Differential L1 regulation in pluripotent stem cells of humans and apes.

Identifying cellular and molecular differences between human and non-human primates (NHPs) is essential to the basic understanding of the evolution and diversity of our own species. Until now, preserved tissues have been the main source for most comparative studies between humans, chimpanzees (Pan troglodytes) and bonobos (Pan paniscus). However, these tissue samples do not fairly represent the distinctive traits of live cell behaviour and are not amenable to genetic manipulation. We propose that induced pluripotent stem (iPS) cells could be a unique biological resource to determine relevant phenotypical differences between human and NHPs, and that those differences could have potential adaptation and speciation value. Here we describe the generation and initial characterization of iPS cells from chimpanzees and bonobos as new tools to explore factors that may have contributed to great ape evolution. Comparative gene expression analysis of human and NHP iPS cells revealed differences in the regulation of long interspersed element-1 (L1, also known as LINE-1) transposons. A force of change in mammalian evolution, L1 elements are retrotransposons that have remained active during primate evolution. Decreased levels of L1-restricting factors APOBEC3B (also known as A3B) and PIWIL2 (ref. 7) in NHP iPS cells correlated with increased L1 mobility and endogenous L1 messenger RNA levels. Moreover, results from the manipulation of A3B and PIWIL2 levels in iPS cells supported a causal inverse relationship between levels of these proteins and L1 retrotransposition. Finally, we found increased copy numbers of species-specific L1 elements in the genome of chimpanzees compared to humans, supporting the idea that increased L1 mobility in NHPs is not limited to iPS cells in culture and may have also occurred in the germ line or embryonic cells developmentally upstream to germline specification during primate evolution. We propose that differences in L1 mobility may have differentially shaped the genomes of humans and NHPs and could have continuing adaptive significance.

Humans, chimpanzees and bonobos are genetically very similar, sharing nearly 98% of their alignable genomic sequence[1-3]. However, cellular and molecular phenotypes, especially at identical stages of development, are difficult to establish, mainly due to limited access to embryonic material from human and NHPs[8]. We reprogrammed fibroblasts from two bonobos and two chimpanzees into iPSCs as previously described[9,10] (). After culture in human ESC (hESC)-supporting conditions, NHP iPSC colonies could be distinguished by the high nucleus-to-cytoplasm ratio morphology. iPSC clones from both species continuously expressed pluripotency markers, retained undifferentiated morphology in culture, and maintained a normal karyotype (). After embryoid body (EB)-mediated differentiation in vitro, clones contained tissue derivatives from the three embryonic germ layers and down-regulated expression of pluripotency markers (). iPSCs selected clones were also able to differentiate into the three embryonic germ layers in vivo as shown by analysis of teratomas in nude mice (). Together, these data demonstrate that NHP iPSC clones re-established pluripotency at the molecular and cellular levels.

To gain insight into differences in gene expression between human and NHP iPSCs, we performed RNA-seq analyses on four human, two chimpanzee and two bonobo iPSC lines (. The expression profiles of iPSCs from the three species clustered together with hESCs (HUES6 and H1) and were distinguishable from hESC-derived neural precursor cells (NPCs) (), chimpanzee and bonobo iPSCs clustered closer to each other than to human iPSCs (). We then performed pairwise comparisons of protein-coding gene expression levels (. Venn diagrams represent expressed genes with non-significant differences between species (purple) and up-regulated genes with estimated false discovery rates (FDR) of less than 5% and a fold change greater than 2-fold (pink and blue). Comparison between human and NHP ( bottom-right) revealed 1,376 genes with elevated expression in human iPSCs and 1,042 common genes with elevated expression in NHP iPSCs, whereas no significant differences were observed in 11,585 protein-coding genes. Next, we focused on genes differentially expressed between human and NHP iPSCs (, and found, among the top 50 genes with elevated expression in human iPSCs compared to NHP iPSCs, two genes involved in restriction of L1 retrotransposition, namely A3B and PIWIL2 ().

Active, full-length L1 elements have the ability to move from one location in the genome to another by a copy-paste mechanism called retrotransposition[11]. Active L1 elements have been detected in both germline and somatic tissues, and can impact genome integrity[12,13]. As uncontrolled retrotransposition activity can be deleterious to the host[14], organisms have evolved mechanisms to control L1 mobility[11]. A3B is a member of the APOBEC3 family of cytidine deaminases that can inhibit L1 mobility in different cell types, including human embryonic stem cells (hESCs) and iPSCs, via a still unclear mechanism[6,15,16]. PIWIL2 is an effector of the piRNA pathway involved in L1 silencing mainly in germ line[7].

To confirm differences in A3B and PIWIL2 in human versus NHP iPSCs, we first cloned their cDNAs from the three species, and found a high degree of conservation between human and NHPs (). Quantification of A3B mRNA levels by qPCR confirmed significantly higher levels (~30-fold) of A3B in both human iPSC lines compared to NHP iPSCs (). Levels of PIWIL2 mRNA were 16-fold higher in human iPSC lines than in NHP iPSCs (). PIWIL2-mediated control of transposons is most active in germline and we observed that levels of PIWIL2 mRNA are 20 to 40-fold lower in human iPSCs than in testis (). The increased expression observed in human iPSCs appears to be specifically restricted to A3B and PIWIL2 compared to other members of these protein families ( Differences in A3B and PIWIL2 mRNA levels reflected higher A3B and PIWIL2 protein levels in human versus NHP iPSCs ().

Ectopic expression of A3B has been shown to inhibit the mobility of human L1 reporter elements[6,17] ()[18,19]. In 293T cells, ectopic expression of human A3B significantly reduced L1-expressing firefly luciferase (L1-Luc)[18,19] mobility by 5-fold compared to control plasmid or a plasmid expressing A3G, another APOBEC3 protein that lacks anti-L1 activity ([17]. We also found a significant decrease in L1-Luc retrotransposition in cells overexpressing PIWIL2 compared to control transfected cells (). We then confirmed that human L1 can retrotranspose in hESCs and hiPSCs under our culture conditions, as previously shown[15,20,21] ( Since we found reduced levels of L1 restriction factors A3B and PIWIL2 in NHPs, we compared L1 activity in human versus NHP iPSCs using the human L1-EGFP reporter element[18,19]. L1 retrotransposition was significantly higher in NHP compared to human iPSC lines, with 10- and 8-fold increases in EGFP-positive cells in chimpanzee and bonobo iPSCs, respectively (. To test whether the differential L1 regulation in iPSCs is specific to human L1, we measured the activity of a NHP L1 element in iPSCs. We generated a retrotransposition-competent chimpanzee L1-EGFP reporter element (Chimp-L1) (, and observed that Chimp-L1 was significantly more active in NHP iPSCs than in human iPSCs ( suggesting that the decreased L1 activity in human iPSCs is not specific to human L1 element, and that human iPSCs are more efficient in repressing L1 retrotranspostion than NHP iPSCs.

To analyze the contribution of endogenous A3B to the differential L1 activity observed among primate cells, we generated human ESCs and iPSCs with decreased levels of A3B (). Stable knockdown of A3B (shA3B) resulted in a significant increase in L1-EGFP activity compared to shScramble (shScr) control cells in human iPSCs ( Knockdown was specific to A3B, not affecting other APOBEC3s (. As expected, L1 mobility was significantly decreased in both chimpanzee and bonobo iPSC lines after A3B overexpression (. Similarly, overexpression of chimpanzee or bonobo PIWIL2 in NHP iPSCs reduced L1 mobility to levels detected in human iPSCs (. We did not observe differences in the levels of L1-EGFP mRNA expressed from transfected L1 plasmid or in L1 promoter activity between human and NHP iPSCs (). Together these results suggest that differences in A3B and PIWIL2 expression levels contribute to higher L1 retrotransposition in NHP iPSCs compared to human iPSCs

PIWIL2 repression of transposons is mediated through piRNAs[22]. Thus, we analyzed the presence of PIWIL2-bound piRNAs in doxycycline-inducible human iPSCs expressing Flag-tagged PIWIL2 by immunoprecipitation and subsequent 5’ end labeling. Analysis of PIWIL2-associated small RNAs revealed the presence of ~26-30 nt RNAs only in cells expressing Flag-PIWL2 but not in control cells or in pull-downs with control antibody (). Next, to probe for the presence of L1-targeting-piRNAs we characterized the small RNA populations in human iPSCs by small RNA-seq analysis (Extended Data Fig. 6c,d, Supplementary Table 1 and 2). We detected 272 and 229 annotated piRNAs in human iPSC1 and iPSC2, respectively (. In addition, we observed a number of 26-33 nt small RNAs mapping to the consensus human-specific L1 element (L1Hs) sequence (), including 12 and 10 out of the 37 annotated piRNAs mapping to L1Hs in piRNAbank (http://pirnabank.ibab.ac.in/) in hiPSC1 and hiPSC2, respectively (). Together these results demonstrate the presence of piRNAs complementary to L1Hs in human iPSCs.

Extended Data Figure 6

Immunoprecipitation of piRNAs associated with PIWIL2 in human iPSCs and annotated piRNAs mapping to consensus L1Hs in iPSCs. (a) Immunoprecipitation of PIWIL2 RNPs using Flag-tag antibodies from Tet-inducible Flag-tagged PIWIL2 human iPSCs after addition of Doxyclycine to the culture media. HA-tag antibody was used as control. (b), γ32P-ATP end labeling of RNAs associated with Flag-PIWIL2 RNPs. Signal in the piRNAs size range is detected only in anti-Flag but not in control antibody anti-HA immunoprecipitates. (c) Size distribution of RNA reads detected by small RNA-seq from small RNAs samples extracted from human iPSC cell lines (hiPSC1 and hiPSC2). (d) Number of mapped reads per sample in small RNA-seq. (e) Number of annotated piRNAs (piRNAbank) detected by RNA-seq in human iPSC1 and iPSC2. (f) Characterization of 5’ end of piRNAs detected in human iPSCs relative to annotated piRNAs. Read count distribution relative to piRNA 5' ends (piRNAbank). (g) Sequences of annotated piRNAs (piRNAbank) mapping to consensus L1Hs detected in human iPSC1 and iPSC2. 26-33nt RNA reads from human iPS cell lines 1 and 2 (hiPSC1 and hiPSC2) characterized by RNASeq are aligned to annotated piRNAs mapping to the consensus L1Hs sequence. Analysis of mapping sequences was performed allowing 2 mismatches.

We then asked whether different levels of L1 reporter mobility between human and NHP iPSCs reflect differences in endogenous L1 activity. First, we analyzed endogenous L1 RNA levels by qPCR and found higher levels of endogenous L1 mRNA in chimpanzee and bonobo compared to human iPSCs (). Next, we examined the number of L1 elements in human and chimpanzee genomes to assess differences in recent L1 mobility. We did not observe major differences in the number of L1 elements for older families (L1PA4, L1AP3 and L1PA2; approximately 18, 12.5 and 7.6 Million year old, respectively)[23,24] (). However, we did observe significantly higher numbers of chimpanzee-specific L1 elements (L1Pt) compared to L1-Hs elements (L1Hs)[25,26]( Differences in expression of A3B and PIWIL2 suggest that L1 mobility may have been altered at a relatively recent evolutionary divergence. Therefore, using divergence as a measurement of L1 age, we estimated the number of species-specific L1 loci and found that the number of chimpanzee-specific loci was significantly higher than the number of human-specific loci (. The increased number of species-specific L1 loci in chimpanzee suggests that endogenous L1 has been more active in NHP genomes, correlating with the decreased levels of A3B and PIWIL2.

Here we report that iPSCs from both chimpanzees and bonobos have increased L1 mobility. Different rates of L1 activity could lead to significant changes in genomic structure and function, and could potentially impact adaptation. The human population has gone through one or more bottlenecks throughout evolution that might have contributed to decreased genetic diversity[27]. Chimpanzees and bonobos, in contrast, have increased levels of genetic diversity when compared to humans[28]. This idea is also supported by data showing that there is substantially more genetic difference among individuals within chimpanzee troops in West Africa than among all living humans[29]. While it remains unclear what the main generators of the phenotypic differences between us and our closest living relatives are (despite the extreme genetic similarity), we propose that L1 mobility could be playing a role in differentially shaping the genomes of humans and NHPs, providing an extra layer of variability to the latter. In fact, recent studies have suggested that ongoing L1 retrotransposition may contribute to inter individual genetic variation[30]. In this work we also present a new perspective on the use of iPSC technology as a powerful tool for the study of early stages of development and possible validation of evolutionary genomic and transcriptomic modifications that identify humans as outliers among primates. The great apes iPSCs that we describe here can also be used for comparative studies of any derivative pluripotent or terminally differentiated cell types, limited solely by the availability of differentiation protocols.

METHODS

Cell culture and retrovirus infection

hESC Hues6 and H1, human iPSC lines WT-33, ADRC-40 (human IPSC1 and IPSC2 in this work, respectively) and WT-126 were previously described[31]. Fibroblasts from human GM22159 (WT-9), Pan troglodytes (Chimpanzees: PR00818 and PR01209) and from Pan paniscus (Bonobos: AG05253 and PR01086) were acquired from Coriell Cell Repositories (NJ) (). All fibroblasts were cultured in Minimum Essential Medium (Invitrogen) supplemented with 10% fetal bovine serum (HyClone Laboratories). Retroviral vectors expressing Oct4, c-Myc, Klf4 and Sox2 human cDNAs from Yamanaka's group[9] were obtained from Addgene. Recombinant viruses were produced by transient transfection in 293T cells, as previously described[32]. Two days after infection, cells were plated on mitotically inactivated mouse embryonic fibroblasts (Chemicon) with hESC medium. After 2 to 4 weeks, iPSCs colonies were picked manually and directly transferred to feeder-free conditions on matrigel-coated dishes (BD) using mTeSR™1 (StemCell Technologies). Established iPSCs colonies were kept in feeder-free conditions indefinitely and passed using mechanical dissociation. Embryoid body (EB)-mediated differentiation in suspension was carried out for 10 days in the absence of growth factors. The use of chimpanzee and bonobo fibroblast samples was approved by the US Fish and Wildlife Service, under the permit MA206206. Protocols describing the use of iPSCs and hESCs were previously approved by the University of California San Diego and Salk Institute Institutional Review Board and the Embryonic Stem Cell Research Oversight Committee[31]. To generate stable shA3B cells lines, Hues6, WT33 and ADRC-40 cells were transduced with lentiviruses expressing shRNAs and selected for puromycin resistance. pLKO.1-based lentiviral plasmids encoding shRNAs against A3B (RHS3979-99216651 and RHS3979-99216658) were obtained from Open Biosystems. Recombinant lentiviruses were produced by transient transfection on 293T cells as previously described [33].

Teratoma formation in nude mice

Around 1-3 ×106 cells were injected subcutaneously into the dorsal flanks of nude mice (CByJ.Cg-Foxn1nu/J) anesthetized with isoflurane. Five to 6 weeks after injection, teratomas were dissected, fixed overnight in 10% buffered formalin phosphate and embedded in paraffin. Tissues were then prepared for histopathologic analysis by the UCSD Mouse Phenotyping Services (UCSD, CA http://mousepheno.ucsd.edu). Briefly, the tissue was sectioned and stained with hematoxylin and eosin. Control mice injected with fibroblasts failed to form teratomas.

Karyotyping

Standard G-banding chromosome analysis was performed by Cell Line Genetics (CLG, Madison, WI). Diploid human cells with 2n=46 chromosomes; bonobo and chimpanzee with 2n=48 chromosomes.

RNA extraction and RT-PCR

Total cellular RNA was extracted from ~5×106 cells using the RNeasy Protect Mini kit or RNeasy Plus kit (Qiagen, Valencia, CA), according to the manufacturer's instructions, and was reverse transcribed using the SuperScript III First-Strand Synthesis System RTPCR from Invitrogen. For iPSC markers, cDNA was amplified by PCR using Accuprime Taq DNA polymerase system (Invitrogen). Primer sequences are shown in . PCR products were separated by electrophoresis on a 2% agarose gel, stained with ethidium bromide and visualized by UV illumination. Total RNA samples from human testis were obtained from Clontech. Small RNA was extracted using the mirVana kit (Ambion).

Quantitative RT-PCR

RNA was extracted using a QIAGEN RNeasy Plus kit or TRIzol (Life Technologies) and then retrotranscribed to cDNA with the Superscript III First-Strand synthesis system (Invitrogen). qRT-PCRs were carried out using SYBR Green mix (Roche) or TaqMan Assays (Life Technologies) using ABI Prism 7900HT sequence detection system (Applied Biosystems). The primers and Taqman sets used in this work are described in . Data analysis was performed with SDS 2.3 software (Applied Biosystems). Primer efficiency was verified by linear regression to the standard curve. Values were normalized to GAPDH, HPRT or b-Actin. Reactions were carried out in triplicate and data were analyzed using the comparative (delta-delta Ct) method. For A3B and PIWIL2, RNA levels were normalized to GAPDH or b-Actin and represented as relative to IPSC-1. Relative A3B and PIWIL2 mRNA levels normalized to GAPDH for each individual iPSC line and fibroblasts are shown in The reduced levels of A3B in NHP iPSC were not due to an A3B deletion polymorphism previously described in human individuals[34] (data not shown). For L1 RNA quantitative RT-PCR, values representing the average of relative levels for L1 RNA (5’UTR, ORF1 and ORF2) were calculated, and were normalized to Actin mRNA levels. L1 levels are shown as relative to IPSC-1. Quantitative RT-PCR analysis of EGFP expression in iPSC lines transfected with L1-EGFP plasmid was carried out 60-72 hours post-transfection. At this time after transfection, EGFP RNA expressed from retrotransposed L1-EGFP will be insignificant compared to L1-EGFP plasmid-driven expression. EGFP levels were normalized to GAPDH or puromycin. L1-EGFP contains a puromycin expression cassette under PGK promoter control. Thus, puromycin expression can be used as normalizer for transfection. IPSCs from two different individuals per species were transfected, and EGFP levels are shown as relative to human iPSCs.

Plasmids

Human APOBEC3B cDNA from WT-33 and ADRC-40 iPSCs was amplified using Phusion high-fiidelity polymerase (New England BioLabs) and primers described in . A3B cDNA fused to HA-tag was then inserted into KpnI/XbaI digested pcDNA3.1+ (pcDNA3-A3B) as previously described [35]. Similarly, PIWIL2 cDNAs were amplified from human and NHP iPSCs as described above and inserted into pEF-BOS-EX using EcoRI/SalI[36]. The plasmid expressing APOBEC3G was previously described [35].

RNA library generation and deep sequencing

PolyA+ RNA was fragmented and prepared into sequencing libraries using the Illumina TruSeq RNA sample preparation kit and analyzed on an Illumina HiSeq 2000 sequencer at the UCSD Biomedical Genomics Laboratory (BIOGEM). cDNA libraries were prepared from four human, two chimpanzee and two bonobo iPSC lines derived from fibroblasts (two clones each, except for human WT9 and WT126), and two hESCs (Hues6 and H1). Libraries were sequenced using paired-end 2× 100 bp reads at a depth of 15-30 million reads per library (250 ± 25 bp fragments) (). Paired end reads from all libraries were mapped to both the human (hg19, GRCh37) and chimpanzee (panTro3, CGSC 2.1.3) genomes using STAR (v2.2.0c)[37]. To compare gene expression between human and NHP iPSCs, we first mapped paired end reads from all libraries to both human and chimpanzee genomes and then calculated gene expression read counts relative to human RefSeq transcripts. Due to the lack of annotation in the chimpanzee genome, human gene models (RefSeq) were used to quantify gene expression. To avoid bias introduced by genome insertions and deletions, only reads mapping to both the human and chimpanzee genomes uniquely were used from each sample when comparing gene expression values (~4% of reads mapped to only one genome per sample). To calculate gene expression, read counts in the exons of RefSeq transcripts where calculated using HOMER[38]. Gene expression clustering was carried out using Gene Cluster 3.0 and visualized with Java Tree View[39,40]. EdgeR was used to identify differentially expressed genes comparing human samples with NHPs, and pairwise between bonobo, chimpanzee, and human[41]. Functional enrichment analysis was restricted to differentially expressed coding genes with false discovery rates less than 5% and a fold change greater than 2-fold. We further restricted genes to contain an average of 10 normalized reads across sample groups to remove genes with very low expression. Gene Ontology functional enrichment for biological processes (level 2) was carried out using DAVID[42], Homo sapiens whole genome was set as background.

Small RNA library generation and deep sequencing

Small RNA (15-40 nt) libraries were prepared using the Illumina TruSeq Small RNA sample preparation kit and analyzed on an Illumina HiSeq 2000 sequencer at the Beijing Genomics Institute. Libraries were sequenced using single-end reads at a depth of 15-25 million reads per library. Adapter sequences were clipped from the 3’ end of each read and then aligned to the human (hg19, GRCh37) genome or to the L1Hs consensus sequence (Repbase[43]) using Bowtie2 (v4.1.2)[44]. Reads aligning to miRBase-defined miRNA transcripts were quantified using HOMER. Matches to previously identified human piRNAs were restricted to small RNAs with lengths between 26 and 33 nt with 5’ ends within 2 nt of previously identified piRNA 5’ ends based on piRNABank (http://pirnabank.ibab.ac.in/)[45].

L1 retrotransposition

Reporter L1 elements are tagged with a reporter gene (EGFP or firefly Luciferase (Luc)) such that only cells that complete a round of retrotransposition will express the reporter gene[46]. Three L1 reporter plasmids were used in this work. L1-EGFP[18,46] was previously described and was a gift from Dr. J.V. Moran. L1 luciferase-tagged plasmids (pYX014 and pYX017)[19] were obtained from W. An. In pYX014, L1 is regulated by its native promoter (5’-UTR) and, in pYX017, by the heterologous promoter CAG. pYX014 and pYX017 plasmids contain a renilla luciferase expression cassette that allows for control of transfection efficiency. L1 assays in 293T cells were carried out as previously described [47]. 293T cells were transfected with L1 reporter plasmid together with control plasmid or plasmids expressing A3B, A3G or PIWIL2 using polyethylenimine (PEI). L1-Luciferase retrotransposition was measured by quantification of luciferase activity using the Dual-Glo luciferase reporter assay (Promega) and normalized to renilla luciferase. L1-Luc inhibition in the presence of A3B or PIWIL2 was independent of the promoter driving L1 expression (data not shown). Inhibition levels of L1 retrotransposition by A3B and PIWIL2 were comparable between the three L1 reporter plasmids used in this study. Plasmid transfections of iPSCs were performed by electroporation of L1-EGFP plasmid following the manufacturer's instructions (Lonza/Amaxa Nucleofactor, Kit V). The cells were then cultured under normal conditions for 10 days and the percentage of retrotransposition was measured by fluorescent activated cell sorting (FACS) of EGFP-positive cells. Electroporation efficiency of the L1-EGFP plasmid in human and NHP iPSC was controlled by transfecting a cassette expressing EGFP and analyzed by FACS after 48h. Human and NHP iPSC lines had similar transfection efficiency rates. To test the effect of A3B and PIWIL2 overexpression on L1 activity in NHP iPSCs, human A3B and human, chimpanzee or bonobo PIWIL2 cDNAs were electroporated. All experiments were performed at least three times independently. L1 mobility assays are shown as relative value compared to mock transfections or human IPSC-1 and represented as mean±SEM of at least three independent experiments.

Identification and cloning of a retrotransposition-competent Chimpanzee L1

To clone an intact L1 and generate a chimpanzee L1-EGFP reporter plasmid, we followed a modification of the strategy previously described[48]. Intact L1Pt elements were identified in the Chimpanzee genome (CSAC 2.1.4/panTro4, UCSC) through Blat and L1Xplorer analyses[49]. Among the identified intact full-length L1Pt elements, we amplified the L1 element located in chromosome 7: 11771100-11777132 of the chimpanzee genome from 0.2 ng of genomic DNA extracted from chimpanzee iPSC1. Primers were designed to match unique sequences flanking 5’ and 3’ of the L1Pt and PCR reactions were performed using Phusion High-Fidelity polymerase (NEB). PCR product was sequenced to confirm intactness (). A second PCR was performed using the first PCR product as template to introduce a NotI site upstream the 5’end of L1Pt. The second PCR product was digested with NotI/BstZ17I (New England Biolabs) and inserted into NotI/BstZ17I-digested pL1-EGFP replacing the human-L1 element to generate L1IN71 by using Rapid Ligation kit (Roche). L1IN71 contains a full-length L1Pt element tagged with the EGFP retrotransposition reporter cassette. Primers used for cloning L1IN71 are show in

L1 promoter activity

Human and chimp L1 promoters (L1 5’UTR) were amplified by PCR from L1-EGFP and L1IN71 plasmids, and inserted into XhoI/HindIII digested pGL4.10 (Promega) upstream firefly luciferase cDNA (L15’UTR plasmids). To quantify L1 promoter activity, L15’UTR plasmids were co-transfected into human and NHP iPSC lines with a plasmid expressing Renila luciferase. 72 hours post-transfection luciferase activity was quantified and firefly luciferase signal were normalized to renilla luciferase signal. Results are shown as relative to human L1 5’UTR activity in human iPSCs. Two iPSC lines from different individuals (iPSC1 and iPSC2) per species were transfected. Primers used for cloning L1 5’UTRs are show in

Quantification of reference genome-encoded L1 insertions

Quantification of L1-derived genomic sequences was based on Repbase defined elements annotated by RepeatMasker (http://www.repeatmasker.org). L1 genomic positions for human (hg19, GRCh37) and chimpanzee (panTro3, CGSC 2.1.3) genomes were downloaded from the UCSC Genome Browser annotation database[50]. Due to the large number of unfinished gaps in the chimpanzee genome assembly greater than 2 kb in size, only truncated L1 elements between 100 bp and 1 kb in length were considered in this analysis. Most of these represented the 3’ end of L1 elements. L1 elements were separated based on their annotation as L1HS, L1Pt, L1PA2, L1PA3, or L1PA4 and were plotted as a histogram relative to their divergence values, which indicates the fractions of nucleotides that are mutated relative to the consensus element for each family. To estimate the variability of L1 coverage across the genome, each genome was fragmented into 1 Mb sections and then was randomly sampled in ten separate groups to calculate the standard deviation in number of L1 elements across different regions of the genome. A strong concentration of L1 elements in a few specific regions of the genome would result in a very high variance between groups, whereas uniform insertion across the genome would result in a low variance. This standard deviation between each sampling was reported as a function of divergence for each class of L1 elements.

To identify reference L1 elements that were inserted into the genome after the last common ancestor for human and chimpanzee, L1 elements were mapped between homologous regions of each genome using the UCSC LiftOver tool. If an element failed to map between genomes, the 100 bp regions immediately upstream and downstream of the L1 element were also mapped between genomes using the LiftOver tool. If the upstream and downstream regions both mapped to the other genome, then the L1 element was mostly likely a result of a recent retroviral insertion. If only one or neither of the upstream and downstream regions mapped between genomes, the region was more likely the result of a genomic duplication or deletion and was discarded from the analysis. Error bars (standard deviation) represent the differences in L1 density based on the sampling of different genomic regions and represents the variability of L1 coverage across the genomes.

Immunocytochemistry

Cells were fixed in 4% paraformaldehyde and then permeabilized with 0.5% Triton-X100 in PBS. Cells were then blocked in 5% donkey serum for 1 hour before incubation with primary antibody overnight at 4°C. After three washes with PBS, cells were incubated with secondary antibodies conjugated to fluorophors (Jackson Immuno Research) for 1 hour at room temperature. Fluorescence was detected using a Zeiss inverted microscope.

Immunoblotting

Immunoblotting was performed as previously described (Narvaiza et al. 2009). Cell pellets were lysed in lysis buffer supplemented with Complete protease inhibitor cocktail (Roche) for 30 min on ice[35]. Protein concentrations from whole cell lysates were quantified by BCA assay (Bio-Rad). Proteins were separated in 4–12% Acrylamide Bis-Tris NuPage gels in MOPS buffer (Invitrogen) and transferred onto Hybond nitrocellulose membranes (Amersham Biosciences).

Antibodies

Primary antibodies used in this study were: TRA-1-81 (1:100, Millipore, MAB4381), Nanog (1:500, R&D Systems, AF1997), APOBEC3B (D-15) (1:500, Santa Cruz, sc-86289), PIWIL2 (1:1000, R&D Systems, AF6558), GFP (1:200, Molecular Probes-Invitrogen, A-6455), Flag (1:1000 Sigma, F7425) and HA (1:1000 Sigma, H6908). All secondary antibodies were purchased from Jackson ImmunoResearch.

PIWIL2 RNPs immunoprecipitation and end labeling

Tetracycline-inducible human iPSCs expressing flag-tagged PIWIL2 were generated by transduction with lentiviruses (Lv)[51]. Cells were first transduced with a Lv expressing tetracycline transactivator rtTA (LvXEtO). After 10 days of culture in growth media with Neomycin (Neo), Neo-resistant colonies were then transduced with a Lentivirus expressing Flag-PIWIL2 under the control of a tetracycilne-inducble promoter (LvXTPFlagPIWIL2) and selected for resistance to puromycin. For PIWIL2 RNP immunoprecipitation, ~3×107 human iPSCs were treated with doxycycline for 72 hrs, and pelleted cells were resuspended in 1 ml of Lysis Buffer 1 (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM MgCl2, 0.5% NP40, 1% glycerol, 1 mM DTT, 0.1 U/μl RNAse inhibitor (Ambion), Complete EDTA-free protease inhibitor (Roche)). Cell lysates were cleared by centrifugation at 20,000 g for 20 min at 4°C. Cleared lysates were incubated with EZview Red FLAG M2 Affinity Gel (Sigma) for 3 hrs at 4°C and washed 5 times with wash buffer (Lysis Buffer 1 without glycerol). Co-immunoprecipitated RNAs were extracted with Trizol, followed by precipitation with isopropanol and glycogen (Ambion). Isolated RNA was 5’ labeled with [γ-32P]-ATP using T4 polynucleotide kinase (NEB), resolved on 15% PAGE TBE urea gels along with radiolabeled Decade size marker (Ambion) and visualized in a Typhoon phosphorimager (Amersham Biosciences). Control immunoprecipitations were carried out with lysates from cells without doxycycline induction, from doxycycline-induced GFP expressing hiPS cells or with control antibody (anti-HA).

Data deposition

RNA-Seq and small RNA-seq data have been deposited in the Gene Expression Omnibus under accession number GSE47626. GenBank accession numbers: KF651164 (P. paniscus PIWIL2), KF651165: (H. sapiens PIWIL2), KF651166 (P. troglodytes PIWIL2), KF651167 (H. sapiens APOBEC3B), KF651168 (P. troglodytes APOBEC3B), KF651169 (P. paniscus APOBEC3B) and KF661301 (L1Pt in Chimp-L1 plasmid).

Extended Data Table 1

List of primers used in this study.

Primers	Sequence	Use
Nanog-F	5’- CCTATGCCTGTGATTTGTGG -3’	PCR
Nanog-R	5’- CTGGGACCTTGTCTTCCTTT -3’	PCR
AFP-F	5’- AAAAGCCCACTCCAGCATC -3’	PCR
AFP-R	5’- CAGACAATCCAGCACATCTC -3’	PCR
Musashi-F	5’- AAAGGAGGTGATGTCGCCAA -3’	PCR
Musashi-R	5’- TGGTCCGTAGGCAGTGAGA -3’	PCR
Brachyury-F	5’- GCCCTCTCCCTCCCCTCCACGCACAG -3’	PCR
Brachyury-R	5’- CGGCGCCGTTGCTCACAGACCACAGG -3’	PCR
B-Actin-F	5′- TGTTTTCTGCGCAAGTTAGGTTTT -3′	PCR
B-Actin-R	5′- GCCGACAGGATGCAGAAGGAGAT -3′	PCR
APOBEC3B (20-40)	5′-GCGGGACAGGGACAAGCGTAT-3′	Cloning
APOPEC3B (1250-1228)	5′-CTGCTCAACCCAGGTCTCTGCCT-3′	Cloning
APOBEC3B (19-41)	5′-AGCGGGACAGGGACAAGC GTATC-3′	Cloning
APOBEC3B(1309-1288)	5′-AGCTGGAGATGGTGGTGAACGG-3′	Cloning
L1Pt ch7 11 F	5′-TTGCAGGTACTCTGAGCTTCAC-3′	Cloning
L1Pt ch7 11 R	5’-AAGGAGAAGCACCTGCATGA-3’	Cloning
Not-L1 F	5’-ATAAGAATGCGGCCGCGGGGGAGGAGCCAAGATG-3’	Cloning
XhoINotI L1 5UTR	5’-CCGCTCGAGCGGCCGCGGGGGAGGAG-3’	Cloning
L1 5UTRHindIIIATG	5”-TTTTTAAGCTTCCATCTTTGTGGTTTTATCTAC-3”	Cloning
APOBEC3B-F	5’-CGCCAGACCTACTTGTGCTAT-3’	qPCR
APOBEC3B-R	5’-CATTTGCAGCGCCTCCTTAT-3’	qPCR
GAPDH-F	5’- CATGTTCCAATATGATTCCACC-3’	qPCR
GAPDH-R	5’- CTCCACGACGTACTCAGCG-3’	qPCR
PIWIL2-F	5′- TTGTGGACAGCCTGAAGCTA -3′	qPCR
PIWIL2-R	5′- CCATCAGACACTCCATCACG -3′	qPCR
L1 5’UTR set1-R	5’-AAGATGGCCGAATAGGAACA-3’	qPCR
L1 5’UTR set1-R	5’-GATGAACCCGGTACCTCAGA-3’	qPCR
L1 5’UTR set2-R	5’-GAGATCTGAGAACGGGCAGA-3’	qPCR
L1 5’UTR set2-R	5’-AGCTGCAGGTCTGTTGGAAT-3’	qPCR
L1 ORF1 set1-F	5’-GCTACGGGAGGACATTCAAA-3’	qPCR
L1 ORF1 set1-R	5’-TTCAGCTCCATCAGCTCCTT-3’	qPCR
L1 ORF1 set2-F	5’-ATGAGCAAAGCCTCCAAGAA-3’	qPCR
L1ORF1 set2-R	5’-TTCTCCCCATCACTTTCAGG-3’	qPCR
L1 ORF2 set1-F	5’-TGACAAACCCACAGCCAATA-3’	qPCR
L1 ORF2 set1-R	5’-CCCTGTCTTGTGCCAGTTTT-3’	qPCR
L1 ORF2 set2-F	5′-TGGAGGCATCACACTACCTG-3′	qPCR
L1 ORF2 set2-R	5′-ATGCGGCATTATTTCTGAGG-3′	qPCR
Actin-F	5′- TACAATGAGCTGCGTGTGG-3′	qPCR
Actin-R	5′- TAGCACAGCCTGGATAGCAA-3′	qPCR
GFP F2	5′- GGGTGTTCTGCTGGTAGTGG-3′	qPCR
GFP R2	5′- TATATCATGGCCGACAAGCA-3′	qPCR
PURO F	5′- CTCGACATCGGCAAGGTGTG-3′	qPCR
PURO R	5′- GCCTTCCATCTGTTGCTGCG-3′	qPCR
APOBEC3A	TaqMan Assay (Life Technologies) Hs00377444	qPCR
APOBEC3B	TaqMan Assay (Life Technologies) Hs00358981	qPCR
APOBEC3C	TaqMan Assay (Life Technologies) Hs00828074	qPCR
APOBEC3D	TaqMan Assay (Life Technologies) Hs00537163	qPCR
APOBEC3F	TaqMan Assay (Life Technologies) Hs01665324	qPCR
APOBEC3G	TaqMan Assay (Life Technologies) Hs00222415	qPCR
APOBEC3H	TaqMan Assay (Life Technologies) Hs00962174	qPCR
PIWIL2	TaqMan Assay (Life Technologies) Hs01032720	qPCR
GAPDH	TaqMan Assay (Life Technologies) Hs03929097	qPCR
HPRT	TaqMan Assay (Life Technologies) Hs01003267	qPCR