The data presented here provide an interesting contrast between the synthetic versions of human and mouse retrotransposons, ORFeus
-Hs and ORFeus
-Mm. Our previous data showed that the generation of ORFeus
-Mm with optimized codons, which were presumably free of sequences that might hamper transcription, resulted in a highly active element with levels of retrotransposition that were as much as 200-fold higher than the native element [16
]. These were shown to be in part due to higher levels of mouse L1 transcripts [15
], and presumably correspondingly higher levels of protein products. However, when similar techniques were attempted in order to develop a highly active human retrotransposon, we were only able to increase levels of retrotransposition by a maximum of two to three times. Contrary to the findings in mouse L1 elements, the synthetic sequences did not increase human L1 protein and retrotransposition levels by the same margin.
Native L1 elements contain premature polyadenylation sites [15
] and cryptic splice sites [25
] that produce premature polyadenylated and spliced form L1 RNAs. These isoform RNAs could limit full-length L1 RNA production or compete for L1-encoded ORFs [25
]. In our recoding process, most or all of these signals were removed, and this probably contributed to the increased abundance of full-length L1 RNA (Figure , Figure ). Although the function of these signals in nature remains unknown, they are dispensable for L1 retrotransposition in tissue-culture assays.
One obvious reason that ORFeus
-Hs did not increase retrotransposition frequency by 200 times is that the native mouse L1 element (L1spa
) has much lower activity than the native human L1RP
]. In fact, codon optimization of both mouse and human L1 elements increased their retrotransposition abilities to a similar level (Table ). This could represent an upper limit of L1 retrotransposition that can be readily tolerated by tissue-culture cells and/or a rate-limiting step(s) during the process of retrotransposition. Elevated levels of L1 RNA/protein and shuttling between nucleus and cytoplasm may have a strong effect on the cell, perhaps overloading its full capacity to process L1-RNP retrotransposition intermediates [26
]. Consistent with this, we observed that cells overexpressing either ORFeus
-Mm or ORFeus
-Hs displayed considerably higher sensitivity to antibiotics than those transfected with native L1s. For example, HEK-293T cells transfected with ORFeus
-Hs grew more slowly at a concentration of 2 μg/ml puromycin than did cells overexpressing native L1RP
(see Additional file 1
, Figure S4). These results are consistent with studies that reported effects on L1 protein expression leading to high levels of double-strand breaks and/or apoptosis and/or cellular senescence [27
]. It is formally possible that codon optimization changes made in ORFeus
corrected a mutation(s) in a cis
element(s) that hampers retrotransposition efficiency of native L1, but results from both the study of Han et al
] on the mouse element and the present study on the human element show that L1 activity increases progressively as larger proportions of the native sequence are recoded, consistent with the reported elongation defect.
We noted that the three sets of L1 constructs driven by CMV only, CMV plus 5' UTR or the 5' UTR only had very different trends of retrotransposition frequency changes. This suggests that different promoters somehow produce RNAs that are of different 'quality'. One difference in quality is the structure of the RNA 5' end; the CMV promoter fragment also contains the 51 bp viral 5' UTR upstream of the L1 ORF1 AUG motif, whereas the native L1 promoter introduces the native 907 bp L1 5' UTR in its place. In the double-promoter construct, there are thought to be two transcription start sites, one identical to the native site, and one extended on its 5' end by the CMV 5' UTR. We have not directly examined the relative abundance of these two RNA forms. Thus, the 5' UTR sequences differ between the three types of elements, and it is possible that the interactions between the 5' UTRs and the rest of the RNA sequence influence retrotransposition efficiency.
Unexpected discrepancies in the increases in RNA abundance, protein abundance and retrotransposition frequency were noted between the various constructs, with RNA increasing much more dramatically than protein, and protein increasing more than retrotransposition frequency. This suggests that when comparing native versus synthetic RNA sequences, the latter either interferes with translational efficiency, decreases stability of the encoded protein, or both. The larger relative increase in RNA suggests that the primary effect of the codon optimization was to improve levels of full-length L1 mRNA. Because codon optimization is predicted to increase translational efficiency, it is surprising that protein levels actually decreased relative to RNA template abundance. It is formally possible that recoding leads to enhanced protein degradation. Native codon usage may provide signals for proper folding of the nascent protein [31
]. Alternatively, if the interaction between the RNA and the protein to form RNP intermediates is abrogated in the fully or mostly synthetic elements, any ORF1 protein that does not get incorporated into RNPs may become very unstable, potentially explaining the RNA-protein discrepancy observed here.
Perhaps most surprisingly, we found that changes in ORF2 sequences in chimera B had a significant effect on the expression of ORF1p (compare pLD225 and pLD165 in Figure ). This is consistent with models in which ORF2 protein, or the RNA sequence encoding it, might participate in the regulation of ORF1 protein translation or stability. Finally, although recoding of the human element did not produce an increase in retrotransposition frequency that was as dramatic as with ORFeus-Mm, ORFeus-Hs and its chimeric derivatives remain useful tools for L1 studies, and have the highest retrotransposition frequency of the available human L1s reported. Higher levels of full-length L1 mRNA and ORF1p provide a convenient and rapid marker for the early stages of the L1 replication cycle by various methods such as immunofluorescence and immunoprecipitation.