Requirement of the Activation-Induced Deaminase (AID) Gene for Immunoglobulin Gene Conversion

See allHide authors and affiliations

Science  15 Feb 2002:
Vol. 295, Issue 5558, pp. 1301-1306
DOI: 10.1126/science.1067308


Three phenotypically distinct processes—somatic hypermutation, gene conversion, and switch recombination—remodel the functionally rearranged immunoglobulin (Ig) loci in B cells. Somatic hypermutation and switch recombination have recently been shown to depend on the activation-induced deaminase (AID) gene product. Here, we show that the disruption of the AID gene in the chicken B cell line DT40 completely blocks Ig gene conversion and that this block can be complemented by reintroduction of theAID complementary DNA. This demonstrates that theAID master gene controls all B cell–specific modifications of vertebrate Ig genes.

The formation of a large antigen receptor repertoire by DNA rearrangement and hypermutation is unique to the immune system. Early lymphocytes first assemble their antigen receptor genes from different V, D, and J segments by site-specific V(D)J recombination. B cells then further modify the rearranged V segments by untemplated hypermutation (1) or pseudogene templated gene conversion (2). Some species (such as sheep) exclusively use somatic hypermutation for V segment diversification, whereas others (such as chickens, rabbits, cattle, and pigs) rely predominantly on Ig gene conversion (3). Concomitant with antigen stimulation, the Ig heavy chain locus is further reshuffled by switch recombination to yield different Ig isotypes (4).

It was reported recently that mutations of the so-called AIDgene, which is induced by B cell stimulation and is homologous to the cytidine deaminase APOBEC-1 gene (5), abolish switch recombination and severely reduce somatic hypermutation in mice (6) and humans (7). It is speculated that the AID protein acts by sequence-specific mRNA editing in a manner similar to that of the APOBEC-1 protein. The AID mutant phenotypes raise the question whether AID might regulate Ig gene conversion as well. The best characterized model for Ig gene conversion is the chicken Ig light chain locus, for which all V pseudogenes have been sequenced (2).

To investigate the role of AID in Ig gene conversion, we cloned the chicken AID homolog and disrupted it in the bursal B cell DT40 cell line, where light chain gene conversion is preserved (8). An expressed sequence tag (EST) (riken1_1b7r1) of the likely 5′ end of the chicken AID cDNA was identified in the bursal EST database (9) and extended by primer walk. The corresponding cDNA insert contains an open reading frame of 198 amino acids with over 85% identity to the murine and human AID sequence (Fig. 1A). Expression of this bonafide chicken AID homolog was analyzed by semiquantitative reverse transcription–polymerase chain reaction (RT-PCR) in different cell types (Fig. 1B). AID is highly expressed in the bursa of Fabricius and DT40 cells; weakly expressed in the spleen, thymus, and testis; but no expression is detected in the liver or brain. Because the bursa of Fabricius is the main site of Ig gene conversion, this expression pattern is compatible with a role forAID in this process. Two AID knockout constructs (pAidBsr and pAidPuro) were made by cloning genomic fragments from the 5′ end and the 3′ untranslated region of the AIDlocus upstream and downstream of loxP-flanked drug-resistance markers (Fig. 2A). Targeted integration of these constructs deletes the AID coding region from the third codon to the end, resulting in an AID null mutation.

Figure 1

Sequence conservation and expression of the chicken AID gene. (A) Alignment of the chicken (Gd), mouse (Mm) (NM_009645), and human (Hs) (NM_020661)AID amino acid sequences. The conserved cytidine deaminase motif is marked by a line. Asterisks denote conserved leucine residues in the COOH-terminal region. (B)AID expression analyzed by semiquantitative RT-PCR.AID specific transcripts were amplified as described in (17, 18) with dilutions of total cDNA as indicated on top of the lanes. Amplifications of β-actin transcripts served as controls (18).

Figure 2

Disruption of the chicken AID gene (20, 21). (A) A physical map of theAID locus, the two knockout constructs, and the targeted loci. Open boxes represent transcribed 5′ and 3′ untranslated regions. Solid boxes represent coding regions. The positions of primers for RT-PCR and the identification of targeted integration events are indicated by arrowheads. (B) Southern blot analysis of wild-type and transfected clones using theprobe shown in (A) after Xba I digestion. The wild-type locus hybridizes as a 6-kb fragment, whereas the targeted locus hybridizes as a 4.5-kb fragment and a 4-kb fragment before and after removal of the drug resistance marker cassettes, respectively. (C) AID expression. Transcripts of AID were amplified by RT-PCR (17,18) to verify the gene disruptions and the complementation.

Ig gene conversion rates can be measured easily by quantifying cell surface(s) IgM reversion rates in the predominately sIgM(–) CL18 cell line (8). This spontaneous DT40 variant has a frameshift in its rearranged V segment, which can be repaired by homologous pseudogene sequences (Fig. 3A). Because CL18 shows rather low light chain conversion frequencies, a v-myb transfectant of CL18, which has five times higher rates of sIgM reversion (10), was used. This clone, DT40Cre1, also contains a transfected MerCreMer recombinase gene, which can be induced to excise loxP-flanked cassettes (11). After transfection of pAidBsr into DT40Cre1, a heterozygous AIDclone was identified (DT40AID+/−) which was subsequently transfected by pAidPuro to produce a homozygous AID knockout clone (DT40AID−/−) (Fig. 2A). Both the blasticidin and the puromycin resistance markers were then removed from DT40AID−/− by tamoxifen induction of the MerCreMer recombinase (12), yielding the clone DT40AID–/–E. The disruption of AID in DT40AID+/− and DT40AID−/−, and the excision of the resistance marker cassettes in DT40AID–/–E, were confirmed by Southern blot analysis (Fig. 2, A and B). To reconstitute the AID disruption, DT40AID–/–E was transfected by an AID expression vector in which theAID cDNA and the puromycin gene were under the control of the β-actin promoter and were flanked by loxP sites. One of eight stable puromycin resistant transfectants (DT40AID–/–R) expressed the AID cDNA (Fig. 2C).

Figure 3

Ig gene conversion quantified by sIgM expression. (A) Principle of the Ig reversion assay. The predominantly sIgM(–) DT40Cre1 clone contains a frameshift in its rearranged V segments, which can be repaired by pseudogene templated conversion events. The rate of Ig gene conversion can be measured in subclones by FACS analysis after staining for sIgM. (B) FACS profiles of representative subclones (22). (C) Average percentages of events falling into the sIgM(+) gates. Forty-eight subclones of each clone were analyzed (22). The measurements for individual subclones are shown in Table 1. The percentages in brackets are the values after subtraction of the background. (D) Sequences of the CDR1 regions from the rearranged V segments of DT40 Cre1 and AID–/–E clones 6 weeks after subcloning (23). Dashes indicate nucleotides identical to the sequence of DT40 Cre1 sIgM(–), and brackets mark the borders of deletions. Gene conversion tracts are underlined and annotated with the name of the likely pseudogene donor.

To quantify the Ig gene conversion rates, 48 subclones of the wild-type positive control and the AID mutant clones were analyzed for sIgM expression by fluorescence-activated cell sorting (FACS) after 18 days of being cultured. Subclones of the stable sIgM(–) light chain knockout DT40IgL were included as negative controls. Representative FACS anti-IgM stainings of subclones are shown in Fig. 3B, and the average number of events in the positive gate from the 48 subclones is displayed in Fig. 3C (the individual clone measurements are in Table 1). This analysis indicates that the disruption of a single AIDallele reduces the sIgM reversion rate of DT40AID+/− by more than 10 times, although the reversion is not completely blocked, because at least 2 of the 48 DT40AID+/− subclones showed distinct sIgM(+) populations. The disruption of both AIDalleles in DT40AID–/–E reduces the reversion frequency by at least 100 times. Reexpression of the AID gene in DT40AID–/–R increases the frequency of sIgM reversion above that measured for the parental DT40Cre1 clone. This strongly suggests that AID expression is essential for Ig gene conversion. To increase the sensitivity of the assay, we cultured 24 DT40AID–/–E subclones for 6 weeks in parallel with DT40Cre1 subclones. Whereas, on average, more than 60% of DT40Cre1 cells scored sIgM(+), no sIgM(+) populations were detectable in the DT40AID–/–E subclones.

Table 1

Percentages of events falling into the sIgM(+) gates for individual subclones.

View this table:

Oligonucleotide primer sequences.

View this table:

There remained the possibility that the lack of sIgM(+) cells in theAID negative clones was not due to a block of Ig gene conversion but was caused by a lack of Ig light or heavy chain transcription. We therefore confirmed by RT-PCR that the light and heavy chain Ig mRNA was present in the DT40AID−/− cells (10). To rule out the possibility that the lack ofAID expression indirectly affects sIgM expression, we removed the loxP-flanked AID expression cassette from a predominantly sIgM(+) DT40AID−/−R subclone by tamoxifen induction. FACS analysis confirmed that the level of sIgM expression in the resulting AID −/− subclones was identical to the level of sIgM in wild-type DT40Cre1 sIgM(+) cells. Together, these results indicate that AID −/− sIgM(+) cells are viable and that the lack of sIgM reversion inAID-negative clones is not caused by a defect in sIgM expression but by inhibition of Ig gene conversion. To confirm by sequence analysis that AID disruption prevents gene conversion, we compared the light chain genes of DT40Cre1 and AID−/−E clones 6 weeks after subcloning. The CL18 frameshift was replaced by gene conversion tracts in the majority of wild-type cells, but no evidence for a gene conversion event was found in 80 light chain sequences from AID-negative cells (Fig. 3D).

The AID disruption in DT40 produces a complete block of Ig gene conversion. We have tested a number of DNA repair and recombination candidate genes over the years, but these mutants either maintain Ig gene conversion at wild-type levels [RAG-2 (13), MSH4,MSH3, MSH6, and POLλ (10)] or show a modest reduction, most likely due to defects in general homologous recombination [RAD54(14) and RAD52 (10)]. The abolition of all detectable frameshift repair in theAID −/− DT40 mutant is remarkable, because a low level of gene conversion can be easily detected in non–B cell lines if an appropriately designed transgene with a donor and recipient sequence is provided (15). The dependence on AIDexpression strongly supports the notion that the repair of the V segment frameshift in DT40 reflects B cell–specific Ig gene conversion. The reduction of Ig gene conversion in the heterozygous clone suggests that AID acts in a dose-dependent manner. No clear reduction in somatic hypermutation or switch recombination was reported for the AID +/− heterozygous mice (6) and human carriers (7), although the results of transfection experiments with the CH12F3-2 cell line also suggest a dose-dependent stimulation of switch recombination (6).

The AID −/− phenotype links Ig gene conversion to the other two AID-regulated processes: somatic hypermutation and switch recombination. Although it had been speculated for a long time that Ig gene conversion and somatic hypermutation are related and possibly initiated by the same lesion, the DT40AID mutant provides direct genetic evidence for such a relationship. Because somatic hypermutation and switch recombination can be dissociated in certain cell types, and because DT40 does not seem to undergo switch recombination (10), events downstream of the AID action must determine which of the three processes are executed. The recent finding that mutations in theRAD51 homologs XRCC2 and XRCC3activate somatic hypermutation in DT40 (16) suggests that the decision between Ig gene conversion and somatic hypermutation is influenced by the availability of a specific homologous recombination pathway.

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


View Abstract

Navigate This Article