We performed gene targeting in ES cells using a Dicer gene replacement vector, and two Dicer alleles were recovered after Cre excision of the drug selection markers (). ES clones bearing either a Dicer allele with exons 15–17 flanked by loxP sites, or a Dicer allele deleted for exons 15–17 were isolated. Excision of these exons should result in a Dicer-null allele, as these sequences encode the PAZ domain required for Dicer recognition of pre-miRNA molecules. Furthermore, Cre deletion places the remainder of coding sequences, including those encoding the RNase III domains, in an incorrect translational frame. Blastocyst injections were performed and germline transmission of either the Dicer-null (ΔC) allele or the Dicer-conditional (C) allele was achieved ().
Figure 1. Generation of Dicerc/c mice and embryonic fibroblasts. (A) Dicer exons 15–17 flanked with loxP sites in targeted ES cells. Cre-excision generates a conditional allele (middle) and a null allele deleted for Dicer exons 15–17 (bottom). (B) (more ...)
To confirm that deletion of the floxed exons results in loss of Dicer function, we intercrossed Dicer+/ΔC
mice. Genotyping of embryos harvested from the Dicer+/ΔC
crosses at varying times during gestation indicated lethality of DicerΔC/ΔC
mice before E8.5 (Fig. S1 A, available at http://www.jcb.org/cgi/content/full/jcb.200802105/DC1
), and recovery of E7 embryos revealed that 25% (5/20) of the embryos were smaller than wild-type (wt) E7 embryos (). This phenotype is similar to that reported for a loss of function Dicer allele in mice (Bernstein et al., 2003
), confirming that deletion of the floxed region generates a Dicer-null allele.
Mice heterozygous for the Dicer-conditional allele (Dicer+/c) were crossed to produce Dicerc/c mice. The expected Mendelian ratio for each genotype was recovered, and Dicerc/c mice were indistinguishable from wt mice. In addition, multiple lines of Dicerc/c or Dicer+/c MEFs were generated from these matings.
To ablate Dicer function, Cre recombinase was transiently expressed in Dicerc/c MEFs by infecting cells with recombinant adenovirus (Ad) encoding Cre. Infection with 200 plaque-forming units of Ad-Cre failed to alter the growth of wt MEFs, yet resulted in a greater than 95% transduction of Cre activity as judged by excision of flox-stop reporter gene alleles in control Gt(ROSA)26 MEFs (unpublished data). Dicerc/c MEFs or control wt and Dicerc/+ MEFs were plated 24 h before Ad-Cre or Ad-control (Ad-βgal reporter virus) infections, and cells maintained at subconfluent densities through 12 or longer days post-induction (dpi) (). Southern analysis demonstrated Cre deletion of the floxed exons at 3 dpi (Fig. S1 B). RT-PCR performed on RNA isolated from Dicerc/c cells revealed loss of exons 15–17 and reduced amounts of Dicer mRNA, suggesting that deletion of exons 15–17 also reduced message stability (Fig. S1 C). To confirm loss of miRNA maturation, we examined the global level of mature miRNAs in Dicer-ablated MEFs. miRNA microarray data revealed that ~80% of the total mature miRNA molecules were dramatically reduced by 3 dpi, and almost all of the mature miRNA were depleted by day 5 (unpublished data). Quantitative PCR (qPCR) and Northern analysis revealed that the level of select mature miRNAs in Dicerc/c MEFs transduced with exogenous Cre was greatly reduced at 3 dpi (Fig. S1, D and E). Collectively, these results indicate that the addition of Cre to Dicerc/c MEFs efficiently depletes Dicer activity and mature miRNAs.
MEFs were infected with Ad-Cre or Ad-control (Ad-βgal) and recovered at 9 dpi. As a control, wt MEFs were also infected with Ad-Cre or Ad-control. Fluorescence-activated cell sorting (FACS) analysis of asynchronously growing MEFs revealed that Dicer-ablated MEFs were undergoing less DNA replication than Dicerc/c
MEFs infected with control virus, whereas no difference was seen in the growth rate of wt MEFs transduced with Ad-Cre or with Ad-control (). Furthermore, Dicer loss results in a reduction in S phase cells without inducing apoptosis (Fig. S2 A, available at http://www.jcb.org/cgi/content/full/jcb.200802105/DC1
Figure 2. Reduced primary cell proliferation and premature senescence in Dicer-ablated primary MEFs. (A) FACS indicates fewer Dicer-ablated cells undergo DNA replication. (B) Proliferation assay of Dicer-wt and Dicer-ablated cells reveals that Dicer loss inhibits (more ...)
To confirm the reduced growth of MEFs after Dicer ablation, a proliferation assay was performed. MEFs lacking Dicer proliferate far slower than control-infected MEFs or MEFs retaining Dicer (). Similar results were obtained in Dicerc/c
MEFs after tamoxifen-induced Dicer ablation in Dicerc/c
MEFs bearing the CAG-CreER transgene (Hayashi and McMahon, 2002
). Furthermore, induction of Cre activity in Dicerc/wt
MEFs had no effect on the proliferation of these cells, demonstrating that haploinsufficiency for Dicer does not alter cell growth ().
Although Dicer-ablated MEFs grow more slowly than wt MEFs, the Dicer-ablated cells could be passaged several times before the cells stopped dividing and assumed a large, flattened morphology. To determine if these cells were becoming senescent, Dicerc/c MEFs were either mock infected (no virus) or Ad-Cre infected, passaged once at a lower density on 3 dpi, and the media changed every third day. Mock-infected Dicerc/c MEFs displayed normal growth characteristics, whereas Dicer-ablated cells displayed flat cell morphology after 9 d in culture. Harvesting and replating cells in fresh media at higher or lower densities failed to stimulate their growth, and approximately half of the Dicer-null cells failed to attach when replated. Increasing the serum content from 10% to 20% also had no effect on Dicer-null MEFs. After 3 wk in culture, foci of cells would appear and begin to expand on the plates. However, genotyping of foci cells revealed that these were MEFs that had escaped Cre-mediated deletion of the Dicer-conditional allele (unpublished data).
To confirm that Dicer-ablated MEFs were undergoing senescence, the cells were fixed and stained for senescence-associated β-galactosidase (SA-βgal) activity (Campisi, 2003
). Little or no staining was detected in the control Dicerc/c
plates at 9 dpi, whereas SA-βgal activity was readily apparent in the Dicer-ablated cells (). Repeat experiments using triplicate lines of MEFs were performed, and the percentage of cells showing SA-βgal activity was counted in mock-infected or Ad-Cre infected Dicerc/c
MEFs (). Over 40% of the Dicer-ablated MEFs underwent senescence by day 9, compared with less that 5% senescence in nondeleted MEFs, and Dicer-ablated MEFs displayed a 10-fold increase in senescent cells by day 12. Control experiments revealed no difference in SA-βgal activity in wt MEFs at early or late times post-infection using either Ad-Cre or with Ad-control virus (Fig. S2 B).
Senescent human cells are characterized by the formation of senescence-associated heterochromatin foci (SAHF) (Dimri et al., 1995
; Narita et al., 2003
). In SAHF, chromatin is dramatically reorganized in an Rb- and p53-dependent manner, resulting in striking DAPI-dense accumulations of heterochromatin (Ye et al., 2007
). As expected, the nuclei of Dicerc/c
MEFs show DAPI-dense chromatin associated with mouse chromocenters (clustered centromeres from multiple chromosomes) before or after mock infection (, left). In contrast, Cre-mediated deletion of Dicer in Dicerc/c
cells resulted in 20–30% of the cells displaying very large aberrant clumps of coalesced heterochromatin (, right), which is observed only in 1–5% of mock-infected Dicerc/c
cells. Although SAHF has not been fully characterized in mouse cells, these DAP1-dense structures in Dicer-ablated cells are reminiscent of the SAHF observed in senescent human cells.
Senescence can be induced by oncogene activation or by DNA damage (Zhang, 2007
), and biogenesis of miRNAs in cell lines has been reported to regulate the expression of oncogenes involved in cell growth control (Johnson et al., 2005
). To examine the effects of Dicer ablation on Myc and Ras levels, we performed Westerns using Dicerc/c
MEFs infected with either Ad-Cre or Ad-control. However, little difference was observed in the levels of the Myc or Ras (). In contrast, an increase was readily detected in p53 levels in the Dicer-ablated cells.
Figure 3. Dicer ablation induces DNA damage and activates the p19Arf-p53 signaling pathway. (A) Western analysis of Ras, Myc, and p53 levels in Dicerc/c MEFs either mock infected (−) or Ad-Cre infected (+) at 6, 11, and 15 dpi. Elevated p53 is observed (more ...)
To determine if Dicer loss induced DNA damage, we performed immunofluoresence staining with an antibody to histone H2A.X. Numerous H2A.X-positive foci were seen in Dicer-ablated cells at 12 dpi, but not in control-infected Dicerc/c MEFs (). To determine if DNA damage coincides with the onset of senescence, we performed H2A.X staining on Dicerc/c MEFs and Dicer-heterozygous MEFs at various times after Ad-Cre infection. Increased DNA damage could be readily detected at 6 dpi in Dicerc/c MEFs after infection with Ad-Cre. No difference was seen in the levels of DNA damage in control-infected Dicer-heterozygous MEFs, confirming that the DNA damage in Dicer-ablated MEFs was due to loss of Dicer function ().
DNA damage–induced senescence is mediated, in part, by the p19Arf
and p53 tumor suppressors, and by the cdk inhibitor p21, a downstream mediator of p53 activity. Therefore, we examined the levels of these proteins in Dicer-ablated MEFs (). The data reveal an increase in p19Arf
levels over time, with higher levels of p19Arf
consistently found in cells lacking Dicer. Similarly, p21 protein levels are higher in Dicer-ablated MEFs, suggesting that premature senescence in Dicer-ablated MEFs may be due to induction of p19Arf
-p53-p21 signaling. In addition, activation of p53 could also be detected in Dicer-ablated MEFs as judged by p53-serine18 phosphorylation (Webley et al., 2000
). Control experiments with wt MEFs confirmed that Ad-Cre transduction alone does not alter p19 or p53ser18 phosphorylation (Fig. S2 C). Furthermore, Dicer loss resulted in increased expression of p53 target genes associated with p53-mediated cell senescence, including p21
, and CylinD2
Thus, loss of miRNA biogenesis in primary cells results in increased DNA damage, up-regulation of p19Arf-p53 signaling, p53 activation, induction of p53 target genes, and subsequent cell senescence. To confirm that MEFs lacking Dicer could undergo DNA damage–induced growth arrest, we examined the response of Dicer-ablated MEFs to adriamycin. FACS analysis in Dicer-ablated cells revealed a large reduction in S phase after DNA damage (). As expected, DNA replication in p53-null MEFs was only slightly reduced by adriamycin. In contrast, there was no difference in the growth arrest of MEFs lacking Dicer.
To confirm Dicer loss induces premature senescence by activation of the p19Arf
-p53 pathway, we bred Dicerc/c
mice with either Ink4a/Arf-null mice (Serrano et al., 1996
) or with p53-null mice (Donehower et al., 1992
) and generated Dicerc/c
, Ink4a/Arf-null MEFs and Dicerc/c
, p53-null MEFs. As expected, Dicerc/c
MEFs lacking either Ink4a/Arf or p53 proliferated faster than Dicerc/c
MEFs. Deletion of Dicer by Ad-Cre again induced senescence in a majority of Dicerc/c
MEFs by 12 dpi. However, absence of either Ink4a/Arf or p53 in Dicerc/c
cells fully rescued the Dicer-ablated MEFs from premature senescence (). Dicer-ablated cells lacking either Ink4a/Arf or p53 expanded to fill the plates and could be passaged multiple times in culture. Analysis of senescence marker gene expression was performed using total RNA isolated from wt and Ink4a/Arf-null MEFs that retained or lost Dicer. Increased p16 and p19Arf
expression and expression of the senescence-associated, p53-target genes p21
, and PAI-2
was observed in MEFs at 10 d post ablation (Fig. S2 D). Dicer ablation in Ink4a/Arf-null MEFs did not alter the expression of p21
, or PAI-2
. As p19Arf
and the p53 downstream effectors p21 and PAI-1 are major regulators of senescence in mouse cells (Sharpless et al., 2004
; Jackson and Pereira-Smith, 2006
; Kortlever et al., 2006
), these data provide genetic proof that a global reduction of miRNAs induces a p19Arf
-p53–dependent premature senescence in primary cells.
Dicer regulates cell proliferation and morphogenesis in several tissues in mice, including the developing limb (Harfe et al., 2005
) and hair follicle (Andl et al., 2006
). To confirm that Dicer prevents cell senescence in vivo, we crossed Dicerc/c
female mice with CAG-CreER transgenic Dicerc/c
males, and induced embryonic Cre activation by intraperitoneal tamoxifen injection (3 mg/40 g bodyweight) in females at day 8.5 and 11.5 of pregnancy. Embryos (n
= 26) were isolated at E14–E16, genotyped, and stained for SA-βgal activity. Cre-deletion of Dicer in CAG-CreER transgenic, Dicerc/c
embryos was confirmed by PCR (Fig. S3 A, available at http://www.jcb.org/cgi/content/full/jcb.200802105/DC1
). The results demonstrate senescence in cells of the developing limbs in 13 of 13 Cre-transgenic Dicerc/c
mice (). No senescence was detected in the 13 Dicerc/c
embryos lacking the Cre transgene, or in Cre-transgenic Dicerc/wt
heterozygous mice. In addition, we crossed Dicerc/c
mice with K5-Cre transgenic mice that express Cre in keratinocytes and in skin epidermis (Zhou et al., 2002
), and examined the follicular cells of K5-Cre+, Dicerc/c
mice for signs of senescence. All K5-Cre Dicerc/c
mice displayed loss of hair and a roughening of the epidermis after 2–3 mo of age (), in keeping with a proposed role for miRNA in the maintenance of hair follicles (Andl et al., 2006
). K5-Cre transgene function in skin was confirmed using the Gt(ROSA)26Sortm1Sor
/J mouse (Fig. S3 B). Skin was isolated from K5-Cre+, Dicerc/c
mice and from K5-Cre+, Dicer+/c
control mice, fixed, and stained for histological analysis, and for SA-βgal activity (). Analysis of genomic DNA in K5-Cre+, Dicerc/c
mice documented ablation of Dicer specifically in skin (Fig. S3 C). Fewer and much larger cells were present in the follicular epidermis of mice ablated for Dicer, and SA-βgal activity could be detected in these cells (). Collectively, these data reveal that loss of Dicer can induce a cell senescence phenotype in vivo in embryos and in adult mice, confirming that the premature senescence observed in MEFs after inhibition of miRNA maturation is not an artifact of culture.
Figure 4. Deletion of Dicer induces cellular senescence in vivo. (A) SA-βgal staining of embryos after tamoxifen-induced Dicer ablation in utero reveals that loss of Dicer promotes senescence in E16 developing limbs. No senescence is seen in Dicerc/wt embryos (more ...)
Recent reports have prompted interest in the potential role of miRNA in cancer. Overexpression of specific miRNAs has been associated with several types of human cancer (Volinia et al., 2006
; Voorhoeve et al., 2006
; Brueckner et al., 2007
). In contrast, impaired miRNA processing increased the tumorigenic potential of mouse adenocarcinoma cells in orthotopic transplantation experiments, and increased the tumor burden of mice bearing a conditional activated allele of K-Ras (Kumar et al., 2007
). These findings indicate that loss of miRNA processing and an overall reduction in miRNA enhances tumorigenesis. Very recently, the Myc oncoprotein was found to inhibit expression of select miRNAs that have anti-tumorigenic properties in lymphoma cells, offering further support for a role for miRNAs in tumor suppression (Chang et al., 2008
Our data reveal that loss of miRNA biogenesis induces a DNA damage checkpoint in certain primary cells, which promotes p19Arf
-p53–dependent cellular senescence. As p53-mediated senescence has been reported to be an important facet of p53 tumor suppression (Braig et al., 2005
; Chen et al., 2005
), loss of p53 signaling should greatly facilitate the tumorigenic potential of cells with reduced presence of mature miRNAs. Further experiments exploring the effects of p19 or p53 loss on the role of Dicer and miRNA processing in normal cell growth and in tumorigenesis are presently ongoing.