containing 10 mM HEPES, 3% FBS, and 0.1% NaN₃). The cells were then mixed with a 1 ml syringe and treated with 0.165 M NH₄Cl to eliminate erythrocytes. After washing with staining medium, the bone marrow cells were incubated with FITC-RA3.6B2 anti-B220 mAb (Southern Biotechnology Associates, Birmingham, AL) and PE-goat anti-mouse IgM (Jackson ImmunoResearch, West Grove, PA) for 15 min at 4°C, followed by 3 washes in staining media. Flow cytometric analysis and sorting of the B220⁺ IgM⁻ subpopulation was conducted at the Temple University FACS core facility using an Epics Elite equipped with an autocloning attachment. The sorted fractions were reanalyzed to verify that their purity exceeds 95%. We further verified that there was virtually no (<0.34%) light chain expression in the sorted pre-B cell populations (data not shown; Ref. 19).

DNA was extracted from the B220⁺ IgM⁻ bone marrow cells by proteinase K digestion and was then subjected to PCR amplification. The PCR primers and protocols were adapted from previously described methods (20, 21, 22). Amplifications were conducted using nested PCR as follows. In the first PCR, DNA was amplified with a mixed set of antisense J_H primers (15) and either a sense J558 primer (GGGCAAGGCCACATTGACTGTAG) or a sense 7183 primer (GGGCCGATTCACCATCTCCAGAG). In the second PCR, 10 μl of the first PCR product was reamplified with primers that were internal to those used in the first PCR. The internal J_H primers were identical with those previously used (15), while the internal V_H primers were as follows: J558 sense (GTAGACAAATCCTCCAGCACAGC) and 7183 sense (GAGACAATGCCAAGAACACCCTG). All PCR were conducted in a volume of 50 μl containing 1.5 mM MgCl₂, dNTPs (200 μM each), primers (0.4 μM each), and 2 units Taq DNA polymerase. Cycling conditions were as follows: an initial 5-min denaturation at 94°C; 35 cycles (1 min at 94°C, 2 min at 50°C, 1 min at 72°C); and a final 5-min extension at 72°C. After the second (internal) PCR, amplification products were separated on a 1% agarose gel and bands of the appropriate size were isolated with the QIAEX II agarose gel extraction kit (Qiagen, Chatsworth, CA). The purified products were directly ligated into the pGEM-T vector (Promega, Madison, WI) and transformed into JM109 cells. After blue/white selection, positive colonies were grown in 5 ml Luria-Bertani/Amp for plasmid purification. The inserts were sequenced using the fmol DNA sequencing system (Promega) with a ³²P-labeled primer complementarity to the T7 promoter (GTAATACGACTCACTATAGGGC). Virtually all plasmids contain a V_H-D-J_H rearrangement and their sequences were analyzed using the GCG program by comparison with known D and J_H germline sequences (23, 24, 25). A minimum of four contiguous identical nucleotides was required for assignment to a germline D sequence.

All analyses were conducted with the Prism software (Release 2.01; GraphPad, San Diego, CA).

B220⁺IgM⁻ pre-B cells were sorted from MRL and C3H bone marrow. Ig gene rearrangements were amplified using a nested PCR with primers specific for the 7183 and J558 V_H gene families and for each of the four J_H genes. A total of 180 MRL and 153 C3H V_H-D-J_H clones were sequenced and analyzed for this study. There was no significant difference in the frequency of productive vs nonproductive rearrangements between both strains (73% productive rearrangements in MRL mice and 67% in C3H mice). These sequences are available from GenBank under accession numbers AF265709 to AF266041.

Conventional heavy chain Ab gene rearrangements result from the combining of single V_H, D, and J_H genes in the appropriate orientation, whereas we define atypical rearrangements as D-D fusions and D inversions. D-D fusions are possible due to an cryptic heptamer embedded within most D genes that simulates a 23-bp recombination signal sequence (26). Further, D inversions result from the fact that D genes are flanked by symmetric recombination signal sequences with a 12-bp spacer that allows recombination in either orientation.

As previously observed in a newborn V_H-D-J_H rearrangement library (15), MRL mice displayed more atypical rearrangements than their Igh allotype-matched C3H controls (48/180 for MRL vs 25/153 for C3H) (p = 0.016 using the Fisher’s exact test; Table I). This difference was mostly due to D-D fusions in that MRL mice exhibited 44 D-D fusions, whereas C3H mice had only 15 D-D fusions. The increased frequency of atypical junctions in MRL mice was observed for both productive and nonproductive rearrangements (Table I).

The D genes at the heavy chain locus can be divided into four gene families. The DFL16 family is comprised of two D genes, DFL16.1 and DFL16.2, which are V_H proximal. The next D family, DSP2, is the largest and is composed of 10 known members, including DSP 2.x, which we have observed to be frequently used in MRL and C3H mice (15). The last two families are composed of single genes, DST4 and DQ52. The DQ52 gene is the most J_H-proximal D gene, mapping only 700 bp upstream of J_H 1.

In a previous study of newborn MRL and C3H mice, we observed a clear difference in the pattern of D gene usage (15). Although C3H mice tended to use DQ52 in the majority of their heavy chain rearrangements, MRL mice used most often members of the more upstream D gene family, DSP2 (15). In the present study, there was also a significant difference in the overall distribution of D genes used between the MRL and C3H pre B cells (Table II), although the difference was less pronounced than in the newborn repertoire. Most notably, MRL mice used the J_H-distal DFL16 gene family more often than their C3H counterparts. The pattern of D gene usage was similar for productive and nonproductive rearrangements in MRL and C3H mice (Table II).

As in the newborns, both strains again frequently recombined a particular DSP family member, DSP 2.x, in most of their normal and atypical heavy chain gene rearrangements. MRL mice used this particular D segment in 16% of both types of rearrangements, whereas the C3H mice recombined this particular gene in 23% of conventional rearrangements and 28% of their atypical rearrangements (data not shown). The predominance of the DSP 2.x gene is interesting in that it has frequently been observed in Abs with autoreactive specificities (27, 28, 29, 30, 31). The repeated usage of DSP 2.x may be a property of the Igh j heavy chain allotype shared between the strains. DSP2.x may be unique to this allotype or it may possess distinct recombination signal sequences or regulatory elements that favor recombination to this particular gene.

Productively rearranged heavy chain D genes can be read in three different reading frames (RF; Ref. 32, 33). The majority of rearrangements use RF1 which encodes a glycine-tyrosine-rich neutral amino acid sequence. Often, RF2 usage results in the expression of a truncated Dμ protein that is selected against, whereas RF3 frequently contains a premature stop codon (34). As in the newborn library, both MRL and C3H mice favored RF1 and there was no significant difference between the two strains (Table III).

There was also an overall significant difference in J_H gene usage in the MRL and C3H libraries (Table IV). Most notably, both strains overused J_H 3, a bias that was not observed in the newborn library (15). The J_H 3 over-representation could be due to the fact that different primers were used in the present study and that certain primer pairs may favor amplification over others. This assay bias, although theoretically possible, is unlikely in that the adult library was created using the same primers for the J_H genes and only those used to amplify V_H gene families were changed. Biased J_H 3 usage has also been observed in other repertoire studies in autoimmune and nonautoimmune mouse strains, although it is yet unclear why the over-representation was detected (35, 36).

The junctional diversity of Ab genes is enhanced through the addition of N and P nucleotides. N nucleotides are added by TdT, which is absent in newborn mice and up-regulated in the adult (20, 37). As expected, almost all of the adult pre-B cell sequences contained N nucleotide additions (Table V), whereas newborn MRL and C3H mice showed a limited number of N nucleotides (15). Further, there was no significant difference in the amount of N nucleotide additions between the two strains. MRL mice possessed an average of 5.71 (± 3.54) N nucleotides per CDR3, and C3H mice had an average of 5.34 (±3.31) N nucleotides per CDR3. P nucleotides are presumably created by uneven cutting of DNA that resolves hairpins created by intermediate coding joints of the Ab genes (38). As with N nucleotide additions, there was no significant difference between MRL and C3H adult pre-B cells in the frequency of P nucleotide additions (Table V).

The deduced amino acid sequences for the productive MRL and C3H V_H-D-J_H rearrangements are listed in Fig. 1. The heavy chain CDR3 is critical for interaction with Ag. In particular, specific residues such as arginine have been linked to mediating CDR3 binding of Abs to DNA or DNA/histone complexes (4, 10, 29). Overall, the CDR3 sequences deduced from productive pre-B cell rearrangements displayed similar properties for both strains (Table V, Fig. 1). For instance, there was no significant difference in length between the MRL and C3H CDR3 sequences (12.14 ± 2.562 amino acids for MRL and 11.29 ± 2.768 for C3H) and their average arginine contents were similar (0.37 ± 0.62 arginine per CDR3 for MRL and 0.31 ± 0.51 arginine per CDR3 for C3H). Taken together, these data suggest that at least some of the events leading to the presence of antichromatin CDR3 sequences in MRL mice occur later than the pre-B cell stage.

The development of systemic lupus erythematosus is characterized by the presence of antinuclear Abs capable of binding DNA and nucleosomes (1, 2, 3, 4). These Abs differ from conventional Abs in that they frequently possess positively charged amino acids such as arginine in the CDR3 of their heavy chain (4, 10). Atypical V_H-D-J_H rearrangements such as D-D fusions and D inversions favor the presence of these residues allowing binding to nuclear epitopes (10). Previous results in our laboratory revealed that in the lupus-prone MRL mouse strain, an increased number of B cell precursors with atypical V_H-D-J_H rearrangements may favor the production of antinuclear Abs (15). The analysis of PCR-amplified heavy chain rearrangements from newborn mice showed that this strain possesses more unusual rearrangements of D-D fusions and D inversions than C3H controls (15).

In this study, we wanted to evaluate the V_H-D-J_H rearrangements amplified from adult MRL and C3H B220⁺IgM⁻ pre-B cells which have not yet undergone Ag selection. Our current results confirm our earlier study (15) and suggest that MRL mice indeed exhibit an intrinsic defect that leads to the development of B cell precursors with atypical Ig heavy chain gene rearrangements. It is interesting that most of the difference between the strains is due to D-D fusions in that MRL mice had 44 D-D fusions compared with 15 in C3H. This observation is revealing in that while D inversions are uncommon, they are “conventional” in that they follow established recombination rules. However, D-D fusions may be more predictive of an inherent recombination defect because of the unconventional nature of this particular rearrangement which breaks the 12/23 rule.

When comparing the current results to our previous report (15), we observed that the bias in upstream D and downstream J_H gene usage previously seen in newborn MRL rearrangements was not present in the adult libraries. A significant difference is that the neonatal libraries were unselected and contained a large proportion of mature B cells, whereas the adult libraries were restricted to pre-B cells. Therefore, the difference in D and J_H usage pattern between the newborn and adult libraries in MRL mice may be related to the stage of B cell maturation. Our results with the newborn library led us to propose that MRL mice may undergo more secondary D-J_H rearrangements because they typically recombine more V_H-proximal D genes to more distal J_H genes when compared with C3H controls. Because of this biased gene usage and the presence of atypical rearrangements in MRL mice, we hypothesized that the mechanisms of atypical heavy chain gene rearrangement and secondary gene rearrangements were related. In a conventional secondary rearrangement, an upstream D gene recombines to a downstream J_H gene, thus eliminating the intervening DNA sequence and any evidence of a primary rearrangement. In a D-D fusion, an upstream D gene combines with a preformed D-J_H rearrangement forming a D-DJ_H complex. However, although our MRL pre-B cells contained a significant number of D-D fusions, we did not observe the strong bias in D and J_H gene usage seen in the newborn library (15). This suggests that during autoimmunity D-D fusions may also arise from mechanisms other than incomplete secondary rearrangements. For example, V_H genes could directly recombine with downstream D genes in accordance with the 12/23 rule. This V_H-D product may then rearrange to a downstream D-J_H complex resulting in a heavy chain with a D-D fusion (26). It has also been suggested that TdT may play a role in both the production of anti-DNA Abs and Ig gene usage. TdT-mediated N nucleotide additions increase the length of the CDR3 thus increasing the potential to generate arginine residues in the Ag binding site (39). C57BL/6-Fas(lpr) mice deficient in TdT had significantly fewer arginines in their CDR3 and a lower frequency of anti-DNA Abs (39). Further, Tuaillon and Capra (40) recently demonstrated that the presence or absence of TdT can modulate the choice of V_H, D, and J_H use independent of B cell Ag stimulation. Nevertheless, the mechanism by which TdT could affect Ig gene choice during rearrangement remains unknown and our data do not indicate a quantitative difference in TdT activity between MRL and C3H mice.

The analysis of the V_H-D-J_H libraries from newborn (15) and adult pre-B cells (this study) leads us to propose a model of D-J_H revision in the periphery of autoimmune mice (Fig. 2). After D to J_H rearrangements on both alleles at the pro-B cell stage, the first allele undergoes a V_H to DJ_H rearrangement (Fig. 2, A and B). If the rearrangement on the first heavy chain allele is productive, the B cell will proceed through normal maturation stages. However, subsequent D to J_H rearrangements may continue on the other allele, resulting in a biased usage of upstream D and downstream J_H sequences or in D-D fusions (Fig. 2,B; Ref. 41). Later in B cell life, the productive rearrangement on the first allele is inactivated (for instance by a nonproductive V_H replacement), but the B cell may be rescued by a productive V_H to DJ_H (or V_H to D-DJ_H) rearrangement on the second allele (Fig. 2,D; Ref. 42). The B cell now possesses a new productive rearrangement on the second allele that will be biased in its D and J_H usage and may also contain D-D fusions that will predispose the B cell to react with DNA or DNA-protein complexes (Fig. 2 E).

This model obviously applies only to situations where the first V_H-D-J_H rearrangement is productive. This situation occurs often enough to make our model meaningful and biologically relevant. The model also proposes that B cell development in autoimmune mice will manifest some unique properties. The first is that D to J_H rearrangements will be ongoing on the second allele, even after a successful V_H to D-J_H rearrangement on the first allele. This is not unlikely, especially when one considers that secondary D to J_H rearrangement is mechanistically similar to a secondary Vk to Jk rearrangement (8, 9). Secondary light chain rearrangements occur routinely during the process of receptor editing, and recent work suggests that both central and peripheral editing can occur more frequently in lupus patients (43, 44, 45). Further, D-J_H replacements were observed in an Abelson murine leukemia virus-transformed B cell line providing evidence that an initial D-J_H recombination does not prevent secondary recombinations (41).

Another critical element of our model is the rescue of a doomed B cell by a V to D-J_H rearrangement on the second heavy chain allele. In our model, the B cell should be doomed because its initial V_H-D-J_H rearrangement has become inactivated. A likely reason for this event could be a stop codon resulting from the somatic hypermutation process, but this may also result from an additional rearrangement event within the V_H-D-J_H region. Most V_H genes contain embedded heptamers that may serve to mediate V(D)J recombination-type events (46). When the heptamer located in the 3′ region of the V_H segment is used, the rearrangement may be functional, leading to a so-called V_H replacement, but rearrangements at other heptamers will be nonfunctional and remain undetected because they result in the death of the B cell (47, 48). Studies with transgenic mice as well as recent human work suggest that such recombination within pre-existing V_H-D-J_H rearrangements are more common than previously thought (49, 50, 51, 52). RAG1 and RAG2 are likely to be required for these types of V_H replacements. This is supported by studies showing that these enzymes can be expressed in the periphery although it is unclear whether this is due to re-activation or continuous expression in B cells that have left the bone marrow only recently (53, 54, 55, 56). Irrespective of the mechanism, the involvement of RAG1 and RAG2 in the recombination on the first allele would indicate that they are also available to mediate the recombination on the second allele.

Another critical factor is the existence of apoptosis defects during lupus. Although no single consistent defect has been identified, there is evidence that apoptosis is impaired in lupus individuals (57, 58, 59, 60). This decrease in apoptosis efficiency may be critical in this situation because it would allow enough survival time for the productive rearrangement to take place on the second allele, whereas a normal B cell would otherwise die before being rescued. An additional important element for autoimmunity is the timing of this secondary rearrangement. As mentioned above, the rearrangements on the second allele will be biased toward the use of certain D and J_H genes. This is of particular relevance during systemic autoimmunity where self-reactive Abs often express the downstream J_H 4 gene, an observation consistent with our hypothesis that successive rearrangements may have taken place on the second allele (61, 62, 63, 64). Further, these secondary rearrangements may include D-D fusions that favor reactivity with chromatin Ags. Because these “rescue-type” rearrangements will take place in the periphery, they will not be subject to the rigorous mechanisms of central tolerance and they may even occur within a context of B cell activation, resulting in the expansion of potentially self-reactive clones.

We are grateful to Dr. Martin Weigert for his comments on the manuscript, to Dr. Reiner Class for help with the flow cytometry, and to Mike Fowler for his technical assistance.