Fragile X Syndrome (FXS) is the most common single-gene cause of autism spectrum disorder and intellectual impairment (Wang et al., 2012), with a frequency of about 1 in 7,100 males and 1 in 11,100 females (Hunter et al., 2014). FXS is associated with impairments in learning and memory, hyperactivity, hypersensitivity to sensory stimuli, increased susceptibility to seizures, and autistic behaviors (Penagarikano et al., 2007). The majority of FXS cases result from the epigenetic transcriptional silencing of the Fragile X Mental Retardation 1 gene (FMR1), which subsequently prevents the expression of fragile X mental retardation protein (FMRP) (Pieretti et al., 1991; Verkerk et al., 1991; Coffee et al., 1999, 2002; Eiges et al., 2007). FMRP is responsible for the translational regulation of many mRNAs (Miyashiro et al., 2003; Darnell et al., 2011; Ascano et al., 2012) including those involved in synaptic plasticity (Bassell and Warren, 2008; Darnell et al., 2011), and has been implicated in regulating the proliferation and differentiation of embryonic and adult neural precursor cells (NPCs) (Castrén et al., 2005; Eadie et al., 2009; Tervonen et al., 2009; Luo et al., 2010; Guo et al., 2011). A loss of FMRP and the associated memory and learning impairments observed in FXS, may therefore be linked to aberrant regulation of NPCs in the hippocampus.

There is a strong correlation between learning and memory capacity and hippocampal neurogenesis. A hallmark of the dentate gyrus (DG) of the hippocampal formation is lifelong neurogenesis to which NPCs contribute (Altman and Das, 1965). The enhanced plasticity of immature newborn neurons is suggested to enable learning and memory (Deng et al., 2010; Snyder and Cameron, 2012). In fact, many neurological diseases marked by cognitive decline such as Alzheimer’s, Parkinson’s, depression, and epilepsy, all show alterations in hippocampal neurogenesis (Zhao et al., 2008). Similarly, conditional knockout of FMRP in adult NPCs results in increased NPC proliferation, and the consequent impairment of hippocampus dependent learning (Guo et al., 2011). Notably, the majority of DG granule cells are born postnatally in the rodent brain (Altman and Bayer, 1990), and most adult born neurons are generated from early-born NPCs that reside in the DG as it is forming (Mathews et al., 2010). It is unknown whether early postnatal dysregulation of NPCs has ramifications on neurogenesis throughout life, or whether later interventions can correct abnormalities. Regulation of neurogenesis via FMRP is likely to impact early-born NPCs in the developing DG since expression of FMRP peaks at postnatal day 7 (Lu et al., 2004), which may underlie abnormal hippocampus-dependent memory in 3-week-old Fragile X mice (Bilousova et al., 2009).

The role of FMRP and its downstream effects have been largely limited to studies of neural populations in isolation. However, emerging evidence suggests that the glial environment and the role of FMRP in these cells are just as critical for proper brain development. Astrocytes for example, previously believed to be only support cells, secrete factors that actively regulate neurogenesis, neural function and communication (Barres, 2008). When cortical astrocytes devoid of FMRP are co-cultured with wild type (WT) neurons, the growth of the neurons is affected and normal synaptic formations among them are limited (Jacobs and Doering, 2010a). Of particular interest, the astrocytes within the hippocampus have been shown to promote NPC proliferation and neuronal differentiation (Song et al., 2002), and we hypothesized that a lack of FMRP in astrocytes may contribute to altered neural precursor/stem cell production.

To test this hypothesis, we used the Neural Colony Forming Cell (NCFC) Assay to examine the proliferation of neural stem cells within WT or Fragile X mice (Fmr1-KO) hippocampal neurospheres cultured in the presence of astrocyte conditioned media (ACM). The NCFC assay is specifically designed to distinguish between functional stem cells that form neurospheres more than 2 mm in diameter from neurospheres produced by the general precursor cell population with limited proliferative capacity (Louis et al., 2008). We found that Fragile X progenitor-derived neurospheres showed restricted proliferation in the presence of both WT and Fmr1-KO ACM. On the other hand, the proliferation of WT neural progenitor-derived neurospheres selectively decreased with Fmr1-KO ACM, while that of WT stem cell-derived neurospheres was enhanced. Interestingly, one population of WT neural progenitor-derived neurospheres showed decreased proliferation in the presence of cortical Fmr1-KO ACM, but not hippocampal Fmr1-KO ACM, highlighting a regional difference in astrocyte-secreted factors. Additionally, we compared the protein expression profiles of secreted factors from cortical and hippocampal astrocytes between WT and Fmr1-KO brains. Of the multiple proteins with differential expression that we detected, we identified the following proteins using mass spectrometry: multidrug resistance protein 1B, microphthalmia-associated transcription factor (MITF), haptoglobin, fasciculation, and elongation protein zeta 2, antithrombin III, and serum albumin. Together, these results indicate that signals derived from Fmr1-KO astrocytes affect the proliferation of WT neurospheres, and that Fmr1-KO neurospheres have intrinsic deficits in responding to environmental cues as evident in their indiscriminate response to WT versus Fmr1-KO ACM.

In this study, we examined potential differences in the proportions of neural progenitor and neural stem cells between WT and Fmr1-KO NPCs in vitro using the NCFC Assay. Our results confirmed that the NCFC Assay generates four size categories of neurospheres, in turn reflecting different proliferative potentials in the NPC population (Louis et al., 2008). We also found that Fmr1-KO neural progenitor cells had a non-autonomous cell deficit in correctly responding to astrocyte secreted factors. This was evident in the indiscriminate response of Fmr1-KO neurospheres to ACM from WT or Fmr1-KO astrocytes, which contrasted with the specific response of WT neurospheres to Fmr1-KO, but not WT, ACM (Table 2). Interestingly, Fmr1-KO ACM increased the proliferation of WT functional neural stem cells, highlighting the abnormal signals that astrocytes send to NPCs in the absence of FMRP.

NPCs comprise populations of neural stem and progenitor cells. Neural stem cells self-renew and are multipotent, giving rise to all neural lineages, namely neurons, astrocytes, and oligodendrocytes. Neural progenitor cells, on the other hand, have limited proliferative capacity and restricted potential in differentiating to distinct lineages. Previous research has reported enhanced NPC proliferation in the embryonic SVZ (Castrén et al., 2005), the embryonic VZ and SVZ combined (Tervonen et al., 2009), and the adult DG (Luo et al., 2010; Guo et al., 2011) in association with lack of FMRP. Given this, we expected to observe differences between WT and Fmr1-KO cultures in the proportions of neural progenitor- or stem cell-derived neurospheres generated. Our results alternatively show that there were in fact no inherent differences between the proliferative propensities of the neurospheres of each genotype, in line with Eadie et al. (2009), who reported no differences in the proliferation of NPCs in the DG of adult Fmr1-KO mice in vivo (Eadie et al., 2009). It is noteworthy to mention that our study is the first to examine hippocampal NPCs from the Fmr1-KO mouse brain on an FVB genetic background. While we detected no differences in neural progenitor and stem cell proliferation in our culture system between WT and Fmr1-KO cultures after 21 DIV, differences may exist at other time points especially since FMRP is developmentally regulated (Till et al., 2012). Till et al. (2012) indicate that Fmr1-KO mice show delays in cortical development at time points that coincide with the highest FMRP levels in normal brains. Notably, astrocytes of the developing hippocampus express FMRP (Pacey and Doering, 2007), suggesting that astrocytes contribute to the hippocampal impairments seen in Fragile X.

We identified a subset of astrocyte-secreted factors whose expression was up- or down-regulated by at least 50% in Fmr1-KO samples. These included multidrug resistance protein 1B, MITF, haptoglobin, and FEZ2 in cortical ACM, and antithrombin III and serum albumin in hippocampal ACM. These factors were all downregulated with the exception of antithrombin III (Supplementary Tables S1 and S2). Notably, the expression of multidrug resistance protein 1 has been reported in human NPCs where it promotes the proliferation of NPCs (Yamamoto et al., 2009). Thus, our finding of decreased multidrug resistance protein 1 expression in Fmr1-KO cortical ACM is in agreement with the decreased neural progenitor proliferation detected in the presence of Fmr1-KO cortical ACM. Unlike the multidrug resistance family of proteins, the function of the FEZ family of proteins is elusive. Previous research has found that FEZ2 interacts with 59 proteins in categories that include transcription, translation, apoptosis, signal transduction, neuronal cell development, and cytoskeleton and centrosome (Alborghetti et al., 2011). The function of FEZ2 remains poorly understood at this time. The expression of albumin and antithrombin III in the developing brain under normal conditions suggests that they play an important role in brain development (Dziegielewska et al., 1984; Deschepper et al., 1991). In fact, astrocytes of the developing brain have been shown to secrete antithrombin III in culture, similar to our findings (Deschepper et al., 1991). Albumin is also necessary for astrocytes to provide energy to the developing brain (Tabernero et al., 1999) and to synthesize oleic acid (Tabernero et al., 2001). In turn, oleic acid promotes neuronal differentiation (Tabernero et al., 2001). Given that astrocytes show developmental delays in FXS (Jacobs et al., 2010), we suggest that the expression levels of albumin and antithrombin III that we detected in Fmr1-KO hippocampal ACM may mirror the expression levels in ACM from the WT hippocampus at earlier time points. We are currently studying the role of the astrocyte secreted factors in Fragile X in order to better understand the extent of astrocyte involvement in Fragile X and associated neurodevelopmental disorders.

Astrocyte-secreted factors, such as thrombospondin-1 (Lu and Kipnis, 2010), fibroblast growth factor-2 (Kirby et al., 2013), and clusterin (Cordero-Llana et al., 2011) promote NPC proliferation, neurogenesis, or both. Indeed, astrocytes are an important component of the hippocampal neurogenic niche (Song et al., 2002). Our previous work has shown that signals from Fmr1-KO astrocytes can alter the development and growth of co-cultured WT neurons, an effect not seen with WT astrocytes (Jacobs and Doering, 2010a). Thus, we hypothesized that ACM from WT astrocytes would correct aberrant proportions of progenitor and stem cell-derived Fmr1-KO neurospheres, whereas Fmr1-KO ACM would alter the proportions of each of the WT neurosphere size categories. Our results demonstrated that ACM from Fmr1-KO astrocytes did alter the proliferation of WT neural progenitor and functional stem cells, restricting the proliferation of neural progenitors while increasing the proliferation of neural stem-derived neurospheres. It remains to be determined why Fmr1-KO ACM exerts opposite effects on neural progenitor compared to neural stem cells.

Intriguingly, the percentage of WT neurospheres measuring ≤0.5 mm increased in the presence of WT and Fmr1-KO ACM from the cortex and the hippocampus, similar to Fmr1-KO neurospheres of the same size category. These neurospheres have a limited ability for self-renewal (Louis et al., 2008). The generic response of WT neurospheres to ACM is interesting, as it was only detected in this size category, and mirrors the effect of ACM on Fmr1-KO neurospheres. This suggests a non-cell autonomous response of WT NPCs of limited potential as demonstrated by their generic response to astrocyte-secreted factors regardless of FMRP expression.

Lack of FMRP caused a defect in Fmr1-KO NPCs to respond correctly to environmental cues. Fmr1-KO neurospheres showed an indiscriminate response toward WT and Fmr1-KO ACM. We found that ACM from WT and Fmr1-KO astrocytes caused a global increase in the proportion of Fmr1-KO neurospheres ≤0.5 mm. In parallel, ACM decreased the proportion of Fmr1-KO neurospheres 1–2 mm in diameter relative to Fmr1-KO neurospheres of the same size without ACM. Collectively, this indicates that ACM restricted the proliferation of Fmr1-KO neural progenitors without affecting the proliferation of stem cells. The fact that the percentages of some Fmr1-KO neurosphere size categories (i.e., 0.5–1 mm and ≥2 mm) did not change in the presence of ACM suggests that not all NPC populations are targets of astrocyte-secreted factors in Fragile X.

In contrast to Fmr1-KO NPCs, only ACM from Fmr1-KO astrocytes decreased the percentage of WT neurospheres 1–2 mm, and increased that of neurospheres ≥2 mm in diameter. The selective response of WT neurospheres to Fmr1-KO ACM points to a non-cell autonomous defect in NPC proliferation in the absence of FMRP caused by abnormal Fmr1-KO astrocytic cues. This was particularly evident in the effect of ACM on neurospheres ≥2 mm in diameter, where the percentage of WT neurospheres was increased in the presence of Fmr1-KO ACM, while that of Fmr1-KO neurospheres was not. This population of neurospheres represents multipotent and proliferative functional stem cells (Louis et al., 2008). The apparent aberrant signaling of Fmr1-KO astrocytes may be due to compensatory measures for the lack of responsiveness of Fmr1-KO neurospheres to cellular or environmental cues.

The addition of Fmr1-KO cortical ACM to WT neurospheres caused the most robust changes, affecting the total number of neurospheres generated (Supplementary Figure S1) as well as the percentages of neurospheres in all size categories relative to WT neurospheres without ACM (Supplementary Figure S2), which highlights the heterogeneity of astrocytes (Matyash and Kettenmann, 2010). This is likely due to the genotypic-spatial mismatch between hippocampal WT neurospheres and cortical Fmr1-KO ACM. Indeed, of all combinations tested, ACM from Fmr1-KO cortical astrocytes decreased the percentage of WT colonies measuring 0.5–1 mm. Thus, the proliferation of NPCs that make up colonies 0.5–1 mm is not solely changed by astrocyte-secreted factors from Fragile X brains. Rather, the enhanced effect of Fmr1-KO cortical ACM on WT hippocampal neurospheres resulted in our finding, which seems to only apply to neurospheres 0.5–1 mm. In agreement with the specificity of astrocytes, hippocampal astrocytes are shown to promote NPC proliferation and neurogenesis, whereas spinal cord astrocytes do not (Song et al., 2002). Similarly, ACM from the hippocampus increases the number of neurons differentiated from human NPCs, whereas cortical ACM does not (Cordero-Llana et al., 2011).

One of the functions of hippocampal neurogenesis is learning and memory (Deng et al., 2010). We thus hypothesized that impaired learning, characteristic of FXS, is attributed to aberrant NPC proliferation and neurogenesis. Fmr1-KO mice have demonstrated deficits in hippocampus-dependent learning such as exaggerated inhibitory avoidance extinction (Dölen et al., 2007) and trace conditioning tasks (Hayashi et al., 2007). Interestingly, conditional ablation of FMRP from adult NPCs results in learning deficits on the trace conditioning and delayed non-matching-to-place radial arm maze tasks, both of which require an intact hippocampus, and conditional restoration of FMRP results in improved performance on these tasks (Guo et al., 2011). Therefore, functional restoration of hippocampal neurogenesis may alleviate learning and memory impairments in Fragile X.

In this study, we examined the proliferation of hippocampal neural progenitor and stem cells from the early postnatal Fmr1-KO mouse brain. We found that Fmr1-KO progenitor-derived neurospheres showed decreased proliferation in response to WT and Fmr1-KO ACM, thereby demonstrating non-cell autonomous defects in responding correctly to astrocyte-secreted factors. In contrast, WT neurospheres showed a specific response to Fmr1-KO ACM, where Fmr1-KO ACM restricted the proliferation of neural progenitors, and enhanced the proliferation of stem cells. We also identified some of the astrocyte secreted factors with ± 1.5 fold expression difference in Fmr1-KO brains. To our knowledge, this is the first time that astrocyte-secreted factors are assayed and a number of them identified with mass spectrometry in relation to Fragile X. In turn, this work opens up new possibilities in investigating the functions of these factors and uncovering the role of astrocytes in Fragile X and neurodevelopmental disorders.

MS conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript. LD conception and design, financial support, provision of study material or patients, data analysis and interpretation, manuscript writing, final approval of manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.