Activation-induced cytidine deaminase induces DNA break repair events more frequently in the Ig switch region than other sites in the mammalian genome

Experimental system and rationale

This study examines the capacity of AID-expressing hybridoma cells to stably incorporate transfected enhancer-trap plasmid DNA, with the additional objective of mapping AID-induced vector integration sites within the genome. The enhancer-trap plasmids used in this study are derivatives of pSV2neo 32 in which the SV40 early region enhancer facilitating neomycin phosphotransferase (neo) gene expression is deleted 33. The enhancer-trap feature enriches for genomic integration sites that are active in promoting neo gene expression 33. Since DSB stimulate integration of transfected plasmid DNA 34, 35 and AID is thought to target DSB to actively expressed genes 21, 22, 36, we reasoned that this vector system would yield an increased frequency of stable transformants in AID-expressing cells, and that analysis of vector integration events would identify genomic sites that are subject to AID cleavage.

Increased stable transformation frequency in hybridoma cells expressing wild-type AID

Recipient Sp6/HL hybridoma cells were transfected with pCEP4hAID 8 or pCEP4hAIDC90A, a construct containing a catalytically inactive mutant of AID 37, and hygromycin-resistant (HYG^R) transformants were screened by RT-PCR to identify those expressing AID. Twelve AID-expressing clones were recovered (data not shown). Western analysis (Fig. 1) confirmed the expected AID protein in three cell lines expressing wild-type AID mRNA (Sp6-10, Sp6-12, Sp6-22; lanes 3–5), and in three cell lines expressing catalytically inactive AID C90A mRNA (Sp6-3, Sp6-8, Sp6-31; lanes 6–8), but not in recipient Sp6/HL cells (lane 1), or in a HYG^R transformant in which AID mRNA is absent (Sp6-1, lane 2). The levels of wild-type or mutant AID expression were quantified using densitometry and the respective AID/β-actin ratios are shown beneath each lane.

Cell lines expressing wild-type or mutant AID were transfected with the uncut 9.6-kb enhancer-trap plasmid vector, pTCμEn^– (Fig. 2) 38. Initial experiments utilized this vector in order to assess whether AID-induced breaks in the Cμ region stimulate vector integration by homologous recombination. In later studies (described below), transfections were performed with the enhancer-trap pSV2neo vector alone. Following transfection with uncut pTCμEn^–, the frequency of stable G418^R transformants was determined (Fig. 3A). Wild-type AID-expressing cell lines (Sp6-10, Sp6-12 and Sp6-22) revealed a significant stimulation in the frequency of stable transfection compared to non-AID-expressing Sp6-1 cells or recipient Sp6/HL cells. In contrast, no significant differences in transfection efficiencies were observed between Sp6/HL and cell lines Sp6-3, Sp6-8 and Sp6-31 expressing catalytically inactive AID. In general, the stable incorporation of transfected vector correlates with the level of expression of wild-type AID: cell line Sp6-10, in which densitometric analysis reveals an AID/β-actin ratio of 1.3, features a 29-fold stimulation in transfection efficiency, followed by cell line Sp6-22 (AID/β-actin ratio = 0.4) and cell line Sp6-12 (AID/β-actin ratio = 0.3) with 18-fold and threefold stimulations, respectively.

While consistent with AID-induced genomic DSB stimulating incorporation of transfected uncut vector, the possibility that AID cleavage of the circular vector might enhance integration into pre-existing genomic breaks was considered. In this case, little or no stimulation in transformation efficiency would be expected following transfection of AID-expressing cells with linearized (ScaI-digested) vector. Although somewhat lower in comparison to uncut DNA, in AID-expressing Sp6-10 cells, a significant 13-fold higher frequency of stable transformants was observed compared to control Sp6/HL or Sp6-1 cells (Fig. 3B). Thus, we conclude that AID-induced cleavage of the genome provides sites for integration of transfected vector DNA.

The chromosomal IgH μ gene switch region is a preferred site of vector integration

To investigate whether genomic integration sites include the chromosomal IgH μ locus, 48 and 25 G418^R transformants recovered from the AID-expressing Sp6-10 and Sp6-22 hybridoma cell lines, and 55 G418^R transformants recovered following transfection of Sp6/HL and Sp6-1 hybridoma cell lines were examined by Southern analysis 39 according to the mapping strategy in Fig. 4. The enzymes PacI and PaeR7I do not reside within the pTCμEn^– vector, and therefore, vector integration into the IgH µ gene is expected to alter the size of the endogenous 14.7-kb PacI/PaeR7I fragment detected with µ-specific probe N.

Southern analysis revealed disruption of the endogenous IgH μ locus in only 6/55 G418^R transformants (11%) recovered from the Sp6/HL or Sp6-1 cell lines not expressing AID. In contrast, in the AID-expressing Sp6-10 and Sp6-22 cell lines, the IgH μ locus is disrupted in 25/48 (52%) and 8/25 (32%) of the G418^R transformants, respectively, i.e. an approximately fourfold increase. Thus, when considered with the aforementioned 29-fold stimulation in transformation efficiency, AID-expressing cells display a preference for vector integration into the chromosomal IgH μ locus by as much as 116-fold. In the remaining transformants recovered from either AID-expressing or AID-non-expressing cell lines, the endogenous IgH μ locus is intact, suggesting vector integration in another position in the genome. For a representative Southern blot, refer to Supporting Information Fig. 1.

More detailed analysis of vector integration sites within the IgH μ locus was conducted on the 25 and eight G418^R transformants recovered from the AID-expressing Sp6-10 and Sp6-22 hybridoma cell lines, respectively. As shown in Fig. 4, the enzymes XbaI and EcoRI cleave the endogenous IgH μ locus into TNP-specific Ig V_H (V_HTNP), Sμ and Cμ region segments, whose disruption by pTCμEn^– integration can be conveniently detected in Southern analysis with μ-specific probes E, N or F, respectively. Table 1 presents a summary of the mapping information. For representative Southern blots, refer to Supporting Information Fig. 2.

On average, the Sµ region is disrupted in 73% of transformants recovered from the AID-expressing Sp6-10 and Sp6-22 hybridoma cell lines. In contrast, the adjacent V_HTNP and Cμ regions are disrupted in only 15% and 6% of the transformants, respectively. More complex rearrangements involving the Sμ and adjacent V_HTNP or Cμ regions are observed in a minor fraction of Sp6-22 transformants (6%). Therefore, we conclude that the IgH Sμ region is a preferred site of vector integration in AID-expressing cells, being stimulated five- and 12-fold compared to the adjacent V_HTNP or Cμ regions, respectively. Further, use of the pTCμEn^– vector revealed no obvious enhancement of AID-induced homologous recombination between vector-borne and chromosomal Cμ regions.

The IgH Sμ region occupies ∼4.7 kb of the 6.2-kb XbaI fragment (Fig. 4) shown in the mapping studies above to harbor a substantial fraction of IgH μ locus vector integrations. To verify vector integration directly into the Sμ region, 20 G418^R Sp6-10 transformants that according to Southern analysis were disrupted for Sµ were subjected to PCR using primer combinations specific for the pSV2neo vector backbone (Fig. 2) and chromosomal Sμ region (Fig. 4). A PCR product was obtained with at least one pair of primers in 15 transformants (refer to Fig. 5A for representative cell lines), and of these, four were sequenced. In each clone, vector sequences adjoin Sμ sequences confirming vector integration directly into the 4.7-kb Sμ region (Fig. 5B).

Analysis of vector integration sites by digestion-circularization PCR

As suggested from the Southern analysis, a sizeable fraction of transformants from AID-expressing and AID-non-expressing cells bear vector integrations in unknown sites outside the chromosomal IgH μ locus. Therefore, digestion-circularization PCR (DC-PCR) was adopted as an alternate means of surveying vector integration sites (Fig. 6). For DC-PCR, the enhancer-trap pSV2neoEn^– vector was used, which is identical to pTCμEn^– (Fig. 2) except that it lacks the IgH Cμ region. AID-expressing Sp6-10 and control Sp6/HL hybridoma cell lines were electroporated with uncut pSV2neoEn^– and genomic DNA isolated from independent G418^R transformants was digested with either MscI or BglII. The enzymes cut within or near the intact vector-borne neo gene in each transformant and provide a second cut site in genomic DNA surrounding the integrated vector, thus creating a linear DNA fragment containing both vector and chromosomal sequences. The linear products were self-ligated, PCR-amplified using vector-specific primers, and the products sequenced to determine the integration site.

Of 94 independent G418^R Sp6-10 transformants screened, 29 yielded DNA products that could be sequenced. Likewise, of 58 independent G418^R Sp6/HL transformants screened as controls, 16 were successfully amplified and sequenced. The failure to amplify more independent transformants is likely a function of the unknown distance between the vector-borne and chromosomal MscI or BglII restriction enzyme sites and the difficulty of amplifying large ligation products with primers specific to the vector backbone.

As summarized in Table 2, 16 of the 29 Sp6-10 clones were characterized by vector integration into a c-myc/IgH translocation in the recipient Sp6/HL hybridoma. The Sp6/HL hybridoma is derived from a fusion involving the myeloma P3-X63-Ag8 40 in which the c-myc/IgH translocation has been characterized and contains the 3′ portion of c-myc fused end-to-end with the IgH Sμ region (41, reviewed in 42). Interestingly, of the 16 vector insertions that occurred in the translocated chromosome, 13/16 integrations occurred in the IgH Sμ portion, 1/16 occurred between exons 1 and 2 of the translocated c-myc portion, and 2/16 occurred in the Sμ region, but with insertion of additional (probably) filler nucleotides at the vector: DNA junction. Furthermore, in an additional three clones, vector integration occurred in the endogenous IgH Sμ region. Thus, 18/29 vector integrations occurred in Sμ region sequences (Table 2). The remainder of the vector integrations in the Sp6-10 transformants occurred in other IgH genes, the κ light chain variable region, or in other locations in the hybridoma genome (Table 2).

None of the control Sp6/HL clones revealed vector integrations in Sμ, c-myc or c-myc:Sμ sequences. Instead, the majority showed vector integration into repetitive DNA sequences, and other locations in the hybridoma genome including a single integration in the Ig κ gene (Table 2). Comparison of the data reveals that AID expression elevates vector integration into chromosomal IgH genes (the endogenous Sμ and the IgH γ genes) compared to Ig κ genes by a ratio of 3:1. To a first approximation, the frequency of vector integration into the Ig κ genes in AID-expressing Sp6-10 cells (2/29) appears similar to that in Sp6/HL control cells (1/15). However, when considered in relation to the aforementioned 29-fold stimulation in vector integration in Sp6-10 cells (Fig. 3A), it is clear that AID-induced vector integration into the Ig κ genes is enhanced 30-fold compared to control Sp6/HL cells. The sequences of vector:chromosome junctions obtained by DC-PCR analysis are shown in Supporting Information Fig. 3.

The remaining 66 Sp6-10 and 42 Sp6/HL G418^R transformants that could not be amplified by DC-PCR were tested for their capacity to produce normal TNP-specific IgM by lysis of TNP-coupled sheep red cells in agarose plate spot tests 40, 43. The rationale was that if AID-induced vector insertion occurred in the chromosomal IgH μ locus, mainly by disruption of the V_HTNP or Sμ regions as suggested from the Southern analysis of AID-induced Sp6-10 and Sp6-22 clones described above, then a significant fraction of the Sp6-10 transformants would fail to make normal TNP-specific IgM and hence would be non-cytolytic in this assay. The analysis revealed that 52% (34/66) and 17% (7/42) of the independent Sp6-10 and Sp6/HL transformants were non-cytolytic, suggesting vector integration into the IgH μ locus, while the rest made normal cytolytic TNP-specific IgM consistent with vector integration outside the IgH μ locus. These results confirm the earlier Southern analysis, which revealed vector integration into the IgH μ locus in 52 and 11% of transformants isolated from the Sp6-10 and Sp6/HL hybridoma cell lines, respectively.

To determine whether the non-cytolytic phenotype is the result of vector integration into the Sμ region, a subset of ten non-cytolytic Sp6-10 transformants was examined by PCR using vector and Sμ-specific primers similar to that described in Fig. 5A. A specific PCR product was observed in 7/10 cell lines (data not shown). Thus, like the previous Southern analysis of transformants with vector insertions in the IgH μ locus, the PCR analysis suggested that the IgH Sμ region is the site of vector integration in a large fraction of the non-cytolytic transformants. Thus, the negative outcome of the initial DC-PCR analysis utilizing vector-specific primers is likely attributable to the length or repetitive nature of the Sμ region, which is known to be problematic for PCR amplification 44.

Hybridoma cells and culture conditions

The mouse hybridoma Sp6/HL bears a single copy of the TNP-specific chromosomal IgH μ gene and makes normal, cytolytic TNP-specific IgM (κ-chain) 38, 40. The igm10 hybridoma is an Sp6/HL-derived mutant that has lost the TNP-specific μ gene 38. The conditions for hybridoma growth in Dulbecco's modified Eagle's medium are described elsewhere 38, 40.

Construction of stable transformants expressing human AID

The vector pCEP4hAID contains the 560-bp cDNA segment encoding human AID inserted into the backbone of pCEP4, which contains the HYG^R gene 8. To create the vector pCEP4hAIDC90A, the wild-type AID cDNA in pCEP4hAID was replaced with a catalytically inactive form of AID containing the substitution mutation C90A 37. 50 μg of the respective vectors were digested with EcoRV/NruI, and transfected into 2×10⁷ Sp6/HL hybridoma cells by electroporation 57. Selection for HYG^R cells was performed in batch culture, and limited dilution cloning (0.1 cell/well) in 96-well tissue culture plates was used to recover independent HYG^R transformants, which were saved for analysis of AID expression levels.

Routine screening of HYG^R transformants for AID expression was performed by RT-PCR. RNA was prepared by Trizol extraction (Invitrogen). DNaseI-treated RNA (3 mg) were reverse-transcribed using Superscript (Invitrogen) according to manufacturer's specifications. Each sample (5 µL) was amplified with Biotools Taq polymerase (Interscience) using forward and reverse primers (5′-TAGACCCTGGCCGCTGCTACC-3′ and 5′-CAAAAGGATGCGCCGAAGCTGTCTGGAG-3′, respectively), derived from a 316-bp segment specific to the human AID sequence 58. The conditions used for PCR have been described previously 59.

Western blot analysis of AID expression was performed using standard procedures. To detect the 24-kDa protein, the membrane was probed with anti-mouse AID antibody (AID 7E7; Cell Signaling Technology Inc., Danvers, MA), and then with HRP-conjugated secondary antibody (goat anti-mouse IgG, γ chain-specific; SouthernBiotech) as suggested by the manufacturer. Immunoblot signals were detected using ECL reagent (Amersham Biosciences). Quantification of protein bands in Western blots was performed by densitometry using Scion Image software (Scion Corporation, Frederick, MD).

Assay for stable transfection

The pTCμEn^– vector contains a 4.3-kb segment from the IgH Cμ region inserted into a derivative of the vector pSV2neo 32 in which the 372-bp Nsi/NdeI SV40 early region enhancer segment responsible for expression of the neo gene was deleted 33. Uncut or linear pTCμEn^– vector was transfected into 5×10⁶ wild-type or mutant AID-expressing or -non-expressing Sp6/HL mouse hybridoma cells by electroporation as described 57. Immediately following electroporation, the hybridoma cells were distributed in tissue culture plates at low density (∼3000 viable cells/well) in Dulbecco's modified Eagle's medium supplemented with 1.2 mg/mL G418 to recover G418^R transformants 33, 59. The low frequency of transfection with the enhancer-trap vector system (∼10^–5/cell) makes it highly likely that each G418^R transformant represents the progeny of an independent G418^R cell 33, 59.

Digestion-circularization PCR

To investigate AID-induced vector integration sites, a DC-PCR strategy was used. Independent stable G418^R transformants were generated as described above using an uncut, SV40 enhancer-deleted version of the vector pSV2neo 32 (GenBank accession number U02434), which we refer to as pSV2neoEn^–. Genomic DNA was prepared from independent transformants and 30 μg were digested with either MscI or BglII and joined with T4 DNA ligase using conditions suggested by the manufacturer (Invitrogen). The ligation products were PCR-amplified and sequenced on an ABI Prism automated sequencer (Model 3730, ABI Inc.) using standard procedures (College of Biological Science DNA Facility, University of Guelph) to determine the site(s) of vector integration. Vector-specific primers 5190 (5′-CCAGTAGCTGACATTCATCC-3′) and 4897 (5′-AGAGGCTATTCGGCTATGAC-3′) were used to amplify and sequence ligation products generated by MscI cleavage, whereas vector-specific primers 4502 (5′-GAAGCCGGTCTTGTCGATCA-3′) and 4837 (5′-AACACGGCGGCATCAGAGCA-3′) were used to amplify and sequence products generated by BglII cleavage.

Sequences were compared to mouse genomic sequences using a blastn comparison (http://www.ncbi.nlm.nih.gov). Alignments with the highest bit score and lowest E value were used to identify vector integration sites. Sequences were further characterized using Repeat Masker software to identify repetitive elements (http://www.repeatmasker.org).

IgM analysis

Culture supernatants from individual G418^R transformants were examined for secreted TNP-specific IgM production by lysis of TNP-coupled sheep red cells by an agarose plate spot test as described previously 40, 43.

General techniques

Genomic DNA was prepared by SDS-proteinase K treatment as described by Gross-Bellard et al. 60. Plasmid DNA was propagated in Escherichia coli DH5α, and isolated by column purification (Marligen Biosciences). Restriction enzymes were purchased from New England Biolabs and used in accordance with manufacturer's specifications. For Southern analysis, restriction-enzyme-digested genomic DNA was electrophoresed through 0.7% agarose gels, transferred to nitrocellulose membrane, hybridized with ³²P-labeled probe, and band sizes visualized by autoradiography according to standard procedures 61. PCR amplification of Sμ:vector junctions exploited various combinations of pSV2neo and chromosomal Sμ-specific primer pairs. Vector primers 5190, 4897 (described above), neoF20 (5′-GCTGACTGTCAACTGTAGCA-3′) and ampR20 (5′-AAGTGCCACCTGACGTCTAA-3′) are specific to the pSV2neo backbone 32 of the transfected enhancer-trap pTCμEn^– vector (refer to Fig. 2 for nucleotide binding positions). Chromosome-specific primers 5989 (5′-GACAGAGAAGGCCAGACTCA-3′) and 9797 (5′-CTCAGCTCAGCTCACTCCAG-3′) bind 5′ and 3′ of the mouse chromosomal Sμ region (refer to Fig. 4 for nucleotide binding positions). PCR products were sequenced on an ABI Prism sequencer Model 3730 using standard procedures (CBS DNA Facility, University of Guelph).