(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



RIOK2 phosphorylation by RSK promotes synthesis of the human small ribosomal subunit

['Emilie L. Cerezo', 'Molecular', 'Cellular', 'Developmental Biology Department', 'Mcd', 'Centre De Biologie Intégrative', 'Cbi', 'Université De Toulouse', 'Cnrs', 'Ups']

Date: 2021-08

Ribosome biogenesis lies at the nexus of various signaling pathways coordinating protein synthesis with cell growth and proliferation. This process is regulated by well-described transcriptional mechanisms, but a growing body of evidence indicates that other levels of regulation exist. Here we show that the Ras/mitogen-activated protein kinase (MAPK) pathway stimulates post-transcriptional stages of human ribosome synthesis. We identify RIOK2, a pre-40S particle assembly factor, as a new target of the MAPK-activated kinase RSK. RIOK2 phosphorylation by RSK stimulates cytoplasmic maturation of late pre-40S particles, which is required for optimal protein synthesis and cell proliferation. RIOK2 phosphorylation facilitates its release from pre-40S particles and its nuclear re-import, prior to completion of small ribosomal subunits. Our results bring a detailed mechanistic link between the Ras/MAPK pathway and the maturation of human pre-40S particles, which opens a hitherto poorly explored area of ribosome biogenesis.

Ribosomes are universal molecular machines composed of ribosomal RNAs (rRNAs) and ribosomal proteins (RPs) responsible for the synthesis of cellular proteins. In eukaryotic cells, ribosome biogenesis relies on the assembly of large precursor particles (pre-ribosomes) containing, in addition to rRNAs and RPs, more than 200 accessory proteins named assembly and maturation factors (AMFs). Ribosome biogenesis is a highly energy demanding process which is tightly regulated to accommodate protein synthesis with cell growth requirements. In humans, the Ras/MAPK regulatory pathway is one of the main systems controlling cell growth and proliferation. Under favorable growth conditions, the Ras/MAPK pathway activates several aspects of ribosome production, such as synthesis of rRNAs and RPs. Our study highlights that the MAPK pathway also directly stimulates the assembly and maturation of pre-ribosomes. We identify the kinase RIOK2, an AMF involved in the synthesis of the small (40S) ribosomal subunit, as a new target of the MAPK-activated kinase RSK. We describe molecular events promoted by the MAPK pathway to finely tune the activity of RIOK2 during pre-40S particle maturation. This work is particularly important because of the current lack of mechanistic connections between growth-promoting signaling pathways and the assembly and maturation of pre-ribosomes in human cells.

Funding: The Henry/Henras group is supported by grants from ANR (to A.H.) and Ligue Contre le Cancer (to Y.R.). The work was also supported in part by the Région Midi-Pyrénées, European funds (Fonds Européens de Développement Régional, FEDER), Toulouse Métropole, and by the French Ministry of Research with the Investissement d’Avenir Infrastructures Nationales en Biologie et Santé program (ProFI, Proteomics French Infrastructure project, ANR-10-INBS-08, to O. B.-S.). This work was also supported by grants from the Canadian Institutes for Health Research (MOP-142374 and PJT-152995 to P.P.R.), and the Natural Sciences and Engineering Research Council of Canada (to P.P.R.). P.P.R. is a scholar of the Fonds de la recherche du Québec - Santé (FRQS). E.C. is supported by a fellowship from the French Ministry of Higher Education and Research. URLs: https://anr.fr/ , https://www.ligue-cancer.net/ , https://cihr-irsc.gc.ca/e/193.html , http://www.frqs.gouv.qc.ca/ The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Here we report that the MAPK pathway directly regulates discrete molecular events during the post-transcriptional steps of ribosome biogenesis, which fills a major gap between currently known functions of MAPK signaling in Pol I transcription and cytoplasmic translation. Our results unravel a link between RSK signaling and the maturation of human pre-40S particles, through the regulation of RIOK2, an atypical protein kinase of the RIO family involved in the synthesis of the small ribosomal subunit [ 44 ].

By co-activating the syntheses of rRNAs, RPs and AMFs, MAPK and mTORC1 pathways participate in stoichiometric expression of ribosome components and ribosome synthesis machinery, and thereby coordinate the initial stages of ribosome production. Evidence in the literature suggests that these two pathways also regulate the post-transcriptional steps of ribosome synthesis. MAPK and mTORC1 pathways seem to promote pre-rRNA processing [ 38 , 39 ], and rapamycin treatment induces the mislocalisation of several AMFs both in yeast and human cells [ 40 – 43 ]. Although these data suggest that pre-ribosome assembly and maturation are also under the control of signaling pathways, no search has been undertaken so far to identify direct targets of the MAPK and mTORC1 signaling pathways among the large number of pre-ribosome AMFs.

Sustained activation of both MAPK and mTORC1 pathways increases global translation to feed the protein needs under conditions of growth and proliferation. This is achieved in part by promoting ribosome biogenesis, through the regulation of multiple stages of the process (a global view of the function of ERK and RSK kinases in this process is shown in S1A Fig ). They stimulate rDNA transcription by RNA polymerases I and III [ 22 ]. ERK1/2 and RSK, as well as mTORC1 and its downstream target S6K, phosphorylate major Pol I (RRN3/TIFI-A and UBF) and Pol III (TFIIIB) transcription factors, and thereby increase synthesis of the 47S pre-rRNA (Pol I) and the 5S rRNA (Pol III) [ 23 – 30 ]. In addition, mTORC1 stimulates transcription of genes encoding RPs and AMFs through the activity of S6Ks [ 31 ]. Both pathways also promote translation of mRNAs encoding RPs and AMFs, thereby increasing the supply in ribosome components and ribosome biogenesis factors [ 32 – 36 ]. Especially, mTORC1 regulates translation of a subset of mRNAs possessing a 5′ terminal oligopyrimidine (TOP) motif, which encode components of the translation apparatus, including RPs [ 37 ].

In yeast, ribosome biogenesis is one of the most energetically demanding cellular activities [ 11 , 12 ]. This process also represents a substantial expenditure of energy in human cells since growing HeLa cells produce around 7500 ribosomal subunits per minute [ 13 ]. This process needs to be tightly and dynamically regulated to accommodate cell growth and proliferation under favorable conditions, and to prevent energy waste under limiting conditions. The signal transduction cascades Ras/mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase (PI3K)/AKT pathways activate ribosome biogenesis in response to external growth factors, mitogens and hormones or changes in intracellular nutrients [ 14 – 17 ]. Upon activation of the Ras/MAPK pathway ( S1B Fig ), the MAPKs ERK1/2 phosphorylate scores of substrates [ 18 ], including members of the RSK (p90 ribosomal S6 kinase) family of protein kinases [ 19 , 20 ], which collectively promote cell growth and proliferation. The PI3K/AKT signaling pathway activates the mechanistic target of rapamycin (mTOR), a conserved Ser/Thr kinase that is found within two different complexes, referred to as mTORC1 and mTORC2. They are distinguished by their accessory proteins, their differential sensitivity to rapamycin inhibitor, and their largely non-overlapping substrates [ 21 ].

Assembly and maturation of pre-ribosomes is promoted by scores of accessory factors associating transiently with the precursor particles and collectively referred to as assembly and maturation factors (AMFs) [ 1 – 4 ]. Some of these factors participate in the structuring of pre-ribosomes via RNA-binding and/or protein-protein interaction domains, whereas others carry different enzymatic activities, such as nucleases, nucleotide modifying enzymes, putative RNA helicases, kinases, ATPases or GTPases. Ribosome AMFs have been rather exhaustively identified in yeast and the function of some of them is quite well characterized [ 5 ]. In human cells, ribosome synthesis appears more complex since large-scale studies revealed that it mobilizes an increased number of AMFs [ 6 – 9 ] and pre-rRNA processing involves a more complex series of events compared to yeast [ 1 , 10 ].

Ribosomes are the universal macromolecular machines responsible for protein synthesis. Eukaryotic ribosomes consist of two subunits (40S and 60S) containing four ribosomal RNAs (rRNAs) and approximately 80 ribosomal proteins (RPs). Ribosome biogenesis begins in the nucleolus with the synthesis by RNA polymerase I (Pol I) of a polycistronic transcript precursor to the 18S, 5.8S and 28S rRNAs and of a precursor to the 5S rRNA by Pol III ( S1A Fig ). The nascent Pol I transcript is co-transcriptionally packaged into a pre-ribosomal particle that undergoes a series of maturation steps comprising folding and nucleolytic processing of the precursor RNA, chemical modifications of selected nucleotides, and incorporation of RPs and the 5S ribonucleoprotein particle (RNP). Early in the pathway, two discrete particles are generated, the pre-40S and pre-60S particles, precursors to the small and large subunits, respectively. These particles undergo independent maturation steps in the nucleolus and nucleoplasm, before being exported through the nuclear pore complexes. Once in the cytoplasm, pre-40S and pre-60S particles undergo final maturation steps before entering the pool of translation-competent subunits [ 1 , 2 ].

We anticipated that defects in RIOK2 dissociation from cytoplasmic pre-40S particles would slow down its recycling to the nucleus. To investigate the dynamics of RIOK2 shuttling, we compared the kinetics of RIOK2 WT or RIOK2 S483A nuclear import upon inhibition of pre-40S particle export using leptomycin B (LMB), an inhibitor of CRM1 exportin. We monitored the progressive transfer of fluorescence from the cytoplasm to the nucleus following LMB treatment as a readout of the rate of dissociation of RIOK2 from cytoplasmic pre-40S particles and re-import in the nucleus. RIOK2 WT or RIOK2 S483A eHAP1 cell lines were treated with LMB and the nucleo-cytoplasmic distribution of RIOK2 was analyzed by IF during a time-course of LMB treatment ( Fig 6C and quantifications in Fig 6D ). In cells expressing RIOK2 WT , the vast majority of cytoplasmic RIOK2 was imported back to the nucleus by 60 min of LMB treatment and the cytoplasmic signal became barely detectable by 90 min ( Fig 6C ). In contrast, RIOK2 S483A remained mostly cytoplasmic after 60 minutes of LMB treatment and a stronger signal remained in the cytoplasm after 90 min compared to RIOK2 WT . Quantification of the nucleo-cytoplasmic ratios of the IF signals confirmed the slower nuclear import rate of RIOK2 S483A compared to RIOK2 WT ( Fig 6D ). We obtained similar results for NOB1, whose nuclear import following LMB treatment occurred more slowly in cells expressing RIOK2 S483A compared to control cells ( S4A Fig and quantifications in S4B Fig ). We concluded that phosphorylation at Ser483 facilitates the release of RIOK2 and NOB1 from cytoplasmic pre-40S particles and allows their recycling into the nucleus.

(A) HEK293 cells were co-transfected with plasmids expressing HA-NOB1 and Flag-RIOK2 WT , Flag-RIOK2 S483A or Flag-RIOK2 S483D . Pre-40S particles were immunopurified via HA-NOB1 and subsequently incubated for 45 or 90 min with a buffer inducing RIOK2 release at 16°C. The presence of RIOK2, LTV1 and RPS7 proteins in supernatants (released proteins) and on beads (pre-40S-bound proteins) were analyzed by WB. Experiments with Flag-RIOK2 S483A and Flag-RIOK2 S483D were performed with different sets of Flag-RIOK2 WT as controls. A representative WB experiment for Flag-RIOK2 WT is shown. (B) Quantification of WB signals from (A) using ImageLab software and expressed as released over bound RIOK2 ratios. Statistically significant differences are indicated by asterisks (***: P≤0.001, **: P≤0.01, Two-way ANOVA test, Bonferroni posttests). (C) RIOK2 WT and RIOK2 S483A eHAP1 cells were treated with Leptomycin B (LMB, 20 nM) for the indicated times. Subcellular localization of RIOK2 was monitored by immunofluorescence microscopy using specific antibodies. Nuclei were visualized by DAPI staining. (D), Quantification of nuclear to cytoplasmic fluorescence ratios at the indicated time points obtained in (C) using ImageJ software (n = 100 cells from different fields), as described in S7 Fig . Statistically significant differences are indicated by asterisks (***: P≤0.001, *: P≤0.05, 2way ANOVA tests, Bonferroni posttests).

To understand the molecular determinants accounting for the higher steady-state association of RIOK2 S483A with pre-40S particles, we performed in vitro RIOK2 dissociation assays from purified pre-40S particles ( Fig 6A and quantifications in Fig 6B ). For this, we purified pre-40S particles using HA-NOB1 as a bait from HEK293 cells also expressing Flag-tagged versions of either RIOK2 WT , RIOK2 S483A or RIOK2 S483D . The presence of the pre-40S particle AMF LTV1 and ribosomal protein RPS7 in the purified material besides HA-NOB1 and Flag-RIOK2 ( Fig 6A ) confirmed efficient purification of pre-40S particles. These pre-40S particles bound to the anti-HA affinity matrix were then incubated in a buffer intended to promote RIOK2 catalytic activity, which is required for its dissociation from pre-40S particles [ 57 , 67 ]. Using this approach, we found that the dissociation kinetics of RIOK2 S483A was significantly less efficient than RIOK2 WT , as only ~30% of the mutant protein was dissociated from pre-40S particles compared to RIOK2 WT following a 90 min incubation ( Fig 6B ). In contrast, the phosphomimetic RIOK2 S483D mutant was found to display dissociation kinetics quite similar to RIOK2 WT , since by 90 min of incubation, the same amounts of RIOK2 S483D and RIOK2 WT were released from pre-40S particles. Importantly, we noticed that the majority of LTV1 and RPS7 remained associated to NOB1-immunoprecipitated pre-40S particles, indicating that dissociation of RIOK2 is not due to artifactual release of pre-40S particles during incubation time. We concluded from these experiments that RIOK2 S483A is more stably associated to pre-40S particles compared to RIOK2 WT or RIOK2 S483D , which suggests that phosphorylation of RIOK2 at Ser483 facilitates its release from cytoplasmic pre-40S particles.

To more directly assess the physical association between RIOK2 and pre-40S particles, we performed immunoprecipitation (IP) experiments. We purified pre-40S particles from HEK293 cells expressing HA-tagged versions of RIOK2 S483A or RIOK2 WT , and quantified the levels of co-purified 18S-E pre-rRNA by Northern blotting. Using this approach, we found that the 18S-E pre-rRNA co-immunoprecipitated with ~2-fold increased efficiency with RIOK2 S483A compared to RIOK2 WT ( Fig 5E and quantifications in Fig 5F ). Importantly, ectopic expression of RIOK2 S483A in these experiments did not change the global level of 18S-E pre-rRNA ( Fig 5E , Cell Lysate), most likely due to the presence of endogenous RIOK2, strengthening the conclusion that 18S-E pre-rRNA is more efficiently co-purified with RIOK2 S483A . We also monitored by western blot the presence in the precipitated particles of endogenous NOB1, another component of cytoplasmic pre-40S particles whose release occurs after RIOK2 dissociation. We found that NOB1 also co-immunoprecipitated with increased efficiency with RIOK2 S483A ( Fig 5E and quantifications in Fig 5F ). These results were confirmed by immunoprecipitating HA-tagged NOB1 from HEK293 cells also expressing Flag-tagged versions of either RIOK2 WT or RIOK2 S483A ( Fig 5G ). We observed that Flag-RIOK2 S483A was more efficiently co-purified with HA-NOB1 compared to Flag-RIOK2 WT . Interestingly, we noticed in this experiment a shift in the electrophoretic mobility of Flag-RIOK2 WT that is lost with Flag-RIOK2 S483A mutant, supporting that RIOK2 is indeed post-translationally modified at Ser483. We conclude that the loss of RSK-mediated phosphorylation at Ser483 increases the steady-state association of RIOK2 with pre-40S particles.

(A) RIOK2 localization was analyzed by IF microscopy using anti-RIOK2 antibodies in RIOK2 WT (Ctl) and RIOK2 S483A eHAP1 cell lines. Nuclei were visualized by DAPI staining. (B) Quantification of fluorescence observed in (A) using ImageJ software, as described in S7 Fig , and expressed as fold change relative to Ctl (n = 100 cells from different fields). Statistically significant differences are indicated by asterisks (***: P<0.0001, One-tailed Mann Whitney test). (C) Nucleo-cytoplasmic fractionation of serum-growing RIOK2 WT (Ctl) and RIOK2 S483A eHAP1 cell lines. RIOK2 levels in the different fractions (total cell extract, cytoplasm, nucleus) were analyzed by WB. Fractionation quality was validated using antibodies detecting tubulin (cytoplasmic protein) and fibrillarin (nuclear protein). (D) WB signals obtained in the cytoplasmic and nuclear fractions from (C) were quantified. The graph represents cytoplasmic/nuclear intensity ratios. Statistically significant differences are indicated by asterisks (*: P≤0.05, One-tailed Mann Whitney test). (E) HEK293 cells were transfected with plasmids expressing HA-tagged RIOK2 WT or RIOK2 S483A , or with empty vector (-). HA-RIOK2 was immunoprecipitated and co-immunoprecipitated proteins and 18S-E pre-rRNA were analyzed by WB and NB, respectively. (F) Quantification of the WB and NB signals obtained in (E) expressed as fold change compared to immunoprecipitated HA-RIOK2 WT . (G) HEK293 cells were co-transfected with plasmids expressing HA-NOB1 and either Flag-RIOK2 WT or Flag-RIOK2 S483A . HA-NOB1 was immunoprecipitated and the co-immunoprecipitated proteins were analyzed by WB.

To elucidate the molecular impact of RSK-dependent RIOK2 phosphorylation during the maturation of pre-40S particles, we first compared the nucleo-cytoplasmic distribution of RIOK2 WT and RIOK2 S483A in eHAP1 cells using immunofluorescence (IF) microscopy ( Fig 5A ). Our results indicated that RIOK2 S483A accumulates in the cytoplasm to a greater extent than RIOK2 WT . Quantification of the nuclear and cytoplasmic signals revealed a significant increase in the cytoplasmic to nuclear localization ratio of RIOK2 S483A compared to RIOK2 WT ( Fig 5B ). This observation was independently confirmed using cell fractionation assays. Protein samples were prepared from either total cell extracts or from the isolated cytoplasmic and nuclear fractions of eHAP1 cells expressing RIOK2 WT and RIOK2 S483A . Western blotting analyses revealed that RIOK2 S483A was found more abundant in the cytoplasmic fraction than RIOK2 WT ( Fig 5C and quantifications in Fig 5D ). Altogether, these results indicate that in RIOK2 S483A -expressing cells, both the mutant RIOK2 protein and pre-40S particles ( Fig 4E and 4F ) accumulate in the cytoplasm.

Rio2/RIOK2 is incorporated into pre-40S particles in the nucleus, participates to their export through direct binding to the CRM1 exportin, and dissociates from cytoplasmic pre-40S particles to get recycled back into the nucleus [ 52 , 57 , 65 ]. Rio2/RIOK2 catalytic activity contributes to its recycling into the nucleus [ 57 , 59 , 66 , 67 ]. Furthermore, a defect in RIOK2 release is correlated with aberrant retention within pre-40S particles of other late AMFs, such as ENP1/Bystin, PNO1/DIM2, LTV1 and NOB1, the endonuclease responsible for conversion of the 18S-E precursor into mature 18S rRNA [ 57 ].

Pre-40S particles containing the 18S-E pre-rRNA are generated in the nucleolus [ 64 ]. They undergo maturation steps in the nucleoplasm before being exported to the cytoplasm, where final maturation events lead to the production of the mature 18S rRNA of the 40S subunit [ 4 ]. To delineate more precisely which stage of pre-40S particle maturation is delayed in RIOK2 S483A cells, we performed fluorescence in situ hybridization (FISH) experiments to detect precursors of the 18S rRNA in situ in fixed cells. We used a probe detecting all pre-rRNAs of the small subunit maturation pathway (ITS1 probe, S1A Fig ) in the nucleoli, nucleoplasm and cytoplasm. FISH signal observed in the cytoplasm corresponds exclusively to the detection of the 18S-E pre-rRNA, the only precursor retaining part of the ITS1 sequence in this compartment. Cells expressing RIOK2 S483A displayed a stronger FISH signal in the cytoplasm compared to control cells ( Fig 4E and quantifications in Fig 4F ), indicating that these cells accumulate the 18S-E precursors, therefore pre-40S particles, in the cytoplasm. Collectively, our results suggest that phosphorylation of RIOK2 at Ser483 facilitates late, cytoplasmic stages of pre-40S particle maturation.

To delineate the molecular mechanism at the origin of this phenotype, we analyzed pre-rRNA processing in RIOK2 WT , RIOK2 S483A and RIOK2 S483D cell lines. Total RNAs were extracted from these cells and rRNA precursor levels were analyzed by Northern blotting ( Fig 4C upper panel and RAMP quantifications in Fig 4D ). Interestingly, we found a significant accumulation of the 18S-E precursor in all cell lines expressing RIOK2 S483A , indicating that the maturation of pre-40S particles is affected by the loss of Ser483 phosphorylation. RIOK2 knockdown in eHAP1 cells resulted in a similar accumulation of 18S-E pre-rRNA ( S3E Fig and RAMP quantifications in S3F Fig ), suggesting that Ser483 contributes to the regulation of RIOK2 function. Importantly, expression of the RIOK2 S483D phospho-mimetic mutant induced distinct processing defects compared to RIOK2 S483A . Instead of a marked increase in the late 18S-E precursor, we observed on the contrary a slight reduction of this precursor levels in cells expressing RIOK2 S483D and an accumulation of the 30S precursor ( Fig 4C upper panel and RAMP quantifications in Fig 4D ), consistent with processing defects at earlier stages. Since RIOK2 is recruited in nuclear pre-ribosomal particles, these data may suggest that expression of a mutant version of RIOK2 mimicking a constitutive phosphorylation at S483 impairs RIOK2 recruitment into pre-ribosomes. In both cases, no significant change in mature 18S rRNA levels was observed in these cell lines (Figs 4C , middle panel, and S3G and quantifications in S3H Fig ), suggesting that the steady state levels of mature 40S ribosomal subunits are not altered. It is worth noting that the S483A and S483D mutations did not decrease the steady-state levels of the mutant RIOK2 proteins ( Fig 4C , lower panel), indicating that the processing defects observed in these cell lines do not result from a shortage in cellular RIOK2 protein.

(A) MTS assays were performed on Control (Ctl) and RIOK2 S483A (1 to 3) eHAP1 cell lines at the indicated time points. ODs at 490 nm were measured using Spectramax. (B) Control (Ctl) and RIOK2 S483A (1 to 3) eHAP1 cell lines were incubated with 1 μM puromycin for the indicated times. Levels of puromycin-labelled peptides were monitored by WB using anti-puromycin antibodies. WB signals were quantified using ImageLab software and expressed as arbitrary units (a.u.). (C), Total cellular RNAs were extracted from control (Ctl), RIOK2 S483A (1 to 3) or RIOK2 S483D eHAP1 cell lines. Accumulation levels of pre-rRNAs and mature rRNAs were analyzed by NB as in Fig 1 . (D) RAMP analyses of pre-rRNA levels obtained in (C) normalized to the 26S signals after quantification using MultiGauge software (Fujifilm). Graphical representations show fold changes compared to the control condition (Ctl). Statistically significant differences are indicated by asterisks (***: P≤0.001, *: P≤0.05, Two-way ANOVA, Bonferroni posttests). (E) FISH experiments performed on RIOK2 WT (Ctl) and RIOK2 S483A eHAP1 cell lines. Pre-rRNAs were detected using a Cy5-labeled 5’-ITS1 probe. Cells were stained with DAPI to visualize nuclei, and images were captured in identical setting conditions. (F) Nucleolar, nuclear and cytoplasmic fluorescence signals were quantified using ImageJ software, as described in the “Materials and Methods” section and S7 Fig . Graph representations show fold change in RIOK2 S483A relative to RIOK2 WT cell line (n = 100 cells from different fields). Statistically significant differences are indicated by asterisks (***: P≤0.001, **: P≤0.01, *: P≤0.05, One-tailed Mann Whitney test).

To assess the functional relevance of RIOK2 phosphorylation at Ser483, we used a CRISPR/Cas9-based knock-in approach to generate human eHAP1 haploid cell lines expressing mutant versions of RIOK2, bearing a substitution of Ser483 for either a non-phosphorylatable alanine (RIOK2 S483A ) or a phosphomimetic aspartic acid (RIOK2 S483D ) [ 62 ]. Notably, we found that RIOK2 S483A mutant cell lines displayed a significantly decreased proliferation rate, as assessed by both MTS assay ( Fig 4A ) and cell counting ( S3A Fig ), indicating that phosphorylation of RIOK2 at Ser483 is required for optimal cell proliferation. This proliferation defect was found to be less pronounced than that resulting from treatment with PD184352, LJH685 or BI-D1870 inhibitors ( S3B Fig ), likely because the latter affect ribosome synthesis at multiple levels (Figs 1 and S1A ) and probably also affect several other MAPK-dependent cellular processes. We further observed that the slower proliferation rate of RIOK2 S483A -expressing mutant cells was not correlated with a significant increase in cell death mechanisms, such as apoptosis ( S3C Fig ). Interestingly, abrogation of RIOK2 phosphorylation at Ser483 significantly impaired global protein synthesis, as measured using the surface sensing of translation (SUnSET) method [ 63 ] (Figs 4B and S3D ). Since RIOK2 functions in the last stages of pre-40S particle maturation, these results suggest that RIOK2 phosphorylation at Ser483 is important for production of translation-efficient ribosomes.

Both yeast Rio2 and human RIOK2 are essential for cell viability [ 52 , 53 ]. In agreement with this, RIOK2 is required for cell proliferation, migration and survival of glioblastoma [ 54 , 55 ]. RIOK2 has been suggested to function in mitotic progression [ 56 ] but its best documented molecular function is linked to the synthesis of the small ribosomal subunit. Depletion of Rio2/RIOK2 prevents processing of the last precursor to the mature 18S rRNA (20S pre-rRNA in yeast or 18S-E pre-rRNA in human) within cytoplasmic pre-40S particles, and therefore inhibits production of the 40S subunit [ 52 , 57 , 58 ]. The precise function of RIOK2 in the maturation of pre-40S particles remains ill-defined. Yeast Rio2 features autophosphorylation and ATPase activities in vitro and has been suggested to function as an ATPase in the maturation process rather than a bona fide kinase [ 59 ]. In contrast, human RIOK2 was recently shown to phosphorylate DIM1 in vitro, which is a component of nuclear pre-40S particles [ 60 ]. Human RIOK2 forms catalytically inactive homodimers in vitro, suggesting that some aspects of RIOK2 regulation in vivo may involve dimerization [ 61 ].

To determine if RIOK2 is a direct RSK substrate, we performed in vitro phosphorylation assays using a GST-tagged C-terminal fragment of RIOK2 spanning residues Asp443 to Glu552 purified from E. coli ( Fig 3G and 3H ). Upon incubation with active, recombinant human RSK1 produced in insect cells and γ[ 32 P]-ATP, this peptide became efficiently phosphorylated ( Fig 3G ). As specificity controls, we showed that GST alone or a mutant RIOK2 peptide in which Ser483 was replaced by a non-phosphorylatable alanine were poorly phosphorylated by RSK. Addition of ATP-competitive RSK inhibitors (BI-D1870 or LJH685) to the in vitro assay compromised [ 32 P] incorporation into the RIOK2 peptide, demonstrating that this event requires RSK catalytic activity ( Fig 3H ). Together, these experiments strongly suggest that RSK directly promotes RIOK2 phosphorylation at Ser483.

The RSK family comprises four closely related Ser/Thr kinases (RSK1-4) expressed from independent genes. Both RSK1 and RSK2 promote cell growth and proliferation, and are the predominant RSK isoforms expressed in HEK293 cells. We next investigated which ones are involved in RIOK2 phosphorylation at Ser483. HEK293 cells over-expressing each of the four RSK isoforms were stimulated with PMA and RIOK2 phosphorylation at Ser483 was analyzed by immunoblotting ( Fig 3C and quantifications in Fig 3D ). Overexpression of RSK1, and RSK2 to a lesser extent, increased RIOK2 phosphorylation at Ser483 upon PMA treatment. Signal quantifications and associated statistical tests indicated that overexpression of RSK3 or RSK4 did not significantly contribute to RIOK2 phosphorylation. These data suggest that RSK1 and RSK2 are the predominant isoforms involved in the regulation of RIOK2. Of note, we detected lower levels of RSK3 protein in the soluble extract because the protein is largely insoluble when overexpressed, as previously reported [ 47 ]. The role of RSK1 and RSK2 was further confirmed through knockdown experiments using stable shRNAs. Consistent with a dual role for RSK1 and RSK2, we found that knockdown of both isoforms resulted in a reduction of RIOK2 Ser483 phosphorylation upon activation of MAPK signaling, compared to control cells ( Fig 3E and quantifications in Fig 3F ). We noted that RSK knockdown ( Fig 3E ) or inhibition ( Fig 3A ) slightly increased ERK phosphorylation levels, most likely due to a previously described negative feedback loop in the MAPK pathway, whereby RSK activation partially inhibits ERK phosphorylation [ 48 – 51 ].

(A) Serum-starved (Ctl) HEK293 cells were treated or not (-) with PD184352 (PD), or LJH685 (10 μM) (LJH) for 1 h prior to PMA stimulation (100 ng/ml, 20 min). Phosphorylation of endogenous RIOK2 at Ser483 ((P)-RIOK2) was monitored by WB using specific antibodies. (B) (P)-RIOK2 and total RIOK2 signals obtained in (A) were quantified using ImageLab software and expressed as fold change relative to the PMA condition. Statistically significant differences are indicated by asterisks (*: P≤0.05, One-tailed Mann Whitney test). (C) HEK293 cells were transfected with vectors over-expressing HA-tagged RSK1, RSK2, RSK3 or RSK4, or the empty vector (Ctl). Following serum-starvation and PMA stimulation (100 ng/ml, 20 min), samples were analyzed by WB using the indicated antibodies. (D) (P)-RIOK2 and total RIOK2 signals obtained in (C) were quantified using ImageLab software and expressed as fold change relative to starved Ctl cells. Statistically significant differences between starved conditions relative to starved Ctl cells are indicated by hashes, and between PMA stimulated conditions relative to stimulated Ctl cells by asterisks (#/*: P≤0.05, ns: not statistically significant, One-tailed Mann Whitney test). (E) HEK293 cells expressing shRNA targeting an irrelevant sequence (Ctl) or both RSK1 and RSK2 (RSK1/2) were processed as in (C). (F) (P)-RIOK2 and total RIOK2 signals obtained in (E) were quantified using ImageLab software and expressed as fold change relative to the starved condition. Statistically significant differences are indicated by asterisks (*: P≤0.05, One-tailed Mann Whitney test). (G) Human activated RSK1 was incubated in the presence of γ[ 32 P]-ATP with either GST alone, or a GST-RIOK2 peptide (D443-E552) containing either S483 (RIOK2 WT ) or the non-phosphorylatable version (RIOK2 S483A ). The resulting samples were analyzed by SDS-PAGE and revealed by autoradiography or Coomassie blue staining. Quantification of [ 32 P] incorporation within each peptide is expressed as n-fold change compared to the absence of RSK1. (H) In vitro kinase assays performed as in (G) in the presence or not of RSK inhibitors BI-D1870 or LJH685 (10 mM).

To test whether RSK is responsible for RIOK2 phosphorylation at Ser483, we treated serum-starved HEK293 cells with PMA, with or without prior treatment with the MEK1/2 (PD184352) or RSK (LJH685) inhibitors and analyzed RIOK2 phosphorylation at S483 ( Fig 3A and quantifications in Fig 3B ). As LJH685 targets the N-terminal kinase domain of RSK, it does not prevent its phosphorylation at S380 by the C-terminal kinase domain. As in Fig 1 , the efficiency of the inhibitors was attested by the strong decrease in RPS6 phosphorylation at Ser235/236. Consistent with a role of RSK in RIOK2 phosphorylation, we observed that treatment of cells with LJH685 inhibitors significantly reduced Ser483 phosphorylation upon PMA stimulation.

To validate these results, we generated antibodies specifically directed against the Ser483-phosphorylated RIOK2 peptide and monitored the phosphorylation status of endogenous RIOK2 in serum-starved HEK293 cells or in response to agonists of the MAPK pathway ( Fig 2F and quantifications in Fig 2G ). Phosphorylation of RIOK2 at Ser483 increased upon stimulation with serum, EGF or PMA, indicating that activation of MAPK signaling results in phosphorylation of endogenous RIOK2 at Ser483. In addition, RIOK2 phosphorylation levels correlated well with those of ERK and RSK kinases, suggesting a direct regulation of RIOK2 by at least one of these kinases.

To further confirm that the MAPK pathway induces RIOK2 phosphorylation at Ser483, we transiently expressed in HEK293 cells HA-tagged versions of WT or mutant RIOK2 in which Ser483 or Thr481 were substituted for a non-phosphorylatable alanine (RIOK2 S483A , RIOK2 T481A ) ( Fig 2D ). These cells were serum-starved, stimulated with PMA, and HA-RIOK2 was purified from cell lysates and analyzed by immunoblotting. Mutation of Ser483 almost completely prevented RIOK2 phosphorylation after PMA stimulation. Quantifications of phosphorylation signals ( Fig 2E ) showed a residual phosphorylation of RIOK2 S483A that may be due to other basophilic kinases, which phosphorylate Ser/Thr in the vicinity of Arg residues, or compensatory phosphorylation effects at sites not normally recognized by RSK. Since mutation of Thr481 alone did not reduce RIOK2 phosphorylation and since we did not detect Thr481 phosphorylation in vivo, we concluded that activation of MAPK signaling induces RIOK2 phosphorylation specifically at Ser483 and we therefore focused solely on this site in the functional study described below.

We next attempted to identify RIOK2 phosphorylation sites that are regulated by MAPK signaling using a label-free quantitative mass spectrometry (MS) approach. HEK293 cells expressing HA-tagged RIOK2 were serum-starved overnight and pre-treated or not with MEK1/2 inhibitor (PD184352) prior to PMA stimulation ( S2C Fig ). Immunoprecipitated HA-RIOK2 was isolated using SDS-PAGE and digested in-gel with trypsin. Samples were analyzed by nano-liquid chromatography-tandem MS (nanoLC-MS/MS) to detect the presence of putative phosphorylation sites. We evaluated the relative abundance of all phosphopeptides identified in our study, corresponding to 12 phosphorylation sites ( S1 Table ). Among these, Ser483 was the only RIOK2 residue whose phosphorylation robustly increased upon PMA stimulation and returned to basal levels upon MEK1/2 inhibition (Figs 2C and S2D ). Phosphorylation of the other predicted RXRXXpS/T phosphorylation site (Thr481) was not detected, suggesting that this residue is not phosphorylated in HEK293 cells. Altogether, these results suggest that Ser483 is the only residue of RIOK2 whose phosphorylation is under the control of MAPK signaling. Interestingly, the RSK phosphorylation motif containing Ser483 is conserved within vertebrates, suggesting that it is involved in an important biological function ( S2E Fig ).

(A) HEK293 cells were transfected with a plasmid expressing HA-RIOK2 or an empty vector (Ctl). HA-RIOK2 was immunoprecipitated from serum-starved cells treated (+) or not (-) with PD184352 for 1h (10 μM) prior to PMA (100 ng/ml, 20 min) or EGF (25 μg/ml, 10 min) stimulation (+). Samples were analyzed by WB using anti-RXRXXpS/T or anti-HA antibodies. (B) RXRXXpS/T ((P)-RIOK2) and total RIOK2 signals obtained in (A) were quantified using ImageLab software and expressed as fold change relative to the PMA condition. Statistically significant differences are indicated by asterisks (*: P≤0.05, One-tailed Mann Whitney test). (C) HA-RIOK2 was immunoprecipitated from serum-starved (Ctl) HEK293 cells treated (PMA+PD) or not (PMA) with PD184352 (10 μM, 1h) prior to PMA stimulation (100 ng/ml, 20 min). Purified HA-RIOK2 was isolated following SDS-PAGE, in gel digested with trypsin and the resulting peptides were submitted to nano-LC-MS/MS analysis. Label-free quantitative analysis of phosphorylation of the different RIOK2 phospho-peptides was performed as specified in the Materials and Methods section. Data are representative of triple biological replicate experiments for each condition. (D) HEK293 cells expressing HA-tagged versions of WT or mutant versions of RIOK2 (T481A or S483A) were serum starved (-) prior to PMA (100 ng/ml, 20 min) stimulation (+). HA-RIOK2 was immunoprecipitated and samples were analyzed by WB as in (A). (E) RXRXXpS/T ((P)-RIOK2) and total RIOK2 signals obtained in (D) were quantified using ImageLab software and expressed as fold change relative to the WT + PMA condition. Statistically significant differences are indicated by asterisks (*: P≤0.05, One-tailed Mann Whitney test). (F) Serum-starved HEK293 cells were stimulated with different agonists of the MAPK pathway. Phosphorylation of endogenous RIOK2 at Ser483 ((P)-RIOK2) was monitored by WB using specific antibodies. (G) (P)-RIOK2 and total RIOK2 signals obtained in (F) were quantified using ImageLab software and expressed as fold change relative to the starved condition. Statistically significant differences are indicated by asterisks (*: P≤0.05, One-tailed Mann Whitney test).

To determine whether RIOK2 is phosphorylated at RXRXXpS/T motifs, endogenous RIOK2 was immunoprecipitated from serum-growing HEK293 cells and its phosphorylation was analyzed by immunoblotting using antibodies detecting the phosphorylated consensus RXRXXpS/T motif ( S2B Fig ). We observed that RIOK2 was indeed phosphorylated at one or several RXRXXpS/T motif(s) and that the signal disappeared upon treatment of the immunoprecipitate by λ phosphatase, attesting for the presence of a phosphate moiety. To determine whether this phosphorylation event responds to MAPK pathway activation, HEK293 cells transiently expressing HA-tagged RIOK2 were serum-starved and stimulated with EGF or PMA agonists, and phosphorylation of immunoprecipitated HA-RIOK2 was analyzed by immunoblotting ( Fig 2A and quantifications in Fig 2B ). Notably, we found that activation of the MAPK pathway stimulated RIOK2 phosphorylation at the RXRXXpS/T motif. Pre-treatment of starved cells with MEK1/2 inhibitor (PD184352) abrogated the induction of RIOK2 phosphorylation in response to EGF or PMA, indicating that RIOK2 is phosphorylated on RXRXXpS/T consensus sites in a MAPK-dependent manner.

RSK is the most prominent effector kinase that operates downstream of ERK1/2, suggesting that RSK could regulate post-transcriptional steps of ribosome synthesis, possibly by modulating the activity of selected pre-ribosome AMFs. To address this hypothesis, we used the Scansite bioinformatics tool ( https://scansite4.mit.edu/4.0/ ) to search for the canonical Arg/Lys-X-Arg/Lys-X-X-pSer/Thr (RXRXXpS/T) motif found in RSK substrates, within the sequences of an exhaustive list of human AMFs [ 5 , 7 , 8 ]. We found that the 552 amino acid protein kinase RIOK2 features high confidence RXRXXpS/T phosphorylation motifs. Numerous phosphorylated sites have been detected in RIOK2 ( https://www.phosphosite.org/proteinAction.action?id=2360&showAllSites=true ) but the protein only contains two predicted RXRXXpS/T sites in its C-terminal region: a high-stringency site predicting phosphorylation at Ser483 (S483) and a medium-stringency site at Thr481 (T481) ( S2A Fig ).

Northern blotting experiments did not show any drastic change in mature 18S and 28S rRNA steady-state levels in these cell lines ( Fig 1A , mature rRNAs, and RAMP quantifications in S1C Fig ). To get a more dynamic assessment of the role of the MAPK signaling pathway in rRNA production, we performed pulse-chase analyses of rRNA synthesis using [ 32 P]-ortho-phosphate. Treatment of serum-starved HEK293 cells with PMA increased the levels of the 45S pre-rRNA, consistent with an activation of Pol I transcription. All downstream rRNA precursors of both pre-40S and pre-60S maturation pathways also accumulated to higher levels upon PMA treatment, and consequently, production of the mature 18S and 28S rRNAs was remarkably increased ( Fig 1C and 1D ). Inhibition of MEK1/2 induced a marked delay in the production of both the 18S and 28S rRNAs. Importantly, and consistent with the Northern blotting experiments presented in Fig 1A , cells treated with MEK1/2 inhibitors (PD184352) displayed a pronounced accumulation of the early 30S precursor to the 18S rRNA, which was barely detected in PMA-treated cells, indicating that the maturation of this precursor is affected upon inhibition of the MAPK pathway. Altogether, these data show that the MAPK pathway regulates both the synthesis of the 45S pre-rRNA and subsequent maturation events leading to the biogenesis of both ribosomal subunits in human cells.

(A) Serum-starved HEK293, eHAP1 or HeLa cells were stimulated (PMA) or not (Ctl) with PMA (100 ng/ml) for 30 min, then treated (PD) or not (-) with 10 μM of PD184352 for 3h. Total RNAs were extracted and the levels of precursor and mature rRNAs and 7SL1 RNA were monitored by Northern blotting (NB, see S1A Fig for a detailed representation of the human pre-rRNA processing pathway and position of the probes). Total proteins were extracted and analyzed by Western blotting (WB) using the indicated antibodies. (B) RAMP analyses of pre-rRNA levels obtained in (A) normalized to the 7SL1 signals after quantification using MultiGauge software (Fujifilm). Graphical representations show fold changes compared to the starved conditions (Ctl). Statistically significant differences are indicated by asterisks (***: P≤0.001, **: P≤0.01, *: P≤0.05, Two-way ANOVA, Bonferroni posttests). (C) Serum-starved (Ctl) HEK293 cells were treated (PD+PMA) or not (PMA) with MEK 1/2 inhibitor PD184352 prior to PMA stimulation. After a 1 h phosphate deprivation, neo-synthesized RNAs were labeled for 1 h with 32 P-labeled orthophosphate. Cells were harvested 0, 60 and 180 min following addition of cold orthophosphate. Total RNAs were extracted, separated by electrophoresis and transferred to nylon membranes. The labeled (pre-)rRNAs indicated on the left and on the right were detected by autoradiography. (D) Graphical representation of radioactive signals obtained in (C) after quantification using MultiGauge software (Fujifilm).

To assess whether the MAPK signaling pathway regulates pre-rRNA processing in human cells, we first examined the levels of various rRNA precursors to both the small and large ribosomal subunits (maturation pathway depicted in S1A Fig ) upon pharmacological inhibition of ERK1/2 kinases in three different human cell lines (HEK293, eHAP1 and HeLa). Cells were serum-starved to reduce signaling pathways to basal levels and incubated with PMA (phorbol 12-myristate 13-acetate) to stimulate the MAPK pathway prior to treatment with MEK1/2 (PD184352) inhibitor ( S1B Fig ). In this and all subsequent experiments, we have performed western blotting to assess the activation levels of ERK1/2 and RSK kinases ( Fig 1A , lower panels). Efficient activation of the MAPK pathway by PMA, serum or epidermal growth factor (EGF) was shown using phospho-specific antibodies targeted against phosphorylated ERK1/2 (T202/Y204) and RSK (S380). Using these tools, we found that treatment with MEK1/2 inhibitor (PD184352) abrogated ERK1/2 and RSK phosphorylation. The inhibitor efficiency was also attested by the strong decrease in RPS6 phosphorylation at Ser235/236, which is known to be regulated by RSK [ 45 ]. In all cell lines, treatments with MEK1/2 inhibitor induced changes in the processing pathways leading to the production of both the 18S and 28S rRNAs. In the small subunit biogenesis pathway, inhibition of MEK1/2 induced a strong accumulation of the 30S precursor and a marked reduction in the production of all downstream intermediates, in particular the 18S-E pre-rRNA, which is the ultimate precursor to the mature 18S rRNA ( Fig 1A , ITS1 probe). In this and all subsequent Northern blotting experiments, the relative abundance of the pre-rRNA species was quantified using the “Ratio Analysis of Multiple Precursors” (RAMP) procedure [ 46 ] ( Fig 1B ). With regards to the large ribosomal subunit processing pathway, production of the 32S and 12S precursors was affected to some extent by MAPK pathway inhibition ( Fig 1A , ITS2 probe and RAMP quantifications in Fig 1B ). These profiles indicate that inhibition of MAPK signaling induces changes in the steady-state levels of several intermediates in the maturation pathways leading to both ribosomal subunits.

Discussion

The MAPK signaling pathway ensures coordinated expression of ribosome components and of the machinery involved in pre-ribosome assembly and maturation [22]. Upon activation, the MAPK pathway promotes both rDNA transcription by Pol I and Pol III in the nucleus, and translation of mRNAs encoding RPs and AMFs in the cytoplasm. Our study provides evidence that MAPK signaling applies another level of coordination during ribosome biogenesis, by directly regulating pre-40S particle assembly and maturation. In addition, since ERK and RSK inhibition induce processing defects at different stages of the maturation process (for example a clear delay in the maturation of the 30S precursor, Fig 1), both kinases may also regulate several other steps of pre-rRNA processing.

We report direct evidence showing that RSK stimulates the maturation of pre-40S particles, most likely by improving the dissociation efficiency of RIOK2 and of the other AMFs whose release depends on RIOK2 (such as NOB1 as shown in our study). These events are expected to facilitate the subsequent maturation steps towards completion of small subunit biogenesis (Fig 7). We identified RIOK2 as a direct phosphorylation target of RSK and used a combination of approaches to gain in-depth functional understanding of how phosphorylation by RSK influences RIOK2 function.

PPT PowerPoint slide

PowerPoint slide PNG larger image

larger image TIFF original image Download: Fig 7. RIOK2 phosphorylation at Ser483 by RSK facilitates late stages of pre-40S particle maturation. RIOK2 is incorporated into pre-40S particles in the nucleus and participates to their export to the cytoplasm. Phosphorylation of RIOK2 by RSK at Ser483 facilitates its dissociation from pre-40S particles, which allows simultaneous or subsequent dissociation of other factors (ENP1, LTV1, DIM2, NOB1) to promote efficient maturation of the 18S-E pre-rRNA. This phosphorylation event is required for optimal protein synthesis and cell proliferation. https://doi.org/10.1371/journal.pgen.1009583.g007

We demonstrated that RIOK2 phosphorylation at Ser483 stimulates the maturation of the 18S-E pre-rRNA by facilitating RIOK2 release from pre-40S particles and promoting its recycling to the nucleus. It remains unclear whether RSK phosphorylates RIOK2 within pre-40S particles or before its association. A recent cryo-electron microscopy (cryo-EM) structure of a late human pre-40S particle containing RIOK2 has been reported [68]. The domain of RIOK2 containing Ser483 is not resolved in the structure (from V300 to C493), suggesting that this domain is highly flexible. Given that the flanking residues (E299 and S494) are located on the external surface of the pre-40S particle, it seems that this domain is protruding outside of the protein, and would therefore be accessible to RSK kinase for phosphorylation (S5 Fig). These observations suggest that RSK could phosphorylate RIOK2 once incorporated into pre-40S particles to stimulate its dissociation. RSK has not been detected in pre-40S particles purified using different baits [69], suggesting that its interaction with pre-40S particles may be very transient or labile. Consistent with this, sucrose gradient experiments showed that RSK can be detected to low levels in the fractions containing pre-40S particles (S6 Fig).

The molecular mechanisms underlying yeast Rio2 release from pre-40S particles have been recently investigated [67,70]. Rio2 binds to pre-40S particles in a catalytically inactive conformation with its catalytic P-loop lysine (Lys106) bound to the pre-rRNA [59]. Following conformational rearrangements within the pre-40S particles involving the Rps20 connecting loop, release of the P-loop Lys106 allows Rio2 activation and dissociation from the particles. In the human pre-40S particle, RIOK2 seems to be positioned in a similar way compared to the yeast particle, with its P-loop Lys105 in close contact to helix 30 of the 18S rRNA (S5 Fig). However, as the C-terminal extension of RIOK2 containing Ser483 is not present in yeast Rio2, some aspects of the function or regulation of the protein are likely different in yeast and human cells. We propose that phosphorylation of RIOK2 at Ser483 by RSK could participate in conformational rearrangements that trigger RIOK2 catalytic activity by releasing its P-loop lysine and/or weakening its association with pre-40S particles, both favoring RIOK2 dissociation from pre-40S particles. Interestingly, although in yeast Ltv1 has been proposed to dissociate from pre-40S particles concomitantly to Rio2 through the connecting Rps20 loop [70], RIOK2 dissociated independently from LTV1 in our in vitro assay (Fig 6A). This result suggests that additional factors, not present in the pre-40S particles purified in our assay, may coordinate the dissociation of LTV1 and RIOK2. RIOK1 could be a candidate since it has been proposed that RIOK1 position on pre-40S particles overlaps the binding sites of both LTV1 and RIOK2 [68]. RIOK1 could therefore trigger structural rearrangements within pre-40S particle inducing LTV1 and RIOK2 release.

Part of the data supporting our conclusion that RIOK2 phosphorylation at Ser483 is required for optimal maturation of pre-40S particles stems from in vivo experiments where RIOK2 Ser483 was substituted to an alanine using the CRISPR/Cas9 approach. Although these mutant cells display phenotypes consistent with a partial loss of RIOK2 function, we cannot exclude that these effects are not due to the loss of phosphorylation, therefore that what is important for efficient 18S-E pre-rRNA processing is Ser483 per se, and not its phosphorylation status. However, we consider this hypothesis unlikely for several reasons. Substitution of Ser483 to an aspartic acid instead of an alanine (RIOK2S483D mutant) does not induce the same processing defects. In particular, accumulation of 18S-E pre-rRNA is not observed (Fig 4C and 4D), indicating that the serine residue can be substituted by a residue mimicking the phosphorylated state without significant consequences on pre-40S particle maturation. Interestingly, the RIOK2S483D mutant displays a slight accumulation of the 30S intermediate, suggesting that earlier stages of the pre-40S pathway are impaired by this mutation, possibly at the stage of RIOK2 recruitment into pre-ribosomes in the nucleus. Furthermore, in our in vitro dissociation assay, RIOK2S483D behaved like the wild-type protein, indicating again that a negative charge at position 483 is sufficient to confer close to wild-type physical properties to RIOK2. Finally, release of assembly factors by phosphorylation events is not unprecedented. In yeast, Hrr25 kinase phosphorylates Rps3, Enp1 and Ltv1 within nuclear pre-40S particles and thereby weakens their association with the particle to promote conformational rearrangements necessary for formation of the beak structure [70–72]. Hrr25/CK1 also promotes the dissociation of Ltv1 from pre-40S particles both in human and yeast [73,74]. These data combined with our study suggest that phosphorylation events are a common theme in the regulation of AMF dissociation from pre-ribosomal particles.

Our results show that mutation of RSK target site within RIOK2 (RIOK2S483A) induces processing defects that differ from those observed upon MAPK inhibition. RIOK2 is specifically involved in the late, cytoplasmic stages of the maturation of the pre-40S particles containing the 18S-E pre-rRNA, and alterations of RIOK2 function are therefore expected to impair the maturation of this precursor. In contrast, the MAPK pathway regulates ribosome biogenesis at multiple levels (S1A Fig). ERK and/or RSK inactivation is therefore expected to induce pleiotropic defects in ribosome biogenesis. Our data show that ERK inhibition (and consequently RSK inhibition) leads to the accumulation of the 30S intermediate, a phenotype similar to those obtained following depletion of early-associating ribosomal proteins, in particular RPS6 [75]. Since RPS6 is a known target of RSK kinase [45], we hypothesize that phosphorylation of RPS6 by RSK may be important for the incorporation of RPS6 into pre-ribosomes and that this may constitute another important regulation of pre-ribosome assembly by the MAPK pathway. This assumption is supported by recent findings in yeast, where inhibition of RPS6 phosphorylation was shown to prevent its incorporation into pre-ribosomes [76]. The accumulation of the 30S intermediate upon inactivation of ERK and RSK affects production of the downstream intermediates and thus precludes accumulation of 18S-E even if the activity of RIOK2 is affected. It is therefore biologically relevant to observe that inactivation of the MAPK pathway or RIOK2 does not result in the same processing defects, because the former has a broader impact on the maturation pathway than the latter.

Inhibition of RSK-dependent RIOK2 phosphorylation, although it delays 18S-E pre-rRNA processing, does not affect the steady-state amount of the small ribosomal subunit (18S rRNA levels). This possibly means that point mutation of S483 only partially affects RIOK2 function and delays the maturation of pre-40S particles to an extent that is not sufficient to impact mature 18S rRNA levels, as previously observed in other studies [77,78]. However, we do observe a global defect in translation correlated with a decrease in cell proliferation. Several hypotheses can be proposed to account for this apparent paradox. For technical reasons, we may have been unable to detect minor reductions in mature 18S rRNA levels. Another hypothesis could be that compensation mechanisms may operate to maintain mature 18S rRNA levels despite RIOK2 mutation. For example, mutated cells may extend the half-life of the ribosomes produced in these conditions to counterbalance a production defect. These ribosomes may become partially dysfunctional due to increased exposure to reactive oxygen species, leading to stalled ribosomes or decreased translation efficiency [79].

RIOK2 is the first pre-ribosome AMF shown to be regulated by the MAPK pathway. As RSK kinases share several phosphorylation targets with S6K and AKT, which belong to the PI3K/AKT/mTOR signaling pathway, it would be interesting to assess also the contribution of this pathway in the regulation of AMF functions. Since ribosome synthesis involves hundreds of AMFs, potentially many other MAPK- or PI3K/AKT/mTOR-driven regulatory mechanisms operate. Like RIOK2, a significant proportion of other AMFs corresponds to energy-consuming enzymes such as kinases, ATPases, GTPases or RNA helicases [1,3,4]. These enzymes have been suggested to provide directionality, accuracy and quality control to the process. They are believed to provide energy to overcome thermodynamically unfavorable steps of the process such as disruption of stable RNA helices, protein/RNA or protein/protein interactions [80,81]. Our in silico screen identified several other potential substrates of RSK and among these, energy-consuming enzymes are particularly represented. We therefore propose that MAPK signaling may participate in the coordination of the series of events occurring during pre-ribosome assembly and maturation by stimulating the activity of selected energy-consuming enzymes, thereby allowing to overcome rate-limiting steps. Interestingly, mammalian RIOK1 and RIOK3, also involved in pre-40S particle maturation [82,83] are not phosphorylated within a RSK consensus motif suggesting a specific regulation of RIOK2 by RSK.

Our study paves the way for future exploration of the regulation of ribosome biogenesis at the post-transcriptional level by the MAPK pathway. Identification of key limiting steps, in particular those catalyzed by discrete catalytic activities, would help in designing innovative molecules aimed at counteracting MAPK-driven deregulated ribosome production in pathologies, such as cancers or RASopathies.

[END]

[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1009583

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.