Mammalian genomes are littered with enormous numbers of transposable elements interspersed within and between single-copy endogenous genes. The only presently spreading class of human transposable elements comprises non-LTR (long terminal repeat) retrotransposons, which cover approx. 34% of the human genome. Non-LTR retrotransposons include the widespread autonomous LINEs (long interspersed nuclear elements) and non-autonomous elements such as processed pseudogenes, SVAs [named after SINE (short interspersed nuclear element), VNTR (variable number of tandem repeats) and Alu] and SINEs. Mobilization of these elements affects the host genome, can be deleterious to the host cell, and cause genetic disorders and cancer. In order to limit negative effects of retrotransposition, host genomes have adopted several strategies to curb the proliferation of transposable elements. Recent studies have demonstrated that members of the human APOBEC3 (apolipoprotein B mRNA editing enzyme catalytic polypeptide 3) protein family inhibit the mobilization of the non-LTR retrotransposons LINE-1 and Alu significantly and participate in the intracellular defence against retrotransposition by mechanisms unknown to date. The striking coincidence between the expansion of the APOBEC3 gene cluster and the abrupt decline in retrotransposon activity in primates raises the possibility that these genes may have been expanded to prevent genomic instability caused by endogenous retroelements.
- apolipoprotein B mRNA editing enzyme catalytic polypeptide 3 (APOBEC)
- intracellular defence
- intrinsic immunity
- long interspersed nuclear element 1 (LINE-1)
Destructive and constructive effects of LINE (long interspersed nuclear element)-1 activity
All vertebrate genomes include large numbers of retroelements which can be divided into three classes: exogenous retroviruses, retrotransposons containing LTRs (long terminal repeats) and retrotransposons without LTRs, named non-LTR retrotransposons. Only members of the class of non-LTR retrotransposons are currently spreading in the human genome, and they can be subdivided further into autonomous LINEs and non-autonomous non-LTR retrotransposons, such as SINEs (short interspersed nuclear elements), SVAs [named after SINE, VNTR (variable number of tandem repeats) and Alu] and processed pseudogenes (Figure 1A). Alu elements represent a highly active family of human SINEs and constitute ∼11% of the entire genome . LINE-1, the most common and only active human LINE, is named autonomous because it has coding capacity for the protein machinery required for its own transposition. In contrast, non-autonomous non-LTR retrotransposons lack any protein-coding capacity and have evolved refined parasitic strategies to efficiently bypass the cis-preference of LINE-1 for their own retrotransposition. Therefore more than one-third of the human genome is the result, directly or indirectly, of LINE-1 retrotransposon activity.
LINE-1-mediated retrotransposition can occur in germ cells and/or during early embryogenesis, before the germline becomes a distinct lineage [2,3]. Although retrotranspostion events also occur occasionally in specific somatic cell types, only those events that take place in cells destined for the next generation are evolutionarily successful. LINE-1 elements continuously alter the genome structure in many ways, both destructive and constructive, and their activity can lead to genomic instability and sometimes cause human disease [4–7]. LINE-1s and Alus can be destructive by causing insertions and by providing material for homologous recombination. The average human diploid genome has 80–100 active LINE-1s . At least 1 in every 50 humans has a new genomic LINE-1 insertion that occurred in parental germ cells or in early embryonic development [2,3,9]. LINE-1 insertions account for approx. 1 in 1200 human mutations, some of which cause disease [10–12]. LINE-1s cause genome instability by generating substantial deletions during the LINE-1 integration process [4,5]. They have a marked cis preference, whereby their proteins greatly prefer to act on the RNA that encodes them . Nevertheless, LINE-1s are still able on occasion to mobilize non-autonomous sequences in trans [14,15]. Trans-mobilization of Alu elements resulted in the expansion of Alu sequences to 1.1 million copies in the human genome. Alu insertions have accounted for over 20 cases of human genetic disease, and Alu retrotransposition events occur in at least 1 in every 30 individuals . Processed pseudogenes and SVA elements are two other non-autonomous retrotransposons that are probably mobilized by human LINE-1s [15,16].
Nevertheless, there are several effects of LINE-1 activity that could also be constructive for the host organism. First, LINE-1s occasionally repair double-strand breaks in DNA by inserting into the genome via an endonuclease-independent pathway . Secondly, LINE-1 can retrotranspose non-LINE-1 DNA from its 3′ flank to new genomic locations in a process termed 3′ transduction . Thirdly, retrotransposons provide their sequences for a number of protein-coding exons of genes. In the human genome, LINE-1 or Alu sequences are present in nearly 200 confirmed and 2400 predicted protein-coding sequences . Fourthly, LINE-1 retrotransposition can produce new chimaeric retrogenes that are probably generated through template switching of LINE-1 RT (reverse transcriptase) from LINE-1 RNA or Alu RNA to other small nuclear RNAs . Fifthly, LINE-1 retrotransposons can also affect gene expression. They contain an antisense promoter in their 5′-UTR (untranslated region), and a number of expressed genes located 5′ to full-length LINE-1s have alternative transcription start sites in this LINE-1 region . Also, intronic LINE-1 insertions in either orientation can affect the RNA production of endogenous genes, both qualitatively and quantitatively .
Owing to their deleterious effects, mobile genetic elements have been in conflict with host genomes for over a billion years. Our own genomes reveal the remarkable effects of retrotransposition, as approx. 45% of our genomic DNA results directly from this process . This perennial state of conflict has led eukaryotes to adopt several strategies to curb the proliferation of transposable elements and viruses. These include transcriptional silencing through DNA and histone methylation or RNA interference, and even directed mutagenesis of mobile elements.
Several host-encoded proteins have been identified to influence LINE-1 transcription [23–27], and it has been suggested that the human LINE-1 element is susceptible to RNA interference [28,29]. However, the modulating potential of these factors does not sufficiently explain the well-controlled cell-type-specific LINE-1 expression and retrotransposition observed in mammals. Therefore additional host-encoded strategies were postulated to keep LINE-1-mediated retrotransposition in check and modulate the transposition frequency so that it is not harmful to the host.
The APOBEC (apolipoprotein B mRNA editing enzyme catalytic polypeptide) 3 protein family
The ability of human APOBEC3 protein family members to function as innate inhibitors of a wide range of exogenous retroviruses was first noticed during studies analysing the HIV-1 Vif protein [30,31]. APOBEC3G (hA3G), one member of the APOBEC3 protein family, was shown to be a potent inhibitor of Vif-deficient replication. In the absence of Vif, hA3G is specifically packaged into progeny virion particles and then interferes with reverse transcription during subsequent infections [30,31]. hA3G is a CDA (cytidine deaminase) that edits dC residues to dU on nascent DNA-minus strands during reverse transcription [32–34]. This activity induces extensive mutagenesis of the HIV genome and was suggested to destabilize incomplete reverse transcripts.
APOBEC3 proteins belong to a family of CDAs with diverse functions . This family was named after the first of these proteins to be discovered, APOBEC1. All proteins in this family have one or two CDA motifs comprising the sequence His-Xaa-Glu-Xaa23–28-Pro-Cys-Xaa2–4-Cys (Figure 1B). Catalysis results in hydrolytic deamination at the C4 position of the cytosine base, thereby converting cytidine (C) into uridine (U) in a process called ‘editing’. In humans, the APOBEC family includes AID (activation-induced CDA), APOBEC1, APOBEC2, APOBEC3A–APOBEC3H (hA3A–hA3H) (Figure 1B) and APOBEC4. The APOBEC3 family has expanded from one gene in the mouse to as many as seven genes in primates . APOBEC3 proteins can confer innate immunity on a wide range of exogenous retroviruses. Individual APOBEC3 members can also block the replication of hepatitis B virus, and inhibit the replication of endogenous retroviruses . APOBEC3 genes are expressed in almost every human tissue, cell type and cell line analysed so far (Table 1). The genes encoding APOBEC3 family members are clustered on chromosome 22 . During mammalian evolution, this locus expanded from a single gene in mice to seven genes (hA3A–hA3H) in primates [35,36] (Figure 1B).
Inhibition of LINE-1 retrotransposition
Effects of APOBEC3 proteins on human LINE-1 retro-transposition were assessed applying LINE-1 retrotransposition reporter assays [37,38] in human HEK-293T (human embryonic kidney), BTK-143 osteosarcoma and HeLa cells [39–44]. Co-expression of LINE-1 reporter constructs with APOBEC3 proteins in these cells revealed that hA3A, hA3B and hA3F are effective inhibitors of LINE-1 retrotransposition causing a reduction of LINE-1 transposition frequency by 85–99, 75–90 and 70–80% respectively (Figure 2) [39,42–44]. Notably, the hA3F protein used in those experiments differs at two amino acid residues (108 and 231) from the hA3F protein used in two additional studies, demonstrating only a negligible reduction of LINE-1 transposition by ∼20%  and even an increase by 30–75% . This suggests that both amino acid residues are critical for the LINE-1-inhibiting activity of hA3F. There is also the possibility that variations in the retrotransposition assay used in the different studies account for the differences which are also observed in the case of hA3C and hA3G. Although two studies report a reduction of LINE-1 transposition by ∼75% in the presence of hA3C [43,44], a third study demonstrates an inhibition by only ∼40% .
In the case of hA3G, several reports show a minor decrease by 12–20% [42–44] or no effect all . Two other studies describe an inhibition by ∼40%  and a negligible increase by ∼10%  respectively. Taken together, hA3A, hA3B, hA3C and hA3F are major inhibitory factors of human LINE-1 retrotransposition, whereas hA3G, hA3H and hA3D have only a negligible impact on LINE-1 mobilization (Figure 2). Although initial reports on the inhibition of mobility of diverse retroviruses and retrotransposons by APOBEC3 proteins proposed that these effects were due to the DNA-editing capabilities of these enzymes, there is no evidence for editing-dependent inhibition of non-LTR retrotransposon mobility so far. Non-LTR retrotransposon RNA is reverse-transcribed at the genomic integration site through a process known as target-primed reverse transcription  and thus takes place in the nucleus. The identification of hA3A, hA3B and hA3C proteins in the nucleus (Figure 1B) [39,42,43] would therefore be consistent with the hypothesis of LINE-1 inhibition by DNA editing, especially because these proteins have been shown to function as active CDAs on other substrates [42,44,48]. However, although LINE-1-inhibiting activity of hA3A was shown to be dependent on an intact CDA site , inhibition by hA3B was independent of CDA activity . Also, mutations in DNA sequences that are retrotransposed in the presence of any of the four LINE-1-inhibiting APOBEC3 proteins (Figure 2) could not be detected [39,42,43]. It is possible that the single-stranded nucleic acid substrate is not available for editing during target-primed reverse transcription or that edited transcripts are rapidly degraded and are not available for reverse transcription and integration into the genomic DNA. The cytoplasmic localization of hA3F is consistent with a LINE-1-inhibiting mechanism that does not interfere with reverse transcription.
Inhibition of Alu retrotransposition
Retrotransposition of Alu elements is mediated by the LINE-1 ORF 2p [ORF (open reading frame) 2 protein] (Figure 1A) , which has RT and endonuclease activities , but does not require the LINE-1 ORF1 RNA-binding protein . hA3A, hA3B and hA3G acted as potent inhibitors of Alu retrotransposition causing a 75–90% reduction of transposition frequency regardless of whether ORF2p alone or both ORF1p and ORF2p were expressed [42,45,50].
The inhibition of Alu retrotransposition by hA3G, a protein that does not affect LINE-1 retrotransposition, is surprising [45,50]. The fact that an active CDA site is not required for restriction of Alu retrotransposition by hA3G implies that at least hA3G clearly inhibits Alu retrotransposition independent of hypermutation [45,50]. It has been suggested that hA3G achieves this inhibition by sequestering Alu RNA in cytosolic ribonucleoprotein complexes, where it is unavailable for targeting by the LINE-1 RT and thus cannot be reverse-transcribed . This would imply that the inhibitory mechanism inflicted by hA3G is not only distinct from editing, but also differs from the hA3A- and hA3B-mediated inhibition that has been proposed to target Alu elements indirectly by inhibiting the LINE-1 transposition machinery. This possibility seems to represent one example for a mechanism of editing-independent inhibition of retroelement mobility.
Taken together, evidence is accumulating that the APOBEC3 family of CDAs can inhibit retroelements by several mechanisms, including editing-dependent and editingindependent processes. It is striking that mice which are equipped with only one single APOBEC3 gene carry 50–60 times more active LINE-1 elements in their genomes compared with humans and that the proportion of LINE-1 insertions causing mouse disease is also approx. 35-fold higher . The observation that the abrupt decline in retrotransposition activity in primates coincides with the expansion of the APOBEC3 gene cluster during evolution implies an essential role of APOBEC3 proteins in the protection of host genomes from invading mobile DNA.
I thank Carsten Münk for comments on this manuscript. Work in my laboratory is supported by the Deutsche Forschungsgemeinschaft (DFG).
The Testis as a Conduit for Genomic Plasticity: Independent Meeting held at University of Leeds, U.K., 15–18 November 2006. Organized and Edited by M. Brinkworth (Bradford, U.K.), J. Cummins (Murdoch University, Australia), S. Krawetz (Wayne State University, U.S.A.), D. Miller (Leeds, U.K.) and C. Spadafora (Institutio Superiore di Sanità, Italy).
Abbreviations: APOBEC, apolipoprotein B mRNA editing enzyme catalytic polypeptide; CDA, cytidine deaminase; hA3A etc., human APOBEC3A etc.; LINE, long interspersed nuclear element; LTR, long terminal repeat; ORF, open reading frame; ORF1p etc., ORF1 protein etc.; RT, reverse transcriptase; SINE, short interspersed nuclear element; UTR, untranslated region; VNTR, variable number of tandem repeats; SVA SINE, VNTR and Alu
- © 2007 Biochemical Society