MiR-662 impairs C/D box snoRNA expression and rRNA 2′-O-ribose methylation marks Authors: Charlotte Filossi, Caroline Isaac, Mariam Jaafar, Philippe Clézardin, Caroline Moyret-Lalle, Virginie Marcel, Margherita Puppo Published: 2026 ABSTRACT BackgroundMiR-662 overexpression has been reported to promote breast cancer metastatic progression and to impair the expression of genes encoding for proteins involved in translation, ribosome biogenesis and ribosome processing (Puppo et al., 2023); however, the relationship between miR-662 and the translational machinery was not further investigated at the time.MethodsMiR-662 was overexpressed in MDA-MB-231-luc2 (NW1) human breast cancer cells. Global protein synthesis was analyzed using polysome profiles. Potential defects in rRNA 47S precursor biogenesis were evaluated by Northern blot. C/D box snoRNA (SNORD) expression was quantified by a medium-throughput RT-qPCR microfluidic dynamic array. rRNA 2’O-ribose methylation (2’Ome) was profiled using RiboMethSeq.ResultsNo changes in rRNA synthesis, processing, or maturation were observed upon miR-662 overexpression. In contrast, a marked reduction in the ratio of polysomal to free ribosomal fractions was observed, indicating an impaired mRNA engagement into translation. Interestingly, a drastic decrease in SNORD levels (but not of their host genes) and a global decrease in rRNA 2’Ome was observed upon miR-662 overexpression.ConclusionsThe involvement of miRNAs in snoRNA-induced modulation in rRNA epitranscriptomic marks might be a novel mechanism of fine regulation of the ribosome composition with possible repercussions on breast cancer metastatic progression. Introduction We have previously reported an association between the small non-coding RNA (sncRNA) miR-662 –an onco-miR that contributes to breast cancer metastatic progression–, and the translational machinery as a result of the GSEA analysis with RNA-seq data (available on GEO dataset, #GSE267064) on an engineered breast cancer cell model [1, 2]. Specifically, gene pathways related to translation, ribosome biogenesis and ribosomal RNA (rRNA) processing were repressed upon miR-662 overexpression in MDA-MB-231-luc2 (NW1) cells [1], suggesting a link between miR-662 biological activities, global protein synthesis, and ribosome biogenesis that warrants further investigation. In the last decades, the pivotal influence of translational regulation on cell phenotype determination, particularly via the fine-tuning modulation of ribosome composition, has been demonstrated [3, 4]. Depending on the physio-pathological contexts, ribosomes can exhibit different compositions with distinct structures and functions [5]. Interestingly, another class of sncRNAs, the small nucleolar RNAs (snoRNAs), collaborates to shape ribosome composition by being involved at both rRNA processing (i.e., cleavage of rRNA precursors) and rRNA maturation (i.e., addition of rRNA chemical modifications) in physio-pathological contexts [6, 7]. We hypothesized that miR-662 might be able to regulate protein synthesis and ribosomes biogenesis and/or composition. Here, we report a novel regulatory mechanism that links miR-662 biological activities to rRNA 2’-O-ribose methylation (2’Ome) composition probably through the regulation of C/D box snoRNA expression levels. Materials and methods MiRNA overexpression in cell models Stable human MDA-MB-231-luc2 (NW1) breast cancer cells overexpressing miR-662 (NW1/LENTI-662) and control cells (NW1/LENTI-Ctrl) were cultured as previously described [1]. Transient miR-662-overexpressing (MIMIC-miR-662) or control (MIMIC-Ctrl) NW1 cells have been obtained as previously described by miRNA-mimic transfection [1]. Polysome profiles Polysome profiles were performed to analyze changes in ribosomal (i.e., 40S and 60S ribosomal subunits, and 80S ribosomes) and polysomal fractions (i.e., mRNAs associated with >1 actively translating ribosome) in stable miR-662-expressing or control NW1 cell lines. NW1/LENTI-Ctrl and NW1/LENTI-662 cells were plated in advance to reach ~70-80% confluence on the day of the experiment. One hour before lysis, cells were incubated in pre-warmed medium containing Emetine (25 µg/mL, Sigma-Aldrich, E2375) to inhibit ribosome run-off. In parallel, 10–50% sucrose gradients were prepared using filtered sucrose solutions in the lysis buffer (10 mM HEPES-KOH pH 7.9, 100 mM KCl, 1.5 mM MgCl₂). Gradients were generated using a Gradient Master (Biocomp, SW41 program) in 13.2 mL open-top polyallomer tubes. Cells were washed with 1X PBS, trypsinized, pooled, and centrifuged at 500 ×g at 4 °C. Pellets were sequentially washed with cold 1X PBS and lysis buffer, then lysed in ice-cold lysis buffer supplemented with 1 mM DTT, 0.5 U/µL RNasin (Promega, N2515), and protease inhibitor cocktail (ThermoFisher Scientific, 78410). After 20 min incubation on ice, lysates were homogenized using a Precellys homogenizer (2 mL CK14 tubes, 10 s, 4500 RPM, Bertin) and clarified via subcellular fractionation: centrifugation at 800 × g (10 min, nuclei), then 12,000 × g (10 min, mitochondria). The resulting cytoplasmic supernatant was quantified by Nanodrop, and 300 µg of RNA was loaded per gradient. Gradients were centrifuged at 39,000 RPM for 2.5 hrs at 4 °C using a Beckman SW41 Ti rotor. Polysome profiles were collected using an automated fraction collector coupled to UV absorbance detection at 254 nm. RNA was extracted from polysome gradient fractions using a two-step acid phenol–chloroform protocol under a chemical hood at room temperature. Fractions (~700 µL) were split into two equal volumes (∼350 µL) to allow parallel extractions, later pooled. Each aliquot was mixed with 700 µL of TRIzol Reagent (Thermo Fisher Scientific), vortexed for 5 s, and incubated for 5 min at room temperature. Then, 200 µL of chloroform was added, and samples were gently inverted 15 times, incubated for 3 min, and centrifuged (12,000 RCF, 15 min, 4 °C). The aqueous phase (~620 µL) was transferred to a new tube, and a second chloroform extraction was performed (200 µL chloroform, same steps as above). After centrifugation, ~500 µL of the upper phase was collected and mixed with 300 µL isopropanol and 2 µL GlycoBlue (ThermoFisher). RNA was precipitated for 10 min at room temperature and pelleted by centrifugation (12,000 RCF, 15 min, 4 °C). The RNA pellet was washed twice with 1 mL of 75% ethanol, each time vortexed briefly and centrifuged (12,000 RCF, 5 min, 4 °C). Residual ethanol was removed and pellets were air-dried for 20 min. Each was resuspended in 5–6 µL of RNAse-free water without pipetting up and down and left to dissolve at 4 °C for several hrs or overnight. Paired extractions were pooled, and RNA concentration was measured using a Nanodrop spectrophotometer. Northern blot Potential defects in processing of rRNA 47S precursor (pre-47S) were evaluated by Northern blot. Cells were collected with TRIzol and RNA were isolated using TRIzol/Chloroform as detailed in the polysome profiles section (Life Technologies, 15596018). RNA pellets were suspended in the formamide buffer (Deionized formamide, 5 mM EDTA, 0.1% Bromophenol blue). Total RNA was mixed at 1:1 in MOPS:formaldehyde buffer (0.04 M MOPS, 30% formaldehyde, 5 mM EDTA, 0,1% Bromophenol Blue), denatured at 95 °C for 2 min and then resolved by electrophoresis on 1.2% agarose, 6% formaldehyde, 0.02 M MOPS gel. Transfers were performed by capillarity on a nylon membrane (Whatman™ Nytran® SPC 0.45 µm) with 10X SSC (1.5 M sodium chloride, 0.15 M sodium citrate, pH 7.0). Membranes were incubated overnight at 42 °C with the hybridization buffer (ULTRAhyb-Oligo, Invitrogen containing ITS1 or ITS2 probes) (Table S1). Membranes were washed four times with 0.1X SSC and 0.1% SDS before signal exposure on the ChemiDoc MP system (Bio-Rad). Quantification was performed using Image Lab software (Bio-Rad). The results are presented as Ratio Analysis of Multiple Precursors, with each intermediate normalized to its most stable precursor. Low and medium-throughput RT-qPCR Total RNA was extracted using a miRNeasy kit (Qiagen) and manufacturers’ instructions. RNA concentration and purity were evaluated using a Nanodrop™ 2000 spectrophotometer (Thermo Fisher Scientific). PrimeScript RT Reagent Kit (Takara) was used for reverse transcription. A negative control using RNase/DNase-free water, as well as a positive control using reference RNA (XpressRef Universal, Qiagen), were included. The thermal cycler program followed was the one recommended by the manufacturer (Takara). For low-throughput RT-qPCR, the resulting cDNAs were diluted by adding 190 µL of RNase/DNase-free water. Real-time qPCR was performed using the LightCycler 96 (Roche) and the LightCycler 480 SYBR I Master (Roche, 4887352001). Gene expression was assessed by analyzing the cycle threshold (Ct) values of the genes (RPL5, RPS8, RPS3, RPS12, RPL13A, RPL4, miR-662) relative to housekeeping genes (GAPDH, HPRT1) to obtain delta Ct values (ΔCt). Relative expression levels were calculated using the 2−ΔCt method [4]. SNORD expression was quantified using a 48 × 48 microfluidic dynamic array on the Biomark HD system (Standard BioTools) (Table S2) [8]. This platform allows simultaneous quantification of 48 targets across 48 samples using EvaGreen dye-based real-time PCR. Medium-throughput RT-qPCR was performed as previously detailed [5, 6]. Briefly, after reverse transcription using Takara (RR037A), a pre-amplification step was performed to increase the DNA matrix using Preamp Master Mix (Standard BioTools) prior to perform EvaGreen dye-based RT-qPCR using SsoFast EvaGreen Supermix with Low Rox (Bio-Rad). For each amplification cycle, fluorescence signals were recorded and analyzed using the Fluidigm Real-Time PCR Analysis software. Quality control steps included assessment of uniform sample loading, evaluation of no-template controls for contamination, and verification of amplicon specificity via melting temperature (Tm) analysis were performed. Ct values were normalized using the median of two reference small RNAs (RNU6B, U3), and relative expression levels were calculated using the 2–∆Ct method [4]. RiboMethSeq rRNA 2′Ome was profiled at 112 sites using 100 ng of RNA by RiboMethSeq. Briefly, the presence of 2′Ome protects the phosphodiester bond located at the 3′ of the methylated site from alkaline hydrolysis. Thus, the presence of 2′Ome at the given nucleotide n induces a decrease in RNA fragments starting at the nucleotide n + 1, allowing the calculation of the 2′Ome level at the corresponding site (or C-score). RiboMethSeq was performed as previously described using Illumina technology on the RibosOMICS and PGC/PBGT platforms of the CRCL [9, 10]. After sequencing, adapter sequences were trimmed off raw reads using Trimmomatic, the extracted reads were aligned on a human rRNA sequence (NR_046 235) using Bowtie 2, and 5′ end reads were counted at each rRNA nucleotide position (https://github.com/RibosomeCRCL/ribomethseq-nf). Using the rRMSAnalyzer package (https://github.com/RibosomeCRCL/rRMSAnalyzer), the quality control was performed to identify putative outliers, the adjustment of end read counts was applied using the CombatSeq algorithm when required, C-scores were computed (flanking window of ±6 nucleotides), the identification of the most variable rRNA 2′Ome sites was performed using the C-score interquartile range (IQR), and data visualization was drawn. Statistical analysis Statistical tests are described above and in figure legends. Biological replicates are shown in the figures, and sample sizes are indicated in the figure legends. Statistical analyses were performed using GraphPad Prism 10.1.1. Polysome quantification was assessed using an unpaired parametric t-test (p.value < 0.05), while quantifications from northern blot, Fluidigm and supplementary data were analyzed using the non-parametric Mann–Whitney test (p.value < 0.05). Results To first assess a possible impact of miR-662 overexpression on ribosome biogenesis (i.e., rRNA synthesis and processing), we performed a Northern blot analysis using ITS1 and ITS2 probes that hybridize rRNA internal transcribed spacer (ITS) regions on total RNA from MIMIC-miR-662 transfected NW1 cells compared to MIMIC-Ctrl at 40 hrs post-transfection. No changes in rRNA synthesis or processing were observed upon miR-662 overexpression compared to control (Fig. 1a–c). In parallel, we compared polysome profiles between stably miR-662-overexpressing (NW1/LENTI-662) and control (NW1/LENTI-Ctrl) cells. We found that NW1/LENTI-662 cells showed a significant decreased in the ratio of polysomal fractions (7-12) to free ribosomal fractions (5–7, that include 40S, 60S, and 80S) compared to the control (Fig. 1d, e). As displayed in the representative polysome profile (Fig. 1d), this decrease in polysome/ribosome ratio is likely attributable to both a reduction in the polysome fraction and a concomitant increase in the free ribosomal fractions under miR-662 overexpression, which is consistent with impaired mRNA engagement into translation. These results are in accordance with the previously described miR-662-induced decrease in translation [1], suggesting that miR-662 reduces translational process.Fig. 1Reduced polysomes upon miR-662 overexpression are associated with decrease in both C/D box snoRNA and rRNA 2’Ome levels.a–c Representative Northern blots probed with ITS1 (a) and ITS2 (b) at 40 hrs post-transfection with a miR-662-mimic (MIMIC-miR-662) or a negative control miRNA-mimic (MIMIC-negCtrl) are shown as well as their quantification (c). d Representative polysome profiles from miR-662-overexpressing cells (NW1/LENTI-662, pink) and negative control cells (NW1/LENTI-Ctrl, grey) are shown. e Quantification of polysome profiles was presented as the ratio of area under the curve (AUC) of polysomal fractions (7-12) compared to respective ribosomal fractions (5-7). f Heatmap showing the mean log2(FC) of normalized relative expression ratios for 37 C/D box snoRNAs (SNORDs) after transient (MIMIC-miR-662, 24 hrs post-transfection) or stable (NW1/LENTI-662) miR-662 overexpression, compared to corresponding controls (MIMIC-negCtrl, NW1/LENTI-Ctrl). g Expression levels of the SNORD host gene expression were analysed by RT-qPCR in NW1 cells at 96 hrs post-transfection with negative mimic control (MIMIC-negCtrl) or miR-662-mimic (MIMIC-miR-662). h rRNA 2’Ome levels were presented as C-scores, either as median C-scores in each condition (upper panel) or median C-score differences between NW1/LENTI-662 versus NW1/LENTI-Ctrl (lower panel). Reduced polysomes upon miR-662 overexpression are associated with decrease in both C/D box snoRNA and rRNA 2’Ome levels. a–c Representative Northern blots probed with ITS1 (a) and ITS2 (b) at 40 hrs post-transfection with a miR-662-mimic (MIMIC-miR-662) or a negative control miRNA-mimic (MIMIC-negCtrl) are shown as well as their quantification (c). d Representative polysome profiles from miR-662-overexpressing cells (NW1/LENTI-662, pink) and negative control cells (NW1/LENTI-Ctrl, grey) are shown. e Quantification of polysome profiles was presented as the ratio of area under the curve (AUC) of polysomal fractions (7-12) compared to respective ribosomal fractions (5-7). f Heatmap showing the mean log2(FC) of normalized relative expression ratios for 37 C/D box snoRNAs (SNORDs) after transient (MIMIC-miR-662, 24 hrs post-transfection) or stable (NW1/LENTI-662) miR-662 overexpression, compared to corresponding controls (MIMIC-negCtrl, NW1/LENTI-Ctrl). g Expression levels of the SNORD host gene expression were analysed by RT-qPCR in NW1 cells at 96 hrs post-transfection with negative mimic control (MIMIC-negCtrl) or miR-662-mimic (MIMIC-miR-662). h rRNA 2’Ome levels were presented as C-scores, either as median C-scores in each condition (upper panel) or median C-score differences between NW1/LENTI-662 versus NW1/LENTI-Ctrl (lower panel). While the role of miR-662 in translation mechanisms remains to be clarified, this process is known to be regulated by several families of sncRNAs. In particular, C/D box snoRNAs (SNORDs) play an established, direct role in rRNA 2’Ome and ribosome assembly by being key primary actors of rRNA maturation through rRNA 2’Ome guidance [5]. This led us to investigate the expression levels and functions of SNORDs in our engineered cell lines. Using medium-throughput RT-qPCR, we quantified the expression levels of 37 SNORDs in response to miR-662 overexpression. These 37 SNORDs have been specifically selected for their role in guiding rRNA 2’Ome, whose levels have previously been reported to exhibit variation in breast tumors [11]. Besides the intrinsic differences of the two miR-662 overexpression methods that might results in a different biological output and temporal regulation, a global decrease in SNORD expression levels was observed in both stable (NW1/LENTI-662) and transiently (MIMIC-miR-662, 24 h) miR-662-overexpressing cell lines compared to respective controls (Fig. 1f). Notably, no changes in the expression levels of their relative host genes were observed (Fig. 1g). Furthermore, these changes in SNORD levels were robust across biological replicates and independent of the overexpression strategy, indicating a reproducible regulatory effect of miR-662 on SNORD expression. A global decrease in SNORD expression levels was also observed in response to miR-24-2-5p transient overexpression (data not shown), which shares with miR-662 common biological activities at an early stage of the metastatic progression [2], suggesting a general response to miRNAs involved in metastatic progression. These data suggest that, by altering SNORD expression, miR-662 might impair rRNA 2’Ome. To test this hypothesis, we have profiled rRNA 2’Ome using RiboMethSeq. A global decrease in rRNA 2’Ome level was observed in NW1/LENTI-662 cells compared to NW1/LENTI-Ctrl cells (Fig. 1h). Some sites were more affected than others and could be explained by variations in the expression levels of individual guiding SNORDs, differences in the efficiency of snoRNA incorporation into snoRNP complexes, variability in snoRNA–rRNA binding affinity, and potential competition between nearby rRNA modification sites. Overall, these data suggest that miR-662 reduces both rRNA 2′Ome and guide SNORD expression levels. These results are in accordance with the previously described miR-662-induced defects in ribosome biogenesis at rRNA level [1]. Discussion and future perspectives In agreement with the observation that miR-662 impaired translation and/or ribosome biogenesis [1], here, we have now demonstrated that miR-662 overexpression does not affect rRNA synthesis or processing, but rather rRNA maturation by drastically reducing rRNA 2′Ome altogether with global protein synthesis. Our data illustrate the close relationship between miRNA, snoRNAs, ribosome biogenesis and protein synthesis. Interestingly, it was recently shown that some nucleolar localized miRNAs were identified as negative regulators of ribosome biogenesis since they impaired pre-rRNA processing, induced dysregulation of the 30S pre-rRNA precursor maturation and significantly decreased global protein synthesis [12]. We have shown that miR-662 downregulates SNORD expression levels, potentially at the origin of the decrease in rRNA 2’Ome patterns and protein synthesis. It has to be noted that we have observed changes of SNORD expression levels, but not of the host genes from which they originate, suggesting that miR-662 might regulate SNORD processing rather than their expression levels. Therefore, it is likely that miR-662 regulates genes encoding RNA-binding proteins involved in either snoRNA processing and stability or snoRNP formation. In fact, the involvement of miRNAs in the expression and/or processing of the different classes of snoRNAs is still to be explored as a novel mechanism of fine regulation of ribosome composition, warranting further investigation into the epitranscriptomic reprogramming of ribosomes driven by miRNAs, which can then influence cancer progression. Supplementary information Supplementary Material Supplementary Material