Cell cycle-targeting microRNAs promote differentiation by enforcing cell-cycle exit

Significance

The interplay between microRNAs and the cell-cycle machinery in vivo remains poorly understood. Here we report that the microRNA family miR-34/449 plays an essential and rate-limiting role in repressing cell-cycle proteins and enforcing cell-cycle exit during epithelial cell differentiation. We demonstrate that genetic ablation of the entire miR-34/449 family leads to derepression of cell cycle-promoting proteins in differentiating epithelial cells, thereby preventing their timely cell-cycle exit. This, in turn, impairs epithelial ciliation and leads to profound developmental defects. Hence, this study describes a function of the miR-34/449 family in linking cell proliferation and differentiation.

Abstract

MicroRNAs (miRNAs) have been known to affect various biological processes by repressing expression of specific genes. Here we describe an essential function of the miR-34/449 family during differentiation of epithelial cells. We found that miR-34/449 suppresses the cell-cycle machinery in vivo and promotes cell-cycle exit, thereby allowing epithelial cell differentiation. Constitutive ablation of all six members of this miRNA family causes derepression of multiple cell cycle-promoting proteins, thereby preventing epithelial cells from exiting the cell cycle and entering a quiescent state. As a result, formation of motile multicilia is strongly inhibited in several tissues such as the respiratory epithelium and the fallopian tube. Consequently, mice lacking miR-34/449 display infertility as well as severe chronic airway disease leading to postnatal death. These results demonstrate that miRNA-mediated repression of the cell cycle is required to allow epithelial cell differentiation.

MicroRNAs (miRNAs) are ∼22-nt-long RNAs that participate in numerous cellular functions by posttranscriptionally regulating gene expression. Each miRNA is capable of repressing hundreds of mRNAs through pairing to sites in the 3′ untranslated regions of target transcripts (1). Each individual target is typically repressed only modestly (2), and regulation of cellular processes frequently depends on coregulation of several genes within one particular pathway. Understanding the physiological functions of miRNAs is frequently hampered by the extensive redundancy of miRNA regulatory networks.

Several miRNAs have been described as negative regulators of cell proliferation and potential suppressors of tumorigenesis. However, the biological relevance of such an miRNA-mediated regulation of the cell-cycle machinery in vivo remains unclear. One of the best-described examples of miRNAs that target cell-cycle proteins is the miR-34 family, which includes family members miR-34a, miR-34b, and miR-34c. These miRNAs have been described as targeting multiple components of the cell-cycle machinery including cyclin D1, cyclin E2, CDK4, CDK6, CDC25A, and E2F3 (3–7), and trigger cell-cycle arrest (mainly during G1 phase) and apoptosis (4, 8–11). Furthermore, miR-34 family members are frequently down-regulated in many human cancers, suggesting a potential role of this miRNA family as tumor suppressors (12).

Although originally designated as a separate miRNA family, miR-449a, miR-449b, and miR-449c possess identical seed sequences (i.e., nucleotides 2 to 7) as members of the miR-34 family. Indeed, all these miRNAs were shown to regulate an overlapping set of targets, and can hence be considered as one miRNA family, named miR-34/449 (13). miR-449a also induces cell-cycle arrest and apoptosis in vitro, and its expression is down-regulated in prostate cancer (14).

Most attempts to understand the function of miR-34/449 family members have relied on overexpression in cultured cells. In contrast, ablation of individual members of the miR-34/449 family in mice did not uncover any essential functions in vivo (15–17), apart from minor defects in bone development (18) and improved heart function during aging (19). Combined ablation of the miR-34/449 family revealed an essential role of these miRNAs during ciliogenesis in vivo (20, 21). This function was ascribed either to the ability of the miR-34/449 family to regulate the centrosomal protein CP110, thereby affecting docking of basal bodies to the cell membrane (20), or to regulation of Notch1 signaling (22). Independent of these studies, we generated mice with combined deletion of all three genomic loci of the miR-34/449 family to investigate the physiological role of these miRNAs in vivo. Our analyses uncovered an essential function of the miR-34/449 family in controlling the balance between cell proliferation and differentiation.

Results

Ablation of All miR-34/449 Members Causes Postnatal Lethality in Mice.

We started our analyses by comparing the expression levels of all major transcripts of the miR-34/449 family in tissues from wild-type adult mice (Fig. S1A). We found that miR-34a was expressed ubiquitously, whereas miR-34b/c and miR-449a/b/c expression was restricted mainly to lungs, female reproductive organs (ovaries and fallopian tubes), and testes.

Fig. S1.

Generation and analysis of miR-34/449 TKO mice. (A) Expression levels of miR-34a (miR-34a-5p), miR-34b/c (miR-34b-5p, miR-34b-3p, miR-34c-5p, and miR-34c-3p combined), and miR-449a/b/c (miR-449a-5p, miR-449b-5p, and miR-449c-5p combined), relative to the geometric mean of the expression level of U6-snRNA and snoRNA202, in 17 organs of 8-wk-old wild-type mice (n = 3; for testis, n = 1). Mammary glands were isolated from female mice 1 d postpartum. Data are presented as mean ± SD. (B) Gene targeting strategy to generate miR-34a knockout mice (red symbols, loxP sites; green symbols, FRT sites; gray boxes, neomycin resistance gene; purple symbols, miRNA loci; orange symbols, 5′ probes). (B, Right) Southern blot analysis of DNA isolated from mouse tails and mouse embryonic stem cells, digested with BclI and hybridized with a 5′ probe to confirm successful targeting. B, BclI; N, NsiI; P, PsiI. (C) Gene targeting strategy to generate miR-34b/c knockout mice (symbols as in B). (C, Right) Southern blot analysis of DNA isolated from mouse tails and mouse ESCs, digested with BstEII, and hybridized with a 5′ probe to confirm successful targeting. Bg, BsgI; Bt, BstEII; X, XmnI. (D) Gene targeting strategy to generate miR-449a/b/c knockout mice (symbols as in B; blue symbols denote the first two exons of the Cdc20b host gene). (D, Right) Southern blot analysis of DNA isolated from mouse tails and mouse embryonic stem cells, digested with ApaLI and hybridized with a 5′ probe to confirm successful targeting. A, ApaLI; B, BsmI; Cd, Cdc20b; K, KpnI. (E) Survival of adult mice. Comparison of mice lacking all six miR-34/449 members (miR-34/449 TKO; n = 58) with wild-type mice (n = 51). Log-rank test. (F) Tumor incidence in wild-type (n = 17), miR-34 DKO (n = 39), and miR-34/449 TKO mice (n = 17) until the age of 20 mo. Benign or malignant tumors were identified macroscopically in mice that were either moribund or had reached the age of 20 mo, and confirmed by histology. ns, not significant; χ² test.

We then generated knockout strains of mice lacking miR-34a, miR-34b/c, or miR-449a/b/c (Fig. S1 B–D). None of these single-miRNA locus knockout (KO) as well as miR-34 “double-knockout” (DKO) mice displayed any apparent abnormalities.

Subsequently, we obtained “triple-knockout” (TKO) mice by interbreeding mice lacking miR-34a, miR-34b/c, and miR-449a/b/c loci. Although TKO mice were born at the expected frequency [expected, 25%; observed, 78 out of 284 (27%) newborn mice], almost 80% of mice died within the first 4 wk (Fig. 1A). Knockout neonates failed to thrive and exhibited reduced body weights (Fig. 1B), severe inflammation (with accumulation of neutrophil granulocytes and debris) in the nasal cavity (Fig. 1C), as well as opportunistic bacterial infection with gram-negative Pasteurella pneumotropica (SI Materials and Methods).

Fig. 1.

miR-34/449 TKO mice display postnatal lethality resulting from chronic airway disease. (A) Survival of mice lacking all miR-34/449 members (miR-34/449 TKO; n = 17) compared with littermates lacking miR-34a/b/c (miR-34 DKO; n = 28). Log-rank test. (B) Body weight of miR-34 TKO mice (n = 11) compared with miR-34 DKO littermates (n = 26). Data are presented as mean ± SD; ****P < 0.0001 (for days 4 to 21); two-way ANOVA with Bonferroni’s multiple comparisons test (for each day). (C) Histological sections of the nasal cavities from 2-d-old miR-34 DKO and miR-34/449 TKO littermates, stained with hematoxylin and eosin (HE). Arrows indicate accumulation of debris and neutrophils. [Scale bars, 500 µm (Left for each group) and 100 µm (Right for each group).] (D) Survival of miR-34/449 TKO mice (n = 42) compared with miR-34 DKO littermates (n = 107). Both groups were treated with the antibiotic enrofloxacin until weaning (day 21), as indicated by a green line. Log-rank test. (E) Body weight of miR-34 TKO mice (n = 9) compared with miR-34 DKO littermates (n = 17), treated as in D. Data are presented as mean ± SD; ns, not significant; two-way ANOVA with Bonferroni’s multiple comparisons test.

To confirm that the bacterial infection was responsible for the observed lethality, we treated pregnant females and newborn mice with enrofloxacin, a broad-spectrum antibiotic. Indeed, this treatment led to an almost complete rescue of postnatal lethality (Fig. 1D), as well as a rescue of the weight loss in neonatal TKO mice (Fig. 1E). Moreover, enrofloxacin-treated TKO mice that survived the first 5 wk showed essentially normal life spans (Fig. S1E). Despite reports that miR-34/449 family members have tumor-suppressive properties (12), we observed no increase in tumor incidence in TKO animals (Fig. S1F).

TKO Mice Show Defective Multiciliogenesis in Several Tissues.

We next investigated the cause of the infections in TKO mice. An important mechanism that prevents infections by pathogens in the airways is the coordinated beating of motile multicilia that cover the respiratory epithelium. Therefore, we analyzed the respiratory epithelium for the presence of multicilia. As expected, the respiratory epithelium of wild-type and miR-34 DKO mice exhibited an almost complete coverage with multicilia (Fig. 2A). In contrast, the extent of ciliation in the nasal respiratory epithelium of TKO mice was severely reduced (Fig. 2A). Similarly, we observed an absence of normal ciliation in the respiratory epithelium of the trachea, bronchi, and bronchioles (Fig. 2B). Importantly, the typical “9 + 2” microtubular structure of these motile cilia was unaltered in TKO embryos (Fig. S2A). The lack of multicilia in the respiratory epithelium most likely explains the postnatal lethality of TKO mice, since continuous beating of cilia is required for proper elimination of debris and bacteria.

Fig. 2.

Impaired multiciliogenesis in multiple tissues of miR-34/449 TKO mice. (A) Histological sections of the nasal respiratory epithelium from wild-type (wt), miR-34 DKO, and miR-34/449 TKO embryos at E18.5, stained for the cilia marker K40-acetylated α-tubulin (brown). [Scale bars, 100 µm (Upper) and 20 µm (Lower).] (A, Right) Quantification of ciliation as the percentage of epithelium covered with cilia using K40-acetylated α-tubulin–stained sections from wt (n = 3), miR-34 DKO (n = 3), and miR-34/449 TKO (n = 3) embryos. Data are presented as mean ± SD; ****P < 0.0001; ns, not significant; one-way ANOVA with Tukey’s multiple comparisons test. (B) Histological sections of the trachea (Upper), bronchus (Middle), and bronchiole (Lower) from wild-type, miR-34 DKO, and miR-34/449 TKO embryos at E18.5, stained as in A. [Scale bars, 20 µm (Upper and Lower) and 50 µm (Middle).] (B, Right) Quantification as in A. Data are presented as mean ± SD; ****P < 0.0001, ***P < 0.001, *P < 0.05; ns, not significant. (C) Histological sections of the fallopian tube epithelium from 6-wk-old wt and miR-34/449 TKO mice, stained as in A. (Scale bars, 20 µm.) (C, Right) Quantification as in A using histological sections from wt (n = 10) and miR-34/449 TKO (n = 5) mice. Data are presented as mean ± SD; ****P < 0.0001; unpaired t test. (D) Histological sections of testicular efferent ducts from 6-wk-old w-t and miR-34/449 TKO mice, stained as in A. (Scale bars, 50 µm.) (D, Right) Quantification of ciliation as the total amount of brown staining per image using K40-acetylated α-tubulin–stained sections from wt (n = 6) and miR-34/449 TKO (n = 4) mice. Data are presented as mean ± SD, relative to the mean amount of cilia staining in the wt group; *P < 0.05; unpaired t test with Welch’s correction.

Fig. S2.

Lack of miR-34/449 does not affect cilia structure, other ciliated tissues, ovulation, or cell-fate specification. (A) Representative transmission electron microscopy images of the nasal respiratory epithelium from miR-34/449 TKO and miR-34 DKO embryos at E18.5. [Scale bars, 1 µm (Upper) and 200 nm (Lower).] (B) Representative histological sections of the nasal olfactory epithelium from a wild-type and an miR-34/449 TKO embryo at the E18.5, stained for the cilia marker K40-acetylated α-tubulin (brown). (Scale bars, 20 µm.) (C) Representative histological sections of the kidney from a wild-type and an miR-34/449 TKO mouse at the age of 4 wk, stained for the cilia marker K40-acetylated α-tubulin (green fluorescence) and for nuclei (blue fluorescence). (Scale bars, 20 µm.) (D) Representative histological sections of the third ventricle of the brain from a wild-type and an miR-34/449 TKO mouse at the age of 8 wk, stained for the cilia marker K40-acetylated α-tubulin (brown). (Scale bars, 50 µm.) (E) Representative histological sections of the lateral ventricle of the brain from a wild-type and an miR-34/449 TKO mouse at the age of 8 wk, stained for the cilia marker K40-acetylated α-tubulin (brown). (Scale bars, 50 µm.) (F) Representative histological sections of the ovaries from an miR-34 DKO and an miR-34/449 TKO mouse at the age of 8 wk, stained with hematoxylin and eosin. (Scale bars, 500 µm.) Arrows point to corpora lutea. (G) Representative histological sections of the nasal respiratory epithelium from a wild-type and an miR-34/449 TKO embryo at E16.5, stained for the multiciliated progenitor cell marker FOXJ1 (red/brown). (Scale bars, 20 µm.) (G, Right) Quantification of the percentage of epithelial cells that exhibited nuclear FOXJ1 staining (FOXJ1⁺) using histological sections from wt (n = 3) and miR-34/449 TKO (n = 8) embryos. Data are presented as mean ± SD; ns, not significant; Mann–Whitney test. (H) Representative histological sections of the nasal respiratory epithelium from a wild-type and an miR-34/449 TKO embryo at E18.5, stained for mucus-producing goblet cells using Alcian blue (blue; positive cells are indicated by arrows). (Scale bars, 50 µm.) (H, Right) Quantification of the percentage of epithelial cells that exhibited Alcian blue staining (Alcian Blue⁺) using histological sections of wt (n = 3) and miR-34/449 TKO (n = 5) embryos. Data are presented as mean ± SD; *P < 0.05; unpaired t test. (I) Representative histological sections of the nasal respiratory epithelium from a wild-type and an miR-34/449 TKO embryo at E16.5, stained for the basal cell marker p63 (brown). (Scale bars, 20 µm.) (I, Right) Quantification of the percentage of epithelial cells that exhibited nuclear p63 staining (p63⁺) using histological sections of wt (n = 8) and miR-34/449 TKO (n = 8) embryos. Data are presented as mean ± SD; *P < 0.05; unpaired t test.

In addition to the respiratory epithelium, motile cilia cover the epithelial surface of several other compartments. We found that the fallopian tube epithelium of TKO females displayed a severe reduction of multicilia (Fig. 2C). We also observed a strong reduction in the amount of multicilia in the efferent ducts of the testes (Fig. 2D). In contrast, other ciliated tissues (including the multiciliated ependymal cells lining the brain ventricles and the monociliated cells of the olfactory epithelium and kidney) did not display any apparent ciliation defects (Fig. S2 B–E), consistent with the tissue-restricted expression pattern of most miR-34/449 family members (Fig. S1A).

Ablation of miR-34/449 Members Leads to Infertility in Female and Male Mice.

The observation of ciliogenesis defects in the reproductive organs prompted us to investigate the fertility of TKO mice. When mated with wild-type males, female TKO mice were unable to produce offspring. Importantly, TKO females showed unperturbed ovarian histology, with the presence of corpora lutea indicating normal ovulation (Fig. S2F). To further investigate the reason for infertility, we superovulated animals and analyzed the fate of ovulated oocytes by histology. As expected, by 2.5 d postsuperovulation, oocytes in control mice reached the ampulla of the fallopian tube, a site where fertilization takes place (Fig. 3 A and B). In contrast, in TKO females, the ovulated oocytes were trapped in the bursa of the ovary (Fig. 3 A and B). Since oocyte migration critically depends on cilia function, the loss of multicilia in the fallopian tube is most likely responsible for the infertility of female TKO mice.

Fig. 3.

miR-34/449 TKO mice exhibit defects in female and male reproductive organs. (A) Histological sections showing oocytes in the fallopian tube (ampulla) of an miR-34 DKO mouse and in the ovarian bursa of an miR-34/449 TKO mouse, 16 h after injection with human chorionic gonadotropin (hCG). Sections were stained with HE. (Scale bars, 200 µm.) Arrows point to oocytes. (B) Number of oocytes detected in histological sections of ovaries and fallopian tubes from 4-wk-old miR-34/449 TKO (n = 4) and miR-34 DKO mice (n = 5), 16 h after injection with hCG. Data are presented as mean ± SD; *P < 0.05; ns, not significant; Mann–Whitney test. (C) Histological sections of testes from 5-wk-old miR-34 DKO and miR-34/449 TKO littermates, stained with HE. (Scale bars, 100 µm.) (D) Number of sperm cells isolated from the cauda epididymis of 8-wk-old miR-34 DKO (n = 12) and miR-34/449 TKO (n = 9) mice. Data are presented as mean ± SD; **P < 0.01; Mann–Whitney test.

Cilia are also essential for maintaining male fertility. We found that seven out of eight TKO males were incapable of impregnating wild-type females. Histological analysis of testes demonstrated atrophy of the testicular tubules starting at the age of 5 wk (Fig. 3C). Furthermore, analyses of sperm content within the cauda epididymis revealed a strong reduction of sperm counts in TKO males (Fig. 3D). These observations suggest that the ciliogenesis defect in the efferent ducts of the testes compromises the movement of sperm from the testis into the epididymis, and likely contributes to infertility in TKO males.

TKO Animals Display Normal Cell-Fate Specification.

We next investigated the cellular basis of the ciliogenesis defects in TKO animals. Multicilia formation in the respiratory epithelium requires cell-fate specification into progenitors of multiciliated columnar epithelial cells, which are characterized by expression of the transcription factor FOXJ1. Staining with an anti-FOXJ1 antibody revealed an unperturbed proportion of FOXJ1-positive progenitor cells in TKO respiratory epithelium (Fig. S2G). Likewise, we observed only minor changes in the proportion of mucus-producing goblet cells and undifferentiated basal cells (Fig. S2 H and I). We concluded that the mutant epithelium undergoes normal cell-fate specification and hence the observed defects cannot be explained by the absence of cilia-producing cells.

miR-34/449 Ablation Causes Up-Regulation of Cell-Cycle Genes in the Respiratory Epithelium.

To understand the molecular basis of the defect in ciliogenesis in TKO mice, we analyzed gene expression changes in the respiratory epithelium upon miR-34/449 ablation. For analysis, we chose embryos at embryonic day 16.5 (E16.5), that is, shortly after the onset of ciliogenesis, to detect primary expression changes in the mutant epithelium. We isolated nasal respiratory epithelium from wild-type and TKO embryos using laser capture microdissection and compared global gene expression using cDNA microarrays. We detected significant expression changes of 480 mRNAs (Fig. 4A). As expected from the repressive function of miRNAs, most of these deregulated mRNAs (398 out of 480; i.e., 83%) displayed up-regulated levels upon miRNA ablation. Notably, mRNAs up-regulated in TKO epithelium were enriched for transcripts containing predicted target sites of the miR-34/449 major strand (5p) seed sequence in their 3′ UTR, indicating their direct regulation by miR-34/449 (Fig. S3 A and B).

Fig. 4.

Ablation of miR-34/449 causes up-regulation of cell-cycle genes in the respiratory epithelium. (A) A heat map showing gene expression changes in the nasal respiratory epithelium from miR-34/449 TKO (n = 5) compared with wild-type (n = 5) embryos at E16.5. Expression of 480 significantly deregulated genes (columns) across the analyzed embryos (rows) identified by cDNA microarray, using a cutoff for significance (false discovery rate < 0.05) and expression (fold change > 1.2), sorted according to supervised hierarchical clustering, with expression changes indicated as log2 values relative to the mean expression for each gene (red, up-regulated; green, down-regulated). (B) A heat map showing gene expression changes in the nasal respiratory epithelium from miR-34/449 TKO (n = 5) compared with wild-type (n = 5) embryos at E16.5, presented as in A, showing only 40 genes enriched in the Gene Ontology process “cell cycle.” Expression changes are indicated as in A. (C) RT-qPCR validation of gene expression changes for selected cell-cycle genes in the nasal respiratory epithelium from miR-34/449 TKO (n = 4) compared with wild-type (n = 4) embryos at E16.5. Data are presented as mean expression ± SD, normalized to Actb and Gapdh transcript levels, relative to the mean of the wt group; **P < 0.01, *P < 0.05; ns, not significant; unpaired t tests with multiple comparison correction using the Holm–Sidak method.

Fig. S3.

Gene expression analysis in miR-34/449 TKO epithelium. (A) Analysis of the 3′ UTR of genes (either from whole mouse genome or among genes found to be up-regulated/down-regulated in the nasal respiratory epithelium of miR-34/449 TKO embryos) for the presence of predicted target sites for the major strand (5p) of the miR-34/449 family (miR-34/449–5p) using the software TargetScan. Note that mRNAs up-regulated in TKO epithelium were enriched for transcripts containing predicted target sites of the miR-34/449 major strand (5p) seed sequence in their 3′ UTR. In contrast, no such enrichment was observed among mRNAs down-regulated in TKO embryos. ****P < 0.0001; ns, not significant; Fisher’s exact test. (B) Analysis as in A for the minor strands (3p) of the miR-34/449 family members (miR-34/449-3p). Note that no significant enrichment for predicted targets of the minor strand (3p) of the miR-34/449 family members was detected among mRNAs up-regulated or down-regulated in TKO embryos. ns, not significant; Fisher’s exact test. (C) A heat map showing gene expression changes in the nasal respiratory epithelium of miR-34/449 TKO (n = 5) compared with wild-type (n = 5) embryos at E16.5. Shown are genes involved in either lung/multiciliated epithelial cell differentiation or regulation of ciliogenesis (according to biological processes defined by Gene Ontology). Data are presented as log2 values relative to the mean expression for each gene (red, up-regulated; green, down-regulated). Asterisks indicate genes significantly deregulated in the microarray (FDR < 0.05, fold change > 1.2). (D) Validation of gene expression changes in the nasal respiratory epithelium of miR-34/449 TKO (n = 4) compared with wild-type (n = 4) embryos at E16.5 for selected genes from C by RT-qPCR. Data are presented as mean expression ± SD, normalized to Actb and Gapdh transcript levels, relative to the mean of the wt group; *P < 0.05; ns, not significant; unpaired t tests with multiple comparisons correction using the Holm–Sidak method. (E) A heat map showing gene expression changes in the nasal respiratory epithelium of miR-34/449 TKO (n = 5) compared with wild-type (n = 5) embryos at E16.5 for Notch1 target genes (according to a published dataset; ref. 24). Data are presented as log2 values relative to the mean expression for each gene (red, up-regulated; green, down-regulated). Asterisks indicate genes significantly deregulated in the microarray (FDR < 0.05, fold change > 1.2).

We next analyzed these deregulated mRNAs for enrichment of biological processes. Strikingly, the cell cycle represented the only biological process/pathway significantly enriched among deregulated mRNAs (Table S1). Indeed, 40 out of 457 deregulated mRNAs were functionally associated with the cell cycle, and most of these cell-cycle genes (37 out of 40) were up-regulated in the respiratory epithelium upon miR-34/449 ablation (Fig. 4B). We confirmed gene expression changes for 10 of these cell-cycle genes using RT-qPCR (Fig. 4C). In contrast, genes functionally associated with “lung/multiciliated epithelial cell differentiation” or “ciliogenesis regulation” were overall not deregulated, including no significant changes in mRNA levels of Ccp110, a gene implicated in negative regulation of ciliogenesis (20) (Fig. S3 C and D). Since Notch1 pathway activation is well-known to affect cell-fate specification (23), we searched our mRNA expression data for an indication of Notch1 activation. However, among 188 genes described as Notch1 target genes (24), only 5 were significantly deregulated in the respiratory epithelium of TKO embryos, indicating that Notch1 signaling is not activated upon miR-34/449 ablation (Fig. S3E).

Table S1.

Enrichment of biological processes and pathways in the miR-34/449 TKO respiratory epithelium

Category	Term	Count	%	Benjamini q value	Genes
GO BP	Cell cycle	40	8.8	2.3E-07	Aurka, Ccnb1, Ccnb2, Ccnd3, Ccndbp1, Ccne2, Cdc20, Cdca8, Cdk4, Cdk6, Cdkn2c, Cdkn3, Cenpe, Cenpf, Cenpv, Chmp1b, Ckap2, Cks2, Dbf4, Dlgap5, E2f5, E2f7, Fancd2, Fbxo43, Gas2l3, Gp1bb, Gsg2, Kif18a, Nedd9, Pcnt, Pmf1, Poln, Rad17, Sac3d1, Sgol2, Siah1a, Swt1, Syce2, Tfdp2, Zwilch
GO BP	Cell-cycle process	25	5.5	4.2E-04	Ccnb1, Ccnb2, Cdc20, Cdca8, Cdkn3, Cenpe, Cenpf, Cenpv, Cks2, Dbf4, Dlgap5, E2f5, Fancd2, Fbxo43, Gas2l3, Kif18a, Nedd9, Pcnt, Pmf1, Poln, Sac3d1, Sgol2, Siah1a, Syce2, Zwilch
GO BP	M phase	21	4.6	4.3E-04	Ccnb1, Ccnb2, Cdc20, Cdca8, Cenpe, Cenpf, Cenpv, Cks2, Dlgap5, Fancd2, Fbxo43, Kif18a, Nedd9, Pcnt, Pmf1, Poln, Sac3d1, Sgol2, Siah1a, Syce2, Zwilch
GO BP	Cell division	21	4.6	5.4E-04	Ccnb1, Ccnb2, Ccnd3, Ccne2, Cdc20, Cdca8, Cdk4, Cdk6, Cenpe, Cenpv, Chmp1b, Cks2, Nedd9, Pmf1, Poln, Sac3d1, Sgol2, Stx2, Syce2, Top2a, Zwilch
GO BP	Cell-cycle phase	23	5.0	6.1E-04	Ccnb1, Ccnb2, Cdc20, Cdca8, Cenpe, Cenpf, Cenpv, Cks2, Dbf4, Dlgap5, E2f5, Fancd2, Fbxo43, Kif18a, Nedd9, Pcnt, Pmf1, Poln, Sac3d1, Sgol2, Siah1a, Syce2, Zwilch
KEGG	Cell cycle	11	2.4	1.8E-03	Ccnb1, Ccnb2, Ccnd3, Ccne2, Cdc20, Cdk4, Cdk6, Cdkn2c, Dbf4, E2f5, Tfdp2
GO BP	Mitotic cell cycle	18	3.9	2.3E-03	Aurka, Ccnb1, Ccnb2, Cdc20, Cdca8, Cenpe, Cenpf, Cenpv, Dbf4, Dlgap5, E2f5, Kif18a, Nedd9, Pmf1, Poln, Sac3d1, Swt1, Zwilch
KEGG	p53 signaling pathway	7	1.5	2.8E-02	Ccnb2, Cdk6, Cdk4, Siah1a, Ccne2, Ccnb1, Ccnd3
GO BP	M phase of mitotic cell cycle	14	3.1	3.2E-02	Ccnb1, Ccnb2, Cdc20, Cdca8, Cenpe, Cenpf, Cenpv, Dlgap5, Kif18a, Nedd9, Pmf1, Poln, Sac3d1, Zwilch

Enrichment of biological processes and pathways was analyzed among those 480 genes deregulated in the nasal respiratory epithelium from miR-34/449 TKO compared with wild-type embryos at E16.5 (Fig. 4A) using Gene Ontology Biological Processes (GO BP) and KEGG pathways (KEGG). For each enriched pathway (term), the number of deregulated genes, their percentage among all 457 analyzed deregulated genes, significance using a Benjamini-adjusted q value, and gene symbols of the deregulated genes are shown.

miR-34/449 Ablation Increases the Levels of Cell-Cycle Proteins and Enhances Proliferation in Multiciliated Epithelia.

Given the well-established notion that cells must exit the cell cycle to start ciliogenesis, we hypothesized that increased levels of cell-cycle proteins might retain differentiating cells in a proliferative state and prevent cell-cycle exit, thereby blocking normal ciliogenesis. To test this, we analyzed protein levels of some of the up-regulated cell-cycle genes in the upper cell layer of the respiratory epithelium, which almost exclusively consists of multiciliated columnar epithelial cells. Consistent with changes in mRNA levels, we observed a strong increase of cyclin D3 and cyclin B1 protein levels in the respiratory epithelium of TKO embryos (Fig. 5A), as well as an up-regulation of cyclin B1, cyclin A2, CDC25A, N-MYC, Aurora A, and CDK1 in the fallopian tube of TKO females (Fig. 5B). In contrast, we observed no increase of CP110 (encoded by Ccp110) in the fallopian tube (Fig. 5B) or trachea of TKO mice (Fig. S4A). Collectively, these results established a widespread up-regulation of cell cycle-regulating proteins in multiciliated epithelia of TKO mice.

Fig. 5.

Ablation of miR-34/449 activates the cell-cycle machinery, resulting in abnormal proliferation. (A) Histological sections of the nasal respiratory epithelium from wild-type and miR-34/449 TKO embryos at E16.5, stained for cyclin D3 (Upper, brown) and cyclin B1 (Lower, brown). (Scale bars, 20 µm.) (A, Right) Quantification of average staining intensity per nucleus among columnar epithelial cells using histological sections from wt (n = 10) and miR-34/449 TKO (n = 10) embryos. Data are presented as mean ± SD; ***P < 0.001, *P < 0.05; Mann–Whitney test (Upper), unpaired t test with Welch’s correction (Lower). (B) Western blot analysis of fallopian tubes from 6-wk-old wild-type and miR-34/449 TKO mice, probed with antibodies against the indicated proteins. GAPDH served as a loading control. Asterisks indicate nonspecific bands. (C) Histological sections of the nasal respiratory epithelium as in A, stained for the proliferation marker Ki67 (brown). (Scale bars, 20 μm.) (C, Right) Quantification of the percentage of columnar epithelial cells that exhibited nuclear Ki67 staining (Ki67⁺) using histological sections from wild-type (n = 3) and miR-34/449 TKO (n = 8) embryos. Data are presented as mean ± SD; **P < 0.01; unpaired t test. (D) Quantification of the percentage of columnar epithelial cells with nuclear BrdU staining (BrdU⁺) using sections from wild-type (n = 8) and miR-34/449 TKO (n = 7) embryos (E16.5). Data are presented as mean ± SD; ***P < 0.001; unpaired t test. (E) Histological sections of fallopian tubes from 6-wk-old wild-type and miR-34/449 TKO mice, stained for Ki67 (brown). (Scale bars, 20 µm.) (E, Right) Quantification of the percentage of fallopian tube epithelial cells with nuclear Ki67 staining (Ki67⁺) using sections from wt (n = 10) and miR-34/449 TKO (n = 5) mice. Data are presented as mean ± SD; *P < 0.05; unpaired t test with Welch’s correction. (F) Quantification of the percentage of fallopian tube epithelial cells with nuclear PCNA staining (PCNA⁺) using sections from 6-wk-old wild-type (n = 9) and miR-34/449 TKO (n = 4) mice. Data are presented as mean ± SD; ****P < 0.0001; unpaired t test.

Fig. S4.

Ablation of miR-34/449 triggers abnormal proliferation in selected epithelial compartments. (A) Western blot analysis of the respiratory epithelium from the trachea of wild-type and miR-34/449 TKO mice at the age of 4 wk, probed with an antibody against CP110. GAPDH served as a loading control. (B) Inverse correlation between the percentage of proliferating cells and the degree of ciliation. The extent of proliferation was determined by quantification of the percentage of columnar epithelial cells that exhibited nuclear Ki67 staining (Ki67⁺) using histological sections of wild-type (n = 3), miR-34 DKO (n = 8), and miR-34/449 TKO (n = 8) embryos (data as in Fig. 5C). The degree of ciliation in the same embryos was defined as the percentage of the epithelium covered with cilia using histological sections of wt (n = 3), miR-34 DKO (n = 3), and miR-34/449 TKO (n = 3) embryos, stained with the cilia marker K40-acetylated α-tubulin. The linear regression curve depicts the correlation. r, Pearson correlation coefficient (with P value). (C) Representative histological sections of the small intestinal epithelium from a wild-type and an miR-34/449 TKO embryo at E18.5, stained for the proliferation marker Ki67 (brown). Note that ablation of miR-34/449 did not affect the proliferation in the intestinal epithelium. (Scale bars, 50 µm.)

We next asked whether the increased levels of cell-cycle proteins led to unscheduled proliferation of epithelial cells in TKO mice. Indeed, the fraction of proliferating cells was significantly elevated in the respiratory and fallopian tube epithelium of TKO mice, as revealed by Ki67 staining, proliferating cell nuclear antigen (PCNA) staining, as well as BrdU incorporation (Fig. 5 C–F). Moreover, the percentage of proliferating cells inversely correlated with the extent of ciliation (Fig. S4B). In contrast, ablation of miR-34/449 members did not affect the proliferation in the intestinal epithelium (Fig. S4C). Taken together, these analyses revealed that mice lacking the miR-34/449 family display elevated levels of several cell cycle-regulating proteins in specific epithelial compartments, thereby retaining epithelial cells in a proliferative state and preventing their exit from the cell cycle.

Increased Activity of Cyclin-Dependent Kinases Is Responsible for the Multiciliogenesis Defect in TKO Mice.

We hypothesized that the inability of TKO respiratory epithelial cells to normally exit the cell cycle, due to an up-regulation of cell-cycle proteins, represented the direct cause of the ciliogenesis defect. To test this, we asked whether inhibition of the hyperactivated cell-cycle machinery might correct the phenotypic abnormalities seen in TKO animals. We treated newborn TKO mice with a panel of inhibitors of cyclin-dependent kinases (CDKs): ribociclib (which inhibits CDK4 and CDK6), roscovitine (which inhibits CDK1, CDK2, and CDK7), and R547 (which inhibits CDK1, CDK2, and CDK5). In the case of ribociclib, we confirmed that administration of this inhibitor to TKO animals extinguished aberrant cell proliferation in the respiratory epithelium (Fig. 6A and Fig. S5A). Prolonged treatment with each of these inhibitors fully corrected the phenotypic abnormalities seen in TKO animals (Fig. 6B). We found that the extent of ciliation in the respiratory epithelium of inhibitor-treated TKO mice was identical to that of untreated wild-type mice (Fig. 6B). We concluded that an increased CDK activity resulting from increased levels of cyclins and CDKs represented the direct cause of the ciliogenesis defect in TKO mice. Collectively, our results reveal that members of the miR-34/449 family play an essential and rate-limiting role during epithelial differentiation, by repressing the cell-cycle machinery and promoting cell-cycle exit (Fig. S5C).

Fig. 6.

CDK inhibition rescues the ciliogenesis defect in miR-34/449 TKO mice. (A) Histological sections of the nasal respiratory epithelium from wild-type and miR-34/449 TKO mice treated for 48 h with PBS, and miR-34/449 TKO mice treated for 48 h with ribociclib, stained for the proliferation marker Ki67 (brown). (Scale bars, 20 µm.) (A, Right) Quantification of the percentage of columnar epithelial cells with nuclear Ki67 staining (Ki67⁺) using sections from at least three mice per group. Data are presented as mean ± SD; ***P < 0.001, *P < 0.05; ns, not significant; one-way ANOVA with Tukey’s multiple comparisons test. (B) Histological sections of the nasal respiratory epithelium from wild-type and miR-34/449 TKO mice treated for 7 d with PBS, as well as miR-34/449 TKO mice treated for 7 d with the CDK inhibitors ribociclib, roscovitine, or R547. Sections were stained for the cilia marker K40-acetylated α-tubulin (brown). (Scale bars, 50 µm.) (B, Right) Quantification of ciliation as the percentage of epithelium covered with cilia using sections from at least eight mice per group. Data are presented as mean ± SD; ****P < 0.0001; ns, not significant; one-way ANOVA with Tukey’s multiple comparisons test.

Fig. S5.

CDK inhibition rescues the ciliogenesis defect in miR-34/449 TKO mice. (A) Representative histological sections of the nasal respiratory epithelium from wild-type and miR-34/449 TKO mice treated for 48 h with PBS, as well as miR-34/449 TKO mice treated for 48 h with the CDK inhibitor ribociclib. Mice were injected with BrdU 6 h before tissue collection. Sections were stained for BrdU incorporation (brown). (Scale bars, 20 µm.) (A, Right) Quantification of the percentage of columnar epithelial cells that exhibited nuclear BrdU staining (BrdU⁺) using histological sections from at least three mice per group. Data are presented as mean ± SD; **P < 0.01; ns, not significant; one-way ANOVA with Tukey’s multiple comparisons test. (B) A model summarizing the changes detected in mice lacking all miR-34/449 family members and their contribution to the observed phenotypes.

SI Materials and Methods

Animals.

For the in vivo animal studies presented here, we used mice (Mus musculus) that were generated from V6.5 embryonic stem cells (ESCs), a murine ESC line with a mixed C57BL/6 and 129/Sv background. Chimeric mice from this mixed background were then backcrossed once with C57BL/6J mice. Therefore, the mice used in this study were of a mixed genetic background. Unless otherwise stated, we used both male and female embryos and mice for the experiments, and made sure that the fraction of females and males was very similar in experimental and control groups. We used embryos and mice at various stages of development and age, which is indicated in the figure legends for the respective experiments. All animals were held in the same animal room in individually ventilated cages with a 12-h day/night cycle. Except for the first three experiments (shown in Fig. 1 A–C), all animals were treated with antibiotics (for details, see Treatment of Animals) and housed with “Enrich-n’Pure” bedding and “Enviro Dri” nesting material (to reduce dust formation) to improve survival of miR-34/449 TKO mice, which exhibited severe chronic airway disease. Mice were not used for any previous procedures. For control groups, we used miR-34 DKO littermates, which were housed together with the experimental miR-34/449 mice, as well as wild-type mice (of the identical genetic background), which were bred separately. All animal experiments were approved by the Institutional Animal Care and Use Committee of the Dana-Farber Cancer Institute and were carried out in accordance with approved protocols and regulatory standards.

Generation of Constitutive Knockout Mice.

Targeting constructs were assembled by inserting loxP sites ≈500 bp upstream and downstream of the miRNA loci for each of the three miRNA loci. Targeting constructs harbored a selectable marker, neomycin resistance (Neo) cassette, flanked by FRT recombination sites. Targeting constructs were introduced into V6.5 ES cells, and heterozygous ES cells were derived through homologous recombination and verified by Southern blotting using 5′ probes and 3′ probes. Recombinant ES cell clones were electroporated with Cre recombinase to excise the Neo cassette and the miRNA locus, and verified by Southern blotting using 5′ probes (Fig. S1 B–D).

ES cells with heterozygous deletion of the miRNA locus were then injected into blastocysts derived from C56BL/6J females and implanted into pseudopregnant foster females. The resulting male chimeric mice were then crossed with C57BL/6J females to obtain germline transmission. Deletion of the miRNA loci in these heterozygous mice was confirmed in DNA from tail biopsies by PCR and Southern blot using 5′ probes (Fig. S1 B–D).

Genotyping of mice was performed by PCR from tail biopsies using specific primer combinations for miR-34a (wt allele, 160 bp; KO allele, 270 bp), miR-34b/c (wt, 210 bp; KO, 320 bp), and miR-449 (wt, 250 bp; KO, 360 bp).

Generation of miR-34/449 TKO and Control Mice.

To obtain mice with constitutive deletion of all three miRNA loci of the miR-34/449 family (miR-34/449 TKO), we intercrossed the three knockout strains. This way, we generated mice that only retained one wild-type allele of the miR-34/449 family (i.e., miR-34a^−/−;miR-34b/c^−/−;miR-449a/b/c^+/−). These mice were used to set up breedings that generated both miR-34/449 TKO offspring (i.e., miR-34a^−/−;miR-34b/c^−/−;miR-449a/b/c^−/−) and miR-34 DKO offspring, which were used as littermate control animals in some of the experiments (as indicated).

To obtain mice that were wild-type for all three miRNA loci, we also intercrossed the three single-knockout strains and generated control mice with the same genetic background as miR-34/449 TKO mice but wild-type for all alleles of the miR-34/449 family (i.e., miR-34a^+/+;miR-34b/c^+/+;miR-449a/b/c^+/+). These wild-type mice did not show any differences from the miR-34 DKO mice, and hence both genotypes were used as controls in this study.

Mouse Survival, Weight, Bacteriology, and Tumor Analysis.

For analysis of survival and body weight of newborn miR-34/449 TKO mice and control mice (miR-34 DKO littermates), we monitored these mice every day (in the morning) until the age of 35 d, assessed their overall health, and determined their body weight. Mice that showed signs of morbidity were euthanized and scored as death event for survival analysis.

For analysis of tumor incidence in aging mice (wild type, miR-34 DKO, and miR-34/449 TKO), we monitored a cohort of adult mice starting at the age of 6 wk, three times per week, and assessed their overall health. Mice that showed signs of morbidity were euthanized, and upon a gross examination of all organs any abnormal organs were isolated and processed for histological analysis. Histological analysis was performed by an experienced pathologist (R.T.B.) to confirm the presence of benign or malignant tumors. In addition, all mice in this cohort that remained healthy until the age of 20 mo were also euthanized and analyzed in the same way to identify the occurrence of tumors.

For bacteriological analysis, we obtained nasal swabs from four TKO mice at the age of 4 d. These were sent for analysis to Charles River Research Animal Diagnostic Services. Upon aerobic culture on various agar types, two out of four samples showed bacterial growth on tryptose blood agar. These bacterial colonies were identified as Pasteurella pneumotropica Heyl strain by PCR.

For analysis of survival in aging mice, we monitored a cohort of mice for tumor incidence analysis, as well as an additional cohort of adult mice starting at the age of 6 wk, three times per week, and assessed their overall health. Mice that showed signs of morbidity were euthanized, and this day was scored as death event for survival analysis.

Treatment of Animals.

Mating pairs and all newborn mice (miR-34/449 TKO as well as all other control mice, until weaning at the age of 21 d) were treated with enrofloxacin. Enrofloxacin was administered in the drinking water at a concentration of 0.12 mg/mL, leading to an approximate intake of 0.6 mg/d for an adult mouse (assuming 5-mL water consumption). Upon occurrence of opportunistic infections resistant to enrofloxacin (by Staphylococcus xylosus), we changed the antibiotic to amoxicillin, another broad-spectrum antibiotic, which was included in the diet at a concentration of 0.2%.

For superovulation of female mice, we injected 4-wk-old female mice intraperitoneally with 6.25 IU pregnant mare’s serum gonadotropin (PMSG; 250 µL of a 25 IU/mL solution in 0.9% NaCl). This was followed by an i.p. injection of 6.25 IU hCG (250 µL of a 25 IU/mL solution in 0.9% NaCl) 48 h after PMSG. Mice were then euthanized 16 h after hCG injection, and ovaries and fallopian tubes were processed for histological analysis.

For labeling of proliferating cells in embryos, we injected pregnant females intraperitoneally with 5-bromo-2′-deoxyuridine (BrdU) at a dose of 60 mg/kg body weight. For labeling of proliferating cells in neonatal mice, we injected mice intraperitoneally with BrdU at a dose of 100 mg/kg body weight. Six hours after BrdU injection, pregnant females or neonatal mice were euthanized, embryos were isolated and euthanized, and samples were processed for histological analysis.

For inhibition of CDK activity in neonatal mice, we treated newborn mice for 48 h, starting on the day of birth (for analyses of proliferation, shown in Fig. 6A and Fig. S5A), or for 7 d, starting 1 d after birth (for analyses of ciliation, shown in Fig. 6B), by daily i.p. injection with one of the following CDK inhibitors: ribociclib (also known as LEE011; 200 mg/kg body weight, by injection of 25 mL/kg body weight using a solution with 8 mg/mL ribociclib in PBS; Novartis), roscovitine (R-1234; 5 mg/kg body weight, by injection of 25 mL/kg body weight using a solution with 0.2 mg/mL roscovitine in 30% polyethylene glycol 300, 1% Tween 80, in PBS; LC Laboratories), and R547 [1983; 2 mg/kg body weight, by injection of 25 mL/kg body weight using a solution with 0.08 mg/mL R547 in 0.5% (hydroxypropyl)methyl cellulose, 0.2% Tween 80, in PBS; Axon Medchem]. For control, we treated newborn mice in the same way by daily i.p. injection with PBS (25 mL/kg body weight).

Histological Sample Processing.

For histological analysis and subsequent immunohistochemistry staining, whole embryos or individual mouse organs were fixed in 10% phosphate-buffered formalin solution for 24 h at 4 °C, transferred into 70% ethanol, and stored at 4 °C. For histological analysis of the brain or kidneys, a cardiac perfusion with cold PBS followed by cold 10% phosphate-buffered formalin solution was performed, followed by fixation in 10% phosphate-buffered formalin solution for 24 h at 4 °C, transfer into PBS, and storage at 4 °C. In cases where only histological analysis without subsequent immunohistochemistry staining was performed, mouse organs were fixed in Bouin’s solution for 3 d at room temperature, transferred into 70% ethanol, and stored at 4 °C. Subsequently, tissues were processed and embedded in paraffin (by the Rodent Histopathology Core Facility, Harvard Medical School); 4- to 5-µm paraffin sections were stained with hematoxylin and eosin or used for specific immunohistochemistry staining (see section below).

Alcian Blue Staining.

For identification of mucus-secreting goblet cells, formalin-fixed paraffin-embedded tissue sections were deparaffinized and incubated in Alcian blue solution (pH 2.5) for 30 min at room temperature. The slides were then washed with tap water and counterstained with Vector Nuclear Fast Red (H-3403; Vector Laboratories), according to the manufacturer’s instructions.

Immunohistochemistry Staining.

Formalin-fixed paraffin-embedded tissue sections were deparaffinized, boiled with 10 mM sodium citrate (pH 6.0) for 10 min, and treated with 3% H₂O₂ for 10 min. In the case of staining for BrdU, sections were incubated in 1 M HCl for 10 min on ice, 2 M HCl for 10 min at room temperature, 2 M HCl for 20 min at 37 °C, and 0.1 M sodium tetraborate (pH 9.0) for 10 min at room temperature. For staining with mouse primary antibodies, blocking and antibody incubations were performed according to the manufacturer’s instructions using the Vector M.O.M. Immunodetection Kit (BMK-2202; Vector Laboratories). For staining with rabbit primary antibodies, blocking was performed with 10% goat serum [in Tris-buffered saline with 0.1% Tween 20 (TBS-T) with 1% BSA], followed by incubation with the primary antibody (diluted in TBS-T with 5% goat serum and 1% BSA) overnight at 4 °C, and incubation with a biotinylated secondary antibody (diluted in TBS-T with 1% BSA) for 30 min at room temperature. We used the following primary antibodies: K40-acetylated α-tubulin (T6793; 1:10,000; Sigma-Aldrich), cyclin D3 (sc-182; 1:500; Santa Cruz Biotechnology), cyclin B1 (SAB4503501; 1:500; Sigma-Aldrich), Ki67 (VP-RM04; 1:500; Vector Laboratories or 12202; 1:400; Cell Signaling Technology), BrdU (347580; 1:100; BD Biosciences), PCNA (sc-56; 1:5,000; Santa Cruz Biotechnology), FOXJ1 (14-9965; 1:200; eBioscience), and p63 (sc-25268; 1:200; Santa Cruz Biotechnology). Subsequently, antibody binding was visualized using the Vectastain ABC Kit (PK-4000; Vector Laboratories) and ImmPACT DAB substrate (SK-4105; Vector Laboratories), and counterstained using Vector Hematoxylin QS (H-3404; Vector Laboratories), all according to the manufacturer’s instructions. Sections were then dehydrated and mounted using Permount. In the case of staining for FOXJ1, sections were also stained with Alcian blue and counterstained with Hematoxylin QS, leading to a more reddish appearance of the DAB substrate.

Immunohistochemistry/Immunofluorescence Staining of Kidneys.

Formalin-fixed paraffin-embedded kidney sections were deparaffinized and boiled with 10 mM sodium citrate (pH 6.0) for 10 min. Blocking and primary antibody incubation were performed according to the manufacturer’s instructions using the Vector M.O.M. Immunodetection Kit (BMK-2202; Vector Laboratories). We used the following primary antibody: K40-acetylated α-tubulin (T6793; 1:5,000; Sigma-Aldrich). Incubation with an Alexa Fluor 488-conjugated secondary antibody (diluted in TBS-T with 1% BSA) was performed for 60 min at room temperature. Nuclei were stained by incubation with Hoechst 33342 (0.5 µg/mL) for 5 min at room temperature. Sections were then mounted using the water-soluble mounting medium Fluoromount-G. Pictures were taken using a Nikon Eclipse E600 microscope with suitable filters for the green fluorescence of Alexa Fluor 488 (FITC-HYQ filter, EX460-500, DM505, BA510-560) and the blue fluorescence of Hoechst 33342 (UV-2A filter, EX330-380, DM400, BA420).

Quantification of Immunohistochemistry Staining.

The investigators performing the quantification of immunohistochemistry stainings were blinded to the identity of the samples. For analysis of the fraction of Ki67-, BrdU-, PCNA-, FOXJ1-, Alcian blue-, or p63-positive cells, the quantification was done by manual counting. For analysis of the nuclear protein amount (e.g., cyclin D3 or cyclin B1), the ratio of red color intensity (DAB, indicating staining for the protein of interest) to blue color intensity (hematoxylin, indicating no staining for the protein of interest) within 30 nuclei per sample was determined using the software ImageJ (version 1.46r; National Institutes of Health). For analysis of ciliation, the cumulative length of the epithelial surface covered by cilia (detected by brown staining for K40-acetylated α-tubulin) and the total length of the epithelial surface were measured with ImageJ and used to calculate the percentage of epithelium covered with cilia.

Electron Microscopy.

For structural analysis of the cilia in the nasal respiratory epithelium, we euthanized embryos (E18.5) and immediately fixed the heads in a fixation solution containing 1.25% paraformaldehyde, 2.5% glutaraldehyde, 0.03% picric acid, and 0.1 M cacodylate buffer (pH 7.4) for 20 h at 4 °C. Subsequently, slices from the region of the head encompassing the nasal cavity were cut with a razor blade, incubated for at least 2 h at room temperature in the above fixation solution, washed in 0.1 M cacodylate buffer, incubated in a solution with 1% osmium tetroxide (OsO₄) and 1.5% potassium ferrocyanide (KFe[CN]₆) for 1 h, washed three times in water, and incubated in 1% aqueous uranyl acetate for 1 h, followed by two washes in water and subsequent dehydration in grades of alcohol. The samples were then transferred into propylene oxide for 1 h and infiltrated overnight in a 1:1 mixture of propylene oxide and TAAB Epon. The next day, the samples were embedded in TAAB Epon and polymerized at 60 °C for 48 h. Ultrathin sections (about 80 nm) were cut on a Reichert Ultracut-S microtome and stained with lead citrate. Samples were examined using a JEOL 1200EX transmission electron microscope at 10,000× or 50,000× magnification, and images were recorded with an AMT 2k CCD camera. This procedure was performed at the Harvard Medical School Cell Biology Conventional Electron Microscopy Facility.

Laser Capture Microdissection and Microarray Analyses.

To analyze gene expression changes specifically in the respiratory epithelium, we used laser capture microdissection (LCM). We prepared sequential 8-µm sections of formalin-fixed paraffin-embedded samples from the region of the head encompassing the nasal cavity of embryos at E16.5. Initially, 4-µm sections were stained with hematoxylin and eosin, and the region of interest for microdissection was annotated. Subsequently, we prepared up to 10 sections (8-µm-thick) on polyethylene naphthalate membrane slides for each sample and microdissected the respiratory epithelial cell layer using an Arcturus instrument.

RNA was then extracted using the AllPrep DNA/RNA FFPE Kit (80234; Qiagen). Quality control was performed using the Bioanalyzer Pico Chip (Agilent). RNA was quantitated by RiboGreen assay and used for gene expression analysis by cDNA microarray as well as by RT-qPCR (see section below).

For cDNA microarray analysis, we used 33 ng of RNA from five wild-type and five miR-34/449 TKO embryos (E16.5). cDNA was prepared and labeled using the SensationPlus FFPE Amplification and WT Labeling Kit (902042; Affymetrix). cDNA was then hybridized onto Mouse Gene v2.0 ST array chips (90119; Affymetrix) in an Affymetrix Hybridization Oven 645 at 45 °C for 16 h. The chips were washed on an Affymetrix Fluidics Station 450 and scanned on a 7G GeneChip Scanner.

RNA Isolation and RT-qPCR.

For verification of gene expression changes in the respiratory epithelium of miR-34/449 TKO embryos, we used RNA from LCM-isolated tissue (see section above). We then prepared cDNA using the High Capacity cDNA Reverse Transcription Kit (4374966; Applied Biosystems) and performed real-time qPCR with 0.2 ng cDNA and specific primers (Table S2) using Power SYBR Green PCR Master Mix (4367659; Applied Biosystems) and an ABI 7300 real-time cycler, according to the manufacturer’s instructions.

Table S2.

Primer sequences for genotyping and RT-qPCR

Primer name	Primer sequence
Primer for genotyping of knockout mouse strains
miR-34a fwd	CTCAGCTAGGGGTGGGACT
miR-34a rev1	TCAGTGGCTGTGACTGTATTCAT
miR-34a rev2	ACCTGCAAGGTGACTCACAA
miR-34b/c fwd	AGCTCTTGTTGGCACACCTC
miR-34b/c rev1	TCTCACTTTGGGCTCCTGTT
miR-34b/c rev2	TGTCGAGGTCACAATTTTCG
miR-449a/b/c fwd	GGCATTTGCAGGAATTATGG
miR-449a/b/c rev1	CTTTGCATGCTTCTGAGTGC
miR-449a/b/c rev2	CAGCAACACCAAGGATGAGA
Primer for RT-qPCR
Ccnd3 fwd	GGCCCTCTGTGCTACAGATTA
Ccnd3 rev	GTGGCGATCATGGATGGAGG
Ccne2 fwd	GGAGACGTTCATCCAGATAGCTC
Ccne2 rev	AGTCGATGGCTAGAATGCACAG
Ccnb1 fwd	ATGATGGGGCTGACCCAAAC
Ccnb1 rev	TCCAGTCACTTCACGACCCT
Ccnb2 fwd	AACCCACAGCCTCTGTGAAA
Ccnb2 rev	CTTGCAGAGCAGAGCATCAGA
Cdk4 fwd	CGGCCTGTGTCTATGGTCTG
Cdk4 rev	GAAGCAGGGGATCTTACGCT
Cdk6 fwd	CGCAGAAAGCCTCTTTTTCGT
Cdk6 rev	GTCCCTAGGCCAGTCTTCCT
E2f5 fwd	ACCTTGGCTGTGAGGCAAAA
E2f5 rev	CAGCACCTACACCCTTCCAC
E2f7 fwd	GTGAACTCCCTGCAGCTTGA
E2f7 rev	CCGTTGACAAGGGGTAGCTC
Aurka fwd	CATGACGCCACCCGAGTTTA
Aurka rev	GAGCGTTTGCCAACTCAGTG
Cdkn2c fwd	GGGGGACCTAGAGCAACTTAC
Cdkn2c rev	TAGCACCTCTGAGGAGAAGCC
Ccdc67 fwd	AGGAATTAAGCAAGGCTGTGGA
Ccdc67 rev	TCGCTTCGGAGTCTTTCGTT
Ccno fwd	CCTTCCGAGAATACGGCCAG
Ccno rev	GATTCGGCAGTCACTTGTGG
Foxj1 fwd	CTGAGCCAGGCCTCACATTC
Foxj1 rev	TCACTTCCATTCTGCGACCC
Ccp110 fwd	AAGATGCTAGAAACTAGCCCCAA
Ccp110 rev	GGCATGCTGTTCCTCTAGTCTC
Kctd17 fwd	CTTCGAGCAGCTGGTGAACAT
Kctd17 rev	GACCTTGGCCTTGCGAGT
Cep135 fwd	GTGGACAGTCTGCCTTTGGT
Cep135 rev	CGTATGAACGAGGTCACTGAA
Cep97 fwd	AGAACGGCAGAAGACGGTTG
Cep97 rev	CAGGGCAAGTTGGCACCTAA
Gapdh fwd	TCAAGCTCATTTCCTGGTATGAC
Gapdh rev	CTTGCTCAGTGTCCTTGCTG
Actb fwd	CAGCTTCTTTGCAGCTCCTT
Actb rev	ATCCATGGCGAACTGGTG

Fwd, forward; rev, reverse.

For analysis of miRNA expression changes in mouse organs, tissues were dissected and immediately frozen using liquid nitrogen. Upon thawing, RNA was isolated according to the manufacturer’s instructions using the mirVana miRNA Isolation Kit (AM1560; Applied Biosystems). We then used TaqMan MicroRNA Assays (4427975; Applied Biosystems), which use specific primers for reverse transcription of the miRNAs into cDNA, and specific primers and TaqMan probes for detection and quantification of the cDNA by qPCR (Table S2). For reverse transcription, we multiplexed the reaction with primers for four different miRNAs (using 2 µL each RT primer, 100 ng RNA, and 1.33 µL MultiScribe Reverse Transcriptase in a total volume of 20 µL). The real-time qPCR was then performed with 5 ng of cDNA using TaqMan MicroRNA Assays and an ABI 7300 instrument (according to the manufacturer’s instructions).

Western Blotting.

For Western blot analysis of fallopian tubes or trachea, the organs were homogenized and whole-cell extracts were prepared using 50 µL RIPA buffer per sample. We then used 20 µg of protein for Western blot analysis and probed the membranes using the following dilutions for the primary antibodies: cyclin D3 (sc-182; 1:500; Santa Cruz Biotechnology), cyclin B1 (MAB3684; 1:500; Millipore), CDC25A (sc-7389; 1:200; Santa Cruz Biotechnology), CP110 (PA5-34380; 1:1,000; Thermo Scientific), GAPDH (5174; 1:1,000; Cell Signaling Technology), cyclin A2 (ab13824; 1:1,000; Abcam), CDK1 (ab71939; 1:1,000; Abcam), N-MYC (sc-53993; 1:1,000; Santa Cruz Biotechnology), and Aurora A (ab13824; 1:1,000; Abcam).

Computational and Statistical Analysis.

For analysis of the cDNA microarray, we used the software Omics Explorer (version 3.2; Qlucore). Intensity values were normalized using RMA (Affymetrix). Probes mapping to the same UniGene cluster were collapsed using the average intensity values. This way, we analyzed expression changes of 26,542 unique UniGene clusters. As a significance threshold, we used a false discovery rate (FDR) of less than 0.05 and an expression change of more than 1.2-fold. For visualization, we normalized the gene expression level to the mean expression level for each gene and displayed the gene expression as log2 values of the expression, with genes sorted using supervised hierarchical clustering. For subsets of genes analyzed in this microarray (e.g., Notch1 target genes or genes involved in cell-cycle, differentiation, or ciliogenesis regulation), we displayed gene expression changes in the same way using the software Prism (version 7.01; GraphPad), that is, log2 values of gene expression level, normalized to the mean expression level for each gene.

For analysis of enrichment of biological processes and pathways among the significantly deregulated genes identified in the cDNA microarray, we used the internet-based software DAVID (version 6.7; National Institutes of Health). Among the 480 significantly deregulated genes, 457 mouse genes were found in this database. We then searched for enrichment of biological processes and pathways using the databases Gene Ontology Biological Processes (GOTERM_BP_FAT) and Kyoto Encyclopedia of Genes and Genomes (KEGG_PATHWAY). A Benjamini-adjusted q value of less than 0.05 was considered statistically significant.

For the detection of predicted miRNA target sites in 3′ UTRs of mRNAs, we used the software TargetScanMouse (release 7.1; Whitehead Institute for Biomedical Research).

For all other statistical analyses, we used the software GraphPad Prism (version 7.01). For each experiment, the number of animals (n) and the statistical test are indicated in the figure legend. For experiments involving a comparison between two groups, we also tested whether the data were following a normal distribution using the D’Agostino and Pearson normality test (for n ≥ 7) or the Shapiro–Wilk normality test (for n < 7), and used this information to decide between an unpaired t test (if both groups followed a normal distribution) or a Mann–Whitney test for determination of significance. In cases with significant differences in the variance between two groups that followed a normal distribution (as determined by an F test), we used an unpaired t test with Welch’s correction to determine significance. For all experiments, P < 0.05 was considered statistically significant and the level of significance is indicated in each figure (i.e., ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05; ns, not significant). For all figures, data are presented as mean ± SD.

Data Resources.

The cDNA microarray data have been deposited in the GEO database under accession no. GSE92296 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE92296).

Discussion

Since their discovery, miRNAs have been implicated in a wide range of biological processes. Early studies revealed that expression of miRNAs is relatively tissue-specific, correlates with the differentiation state, and is generally down-regulated in human cancers (25). These observations suggested that miRNAs may function to restrain cell proliferation, to prevent tumorigenesis, and to promote differentiation. Members of the miR-34/449 family represent one of the best-studied examples of miRNAs with potentially antiproliferative roles. These miRNAs possess various growth-suppressive functions such as induction of apoptosis and cell-cycle arrest in vitro (4, 8–11, 26, 27). Consistent with a growth-suppressive role for miR-34a, its genetic ablation was shown to enhance tumorigenesis in mouse models of KRAS^G12D;p53^+/− lung cancer (28), p53^−/− prostate cancer (29), carcinogen-induced colon cancer (30), and SmoA1-overexpressing medulloblastoma (31). In contrast, we observed that mice lacking all members of the miR-34/449 family did not display any increase in spontaneous tumorigenesis. These results suggest that while miR-34a suppresses tumorigenesis in the context of certain oncogenic lesions, deletion of all miR-34/449 members is insufficient to cause cancer formation.

We report here that deletion of all members of the miR-34/449 family results in derepression of cell-cycle proteins and triggers unscheduled proliferation in several differentiating tissues. Furthermore, we demonstrate that failure of cells to timely exit the cell cycle is responsible for impaired epithelial cell differentiation, namely it blocks formation of motile multicilia. It is well-established that ciliogenesis requires cells to exit the cell cycle. Once rendered quiescent, cells then assemble cilia in an ordered process. This can be reversed by ciliary resorption, which occurs upon cell-cycle reentry and allows the centrioles to be used as spindle poles during mitosis (32). Therefore, the lack of cilia in proliferating cells likely represents a combined result of suppression of cilia formation as well as enhanced ciliary resorption.

The defect in ciliogenesis observed upon genetic deletion of all miR-34/449 family genes in mice resembles some symptoms of human primary ciliary dyskinesia (PCD), such as chronic airway disease and reduced fertility in females and males. However, unlike TKO mice, PCD patients suffer from reduced cilia motility, commonly caused by mutations that trigger ultrastructural defects of the ciliary axoneme. Moreover, abnormalities such as situs inversus and hydrocephalus, which also occur in a subset of PCD patients, were not observed in TKO mice. It should be noted that expression of the miR-34b/c and miR-449a/b/c genes is restricted to only a few tissues, including respiratory epithelium, fallopian tubes, and testes, and their expression increases during multiciliogenesis in vitro (33). The tissue-specific expression of these miRNAs likely explains why ciliogenesis in other tissues is not affected. We speculate that in these tissues, other microRNAs are responsible for repression of the cell-cycle machinery during differentiation and ciliogenesis.

Two earlier studies implicated the miR-34/449 family in regulation of ciliogenesis in vivo. The first study reported that miR-449 regulates ciliogenesis by repressing Notch1 signaling (22). However, we did not observe any evidence of activation of Notch1 signaling in the respiratory epithelium of TKO mice (Fig. S3E). Moreover, Notch1 activation is well-known to affect cell-fate specification by favoring the formation of mucus-producing goblet cells at the expense of cilia-generating columnar epithelial cells (23), a phenotype not seen in TKO mice (Fig. S2 G and H). The second report implicated CP110 as a regulator of ciliogenesis operating downstream of miR-34/449 (20). The authors proposed that increased levels of CP110 prevent docking of the basal bodies to the cell membrane. However, we found that ablation of miR-34/449 had no effect on the levels of CP110 transcripts and protein in the nasal respiratory epithelium at the time of multicilia formation (Fig. S3 C and D), in the trachea (Fig. S4A), or in the fallopian tube epithelium (Fig. 5B). Therefore, we conclude that, while alterations of Notch1 signaling and CP110 may contribute to the observed phenotype, it is unlikely that they represent the major primary molecular lesion responsible for the differentiation defects seen in TKO mice.

In contrast to these findings, we observed that ablation of miR-34/449 triggered an enhanced expression of almost 40 cell cycle-promoting genes. Of note, the cell cycle was the only biological process significantly enriched among genes deregulated in the respiratory epithelium of TKO mice at the time when ciliogenesis takes place in vivo. These cell cycle-promoting proteins are well-known to drive cell-cycle progression by increasing the activity of cyclin-dependent kinases. Indeed, we observed that ablation of miR-34/449 increased proliferation of epithelial cells and prevented cell-cycle exit. Importantly, inhibition of abnormal CDK activity corrected the phenotype seen in TKO mice and restored normal ciliogenesis in miR-34/449–deficient animals. These results indicate that expression of miR-34/449 is essential for these cells to exit the cell cycle and enter quiescence, or to maintain cells in a quiescent state. Several cell-cycle genes were previously shown to represent direct targets of miR-34/449 family members, including cyclin D3 (34), cyclin B1 (22), CDC25A (7), and N-MYC (35). Hence, increased expression of these cell-cycle proteins in TKO mice represents a direct effect of their derepression upon miR-34/449 ablation.

miRNAs of the miR-34/449 family were also reported to be involved in spermatogenesis (21, 36, 37). In particular, miR-34b/c and miR-449a/b/c are required for proper execution of meiosis in male gametes and during the maturation of spermatids (36). It is likely that both the involvement of this miRNA family in spermatogenesis as well as their role in multicilia formation in the efferent ducts of the testes, described in our study, contribute to the infertility observed in male TKO mice. Further experiments using tissue-specific deletion of these miRNAs will allow dissecting the relative contribution of these defects to male infertility.

In summary, our study establishes the miR-34/449 family as an essential and rate-limiting regulator of epithelial differentiation in vivo, acting via suppression of cell-cycle genes. Ablation of miR-34/449 derepresses the cell-cycle machinery, thereby preventing epithelial cell differentiation. In addition to uncovering a novel function for cell cycle-targeting miRNAs during normal development, these findings may have implications for the mechanisms responsible for human ciliopathies as well as for novel therapeutic strategies for patients with cilia defects.

Materials and Methods

Treatment of miR-34/449 TKO Mice.

For inhibition of CDKs, newborn mice were treated by daily i.p. injection with the CDK inhibitors ribociclib (200 mg/kg body weight), roscovitine (5 mg/kg), R547 (2 mg/kg), or, for control, PBS (25 mL/kg). All animal experiments were approved by the Institutional Animal Care and Use Committee of the Dana-Farber Cancer Institute and were carried out in accordance with approved protocols and regulatory standards. Details are provided in SI Materials and Methods and Table S2.

Laser Capture Microdissection and Microarray Analysis.

The respiratory epithelium cell layer was isolated from formalin-fixed paraffin-embedded sections of embryonic heads using laser capture microdissection. Subsequently, RNA was extracted and cDNA microarray analysis was performed using the SensationPlus FFPE Amplification and WT Labeling Kit and Mouse Gene v2.0 ST array chips (Affymetrix). Details are provided in SI Materials and Methods.

Histology, RT-qPCR, and Western Blot Analysis.

These procedures were performed according to established protocols. Details are provided in SI Materials and Methods and Table S2.