A UTP3-dependent nucleolar translocation pathway facilitates pre-rRNA 5′ETS processing

Abstract

The ribosome small subunit (SSU) is assembled by the SSU processome which contains approximately 70 non-ribosomal protein factors. Whilst the biochemical mechanisms of the SSU processome in 18S rRNA processing and maturation have been extensively studied, how SSU processome components enter the nucleolus has yet to be systematically investigated. Here, in examining the nucleolar localization of 50 human SSU processome components, we found that UTP3, together with another 24 proteins, enter the nucleolus autonomously. For the remaining 25 proteins we found that UTP3/SAS10 assists the nucleolar localization of five proteins (MPP10, UTP25, EMG1 and the two UTP-B components UTP12 and UTP13), likely through its interaction with nuclear importin α. This ‘ferrying’ function of UTP3 was then confirmed as conserved in the zebrafish. We also found that knockdown of human UTP3 impairs cleavage at the A₀-site while loss-of-function of either utp3/sas10 or utp13/tbl3 in zebrafish causes the accumulation of aberrantly processed 5′ETS products, which highlights the crucial role of UTP3 in mediating 5′ETS processing. Mechanistically, we found that UTP3 facilitates the degradation of processed 5′ETS by recruiting the RNA exosome component EXOSC10 to the nucleolus. These findings lay the groundwork for studying the mechanism of cytoplasm-to-nucleolus trafficking of SSU processome components.

Graphical Abstract

Graphical Abstract.

Introduction

Precursor ribosomal RNA (pre-rRNA) (containing 18S, 5.8S and 28S rRNA in the 5′ to 3′ order) is transcribed by RNA polymerase I (RNAPI) in the nucleolus (1–3). Processing and maturation of 18S rRNA from pre-rRNA is accomplished by the ribosome small subunit (SSU) processome which removes the 5′-external (5′ETS) and internal transcriptional spacers 1 (ITS1) from the pre-rRNA through cleavage at specific sites. The function of the SSU processome is relatively conserved across species ranging from yeast, zebrafish, mice to humans. However, due to differences in the length and sequences of the 5′ETS and ITS1, the exact endonuclease cleavage sites are not precisely identical between different species (4–9).

The SSU processome consists of ∼70 non-ribosomal proteins, along with U3 snoRNA. Extensive studies have established that the UTP-A subcomplex, made up of seven proteins, is the first to be recruited to the 5′ETS region of pre-rRNA, marking the beginning of the assembly of the SSU processome. Next, the UTP-B subcomplex, composed of six proteins and U3 snoRNP (containing U3 snoRNA), are recruited, followed by the addition of the MPP10-IMP3-IMP4 subcomplex. As pre-rRNA transcription progresses, several additional protein complexes, including BMS1-RCL1, NAT10-(AATF-NGDN-NOL10) (ANN) and UTP-C, are successively recruited to the newly produced pre-rRNA. In this process U3 snoRNA binds to the pre-rRNA to constrain its folding, while the SSU processome components bind to multiple sites of the pre-rRNA, ensuring its structure. This then facilitates the precise cleavage of the pre-rRNA by endonucleases such as UTP24 and MRP (4,10–18).

The processed 5′ETS fragment is degraded by the RNA Mtr4-exosome which is recruited by Utp18, a component of the Utp-B subcomplex (18–20). In human cells, the rapid degradation of 5′ETS occurs exclusively in the 3′→5′ direction after A₀ cleavage, involving two nucleases EXOSC10/RRP6 and DIS3/RRP44 associated with the exosome ring. This process is primarily carried out by EXOSC10, with some contribution from DIS3 particularly at the final phase (21).

The SSU processome components, after being synthesized in the cytoplasm, require initial entry into the nucleus prior to being recruited to the nucleolus to exert their specific functions. Differing SSU processome components may adopt different approaches to enter the nucleus. Some proteins enter through a process involving nuclear localization sequence (NLS) (22), whilst others lacking NLS may utilize conformational changes triggered by post-translational modification (PTM) (23) or interaction with another nuclear-localized protein (24). After entering the nucleus, they can proceed to enter the nucleolus autonomously if harboring a nucleolar localization sequence (NoLS), or else be recruited for entry by other nucleolar proteins. The nucleolar protein NPM1 is a good example of this, interacting with certain nuclear transporting factors to facilitate the entry of other interacting proteins into the nucleus or nucleolus (25,26). However, specifically how the SSU processome components are translocated to the nucleolus has been seldom investigated.

Sas10 (Something About Silencing 10), also called Utp3, was first identified in yeast based on a gene screening revealed that, upon its overexpression, it could derepress silenced genes at different loci (including the rDNA locus) (27). Later studies revealed that UTP3 is a conserved protein essential for the maturation of 18S rRNA in different species (28–31). Structural analysis of the human SSU processome revealed that the UTP3 C-terminus is localized on the surface of the SSU processome at the pre-A1 and pre-A1* stage. However, its more flexible N-terminus and mid-located conserved C1D domain were not modeled in such studies. Interestingly, at later stages UTP3 appeared not to be always present in the SSU processome (12,13,32,33), leaving open the possibility that UTP3 might play its primary role in assisting the earlier assembly of the SSU processome. This hypothesis is supported by our previous finding that zebrafish Sas10/Utp3 mediates the nucleolar-localization of Mpp10, a core component of the SSU processome (31). Another recent study has shown that UTP3 activates the expression of VAMP3 by recruiting c-Myc (34). In addition to the nucleolus, Utp3 has been detected in the cytoplasm, nucleoplasm and vesicles (35). These findings suggest possible roles of UTP3 in protein transport and gene expression regulation.

This work explores the role of UTP3 in regulating protein transport, specifically in the context of the SSU processome. To this end, we examined 49 human SSU processome components for their UTP3-dependent nucleolar-localization. We successfully determined that, in addition to MPP10, two UTP-B core components UTP12 (also called WDR3) and UTP13 (also called TBL3) and two other nucleolar proteins UTP25 (also called DEF) and EMG1, exhibit clear UTP3-dependent nucleolar localization. We also show that this ferrying function of UTP3 is conserved in zebrafish. The knockdown of human UTP3 or depletion of either zebrafish Utp3/Sas10 or Utp13/Tbl3 yielded improperly processed pre-rRNA intermediates containing the 5′ETS and impaired the degradation of the processed 5′ETS. Overall, our results reveal that UTP3 is essential for the nucleolar localization of a subset of the SSU processome components, especially components in the UTP-B and MPP10 subcomplexes, to facilitate pre-rRNA processing. Furthermore, we show that UTP3 facilitates the delivery of EXOSC10, a catalytic subunit of the exosome, into the nucleolus to then exert its role in the degradation of the processed 5′ETS products.

Materials and methods

Human cell culture and transient transfection

Human HeLa cells were provided by Dr Jun Huang from Zhejiang University. Human 293T (HEK293T) and MCF-7 cell lines were purchased from the Cell Bank of the Chinese Academy of Sciences in Shanghai, China. All cells were cultured at 37°C with 5% CO₂ in Dulbecco's modified Eagle's medium (DMEM, high glucose; BioChannel, BC-M-005) supplemented with 10% fetal bovine serum (ExCell Bio, FSP025) and 1× penicillin–streptomycin solution (Beyotime, ST488S). Transient transfection of the cultured cells with plasmids was performed using PolyJet transfection reagent (SignaGen Laboratories, SL100688), following the manufacturer's instructions. Briefly, 1 μg of plasmids was mixed and incubated with 3 μl of PolyJet in 100 μl of DMEM and then added to the cultured cells in six-well plates. The transfected cells were harvested for protein extraction or immunofluorescence staining at desired time points as stated in the results.

Construction of expression plasmids

In screening for SSU processome components showing UTP3-dependent nucleolar localization, the coding sequences of 50 SSU processome components and 3 exosome components were successfully obtained through PCR amplification of the cDNA prepared from HeLa or 293T cells using gene specific primers (Supplementary Table S1). The coding sequences of these genes were cloned, into their respective FLAG-tagged pCS2⁺ vectors using ClonExpress® II One Step Cloning Kit (Vazyme, C112). UTP3 was cloned into the HA-tagged pCS2⁺ vector.

For preparing expression plasmids for the protein–protein interaction or co-immunostaining study, derivatives of UTP3, UTP13, UTP25, KPNA3 and EXOSC10 were obtained from the full-length cDNA of corresponding genes through PCR amplification using gene specific primers (Supplementary Table S2). The resultant cDNAs were cloned, into their respective pCS2⁺ vectors harboring an HA- or FLAG-tag as indicated.

Preparation of UTP3-knocdown cells

A CS-RfA-ETBsd doxycycline-inducible (DOX-inducible) knockdown lentivirus vector was used to express short hairpin RNA (36). This lentivirus vector was transfected into 293T cells along with packaging plasmids, and the virus-containing medium was harvested at 24, 48 or 72 h post-transfection. HeLa cells were then infected with the virus media for 72 h and positive cells were selected using 5μg/ml blasticidin. Two inducible shRNAs against UTP3: UTP3 sh-1 (5′-GGCTAAAGTTTCAGTGAAA-3′) and UTP3 sh-2 (5′-GGAGGTGCTAGCCCTAGATAT-3′) were employed in this study.

Immunofluorescence staining of the cultured human cells

Cells were cultured in six-well plates containing pre-placed coverslips. For overexpression experiments, cells were harvested 48 h post-transfection. In the knockdown experiment, cells were collected 4 days after DOX treatment. The collected cells were rinsed twice with PBS and fixed with 4% paraformaldehyde at room temperature (RT) for 15 min. Afterward, the cells were washed three times with PBS and permeabilized with PBS-Triton (0.2% Triton X-100 in PBS) for 30 min at RT. After being blocked in FDB (1× PBS containing 5% goat serum, 2% newborn calf serum, and 5% fetal bovine serum) for 30 min at RT, the coverslips were incubated with the primary antibody for 2 h at RT. The coverslips underwent three 5-min washes with PBS-Triton. Secondary antibodies (diluted 1:400 in FDB) and DAPI (diluted 1:400 in FDB) were then added, and the coverslips were incubated for 1 h at RT. After being washed three times with PBS-Triton, the coverslips were mounted in 80% glycerol for image acquisition. Antibodies against FLAG (CST, 8146) (1:100), HA (Sigma, H6908) (1:100), UTP3 (Abclonal, customer order) (1:100), MPP10 (HUABIO, customer order) (1:100), FBL (rabbit: CST, C13C3; mouse: abcam, ab4566) (1:100), UTP13 (abcam, ab228932) (1:200), UTP25 (HUABIO, customer order) (1:100), NPM1 (Thermo Fisher, FC-61991) (1:100) and EMG1 (Atlas Antibodies, HPA039034) (1:50) were used for immunofluorescence staining. Images were taken under an Olympus FV1000 confocal microscope (Olympus, BX61WI).

Preparation of UTP3 overexpression cells

To establish a doxycycline-inducible stable cell line expressing the UTP3 coding sequence, the N-terminal HA-tagged UTP3 sequence was cloned into the pRetroX-TetOne-Puro vector. The retrovirus vector was subsequently co-transfected into 293T cells along with two packaging plasmids, VsVg and pMD-MLV, to generate the virus. After harvesting the virus-containing medium, this was transferred to HeLa cells for a 2-day incubation period. Puromycin (2 ug/ml) was then added to select infected HeLa cells for a minimum of 5 days, with doxycycline (1.5ug/ml) subsequently introduced to assess the efficiency of HA-UTP3 overexpression.

Co-immunoprecipitation (Co-IP) and western blot analysis

For Co-IP analysis, cells were harvested 48 h post-transfection with total proteins extracted using IP lysis buffer (Beyotime, P0013) supplemented with protease inhibitor cocktail (cOmplete, Roche, 11836145001). Co-immunoprecipitation (Co-IP) was performed using HA-tagged (Abmart, M20013M) or FLAG-tagged (bimake, B23101) agarose beads.

For the Co-IP experiment coupled with RNase treatment, cells were treated with or without 1.5 μg/ml doxycycline for 48 h to induce the expression of HA-tagged UTP3. Cells were harvested and pre-treated with 100 mM NaCl and then centrifuged at 2000 × g at 4°C for 5 min. Cells were lysed by sonication in a buffer containing 10 mM HEPES/KOH pH 7.6, 10 mM KOAc, 0.5 mM MgOAc, 5 mM DTT. After removing cell debris by centrifugation at 20,000 × g for 15 min, the soluble lysate was incubated with 10ug/ml RNase A (Beyotime, ST576) and anti-HA agarose beads for 5 h at room temperature (37).

Western blot analysis was conducted on both input and immunoprecipitated protein samples using the corresponding antibodies. For detecting target proteins in cultured cells using western blot, total proteins were extracted using the SDS lysis buffer (40 mM Tris–HCl, pH 6.8, 10% glycerol, 3% SDS, 0.05% bromophenol blue, 5% 2-mercaptoethanol). Both western blotting and Co-IP procedures were carried out according to previously described methods (31). Antibodies against FLAG (CST, 8146) (1:1000), HA (Sigma, H6908) (1:1000), UTP3 (Abclonal, customer order) (1:1000), MPP10 (HUABIO, customer order) (1:1000), UTP12 (Atlas Antibodies, HPA027509) (1:1000), UTP13 (abcam, ab228932) (1:1000), UTP25 (HUABIO, customer order) (1:1000), EMG1 (Atlas Antibodies, HPA039034) (1:1000), LAMINB (HUABIO, ET1606-27) (1:1000), alpha-TUBULIN (HUABIO, ER130905) and beta-TUBULIN (Abclonal, ac012) were used for western blot.

Protein expression, purification and GST-pull down

We utilized a bacterial expression system for expressing UTP3, EXOSC10 and UTP13 proteins. For UTP25 we employed the Bac-to-Bac Baculovirus Expression System. The primers utilized are listed in Supplementary Table S2.

To express HIS-tagged UTP25, the coding sequence was cloned into pFastBacHT C vector and the recombinant plasmid was transformed into DH10Bac. Recombinant bacmids were then isolated and transfected into SF9 cells to produce baculovirus. SF9 cells were infected with baculovirus for 48 h to express HIS-UTP25. Cells were subsequently harvested, sonicated in binding buffer (25 mM HEPES 7.4, 150 mM NaCl, 3% v/v glycerol, 1 mM DTT) supplemented with 10 mM imidazole and 1 mM PMSF, and the cell debris was removed by high-speed centrifugation. The lysate supernatant was incubated with HIS beads (YEASEN, 20503ES50) at 4°C for 2 h on a turning wheel. After washing the beads with the washing buffer (binding buffer supplemented with 20 mM imidazole), proteins were eluted using elution buffer (binding buffer supplemented with 250 mM imidazole). HIS-UTP25 was further purified using a RESOURCE Q affinity column (cytiva, 17 117 701, Buffer A: 25 mM HEPES 7.4, 150 mM NaCl, 3% v/v glycerol, 1 mM DTT; Buffer B: 25 mM HEPES 7.4, 1 M NaCl, 3% v/v glycerol and 1 mM DTT). Peak fractions were purified using a Superdex 200 Increase 10/300GL column (GE Healthcare, 28 990 944) equilibrated with the binding buffer.

For HIS-tagged UTP3^1–319 and EXOSC10^601-885 expression, the coding sequences were cloned into a pET30ax vector, while GST-tagged UTP3^1–319 and UTP13^1–320 were cloned into a pGEX6P-1 vector. An empty vector expressing GST only served as a negative control. The expression plasmids were transformed into BL21-Star (Weidi, EC1005), and protein expression was induced by adding IPTG to a final concentration of 0.5 mM at 18°C overnight.

For the purification of HIS tagged proteins (HIS-UTP3^1–319 and HIS- EXOSC10^601-885), E. coli pellets were resuspended and sonicated in Buffer H (20 mM Tris 7.0, 500 mM NaCl, 3% v/v glycerol, 1 mM DTT) containing 10 mM imidazole, 1 mM PMSF, 100 μg/ml lysozyme and 10 μg/ml DNase. The lysate was clarified by centrifugation, and the supernatant was incubated with pre-equilibrated HIS beads at 4°C for 2 h. After washing the beads with the washing buffer (Buffer H supplemented with 20 mM imidazole), proteins were eluted using elution buffer (Buffer H supplemented with 250 mM imidazole). For the GST-pull down experiment, the HIS-tagged proteins were dialyzed in binding buffer at 4°C overnight. For the purification of GST, GST-UTP3^1–319 and GST-UTP13^1–320, cells were harvested and sonicated in binding buffer containing 1 mM PMSF, 100 μg/ml lysozyme and 10 μg/ml DNase. The supernatant of the cell lysate was incubated with pre-equilibrated GST beads (GE Healthcare, 17 075 601) at 4°C for 2 h. After washing the beads with binding buffer, GST-tagged proteins were incubated with the purified HIS-tagged proteins at 4°C overnight. Following a washing step with binding buffer, GST beads were eluted using elution buffer (binding buffer supplemented with 50 mM GSH). Samples were analyzed by SDS-PAGE and Coomassie staining.

Bimolecular fluorescence complementation assay

pBiFC-VC155 and pBiFC-VN155(I152L) was a gift from Chang-Deng Hu (Addgene plasmid # 22 011; Addgene plasmid # 27 097). Bimolecular fluorescence complementation (BiFc) was based on protein-protein interaction bringing together the nonfluorescent Venus N-terminal (VN-) and C-terminal (VC-) fragments to reconstitute Venus fluorescent protein. This system is sometimes used for studying the interaction between nuclear transport protein KPNAs and its cargos (38). KPNA3 was constructed into pBiFC-VC155 which was fused to the Venus C-terminal and UTP3 and its derivatives U1 and U2 were constructed into pBiFC-VN155 (I152L) which was fused to the Venus N-terminal. PCR-amplified sequences for the KPNA3 and UTP3 derivatives were inserted into NotI restriction sites of the plasmid vectors pBiFC-VC155 (39) and pBiFC-VN155(I152L) (40) using a ClonExpress^® II One Step Cloning Kit (Vazyme, C112). HeLa cells were cultured and transfected with pBiFC plasmids (VN- and VC-), live cell images were captured 48 h post-transfection by Keyence BZ-800E with immunostaining performed using corresponding tag antibodies.

Zebrafish lines and maintenance

The zebrafish sas10^zju2/zju2 mutant line was identified as previously described (31). The zebrafish AB line was used as the wild-type (WT) for generating the utp13/tbl3 mutant lines in this study. To generate the tbl3 mutant, we designed a gRNA (5′-TGGGATCTGCCCGAGACTACTGG-3′) against exon 14 of the zebrafish tbl3 gene using the software tool in http://chopchop.cbu.uib.no/. The mutant lines were genotyped by comparing the polymerase chain reaction (PCR) product using the tbl3 ID Fw and tbl3 ID Rv primer pair (Fw: CTCTCAGTGGATCGAGCCTTC; Rv: AATGTGTCTCAGGATGAAGCAG). Zebrafish were raised and maintained in accordance with the guidelines provided by http://zfin.org/. All animal procedures were performed in compliance with the Regulation for the Use of Experimental Animals in Zhejiang Province, and approved by the Animal Ethics Committee of the School of Medicine, Zhejiang University (ETHICS CODE Permit NO. ZJU2011-1-11-009Y, issued by the Animal Ethics Committee of the School of Medicine, Zhejiang University).

Immunofluorescence staining on zebrafish cryosections

Cryosections of zebrafish embryos were prepared as previously described (31), followed by permeabilization with PBS-Triton (PBS plus 0.2% Triton X-100) for 30 min. The sections were then washed with PBB (PBS-Triton containing 0.5% bovine serum albumin) and blocked with 20% goat serum in PBB. After washing again with PBB, primary antibodies were diluted in PBB and incubated with the sections overnight at 4°C. Following a further three 10-minute washes with PBB, sections were then incubated with secondary antibodies (1:400) and DAPI (1:500) in PBB for 1 h. After another three washes with PBB, sections were mounted in 80% glycerol and covered with coverslips for image acquisition. Antibodies against Sas10 (rabbit polyclonal antibody, 1:200), Sas10 (mouse monoclonal antibody, 1:200), Tbl3 (rabbit polyclonal antibody, 1:200), and Utp25 (rabbit polyclonal antibody, 1:200) were generated by the HUABIO Company by customer order. FBL (mouse monoclonal antibody, 1:500) was purchased from Abcam. Images were taken under an Olympus confocal microscope (Olympus, BX61WI).

Whole-mount RNA in situ hybridization (WISH)

The fabp10a, fabp2 and trypsin probes (Supplementary Table S3) for WISH experiment were labelled with DIG (Roche Diagnostics, 11 277 073 910). The preparation of these probes and subsequent WISH analysis were conducted as previously described (31,41). Embryos were genotyped individually and the images of these embryos were taken after WISH under a Nikon microscope (Nikon, AZ100).

Northern blot

Total RNA was extracted from various samples using TRIpure Reagent (Aidlab, RN0102), following the manufacturer's instructions. Probes for the Northern blot analysis were designed as previously reported (9,21). Biotin-labelled zP1∼zP5 and hP1∼hP5 probes (Supplementary Table S3) were synthesized by GeneRay Biotech (China). DIG-labelled 5′-ETS-1, 5′-ETS-2, 5′-ETS-3 and ITS1-1 probes were obtained by PCR with specific primers (Supplementary Table S3) using corresponding plasmid DNA as the template together with the DIG DNA Labeling Mix (Roche, 11 277 065 910). Northern blot hybridization was performed using the labelled probes as previously described (9). Ratio analysis of multiple precursors (RAMP) was performed after Northern blot hybridization as previously reported (42).

RNA sequencing (RNA-seq) and data analysis

Total RNA was extracted from WT and sas10^zju2/zju2 mutant embryos at 5dpf using TRIpure Reagent (Aidlab, RN0102), with three independent biological repeats each, and was treated with DNase I (New England Biolabs, M0303S) to remove genomic DNA contamination. Qualified RNA was subjected to the RNA-seq analysis. RNA library construction, high-throughput sequencing and data filtering were accomplished by Annoroad Gene Technology (Beijing) Co., Ltd. Clean reads were mapped to the zebrafish genome (GRCz11) via HISAT2 (version 2.2.1) (43). Transcripts per million (TPM) for each gene was obtained through featureCounts and adopted to quantify the expression level (version 2.0.1) (44). TPM-based quantification of different regions across the pre-rRNA transcript, including 5′ETS, ITS1, 18S, ITS2, 5.8S and 28S, was achieved using Salmon (version 0.12) (45). The DESeq2 R package (version 1.40.2) was used to obtain differentially expressed genes (DEGs) (46). Gene Ontology (GO) analysis was performed using DAVID Bioinformatics resources (version DAVID 2021) (47).

3′RACE

High-throughput analysis of interested RNA's 3′-ends was achieved by 3′RACE-seq as previously described (48), with some modifications. Briefly, 10μg total RNA extracted from 5dpf WT or sas10^zju2/zju2 mutant embryos was used as a starting material. Following 3′RACE_Adapter ligation by T4 RNA Ligase 2 truncated KQ (NEB, M0373) at 4°C overnight, ligated RNA was purified using lithium chloride-ethanol and reverse transcribed with a 3′RACE_RT primer (Supplementary Table S3) using M-MLV Reverse Transcriptase (Invitrogen, 28 025 013).

In order to prepare the libraries for Illumina Novaseq X platform, the cDNA was amplified by KOD DNA polymerase (TOYOBO, KMM-201) using a forward primer (3′RACE_Fw_zf_5′ETS) comprised of the Illumina P5 sequence linked to 20 nucleotides of zebrafish 5′ETS, and a reverse primer (TruSeq RNA PCR index primer (RPI)) comprised of Illumina P7 sequence linked to 33 nucleotides complementary to the 3′ end of the 3′RACE_Adapter (Supplementary Table S3). After amplification, PCR products were purified using VAHTS DNA Clean Beads (Vazyme, N411).

High-throughput sequencing was performed by Annoroad Gene Technology (Beijing) Co., Ltd. Data filtering was achieved through Cutadapt (version 4.7.0) (49). Further detailed analysis was performed as previously described (48).

Real-time quantitative PCR (qPCR)

For qPCR verification of the RNA-seq data, total RNA was treated with DNase I (New England Biolabs, M0303S) prior to reverse transcription. First-strand complementary DNA (cDNA) was synthesized using M-MLV Reverse Transcriptase (Invitrogen, 28 025 013) and with oligo-dT as the primer according to the manufacturer's instruction. The above cDNA was used as the template for qPCR that was performed with AceQq PCR SYBR Green Master Mix (Vazyme, Q111-02) following the manufacturer's instructions. gapdh, rpl32 or actb2 were used as the reference genes, and we considered the data being a credible data only when any one of these three genes differed by less than 0.5 cycles between the WT and mutant in our experiments. Statistical data were based on triplicate reactions. The gene specific primer pairs used for the above qPCR are listed in Supplementary Table S3.

Protein structure prediction with the AlphaFold tool

As no experimentally determined structures for UTP25 were available, we employed the AlphaFold predicted structure of the human UTP25/DEF protein (accession code: Q68CQ4) obtained from the AlphaFold database (https://alphafold.ebi.ac.uk) (50,51). For the UTP25-UTP3 complex, we assessed the predicted structures of UTP25 (amino acids 157–756) and UTP3 (amino acids 100–200) using AlphaFold-Multimer (alphafold2_multimer_v3) within ColabFold (version 1.3.0) (52–55). In our analysis, default settings were employed, resulting in the generation of five models. Among these, we selected the top-ranked prediction for further investigation. This exhibited a pTM score of 0.89 and an ipTM score of 0.85. These scores offered valuable insights into the quality and reliability of our chosen model (56). All structural visualization and rendering were conducted using PyMOL (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC).

Results

Nucleolar localization of human MPP10 is UTP3/SAS10-dependedent

Our previous study showed that the stability of zebrafish Utp3/Sas10 and Mpp10 is mutually dependent and that Utp3/Sas10 facilitates the nucleolar localization of Mpp10 (31). To determine whether the UTP3/SAS10 plays a similar function in humans, we examined the sub-cellular localization of human MPHOSPH10 (MPP10) when overexpressed alone or co-overexpressed with UTP3, in the cultured HeLa, HepG2, 293T or MCF7 human cells, using immunofluorescence staining. Consistent with our observations for zebrafish Mpp10, overexpressed human MPP10 was mainly localized in the cytoplasm when expressed alone, especially in the HeLa and HepG2 cells, but enriched in the nucleolus in all cell lines tested when co-expressed with UTP3 (Figure 1A and Supplementary Figure S1A–C). This result suggested that the UTP3-dependent nucleolar localization of MPP10 is a conserved feature from zebrafish to humans.

Figure 1.

Nucleolar localization of human MPP10, UTP25/DEF, UTP13/TBL3, UTP12/WDR3 or EMG1 is UTP3/SAS10-dependent. (A) Nucleolar localization of human MPP10 is UTP3-dependent. HeLa cells were transfected with FLAG-tagged MPP10 plasmid alone or together with HA-tagged UTP3 plasmid for 48 h and were then subjected to immunofluorescence staining (IF) using the FLAG, MPP10 or HA antibodies as indicated. Scale bar, 5 μm. (B) Outline of the flowchart of the screening for SSU processome components (SSU_PCs) whose nucleolar localization is UTP3-dependent. (C–G) Showing the IF images of NOP56 representing the group I proteins (C), GFM1 representing the group II proteins (D), FAF1 representing the group III proteins (E), UTP25, UTP12 and EMG1 (group VI) (F), and UTP13 (group V) (G) expressed alone or together with UTP3 in HeLa cells 48 h post-transfection. IF was performed using FLAG, HA or UTP3 antibodies as indicated. DAPI (blue), staining the nuclei.

Screening of 48 human SSU processome components identifies UTP12/WDR3, UTP13/TBL3, UTP25/DEF and EMG1 to require UTP3 for their nucleolar localization

Based on the above observation, we hypothesized that UTP3 might serve as a mediator to facilitate the nucleolar localization of some other SSU processome components in addition to MPP10. The human SSU processome contains over 60 components which are sub-grouped into UTPA, UTPB, U3 snoRNP, MPP10, BMS1 complexes and etc (32) (Supplementary Figure S2). Therefore, in addition to UTP3 and MPP10,we successfully cloned the full-length transcripts encoding 48 of the SSU processome components in the human SSU processome (Supplementary Figure S2). We then examined the UTP3-dependency of nucleolar localization of these 48 proteins (tagged with a FLAG-epitome) based on co-immunofluorescence staining in HeLa cells (Figure 1B). Considering our previous observation that Utp3/Sas10 stabilizes Mpp10 in zebrafish (31), we first examined whether the expression levels of these 49 proteins would be affected by UTP3 co-expression. Western blot analysis showed that, in addition to MPP10, at least 8 other proteins (AATF, NOC4L, NOP14, UTP1, UTP4, UTP14, UTP15, UTP24) displayed protein levels that were obviously elevated by UTP3 co-expression (Supplementary Figure S3A–E). The remaining proteins, with the exception of UTP17 which displayed reduced levels upon UTP3 co-expression, appeared relatively unaffected by UTP3 co-expression (Supplementary Figure S3A-E). Immunofluorescence staining showed that 25 proteins, including UTP3 itself, together with NOP56, DDX10, DDX52 and RRP8, entered the nucleolus independently of the absence or presence of ectopically expressed UTP3 (Figure 1C and Supplementary Figure S4A) and were designated here as Group I. Eight proteins, including GFM1, AATF, DCAF13 and DDX49, entered the nucleus but lacked an obvious nucleolar enrichment regardless of whether they are overexpressed alone or co-expressed with UTP3 (Figure 1D and Supplementary Figure S4B), designated here as Group II. Twelve other proteins, including FAF1, RPH3AL, UTP4 and UTP17, were localized in the cytoplasm or in both cytoplasm and nucleus either overexpressed alone or co-expressed with UTP3 (Figure 1E and Supplementary Figure S4C), designated here as Group III. Nucleolar localization of the above 44 SSU processome components was considered to be UTP3-inpendent. Three proteins, namely UTP25/DEF, UTP12/WDR3 and EMG1, entered the nucleus when over-expressed alone and were enriched in the nucleolus upon co-expression with UTP3 (Figure 1F). These were marked as Group IV. Two proteins, MPP10 and UTP13/TBL3, were localized in the cytoplasm when overexpressed alone but was obviously localized in the nucleolus when co-expressed with UTP3 (Figure 1A and G). These were marked here as Group V. These observations strongly suggest that the nucleolar localization of the five proteins in Group IV and V was UTP3-dependent, for which we noticed that their endogenous expression appeared not to be dramatically affected by UTP3 overexpression (Supplementary Figure S3D and E).

Nucleolar localization of the endogenous human MPP10, UTP13, UTP25 and EMG1 is UTP3-dependent

To confirm whether endogenous MPP10, UTP12, UTP13, UTP25 and EMG1 truly exhibits a UTP3-depepdent nucleolar localization, we knocked down the expression of the endogenous UTP3 in HeLa cells using two doxycycline (DOX) inducible UTP3-specific shRNAs (shUTP3-1 and shUTP3-2) (Supplementary Figure S5A). Western blot analysis showed that the DOX-induced shUTP3-1 and shUTP3-2 effectively knocked down the endogenous UTP3 protein levels when compared with the cells treated with the control shRNA (shLUC) (Supplementary Figure S5B). Western blot also showed that the MPP10 protein levels were obviously reduced whilst the UTP25, UTP13, UTP12 and EMG1 protein levels were not drastically affected in the UTP3-knockdown cells (Supplementary Figure S5C). This result suggested that the stability of endogenous UTP25, UTP13, UTP12 and EMG1 was likely to be independent of UTP3.

We then performed a series of co-immunostaining of proteins in both control and UTP3-konckdown cells. We observed that the immunostaining signal of the endogenous UTP3 was low in some cells but high in other cells even in the same batch of treated cells, suggesting that the UTP3-shRNA knockdown efficiency varied between each individual cell. Taking this as an advantage, in addition to the LUC-shRNA treated control cells, we also adopted the UTP3-high cells in the UTP3-shRNA treated cells as the internal control for comparing the effect of UTP3 on the nucleolar localization of UTP25, UTP13, MPP10 and EMG1. The result showed that the nucleolar localization of FBL and NPM1 remained unaffected in all cells (Figure 2A and B). By contrast, the nucleolar signals of the endogenous MPP10, UTP13, UTP25 and EMG1 were obviously reduced, or even absent, in the UTP3-low cells (Figure 2C-F). These results strongly support the understanding that the nucleolar localization of UTP13, UTP25, MPP10 and EMG1 is indeed UTP3-dependent. Due to the lack of a suitable antibody for immunostaining, we were not able to examine this situation for endogenous UTP12.

Figure 2.

Nucleolar localization of the endogenous MPP10, UTP13, UTP25 and EMG1 is UTP3-dependent. (A–H) Representative images showing the effect of knockdown of the endogenous UTP3 on nucleolar localization of FBL (A), NPM1 (B), MPP10 (C), UTP13 (D), UTP25 (E), EMG1 (F), UTP18 (G) and BMS1 (H) in HeLa cells. The endogenous UTP3 protein was knocked down using the DOX-inducible UTP3-specific shRNA (shUTP3). The LUC shRNA (shLUC) was used as a negative control. Immunofluorescence staining using antibodies against FBL, NPM1, UTP3, MPP10, UTP13, UTP25 and EMG1 were performed 96 h after DOX treatment. Nuclei (blue) were stained with DAPI. Scale bar, 5μm. Arrow in each panel indicates a representative UTP3-knockdown cell.

The absence of nucleolar signals for UTP13, UTP25, MPP10 and EMG1 in the UTP3-depleted cells could be attributable either to UTP3-dependent nucleolar translocation or to UTP3-dependent nucleolar retention. To investigate this, we examined the subcellular localization of UTP18 (an essential component of the UTP-B complex) and BMS1 (a core component of the BMS1-RCL1 subcomplex). We found that the nucleolar localization signals of both UTP18 and BMS1 were not obviously affected in the UTP3-depleted cells (Figure 2G and H). These observations indicated that the integrity of the SSU processome was likely to be largely retained following UTP3 depletion, thus suggesting that the exclusion of UTP13, UTP25, MPP10 and EMG1 from the nucleolus in the UTP3-depleted cells might be due to a lack of UTP3-dependent nucleolar translocation. However, we cannot conclusively dismiss the possibility of some level of UTP3-dependent nucleolar retention since we do not know whether UTP18 and BMS1 are exclusive to the SSU processome or also serve functions as part of other nucleolar complexes as does FBL (FBL being not only a component of the U3 snoRNP but also of other C/D box snoRNPs (57)).

Human MPP10, UTP12, UTP13, UTP25 and EMG1 are co-precipitated with UTP3

Whether directly or indirectly, the UTP3-dependent nucleolar localization of MPP10, UTP12, UTP13, UTP25 and EMG1 is likely achieved through protein-protein interaction with UTP3. Our previous studies had demonstrated that zebrafish Utp3 interacts with Mpp10 and Utp25/Def (31). Utp3 is also known to interact with the N-terminus of Mpp10 through the Utp3/C1D domain (31). However, whether human UTP3 was able to interact with human MPP10, UTP12, UTP13, UTP25 and EMG1 has not yet been reported. For this, we expressed FLAG- or HA-tagged human UTP3 alone or together with each of these five proteins and then extracted the proteins for co-immunoprecipitation (Co-IP) analysis. MPP10, UTP25 and EMG1 were robustly co-precipitated with UTP3 (Figure 3A and B and Supplementary Figure S6A and B). Interestingly, UTP12 pulled down UTP3 more efficiently as a bait than it did as a prey (Supplementary Figure S6B), probably due either to UTP3 having multiple interacting proteins or that its interaction with UTP3 may be indirect. These results suggest that the interaction of UTP3 with MPP10 and UTP25 is conserved in human cells. Meanwhile, UTP12, UTP13 and EMG1 were identified to be novel UTP3-interacting proteins, acting either directly or indirectly.

Figure 3.

UTP3 complexes with MPP10, UTP25, UTP12, UTP13 and EMG1. (A and B) Western blot showing the Co-IP product pulled down using a FLAG antibody (FLAG fused to UTP3) (A) or an HA antibody (HA fused to UTP3) (B). In (A), MPP10 and UTP25 were fused with an HA tag, and in (B), UTP12, UTP13 and EMG1 were fused with a FLAG tag. TUBULIN was used as loading control. (C) Western blot showing the effect of RNase treatment on the interaction between UTP3 and the endogenous MPP10, UTP25, UTP12, UTP13 or EMG1. Total proteins were extracted from the DOX-inducible HA-UTP3 overexpression cells. UTP3 was detected using an HA antibody, the endogenous MPP10, UTP25, UTP12, UTP13 and EMG1 were detected using their corresponding specific antibodies. Considering the weak non-specific binding of MPP10, UTP12 and UTP13 by the beads, non-specific beads-bound LAMINB in IP products was used as a loading control.

We considered that UTP3-mediated nucleolar localization of these five proteins may begin with interactions in the cytoplasm. To this end, we overexpressed FLAG-UTP3 in 293T cells (using the FLAG-Vector as the negative control) and extracted the cytoplasmic protein for co-immunoprecipitating UTP3 interacting proteins. Mass spectrometry analysis identified 21 out of 60 SSU processome components in the UTP3 Co-IP products (Supplementary Figure S6C and Supplementary Table S4), including MPP10, UTP12, UTP13, UTP25 and EMG1 the five proteins. To validate these results, we generated a stable and DOX-inducible HA-tagged UTP3 overexpression cell line (Supplementary Figure S6D). Cytoplasmic protein samples were extracted from the cells 48 h after DOX treatment and were subjected to Co-IP analysis using an HA antibody. Results showed that the endogenous MPP10, UTP12, UTP13, UTP25 and EMG1, but not the negative control BMS1 (not in the list of proteins identified by mass spectrometry analysis), were each co-precipitated with HA-UTP3 (Supplementary Figure S6E).

As the SSU processome is basically a protein–RNA complex, we needed to exclude the possibility that the complexing of UTP3 with MPP10, UTP12, UTP13, UTP25 and EMG1 resulted from using RNA as a scaffold. For this, we extracted proteins from the DOX-treated or -untreated cells and then treated the cell extract with or without RNase followed by conducting a Co-IP experiment. The result showed that, among these five proteins, the interaction between UTP3 and endogenous MPP10, UTP12 or UTP13 was not obviously affected by the RNase treatment (Figure 3C). Interestingly, the RNase treatment appeared to enhance the interaction between UTP3 and endogenous UTP25, but greatly reduced the interaction between UTP3 with EMG1 (Figure 3C). This suggested that RNA might facilitate the association between UTP3 and EMG1, but that this was not the case for MPP10, UTP12, UTP13 or UTP25. Given that the interaction between EMG1 and UTP3 also takes place in the cytoplasm (Supplementary Figure S6E), it is unclear upon which type of RNA this interaction is dependent.

Since UTP13 represents a group of proteins which rely on UTP3 to enter the nucleolus from the cytoplasm, and UTP25 represents the group of proteins whose nuclear localization is UTP3-independent but nucleolar localization is UTP3-dependent, we chose these two contrasting proteins for the following detailed studies.

The N-terminal 1–319aa of UTP3 directly interacts with UTP25

We first investigated which regions enabled the interaction between human UTP3 and UTP25. Structural predictions using the AlphaFold tool (50,51,58) showed that human UTP25/DEF contains two structured globular domains (domain I: 157–559aa; domain II: 567–756aa) linked by a short hinge region (560–566aa) (Supplementary Figure S7A–C). Based on the predicted structure, we generated plasmids expressing different FLAG-tagged UTP25 derivatives, including D0 (the full-length WT UTP25, 1–756aa), D1 (1–156aa), D2 (1–562aa), D3 (157–756aa) and D4 (563–756aa) (Figure 4A). We co-transfected these FLAG-tagged UTP25-derived plasmids with HA-tagged full length UTP3 into 293T cells and determined the UTP3-interacting regions in UTP25 by co-IP experiments. The Co-IP results showed that UTP3 failed to pull down D1 and D4 but successfully pulled down D2 and D3 (Figure 4B). It appeared that UTP3 pulled down D3 more efficiently than D2 (Figure 4B). From a structural point of view, the main difference between D2 and D3 was that D2 contained an N-terminal disordered region (1–156aa) and domain I while D3 contained both domain I and II together with the hinge region (Supplementary Figure S7A–C). This indicated that the structure formed by the domain I and II might be more favorable for the UTP3 interaction.

Figure 4.

The N-terminus of UTP3 is required to complex with both UTP25 and UTP13. (A) Diagram showing the different UTP25 derivatives (D0 to D4) generated by N- or C-terminal truncation. (B) Western blot of the Co-IP products pulled down using the HA antibody (HA was tagged to UTP3) for analyzing the interaction of the full length UTP3 with different UTP25 derivatives as indicated. UTP25 derivatives were detected using a FLAG antibody. TUBULIN was used as a loading control for the input. (C) Diagram showing the different UTP3 derivatives (U0 to U2) generated by N- or C-terminal truncation. (D) Western blot of the Co-IP products pulled down using the HA antibody (HA was tagged to the UTP3 and its derivatives) for analyzing the interaction of D3 (a UTP25 derivative) with different UTP3 derivatives as indicated. TUBULIN was used as a loading control for the input. Note, the membrane after blotting with the HA antibody was re-used for blotting of TUBULIN, which explains the bands observed on the right side of the Co-IP product because the striping was not complete. (E) Predicted structure of the complex formed by D3 (green, UTP25^F157-K756) and UTP^E100-E200 (cyan) by the AlphaFold-Multimer tool. The N- and C-termini of D3 and UTP^E100-E200 were indicated. (F) Diagram showing the different UTP13 derivatives (T1 to T4) generated by N- or C-terminal truncation. (G and H) Western blot of the Co-IP products pulled down using the HA antibody (HA was tagged to UTP3 and its derivatives) for identifying the interacting domain in UTP13 (G) and UTP3 (H) for their interaction as indicated. UTP13 derivatives were detected using a FLAG antibody. TUBULIN was used as a loading control for the input. Note, the membrane after blotting with the FLAG antibody was re-used for blotting of TUBULIN, which explains the bands observed on the right side of the Co-IP product because the striping was not complete.

In yeast, Utp25 has been reported to interact with the N-terminus of Utp3/Sas10 (1–227aa) (59). To confirm the necessity of the N-terminus of UTP3 for its interaction with UTP25, we generated plasmids expressing different HA-tagged UTP3 derivatives, including U0 (the full-length WT UTP3, 1–479aa), U1 (1–319aa) and U2 (320–479aa) (Figure 4C). We then examined their interactions with the UTP25 derivative D3. Co-IP results showed that U0 and U1 robustly pulled down D3, whilst U2 did so only weakly (Figure 4D). Prediction of the complex structure of D3 (UTP25 derivative) and N-terminal of UTP3 (100–200aa) using the AlphaFold-Multimer tool showed that the ‘groove’ formed by domain I and domain II of UTP25 clamped the α-helix formed by Q151-A173aa of UTP3 (Figure 4E and Supplementary Figure S7A–C). This may explain the observed stronger interaction between D3 and UTP3 than that between D2 and UTP3 (Figure 4B) since D2 contained only domain I and thus lacked such a ‘groove’.

Next, we conducted a GST-pull down experiment to determine whether these two proteins directly bind to each other. We fused full-length UTP25 with 6XHIS tag as this exhibited the highest affinity for UTP3 binding. Based on the peptide mapping result, we opted for the N-terminal derivative of UTP3 (U1, UTP3^1–319) fused with GST to perform the experiment. It became evident from Coomassie brilliant blue staining that HIS-UTP25 was successfully pulled down by GST -UTP3^1–319 (Supplementary Figure S8A). The identities of HIS-UTP25 and GST-UTP3^1–319 were further confirmed by the western blot analysis using the HIS- and UTP3-specific antibodies, respectively (Supplementary Figure S8B). This result suggests that UTP3-meidated nucleolar localization of UTP25 likely occurs through direct interaction.

The N-terminal 1–319aa of UTP3 directly interacts with the N-terminal 1–320aa of UTP13

Co-IP experiments were also performed to identify the domains for the interaction between UTP3 and UTP13. As a member of the UTP-B subcomplex in the SSU processome, UTP13 has been reported to use its C-terminus to interact with UTP12 to form a tetrameric complex (UTP1-UTP21-UTP12-UTP13) in the UTP-B subcomplex (18). We speculated that UTP13 may use its N-terminus to interact with UTP3. To test this hypothesis, we generated plasmids expressing the different UTP13 derivatives (T1 to T4) (Figure 4F). Clearly, the UTP13 derivatives containing the N-terminus (T1 and T3) were successfully pulled down by UTP3 with a similar efficiency (Figure 4G). Both T2 and T4 displayed only a very weak UTP3 signal which might be an outcome of a weak interaction or simply due to a non-specific binding of these peptides to the beads (Figure 4G). The Co-IP experiment also showed that the UTP3 derivative U1 (1–319aa) robustly pulled down the UTP13 derivative T3 (Figure 4H). We then expressed and purified HIS-tagged UTP3^1–319 (U1) and GST-tagged UTP13^1–320 (T3) peptides from E. coli and performed a GST-pull down experiment. The result showed that GST-UTP13^1–320 successfully pulled down HIS-UTP3^1–319 (Supplementary Figure S8C).

The UTP3 nucleolar localization domain is indispensable for translocating UTP25 and UTP13 to the nucleolus

Currently, it is not known which region in UTP3 dictates its nucleolar localization. Therefore, we transfected HeLa cells with UTP3 (U0) and its derivatives U1 and U2 (Figure 4C) and checked their subcellular localization. We found that U1 (1–319aa) was mainly present in the cytoplasm (Figure 5A and B) whilst U2 was localized in the nucleolus in the majority of the U2-positive cells but was occasionally also localized in the nucleoplasm (Figure 5A and B).

Figure 5.

UTP3 nucleolar localization domain is indispensable for translocating UTP25 and UTP13 to the nucleolus. (A and B) Immunostaining of UTP3 derivative U0, U1 and U2 in HeLa cells 48 h after transfecting with corresponding plasmids (A). U0 and U2, but not U1, were able to enter the nucleolus (A). The ratio of the cells exhibiting the defined subcellular localization was shown in (B). (C–F) Co-immunostaining of U0 or U1 with UTP25 (C) or UTP13 (E) in HeLa cells 48 h after transfecting with corresponding plasmids as indicated. The ratios of the cells exhibiting the defined subcellular localization for UTP25 (D) or UTP13 (F) were provided. In (B), (D) and (F), CP, cytoplasm; NP, nucleoplasm; NO, nucleolus. (G) Western blot analysis of the Co-IP product pulled down using the HA antibody for analyzing the interaction between UTP3 and UTP13 in the absence or presence of UTP12. UTP3 was detected by the HA antibody, UTP13 by the FLAG antibody and UTP12 by the UTP12 antibody, respectively.

In the above experiments we have shown that both UTP25 and UTP13 interact with the UTP3 N-terminal region 1–319aa (i.e. U1) (Figure 4). Using co-immunostaining, here we checked whether U1 was able to guide the nucleolar localization of UTP25 and UTP13. The result showed that U1 failed to translocate either UTP25 or UTP13 to the nucleolus (Figure 5C–F). Interestingly, while it was mainly localized in the cytoplasm when overexpressed alone (Figure 5A and B), U1 was localized in the nucleoplasm rather than the cytoplasm when co-expressed together with UTP25 (Figure 5C and D). These observations suggest that the interaction between U1 and UTP25 changed the subcellular localization of U1. However, the UTP3 nucleolar localization domain (contained in U2) is indispensable for the nucleolar localization of UTP25. Surprisingly, co-expressing U1 and UTP13 resulted in the nuclear localization of both, whilst they remained mainly localized in the cytoplasm when expressed alone (Figure 5A, B, E and F). This result suggests that the U1 and UTP13 likely form a complex with other protein(s) which would thereby enable their autonomous entrance to the nucleus. Considering the facts that (i) UTP12 can enter the nucleus autonomously (Figure 1F) and (ii) UTP12 interacts with UTP13 directly (18,19), we speculated that UTP12 might facilitate the nuclear localization of the UTP3 and UTP13 complex. Indeed, the addition of UTP12 greatly increased the level of immunoprecipitated UTP3 by UTP13 (Figure 5G). Such a clear colocalization of UTP3^1–319 with UTP25 or UTP13 in the nucleoplasm also suggest that the failed nucleolar localization of UTP25 and UTP13 in UTP3-depleted cells is likely due to the lack of assistance from UTP3.

The C-terminus of UTP3 interacts with the nuclear transport protein KPNA3

We were intrigued how UTP3 and its C-terminus (UTP3^320–479, i.e. U2) entered the nucleus (Figure 5A). To this end, we reviewed the protein list identified by the mass spectrometry analysis of the UTP3 Co-IP products in the cytoplasm. We found that the list contained some nuclear importins, including KPNAs (Impα) and KPNB1 (Impβ) (Supplementary Table S4). This suggests the potential use of the Impα-Impβ nuclear transporting pathway by UTP3 (60). Impα binds with its cargo via NLS. Interestingly, when using the NLS Mapper (61) for predictive analysis, we found that UTP3 contained multiple predicted Nuclear Localization Sequences (NLSs), all located within its C-terminus (Figure 6A). To explore whether UTP3 enters the nucleus via the Impα-Impβ pathway, we performed a Co-IP experiment to validate the interaction between UTP3 and KPNA3, the most prominent Impα detected by the mass spectrometry analysis. The result clearly showed that the C-terminus of UTP3 (UTP3^320–479), but not the N-terminal UTP3^1–319, interacted with KPNA3 (Figure 6B).

Figure 6.

The C-terminus of UTP3 interacts with KPNA3. (A) Diagram showing predicted Nuclear Localization Sequences (NLSs), determined via the NLS Mapper tool, with a set cut-off score of 5. MP-NLS, predicted monopartite NLS; BP-NLS, predicted bipartite NLS. (B) Western blot of the Co-IP products pulled down using the FLAG antibody (FLAG was tagged to the KPNA3) for analyzing the interaction of KPNA3 with different UTP3 derivatives as indicated. TUBULIN was used as a loading control for the input. Star denotes unspecific bands. (C) Immunostaining of bimolecular fluorescence complementation experiment. Co-immunostaining of KPNA3 with UTP3, UTP3 derivatives or vector in HeLa cells 48 h after transfecting with corresponding plasmids. KPNA3 were detected using an HA antibody, UTP3 derivatives or vector were detected using a MYC antibody. Scale bar, 5 μm. (D) Live cell images of the bimolecular fluorescence complementation system were captured 48 h post-transfection. The plasmids used were indicated in the figure panels. The magnification view of black box is shown on the right. Scale bar, 50 μm.

To further confirm the interaction between UTP3^320–479 and KPNA3, we adopted a bimolecular fluorescence complementation system which yields a strong green color once two molecules interact with each other irreversibly (38–40). UTP3, UTP3^1–319 and UTP3^320–479 were fused with N-terminal Venus (VN-) while KPNA3 was fused with C-terminal Venus (VC-). UTP3 and its derivatives were co-expressed with KPNA3 in HeLa cells. Both live cell images and immunofluorescence examinations showed that neither the negative control nor UTP3^1–319 interacted with KPNA3 since they are located in different subcellular regions (Figure 6C and D). In contrast, the C-terminal UTP3^320–479 displayed an obvious co-localization with KPNA3 in both the nucleus and nucleolus (Figure 6C and D). Strikingly, KPNA3 was co-localized with the full length UTP3 mainly in the nucleolus (Figure 6C), an outcome explained by the nature of irreversible target binding of this assay system. Taken together, this finding implies that UTP3 has the potential to transport a bound protein to the nucleoli, regardless of whether that which is bound is originally a non-nucleolar protein.

The ferrying function of UTP3 is conserved in zebrafish

To find out whether the UTP3-dependent nucleolar localization of UTP25 and UTP13 is conserved in zebrafish, we performed co-immunostaining experiments to examine the nucleolar localization of Utp25 and Utp13 in the liver bud of the 5dpf-old zebrafish utp3 knock-out mutant sas10^zju2/zju2 (31). We first performed western blot analysis and observed that the Utp13 level was greatly, and Utp25 slightly, elevated in sas10^zju2/zju2 (Figure 7A). This differed from that observed in the UTP3 knockdown cells where no obvious change was noted for either protein (Supplementary Figure S5C). The weak band in the sas10^zju2/zju2 lane in the α-Sas10 panel probably resulted from a non-specific reaction, some residual maternal Utp3 protein, or via contamination of a few of WT or heterozygous embryos that had been brought in by a genotyping error. The latter possibility may have occurred because the sas10^zju2/zju2 mutant carries a 2 bp deletion in exon 4 and that genotyping is assisted by cleaving the PCR product using the Acil restriction enzyme (31). Either way, the co-immunostaining results showed that Utp3 was co-localized with the nucleolar marker Fbl in wild type (WT) whilst in the sas10^zju2/zju2 mutant Utp3 remained undetectable in cells and Fbl remained localized in the nucleolus (Figure 7B and C). This observation is consistent with our previous report (31). Immunofluorescence staining also showed Utp25 and Utp13 to be both co-localized with Utp3 in the nucleolus in the WT (Figure 7B). However, the signals of Utp25 in sas10^zju2/zju2 were no longer enriched in the nucleolus whilst being highly so in the nucleus as indicated by co-staining with DAPI (Figure 7C). The signals of Utp13 in sas10^zju2/zju2 were distributed in both cytoplasm and nucleoplasm (Figure 7C). These results suggested that, in the zebrafish, depletion of Utp3 prevented Utp25 and Utp13 from entering the nucleolus. We therefore concluded that, as in human cells, depletion of Utp3 prevented Utp25 and Utp13 from entering the nucleolus in zebrafish.

Figure 7.

The ferry function of UTP3 is conserved in zebrafish. (A) Western blot analysis of Utp3, Utp25 and Utp13 in WT and sas10^zju2/zju2. (B and C) Representative images showing the co-immunostaining of Utp3 (green) with Fbl, Utp25 or Utp13 (red) in the embryonic liver of WT (B) and sas10^zju2/zju2 (C) at 5dpf. DAPI, staining the nuclei. Scale bar, 5μm.

Loss-of-function of utp3 upregulates the expression of a cohort of genes related to protein transport and ribosome biogenesis in zebrafish

In yeast, Upt3 overexpression caused a de-repression of the silenced genes at different loci (27). This suggested that Utp3 might be involved in regulating gene expression. To this end, at 5dpf we extracted total RNA from the WT and sas10^zju2/zju2 embryos, three biological repeats for each, and carried out RNA-seq analysis. Data related to the total number of clean bases, clean Q30 bases rate, genome mapping rates, and principal component analysis (PCA), suggested that the RNA-seq data met the requirement for further data analysis (Supplementary Tables S5 and Supplementary Figure S9A). Transcript per million (TPM) was adopted to represent the expression levels of individual genes in each sample (62,63). Comparing the gene expression between the WT and sas10^zju2/zju2 using DESeq2 identified a total of 8574 differentially expressed genes (DEGs), including 4457 downregulated and 4117 upregulated DEGs in the sas10^zju2/zju2 mutant (criteria adopted: TPM >5 in at least one replicate, fold-change ≥1.5, P-adjust < 0.05) (Supplementary Figure S9B,C and Supplementary Tables S6 and S7). The expression of 11 up-regulated, 8 no-change and 17 down-regulated genes identified by the RNA-seq analysis was confirmed by qPCR analysis (Supplementary Figure S9D).

Considering the phenotype of severe hypoplastic digestive organs observed in the sas10^zju2/zju2 mutant (31), we cross-compared the 4457 downregulated DEGs with the 308 liver-specific (based on liver against intestine) and 385 intestine-specific (based on intestine against liver) genes identified in our previous studies (64). 150 out of 308 liver- and 185 out of 385 intestine-specific genes were identified to be among the downregulated DEGs (Supplementary Table S6), confirming the link to a malfunctioning digestive system. Next, we carried out a GO analysis of the up-regulated DEGs in sas10^zju2/zju2. The result showed that two of the top 10 most significantly affected categories under the GO biological process (GO_BP) term were related to protein transport, including terms ‘protein transport’ and ‘protein import into nucleus’ (together 151 genes). In addition, 4 of the top 10 categories were related to the ribosome biogenesis and function, including terms ‘translation’, ‘rRNA processing’, ‘ribosome biogenesis’, and ‘translational initiation’ (together 200 genes) (Supplementary Figure S9E, BP panel; Supplementary Table S7).

Among the ribosome biogenesis related genes altered in the GO_BP, we found that 64 genes involved in the encoding of SSU processome components, including the UTP-B genes wdr36, utp6, tbl3, pwp2h and utp18, were upregulated in sas10^zju2/zju2 (Supplementary Figure 9F and Supplementary Table S7). Several reports have shown that loss-of-function of the SSU processome components often causes an upregulation of the expression of ribosome biogenesis genes (9,65,66). Our observation suggests that there might be a common mechanism to trigger a feedback upregulation of the ribosome biogenesis genes in these cases. Such a possibility should be investigated in future research.

Loss-of-function of utp3 causes the accumulation of aberrant products containing 5′ETS in zebrafish

We had previously defined the exact cleavage positions for A′, A₀, A₁, E and 2 sites in the zebrafish pre-rRNA (9). To determine exactly which cleavage site(s) was/were affected in the sas10^zju2/zju2, we designed five biotin-labelled probes targeting distinct regions in the zebrafish pre-rRNA for Northern blot analysis (Figure 8A and B) (9). We then used ‘ratio analysis of multiple precursors (RAMP)’ (42) to quantify various intermediates detected from three independent experiments. Through the comparison of sas10^zju2/zju2 versus WT we observed that, two aberrant intermediates, ‘x₁’ and ‘x₂’, occurring in sas10^zju2/zju2were detected only when using the probes zP1 and zP2 derived from 5′ETS, but not the ITS1 probes zP3 and zP4 (Figure 8B and C, probe zP1, zP2). This was consistent with our previous report (9). Furthermore, we noticed that the processed product ‘e’, as detected by the zP1 probe, was obviously decreased in sas10^zju2/zju2 (Figure 8B and C). These observations suggest that the cleavage of the A₀-site is impaired in sas10^zju2/zju2. In zebrafish, cleavage at the 2-site is proposed to be the key step involved in separating 18S rRNA from 28S and 5.8S rRNA (9). Intriguingly, we observed a drastic decrease in the intermediate ‘h’ while the ‘b’ appeared to be normal in the sas10^zju2/zju2when using the ITS1 probes zP3 and zP4 (Figure 8B and C, probe zP3, zP4), suggesting that the depletion of Utp3/Sas10 does not affect cleavage at the 2-site, but does impair cleavage at the E-site. Since the intermediate ‘h’ is likely yielded from cleavage at the E-site and contains the 5.8S and 28S rRNA, here we propose that the ‘h’ product represents the minor pathway for the pre-rRNA processing in zebrafish. Impaired cleavage at the A₀- and E-sites might explain why the intermediate ‘d’ product was almost undetectable in sas10^zju2/zju2 (Figure 8B and C).

Figure 8.

Loss-of-function of utp3 causes an accumulation of aberrant products containing 5′ETS. (A) Diagram showing the zebrafish rDNA genomic structure and positions of the five probes (zP1 to zP5) for Northern blot. (B) Comparison of the processed rRNA species in WT and sas10^zju2/zju2 by Northern blot using two 5′ETS (zP1, and zP2) probes, two ITS1 (zP3 and zP4) probes and one ITS2 (zP5) probe. The intermediate products of pre-rRNA processing in zebrafish were highlighted according to (A). WT, wild type; x₁, aberrant band 1, x₂, aberrant band 2. Stars denote unspecific bands, with the larger one representing 28S and the smaller one representing 18S, caused by the oversensitivity of biotin probes. (C) Ratio analysis of multiple precursors (RAMP) derived from three independent Northern blot experiments. Different colors of bars represent analyses derived from distinct probes as indicated. (D) Graph showing the statistical analysis of the TPM-based sequence reads derived from the total RNA-seq data for 5′ETS, 18S, ITS1, 5.8S, ITS2, 28S and 3′ETS in WT and sas10^zju2/zju2. (E) Statistical analysis of reads encompassing A′, A₀, and A₁ cleavage sites in WT and sas10^zju2/zju2. (F) 3′RACE analysis to determine the enrichment of 3′-end reads along the pre-rRNA in WT and sas10^zju2/zju2. Left panel: comparison of the 3′-end reads along the A′ (−447nt) to 18S rRNA (1887nt) region between WT (black peak line) and sas10^zju2/zju2 (red peak line). The region after A₀ (including the A₀ to A₁ and the 18S rRNA regions) was magnified to facilitate the visualization and comparison of the distribution of 3′-end reads in WT and sas10^zju2/zju2. The 3′RACE_Fw_zf_5′ETS primer was used in 3′RACE to generate the library for sequencing. Right panel: statistical analysis of the distribution of 3′-end reads encompassed in the regions of before A₀ site, between A₀ and A₁, and after A₁ in the WT and sas10^zju2/zju2.

Figure 9.

Depletion of human UTP3 mainly impairs the 5′ETS processing at the A₀-site. (A) Diagram showing the human rDNA genomic structure and positions of the five probes (hP1 to hP5) for Northern blot. Short-lived precursors are represented with dotted lines. (B) Comparison of the processed rRNA species in shLUC and shUTP3 cells by Northern blot using three 5′ETS (hP1, hP2, hP3) probes, one ITS1 (hP4) probe and one ITS2 (hP5) probe. The intermediate products of pre-rRNA processing in cells were highlighted according to (A). PTP: primary transcript plus, co-migrating 47S and 45S pre-rRNAs; x: aberrant band. (C) Ratio analysis of multiple precursors (RAMP) derived from three independent Northern blot experiments. Different colors of bars represent analyses derived from distinct probes as indicated. (D) Graph showing the statistical analysis of the TPM-based sequence reads derived from the total RNA-seq data for 5′ETS, 18S, ITS1, 5.8S, ITS2, 28S and 3′ETS in shLUC and shUTP3 cells. (E) Statistical analysis of reads encompassing A′, A₀ and A₁ cleavage sites in shLUC and shUTP3 cells.

To further substantiate the accumulation of the processed 5′ETS, considering a lack of polyA tail for rRNA, we conducted ‘total RNA-seq’ (a methodology not reported previously). The primary distinction between total RNA-seq and mRNA-seq lies in the absence of the ‘enriching mRNA’ step in the former. The subsequent procedures, including segmentation, cDNA library construction, and high-throughput sequencing, then remain identical to those in normal RNA-seq. The quality of the total RNA-seq data for WT and sas10^zju2/zju2 embryos at 5 dpf (three repeats for each genotype) was satisfactory based on criteria of the total number of clean bases, the Clean Q30 Bases Rate, and pre-rRNA mapping rates (Supplementary Table S5). Next, we adopted TPM to quantify the reads encompassed in 5′ETS, 18S rRNA, ITS1, 5.8S rRNA, ITS2 and 28S rRNA regions. Results revealed the decreased reads of 18S rRNA and increased reads within 5′ETS, 5.8S and ITS2, but no significant change for ITS1 and 28S rRNA (Figure 8D). This remained consistent with our Northern blot analysis (Figure 8B and C). In light of our Northern blot results revealing impaired cleavage at the A₀-, E-site and subsequently A₁-site in sas10^zju2/zju2, we conducted TPM-based quantification of sequence reads mapped to the uncleaved A′-, A₀-, and A₁- sites, respectively, in both WT and sas10^zju2/zju2. In this, each of the RNA-seq reads is usually 150 nt in length. For our analysis, we arbitrarily defined that a qualified sequence read must contain sequence from both sides of a cleavage site with a minimum of 6 nt on one side. The result showed that the reads covering the uncleaved A₀-site (2.14-fold) and A₁-site (2.11-fold) were significantly higher in sas10^zju2/zju2 (Figure 8E). It also showed that there were more sequence reads containing uncleaved A′-site (3.07 folds) in sas10^zju2/zju2 than in the WT (Figure 8E). Due to the lack of a suitable probe before A′-site, we were unable to confirm this result through Northern blot analysis.

The length of 5′ETS in zebrafish is only 631 nt (9), indicating that the aberrant intermediates ‘x₁’ and ‘x₂’ (∼1.5kb and 0.9kb, respectively), as detected by the 5′ETS probes in sas10^zju2/zju2, might contain partial 18S rRNA. To validate our speculation, we performed a 3′RACE coupled with sequencing analysis using a primer just after the A′-site (Figure 8F). Compared with WT, the sas10^zju2/zju2 had significantly fewer reads located within the A′-A₀ region, but had drastically more reads containing different lengths of 18S rRNA sequence (Figure 8F), suggesting that Utp3-depletion mainly impairs the 5′ETS processing that finally leads to abnormal cleavages within the 18S rRNA.

Loss-of-function of the Utp-B component Utp13 causes an accumulation of the aberrant products containing 5′ETS

The Utp-B subcomplex is known to ensure the pairing of 18S pre-rRNA with U3 snoRNA for proper 5′ETS processing and also to mediate the degradation of the processed 5 ′ETS by recruiting the RNA exosome through Mtr4 (19). The depletion of Utp3 impairs the nucleolar localization of Utp12 and Utp13, two proteins in the Utp-B subcomplex in the SSU processome. This suggests that the dysfunction of Utp-B might contribute to the abnormal cleavage in sas10^zju2/zju2.

To establish the relationship between the two aberrant bands detected by the 5′ETS probes in sas10^zju2/zju2 and Utp-B function, we generated two utp13 mutant alleles, one harboring a 58bp insertion and another 19bp deletion, via the CRISPR-Cas9 technique (Supplementary Figure S10A). The two utp13 mutants (designated as tbl3^+58/+58 and tbl3^Δ19/Δ19, respectively), like sas10^zju2/zju2, also displayed a phenotype of severe hypoplastic digestive organs as revealed through the whole-mount in situ hybridization (WISH) using the liver marker fabp10a, pancreatic marker trypsin, and intestinal marker fabp2 (Supplementary Figure S10B). Northern blot analysis clearly identified the two aberrant bands when using DIG-labelled three different 5′ETS probes (5′ETS-1, 5′ETS-2 and 5′ETS-3) but not the ITS1-1 probe (Supplementary Figure S10C). We also conducted a total RNA-seq to quantify the sequence reads encompassed in the 5′ETS, 18S rRNA, ITS1, 5.8S rRNA, ITS2 and 28S rRNA regions and the sequence reads containing uncleaved A′-, A₀- and A₁-site in WT and tbl3^+58/+58. The data of tbl3^+58/+58 versus WT comparison showed a tendency to closely resemble that observed in in sas10^zju2/zju2 (Supplementary Figure S10D and E). Specifically, the TPM of 5′ETS significantly increased, while the 18S was decreased (Supplementary Figure S10D and E), suggesting that Utp3/Sas10 functions, at least partially, via assisting the assembly of the Utp-B subcomplex.

Depletion of human UTP3 mainly impairs the 5′ETS processing at the A₀-site

To determine the effect of the depletion of human UTP3 on 47S pre-rRNA processing, we compared the processed intermediates between shLUC and shUTP3 treated HeLa cells by Northern blot analysis using various probes followed by RAMP analysis (Figure 9A–C). The UTP3-depletion greatly increased the 30S pre-rRNA, as detected by the 5′ETS-derived hP1, hP2, hP3 and ITS1-derived hP4 probes, and decreased the 21S and 18S-E intermediates, as detected by the hP4 probe (Figure 9B and C). This result strongly suggested an impaired cleavage at the A₀-site after the depletion of UTP3. In contrast, no obvious difference was observed for the 32S intermediate when using the ITS2-derived hP5 probe (Figure 9B and C), suggesting that cleavage at the 2-site was fairly normal after the depletion of UTP3. Interestingly, we observed that a weak aberrant band of ∼3.6k nt was detected by the 5′ETS probes hP1, hP2 and hP3, but not the ITS1 probe hP4, suggesting an inefficient degradation of the processed 5′ETS product in UTP3-depleted cells (Figure 8B and C).

The total RNA-seq analysis (three repeats each for shLUC and shUTP3 treatment) revealed that sequence reads encompassing 5′ETS, ITS1 and 5.8S rRNA were significantly increased while the 18S rRNA encompassed reads were significantly decreased in the UTP3-depleted cells 144 h post DOX treatment (Figure 9D). On the other hand, no significant changes were observed for the ITS2, 28S and 3′ETS regions (Figure 9D). Further analysis showed that the reads containing the uncleaved A′, A₀ and A₁ sites all were significantly higher in the UTP3-depleted cells (Figure 9E). These results suggest that, as that observed in the zebrafish sas10^zju2/zju2, the depletion of UTP3 also leads to an impaired A′, A₀ and A₁ cleavage together with inefficient degradation of the processed 5′ETS products in human cells.

UTP3 regulates exosome function by mediating the nucleolar localization of EXOSC10

All the previous results suggest a potential dysfunction of exosomes related to the depletion of UTP3. We checked the RNA-seq data and found 16 out of 17 genes which encode RNA exosome proteins essential for ribosome biogenesis (65) were among the upregulated genes in sas10^zju2/zju2 (Supplementary Figure S11A). We focused on DIS3, EXOSC10 and MTR4, where the first two possess catalytic activity for rRNA degradation, the depletion of which might lead to an inefficient degradation of the processed 5′ETS products, while MTR4 mediates the recruitment of exosome through UTP-B (20,67,68). In addition, Dis3 was previously reported to be a potential interacting candidate of Utp3 in yeast (69). Co-IP assays showed that, when co-expressed in 293T cells, neither FLAG tagged MTR4 nor DIS3 was able to pull down HA-tagged UTP3 (Supplementary Figure S11B). In contrast, HA tagged EXOSC10 successfully co-precipitated FLAG tagged UTP3 when co-expressed in 293T cells (Figure 10A). Further co-immunofluorescence staining analysis showed that the endogenous DIS3 was mainly localized throughout the nucleus while MTR4 was localized mainly in the nucleoplasm rather than nucleolus (Figure 10B). Depleting the endogenous UTP3 by shUTP3 did not obviously alter the subcellular localization of either MTR4 or DIS3 (Figure 10B). In contrast, the endogenous EXOSC10 was mainly co-localized with UTP3 in the nucleolus, but was localized with a reduced signal intensity in the nucleoplasm of UTP3-depleted cells (Figure 10B).

Figure 10.

UTP3 directly interacts with and mediates the nucleolar localization of EXOSC10. (A) Western blot showing the Co-IP product pulled down using an HA antibody (HA fused to EXOSC10). TUBULIN was used as the loading control. (B) Representative images showing the effect of knockdown of the endogenous UTP3 on nucleolar localization of FBL, EXOSC10, MTR4 and DIS3 in HeLa cells. The shLUC was used as a negative control. Scale bar, 5μm. Arrow in each panel indicates a representative UTP3 knockdown cell. (C) Western blot of the Co-IP products pulled down using the FLAG antibody (FLAG was tagged to derivatives of EXOSC10) for identifying the interaction domain between EXOSC10 and UTP3. Co-IP result showed that UTP3 interacted with the C-terminal region of EXOSC10 (E10^601-885), as indicated. (D) GST-pull down experiment showing the direct interaction between UTP3 and EXOSC10. Recombinant GST (black arrow), GST-UTP3 (yellow arrow) and HIS-EXOSC10^601-885 (red arrow) were expressed and purified. Since GST-UTP3 lane displayed a non-specific band of similar size to HIS-EXOSC10^601-885, the flowthrough (FT) fraction after GST resin incubation was run in parallel to show that the EXOSC10^601-885 was retained by the beads when being co-incubated with GST-UTP3 but not with GST.

EXOSC10 has been reported to be specifically localized in nucleoli (70,71) and to act prior to DIS3 in 5′ETS degradation (21). To further validate the interaction between UTP3 and EXOSC10, we conducted a GST-pull down experiment. Since it was difficult to obtain the full-length EXOSC10 (as also noted in previous reports (70,72,73)), we opted for truncated fragments. One previous report had shown that the C1D protein interacts with the N-terminal region of EXOSC10 (74,75). However, our Co-IP result showed that UTP3, containing the C1D domain interacted with the C-terminal region of EXOSC10 (Figure 10C) which is previously known to interact with the exosome core (70,76). Finally, we expressed and purified the C-terminus of EXOSC10 fused with 6XHIS-tag and the GST tagged UTP3. The GST-pull down experiment demonstrated a direct interaction between UTP3 and the C-terminus of EXOSC10 (Figure 10D). These results suggest that a compromised recruitment of EXOSC10 to the SSU processome upon Utp3/UTP3 depletion likely contributed to the inefficient degradation of the processed 5′ETS products.

Discussion

Cytoplasm-to-nucleolus trafficking is a prerequisite for the nucleolar proteins to exert ribosome biogenesis and many other functions in the nucleolus. In this study, we carried out a systematic examination of the subcellular localization of 50 proteins forming the SSU processome. Based on their subcellular distribution patterns these proteins were classified into five groups (group I to V). We performed a search for the NoLS for these 50 SSU processome components using an online tool (http://www.compbio.dundee.ac.uk/www-nod/index.jsp) (22,77,78). The prediction outcome showed that 20 out of the 25 proteins in the group I harbor at least one putative NoLS (Supplementary Table S8), thus supporting our experimental data that these proteins entered the nucleolus autonomously. We speculate that, for the five remaining proteins in the group I which do not have an obvious NoLS, these might harbor other unidentified NoLSs or that their nucleolar localization is facilitated by other proteins. Nevertheless, we believe our experimental data can serve as a new initial reference for future refinement of the NoLS predicting tool.

Our search also showed that only 3 out of the 8 proteins in the group II and 4 out of the 12 proteins in the group III appeared to have one or more putative NoLS (Supplementary Table S8). For the group IV and V proteins, no NoLSs were predicted for UTP13, UTP25 or EMG1 (Supplementary Table S8). Interestingly, MPP10 and UTP12 appeared to harbor two NoLSs each. The 25 proteins in the group II to V, which failed to enter the nucleolus autonomously, likely require the assistance of other proteins to facilitate their entry to the nucleolus. It is envisaged that two types of facilitators are needed. The first one is the machinery which helps these proteins to cross the nuclear membrane barrier. The second is that which guides these proteins to the nucleolus. Our studies reveal that a subgroup among the SSU processome components, namely MPP10, UTP25, UTP13, UTP12 and EMG1, rely on UTP3 for their nucleolar localization. The mass spectrometry, Co-IP and GST-pull down experiments suggested that MPP10, UTP25, UTP13, UTP12 and EMG1 might initially complex with UTP3 individually or in group in the cytoplasm, from there the UTP3 complex is transported, likely via the Impα-Impβ pathway, to the nucleolus. Moreover, RNase pre-treatment followed by Co-IP experiments revealed that, with the exception of EMG1, the interaction between the remaining four proteins and UTP3 was RNA-independent.

Based on the available structural data related to the SSU processome, it appears that UTP3 itself may not be a structural skeleton member of the SSU processome as its presence could not be identified in 3 out of 12 studies on the SSU processome (11–14,16,19,32,33,79–82). Considering our experimental findings, we hypothesize that UTP3 primarily acts as a scaffold for specific SSU processome components, thus facilitating the creation of a subcomplex within the cytoplasm. However, the exact interaction dynamics of UTP3, whether it binds these specific proteins individually or as a collective, remains a matter for ongoing studies. The observed interaction between UTP3 and the transport protein KPNA3 suggests that UTP3 adopts the Impα-Impβ pathway to facilitate the translocation of this subcomplex into the nucleus. Thus, we conceptually propose a ‘ferry function’ to elucidate UTP3’s possible role in enabling the transport of the subcomplex into the nucleus. Our data provides persuasive support for ‘UTP3-dependent translocation’ from several vantage points, including: (i) that UTP3 complexes with the endogenous proteins MPP10, UTP25, UTP13, UTP12 and EMG1 within the cytoplasm and (ii) that UTP3 binds to the transport protein KPNA3 through its C-terminus and to UTP25 and UTP13 through its N-terminus (Intriguingly, here UTP3 also displays the capability to transport non-nucleolar proteins to the nucleoli, once these proteins are bound to it irreversibly). (iii) Notwithstanding that UTP3^1–319 retains its capacity to interact directly with UTP25 and UTP13, the loss of its C-terminus results in a failure to achieve nucleolar localization.

In this process, we cannot dismiss the possible contribution of ‘UTP3-dependent nucleolar retention’ to the phenotype observed in shUTP3 cells and sas10^zju2/zju2 mutants. It remains plausible that UTP3 offers a platform for specific protein interactions within nucleolus. Consequently, the depletion of UTP3 could result in the diffusion or degradation of these proteins, leading to a loss of nucleolar signals.

The conserved function of Utp3 in mediating the nucleolar localization of Mpp10, Utp25 and Utp13 in zebrafish provides us an ideal genetic model to investigate the function of Utp3. The major pre-rRNA processing pathway in zebrafish includes the procession of the full length pre-rRNA at the A′, A₀ and A₁ sites to remove the 5′ETS (9). The facts that (i) only the 5′ETS probe, but not the ITS1 probe, detected the two extra bands ‘x₁’ and ‘x₂’, (ii) total RNA-seq analysis revealed a significant elevation of sequence reads within the 5′ETS region and of sequence reads containing the uncleaved A′, A₀ and A₁ sites and (iii) 3′-RACE analysis identified abnormal cleavage within the 18S rRNA in sas10^zju2/zju2, all strongly suggests that depletion of Utp3 impairs the processing of the 5′ETS and the degradation of the processed products containing 5′ETS. Similarly, Northern blot and total RNA-seq analysis revealed that an impairment in A₀ cleavage led to subsequent impaired processing of 30S pre-rRNA in the UTP3-depeleted human cells. UTP12 and UTP13/TBL3, two targets of the UTP3-mediated nucleolar localization pathway, are components of the UTP-B subcomplex (18,19). It was of great interest to find that the results obtained from the Northern blot and total RNA-seq analysis displayed similar patterns between sas10^zju2/zju2 and tbl3^+58/+58. Based on the above, we conclude that UTP3/Utp3 executes its function during the 5′ETS processing, especially at the A₀ cleavage site, occurring at least partly through facilitating the assembly of the UTP-B subcomplex.

Recent structural studies have revealed that Utp18, a component of the Utp-B subcomplex, helps to recruit the Mtr4-exosome complex, facilitating the degradation of the processed 5′ETS (19,20). We show here that the nucleolar localization of UTP12, UTP13 and EXOSC10 is UTP3-dependent, suggesting that UTP3 is not only involved in facilitating the cleavage at the A₀-site but promoting the degradation of the processed 5′ETS products as well. This hypothesis is supported by the detection of the aberrant products ‘x₁’ and ‘x₂’ in the zebrafish mutants sas10^zju2/zju2 and tbl3^+58/+58 and ‘x’ in the UTP3-depleted human cells by the probes derived from the 5′ETS only. In yeast, the structure of the Rrp6 (hu-EXOSC10)-RNA exosome complex shows that Rrp6 uses its N-terminus to interact with Rrp47 (hu-C1D), while its PMC2NT domain is placed over the Exo9 central channel to facilitate substrate recruitment, and the C-terminal region of Rrp6 associates with the Exo9 core. Considering that UTP3 interacts with the C-terminal region of EXOSC10, we speculate that upon UTP3 delivering EXOSC10 into the nucleolus, the complex would then dissociate to release the C-terminal region of EXOSC10 for its interaction with EXOSC9. Alternatively, UTP3 may help to anchor the EXOSC10-exosome to the precise accession position to degrade the 5′ETS. In conclusion, we have identified the dual function of UTP3 in regulating exosome-mediated 5′ETS degradation.

Although we have established a ferrying function for UTP3 in facilitating 5′ETS processing at the A₀-site and the degradation of the processed 5′ETS products in both human cells and zebrafish, there still many questions remain unresolved. For example, does an impairment in the cleavage at the A₀-site in human cells and zebrafish induce a cleavage at the Q1-site within 18S rRNA to produce the aberrant products as that seen in the yeast (83)? How is UTP3 dissociated from its interacting proteins once they are delivered to the nucleolus? Which protein(s) mediate(s) the nucleolar localization of the remaining 20 proteins in groups II and III? In addition to the Impα-Impβ pathway, do other nuclear transport machineries help the SSU processome components to cross the nuclear membrane barrier prior to entering to the nucleolus? Finally, it is notable that depletion of genes encoding SSU processome components often upregulate a cohort of genes for ribosome biogenesis, suggesting that there might be a unique genetic compensation mechanism behind this process. This would also be of interest as a target for further investigation.

Data availability

All data have been included in the manuscript. The mass spectrometry data has been deposited in the iProX (an integrated proteome resources center in China) at https://www.iprox.cn, with project ID: IPX0009143000. The RNA-seq and 3′RACE-seq data have been deposited in Gene Expression Omnibus (GEO) at https://www.ncbi.nlm.nih.gov/geo, with GEO accessions: GSE244191 (private token: yhwnmgoojdojfol) for RNA-seq data and GSE262698 (private token: gfermgiydbervif) for 3′RACE data. Additionally, the total RNA-seq data are available in the Sequence Read Archive (SRA).

For access to the zebrafish data the URL is as follows: https://dataview.ncbi.nlm.nih.gov/object/PRJNA1092384?reviewer=5s4g65bd949s4m9csanmnfuo8d; For human cell data, the assess URL is: https://dataview.ncbi.nlm.nih.gov/object/PRJNA1092380?reviewer=8c2jveaavb3unp1rk96rciscrg.

Supplementary data

Supplementary Data are available at NAR Online.

Funding

National Natural Science Foundation of China [U21A20198, 32 200 672]; National Key R&D Program of China [2018YFA0800502]. Funding for open access charge: National Natural Science Foundation of China [32200672].

Conflict of interest statement. None declared.