Fragile X Mental Retardation Protein Regulates Proliferation and Differentiation of Adult Neural Stem/Progenitor Cells

1Fragile X syndrome, one of the most common forms of inherited mental retardation, is caused by the functional loss of fragile X mental retardation protein (FMRP/Fmrp). Patients with fragile X syndrome show an array of deficits in motor control, cognition, learning, and memory, although their overall brain morphology is generally normal. Fmrp is a selective RNA-binding protein that forms a messenger ribonucleoprotein (mRNP) complex that can associate with polyribosomes. Evidence suggests that Fmrp is involved in the post-transcriptional regulation of protein synthesis –. Studies from both human patient brain tissues and Fmrp mutant mice suggest that Fmrp is involved in synaptic plasticity and dendritic development. Fmrp mutant mice are found to perform poorly in highly challenging learning tests, particularly the hippocampus-dependent trace learning test,, suggesting that Fmrp is necessary especially for complex learning that requires an intact hippocampus. However, how the functional deficiency of Fmrp results in learning and memory deficits remains unclear.

2Neurogenesis persists throughout life in two germinal zones, the subgranular zone (SGZ) in the dentate gyrus (DG) of the hippocampus and the subventricular zone (SVZ) of the lateral ventricles. The neurons produced in the DG during adulthood are known to integrate into the existing circuitry of the hippocampus, and young neurons show greater synaptic plasticity than mature neurons under identical conditions,. Although the specific purpose of adult neurogenesis is still being debated, mounting evidence points to an important role in adult neuroplasticity –. It has been suggested that new neurons in the DG are critical for hippocampus-dependent learning,,. Indeed, blocking of adult neurogenesis using generic anti-proliferative drugs or radiation can lead to deficits in learning and memory –. More recent direct evidence has come from inducing the death of new neurons in the hippocampus – and from inhibiting the Wnt signaling pathway in the hippocampus using retrovirus. Adult neurogenesis is regulated at many levels by both extrinsic factors, such as physiological and pathological conditions, and intrinsic factors, such as genetic and epigenetic programs. Although both adult hippocampal neurogenesis and learning are altered in several pathological conditions, such as stress, diabetes, neurological diseases, strokes, and traumatic injuries, the link between adult neurogenesis and mental retardation, a deficiency in learning and memory, remains elusive –.

3The cellular basis of adult neurogenesis is adult neural progenitor/stem cells (aNPCs). The maintenance and differentiation of aNPCs are tightly controlled by intricate molecular networks. Despite exhaustive efforts devoted to understanding transcriptional regulation in adult neurogenesis, the role of translational control by RNA-binding proteins, such as Fmrp, has gone largely unexplored. Recently, Fmrp was found to be required for the maintenance of Drosophila germline stem cells; however, its function in mammalian embryonic neurogenesis is controversial,. Whether and how Fmrp regulates neural stem cells in the adult mammalian brain and the implications for learning and memory have not been established.

4Here we show that loss of Fmrp in vitro and in vivo led to altered adult neurogenesis and impaired learning. Fmrp-deficient aNPCs displayed increased proliferation and decreased neuronal differentiation, but increased glial differentiation. We identified specific mRNAs regulated by Fmrp in stem cell proliferation and differentiation, including glycogen synthase kinase 3β (GSK3ß), a negative regulator of ß-catenin and the canonical Wnt signaling pathway that has been implicated in adult neurogenesis,. The loss of Fmrp resulted in reduced ß-catenin levels and a defective Wnt signaling pathway, which in turn led to the downregulation of neurogenin1 (Neurog1), which is an early initiator of neuronal differentiation and an inhibitor of astrocyte differentiation,. These data not only reveal a novel regulatory role for Fmrp in adult neurogenesis, but also provide direct evidence that adult neurogenesis could be a factor in the pathogenesis of fragile X mental retardation.

5To investigate the role of Fmrp in adult neurogenesis, we determined the expression pattern of Fmrp in the dentate gyrus (DG) of the adult hippocampus using cell type-specific markers. Consistent with published literature,, Fmrp was enriched in a majority of the granule neurons in the DG (Figure S1A), but was undetectable in either GFAP-positive or S100β-positive astrocytes (Figure S1B and S1C). Using markers specific to immature neural progenitors (NPCs) and young neurons, we discovered that Fmrp was also expressed in Sox2 and Nestin double-positive NPCs (Figure 1A), as well as in either NeuroD1-postive or doublecortin (DCX)-positive newly generated neurons (Figure 1B and 1C). The presence of Fmrp in these immature cells supports a potential function of this protein in adult neurogenesis.

6To determine the functions of Fmrp in aNPCs, we isolated aNPCs from both the forebrain and the dentate gyrus (DG) of adult Fmr1 knockout (KO) mice and wild-type (WT) controls. Due to the difficulty of obtaining large numbers of the DG aNPCs, we performed all functional assays first using forebrain aNPCs, and then confirmed our findings using the DG aNPCs. As shown below, we found that both the forebrain aNPCs and the DG aNPCs yielded similar results. Nearly all cultured aNPCs were positive for the progenitor markers Nestin and Sox2 (Figure 1D), suggesting a relative homogeneity of these primary aNPCs. Fmrp was expressed in WT aNPCs, but not in Fmr1 KO aNPCs (Figure 1E). We pulsed the cells with BrdU for eight hours to assess the proliferation of these aNPCs (Figure 1F) and found that Fmr1 KO aNPCs exhibited twice as much BrdU incorporation as WT aNPCs (Figure 1G). We further analyzed the cell cycle profiles of aNPCs and found that more Fmr1 KO cells were in mitotic (G2/M) phase compared with WT controls (Figure S2, 11% higher; n = 3, p<0.02). Hence a lack of functional Fmrp led to a rise in the proliferative capability of aNPCs.

7To assess the effect of Fmrp on aNPC differentiation, both WT and Fmr1 KO forebrain aNPCs were differentiated for three days, and the phenotypes of differentiated cells were determined using several independent assays. First, differentiated cells were stained using cell lineage-specific antibodies, β-III tubulin (Tuj1) for neurons and glial fibrillary acidic protein (GFAP) for astroglia,. Both WT and Fmr1 KO aNPCs could be induced to differentiate into neurons and astrocytes (Figure 2A and 2B); however, Fmr1 KO aNPCs exhibited a 60.4% decrease in neuronal differentiation (Figure 2C) and a 74.9% increase in astrocyte differentiation (Figure 2D) compared with WT aNPCs. Under our culture conditions, only differentiated astrocytes, not proliferating aNPCs, expressed GFAP (data not shown). To validate our immunocytochemical data, we then assessed the neuronal differentiation of aNPCs by measuring the promoter activity of a pan-neuronal transcription factor, neurogenic differentiation 1 (NeuroD1), and astrocyte differentiation by measuring the promoter activity of GFAP using two well-characterized promoter constructs –. We found that in Fmr1 KO aNPCs, NeuroD1 promoter activity decreased by 31.4% (Figure 2E), whereas GFAP promoter activity increased by 73.4% (Figure 2F), which is consistent with our immunocytochemistry results. Finally, using real-time quantitative PCR, we further demonstrated that differentiating Fmr1 KO aNPCs had 17.8% reduced NeuroD1 mRNA (Figure 2G, n = 3, p<0.05), but 1.5×-fold increased GFAP mRNA (Figure 2H; n = 3, p<0.05) levels. Since the above three methods, immunostaining, promoter activity assay, and real-time PCR, yielded consistent results, we used these assays as interchangeable methods for assessing aNPC differentiation in subsequent experiments. The increased proportion of astrocytes in differentiating Fmr1 KO aNPCs was not due to an increased proliferation of newly differentiated astrocytes, because GFAP+ astrocytes differentiated from Fmr1 KO aNPCs did not incorporate more BrdU compared with those from WT aNPCs (data not shown). The differentiation to oligodendrocytes was no different between Fmr1 KO and WT aNPCs (data not shown).

8To confirm that the altered fate specification of Fmr1 KO aNPCs was due to the loss of functional Fmrp, we used siRNA (Fmr1-siRNA, Figure S3) to knock down Fmrp expression in WT aNPCs. We found that acute knockdown of Fmrp expression in WT aNPCs led to both reduced NeuroD1 (Figure 2I, left, n = 4, p<0.05) and Tuj1 (Figure 2I, middle, n = 4, p<0.001) mRNA levels, as well as diminished NeuroD1 promoter activity (Figure 2I, right, n = 6, p<0.05) compared with aNPCs transfected with a nonsilencing control siRNA (NC-siRNA). On the other hand, acute knockdown of Fmrp resulted in increased mRNA levels of both GFAP (Figure 2J, left; n = 4 p<0.01) and another astrocyte marker aquaporin4, (Figure 2J, middle, n = 4, p<0.001), as well as enhanced GFAP promoter activity (Figure 2J, right, n = 6, p<0.05). Furthermore, exogenously expressed WT Fmrp, but not mutant (I304N) Fmrp, which is unable to bind polyribosomes, rescued both the neuronal (Figure 2K) and the astrocyte (Figure 2L) differentiation deficits associated with Fmr1 KO cells.

9We then confirmed that aNPCs isolated from Fmr1 KO DG had similar reductions in neuronal differentiation and increases in astrocyte differentiation (Figure S4A, S4B, S4C, S4D) as Fmr1 KO aNPCs derived from forebrain. In addition, acute knockdown of Fmrp in the WT DG aNPCs resulted in phenotypes in neuronal and astrocyte differentiation (Figure S4E, S4F, S4G, S4H) similar to those we observed in forebrain aNPCs. Together, these results suggest that the loss of Fmrp alters both the proliferation and fate specification of aNPCs.

10To investigate the role of Fmrp in adult neurogenesis in vivo, we assessed the proliferation, survival, and differentiation of endogenous aNPCs in both WT and Fmr1 KO mice. Newborn cells were distinguished by the incorporation of BrdU administered through intraperitoneal injections into adult mice using two cohorts of mice (Figure 3A). Cohort 1 animals (Figure 3C) had the same injection paradigm as those mice used for the differentiation assay (Figure 4); therefore, they were used to assess new cell survival. Cohort 2 animals were used to evaluate cell proliferation in the DG. Quantitative histological analysis at one day following a seven-day regimen of daily BrdU injection (Cohort 1) showed that Fmr1 KO mice had 52.0% more BrdU-positive cells compared with WT mice (Figure 3C). To further assess the proliferation of aNPCs without the confound of cell survival in Fmr1 KO mice, we gave mice six doses of BrdU injection within 24 hours to label the entire proliferating population in the DG based on a published paradigm and analyzed the mice at four hours after the last BrdU injection (Figure 3D, Cohort 2). We found that Fmr1 KO mice had 53.2% more BrdU-positive cells compared with WT mice (Figure 3D, p<0.001). Since the volume of the DG is also increased in Fmr1 KO mice (Figure 3E, p<0.05) and the above data were normalized to the DG volume, the total number of BrdU-positive cells was even higher in KO mice compared with WT controls. It has been shown that the adult DG contains at least two types of proliferating immature cells that can be labeled by BrdU: one type is GFAP+ and Nestin+ (Figure 3F lower panel) and might be stem cells, whereas the other type is GFAP− and Nestin+ (Figure 3F upper panel) and more likely to be progenitor cells,. To determine which types of cells exhibited increased BrdU incorporation in Fmr1 KO mice, we stained the brain sections with antibodies against BrdU, GFAP, and Nestin (Figure 3F). We found that the Fmr1 KO DG had increased BrdU incorporation in both the Nestin+/GFAP− cell population (Figure 3G, 40.8% increase, p<0.05) and the Nestin+/GFAP+ cell population (Figure 3H, p<0.001, 1.2-fold increase). The proliferation of astrocytes (BrdU+, GFAP+, Nestin− cells) was no different between WT and Fmr1 KO mice (data not shown). Cell proliferation in the SVZ was also 1.1-fold higher in Fmr1 KO mice (p<0.05). Thus Fmrp deficiency may lead to increased proliferation of both stem and progenitor cells.

11The long-term survival and differentiation of BrdU-labeled cells was evaluated by analyzing the labeled cells at four weeks after BrdU injections (Figure 4A–4C). The number of BrdU+ cells at four weeks post-injection was no different between WT and Fmr1 KO mice (Figure 4D); therefore, the percentage of BrdU+ cells that survived from one day to four weeks post-BrdU administration is significantly lower in Fmr1 KO mice compared with WT mice (Figure 4E, p<0.05). Hence Fmrp deficiency may also lead to reduced survival of young neurons.

12Since we observed altered neuronal and astrocyte differentiation of Fmr1 KO aNPCs in vitro (Figure 2), we then used triple fluorescence immunostaining with antibodies for mature neurons (NeuN) and astrocytes (S100β) to further determine the fate of differentiated aNPCs in vivo (Figure 4B and 4C). Consistent with our in vitro observation, we found that in Fmr1 KO mice, the percentage of BrdU+ cells that are NeuN+ neurons was 10.4% lower (Figure 3F, p<0.05), whereas the percentage of BrdU-positive cells that are S100β+ astrocytes was 75.7% higher compared with WT mice (Figure 4G, p<0.05). In addition, the expression levels of NeuroD1 and Neurog1, two transcription factors expressed in new neurons, were also reduced in the hippocampus of Fmr1 KO mice (Figure S5, n = 3, p<0.05). Therefore, the loss of Fmrp leads to reduced neuronal differentiation but greater glial differentiation in aNPCs residing in the DG. These in vivo data along with our in vitro results suggest that Fmrp indeed plays important roles in regulating the differentiation and proliferation of aNPCs.

13As an RNA-binding protein, Fmrp is known to bind to a subset of specific mRNAs and suppress their translation. To identify the mRNAs that are regulated by Fmrp in aNPCs, we employed the strategy of specifically immunoprecipitating Fmrp-containing mRNP particles and identifying the copurified mRNAs by probing expression microarrays, which we established previously. Due to the large quantity of cells needed, we only used forebrain aNPCs derived from WT and Fmr1 KO mice for immunoprecipitation with an antibody that could specifically precipitate Fmrp (Figure 5A). Both immunoprecipitated and input RNAs were used to probe Affymetrix arrays (data not shown). The mRNAs of interest were further confirmed to be associated with Fmrp by independent IP and real-time PCR (Figure 5B). Among these mRNAs, we found several already known to be regulated by Fmrp, such as MAP1B and EF1α, confirming the specificity of our assay (Figure 5B and 5C). Also among the identified mRNAs, we found two key factors well established as enhancers of cell cycle progression, cyclin-dependent kinase 4 (CDK4) and cyclin D1. Their specific association with Fmrp was further confirmed by additional IP and RT-PCR (Figure 5B). We therefore examined the expression levels of CDK4 and cyclin D1 in both WT and Fmr1 KO aNPCs. Though there was no significant change in the mRNA levels (Figure S6B), the loss of Fmrp led to higher protein levels of both genes (Figure 5C, Figure S6). Both CDK4 and cyclin D1 expression levels are important for the proliferation of neural progenitors,,,. We found that a chemical inhibitor of CDK4 could partially rescue the proliferation phenotype of Fmr1 KO aNPCs (Figure S6C). Hence increased expression of CDK4 and cyclin D1 as a result of Fmrp deficiency could be responsible for the increased proliferation of Fmr1 KO aNPCs.

14We also noticed that the mRNA of GSK3ß, known to be involved in the Wnt signaling pathway, could be coimmunoprecipitated with Fmrp from aNPCs. We confirmed the specific association between Fmrp and the mRNA of GSK3ß using additional Fmrp IP coupled to real-time PCR (Figure 5B). Furthermore, we confirmed that the loss of Fmrp led to increased protein levels of GSK3β (Figure 5C) and reduced protein levels of ß-catenin (Figure S7), a downstream target of GSK3β in proliferating Fmr1 KO aNPCs.

15To determine whether Fmrp could regulate the translation of GSK3β protein, we cloned the 3′ untranslated region (3′UTR) of GSK3β and inserted it into the 3′ region of the Renilla luciferase coding sequence, such that the translation of Renilla luciferase could be regulated by the 3′UTR of GSK3β. Upon transfection of this construct into Fmr1 KO and WT aNPCs, we observed significantly higher Renilla luciferase activity in Fmr1 KO aNPCs compared with WT aNPCs, suggesting that the 3′UTR of GSK3ß leads to increased translational activity in Fmr1 KO cells (Figure S7A). To further ensure that this increased protein level was due to increased translation rather than reduced protein stability of GSK3β in Fmr1 KO cells, we treated Fmr1 KO and WT aNPCs with the protein synthesis inhibitor cycloheximide over a 24-hour period. We found that, even though the GSK3β protein level was higher in the KO cells (time 0 h), there was no significant difference in the rate of GSK3β protein degradation between WT and KO aNPCs (Figure S7B). Therefore, these data suggest that Fmrp regulates the protein translation of GSK3β.

16The canonical Wnt pathway is known to be critical for adult neurogenesis, but the downstream effectors have been a mystery,. Since Fmrp was able to regulate the translation of GSK3β, we further investigated whether the activity of the Wnt pathway was altered in Fmr1 KO aNPCs. GSK3β is known to phosphorylate and promote the proteasome degradation of ß-catenin, a central player in the Wnt signaling pathway. We therefore chose to examine the expression of ß-catenin in aNPCs. In both proliferating and differentiating aNPCs, we observed increased GSK3β protein levels (Figure 5C) and decreased expression of ß-catenin (Figure 6A and Figure S7C). Hence Fmrp may promote adult neurogenesis by regulating the expression of GSK3β and subsequently ß-catenin.

17In the absence of Wnt, ß-catenin is known to be held in cytosol and degraded by a collection of regulatory factors, such as GSK3β. The activation of Frizzled by Wnt leads to stabilization and nuclear translocation of ß-catenin, which forms a complex with TCF/LEF transcription factors and induces the expression of downstream target genes. To confirm that loss of Fmrp led to the deficit in the Wnt signaling pathway, we used a well-characterized luciferase reporter system for monitoring the activity of the Wnt signaling pathway,. Upon growth factor withdrawal and activation by cotransfected Wnt3a expression vector, Fmr1 KO aNPCs exhibited significantly reduced luciferase activity compared with WT aNPCs (Figure 6B). In addition, expression of Axin2, a downstream effector of the Wnt signaling pathway, was reduced in the hippocampus of Fmr1 KO mice (Figure S8). Therefore, the Wnt signaling pathway is indeed defective in the absence of Fmrp. In addition, treatment of Fmr1 KO aNPCs with a well-established GSK3β inhibitor SB216763 could enhance the Wnt signaling pathway and partially rescue the neuronal (Figure 6C and 6D) and astrocyte (Figure 6E and 6F) differentiation deficits in aNPCs. Similar results were also obtained using the DG aNPCs (Figure S9B and S9C). Interestingly, SB216763 also repressed aNPC proliferation without affecting cyclin D1 expression levels (Figure S10). Therefore, Fmrp deficiency leads to reduced Wnt signaling, which could be responsible for altered aNPC differentiation.

18The basic helix-loop-helix family transcription factor neurogenin1 (Neurog1) can be regulated by Wnt signaling, and its promoter contains one single classic TCF/LEF binding element. We therefore assessed the mRNA levels of Neurog1 in Fmr1 KO proliferating and differentiating aNPCs. Neurog1 was transiently expressed in differentiating WT aNPCs (Figure 6G), as shown previously. We found that Neurog1 mRNA levels indeed decreased in Fmr1 KO differentiating aNPCs (Figure 6G). To determine whether the altered Neurog1 expression resulted from a Wnt signaling deficit in Fmr1 KO aNPCs, we created a reporter construct that has a mouse native Neurog1 promoter driving the expression of luciferase. When transfected into Fmr1 WT and KO aNPCs that were subjected to differentiation, the Neurog1-luciferase reporter yielded detectable luciferase activity only in the presence of Wnt3a (Figure 6H), indicating that this promoter is activated by Wnt signaling. As expected, we found that Neurog1 promoter activity was significantly reduced in differentiating Fmr1 KO aNPCs compared with WT cells (Figure 6H). Furthermore, we could rescue the Neurog1 promoter activity by expressing the wild-type but not the mutant Fmr1 in Fmr1 KO aNPCs (Figure 6I). Taken together, these data suggest that the expression of Neurog1 is controlled by Fmrp through the Wnt signaling pathway in aNPCs.

19Since Neurog1 is an early initiator of neuronal differentiation and an inhibitor of glial differentiation, its downregulation could be responsible for the reduced neuronal differentiation and increased glial differentiation seen in Fmr1 KO aNPCs. To test this possibility, we expressed exogenous Neurog1 in Fmr1 KO forebrain aNPCs and found that exogenously expressed Neurog1 could rescue the altered fate specification of Fmr1 KO aNPCs, as assessed by the mRNA levels of neuronal genes (Figure 7A, NeuroD1 and Tuj1) and astrocytic genes (Figure 7B, GFAP and aquqporin4), as well as the promoter activity of NeuroD1 and GFAP (data not shown) in differentiating cells. To further validate the role of Neurog1 in aNPC differentiation, we acutely knocked down Neurog1 expression in aNPCs using siRNA (Figure 7C) and found that acute knockdown of Neurog1 in aNPCs led to decreased neuronal differentiation (Figure 7D), but increased astrocyte differentiation (Figure 7E), reminiscent of what we found in Fmr1 KO aNPCs. Similar results were also obtained using the DG aNPCs (Figure S11). Therefore, our findings suggest that Fmrp regulates aNPC fate specification by modulating the activity of the Wnt/β-catenin signaling pathway and subsequently its downstream effector, Neurog1 (Figure 7F).

20In this study we demonstrate that the loss of functional Fmrp in aNPCs leads to reduced neurogenesis both in vitro and in vivo. We show that Fmrp regulates the translation of several factors involved in stem cell proliferation and differentiation, including CDK4, cyclin D1, and GSK3β. As a result of dysregulation of GSK3β and the Wnt signaling pathway, the expression level of Neurog1, one of the Wnt-regulated genes, is reduced, which is likely responsible for the reduced neuronal differentiation and increased astrocyte differentiation seen in Fmr1 KO aNPCs. Our data demonstrate that Fmrp plays profound regulatory roles in adult neurogenesis.

21Despite exhaustive efforts devoted to understanding transcriptional regulation in adult neurogenesis, the role of translational control in adult neurogenesis has gone largely unexplored; yet our results indicate that translational control is just as important, if not more so, in the regulation of aNPC functions. We have identified the molecular pathways by which Fmrp regulates aNPC proliferation and fate specification. Both a previous study from another group and our current study found that the mRNAs of both CDK4 and cyclin D1 could be bound by Fmrp. CDK4 and cyclin D1 are well-characterized cell-cycle regulators in many cell types. In mammalian neural progenitor cells, increased cyclin D1 expression is positively correlated with their proliferation, and reduced cyclin D1 levels result in decreased proliferation. CDK4 has been shown to regulate the proliferation of neural progenitors in adult brains, and inhibition of CDK4 activity leads to growth arrest in neural progenitors. The fact that we could rescue the proliferation deficits of Fmr1 KO aNPCs using a chemical inhibitor of CDK4 supports our model that Fmrp regulates aNPC proliferation in part through CDK4.

22We also found here that Fmrp could bind and regulate the translation of GSK3β mRNA. As a negative regulator of the Wnt signaling pathway, GSK3β promotes the degradation of β-catenin and inhibits the activity of the canonical Wnt signaling pathway. The Wnt signaling pathway has been shown to promote the proliferation of a number of cell types, including hematopoietic stem cells. Although one study suggests that Wnt signaling can also promote cell proliferation in the DG, other publications clearly point out the function of the Wnt signaling pathway in activating neuronal differentiation during adult neurogenesis, and inhibiting this pathway results in hippocampus-dependent learning deficits,,. Our data show Fmr1 KO aNPCs had reduced Wnt signaling, and we identified Neurog1 as one of the downstream targets of Fmrp and Wnt. Neurog1 is a transcription factor expressed only at the early stage of differentiation, and it promotes neuronal differentiation while inhibiting astrocyte differentiation,. Neurog1 contains a conserved Tcf/Lef binging site in its promoter, allowing it to sense the levels of Wnt signaling. Although the Wnt signaling pathway has been found to enhance cyclin D1 transcription in HeLa cells and several other cell types, we saw no such activation in aNPCs. Interestingly, enhancing Wnt signaling via a Gsk3β inhibitor repressed proliferation of Fmr1 KO aNPCs, possibly due to the neuronal differentiation effect of Wnt signaling. It is likely that in aNPCs, Wnt signaling and cyclin D1 act independently on cell proliferation, and they are both also regulated by Fmrp.

23Several studies have examined embryonic and early postnatal neurogenesis in mice and humans. One study found that the loss of Fmrp led to increased neuronal differentiation and reduced glial differentiation in mice. Due to the large scale of embryonic neurogenesis, factors affecting aNPCs would also be expected to affect both the overall number of neurons, as well as brain size. However, neither adult fragile X patients nor adult Fmr1 KO mice show any differences in the number of neurons and glia compared with controls, raising questions about the potential significance of increased early neurogenesis to the pathogenesis of fragile X syndrome. Another study found no alteration in the differentiation of embryonic NPCs (eNPCs) isolated from one human embryo diagnosed with a fragile X mutation. While the discrepancies between human and mouse studies require further confirmation using additional human tissues, the different phenotypes observed in Fmrp-deficient eNPCs versus aNPCs support the idea that adult neurogenesis is subjected to regulatory mechanisms distinct from those in embryonic neurogenesis. First, during adult neurogenesis, multipotent aNPCs are in intimate contact with the surrounding mature neurons and glia, and the fate of aNPCs can be affected by their microenvironment,,,. Mice that lack Sonic hedgehog, Tlx, Bmi1,, and Mbd1 have all exhibited profound deficits in postnatal neurogenesis, but not in their embryonic neural development. In fact, in prenatal and early postnatal developing brains, Fmrp is widely expressed in neural cells, including glia and glial precursors, with the levels of Fmrp decreasing during oligodendrocyte differentiation,, whereas in adult brains, Fmrp is expressed predominantly in neurons, with negligible expression in mature glia,. Further studies into the role of Fmrp in both embryonic and adult neurogenesis would facilitate our understanding of the unique molecular networks that regulate eNPCs and aNPCs at the level of translational control.

24Hippocampal neurogenesis has been associated with hippocampus-dependent learning,, and blocking neurogenesis using methods nonexclusive to adult NPCs or new neurons has supported this model –,. Altered adult hippocampal neurogenesis and impaired learning have been found in several pathological conditions,; however, the possibility of a link between adult neurogenesis and human mental retardation disorders, though recently put forward, has not been studied well. Although there is a low level of DG neurogenesis in adults, mounting evidence points to its potentially important role in neuroplasticity, emotional behavior, and the higher cognitive functions of adult brains. It has been proposed that adult neurogenesis enables the lifelong adaptation of the hippocampal network to the levels of novelty and complexity a person experiences. Using precise and unbiased stereological methods coupled with confocal microscopy, we observed a mild but significant reduction in the number of new neurons in Fmr1 KO mice, which could easily have been missed by others who employed non-stereology quantification methods. Due to the restricted nature and low level of adult neurogenesis, a lack of Fmrp may not affect the total number of neurons in adult brains, but it can contribute to pathological conditions linked to higher cognitive functions and learning abilities,. The learning deficits of the Fmr1 KO mice may be the result of both reduced neurogenesis and defective neuronal maturation. Consistent with the literature, we have observed that Fmr1 KO aNPC-differentiated neurons had reduced dendritic complexity and length (data not shown), which could also contribute to behavioral deficits. In addition, although Fmr1 KO mice have increased proliferation, at four weeks post-BrdU labeling, both KO and WT mice had similar numbers of surviving new cells, possibly due to the decreased survival of new cells in KO mice. How Fmrp regulates the survival of young neurons is another interesting question that is currently being pursued as an independent study. One mystery that remains to be cleared up is why the size of the DG in the adult Fmr1 KO mice is bigger than in controls. Since the new cells generated in the adult DG account for only a small portion of the total DG cells, increased proliferation of these new cells may not contribute much to the increased size of the DG. Castren et al. have shown that Fmr1 KO mice exhibit increased cell proliferation in the subventricular zone during embryonic development (E13). The mammalian DG is formed during the postnatal period, with P7 as the peak of cell genesis. It is possible that increased cell proliferation during DG formation results in an increased DG volume that persists into adulthood. In addition to its function in the initial stage of neurogenesis, Fmrp-deficient neurons are known to have reduced dendritic complexity,; therefore, it is possible that new neurons generated in the adult DG also have reduced dendritic complexity. Hence deficits in several stages of adult neurogenesis could contribute to the higher brain functions, such as the learning and emotional disabilities associated with fragile X patients, without significantly affecting the gross brain structure of human patients.

25Our results suggest that translational regulation by Fmrp in aNPCs and young neurons is essential for learning and memory, and the reduced number of new neurons together with defective maturation of these new neurons may contribute to the cognitive deficiency seen in fragile X patients. This is a facet of the etiology of fragile X syndrome that has not been recognized before.

26All animal procedures were performed according to protocols approved by the University of New Mexico Animal Care and Use Committee. The Fmr1 KO mice bred onto the C57B/L6 genetic background were as described previously.

27Adult aNPCs used in this study were isolated from 8- to 10-week-old male Fmr1 KO mice and wild-type (WT) controls based on published methods: for the forebrain aNPCs and for the DG aNPCs. (See Text S1 for details.)

28These analyses were carried out using our established method,. (See Text S1 for details.)

29In vivo neurogenesis analyses were performed essentially as described previously,. These experiments have been performed using 3 different batches of animals, with n = 4–6/genotype each batch. For the first two batches, BrdU (50mg/kg) was injected into 8-week-old mice daily for 7 consecutive days to increase the amount of labeling. Mice were then euthanized 1 day post-injection to assess the in vivo proliferation (and early survival) of labeled cells. For cell survival analysis, another group of mice was injected with BrdU at 8 weeks of age and euthanized 4 weeks post-injection. The third batch of mice, on the other hand, were given 6 injections of BrdU (50 mg/kg) within 24 hours to label all dividing cells in the DG within this time period and sacrificed at 4 hours post-last injection based on a published protocol. Mice were euthanized by intraperitoneal injection of sodium pentobarbital, and then transcardially perfused with saline followed by 4% PFA. Brains were dissected out, post-fixed overnight in 4% PFA, and then equilibrated in 30% sucrose. Forty-µm brain sections were generated using a sliding microtone and stored in a −20°C freezer as floating sections in 96-well plates filled with cryoprotectant solution (glycerol, ethylene glycol, and 0.1 M phosphate buffer, pH 7.4, 1∶1∶2 by volume). We performed immunohistological analysis on 1-in-6 serial floating brain sections (240 µm apart) based on the published method. (Please see Text S1 for more details.)

30The DNA plasmids carrying 2.5 kb of glial fibrillary acidic protein (GFAP) promoter-firefly luciferase reporter gene (GF1L-pGL3) or its mutant version, with the STAT3 binding site mutated (GF1L-S-pGL3), and an internal control plasmid containing sea pansy luciferase driven by human elongation factor 1α promoter (EF1α-Luc) were as described previously,. NeuroD1-luciferase, a gift from Dr. F.H. Gage, was then cloned into pGL3 plasmid. Fmr1-siRNA, control-siRNA, and mouse Neurog1 expression vector were purchased from Open Biosystems (www.openbiosystems.com). Neurog1 siRNA was purchased from SABiosciences (Frederick, MD). Wild-type Wnt reporter construct pTOPFLASH containing 8 TCF/LEF binding sites and mutant reporter construct pFOPFLASH were gifts from R.T. Moon (University of Washington) as described. Wnt3a expression plasmid was a gift from Dr. D.C. Lie (Institute of Developmental Genetics, Germany) as described. Wild-type FLAG-Fmrp was cloned into pDEST-27 vector, and mutant FLAG-I304N was generated by site-directed mutagenesis (Stratagene). All the constructs were verified by DNA sequencing. Myelin basic protein (MBP) promoter was cloned from mouse genomic DNA based on published information and cloned into pGL3 plasmid. The mouse Neurog1 promoter, containing its native TCF/LEF binding site “cctttgaa,” was cloned by PCR based on the GenBank sequence (GenBank ID #18014) using the following primers: 5′-GTCTGACTCTGAAGCCATCTCTGA-3′ (forward) and 5′ -ACGCGCCGGGCTGGTCTCCT-3′ (reverse). The PCR product was then subcloned into the pCRII-TOPO plasmid, sequenced, and inserted into the KpnI-XhoI site of the pGL2 basic vector to yield Neurog1-luciferase reporter construct. The full-length 3′-UTR of GSK-3ß mRNA was PCR-amplified directly from proliferating aNPC first-strand cDNA generated from 5 µg TRIzol-isolated total RNA using oligo-dT SuperScript III reverse transcription according to the manufacturer's protocol (Invitrogen, Cat. #1808-093). It was cloned into pIS2 Renilla luciferase vector, and pIS0 firefly luciferase was used as a transfection control.

31Electroporation of plasmid DNA into aNPCs and the luciferase assay were carried out using an Amaxa Nucleofector electroporator based on the manufacturer's protocol (Amaxa, #VPG-1004) with modifications. Briefly, 2×106 cells were trypsinized, resuspended in Nucleofector solution, mixed with DNA, and electroporated using a preset program for mouse NPCs (#A033). The cells were then plated onto polyornithin/laminin-coated 24-well plates in proliferation medium. After 24 h, cells were changed into differentiation medium for 48 h. Transfection of aNPCs was carried out using Stemfect (Stemgent, San Diego, CA) based on the manufacturer's protocol with modifications. Briefly, aNPCs were plated into 24-well P/L-coated plate for 24 hours. Then 3 µg DNA and 0.9 µl Stemgent reagent were mixed, incubated for 10 minutes, and then added to the cells. Sixteen hours later, the transfected cells were changed into differentiation medium for 48 hours. The cells were then collected and luciferase activity was detected using the Dual-Luciferase Reporter 1000 System (Promega, Cat# E1980) based on the manufacturer's protocol. Briefly, collected cells were lysed in 100 µl of 1× passive lysis buffer at room temperature for 15 minutes. Then 20 µL of the lysate was added to 100 µl of Luciferase Assay Buffer II and mixed briefly. Firefly luciferase (F-luc) activity was immediately read using a SpectraMax M2E plate reader (Molecular Devices Corp.). Next, 100 µl of Stop & Glo Buffer with Stop & Glo substrate was added and mixed briefly. Renilla luciferase (R-luc) activity was immediately read. F-luc activity was normalized to R-luc activity to account for variation in transfection efficiencies. Each experiment was independently repeated 3 times. For each electroporation, 3 µg (NeuroD1− or GFAP−) luciferase DNA, 5 µg Neurog1-luciferase DNA, 0.2 µg R-Luc, and 0.004–2 µg Fmr1, Neurog1, or control expression plasmids were used.

32These procedures were carried out as described. (Please see Text S1 for details.)

33Twenty-mg protein samples were separated on SDS-PAGE gels and then transferred to PVDF membranes (Millipore). Membranes were processed following the ECL western blotting protocol (GE Healthcare). anti-MAP1B (a gift from I. Fischer, Drexel University, Philadelphia), anti-Nestin (Millipore), anti-Fmrp (7G1-1), anti-Fmrp (John Louis), anti-β-catenin (Millipore), anti-CDK4 (Millipore), anti-Cyclin D1 (Upstate), anti-TCF4 (Abcam), GSK3ß (Abcam), anti-EF1α (ATCC), anti-Neurog1 (Millipore), anti-NeuroD1 (Santa Cruz), anti-Axin2 (Cell Signaling) and anti-β-Actin (Abcam) were used as primary antibodies at the concentrations recommended by the manufacturers. HRP-conjugated secondary antibodies were obtained from Sigma. For loading controls, membranes were stripped and reprobed with the antibody against eIF5α (Santa Cruz Biotechnology), anti-GAPDH (Ambion), or eIF4E (Transduction Laboratories). To test the efficiency of Fmr1-siRNA, Fmrp expression plasmid and siRNA expression plasmid were cotransfected into HEK293 cells, and the mRNA and protein expression levels of Fmrp were analyzed using PCR and western blot, respectively.

34Statistical analysis was performed using ANOVA and Student's t-test, unless specified with the aid of SPSS v.17. All data were shown as mean with standard error of mean (mean ± SEM). Probabilities of P<0.05 were considered significant.