miR-17-92 fine-tunes MYC expression and function to ensure optimal B cell lymphoma growth

Abstract

The synergism between c-MYC and miR-17-19b, a truncated version of the miR-17-92 cluster, is well-documented during tumor initiation. However, little is known about miR-17-19b function in established cancers. Here we investigate the role of miR-17-19b in c-MYC-driven lymphomas by integrating SILAC-based quantitative proteomics, transcriptomics and 3' untranslated region (UTR) analysis upon miR-17-19b overexpression. We identify over one hundred miR-17-19b targets, of which 40% are co-regulated by c-MYC. Downregulation of a new miR-17/20 target, checkpoint kinase 2 (Chek2), increases the recruitment of HuR to c-MYC transcripts, resulting in the inhibition of c-MYC translation and thus interfering with in vivo tumor growth. Hence, in established lymphomas, miR-17-19b fine-tunes c-MYC activity through a tight control of its function and expression, ultimately ensuring cancer cell homeostasis. Our data highlight the plasticity of miRNA function, reflecting changes in the mRNA landscape and 3' UTR shortening at different stages of tumorigenesis.

Figure 1. Mild overexpression of miR-17-19b in B cell lymphoma triggers a global proteomic response.

(a) Expression levels of MYC (left panel) and the mature forms of miR-17-19b components (right panel) were profiled in Epstein–Barr virus (EBV) positive (Daudi and Raji) and EBV negative (CA46 and Ramos) human Burkitt lymphoma cell lines, as well as in primary lymphomas and pre-tumoral (PT) B cells isolated from λ-MYC transgenic mice. Wild-type mouse splenic B cells were used as controls. (b) Scheme of the SILAC experiment. Control and miR-17-19b overexpressing cells (miR cells) were cultured in light (L) and heavy (H) media, respectively. Labelled cells were combined in 1:1 ratio and analysed by liquid chromatography–tandem mass spectrometry. In parallel, from the same samples, total RNA was prepared and analysed by microarray. Data were subjected to statistical and functional analysis. Intensity peak ratios between heavy and light peptides (H/L ratio) reflect changes in protein expression. (c) Reproducibility of the SILAC proteome in the two experimental data sets, #2567-1 and #2567-2 (d) Log₂-transformed H/L protein ratio distributions of the two functional experiments (miR cells versus control for #2567-1 in red and #2567-2 in blue) and of the control experiment (control cells versus control cells, in black) indicate a strong proteome response upon induction of the cluster, with both, up- and downregulated proteins. (e) The comparison of log₂ fold changes at protein and mRNA levels shows low correlation (r=0.51), with significantly larger protein dispersion. (See related Supplementary Figs 1 and 2).

Figure 2. 3′ UTR shortening alters the pool of miR-17-19b targets in full-blown lymphoma.

(a) Cumulative distributions of normalized protein H/L ratios (left panel) and mRNA fold changes (right panel) upon miR-17-19b overexpression, shown for non-targets (black), miR-17-19b targets predicted by TargetScan (blue), an in-house algorithm for unbiased searching of sites corresponding to 8mer (red) and 7mer-m8 seeds (green). (b) Western blot validation of a set of known miR-17-19b targets (left panel; see related Supplementary Fig. 3a). Dotted line separates targets that are confirmed (upper part) from Pten, which is not downregulated in our model (lower part). H2A.X and Vcl were used as loading controls. SILAC-based MS analysis indicates that Pten is not downregulated upon miR-17-19b overexpression (right panel). (c) Analysis of proximal poly(A) site usage in two samples from primary λ-MYC B-lymphoma cells (#2567; in blue) reveals a trend towards the usage of shorter 3′ UTRs in a full-blown lymphoma relative to B cells (in black; ENCODE project). The frequency of proximal poly(A) site usage is plotted on the x-axis, while number of transcripts is plotted on the y-axis (left panel). Examples of distal (Ccnd1 mRNA) and proximal (Gnl1 mRNA) poly(A) signal usage, which generate longer and shorter 3′ UTRs, respectively (right panel). Red arrows indicate the poly(A) sites that are used. UTR, untranslated region; ORF, open reading frame; p(A), poly(A). See related Supplementary Fig. 3b. (d) Cumulative distributions of normalized protein H/L ratios for the cluster and for miR-17 and miR-19 families, upon manual filtering for the presence of miR-17-19b seeds within 3′ UTRs. Non-targets are shown in black, miR-17-19b targets predicted by TargetScan are in blue and those predicted by the in-house algorithm for unbiased searching of sites corresponding to 8mer in red and 7mer-m8 seeds in green. (See related Supplementary Fig. 3c). (e) Log₂ fold changes of 148 significantly downregulated miR-17-19b targets at the protein level (blue, for the definition of the statistical cut-off see Supplementary Fig. 4) and mRNA levels (red, #2567-1). The dotted lines represent the cut-off values of ±1.5-fold change that set significant outliers in the transcriptome. (See related Supplementary Table 2).

Figure 3. miR-17-19b counterbalances MYC expression and function.

(a) Ingenuity pathway analysis (IPA) for the newly identified miR-17-19b targets (high and low confidence targets, Supplementary Table 2), left panel. Graphical representation of the MYC-centred regulatory network of the targets co-regulated by MYC and miR-17-19b (right panel). (b) Western blot analysis of MYC expression in control and miR cells (upper panel). Real-time PCR analysis of MYC mRNA in control and miR cells (middle panel). Histogram represents the averages from three independent experiments with error bars representing s.e.m. MYC expression is normalized to B2m mRNA. Western blot analysis of MYC stability in cycloheximide-treated control and miR cells (lower panel; Vcl is used as negative control). (c) Polysomal analysis indicates less efficient translation of MYC mRNA in miR cells relative to control. Equal amount of the control and miR cells lysates were fractionated through sucrose gradient to generate polysome profiles by measuring absorbance at 254 nm. The relative distribution of MYC and Mfge8 mRNAs (as control) on polysome gradients was assessed by RT–qPCR analysis of the RNA present in all 11 fractions, and displayed as percentage of total mRNA (% of mRNA). Arrow indicates the direction of sedimentation; 40S and 60S, small and large ribosomal subunits, respectively; 80S, monosomes; LMWP (fractions 6–8) and HMWP (fractions 9–11), low- and high-molecular weight polysomes, respectively. The polysome profile from control cells is in pink while that from the miR-17-19b-overexpressing cells is in blue. (d) Western blot analysis of MYC protein levels upon enforced expression of constructs bearing a MYC-coding region with either a 5′ UTR, 5′ and 3′ UTR or 3′ UTR regions in miR cells (left panel) and the quantification (n=3; t-test, equal variances, one-tail, *P≤0.05; middle panel) show impaired expression of the construct bearing the 3′ UTR. A schematic representation of the constructs is displayed in the right panel. (See related Supplementary Fig. 5).

Figure 4. HuR binds more MYC mRNA in miR cells.

(a) HuR-immunoprecipitations (HuR-IPs) coupled with RT–qPCR analysis revealed higher level of MYC mRNA precipitated by the HuR in miR cells compared with control. Histogram represents mean±s.e.m. (the control is set as 1; n=3; t-test, unequal variances, one-tail, **P≤0.01). (b) Western blot analysis of HuR, Ago2, MYC and Vcl (negative control) upon RNA pull-down assay. Biotinylated RNAs corresponding to the fragments B and D of MYC 3′ UTR (see Supplementary Fig. 6a) were used for pull-down assays with extracts from control and miR cells. Different exposure times were used for input and pull-down results. (c) Cytoplasmic extracts from control and miR-17-19b-overexpressing cells were subjected to IPs using anti-HuR antibody (ab). Each HuR-IP was loaded as two samples: 50% IP was probed with anti-HuR and 50% with pS/T ab. The anti-HuR ab precipitated both, the full length (35 kDa) and cleaved form 1 (CP-1, 25 kDa), of which the last is detected as predominantly phosphorylated. A mild, yet reproducible decrease of phospho-HuR was measured in miR cells when compared with control (histogram, right panel). Quantification represents mean of pS/T signal normalized to CP-1±s.e.m. (control cells are set as 1; n=3; t-test, one-tail, unequal variances, *P≤0.05). See related Supplementary Fig. 6.

Figure 5. A novel miR-17-19b target Chek2 regulates MYC expression.

(a) Chek2 is downregulated upon miR-17-19b overexpression, as detected by the SILAC proteomics and western blot analysis (left and right panels, respectively). (b) Scheme of Chek2 3′ UTR as obtained from the analysis of RNA-Seq data, with miR-17/20-binding site, proximal and distal p(A) annotated. (c) Luciferase reporter assay performed in HeLa cells for a wild-type 3′ UTR Chek2 (WT) or a mutant for the miR-17/20-binding site (left panel). The histogram represents mean±s.e.m. (two biological replicates, done in triplicate; t-test, one-tail, equal variances, **P≤0.01). Chek2 mRNA translatability was assessed using the same polysome gradients as for the MYC mRNA. Accumulation of Chek2 mRNA in sub-monosomal (40S, 60S and 80S) and LMWP fractions indicate translational repression in the cells overexpressing miR-17-19b (right panel). LMWP and HMWP, low- and high-molecular weight polysomes, respectively. (d) Increasing amounts of the Chek2-inhibitor PV1019 progressively downregulate MYC expression in primary B-lymphoma cells. Vinculin (Vcl) was used as a loading control (control cells; left panel). PV1019 time course in control and miR cells revealed stronger response in miR-17-19b-overexpressing cells, where MYC downregulation was detectable already after 3 h; H3 was used as a loading control (right panel). (e) PV1019 negatively regulates MYC expression in human BL cells (Ramos and Daudi). Western blot was used to assess levels of total and phosphorylated form of Chek2 (Chek2 and pChek2, respectively). Vinculin was used as a loading control. (f) miR-17-19b-dependent regulation of MYC requires Chek2 activity. miR cells, transiently transfected with constructs bearing a MYC-coding sequence with either a 5′- or 3′ UTR were treated with PV1019 for 3 h. H3 was used as a loading control.

Figure 6. A modest miR-17-19b induction reduces the aggressiveness of MYC-dependent Burkitt lymphoma.

(a) Phenotypic characterization of miR-17-19b-overexpressing B-lymphoma cells. In vitro growth of λ-MYC lymphoma lines (#2567 and #2646) infected with retroviruses expressing miR-17-19b or control virus (left panel). Cell cycle profiles of #2567 cells, assayed by flow cytometry after a short BrdU pulse (middle panel). Apoptosis rate in control and miR cells evaluated by caspase-3/7 activity assay (right panel). Results represent the averages±s.e.m. from three independent experiments (t-test, one-tail, equal and unequal variances for cell cycle and apoptosis, respectively, **P≤0.01; ***P≤0.001). (b,c) miR-17-19b-overexpressing cells are outcompeted by control, when co-cultured both in vitro (b) and in vivo (c). Results are represented as % of miR cells relative to the entire population, normalized to the input (see Methods for details of the experiment). For in vitro experiments, bar graphs represent the averages±s.e.m. from three independent experiments, while for in vivo experiment one sample per animal was analysed in triplicate.

Figure 7. The regulatory loop involving miR-17-92 and MYC maintains lymphoma homeostasis.

(a) A modest increase of MYC protein level restores growth rate of miR-17-19b-overexpressing cells. Growth rate of the cells is displayed as line plots, which are the mean±s.e.m. from three independent experiments. (b) Frequency of poly(A) site usage analysis for ‘regulated' and ‘non-regulated' miR-17 targets. The comparison between regulated and non-regulated targets reveals longer 3′ UTRs for the class of regulated targets, but only when they are co-regulated by MYC. A Mann–Whitney U test was used to determine statistical differences between the two distributions (*P≤0.05). (See related Supplementary Fig. 8). (c) Proposed model for miR-17-92 role in maintaining lymphoma homeostasis is schematized (modified from the study by Aguda et al.9). A ‘cancer zone' is defined by a dynamic equilibrium between proliferation and apoptosis. During B cell lymphomagenesis, miR-17-92 antagonizes MYC-induced apoptosis by downregulation of Pten, operating as an oncogene. In full-blown lymphoma, instead, miR-17-92 inhibits cell cycle progression and increases apoptosis through a tight regulation of MYC expression and function, acting as a tumour suppressor.