The Zinc-Finger Antiviral Protein ZAP Inhibits LINE and Alu Retrotransposition

Long INterspersed Element-1 (LINE-1 or L1) is the only active autonomous retrotransposon in the human genome. To investigate the interplay between the L1 retrotransposition machinery and the host cell, we used co-immunoprecipitation in conjunction with liquid chromatography and tandem mass spectrometry to identify cellular proteins that interact with the L1 first open reading frame-encoded protein, ORF1p. We identified 39 ORF1p-interacting candidate proteins including the zinc-finger antiviral protein (ZAP or ZC3HAV1). Here we show that the interaction between ZAP and ORF1p requires RNA and that ZAP overexpression in HeLa cells inhibits the retrotransposition of engineered human L1 and Alu elements, an engineered mouse L1, and an engineered zebrafish LINE-2 element. Consistently, siRNA-mediated depletion of endogenous ZAP in HeLa cells led to a ~2-fold increase in human L1 retrotransposition. Fluorescence microscopy in cultured human cells demonstrated that ZAP co-localizes with L1 RNA, ORF1p, and stress granule associated proteins in cytoplasmic foci. Finally, molecular genetic and biochemical analyses indicate that ZAP reduces the accumulation of full-length L1 RNA and the L1-encoded proteins, yielding mechanistic insight about how ZAP may inhibit L1 retrotransposition. Together, these data suggest that ZAP inhibits the retrotransposition of LINE and Alu elements.


Introduction
Long INterspersed Element-1 (LINE-1, also known as L1) sequences comprise~17% of human DNA and represent the only class of autonomously active retrotransposons in the genome [1]. L1s mobilize (i.e., retrotranspose) throughout the genome via an RNA intermediate by a copyand-paste mechanism known as retrotransposition [reviewed in 2]. The overwhelming majority of human L1s are retrotransposition-deficient because they are 5' truncated, contain internal rearrangements (i.e., inversion/deletion events), or harbor point mutations that compromise the functions of the L1-encoded proteins (ORF1p and ORF2p) [1,3]. Despite these facts, it is estimated that the average diploid human genome contains~80-100 L1 elements that are capable of retrotransposition [4][5][6]. It is estimated that a new L1 insertion occurs in approximately 1 out of 200 live human births [reviewed in 7]. On occasion, L1 retrotransposition events can disrupt gene expression, leading to diseases such as hemophilia A [8], Duchenne muscular dystrophy [9], and cancer [10,11]. Indeed, L1-mediated retrotransposition events are responsible for at least 96 disease-producing insertions in man [reviewed in 12].
To gain a more complete understanding of the interplay between the L1 retrotransposition machinery and the host cell, we used liquid chromatography-tandem mass spectrometry (LC-MS/MS) to identify proteins that co-immunoprecipitate with L1 ORF1p in HeLa cells, reasoning that some of these proteins may affect L1 retrotransposition. We next analyzed the effects of ORF1p-interacting proteins on L1 retrotransposition by overexpressing a subset of them in a cultured cell retrotransposition assay [30,71]. Here, we report that the zinc-finger antiviral protein ZAP [72] interacts with L1 RNPs and inhibits L1 retrotransposition in cultured cells. ZAP also inhibits human Alu retrotransposition and the retrotransposition of mouse and zebrafish LINE elements. Molecular genetic and biochemical analyses suggest that ZAP inhibits retrotransposition by suppressing the accumulation of full-length L1 RNA and L1-encoded proteins in the cell.
To determine the identity of cellular proteins that associated with ORF1p-FLAG, the bands from the lanes corresponding to the pJM101/L1.3FLAG and pJM101/L1.3 immunoprecipitation  [5] containing an mneoI retrotransposition indicator cassette within the L1 3' UTR. The pJM101/L1.3FLAG construct is identical to pJM101/L1.3, but contains a single FLAG epitope on the carboxyl-terminus of ORF1p. Both constructs were cloned into a pCEP4 mammalian expression vector. A CMV promoter augments L1 expression and an SV40 polyadenylation signal (pA) is located downstream of the native L1 polyadenylation signal. (B) Results of immunoprecipitation experiments: Whole cell lysates from HeLa cells transfected with either pJM101/L1.3 or pJM101/L1.3FLAG were subjected to immunoprecipitation using an anti-FLAG antibody. The proteins then were separated by SDS-PAGE, visualized by silver staining, and subjected to LC-MS/ MS. An~40 kDa band corresponding to the theoretical molecular weight of ORF1p is visible in the pJM101/L1.3FLAG lane (*). Black bars indicate the approximate molecular weights of the ORF1p-FLAG interacting proteins. Molecular weight standards (kDa) are shown on the left hand side of the gel. (C) Validation of the ORF1p-FLAG immunoprecipitation: Western blot experiments using an antibody specific to amino acids 31-49 of L1.3 ORF1p verified the enrichment of ORF1p-FLAG in pJM101/L1.3FLAG, but not pJM101/L1.3 immunoprecipitation reactions. Cells transfected with the pCEP4 vector served as a negative control. (D) Validation of putative ORF1p-FLAG interacting proteins: Western blot images of the pJM101/L1.3FLAG and pJM101/L1.3 immunoprecipitation (IP) reactions. The pCEP4 lanes denote whole cell lysates derived from HeLa cells transfected with an empty pCEP4 vector (~1.0% input). Primary antibodies used to probe western blots are indicated to the left of the images. Immunoprecipitation reactions were conducted in either the experiments were excised from SDS-PAGE gels and submitted for LC-MS/MS (see Methods). An LC-MS/MS-identified protein was selected as an ORF1p-interacting candidate if it met the following criteria: 1) the protein was unique to the pJM101/L1.3FLAG immunoprecipitation, and 2) the protein was identified by two or more unique peptide sequences (peptide error rate 0.05; protein probability 0.95) (S1 Table and Methods). Thirty-nine ORF1p-interacting protein candidates were identified that met these criteria (S1 Table).

ORF1p-interacting proteins restrict L1 retrotransposition in HeLa cells
We next investigated whether overexpression of nine of the validated ORF1p-interacting proteins, as well as nine unvalidated ORF1p-interacting proteins, affects L1 retrotransposition [30,71]. Briefly, HeLa cells were co-transfected with a cDNA plasmid expressing one of the ORF1p-FLAG interacting proteins and an engineered human L1 construct (pJJ101/L1.3; [85]) marked with a blasticidin retrotransposition indicator cassette (mblastI) (Fig 2A and 2B; top panel). The mblastI cassette contains an antisense copy of the blasticidin deaminase gene, which is cloned into the L1 3' UTR. The blasticidin deaminase gene also is interrupted by an intron in the same transcriptional orientation as L1. This arrangement ensures that the blasticidin deaminase gene is expressed only when the L1 transcript is spliced, reverse transcribed, and inserted into genomic DNA. The resulting blasticidin-resistant foci then provide a visual, quantitative readout of retrotransposition activity [30,45].
To monitor potentially toxic side effects of cDNA overexpression, HeLa cells also were cotransfected in a parallel assay with a cDNA expression vector and a control plasmid (pcDNA6/ TR) that expresses the blasticidin deaminase gene (Fig 2B; bottom panel). Following blasticidin selection, the resulting foci provide a visual, quantitative readout of the effect of cDNA overexpression on colony formation (Fig 2B; bottom panel). This control is essential to determine if a cDNA affects L1 retrotransposition or cell viability and/or growth.
We co-transfected HeLa cells with each of the 18 ORF1p-FLAG interacting candidates and pJJ101/L1.3 ( Fig 2C). An empty pCEP4 vector that was co-transfected with pJJ101/L1.3 served as a normalization control (Fig 2B and 2C). As a negative control, we demonstrated that a plasmid that expresses the humanized renilla green fluorescence protein (pCEP/GFP) did not affect pJJ101/L1.3 retrotransposition. As a positive control, we demonstrated that a plasmid that expresses human APOBEC3A (pK_A3A) reduced pJJ101/L1.3 retrotransposition to~18% of control levels (Fig 2C), which is in agreement with previous studies [61,[86][87][88]. Four of the absence (left) or presence (right) of RNaseA (10 μg/mL). The putative cellular functions of the ORF1p-FLAG interacting proteins are indicated on the right hand side of the blots. construct was cloned into a pCEP4 mammalian expression vector. A CMV promoter augments L1 expression and an SV40 polyadenylation signal (pA) is located downstream of the native L1 polyadenylation signal. The mblastI cassette is cloned into the L1 3' UTR antisense to the L1 and contains a blasticidin deaminase gene that is disrupted by an intron in the L1 sense orientation. The blasticidin deaminase gene can only be expressed when the L1 transcript is spliced, reverse transcribed, and inserted into genomic DNA [30,71]. (B) Schematic of the pJJ101/L1.3 retrotransposition screen: To analyze the effect of the ORF1p-FLAG interacting proteins on L1 retrotransposition, HeLa cells were co-transfected with equal amounts of pJJ101/L1.3 and a cDNA plasmid expressing one of the candidate ORF1p-FLAG interacting proteins or a pCEP4 empty vector. To control for potential off-target effects, HeLa cells also were co-transfected with a control plasmid (pcDNA6/TR) that expresses the blasticidin deaminase gene and a cDNA plasmid expressing one of the candidate proteins or a pCEP4 empty vector. Both assays were subjected to the same blasticidin selection regimen. The resultant number of blasticidin-resistant colonies in pcDNA6/TR control assays provides a visual, quantitative readout of the effect of cDNA overexpression on the ability of cells to grow in the presence of blasticidin. (C) Results of pJJ101/L1.3 retrotransposition screen: HeLa cells were co-transfected with pJJ101/L1.3 and each of the indicated cDNA expressing plasmids. L1 retrotransposition was assayed in 6-well tissue culture plates. The X-axis indicates the cDNA that was co-transfected with cDNA-expressing plasmids that we tested (ZAP-S (ZAP short isoform), hnRNPL, MOV10, and PURA) each reduced pJJ101/L1.3 retrotransposition to less than 50% of pCEP4 control levels. Notably, ZAP-S (~30% of control), hnRNPL (~30% of control), MOV10 (~13% of control), and PURA (~10% of control) inhibited retrotransposition to levels similar to that of pK_A3A (~18% of control) (Fig 2C). By comparison, the majority of the cDNA-expressing plasmids (14/18) did not significantly affect pJJ101/L1.3 retrotransposition levels (less than 50% inhibition when compared to pCEP4 control levels) (Fig 2C). Thus, the data suggest that ZAP-S, hnRNPL, MOV10, and PURA inhibit L1 retrotransposition in cultured cells.
ZAP is a poly (ADP-ribose) polymerase (PARP) family member [90] initially characterized as an antiviral protein that inhibits murine leukemia virus (MLV) replication in cultured rat cells [72]. Previous studies identified two human ZAP isoforms that resulted from alternative splicing [90] (Fig 3A; top panel). The long ZAP isoform (ZAP-L) is 902 amino acids in length and contains an amino-terminus CCCH zinc-finger domain and an inactive carboxyl-terminal PARP-like domain [90]. The short ZAP isoform (ZAP-S) is 699 amino acids in length and lacks the carboxyl-terminal PARP-like domain [90]. The HA-tagged human ZAP-L isoform restricted pJJ101/L1.3 retrotransposition to~40% of control levels ( Fig 3A; black bars) and the human ZAP-S isoform restricted pJJ101/L1.3 retrotransposition to~30% of control levels (Figs 2C and 3A; black bars). Notably, overexpression of ZAP-L or ZAP-S did not dramatically affect the ability of HeLa cells to form blasticidin-resistant colonies in pcDNA6/TR control assays Putative ZAP orthologs are present in several species [90]; thus, we tested whether a rat ZAP cDNA (rZAP) [72], that is orthologous to human ZAP-S [72,90] could restrict pJJ101/ L1.3 retrotransposition. Overexpression of rZAP efficiently reduced retrotransposition tõ 40% of control levels ( Fig 3A; black bars). Thus, the ability to restrict L1 retrotransposition is not limited to human ZAP.

The ZAP zinc-finger domain is necessary and sufficient to inhibit L1 retrotransposition
The ZAP zinc-finger domain binds to RNA and is required for antiviral activity [91,92]. To analyze the role of the ZAP zinc-finger domain in L1 restriction, we tested the effects of a truncated ZAP-S mutant that expresses the ZAP zinc-finger domain (ZAP-S/1-311; containing amino acids 1-311) as well as a ZAP-S mutant that lacks the zinc-finger domain (ZAP-S/Δ72-372; lacking amino acids 72-372) in pJJ101/L1.3 retrotransposition assays (Fig 3A; above graph). pJJ101/L1.3. The bracketed number next to each cDNA indicates the number of independent experiments. The Y-axis indicates L1 retrotransposition activity after accounting for cDNA toxicity (see Fig 2B). Retrotransposition activity (black bars) is normalized to the pCEP4 empty vector control. Error bars represent the standard deviation for each set of experiments. The red dotted line indicates a 50% inhibition of retrotransposition activity.   Fig  3A; black bars). The overexpression of the wild type or mutant ZAP-S/Δ72-372 expression constructs did not adversely affect the ability of HeLa cells to form blasticidin-resistant colonies in pcDNA6/TR control assays ( Fig 3A; white bars). Notably, transfection with ZAP-S/1-311 resulted in an~50% decrease in the ability of HeLa cells to form blasticidin-resistant colonies; however, this effect has been accounted for through normalization ( Fig 2B) and thus is independent of the ability of ZAP-S/1-311 to restrict L1 retrotransposition. Indeed, similar offtarget effects have been reported for A3A cDNA expressing plasmids in HeLa cell-based L1 retrotransposition assays [61]. Western blot control experiments revealed that wild-type ZAP-S and the two mutant ZAP-S isoforms were expressed at similar levels~48 hours post-transfection (S2B and S2C Fig). Thus, the ZAP zinc-finger domain is necessary and sufficient to inhibit L1 retrotransposition.

ZAP restricts the retrotransposition of various non-LTR retrotransposons
To determine if ZAP-S was able to restrict other non-long terminal repeat (non-LTR) retrotransposons, we tested whether ZAP-S expression affected human Alu retrotransposition. Unlike L1, Alu is a 7SL-derived non-autonomous retrotransposon that does not encode its own proteins [93]. Instead, Alu elements must parasitize L1 ORF2p in trans to mediate their retrotransposition [49]. Briefly, HeLa cells were co-transfected with a full-length L1 element (pJM101/L1.3Δneo), an Alu retrotransposition reporter plasmid (pAluneo Tet ), and a ZAP-S expression plasmid. Notably, ZAP-S potently reduced Alu retrotransposition to~25% of control levels ( Fig 3B). In contrast, the expression of the L1 restriction-deficient ZAP-S/Δ72-372 mutant did not negatively affect Alu retrotransposition ( Fig 3B). Thus, ZAP-S is able to restrict the mobility of the two most prolific retrotransposons present in the human genome.
We next tested if human ZAP-S could restrict the retrotransposition of a natural mouse L1 (pG F 21) [94], a zebrafish LINE-2 (pZfL2-2) [95], or a synthetic mouse L1 (pCEPsmL1) [96] that has been extensively mutagenized to alter 24% of the nucleic acid sequence without disrupting amino acid sequence. Human ZAP-S inhibited the retrotransposition of human L1 (pJM101/L1.3;~43% of control levels), natural mouse L1 (pG F 21;~24% of control levels), zebrafish L2 (pZfL2-2;~19% of control levels), and synthetic mouse L1 (pCEPsmL1;~70% of control levels) (Fig 3C). The restriction-defective ZAP-S mutant, ZAP-S/Δ72-372, did not significantly affect the retrotransposition activity of these retrotransposons ( Fig 3C). Notably, the pCEP4 empty vector control (100%). The numbers above the bar graphs indicate the number of independent experiments performed with each cDNA expression construct. Error bars represent standard deviations. Bottom panel: A single well of a representative six-well tissue culture plate, displaying blasticidin-resistant colonies from the pJJ101/L1.3 retrotransposition assay (top, black rectangle) and the pcDNA6/TR control assay (bottom, white rectangle). (B) ZAP inhibits Alu retrotransposition: The X-axis indicates the cDNA co-transfected with pJM101/L1.3Δneo and pAluneo Tet . The Y-axis indicates the retrotransposition efficiency. All values are normalized to the pCEP4 empty vector control (100%). Control assays using a plasmid that expresses the neomycin phosphotransferase gene (pcDNA3) were conducted similarly to pcDNA6/TR control assays as outlined in Fig 2B. Representative images of G418-resistant HeLa foci from the Alu retrotransposition assay are shown below the bar graph. The results are the average of three independent experiments. Error bars indicate standard deviations. (C) ZAP inhibits the retrotransposition of mouse and zebrafish LINE elements. The X-axis indicates the cDNA that was co-transfected with human L1 (pJM101/L1.3 (black bars)), mouse L1 (pG F 21 (dark grey bars)), zebrafish L2 (pZfL2-2 (light grey bars)), or synthetic mouse L1 (pCEPsmL1 (white bars)). The Y-axis indicates the retrotransposition efficiency. Representative images of G418-resistant HeLa cell foci are shown below the bar graph. Control assays using a plasmid that expresses the neomycin phosphotransferase gene (pcDNA3) were conducted similarly to pcDNA6/TR control assays outlined in Fig  milder inhibition of ZAP-S on pCEPsmL1 may be due to the elevated efficiency of pCEPsmL1 retrotransposition, the increased steady-state level of pCEPsmL1 mRNA and proteins, and/or the GC-rich nature of pCEPsmL1 [96]. Thus, ZAP-mediated restriction of retrotransposition is not specific to human non-LTR retrotransposons.

Depletion of endogenous ZAP enhances L1 retrotransposition
To test if endogenous ZAP restricts L1 retrotransposition, we used small interfering RNA (siRNA) to deplete endogenous ZAP from HeLa cells. Following siRNA treatment, cells were transfected with an L1 plasmid (pLRE3-mEGFPI) tagged with an EGFP indicator cassette (mEGFPI), which allows retrotransposition activity to be detected by EGFP fluorescence [97]. As a negative control, HeLa cells were transfected with the L1 retrotransposition-defective plasmid pJM111-LRE3-mEGFPI, which carries two missense mutations that adversely affect ORF1p RNA binding [22,30,98]. Treatment of HeLa cells with an siRNA pool against ZAP resulted in an~80% and~90% reduction of ZAP-L and ZAP-S protein levels, respectively, when compared to HeLa cells treated with a non-targeting control siRNA pool ( [64,65]. These data suggest that endogenous ZAP may restrict L1 retrotransposition.

ZAP-S inhibits the accumulation of full-length LINE-1 mRNA
To investigate how ZAP restricts L1 retrotransposition, we analyzed the effect of ZAP-S expression on the accumulation of the L1 RNA. HeLa cells were co-transfected with pJM101/ L1.3Δneo and either ZAP-S or ZAP-S/Δ72-372. Polyadenylated RNA from whole cell extracts then was analyzed by northern blot using RNA probes complementary to sequences within the L1.3 5' UTR (5UTR99) and ORF2 (ORF2_5804) (Fig 4A). Co-transfection with ZAP-S resulted in a reduction of full-length polyadenylated L1 RNA levels (~13% of pCEP4 control) compared to cells co-transfected with either the restriction-defective ZAP-S/Δ72-372 (~47% compared to pCEP4 control) or an empty pCEP4 control vector ( Fig 4B; black arrow in blot; black bars in graph). Interestingly, ZAP-S expression did not have a pronounced effect on the accumulation of smaller L1 RNA species, which may have resulted from cryptic splicing and/or premature polyadenylation (Fig 4B; top panel: blue and yellow arrows, bottom panel: blue and yellow bars) [99][100][101]. Finally, control experiments revealed that ectopic ZAP-S expression did not affect endogenous actin RNA levels ( Fig 4B). Thus, ZAP-S expression reduces the accumulation of full-length L1 mRNA in cultured cells.

ZAP-S inhibits the accumulation of ORF1p and ORF2p
We next examined the effect of ZAP-S expression on the accumulation of ORF1p and ORF2p. We co-transfected HeLa cells with either ZAP-S or ZAP-S/Δ72-372 and the L1 plasmid, pJBM2TE1, which expresses an L1.3 element marked with a T7 gene10 epitope tag on the carboxyl-terminus of ORF1p and a TAP epitope-tag on the carboxyl-terminus of ORF2p ( Fig  4C). Following co-transfection, HeLa cells were treated with puromycin to select for cells expressing pJBM2TE1. Both whole cell lysates (WCL) and RNP fractions were collected 5 days post-transfection and subjected to western blot analyses to monitor ORF1p and ORF2p expression levels. Expression of ZAP-S led to a decrease in the level of ORF1p and ORF2p in both WCL and RNP fractions, whereas the expression of the restriction-defective ZAP-SΔ/72-372 mutant or an empty pcDNA3 vector did not dramatically affect ORF1p or ORF2p expression levels ( Fig  4D). The reduction in ORF1p and ORF2p was most evident in the RNP fraction, likely because both ORF1p and ORF2p are enriched in RNPs [19,26,37,38,102]. Control experiments revealed that ZAP-S expression did not affect the level of eIF3 protein ( Fig 4D) and that ZAP-S and ZAP-S/Δ72-372 are expressed at similar levels in whole cell lysates (Fig 4D: top WCL panel). By comparison, ZAP-SΔ/72-372 is present at much lower levels in the RNP fraction compared to wild-type ZAP-S (Fig 4D; bottom RNP panel), suggesting that the zinc-finger domain is responsible for ZAP-S localization to the RNP fraction.
To determine if ZAP-S affects the expression of non-L1 proteins, we examined the effect of ZAP-S on EGFP expression. We co-transfected ZAP-S with an L1 plasmid (pLRE3-EF1-mEGFPΔIntron) [ ZAP co-localizes with ORF1p and L1 RNA in the cytoplasm ORF1p, ORF2p, and L1 RNA form RNP complexes that appear as discrete cytoplasmic foci when visualized by fluorescence microscopy [25,26,104]. Notably, previous studies have shown that ZAP predominantly is localized in the cytoplasm [105] and that ZAP antiviral activity also is localized to the cytoplasm [72]. To determine if ZAP co-localizes with ORF1p, we cotransfected HeLa cells with pJM101/L1.3Δneo and a plasmid that expresses a carboxylterminus turbo-GFP tagged ZAP-S protein (ZAP-S-tGFP). Control experiments showed that ZAP-S-tGFP restricted pJJ101/L1.3 retrotransposition to~55% of control levels (S4A Fig). Confocal fluorescence microscopy revealed that ORF1p and ZAP-S-tGFP co-localized in discrete cytoplasmic foci in~68% of cells that co-expressed both ORF1p and ZAP-S-tGFP ( Fig  5A). To test if transfected ORF1p co-localizes with endogenous ZAP, we transfected HeLa cells with pAD2TE1, which expresses a human L1 (L1.3) containing a T7 gene10 epitope-tag on the corner of the respective blots. The black arrow indicates the position of the full-length L1 RNA. The blue and yellow arrows indicate shorter L1 RNA species. The experiment was repeated three times with similar results. Actin served as a loading control. RNA size standards (~kb) are shown at the right of the blot image. Bottom panel: Quantification of northern blot bands. The X-axis indicates the cDNA expression construct that was co-transfected with pJM101/ L1.3Δneo. The Y-axis indicates relative band intensity normalized to pCEP4 controls (100%). Black bars represent the full-length L1 band. Blue and yellow bars represent the smaller L1 RNA bands, corresponding to the colored arrows, respectively, in the top panel. The results are the average of three independent experiments. Error bars represent standard deviations. (C) Schematic of pJBM2TE1: The construct contains a T7 epitope tag on the carboxylterminus of ORF1p and a TAP tag on the carboxyl-terminus of ORF2p. An mneoI retrotransposition indicator cassette is present in the 3' UTR. pJMB2TE1 is expressed from a pCEP4 backbone, which has been modified to contain a puromycin selectable marker. A CMV promoter augments L1 expression and an SV40 polyadenylation signal (pA) is located downstream of the native L1 polyadenylation signal.  carboxyl-terminus of ORF1p [26]. Confocal microscopy revealed that endogenous ZAP colocalized with ORF1p-T7 in cytoplasmic foci in~91% of cells that contained ORF1p-T7 foci ( Fig 5B). Next, to test if endogenous ORF1p co-localizes with transfected ZAP-S, we transfected PA-1 cells (a human embryonic carcinoma-derived cell line that expresses endogenous ORF1p [106,107]) with ZAP-S-tGFP. Confocal microscopy demonstrated that endogenous ORF1p colocalized with ZAP-S-tGFP in~89% of PA-1 cells that expressed ZAP-S-tGFP foci (Fig 5C). Thus, ORF1p and ZAP generally localize to the same region of the cytoplasm.

Discussion
In this study, we identified 39 cellular proteins that interact with L1 ORF1p and validated 13 of these interactions in biochemical assays. Our data showed that the 13 validated ORF1pinteracting proteins associate with ORF1p via an RNA bridge (Fig 1D). Notably, 33 out of 39 of the ORF1p-interacting proteins also were detected in recent studies (S1 Table; [47,67,89]). Importantly, we discovered that ZAP restricts human L1 and Alu retrotransposition. We also showed that hnRNPL, MOV10, and PURA inhibit L1 retrotransposition, which is in agreement with previous studies [64][65][66][67]89]. Thus, our data both confirm and extend those previous analyses and will help guide future studies that endeavor to determine how L1 retrotransposition impacts the human genome.
ZAP inhibits the mobility of both human and non-human non-LTR retrotransposons. The overexpression of the human and rat orthologs of ZAP restricts human L1 retrotransposition ( Fig 3A). Human ZAP-S overexpression restricts the retrotransposition of an engineered human SINE (Alu) (Fig 3B), an engineered mouse L1 (G F 21), and an engineered zebrafish LINE-2 element (ZfL2-2) (Fig 3C). Although our studies primarily involved the overexpression of ZAP, we also demonstrated that the depletion of endogenous ZAP in HeLa cells led to añ 2-fold increase in L1 retrotransposition (Fig 3D). This observed increase in L1 retrotransposition activity is similar to increases in L1 activity that were observed upon depletion of MOV10 and hnRNPL proteins in other studies [64,65,67]. Thus, in principle, physiological levels of ZAP may be sufficient to influence retrotransposition in certain cell types.
The ZAP CCCH zinc-finger domain is required to both bind and mediate the degradation of viral RNA [92,109]. Our data indicate that ZAP binding to L1 RNA is critical for L1 restriction. We demonstrated that ORF1p-FLAG and ZAP interact via an RNA bridge (Fig 1D). Moreover, we showed that overexpression of the ZAP zinc-finger domain more potently inhibits L1 retrotransposition than overexpression of wild type ZAP-L or ZAP-S (Fig 3A), and that the ZAP-S zinc-finger domain is required to inhibit L1 retrotransposition (Fig 3A and S4A  Fig). In addition to our genetic and biochemical data, fluorescence microscopy revealed that: 1) co-transfected ZAP-S, L1 RNA, and ORF1p co-localize in the cytoplasm of HeLa cells; 2) the ZAP-S zinc-finger domain is necessary and sufficient for the co-localization of ZAP-S, L1 RNA, and ORF1p; 3) endogenous ZAP co-localizes with transfected ORF1p in HeLa cells; and 4) endogenous ORF1p interacts with transfected ZAP-S in human PA-1 embryonic carcinoma cells (Figs 5A-5E and 6A-6D). Thus, the data suggest that ZAP interacts with L1 RNA in order to mediate L1 restriction.
Notably, the zebrafish ZfL2-2 retrotransposon lacks a homolog to ORF1 and only encodes a single ORF that contains an apurinic/apyrimidinic endonuclease-like (EN) and a reverse transcriptase (RT) domain [95]. The finding that ZAP-S efficiently restricts ZfL2-2 retrotransposition further indicates that ZAP-S likely restricts retrotransposition by interacting with LINE RNA. Although a ZAP consensus RNA target sequence/motif has not yet been identified, evidence suggests that ZAP recognizes long RNA stretches (>500 nucleotides) and/or specific RNA tertiary structure [91,92]. The ability of ZAP to inhibit non-human LINE elements suggests that ZAP may not recognize a particular LINE linear consensus RNA sequence, but instead may recognize an unidentified structural feature common to certain LINE RNAs [91,92].
Evidence suggests that ZAP prevents the accumulation of viral RNAs in the cytoplasm [72]. ZAP-S overexpression significantly reduced the amount of polyadenylated, full-length L1 RNA (Fig 4B), which would be expected to inhibit retrotransposition by limiting the supply of L1 mRNA available for translation and as a template for TPRT. Notably, while ZAP-S selectively inhibited the accumulation of full-length L1 transcripts, it did not dramatically affect the accumulation of shorter, spliced and/or polyadenylated L1 RNAs (Fig 4B) [99][100][101]110]. Thus, ZAP does not appear to affect L1 transcription per se, but instead likely affects the post-transcriptional processing of full-length L1 mRNA. In addition to biochemical data, fluorescence microscopy revealed that L1 RNA was depleted from L1 ORF1p cytoplasmic foci in the presence of ZAP (S5A Fig). The depletion of RNA from L1 cytoplasmic foci was dependent on the ZAP-S zinc-finger domain. Based on these data it is likely that ZAP prevents the accumulation of cytoplasmic L1 mRNA.
Previous studies have shown that ZAP also suppresses the expression of viral proteins [111][112][113]. Western blot experiments demonstrated that the overexpression of ZAP-S inhibited the accumulation of L1 ORF1p and L1 ORF2p in whole cell lysates and RNPs derived from transfected HeLa cells (Fig 4D and S3B Fig). In agreement with western blot experiments, confocal microscopy experiments showed that tGFP-tagged ZAP-S inhibited the expression of ORF1p in transfected HeLa cells and that the ZAP-S zinc-finger domain is critical for the inhibition of ORF1p expression (S5B Fig). ZAP-S expression also inhibited Alu retrotransposition (Fig 3B), which depends on ORF2p to be supplied in trans by L1 [49]. In contrast to these data, ZAP-S overexpression did not significantly affect the expression and/or accumulation of EGFP or other endogenous proteins (e.g., eiF3 and tubulin) (Fig 4D and S3B Fig). Thus, ZAP may preferentially limit the accumulation of ORF1p and ORF2p by interacting with L1 mRNA.
In sum, our data suggest that ZAP restricts L1 retrotransposition by preventing the accumulation of cytoplasmic L1 RNA. Notably, a recent study suggests that ZAP may interfere with translation of viral RNA, and that translation inhibition may precede viral RNA destruction [113]. Although a reduction in L1 RNA could explain the observed decrease in L1 protein expression, it also is conceivable that the interaction between ZAP and L1 RNA could interfere with L1 translation (Fig 7).
It remains unclear how ZAP might destabilize full-length L1 RNA to restrict retrotransposition. Evidence suggests that ZAP recruits exosome components [109] along with other proteins involved in RNA degradation [111,114] to destroy viral RNA. Interestingly, immunofluorescence microscopy experiments revealed that ORF1p and ZAP co-localize with components of cytoplasmic SGs (S4B and S4D Fig), which contain numerous RNA binding proteins involved in cytosolic RNA metabolism [reviewed in 115]. Indeed, ZAP previously has been shown to localize to SGs [108] and SGs have been suggested to play a role in regulating L1 retrotransposition [104] and viral pathogenesis [reviewed in 116]. The co-localization of ORF1p and ZAP with SGs also suggests that ZAP may possibly inhibit L1 translation, as SG assembly is stimulated by translational arrest [reviewed in 115]. Thus, we propose that ZAP interacts directly with L1 RNA in the cytoplasm, which likely results in the recruitment in SG components and/or other cellular factors involved in RNA metabolism to destabilize L1 RNA and/or block translation (Fig 7).

Plasmids
Oligonucleotide sequences and cloning strategies used in this study are available upon request. All human L1 plasmids contain the indicated fragments of L1.3 (accession no. L19088) [5] DNA cloned into pCEP4 (Invitrogen) unless otherwise indicated. A CMV promoter augments expression of all L1 and cDNA expressing plasmids unless noted otherwise. L1 plasmids also contain an SV40 polyadenylation signal that is located downstream of the native L1 polyadenylation signal. All plasmid DNA was prepared with a Midiprep Plasmid DNA Kit (Qiagen).
pAluneo Tet : expresses an Alu element cloned from intron 5 of the human NF1 gene [129] that is marked with the neo Tet reporter gene. The reporter [130] was subcloned upstream of the Alu poly adenosine tract [49]. pCEP/GFP: is a pCEP4 based plasmid that expresses the humanized renilla green fluorescent protein (hrGFP) coding sequence from phrGFP-C (Stratagene), which is located downstream of the pCEP4 CMV promoter [33]. pJJ101/L1.3: is a pCEP4 based plasmid that contains an active human L1 (L1.3) equipped with an mblastI retrotransposition indicator cassette [85]. pJJ105/L1.3: is similar to pJJ101/L1.3, but contains a D702A missense mutation in the RT active site of L1.3 ORF2 [85]. pJM101/L1.3Δneo: is a pCEP4 based plasmid that contains an active human L1 (L1.3) [35]. pLRE3-EF1-mEGFPΔIntron: is a pBSKS-II+ based plasmid that expresses an active human L1 (LRE3) that is tagged with an EGFP cassette (mEGFPI) containing an antisense, intronless copy of the EGFP gene. A UbC promoter drives EGFP expression. An EF1α promoter drives L1 expression [103]. pAD2TE1: is similar to pJM101/L1.3 except that it was modified to contain a T7 gene10 epitope-tag on the carboxyl-terminus of ORF1p and a TAP epitope-tag on the carboxylterminus of ORF2p. The 3 0 -UTR contains the mneoI retrotransposition indicator cassette [26]. pJBM2TE1: is similar to pAD2TE1 except that the pCEP4 backbone was modified to contain the puromycin resistance (PURO) gene in place of the hygromycin resistance gene. pLRE3-mEGFPI: is a pCEP4 based plasmid that contains an active human L1 (LRE3) equipped with an mEGFPI retrotransposition indicator cassette [97,106]. The pCEP4 backbone was modified to contain a puromycin resistance (PURO) gene in place of the hygromycin resistance gene. The CMV promoter also was deleted from the vector; thus, L1 expression is driven only by the native 5 0 UTR [97].
pK_A3A: expresses HA-tagged APOBEC3A and was a generous gift from Dr. Brian Cullen [131]. pDCP1α-GFP: expresses a GFP-tagged version of DCP1α and was a generous gift from Dr. Gregory Hannon [132]. pG3BP-GFP: expresses a GFP-tagged version of G3BP and was a generous gift from Dr. Jamal Tazi [133]. pcDNA6/TR: expresses the blasticidin resistance gene and was obtained from Invitrogen.

Immunoprecipitation of L1 ORF1p
HeLa-JVM cells were seeded in T-175 flasks (BD Falcon) at~6-8×10 6 cells/flask and transfected the next day with 20 μg of plasmid DNA using 60 μL of FuGENE HD (Promega). Approximately 48 hours post-transfection, hygromycin B (Gibco) (200 μg/mL) was added to the medium to select for transfected cells. After approximately one week of hygromycin selection, cells were washed 3 times with ice cold PBS and collected with a rubber policeman into 50 mL conical tubes (BD Falcon). Cells were then pelleted at 1,000×g and frozen at -80°C. To produce whole cell lysates (WCL), frozen cell pellets were rapidly thawed and then lysed in~3 mL (1 mL lysis buffer per 100 mg of cell pellet) of lysis buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 10% glycerol, 1 mM EDTA, 0.1% IGEPAL CA-630 (Sigma), 1X complete EDTA-free protease inhibitor cocktail (Roche)) on ice for 30 minutes. WCLs were then centrifuged at 15,000×g for 15 minutes at 4°C. Supernatants were transferred to a clean tube and protein concentration was determined using the Bradford reagent assay (BioRad). For the IP,~1 mL of the supernatant (~3 mg total protein) was pre-cleared with~15 μL (packed gel volume) of mouse IgG-agarose beads (Sigma) for 4 hours at 4°C. Pre-cleared supernatants were then mixed with 15 μL (packed gel volume) of EZview Red ANTI-FLAG M2 Affinity Gel (Sigma) and incubated overnight with rotation at 4°C. The beads then were rinsed 3x with 0.5 mL of lysis buffer, and then washed 3 times with 0.5 mL of lysis buffer for 10 minutes per wash on ice with gentle agitation. Protein complexes were eluted from the beads by adding~70 μL of 2X NuPAGE LDS Sample Buffer (Novex), supplemented with NuPAGE Sample Reducing Agent (Novex), directly to the washed beads and incubating for 10 minutes at 70°C. Following incubation, the beads were pelleted and the sample was transferred to a fresh tube. For SDS-PAGE analysis, 20 μL of the IP were loaded onto a 4-15% gradient midi-gel (BioRad) and run under reducing conditions. Gels were silver stained using the SilverQuest Silver Staining Kit (Novex) to visualize proteins.

Protein identification by LC-MS/MS
The Proteomics Facility at the Fred Hutchinson Cancer Research Center (Seattle, WA) conducted protein identification experiments. Excised silver-stained gel slices were destained and subjected to in-gel proteolytic digestion with trypsin as described [134]. Following gel-slice digestion, the digestion products were desalted using C18-micro ZipTips (Millipore) and were dried by vacuum centrifugation. The resultant peptide samples were resuspended in 7 μL of 0.1% formic acid and 5 μL were analyzed by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). LC-MS/MS analysis was performed using an LTQ Orbitrap XL mass spectrometer (Thermo Scientific). The LC system, configured in a vented format [135], consisted of a fused-silica nanospray needle (PicoTip emitter, 50 μm ID) (New Objective) packed in-house with Magic C18 AQ 100A reverse-phase medium (25 cm) (Michrom Bioresources Inc.) and a trap (IntegraFrit Capillary, 100 μm ID) (New Objective) containing Magic C18 AQ 200A reverse-phase medium (2 cm) (Michrom Bioresources Inc.). The peptide samples were loaded onto the column and chromatographic separation was performed using a two mobile-phase solvent system consisting of 0.1% formic acid in water (A) and 0.1% acetic acid in acetonitrile (B) over 60 min from 5% B to 40% B at a flow rate of 400 nL/minutes. The mass spectrometer operated in a data-dependent MS/MS mode over the m/z range of 400-1800. For each cycle, the five most abundant ions from each MS scan were selected for MS/MS analysis using 35% normalized collision energy. Selected ions were dynamically excluded for 45 seconds.
For data analysis, raw MS/MS data were submitted to the Computational Proteomics Analysis System (CPAS), a web-based system built on the LabKey Server v11.2 [136] and searched using the X!Tandem search engine [137] against the International Protein Index (IPI) human protein database (v3.75), which included additional common contaminants such as BSA and trypsin. Search results were compared between the pJM101/L1.3FLAG lane and the pJM101/ L1.3 lane to generate a list of candidate L1 ORF1p associated proteins unique to the pJM101/ L1.3FLAG immunoprecipitation. The search output files were analyzed and validated by Pro-teinProphet [138]. Peptide hits were filtered with PeptideProphet [139] error rate 0.05, and proteins with probability scores of 0.95 were accepted. Suspected contaminants (e.g. keratin) were filtered from the final L1 RNP candidate list.

L1 retrotransposition assays
The cultured cell retrotransposition assay was carried out essentially as described [30,71]. For retrotransposition assays with L1 constructs tagged with mblastI, HeLa-JVM cells were seeded at~1-2×10 4 cells/well in a 6-well plate (BD Falcon). Within 24 hours, each well was transfected with 1 μg of plasmid DNA (0.5 μg L1 plasmid + 0.5 μg cDNA plasmid or pCEP4) using 3 μL of FuGENE 6 transfection reagent (Promega). Four days post-transfection, blasticidin (EMD Millipore) containing medium (10 μg/mL) was added to cells to select for retrotransposition events. Medium was changed every two days. After~8 days of selection, cells were washed with PBS, fixed, and then stained with crystal violet to visualize colonies. To control for transfection efficiency and off-target effects of cDNA plasmids, in parallel with retrotransposition assays, HeLa-JVM cells were plated in 6-well plates at 500-1,000 cells/well and transfected with 0.5 μg pcDNA6/TR (Invitrogen) plasmid + 0.5 μg cDNA plasmid using 3 μL of FuGENE 6 transfection reagent (Promega). The pcDNA6/TR control assays were treated with blasticidin in the same manner as for retrotransposition assays.
For retrotransposition assays with L1 constructs tagged with mneoI, HeLa-JVM cells were transfected as described above. Two days after transfection, cells were treated with medium supplemented with G418 (Gibco) (500 μg/mL) for~10-12 days. As a control, HeLa cells were plated at~2×10 4 cells/well in a 6-well plate and transfected with 0.5 μg pcDNA3 (Invitrogen) plasmid + 0.5 μg cDNA plasmid using 3 μL of FuGENE 6 transfection reagent (Promega). The pcDNA3 control assays were treated with G418 in the same manner as for retrotransposition assays.
After 4 days of puromycin selection, the percentage of GFP positive cells was determined by flow cytometry using an Accuri C6 flow cytometer (BD Biosciences).

RNP isolation
RNPs were isolated as previously described [37]. Briefly, HeLa-JVM cells were seeded onto 60 mm tissue culture dishes (BD Falcon) and 24 hours later cells were co-transfected with 2.5 μg of pJBM2TE1 and 2.5 μg of the indicated cDNA plasmid using 15 μL of FuGENE HD (Promega). Approximately two days after transfection, puromycin (5 μg/mL) was added to culture medium to select for cells transfected with pJBM2TE1. After~3 days of puromycin selection (5 days after transfection), cells were lysed in RNP lysis buffer (150 mM NaCl, 5 mM MgCl 2 , 20 mM Tris-HCl (pH 7.5), 10% glycerol, 1mM DTT, 0.1% NP-40, and 1x complete EDTA-free protease inhibitor cocktail (Roche)). Following lysis, whole cell lysates were centrifuged at 12,000xg for 10 minutes at 4°C, and then the cleared lysate was layered onto a sucrose cushion (8.5% and 17% sucrose) and subjected to ultracentrifugation at 4°C for 2 hours at 178,000xg. The supernatant was discarded and the resulting pellet was resuspended in water supplemented with 1x complete EDTA-free protease inhibitor cocktail (Roche). Approximately 20 μg (total protein) of the RNP sample or~30 μg (total protein) of the cleared whole cell lysate (supernatant post 12,000xg centrifugation) were then analyzed by western blot. Blots were analyzed using an Odyssey 1 CLx (LI-COR) with the following secondary antibodies: IRDye 800CW Donkey anti-Rabbit IgG (1:10,000) (LI-COR) and IRDye 680RD Donkey anti-Mouse IgG (1:10,000) (LI-COR).
To simultaneously analyze the effects of ZAP-S on ORF1p and EGFP protein expression, HeLa-JVM cells were seeded onto 10 cm dishes (~2.7×10 6 cells/dish) (BD Falcon) and transfected with 10 μg of plasmid DNA (5.0 μg pLRE3-EF1A-mEGFPΔIntron + 5.0 μg cDNA plasmid or pCEP4) using 30 μL of FuGENE HD. After 48 hours, cells were harvested with trypsin and then subjected to flow cytometry to isolate GFP expressing cells. Approximately 1.2-1.7×10 6 GFP positive cells were collected for each transfection condition using a MoFlo Astrios cell sorter (Beckman Coulter). The GFP gate was set using untransfected HeLa-JVM cells. The sorted cells were lysed as described in the IP procedure and lysates were then subjected to western blotting using standard procedures. For all other protein expression analyses, HeLa-JVM cells were seeded at~4×10 5 cells/well in 6-well plates and transfected with 2 μg of plasmid DNA with 6 μL of FuGENE HD. Cells were collected 48 hours after transfection using a rubber policeman and lysates were prepared as described above. Western blots were visualized using either the SuperSignal West Femto Chemiluminescent Substrate (Pierce) or SuperSignal West Pico Chemiluminescent Substrate (Pierce) and Hyperfilm ECL (GE Healthcare).

Northern blots
HeLa-JVM cells were seeded in T-175 flasks (BD Falcon) and transfected with 20 μg of plasmid DNA (10 μg pJM101/L.13Δneo + 10 μg cDNA plasmid) using 60 μL FuGENE HD. Two days after transfection, cell pellets were collected and frozen at -80°C. Frozen cell pellets were then thawed and total RNA was extracted with TRIzol reagent (Ambion), and then poly(A)+ RNA was prepared from total RNA using an Oligotex mRNA kit (Qiagen). Each sample (~1.5 μg of poly(A)+ RNA) was subjected to glyoxal gel electrophoresis and northern blotting using the NorthernMax-Gly Kit (Ambion) according to the manufacturer's protocol. Following electrophoresis, RNA was transferred to BrightStar Nylon membranes (Invitrogen) and then crosslinked using UV light. For northern blot detection, membranes were prehybridized for~4 hours at 68°C in NorthernMax Prehybridization/Hybridization Buffer (Ambion), and then incubated with a strand specific RNA probe (final concentration of probe~3×10 6 cpm ml -1 ) overnight at 68°C. For band quantification, northern blot films were analyzed using ImageJ software [140].
The T3 promoter sequence (underlined) was added to the reverse primer of each primer pair. The pTRI-β-actin-125-Human Antisense Control Template (Applied Biosystems) was used in T3 reactions as a template to generate the β-actin RNA probe. Each northern blot experiment was independently repeated three times with similar results.

Immunofluorescence microscopy
Immunofluorescence microscopy was performed essentially as described [26] with modifications. Briefly, cells were plated on round glass cover slips (Fisher) in a 12-well plate or into 4-well chambered glass slides (Fisher) and transfected~24 hours later with 0.5 μg of plasmid DNA using 1.5 μL of FuGENE 6 transfection reagent. To visualize proteins, approximately 48 hours post-transfection cells were washed with 1x PBS, fixed with 4% paraformaldehyde for 10 minutes and then treated with ice-cold methanol for 1 minute. Next, cells were incubated for 30 minutes at 37°C in 1x PBS + 3% BSA. Cells then were incubated with primary antibodies in 1x PBS + 3% BSA for 1 hour at 37°C. Cells were washed three times with 1x PBS (10 minutes per wash) and then incubated with appropriate, fluorescently-labeled secondary antibodies diluted in 1x PBS for 30 minutes at 37°C. The following secondary antibodies were used for indirect immunofluorescence: Alexa Fluor 488 conjugated Goat anti-Mouse and Goat anti-Rabbit (Invitrogen) (1:1000), Alexa Fluor 546 conjugated Goat anti-Mouse and Goat anti-Rabbit IgG (Invitrogen) (1:1000), and Cy5 conjugated Donkey anti-Rabbit IgG (H+L) (Jackson ImmunoResearch) (1:100). To obtain images, a cover slip and/or slide was visually scanned and representative images were captured using a Leica SP5X confocal microscope (63x/1.4 objective; section thickness 1 μm).
Supporting Information S1 Fig. Supporting data for Fig 1. (A) The FLAG epitope on ORF1p is compatible with retrotransposition: Constructs were tested in a transient HeLa cell retrotransposition assay [30,71]. The X-axis indicates the L1 plasmid. The Y-axis indicates the retrotransposition efficiency. Retrotransposition assays were normalized to pJM101/L1.3 (100%). The pJM105/L1.3 plasmid serves as a negative control and harbors a point mutation in the ORF2p RT domain that renders the element inactive [30]. Representative results from a single experiment are depicted below the graph. The assay was repeated two times with similar results. (B) Immunoprecipitation reactions conducted using various wash conditions: Top panel: HeLa cells were transfected with pCEP4, pJM101/L1.3, or pJM101/L1.3FLAG and were subjected to lysis using two different salt concentrations (500 mM NaCl (left gel) or 150 mM NaCl (right gel)). Shown are the images of silver stained gels from immunoprecipitation reactions. The black rectangles indicate the cropped image depicted in Fig  (A) A schematic of the pLRE3-EF1-mEGFPΔIntron: pLRE3-EF1-mEGFPΔIntron expresses a human L1 (LRE3) that is tagged with an mEGFPI expression cassette that lacks an intron. The human elongation factor-1 alpha (EF1α) promoter (arrow) augments L1 transcription. The ubiquitin C (UbC) promoter (upside down arrow) drives EGFP transcription. (B) ZAP-S inhibits ORF1p expression: Western blots were conducted using whole cell lysates derived from cells co-transfected with pLRE3-EF1-mEGFPΔIntron and the ZAP-S expression plasmid or pCEP4 indicated above each lane. UTF indicates whole cell lysates from untransfected HeLa cell. Antibodies are indicated on the right side of each blot. Tubulin is used as a loading control. Western blot images depict a representative experiment that was repeated three times with similar results. Notably, upon extended exposure times ORF1p was able to be visualized in the ZAP-S lane. (C) ZAP-S does not inhibit EFGP expression and/or accumulation: HeLa cells were cotransfected with pLRE3-EF1-mEGFPΔIntron and the indicated expression plasmids (noted above the plots). Flow cytometry was used to determine the percentage of EGFP-positive, livegated cells for each condition. UTF indicates untransfected HeLa cells. The EGFP-positive gate was set using the UTF sample as a negative control. The X-axis depicts the percentage of EGFP positive cells. The Y-axis indicates the side scattering profile (SSC). Approximately 1.2-1.7 x 10 6 GFP positive cells were collected and analyzed for each transfection condition. (TIF) Results of pJJ101/L1.3 retrotransposition assays. The X-axis indicates the cDNA co-transfected with pJJ101/L1.3 or pcDNA6/TR. The Y-axis indicates pJJ101/L1.3 retrotransposition activity (black bars). All values have been normalized to the pCEP4 empty vector control (100%). The numbers above the bar graphs indicate the number of biological replicates performed with each cDNA expression construct. Error bars represent standard deviations. (B-E) ORF1p and ZAP co-localize with stress granules in HeLa cells: HeLa cells were co-transfected with pJM101/ L1.3Δneo and ZAP-S-tGFP; proteins were visualized by direct immunofluorescence. ORF1p co-localizes with ectopic ZAP-S-tGFP and eIF3 in cytoplasmic foci (panel B). ORF1p co-localizes with ectopic ZAP-S-tGFP (panels A and B), but not with tubulin (panel C). GFPtagged G3BP co-localizes with endogenous ZAP in cytoplasmic foci (panel D). GFP-tagged DCP1α forms cytoplasmic punctate structures, which do not appear to co-localize with endogenous ZAP (panel E). The right-most image in each panel represents a merged image. The cell type is indicated at the top left (yellow), the protein name is listed on the bottom left, and the name of the primary antibody used (italicized) is annotated at the bottom right. Nuclei were stained with DAPI (blue) and the scale bar represents 25 μM. (TIF) S5 Fig. Supporting data for Fig 6. (A) Fluorescence microscopy was used to determine the percentage of cells that contained L1 RNA in cytoplasmic ORF1p foci. The X-axis indicates the plasmid that was co-transfected with pJM101/L1.3. The Y-axis of the graph depicts the percentage of cells where L1 RNA was detected in cytoplasmic ORF1p foci. Experiments were repeated three times. A total of~60 visual fields (~1600 cells) were examined amongst all three experiments and~33-41 ORF1p foci containing cells were evaluated for each experimental condition. Error bars represent standard deviations. (B) Confocal microscopy was used to determine the number of ORF1p-expressing cells~48 hours post transfection. The X-axis indicates the plasmid that was co-transfected with pJM101/L1.3Δneo. The Y-axis of the graph depicts the percentage of cells that express ORF1p. Experiments were repeated twice. Each experiment contained two biological replicates and~1100-1500 cells were enumerated amongst all experiments for each condition. Error bars indicate standard deviations. (TIF) S1 Table. L1 ORF1p-interacting protein candidates identified by LC-MS/MS. ORF1pinteracting proteins were selected based on the criteria that the protein was unique to the pJM101/L1.3FLAG IP and was identified by two or more unique peptides (peptide error 0.05; protein probability 0.95). Column 1 = protein name. Column 2 = protein mass. Column 3 = total number of identified peptides. Column 4 = number of unique peptides. Column 5 = the percentage of amino acid coverage for each of the respective proteins. Columns 6-8 = whether the proteins were identified in the indicated studies [47,67,89]. Green highlighting indicates ORF1p-interacting candidates that were verified by western blot (Fig 1D). "Y" in columns 6-8 indicate that the protein was identified as a significant L1-interacting protein by statistical and/or direct biochemical methods; "n.s." in column 7 indicates that the protein was identified in the study, but that it did not reach the significance threshold set by the authors. (TIF)