BioID reveals novel and diverse associations with ATRX

We theorized that a proximity-based protein labeling methodology in live cells would capture ATRX protein-protein associations that are not represented when using biochemical means. We therefore made use of proximity-dependent biotinylation (BioID) [42, 43] to identify proteins that directly or indirectly associate with the ATRX protein (Fig 1A). ATRX was fused to a FLAG-tagged BirA* biotin ligase at either the N- or C-terminus, and expressed from isogenic, tetracycline-inducible HEK293 Flp-In T-REx cells (herein denoted as HEK293 Flp-In). This produced nuclear bait proteins capable of biotinylating associating proteins in living cells, regardless of whether they are direct or indirect, or static or dynamic. Cells were lysed under stringent conditions, and the proteins purified using streptavidin beads (S1(A) –S1(C) Fig). To identify ATRX-associating proteins, we rigorously filtered the ATRX BioID against a total of 20 negative controls consisting of untransfected cells and isogenic HEK293 Flp-In cells expressing BirA* alone. This identified 36 ATRX proximal associations with a Bayesian false discovery rate (BFDR) ≤ 5% and at least 5 peptides detected across biological duplicates (Fig 1B). BirA* fused to a nuclear localization sequence (NLS) was included for comparison. The DAXX histone chaperone, NBS1 component of the MRN complex, PML protein, BLM helicase, IMP3 U3 snoRNP protein, and the SUPT16H and SSRP1 subunits of the FACT histone chaperone were common to our BioID and published AP-MS experiments in HeLa and HEK293T cells [30, 31, 41, 51], or the curated BioGRID database [52]. Affinity-purified ATRX predominantly captured DAXX, the MRN complex, and the FACT histone chaperone by mass spectrometry [30, 31, 41]. BioID expands this repertoire by considering proximal associations and biochemically labile interactions. Gene ontology analysis of the ATRX BioID showed an enrichment of proteins relating to chromosome organization, chromatin remodelling, and transcriptional regulation (S1(D) Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Proximity-dependent biotin identification (BioID) of ATRX-associating proteins.

A: Experimental pipeline. The BirA* biotin ligase was fused to ATRX to biotinylate proximal proteins. Labeled proteins were captured on streptavidin beads and analyzed by mass spectrometry. B: Dot plot showing prey proteins identified with ATRX-BirA* that were enriched over endogenous biotinylation (untransfected) and unspecific pan-cellular biotinylation (BirA* alone—BFDR ≤ 5%, SAINT [53]). Data represent two biological replicates. Proteins in boldface remained statistically enriched when the nuclear localization signal (NLS)-BirA* control was used to further filter the ATRX BioID data. C-D: Proximity-ligation assay (PLA) showing proximal associations between endogenous proteins in the MO3.13 human glial cell model. PLA data plots account for ∼100 nuclei in three independent experiments, with the exact number of nuclei assessed indicated in brackets. The p-values were obtained using a 2-sided Student’s t-test with unequal variance. Scale bars = 4 μm.

https://doi.org/10.1371/journal.pgen.1009909.g001

To identify the most robust associations, we further filtered our ATRX-associating proteins against proteins identified with the NLS-BirA* dataset (Fig 1B and S1 Data). CCDC71, DAXX, FAM207A, IMP3, NBS1, SLF2, ZBTB4, and ZFHX4 remained enriched over all three negative controls (untransfected, BirA* alone, NLS-BirA*). Subcellular fractionation experiments were performed, and while some tubulin was detected in the nuclear fraction, the ATRX-associating proteins that we probed for were all present in the nucleus (S1(E) Fig). To visualize the endogenous protein associations, a proximity-ligation assay (PLA [53]) was performed. PLA generates a fluorescent signal when antibodies that recognize associating proteins are in close proximity to one another (Fig 1C and 1D and S1(F)–S1(I) Fig). Endogenous proteins were followed in the human oligodendrocytic MO3.13 cell model [54] because ATRX is highly expressed in brain tissue and is often deregulated in brain tumours. PLA experiments probing for any one protein alone served as a control. PLA signals were obtained for all but one of the eight proteins listed above. Endogenous FAM207A was excluded from our endogenous PLA analyses because commercial antibodies did not recognize the protein by western blotting.

The endogenous PLA signals for CCDC71 and SLF2 in MO3.13 cells were done using the same antibodies previously used by the Human Protein Atlas [55], but we found these two antibodies also recognized additional proteins by western blotting. The endogenous PLA signals likely reflect true associations, because proteins that cross-react with the antibodies would also need to be in close proximity to generate a signal. Nevertheless, to further ascertain these associations, PLA signals were further obtained between endogenous ATRX and tagged, exogenous CCDC71, FAM207A, and SLF2 (S1(G)–S1(I) Fig). The FLAG tag increased the abundance of PLA background signals, but stark differences in signal intensity were observed when protein associations occurred. Because we lacked PLA data for the endogenous FAM207A protein, we further validated this interaction in vitro. Recombinant FAM207A was expressed using a coupled in vitro transcription/translation reaction in the presence of biotinylated lysine. FAM207A was immobilized on streptavidin beads, washed, and incubated with recombinant ATRX expressed in SF9 insect cells. Further washing of the beads showed that the two proteins bind one another (S1(J) Fig).

Altogether, the data show that ATRX forms several uncharacterized associations with a diverse group of proteins. To gauge the breadth of functions imparted by ATRX proximal associations, we overviewed three novel ATRX-associating partners for whom there is limited information namely, FAM207A, CCDC71, and SLF2.

FAM207A is associated with ribosome biogenesis

An estimated 273 genes orthologous to FAM207A (SLX9) exist across 256 metazoan species [56]. Subcellular fractionation of MO3.13 and HEK293 Flp-In cells expressing the exogenous protein shows varying levels of FAM207A across the fractions of the two lineages (Fig 2A and S2(A) Fig). This may be due to the exogenous protein expression, but possibly also reflects tissue-specific differences. Immunolabeling of detergent-extracted cells showed that FAM207A strongly localized to the cell nucleoli (Fig 2B). To further ascertain this, cells were then co-labeled for FLAG-FAM207A and IMP3, which is not only an ATRX-associating protein (Fig 1B), but also an established nucleolar protein [57]. Both proteins perfectly co-localized with one another (S2(C) Fig). Curiously, the ATRX-FAM207A PLA signals were not confined to the nucleolus (S1(H) Fig). Because PLA signals form when there is a protein-protein interaction, the signals outside the nucleolus could suggest role(s) for ATRX-FAM207A distinct from those in ribosome biogenesis. It is also worth noting that the Human Protein Atlas database suggests that the protein resides in the Golgi apparatus [58]. However, despite high spectral counts for the protein in our BioID, we were not able to detect the endogenous protein by western blotting when we used the same antibody, even under low stringency conditions (e.g., no tween, high antibody concentration).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. FAM207A is a nucleolar protein involved in ribosome biogenesis.

A: Subcellular fractionation of MO3.13 cells expressing FLAG-FAM207A. B: Immunolabeling of exogenous FAM207A in HEK293 Flp-In and MO3.13 cells. C: Fold enrichment of FAM207A BioID prey (BFDR ≤ 5%, SAINT [103]) over negative controls. Data were obtained from biological duplicates. D: Gene ontology terms represented in the FAM207A BioID. Cy—cytoplasm; Nu—nucleus; Ch—chromatin. Scale bars = 4 μm.

https://doi.org/10.1371/journal.pgen.1009909.g002

A prior affinity purification of the nucleolar NOC4 protein identified FAM207A as a likely component of pre-ribosomal particles [59] and further work suggests a role in the export of 40S pre-ribosomes [60]. Our FAM207A BioID results confirm an association with proteins involved in ribosome maturation (Fig 2C and 2D and S2(D) Fig). Twenty-nine proteins were enriched in the FAM207A BioID (BFDR ≤ 1%) in comparison to the BirA* and NLS-BirA* controls (Fig 2C). A much larger number of proteins were retained when NLS-BirA* was used as a comparative bait instead of a control (S2(D) Fig and S1 Data). We believe that the large number of associated proteins reflects the protein’s involvement in macromolecular assemblies. While ATRX was detected in the FAM207A BioID with a relatively high average spectral count of 27.5, it was filtered out by the negative controls in the SAINT analysis. This perhaps indicates that ATRX only constitutes a minor proportion of the protein associations made by FAM207A, since we otherwise detected the ATRX-FAM207A association through the ATRX BioID and PLAs. Gene ontology groups relating to ribosome biogenesis and rRNA processing predominated (Fig 2D), as previously suggested [59]. FAM207A, however, also associated with proteins involved in chromosome and transcriptional regulation, again suggesting more than one biological function.

CCDC71 is a chromatin-bound protein that associates with HP1

Current databases and our own experimental data implicate CCDC71 in chromatin organization. CCDC71 belongs to the coiled-coil domain 71/71L family found within vertebrates (Fig 3A). In addition to a coiled-coil region, the protein harbours an NLS and several regions that are predicted to be disordered (Fig 3B). The mouse protein contains two canonical PxVxL sequences, a motif found in proteins that bind HP1 [35]. Variation within the sequence is tolerated to various degrees [35, 61], and both the human and the consensus sequence from 254 putative orthologs contain PxVxL-like sequences at these two locations (Fig 3C). A canonical PxVxL sequence is also found in the human CCDC71L protein, a shorter family member that shares a high degree of identity within the N-terminal half of CCDC71 and its C-terminal extremity (S3(A) Fig). It was, therefore, not surprising to see that the exogenous protein was almost exclusively found in the chromatin fraction of MO3.13 and HEK293 Flp-In cells (Fig 3D and S3(B) Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. CCDC71 is a chromatin-bound protein that associates with HP1.

A: Hierarchical catalogue of orthologs (OrthoDB v10.1) [56] diagram showing 262 putative genes within the CCDC71/71L family. B: Graphical depiction of the CCDC71 primary structure and features identified using Pfam [110] and Prosite [111]. Putative PxVxL-like motifs are also shown. The CCDC71/71L domain is shown in light grey and putative disordered regions in dark red. C: Sequence analysis of 254 CCDC71 orthologs aligned around the two PxVxL-like motifs. The sequence alignment was generated using COBALT [112] and visualized using WebLogo [113]. D: Subcellular fractionation of MO3.13 cells expressing FLAG-CCDC71. E: Immunolabeling of exogenous CCDC71 in HEK293 Flp-In and MO3.13 cells. F: Fold enrichment of CCDC71 BioID prey (BFDR ≤ 5%, SAINT [103]) over negative controls. Data were obtained from biological duplicates. G: Gene ontology terms represented in the CCDC71 BioID. Scale bars = 4 μm.

https://doi.org/10.1371/journal.pgen.1009909.g003

To our knowledge, there are no reports on CCDC71, but data gathered by the Human Protein Atlas [55] suggest that the protein locates to the nuclear periphery in HeLa cells, with some cell types showing a more nuclear signal. Immunolabeling in HEK293 Flp-In cells showed pan-nuclear granular signals (Fig 3E), a pattern also seen in the ATRX-CCDC71 PLA experiments in HEK293 Flp-In cells (S1(G) Fig). Putative CCDC71 protein-protein interactions reported in BioGrid [52] suggest a role in chromatin compaction. They include histone H3.3, the EED subunit of the polycomb repressive complex 2 (PRC2), the RbAp46 histone chaperone, and the HDAC1 histone deacetylase. We could indeed co-immunoprecipitate H3.3, EZH2 (the catalytic subunit of PRC2), and HDAC1 with CCDC71 (S3(C) Fig). To obtain a clearer portrait of CCDC71 functions and its proximal protein associations, a BioID analysis was also performed for this protein. While HDAC1 was represented, the most prevalent associations included ATRX, all three HP1 isoforms, the NAP1 histone chaperone, and components of the NuRD chromatin remodelling protein complex (also represented in the ATRX BioID; Fig 3F, S3(E) Fig and S1 Data). As expected, a gene ontology analysis of the CCDC71 BioID results included terms related to chromosome and chromatin organization (Fig 3G). This shows that ATRX and CCDC71 share important functions in chromatin regulation.

SLF2 prevents spurious telomere recombination

The SLF2 proteome was previously described [44], but the association with ATRX is novel. It was found that SLF1/2 loads the cohesin-like SMC5/6 proteins at sites of DNA damage [44]. A genetic interaction between ATRX and SMC5 and SMC6 was recently reported [62] and deregulation of SMC5/6 in ALT-positive cells promotes APB formation and telomere exchanges [49]. To determine if ATRX influences SLF1/2 and SMC5/6 recruitment to telomeres, we disrupted ATRX expression in HEK293 Flp-In cells using CRISPR-Cas9 (Fig 4A), and quantified SLF2 occupancy at telomeres. Exogenous SLF2 and endogenous SMC5 were imaged by immunofluorescence and telomeric regions identified using fluorescence in situ hybridization (IF-FISH). SLF2 and SMC5 formed foci whose signals strongly overlapped with one another (S4(A) and S4(B) Fig). Most SLF2 and SMC5 foci formed outside telomeres, but approximately 20% of the signals coincided with those from the telomeric FISH probe (Fig 4B). The proportion of SLF2/SMC5 overlap with telomeres decreased in the ATRX-null cells. G-quadruplex (G4) structures can form at telomeres, where they cause DNA damage if not properly resolved. The analysis was therefore repeated in cells exposed to the G4-stabilizing small molecule, pyridostatin [63], but the treatment had no impact on the SLF2/SMC5—telomere signal overlap (Fig 4B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Loss of ATRX and SLF2 enables telomere exchanges.

A: DNA sequencing and western blotting confirming the disruption of ATRX expression in CRISPR-Cas9-edited HEK293 Flp-In T-REx cells inducibly expressing RAP1-BirA*. B: Relative signal overlap between immunolabeled SLF2 and telomere fluorescence in situ hybridization (IF-FISH), in ATRX-expressing or -null HEK293 Flp-In cells. Cells were exposed to 10 μM pyridostatin (PDS, G4 stabilizer) for 60 min. At least 40 cells per condition were analyzed using a CellProfiler colocalization pipeline [106]. ATRX loss decreased SLF2 recruitment to telomeres in the absence of pyridostatin. C: Schematic of dual-colour chromosome-orientation fluorescence in situ hybridization (CO-FISH) [64] using PNA probes to label C- and G-rich telomere strands (left). Examples of normal CO-FISH signals and of a chromosome end with a telomere exchange are shown (right). D: Relative amount of telomere exchanges in ATRX, SLF2, and ATRX/SLF2 KO HEK293 Flp-In cells. At least 30 mitotic spreads per condition were counted and the percent of telomeric exchanges plotted. p-values were obtained using a 2-sided Student’s t-test with unequal variance.

https://doi.org/10.1371/journal.pgen.1009909.g004

ATRX expression restrains telomere exchanges [9, 15, 40]. We therefore wanted to determine if ATRX and SLF2 together repress t-SCEs. To investigate this, we disrupted SLF2 in ATRX-expressing and -null HEK293 Flp-In cells (S4(C) and S4(D) Fig), and performed chromosome-orientation fluorescent in situ hybridization (CO-FISH) [64] to visualize and quantify telomere exchanges (Fig 4C and 4D). As recently reported [15], we observed increased levels of t-SCEs (dual signal on matching sister chromatid ends), but also a high proportion of exchanges on telomere ends of single chromatids in our ATRX-null cells (Fig 4D). This can indicate non-allelic exchanges and the events were therefore grouped as ‘telomere exchanges,’ as previously done [15]. Low p-values were obtained when comparing telomere exchange rates between the single null cells and the parental HEK293 Flp-In (e.g., 0.059 and 0.15 for ATRX- and SLF2-null cells, respectively), but significance was reached in double ATRX/SLF2-null cells (Fig 4D). This suggests that both proteins inhibit spurious exchanges. However, the lack of a clear synergistic effect may signify that the proteins do so independently of one another. In the ALT-positive (ATRX-null) U2OS cells, an SLF2 gene disruption again caused a significant increase in telomere exchanges (S4(E) Fig). Altogether, this shows that ATRX has some influence on SLF2 recruitment, and a loss of both proteins permits high levels of telomere exchanges.

ATRX and SLF2 depletion alters protein enrichment at telomeres

A loss of ATRX causes gradual changes on telomeres [20], so we wanted to determine if its loss, or that of SLF2, caused immediate observable changes in protein enrichment at telomeres. To examine this, we performed BioID on the RAP1 (TERF2IP) subunit of the shelterin complex, in parental HEK293 Flp-In cells and compared the results to that of the same BioID done in isogenic SLF2-null, ATRX-null, or cells stably expressing ATRX shRNA (Fig 5 and S5 Fig). The same BioID strategy, but using TERF1 as a bait, previously identified changes in the telomeric proteome of ALT-negative and ALT-positive cells [65]. While we did not enrich for all the shelterin components, TERF2 was one of the most abundant prey proteins identified in the RAP1 BioID (S1 Data). Importantly, 38% of the 343 proteins identified by our RAP1 BioID in HEK293 Flp-In cells were also identified by the previously reported TERF1 BioID in HeLa cells [65], and 55% our of preys were also identified by other proteomic techniques used to identify telomeric proteins (namely, PICh [66], QTIP [67], C-BERST [68], and TERF1 BioID [65]; S5(C) Fig). A similar overlap was seen when using NLS-BirA* as an additional control in our BioID, but we first omitted the control from this analysis to identify as many changes as possible (data were still filtered against untransfected cells and BirA*). As expected, RAP1 BioID experiments performed in parental and ATRX- or SLF2-depleted HEK293 Flp-In cells were far more similar to one another than our other BioID experiments with other baits (Fig 5A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Loss of ATRX changes the enrichment of a subset of proteins at telomeres.

A: Correlation between BioID bait proteins. RAP1 BioID experiments strongly correlated with one another independently of ATRX expression. B: Relative ATRX detection by RAP1 BioID in ATRX-expressing and isogenic cells where ATRX expression was knocked-down (KD) using shRNA or knocked-out (KO) using CRISPR-Cas9. C: Fold change of prey proteins identified by RAP1 BioID (BFDR ≤ 5%, SAINT [103]) in ATRX-expressing vs. ATRX-null HEK293 Flp-In cells. Labeled proteins had ≥2.5 average spectra and ≥1.5-fold change between conditions. Proteins labeled in the scatter plot remained significant when NLS-BirA* was used as a control. Inset box represents additional proteins identified with NLS-BirA* as a bait (less stringent).

https://doi.org/10.1371/journal.pgen.1009909.g005

The ATRX knockout (Fig 5C) and knockdown (S5(D) Fig) caused surprisingly vast changes at telomeres, with proteins involved in chromatin maintenance, DNA replication, and repair showing both gain and loss of abundance. The relative abundance of some the proteins that associated with ATRX (such as NBS1 and PML) also changed in relative abundance at telomeres, but the most prominent changes involved other non-associating proteins. Loss of ATRX therefore carries indirect downstream changes that alter telomeres. We then examined the changes that occur in SLF2-null and isogenic SLF2/ATRX double null cells (Fig 6). In comparison to ATRX, SLF2 depletion had a far more limited effect on telomeric proteins, although chromatin remodelling and DNA repair proteins again changed in abundance at telomeres (Fig 6A). The double-null cells largely recapitulated the effects seen in the individual gene knockouts, suggesting that most of the effects are additive.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Changes at telomeres caused by a loss of SLF2 and ATRX.

Fold change of prey proteins identified by RAP1 BioID (BFDR ≤ 5%, SAINT [103]) in A: SLF2-null, and B: ATRX/SLF2-null vs. ATRX- and SLF2-expressing HEK293 Flp-In cells. Labeled proteins had ≥2.5 average spectra and ≥1.5-fold change between conditions. Proteins labeled in the scatter plot remained significant when NLS-BirA* was used as a control. Inset box represents additional proteins identified with NLS-BirA* as a bait (less stringent).

https://doi.org/10.1371/journal.pgen.1009909.g006

The increase in PIN2/TERF1-interacting telomerase inhibitor 1 (PINX1) at telomeres upon SLF2 loss is interesting because the protein inhibits telomerase activity [69] and promotes TERF1 binding to telomeres [70]. Proteomic changes caused by a loss of ATRX and/or SLF2 therefore likely destabilize telomeres.

Antibodies

The following primary antibodies were used: ATRX (Santa Cruz, sc-55584, PLA: 3 μg/mL, WB: 1:1000), ATRX (Santa Cruz, sc-15408, WB: 1:1000), ATRX (Thermo Fisher Scientific, PA5–21348, PLA: 1:1000–1:5000, WB: 1:1000), biotin (Cell Signaling Technology, 5597S, WB: 1:500), CCDC71 (Thermo Fisher Scientific, PA5–61543, PLA: 1:250), DAXX (Santa Cruz, sc-7152, PLA: 1:250, WB: 1:500), EZH2 (Sigma-Aldrich, MABE362, WB: 1:1000), FLAG (Sigma-Aldrich, F1804, IP: 2 μg/mg lysate, PLA: 1:1000, WB: 1:1000), GAPDH (Santa Cruz, sc-25778, WB: 1:1000), H3.3 (Abcam, ab176840, WB: 1:5000), HDAC1 (Thermo Fisher Scientific, PA1–860, WB: 1:1000), HP1 (Cell Signaling Technology, 2616S, WB: 1:1000), IMP3 (Thermo Fisher Scientific, PA5–26897, IF: 1:500, PLA: 1:500, WB: 1:1000), LaminB1 (Proteintech, 66095–1-Ig, WB: 1:50000), c-myc (Invitrogen, MA1–980, PLA: 1:2000), NBS1 (Novus Biologicals, NB100–143, PLA: 1:200, WB: 1:1000), SLF1 (Sigma-Aldrich, SAB2701555, WB: 1:1000), SLF2 (Thermo Fisher Scientific, PA5–66091, PLA: 2 μg/mL), SMC5 (Bethyl Laboratories, A300–236A, IF: 1:1000, WB: 1:1000), beta-tubulin (Developmental Studies Hybridoma Bank, E7, WB: 1:10000), ZBTB4 (Novus Biologicals, NBP1–76517, PLA: 1:10000), and ZFHX4 (Sigma-Aldrich, HPA023837, PLA: 1:500, WB: 1:500). The following secondary antibodies were used: anti-rabbit IgG HRP (Invitrogen, 0031458, WB: 1:10,000), anti-mouse kappa light chain HRP (Abcam, AB99617, WB: 1:10,000), anti-Rabbit IgG Alexa Fluor 594 (Jackson ImmunoResearch, 111–585-008, IF: 1:250), anti-rabbit IgG ATTO 647N (Rockland, 611–156-122S, IF: 1:250), anti-mouse IgG ATTO 488 (Rockland, 610–152-121S, IF: 1:250), and anti-mouse IgG AlexaFluor 647 (BioLegend, 405322, IF: 1:500).

cDNA constructs

ATRX cDNA was amplified by PCR and the DNA band was excised and purified using the PureLink Gel Extraction Kit (Thermo Fisher Scientific, K210012) and introduced into pDONR223 using Gateway BP Clonase II Enzyme mix (Thermo Fisher Scientific, 8602289719), as well as a pFastBac vector (ThermoFisher Scientific) for expression in insect cells. pDONR223 plasmids containing CCDC71 and FAM207A were generated in the laboratory of Dr. Anne-Claude Gingras. Plasmids were prepared using Stable Competent E. coli (NEB, C3040I). Clones were selected from LB agar plates (1% tryptone, 0.5% yeast extract, 170 mM NaCl, pH 7.5) containing 50 μg/mL spectinomycin. ATRX was subcloned into a pDEST vector containing an N- or C-terminal FLAG-BirA* using Gateway LR Clonase II Enzyme mix (Thermo Fisher Scientific, 11791020). Transformed cells were selected on LB agar plates with 50 μg/mL ampicillin. The pDEST-pcDNA5-RAP1-BirA*-FLAG-C-term plasmid was prepared in the laboratory of Dr. Anne-Claude Gingras. Lentiviral particles encoding myc-tagged SLF2 were prepared in the laboratory of Dr. Grant Stewart.

Cell lines and cell culture

HEK293 Flp-In T-REx cells (Thermo Fisher Scientific) were cultured in DMEM (Corning, 10–017-CV) supplemented with 10% FBS (Wisent, 080–150), 1X penicillin/streptomycin, 5 μg/mL blasticidin, and 75 μg/mL zeocin. HeLa S3 cells were grown in Joklik MEM (Caisson Labs), and MO3.13 (Cellutions Biosystems) in DMEM. Both were supplemented with bovine serum and penicillin/streptomycin. All cell lines were regularly tested for mycoplasma. Transfections were performed with lipofectamine 3000 (Thermo Fisher Scientific, L3000008) according to manufacturer’s instructions. For HEK293 Flp-In T-REx cells, transfections were performed at a 9:1 ratio of pOG44 to pDEST plasmid. Transfected HEK293 Flp-In cells (BirA*, NLS-BirA*, ATRX-BirA*, and RAP1-BirA*) were cultured in DMEM supplemented with 10% FBS, 1X penicillin/streptomycin, 5 μg/mL blasticidin, and 200 μg/mL hygromycin (for selection, or 100 μg/mL for maintenance). To generate ATRX KO lines, an sgRNA targeting exon 9 (5’- AAATTCCGAGTTTCGAGCGA -3’) was cloned into pSpCas9(BB)-2A-Puro (pX459) [97], a gift from Dr. Feng Zhang, Addgene plasmid 48139. For the SLF2 KO lines, sgRNA guides targeting exon 5 (5’- AGTTTCATCACTCGGTTCCT, GGCTTGGCACCTTCAAATTC -3’) were cloned into pSpCas9n(BB)-2A-GFP (pX461). Double ATRX/SLF2 KO cells were obtained by disrupting the SLF2 gene in the ATRX-null cells. Constructs were transfected into HEK293 Flp-In RAP1-BirA* cells as described above. To validate the gene disruptions, genomic DNA was isolated by suspending cell pellets in genomic DNA extraction buffer (20 mM Tris pH 7.6, 150 mM KCl, 1 mM EDTA, 20% glycerol, 0.5% NP-40) supplemented with 1 mM DTT, 25 μg/mL RNase A, and 20 U/mL proteinase K and incubated at 60°C for 2 hr. DNA was extracted with phenol:chloroform and the disrupted genes were PCR amplified. PCR products were purified using PureLink PCR Purification Kit (Invitrogen, K310002) following the manufacturer’s protocol and prepared for Sanger sequencing using the BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher, 4337457). Samples were sequenced on the Applied Biosystems SeqStudio Genetic Analyzer (Thermo Fisher). ATRX-null cells were further verified by western blotting. For SLF2, validation at the RNA level was done using PowerUp SYBR Green Master Mix (Applied Biosystems, A25742) according to manufacturer’s instructions, with the following primers: SLF2-RT-F 5’-AAACACTTTGTGCTACTCTGTGG-3’; SLF2-RT-R 5’-GTATCCTGGCGACCAAGTCTTTCA-3’; GAPDH-F 5’-CAATGACCCCTTCATTGACCTC-3’; and GAPDH-R 5’-GATCTCGCTCCTGGAAGATG-3’. Samples were analyzed with the QuantStudio 3 Real-Time PCR System (Applied Biosystems, A28567). Pyridostatin was purchased from Cayman Chemical (18013).

BioID

Protocol was adapted from previously published work [98]. All BioID runs were performed in biological duplicates as previously described [98], but with results compared to 6 replicates of untransfected HEK293 Flp-In T-REx, 14 replicates of cells expressing BirA* and, where indicated, also 14 replicates of cells expressing BirA*-NLS. Less than 20 cell passages elapsed between the CRISPR-Cas9 gene disruptions and the RAP1 BioID experiments. Cells were seeded at 70% confluency and induced with 1 μg/mL tetracycline (or doxycycline concentrations yielding near equal protein expression across samples) for 25 hr and 50 μM biotin added during the last 8 hr. Cells were pelleted with at least 0.1 g per sample, and snap frozen. Upon processing, cells were lysed in modified RIPA buffer (50 mM Tris pH 7.5, 150 mM NaCl, 1.5 mM MgCl₂, 1 mM EGTA, 0.1% SDS, 1% IGEPAL CA-630) with freshly added sodium deoxycholate (0.4%) and protease inhibitors (Sigma-Aldrich P8340) at 400 μL/0.1 g cells and solubilized for 20 min, gently rotating at 4°C (all the following 4°C incubations were also done with gentle rotation). Samples were sonicated at 25% amplitude for 5 sec on, 3 sec off cycles, for 3 cycles, using a Qsonica sonicator with CL-18 probe. Benzonase (Millipore 70746) was added and incubated for 15 min, 4°C (1.5 μL or 375 units per sample). Samples were spiked with additional SDS to a final concentration of 0.4% and incubated 15 min, 4°C. Samples were spun 16,000 x g for 20 min and the cleared lysates (supernatant) was transferred to a new tube. Streptavidin sepharose beads (GE 17–5113-01) were washed 3X with modified RIPA buffer with 0.4% SDS, and 30 μl (bed volume) was added to each sample and incubated 3 hr, 4°C. Samples were washed once with wash buffer (2% SDS, 50 mM Tris pH 7.5), 2X with modified RIPA buffer with 0.4% SDS, and 3X with ABC buffer (50 mM ammonium bicarbonate pH 8.5). Samples were spun, supernatant removed, and on-bead trypsin digest of peptides was performed by incubating with 1 μg trypsin dissolved in ABC buffer, rotating overnight at 37°C. An additional 0.5 μg trypsin per sample was added the next day and samples incubated for 2 hr at 37°C. Samples were gently vortexed, spun down, and supernatant transferred to new tube, with additional washing of beads and collection of supernatant performed twice (30 μL of HPLC grade water). Fresh 50% formic acid was added to samples to a final concentration of 2% prior to drying by vacuum centrifugation and subsequent storage at -80°C.

Mass spectrometry acquisition

Each sample (6 μL in 2% formic acid; corresponding to 1/6th of a 15 cm tissue culture dish) was directly loaded at 800 nL/min onto an equilibrated HPLC column (pulled and packed in-house). The peptides were eluted from the column over a 90 min gradient generated by a Eksigent ekspert nanoLC 425 (Eksigent, Dublin CA) nano-pump and analysed on a TripleTOF 6600 instrument (AB SCIEX, Concord, Ontario, Canada). The gradient was delivered at 400 nL/min starting from 2% acetonitrile with 0.1% formic acid to 35% acetonitrile with 0.1% formic acid over 90 min followed by a 15 min clean-up at 80% acetonitrile with 0.1% formic acid, and a 15 min equilibration period back to 2% acetonitrile with 0.1% formic acid, for a total of 120 min. To minimize carryover between each sample, the analytical column was washed for 2 hr by running an alternating sawtooth gradient from 35% acetonitrile with 0.1% formic acid to 80% acetonitrile with 0.1% formic acid at a flow rate of 1500 nL/min, holding each gradient concentration for 5 min. Analytical column and instrument performance was verified after each sample by loading 30 fmol bovine serum albumin (BSA) tryptic peptide standard with 60 fmol alpha-casein tryptic digest and running a short 30 min gradient. TOF MS mass calibration was performed on BSA reference ions before running the next sample to adjust for mass drift and verify peak intensity. Samples were analyzed with two separate injections with instrument method set to data dependent acquisition (DDA) mode. The DDA method consisted of one 250 milliseconds (ms) MS1 TOF survey scan from 400–1800 Da followed by ten 100 ms MS2 candidate ion scans from 100–1800 Da in high sensitivity mode. Only ions with a charge of 2+ to 5+ that exceeded a threshold of 300 cps were selected for MS2, and former precursors were excluded for 7 seconds after one occurrence.

Data-dependent acquisition data search

Mass spectrometry data generated were stored, searched, and analyzed using ProHits laboratory information management system (LIMS) platform [99]. Within ProHits, WIFF files were converted to an MGF format using the WIFF2MGF converter and to an mzML format using ProteoWizard (V3.0.10702) and the AB SCIEX MS Data Converter (V1.3 beta). The data were then searched using Mascot (V2.3.02) [100] and Comet (V2016.01 rev.2) [101]. The spectra were searched with the human and adenovirus sequences in the RefSeq database (version 57, January 30th, 2013) acquired from NCBI, supplemented with “common contaminants” from the Max Planck Institute (http://maxquant.org/contaminants.zip) and the Global Proteome Machine (GPM; ftp://ftp.thegpm.org/fasta/cRAP/crap.fasta), forward and reverse sequences (labeled “gi|9999” or “DECOY”), sequence tags (BirA, GST26, mCherry and GFP) and streptavidin, for a total of 72,481 entries. Database parameters were set to search for tryptic cleavages, allowing up to 2 missed cleavages sites per peptide with a mass tolerance of 35 ppm for precursors with charges of 2+ to 4+ and a tolerance of 0.15 amu for fragment ions. Variable modifications were selected for deamidated asparagine and glutamine and oxidized methionine. Results from each search engine were analyzed through TPP (the Trans-Proteomic Pipeline, v.4.7 POLAR VORTEX rev 1) via the iProphet pipeline [102].

SAINT analysis

SAINTexpress version 3.6.1 [103] was used as a statistical tool to calculate the probability of potential protein-protein associations compared to background contaminants using default parameters, with bait compression set to 2 and control compression set to 4. A 95% FDR iProphet filter was used. SAINT scores with a Bayesian false discovery rate (BFDR) ≤ 1% were considered high-confidence protein interactions. All non-human protein interactors (did not start with “NP” in Prey column) were removed from the SAINT analysis, except for BirA R118G H0QFJ5. Dot plots were generated using the “Dot plot generator” tool in ProHits-viz [104] using SAINTexpress file generated from ProHits. Data were normalized by total abundance. Gene ontology term analysis for biological process (GO:BP) was done using g:Profiler [105].

Proximity ligation assay

Cells were seeded to a density of 50–70% on 12 mm glass coverslips coated in poly-L-lysine (Sigma-Aldrich, P8920). Cells were fixed with a 2% PFA solution (2% PFA, 0.2% Triton-X-100, pH 8.2), washed with 1X PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄), permeabilized with 0.5% tergitol (Sigma-Aldrich, NP40S), and washed with 1X PBS. PLAs were performed according to manufacturer’s instructions using the Duolink In Situ Detection Kit (Sigma-Aldrich, DUO92013). Confocal microscopy was done at the SickKids Imaging Facility and foci counted using CellProfiler [106]. Signal intensity was obtained using ImageJ, with background signal subtracted for each figure [107].

Cell lysis & subcellular fractionation

All steps below were performed at 4°C. Subcellular fractionation was done by gently dounce homogenizing PBS-washed cell pellets in 2 pellet volumes of cytosolic lysis buffer (20 mM Tris pH 7.6, 10 mM NaCl, 1 mM EDTA, 0.5% IGEPAL, 0.5 mM DTT, protease inhibitors) using a loose pestle. Nuclei were recovered by centrifugation at 500 x g for 5 min, and the supernatant containing the cytoplasmic fraction was transferred to a new tube. Nuclei were then resuspended in 3 pellet volumes of nuclear extraction buffer (20 mM Tris pH 7.6, 300 mM NaCl, 1 mM EDTA, 20% glycerol, 0.5 mM DTT) with protease inhibitors and dounce homogenized with a tight pestle. The mixture was spun down at 30,000 x g for 15 min, and the supernatant containing nuclear extracts was transferred to a fresh tube. The remaining insoluble pellet was resuspended in 3 volumes of BC100 (20 mM Tris pH 7.6, 100 mM NaCl, 0.2 mM EDTA, 20% glycerol, 0.5 mM DTT) with protease inhibitors. The sample was then sonicated and digested with 75 U/μL MNase in the presence of 5 mM CaCl₂ for 2 hr, with gentle rotation. The subcellular fractions were dialyzed against BC100 and insoluble material removed by centrifugation as per conditions above. Whole cell extracts were generated by incubating PBS-washed cell pellets with 3 pellet volumes of lysis buffer (50 mM Tris pH 7.6, 100 mM KCl, 2 mM EDTA, 0.1% NP-40, 10% glycerol, 0.5 mM DTT). Samples were then sonicated, digested and processed as per the solubilized nuclear fraction above.

Immunoprecipitation

Cells were seeded at 4 x 10⁶ cells per plate and induced with 1 μg/mL doxycycline for 24 hr. The cells were harvested by first washing with cold 1X PBS, scraping, and collecting the cells in 1 mL 1X PBS. The cells were spun down at 500 x g for 5 min at 4ºC, and lysed in 1X IP lysis buffer (50 mM Tris pH 7.6, 100 mM KCl, 2 mM EDTA, 0.1% IGEPAL, 10% glycerol). The lysate was flash frozen on dry ice for 10 min and then allowed to thaw at room temperature. The lysate was then sonicated 3X for 10 sec at 1.5 amp. After sonication, the lysate was digested with 1 U/μL MNase and the buffer supplemented with 5 mM CalCl₂. This was left to incubate overnight with rocking at 4ºC. Four milligrams of protein was incubated with anti-FLAG (1 μg/mg lysate) overnight at 4ºC, rocking. The following day, magnetic Protein G beads (Cytiva, 28967070) were washed three times in 1X IP lysis buffer and the lysate and antibody mix was added onto the beads and incubated at 4ºC for 4 hr with rocking. The supernatant was removed, and the beads were washed three times in 1X IP lysis buffer. Proteins were eluted with 40 μL 1X Laemmli (2% SDS, 0.1% bromophenol blue, 10% glycerol, 62.5 mM Tris pH 6.8, 100 mM DTT) and boiled at 95ºC for 15 min.

In vitro transcription/translation & pulldown assays

FLAG-ATRX was expressed from baculovirus infected SF9 insect cells as per previous descriptions [108]. The protein was immobilized on an M2 resin (MilliporeSigma), washed, and eluted with 25 μg/ml 3X FLAG peptide (MilliporeSigma). The TnT Quick Coupled Transcription/Translation System (Promega, L1170) was used to generate FAM207A in accordance with manufacturer’s instructions, with the addition of Transcend tRNA (Promega L506A) and 1 μg/mL leupeptin. Recombinant FAM207A was incubated with Strep-Tactin MacroPrep resin (IBA LifeSciences 2–1505-010) in BC150 (20 mM Tris pH 7.6, 150 mM NaCl, 0.2 mM EDTA, 20% glycerol) for 30 min at 4°C prior to the addition of 20 pmol of recombinant ATRX. Proteins were incubated together at room temperature for 1 hr before washing with BC150 + 0.1% NP40, BC300 (20 mM Tris pH 7.6, 300 mM NaCl, 0.2 mM EDTA, 20% glycerol) + 0.1% NP40, and BC150. Samples were spiked with 1X Laemmli buffer and boiled at 95°C for 10 min prior to Western blotting.

Western blotting

Protein lysates were diluted in 1X Laemmli buffer and boiled at 95°C for 5 min. For streptavidin pull downs, 2 mM biotin was added to 1X Laemmli prior to boiling. Protein lysates were loaded onto 8–12% Tris-Glycine gels and run at 100 V for 90 min in 1X running buffer (25 mM Tris, 192 mM glycine, 0.1% SDS). Proteins were then transferred onto activated PVDF membranes for 2 hr at 270 mA in 1X transfer buffer (200 mM glycine, 25 mM Tris) using a Mini-protean wet transfer system (BioRad). All samples were run this way, with the exception of ZFHX4 fractionations, which were loaded onto a 3–8% Tris-Acetate gel (Thermo Fisher, EA0375PK2), run at 125V for 1.5 hr in 1X running buffer (Invitrogen, LA0041) supplemented with 500 μL antioxidant (Thermo Fisher, NP0005). Gels were washed in 20% ethanol for 10 min prior to transfer to PVDF with the iBlot (Thermo, IB21001) at 20V for 2 min, 23V for 6 min, and 25V for 4 min. All membranes were blocked in 5% milk in 1X TBST (20 mM Tris, 150 mM NaCl, 0.1% Tween) and incubated with primary antibody diluted in 1X TBST and 0.04% NaN₃ for 1 hr at room temperature, or overnight at 4°C with gentle rotation. Membranes were then washed 3X for 5 min with 1X TBST and incubated with HRP-conjugated secondary antibodies diluted in 1X TBST with 5% milk at room temperature, rocking for 1 hr. Membranes were then washed 3X 5 min with 1X TBST. To image, a 5X luminol solution (100 mM Tris pH 8.8, 1.25 mM luminol, 0.2 mM coumaric acid) was diluted as needed and supplemented with 5 μL 10% H₂O₂ per mL, and poured over the blot. For westerns of the TnT reactions, samples were treated as described above, with the exception of the luminol solution. Instead, blots were washed treated with streptavidin-AP (Promega V5591) according to manufacturer’s instructions, rocking at room temperature for 1 hr. Blots were washed twice with 1X TBST and then twice with water. Blots were then incubated in Western Blue Stabilized Substrate (Promega S384C) according to manufacturer’s instructions. Images were captured using a BioRad ChemiDoc XRS+ system.

CO-FISH

CO-FISH was performed as previously described [109] with the following modifications. Cells were treated with 10 μM BrdU for 18 hr and 0.2 μg/mL KaryoMAX Colcemid (Thermo Fisher, 15212012) for the final 4 hr. Metaphase spreads were prepared using a CDS 5 Cytogenetic Drying chamber (Thermotron) according to standard methods, rehydrated with 1X PBS, and slides were treated with 0.5 mg/mL RNase A (Life Technologies, 12091021) at 37°C for 10 min. Slides were stained with 0.5 μg/mL bisbenzimide Hoechst 33258 (Sigma-Aldrich, 14530) in 2X SSC (300 mM NaCl, 30 mM sodium citrate) for 15 min at room temperature. Slides were exposed to 365 nm UV, 5.4 x 10³ J/m2 twice. Next, slides were treated with 10 U/μL exonuclease III (NEB, M0206L) for 40 min at 37°C. Cells were washed in 1X PBS and then dehydrated in a series of 70, 95, and 100% ethanol. TelG-Cy3 (PNA Bio, F1006) and TelC-ALEXA 488 (PNA Bio, F1004) probes were heated at 70°C for 30 min, diluted in hybridization buffer [20 mM Tris pH 7.2, 60% formamide (Thermo Fisher, 15515026), 0.5% block (Roche, 11096176001)], and applied to slides for 2 hr each at room temperature, sequentially. Slides were washed with Wash Buffer 1 (10 mM Tris pH 7.2, 70% formamide, 0,1% BSA) and Wash Buffer 2 (100 mM Tris pH 7.2, 150 mM NaCl, 0.08% Tween-20) three times each. DAPI (Sigma-Aldrich, D9542) was added at final concentration of 1 μg/mL to the second wash with Wash Buffer 2. Then, slides were dehydrated in an ethanol series, as before, and mounted with ProLong Gold Antifade Mountant (Invitrogen, P36930). Slides were imaged with an Olympus BX61 microscope with the CytoPower automated imaging system (Applied Spectral Imaging). Telomere exchanges were manually scored in a blinded fashion, with sample identity only revealed after all samples were scored.

Immunofluorescence

Cells were seeded on coverslips and fixed in a 2% PFA solution for 20 min at room temperature after the desired treatment. Cells were washed 3X in 1X PBS and permeabilised with 0.5% tergitol in 1X PBS for 10 min at room temperature. Coverslips were washed 3X in 1X PBS and incubated in blocking buffer (3% BSA, 1% NGS, 1x PBS) for 1 hr at room temperature. Cells were then incubated in primary antibody diluted in blocking buffer for 1.5 hr at room temperature in a humidity chamber, washed 3X in 1X PBS, and incubated in secondary antibody diluted in blocking buffer for 30 min at room temperature. Coverslips were washed 3X with 1X PBS and treated with 1 μg/mL DAPI diluted in 1X PBS for 10 min at room temperature. Coverslips were washed once more in 1X PBS, mounted onto slides and imaged as described for the PLAs.

IF-FISH

Cells were seeded on slides, fixed in a 2% PFA solution for 10 min at room temperature, washed 3X in 1X PBS, and treated with 0.5% tergitol in 1X PBS for 10 min at room temperature. Cells were washed 3X in 1X PBS and incubated in blocking buffer for 1 hr at room temperature. Cells were then incubated in primary antibody diluted in blocking buffer for 1.5 hr at room temperature in a humidity chamber, washed 3X in 1X PBS, and incubated in secondary antibody diluted in blocking buffer for 30 min at room temperature. Cells were washed 2X with 1X PBS and fixed in 2% PFA in 1X PBS for 10 min at room temperature. Slides were dehydrated in an ethanol series of 70, 95, and 100% for 5 min each. Slides were air dried and TelC-488 probe (PNA Bio, F1004) was heated at 72°C for 30 min before being diluted 1:500 in hybridization buffer [70% formamide, 0.5% blocking reagent (Roche, 11096176001), 10 mM Tris pH 7.2] and being applied to slides. Slides were heated at 72°C for 10 min, sealed with rubber cement, and placed in a humidity chamber to allow probe binding overnight. The next day, slides were washed 2X, 15 min each at room temperature in wash buffer 1 (70% formamide, 10 mM Tris pH 7.2). Slides were then washed 3X, 5 min each at room temperature in wash buffer 2 (100 mM Tris pH 7.2, 150 mM NaCl, 0.08% Tween 20), with 1 μg/mL of DAPI being added to the second wash. Slides were dehydrated in an ethanol series as before, allowed to air dry, and mounted before being imaged. Colocalization was scored using the Cellprofiler Colocalization pipeline [106] scoring foci that were 2–15 pixels in diameter and greater than 0.15 units of intensity.