miR-34 and p53: New Insights into a Complex Functional Relationship

Abstract

miR-34, a tumor suppressor miRNA family transcriptionally activated by p53, is considered a critical mediator of p53 function. However, knockout of the mouse miR-34 family has little or no effect on the p53 response. The relative contribution of different miR-34 family members to p53 function or how much p53 relies on miR-34 in human cells is unclear. Here we show that miR-34a has a complex effect on the p53 response in human cells. In HCT116 cells miR-34a overexpression enhances p53 transcriptional activity, but the closely related family members, miR-34b and miR-34c, even when over-expressed, have little effect. Both TP53 itself and MDM4, a strong p53 transactivation inhibitor, are direct targets of miR-34a. The genes regulated by miR-34a also include four other post-translational inhibitors of p53. miR-34a overexpression leads to variable effects on p53 levels in p53-sufficient human cancer cell lines. In HCT116, miR-34a overexpression increases p53 protein levels and stability. About a quarter of all mRNAs that participate in the human p53 network bind to biotinylated miR-34a, suggesting that many are direct miR-34a targets. However, only about a fifth of the mRNAs that bind to miR-34a also bind to miR-34b or miR-34c. Two human cell lines knocked out for miR-34a have unimpaired p53-mediated responses to genotoxic stress, like mouse cells. The complex positive and negative effects of miR-34 on the p53 network suggest that rather than simply promoting the p53 response, miR-34a might act at a systems level to stabilize the robustness of the p53 response to genotoxic stress.

Fig 1. Overexpression of miR-34a, but not miR-34b/c, enhances p53 transcription in HCT116 cells.

(A) qRT-PCR analysis of mRNA levels of p53 transcriptional targets, normalized to GAPDH, in miR-34a or cel-miR-67 (control) overexpressing WT and TP53 ^-/- HCT116 cells. (B) Immunoblot showing protein levels of some p53 transcriptional targets. (C) Effect of miR-34 OE on Firefly luciferase reporters driven by the promoters of the p53 targets PUMA, CDKN1A and BAX or a tandem repeat of 13 p53 binding consensus sites (pG13-luc). Normalized Firefly luciferase activity, relative to Renilla luciferase activity, after miR-34 transfection is plotted as fold change relative to control miRNA-transfected sample. Alignment of the miR-34 family with the seed sequence highlighted in red is shown at top. (D) qRT-PCR analysis of p53 transcriptional target mRNAs after transfecting control or miR-34 mimics into HCT116 cells. Bar graphs show mean +/- SD of at least three independent experiments (*, p<0.05; **, p<0.01, relative to control miRNA-transfected cells, 2-tailed Student’s t-test).

Fig 2. Genome-wide transcriptome analysis of miR-34 OE HCT116 cells.

(A) Overlap of genes down-regulated ≥ 1.5 fold in miR-34 OE HCT116 cells compared to control-transfected cells. (B-D) Top enriched biological processes (p<0.05) of all down-regulated genes in miR-34 overexpressing HCT116 cells, as determined using the DAVID tool. A complete list of significantly enriched biological processes is provided in S2 Table.

Fig 3. Overexpression of miR-34a-resistant SIRT1 or MDM4 does not inhibit miR-34a-mediated p53 transcriptional activation.

(A-C) Overexpression of miR-34a resistant SIRT1 (lacking its 3’UTR, immunoblot (B)) does not reduce the miR-34a-mediated increase in the activity of a luciferase reporter driven by a promoter containing 13 p53 consensus binding sites (pG13-luc) (A) or the miR-34a-mediated increase in mRNA of the p53 transcriptional targets CDKN1A, PUMA and TP53INP1, as measured by qRT-PCR (C). (D-F) Same analysis as in (A-C) performed after co-transfecting an HA-tagged MDM4 gene with no 3’UTR and containing synonymous mutations of the miR-34a CDS MREs. All graphs show the mean +/- STDEV of at least three independent experiments (*, p<0.05; **, p<0.01, relative to control miRNA-transfected cells, by 2-tailed Student’s t-test).

Fig 4. TP53 is a direct miR-34a target.

(A) TP53 mRNA is enriched in Bi-miR-34a PDs in HCT116 cells. mRNA levels were determined by qRT-PCR and plotted as fold change relative to mRNAs pulled down with the Bi-control miRNA (Bi-ctl-miRNA). The housekeeping genes UBC and SDHA were used as negative controls. An additional control was PD of unbiotinylated miR-34a. (B) miR-34a does not affect the activity of a luciferase reporter containing the full length 3’UTR of TP53. A reporter containing the 3’UTR of MYB was used as positive control. Luciferase activity is relative to cells transfected with the control miRNA. (C) Pairing of miR-34a to predicted TP53 MREs. The miR-34a seed region is highlighted in blue, while mutations (mt) introduced in the MREs are highlighted in red. Black dashes indicate Watson-Crick base pairing and red dashes G:U base pairing. The numbers in parentheses indicate the position of the MRE in the mRNA. (D) miR-34a recognition of predicted wild-type (wt) or mutant (mt) TP53 miR-34a MREs cloned into the 3’UTR of Renilla luciferase were assessed in dual luciferase reporter assays in cells transfected with miR-34a relative to cells transfected with control miRNA. (E,F) The function of predicted TP53 miR-34a MREs (wt or mt) in their native location in full length TP53 mRNA was assessed in TP53 ^-/- HCT116 cells cotransfected with control miRNA or miR-34a mimics and wt or mt p53 cDNA. p53 protein was analyzed by p53 vs β-actin immunoblot 48 hr later. A representative blot is shown in (E) and densitometry of p53 relative to β-actin signal in 3 independent experiments in cells transfected with miR-34a relative to cells transfected with control miRNA is shown in (F). (G) miR-34a over-expression in p53-proficient HCT116 cells increases p53 protein. WT HCT116 cells were transfected with control or miR-34a mimics and protein levels were analyzed by immunoblot 48 hr post-transfection. CDK6 and β-actin immunoblots are shown as controls. (H) qRT-PCR analysis of CDK6, TP53 and CDKN1A mRNA in samples from (G). Levels are normalized to expression in control (ctl) miRNA-transfected cells. (I) miR-34a overexpression in HCT116 cells increases p53 protein stability. Pulse-chase analysis of p53 protein in HCT116 cells transfected with control or miR-34a mimics. DOX-treated HCT116 cells are a positive control. All graphs show the mean +/- STDEV of at least three independent experiments (*, p<0.05; **, p<0.01, relative to control miRNA-transfected cells, 2-tailed Student’s t-test). Immunoblots are representative of at least 3 independent experiments.

Fig 5. miR-34a overexpression differentially affects p53 levels in p53-sufficient cancer cell lines.

(A) qRT-PCR analysis of the enrichment of TP53 (top) and CDK4 (bottom) mRNAs in Bi-miR-34a PDs in tumor cell lines of different origin. PD mRNA levels were plotted as fold change relative to mRNAs pulled down with the control Bi-miRNA (Bi-ctl-miRNA). The housekeeping genes UBC and SDHA served as negative controls. (B) Immunoblot of p53 and CDK4 in cells transfected with control or miR-34a mimics relative to β-actin (top). Protein levels were quantified by densitometry and the relative ratio of protein/β-actin was normalized to the value in cells transfected with control miRNA (bottom). (C) qRT-PCR analysis of p53 transcriptional target CDKN1A, TP53INP1 and PUMA mRNAs after transfecting control or miR-34a mimics in HepG2 cells. mRNA levels of the miR-34a targets CDK4 and CDK6 are shown as controls. All experiments were performed at least 3 times and the graphs show mean +/- STDEV of replicate experiments (*, p<0.05; **, p<0.01, relative to control miRNA-transfected cells, 2-tailed Student’s t-test).

Fig 6. miR-34a-KO HCT116 cells have a normal p53 response to genotoxic stress.

(A) Sequence of the miR-34a region in chromosome 1 in miR-34a-KO HCT116 cells compared to wild-type (WT) cells. The mature miR-34a sequence is in red, with the seed sequence underlined. The bottom panel shows the Northern blot for miR-34a in WT and 34a-KO HCT116 cells. U6 is a loading control. (B) miR-34a levels by qRT-PCR in WT or 34a-KO cells, untreated or DOX-treated for 16 hr. N/D, non-detectable. (C) Induction of miR-34b and miR-34c in WT and 34a-KO HCT116 cells after DOX treatment. Cells were treated with DOX as in (B). miR-34b/c levels were analyzed by qRT-PCR. (D) Proliferation of WT, p53-KO and 34a-KO cells by MTT cell proliferation assay. (E) qRT-PCR analysis of mRNA levels of p53 transcriptional targets in WT or 34a-KO HCT116 cells treated or not with DOX for 16 hr, relative to the untreated WT sample. The inset shows a representative immunoblot for p53 and some p53 transcriptional targets. α-tubulin is a loading control. (F) Apoptosis, assessed by annexin V/PI staining, in untreated or DOX-treated (48 hr) WT or 34a-KO HCT116 cells (DOX fluorescence in the PI channel increases “PI staining” in non-apoptotic cells). Representative dot plots are at left and the mean +/- STDEV of 3 independent experiments is at right. (G) Immunoblot of miR-34a target proteins in WT and 34a-KO HCT116 cells. The numbers on the left indicate the average signal intensity, normalized to the loading control, in 34a-KO/WT cells from 3 independent experiments. All experiments were performed at least 3 times and the graphs show the mean +/- STDEV from replicate experiments (*, p<0.05; **, p<0.01, relative to WT cells, 2-tailed Student’s t-test).

Fig 7. Unimpaired p53 response in miR-34a-KO MCF7 cells.

(A) Sequence of the miR-34a region in chromosome 1 in miR-34a-KO MCF7 cells compared to WT cells. The mature miR-34a sequence is in red, with the seed sequence underlined. The bottom panel shows the Northern blot for miR-34a in WT and 34a-KO MCF7 cells. U6 is a loading control. (B) qRT-PCR analysis of mRNA levels of p53 transcriptional targets in WT or 34a-KO MCF7 cells treated or not with DOX for 16 hr. The data are plotted as the fold change normalized to the untreated WT sample. (C) Immunoblot of miR-34a target proteins in WT and 34a-KO MCF7 cells. The numbers on the left indicate the average relative signal intensity, normalized to the loading control, in 34a-KO/WT cells from three independent experiments. (D) Apoptosis, assessed by annexin V/PI staining, in untreated or DOX-treated (48 hr) WT or 34a-KO MCF7 cells. (E) Cell cycle analysis of untreated or DOX-treated (48 hr) WT or 34a-KO MCF7 cells. All experiments were performed at least 3 times and the graphs show the mean +/- STDEV of replicate experiments (*, p<0.05; **, p<0.01, relative to WT cells, 2-tailed Student’s t-test).

Fig 8. Model of miR-34a and p53 interactions.

Competing negative and positive feedback loops determine the net effect of miR-34a on p53 function. Highlighted in red are the new layers of regulation revealed by our data. Activation of p53 by cellular stress leads to transcription of miR-34 miRNAs, which in turn can enhance p53 function by: (1) miR-34a-mediated inhibition of multiple negative regulators of p53 to further increase p53 transcriptional activity; and (2) miR-34a-mediated increase of p53 protein stability (miR-34a feed-forward loops); or inhibit p53 function by: (3) direct miR-34a-mediated inhibition of TP53; and (4) direct miR-34 inhibition of many p53-activated genes (negative feedback loops). The mechanism by which miR-34a increases p53 half-life is not known, but its suppression of YY1, whose gene product is known to enhance p53-MDM2 interactions may contribute. The net effect of miR-34a on the p53 response will depend on the relative importance of these pathways, which will be determined by differences in gene expression in each cell.