APOBEC3 proteins and genomic stability

Abstract

The human APOBEC3 family of cytidine deaminases constitutes a cellular intrinsic defense mechanism that is effective against a range of viruses and retro-elements. While it is well established that these enzymes are powerful mutators of viral DNA, the possibility that their activity could threaten the integrity of the host genome has only recently begun to be investigated. Here, we discuss the implications of new evidence suggesting that APOBEC3 proteins can mediate the deamination of cellular DNA. The maintenance of genomic integrity in the face of this potential off-target activity must require high fidelity DNA repair and strict regulation of APOBEC3 gene expression and enzyme activity. Conversely, the ability of specific members of the APOBEC3 family to activate DNA damage signaling pathways might also reflect another way that these proteins contribute to the host immune response.

APOBEC3 Proteins in Innate Immunity

The AID/APOBEC3 family of cytidine deaminases has emerged as an important component of the immune system in vertebrates, acting in both innate and adaptive pathways.1 The AID/APOBEC3 enzymes share the ability to catalyze the deamination of cytidine to uracil and are thus capable of functioning as powerful DNA mutators.2 These cytidine deaminases have evolved to target the genome of both the host and its invading pathogens. Activation-induced cytidine deaminase (AID), a key mediator of the adaptive humoral immune response, functions in germinal center B cells to diversify antigen receptor genes through the processes of somatic hypermutation (SHM) and class switch recombination (CSR).3,4 In contrast, the human apolipoprotein-B mRNA-editing catalytic polypeptide-like 3 (APOBEC3) family of cytidine deaminases, which comprise seven host restriction factors (APOBEC3A-H), constitute an innate barrier to retroviruses, endogenous retro-elements and DNA viruses.5

APOBEC3 proteins have been shown to exert selective antiviral activity against a range of viruses and to act at specific steps within the viral infection cycle.6 Several studies have demonstrated that the restricting action of APOBEC3 proteins can result from both deaminase-dependent and deaminase-independent mechanisms.7 Nevertheless, in many cases, APOBEC3-mediated antiviral activity correlates with the detection of hypermutated viral genomes.8–13 Hypermutation of viral genomes by APOBEC3 proteins has been suggested to compromise viral infectivity by altering the integrity of open reading frames. Alternatively, uracilcontaining viral DNA could represent a substrate for the cellular DNA cellular repair machinery. Deamination of cytosines creates uracil lesions that can be processed by the cellular DNA glycosylases, which catalyze the hydrolysis of the glycosylic bond between the base and the deoxyribose backbone, leaving an abasic site. It has been suggested that the concerted action of uracil-DNA glycosylase (UNG), the main glycosylase initiating the base-excision repair of uracil in DNA, and apurinic/apyrimidinic endonuclease1 (APE1), which catalyzes strand cleavage at the lesion, can trigger the degradation of deaminated viral DNA.8,14 However, this hypothesis has been challenged,15–17 and it remains unclear whether these repair enzymes contribute to APOBEC3-mediated antiviral activity.

In addition to their role as antiviral factors, it has been recently proposed that APOBEC3 proteins can mediate the extensive deamination and degradation of transfected plasmid DNA.18 A study by Stenglein et al.18 described the ability of several APOBEC3 members to restrict gene transfer and suggested that clearance of foreign DNA could be a distinct physiological function of APOBEC3 proteins. Among all APOBEC3 proteins, APOBEC3A (A3A) had the strongest DNA-restricting ability,18 consistent with its previously reported potent cytidine deaminase activity.19 Using selective amplification of edited DNA through differential denaturation (3D)-PCR,20 deamination marks were detected in plasmids recovered from transfected primary macrophages that were induced to express A3A. The authors were, however, unable to detect editing in genomic DNA, leading them to suggest that the host nuclear DNA may be less susceptible to A3A-mediated deamination than plasmid DNA.

Collateral Damage from Cytidine Deaminases

The AID and APOBEC3 proteins have been assigned to specific physiological functions through their activities on defined substrates, i.e., AID functions at the immunoglobulin loci to promote antibody gene diversification, while APOBEC3 proteins act upon viral nucleic acids as a form of intrinsic host defense.21 Their ability to trigger nucleotide alterations confers intrinsic mutagenic potential, implying that deamination of DNA outside of their intended targets could cause collateral damage to the cellular genome with dire consequences.22 The precise details of how AID is targeted to the immunoglobulin genes are still being deciphered, but it is now clear that AID deaminates many more genes than originally suspected and can cause off-target mutations throughout the genome.23,24 Recent work has demonstrated that AID can deaminate many oncogenes involved in B-cell tumorigenesis, highlighting the role of base excision repair (BER) and mismatch repair (MMR) pathways in preventing the accumulation of deleterious mutation.25 AID can also generate double-strand breaks (DSBs) at many non-immunoglobulin genes.26 Processing of these breaks can lead to rare translocations through joining of DSBs generated during antigen receptor diversification to breaks that are located near oncogenes.27,28 In this regard, it has been shown that homologous recombination repair proteins act as safeguards to prevent genomic instability by the widespread genomic breaks that can be induced by off-target activity of AID in B cells.29 Interestingly, it has been hypothesized that over the course of evolution, AID activity has been restricted to minimize the risk of genomic stability.30

Evidence for APOBEC3-Mediated Deamination of Cellular DNA

The studies of APOBEC3 proteins have so far predominantly focused on understanding their antiviral activities and biochemical properties. In contrast, little is known about the possible detrimental consequences of expressing these mutagenic enzymes and their propensity to act in an “off-target” fashion on host rather than viral genomes. In a recent study, we reported that A3A expression can induce DNA breaks and activate the cellular DNA damage response (DDR) in a deaminasedependent fashion.31 The DDR induced by A3A was revealed using antibodies generated to specific phosphorylation marks on DNA damage/repair proteins (such as the histone variant H2AX),32 and DNA breaks were detected using the terminal deoxynucleotidyl transferase dUTP nick end-labeling (TUNEL) assay. In addition, using a UNG inhibitor (UGI), we demonstrated that UNG activity is required for both A3A-induced DNA breaks and activation of the DDR, suggesting that DNA damage signaling occurs as a consequence of cellular processing of A3A-induced uracil lesions. Importantly, using stable cell lines expressing A3A under the control of a Tet-inducible promoter, we showed that activation of the DDR by induction of A3A expression is independent of the presence of foreign DNA. This observation implies that uncontrolled A3A expression can damage the cellular DNA in the absence of foreign DNA or virus infection. In our study, we also reported that ectopic A3A expression leads to cell cycle arrest in S phase. Similar observations have been made when B cells deficient for homologous recombination (HR) were induced to express AID.29 Although it is unclear whether endogenous A3A can affect cell cycle progression, our results suggest that the effects of A3A expression on the host cell could have been a contributing factor in previous experiments focused on understanding A3A function as a restriction factor.

The precise series of events that results in activation of the DDR we detected with overexpressed A3A remains to be determined. Since A3A and the other AID/APOBEC3s have been well-documented to have a preference for single-stranded DNA substrates,19,33–35 the initiating deamination must occur on genomic DNA that is single-stranded and accessible. Active transcription is a prerequisite for somatic hypermutation induced by AID,36–39 and thus, it is possible that the genomic instability induced by A3A is also associated with transcription. In this case, DNA damage could result from deamination on the coding DNA strand that transiently adopts a single-stranded conformation during the passage of a transcription bubble.40 Another possibility is that the DNA damage induced by A3A occurs in S phase, when single-strand DNA is exposed during cellular DNA replication. This conclusion is consistent with our recent findings showing that when we induce A3A expression in cells arrested in G₁, we do not detect H2AX phosphorylation (unpublished observations). Alternatively, it is possible that deaminated residues induced in G₁ are not processed into breaks or are somehow protected from activation of the DDR. Identification of the nature of the genomic DNA substrate targeted by A3A will provide information that could lead to a better understanding of its physiological function.

While our work demonstrated that ectopic A3A expression is detrimental to cellular DNA integrity, we did not obtain conclusive evidence for activation of DNA damage signaling by endogenous A3A. Using an approach based on 3D-PCR, Suspene et al.41 reported that mitochondrial DNA containing signs of deamination in the form of G→A and C→T hypermutations can be amplified from peripheral blood mononuclear cells expressing A3 proteins. In addition, the authors detected hypermutated nuclear DNA in cells from patients with a specific form of hyper-IgM syndrome that is associated with UNG deficiency.42 These intriguing results suggest that cellular DNA is sensitive to endogenous levels of APOBEC3 proteins. However, further studies are required to evaluate the extent of APOBEC3-mediated mutagenesis of cellular DNA under physiological conditions.

Regulation of APOBEC3 Activity

The finding that APOBEC3 proteins are capable of editing cellular DNA provides a compelling reason to suggest that their deaminase activity must be tightly regulated. Since the discovery of AID and its role in SHM and CSR,43–45 an immense amount of effort has been directed toward deciphering how the enzyme functions and determining how it is regulated.46 Regulation has been proposed at many different levels, including transcription,47 mRNA degradation by miRNAs,48 cellular localization and active nuclear shuttling,49 post-translational modifications,50,51 proteasomal degradation52 and by interaction with various cellular partners.53–55 It is intriguing to note that AID expression is augmented in B cell-derived lymphomas and leukemias,56 and there is increasing evidence that AID upregulation by proinflammatory cytokines in tissues other than B cells can promote tumorigenesis.57

In the case of APOBEC3 proteins, little is currently known about their regulation. APOBEC3 proteins have been reported to be expressed in a range of hematopoietic cells and to be induced in response to different cytokines and hormones.19,58–63 Notably, APOBEC3 gene expression was shown to be induced by type-I interferon (IFN) stimulation, consistent with their roles in the antiviral immune response. A limited number of studies have investigated the transcriptional regulation of the APOBEC3 gene,64–67 and it is still unclear which transcription factors regulate its expression in response to IFN. A3A gene expression is very sensitive to INFα61 and contains an interferon-stimulated response element (ISRE) in its potential promoter region.58 Characterization of the APOBEC3 gene promoter regions is required, since a better understanding of their transcriptional regulation might provide crucial information on how the expression of A3 proteins can be modulated to reach a precarious costbenefit balance.

While APOBEC3G (A3G) activity has been shown to be regulated by several mechanisms,5,62,68–80 the mechanisms controlling the activity of APOBEC3 proteins that localize to the nucleus have yet to be described. Is it, indeed, surprising that despite being the most potent deaminase of the APOBEC3 family, A3A partially localizes to the nucleus.19,81 While it has been reported to co-purify with complexes containing LINE-1 RNA,82 interacting proteins that might regulate A3A activity have yet to be identified. It is indeed possible that A3A DNA binding affinity and/or activity are modulated by its interaction with a cellular co-factor(s), as has been proposed for other cytidine deaminases.55,83–86 Human A3A is expressed as two different isoforms, the smaller form resulting from the presence of an internal translation initiation codon.87,88 In a recent study, Thielen et al.88 showed that both A3A isoforms appear to be enzymatically active. However, it remains unclear if the expression of two A3A isoforms represents a strategy to regulate its activity and whether the acquisition of an alternative start codon is a consequence of recent evolution.88,89

The finding that A3A-induced mutations in nuclear DNA are detectable exclusively in UGI-expressing cells suggests that the BER machinery is normally able to limit the deleterious effects of A3A activity.41,90 According to a recent analysis of APOBEC3 gene expression in hematopoietic cells, A3A is the sole member of the family whose expression is restricted to the myeloid lineage.91 These observations support the hypothesis that restricting A3A expression to non-proliferating cells could constitute a strategy to limit the negative consequences of its off-target activity. In this regard, it has been reported that the A3A gene is under negative selection,92 supporting the idea that, at some point during vertebrate evolution, the collateral damage resulting from A3A expression might have exceeded its beneficial properties. Similar hypotheses have been proposed to explain the frequent deletion of the APOBEC3B gene in humans93 and the recent acquisition of destabilizing mutations affecting the anti-retroelement activity of APOBEC3H.94

Linking the DDR and the Immune Response

Our recent analysis of A3A-induced DDR revealed that activation of many DNA damage mediators is exquisitely sensitive to the presence of A3A and can be observed in response to protein levels that are comparable to physiological A3A levels. It has been proposed that activation of the DDR could participate in the immune response by inducing the expression of ligands for the activating receptor NKGD2.95,96 In addition, Gourzi et al.97 have shown that expression of AID in response to the transforming retrovirus Abelson murine leukemia virus (Ab-MLV) can activate DNA damage signaling and promote the expression of the NKG2D ligand Rae-1. The authors also found that virus-induced AID expression restricts the proliferation of Ab-MLV-infected cells. More recently, A3G expression in HIV-infected T cells has been shown to correlate with the upregulation of natural killer (NK) cell-activating ligands and activation of DNA damage signaling.98 Interestingly, this study reported that uracil residues can be generated in both viral and cellular DNA in infected cells, and that this process is inhibited in the presence the viral Vpr and Vif proteins. Although it is unclear how A3G can access nuclear DNA in HIV-infected cells, these observations suggest that activation of the DDR by APOBEC3 proteins can promote the recognition of infected cells by NK cells.

In light of these observations, it is tempting to speculate that the controlled ability of A3A to activate the cellular DDR could similarly represent a strategy that contributes to the immune response. AID expression in germinal center B cells can affect cell viability, and this appears to correlate with its ability to induce DNA breaks.99 This effect has been proposed to play a role in maintaining the integrity of the germinal center compartment and preventing the development of autoimmune B cells. It is thus possible that induction of DNA damage by members of the APOBEC3 family could contribute to cell cycle checkpoint activation and promote apoptosis in physiological conditions, such as in virus-infected cells.96,100 We have observed that prolonged induction of A3A expression in an inducible cell line leads to cell death (unpublished observations). It will be important to determine if endogenous A3A retains its damaging ability in certain cell types and whether it proves to be beneficial under specific circumstances.

Conclusions

The finding that APOBEC3 proteins can deaminate cellular DNA suggests they may contribute to genomic instability. The new observations we have here discussed, combined with the established links between AID and genomic instability, highlight the need to carefully investigate the potential role of APOBEC3 proteins in tumorigenesis and cancer. Genome protection from deaminase-induced mutagenesis requires tight regulation of the activity of the APOBEC3 proteins. It will be critical to identify the cellular mechanisms regulating APOBEC3 protein activity and determine whether uncontrolled deaminase activity can contribute to malignancy.101–103 Stimulation of A3A expression as part of the immune response to pathogens provides necessary intrinsic antiviral defenses but could also promote tumorigenic pathways. Several mechanisms could participate in protecting genomic DNA from destructive deamination, including chromatin structure, subnuclear compartmentalization or steric hindrance by single-stranded DNA binding proteins that cover the DNA substrate. Little is currently known about A3A regulation by either transcriptional control or post-translational modifications, and cellular factors that associate with the protein to influence its activity remain to be identified. In addition to their well-documented antiviral activities, APOBEC3 proteins may provide some additional physiological function that requires expression in particular cell types. This might be restriction of foreign DNA18 or a recently suggested role in DNA demethylation.104,105 Although more work is needed to clearly establish the role of APOBEC3 proteins in these processes, it is also intriguing to consider that deamination of the host cell DNA could be advantageous, and that APOBEC3 function in the immune system might not be restricted to their previously described antiviral and anti-retroelement activities. While it remains unclear whether the APOBEC3 proteins contribute to abundant genetic changes that accumulate in tumor cells, the recent findings suggest that at least some members of the family constitute a threat to cellular genomic integrity, and that their selection through evolution must have been governed by a cost-benefit balance.

Acknowledgments

We thank C. Lilley and B. Lamarche for helpful discussions. Work in the Weitzman lab was partially supported by NIH grants AI067952 and AI074967 (M.D.W.). This work was also funded by fellowships from the Instituto de Salud “Carlos III”/Consejo Superior de Investigaciones Científicas/Salk Institute and the Lynn Streim Postdoctoral Endowment Fellowship (I.N.), and the Natural Sciences and Engineering Research Council of Canada (S.L.).