Recombinogenic Phenotype of Human Activation-Induced Cytosine Deaminase

During the T cell-dependent humoral response, the repertoire of Abs undergoes specific molecular processes that increase their affinity and specificity (1, 2, 3, 4). Somatic hypermutation (SHM),³ gene conversion, and class switch recombination (CSR) are known to be dependent on activation-induced cytosine deaminase (AID) activity. When it was found that AID deficiency causes defects in these processes, the leading explanation was RNA deamination, as AID is homologous to known RNA-modifying enzymes (5, 6, 7). Another hypothesis, deamination of cytosines in DNA, was considered, but to a lesser extent, because of the lack of evidence that AID has the capability to deaminate cytosine in DNA. Last year new data emerged that support a direct cytosine deamination model for SHM. First, AID expression in nonimmune system cells, e.g., fibroblasts and Escherichia coli, was found to be mutagenic (8, 9, 10). This effect of AID expression was dependent on the uracil-DNA glycosylase (UDG) status of the cells. Thus, the mutator effect of AID was enhanced in an ung1⁻ bacterial strain, suggesting a direct role of uracil in generation of the mutator effect (8). Second, inhibition of UDG changed the specificity of AID-induced mutations in the DT40 chicken cell line and in UDG-null mice (11, 12). These results were interpreted as the manifestation of a switch from mutagenic abasic site repair to mutagenesis during replication of deaminated cytosines. Finally, direct cytosine deamination by AID was demonstrated in vitro for a DNA substrate (13, 14, 15, 16), and it was also shown that AID-induced deamination of cytosine occurs predominantly in GYW/WRC motifs (13) that correspond to the RGYW/WRCY sequence (mutable nucleotides are underlined), a hotspot of SHM (17, 18). Therefore, SHM appears to be due to direct cytosine deamination, that leads to G-C to A-T mutations after replication, or to base substitutions at G-C pairs introduced by error-prone base excision repair (BER) (8, 19, 20) (Fig. 1). In contrast, the mechanisms by which AID might induce CSR and gene conversion remain unknown (21).

AID expression led to induction of CSR in fibroblasts, suggesting that AID is involved in both SHM and CSR (10, 22). Based on the reduction of CSR in UDG-null mice, it was proposed that the U/G mispair might be converted to a DNA strand break, either by the combined action of UDG and apurinic apyrimidinic endonuclease or by mismatch repair (23, 24). Such a strand break could initiate recombination (Fig. 1). It is well known that the initial steps of BER to remove alkylation damage can lead to recombination in E. coli (25, 26), yeast (27, 28), and mammalian cells (29). However, it has also been shown that de novo protein synthesis is required for AID-induced CSR, and this was interpreted in favor of an RNA-editing model for AID involvement in recombination (30), rather than a DNA repair-linked model. It was also suggested that AID expression, by an unspecified mechanism, leads to double-strand breaks, which are recombinogenic (31). In this study we conducted experiments to determine whether AID-mediated DNA cytosine deamination can lead to recombinogenic events. We expressed human AID in yeast and examined the effect on mitotic recombination as a function of UNG expression.

We used haploid yeast strains BY4741 (Research Genetics, Inc., Huntsville, AL), 1B-D770, E134, and 29RL-CG379, described previously (32). Their ung1 derivatives were obtained by complete substitution of the UNG1 open reading frame with the hygB marker. 1B-D770 and E134 and their ung1 derivatives were crossed to obtain two diploids: YUNI500 (MATα/ΜΑΤa ura3–4/ura3–52 leu2–3,112/leu2–3,112 trp1–289/trp1–289 his7–2/his7–2 ade5–1/ade5–1 lys2-Tn5–13/lys2::InsE_A14) and YUNI501 (the same, except for ung1:hygB/ung1::hygB) that were used for intrachromosomal recombination studies. We also crossed 1B-D770 and 29RL-CG379 and their ung1 derivatives to obtain strains YUNI400 (MATα/MATa ura3–4/ura3Δ ura3–29 leu2–3,112/leu2–3,112 trp1–289/trp1–289 bik1::ura3–29RL his7–2/his7–2 ade5–1/ade5–1 lys2-Tn5–13/lys2-Tn5–13) and YUNI402 (MATα/MATa ura3–4/ura3Δ ura3–29 leu2–3,112/leu2–3,112 trp1–289/trp1–289 bik1::ura3–29RL his7–2/his7–2 ade5–1/ade5–1 lys2-Tn5–13/lys2-Tn5–13 ung1::hygB/ung1::hygB), which were used for interchromosomal recombination studies.

cDNA created from total RNA, isolated from the BL2 cell line, was amplified in PCR with Advantage2 Polymerase Mix (Clontech Laboratories, Palo Alto, CA). Oligonucleotides based on the human AID cDNA sequence (AIDYSTF, 5′-TTGCGGCCGCAAGAAGACACTCTGGACACCA-3′; AIDYSTR, 5′-CCCACTAGTAGTCCCAAAGTACGAAATGCGTCG-3′) were used for amplification of the full-length open reading frame. Reactions were conducted with denaturation at 94°C for 5 min, followed by 20 cycles of denaturation at 94°C for 30 s, annealing at 58°C for 30 s, and strand extension for 2 min at 72°C. The product of reactions was digested with NotI/SpeI restriction enzymes and ligated into the pESC-LEU vector (Stratagene, La Jolla, CA). The corresponding human (h) plasmid (p) phAID was sequenced to verify the construct.

DNA from individual clones was isolated with the Yeast DNA Purification Kit (Epicentre Technologies, Madison, WI) according to the manufacturer’s recommendations. The CAN1 sequence was amplified with primers canF (5′-CAGACTTCTTAACTCCTG-3′) and CANR (5′-GGAATGTGATTAAAGGTAATAAAACG-3′).

The reaction was conducted with denaturation at 94°C for 5 min, followed by 30 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, and strand extension for 2 min at 72°C. The sequence of the purified PCR product was verified using primers canF, can2F, can3F, can1R, can2R, and can3R, as previously described (33).

Yeast extracts were prepared as described previously (32). Briefly, cells transformed by phAID or pESC-LEU were grown in YP medium supplemented with 2% galactose to an OD at 600 nm of 0.8–0.9. Cells were collected by centrifugation, resuspended in an equal volume of lysis buffer (25 mM Tris-HCl (pH 7.6), 1 mM EDTA, 100 mM NaCl, 10 mM 2-ME, and 1 mM PMSF), and disrupted by vortexing with the equal volume of glass beads (0.5 ml). Cell debris was removed by centrifugation. Extract protein (20 μg) was applied to NuPAGE 4–12% bis-Tris gels (Invitrogen, Carlsbad, CA), and proteins were transferred to a nitrocellulose membrane. Western blots were performed according to the manufacturer’s recommendations, using a 1/2000 dilution of M2 anti-FLAG Ab (Sigma-Aldrich, St. Louis, MO). The bands were visualized by Western Lightning Chemiluminescence Reagent Plus (PerkinElmer, Wellesley, MA).

In a patch test, the tested strains were grown in medium containing glucose. Then, yeast were replicated with velvets to medium containing galactose. After 2 days of induction, plates were replicated to glucose-containing medium, selected for a plasmid marker and revertants. We used two methods to study recombination/mutation rates. In one, transformants by vector or galactose-inducible hAID-bearing plasmids were selected and colony-purified on appropriate selective medium on glucose-containing plates. Then, 9–12 colonies were transferred with a microbiological loop into 5 ml of corresponding liquid medium selective for a plasmid marker and containing galactose. Tubes were incubated with aeration for 3 days. Viability and mutant frequencies were determined as described previously (18). In another method, which allowed only very limited growth on selective medium, cells grown in plasmid selective medium in glucose were plated onto galactose-containing medium selective for revertants. Mutation/recombination rates were then measured using a replicator-pronger device as previously described (34, 35).

To study the effects of DNA cytosine deamination by AID in a yeast system, we studied mutations rates in yeast strains overexpressing hAID. The hAID gene was cloned in a yeast galactose-inducible expression vector in-frame with a C terminus FLAG epitope. BY4741, E134, and 1B-D770 strains were used, which allow concomitant measurement of mutation rates at several loci. These included: forward mutations at the CAN1 locus, where mutation reflects a variety of substitution, frameshift, and more complex events; reversion of the trp1–289 (TAG) (36) and ade5–1 (TAA) nonsense mutations; reversion of the ura3–4 missense mutation (strain 1B-D770), which reverts via a T to C change (position 605; Fig. 3,A) in a CTC codon, resulting in a CCC codon encoding proline (32); and Ty1 transposon element insertion (Fig. 3) (37) (strain E134) in the ura3–52 gene.

The expression of AID was confirmed by Western blotting (Fig. 2,A). In a patch test with strain BY4741, AID expression led to a moderate increase in Can^r mutations in the UNG1⁺ strain and a strong increase in Can^r mutants in the BY4741 ung1::KanMX strain (Fig. 2,B). Mutation rates were determined in growing cultures of haploid stains and the corresponding ung1 derivatives (Table I). In UNG1⁺ strains, AID expression lead to a 10-fold increase in forward mutation and a 5- to 10-fold increase in nonsense mutation reversion. The ung1 mutation led to approximately the same increase in mutation rate. When AID was expressed in the ung1 strain, the mutator effect was higher for Can^r forward mutations (160- to 190-fold increase over wild-type strain) and much higher for nonsense mutation reversion (720- to 3800-fold increase over wild-type strain). Genetic analysis of revertants showed that they were dominant suppressors, most likely representing mutations in anticodons of tRNA genes. The high response of the ade5–1 and trp1–289 markers to AID expression may reflect a role of transcription in AID-induced hypermutation in yeast, as tRNA genes are transcribed differently from metabolic genes by RNA polymerase III.

AID expression did not induce frameshift mutations (data not shown). Also, no induction of reversion of the ura3–4 marker was observed, confirming that mutations at A-T pairs were not induced; in contrast, ura3–29 was revertible due to C-G to T-A transitions, albeit moderately. Sequencing of 28 Can^r mutants revealed that all were due to single base changes in G-C pairs, 26 being transitions consistent with the specificity of mutagenesis due to cytosine deamination (Table II) (38). Both DNA strands were affected equally, as we observed 13 G-C to A-T mutations and 15 C-G to T-A mutations. This is similar to observations for SHM (18, 39, 40, 41) and different from recent data in E. coli that suggested asymmetric deamination of one DNA strand (13, 14, 15, 16, 42). It has been proposed that mutations at A-T base pairs during SHM are generated with the assistance of mismatch repair proteins, Msh2 and Msh6 (23). Although, the strains used in this study were MMR proficient, this mechanism was not operating during AID expression-induced mutagenesis, as mutations at A-T base pairs were not observed. We conclude that AID expression was mutagenic due to DNA cytosine deamination.

Next, we crossed the E134 and 1B-D770 strains to obtain diploid strains possessing homozygous trp1–289 and ade5–1 markers, which were heteroallelic for ura3–4 and ura3–52 (Fig. 3,A). We examined the mutagenic and recombinogenic effects of AID in these diploids under conditions where expression of AID did not increase the reversion of either ura3–4 or ura3–52 (Table I). All Ura⁺ revertants in these experiments were due to mitotic recombination and gene conversion (43).

Cells from early stationary phase, grown in glucose, were plated onto medium selective for mutants and containing the inducing agent galactose. This medium permitted only limited growth of the parent strain due to trace amounts of required nutrients, but supported the growth of revertants. Under these conditions, a strong AID expression effect was detected for mutagenesis in the ung1 strain (Fig. 3,B). Note that the rate of mutation in the Trp locus in diploid strains was lower than that in haploid strains (Table I). There was a strong effect of AID expression on recombination (>10-fold) in the UNG1⁺ strain (Fig. 3,C). Similarly, AID expression stimulated recombination between URA3 alleles placed on different chromosomes, V and III (diploids resulting from crosses of 29RL-CG379 and 1B-D770; results not shown). Remarkably, the recombinogenic effect was manifested only in the UNG1⁺ strain (Fig. 3 C). Thus, AID-generated uracil removal by UDG was essential for initiation of recombination and prevention of mutagenesis.

The somatic hypermutation and class switch recombination that occur in B cells increase the diversity of Abs as a part of humoral T cell-dependent immune response. Both these processes are strictly dependent on AID protein, the function of which is to deaminate cytosine in RNA or DNA substrates. In this study we studied the effect of this enzyme using a heterologous yeast system. We found that the expression of AID resulted in increased mutagenesis and recombination. Both these end points were dependent on the background concerning expression of UDG1 activity. As predicted from the provided model (Fig. 1), inactivation of UDG1 resulted in increased rates of mutagenesis, whereas recombination was detected only in strains possessing normal UDG1 activity. Our results support the conclusion that AID acts through direct modification of DNA bases, rather than RNA.

The use of a yeast model provides several advantages. This lower eukaryote organism has been successfully used in unfolding the function of number of human proteins. It is relatively easy to create gene deletions and to estimate mutation rates. Typical experiments involve measurement of forward mutation rates in the canavanine locus. In UNG1 and ung1⁻ strains, mutations were observed in G/C base pairs. These results were similar to those obtained in mammalian cells, i.e., B cells (44), fibroblasts (45), and CHO cells expressing AID protein (44), as mutations were observed predominantly in G/C pairs. For example, with green fluorescence protein expressed in fibroblasts, only three of 247 mutations occurred in A/T pairs (45). In E. coli, ung1⁻ mutagenesis also occurs in G/C pairs (46).

In the UNG1-expressing strain, we observed two C to G transversion mutations (Table II). This could reflect Rev1 polymerase translesion synthesis activity, bypassing the abasic site with C insertion (47). In contrast to the picture in yeast, mammalian cell cytosines could be replaced by any nucleotide (9), as additional translesion polymerases could be involved, such as members of Y family (48). Disruption of Pol ι in human B cells significantly impaired mutagenesis in G/C pairs (49), and inactivation of Rev1 in chicken B cells resulted in less nontemplated mutations in Ig locus (50).

In this study we did not observe AID-induced strand bias mutations; thus, for 28 mutations observed, 13 were substitutions in guanine, and 15 were substitutions in cytosine. In contrast, it was recently reported that AID expression in E. coli resulted in preferential deamination of the nontranscribed DNA strand (13, 14, 15, 16). Analysis of the Ig locus in B cells revealed that both strands are equally selected for somatic hypermutations, at least for G/C pairs (39, 40, 51). Moreover, expression of AID in mammalian cells did not reveal any strand specificity in mutations in targeted genes. For example, expression of AID in human fibroblasts revealed 121 substitutions of guanine and 123 of cytosine (45); in B cell hybridoma, 14 mutations were substitutions of guanine, and 13 were of cytosine (9).

Ig CSR is a site-specific recombination between the switch (S) regions located upstream of the μ H chain and one of the downstream S regions, leading to elimination of the intervening regions of DNA. This process results in creating new Ig genes with preexisting Ag specificity encoded by a variable region and a constant region providing the new effector functions. The process starts with the initiation of transcription of the sterile germline IgH gene. In the next step, double-strand breaks are formed. Finally, in the third step, the ends are joined through the nonhomologous end-joining pathway.

It has been proposed that AID participates in CSR as an RNA-editing enzyme. According to this model, AID, as an RNA-modifying enzyme, could modify RNA novel transcriptional factors that will accelerate general transcription of the IgH locus, thereby increasing accessibility of the S region for recombination. AID also may be involved in modulation of the activity of a novel DNA endonuclease that will cleave accessible S regions. In contrast, AID could modulate the activity of as yet unidentified proteins involved in resolving double-strand break. If AID functions as a DNA-modifying enzyme, it would be expected to participate in the second stage of CSR, initiating the strand-nicking process by deaminating cytosines. If the initial BER steps are required for recombination, this process should be dependent on UDG1. As noted above, we demonstrated that expression of hAID in yeast increased homologous recombination in the URA3 locus, and this recombination was strictly dependent on UDG1 activity, showing that the initial events of BER are crucial for recombination. The intermediate steps of BER provide a single-strand break, whereas recombination and particularly CSR require the double-strand break. Such double-strand breaks can be formed during replication when the replication fork encounters and collapses at a nick (52). Alternately, the repair of two deaminated cytosines located in close proximity could lead to a double-strand break. The Sμ region contains GC repeats that could be the target for AID-dependant deamination. In particular, the region for study of recombination used in this study contains 13 GGGT and GAGC sequences that are the part of the consensus sequence of the mouse Sμ repeats (53). We also observed mutations in two bases in close proximity, at positions 979 and 980. If alterations at these sites occurs simultaneously in the same nucleus, BER will create a double-strand break.

After this article was prepared, other studies describing a role for uracil-DNA glycosylase in SHM and CSR were reported (54, 55). Analysis of B cells isolated from hyperIgM syndrome patients demonstrated that the cells are deficient in uracil-DNA glycosylase. Detailed analysis showed that all patients expressed a truncated form of UDG. Analysis of mutations in hyperIgM syndrome patients revealed that cytidine deamination alone is not sufficient to explain CSR and SHM (54). In particular, B cells carrying mutations outside the deaminase active site had normal SHM, but were impaired in CSR.

In summary, the expression of hAID in yeast represents a novel model system with which to study the biochemistry and genetics of immunological genes. The results presented in this study provide evidence that an AID-mediated step in SHM, gene conversion, and CSR is indeed DNA cytosine deamination. Because of the dependence of the AID effect on UDG activity, we conclude that BER intermediates are involved. An RNA-editing model for the role of AID in the recombination observed in this study is unlikely. Therefore, our results strongly support the cytosine deamination-induced model for SHM, gene conversion, and CSR. It remains to be determined how AID is targeted to specific chromosomal regions during the T cell-dependent humoral response that leads to either increased mutagenesis in rearranged variable regions or increased recombination in constant regions. Future studies on AID-interacting proteins may shed light on details of multiple functions of this protein.

We thank J. Myers for editorial assistance, and J. Horton for critical reading of the manuscript.