Dissection of affinity captured LINE-1 macromolecular complexes

  1. Martin S Taylor
  2. Ilya Altukhov
  3. Kelly R Molloy
  4. Paolo Mita
  5. Hua Jiang
  6. Emily M Adney
  7. Aleksandra Wudzinska
  8. Sana Badri
  9. Dmitry Ischenko
  10. George Eng
  11. Kathleen H Burns
  12. David Fenyö
  13. Brian T Chait
  14. Dmitry Alexeev
  15. Michael P Rout
  16. Jef D Boeke
  17. John LaCava  Is a corresponding author
  1. Massachusetts General Hospital, United States
  2. Moscow Institute of Physics and Technology, Russia
  3. The Rockefeller University, United States
  4. NYU Langone Health, United States
  5. Johns Hopkins University School of Medicine, United States
  6. Novosibirsk State University, Russia
16 figures, 2 tables and 8 additional files


RNase sensitivity and split-tandem affinity capture of L1 ORF2p RNPs.

(A) On-bead RNase-sensitivity assay: L1 complexes were affinity captured by ORF2p-3xFLAG. The magnetic media were then treated with a solution containing either a mixture of RNases A and T1 or BSA. After treatment, the supernatants were removed and the remaining bound material was released with LDS. Proteins requiring intact RNA to maintain stable interactions with immobilized ORF2p were released from the RNase-treated medium, while the BSA-treated sample controlled for the spontaneous release of proteins from the medium. Representative SDS-PAGE/Coomassie blue stained gel lanes are shown for each fraction. (B) The experiments described above were carried out in duplicate, once with light isotopically labeled cells (L) and once with heavy isotopically labeled cells (H), resulting in four label-swapped, SILAC duplicates (one light set and one heavy set). The four fractions were cross-mixed and the differential protein retention upon the affinity medium during the treatments (BSA vs. RNase) was assessed by quantitative MS. (C) Results from the RNase-sensitivity assay graphed as the fraction of each detected protein present in the BSA-treated sample (RNase-sensitive proteins are more present in the BSA treated sample), normalized such that proteins that did not change upon treatment with RNases are centered at the origin. A cut-off of p=10−3 for RNase-sensitivity is indicated by a light gray circle; proteins that are RNase-sensitive with a statistical significance of p<10−3 are outside the circle. Proteins previously ranked significant by I-DIRT analysis (Table 1) are labeled and displayed in blue or magenta (as indicated); black nodes were RNase-sensitive but not significant by I-DIRT; gray, unlabeled nodes were neither RNase-sensitive nor significant by I-DIRT. (D) Split-tandem affinity capture: L1 complexes were affinity captured by ORF2p-3xFLAG. After native elution with 3xFLAG peptide, this fraction was depleted of ORF1p-containing complexes using an α-ORF1 conjugated magnetic medium, resulting in a supernatant fraction depleted of ORF1p-containing complexes. The α-ORF1 bound material was then released with LDS, yielding an elution fraction enriched for ORF1p-containing complexes. Representative SDS-PAGE/Coomassie blue stained results for each fraction are shown. (E) SILAC duplicates, two supernatants and two elutions, were cross-mixed to enable an assessment of the relative protein content of each fraction by quantitative MS. (F) The results from split-tandem affinity capture graphed as the fraction of each protein observed in the elution sample. In order to easily visualize the relative degree of co-partitioning of constituent proteins with ORF1p, these data were normalized, setting the fraction of ORF1p in the elution to 1. Proteins which were previously ranked significant by I-DIRT analysis are labeled and displayed in blue or magenta (as indicated); gray, unlabeled nodes were not found to be significant by I-DIRT. MOV10 is marked with a dagger because in one replicate of this experiment it was detected by a single unique peptide, whereas we have enforced a minimum of two peptides (see Materials and methods) for all other proteins, throughout all other proteomic analyses presented here.

Transcriptomic and enzymatic analysis of split-tandem RNP fractions.

(A) RNA sequencing affinity captured L1s: L1 complexes were obtained by split-tandem affinity capture, as in Figure 1D (simplified schematic shown); RNA extracted from these three fractions was subjected to next-generation sequencing. The results are summarized with respect to coverage of the synthetic L1 sequence (see schematic with nucleotide coordinates) as well as the relative quantities of mapped, annotated reads (pie charts; the mean of duplicate experiments is displayed). (B) Summary of sequencing reads: displays the total number of sequencing reads that mapped to our reference library, the subset of mapped reads carrying a genome annotation, and the number of reads that corresponding to L1, both raw and normalized (see Materials and methods and Supplementary file 4). The mean of duplicate experiments is displayed; ±indicates the data range. (C) LINE-1 element amplification protocol (LEAP) of affinity captured L1s: L1 complexes were obtained from full length synthetic L1 (pLD401) and an otherwise identical ΔORF1 construct (pLD561) following the same experimental design as in (A), except that elution from α-ORF1p affinity medium was done natively, by competitive elution. In this assay, L1 cDNAs are produced, in cis, by ORF2p catalyzed reverse transcription of L1 RNAs; the resulting cDNAs by were measured by quantitative PCR and presented as relative quantities normalized to pLD401 input (Supplementary file 4). The mean of duplicate experiments is displayed; error bars indicate the data range.

Immunofluorescent imaging reveals ORF1p expression is required for nuclear ORF2p staining.

(A) Puromycin-selected HeLa-M2 cells containing pLD401 (Tet promoter, [ORFeus-Hs] full L1 coding sequence, ORF2p-3xFLAG, top two rows) or pLD561 (Tet promoter, ΔORF1, ORF2p-3xFLAG, bottom row) were plated on fibronectin-coated coverslips and induced for 24 hr with doxycycline prior to fixation and staining. With pLD401, the previously-observed pattern of cytoplasmic-only ORFs (top row) and a new pattern of pairs of cells displaying ORF2p in the nucleus (middle row) were apparent. When ORF1p was omitted from the construct (pLD561, bottom row), nuclear ORF2p was not apparent. Scale bars: 10 μm. (B) Statistical analysis of the distances between pairs of ORF2p + nuclei as compared to random: Violin plots of the distributions of shortest distances between 1000 pairs of randomly selected nuclei (‘no’) and the observed pairs of ORF2p + nuclei (‘yes’) in cells transfected with pLD401; n = 262 cells, 47 nuclear ORF2+. ***p=3.955 × 10−11 (Welch's t-test).

Figure 3—source data 1

Source data used in the analysis of ORF2p+ inter nuclear distance analysis: Figure 3 and Supplementary file 5.

Figure 4 with 1 supplement
Catalytic inactivation of ORF2p alters the L1 interactome: L1s were affinity captured from cells expressing enzymatically active ORF2p-3xFLAG sequences (pLD401, WT), a catalytically inactivated endonuclease point mutant (pLD567; H230A, EN-), and a catalytically inactivated reverse transcriptase point mutant (pLD624; D702Y, RT-). These were analyzed by next-generation RNA sequencing and quantitative MS.

(A) Proteomic workflow: WT L1s were captured from heavy-labeled cells, EN- and RT- L1s were captured from light-labeled cells. WT and either EN- or RT- fractions were mixed after affinity capture, in triplicate, and the relative abundance of each co-captured protein in the mixture was determined by quantitative MS. (B) L1 RNA yield and coverage between different preparations: As in Figure 2A, RNA extracted from 3xFLAG eluates originating from pLD401, pLD567, and pLD624 were subjected to next-generation sequencing. The results are summarized with respect to coverage of the synthetic L1 sequence (see schematic with nucleotide coordinates) as well as the relative quantities of mapped, annotated reads. The mean of duplicate experiments is displayed. (C) I-DIRT significant proteins displayed were detected in at least two replicates. All values were normalized to ORF2p. Data are represented as mean ±SD. Triangles (▵) mark proteins whose levels of co-capture did not exhibit statistically significant differences in the mutant compared to the WT. A single or double asterisk denotes a statistically significant difference between the relative abundances of the indicated protein in EN- and RT- mutants: p-values of between 0.05–0.01 (*) and below 0.01 (**), respectively. Gray horizontal bars on the plot mark the 2x (upper) and 0.5x (lower) effect levels. (D) The double histogram plot displays the distributions of all proteins identified in at least two replicates, in common between both EN/WT (TOP) and RT/WT (LOWER) affinity capture experiments. The x-axis indicates the relative recovery of each copurifying protein and the y-axis indicates the number of proteins at that value (binned in two unit increments). The data are normalized to ORF2p. The relative positions of ORF2p and ORF1p are marked by colored bars. Differently colored lines illustrate the relative change in positions of the proteins within the two distributions (as indicated). Colored lines denote I-DIRT significance, with magenta lines indicating a statistically significant shift in position (p≤0.05) within the two distributions and green lines indicating that statistical significance was not reached (entities labeled in Figure 4—figure supplement 1). A cluster of magenta lines can be seen to track with ORF1p (red line, upper and lower histogram), and another cluster can be seen to behave oppositely, creating a crisscross pattern in the center of the diagram. A similar crisscross pattern is exhibited by many gray lines.

Figure 4—figure supplement 1
Double histogram plot with entities labeled.

In the top plot (A), entities crossing from left to right, increasing between EN- and RT- mutants (and TOP1, in green), are labeled. In the bottom plot (B), entities crossing from right to left, decreasing between EN- and RT- mutants (and TROVE2 and YMEL1, in green).

Monitoring coordinated dissociation and exchange exhibited by L1 interactors in vitro: L1s were affinity captured from heavy-labeled cells expressing ORF2p-3xFLAG in the context of the naturally occurring L1RP sequence (pMT302); the stabilities of the protein constituents of the captured heavy-labeled L1 population were monitored in vitro by competitive exchange with light-labeled cell extracts containing untagged L1s (pMT298) (Taylor et al., 2013).

(A) 3xFLAG-tagged L1s were captured from heavy-labeled cells and then, while immobilized on the affinity medium, were treated with an otherwise identically prepared, light-labeled, untagged-L1-expressing cell extract. Untreated complexes were compared to independently prepared complexes incubated for 30 s, 5 min, and 30 min, (respectively) to determine the relative levels of in vivo assembled heavy-labeled interactors and in vitro exchanged light-labeled interactors, using quantitative MS. (B) The results were plotted to compare the percentage of heavy-labeled protein versus time. I-DIRT significant proteins from Table 1 are highlighted if present. Three clusters were observed (as indicated). (C) The cosine distance between the observed I-DIRT significant proteins was plotted along with time.

Interactomic data integration

(A) All MS-based affinity proteomic experiments presented were combined and analyzed for similarities across all I-DIRT significant proteins, producing five groupings. Distance are presented on a one-unit arbitrary scale (see Materials and methods: Mass Spectrometry Data Analysis). (B) The traces of each protein in each cluster, across all experiments, are displayed. The y-axis indicates the raw relative-enrichment value and the x-axis indicates the categories of each experiment-type. Each category is as wide as the number of replicates or time-point samples collected.

Refined interatomic model: Our results support the existence of distinct cytoplasmic and nuclear L1 interactomes.

Affinity capture of L1 via 3xFLAG-tagged ORF2p from cell extracts results in a composite purification consisting of several macromolecular (sub)complexes. Among these, we propose a canonical cytoplasmic L1 RNP (depicted) and one or more nuclear macromolecules. UPF1 exhibited equivocal behavior within our fractionations and was also co-captured with chromatin associated ORF2p, suggesting it participates in both cytoplasmic and nuclear L1 interactomes. Within the nuclear L1 interactome, our data support the existence of a physically linked entity consisting of (at least) PCNA, PURA/B, TOP1, and PARP1 (depicted).

Appendix 1—figure 1
SILAC suspension expression of L1 constructs: western blotting.
Appendix 1—figure 2
RNase sensitivity affinity capture: data normalization.
Appendix 1—figure 3
RNase sensitivity affinity capture: normality test.
Appendix 1—figure 4
Split-tandem affinity capture: data normalization.
Appendix 1—figure 5
Distances between two-node groups.
Appendix 1—figure 6
Distances between three-node groups.
Appendix 1—figure 7
Efficacy of elution using ORF1p peptides: Coomassie blue stained gel .
Appendix 1—figure 8
Retrotransposition mutants affinity capture: distributions of normalized affinities.
Appendix 1—figure 9
Protein in vitro exchange: the distributions of H/(H+L) values.


Table 1
Putative L1 interactors: Through a series of affinity capture experiments (co-IP) using I-DIRT, we characterized a set of putative host-encoded L1 interactors (Taylor et al., 2013). 

The proteins observed were associated with both ORF1p and ORF2p (highlighted in blue), only in association with ORF2p (highlighted in magenta), or only in association with ORF1p (no highlight). The two highlighted populations are the central focus of this study.

Gene symbolUniprot symbolProteinco-IP with
MOV10Q9HCE1Putative helicase MOV-10ORF1/2
PABPC1P11940Polyadenylate-binding protein 1ORF1/2
PABPC4Q13310Polyadenylate-binding protein 4ORF1/2
UPF1Q92900Regulator of nonsense transcripts 1ORF1/2
ZCCHC3Q9NUD5Zinc finger CCHC domain-containing protein 3ORF1/2
FKBP4Q02790Peptidyl-prolyl cis-trans isomerase FKBP4ORF2
HAX1O00165HCLS1-associated protein X-1ORF2
HMCESQ96FZ2Embryonic stem cell-specific 5-hydroxymethylcytosine-binding proteinORF2
HSP90AA1P07900Heat shock protein HSP 90-alphaORF2
HSP90AB1P08238Heat shock protein HSP 90-betaORF2
HSPA1AP0DMV8Heat shock 70 kDa protein 1AORF2
HSPA8P11142Heat shock cognate 71 kDa proteinORF2
NAP1L1P55209Nucleosome assembly protein 1-like 1ORF2
NAP1L4Q99733Nucleosome assembly protein 1-like 4ORF2
PARP1P09874Poly [ADP-ribose] polymerase 1ORF2
PCNAP12004Proliferating cell nuclear antigenORF2
PURAQ00577Transcriptional activator protein Pur-alphaORF2
PURBQ96QR8Transcriptional activator protein Pur-betaORF2
RPS27AP62979Ubiquitin-40S ribosomal protein S27aORF2
TIMM13Q9Y5L4Mitochondrial import inner membrane translocase subunit Tim13ORF2
TOP1P11387DNA topoisomerase 1ORF2
TOMM40O96008Mitochondrial import receptor subunit TOM40 homologORF2
TUBBP07437Tubulin beta chainORF2
TUBB4BP68371Tubulin beta-4B chainORF2
YME1LQ96TA2ATP-dependent zinc metalloprotease YME1L1ORF2
DDX6P26196Probable ATP-dependent RNA helicase DDX6ORF1
ERAL1O75616GTPase Era, mitochondrialORF1
HIST1H2BOP23527Histone H2B type 1-OORF1
LARP7Q4G0J3La-related protein 7ORF1
MEPCEQ7L2J07SK snRNA methylphosphate capping enzymeORF1
PABPC4LP0CB38Polyadenylate-binding protein 4-likeORF1
TROVE2P1015560 kDa SS-A/Ro ribonucleoproteinORF1
YARS2Q9Y2Z4Tyrosine--tRNA ligase, mitochondrialORF1
Key resources table
Reagent type (species)
or resource
DesignationSource or referenceIdentifiersAdditional information
gene (human) LINE-1ORFeus-Hs; L1RP10.1016/j.cell.2013.10.021;
cell line (human)HEK-293T_LD10.1016/j.cell.2013.10.021;
Mycoplasma testing was done
regularly and was negative. We received
an authenticated cell line from the ATCC
and subsequently made them blastomycin
resistant so we validated cells by
blastomycin resistance.
transfected construct
pLD401; pLD561; pLD567;
pLD624; pMT302; pMT289
antibodyanti-FLAG; anti-ORF1p10.1016/j.cell.2013.10.021Sigma-Aldrich
Cat# F1804,
custom made,
Abmart: 4H1
peptide, recombinant
ORF1p N-terminal
this paper
software, algorithmScripts for IF (Figure 3);
formal analysis used custom
R code throughout
this paperScripts are in Supplementary file 5;
R code in – https://bitbucket.org/altukhov/line-1/

Additional files

Supplementary file 1

Supporting and supplemental data for the figures and experiments: RNase – Figure 1C; Tandem – Figure 1F; Mutants – Figure 4; Exchange – Figure 5; Exchange distance matrix – Figure 5C; Integration – Figure 6; Integration distance matrix – Figure 6B; Raw – unnormalized values extracted from the MaxQuant proteinGroups.txt file for the experiments presented in this study.

Supplementary file 2

Supporting and supplemental data for overexpression and siRNA knockdown experiments.

Supplementary file 3

Supporting and supplemental data for affinity capture of ORF2p-3xFLAG from fractionated chromatin.

Supplementary file 4

Supporting and supplemental data for figures and experiments: RNA sequencing and LEAP assays – Figure 2.

Supplementary file 5

Supporting and supplemental data for Nuclear Location and ORF2p status: Figure 3.

Supplementary file 6

Supporting and supplemental data for GO analysis.

Supplementary file 7

ORFeus-Hs sequence from pLD401 included in our reference FASTA file: Figure 2 and Supplementary file 4.

Transparent reporting form

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Martin S Taylor
  2. Ilya Altukhov
  3. Kelly R Molloy
  4. Paolo Mita
  5. Hua Jiang
  6. Emily M Adney
  7. Aleksandra Wudzinska
  8. Sana Badri
  9. Dmitry Ischenko
  10. George Eng
  11. Kathleen H Burns
  12. David Fenyö
  13. Brian T Chait
  14. Dmitry Alexeev
  15. Michael P Rout
  16. Jef D Boeke
  17. John LaCava
Dissection of affinity captured LINE-1 macromolecular complexes
eLife 7:e30094.