Nopp140-mediated concentration of telomerase in Cajal bodies regulates telomere length

CRISPR/Cas9 targeting of NOLC1 yields stable Nopp140 KD cell lines

Nopp140 is the only protein that concentrates in both nucleoli and CBs without being an integral part of an RNP (see Figure 6C later in this article). Nevertheless, Nopp140 associates with both H/ACA and C/D RNPs. For functional analysis, we targeted the Nopp140 gene NOLC1 with CRISPR/Cas9 technology. Other studies using genomewide CRISPR screens identified Nopp140 as an essential core fitness gene in most cells, except HeLa, possibly due to high ploidy (Hart et al., 2015; Wang et al., 2015). Indeed, after induction of Cas9 expression, we cloned three viable Nopp140 KD cell lines from HeLa but not HEK-293T cells transduced with two single guide (sg)RNAs complementary to exon 1 or 3 of NOLC1. Parent 1 cells (P1) were transduced with sgRNA 1 giving rise to the clonal Nopp140 KD lines KD1a and KD1b, whereas parent 2 cells (P2) were transduced with sgRNA 2 yielding the KD2 line. Western blots (WBs) of whole cell extracts showed Nopp140 KD cells to express little Nopp140 (Figure 1A). The low Nopp140 amounts for each cell line remains stable for years. KD1a and KD2 expressed ∼1% of Nopp140 relative to their parent cells, whereas some 7% of Nopp140 remained in KD1b cells (Figure 1B). In contrast, expression of the enzymes of H/ACA and C/D RNPs, NAP57 and fibrillarin, and the scaRNP and Cajal body (CB) marker proteins, WDR79 and coilin, respectively, remained unaffected in the KD cells (Figure 1, A and B). By indirect immunofluorescence (IF), Nopp140 was barely detectable in nucleoli and not at all in CBs of KD cells (Figure 1C). Expression levels of Nopp140 are uniform for each cell line, whether parent or KD cells. Importantly, there is no significant growth difference between any of the cells, which have been dividing with the same frequency for at least 2 yr (Supplemental Figure S1). Further, there are no gross phenotypes (Figure 1C, phase contrast). In summary, we generated stable cell lines expressing minimal levels of the essential Nopp140.

Nopp140 KD selectively displaces scaRNPs from CBs

Although overall cellular levels of Nopp140-associated proteins are unaffected by Nopp140 KD (Figure 1A), we asked whether the KD impacted their subcellular distribution. We first tested the major core protein of H/ACA RNPs, NAP57, which was identified as a Nopp140-associated protein (Meier and Blobel, 1994). In parent cells, NAP57 fluorescence in CBs (small arrows) marked by coilin is equal or greater to that in nucleoli (Figure 2A, Parent 1). In contrast, in Nopp140 KD cells, NAP57 is visibly diminished in CBs (small arrows) while nucleolar fluorescence intensity remains unaffected if not increased (Figure 2A, Knockdown 1a). This is confirmed by quantification of the fluorescence signal in CBs for NAP57 and all other proteins (Figure 2H). No difference is detected between the phase dense CBs of P1 versus those of KD1a cells (Figure 2A, phase contrast) indicating the absence of gross effects. To ensure that the difference in staining between P1 and KD1a was not due to staining artifacts, we mixed P1 and KD1a cells 1:1 and costained them for NAP57 (Figure 2B, green) and Nopp140 (red). Whereas the overlapping red–yellow staining was seen in both nucleoli and CBs of P1 cells, KD1a cells only showed NAP57 staining in nucleoli confirming the results from independent staining (Figure 2, A and B). In contrast, coilin was unaffected by Nopp140 KD and equally present in CBs of P1 and KD1a cells (Figure 2C, see also Figures 2, A, D, and G, and 1C). The major box C/D core protein fibrillarin is not as concentrated in CBs as NAP57 because there are less C/D than H/ACA scaRNPs (Meier, 2016), but it was still reduced in Nopp140 KD cells (Figure 2D). To test whether components of the other major RNPs of CBs, the spliceosomal snRNPs, were impacted by Nopp140 KD, we tested for the presence of their Sm core ring antigens (Figure 2E) and their assembly factor SMN (Figure 2F). Both were equally present in CBs of cells with and without Nopp140. However, WDR79, the scaRNP-specific protein responsible for localization of H/ACA and C/D RNPs to CBs, was also reduced in CBs of Nopp140 KD cells (Figure 2G), as were the H/ACA core proteins GAR1 and NHP2 (unpublished data). We quantified the fluorescence of these antigens in CBs of Nopp140 KD cells relative to parent cells (Figure 2H). The large variability is due to the intra- and intercell differences in size and fluorescence signal of CBs. Nevertheless, the results show that, in addition to Nopp140, the overall amount of scaRNP proteins, NAP57, fibrillarin, and WDR79, was reduced by more than 50% in CBs, whereas that of coilin and the snRNP proteins Sm and SMN varied little (Figure 2H). We conclude that Nopp140 KD leads to a specific loss of scaRNP proteins from CBs (NAP57, fibrillarin, and WDR79) without affecting the localization of other CB components (SMN, Sm, and coilin). Meanwhile, nucleolar localization of snoRNP proteins remained unchanged.

To test whether the loss of RNP proteins from CBs reflected that of intact protein–RNA complexes, we also interrogated the presence of scaRNAs and snRNAs in CBs using RNA fluorescence in situ hybridization (FISH) followed by indirect IF. The RNA FISH signal of various scaRNAs, hybrid C/D-H/ACA (Figure 3A), tandem H/ACA-H/ACA (Figure 3B), and the H/ACA telomerase RNA (Figure 3C), were reduced or lost from CBs (small arrows) of KD compared with those of parent cells. This was confirmed for SCARNA5 (U87, green) in CBs (coilin, red) of mixed P2 (yellow), and KD2 (red/orange) cells (Figure 3D). In contrast, U2 snRNA staining was indistinguishable in a mixed population of P1 and KD1a cells (Figure 3E, left panel) identified by Nopp140 (green) and coilin (red) presence (Figure 3E, right panel). The results for these and other scaRNAs were confirmed by quantification of hundreds of cells for the presence of the RNAs in CBs (Figure 3F). In all cases, KD1a cells exhibited significantly fewer cells with scaRNAs in CBs, whereas little difference was noted for the presence of the spliceosomal snRNA U2 (Figure 3F). Additionally, the C/D box snoRNA U3 was observed in significantly fewer CBs of KD2 compared with P2 cells (Figure 3F). Nevertheless, U3 was barely detectable in CBs and its CB levels were very low next to those in nucleoli and of bona fide scaRNAs (Supplemental Figure S2). The purpose for U3 transiting through CBs could be its need for cap hypermethylation by TGS1 and its pseudouridylation of at least two uridines (Reddy et al., 1979, 1980; Verheggen et al., 2002). Perhaps only U3 precursors make their way to CBs, explaining the low levels and detection in only about half of CBs. Regardless, the localization of U3 to nucleoli was unaffected by Nopp140 KD (Supplemental Figure S2).

The overall levels of scaRNAs did not change between P1 and KD1a cells as assessed by RT-qPCR (quantitative reverse transcription PCR) (Figure 3G), indicating that only their localization, but not their integrity, was affected by Nopp140 KD. Importantly, scaRNPs remained fully intact, that is, associated with WDR79, because immunoprecipitation (IP) with WDR79 antibodies followed by RT-PCR confirmed coprecipitation of equal amounts of SCARNA5/U87 and hTR from both parent and Nopp140 KD cells (Figure 3H). Taken together, our data document that Nopp140 KD displaces intact scaRNPs, but not snRNPs, from CBs.

Nopp140 reexpression rescues the effects of Nopp140 KD

Although three CRISPR cell lines generated with two different sgRNAs targeting NOLC1 exon 1 or 3 produced identical results, we confirmed the specificity for Nopp140 KD by rescue of the KD cells through Nopp140 reexpression. Overexpression of Nopp140 generates intranuclear membrane tubules and is toxic to cells (Isaac et al., 1998, 2001; Chen et al., 1999). Therefore, we expressed human Nopp140 from the low-expressing UBC promoter. Over time, exogenously expressed Nopp140 first accumulated in CBs and subsequently in nucleoli of the Nopp140 KD cells (Figure 4, A–C, first panels). Unlike in untransfected cells (stars), endogenous NAP57 and SCARNA5/U87 followed the transfected Nopp140 back into CBs (Figure 4, A and B, arrows). Expression of GFP-tagged Nopp140 also restored CB localization of NAP57 and WDR79 (Figure 4C, arrows). Similar results were obtained with fibrillarin, NHP2, and GAR 1 (unpublished data) and with SCARNA19/hTR and U93 (Supplemental Figure S3). We conclude that Nopp140 reexpression in CBs restores scaRNPs to CBs and that their depletion from CBs is not due to CRISPR/Cas9 off-target effects but specific to Nopp140 loss.

ScaRNP localization to CBs depends on two factors, the CAB box of scaRNAs and the WDR79 which binds to it (Darzacq et al., 2002; Richard et al., 2003; Mahmoudi et al., 2009; Tycowski et al., 2009; Venteicher et al., 2009). This is illustrated by DC causing mutations in WDR79 that lead to scaRNP (including hTR) loss from CBs (Zhong et al., 2011). Therefore, we tested whether overexpression of WDR79 in the Nopp140 KD cells might also restore scaRNP localization to CBs. Indeed, expression of GFP-tagged WDR79 in Nopp140 KD cells (arrows), restored NAP57 (as a marker for scaRNPs) to CBs (Figure 4D, small arrows). In contrast, residual Nopp140 in the KD cells failed to reaccumulate in CBs (Figure 4D). Together with the fact that the Nopp140 loss from CBs is accompanied by a loss of WDR79 (Figure 2G), which is restored upon Nopp140 reexpression (Figure 4C), these data indicate that Nopp140 acts upstream of WDR79 in scaRNP localization to CBs.

While this work was in progress, TDP43 was shown to regulate the CB localization of a varying subset of four C/D scaRNAs (Izumikawa et al., 2019). Transient KD of TDP43 displaced some of these scaRNAs from CBs. A loss of TDP43 from CBs in the Nopp140 KD cells suggests that Nopp140 is also responsible for the TDP43-­mediated localization of scaRNPs to CBs (Supplemental Figure S4). Finally, TGS1, the trimethyl transferase responsible for hypermethylating the monomethyl caps of snRNAs in the cytoplasm and hTR and snoRNA U3 possibly in CBs, is also concentrated in CBs (Verheggen et al., 2002) (Supplemental Figure S5). However, CB localization of TGS1 is reduced by about half in Nopp140 KD cells similar to the loss of hTR and U3 (Figure 3, C and F, and Supplemental Figure S5).

To further cement the role of Nopp140 in scaRNP localization to CBs, we replicated the effects of our stable Nopp140 KD cell lines by transient KD of Nopp140 using small interfering RNAs (siRNAs). Indeed, in regular HeLa cells transfected with Nopp140 siRNAs (Figure 4E, arrows), NAP57 was lost from CBs but not nucleoli and in contrast to CBs of untransfected cells (Figure 4E, small arrows). Similar results were observed in U2OS cells showing this to be a more general effect not unique to HeLa cells (unpublished data). Finally, we replicated the previously reported effect of transient WDR79 KD (Zhong et al., 2011) documenting a specific loss of NAP57 from CBs but not nucleoli in cells transfected with WDR79 siRNA (Figure 4F). In both cases, KD was confirmed by WB of whole cell extracts (Figure 4G). Overall, we conclude that Nopp140 is responsible for scaRNP accumulation in CBs upstream of WDR79. Consequently, we are left with CBs that still accumulate coilin, SMN, and snRNPs, but lack Nopp140 and scaRNPs.

Nopp140 loss changes the ultrastructure of nucleoli and CBs

By light microscopy, Nopp140- and scaRNP-less CBs in KD cells look indistinguishable from their normal counterparts (Figures 1C and 2A). But might differences be detected by electron microscopy on an ultrastructural level? Nopp140 is equally concentrated in the DFC of nucleoli and in CBs (Meier and Blobel, 1992, 1994; Thiry et al., 2009). Transmission electron microscopic analysis of nucleoli in parent cells (Figure 5A, outlined) revealed the canonical tripartite structure of fibrillar centers (FCs, asterisk) surrounded by DFCs (dark, circled area) altogether embedded in the granular component (GC). KD1a cells showed the same nucleolar hallmarks, except for the dark DFCs (Figure 5B). In fact, whereas 46 of 47 nucleoli of P1 cells exhibited distinct DFCs, none of 31 KD1a nucleoli did (Figure 5E). We conclude that the loosening of DFCs is a direct consequence of low Nopp140 levels in the KD cells. CBs of parent cells can be identified by their characteristic coils of dark granules (Figure 5C). The only visible difference between CBs of KD1a to parent cells was a reduction in granule size by about half (Figure 5D, quantified in Figure 5F). In contrast, granule number did not vary significantly (Figure 5G) nor did the overall shape and size of CBs (compare Figure 5, C and D). The reduction of granule size is the only visible difference between KD and parent cells. This observation, together with the loss of Nopp140 and scaRNPs from CBs being the only physiological difference between the KD and parent cells, strongly implies that these granules are the CB residences of scaRNPs and, likely, their targets, snRNPs.

Telomerase relocation from CBs to the nucleoplasm causes telomere lengthening

One of the scaRNPs dislocated from CBs in the Nopp140 KD cells is telomerase (Figure 3, C and F), which is responsible for maintaining telomere length. If telomerase is lost from CBs due to mutation or KD of WDR79, as is observed in some cases of DC, telomeres shorten critically (Zhong et al., 2011; Vogan et al., 2016; Chen et al., 2018). Hence, we assessed the telomere length in our cell lines using telomere restriction fragment length analysis. Surprisingly, in all three Nopp140 KD clones, telomeres lengthened over time (Figure 6A). Note that the parental cells are a mixture of cells, whereas the KD cells are individual clones with shorter initial telomere lengths that vary according to clone. Importantly, the lengths of telomeres in all three KD clones gradually increase, whereas those of the parent cells remained stable for more than 1 yr (Figure 6B). This gradual increase is in stark contrast to telomere lengthening caused by hTR and telomerase overexpression, which proceeds significantly faster (Cristofari and Lingner, 2006).

Nopp140 acts upstream of WDR79 in CB localization of scaRNPs

Our results show a clear dichotomy of RNP localization to CBs, that of scaRNPs and that of snRNPs, both accumulating independently. Indeed, KD of Nopp140 only affects localization of scaRNPs, whereas KD of coilin affects only that of snRNPs (Tucker et al., 2001). Therefore, and as previously suggested (Lemm et al., 2006), the congregation in CBs of the two types of RNPs, the scaRNPs and their modification targets, snRNPs, is handled independently by the cell. Localization of scaRNPs depends on their CB localizing elements, the CAB box and the GU/UG wobble stem (Darzacq et al., 2002; Richard et al., 2003; Marnef et al., 2014). Both elements are recognized by the protein WDR79, which is required for scaRNP accumulation in CBs (Tycowski et al., 2009; Venteicher et al., 2009). We now show that Nopp140 acts upstream of WDR79 in scaRNP localization to CBs because KD of Nopp140 prevents CB localization of WDR79, whereas overexpression of WDR79 fails to rescue CB localization of Nopp140 while rescuing that of scaRNPs. The latter may be achieved by excess binding of WDR79 to scaRNPs, for example, at two CAB boxes instead of one, or by weak binding to coilin (Mahmoudi et al., 2010; Enwerem et al., 2014). As it is, little is known about the mechanism of WDR79 association with scaRNPs, that is, are there assembly factors involved or are all sites on scaRNAs occupied? More importantly, how does Nopp140 concentrate in CBs in the first place? Liquid–liquid phase separation is the most likely answer (Zhu and Brangwynne, 2015). Nopp140 with its large central intrinsically disordered domain and its ability to bind both C/D and H/ACA scaRNPs, as well as coilin, is an ideal molecule to participate in the formation of CBs as phase-separated organelles in the membraneless nucleoplasm. Indeed, in the absence of coilin, Nopp140 together with scaRNPs forms residual CBs without snRNPs (Tucker et al., 2001). Hence Nopp140 is one of the core molecules underlying CB formation.

Safeguarding telomerase in CBs is required for telomere homeostasis

Telomerase is a specialized H/ACA scaRNP and concentrates with them in CBs. Although dependent on the CAB box of hTR and on WDR79, recruitment to CBs occurs through Nopp140. Mutation or ablation of the hTR CAB box or of WDR79 displaces telomerase from CBs and causes severe telomere shortening (Cristofari et al., 2007; Zhong et al., 2011; Vogan et al., 2016; Chen et al., 2018). However, displacement from CBs of the holo-telomerase scaRNP (with WDR79) leads to gradual telomere lengthening. How can this surprising and apparent contradiction be explained? We offer two possibilities.

First, CBs shelter telomerase during most of the cell cycle (Jády et al., 2004, 2006). However, during S-phase, telomerase shuttles from CBs to some telomeres to extend them (Jády et al., 2006; Cristofari et al., 2007). After extension, telomerase apparently returns to its harbor in CBs. We now show that Nopp140 is required as a landing pad or glue in CBs for all scaRNPs, including telomerase, which leads us to the following model. In the absence of its landing pad and CB shelter, telomerase concentration in the nucleoplasm slightly increases. Thus, telomerase may linger at telomeres for a prolonged period and slowly extend them (Figure 6D). There is an extreme precedence for such a mechanism when TERT and hTR are concomitantly overexpressed, telomeres extend severalfold beyond their normal length (Cristofari and Lingner, 2006). When telomerase is displaced from CBs, its nucleoplasmic concentration increases only slightly, explaining the observed slow but steady lengthening of telomeres.

Second, hTR may get modified in CBs, which would not happen or be reduced if it was displaced. For example, the monomethyl cap of hTR appears to be hypermethylated by TGS1 in CBs (Verheggen et al., 2002). Additionally, hTR is pseudouridylated attenuating the activity of telomerase while increasing its processivity (Kim et al., 2010). Thus, telomerase with unmodified hTR would exhibit heightened activity toward telomeres. If modification of hTR occurred in CBs, the absence of Nopp140 and of telomerase recruitment to CBs would result in telomerase with increased activity. This in turn could explain the observed slow telomere lengthening. Of course, both mechanisms could operate in parallel and additional explanations are possible. Regardless, the main difference between prior observations and our studies is that in our cells the holo-telomerase scaRNP remains intact; only its localization is affected.

Mechanism of scaRNP localization to CBs

Why is it particularly important to study effects on endogenous rather than exogenously expressed RNAs and proteins in the case of CBs? Recent evidence suggests CBs, like nucleoli, to be liquid–liquid phase-separated organelles (Zhu and Brangwynne, 2015). For example, expression of poly-arginine-glycine or -proline dipeptides disrupt nuclear phase-separated, membraneless organelles including CBs (Lee et al., 2016). Such organelles form in a manner that is exquisitely sensitive to the concentration of its constituents. Hence, exogenous expression of RNAs or proteins could change the partition of CB residents. Liquid–liquid phase separation is further stimulated by molecules that are intrinsically disordered and offer multiple binding sites, such as Nopp140 (Na et al., 2018). Nopp140, with its 10 alternating acidic and positively charged repeats, forms an ideal landing and neutralizing pad for sno- and scaRNPs with their acidic RNAs and positively charged proteins. Nopp140 can be considered as sca- and snoRNP glue. Why is it then that Nopp140 loss affects scaRNPs in CBs but not snoRNPs in nucleoli? Both organelles apparently form by liquid–liquid phase separation. Nucleoli nucleate around active clusters of rDNA transcription, whereas CBs may form spontaneously due to the high concentration of scaRNPs and snRNPs with the help of Nopp140 and coilin (Sawyer et al., 2019). There are several issues to be kept in mind when considering the differential effect of Nopp140 KD on nucleoli and CBs. First, these are KD and not knockout cells, so probably residual, if undetectable, levels of Nopp140 remain also in CBs. Second, these are clonal cells selected for survival, which depend on ribosome biogenesis in nucleoli but not on visible CBs, which are not essential and even undetectable in some cells. Third, the amounts of snoRNPs (and with it those of Nopp140) in the nucleolus is significantly higher than that of scaRNPs in CBs. Finally, the DFC of the nucleolus is not actually lost but its contrast in electron micrographs is reduced. In fact, the strong contrast of DFC and CBs may be partially caused by the extreme Nopp140 phosphorylation. Thus, even at reduced levels of Nopp140 in nucleoli (Figure 1C), evidenced by the loosened, less contrasted DFC in electron micrographs, RNA polymerase I transcription continues. Whether modification of rRNA is affected is currently being investigated and will be reported in a forthcoming publication. Regardless, the levels of Nopp140 are insufficient to keep scaRNPs focused in CBs. Interestingly, the Nopp140 organizing principle seems to be evolutionarily conserved as deletion of its yeast homologue Srp40p effects a loss of C/D RNAs from the nucleolar body, the closest relative to a CB in yeast (Meier, 1996; Verheggen et al., 2002).

Finally, the apparently well-tolerated consequences of Nopp140 KD on cell growth and morphology offer a reduction of Nopp140 levels as a therapeutic avenue to rescue telomere length in DC patients and aging.

Cell culture and genome engineering

HeLa cells and different clones were cultured in DMEM and 10% fetal bovine serum at 37°C under 5% CO₂ in air. Genome engineering was done using the CRISPR/Cas9 system. HeLa cells were transduced with Cas9 lentivirus, selected for 3 d with 1 µg/ml puromycin, then transduced with lentivirus containing two NOLC1-sgRNAs, sgRNA-Nopp140-Exon1, and sgRNA-Nopp140-Exon3 (Table 1) to obtain P1 and P2 cells, respectively, and selected for 7 d with 5 µg/ml blasticidin. Subsequently, Cas9 expression was induced for 7 d with 500 ng/ml doxycycline while maintaining puromycin and blasticidin selection. Single clones were obtained by limited dilution and tested by indirect IF and WB for Nopp140 loss. Clones with significantly decreased expression of Nopp140 underwent another round of limited dilution cloning and were used for further analysis. From P1, resulted the KD1a and KD1b clones, and from P2, the KD2 clone. For growth evaluation, 12,000 cells of each clone were plated on day 1 and expansion assessed on 4 subsequent days using a Countess II automated cell counter (Thermo Fisher Scientific). Correct targeting by the sgRNAs was confirmed by DNA amplification of the target sequences for each cell line (using primers U659 and U660 for KD1a/KD1b and primers U656 and U657 for KD2; Table 1) followed by Sanger sequencing (with primer U661 for KD1a/KD1b and U658 for KD2; Table 1). Proper targeting of the sgRNAs was also evident from RNA-Seq data of the cell lines (unpublished results). Even reinduction of Cas9 in our clones failed to reduce Nopp140 expression further. As in a prior study, the reasons for the residual expression are not entirely clear, but could be due to polyploidy (Hart et al., 2015). Perhaps one allele is located in a mostly silenced region of the chromatin preventing Cas9 access while allowing minimal expression. Obviously, single cell cloning would select for those cells that still exhibit some expression of an essential gene.

Plasmids, siRNA, and transfection

Nopp140 plasmids used in this study are based on pNK65 (HA-hNopp140-mGFP under CMV promoter). HA-hNopp140 under the CMV promoter (pJB8) was generated by replacing the mGFP in pNK65 with a STOP codon using primers U667 and U668. HA-hNopp140-mGFP under the UBC promoter (pJB9) was generated by replacing the CMV promoter in pNK65 with the UBC promoter amplified off pGL3-UBC promoter-HSP70 using primers U662 and U663. HA-hNopp140 under the UBC promoter (pJB10) was generated by removal of mGFP from pJB9 like that from pNK65 to obtain pJB8.

Lipofectamine 3000 was used for plasmid and siRNA transfections following the manufacturer’s protocol. pJB9 and pJB10 plasmids were transfected for 24 h, pEGFP-WRAP53beta (WDR79; addgene.org) was transfected for 16 h, and Nopp140 and WDR79 siRNAs were transfected for 72 h before analysis.

Antibodies

Antibodies (dilutions in parentheses) for WB, IP, indirect IF, and FISH followed by IF (FISH-IF) were as follows: anti-Nopp140 rabbit serum (RS8 at 1:5000 for WB; 1:1000 for IF and FISH-IF) (Kittur et al., 2007); anti-NAP57 rabbit serum (RU10 at 1:50 for IP; 1:200 for WB and IF) (Darzacq et al., 2006); anti-NHP2 rabbit serum (p15 at 1:100 for IF) (Pogacic et al., 2000); anti-GAR1 rabbit serum (p16 at 1:100 for IF) (Dragon et al., 2000); anti-WDR79 rabbit serum (1:2000 for WB, 1:300 for IF, and 1:500 for IP; Novus Biologicals); anti-TDP43 rabbit serum (1:200 for IF; Proteintech); anti-TGS1 rabbit serum (1:500 for IF; ABclonal Science); anti-coilin rabbit serum (1:5000 for WB and 1:1000 for IF; Proteintech); anti-coilin mouse ascites fluid (5P10 at 1:1000 for IF and FISH-IF) (Almeida et al., 1998); mouse monoclonal anti-NAP57 immunoglobulin G (IgG) (H3 at 1:500 for IF; Santa Cruz Biotechnology); mouse monoclonal anti-fibrillarin culture supernatant (38F3 at 1:2000 for WB; EnCor Biotechnology); mouse monoclonal anti-Sm antibodies (Y12 at 1:100 for IF; Abcam); mouse anti-SMN ascites fluid (MANSMA1 at 1:100 for IF) (Young et al., 2000); mouse anti–γ-tubulin ascites fluid (GTU-88 at 1:5000 for WB; Sigma); DyLight488 goat anti-mouse IgG (1:500 for IF) and rhodamine (TRITC) goat anti-rabbit IgG (1:500 for IF; both Jackson Immuno Research); Alexa Fluor 647 goat anti-mouse IgG (1:500 for IF), Alexa Fluor 350 goat anti-mouse IgG (1:1500 for FISH-IF), Alexa Fluor 488 goat anti-­rabbit IgG (1:1500 for FISH-IF), and Alexa Fluor 680 goat anti-rabbit IgG (1:10,000 for WB; all four from Thermo Fisher Scientific); IRDyeTM 800 goat anti-mouse IgG (1:10,000 for WB; Rockland Immunochemicals).

WBs

For each experiment, proteins from the same number of cells per condition were extracted into SDS-sample buffer (0.5 M Tris; pH6.8; 12% SDS; 0.05% bromophenol blue). The lysates were tip sonicated and total proteins were loaded (100,000 cell equivalents), separated on 9% SDS–PAGE, followed by transfer to nitrocellulose membrane. Transfer efficiency was confirmed by Ponceau red staining, and membranes were blocked in blocking buffer (Tris-buffered saline, 0.1% Tween, and 2.5% nonfat dry milk) for 30 min before incubation with primary antibodies diluted in blocking buffer overnight at 4°C. After extensive washes in blocking buffer, membranes were incubated with appropriate secondary antibodies diluted in blocking buffer for 1 h at room temperature in the dark. After extensive washes in blocking buffer, membranes were scanned on an Odyssey 9120 Imaging System (LI-COR Biosciences), and protein bands were quantified using Image Studio Lite (LI-COR Biosciences).

Indirect IF

Cells grown on coverslips were fixed in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) for 20 min, permeabilized with 1% Triton X-100 in PBS for 5 min, and blocked with 1% powdered milk in PBS (IF blocking buffer) for 15 min. The cells were then incubated for 2 h with primary antibodies in IF blocking buffer, washed, and incubated for 1 h with secondary antibodies in IF blocking buffer in the dark. This was followed by washing and nuclear staining with 4′,6-diamidino-2-phenylidone (DAPI; 1 µg/ml in PBS). Coverslips were mounted on glass slides using ProLong Diamond Antifade Mount (Thermo Fisher Scientific) and observed using a Zeiss Axio Observer Z1 fluorescence microscope (63× objective, NA 1.4) with filter sets 34-DAPI (Zeiss #000000-1031-334), 10-AF488 (Zeiss #488010-9901-000), 43HE-DsRED (Zeiss #489043-9901-000), and 50-Cy5 (Zeiss #488050-9901-000). Z-stack images in 200-nm steps were acquired with a Zeiss AxioCam MRm camera using Axiovision software (Zeiss). Maximum projections were generated using ImageJ (National Institutes of Health [NIH]). Quantification of protein signals in CBs was done using ImageJ with the help of macros (available upon request). Briefly, coilin images were used to locate the centroids of CBs around which masks were generated of circles with 8 pixel diameter (0.1024 µm/pixel). These masks were applied to the CB protein images to determine their signal intensity in CBs. Background was subtracted individually for each CB and was defined as the pixel with the lowest signal in a 2-pixel circumference of the mask using an ImageJ function. Images for figures were cropped and adjusted using Photoshop CC (Adobe). To compare parent and KD cell images, all images within the same panels and of the same antigens were acquired and adjusted identically. Methods to establish statistical significance are indicated in the figure legends and were calculated and plotted using Prism 8 (GraphPad Software).

FISH-IF

Cells on coverslips were washed with 5 mM MgCl₂ in PBS (PBSM), fixed with 4% PFA in PBSM for 10 min, quenched with 50 mM Glycine in PBSM for 5 min, permeabilized with 0.1% Triton X-100 for 10 min, and blocked with bovine serum albumin (BSA; 0.5 g/l) in PBSM (FISH-IF blocking buffer) for 15 min. Cells were then incubated with primary antibodies in FISH-IF blocking buffer for 1 h followed by secondary antibodies in FISH-IF blocking buffer for 1 h in the dark. After washes in FISH-IF blocking buffer, cells were incubated in 70% ethanol for 10 min followed by prehybridization buffer (2× SSC; 10% formamide; 0.5 g/l BSA; 1:1000 RNase OUT) for 30 min. Probes labeled with Quasar 570 on their 5′ and 3′ ends (Biosearch Technologies, each at 125 nM final; Table 2) were applied in hybridization buffer (2× SSC; 10% formamide; 0.5g/l BSA; 1 µg/µl yeast tRNA; 2 mM vanadyl ribonucleoside complex; 10% dextran sulfate; 1:100 RNase OUT) for 3 h at 37°C in the dark, then washed with wash buffer 1 (2× SSC; 10% formamide), followed by wash buffer 2 (2× SSC), and the coverslips were mounted as described for IF. The samples were observed using an Olympus BX61 epifluorescence microscope with PlanApo 60×, 1.4NA, oil-immersion objective (Olympus). An X-Cite 120 PC (Lumen Dynamics, Canada) light source was used for illumination, with filter sets DAPI-5060C-Zero (Semrock), Cy3-41007a (Chroma), and FITC-5050A-Zero (Semrock). Z-stack images in 200-nm steps were acquired with a CoolSNAP HQ camera (Photometrics) using MetaMorph (Molecular Devices) and processed as described for IF.

Electron microscopy

Monolayers of cells were fixed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, postfixed with 1% osmium tetroxide followed by 2% uranyl acetate, and dehydrated through a graded series of ethanol, and the cells were lifted from the monolayer with propylene oxide and embedded as a loose pellet in LX112 resin (LADD Research Industries, Burlington, VT) in Eppendorf tubes. Ultrathin sections were cut on a Leica Ultracut UC7, stained with uranyl acetate followed by lead citrate, and viewed on a JEOL 1400 Plus transmission electron microscope at 80 kv.

Total RNA extraction

RNA from the different cell lines was extracted using 500 µl TRIzol Reagent (Ambion) directly on 10-cm dishes (cell confluency ∼80%, ∼1,000,000 cells). Lysed cells were scraped into tubes, chloroform was extracted twice, and the RNA was precipitated with 0.7 vol isopropanol and 20 µg glycogen and resuspended in UltraPure distilled water. RNA concentration and quality were determined by Nanodrop (ratio 260/230 and 260/280 above 1.8). Total RNA was used as template for RT-qPCR analysis.

RT-qPCR

Residual DNA contamination was removed from 2 µg of RNA using DNase RQ1. 1 µg RNA (the other half was used as no RT control) was used for cDNA synthesis using random hexamers and SuperScript III reverse transcriptase (Thermo Fisher Scientific). cDNA corresponding to 5 ng equivalents of RNA templated the qPCR reaction with Power SYBR Green PCR master mix and amplified on a QuantStudio 6 Flex real-time PCR system (Applied Biosystems) The gene expression levels were calculated by comparative C_T approach (ΔΔC_T) and normalized for that of GLUD2 mRNA. Primers (Thermo Fisher Scientific) used for RT-qPCR are described in Table 1.

RNA-IP and RT-PCR

Cells (2,000,000 from two 10-cm dishes) were lysed in buffer A (20 mM Tris-HCl, pH 7.9; 150 mM NaCl; 1% Triton X-100; 0.2% SDS; 2 mM EDTA; protease inhibitor, complete EDTA-free cocktail tablet; Roche, 1 tablet/50 ml) and incubated on ice for 10 min. Lysates were cleared by centrifugation at 15,000 rpm (21.1 × g) for 10 min. A 6000 cell equivalent was kept as input control and 60,000 cell equivalents were incubated with anti-WDR79 serum at a dilution of 1:500 for 1 h at room temperature. Incubations were added to 25 µl of protein A sepharose beads (GE Healthcare) for 1 h at room temperature. The beads were washed once with buffer A, three times with buffer B (20 mM Tris-HCl, pH 7.4; 150 mM NaCl; 0.02% Triton X-100), and once with buffer C (20 mM Tris-HCl, pH 7.4; 150 mM NaCl) before RNA extraction using TRIzol (Ambion, following the manufacturer’s instructions). RNA concentrations of the inputs were determined using Nanodrop. Semiquantitative RT-PCR was performed using SuperScript III One-Step RT-PCR System (Thermo Fisher Scientific) on 200 ng of total input RNA and all RNA from the beads (input = 1/10 of precipitates). Primers (Thermo Fisher Scientific) used for RT-PCR are described in Table 1.

Telomere restriction fragment length analysis

DNA was isolated from the different cell lines at different population doublings on 15-cm plates (confluency roughly 80%) and treated overnight with proteinase K (in 10 mM Tris, pH 7.4; 100 mM NaCl; 10 mM EDTA; 1% SDS; 100 µg/ml proteinase K). DNA was extracted with phenol/chloroform and digested overnight with HinfI, Alu1, MboI, and RsaI (NEB enzymes in NEB buffer 2). Approximately 3 µg of the digested DNA was fractionated by electrophoresis on 1% agarose gels. Gels were dried and DNA was denatured in gel for 30 min with 0.5 M NaOH and 1.5 M NaCl. Telomeres were detected by hybridization at 55°C overnight with a ³²P end-labeled oligonucleotide probe (TTAGGG)₄. Probe (2 µg) was prepared with γ-³²P-ATP using T4 Kinase (Thermo Fisher Scientific) in Church buffer (0.5 M sodium phosphate, pH 7.2; 1 mM EDTA; 7% SDS; 10 g/l BSA). Gels were scanned on a Typhoon PhosphorImager (GE Healthcare). Lanes were quantified using ImageJ (NIH) and median telomere length was determined in Excel (Microsoft).