Translation is required for miRNA-dependent decay of endogenous transcripts

Abstract

Post-transcriptional repression of gene expression by miRNAs occurs through transcript destabilization or translation inhibition. mRNA decay is known to account for most miRNA-dependent repression. However, because transcript decay occurs co-translationally, whether target translation is a requirement for miRNA-dependent transcript destabilization remains unknown. To decouple these two molecular processes, we used cytosolic long noncoding RNAs (lncRNAs) as models for endogenous transcripts that are not translated. We show that, despite interacting with the miRNA-loaded RNA-induced silencing complex, the steady-state abundance and decay rates of these transcripts are minimally affected by miRNA loss. To further validate the apparent requirement of translation for miRNA-dependent decay, we fused two lncRNA candidates to the 3'-end of a protein-coding gene reporter and found this results in their miRNA-dependent destabilization. Further analysis revealed that the few natural lncRNAs whose levels are regulated by miRNAs in mESCs tend to associate with translating ribosomes, and possibly represent misannotated micropeptides, further substantiating the necessity of target translation for miRNA-dependent transcript decay. In summary, our analyses suggest that translation is required for miRNA-dependent transcript destabilization, and demonstrate that the levels of coding and noncoding transcripts are differently affected by miRNAs.

Keywords: Dicer knockout mESC; RNA metabolic labelling; long noncoding RNAs; miRNA; translation.

Figure EV1. Characterization of miRNA‐depleted mESC following inducible Dicer knockout

1. A

   Distribution of log₁₀ ratio between nuclear/cytosolic (transcripts per million, TPM) in mESCs for mRNAs (red) and lncRNAs (blue). Median nuclear/cytosolic ratio of mRNAs is represented by dashed line.

2. B

   Distribution of the expression (log₁₀(TPM)) of transcripts with and without experimental evidence for AGO2 binding in mESCs. Central band of boxplot represents median, box depicts 25–75 quantiles of distribution, and whiskers represent the 5^th and 95^th quantiles of the distribution. The data represented are based on analysis of two independent biological replicates.

3. C

   Density of AGO2 wild‐type specific clusters across cytosolic lncRNAs (n = 48, blue) and the 3’untranslated regions of mRNAs (n = 2,355, red) with experimental evidence for AGO2 binding in mESC (>0 AGO2 clusters) based on (Leung et al, 2011). Central band of boxplot represents median, box depicts 25–75 quantiles of distribution, and whiskers represent the 5^th and 95^th quantiles of the distribution.

4. D

   Immunoblot analysis of DICER (DCR) and NANOG in protein extracts from DICER conditional mESCs 8 days after treatment with ethanol (WT) or 4‐OHT (KO) in three independent biological replicates (BR1–3). ACTIN‐β (ACT‐β) was used as an internal control.

5. E

   Densiometric quantification of the relative difference in DCR levels from (D). Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate. Two‐tailed paired t‐test P‐value = 0.010.

6. F

   Percentage of proliferating mESCs after 8 days of treatment with ethanol (WT) or 4‐OHT (KO) in three independent biological replicates. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate. Two‐tailed paired t‐test P‐value = 0.011.

7. G

   Fold change (FC) in Oct4 (two‐tailed t‐test P‐value = 0.34), Nanog (paired two‐tailed t‐test P‐value = 0.084), Myc (paired two‐tailed t‐test ***P‐value = 0.008), Sox2 (paired two‐tailed t‐test P‐value = 0.52) and Dicer (paired two‐tailed t‐test ***P‐value = 4 × 10⁻⁴; x‐axis) expression in mESCs after 8 days of treatment with 4‐OHT (KO) relative to ethanol‐treated cells (WT) measured by RT–qPCR for three independent biological replicates (y‐axis). Transcript expression in WT and KO cells was normalized by Actin‐β and PolymeraseII expression. Horizontal dashed line represents a KO/WT fold change in expression of 1. Each point represents the ratio of one independent biological replicate.

8. H, I

   Western blot using antibodies against mouse OCT‐4 (two‐tailed t‐test P‐value = 0.880) (H) and NANOG (two‐tailed t‐test P‐value = 0.700) (I) in protein extracts from DICER conditional mESCs 8 days after treatment with ethanol (WT) or 4‐OHT (KO) for three independent biological replicates (BR1–3, same samples used in panel (D)). ACTIN‐β (ACT‐β) was used as an internal control and to determine the relative difference in OCT4 and NANOG levels represented in bar plot. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

9. J

   Expression of miR‐295‐3p and miR‐290‐3p relative to sno‐202 (x‐axis) in Dcr conditional mESCs 8 days after treatment with ethanol (WT) or 4‐OHT (KO) (y‐axis) for three independent biological replicates (Paired two‐tailed t‐test, *P‐value = 0.010 for miR‐290‐3P and 0.015 for miR‐295‐3P). Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

10. K

    Fold change (FC) in Cdkn1a, Lats2 and Rbl2 (x‐axis) expression in mESCs after 8 days of treatment with 4‐OHT (KO) relative to WT (ethanol‐treated) measured by RT–qPCR for three independent biological replicates (y‐axis). Transcript expression in WT and KO cells was normalized by Actin‐β and PolymeraseII expression. Horizontal dashed line represents a KO/WT fold change in expression of 1 (paired two‐tailed t‐test WT vs KO, **P‐value = 0.005 for Cdkn1a and 0.002 for Lats2). Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate. Each point represents the ratio of one independent biological replicate.

Data information: Statistics: NS‐P‐value > 0.05, *P‐value < 0.05, **P‐value < 0.01 and ***P‐value < 0.001. Uncropped blots used to assemble panels C, F and G are provided in Fig EV1 Source Data.

Figure 1. Steady‐state abundance of lncRNAs is not directly affected by miRNA loss

1. A

   Percentage of mRNAs (n = 6,701, red) and predominantly cytosolic (n = 57, blue) and nuclear lncRNAs (n = 175, grey) with experimental evidence of binding by ribosomes (Translation Efficiency > 0) in mESCs.

2. B

   Density of HEAP‐AGO2 peaks across cytosolic lncRNAs (n = 62, blue) and the 3’UTR regions of mRNAs (n = 8,798, red) with experimental evidence for AGO2 binding in mESCs (>0 AGO2 peaks). Central band of boxplot represents median, box depicts 25–75 quantiles of distribution, and whiskers represent the 5^th and 95^th quantiles of the distribution.

3. C, D

   Small RNA and Poly(A)‐selected RNA sequencing based estimates of the fold difference (y‐axis) in (C) miRNA and (D) mRNA expression, respectively, relative to day 0, during a 12 days’ time course (x‐axis) following treatment of DTCM23/49XY mESCs with 4‐OHT and consequent loss of DICER function. Points represent the average miRNA or mRNA expression and error bars the standard deviation based on three independent biological replicates.

4. E

   Cumulative distribution plot of the fold difference in expression after 8 days of 4‐OHT treatment for mRNAs, relative to day 0 of treatment (tpm ≥ 1) with (n = 6,034) and without (n = 7,887) evidence of AGO2‐binding by HEAP (Li et al, 2020).

5. F

   Distribution of the relative fold change after 8 days of 4‐OHT treatment in steady‐state abundance, relative to day 0 of treatment, for mESC‐expressed (tpm ≥ 1) mRNAs (n = 19,306, red), cytosolic (n = 445, blue) and nuclear (n = 529, grey) lncRNAs (two‐tailed Mann–Whitney U‐test, mRNAs vs cytosolic lncRNAs P‐value < 2 × 10⁻¹⁶ and cytosolic vs nuclear lncRNAs P = 0.875). Statistics: NS‐P‐value > 0.05 and ***P‐value < 0.001. Central band of boxplot represents median, box depicts 25–75 quantiles of distribution, and whiskers represent the 5^th and 95^th quantiles of the distribution. Two‐tailed Mann–Whitney U‐test.

Figure 2. No evidence for miRNA‐dependent destabilization of cytosolic lncRNAs

1. A

   Schematic representing 4sU metabolic labelling of conditional Dicer knockout and wild‐type cells experiment.

2. B, C

   Correlation (Pearson) between degradation rates (log₁₀) obtained after 10 (x‐axis) and 15 (y‐axis) minutes of 200 µM 4sU labelling in wild‐type (WT) and DICER null (KO) cells.

3. D

   Volcano plot showing the adjusted P‐value (y‐axis) as a function of the fold change in degradation rate estimates, based on the 10 min pulse, between KO and WT cells (x‐axis) for protein‐coding genes (red), cytosolic (blue) and nuclear (grey) lncRNAs. Each point represents a transcript and the horizontal dashed line represents the significance cut‐off.

4. E

   Cumulative distribution plot of the density of AGO2 clusters in the 3’unstralated regions of AGO2 bound mRNAs (AGO2 cluster > 0) whose degradation rates were either significantly (n = 711, red) or not significantly changed (n = 1,127, black) between KO and WT cells, based on the 10 min pulse estimates. Density of clusters presented in this analysis was estimated based on data from (Leung et al, 2011).

5. F

   Distribution of the fold change (FC) in degradation rates of mRNAs (n = 29,900, red), cytosolic (n = 474, blue) and nuclear (n = 2,348, grey) lncRNAs, in 4‐OHT‐treated (KO) relative to ethanol‐treated (WT) cells after 8 days of treatment (estimated based on the 10 min 4sU pulse), horizontal dashed line represents a KO/WT FC in degradation rate of 1 Statistics: NS‐P‐value > 0.05 and ***P‐value < 0.001. Central band of boxplot represents median, box depicts 25–75 quantiles of distribution, and whiskers represent the 5^th and 95^th quantiles of the distribution (two‐tailed Mann–Whitney U‐test).

Figure EV2. 4sU metabolic labelling in wild‐type and miRNA‐depleted mESC

1. A–C

   Principal component analysis of gene expression. The first 2 axes (PC1 and PC2) separate samples into (A) RNA fraction, newly synthesized RNA (NS, red) and pre‐existing RNA (PE blue) and (B) cell type, DICER‐depleted (KO, red) and wild‐type (WT, blue). (C) PC2 and PC3 separate biological replicates (BR1 red and BR2 blue).

2. D

   Fold change in 4sU degradation rate between KO and WT cells (X‐axis) is inversely correlated with the fold change in relative expression between KO and WT after 8 h of treatment with actinomycin‐D relative to 0 h actinomycin‐D treatment timepoint (y‐axis). Points represent the mean and standard deviation based on three independent mESC biological replicates.

3. E

   Volcano plot showing the adjusted P‐value (y‐axis) as a function of the fold change in degradation rate, estimates based on the 15 min pulse, between KO and WT cells (x‐axis) for protein‐coding genes (red), cytosolic (blue) and nuclear (grey) lncRNAs. Each point represents a transcript and the horizontal dashed line represents the significance cut‐off.

4. F

   Cumulative distribution plot of the density of AGO2 clusters in the 3’untranslated regions of AGO‐2 bound mRNAs (AGO2 cluster > 0) whose degradation rates were either significantly (red) or not significantly changed (black) between KO and WT cells, based on the 15 min pulse estimates. Density of clusters presented in this analysis was estimated based on data from (Leung et al, 2011).

5. G

   Distribution of the fold change (FC) in degradation rate of mRNAs (red), cytosolic (blue) and nuclear (grey) lncRNAs in 4‐OHT‐treated (KO) relative to ethanol‐treated (WT) cells after 8 days of treatment (estimated based on the 15 min 4sU pulse), horizontal dashed line represents a KO/WT FC in degradation rate of 1. Central band of boxplot represents median, box depicts 25–75 quantiles of distribution, and whiskers represent the 5^th and 95^th quantiles of the distribution (lower and upper whiskers, respectively). Rate inference was performed based on the results from two independent biological replicates.

6. H

   Distribution of the fold change (FC) in synthesis rate of mRNAs (red), cytosolic (blue) and nuclear (grey) lncRNAs, in 4‐OHT‐treated (KO) relative to ethanol‐treated (WT) cells after 8 days of treatment. Results for the 10‐ and 15‐min pulse are presented separately, horizontal dashed line represents a KO/WT FC in synthesis rate of 1. Central band of boxplot represents median, box depicts 25–75 quantiles of distribution, and whiskers represent the 5^th and 95^th quantiles of the distribution. Rate inference for each labelling duration timepoint was performed based on two independent biological replicates.

Figure 3. Micropeptide‐encoding transcript expression is post‐transcriptionally regulated by miRNAs

1. A

   Distribution of the translational efficiency, in mESCs, of mRNAs (n = 7,156, red), cytosolic (n = 341, blue) and nuclear (n = 1,915, grey) lncRNAs. Two‐tailed Mann–Whitney U‐test: **P‐value < 0.01 and ***P‐value < 0.001. Central band of boxplot represents median, box depicts 25–75 quantiles of distribution, and whiskers represent the 5^th and 95^th quantiles of the distribution.

2. B

   Fraction of cytosolic lncRNAs with experimental evidence for ribosomal binding with (red) or without (blue) an overlapping conserved short open reading frame.

3. C

   Distribution of the translational efficiency, in mESCs, of mRNAs (n = 7,156, red), micropeptide‐encoding transcripts (n = 43, pink) and bona fide cytosolic (n = 298, blue) and nuclear (n = 1,857, grey) lncRNAs. Two‐tailed Mann–Whitney U‐test: *P‐value < 0.05, **P‐value < 0.01 and ***P‐value < 0.001. Central band of boxplot represents median, box depicts 25–75 quantiles of distribution, and whiskers represent the 5^th and 95^th quantiles of the distribution.

4. D

   Distribution of the fold change (FC) in degradation rate of mRNAs (n = 13,296, red), micropeptide‐encoding transcripts (n = 43, pink), bona fide cytosolic (n = 759, blue) and nuclear (n = 4,299, grey) lncRNAs in 4‐OHT‐treated (KO) relative to ethanol‐treated (WT) cells after 8 days of treatment, horizontal dashed line represents a KO/WT FC in degradation rate of 1. Statistics: NS‐P > 0.05, *P‐value < 0.05. Two‐tailed Mann–Whitney U‐test: P‐value = 0.044. Central band of boxplot represents median, box depicts 25–75 quantiles of distribution, and whiskers represent the 5^th and 95^th quantiles of the distribution.

Figure EV3. Selection of lncRNA candidates to test prerequisite of target translation for miRNA‐dependent target destabilization

1. A, B

   Genome browser view of the region encompassing lncRNA‐c1 (A) or lncRNA‐c2 (B) Halo‐enhanced AGO2 pull‐down peaks (Li et al, 2020) and read density for two independent replicates (between 0–127) as well as PhyloCSF scores (between −15 and 15) in all possible reading frames are depicted. Gencode annotated genes are annotated in blue, and the candidates are annotated in black.

2. C

   Distribution of gene expression (log₁₀(TPM), x‐axis) for all mESC‐expressed transcripts. Red dotted horizontal line indicates the expression of lncRNA‐c1 and lncRNA‐c2.

3. D

   Expression of lncRNA‐c1, lncRNA‐c2, Lats2 and Cdkn1a (log₁₀TPM), measured by RNA sequencing, in two independent biological replicates of 4‐OHT‐treated (KO, triangles) and wild‐type (WT, circles) mESCs after 8 days of treatment. Each point represents the expression measured in one biological replicate. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

4. E

   Expression of lncRNA‐c1, lncRNA‐c2, Lats2 and Cdkn1a normalized to ethanol‐treated cells (WT, circles), as measured by RT–qPCR in three independent WT and 4‐OHT‐treated (KO, triangles) mESCs biological replicates following 8 days of treatment. Transcript expression was normalized by expression of Actin‐β and PolymeraseII. Each point represents the normalized expression measured in one biological replicate, horizontal dashed line represents WT‐normalized expression of 1. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

5. F

   Stability measured as the relative amount of transcript detected after 8 h of transcription block using actinomycin‐D, for lncRNA‐c1, lncRNA‐c2, Lats2 and Cdkn1a expression in 4‐OHT‐treated (KO, triangles) cells normalized to ethanol‐treated (WT, circles) cells following 8 days of treatment. Expression was measured by RT–qPCR in three independent b iological replicates of mESCs. Transcript expression was normalized by expression of Actin‐β and PolymeraseII. Each point represents the normalized expression measured in one biological replicate, horizontal dashed line represents WT‐normalized stability of 1. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

6. G

   Log₁₀ of the fold change in absolute expression in the cytosolic fraction relative to the nuclear fraction ((Cyt./Nuc. Expression), y‐axis), as measured by RT–qPCR, for lncRNA‐c1, lncRNA‐c2, nuclear (Malat1) and cytosolic (Gapdh) control transcripts. RT–qPCR analyses of four independent biological replicates of mESC cytosolic fractionation experiments were tested for GapdH, Malat1 and lncRNA‐c1. Three independent mESC biological replicates were tested for lncRNA‐c2. Each point represents the Log₁₀ ratio of expression measured in one independent biological replicate. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

7. H

   Representative Western blot analysis of protein extracts from input, AGO2‐RIP and IgG control. AGO2 was probed with rabbit AGO2 antibody (top panel). After membrane stripping and re‐probing with mouse anti‐AGO2 (middle panel) unspecific band in IgG was cleared. Probing with rabbit antibody confirmed the presence of DICER specifically in the input and AGO2‐RIP samples (lower panel).

8. I

   RT–qPCR quantification of lncRNA‐c1 and Cdkn1a (x‐axis) bound in AGO2‐IP (triangles) relative to input and unspecific IgG (circles) antibody relative to input (y‐axis) in two independent mESC biological replicates. Each point represents the IgG/INPUT % ratio measured in one biological replicate. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate. Data information: Statistics: NS‐P‐value > 0.05, *P‐value < 0.05 and **P‐value < 0.01.

Figure EV4. Candidate lncRNAs are only weakly associated with ribosomes

1. A, B

   Genome browser view of the region encompassing (A) lncRNA‐c1 or (B) lncRNA‐c2. The density of ribosomal profiling (mESC Ribo‐seq, top) and RNA sequencing reads (mESC RNA‐Seq, bottom) is represented in the (y‐axis). GENCODE annotated genes in the depicted regions are annotated in blue, and the candidate annotations are highlighted in grey.

Figure 4. Association of lncRNA‐c1 with translating ribosomes results in its miRNA‐dependent decay

1. A

   Schematic of the construct tested in WT and miRNA‐depleted mESCs.

2. B

   Expression of GFP, lncRNA‐c1, GFP‐lncRNA‐c1, lncRNA‐c2 and GFP‐lncRNA‐c2 (x‐axis) in 8 day 4‐OHT‐treated, miRNA‐depleted cells (KO) relative to ethanol‐treated (WT) mESC (y‐axis) 24h post‐transfection. Four independent biological replicates were treated, transfected and analysed by RT–qPCR. Statistical significance represented on the figure based on comparison of KO/WT fold change in expression of GFP, with GFP‐lncRNA‐c1 and with GFP‐lncRNA‐c2 (paired two‐tailed t‐test P‐value = 0.034 and 0.032, respectively) and based on comparison of KO/WT fold change in expression of GFP‐lncRNAc1 with lncRNA‐c1 and GFP‐lncRNA‐c2 with lncRNA‐c2 (paired two‐tailed t‐test P‐value = 0.012 and 0.018, respectively). Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

3. C

   Expression of GFP‐lncRNA‐c1 ΔMRE relative to GFP‐lncRNA‐c1 (y‐axis) in ethanol‐treated (WT, circles) or 4‐OHT‐treated, miRNA‐depleted cells (KO, triangles, x‐axis). Four independent biological replicates were treated, transfected and analysed by RT–qPCR. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate. Paired two‐tailed t‐test P‐value = 0.0071

4. D

   GFP levels were determined using flow cytometry in mock, GFP and BoxB(−30)‐GFP expressing cells 24 h post‐transfection. For each row, in the left panel we represent side scatter intensity (SSC‐A, y‐axis) as a function of forward scatter intensity (FSC‐A, x‐axis). The percentage of gated events (cells) is show on the lower left corner. Centre panel represents GFP intensity (x‐axis) as a function of FL2 (auto‐fluorescence, y‐axis). Percentage of GFP positive cells is shown on the lower right corner of the panel. Right panel represents the distribution of GFP fluorescence (x‐axis) as a function of the number of cells (count, y‐axis). The Area Under the Curve (AUC) is shown on the top left side of the panel.

5. E

   AUC in mESCs expressing GFP, GFP‐lncRNA‐c1 and GFP‐lncRNA‐c2 wild‐type construct (noBoxB, circles) or with a 5xBoxB cassette insertion 30 nucleotides upstream of the GFP start site (BoxB(−30), triangles). Comparison of AUC between construct with and without (BOxB(−30) paired two‐tailed t‐test P‐value = 8.12 × 10⁻⁴, 0.011 and 0.002 for GFP, GFP‐lncRNA‐c1 and GFP‐lncRNA‐c2, respectively. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

6. F

   Immunoblot analysis of GFP (GFP) in protein extracts from mESCs transfected with mock, BoxB(−30)‐GFP and GFP expressing vectors. ACTIN‐β (ACT‐β) was used as an internal control. One representative blot is depicted.

7. G

   Fold change (FC) in normalized expression of GFP‐lncRNA‐c1, GFP‐lncRNA‐c2 (noBoxB, circles) and BoxB(−30)‐GFP‐lncRNA‐c1, BoxB(−30)‐GFP‐lncRNA‐c2 (BoxB(−30), triangles; x‐axis) 4‐OHT‐treated, miRNA‐depleted mESCs (KO) relative to ethanol‐treated mESCs (WT; y‐axis). Four independent biological replicates were analysed. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate. Paired two‐tailed t‐test P‐value = 0.0494 for GFP‐lncRNA‐c1 and P‐value = 0.0355 for GPF‐lncRNA‐c2

8. H

   Immunoblot analysis of GFP (GFP) in protein extracts from mESCs transfected with mock, BoxB(+339)‐GFP and GFP expressing vectors. ACTIN‐β (ACT‐β) was used as an internal control. One representative blot is depicted.

9. I

   Fold change (FC) in normalized expression of GFP‐lncRNA‐c1, GFP‐lncRNA‐c2 (noBoxB, circles) and BoxB(+339)‐GFP‐lncRNA‐c1, BoxB(+339)‐GFP‐lncRNA‐c2 (BoxB(+339), triangles; x‐axis) in 8 day 4‐OHT‐treated, miRNA‐depleted mESCs (KO) relative to ethanol‐treated mESCs (WT; y‐axis). Four independent biological replicates analysed.

Data information: For all RT–qPCR analyses, transcript expression was first normalized by the amount of Act‐β and PolII and next by the total amount of transfected vectors per cell estimated based on the levels of relative Neomycin expression. Each point corresponds to the results of one independent biological replicate. Statistics: NS‐ P‐value > 0.05, *P‐value < 0.05, **P‐value < 0.01 and ***P‐value < 0.001. Uncropped blots used for assembly of panels (F and H) are provided in Fig 4 Source Data. Source data are available online for this figure.

Figure EV5. Relieving miRNA binding or translation levels of GFP‐lncRNA‐c1 reduces its susceptibility to miRNA‐dependent transcript destabilization

1. A

   Genome browser view of the region encompassing miR‐290 and miR‐295/4 MREs (red) within lncRNA‐c1. Halo‐enhanced AGO2 pull‐down (Li et al, 2020) read density for two independent replicates (between 0–127 is represented in the y‐axis).

2. B

   Pairwise alignment between miR‐290‐3p (top) and miR‐294‐3p (bottom) and respective predicted miRNA response elements (MRE) within lncRNA‐c1. Seed‐complementary MRE start position within annotated lncRNA‐c1 transcript is indicated inside parentheses.

3. C

   GFP‐lncRNA‐c1 expression 24 h following transfection of mESCs with 5, 15 or 30 mM mmu‐miR294‐3p inhibitors (Inhibitor, triangles) or small RNA negative control (NC, squares). Each point corresponds to the results of one technical replicate. Transcript expression was first normalized by the amount of Actin‐β and GapdH and next by the total amount of transfected vectors per cell estimated based on levels of relative Neomycin expression. Comparison between the expression in cells transfected with 5, 15 and 30 nM of Negative control or inhibitor unpaired two‐tailed P‐value = 0.006, 0.001 and 0.003, respectively. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

4. D, E

   Relative Cdkn1a (D) and Lats2 (E) expression following transfection of mmu‐miR‐294‐3p and mmu‐miR‐295‐3p equimolar mixes (Mimic) or negative control small RNA (NC). Each point corresponds to the results of one independent biological replicate. Transcript expression was first normalized by the amount of Actin‐β and PolymeraseII. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

5. F

   Fold change in expression(y‐axis) of GFP, GFP‐lncRNA‐c1 and GFP‐lncRNA‐c1‐MREΔ (x‐axis) in miRNA‐depleted cells transfected with negative control (NC) relative to miRNA‐depleted cells transfected with miRNA mimics (miRNA) (y‐axis). Each point corresponds to the results of one independent biological replicate. Transcript expression was first normalized by the amount of Actin‐β and PolymeraseII and next by the total amount of transfected vectors per cell estimated based on levels of relative Neomycin expression. Two‐tailed paired t‐test P‐value = 0.043. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

6. G

   Schematics of the BoxB insertion in GFP‐lncRNA‐candidate constructs.

7. H

   Relative expression of GFP in mESCs (y‐axis) transfected with GFP‐lncRNA‐c1, GFP‐lncRNA‐c2 (noBoxB, circles) and BoxB(−30)‐GFP‐lncRNA‐c1, BoxB(−30)‐GFP‐lncRNA‐c2 (BoxB(−30), triangles; x‐axis). Four independent mESC biological replicates were transfected and analysed by RT–qPCR. Transcript expression was first normalized by the amount of Actin‐β and PolymeraseII and next by the total amount of transfected vectors per cell estimated based on levels of relative Neomycin expression. Each point corresponds to the results of one independent biological replicate. Comparison of the relative expression in mESC of constructs with or without BoxB(−30) paired two‐tailed t‐test P‐value = 0.004 and 0.200 for GFP‐lncRNA‐c1 and GFP‐lncRNA‐c2, respectively. Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

8. I

   Relative expression of GFP in mESCs (y‐axis) transfected with GFP‐lncRNA‐c1, GFP‐lncRNA‐c2 (noBoxB, circles) and BoxB(+339)‐GFP‐lncRNA‐c1, BoxB(+339)‐GFP‐lncRNA‐c2 (BoxB(+339), triangles; x‐axis). Four independent mESC biological replicates were analysed. Transcript expression was first normalized by the amount of Actin‐β and PolymeraseII and next by the total amount of transfected vectors per cell estimated based on levels of relative Neomycin expression. Each point corresponds to the results of one independent biological replicate. Comparison of the relative expression in mESC of constructs with or without BoxB(+339) paired two‐tailed t‐test P‐value = 0.21 and 0.010 for GFP‐lncRNA‐c1 and GFP‐lncRNA‐c2, respectively). Data are represented as mean ± SD, and each point corresponds to the results of one independent biological replicate.

Data information: Statistics: NS‐P > 0.05, *P < 0.05, **P < 0.01 and ***P < 0.001 two‐tailed paired t‐test.