Regulation of miR-200c/141 expression by intergenic DNA-looping and transcriptional read-through

Abstract

The miR-200 family members have been implicated in stress responses and ovarian tumorigenesis. Here, we find that miR-200c/141 transcription is intimately linked to the transcription of the proximal upstream gene PTPN6 (SHP1) in all physiological conditions tested. PTPN6 and miR-200c/141 are transcriptionally co-regulated by two complementary mechanisms. First, a bypass of the regular PTPN6 polyadenylation signal allows the transcription of the downstream miR-200c/141. Second, the promoters of the PTPN6 and miR-200c/141 transcription units physically interact through a 3-dimensional DNA loop and exhibit similar epigenetic regulation. Our findings highlight that transcription of intergenic miRNAs is a novel outcome of transcriptional read-through and reveal a yet unexplored type of DNA loop associating two closely located promoters. These mechanisms have significant relevance in ovarian cancers and stress response, pathophysiological conditions in which miR-200c/141 exert key functions.

Figure 1. Transcriptional regulation is crucial for miR-200c and miR-141 expression.

(a,b) Kinetics of accumulation of miR-200c (black) and miR-141 (grey; a) and their corresponding primary transcript, pri-miR-200c-141 (b), following H₂O₂ treatment in human ovarian cancer cell lines (SKOV3, OVCA433), human (WI38, MRC5) and mouse fibroblasts, as indicated. qRT–PCR data are means of fold changes (normalized to untreated and expressed as log₂)±s.d. n=3 independent experiments at least, except for OVCA433, n=2 (in this case error bars indicate range of data). Kinetics of accumulation in eight other cell lines can be found in Supplementary Fig. 1. Sequences of the pri-miR-200c-141F/R primers from mouse or human origins are given in Supplementary Table 2. (c,d) Scatter plots showing that levels of miR-200c (left) and miR-141 (right) are correlated with pri-miR-200c-141 primary transcript levels in ovarian (IGROV-1, OV90, OV2008, OVCAR3, OVCA433, RMG1, SHIN3, SKOV3) and breast (SKBr3, MDA-MB-468, MCF7, MDA-MB-231, MDA-MB-436, MDA-MB-453, BT549) cancer cell lines (c) and in HGSOC (d). qRT–PCR data are shown as normalized cycle threshold centred to the mean (ΔΔCt). Spearman correlation coefficients ρ (rho) and corresponding P-values are indicated on each graph. n=3 independent experiments per cell line, except for OVCAR3 cells, n=2. (e,f) Effect of the transcription inhibitor actinomycin D (Act D) on the levels of miR-141 (grey) or miR-200c (black) (e) and their corresponding primary transcript, pri-miR-200c-141 (f), under untreated conditions (NT) or following 3 h of H₂O₂ treatment (+H₂O₂). Cells were pretreated for 4 h with Act D or dimethylsulphoxide (control). qRT–PCR data are shown as fold change±s.d. of treated compared with untreated cells. n≥2 independent experiments per cell line.

Figure 2. pri-miR-200c-141 expression correlates with PTPN6 transcription.

(a) Schematic representation (adapted from http://genome.ucsc.edu) of miR-200c/141 genomic locus showing upstream (PTPN6) and downstream (PHB2, SCARNA12) neighbouring genes in human and mouse species, as indicated. Genomic positions, scale and genome versions used (hg19 and mm10) are indicated on the top. Horizontal arrows indicate the sense of transcription. Coding exons are indicated as blocks connected by horizontal lines representing introns, and arrowheads showing sense of transcription. Untranslated regions are presented by thinner blocks. P1 and P2 indicate the alternative promoters used for PTPN6 transcription in non-haematopoietic and haematopoietic cells, respectively. (b) Kinetics of accumulation of pri-miR-200c-141, pri-PTPN6 and pri-PHB2 primary transcripts following H₂O₂ treatment in human ovarian cancer cells (SKOV3, OVCA433), human (WI38, MRC5) and mouse fibroblasts. qRT–PCR data are means of fold changes (normalized to untreated and expressed as log₂)±s.e.m. n=3 independent experiments at least, except for OVCA433, n=2 (in this case error bars indicate range of data). Sequences of the primers from mouse or human origins (pri-miR-200c-141F/R and pri-PTPN6/Ptpn6(2) F/R) used are given in Supplementary Table 2. (c) Effect of different stresses on pri-miR-200c-141, pri-Ptpn6 and pri-Phb2 primary transcript levels. MSB, menadione sodium bisulfite. qRT–PCR data are shown as fold change±s.d. following treatments, compared with untreated fibroblasts (NT). n=3 independent experiments. (d) Scatter plot showing correlation between pri-PTPN6 and pri-miR-200c-141 primary transcript levels in ovarian and breast cancer cells. qRT–PCR data are shown as normalized cycle threshold centred to the mean (ΔΔCt). (e,f) Scatter plots showing that miR-200c (left) and miR-141 (right) levels are correlated with PTPN6 mRNA levels in NCI-60 panel of cell lines (e) and Sanger Cell Line Project (f). Spearman correlation coefficients ρ (rho) and P-values are indicated on each graph.

Figure 3. miR-200c/141 expression correlates with PTPN6 in HGSOC, independently of genomic alterations.

(a) Scatter plots showing that PTPN6 mRNA levels are correlated with miR-200c (left) and miR-141 (right) levels in ovarian tumours (Institut Curie cohort). qRT–PCR data are shown as normalized cycle threshold centred to the mean (ΔΔCt). (b) Scatter plot showing correlation between levels of PTPN6 and pri-miR-200c-141 primary transcripts in ovarian tumours (Institut Curie cohort). qRT–PCR data are shown as normalized cycle threshold centred to the mean (ΔΔCt). Sequences of the primers (pri-miR-200c-141F/R and pri-PTPN6(1) F/R) used for detection of primary transcripts are given in Supplementary Table 2. (c) Scatter plots showing that PTPN6 mRNA levels are correlated with miR-200c (left) and miR-141 (right) levels in ovarian tumours of the TCGA cohort. Microarray data were obtained from the TCGA portal. pri-miR-200c-141 levels were not available. (d,e) Correlation coefficients between PTPN6 or PHB2 mRNA levels with pri-miR-200c-141 primary transcript levels in ovarian tumours from the Institut Curie (d) and TCGA (e) cohorts. Correlations take into account either all patients (white) or patient subgroups defined according to their genomic status: amplified/deleted (grey), patients with amplifications or deletions in the region of interest; unchanged (black), patients with no genomic alteration in the region of interest. Numbers below indicate Spearman correlation coefficients ρ (rho) and corresponding P-values per gene (PTPN6 or PHB2) for each subgroup of patients in each cohort. NS, not significant.

Figure 4. The usual polyadenylation site of the PTPN6 gene is bypassed.

(a) Schematic representation (adapted from UCSC browser: https://genome.ucsc.edu) of the human genomic locus showing miR-200c/141 and their neighbouring genes, PTPN6 and PHB2. Black and grey boxes represent exons and ESTs, respectively. Polyadenylation sites (pA) are indicated (see details in Supplementary Fig. 5). Set 1, Set 2 and Set 3 primers allow detection of the Intermediate transcript by qPCR (Set 1) and by PCR (Set 2, Set 3), respectively. Sequences are given in Supplementary Table 2. Lengths of the amplified products using Set 1–3 primers are indicated in base pairs (bp) in b,d,e. (b) Kinetics of accumulation of the intermediate transcript detected using Set 1 primers upon H₂O₂ treatment in SKOV3 ovarian cancer cells and compared with pri-PTPN6 (left) or pri-miR-200c/141 (right) primary transcripts. qRT–PCR data are means of fold changes (normalized to untreated and expressed as log₂)±s.e.m. n≥3 independent experiments. (c) Correlation plot showing that RNA levels of the intermediate transcript are correlated with pri-PTPN6 (left) and pri-miR-200c-141 (right) primary transcript levels in ovarian tumours (Institut Curie cohort). (d,e) PCR reactions using Set 2 (d) and Set 3 (e) primers showing intermediate transcripts spanning from PTPN6 3′-end and reaching miR-200c/141 locus. Representative amplifications using cDNA from untreated (NT) or H₂O₂-treated (+H₂O₂) SKOV3 cells (up) and ovarian tumours (bottom). Numbers of PCR cycles performed are indicated on the top. RT- indicates control without the reverse transcriptase enzyme. The amplified fragments have been cloned and sequenced to verify that they correspond to the intermediate transcript.

Figure 5. Quantification of primary transcripts detected along PTPN6 and miR-200c/141 genomic locus.

(a) Schematic representation of the human PTPN6-miR-200c/141 genomic locus, as shown and explained in Fig. 4a. (b) Quantities of primary transcripts detected along PTPN6 and miR-200c-141 genomic locus, assessed by qPCR using cDNA from nuclear RNA of untreated (NT) or H₂O₂-treated SKOV3 cells. pri-PTPN6(1) and pri-PTPN6(2) primers detect the pri-PTPN6 primary transcript, PTPN6 dpA overlaps PTPN6 polyadenylation signal, Set 1 allows detection of the Intermediate transcript and pri-miR-200c-141b detects the pri-miR-200c-141 primary transcript. Quantifications were assessed using standard curves shown in Supplementary Fig. 6. Quantities are expressed in picogram (pg). Data are means of±s.e.m. n=6 independent experiments per cell line. P-values are from Student's t-test. (c,e) Ratios of quantities of intermediate transcript versus pri-PTPN6 primary transcript and pri-miR-200c-141/Intermediate transcript, as indicated in SKOV3 cells (c) and SKOV3+H₂O₂ (e). (d) Ratios of quantities of each primary transcript (as indicated) in SKOV3+H₂O₂ versus SKOV3 cells. P-values are from Student's t-test. Data are means n=6 independent experiments. NS, not significant.

Figure 6. PTPN6 transcription is necessary for pri-miR-200c-141 transcription.

(a) Schematic representation of the genomic deletion of 6,950 bp (grey box) targeting the PTPN6 promoter region using the CRISPR/Cas9 technology. Are represented the P1 and P2 promoters, expressed in epithelial and haematopoietic cells, respectively, as well as the position of the target-specific CRISPR guide RNAs (sgRNA). Double arrows represent the position and orientation of the primers used. (b) Table summarizing the sizes of the PCR- and qPCR-amplification products used for genotyping. (c,d) Characterization of 2 SKOV3-derived cell lines (SKOV3-Δ1, SKOV3-Δ2) among 63 individual clones tested after depletion of PTPN6 promoter genomic region using the CRISPR/Cas9 technology. (c) Fragment of 1,106 pb amplified by PCR from genomic DNA of SKOV3-Δ1 and SKOV3-Δ2 cell lines using 1F1R primers. As expected, no amplification was seen in parental SKOV3 cells. (d) qPCR-based genotyping results from genomic DNA of SKOV3, SKOV3-Δ1 and SKOV3-Δ2 cell lines. Amplified products obtained from primers described in b. Pair 2 and Pair 4 primers are specific of the deleted genomic region; fragments from 2F4R and 3F4R primers can be amplified only in case of deletion. qPCR data are shown as fold change normalized to SKOV3 for Pair2 and Pair4 primers and as normalized cycle threshold for 2F4R and 3F4R primers, as the signal is null in SKOV3 parental cells. (e) qRT–PCR data showing pri-PTPN6, PTPN6 dpA, intermediate transcript and pri-miR-200c-141 primary transcripts. Data are from SKOV3, SKOV3-Δ1 and SKOV3-Δ2 cell lines, following 3 h of H₂O₂ treatment. Data are means of fold changes (normalized to untreated and expressed as log₂)±s.e.m. n=3 independent experiments per cell line. P-values are from Student's t-test.

Figure 7. The same epigenetic marks are associated with PTPN6 and miR-200c/141 promoters.

(a) Schematic representation of PTPN6 and pri-miR-200c-141 genomic organization showing the localization of the primers (referred to as a–h) used for MeDIP and ChIP experiments. Broken arrows represent miR-200c/141 and PTPN6 promoters, PTPN6-P1 and PTPN6-P2 being active in non-haematopoietic and haematopoietic cells, respectively. +1 indicate the start sites of PTPN6 and miR-200c/141 transcription. Positions of the primers relative to each start site are indicated. Sequences of the primers are listed in Supplementary Table 2. (b) Percentage of methylated DNA, defined by MeDIP experiments using primers indicated in a in ovarian cancer cell lines characterized by high (CaOV3, IGROV-1) or low (OVCA433, SKOV3) expression of PTPN6 and miR-200c/141, as indicated. Values are means±s.e.m. n=6 independent experiments per cell line, except for CaOV3, n=2 (in this case error bars indicate range of data). (c,d) ChIP experiment using anti-H3K9Ac- (c) and H3K4me3-specific antibodies (d) in ovarian cancer cell lines, as indicated. Values are presented as percentage (%) of input DNA (amount of DNA extracted per experiment), normalized to DNA amount immunoprecipitated using anti-histone H3 antibody. Values are means±s.e.m. n=3 independent experiments at least per cell line, except for CaOV3, n=2 (in this case error bars indicate range of data).

Figure 8. PTPN6 and miR-200c/141 promoters interact with each other through a DNA loop.

(a,b) Schematic representations of miR-200c/141 genomic locus, DNA fragments generated by chromatin digestion and anchors 1 (a) and 2 (b) used for 3C experiments. On the bottom part, genes surrounding miR-200c/141-coding region are shown as grey boxes. Each vertical line, indicated beneath boxes, corresponds to the middle position of each restriction fragment analysed. The upper parts of the schemas are high magnification of PTPN6-miR-200c-141-PHB2 genomic locus. Black arrows indicate the sense of PTPN6 and PHB2 transcription. DNA restriction fragments, generated by chromatin digestion using NcoI restriction enzyme, are indicated by horizontal lines. Each vertical line, beneath the fragments, shows the middle position of each restriction fragment analysed. Are also indicated the positions of anchors 1 and 2 in miR-200c/141 promoter (a) and PTPN6 promoter (b), respectively. The red and blue shaded areas correspond to the genomic regions that interact with anchors 1 and 2, respectively. (c,d) 3C experiments from IGROV-1 ovarian cancer cells. The curves represent the relative interaction frequency of each DNA fragment with anchors 1 (c) and 2 (d), respectively. X axis represents the genomic distances from the anchor. Values are means±s.e.m. n=4 independent experiments. (e,f) 3C experiments from SKOV3 ovarian cancer cells. (g,h) 3C experiments from SKOV3 cell line exposed during 3 h to H₂O₂ treatment. (i) Quantity of each RNA entity detected along PTPN6 and miR-200c-141 genomic locus (pri-PTPN6, PTPN6dpA, intermediate and pri-miR-200c-141 primary transcripts), assessed by qPCR using cDNA from nuclear RNA of untreated SKOV3 cells, H₂O₂-treated SKOV3 cells and in IGROV-1 cells, as indicated. Quantifications were assessed using the standard curves shown in (Supplementary Fig.6a). Quantities are expressed in picogram (pg). (j,k) Ratios of quantities of Intermediate transcript versus pri-PTPN6 primary transcript and pri-miR(EST) versus pri-miR-200c-141 transcript in SKOV3+H₂O₂ and IGROV-1 cells. Data are means of±s.e.m. n=6 independent experiments per cell line. P-values are from Student's t-test. NS, not significant.

Figure 9. Model: miR-200c and miR-141 are co-transcribed with PTPN6 by two complementary mechanisms, intergenic DNA-looping and transcriptional read-through.

(a) Schematic representation of the PTPN6-miR-200c/141 genomic locus and results obtained by quantitative analyses of the different primary transcripts detected along the locus, including pri-PTPN6 primary transcript, Intermediate transcript, pri-miR-200c-141 primary transcript and pri-miR(EST) exonic isoform, corresponding to previously described ESTs. Double arrows represent the primers used for detection of the transcripts. Quantities (in picograms) of each RNA entity detected along PTPN6 and miR-200c-141 genomic locus are shown in red. Fold changes are expressed in blue. (b) The short-range intergenic DNA-looping formed by the physical interaction between PTPN6 and miR-200c/141 promoters is associated with the transcription rate of the two genes. The grey-shaded ovals indicate the genomic regions, which have been defined as interacting with each other, according to 3C experiments. In cells characterized by high basal levels of PTPN6 and miR-200c/141 transcription, the PTPN6 promoter can interact with a large region including the 3′-end of the PTPN6 gene, the miR-200c/141 promoter and downstream sequences. In cells expressing low basal levels of the two genes, the interaction is restricted to the 3′-end of PTPN6 and does not include miR-200c/141. This conformation is not altered by the increased transcription upon oxidative stress (+H₂O₂), suggesting the existence of another mechanism is responsible for the co-expression of miR-200c/141 and PTPN6 following oxidative stress. (c) The transcriptional co-regulation between PTPN6 and miR-200c/141 can also be mediated by a transcriptional read-through of the PTPN6 gene. This is highlighted by the intermediate transcripts, linking PTPN6 to miR-200c/141, we identified. These intermediate transcripts can result from alternative polyadenylation or late termination of transcription by RNA-polymerase II, as schematically represented.