Although human diseases of retrotransposition-derived etiology have been documented, retrotransposon RNA expression and the occurrence of retrotransposition events in the human oocyte are not studied. We investigated the RNA expression of L1 and HERV-K10 retrotransposons in human oocytes by RT–PCR analysis with designed primers. Using denucleated germinal vesicles (GVs), we detected RT–PCR products of expressed L1, HERV-K10 and, unexpectedly, SINE-R, VNTR and Alu (SVA) retrotransposons. Their transcript specificities were identified as such following RNA-FISH and their origin by cloning and sequence alignment analyses. Assessing the expression level in comparison with somatic cells by densitometry analysis, we found that although in normal lymphocytes and transformed HeLa cells their profile was in an order of L1 > HERV-K10 > SVA, remarkably this was reversed in oocytes. To investigate whether de novo retrotransposition events occur and reverse transcriptases are expressed in the human oocyte, we introduced in GVs either a retrotransposition active human L1 or mouse reverse transcriptase deficient-VL30 retrotransposon tagged with an EGFP-based retrotransposition cassette. Interestingly, in both the cases, we observed EGFP-positive oocytes, associated with an abnormal morphology for L1 and granulation for VL30, and the retrotransposition events were confirmed by PCR. Our results: (i) show that L1, HERV-K10 and SVA retrotransposons are transcriptionally expressed and (ii) provide evidence, for the first time, for retrotransposition events occurring in the human oocyte. These findings suggest that both, network of retrotransposon transcripts and controlled retrotranspositions, might serve important functions required for oocyte development and fertilization while the uncontrolled ones might explain the onset of genetic disorders.
Among retrotransposon families identified by the Human Genome Project, long interspersed nuclear elements-1 (LINE-1 or L1s) and human endogenous retroviruses (HERVs) represent 17 and 8% of the human genome (1,2), respectively.
L1s are non-LTR retrotransposons encoding proteins with RNA-binding, endonuclease and reverse transcriptase activities. Of all L1s, only a small number of 80–100 members are in full length constituting an active subpopulation, competent for retrotransposition events in human somatic cells (2). Remarkably, L1-encoded proteins are responsible for mobilization of Alu retrotransposons (3) and formation of processed pseudogenes (4), which together comprise ∼12% of the human genome (5). Importantly, L1-based recombination and L1-mediated retrotransposition events had a tremendous impact on genome evolution and several cases of de novo mutations are known (2).
Among HERV families, HML-2 group or HERV-K subfamily is classified by the presumed tRNALys primer required for reverse transcription. Among the HERV-K proviruses isolated, HERV-K10 with a typical LTR-gag-prt-pol-env-LTR structure is considered to be the prototype HERV-K genome (6). HERV-K family is of particular interest as it appears evolutionarily much younger (7) and active for particle synthesis observed in teratocarcinoma (8) and melanoma-derived cell lines (9).
Early studies showed preferential expression of L1 transcripts in undifferentiated cells (10), undifferentiated tumor cells (11) and various other cases of human cancer (12). Recently, it was shown that L1s self-suppression by an RNA interference (RNAi) mechanism is achieved through bidirectional transcripts originating from the L1 5′ UTR (13). As regards HERVs, it is known that numerous mutations accumulated in most proviruses have rendered them defective both in coding capacity and in transcriptional activity. HERV-K family is an exception in that some members bear intact ORFs (14) transcribed in a variety of human tissues (15) and enhanced transcriptional activity has been found in breast cancer (16) and germ cell tumors (17).
SINE-R, VNTR and Alu (SVA) is a small family of composite retrotransposons closely related to HERV-K10s (18). SVAs are active mobile elements, receiving research attention lately, whose de novo mobilization is known to cause human genetic disorders (19,20). They are non-autonomous retrotransposons with no reverse transcriptase coding capacity, probably mobilized in trans by active L1s (20); however, their RNA expression profile and role in the genome is not known yet.
Retrotransposons RNA expression is normally restricted to developing germ cells whereas in normal somatic ones their transcriptional activity is suppressed mostly by methylation (21). In support, methylation of L1s at the CpG-rich 5′ UTR results in transcriptional repression in somatic cells, whereas aberrant hypomethylation allows RNA expression in various tumors (21,22). Moreover, it was documented that demethylation following DNA methyltransferase 1 deletion triggered L1 expression (23). Transcriptional activity of HERV-Ks has also been shown to be directly regulated by CpG methylation (24). Conversely, in a hypomethylated status, the reactivation of RNA expression was found for both L1s and HERV-Ks in urothelial cancer (25), and many HERV families were transcriptionally active in placenta (26) characterized by lower DNA methylation.
Oocyte is characterized by methylation acquired during oogenesis by the temporal dynamics of expression of Dnmt3a, Dnmt3b and Dnmt3L methyltransferase (27). During oocyte growth, a large number of genes are transcribed and mRNAs produced are either translated immediately or stored as stable transcripts required for later use during the early stages of embryogenesis (28). Given the temporal hypomethylated status of the oocyte and the mutagenic nature of retrotransposition, an important question concerns the RNA expression levels of retrotransposons and whether retrotransposition events occur in the human oocyte. In this study, we have investigated the RNA expression of two retrotransposons, L1 and HERV-K10, at the germinal vesicle (GV) stage of oocyte development. We have shown that these retrotransposons, and unexpectedly, the HERV-K10-related retrotransposons SVA, are expressed, and that retrotransposition events occur in the human oocyte.
Specificity of designed retrotransposon primers
To study L1 and HERV-K10 retrotransposon RNA expression by RT–PCR in human oocytes, we designed two sets of primers based either on the ORF2 sequence of a retrotransposition active L1 member L1.3 (29) or LTR of HERV-K10 considered to represent the prototype HERV-K genome (6), respectively (Fig. 1A). The latter primers were expected to amplify mostly read-through transcripts originating from full-sized (12%) and truncated (5%) proviruses, as the vast majority of HERV-K sequences are solitary LTRs (83%) in the genome (30).
The functionality of the designed primers was tested by direct PCR using either plasmid DNA harboring L1.2B, a retrotransposition-competent L1 highly homologous to L1.3, as positive control or purified human WI-38T DNA. As seen in Fig. 1B (left panel), L1 primers produced one product of 948 bp, as expected, in reactions containing either L1.2B (lane P) or human DNA (lane 1) as a template. Using WI-38T DNA, HERV-K10 primers produced, also, the expected 606 bp product and an additional one of ∼240 bp representing possible HERV-K10-related sequences (lane 2). In contrast, using mouse DNA as a template, both sets of primers produced multiple bands (Fig. 1B right panel, lanes 2 and 1) with non-compatible molecular sizes to those expected documenting the species-specificity of the designed primers. These data showed that both sets of primers could be used in amplifying specific human retrotransposon sequences.
L1 and HERV-K10 retrotransposons are transcriptionally active in human oocytes
To investigate whether L1 and HERV-K10 retrotransposons were expressed in the human oocyte, we performed RT–PCR analysis with RNA prepared from denucleated GV oocytes. Initially using RNA from 20 denucleated oocytes, amplification with both sets of primers for 36 cycles yielded no products as a result of low template concentration (data not shown). Upon re-amplification for 36 additional cycles, one distinct 948 bp L1 product was detected while HERV-K10 reactions yielded an intense large smear of various molecular-sized products, showing that HERV-K10 transcripts were expressed at a higher level than that of L1s (Fig. 1C). To obtain distinct HERV-K10 products, we applied a total number of 45 amplification cycles under the initial conditions. We found that the aforementioned sequences of 606 and ∼240 bp were amplified, and interestingly, the 240 bp transcript was expressed at a higher level than that of 606 bp (Fig. 1D, left panel).
To assess the relative expression of L1, HERV-K10 and HERV-K10-related transcripts in somatic cells, we performed RT–PCR analysis with cDNAs prepared from isolated normal lymphocytes and HeLa cells. As seen in Fig. 1D (right panel), the expression of both retrotransposons and HERV-K10-related transcripts was higher in the lymphocytes than in the HeLa cells. By densitometry analysis, we found that L1s expression was 1.43- and 1.68-fold higher than HERV-K10 and 240 bp HERV-K10-related transcripts in the lymphocytes. Similarly, L1s expression was found 1.50- and 1.86-fold higher than HERV-K10 and 240 bp HERV-K10-related transcripts in the HeLa cells. Both analyses show an order of expression in the somatic cells such as L1 > HERV-K10 > HERV-K10-related transcripts. Interestingly, the ratio of HERV-K10-related and HERV-K10 transcripts (240/606 bp) for lymphocytes and HeLa cells was found at similar values of 0.85 and 0.80, respectively, indicating that their relative expression in somatic cells was constant. In contrast, the corresponding ratio for oocytes (Fig. 1D, left panel) was found at a value of 2.33, documenting a reversion of the retrotransposon expression order in the human oocyte. Collectively, these data show that L1, HERV-K10 and HERV-K10-related transcripts were expressed in the human oocyte, but in a reverse profile than that of somatic cells.
RT–PCR products detected in human oocytes are specific L1, HERV-K10 and SVA transcripts
The above RT–PCR analysis provided evidence for the RNA expression of L1 and HERV-K10 retrotransposons in the human oocyte. To ensure that the distinct RT–PCR products detected were of retrotransposon origin, and not derived from sequences producing similar mass sized products, we performed sequence analysis to determine their identity.
To this end, the aforementioned RT–PCR L1, HERV-K10 and HERV-K10-related products were purified, cloned into plasmid vectors and five clones per case were sequenced. We found that each group of sequences obtained were both highly homologous themselves (96–98%) and of human retrotransposon origin by comparison with known sequences data (NCBI). Regarding L1-clone sequences, regions of 866–889 nucleotides were found to be 92–97% homologous to the L1.3 element (NCBI accession no: L19088), with which L1 primers were designed.
Likewise, regions of 555–562 nucleotides of HERV-K10-clone sequences had a 93–98% homology to 3′ LTR HERV-K10 (NCBI accession no: M14123) encompassing the U3 region, thus documenting that their transcript origin was from full size or internally truncated HERV-K10 proviruses. Interestingly, all sequences concerning the ∼240 bp product showed a 92–95% homology at two distinct regions of 154 and 42 nucleotides mapped within the U3 and R sequences, respectively, of 3′ LTR HERV-K10. Additional sequence analysis, using the RepeatMasker program, revealed that although no homology was found with SINEs, LINEs, LTR retrotransposons and other mobile elements, an uninterrupted region of 209 nucleotides had a high homology with SVA_B and SVA_D elements (92.9–96.76%). This finding provided strong evidence for the identification of the 240 bp product as SVA transcripts, in agreement with the HERV-K10 origin of SVA elements (18). Hence, the initially termed HERV-K10-related sequences are, from now on, referred to as SVAs. Representative sequences of either L1 or HERV-K10 and SVA transcripts were deposited in the GenBank database with accession numbers FJ435167, FJ435168 and FJ435169, respectively. Collectively, these data show that RT–PCR products detected in the human oocyte (Fig. 1C and D) were specific transcripts of L1, HERV-K10 and SVA elements.
L1 and HERV-K10 transcripts are detected in the ooplasm of human oocytes
To ensure that L1 and HERV-K10 sequences analyzed were not derived from contaminating nuclear DNA, we performed RNA-FISH analysis in oocytes using as probes FITC-labeled PCR products derived either from the respective bacterial clones earlier mentioned or from an EGFP gene as negative control. By UV microscopy, we found that, in contrast to non-specific EGFP probe (Fig. 2b), both L1 and HERV-K10 probes detected respective transcripts both in the nucleus and in the ooplasm of treated oocytes (Fig. 2e and h). Notably, the HERV-K10 signal detected was much stronger than that of L1, consistent with the results of RT–PCR analysis (Fig. 1C). Alternatively, this was explained as an additive cross-hybridization of the HERV-K10 probe with SVA transcripts due to their sequence homology (discussed earlier), and thus an analysis for SVAs was not attempted. These data demonstrate that the sequences studied were transcripts of ooplasmic origin, providing additional evidence for retrotransposon RNA expression in oocytes.
Retrotransposition events occur in the human oocyte
Given that retrotransposon transcripts are required for the occurrence of a retrotransposition event (31), we asked whether the retrotransposon RNA expression found above might be accompanied with retrotransposition events in human oocytes. Towards this, we sought to introduce engineered EGFP-tagged retrotransposons, either a heterologous mouse VL30 transcriptionally expressed in human somatic cells (32) or human L1 into GV oocytes, where retrotransposition events could be detected specifically through the expression of a green fluorescent protein that is solely expressed following a retrotransposition event.
We used the recombinant pNVL-3*/EGFP-INT (33) for VL30 and either an ORF2 mutant, as a negative control, or competent pL1RP-EGFP(Puro) (34,35) for L1, known to produce retrotranspositions in mouse (33,36) and human cells (34), respectively. Each cloned retrotransposon was microinjected into 30 oocytes and following 24 h, the treated oocytes were observed by UV microscopy. Regarding mutant L1, no green fluorescent protein expression was detected (Fig. 3Ad). In contrast, we found that the oocytes microinjected either with the retrotransposition competent VL30 or L1 element exhibited EGFP expression, documenting retrotransposition events that have occurred (Fig. 3Ab and f). Out of the 30 treated oocytes, 21 (70%) and 26 (86%) oocytes were EGFP-positive for VL30 and competent L1, respectively (Fig. 3B). It is noteworthy that the VL30 retrotransposition-positive oocytes exhibited ooplasm granulation (Fig. 3Aa) and those in the case of L1 were characterized by an abnormal morphology (Fig. 3Ae). To provide additional evidence that the EGFP-positive oocytes had copies of recombinant retrotransposons integrated in the genome, microinjected oocytes were analyzed by direct PCR as shown in Fig. 3C. In the case of the mutant L1 recombinant, only a 1243 bp product was detected, showing microinjected plasmid, whereas in both the cases of VL30 and competent L1, the 342 bp retrotransposition diagnostic band (35) was also detected, mirroring the occurrence of retrotransposition events (Fig. 3C). In all, these results showed that retrotransposition events occur in human oocytes, suggesting that the oocyte provides a suitable environment for their generation.
We have shown for the first time that active retrotransposable elements L1, HERV-K10 and SVA, known to be involved in human diseases, are transcriptionally expressed and provided evidence for the occurrence of retrotransposition events in the human oocyte.
Two lines of evidence supported the RNA expression of L1s and HERV-K10s. First, specific transcripts were detected by RT–PCR (Fig. 1C) and, second, RNA-FISH analysis readily revealed a strong hybridization signal in single oocytes (Fig. 2). As for the HERV-K10s and SVAs, we found that these retrotransposons were transcriptionally more active than L1s (Fig. 1C and D). This finding can be indirectly supported by the relative retrotransposon copy number estimated to be ∼520 000 (37) for the L1s, whereas for that of HERV-K10s it is not more than 50 (38) and for SVAs it is 3500–7500 (20) in the human genome. We believe that the retrotransposon RNA expression found is mainly attributed to temporal hypomethylation of the human oocyte (27), as retrotransposon RNA expression is suppressed by methylation in the somatic cells (21). Although little is known about the expression of SVAs, the higher expression of HERV-K10s when compared with L1s might be due to their different levels of methylation, as differential methylation was found between the HERV-K copies and cell lines (24). In this context, the higher expression of HERV-K10 and HERV-K10-derived SVA elements—being younger than L1—is consistent with that of the youngest LTR-retrotransposons found in the mouse oocyte (39), and signifies a basic characteristic of the mammalian oocyte.
We found a higher RNA expression of L1s in comparison with that of HERV-K10s and SVAs in the somatic cells (Fig. 1D, right panel). This signifies rather an established somatic profile of retrotransposon expression as L1 > HERV-K10 > SVA, since the two different cell types of lymphocytes and HeLa cells produced the same profile. Interestingly, we found that this profile was reversed in the oocyte (Fig. 1C and D) rather supporting a previously suggested network of interrelated retrotransposon expression based on the interfered RNA expression of either L1s or HERV-Ks (40). Speculating on the biological significance of retrotransposon expression, we believe that high HERV-K10s RNA expression might concern envelope proteins contributing to the development of oocyte immunosuppression that is required for subsequent sperm–egg binding and fusion (41). Alternatively, the expression of HERV-K10s and SVAs—given their sequence homology—might serve either self or each other’s co-suppression through a high gene copy number-triggered homology-dependent gene silencing mechanism (42) shown for Ty1 retrotransposons (43).
We have shown human L1 and mouse VL30 retrotransposition events in the human oocyte (Fig. 3). These findings signify that retrotransposon transcription, reverse transcription and translation machineries, as well as splicing [based on recombinants structure (33,34)] are active in oocytes, allowing the occurrence of retrotranspositions even for a non-human and reverse transcriptase-deficient (non-autonomous) retrotransposon VL30 (33). In this context, we conclude that oocyte hypomethylation supports VL30 transcription in consistence with the inhibited methylation of mouse cells (44). More importantly, VL30 retrotransposition mirrors enzyme expression of multiple active reverse transcriptases, further supported by an MuLV reverse transcriptase that was found (45). The latter, compatible to that of MoMLV, most probably acted in trans for the VL30 retrotransposition (33). Accordingly, L1 reverse transcriptase may also mobilize non-autonomous retrotransposons SVAs (37) explaining: (i) the occurrence of retrotranspositions in the early stages of human development (46,47) and (ii) an estimation that 2% of newborns may contain a de novo L1-mediated retrotransposition event (2).
In keeping with this, we have observed the L1 retrotransposition-associated abnormal oocyte morphology (Fig. 3Ae). As it is difficult to believe that the L1 retrotransposition solely leads to detrimental effects, the abnormal oocyte morphology was explained by the high quantity of the microinjected L1 DNA. In spite of the L1s RNA expression (shown in Fig. 1C) and the active reverse transcriptases necessary for early oocyte development (48), we believe that the retrotransposition-mediated diseases must be events escaping unknown so far retrotransposition-control mechanism(s). On the other hand, controlled retrotranspositions may be of benefit to the oocyte and serve: (i) anti-sense transcript generation (49) in triggering RNAi-mediated mRNA degradation leading to subsequent methylation (50) and (ii) possible retrotransposon silencing and/or genome remodeling at the MII stage of oocyte maturation.
In conclusion, our results suggest that both a network of retrotransposons’ RNA expression and retrotransposition events might be essential for oocyte development and fertilization.
MATERIALS AND METHODS
Ethics and oocyte preparation
Immature oocytes at GV stage were donated to two IVF centers (Ioannina and Athens) by infertile couples, treated with intracytoplasmic sperm injection (ICSI), following informed consent. The study was approved by the Ioannina University Hospital Ethics Committee (560/2005). Oocyte denudation from cumulus and corona cells before ICSI, using hyalorunidase, allowed identification of the stage of maturation. Nuclei from immature diploid GVs arrested for 20–24 h were carefully aspirated, by micromanipulations under inverted microscope, for subsequent RNA analysis.
RT–PCR analysis, cloning and sequencing
Two sets of primers were designed for amplification of retrotransposon sequences. The design of L1 primers was based on retrotransposition competent LINE-1.3 element (NCBI accession no: L19088) spanning nucleotides 4017–4036 for the forward: 5′-CAG GGC AAT CAG GCA GGA GA-3′ and 4945–4964 for the reverse: 5′-TTG CCC ACG CCT ATG TCC TG-3′ primer, respectively. HERV-K10 primers designed on the HERV-K10 sequence (NCBI accession no: M14123) were: 5′-CCA ACC CCG TGC TCT CTG AA-3′ and 5′-TTG TGG GGA GAG GGT CAG CA-3′ for the forward and the reverse primer, spanning nucleotides 134–153 and 720–739, respectively.
For ooplasmic RNA preparation, denucleated oocytes preserved in Sydney IVF fertilization medium (CooK) were pooled and then centrifuged at 1600g for 1 min, washed and resuspended in 20 µl of distilled H2O. Following membrane disruption by three freeze-thaw cycles in liquid nitrogen/water bath at 37°C, oocytes were incubated with 250 µg/ml of proteinase K for 16 h at 55°C. After heat inactivation at 95°C for 15 min and treatment with RNase-free DNAse (Qiagen), samples were subjected to RT–PCR in a volume of 50 µl (Invitrogen kit). Subsequent PCR reactions were performed in a thermal profile of 94°C for 30 s, 59°C for 30 s and 72°C for 2 min, for 36–45 cycles. Total RNA either from HeLa or from human lymphocytes was prepared using a commercially available kit (Qiagen) and RT–PCR was performed in the same profile as above for 24 cycles. For PCR on bacterial clones, 5 µl of suspension at the exponential growth stage was washed once with 100 µl H2O and resuspended in 80 µl H2O. Following heating at 100°C for 5 min and cooling-down for 30 min, a volume of 50 µl was used as a template.
RT–PCR amplified products were analyzed in 1.2% agarose gels containing ethidium bromide and densitometry band analysis was performed by ImageJ software. For cloning, RT–PCR bands of interest were excised from agarose gels, purified by mini-columns (Qiagen), cloned into pCR 2.1 and the re-amplified products were sequenced (Lark Technologies). Sequence comparison and alignment of L1 948 and HERV-K10 606 bp sequences was performed using BLAST Program (NCBI). Alignment of the 240 bp sequence and characterization was performed using RepeatMasker Program (UCSC Genome Bioinformatics Site, http://genome.ucsc.edu/).
RNA-FISH analysis and fluorescence microscopy
For RNA-FISH analysis, 1 ng either of pre-amplified control pEGFP-N1 (Clontech) 342 bp (33) or of L1 948 or HERV-K10 606 bp PCR products derived from bacterial clones was used for FITC-labeling by PCR Fluorescein Labeling Mix (Roche). Products were purified, as above, and used as probes. Human GV oocytes were fixed (51), hybridized with labeled probes under conditions previously described (52) and stained with DAPI for nuclear DNA. Samples were photographed with a Nikon E400 fluorescence microscope under normal and UV light using B-2A and UV-2B filters for FITC and DAPI, respectively.
Oocyte retrotransposition assay
Human oocytes cultured in medium were microinjected with ∼1–2 pg of engineered EGFP-tagged retrotransposons pNVL-3*/EGFP-INT for mouse VL30 (33) and a retrotransposition-competent pL1RP-EGFP(Puro) as well as an ORF2 mutant for human L1 (34,35). Following 24 h of injection, treated oocytes were fixed (51) and retrotransposition was examined via EGFP expression using fluorescence microscopy. For detection of genomic retrotransposon integrations, six oocytes microinjected per case were lysed 24 h post-treatment under conditions described in (53) and subjected to direct PCR using specific EGFP primers and thermal profile (33).
This research was funded by the Empirikion Foundation, Greece.
The authors are grateful to Dr John Goodier and Prof. Haig H. Kazazian Jr, for providing us with pL1.2B and tagged mutant, and retrotransposition competent pL1RP-EGFP(Puro) constructs.
Conflict of Interest statement. None declared.