The L1 retrotransposition assay: A retrospective and toolkit

1. Process of retrotransposition and structure of products

One of the most useful variations on the vector used in the retrotransposition assay was the addition of a prokaryotic promoter and origin of replication to the mneoI cassette along with removal of a cryptic splice signal [26, 27]. This modification allowed for expression of the neo gene in prokaryotic cells, which caused those cells to be resistant to kanamycin. Therefore, new retrotransposition events containing the spliced, active neo gene could now be rescued in E. coli and the flanking sequence determined. This strategy of creating novel retrotransposition events in vitro became a fruitful area of research, with many studies tracing the structure of retrotransposition insertions, including the target sites for both endonuclease-containing and mutated elements, production of chimeric integrants and assays of the proportion of full-length insertions [26-31]. Hypothesis testing could also be conducted, either in cell culture or in animal models: for instance, to see whether the bias for AT sequences in the genomic DNA flanking preexisting L1 insertions applied equally to newly mobilized elements [25, 32].

Despite the many successes, these studies of insertion sites were inherently limited: only insertions that are longer than the selectable marker could be recovered. L1 retrotransposition often produces short 5′ truncated inserts, and these would be missed. In order to overcome this size bias, Babushok et al. [25] made use of a marker-less cassette in transgenic mice, where removal of the intron was detected purely by a PCR reaction, and insertion site flanking sequences were cloned using TAIL-PCR [33]. This construct allowed for the isolation of insertions as small as 400 bp.

Wei et al.[34] explored whether the neo cassette could be spliced and retrotransposed in trans if provided on a different plasmid from the L1 driver element. They found that retrotransposition was more efficient in cis, where both driver and retrotransposition reporter were provided on the same construct. Nevertheless, transfected L1s are able to drive the retrotransposition of sequences in trans as well, such as cellular genes co-transfected and then assayed by PCR for a loss of an intron sequence [12]. Recently, Garcia-Perez has investigated the formation of L1/snRNA fusion pseudogenes by pooling G418 positive colonies and amplifying products using primers specific for the U6 sequence [35].

In the traditional L1 retrotransposition assay, selection on the drug G418 results in the death of death of those cells that are not carrying de-novo retrotransposition events, with the sole survivors being cells that contain retrotransposition of the mneoI cassette. In order to more accurately trace the kinetics of retrotransposition against a background of retrotransposition negative cells, Ostertag et al. switched out the neo gene with an γ-globin intron-containing EGFP gene [24]. This variant allows for the easy measurement of retrotransposition frequency using a flow sorting of fluorescent cells.

Another development useful for investigating the process of retrotransposition was the creation of a vector which is a hybrid between a retrotransposon and a gutted adenovirus [36, 37]. Use of this gene delivery system demonstrated that retrotransposition could occur in infected primary cells in the absence of cell division. Single events per cell were the norm, with their frequency dependent on the copy number of the construct.

2. Regulation of Retrotransposition

The L1 retrotransposition assay has also proven to be a powerful tool to investigate the sequence requirements for both the repression and promotion of retrotransposition. L1 elements had been known to be cytosine methylated [38]; however, by methylating HpaII sites in the promoter of the L1 construct in vitro, Yu et al. conclusively demonstrated that DNA methylation suppresses retrotransposition [39]. Moreover, adding GAL4 DNA-binding sites upstream of the L1 element on the construct and co-transfection with the methyl-CpG binding protein MeCP2 confirmed that this suppression occurred using a canonical chromatin regulatory pathway. Another cellular strategy to regulate L1s makes use of the APOBEC3 family of cytidine deaminases: here, co-transfections were instrumental in determining that Apo3A and 3B suppressed L1 retrotransposition, although conflicting results were obtained for Apo3F for reasons that remain unclear [40-45]. Alu, a non-autonomous element, could be mobilized when co-transfected with the L1 construct; the retrotransposition of Alu, but not that of the L1 driver, was specifically blocked by Apo3G [41, 46]. Most recently, it was found that the AID (activation-induced deaminase) family of proteins, which are phylogenetically related to APOBEC, can also inhibit L1 retrotransposition; of note, this inhibition does not require deaminase function or DNA binding [47].

SiRNAs derived in vivo from double-stranded RNA antagonize the transcription of retrotransposons in a wide range of organisms, both plants and animals [48]. When it came to studying this phenomenon in mammalian cells, the cell culture assay proved once again to be central When L1 constructs were co-transfected with siRNAs to L1 sequence, retrotransposition was indeed reduced [49]. Subsequently, Yang et al. generated 5′ truncations of the L1 element that abrogated activity of the antisense promoter located ∼ +500 in the 5′UTR [50]. These mutations caused a modest increase in retrotransposition of the construct, which may be due to the loss of double-stranded RNA production.

L1 activity causes genomic instability, and the retrotransposition assay investigated the ways in which the cell responds to various insults and fights back. G418 positive colonies were increased in a line carrying a p53 mutation [51]; the authors of this study suggested that L1 activity may induce apoptosis in p53+ cells. L1 mobilization also precipitated DNA double-stranded breaks which attracted the histone variant γ-H2AX [52, 53], and led to recruitment of the DNA repair pathway [54, 55].

In addition to repression, the cell culture assay has been helpful in characterizing exogenous factors, both genic and environmental, that promote L1 retrotransposition. By cotransfecting Runx3 with an L1 construct carrying mutations at putative RUNX3 binding sites, Yang et al.[56] showed that this transcription factor is able to increase both expression and retrotransposition of the element. Moreover, mutating putative binding sites for the transcription factor YY1 resulted in changes in transcriptional initiation with little difference in retrotransposition [57]. Genotoxins such as γ-radiation and benzopyrene [58] and heavy metals (Hg, Ni, Cd) [59, 60] stimulated increased levels of retrotransposition in the cell culture assay.

3. Finding highly active naturally-occurring L1s

Potentially active L1 elements are occasionally found retrotransposed into protein-coding genes or as polymorphic elements in natural populations. One of the first and more popular ongoing uses of the retrotransposition assay has been to quantify the activity levels of newly isolated polymorphic L1s. In fact, Moran et al.'s [7] original study compared multiple full-length L1 elements; subsequent studies demonstrated that L1RP had a relatively high level of activity [61, 62]. Using flow cytometry to determine EGFP reporter activity, Brouha et al. isolated and measured the activity of another highly active element, LRE3 [63]. L1RP and LRE3, among other elements such as L1.3 [21, 62], remain popular L1 drivers for use in the retrotransposition assay.

The assay was later expanded to look at a large number of full-length L1s from the most evolutionarily recent family, Ta [4, 64, 65]. This survey allowed for an estimate to be made of how many L1s in the entire genome might be active [4]. In addition, taking this assay to naturally-occurring alleles of the same element demonstrated that different individual variants had different levels of activity – a “hot” L1 in one population might therefore not be as active in another [66, 67]. This use of the retrotransposition assay was not limited to testing human elements in human cell lines: mouse elements from different L1 subfamilies have been examined as well [68-70], in both mouse LTK- and human cell lines.

3. Structure/Function Studies

The retrotransposition cell culture assay can test not only natural alleles of L1 elements, but is also an ideal system for examining artificially-generated mutations in the protein-coding regions of the element. These structure/function studies have been central in deciphering the mechanism of retrotransposition. Early uses of this assay showed the importance of expression of L1-encoded ORF1 and ORF2, particularly the endonuclease activity of ORF2p, in effective retrotransposition [7, 71]. The effects of mutations within the endonuclease domain have been investigated more closely recently, including characterizing the structure of altered proteins and cloning out the insertion breakpoints of endonuclease-defective retrotransposition [29, 72].

Knowledge of ORF1p's role as a nucleic acid chaperone has especially benefited from this method [73, 74]. By epitope-tagging mutated versions of the element, Kulpa and Moran [75] were able to show that ORF1 was important for RNP complex formation. Other studies showed that the initial AUG and other intergenic sequences present upstream of ORF2 were dispensable for function [76, 77].

Alterations of the L1 element do not have to be deleterious; indeed, some changes can increase retrotransposition competency. Boeke's lab has synthetically engineering a novel L1 element through codon optimization of mouse L1_Spa [78]. This element, named ORFeus, is 24% divergent but two hundred-fold more active than its L1_Spa element progenitor. An optimized human element has been developed as well which is somewhat more active than its progenitor element [54, 79].

4. Animal Models

Although most research using the retrotransposition assay has been carried out in cell culture (e.g. human HeLa), whole animal transgenics have become a burgeoning area of innovation. Whole animal retrotransposition assays can be used to both determine the stage and tissue-specificity of retrotransposition in vivo and as a mutagenesis tool for gene discovery. In 2002, Ostertag et al. created mice carrying an EGFP-tagged human L1 transgene [80]; these animals expressed green fluorescence upon mobilization of the retrotransposon. Integrated transgenes continued to drive low rates of retrotransposition in the germline of the descendants of the founder animals. A second-generation construct, using a highly active mouse RNA polymerase II large subunit promoter with CMV driving EGFP expression, was used to track retrotransposition visually through monitoring fluorescence in different mouse tissues [81]. These mice featured very little somatic L1 movement, with a strong signal only in the testis. Contrasting results were obtained a couple of years later, when retrotransposition was detected in rat neuronal precursor cells and in the brain of transgenic mice [82]. New data using human elements in mice and rats point to the likelihood that the bulk of L1 mobilization occurs in the early embryo rather than the germ lineage (Hiroki Kano, personal communication). In fact, the codon-optimized, CAG (chicken actin promoter)-driven ORFeus element is retrotranspositionally active in both germline and somatic tissues [32].

A recent development has been the possibility of using L1 as an insertional mutagen for gene discovery [32, 83]. Transposon-tagged genes can be easily cloned based on sequences flanking the element, giving this form of random mutagenesis a distinct advantage over methods based on chemicals such as ENU (ethylnitrosourea). An and Boeke have created an L1 ORFeus construct in which the promoter is separated from the transgene by a β-geo (β-galactosidase/neomycin fusion) cassette flanked by LoxP (locus of X-over P1) sites. Mating mice transgenic for this construct to those expressing the Cre (cyclization recombination) protein results in excision of this cassette and activation of the retroelement in the offspring. The LoxP transgene has been successfully combined in this manner with Pdx1 (pancreatic and duodenal homeobox 1)-Cre and Zp3 (zona pellucida glycoprotein 3)-Cre to drive retrotransposition specifically in the pancreas and female germ cells, respectively [84].

1. Traditional

In the original retrotransposition assay [7], HeLa cells are seeded at 2 × 10⁵ to 4 × 10⁵ cells/well in 6-well plates and grown in DMEM-complete media to 70% confluence. Transfection is performed with 1 μg plasmid DNA and 4 μl lipofectamine reagent (BRL) in 1 ml of Opti-mem (BRL). After 5 hours at 37°C and 7% CO₂, the cultures are supplemented with 1 ml of DMEM containing 20% FCS. This medium is again replaced after 16 hours. Two days later, cells can be trypsinized and seeded into 75 cm² flasks and grown in 200 μg/ml hygromycin to select for the presence of the vector which encodes hygromycin-resistance. (Vectors encoding puromycin instead of hygromycin resistance are selected using 5 μg/ml puromycin.)

Cells are grown under hygromycin or puromycin selection for 10-14 days, after which they are trypsinized, counted with a hemocytometer and re-seeded at known densities in media supplemented with 300-400 μg/ml G418. This procedure selects for cells that have undergone retrotransposition and now express the neo selectable marker. 10-14 days later, cells are fixed and stained with 0.4% Giemsa. Retrotransposition efficiency is calculated based on the number of cells before and following G418 selection. Refer to Moran et al.[7] for more details on this protocol. Kill curves are suggested to optimize the dose and time of the antibiotic selections.

2. Transient Assay

In the transient variation, two sets of transfected cells are plated in parallel. One set serves as a transfection efficiency control, the denominator for retrotransposition efficiency. The second set is grown on G418 to select for cells in which the intron within the neo gene has been removed by splicing and the L1 construct has integrated into the genome by retrotransposition. More information on this assay can be found elsewhere [100]. A brief description of a protocol for HeLa cells follows below. This protocol must be optimized for different cell lines; the reader is encouraged to refer to publications using those cell lines for more details.

HeLa cells are plated at a density of 2×10⁵ per well of a 6-well tissue culture dish. 3 μl of Fugene non-liposomal transfection reagent (Roche) is used for transfection along with 0.5-1.0 μg of plasmid and equal amounts of pGreen Lantern reporter plasmid as a measure of transfection efficiency. Cells are grown for three days and then trypsinized. Flow cytometry is used to measure the amount of green fluorescence. Wells that were not subject to trypsinization (test wells) are exposed to 400 μg/ml G418 for 12 days. G418 positive foci are fixed in 2% formaldehyde, 0.2% glutaraldehyde and 1 × PBS at 4°C for 30 minutes and then stained with 0.1% Bromophenol blue or 0.4% Giemsa for easy visualization. In parallel, single colonies can be propagated and removed for DNA isolation and PCR or Southern blot analysis.

This protocol can be scaled to 75 or 175 cm² tissue culture flasks, using proportionately greater quantities of cells, Fugene and DNA. For instance, 2×10⁶ HeLa cells, 24 μl Fugene and 8 μg DNA are suggested for use in 75 cm² flasks.

3. EGFP selection

Transfections using constructs featuring an EGFP retrotransposition marker were not subjected to G418 selection and are instead examined for fluorescence. Following hygromycin or puromycin selection as described above (approximately day 8 post-transfection), cells are washed with 2 ml 1 × PBS and then incubated in 5 mM EDTA in PBS. The retrotransposition-defective JM111 construct [7] can be transfected in parallel to transfections with the L1-containig construct and used as a negative control to determine background levels of fluorescence. The cell suspension is transferred through cell strainer caps (Falcon) to polystyrene tubes and kept on ice. This solution is analyzed with a FACScan machine (Becton Dickinson) with a blue argon 488nm laser and fluorescein filters of 530.30 bandpass. Typically, between 10,000 and 20,000 events are analyzed per sample. Live-dead gating is performed based on the forward scatter versus side scatter profile. Analysis of living cells for fluorescence intensity can be conducted using the CellQuest software package. An arbitrary threshold of intensity such as <0.05% can be used as a value for background fluorescence, with the proportion of JM111 transfected cells fluorescing above this value set as the false positive rate. Refer to Ostertag et al.[24] for more details on this protocol.

4. Constructs

Traditionally, most constructs used for retrotransposition feature a highly active naturally-isolated L1 element, such as L1RP or LRE3 [7]. A neo gene cassette, mneoI, is disrupted by a γ-globin intron in the opposite orientation [19]. A later variation used an EGFP cassette disrupted by this same intron [24]. Vectors allowing propagation in bacterial cells can be used for easy recovery of novel insertion sites in E. coli [26, 27, 31]. A codon-optimized mouse element, ORFeus, is advantageous for gene-trapping and Cre-Lox conditional activation [32, 78, 84]. An alternative strategy for transfection utilizes a helper-dependent adenovirus backbone for retrotransposon delivery [36, 37]. Other variations include an ORF1 epitope tag [75], an RNAi cassette [85], and the use of a group I self-splicing intron in the retrotransposition marker [22]. There are also versions that include a blasticidin antibiotic selection marker [23] and even no selectable marker [25]. Refer to Figure 1A for a diagram of a canonical construct and to Table 1 for details and complete references.

We thank John Goodier for impassioned discussions and critiques. Research in the Kazazian lab is funded by the National Institutes of Health (NIH).