Effects of L1-ORF2 fragments on green fluorescent protein gene expression

The retrotransposon known as long interspersed nuclear element-1 (L1) is 6 kb long, although most L1s in mammalian and other eukaryotic cells are truncated. L1 contains two open reading frames, ORF1 and ORF2, that code for an RNA-binding protein and a protein with endonuclease and reverse transcriptase activities, respectively. In this work, we examined the effects of full length L1-ORF2 and ORF2 fragments on green fluorescent protein gene (GFP) expression when inserted into the pEGFP-C1 vector downstream of GFP. All of the ORF2 fragments in sense orientation inhibited GFP expression more than when in antisense orientation, which suggests that small ORF2 fragments contribute to the distinct inhibitory effects of this ORF on gene expression. These results provide the first evidence that different 280-bp fragments have distinct effects on the termination of gene transcription, and that when inserted in the antisense direction, fragment 280-9 (the 3' end fragment of ORF2) induces premature termination of transcription that is consistent with the effect of ORF2.

Although elements of L1 can occur almost anywhere in the mammalian genome, their abundance varies among genomic regions. In general, these elements are much more abundant in genomic regions that are AT-rich, have a lowrecombination frequency and are gene-poor (Pavlicek et al., 2001;Yang et al., 2004;Hackenberg et al., 2005;Belancio et al., 2006;Graham and Boissinot, 2006). In human genes, L1s preferentially have an antisense orientation and most copies are truncated (Sassaman et al., 1997;Boissinot et al., 2000;Sheen et al., 2000;Lander et al., 2001), rearranged (Skowronski and Singer, 1986) or both. These findings imply that the length and orientation of L1s have different effects on genes. It would therefore seem highly important to study the effects of L1 fragments and their orientations on gene expression.
L1s can cause the retrotransposition of Alu (Dewannieux et al., 2003) and mediate the cell growth and differentiation associated with this event (Ergün et al., 2004;Sciamanna et al., 2005). Han et al. (2004) reported that L1.2-ORF2 in the sense orientation inhibited GFP expression much more than when in antisense orientation. By using appropriate deletions these authors also showed that the inhibition of gene expression varied with the length of the L1.2-ORF2 fragment.
In this study, we used L1PA3, a subfamily of L1s that shares 96% similarity with L1.2-ORF2, to examine whether L1PA3-ORF2 has the same effect on gene expression as L1.2-ORF2. Seven 280-bp fragments obtained by the polymerase chain reaction (PCR) from different regions of L1PA3-ORF2 (ORF2) were fused in tandem to GFP in order to examine their effect on gene expression. tified by digestion with the restriction enzyme pair HindIII/ Nhe I and DNA sequencing (Generay Co. Shanghai, China). The primers used for PCR are shown in Table 2.

Cell culture and cell transfection
HeLa cells were routinely cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal calf serum (FCS). Aliquots containing 1.8 x 10 5 cells/mL were plated in 12-well plates and then cultured at 37°C in 5% CO 2 for 24 h. At approximately 50%-70% confluence, the cells were transfected with 1.5 mg of plasmid DNA and 3 mL of liposomes (Lipofectamine 2000; Invitrogen, Grand Island, NY) in order to observe fluorescent cells and to generate RNA for subsequent experiments.

Assessment of GFP reporter protein
The expression of GFP reporter protein was assessed by fluorescence microscopy of transfected HeLa cells.

Northern blotting
The GFP probe was labeled with [ 32 P]-deoxycytidine triphosphate (dCTP) via PCR using the primers shown in Table 2. Total RNA was extracted from plasmid-transfected HeLa cells with Trizol reagent (Invitrogen, Inc.) 36 h after transfection. The RNA was electrophoresed in a 1.2% agarose gel denatured with 3% formaldehyde followed by transferring to nylon membranes in 20x salt-sodium citrate (SSC) for 24 h. RNA was cross-linked to the membranes by exposure to UV light and the membranes then incubated with the GFP probe at 42°C followed by autoradiography. The membranes were subsequently stripped by washing twice at 80°C for 1 h with 50 mM Tris, pH 7.4, containing 50% formamide and 5% sodium dodecylsulfate (SDS), and then hybridized with a [ 32 P]-labeled probe for neo mRNA (the cassette for neomycin resistance). This probe was prepared by PCR amplification with the primers shown in Table 2.

Results
Effects of ORF2 in sense and antisense orientations on GFP expression ORF2 (3825 bp) or the lacZ sequence was inserted downstream of GFP in the pEGFP-C1 vector. The insertion of ORF2 in sense or antisense orientation significantly decreased GFP RNA ( Figure 1) and protein (data not shown) expression. To demonstrate this decrease, we used Xho I /Pst I or Apa I restriction enzymes to construct plasmids of pORF2, pORF2as, pORF2Apa and pORF2asApa (see Table 1). When ORF2 was inserted in the sense orientation (pORF2), GFP RNA production was only 3.6% of that seen with ORF2 in the antisense orientation (pORF2as) ( Figure  1, lane 1 vs. lane 2), and when ORF2Apa was inserted in the sense orientation (pORF2Apa) GFP RNA production was 4.2% of that seen with ORF2Apa in the antisense orientation (pORF2asApa) (Figure 1, lane 3 vs. lane 4). Thus, when ORF2 or ORF2Apa was inserted in the antisense orientation most of the decrease in the expression of full-length GFP RNA was related to the generation of low molecular mass RNA species, indicating that antisense Wang et al. 689 Table 1 -Plasmids used in this study.

Plasmids
Fragment inserted into pEGFP-C1 and annotation pORF2, pORF2as ORF2 inserted in sense or antisense (as) orientation downstream of GFP by using the restriction enzymes Xho I /Pst I. pORF2Apa, pORF2asApa ORF2 inserted in sense or antisense orientation downstream of GFP by using the restriction enzyme Apa I.
pLacZ, pLacZas LacZ inserted in sense or antisense orientation downstream of GFP.

690
Distinct effects of L1-ORF2 on GFP ORF2as ( ORF2 caused premature termination of GFP transcription. The insertion of lacZ in either orientation reduced RNA synthesis to low similar levels ( Figure 1, lanes 5 and 6). The insertion of lacZ in antisense orientation caused premature termination of GFP transcription (Figure 1, lane 6), whereas the insertion of this gene in sense orientation induced transcriptional elongation (Figure 1, lane 5). These findings indicated that ORF2 in sense orientation caused much stronger gene inhibition than in antisense orientation, with the latter causing premature transcriptional termination.

Effects of different 280-bp fragments of ORF2 on GFP expression
To study the effects of ORF2 fragments on GFP expression, we obtained seven 280-bp fragments from different regions of ORF2, as shown in Figure 2C and Table 2. Head and tail, tandem 8-sequence repeats (see Table 1) were constructed for each fragment. As shown in Figure 2, all of the inserts inhibited GFP transcription much more strongly in sense than in antisense orientation, which was consistent with the results for full-length ORF2 (Figure 1). Regardless of their orientation (sense or antisense), fragments 280-1 and 280-9 caused premature termination of transcription and produced low molecular mass RNA (Figure 2A, lanes 1 and 7; Figure 2B, lanes 1 and 7). Fragment 280-5 caused premature termination of GFP transcription in sense orientation (Figure 2A, lane 4), whereas fragment 280-4 had the same effect in antisense orientation ( Figure  2B, lane 3). Other ORF2 fragments, including fragments 280-2, 280-7 and 280-8, did not induce premature termination of GFP transcription in either orientation. Thus, in contrast to ORF2 which caused premature termination of GFP transcription when in antisense orientation, the effect of ORF2 fragments on transcriptional elongation were less predictable.

Effects of simple repeats constructed from ORF2 fragments on GFP expression
Since the different 280-bp ORF2 fragments had distinct effects on GFP expression in HeLa cells (Figure 2), we examined the influence of even shorter ORF fragments on gene expression. As shown in Table 3, the ORF2 fragments generally contained more A than T. We chose AAACAAA and AG, which are particularly abundant in ORF2, and constructed 736-bp repeats of these base sequences. The AAACAAA or AG repeats were then inserted into the pEGFP-C1 vector downstream of GFP in sense or antisense orientation. Fragments inserted in sense orientation suppressed transcription more strongly than those in antisense orientation (Figure 3), in agreement with the findings for ORF2 and its 280-bp fragments. Interestingly, AAACAAA repeats in either orientation caused premature transcriptional termination (Figure 3, lanes 1 and 2), whereas AG repeats in antisense orientation resulted in greater synthesis of higher molecular mass transcripts than did AG repeats in sense orientation (Figure 3, lane 4 vs. lane 3).
ORF2 fragment 280-9 is responsible for premature transcriptional termination by ORF2 in antisense orientation As shown above (Figure 2), the ORF2 fragments had distinct effects on GFP transcriptional elongation. Of seven 280-bp fragments, fragment 280-9 (the 3' end sequence of ORF2) caused premature transcriptional termination when inserted in antisense orientation ( Figure 2B), in agreement with the results for ORF2 (Figure 1). These findings implied that when ORF2 is in antisense orientation the 3' end of ORF2 is responsible for premature transcriptional termination. To confirm this, the 3' end of ORF2, including fragment 280-9 and its downstream region, were deleted (the resulting fragment was referred to as 280-1~8). When fragment 280-1~8 was inserted in the antisense orientation downstream of GFP there was no premature transcriptional termination of this gene (Figure 4, lane 2), thus confirming the importance of the 3' end sequence of ORF2 in this phenomenon. However, insertion of fragment 280-1 in sense orientation induced much stronger inhibition of GFP expression than did its insertion in antisense orientation, and when present in either orientation this fragment caused premature transcriptional termination.
Copies of ORF2 fragment 280-4 inserted in sense orientation downstream of GFP decreased RNA transcription ( Figure 5B) and protein expression ( Figure 5C) in a length-dependent manner. The lengths of RNA transcripts increased with increasing numbers of copies of fragment 280-4 ( Figure 5B, lanes 6-10), suggesting that fragment 280-4 did not cause premature termination of transcription.
Alu, used as a control in these experiments, also inhibited gene expression in a length-dependent manner but did not cause premature transcriptional termination ( Figure 5A
As shown here, the ORF2 of L1PA3 in sense orientation inhibited GFP expression much more than in antisense orientation, and caused premature transcriptional termination in the latter orientation, in agreement with previous findings for L1.2-ORF2 (Han et al., 2004). Although the sequences of L1.2-ORF2 and L1PA3-ORF2 are not identical, they had similar effects on gene expression, suggesting that mutation of individual nucleotides does not affect the functions of this ORF.
Different restriction enzymes were used to construct plasmids with ORF2 in sense (pORF2 and pORF2Apa) and antisense (pORF2as and pORF2asApa) orientation. The insertion of ORF2 (Figure 1, lanes 1 and 2) or ORF2Apa (Figure 1, lanes 3 and 4) had the same effect on gene expression as when they were incorporated into plasmids in the same orientation, a finding that increased our confidence in the results of this study. The ORF2 sequence does not inhibit the initiation of transcription and is a poor substrate for transcriptional elongation (Han et al., 2004).
The influence of ORF2 fragments on transcriptional termination and gene inhibition was examined by using seven 280-bp ORF2 fragments lacking restriction enzyme sites that could otherwise disturb the base linkage within the fragments. Each fragment consisted of eight tandem re- Wang et al. 693  peats and was inserted downstream of GFP in sense or antisense orientation. All of the fragments significantly reduced gene expression, with greater inhibition when in sense compared to antisense orientation ( Figure 2); this finding consistent with our observations for ORF2 (Figure 1). Enhanced RNA degradation or decreased RNA production could reduce RNA concentrations. Han et al. (2004) stated that most of the decrease in GFP-ORF2 transcription in the presence of L1.2-ORF2 was not due to transcript degradation. In the present study, the bands seen in northern blots probably reflected the rate of gene transcription.
Mutations involving individual nucleotides did not affect ORF2 function, and each 280-bp fragment continued to have a stronger inhibitory effect in sense compared to antisense orientation. These findings prompted us to investigate the effects of fragments < 280 bp on gene expression. L1 has an adenosine-rich (A-rich) bias in the sense strand (Deininger et al., 2003). We chose small fragments of AAACAAA and AG, which are abundant in ORF2. Fragments containing tandem repeats of AAACAAA and AG were used to ensure a sufficiently large effect on gene expression. As in the experiments with the 280-bp fragments, the AAACAAA and AG repeats showed much stronger inhibition in sense compared to antisense orientation. AAACAAA sequences were certified to be Sox2 protein binding sites. The binding of Sox2 protein to these sites suppresses gene expression driven by L1 5'-UTR (Muotri et al., 2005) and may be one of the mechanisms by which AAACAAA inhibits gene expression. Figure 2 shows that the 280-bp ORF2 fragments had different effects on transcriptional termination. Fragment 280-9 in antisense orientation caused premature transcriptional termination ( Figure 2B, lane 9) that resembled the results obtained with ORF2 in antisense orientation (Figure 1,  lane 2). This finding suggested that fragment 280-9 plays a key role in premature transcriptional termination by ORF2. To confirm this hypothesis, we deleted fragment 280-9 from the 3'end of ORF2 and inserted the resulting fragment (280-1~8) in antisense orientation downstream of GFP. This insert failed to stimulate the production of low molec- 694 Distinct effects of L1-ORF2 on GFP ular mass RNA similar to that seen with ORF2 in antisense orientation (see Figure 4). This result indicated that fragment 280-9 and its 3'end sequence play a key role in the premature transcriptional termination mediated by ORF2.
Since the chromosomal densities of Alu and L1 are negatively correlated with each another (except for the Y chromosome), and since L1 elements are responsible for the retrotransposition of Alu retroelements (Dewannieux et al., 2003), Alu was used as a parallel control in some experiments. The genomic distribution of Alu is suggestive of a possible involvement in enhancing gene expression. However, as shown here, Alu inhibited gene expression in a length-dependent manner ( Figure 5), but had a much weaker effect than ORF2 fragments 280-1 or 280-4. In addition, Alu did not cause premature transcriptional termination. ORF2 may cause premature transcriptional termination (Figure 1, lanes 1-4; Figure 4, lanes 1 and 3) through the presence of multiple functional canonical and noncanonical polyA signals in L1 (Deininger et al., 2003). Such signals are also present in some ORF2 fragments, e.g., fragments 280-9 and 280-1, where they presumably also promote termination. Han et al. (2004) found that tandem L1.2-ORF1 caused length-dependent inhibited of gene expression. As shown here, 280-1, 280-4 and Alu in either orientation also caused length-dependent suppression of gene expression.
In conclusion, we have described a number of potentially important functions of ORF2 and its fragments that affect gene expression. The major findings of this work are that: (1) ORF2 fragments contributed differently to gene transcriptional elongation, with only some fragments inducing the premature transcriptional termination seen with ORF2, (2) in deletion studies, the 3' end sequence of ORF2 (fragment 280-9) is responsible for the premature transcriptional termination observed with ORF2 in antisense orientation, and (3) all of the ORF2 fragments studied here, as well as ORF2 itself, inhibited gene expression much more in sense compared to antisense orientation. The latter observation suggested that small fragments contributed to ORF2-mediated inhibition of gene expression primarily when in sense orientation.