Opossum APOBEC1 is a DNA mutator with retrovirus and retroelement restriction activity

APOBEC3s (A3s) are single-stranded DNA cytosine deaminases that provide innate immune defences against retroviruses and mobile elements. A3s are specific to eutherian mammals because no direct homologs exist at the syntenic genomic locus in metatherian (marsupial) or prototherian (monotreme) mammals. However, the A3s in these species have the likely evolutionary precursors, the antibody gene deaminase AID and the RNA/DNA editing enzyme APOBEC1 (A1). Here, we used cell culture-based assays to determine whether opossum A1 restricts the infectivity of retroviruses including human immunodeficiency virus type 1 (HIV-1) and the mobility of LTR/non-LTR retrotransposons. Opossum A1 partially inhibited HIV-1, as well as simian immunodeficiency virus (SIV), murine leukemia virus (MLV), and the retrotransposon MusD. The mechanism of inhibition required catalytic activity, except for human LINE1 (L1) restriction, which was deamination-independent. These results indicate that opossum A1 functions as an innate barrier to infection by retroviruses such as HIV-1, and controls LTR/non-LTR retrotransposition in marsupials.

The grey, short-tailed opossum, Monodelphis domestica, belongs to the Metatheria, which is one of the three major groups of modern mammals and the closest relative of the Eutheria (Fig. S1a). Although eutherians encode at least one A3 family protein, there is no A3-like gene in the genomes of non-eutherian mammals (Fig. S1b), which was confirmed by the opossum genomic DNA sequence 21 , a pan-species Z1 PCR analysis 22 and a BLAST search in this study (Fig. S1b). These observations suggest that the contribution of A1 to the innate immunity pathways of the marsupials could be greater than their contribution in eutherian mammals.
The opossum A1 is a protein of 235 amino acids that shares ~70% amino acid identity with those encoded by eutherian mammals (Fig. S2). Although the ability of this protein to edit apoB mRNA has previously been characterised 23 , whether it inhibits retroviruses including HIV-1 and LTR/non-LTR mobile elements remains to be determined. We addressed these questions using cell culture-based assays and found that opossum A1 from the small intestine is capable of restricting the infectivity of several retroviruses and the mobility of LTR/non-LTR retrotransposons. Taken together, our data indicate that the ability of A1 enzymes to protect the host genome from the invasion of foreign nucleic acids is conserved in marsupial and eutherian mammals.

Results
Opossum A1 is mutagenic in Escherichia coli. Although the A1 protein encoded in marsupials can edit a synthetic apoB mRNA substrate 23 , its DNA-editing activity has not been determined. The bacterial mutator assay is widely used to evaluate the mutagenic activity of the AID/APOBEC family proteins 17,18,[24][25][26] . In this assay, the expression of these proteins enhances the mutation of the bacterial RNA polymerase beta (rpoB) gene, which is reported as the frequency of rifampicin resistant (Rif R ) colonies. Therefore, we used this system to measure the mutagenic activity of opossum A1 by counting the Rif R mutant colonies that emerged during growth of multiple independent cultures (Fig. 1a). A1 cDNA was isolated from opossum small intestine and cloned into an expression vector. Consistent with previous observations 12,18 , rabbit A1 increased the number of Rif R colonies (24.1-fold relative to the vector control), whereas human A1 had only a slight effect (2.1-fold). Interestingly, the expression of wild-type (WT) opossum A1 in bacteria markedly enhanced the frequency of rpoB mutations (390-fold). However, this mutator phenotype was completely abrogated in catalytic mutant derivatives of opossum A1, E62A and E62Q (0.7-fold), even though they were expressed at levels similar to those of WT opossum A1 (Fig. 1b). These results demonstrate that opossum A1 has greater mutagenic activity than rabbit A1 in this E. coli-based mutation assay.
Opossum A1 partially restricts HIV-1 infection. Several cell-based studies have demonstrated that the A1 proteins of rabbit and rodents inhibit HIV-1 replication independently of HIV-1 Vif expression [9][10][11][12] . To determine whether opossum A1 suppresses HIV-1 infectivity, we performed single-cycle HIV-1 infection assays in HEK293T cells. As expected, rabbit A1 caused a ~200-fold reduction in HIV-1 infectivity, with significant levels of A1 packaged into viral particles (Fig. 2). WT opossum A1 also inhibited HIV-1 replication compared with the vector control (~2-fold; P < 0.01), and the antiviral activity of opossum A1 was dose-dependent ( Fig. 2a,b). The packaging level of the opossum protein was 2-fold lower than that of rabbit A1 (Fig. 2c,d). The antiviral activity of the WT opossum protein was abolished in the catalytic mutants, E62A and E62Q, regardless of the packaging level of the E62A mutant. These results indicate that opossum A1 partially restricts HIV-1 infectivity, and this restriction requires a catalytically active enzyme.
Opossum A1 mainly localises to the nucleus. It has been demonstrated that the degree of A1 cytoplasmic localisation correlates with anti-HIV1 activity 12 . To assess this relationship here, we examined the subcellular localisation of opossum A1 in HeLa and HEK293T cells. The rabbit A1, which is the most active A1 protein against HIV-1 11 , was mainly distributed in the cytoplasm of both HeLa and HEK293T cells, where ~70% of the total signal was observed (Fig. 3a,b, S3). In contrast, ~80% of the total signal for WT opossum A1 was detected in the nuclei of both HeLa and HEK293T cells. The catalytic mutant E62A displayed the same subcellular  (a) Representative single-cycle assays with WT opossum A1 or deaminase-defective mutants (E62A or E62Q). Viral infectivity is plotted relative to vector control and is proportional to luciferase activity (n = 3, average + /− SD). P values are derived by comparisons with vector control data. (b) Representative single-cycle assays with varying concentrations of WT opossum A1 (10, 50 and 250 ng). Data represent the luciferase activity relative to vector control data (n = 3, average + /− SD; P-values as above). (c) Immunoblots of A1 in viral particles and cell lysates. A1 and HIV-1 Gag were detected using anti-HA anti-p24 antibodies, respectively. β -actin expression was used a loading control. (d) A histogram showing the relative packaging efficiency of each A1 construct. Band intensity of each HA-tagged A1 was normalised to the corresponding p24 level. Packaging efficiency is shown graphically relative to the packaging of rabbit A1 (n = 3, average + /− SD). P values represent comparisons with rabbit A1 data. localisation in HeLa and HEK293T cells as the WT protein. However, the E62Q mutation impaired the distribution of the protein, and the mutant protein was detected in the cytoplasm, implying aberrant localisation. These findings were confirmed by immunoblotting fractionated cell extracts (Fig. 3c,d). These results suggest that the moderate antiviral activity of opossum A1 correlates, at least in part, with lower steady state cytoplasmic levels of protein in cells.

Opossum A1 partially restricts SIVs and MLV.
We have also demonstrated that the A1 proteins of rabbit and rodents inhibit the replication of several retroviruses including SIVs from rhesus macaque and African green monkey, and MLV 11 . To assess the capacity of opossum A1 to inhibit other retroviruses in addition to HIV-1, SIV and MLV pseudotyped viruses encoding a reporter gene were produced in the presence of opossum A1, and the effect of opossum A1 on retroviral replication was evaluated by infecting into HEK293T cells. Consistent with previous observation 11 , rabbit A1 inhibited the single cycle infection of SIVs from rhesus macaque and African green monkey (Fig. 4a to c), and MLV (Fig. 4d). A reduction in the infectivity of these viruses was also observed in the presence of opossum A1, and the degree of inhibition was similar to that of the anti-HIV-1 activity (see Fig. 2a,b versus Fig. 4a,c,d). The anti-retroviral activity against SIVs and MLV was similarly diminished by catalytic glutamate mutations (Fig. 4a,c,d), suggesting that catalytic activity is required for the restriction of these retroviruses by opossum A1. Thus, taken together with the HIV-1 data discussed above, these HEK293T-based experiments indicate that opossum A1 may have broad anti-retroviral activity.
Opossum A1 inhibits human L1 retrotransposition using a deaminase-independent mechanism. We have previously shown that the A1 homologues of eutherian mammals use a deaminase-independent mechanism to inhibit L1 retrotransposition in human cell lines 18 . We tested the ability of opossum A1 to inhibit human L1 retrotransposition using enhanced green fluorescent protein (EGFP)-based retrotransposition assays in HEK293T cells 18,27 . Consistent with our previous observation 18 , rabbit A1 suppressed human L1 retrotransposition (~25-fold) (Fig. 5a,b). Similarly, human L1 retrotransposition was reduced ~7.7-fold in the presence of WT opossum A1 (500 ng), and the level of suppression was proportional to the amount of enzyme expression. This anti-L1 activity was not abolished by the inactivation of the deaminase activity of opossum A1 (Fig. 5a, E62A and E62Q). The anti-L1 activity of opossum A1 was also supported by similar results with neo-marked L1 retrotransposition assays in HeLa cells (Fig. 5c,d). These data indicate that opossum A1 inhibits human L1 retrotransposition independently of its catalytic activity.
Opossum A1 suppresses MusD retrotransposition. It has been shown that eutherian mammal A1 proteins also inhibit the mobility of LTR retrotransposons 18 . To test whether this activity is present in the opossum A1, we performed a neo-marked MusD retrotransposition assay in HeLa cells. As previously observed, rabbit A1 reduced MusD retrotransposition by ~4-fold (Fig. 6a). In the presence of opossum A1, a 2-fold reduction in the MusD retrotransposition frequency was observed, and this reduction was abolished when catalytic mutants E62A and E62Q were tested (Fig. 6b). These findings indicate that opossum A1 inhibits MusD retrotransposition in a deaminase-dependent manner.
Opossum A1 mRNA is widely expressed in primary tissues in vivo. A1 mRNAs from rabbit 18 , mouse 18,28 , and rat 28,29 are broadly expressed in primary tissues such as the small intestine, spleen and gonads including both testis and ovary. In contrast, the expression of human A1 mRNA is largely limited to gastrointestinal tissues 30,31 . We quantified the expression level of opossum A1 mRNA in several primary opossum tissues. As anticipated, opossum A1 mRNA was abundantly expressed in the small intestine (280-fold compared with the ovary), where it appears that A1 edits its apoB mRNA (Fig. 7). Interestingly, opossum A1 mRNA was widely expressed in several tissues including the testis (22-fold) and lymphoid tissues (thymus, 4.1-fold; spleen, 3.3-fold) (Fig. 7). This expression pattern appears to be different from that of human A1, which was not reported to be expressed in the testis and lymphoid tissues 30,31 .

Discussion
A3 enzymes are specific proteins of eutherian mammals that are involved in the innate immunity pathways directed against retroviruses and retrotransposons 4-8 . A3 homologs do not appear to exist in non-eutherian mammals such as the opossum 21,22 (Fig. S1b). In this study, we used a series of cell culture studies to examine whether opossum A1 has activity against retroviruses and LTR/non-LTR retrotransposons. Our results show that opossum A1 inhibits retrovirus replication and MusD retrotransposition in a deaminase-dependent manner (Figs 2, 4 and 6). This enzyme also suppresses human L1 retrotransposition in a deaminase-independent manner (Fig. 5). These results suggest that A1 plays role in protecting marsupial host genomes from parasitic genetic elements. Despite potent mutagenic activity in bacteria (Fig. 1), opossum A1 showed modest HIV-1 restriction activity in HEK293T cells (Fig. 2). This relatively weak phenotype may reflect the predominantly nuclear localisation of opossum A1 in comparison to the more potent and predominantly cytoplasmic rabbit A1 enzyme (Figs 3 and S3). These results are consistent with our prior studies reporting that the degree of nuclear localisation of A1 correlates inversely with its antiviral activity 12 . It is likely that the nuclear distribution of opossum A1 prevents interaction with the Gag nucleocapsid protein, which in turn reduces encapsidation into virions. Alternatively, the physiological localisation of opossum A1 could differ from its localisation when overexpressed experimentally because endogenous A1 levels are below the limit of detection with currently available antibodies 32 . Furthermore, the nuclear localisation of A1 enzymes from different species appears to be influenced by cell-type 33 , suggesting that cellular factors regulate A1 localisation. A1s are expected to be more active against HIV-1 when they are in the cytoplasm. It must be noted that a chimeric human A1 fused to the rabbit homologue redistributed to the cytoplasm from the nucleus, was packaged more efficiently, and subsequently inhibited HIV-1 infectivity more effectively than the original human A1 protein 12 . Further studies, including the development of more sensitive antibodies and the identification of cellular factors that regulate the subcellular localisation of A1 will clarify the correlation between the subcellular localisation of A1 and its anti-HIV-1 activity. Our cell-based assays did not use opossum cells and, to our knowledge, vectors have yet to be developed for opossum viruses and mobile elements. Nevertheless, the results presented here suggest that A1 likely functions as an innate barrier to infection by retroviruses and controls LTR/non-LTR retrotransposition in marsupials. First, opossum A1 possesses restriction activity against several different retroviruses (Figs 2 and 4) and LTR/non-LTR retroelements (Figs 5 and 6). Second, ~50% of the opossum genome is composed of repetitive elements (LINEs: ~29%, SINEs: ~10%, endogenous retroviruses: ~11% and DNA transposons: ~2%) 21 . Of note, the proportion of LINEs in the opossum genome is approximately 1.5 times higher than that in human and mouse genomes (~20% LINE). The greater anti-L1 activity of opossum A1 in comparison to anti-retroviral activity might be associated with the relatively higher proportion of LINEs in the opossum genome. Third, A1 mRNA is expressed in many different opossum tissues (Fig. 7), which is distinct from the gastrointestinal expression profile of A1 in human tissues 30,31 . This broad expression profile suggests that opossum A1 has additional functions besides editing of apoB mRNA. Future investigations should include the development of an infectious molecular clone using viruses that naturally infect opossum and assay systems using opossum cells, to determine whether opossum A1 is indeed a bona fide anti-viral restriction factor.

MusD retrotransposition
One of the common functions among AID/APOBEC family proteins is restriction of non-LTR retrotransposon, L1, using a mechanism that is independent of enzymatic activity 17,18,[34][35][36][37][38] . However, it has been proposed that the mechanism by which L1 retrotransposition is suppressed by A1 is different from the deamination-independent mechanisms used by A3 enzymes 17 . In contrast to human A3s that are able to restrict human L1 retrotransposition even in the absence of a L1-encoded ORF1 protein 39 , A1s from human and lizard require the binding to L1 ORF1 protein for the inhibition of human L1 retrotransposition 17 . Interestingly, human and mouse ORF1 proteins form ribonucleoprotein (RNP) complexes with L1 RNA 40,41 , and the targeting of the L1 RNP complexes into cytoplasmic foci like stress granules is likely to control L1 retrotransposition 42,43 . It is noteworthy that human L1 RNA is found in A1 RNP complexes when they are cotransfected into HEK293T cells 18 . These observations suggest that the formation of A1 RNP complexes could sequester L1 RNA and/or ORF1 protein from its appropriate trafficking and translation, and ultimately could interfere with its retrotransposition into genome. This is also in contrast to A3s that are not necessary to form intracellular RNP complexes for L1 inhibition 36 . It is unclear what steps of L1 replication the A1 and A3 enzymes affect using deaminase-independent mechanisms. Elucidation of this deamination-independent repressive activity of APOBECs on L1 retrotransposition may provide important insights into the continuous arms race between viruses and their hosts.
It has been demonstrated that A1s from amniotes, including eutherian mammals, inhibit L1 retrotransposition in vitro 17,18 , and that these A1s also function as DNA mutators 18,44 . However, although A1 is expressed in the lizard, no deamination at the site corresponding to mammalian apoB mRNA editing has been observed 44 , suggesting that RNA editing may not be the ancestral function of A1. Because opossum A1 has RNA editing activity against apoB mRNA 23 , it appears that at least the A1s of mammals acquired the ability to edit apoB mRNA after the divergence of this group from the common ancestor they shared with other vertebrates (i.e., amphibians and birds). Therefore, the most likely original function of A1 was probably DNA editing and protecting cells against mobile elements, because the A1s of amniotes, including mammals, maintain anti-L1 activity in vitro 17,18 .
Although several studies have suggested that non-human A1s (e.g., rabbit A1) may be involved in innate immune pathways 11,13,14,16,18,19 , it seems that the function of human A1 is limited to editing apoB mRNA in the small intestine 30 , and that the functions of A1 in innate immunity have been taken over by the expansion of A3s in the primate lineages such as in humans. Further studies are needed to shed light on the evolutionary history of RNA editing by the A1 deaminases and on their ancestral and physiological functions other than apoB mRNA editing.
Cloning of opossum A1 cDNA. Production, maintenance and experimental manipulations of the animals were approved by the Institutional Animal Care and Use Committee of the Texas Biomedical Research Institute (John L. VandeBerg; previous affiliation). All procedures were carried out in accordance with the approved guidelines. Small intestine has been removed aseptically from euthanized Monodelphis Domestica (opossum). Total RNA was isolated using TRIzol reagent (Invitrogen), and then treated with DNase I (Takara). cDNA encoding the opossum A1 gene was synthesised using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) with a primer set designed so that a single HA epitope tag was attached to the C-terminus of the protein as previously described 23 . (5′ -NNNNNGATATCGAAGCCATGAATTCTAAGACAGGTCCA-3′ and 5′-NGCGGCCGCTCAAGCGTAATCTGGAACATCGTATGGGTATCTCCAGGTCACAAATGGCTGG-3′ , Restriction enzyme sites are underlined). The amplified fragment was cloned into the pCR-Blunt vector (Invitrogen) and sequenced. The amplified product was then inserted into the EcoRV and NotI sites of the pCAGGS expression vector and sequenced. The nucleotide sequence of an open reading frame encoding opossum A1 was identical to that enrolled in GenBank (NM001032982). The catalytic mutants of the opossum A1 were constructed using the oligonucleotide primers. E62A: 5′ -TCTCAACATGCTGCAATCAACTTCATG-3′ and 5′-CATGAAGTTGATTGCAGCATGTTGAGA-3′ . E62Q: 5′ -TCTCAACATGCTCAAATCAACTTCATGG -3′ and 5′-CCATGAAGTTGATTTGAGCATGTTGAGA-3′ . The amplified products were inserted into the pCAGGS vector at the EcoRV and NotI sites and sequenced. Bacterial expression plasmids encoding C-terminal HA-tagged opossum A1 and its catalytic mutants (E62A and E62Q) were created with the following primer set: 5′ -NNNNCTCGAGATGAATTCTAAGACAGGTCCATCAGTAGGAGA-3′ and 5′ -NNNATGCATTCAAGCGTAATCTGGAACATCGTATGGGTA-3′ . The amplified products were then cloned into the pTrcHis A vector (Invitrogen) with restriction enzymes XhoI and NsiI and sequenced.
Immunoblot analysis. Cell and virion lysates were resolved with sodium dodecyl sulphate (SDS) polyacrylamide gel electrophoresis (PAGE), transferred to a polyvinylidene difluoride membrane (Millipore), and blocked with 4% milk in phosphate-buffered saline (PBS) containing 0.1% Tween 20. The membranes were then incubated with a primary antibody, a biotin-conjugated secondary anti-mouse or anti-rabbit IgG antibody (Sigma) and streptavidin-conjugated horseradish peroxidase (Sigma). The signals were visualised with Chemi-Lumi One (Nacalai Tesque) and a VersaDoc 5000 Imager (Bio-Rad).

Subcellular localisation.
An immunofluorescence analysis was performed as previously reported 12,18 .
Briefly, 2 × 10 4 HEK293T or HeLa cells were plated in each well of an eight-well Lab-Tek Chamber Slide (Nalge Nunc International). The cells were transfected with the pCAGGS expression vector (500 ng) encoding either HA-tagged rabbit A1, opossum A1, or its catalytic mutants (E62A or E62Q), using FuGENE HD (Roche). After The intensity of FITC signal overlapped with the portion of DAPI staining was measured to calculate the proportion of nuclear distribution of A1. Then, the proportion of cytoplasmic distribution of each A1 was calculated by subtracting the nuclear FITC signal from the total FITC signal. The densitometric analysis of the blotting bands was quantified using an image analysis system.

Preparation of nuclear and cytoplasmic extracts. The nuclear and cytoplasmic extracts from
HEK293T were prepared with the Subcellular Protein Fractionation Kit for Cultured Cells (Thermo Fisher Scientific), according to the manufacturer's protocol.
Bacterial mutator assay. The generation of bacterial expression plasmids encoding cDNAs encoding HA-epitope-tagged opossum A1 and its derivatives is described above. The Escherichia coli-based DNA mutation assay was performed as described previously 12,18 . The uracil DNA glycosylase (ung)-deficient E. coli strain BW310 was transformed with the parental pTrcHisA plasmid or vectors encoding the various A1 cDNAs. Twenty colonies of the transformed bacteria, selected on plates containing ampicillin, were cultured overnight at 37 °C in 2 mL of LB medium containing ampicillin and 1 mM IPTG. The saturated cultures (100 μ L) were then plated on LB agar containing 100 μ g/mL rifampicin. The total number of Rif R colonies per plate was counted after 24 h. The appropriate dilution of viable cells was plated onto an LB plate containing ampicillin and the mutation frequencies were calculated as Rif R colonies per viable cell. To verify protein expression, 100 μ L of each saturated IPTG-induced culture was lysed and subjected to immunoblotting, as described above.
Single-cycle infection assays. HEK293T cells (2.5 × 10 5 ) were cotransfected with 0.75 μ g of Vif-proficient NL4-3 Luc E − R − proviral DNA construct, together with 0.5 μ g of pVSV-G vector and 0.25 μ g of the control vector or one of several expression vectors encoding C-terminal HA-epitope-tagged A1 protein supplemented with the appropriate concentration of the control vector, using Effectene ® (Qiagen). For SIVs, 0.75 μ g of Vif-proficient SIVmac or SIVagm Luc E − R − proviral DNA construct was cotransfected into HEK293T cells, with 0.5 μ g of pVSV-G vector and 0.25 μ g of the control vector or HA-tagged A1 expression vector. MLV-GFP virus stock was produced as previously described 50,51 . HEK293T cells were cotransfected with 0.3 μ g of MLV genome expressing a GFP reporter M3P-GFP, together with 0.3 μ g of MLV packaging construct pMD-MLV-OGP, 0.1 μ g of pVSV-G and 0.3 μ g of the control vector or HA-tagged A1 expression vector. After 48 h, the virus-containing supernatants were filtered with 0.45 μ m filters and frozen as aliquots. The p24 or p27 content of the viruses was determined with a HIV-1 p24 or SIV p27 antigen ELISA kit (ZeptoMetrix). Fresh target HEK293T cells were infected with the equivalent amounts of the luciferase reporter viruses and after 48 h, the intracellular luciferase activity of the infected cells was measured, as previously described 11,12 . Infectivity of MLV-GFP virus was analysed by flow cytometery. To assess the packaging of the A1 proteins into viral particles, the virus-containing supernatants were spun through a 20% sucrose cushion and the viral pellets were solubilised in 2× SDS sample buffer. Equivalent amounts of p24 or p27 antigen from each solubilised sample were analysed with immunoblotting, as described above. Band intensities were analysed with the ImageJ software.
L1 retrotransposition assays. The EGFP-based human L1 retrotransposition assay was performed as previously described 18 . Briefly, 3 × 10 5 HEK293T cells were cotransfected with 1.5 μ g of pL1 RP -EGFP or the pIRESpuro vector (Clontech) and 0.5 μ g of the control vector or the respective A1 expression vector with the appropriate concentration of a control vector with Effectene. At 24 h posttransfection, the cells were selected with 1.0 μ g/mL puromycin and 7-9 days after selection, the EGFP expression resulting from retrotransposition was measured with flow cytometry. The neo R -gene-marked human L1 retrotransposition assay was performed by cotransfecting HeLa cells (5 × 10 5 cells) with 0.4 μ g of the respective A1 expression plasmid and 1.2 μ g of the human L1-neo reporter vector together with 0.4 μ g of pIRES-EGFP (Clontech) as the transfection control, using FuGENE, as previously described 18 . After 72 h, 5 × 10 5 cells were plated onto 100 mm dishes and selected with 0.75 mg/mL G418. At 12-14 days after selection, the resultant G418 R colonies were stained with crystal violet (Sigma) and counted. The retrotransposition frequency was calculated as the number of G418 R colonies/transfection efficiency (percentage of GFP + cells).
MusD retrotransposition assay. The murine MusD retrotransposition assay was performed as previously described 18 . Briefly, 5 × 10 5 HeLa cells were cotransfected with 0.2 μ g of the control vector or the respective A1 expression plasmid supplemented with the appropriate concentration of the control vector, together with 0.6 μ g of the neo r -based murine MusD reporter vector together with 0.4 μ g of pIRES-EGFP, using FuGENE. After 72 h, 5 × 10 5 cells were reseeded onto 100 mm dishes for G418 (1.0 mg/mL) selection, and the resultant G418 R colonies were counted. The retrotransposition frequency was calculated as described above.
Statistical analysis. The statistical analysis was performed with Student's t test. All data are the means and standard deviations (SDs) of three experiments, unless otherwise stated. *P < 0.05, **P < 0.01, ***P < 0.001.