CRM1 controls the composition of nucleoplasmic pre-snoRNA complexes to licence them for nucleolar transport
2011; Springer Nature; Volume: 30; Issue: 11 Linguagem: Inglês
10.1038/emboj.2011.128
ISSN1460-2075
AutoresBérengère Pradet‐Balade, Cyrille Girard, Séverine Boulon, Conception Paul, Karim Azzag, Rémy Bordonné, Édouard Bertrand, Céline Verheggen,
Tópico(s)RNA regulation and disease
ResumoArticle26 April 2011free access CRM1 controls the composition of nucleoplasmic pre-snoRNA complexes to licence them for nucleolar transport Bérengère Pradet-Balade Bérengère Pradet-Balade Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Cyrille Girard Cyrille Girard Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Séverine Boulon Séverine Boulon Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Conception Paul Conception Paul Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Karim Azzag Karim Azzag Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Rémy Bordonné Rémy Bordonné Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Edouard Bertrand Corresponding Author Edouard Bertrand Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Céline Verheggen Corresponding Author Céline Verheggen Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Bérengère Pradet-Balade Bérengère Pradet-Balade Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Cyrille Girard Cyrille Girard Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Séverine Boulon Séverine Boulon Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Conception Paul Conception Paul Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Karim Azzag Karim Azzag Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Rémy Bordonné Rémy Bordonné Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Edouard Bertrand Corresponding Author Edouard Bertrand Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Céline Verheggen Corresponding Author Céline Verheggen Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France Search for more papers by this author Author Information Bérengère Pradet-Balade1,‡, Cyrille Girard1,‡, Séverine Boulon1, Conception Paul1, Karim Azzag1, Rémy Bordonné1, Edouard Bertrand 1 and Céline Verheggen 1 1Institut de Génétique Moléculaire de Montpellier, UMR 5535 CNRS, Université Montpellier I and II, Montpellier Cedex 5, France ‡These authors contributed equally to this work *Corresponding authors: UMR 5535 CNRS, Institut de Génétique Moléculaire de Montpellier, 1919 route de Mende, 34293 Montpellier Cedex 5, France. Tel.: +33 434359646; Fax: +33 467040231; E-mail: edouar[email protected] or E-mail: [email protected] The EMBO Journal (2011)30:2205-2218https://doi.org/10.1038/emboj.2011.128 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Transport of C/D snoRNPs to nucleoli involves nuclear export factors. In particular, CRM1 binds nascent snoRNPs, but its precise role remains unknown. We show here that both CRM1 and nucleocytoplasmic trafficking are required to transport snoRNPs to nucleoli, but the snoRNPs do not transit through the cytoplasm. Instead, CRM1 controls the composition of nucleoplasmic pre-snoRNP complexes. We observed that Tgs1 long form (Tgs1 LF), the long isoform of the cap hypermethylase, contains a leucine-rich nuclear export signal, shuttles in a CRM1-dependent manner, and binds to the nucleolar localization signal (NoLS) of the core snoRNP protein Nop58. In vitro data indicate that CRM1 binds Tgs1 LF and promotes its dissociation from Nop58 NoLS, and immunoprecipitation experiments from cells indicate that the association of Tgs1 LF with snoRNPs increases upon CRM1 inhibition. Thus, CRM1 appears to promote nucleolar transport of snoRNPs by removing Tgs1 LF from the Nop58 NoLS. Microarray/IP data show that this occurs on most snoRNPs, from both C/D and H/ACA families, and on the telomerase RNA. Hence, CRM1 provides a general molecular link between nuclear events and nucleocytoplasmic trafficking. Introduction Non-coding RNPs perform essential functions in the cell: (i) they are at the heart of the translation apparatus and catalyse the peptidyl-transferase reaction (Simonovic and Steitz, 2009); (ii) they have key roles in the processing of other RNAs (Wahl et al, 2009); and (iii) they have various regulatory functions, best exemplified by miRNAs, which regulate mRNA translation and stability (Chekulaeva and Filipowicz, 2009). To perform their function, non-coding RNAs must be processed and assembled with specific proteins to form stable RNP particles (Staley and Woolford, 2009). This often occurs in specific compartments where the RNAs have to localize transiently before reaching their site of function (Matera and Shpargel, 2006). How these RNAs are transported during their biogenesis is not well understood, in particular in the case of transit between subnuclear compartments. The m2,2,7G-capped U1, U2, U4 and U5 snRNAs are key components of the pre-mRNA splicing machinery and are one of the best understood paradigm to study RNP assembly and transport (for review, see Wahl et al (2009)). In vertebrate cells, the snRNA precursors are transcribed by RNA polymerase II. They receive an m7G-cap, and are exported to the cytoplasm by a complex containing the cap binding complex (CBC; Izaurralde et al, 1995), the phosphorylated export adaptor PHAX (Ohno et al, 2000), Ran and the exportin CRM1 (Fornerod et al, 1997). In the cytoplasm, snRNAs are assembled with the core Sm proteins by the survival of motor neuron (SMN) complex (for review, see Chari et al (2009)). SMN and the Sm core proteins then recruit the cap hypermethylase Tgs1 and this allows the conversion of the m7G-cap to a 2,2,7-trimethylguanosine (TMG; Mouaikel et al, 2002; Lemm et al, 2006; Girard et al, 2008). Consequently, snRNPs are reimported to nuclei and transported to Cajal bodies (CB), to finalize their maturation before being targeted to the nucleoplasm or to speckles (for review, see Chari et al (2009)). SnoRNAs are small, evolutionarily conserved RNAs that localize to nucleoli and function in rRNA biogenesis (for review, see Matera et al (2007)). On the basis of short consensus motifs, snoRNAs can be grouped into two families, the C/D and the H/ACA RNAs. SnoRNAs from each family fold into a similar secondary structure and associate with a small set of common proteins (reviewed in Matera et al (2007) and Reichow et al (2007)): 15.5K, Nop56, Nop58 and fibrillarin for C/D snoRNAs; Nhp2, Nop10, Gar1 and dyskerin for H/ACA snoRNAs. The snoRNAs are transcribed in the nucleoplasm either from their own gene or as part of introns of pre-mRNAs. The intronic snoRNAs are released from the pre-mRNA via a splicing-dependent pathway (Leverette et al, 1992; Hirose et al, 2003), and are uncapped. In contrast, snoRNAs from independent gene, such as U3, U8 and U13, are synthesized from snRNA-type RNA polymerase II promoters (Tyc and Steitz, 1989), and their precursors contain a short 3′ extension and an m7G-cap structure that is hypermethylated into a TMG by Tgs1. SnoRNA precursors are present in large multiprotein complexes (Boulon et al, 2004, 2008; Watkins et al, 2004). In addition to core snoRNP proteins, these complexes contain factors involved in transport (CBC, PHAX and CRM1; Boulon et al, 2004; Watkins et al, 2004), RNP assembly (Nufip, the R2TP complex and Hsp90) (Boulon et al, 2008; Zhao et al, 2008) and RNA processing (Tgs1, La, Lsm, the exosome) (for review, see Matera et al (2007)). Transport of C/D snoRNAs to nucleoli is only partially characterized. It was shown that C/D snoRNAs transit through CB where cap hypermethylation, 3′-end trimming and final assembly steps are thought to occur (Narayanan et al, 1999; Verheggen et al, 2002). More recently, we showed that PHAX and CRM1 are involved in the transport of capped C/D snoRNAs to nucleoli (Boulon et al, 2004). PHAX was shown to direct these complexes to CB, while CRM1 was required for their transport from CB to nucleoli. PHAX and CRM1 are not recruited concomitantly but sequentially: PHAX binds m7G-capped pre-snoRNA, while CRM1 associates predominantly with TMG-capped mature snoRNAs (Boulon et al, 2004). Given that C/D snoRNA transport does not seem to involve a cytoplasmic step (Terns et al, 1995), it is surprising that PHAX and CRM1 are required for targeting snoRNPs to nucleoli. It was, therefore, important to further investigate the role of these transport factors in snoRNP biogenesis and trafficking. In this study, we have elucidated the role of CRM1. We found that it controls the intranuclear trafficking of snoRNPs by promoting the dissociation of Tgs1 from their nucleolar localization signal (NoLS). Results Both functional nuclear pores and CRM1 are required to transport capped and uncapped C/D snoRNAs to the nucleolus We previously showed that CRM1 binds U3 precursor RNAs, and by microinjection experiments, that it was required to transport U3 from CB to nucleoli (Boulon et al, 2004). Indeed, in HeLa cells, inhibition of CRM1 with Leptomycin B (LMB) led to the accumulation of microinjected U3 in CB and prevented its localization to nucleoli. Since CRM1 is an exportin, we wondered whether its role in the trafficking of U3 was dependent on functional nuclear pores or not. To test this, we microinjected HeLa cells with m7G-capped fluorescent U3 in the presence of wheat germ agglutinin (WGA), which binds to the glycosylated residues of nucleoporins and inhibits their function (Yoneda et al, 1987). Upon WGA microinjection, U3 RNAs were not localized to nucleoli but accumulated in CB as seen by colocalization with coilin (Figure 1A). Microinjected RNAs had the same localization as when cells were pre-treated with LMB (Boulon et al, 2004), indicating that CRM1 inhibition and blocking nuclear pores had the same effect on snoRNA transport. To verify that the effect of WGA was specific, we repeated the experiment by coinjecting WGA with N-acetyl glucosamin, which prevents binding of WGA to nucleoporins. In this case, the ability of U3 snoRNA to localize to nucleoli was not affected (Supplementary Figure S1). Figure 1.CRM1 and nuclear pores are required for nucleolar targeting of microinjected C/D snoRNAs. (A) Cy3-labelled (red) fluorescent snoRNA U3 was microinjected in the nucleus of HeLa cells and cells were fixed 2 h later. CB were labelled with anti-coilin antibodies (green). On the merged image, DNA staining with DAPI (blue) is shown. In the control, U3 was localized to CB and nucleoli but after coinjection with WGA, it was only to CB. (B) Microinjection as for (A) but with uncapped U14 snoRNA. In the control, U14 was localized to CB and nucleoli but remained in the nucleoplasm upon LMB pre-treatment or WGA coinjection. The bar represents 5 μm. Download figure Download PowerPoint Since U3 is transcribed from an snRNA-type RNA polymerase II gene and its precursor possess an m7G-cap (Peculis and Steitz, 1993), we wondered whether the effect of LMB would extend to intronic C/D snoRNAs that do not have a cap. To this end, fluorescent uncapped U14 snoRNAs were synthesized in vitro and microinjected into the nuclei of HeLa cells. As shown in Figure 1B, after 2 h, most of the RNAs localized to CB and nucleoli. Interestingly, the nucleolar localization of U14 was lost when cells were treated with LMB or when WGA was coinjected. However, in contrast to capped C/D snoRNAs, U14 did not accumulate in large amount in CB but mostly remained in the nucleoplasm. This could be due to a difference in kinetic properties between U14 and U3 trafficking. Nevertheless, the loss of nucleolar targeting for U14 showed that CRM1 and functional nuclear pores were also required for proper transport of uncapped C/D snoRNAs. Capped and uncapped snoRNAs do not transit through the cytoplasm during their biogenesis One explanation of the above results would be that snoRNAs are recognized by CRM1 and exported to the cytoplasm. Given that trimethylation of capped C/D snoRNAs occurs in the nucleus and that CRM1 is associated with TMG-capped mature snoRNAs (Boulon et al, 2004), such a cytoplasmic step would involve almost completely mature species. To test whether snoRNAs were exported during their biogenesis, we performed heterocaryon assays (Fok et al, 2006). HeLa cells were first transfected with a plasmid coding for a tagged RNA and then fused to mouse Balb C cells, whose nuclei can be easily recognized due to their spotted DAPI staining. The tagged RNAs were then detected by fluorescent in situ hybridization (FISH) to see if they had migrated from human to mouse nuclei, which would provide evidence for a cytoplasmic phase during their biogenesis. In order to verify the fusion of the cytoplasms, HeLa cells were also co-transfected with a plasmid coding for GFP–fibrillarin. Cells were further maintained in nocodazole to prevent entry into mitosis and fusion of nuclei (Figure 2A). To first verify that we could visualize shuttling of neo-synthesized RNAs, human cells were transfected with a plasmid coding for a tagged U4 snRNA and then fused with mouse cells. Export of U4tag could be detected by the heterocaryon assay (Figure 2B). Moreover, we found that shuttling increased with time and that LMB abolished it, as expected for an snRNA (Figure 2B; Supplementary Figure S2). We then did similar assays with plasmids coding for tagged U3 or an artificial intronic C/D snoRNA. In both cases, snoRNAs were detected in the nucleoli of human cells but were absent from mouse nuclei despite high expression levels and incubation times of 16 h (Figure 2C; Supplementary Figure S2B). To rule out the possibility that the tag interfered with RNA export, we also performed the experiment with a non-tagged U3 snoRNA. We transfected mouse cells with a plasmid encoding rat U3B.7 and took advantage of a previously developed probe that specifically detects this U3 variant (Verheggen et al, 2002). Whereas U3B.7 was overexpressed when transfected in mouse nuclei, it was not detectable in the recipient nuclei of heterocaryons (Figure 2D). These experiments demonstrate that capped and uncapped C/D snoRNAs do not have a cytoplasmic phase during their biogenesis. This excludes the possibility that CRM1 acts as an export receptor for these RNAs and suggests that it has an alternative role in their biogenesis. Figure 2.C/D and H/ACA snoRNAs do not transit in the cytoplasm during their biogenesis. (A) Description of the heterocaryon assay. Heterocaryons were cultured for 16 h before fixation. (B) A plasmid encoding a tagged snRNA U4 was used as a positive control. By FISH, we stained U4tag in red. GFP–fibrillarin in green labels nucleoli in all nuclei of the heterocaryon. DNA staining in blue allows human and mouse nuclei to be distinguished. U4tag was localized to CB of both types of nuclei. After LMB treatment, U4tag was only detected in the human nuclei that were initially transfected. Mouse recipient nuclei are rimmed. (C) Legend as in (B) but with cells transfected with plasmids coding for U3tag and C/Dtag snoRNAs. RNAs were detected neither in mouse nuclei nor in non-transfected human nuclei of the heterocaryon. (D) Legend as in (B) except a plasmid coding for rU3B.7, a natural U3 variant, was used, and that mouse nuclei were transfected instead of the human nuclei. No rU3B.7 snoRNAs were seen in the untransfected, recipient nuclei of the heterocaryon (rimmed). (E) Legend as in (B) except that the RNA component of telomerase (hTR) and U64, two H/ACA snoRNAs, were transfected and analysed. Similarly to C/D snoRNAs, H/ACA RNAs did not localize to untransfected nuclei of the heterocaryon. The bar represents 5 μm. Download figure Download PowerPoint We then wondered if the absence of a cytoplasmic step during C/D snoRNA biogenesis would also extend to H/ACA snoRNAs and hTR, the human telomerase RNA. hTR resembles a TMG-capped H/ACA scaRNA that localizes to CB and it associates with PHAX (Kiss, 2002; Terns and Terns, 2002; Boulon et al, 2004), while U64 is a canonical intronic H/ACA snoRNA. We transfected HeLa cells with constructs coding for hTR or U64 and GFP–fibrillarin, generated heterocaryons and looked at their localization by FISH, 16 h after fusion. We detected hTR and U64 in human nuclei that had been transfected but not in the recipient nuclei of the heterocaryon (Figure 2E), indicating that similar to C/D snoRNAs, H/ACA snoRNA and hTR were not exported during their biogenesis. Binding of core proteins to C/D snoRNA is not affected by CRM1 inhibition It was previously shown that nucleolar localization of C/D snoRNAs requires their assembly with all four core proteins (Samarsky et al, 1998; Verheggen et al, 2001). A simple explanation for the effect of CRM1 would thus be that it affects the assembly of snoRNAs with core proteins. To test this, we immunoprecipitated (IP'd) nascent snoRNP complexes with antibodies against Nop56, Nop58 and fibrillarin from LMB- or mock-treated cells which had been metabolically labelled with 32P for snoRNA visualization. As shown in Figure 3A, nascent U3 snoRNAs were efficiently IP'd with the three antibodies, whether cells were treated with LMB or not. These results showed that inhibition of CRM1 neither prevent synthesis of U3 snoRNA nor its assembly with C/D core proteins. Figure 3.CRM1 inhibition does not affect C/D snoRNP assembly but affects the localization of the Tgs1 hypermethylase. (A) RNA metabolic labelling with 32P was performed in HeLa cells in the presence or absence of LMB and specific IPs with antibodies against core proteins Nop56, Nop58 and fibrillarin were carried out. U3 snoRNA was resolved on sequencing gels. For each antibody, the amount of pelleted U3 was the same whether cells were treated or not with LMB. (B) Scheme of Tgs1 LF and Tgs1 SF (Mtase, methyltransferase domain). Positions of NES1 and NES2 in Tgs1 LF sequence are indicated. (C) Constructs coding for GFP–Tgs1 LF were transiently transfected into HeLa cells (green). CB were labelled using anti-coilin antibodies (red). In the merged image, DNA is stained with DAPI (blue). Most of GFP–Tgs1 LF is cytoplasmic but a fraction also colocalized with CB in the nucleus (upper panel). After LMB treatment, the cytoplasmic localization is completely lost. (D) Legend as in (C) except that GFP–Tgs1 NES1mut, NES2mut, NES1Δ4 and NES2Δ5 were transfected. GFP–Tgs1 NES1mut and NES1Δ4 showed nuclear and cytoplasmic localization (upper and middle panels) and the faint cytoplasmic signal was lost after LMB treatment (middle panels). GFP–Tgs1 NES2mut showed pronounced nuclear accumulation (second upper panel). GFP–Tgs1 NES2Δ5 exclusively localized to the nucleus in the absence of LMB treatment (lower panel). The bar represents 5 μm. Download figure Download PowerPoint CRM1 controls the subcellular localization of Tgs1, the sn/snoRNA cap hypermethylase As CRM1 is neither involved in export of snoRNAs nor in their assembly with core proteins, it might control the localization or binding of a key biogenesis factor. Indeed, pre-snoRNPs occur as complex assembly intermediates, and a number of factors involved in C/D snoRNA biogenesis have been characterized (for review, see Matera et al (2007)). Some are nuclear (Nufip, BCD1, Nopp140, La and Lsm) while others are both cytoplasmic and nuclear (hSpagh, hPih1, hRvb1, hRvb2 and Tgs1; data not shown). We tested a possible association of these factors with CRM1 in two ways: (i) for the nuclear factors, we determined in heterocaryon assays if they were shuttling in a LMB-dependent manner; (ii) for those that had a cytoplasmic localization, we treated cells with LMB to see if this led to a nuclear relocalization. This was done for all the factors listed above, but only the localization of Tgs1, the enzyme required for m7G-cap hypermethylation of snRNA and snoRNA (Mouaikel et al, 2002; Verheggen et al, 2002), was affected by CRM1 inhibition (Figure 3). Tgs1 exists as two isoforms in HeLa cells: either as a full-length protein called Tgs1 long form (Tgs1 LF) or as shorter form (Tgs1 SF) produced by proteolytic cleavage of its N-terminal domain (Girard et al, 2008). Tgs1 SF contains the catalytic methyltransferase domain (Mtase, Figure 3B) and interacts with snoRNAs through Nop56 and Nop58 core proteins (Mouaikel et al, 2002; Girard et al, 2008). It is a nuclear protein concentrated in CB and it does not shuttle (Supplementary Figure S3). In contrast, Tgs1 LF is mainly cytoplasmic and its N-terminal domain binds SmB (Girard et al, 2008). GFP–Tgs1 LF is however not only observed in the cytoplasm but also in the nucleus where it accumulates in CB (Verheggen et al, 2002). When cells were treated with LMB, the cytoplasmic localization of GFP–Tgs1 LF was lost and the protein relocalized to the nucleoplasm (Figure 3C). Heterocaryon experiments confirmed that GFP–Tgs1 LF shuttles in a CRM1-dependent manner (Supplementary Figure S3). This suggested that the N-terminal domain of Tgs1 LF, which is removed when the protein is processed into Tgs1 SF, might bind CRM1 via a nuclear export signal (NES). An NES in Tgs1 LF is responsible for its shuttling In order to identify potential NES sequences in Tgs1 LF, we used the online NetNES server (CBS, Technical University of Denmark). To confirm the functionality of the predicted NES sequences, we first constructed truncation mutants of Tgs1 LF fused to GFP and studied their localization and shuttling properties (Supplementary Figure S4). These results pointed to a potential involvement of two NES: (i) one from aa 288 to 295 (NES1) and (ii) another from aa 317 to 326 (NES2; Supplementary Table SI). We tested their role in the export of Tgs1 LF by creating point mutants by site-directed mutagenesis. NES1 was mutated by conversion of Val 290 and Leu 292 to two Ala (NES1mut) or by deletion of Leu 292 to F295 (NES1Δ4). GFP–Tgs1 NES1mut and GFP–Tgs1 NES1Δ4 were both cytoplasmic and nuclear and relocalized to nuclei upon LMB treatment (Figure 3D, upper and middle panels), indicating that NES1 is not essential for shuttling. Conversion of Ile 324 and Leu 326 to two Ala in NES2 (GFP–Tgs1 NES2mut) induced a strong accumulation of the protein in the nucleus, but again LMB treatment led to more complete nuclear accumulation of GFP–Tgs1 NES2mut (Figure 3D, upper panels; and data not shown). However, a deletion of NES2 (GFP–Tgs1 NES2Δ5) led to a complete nuclear relocalization of the mutant protein (Figure 3D, lower panels), and in a heterocaryon experiments, GFP–Tgs1 NES2Δ5 remained in the transfected nuclei (Supplementary Figure S3). Thus, NES2 is the key sequence recognized by CRM1 as it is absolutely required for the nuclear export of Tgs1 LF. Tgs1 directly interacts with CRM1 in a Ran-GTP-dependent manner To test whether Tgs1 LF directly interacts with CRM1, we performed in vitro binding assay using recombinant Tgs1 LF and CRM1 in the absence and presence of the Ran Q69L mutant, which remains in the GTP-bound form and increases the affinity of exportins for their substrates (Fornerod et al, 1997). Specific binding of GST–Tgs1 LF to CRM1 was observed in the presence of Ran-GTP but not in its absence (Figure 4A). Moreover, CRM1 did not associate with GST alone or with the NES mutant GST–Tgs1 NES2Δ5. These results showed that CRM1 binds directly to Tgs1 LF, and that this binding depends on NES2. Figure 4.Tgs1 LF binds CRM1 and snoRNP proteins. (A) His–CRM1, His–RanQ69L and GST–Tgs1 LF were produced in bacteria, purified and used for in vitro binding assay. Western blot using anti-CRM1 revealed direct Ran-dependent binding of CRM1 on immobilized GST–Tgs1 LF. CRM1 binding requires the NES2 in Tgs1 LF, as no binding could be detected with or without RanQ69L using GST–Tgs1 NES2Δ5. Membrane stained with Ponceau shows the material loaded. (B) In HeLa whole cell extract (WCE), both forms of Tgs1 are revealed by western blot using a polyclonal antibody against Tgs1. After IP of fibrillarin-containing complexes, a larger amount of Tgs1 SF was pelleted compared with Tgs1 LF. LMB treatment increased the amount of Tgs1 LF that was co-precipitated. Inputs represent 10% of extracts. (C) Immunoblotting of Nop58 after IP of GFP–Tgs1 LF from nuclear extract (NE). More Nop58 was associated with GFP–Tgs1 LF after LMB treatment. Immunoblotting of Tgs1 showed that GFP–Tgs1 LF was present at similar level as endogenous Tgs1 LF (Input), and the same amount of GFP–Tgs1 LF was IP'd in cells treated or not with LMB (Pellet, Ip). Download figure Download PowerPoint Association of Tgs1 LF with snoRNP is regulated by CRM1 To decipher the links between snoRNPs and CRM1, we tested whether the association of Tgs1 with snoRNPs was affected by CRM1 inhibition. To this end, we IP'd C/D snoRNP complexes from HeLa whole cell extracts in both LMB-treated and non-treated cells using a monoclonal antibody against fibrillarin. As shown in Figure 4B, we found that fibrillarin-containing complexes were strongly associated with Tgs1 SF, and only weakly with Tgs1 LF. Remarkably, however, the association of Tgs1 LF was increased in cells treated with LMB. To confirm these data, we generated a stable cell line expressing GFP–Tgs1 LF. We then IP'd GFP–Tgs1 LF-containing complexes from nuclear extracts and analysed the presence of C/D snoRNP core proteins. We showed that Nop58 associated more strongly with GFP–Tgs1 LF in LMB-treated cells, as compared with non-treated cells (Figure 4C). To confirm that such changes were related to the association of Tgs1 LF with C/D snoRNPs and not unassembled core proteins, we directly measured its association with snoRNAs using RNAse protection assays. To this end, cells were transfected with a rat U3B.7 gene and analysed with a probe covering its 3′-end to discriminate the precursor and mature U3 RNAs (Verheggen et al, 2002). We found that LMB slightly decreased association of Tgs1 with rat U3B.7 RNAs, when both Tgs1 LF and Tgs1 SF were IP'd using anti-Tgs1 antibodies that recognize both isoforms (Figure 5A, left panel). To discriminate Tgs1 LF from Tgs1 SF, we next IP'd GFP-containing complexes from stable cell lines expressing GFP–Tgs1 LF or GFP–Tgs1 SF. The amount of rU3B.7 RNA co-precipitated with GFP–Tgs1 SF in LMB-treated cells was decreased as compared with non-treated cells (Figure 5A, middle panel). In contrast, with GFP–Tgs1 LF, a higher amount of rU3B.7 was found in the pellet from LMB-treated cells, as compared with non-treated cells (Figure 5A, right panel). Thus, CRM1 inhibition led to an increased association of U3 snoRNA with Tgs1 LF but not with Tgs1 SF. Figure 5.Association of Tgs1 LF with capped and uncapped snoRNAs is increased upon CRM1 inhibition. (A) RNAse protection assays were performed to study the association of both forms of Tgs1 with U3 snoRNA. Vertical lines indicate the band corresponding to pre-snoRNAs (pre-U3) and mature snoRNA (U3-m). LMB treatment increases the association of U3 snoRNA with GFP–Tgs1 LF (right panel) but not with GFP–Tgs1 SF (middle panel). This increased association upon LMB is not detectable when both Tgs1 LF and Tgs1 SF were IP'd (left panel). Inputs represent 10% of the extracts. (B) By qRT–PCR, four different RNA species were measured after IP of GFP–Tgs1 LF from LMB treated and untreated. Each value represents the fold enrichment as compared with control beads after normalization to GAPDH, and represents the average of three different IP experiments. Error bars represent s.d. (C) RNAse protection assay using a probe specific to an exogenously expressed intronic snoRNA pCMV-C/DdBB. The snoRNA was IP'd with an antibody recognizing both Tgs1 LF and SF (Ip). Input is 10% of the extract. (D) Isogenic HeLa cells stably expressing tagged 3xFlag–Tgs1 SF and LF were generated with the Flp-In system. The tagged proteins were IP'd from total cellular extract, in presence of LMB when indicated, and the RNA content of each pellet was analysed on microarrays. The control was the parental cell line that did not express any tagged protein. The graph represents the frequency with which the indicated families are found in the 500 RNAs most enriched in the IP (P-value that the enrichment is significan
Referência(s)