Dicer Is Required for Normal Cerebellar Development and to Restrain Medulloblastoma Formation

Dicer, a ribonuclease III enzyme, is required for the maturation of microRNAs. To assess its role in cerebellar and medulloblastoma development, we genetically deleted Dicer in Nestin-positive neural progenitors and in mice lacking one copy for the Sonic Hedgehog receptor, Patched 1. We found that conditional loss of Dicer in mouse neural progenitors induced massive Trp53-independent apoptosis in all proliferative zones of the brain and decreased proliferation of cerebellar granule progenitors at embryonic day 15.5 leading to abnormal cerebellar development and perinatal lethality. Loss of one copy of Dicer significantly accelerated the formation of mouse medulloblastoma of the Sonic Hedgehog subgroup in Patched1-heterozygous mice. We conclude that Dicer is required for proper cerebellar development, and to restrain medulloblastoma formation.


Introduction
MicroRNAs (miRNAs) are short non-coding RNAs derived from pri-miRNAs that are transcribed and processed by Drosha and the microprocessor complex subunit Di George syndrome critical region gene 8 (DGCR8) into pre-miRNAs in the nucleus [1,2]. Pre-miRNAs are then translocated into the cytoplasm to be further processed by Dicer into mature miR-NAs. Mature single-stranded miRNAs are loaded into the RNA-induced silencing complex (RISC) that binds the 3'-UTR of mRNAs to inhibit their translation or induce their degradation [1,2]. Dicer is essential for early mouse development since its loss in the germline induces embryonic lethality at E7.5 [3]. However, conditional deletion of Dicer with Cre transgenic lines in different regions of the nervous system, including inner ear [4,5], retina [6,7], cortex and hippocampus [8][9][10][11][12], Purkinje cells [13], dopaminoceptive neurons [14,15], Schwann cells [16], astrocytes [17] and glia cells [18], reveals a role of Dicer and miRNAs in different developmental processes including cell survival, differentiation and proliferation. In mice, the full loss of Dicer expression is selected against in tumorigenesis [19][20][21], and loss of two copies of Dicer reduces cell proliferation and increases apoptosis [22]. Dicer is haploinsufficient for tumor suppression in some [19,20], but not all cancers [21]. In humans, germline (including DICER1 syndrome) and somatic mutations of one copy of DICER1 has been described in different tumor types, including one human medulloblastoma [19,23,24]. However, loss of heterozygosity is rare, but has been observed in pineoblastoma and pituitary blastoma [24].
Cerebellar granule neuron progenitors (GNPs), born in the rostral rhombic lip (rRL) at embryonic days (E) 11 to E16, migrate on the surface of the developing cerebellum to form the external granule layer (EGL) [25]. After birth, upon Sonic Hedgehog (SHH) stimulation, GNPs rapidly proliferate with maximum proliferation at postnatal days (P) 5 -P7, after which they exit the cell cycle and migrate inward to form the internal granule layer (IGL). In the mouse, the cerebellum is fully formed by 3 weeks after birth [25]. Constitutive activation of the SHH signaling pathway leads to defects in cell cycle exit, migration and differentiation which in turn can induce medulloblastoma.
The miR-17~92 cluster family is expressed in proliferating GNPs [26] and up-regulated in SHH medulloblastoma [26][27][28]. Co-deletion of miR-17~92 with its paralog miR-106b~25 in mice reduces proliferation of GNPs at postnatal age leading to a small cerebellum [29]. While loss of miR-17~92 leads to a relatively mild phenotype, it is absolutely required for SHH medulloblastoma formation [29]. We here show that Dicer that processes pre-miRNAs into mature miRNAs, including those encoded by the miR-17~92 cluster, is required for both normal cerebellar development and medulloblastoma suppression.
For animals that will develop medulloblastoma, we made every effort to minimize suffering by daily observation for clinical signs of sickness. Humane endpoints included dome head, titled head, loss of greater than 20% body weight, slow movements, difficulties with gait and dehydration. In the Ptch1 +/-; Cdkn2c -/genetic background,~60% of the mice will develop medulloblastoma while the other 40% of animals do not develop tumors. Mice with any clinical signs are not kept alive and euthanized immediately, using carbon dioxide (CO 2 ).
Mice were housed in an accredited facility of the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC). This study was carried in strict accordance with the National Institute of Health guidelines for the Care and Use of Laboratory Animals. All procedures in the protocol were approved by the Animal Care and Use Committee (ACUC) of St. Jude Children's Research Hospital (Animal Assurance Number: A3077-01).

Fluorescence in situ hybridization
Mouse embryo fibroblasts (MEF) were isolated from E14.5 Nestin-Cre embryos, as previously described [34]. Fluorescence In Situ hybridization (FISH) was performed as previously described [35]. In brief, metaphases were hybridized with a biotin labeled chromosome 12 BAC probe (RP23-456E21; CHORI, Oakland CA, USA) and a digoxigenin labeled probe containing the rat Nestin promoter. Probes were visualized by incubation with Texas red avidin and fluorescein labeled anti-digoxigenin.

Statistics
Statistical significance was determined using GraphPad Prism software (version 5.0). Data were shown as mean +/-s.e.m. P-values <0.05 were used as significance threshold from unpaired two-tailed Student's t test. For the survival curves, p-values were determined with a logrank (Mantel Cox) test.

Conditional deletion of Dicer in Nestin-positive progenitor cells
To assess the role of mature microRNAs in brain development, we crossed conditional Dicerfloxed mice with the Nestin-Cre transgenic line and generated Dicer floxed/floxed ; Nestin-Cre + mice (referred as Dicer cKO). We found that the Nestin-Cre transgene was located on chromosome 12 (band D1) and genetically linked to Dicer (band E-F1) by~25 megabases (Fig 1A and 1B). Thus, the generation of Dicer cKO animals required that the Nestin-Cre transgene and one floxed allele of Dicer be carried by the same chromosome which necessitated a complex breeding strategy ( Fig 1C). Dicer cKO mice died at birth (15/15) but were alive at E18.5. When compared to control littermates (Fig 1D, 1F and 1H), cerebella from E18.5 Dicer cKO embryos had a reduced size and lacked foliations (Fig 1E, 1G and 1I), a thinner EGL visualized with an antibody to Pax6 ( Fig  1I) and the rRL devoid of Pax6-expressing progenitor neurons ( Fig 1I, white arrows). In contrast, the cerebellum of Dicer floxed/+ ; Nestin-Cre + mice appeared normal (data not shown).
To identify the cause of abnormal cerebellar development, we analyzed embryos at earlier time points. At E14.5, the rRL and the nascent EGL of the Dicer cKO mice appeared similar to control littermates (Fig 2A and 2B). Pax6 positive cells (Pax 6 + ) were similarly present in the exterior face of the rRL and in the nascent EGL of both Dicer cKO and control mice (Fig 2C  and 2D). In contrast, at E15.5, the nascent EGL of the Dicer cKO mice were smaller than in control animals ( Fig 2E and 2F), and the exterior face of the rRL appeared thin (Fig 2G and  2H, white arrows). TUNEL staining revealed few apoptotic cells at E14.5 (Fig 2I, 2J and 2Q). In contrast, massive apoptosis was observed in the rRL but not in the EGL, at E15.5 ( Fig 2K, 2L and 2Q). In Dicer cKO mice at E15.5 by in vivo BrdU labeling we observed decreased proliferation of the Pax6 + cells in the rRL and EGL (Fig 2P and 2R). Because late-born GNPs contribute to the formation of the posterior EGL [36], the loss of Pax6 + cells in Dicer-deficient posterior cerebella at E18.5 could be explained by the apoptosis of cells in the rRL and their reduced  proliferation in the rRL and EGL at E15.5. In addition to the rRL, the Nestin-cre transgene is widely expressed in proliferative zones throughout the central nervous system (CNS). Consistent with this expression pattern, massive apoptosis was observed in the cerebellar ventricular zone (VZ) and caudal rhombic lip (cRL) (Fig 2L and 2Q).
Apoptosis was already detected in the neocortex of Dicer cKO embryos from E14.5 onward (Fig 3A-3D and 3G), as previously reported [10,12] and in the VZ of the colliculus (data not shown). The VZ of the neocortex generates neurons that radially migrate towards the surface of the brain to their final position with a specific address dependent on their birth date resulting in a cortex with characteristic six neuronal layers (Fig 3E). In the Dicer cKO mice, at E18.5, the cortex was less laminated with a thin VZ (Fig 3F).
Apoptosis can proceed by a Trp53-dependent pathway, particularly in the case of induced genome instability, or by Trp53-independent pathways [37,38]. To determine if the cell death resulting from Dicer loss was a result of DNA damage, we asked if the apoptosis was dependent on Trp53 function. Loss of Trp53 did not rescue lethality, nor the phenotypes observed in the developing brain of the Dicer cKO mice (i.e. apoptosis, abnormal cerebellum and less laminated cortex) (Fig 4A-4M), suggesting that apoptosis and other phenotypes did not result from DNA damage.

MicroRNAs are progressively deleted in the cerebellum of Dicer cKO embryos
Cre recombinase activity under the control of the Nestin promoter is initiated between E9.5 and E10.5 [39]. However, we observed massive apoptosis at E15.5 in the rRL and the VZ of the developing cerebella in Dicer cKO mice, suggesting that complete loss of Dicer and therefore of mature microRNAs may have happened progressively from E9.5 to E15.5. We confirmed that the two Dicer alleles were deleted in the cerebellum at E14.5 (Fig 5A). We previously reported that the miR-17~92 cluster family is expressed in the developing cerebellum [29]. To evaluate how the loss of Dicer affects the expression of mature microRNAs, we determined the expression of miR-19a (from the miR-17~92 cluster) and miR-106b (from the miR-106b~25 cluster) in the developing cerebella of control and Dicer cKO embryos at E14.5, E16.5 and E18.5 by Q-RT-PCR (Fig 5B and 5C, lanes 1-6). While the levels of miR-19a and miR-106b were not statistically different at E14.5 (Fig 5B and 5C, lane 1 versus lane 4), their levels were significantly reduced at E16.5 (Fig 5B and 5C, lane 2 versus lane 5, p = 0.0386 and 0.0001, respectively) and at E18.5 (Fig 5B and 5C, lane 3 versus lane 6, p = 0.0005 and 0.0005, respectively). Therefore deletion of Dicer resulted in a delayed decrease of miR-19a and miR-106b microRNAs with significant loss of those microRNAs at E16.5.

Discussion
We here report the role of Dicer and mature miRNAs during cerebellar development and SHH medulloblastoma genesis by conditional deletion of Dicer (Dicer cKO) in Nestin-positive neural progenitors. We found that conditional inactivation of Dicer triggered extensive Trp53-independent apoptosis in all proliferative zones throughout the developing CNS and decreased proliferation of GNPs in the developing cerebellum. Loss of one copy of Dicer accelerated medulloblastoma formation in Ptch1 +/-, Cdkn2c +/animals. Dicer was shown by several groups to be required for proper development of many neural cell populations in the developing CNS [8][9][10][11][12][13][14]18]. Conditional loss of Dicer in the CNS was previously found to induce apoptosis using the same [10,12] or different [8,9,11,13,17,18,40] Cre lines, and to affect cortical development [8][9][10]12]. While using the same Nestin-Cre and Dicer-floxed lines, s, the group of Kawasa-Koga did not find obvious defects cerebellar anomalies in their Dicer cKO mice [10]. They show that while miRNAs are still detected at E15.5 in the cortex of their animals, they are absent at E18.5 correlating with apoptosis. Kawasa-Koga and colleagues remarked that Nestin-Cre; Dicer mice produced very few Dicer conditional knockout mice embyos, 6.8% at E18.5 and that no surviving newborns were detected. This is consistent with our data and the finding that Dicer and Nestin-Cre are genetically linked on mouse chromosome 12. In the Discussion of their paper, Kawase-Koga and colleagues did not find obvious defects in cerebellar development during embryonic stages and yet, they obtained very few and no live embryos while detecting defects in the cortex of Nestin-Cre; Dicer mice. These differences in phenotypes between groups may potentially be due to the genetic background. Using the same Nestin-Cre but different Dicer-floxed mice, McLoughlin and collaborators also describe cortical defects in Dicer cKO mice, but did not report on the status of the mice cerebella [12]. McLoughlin and colleagues used Dicer-floxed animals generated by Harfe and colleagues. In their manuscript, they only described the effect of Dicer deletion in the cortex but not in the cerebellum and found increased apoptosis in cortical regions at E15.5, which is consistent with our data. They also detected low numbers of Dicer conditional knockout mice embryos and no surviving embryos. Kawase-Koga and McLoughlin focused on cortical development while we assessed the role of Dicer in the embryonic cerebellum. To increase the percentage of embryos with the proper genotype, we backcrossed our mice into C57BL/6 background (Fig 1C) that changed the genetic background of our cohort, compared to that of Kawase-Koga and colleagues. This may explain why Kawase-Koga and colleagues do not see apoptosis at E15.5 in the cortices of Nestin-Cre + ; Dicer floxedl/floxed mice; even though they show, by immunoblotting, that the levels of Dicer protein are greatly reduced at E15.5. They also note that Dicer deletion occurs at different times depending on the localization.
Despite these few differences, our results are consistent with these reports since conditional deletion of Dicer in Nestin-positive cells induced substantial apoptosis in all proliferative zones of the CNS.
In Dicer cKO mice, loss of neural progenitors in the rRL led to a diminished pool of GNPs which resulted in an abnormally developed cerebellum with a short and thin EGL, absence of + ; Nestin-Cre + ; Ptch1 +/-; Cdkn2c +/and (E) Dicer +/+ ; Ptch1 +/-; Cdkn2c +/mice. Scale bars = 500μm (C and E) and 50μm (D and F). folia and an aberrant residual rRL at E18.5. This was associated with apoptosis in the rRL and reduced proliferation of GNPs in the developing EGL. Apoptosis was also observed in the ventricular zone of the developing cerebellum. These results suggest that lack of Dicer induced apoptosis in stem / progenitor neural cells, before they commit to their final lineages and controlled the proliferation of GNPs in the developing EGL Although Nestin is expressed as early as E9.5-10.5 in neural progenitors [39], the decrease in microRNAs levels in the developing cerebella of the Dicer cKO mice was delayed, consistent with the relatively long half-life of miRNAs [41]. At E14.5, the cerebella of Dicer cKO and control littermates were normal and the levels of miRs were not significantly reduced. However, at E15.5, massive apoptosis was observed, suggesting that residual levels of microRNAs were insufficient to maintain the proliferation of neural progenitors. Overall, our results suggest that processing of microRNAs by Dicer was required for proper cerebellar development during embryogenesis. Alternatively, the effects observed might be due to miR-independent cell survival function of Dicer, as previously reported [42]. Thus, our study is the first analysis of the role of Dicer during cerebellar development. We consider that our data and that from others illustrate a critical role for this protein, in the regulation of miRNAs during development of the cerebellum.
We found that Dicer restricts SHH medulloblastoma development in Ptch1 +/mice. Our results are consistent with those reported in two mouse models of retinoblastoma and lung cancers [19,20] but not in Eμ-Myc B cell lymphoma [21]. We previously showed that miR-17~92 expression is absolutely required for SHH medulloblastoma initiation [29] and that its silencing inhibits medulloblastoma progression [43]. This implies that complete loss of Dicer is selected against medulloblastoma formation and progression. Here we show that loss of one copy of Dicer in a SHH medulloblastoma mouse model accelerated tumor formation, suggesting that decreased levels of some mature tumor suppressor miRNAs might promote tumorigenesis. We will unable to correlate the phenotypes of Dicer cKO in embryos with the ones we observed with the mir17~92 cluster family because the effects of the loss of the mir17~92 cluster family are only revealed in the cerebellum after birth [29]. The Dicer cKO mice instead die at birth. We previously published that the miR-17~92 cluster family was required for proliferation of GNPs during postnatal cerebellar and medulloblastoma development [29,43]. In the present study, we used two of the microRNAs encoded by the miR-17~92 cluster family (miR-19a and miR106b) to determine the level of microRNA biogenesis. In previous studies, we determined that these two miRNAs were expressed during embryonic cerebellar development [29]. Clearly other microRNAs besides those encoded by the miR-17~92 cluster family are expressed during cerebellar development. Therefore, the apoptosis observed at E15.5 Dicer cKO embryos is most likely due, only in part, to the loss of expression of microRNAs encoded by the miR-17~92 cluster family.
In summary, while Dicer is essential for proper cerebellar development and homeostasis, it restrains SHH medulloblastoma development suggesting that it might be haploinsufficient for medulloblastoma suppression.