Transcription-Dependent Somatic Hypermutation Occurs at Similar Levels on Functional and Nonfunctional Rearranged IgH Alleles¹

Generation of the Ab repertoire in mammalians requires several steps of gene remodeling which finally allow the selection of cells producing a clonally restricted, single sort of Ab. During establishment of the primary repertoire in pro-B cells, DJ rearrangement first occurs on both alleles while by contrast, functional VDJ rearrangement on a single allele later terminates primary recombination of IgH loci. Secondary rearrangements occurring in immature B lymphocytes seem to mostly target the functionally rearranged allele while accessibility of the excluded allele would be down-regulated (1). Evidence has even been provided in the mouse that differential accessibility of both IgH chain loci may be established before entry into the B cell development pathway, as early as day 12 of the embryonic life (2). Differences in the methylation status of alleles have been reported and correlated with asymmetry of both their replication patterns and their nuclear localization (2, 3). This asymmetry likely accounts for the lowered frequency of cells carrying rearrangements on both alleles, which is below what would be expected if any nonfunctional rearrangement on a first allele were indeed followed with a rearrangement attempt on the second allele (4). By contrast to several cytokine genes for which allelic exclusion was clearly demonstrated at the transcriptional level, little is known about a potential asymmetry regarding transcriptional accessibility of Ig alleles (5, 6). That nonfunctionally rearranged V sequences are frequently cloned from genomic libraries but are virtually absent from cDNA libraries or RT-PCR experiments could be taken as an indication of a strong allelic exclusion at the transcriptional level. However, Ig transcripts have been shown to be the targets of mRNA surveillance control pathways that may result in a 4- to 100-fold accelerated degradation of nonfunctional transcripts (7). Because of this efficient nonsense-mediated decay, steady-state levels of mature Ig transcripts cannot provide any evaluation of the transcription rate of the functional vs the nonfunctional allele in a given B cell.

Following Ag encounter, the Ig genes in those B cells triggered by the Ag are subjected to a second wave of diversification by somatic hypermutation (SHM).⁴ This process leads to the introduction of multiple single nucleotide changes in and downstream of the rearranged V(D)J segment, allowing the expression of a secondary repertoire from which those B cells carrying mutated Abs of improved Ag-binding affinity can be selected (reviewed in Ref.8). A first phase of SHM relies on the activation-induced cytidine deaminase (AID) while a second phase may involve mismatch repair enzymes and translesion polymerases, but the molecular mechanisms selectively recruiting this machinery to Ig genes are still unknown (9, 10, 11, 12). In vitro, transcription of DNA substrates allows deoxycytidine deamination by AID (13, 14). In vivo, although not sufficient to ensure mutability, transcription appears as necessary and is directly linked to SHM: the location of the mutation domain is determined by the position of the transcription start site and recruitment of the mutational machinery is affected by the cis-acting promoter and transcriptional enhancer elements (15, 16). In mutable Ig transgenes integrated outside or within the IgH locus, a linear correlation between expression and mutability has been reported (17, 18, 19).

V(D)J rearrangement and presence of a V promoter are known as mandatory for high-level SHM (18, 20). Although unrearranged V segments and incomplete DJ rearrangements are transcribed at low levels in immature B cells, V to DJ approximation boosts transcription by positioning one V_H promoter under the control of the Eμ enhancer (reviewed in Ref.21). Regarding allelic exclusion, the relative transcription rate of Ig loci in wild-type B cells carrying two VDJ rearrangements has not been studied; by contrast, the presence of somatic mutations within out-of-frame rearranged V genes has been documented (22, 23). Since functional and nonfunctional Ig sequences are strongly related and cannot be individually studied by the means of hybridization techniques, we thus decided to use the accumulation of mutations for the quantitative evaluation of the relative transcription rates of both alleles in primary B cells that carry double VDJ rearrangements. Mutations are known to freely accumulate in the framework regions of nonfunctional genes (23), while CDR mutations of the expressed alleles are selected by the Ag-driven cell selection (8). We thus restricted our analysis to intronic sequences located immediately downstream of rearranged J segments, for which the accumulation of mutations only reflected activity of the hypermutation machinery, with no selection for the production of any translated gene product. These experiments were conducted both in wild-type B cells and in B cells carrying an heterozygous targeted disruption that was previously demonstrated to abolish VDJ transcription while preserving all cis-regulatory elements of the IgH locus and still allowing VDJ rearrangement on this nonfunctional allele (24).

C57BL/6 wild-type mice (wt “b” allotype) were mated either with 129/Ola wild-type mice (wt “a” allotype), providing the control wild-type a/b mice. C57BL/6 mice were also mated with the B cell-deficient mouse line J_H-Neo-Eμ, which carries a homozygous disrupted a allotype IgH locus. The J_H-Neo-Eμ mutation features the insertion of a neomycin resistance cassette at the NaeI site immediately downstream of J_H4 (Fig. 1 A). This mutation is known both to preserve the occurrence of V(D)J recombination and to result in a complete transcriptional silencing of rearranged V(D)J segments (25). Heterozygous wt/J_H-Neo-Eμ mice were selected, in which all B cells carried a functional b allotype IgH locus and a nonfunctional a allotype IgH locus.

Peyer patches (PP) were recovered by dissection from the small intestine. Cell suspension was prepared by pressing PP through a nylon mesh and cells were washed three times in cold DMEM containing 10% FCS (Invitrogen, Gaithersburg, MD). Cell suspensions were adjusted at 10⁶ cells/ml after elimination of clumps of dead cells. Cells were incubated with anti-B220-SPRD for 30 min at 4°C. After two washes with PBS containing 10% FCS, FITC-conjugated peanut agglutinin (PNA) was added and incubated on ice for an additional 30 min, before washing and resuspension in medium. Purification of B220⁺PNA^high and B220⁺PNA^low cells was realized on a FACSVantage (BD Biosciences, Franklin Lakes, NJ). The purification rates of these cell populations were constantly above 95% (Fig. 1 B).

Genomic DNA was extracted from PNA^high B cells using the QIAamp Tissue kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. DNA was amplified by PCR in 50 μl of Taq 1× PCR buffer containing 200 μM of each dNTP, 100 ng of each primer, and 1 U Taq polymerase (Pharmacia, Piscataway, NJ), with the following program: 1 cycle at 94°C for 5 min, 35 cycles (94°C for 45 s, 58°C for 45 s, and 72°C for 45 s), and 1 cycle at 72°C for 7 min. The following primers were used: forward V_H7183, 5′-CGGTACCAAGAASAMCCTGTWCCTGCAAATGASC-3′ (complementary to the third framework region of V_H7183 family genes) and various backward primers: 3′J_H2, 5′-CAGTAAGCAAACCAGGCACA-3′, corresponding to an intronic sequence between the J_H2 and J_H3 segments; 3′ NaeI, 5′-TCCTAAAAGGCTCTGAGATCCC-3′, corresponding to an intronic sequence ∼300 bp downstream of J_H4 on a wild-type allele (but 2 kb downstream of J_H4 on a J_H-Neo-Eμ knock-in allele); and N3, 5′-TCTTGACGAGTTCTTCGAGGGGATCGGCA-3′, a primer homologous to the inserted neo gene 3′ end, proximal to the J_H4 element, thus able to specifically amplify those rearrangements occurring on the knock-in allele.

The PCR products were separated on a 1.2% agarose gel and cloned into the pCRII-TOPO vector (Invitrogen) according to the manufacturer’s instructions. Plasmids were isolated using a standard Triton X-100/lysozyme method and inserted sequences were sequenced from both ends using the M13Reverse and M13(−20) forward primers and an automated laser fluorescent DNA ABI PRISM 3100 sequencer (PerkinElmer, Branchburg, NJ). Mutations were analyzed on a limited interval including the rearranged J_H segment and the following 180 bp of intronic sequence, a region known to be strongly targeted by SHM on expressed genes (8, 26). Sequences from wild-type a/b mice and from heterozygous a^wt/b^JH-Neo-Eμ animals were checked for mutation by comparison with the known germline sequences of the J_H region from the a allotype (26) or the b allotype (GenBank accession number AC073553).

PP PNA^high B cells from wt a/b mice were sorted and checked for hypermutation. To appreciate SHM in these cells regardless of their Ag specificity, intronic sequences downstream of rearranged VDJ junctions were amplified using a forward V_H7183 family framework region (FR) 3 consensus primer along with a backward primer located in the intron downstream of J_H2

Sequence analysis focused on the rearranged J_H segment and the following intronic 180-bp sequence, a region known to be strongly targeted by the SHM process (8). This sequencing strategy also avoided ambiguities for the assignment of V sequences to their germline counterparts. Mutations were counted either only in the intronic sequence or in the complete sequence including the J_H-encoded fourth FR (FR4). Sequencing also allowed us to check VDJ junction sequences for functionality and to classify them into two groups: functional sequences featured an in-frame rearrangement with no stop codon created at the VD or DJ junction and nonfunctional sequences either carried out-of-frame junctions or included premature termination codons within the D segment.

Of 50 functional and 23 nonfunctional sequences analyzed from PP PNA^high B cells, somatic mutations were readily observed on both classes of rearranged alleles in wild-type animals (Figs. 2,A and 3). SHM reached the level of 13.1 mutations/1000 bp for functionally rearranged genes and 11.9 mutations/1000 bp for nonfunctionally rearranged alleles (Table I). This difference was not statistically significant. Since our sequencing interval included the J_H region encoding the V domain FR (FR4) for which substitutions might be counterselected in expressed functional genes, we also restricted the comparison to intronic sequences flanking the rearranged J_H. This comparison again indicated similar mutation frequencies of 13.6 and 11.1%, respectively, for the functional and the nonfunctional sequences. In both cases, the frequency was not significantly lower for nonfunctional genes (Student’s t test, p > 0.05).

Although preserving V(D)J recombination to J_H1, J_H2, and J_H3, the J_H-Neo-Eμ mutation blocks transcription of both germline J_H and rearranged DJ_H or VDJ_H segments (24). Although B cell development is arrested in homozygous J_H-Neo-Eμ mutant mice (24), using heterozygous mice with a single knock-in IgH allele allows normal B cell maturation and warrants that all functionally expressed sequences originate from the wild-type allele. Since the J_H-Neo-Eμ knock-in features the insertion of a neomycin resistance cassette at a NaeI site located ∼100 bp downstream of J_H4, it could be used as a “tag” allowing to easily distinguish sequences from the knock-in allele or from the functional allele with a wild-type NaeI site.

As for wild-type mice, PP PNA^high B cells from heterozygous wt/J_H-Neo-Eμ mice were sorted and checked for hypermutation. Although the disrupted IgH locus was from the a allotype, heterozygous mice obtained after breeding with C57/B6 mice provided a wild-type b allotype IgH locus. By force, B cells of such mice all carry a functionally rearranged b allotype J_H region and an allelically excluded a allotype locus. Since a number of B cells carry double complete VDJ rearrangements, it was readily possible from these cells, to amplify rearranged sequences originating from both alleles. To that goal, a unique V_H7183 consensus forward primer was used along with a backward primer either specific for the knock-in neo gene downstream of J_H4 or specific for the wild-type J_H4 3′ flanking intron. In addition, the use of heterozygous IgH a/b allotype mice in which known allelic differences are located in the J_H germline region provided a further control of the origin of cloned sequences. The respective mutation frequencies on both a functional b wt allele and a rearranged but untranscribed a allele carrying the J_H-Neo-Eμ mutation could finally be appreciated in identical conditions and in the very same PNA^high B cell samples (Fig. 2 B).

A complete lack of somatic mutation of rearranged J_H sequences was noticed on the untranscribed neo knock-in allele, reaching 1.1 mutations/1000 bp (i.e., close to the range of mutations introduced by the Taq polymerase enzyme in PCR experiments) instead of a simultaneous 30 mutations/1000 bp on the wild-type b control allele (Table I). The J_H-Neo-Eμ knock-in thus significantly blocked SHM of nonfunctional alleles carrying VDJ rearrangements (Student’s t test, p < 0.02).

In transgenic models, the hypermutation mechanism is known to act only on gene copies that are demethylated as well as transcribed (15, 16, 17, 18, 19, 26). Although endogenous alleles of Ig loci may show some asymmetry with regard to accessibility, DNA methylation, and pattern of early replication, it has not been clearly ascertained whether the processes of primary transcription and of SHM preferentially target the actively expressed allele. Mutations were previously observed in FRs of nonfunctional rearranged Vλ genes but their frequency could not be compared with that of functional genes in which FR mutations are negatively selected (22). By analyzing noncoding sequences which are not subjected to Ag-driven pressure and cell selection, we studied whether the SHM machinery by itself may similarly target sequences appropriately located close to a VDJ junction on both the functional and the allelically excluded alleles of the IgH locus.

The herein reported data show that in a given B cell population, SHM indeed targets both alleles at levels that are roughly similar. These data thus provide indirect but clear evidence that the numerous B cells which spontaneously carry double VDJ rearrangements in wild-type animals also transcribe both alleles at comparable rates in mature B cells, at least in those cells that are undergoing the SHM process. In addition and in agreement with the clear links previously demonstrated by other studies between the rate of transcription and the level of SHM of Ig genes, we show that a mutation blocking V(D)J transcription on the excluded allele also prevents the recruitment of the SHM machinery on that allele. Although a similar reduction in SHM was previously observed upon deletion of the pV_H promoter (18), our data show that not only the presence of the promoter upstream of a rearranged V gene, but also active transcription from this promoter is required for SHM. These results also fall in agreement with recent in vitro observations showing that deamination of cytosines by the AID enzyme needs transcription of its DNA substrate (13, 14).

It may be difficult to reconcile these observations with the previously published data about the early asymmetry of Ig alleles which appears to be established even before commitment to the B cell lineage (2, 3). However despite their asymmetry, κ loci were recently shown to be germline transcribed on both alleles (27). Our data suggest that the asymmetry in replication pattern and nuclear localization that has been observed for IgH alleles is either an inconstant feature of Ig loci or does not result in a significant blockade of transcription of nonfunctionally rearranged alleles. Cells carrying complete VDJ rearrangements of both IgH loci thus show allelic inclusion at the transcriptional level, at least up to the stage of germinal center B cells undergoing SHM. It thus seems that approximation of a V_H promoter to the intronic Eμ enhancer is sufficient for transcription to occur and be maintained, whatever the allele involved, then overriding the potential extinction previously imposed by late replication or nuclear localization allelic asymmetry. Alternatively, it is also conceivable that only a certain proportion of B cells display asymmetric accessibility of Ig genes, while others would undergo symmetric activation of Ig loci. Dual complete VDJ rearrangement would then only occur in those cells that have escaped the rule of allelic asymmetry. Consequently, only the latter cells could have the ability to fully rearrange both IgH loci and to display allelic inclusion at the transcriptional level. However, the observation that knock-in mice carrying two functional VDJ rearrangements express dual BCR on most B cells argues against the idea that allelic asymmetry could definitely retain expression on one allele in a significant proportion of mature or activated B cells (28). Rather, asymmetry could be restricted to the pro-B/pre-B cell stage and only account for the preferential rearrangement of one allele. Both our data and those from dual BCR-expressing mice with a double knock-in of functionally rearranged genes (28) clearly indicate that for IgH genes, double rearrangements result in allelic inclusion at the transcriptional level and that allelic asymmetry is either unable to completely block transcription or that this blockade is relieved by VDJ rearrangement. Importantly and further indicating transcriptional competence of both IgH alleles in mature B cells, germline transcripts from constant genes that precede class switching were also shown to arise from both the expressed and the “allelically excluded” IgH locus (29).

The major conclusion of the present work is that whether they are functional or not, rearranged VDJ genes undergo SHM at roughly similar rates. By contrast, preventing transcription of an allelically excluded VDJ blocks its accessibility to SHM. It can be inferred from these observations that allelic exclusion of IgH expression in normal B cells is not effected at the transcriptional level for the numerous mature cells carrying double VDJ rearrangements. The early asymmetry in the replication pattern or nuclear localization observed for Ig loci could thus have limited effects, acting only by down-regulating V to DJ rearrangement in pro-B cells. Later on and in a number of selected B cells, allelic asymmetry could also be strongly reinforced secondarily to VDJ rearrangement, as a consequence of the asymmetric activation of a single V_H promoter in those cells with a single V to DJ rearrangement. Both our data and those showing allelic inclusion in mice with double knock-in of functional VDJ segments favor the hypothesis that allelic exclusion is mostly effected by the stochastic process of VDJ rearrangement. Since in-frame rearrangements encoding a functional H chain able to pair with its L chain partner constitute rare events, they are highly unlikely to occur on both alleles and to be simultaneously expressed at the protein level. Our findings that nonfunctionally rearranged alleles are not silenced at the level of VDJ transcription further demonstrates that, by contrast to cytokine genes, transcriptional asymmetry plays little role in the allelic exclusion of Ig gene expression. Downstream of transcription, these findings underline the importance of RNA surveillance pathways that degrade nonfunctional mature Ig mRNAs, thus preventing the “wastage” that would result from their translation into truncated Ig chains. They also confirm that in the course of B cell malignancies, both Ig loci may be accessible to mutations, recombinations, and translocations.

We thank Vincent Denis and Claire Carrion for their kind technical assistance.