Address correspondence and reprint requests to Katsuhiko Mikoshiba, MD PhD, Laboratory for Developmental Neurobiology, RIKEN Brain Science Institute and Calcium Oscillation Project, ICORP-SORST, JST, 2-1 Hirosawa, Wako-shi, Saitama 351-0198, Japan. E-mail: firstname.lastname@example.org
During postnatal development of the cerebellum, granule cell precursors (GCPs) proliferate in the external granular layer (EGL), exit the cell cycle, differentiate, and migrate from the EGL to the internal granular layer. In the present study, we report that type 2 and 3 inositol 1,4,5-trisphosphate (IP3) receptors (IP3R2 and IP3R3) regulate the differentiation of GCPs after postnatal day 12 (P12). 5-Bromodeoxyuridine labeling experiments revealed that in mutant mice lacking both of these receptors (double mutants) a greater number of GCPs remain undifferentiated after P12. Consequently, the EGL of the double mutants is thicker than that of control mice at this age and thereafter. In addition, granule cells remain in the EGL of the double mutants at P21, an age when migration has concluded in wild-type mice. Whereas differentiation of GCPs was reduced in the double mutants, the absence of IP3R2 and IP3R3 did not affect the doubling time of GCPs. We conclude that intracellular calcium release via IP3R2s and IP3R3s promotes the differentiation of GCPs within a specific interval of postnatal development in the cerebellum.
terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling
During brain development, the proliferation and differentiation of neuronal precursor cells, together with apoptotic cell death, determines the total number of neurons and brain size (Takahashi et al. 1996; Chenn and Walsh 2002, 2003; Depaepe et al. 2005). Differentiated neurons integrate into neural circuitry after migrate to a final destination within the neuropil.
The cerebellum contains few cell types, a relatively simple architecture, and is an excellent model for studying how neuronal precursors migrate and differentiate during postnatal development. Granule neurons are the predominant cell type in the cerebellum. These neurons synapse upon Purkinje cells, and though they are only present within the cerebellum, they are the most abundant type of neuron in the CNS. The progenitors of granule neurons are generated in the rhombic lip in the anterior hindbrain during embryogenesis. These cells migrate over the cerebellar primordium to form a proliferative zone (PZ) in the external granular layer (EGL). During early stages of postnatal development, granule cell precursors (GCPs) proliferate in the PZ of EGL. Gradually these cells exit the cell cycle, differentiate, and migrate away from the EGL toward the internal granular layer (IGL) (Altman and Bayer 1997). In rodents, this process is complete by the end of the third postnatal week, by which time the EGL has disappeared. The length of the cell cycle (doubling time) for GCPs in the PZ (Fujita et al. 1966; Fujita 1967), and the proportion of daughter cells that exit the cell cycle and differentiate, are critical parameters that determine the total number of granule cells and affect the entire program of the cerebellar development. For example, in humans, mutations in the gene patched (ptc) can lead to unregulated proliferation of GCPs and generate medulloblastomas, a form of cerebellar tumor (Goodrich et al. 1997).
In the present study, we demonstrate that type 2 and 3 inositol 1,4,5-trisphosphate (IP3) receptors (IP3R2 and IP3R3) regulate the differentiation of GCPs in the cerebellum within a specific period of postnatal development. IP3Rs are intracellular Ca2+ release channels located on the endoplasmic reticulum membrane that mobilize Ca2+ from the endoplasmic reticulum to the cytosol (Furuichi et al. 1989). The function of IP3R subtypes has been studied by gene disruption in cell lines (Sugawara et al. 1997) and mice (Matsumoto et al. 1996; Futatsugi et al. 2005), however, their physiological role, especially in the brain, largely remain to be elucidated. This study provides the first evidence implicating IP3R2 and IP3R3 as important regulators of neuronal differentiation. In addition, these findings illustrate the importance of calcium signaling from intracellular stores in the development of the mammalian CNS.
Materials and methods
Inositol 1,4,5-trisphosphate receptor 1 knockout mice were described previously (Matsumoto et al. 1996). The generation of IP3R2 knockout mice and IP3R3 knockout mice, as well as the double mutants has also been described previously (Futatsugi et al. 2005). Mouse genotypes were determined by PCR (data not shown). Mice, including the double mutants, were studied prior to weaning and were fed under standard conditions with their mothers. All experimental procedures were performed in accordance with the Japanese Physiological Society’s guidelines for animal care, and were approved by the committee of the animal facility of the RIKEN Brain Science Institute.
Primary antibodies used for the immunohistological experiments were as follows: an anti-5-bromodeoxyuridine (BrdU) antibody (Dako, Glostrup, Denmark), a polyclonal antibody to proliferating cell nuclear antigen (PCNA) (Santa Cruz Biochemicals, Santa Cruz, CA, USA), an anti-p27/Kip1 antibody (Transduction Laboratories, Lexington, KY, USA), and an anti-calbindin antibody (Sigma, St Louis, MO, USA). Secondary antibodies were biotinylated anti-mouse IgG antibody or biotinylated anti-rabbit IgG antibody (Jackson, West Grove, PA, USA), depending on the primary antibody. Signals were detected using the Vectastain ABC-Peroxidase Kit (Vector, Burlingame, CA, USA). For fluorescent double staining, an anti-rabbit IgG antibody conjugated to Alexa Fluor 488and an anti-mouse IgG antibody conjugated to Alexa Fluor 568 (Molecular Probes, Eugene, OR, USA) were used as secondary antibodies.
Postnatal day 8–28 (P8–28) mice were anesthetized with Nembutal and fixed by cardiac perfusion with 4%p-formaldehyde (PFA) in 0.1 M phosphate buffer. Brains were removed, post-fixed in 4% PFA in phosphate-buffered saline (PBS) overnight, and embedded in paraffin. Serial sagittal sections (7 μm) were cut with a microtome. For general histology, midsagittal sections were stained with cresyl violet or hematoxylin/eosin, and the thickness of the EGL, molecular layer (ML), and IGL of the cerebellum was measured at the midpoint of folia 3, 4/5, 8, and 9. Results were pooled for the analysis within each group of animals. Layer thicknesses in the double mutants and controls at each developmental stage were analyzed with two sample Student’s t-tests.
For immunohistochemical experiments, post-fixed brains were cryoprotected in 30% sucrose, and sagittal sections (14 μm) were cut with a cryostat (Leica Microsystems, Wetzlar, Germany). After inactivating intrinsic peroxidase activity with 3% H2O2 in PBS for 30 min, sections were treated with 0.05% trypsin, 0.025% CaCl2 in PBS for 5 min at 37°C, and then incubated in a blocking solution (1% bovine serum albumin and 0.1% Tween 20 in PBS) followed by incubation with primary antibodies in the blocking solution. The following concentrations were employed for each antibody: antibody to calbindin, 1 : 200; antibody to PCNA, 1 : 200; antibody to BrdU, 1 : 100; and antibody to p27/Kip27, 1 : 200. Incubations were performed at 22°C for 1 h or at 4°C overnight. Sections were then washed for 5 min three times in PBS containing 0.05% Tween 20, and incubated with biotinylated secondary antibody (1 : 100) followed by signal amplification using the Vectastain ABC-Peroxidase Kit (Vector, Burlingame, CA, USA). Bound antibody complexes were detected with the colorimetric reagent diaminobenzidine tetrahydrochloride.
In vivo BrdU labeling and immunohistochemistry
To label proliferating GCPs in the S phase of the cell cycle, mice were injected intraperitoneally with 50 mg/kg body weight of BrdU. To examine the behavior of BrdU-labeled GCPs, mice were killed at the times indicated for each experiment. Mice were perfused with 4% PFA and sagittal sections (14 μm) of brain were cut with a cryostat. Prior to staining with an anti-BrdU antibody, sections were treated with 4 N HCl for 20 min at 22°C. Sections were then incubated with biotinylated or Alexa Fluor 568-linked secondary antibodies, and BrdU-positive cells were visualized. The number of labeled cells/mm in each layer of the cerebellar cortex in the slices was counted in folia 3, 4/5, 8, and 9.
Quantification of dying cells
Dying cells were detected by terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL). TUNEL assays were performed with an ApopTag Kit (Invitrogen, Carlsbad, CA, USA) by following the manufacturer’s instructions. Briefly, midsagittal cerebellar frozen sections (14 μm) from IP3R2/IP3R3 double mutants and control mice were collected, and dying cells were visualized with diaminobenzidine tetrahydrochloride followed by counterstaining with methyl green. TUNEL-positive cells in each layer (EGL, ML, and IGL) in the cerebellar cortex were counted, and the area of each layer were calculated by tracing the outline of a digital picture and determining the area inside the traced region using the program NIH Image (National Institutes of Health, Bethesda, MD, USA). The ratio of dying cells/mm2 in each layer and in the whole cortical layer was calculated for each slice. At least 10 slices from five animals were examined for each genotype. Data was analyzed using Student’s t-test.
Quantitative analysis of GCP proliferation
One hour after BrdU injection, mice were killed and sagittal sections of the brain were prepared. The labeling index (LI), a value that is considered to reflect cell cycle length, was determined for each animal with the methods described by Chenn and Walsh (2002) but adapted for the evaluation of proliferation of GCPs in the cerebellum. To count BrdU-labeled cells and PCNA-positive cells accurately, adjacent sections were examined separately by immunohistochemistry. In each set of adjacent slices, BrdU- and PCNA-positive cells in the EGL of at least four folia, along 550 μm in the middle of each folium, were counted, and the LI calculated. Three slices were analyzed per mouse, and three mice were examined per genotype.
Quantitative analysis of GCP differentiation
Mice were injected with BrdU at P12. At 30, 38, and 48 h after injection, mice were killed and sagittal brain sections were prepared. The exiting fraction was determined as previously described (Chenn and Walsh 2002) with minor modifications to examine GCPs. This metric provides a ratio of cells that have and have not exited cell cycle and reflects GCP differentiation. Double-labeling of BrdU- and PCNA-positive cells was performed with secondary antibodies conjugated to Alexa Fluor 568 and 488, respectively. Only BrdU-positive cells in the EGL were counted. The exiting fraction was calculated independently for six folia in each slice, and three mice were examined per genotype.
Quantification of the number of cells in the proliferative and pre-migratory zone in the EGL
Proliferating cell nuclear antigen-positive cells in the EGL comprise the PZ while cells that are p27/Kip1-positive in the EGL have exited the cell cycle and constitute the pre-migratory zone (PMZ). Accordingly, PCNA- and p27/Kip1-positive cells in the EGL were detected in adjacent sections with the corresponding antibodies and counted as the number of cells in the PZ and PMZ. Cells were counted within the central 550 μm of each folium for six folia in each slice. More than four mice were used per genotype at each developmental stage.
Cerebella were dissected from brains of postnatal mice ranging from age P0 to P20 and homogenized in a solution containing 0.32 M sucrose, 5 mM Tris–HCl, pH 7.4, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 10 μM leupeptin, and 10 μM pepstatin A (homogenizing solution). We separated 1 μg (for IP3R1) or 20 μg (for IP3R2 and IP3R3) of homogenate by sodium dodecyl sulfate–5% polyacrylamide gel electrophoresis, transferred proteins to a polyvinylidene difluoride membrane and probed with antibodies KM1112, KM1083 (Sugiyama et al. 1994), and an anti-IP3R3 antibody (BD Biosciences, San Jose, CA, USA) to detect IP3R1, IP3R2, and IP3R3, respectively. Peroxidase-coupled detection was visualized using enzyme-linked chemiluminescence with the ECL kit (GE Healthcare Life Sciences, Buckinghamshire, England).
Golgi staining of cerebella was performed with the Braitenberg method (Braitenberg et al. 1967). Briefly, post-fixed mouse brains were cut in half at the midline and incubated in 3% K2Cr2O7, 0.5% formalin, 12.5% sucrose in the dark for 7–10 days, and in 0.75% AgNO3 for 2 days. The brains were then incubated in 30% sucrose in PBS for 2–4 days during which the solution was changed several times. Sagittal sections were then prepared with a vibratome (100–150 μm thick) and Purkinje cells imaged with the aid of a light microscope.
Mouse brains were fixed in 4% PFA in PBS for 30 min and then sectioned with a vibratome (400 μm) or cryoprotected in 20% sucrose and sectioned with a freezing microtome (14 μm). Sections were post-fixed in a solution containing 0.2% glutalaldehyde, 2 mM MgCl2, 5 mM EGTA in PBS, and stained in PBS solution containing 1 mg/mL X-gal, 50 mM potassium ferrocyanide, 50 mM potassium ferricyanide, and 2 mM MgCl2.To enhance permeability, 0.01% deoxycholate and 1% Nonidet P-40 were added to the staining solution. Staining was performed at 37°C in the dark for 4 h (vibratome slices) or 24 h (frozen sections).
Fluorescent and bright-field images were obtained with a CoolSNAP CCD camera (Photometrics, Tucson, AZ, USA) on an Olympus (Tokyo, Japan) fluorescence microscope and stereomicroscope. Images were processed with RS Image software (Photometrics, Tucson, AZ, USA).
All data are expressed as the mean ± SEM. Statistical comparison was performed with the Student’s t-test using KaleidaGraph software (Synergy Software, Reading, PA, USA).
EGL is thicker in double mutant mice lacking IP3R2 and IP3R3
In the course of examining brain sections from IP3R2 and/or IP3R3 knockout mice at P14, we discovered that the cerebellar EGL of mutant mice lacking both IP3R2 and IP3R3 was much thicker than that of control mice. We made sure by western blotting that the expression of IP3R2 and IP3R3 were abolished in the cerebellum of knockout mice and that there was no compensatory increase in other subtypes of IP3Rs that were not knocked out (Fig. 1a). Figure 1b and c provide examples of Nissl stained sagittal sections of the cerebellum at P14. Despite the smaller cerebellar size of the double mutants, in parallel with their smaller body size because of the hypofunction of the digestive system (Futatsugi et al. 2005), double mutants possessed a much thicker EGL compared with all other genotypes examined (Fig. 1b). We did not observe such a phenotype in knockout mice lacking either IP3R2 or IP3R3. While IP3R1 knockout mice also display decreased body weights and brain size relative to wild-type mice (Matsumoto et al. 1996), there was no difference between IP3R1 knockout and wild-type mice in terms of EGL thickness. At P21, EGL persists in the double mutants whereas in control mice this layer has disappeared altogether (Fig. 1d). However, by P28, all granule cells had migrated from the EGL to the IGL in double mutants (data not shown).
To determine when the difference in EGL thickness emerges in the double mutants, we measured the thickness of each cell layer of the cerebellum at several time points during the postnatal development of the cerebellum (Fig. 2a, b, and c). The EGL thins after P8 independent the genotype (Fig. 2a). There is no difference in the thickness of the EGL of double mutant mice and control mice until P12. After P12, the EGL is thicker in double mutants. At this age, in control mice the EGL continues to diminish while in double mutants it maintains the same thickness for several days. Consequently, the ML and the IGL tend to be a bit thinner in double mutants after P12 (Fig. 2b and c). The decrease in ML and IGL thickness is transient (lasting about 2 days) presumably because differentiated granule cells are migrating from thickened EGL thereafter.
GCPs are retained in the EGL in mutant mice lacking IP3R2 and IP3R3
We labeled a cohort of proliferating GCPs born on P12 with a pulse of BrdU. We chose to label GCPs at P12 because the thicker EGL observed in double knockouts is first evident at this age. Some labeled cells divided again in the EGL and while others differentiated and migrated out of the EGL (Fig. 3a). We were unable to discriminate cells originally labeled at P12 from daughter cells arising from the division of labeled GCPs. At 20 h post-labeling, BrdU-labeled granule cells had doubled in number (data not shown). These cells were present exclusively in the EGL of both double knockout and wild-type mice and had not started migrating. However, at 48 h post-labeling, migrating cells were observed in the ML and some granule cells had already reached the IGL. The number of BrdU-positive cells in the EGL was greater in double mutants than in control mice. At 96 h post-labeling, most of the BrdU-positive cells had migrated to the IGL both in the double mutants and control mice, but the number of cells still remaining in the EGL was greater in the double knockouts. Figure 3b shows the quantitative analysis of the number of BrdU-positive cells per mm in each cortical layer (EGL, ML, and IGL). These results were similar across several folium examined. These findings coincide with the thickened EGL in the double mutants after P12 and are consistent with either impaired differentiation of GCPs which results in decreased migration from the EGL, enhanced proliferation of GCPs in the EGL, reduced apoptosis of GCPs in the EGL, or some combination thereof.
The fraction of GCPs that differentiate between P12 and P14 is reduced in mutant mice lacking IP3R2 and IP3R3
The thickened EGL in the double mutants could result from an increased fraction of GCPs proliferating and remaining in the EGL rather than exiting the cell cycle and migrating out of the EGL. To test this possibility, we examined the fraction of GCPs that exited the cell cycle 30, 38, and 48 h after cells were labeled with BrdU at P12. We identified GCPs that were not longer proliferating as cells, which were BrdU+ and PCNA-, and proliferating cells as BrdU+ and PCNA+. BrdU+/PCNA− cells in the EGL were present exclusively in the inner part of EGL (Fig. 4a). This region presumably corresponds to the PMZ of the EGL. We counted the number of BrdU+/PCNA− cells and BrdU+/PCNA+ cells in the EGL. We observed that the fraction exiting cell cycle was significantly reduced in the double mutants compared with control mice at 30, 38, and 48 h post-labeling (Fig. 4b). The fraction of GCPs that were no longer proliferative increased with time, suggesting that cells continue to exit the cell cycle during the interval following BrdU labeling. At 48 h post-labeling, labeled GCPs that remained in the PZ may have completed another round of cell division, while a fraction of the differentiated cells have already exited EGL and reached ML and IGL (Fig. 3a and b). Therefore, the fraction of proliferative cells in this experiment does not accurately reflect the number of cells exiting cell cycle. As the thickness of EGL in the double mutants is unaltered between P12 and P14 (Fig. 2a), the actual percentage of GCPs exiting cell cycle during this time is predicted to be about 50%. This fraction should be greater than that in the normal mice, in which the EGL thins over this time interval. From these findings, we conclude that IP3R2 and IP3R3 prompt the differentiation of GCPs in the cerebellum after P12.
The loss of IP3R2 and IP3R3 function increased the number of proliferating cells and decreased the number that differentiate and migrate from the EGL, thereby thickening EGL relative to wild-type mice. Additional support for this hypothesis is contributed by experiments in which we categorized cells in the EGL into those in an outer, PZ which consists of GCPs, and those in an inner, PMZ which contains of post-mitotic, differentiated cells. Cells in the PZ were distinguished by PCNA immunoreactivity. Cells in the PMZ were identified as p27/Kip1-positive, although some cells in the PZ may also express this marker (Miyazawa et al. 2000). At P12, immediately prior to the age when EGL is beginning to appear thicker in the double mutants, the number of cells in PZ is increased in the double mutants (Fig. 4c). Thus, fewer cells may be differentiating and migrating out of the PZ in the double mutants. However, the number of cells in the PMZ of the double knockouts was not decreased thereafter, perhaps because an increased number of cells are migrating from PZ.
Proliferation rate of GCPs is unaltered in mutant mice lacking IP3R2 and IP3R3
The thickened EGL in the double mutants can also be a consequence of enhanced proliferation (i.e. shortened cell cycle time) of GCPs in the EGL. To quantitatively compare cell cycle time, we utilized a LI that is generally considered to provide a good estimate of cell cycle length. We counted the percentage of GCPs labeled by a single pulse of BrdU (1 h). GCPs were identified by immunoreactivity for PCNA (Doughty et al. 1998). We also examined Ki67 immunoreactivity and confirmed that it provided similar results to PCNA staining. As the cell cycle shortens (enhanced proliferation), LI increases. We did not detect a difference in the LI of GCPs between the double mutants and control mice at every postnatal day tested (between P10 and P16) (Fig. 5). Furthermore, in a related BrdU double pulse experiment (1 h plus 3 h), LI was similar between all genotypes of mice tested (data not shown). We conclude that the length of the cell cycle is not perturbed in mutant mice lacking IP3R2 and IP3R3.
Apoptotic cell death is not decreased in the EGL of double mutant mice
To investigate whether the thickened EGL in the double mutants resulted also from decreased programmed cell death (apoptosis), we examined the rate of apoptosis in cerebellar sections of the double knockout and control mice with the TUNEL. The number of apoptotic cells in the EGL is not decreased in the double knockouts. We quantified apoptosis not only in the EGL but also in ML and IGL (Fig. 6). The density of apoptotic cells in the whole cortical layer (EGL, ML, and IGL) was increased in the double mutants relative to control mice between P14 and P16 because of the increased apoptosis in the ML and IGL. From these measurements we conclude that the thickened EGL in the double mutants was not because of decreased cell death. Rather, increased programmed cell death in the double knockouts appears to compensate for the increase in the number of granule cells.
Development of Purkinje cells in the double knockouts
We examined whether the morphology of Purkinje cells was altered in mice lacking IP3R2 and IP3R3, given the reduced differentiation of GCPs. We visualized the cell body and the dendrites of Purkinje cells by calbindin immunostaining of sagittal sections (Fig. 7a) and by Golgi staining (Fig. 7b) at several different time points during development (P13, P14, P15, and P21). The size of Purkinje cells of the double mutants looked smaller at P13 and P14, when ML thickness is decreased (Fig. 2b). We did not observe gross differences in the morphology of Purkinje cell dendrites from the double mutant and control mice, although quantitative analysis would be required to address this issue more in detail.
Expression of IP3R2 and IP3R3 in the cerebellum during the development
While these experiments implicate IP3R2 and IP3R3 as proteins that promote differentiation of GCPs after P12, the mechanism by which these receptors exert this regulation is unclear. To help answer this question, we examined the expression of each type of IP3R in the postnatal cerebellum by western blotting (Fig. 8a). While type 2 was detected at all developmental stages, its expression level was highest between P8 and P12. By comparison, the expression level of type 3 is relatively constant. Unfortunately, as yet we have been unable to determine the specific cell types that express IP3R2 and IP3R3 because our antibodies against these receptors are not compatible with immunohistochemistry. However, as the targeting allele for IP3R2 includes the lacZ gene with a nuclear localization signal under the control of IP3R2 promoter, we utilized lacZ staining to monitor IP3R2 expression in these mice (Sanes et al. 1986). LacZ-positive cells were mainly detected in the deep nucleus of the cerebellum at P10, as well as in the IGL and Purkinje cell layer at P13 and thereafter. Thus, a fraction of granule cells in the IGL and some of Purkinje cells express IP3R2 at this stage (Fig. 8b and c).
The regulation of proliferation and differentiation of GCPs are important for not only determining the total number of granule cells but also for sculpting the cellular organization and functional architecture of the cerebellum. In the present study, we demonstrate the involvement of IP3R2 and IP3R3 in the differentiation of GCPs after P12. At this stage, majority of the GCPs in wild-type mice differentiate and leave EGL whereas in mutant mice lacking both IP3R2 and IP3R3, half of the GCPs do not differentiate and remain in the EGL, resulting in a thicker EGL in the double mutants. The rate of proliferation of GCPs was not altered in these mutants nor was there a decrease in apoptosis in the cerebellum. Interestingly, this abnormality in the differentiation of GCPs was not observed in mice lacking either IP3R2 or IP3R3 alone. Therefore, IP3R2 and IP3R3 perform redundant functions in the developing cerebellum. A similar redundancy of function for IP3R2 and IP3R3 has been reported for in exocrine tissues (Futatsugi et al. 2005).
A number of factors that regulate the proliferation and differentiation of GCPs have been identified (Hatten and Heintz 1995; Goldowitz and Hamre 1998; Wang and Zoghbi 2001). Sonic hedgehog promotes GCP proliferation (Wallace 1999; Wechsler-Reya and Scott 1999) and inhibition of the Sonic hedgehog pathway offer a novel therapy for medulloblastoma (Romer et al. 2004). Factors such as growth arrest specific gene 1 and insulin-like growth factor-1 also promote GCP proliferation (Ye et al. 1996; Lin and Bulleit 1997; Liu et al. 2001), whereas p27/Kip1 suppresses cell division and promotes differentiation of GCPs (Miyazawa et al. 2000). Activation of Notch2 by its ligand Jagged1 enhances GCP proliferation and inhibits GCP differentiation via its downstream target Hairy/Enhancer of Split 1 (Solecki et al. 2001). The chemokine stromal cell-derived factor-1α is secreted from the pia mater and its receptor, CXCR4, is expressed by GCPs. This extracelluar signal retains GCPs in the EGL and enhances proliferation of GCPs in early postnatal cerebellum (Zou et al. 1998; Klein et al. 2001; Reiss et al. 2002). This mechanism is selectively inhibited by Ephrin B2 reverse signaling (Lu et al. 2001). Bone morphogenetic protein 4 and Smad1 signaling have been suggested to promote the differentiation of cells in the EGL (Angley et al. 2003). Neurotrophin-3 accelerates cell cycle exit and differentiation of GCPs (Doughty et al. 1998). While some of these factors are intrinsic, genes expressed by GCPs, others are extrinsic, genes that encode proteins that are secreted and act on GCPs extracellularly.
In knockout mice lacking brain-derived neurotrophic factor (BDNF), another neurotrophin expressed in the cerebellum, a thicker EGL similar to that observed in the double mutants in the present study has been reported (Jones et al. 1994; Schwartz et al. 1997). Secreted BDNF triggers chemotactic response to GCPs through activation of its receptor TrkB on the GCPs to migrate along the BDNF gradient in the cerebellum (Zhou et al. 2007), and Borghesani et al. (2002) have suggested that the thicker EGL in the BDNF mutant mice is caused by defects in initiation of migration of granule cells from the EGL. We postulate that a reduced number of differentiated granule cells that are still in the PMZ in the EGL but are ready for migration underlie this phenomena. Interestingly, in both the BDNF mutants and IP3R2/IP3R3 double mutants, the onset of increased thickness of EGL is similar (P11–12). Similarly, GCP proliferation is not affected in BDNF knockout mice (Choi et al. 2005) and cell death in the cerebellum is increased (Schwartz et al. 1997). Although IP3R2/IP3R3 double mutants do not exhibit obvious neurological phenotypes such as ataxia observed in BDNF mutants (Jones et al. 1994; Schwartz et al. 1997), the remarkable similarities between the cerebellar morphological phenotypes of these mice raise a possibility that there is some functional relationship between BDNF, IP3R2, and IP3R3 in the cerebellum after P12. Enriched distribution of BDNF in the IGL rather than the EGL (Borghesani et al. 2002), which is similar to that of IP3R2 and IP3R3 (see three paragraphs below in discussion) is consistent with this idea. As IP3R2 and IP3R3 mobilize calcium, calcium-dependent expression of BDNF (Shieh et al. 1998; Tao et al. 1998; Chen et al. 2003) or calcium-dependent secretion of BDNF, possibly through calcium-dependent activator protein for secretion-2 (Sadakata et al. 2004) may relate the functions of IP3R2 and IP3R3 to BDNF in this developmental process. Interestingly, calcium-dependent activator protein for secretion-2 knockout mice also exhibit a thicker EGL and increased apoptosis of cerebellar neurons (Sadakata et al. 2007). Alternatively, expression of IP3R2 and IP3R3 may depend on BDNF.
We observed reduced differentiation of GCPs and increased numbers of GCPs in the EGL in the double mutants after P12. As an increased number of GCPs in the EGL proliferate with a doubling time similar to that in the wild-type mice, the corresponding total number of cerebellar granule cells is increased in the double mutants. However, we did not observe an increase in the overall size of the cerebellum or a medulloblastoma-like phenotype in these mice. Instead, we observed increased apoptosis in the cerebellum. Perhaps a number of increased granule neurons produced in the EGL were removed by apoptosis. The observation that increased apoptosis is prominent between P14 and P16 in the ML and IGL (Fig. 6), an age that follows the emergence of a thicker EGL in the double mutants (Fig. 2a), is consistent with this idea.
There are three subtypes of IP3Rs in mammals (Furuichi et al. 1994) and they are expressed in the brain (Yamamoto-Hino et al. 1995; Sharp et al. 1999). Type 1 IP3R (IP3R1) is enriched in cerebellar Purkinje neurons and is also the dominant subtype in cerebellar granule cells at least before the second postnatal week. IP3R1 knockout mice exhibit severe ataxia and epileptic seizures, consistent with a critical role for IP3R1 in the regulation of motor function (Matsumoto et al. 1996; Inoue et al. 1998). IP3R1 also influences the dendritic morphology of Purkinje cells through BDNF production in the granule cells (Hisatsune et al. 2006), but there are no abnormalities in the thickness of the EGL in IP3R1 knockout mice during the development (Matsumoto et al. 1996). Differences in the expression patterns and function of IP3R1, IP3R2, and IP3R3 may underlie these different phenotypes.
We detected IP3R2 expression in the cerebellum only in the deep nuclei before the second postnatal week, and in a fraction of Purkinje cells and granule cells in the IGL (but not in the EGL) thereafter. IP3R3 also is expressed in granule cells (Sharp et al. 1999; Takayama, unpublished observations). Thus, IP3R2 and IP3R3 expressed in granule cells after P12 may promote GCP differentiation. Interestingly, IP3R2 and IP3R3 are expressed by granule neurons in the IGL rather than in the EGL. Therefore, an alternative paracrine or indirect mechanism that attracts GCPs to the IGL may mediate the function of these proteins on GCPs in the EGL. It is reported that IP3R2 and IP3R3 are also expressed in glial cells (Yamamoto-Hino et al. 1995; Sharp et al. 1999). Glial IP3R2 and IP3R3 may contribute to GCP differentiation. Further studies are necessary to clarify the expression pattern and intracellular sites of action of IP3R2 and IP3R3 in the cerebellum and to elucidate the mechanism by which these receptors regulate the differentiation of GCPs.
In summary, we demonstrate that IP3R2 and IP3R3 regulate the differentiation of GCPs in the cerebellum after P12. This is the first report identifying a function for IP3R2 and IP3R3 in the brain in a physiological context. While the underlying molecular mechanisms remain to be elucidated, these results are consistent with intracellular calcium release via IP3R2 and IP3R3 having important roles in postnatal development of the cerebellum by influencing the fate of GCPs. Recently, Brailoiu et al. (2005, 2006) reported that intracellular calcium release from nicotinic acid adenine dinucleotide phosphate-sensitive stores are involved in the neurite extension and PC12 cell differentiation to a neuronal-like phenotype. So, the calcium release from different calcium pools seems to regulate the neuronal differentiation. Further studies will be required to reveal the roles of IP3Rs and other calcium release channels not only during development but also in the adult cerebellum as well as other brain regions.
We are grateful to Dr Toshio Ohshima for helpful discussions and to Mizuki Yamada for technical assistance. We also thank members of the RIKEN animal facility for animal care. The research was supported by the JST and by Grants-in-Aid for scientific research (KM) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.