Diamond-Blackfan anemia (DBA) is a rare congenital disease affecting erythroid precursor differentiation. DBA is emerging as a paradigm for a new class of pathologies potentially linked to disorders in ribosome biogenesis. Three genes encoding ribosomal proteins have been associated to DBA: after RPS19, mutations in genes RPS24 and RPS17 were recently identified in a fraction of the patients. Here, we show that cells from patients carrying mutations in RPS24 have defective pre-rRNA maturation, as in the case of RPS19 mutations. However, in contrast to RPS19 involvement in the maturation of the internal transcribed spacer 1, RPS24 is required for processing of the 5′ external transcribed spacer. Remarkably, epistasis experiments with small interfering RNAs indicate that the functions of RPS19 and RPS24 in pre-rRNA processing are connected. Resolution of the crystal structure of RPS24e from the archeon Pyroccocus abyssi reveals domains of RPS24 potentially involved in interactions with pre-ribosomes. Based on these data, we discuss the impact of RPS24 mutations and speculate that RPS19 and RPS24 cooperate at a particular stage of ribosome biogenesis connected to a cell cycle checkpoint, thus affecting differentiation of erythroid precursors as well as developmental processes.
Although they play a central role in cell metabolism, ribosomal proteins have been linked to genetic diseases only recently. The RPS19 (ribosomal protein S19) gene was first to be associated with a rare congenital red cell disease, Diamond-Blackfan anemia (DBA) (1). This pathology is characterized by a severe erythroblastopenia that results from early arrest of pro-erythroblast differentiation (2–4). DBA patients also exhibit a number of heterogeneous clinical features (short stature, webbed neck, cranio-facial, skeletal, kidney, and heart abnormalities), consistent with the ubiquitous nature of RPS19 expression and of ribosome biogenesis. Heterozygous mutations of RPS19 affect ∼25% of the DBA patients. Why RPS19 deficiency has a prime effect on erythropoiesis remains unclear (5,6). It has been suspected that RPS19 might have an extra-ribosomal function important for pro-erythroblasts, but no solid evidence has substantiated this hypothesis so far, despite the characterization of several proteins interacting with RPS19 (7–9). Alternatively (although not exclusively), pro-erythroblast differentiation may be critically sensitive to partial deficiency of the translation machinery biogenesis or function. Indeed, this latter possibility has recently gained strong support from the finding that two other ribosomal proteins, RPS24 and RPS17, are mutated in ∼2% of DBA patients (10,11). Thus, so far, three autosomal genes encoding ribosomal proteins have been linked to this acute defect in erythropoiesis.
We and others have shown that RPS19 is critical for maturation of the 40S ribosomal subunit in yeast (12) and in mammals (13–15). Knockdown of RPS19 expression with small interfering RNAs (siRNAs) delays pre-rRNA cleavage in the internal-transcribed spacer 1 (ITS1; Fig. 1) and prevents subsequent maturation of the 18S rRNA precursor 3′ end into pre-40S particles. A similar defect, although milder, is detected in cells from patients bearing mutations in RPS19. Consistently, DBA-related mutations in RPS19 hamper its incorporation into ribosomal subunits (16,17). Perturbation of ribosome biogenesis is known to trigger cell-cycle arrest and apoptosis signaling pathways, as observed upon RNA-polymerase I inhibition (18,19) or expression of mutated ribosome biogenesis factors (20,21). A constitutive defect in ribosome production in DBA patients may thus put the cells under constant stress, which would render some cell types, especially pro-erythroblasts, very sensitive to cell-cycle arrest signals.
Sub-optimal production of ribosomes, or alteration of a ribosomal protein function in the 40S subunit, might also have direct consequences on translation efficiency and regulation, which pro-erythroblasts may be particularly sensitive to. Recent data indicate a lower protein synthesis rate in activated lymphocytes from DBA patients (22). In addition, RPS19 was found to interact with the translation initiation factor eIF2-α (23), and with the serine-threonine kinase PIM-1 in translating ribosomes (7), which may be related to a function in translation regulation.
The recent discovery that RPS24 is mutated in a fraction of patients brings an opportunity to further challenge the link between ribosome biogenesis and DBA (10). Although, RPS24 homolog in yeast Saccharomyces cerevisiae was found essential for pre-rRNA processing (24), its function in metazoan cells has not been described yet. In addition, whether there is a functional link between RPS19 and RPS24 is unknown. One may wonder to which degree the two proteins cooperate in ribosome biogenesis, and whether the lack of function of any of them may alter a particular step connected to a cell-cycle checkpoint. The data presented here show that RPS24 bears structural similarity with RNA-binding proteins and is as essential as RPS19 for production of the small ribosomal subunit. However, unlike RPS19, it is involved in processing of the pre-rRNA 5′-external transcribed spacer (5′-ETS). Accordingly, lymphoblastoid cells derived from DBA patients bearing mutations in RPS24 and RPS19 display different pre-rRNA processing defects. Double knockdown of these proteins suppresses the 5′-ETS processing defect observed upon the lack of RPS24, which suggests a functional relationship between the two ribosomal proteins.
Alteration of pre-rRNA processing in DBA patients with mutations in RPS24
To analyze the impact of RPS24 mutation on pre-rRNA processing in DBA, we examined the pre-rRNA maturation pattern in lymphoblastoid cell lines (LCLs) derived from DBA patients with mutations in RPS24 (Q106stop, R16stop, del E2-Q22) (10) and in RPS19 (R62Q, del exon2, del A58-T60) (25). Total RNAs were submitted to northern blotting and analyzed with a probe to the 5′-ITS1 probe specific of the 18S rRNA precursors (Fig. 2). Cells with mutations in RPS19 showed increased levels of 21S and 41S pre-rRNAs when compared with control cells (Fig. 2A lanes 4–6). This phenotype, characterized by high 21S/18S-E and 41S/30S ratios (Fig. 2B), corresponds to delayed maturation of the pre-40S subunits and reproduces previous observations in primary hematopoietic cells or skin fibroblasts from DBA patients affected in this gene, in LCLs bearing mutations in RPS19, or in rps19 siRNA-treated HeLa cells (13–15). The three LCLs with mutations in RPS24 also showed a clear alteration of pre-rRNA processing, but with distinct defects: the level of 41S pre-rRNA was lower than in control cells, whereas the 30S species accumulated (Fig. 2A, lanes 7–9), resulting in a lower 41S/30S ratio (Fig. 2B). In parallel, there was less 21S and 18S-E pre-rRNAs than in control cells. Similar analysis of the 5.8S/28S RNA maturation pathway with a probe complementary to the ITS2 showed no significant difference with control cells (not shown). These data indicate that the mutations in RPS24 found in DBA patients also affect maturation of the 18S rRNA, but not at the same stage as mutations in RPS19.
RPS24 is essential for production of the 40S subunit
To directly study the consequences of RPS24 loss-of-function on ribosome biogenesis, we designed two siRNAs (rps24-1 and rps24-2) that were used to transfect HeLa cells. The RPS24 mRNA level was reduced over 10 times by both siRNAs, as measured by quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) (Fig. 3A). Analysis of the translational machinery of cells depleted of RPS24 on sucrose gradient showed a drastic loss of free 40S subunits paralleled by a strong accumulation of free 60S subunits and reduced levels of 80S monosomes and polysomes (Fig. 3B). This profile is indicative of a strong defect of 40S subunit synthesis. Thus, like RPS19, RPS24 is essential for the production of the small ribosomal subunit.
RPS24 is required for processing of the 5′-ETS
We next examined pre-rRNA processing in cells depleted of RPS24 by pulse-chase labeling with L-[methyl-3H]methionine, which serves as a methyl group donor for methylation of pre-rRNAs by C/D box snoRNPs. Cells were labeled for 30 min, and chase with cold methionine was performed for up to 90 min. As seen in Figure 4A (see also Fig. 6C), cells treated with rps24-1 or rps24-2 siRNAs displayed a different processing pattern when compared with control cells. Most strikingly, there was no formation of the 41S pre-rRNA, which normally results from the cleavage of the 5′-ETS (Fig. 1, pathway A). In addition, the 21S pre-rRNA, whose generation requires cleavage of the 5′-ETS (Fig. 1, pathway B), was also absent, and no 18S rRNA formed in the time-course of the experiment. In contrast, the kinetics of 28S rRNA formation was equivalent to that in control cells. These observations are consistent with a function of RPS24 in the maturation of the 5′-ETS.
This hypothesis was further supported by detection of pre-rRNAs by northern blot with the 5′-ITS1 probe (Fig. 4B). In RPS24-depleted cells, we observed a strong reduction in the levels of several 18S rRNA precursors, namely the 41S, 21S and 18S-E pre-rRNAs. In parallel, there was a build-up of the amount of 30S pre-rRNA, as well as a mild increase in 45S pre-rRNA. Accumulation of 30S pre-rRNA, correlated with the absence of the 41S, 21S and 18S-E species indicates strong inhibition of cleavage of the 5′-ETS, as well as blockage of the subsequent processing of the ITS1 at the 3′ end of the 18S rRNA. Consistent with the absence of 18S-E pre-rRNA, which is normally exported from the nucleus, fluorescence in situ hybridization (FISH) with the same probe showed a strong reduction of the cytoplasmic labeling relative to control cells (Fig. 4C). Pre-18S rRNAs retained in the nucleus were detected in the nucleolus as well as in small nucleoplasmic dots.
The pattern of the precursors to the 60S subunit rRNAs, as detected with a probe to the ITS2, was unchanged (not shown). The amounts of 28S and 5.8S rRNAs were the same in rps24 siRNA-treated cells as in control cells, whereas the 18S/28S ratio was lower, which points towards a ribosome biogenesis defect primarily affecting formation of the 40S subunit. The changes in pre-rRNA pattern observed with rps24 siRNAs are similar, albeit more pronounced, to those observed in cells derived from RPS24-deficient DBA patient, from which we infer that the pre-rRNA processing defect observed in patient-derived cells is the direct consequence of RPS24 loss-of-function.
Down-regulation of RPS19 level in RPS24-depleted cells
Because RPS19 and RPS24 are both involved in DBA, one may suspect a specific functional link between the two proteins. Interestingly, RPS19 protein levels in cells from DBA patients carrying a mutation in RPS24 were reported to be lower than in control cells (10). In accordance to this observation, we found a strong reduction of RPS19 protein level in cells treated with rps24-1 siRNAs on western blots (Fig. 5A). Down-regulation of the protein level was not because of a decrease in RNA synthesis or stability, since the amount of RPS19 mRNA was unchanged in these cells (Fig. 5B). However, this effect was not specific to rps24 siRNAs, since knockdown of RPS15 led to a similar drop in the level of RPS19. Thus, we propose that loss of RPS19 in RPS24-depleted cells is primarily because of impaired synthesis of the 40S ribosomal subunit, which hampers stabilization of ribosomal proteins in mature ribosomal subunits.
RPS24 and RPS19 exert coordinated functions in ribosome biogenesis
The pre-rRNA processing defects observed upon knockdown of RPS19 or RPS24 expression indicate that these two proteins are involved in different cleavage steps: RPS19 is mostly required for processing of the ITS1, whereas RPS24 is necessary to cleave the 5′-ETS. Previous studies in yeast and in vertebrates have established that processing of the 5′-ETS and the ITS1 are spatially and temporally coupled (26–29). To get insights into the functional relationship between RPS24 and RPS19 in ribosome biogenesis, we combined rps19 and rps24 siRNAs to knockdown the expression of both proteins (Fig. 6). As a control, we also transfected each of these siRNAs with siRNAs directed against the ribosomal protein RPS15, which is required for nuclear export of the pre-40S particles and for the last processing step of the 18S rRNA at its 3′ end (30). Indeed, knockdown of RPS24 or RPS19 had a dominant effect on pre-rRNA processing when RPS15 was co-depleted (Fig. 6A, lanes 6 and 7), consistent with RPS19 and RPS24 acting upstream of RPS15 in the 40S subunit biogenesis process. But surprisingly, depletion of RPS19 together with RPS24 suppressed most of the 5′-ETS processing defect observed with rps24 siRNAs, as indicated by the presence of 41S and 21S pre-rRNA (Fig. 6A, compare lanes 3 and 5; quantifications on Fig. 6B). Accumulation of 21S pre-rRNA and low levels of 18S-E pre-rRNA recapitulated the ITS1 processing defect observed upon knockdown of RPS19 expression alone (Fig. 6A, lane 2). This result was reproduced with various combinations of two rps24 and two rps19 siRNAs (not shown). As shown in Figure 5B, the RPS24 mRNA was still reduced by over 10-fold when rps24 and rps19 siRNAs were co-transfected. Restoration of the 5′-ETS processing in cells depleted of both RPS24 and RPS19 was confirmed by pulse-chase experiments with L-[methyl-3H]methionine (Fig. 6C): the cells produced 43S, 41S and 21S pre-rRNAs, but did not process the 21S pre-rRNA into 18S-E RNA, as in rps19 siRNA-treated cells. These data show that depletion of RPS19 suppresses the phenotype induced by RPS24 knockdown, which strongly suggests that RPS19 and RPS24 exert tight inter-dependent functions in pre-rRNA processing.
Structure of RPS24e
The three DBA-related mutations in RPS24 reported so far affect the protein in different ways. The most drastic, R16stop, results in early translation arrest and may be equivalent to a null allele. Deletion N2-Q22 and mutation Q106stop are expected to yield proteins truncated of their N-terminal and C-terminal domains, respectively. To get insights into the structural domains affected by mutations in RPS24, we solved the crystal structure of protein RPS24e from the hyperthermophilic archeon Pyroccocus abyssi. RPS24e is homologous to the human RPS24 (27% identity and 57.5% similarity) and the two proteins are thus expected to harbor a similar fold. The crystal structure was determined at 1.9 Å resolution and refined to an Rfree of 26.6% and an R-factor of 22.1% (Table 1).
|a, b, c (Å)||66.81, 85.63, 42.11|
|α, β, γ (°)||90.0, 90.0, 90.0|
|Total measurements||101,890 (13,468)|
|Unique reflections||19,214 (2,861)|
|Completeness (%)||97.5 (91.3)|
|Rsym (%)||6.2 (31.2)|
|Phi, Psi angles|
|Most favored (%)||99.6|
|Additionally allowed (%)||0.4|
|a, b, c (Å)||66.81, 85.63, 42.11|
|α, β, γ (°)||90.0, 90.0, 90.0|
|Total measurements||101,890 (13,468)|
|Unique reflections||19,214 (2,861)|
|Completeness (%)||97.5 (91.3)|
|Rsym (%)||6.2 (31.2)|
|Phi, Psi angles|
|Most favored (%)||99.6|
|Additionally allowed (%)||0.4|
Numbers in brackets refer to the highest resolution shell (1.90–2.02 Å).
The protein RPS24e is built around a four-stranded anti-parallel β-sheet surrounded by three short α-helices (Fig. 7A). Structural similarity search performed with the DALI program (31) revealed homology with the prokaryotic ribosomal protein L23 (Fig. 7B), a component of the large ribosomal subunit, as observed in the crystal structure of large ribosomal subunits from archaea Haloarcula marismortui (PDB code 1FFK, 1S72) (32,33) and bacteria Deinococcus radiodurans (PDB code 2D3O) (34). The L23 family of proteins includes RPL25 in Saccharomyces cerevisiae and RPL23a in mammals. Prokaryotic L23 binds to the 23S ribosomal RNA through its βαββ structure, which recalls the RNA recognition motif (RRM) of snRNP protein U1A (33), and also interacts with ribosomal protein L29 (Fig. 7C). Presence of a very similar βαββ substructure in RPS24 strongly points towards a putative RRM made of β-strands 2-3-4 and α-helix 1 (amino acids 18 to 79 in human RPS24). The structure of this region is likely to be affected by the deletion of residues N2 to Q22, which compose β-strands 1 and 2 (Fig. 7A and D). In addition, the high sequence conservation of the N-terminal N2-Q22 domain between P. abyssi, S. cerevisiae and Homo sapiens (Fig. 7D) strongly argues for an essential contribution to the protein structure and function. Interestingly, the C-terminal domain missing in the Q106stop mutant is only found in eukaryotes. This domain, which is not present in the structure presented here, may not be essential for protein folding, but rather involved in intermolecular interactions with eukaryote-specific molecules.
The data presented here reinforce the hypothesis that DBA is linked to defective ribosome biogenesis and bring further insights into the mechanisms underlying the physiopathology of this disease. First, we find that mutations in RPS24 in DBA patients are associated with altered pre-rRNA processing. Lymphoblastoid cells derived from DBA patients with mutated RPS24 display a clear defect in pre-rRNA maturation, although distinct from the phenotype observed in cells derived from patients with mutations in RPS19. Consistently, knockdown of RPS24 expression with siRNAs perturbs processing of the 5′-ETS and maturation of the 3′ end of the 18S rRNA. Three cleavage sites were characterized in the 5′-ETS of mammalian pre-rRNA: the early cleavage site to be cleaved first (around nucleotides 415/422 in human 47S pre-rRNA) (35); site 1 at the 5′ end of the 18S; and the A0 site (around nucleotide 1643), which seems to be processed either more slowly or quasi-simultaneously with site 1 (30). Similar to the yeast S. cerevisiae, depletion of Rps24p (RPS24 homolog) blocks maturation of the 5′-ETS, as well as cleavage of the ITS1 at site A2 (24) and results in strong accumulation of ‘23S’ pre-rRNA, which may be seen as the equivalent of the mammalian 30S pre-rRNA. Thus, as for RPS19, the function of RPS24 in ribosome biogenesis is conserved from yeast to mammals.
The fact that the pre-rRNA processing pattern in all lymphoblastoid cells matches the one observed in rps24 siRNA-treated cells strongly suggests that the three mutations affecting RPS24 in DBA patients yield haploinsufficiency. Besides mutation R16stop, which may be equivalent to a null allele, mutation Q106stop and deletion of E2-Q22 also interrupt RPS24 function in ribosome biogenesis. The crystal structure of RPS24e reveals a strong structural homology with the L23 family of ribosomal proteins (L23 in prokaryotes, RPL25 in yeast, RPL23a in mammals): these proteins not only interact with rRNA and ribosomal protein L29, but also provide docking sites for the trigger factor and the signal recognition particle (36,37). Structural superimposition with L23 suggests a putative RNA-binding surface in RPS24. Deletion of 20 amino acids between E2 and Q22 is expected, at least, to alter this surface, hampering association with the pre-rRNA, and most probably to destabilize the structure of the whole protein, leading to its degradation. In contrast, the C-terminal domain absent in mutant Q106stop does not seem to be required for the folding of the protein, since it is absent in the archaeal RPS24e. This module may instead be involved in additional interactions in eukaryotic pre-ribosomes, either with the pre-ribosomal RNA or with some of the numerous pre-ribosomal factors required for ribosome biogenesis. Alternatively, this lysine-rich domain may contain the nuclear localization signal of the protein. In parallel, the premature stop codon in the mRNA encoding this mutant may also trigger nonsense-mediated decay. Further work is needed to sort out these hypotheses.
Loss-of-function of pre-ribosomal factors and ribosomal proteins of the large and the small subunits were shown to activate p53 and trigger cell cycle arrest and apoptosis (18–21). This stress response, together with deficient translation because of a low ribosome number, is likely to play a direct role in a pathological process involving interruption of differentiation. Remarkably, knockdown of different proteins involved in ribosome biogenesis affect distinct cell types in vivo. Hence, homozygous knockout of ribosomal protein RPL22 prevents αβ T-cell development in mouse (38), whereas deficiency of protein SDBS, which is required for maturation of the 60S ribosomal subunit, primarily affects neutrophil differentiation in Shwachman-Diamond syndrome (39,40). Thus, the involvement of RPS19 and RPS24 in DBA, characterized by a differentiation defect in yet another cell lineage (pro-erythroblasts), calls for a closer functional link between the two proteins. By immunoelectron microscopy on isolated rat 40S subunits, RPS19 was localized on the ‘head’ of the particle and RPS24 at some distance on the ‘body’ (41), and the two proteins were found to assemble into pre-40S particles independently from one another in yeast (42). These observations do not support a direct interaction of the two proteins within the ribosome. However, one cannot formally rule out that they interact physically or simply functionally within the pre-ribosomal particles during the maturation process.
In this respect, the connection reported here between the activities of the two proteins in pre-rRNA processing is an interesting clue. Depletion of RPS19 suppresses, in large part, the 5′-ETS processing defect induced by RPS24 siRNAs, which reveals a functional interaction. RPS19 depletion strongly affects processing of the ITS1: slower cleavage at site 2, alternative cleavage of the ITS1, inhibition of the 18S rRNA 3′ end processing (13–15,43). Epistasis of RPS19 on RPS24 is thus counter-intuitive, since RPS19 is not necessary for processing of the 5′-ETS. However, it is well established, in yeast as in vertebrates, that the 5′-ETS and the ITS1 are processed in a coordinated manner (26,44–46). RPS19 and RPS24 may function in concert, together with factors known to be involved in coupling the processing of the 5′-ETS and ITS1, like the snoRNP U3. Different models of the functional interaction between RPS19 and RPS24 may be proposed. For example, the two proteins could chaperone a structural transition of the pre-rRNA. This conformational change would not be directly required for processing of the 5′-ETS, which would thus occur in the absence of the two proteins. However, absence of RPS24 alone would lead to an aberrant structure incompatible with cleavage of the 5′-ETS. Alternatively, RPS19 and RPS24 could regulate the binding to the pre-ribosomal particle of an early pre-ribosomal factor, whose dissociation would be a pre-requisite to processing of the 5′-ETS. Recruitment of this factor would involve RPS19, whereas RPS24 would be needed for its dissociation. Absence of RPS24 would lead to accumulation of aberrant pre-ribosomal particles containing the non-dissociated factor and unable to cleave the 5′-ETS.
These data lead us to propose that RPS19 and RPS24 cooperate at a particular stage in ribosome biogenesis validated by a checkpoint, whose non-compliance would signal cell cycle arrest. This signal, in combination with additional stresses, would be deleterious to differentiation of erythroid progenitors, which require rapid proliferation, or to developmental processes. Analysis of new genes involved in DBA, like RPS17, may confirm or invalidate this hypothesis, depending on their relationship with RPS19 and RPS24. In this respect, it is of note that RPS17 is located on the head of the 40S subunit, in close proximity to RPS19 with which it may directly interact (41).
MATERIALS AND METHODS
Epstein-Barr Virus (EBV)-transformed LCLs were cultured in RPMI 1640 medium supplemented with 15% fetal bovine serum (FBS), 100 U/ml penicillin, 100 μg/ml streptomycin and 0.292 mg/ml l-glutamine (all from Invitrogen, Carlsbad, California) at 37°C in 5% CO2. Total RNAs were isolated from LCLs using RNA isolation kit (Qiagen, Valencia, CA, USA).
Human cervical carcinoma HeLa cells were maintained in culture in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal calf serum (FCS), 1 mm sodium pyruvate, 100 U/ml penicillin and 100 mg/ml streptomycin (all from Invitrogen) in 5% CO2 at 37°C.
Small interfering RNAs
Two different 21-mer siRNAs were designed to knockdown expression of the human RPS24 [GenBank accession no. NM_033022 (full-length transcript)]: 5′-CACCGGAUGUCAUCUUUGUdTdT-3′ (rps24-1 siRNA), 5′-GGAAACAAAUGGUCAUUGAdTdT-3′ (rps24-2 siRNA). The siRNAs targeting expression of RPS15 and RPS19 were described elsewhere (13,30). All siRNAs were purchased from Eurogentec (Seraing, Belgium) and transfected using electrotransformation as described previously (30).
Fractionation and analysis of ribosomes by sucrose density gradient centrifugation
Forty-eight hours after transfection with siRNAs, HeLa cells were treated with 100 mg/ml cycloheximide (Sigma, Sigma-Aldrich Co., Saint-Louis, Missouri) for 10 min, fractionated and the cytoplasmic fraction was analyzed on a sucrose gradient as described previously (30).
Quantitative reverse transcriptase-polymerase chain reaction
The RPS24 gene down-regulation induced by RNA interference was analyzed by quantitative PCR taking GAPDH as an internal control to normalize the RPS24 expression. Total RNAs were isolated using TRIzol reagent (Invitrogen), and first-strand synthesis of cDNAs was performed on 2 µg of RNA at 42°C for 2 h using AMV reverse transcriptase as recommended by the supplier (Promega, Madison, WI, USA). The sequences of primers used for PCR were the following—for RPS24: forward, 5′-CCGACTACTTCAGAGGAAACAAA-3′; reverse, 5′-GCCACCACCAAAATGAGTTCT-3′; for GAPDH: forward, 5′-TGCACCACCAACTGCTTAG-3′; reverse, 5′-GTTCAGCTCAGGGATGACC-3′. Real-time PCR was carried out using Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen) according to the manufacturers′ instructions, on a ICyclerQIM (Bio-Rad, Hercules, CA, USA). After initial denaturation at 95°C for 3 min, 40 cycles were performed with the following conditions: 95°C for 20 s; 56°C for 30 s and 72°C for 20 s. The fluorescent signals were measured after each extension step at 78°C. Triplicate PCR reactions were prepared for each sample and measurements were independently repeated twice. Data were analyzed using the IQCycler version software. The concentration of the gene-specific mRNA in treated cells relative to untreated cells was calculated by subtracting the normalized Ct values obtained for untreated cells from those obtained from treated samples (ΔΔCt = ΔCt, treated—ΔCt, untreated) and the relative concentration was determined (2–ΔΔCt).
Analysis of pre-rRNAs
Pre-rRNA analysis by northern blot was performed as described previously (47) with probes 18S, 28S, 5′-ITS1, ITS2b and ITS2-d/e and actin (30). For detection of the ITS2, the ITS2-b and ITS2-d/e probes were mixed in equal amounts. The protocol of FISH with the 5′-ITS1 probe was reported in Rouquette et al. (30).
For pulse-chase analysis, cells were transfected with rps24-1 and rps24-2 siRNAs using electroporation and plated at a density of 1.5 × 105 cells in 12-well dishes. Two days later, they were pre-incubated for 30 min in serum-free methionine-free medium and then incubated for 30 min in 0.5 ml of the same medium containing 25 µCi L-[methyl-3H]methionine (50 mCi/ml). The cells were then chased in non-radioactive medium containing cold methionine (30 mg/ml) for various times, after which total RNAs were isolated using TRIzol (Invitrogen). RNAs were separated on a 1% agarose gel, transferred to nylon membrane and exposed to a film using an intensifying screen (Transcreen LE, Amersham).
Protein expression, purification and crystallization
The full-length cDNA of P. abyssi RPS24e was cloned into a modified pET-15b allowing production of an N-terminally His-tagged fused protein in Escherichia coli (48). After initial growth at 37°C, cell cultures were cooled down to 15°C and protein production was induced with 0.5 mm isopropyl beta D galactopyranoside (IPTG) overnight. Purification was carried out after cell lysis by centrifugation at 4°C for 1 h at 50,000 g. The supernatant was incubated at 60°C for 30 min and further centrifuged. The supernatant was then incubated in batch with HIS-Select affinity resin (Sigma). The eluate was loaded on a HiQ-Sepharose. The protein was concentrated up to 13 mg/ml in 25 mm Tris–HCl of pH 7.5 and 100 mm NaCl. Crystallization of RPS24e was carried out at room temperature using sitting-drop vapor diffusion by mixing one volume of protein solution with one volume of 40% PEG 400, 100 mm sodium acetate pH 4.2 and 185 mm MgCl2 of reservoir solution (Nextal). Crystals were directly flash-frozen in liquid nitrogen for data collection. Data were processed with XDS (49). Data collection and phasing statistics are shown in Table 1.
Phase information was obtained by a single-wavelength anomalous dispersion (SAD) experiment at a resolution of 1.9 Å on a Se-Met-substituted protein crystal. Seven Se sites were located using SHELXD (50) and initial phases were improved with SHARP/AutoSHARP (51). The model was automatically built with Arp/Warp (52) and further improved manually with Coot (53). Model refinement was achieved with REFMAC5 (54). The final model has a Rfree value of 26.6% and Rwork value of 22.1%. The crystallographic asymmetric unit contains two molecules of RPS24e. Both molecules have a very similar structure with an RMS deviation of 0.9 Å. The final model includes most of the RPS24 amino acids with the exception of the C-terminal residues for both molecules (respectively, seven residues for molecule A and five residues for molecule B). Four amino acids coming from the His-tag are also visible on molecule A. The crystal structure of RPS24e from P. abyssi is similar to the structure of RPS24e from archeon Methanosarcina mazei determined by nuclear magnetic resonance spectroscopy (unpublished, PDB code 1XN9) with the exception of the C-terminal α-helix. This difference could be accounted for protein rearrangement upon crystallization since this area of P. abyssi RPS24e is involved in crystal contacts with neighboring molecules. The atomic coordinates and structure factor have been deposited in the Protein Data Bank (accession code 2v94).
This project was supported by the Agence Nationale de la Recherche (ANR—RIBODBA project), the CNRS and the University of Toulouse. This project was also supported by a research grant from The Diamond-Blackfan Anemia Foundation (to HTG).
The authors wish to thank Coralie Carron and Marlène Faubladier for scientific discussion and technical assistance. The authors are grateful to the Daniella Maria Arturi Foundation for stimulating research collaborations in the field of DBA.
Conflict of interest statement. None declared.