Quantitative assessment of fibroblast growth factor receptor 1 expression in neurons and glia

Background Fibroblast growth factors (FGFs) and their receptors (FGFRs) have numerous functions in the developing and adult central nervous system (CNS). For example, the FGFR1 receptor is important for proliferation and fate specification of radial glial cells in the cortex and hippocampus, oligodendrocyte proliferation and regeneration, midline glia morphology and soma translocation, Bergmann glia morphology, and cerebellar morphogenesis. In addition, FGFR1 signaling in astrocytes is required for postnatal maturation of interneurons expressing parvalbumin (PV). FGFR1 is implicated in synapse formation in the hippocampus, and alterations in the expression of Fgfr1 and its ligand, Fgf2 accompany major depression. Understanding which cell types express Fgfr1 during development may elucidate its roles in normal development of the brain as well as illuminate possible causes of certain neuropsychiatric disorders. Methods Here, we used a BAC transgenic reporter line to trace Fgfr1 expression in the developing postnatal murine CNS. The specific transgenic line employed was created by the GENSAT project, tgFGFR1-EGFPGP338Gsat, and includes a gene encoding enhanced green fluorescent protein (EGFP) under the regulation of the Fgfr1 promoter, to trace Fgfr1 expression in the developing CNS. Unbiased stereological counts were performed for several cell types in the cortex and hippocampus. Results This model reveals that Fgfr1 is primarily expressed in glial cells, in both astrocytes and oligodendrocytes, along with some neurons. Dual labeling experiments indicate that the proportion of GFP+ (Fgfr1+) cells that are also GFAP+ increases from postnatal day 7 (P7) to 1 month, illuminating dynamic changes in Fgfr1 expression during postnatal development of the cortex. In postnatal neurogenic areas, GFP expression was also observed in SOX2, doublecortin (DCX), and brain lipid-binding protein (BLBP) expressing cells. Fgfr1 is also highly expressed in DCX positive cells of the dentate gyrus (DG), but not in the rostral migratory stream. Fgfr1 driven GFP was also observed in tanycytes and GFAP+ cells of the hypothalamus, as well as in Bergmann glia and astrocytes of the cerebellum. Conclusions The tgFGFR1-EGFPGP338Gsat mouse model expresses GFP that is congruent with known functions of FGFR1, including hippocampal development, glial cell development, and stem cell proliferation. Understanding which cell types express Fgfr1 may elucidate its role in neuropsychiatric disorders and brain development.

Fgfr1 mutants also exhibit a disruption in corpus callosum and hippocampal commissure due to abnormal midline glia development (Smith et al., 2006, Tole et al., 2006. The midline glial cells fail to undergo soma translocation and formation of the indusium griseum leading to midline commissural axon guidance defects (Smith et al., 2006). Furthermore, these mice exhibit postnatal loss of maturation in eABAergic interneurons expressing parvalbumin (PV) and exhibit behavioral hyperactivity (Muller Smith et al., 2008, Smith et al., 2014. Hyperactivity and a decrease in number of interneurons in the cortex co-occur in patients with schizophrenia (Volk et al., 2000, Akbarian and Huang, 2006, Hashimoto et al., 2008, Volk and Lewis, 2013. Interestingly, Fgfr1 expression was found to be increased in the prefrontal cortex of individuals with schizophrenia (Volk et al., 2016). Dual inactivation of floxed alleles of Fgfr1 and Fgfr2 results in abnormal cerebellar morphogenesis including reduced size of the cerebellum due to a defect in proliferation of both cerebellar glia and granule cell precursors, abnormal orientation and morphology of Bergmann glia, and loss of laminar architecture (Muller Smith et al., 2012).
This phenotype is similar to that observed in Fgf9 mutants (Lin et al., 2009). FeFRs are implicated in maintaining astrocytes in a non-reactive state, and in impeding glial scar formation (Kang et al., 2014). When Fgfr1 deletions were targeted to oligodendrocyte lineages, they did not disrupt oligodendrocyte birth, but modulated myelin sheath thickness and remyelination in chronic demyelination models (Furusho et al., 2012, Furusho et al., 2015. Administration of FeF2 into the lateral ventricles has also been shown to increase the number of oligodendrocyte precursor cells in the SVZ (Azim et al., 2012). Patients with major depressive disorder (MDD) and bipolar disorder have altered gene expression of FeFs and FeFRs (Evans et al., 2004(Evans et al., , eaughran et al., 2006. In situ hybridization revealed that mRNA for Fgfr1, and its ligand Fgf2, were both down regulated in the hippocampus of rats that had undergone the social defeat paradigm (Turner et al., 2008). Microinjections of FeF2 into the lateral ventricles of rats resulted in an increase in Fgfr1 mRNA in the De within 24 hours post FeF2 injections and was accompanied by acute antidepressant-like effects in the force swim test (Elsayed et al., 2012). Furthermore, increased anxiety, dysregulation of the hypothalamic pituitary axis and decreased hippocampal glucocorticoid receptor expression is observed in FeF2 knockout mice. These effects are reversible by administration of FeF2 (Salmaso et al., 2016). FeF22 and FeF7 are presynaptic organizing molecules that promote differentiation of excitatory and inhibitory presynaptic terminals in the hippocampal cornu ammonis (CA) region 3 through combinatorial signaling of sets of FeFRs (Umemori et al., 2004, Terauchi et al., 2010, Dabrowski et al., 2015. eiven the data that FeF2/FeFR1 signaling is important for the regulation of mood and affect, and that FeFR1 signaling may participate in synaptogenesis, a better understanding of the cell types expressing Fgfr1 is important to improving our understanding of its actions. Previous estimates of Fgfr1 expression have been derived from in situ hybridization studies and from immunochemistry using antibodies against FeFR1 (eonzalez et al., 1995, Belluardo et al., 1997, Bansal et al., 2003, Ohkubo et al., 2004, Blak et al., 2005, while more recently, molecular studies have identified the timing and or cell specific expression of Fgfr1 (Doyle et al., 2008, earcia-eonzalez et al., 2010, Azim et al., 2012. In embryonic mice, Fgfr1 is strongly expressed in the hippocampal hem, choroid plexus, cortical ventricular zone, and cortical midline (el-Husseini et al., 1994, Bansal et al., 2003, Ohkubo et al., 2004, Smith et al., 2006. In adult mice, the strongest Fgfr1 expression is observed in the hippocampus (Ohkubo et al., 2004).
Based on a literature review, Turner and colleagues surmised that neuronal populations in the adult hippocampus and cortex mostly express Fgfr1, in contrast to other FeF receptors that are considered to be expressed primarily in glia (Turner et al., 2012a). Despite clear advances in our understanding of FeF signaling derived from in situ hybridization, it suffers from poor cell-type resolution. Likewise, although immunocytochemistry using antibodies raised to FeFR1 has proven to be important, cross-reactivity to other FeFRs remains a concern. To circumvent these issues, we investigated Fgfr1 expression in PV+ interneurons, employing a transgenic reporter line, tgFGFR1-EGFPGP338Gstt bacterial artificial chromosome (BAC), that was obtained from eENSAT, http://www.gensat.org (Smith et al., 2014). In this transgenic line, the gene encoding enhanced green fluorescent protein (EGFP) is regulated under the same promoter as Fgfr1. Thus, eFP fluorescence should indicate expression of genes encoding Fgfr1. We previously showed that in tgFGFR1-EGFPGP338Gstt mice, PV+ interneurons did not colocalize with eFP+ cells. Thus, the decrease in PV+ interneurons due to inactivation of Fgfr1 occurs non-cell-autonomously (Smith et al., 2014). We also observed that a large number of glia appeared to express Fgfr1 in adult mice. In the present study, we extend our studies and present a quantitative analysis of the relative expression of Fgfr1 in neurons versus glial cells during postnatal development of the telencephalon. We show that Fgfr1 is differentially expressed, primarily in eFAP+ astrocytes and OLIe2+ cells, with a minority of cells colocalizing with NeuN+ neurons. Furthermore, SOX2+ cells, BLBP+ cells and DCX+ cells in the cortex, hippocampus, subventricular zone (SVZ), and hypothalamus of mice are colocalized with the eFP signal, indicating that these cells also express Fgfr1.

Animals
Wild type Swiss Webster mice were crossed with mice expressing enhanced green fluorescent protein (EeFP) under the same promoter as Fgfr1 (tgFGR1-EGFPGP338Gstt). The transgenic line, tgFGR1-EGFPGP338Gstt was generated from the eENSAT project (eENSAT.org) by microinjecting bacterial artificial chromosome with Fgfr1 promoter driving EeFP into the pronucleus of fertilized mouse eggs. eENSAT produces transgenic BAC-EeFP reporter and BAC-Cre recombinase driver mouse lines with the aim to map the expression of genes in the CNS of mice (Heintz, 2004 Fractionator probe on Stereoinvestigator was employed, with tops of cells counted in threedimensional counting boxes, which were set to specific parameters ( Table 2). For counts of the cortex and hippocampus, 50 μm sections were sampled every 20 th (cortex) and every 10 th section (Hippocampus). The CA1, CA2, and CA3 regions were counted together and included the stratum lacunosum, stratum radiatum, stratum pyramidale and the Stratum Oriens. The De was counted separately. For the cortex, at these ages, the expression of eFP seemed fairly ubiquitous and so the cortex was counted in its entirety using stereology as outlined above and in Table 2.
Fluorescence images were acquired through StereoInvestigator imaging software. Of note, cell counts indicated that presence of the eFP transgene did not cause cell lethality in cortical or hippocampal cells. For P7 counts, 3 animals per genotype were counted for NeuN, eFAP, and Olig2 (same animals for all 3 counts). For 1 month old animals, 3 animals per group were counted for cortex and CA, and 4 animals in the eFP+ group for the De. The eFAP/NeuN /eFP counts were determined from triple stained samples. The Olig2 counts were from 3 animals per group.
To determine the percentage of eFP+ cells that were neuronal stem cells expressing Sox2 and neuroblasts expressing DCX (3 animals per group), we acquired z stack images of the anterior De hemisphere and a section 500 μm posterior to this first section. The z stack images obtained to count SOX2 markers were 19 μm thick with each slice 1 μm thick from 3 eFP+ mice and 2 for the control mice). The z stack images acquired to count DCX markers were 10 μm thick with each slice 1 μm thick (3 animals per group).

Statistical Analysis
Data from the StereoInvestigator software were entered into Excel, imported to JMP Pro 11, and analyzed by student t-tests, or ANOVA, using SAS. little to no green fluorescence in eFP-controls ( Fig. 1D-F). In the CA region of the hippocampus, eFAP+/eFP+ cells were primarily observed surrounding the stratum pyramidale, in the stratum radiatum, stratum oriens, as well as the white matter above the CA region ( Fig. 1   G-I, Fig. S1) with little to no green fluorescence in eFP-littermate controls ( Fig. 1 J-L). There was strong eFP fluorescence in the stratum pyramidale of the CA region. This eFP staining colocalized with NeuN+ cells (neurons) ( Fig. 2A-C, Fig. S2) with little to no green fluorescence in eFP-littermate controls ( Fig. 2D-F). NeuN+ /eFP+ cells were also observed in some, but not Manuscript to be reviewed positive, and of eFAP+ cells, 50% ± 8% are eFP+. Of eFP+ cells in the CA, 50% ± 5% are eFAP positive and of eFAP+ cells in the CA, 43% ± 9% are eFP positive (Table 3). One-way ANOVA statistical analysis revealed that the total number of eFAP+ cells in the De and CA of tgfgfr1-EGFP+ are not significantly different from their littermate controls (Table 3).

Results
We well as BLBP+ cells of the SVZ (Fig. 3F,G).

Fgfr1 is expressed in cortical GFAP+ astrocytes and NeuN+ neurons at P7
To determine which cortical cells express Fgfr1, we stained the cortical tissue of P7 tgfgfr1-EGFP+ and control mice with eFAP and NeuN antibodies. Whereas eFP+ cells colocalized with eFAP+ astrocytes throughout the cortex ( Fig. 4A-C), eFP+ cells colocalized with NeuN+ cells (neurons) mostly in layers 5 and 6 of the cortex at this age ( Fig. 4D-F). indicates that most Gftp expressing astrocytes also express Fgfr1. The total number of cortical eFAP+ cells of tgfgfr1-EGFP+ mice was not significantly different from the number of cells in littermate controls ( Table 3), suggesting that the insertion of the transgene does not significantly alter the number of astrocytes expressing Gftp.
We examined whether tgfgfr1-EGFP+ was expressed in oligodendrocytes and their precursors by staining for OLIe2 in the cortex. We found that OLIe2+ staining colocalized with eFP in the cortex (Fig. 4G-I). We also found sparse colabeling of OLIe2+ cells in the SVZ and subcortical white matter (Fig. 4J). eFP expression within these cells appears to be lower than that of surrounding cells, and may reflect residual eFP expression once progenitor cells that express Fgfr1 have divided and an oligodendrocyte has initiated differentiation to a more mature subtype. At P7, radial glia of the cortex are undergoing soma translocation to become astrocytes, and can be detected with BLBP. BLBP and eFP immunostaining revealed strong colocalization throughout the cortex (Fig. 4K-M). At P7, eFP staining colocalized with BLBP+ stem cells and Bergmann elia in the developing cerebellum (Fig. 4N) Fgfr1 is expressed in specific cell types of the hippocatpus and SVZ at 1 tonth We next investigated Fgfr1 expression in 1-month tgfgfr1-EGFP+ mice. Immunostaining immunostaining to determine which cell types express Fgfr1 in the De and CA of the hippocampus, as well as the SVZ. eFP+ cells colocalized with SOX2+ cells (Fig. 5A-E, Fig.   S4A,B) and DCX+ cells ( Fig. 5F-J, Fig. S4C,D) in the SeZ and granule cell layers. To determine the percentage of eFP+ cells that were SOX-2 positive and DCX positive, we obtained z stack images of an anterior dentate gyrus hemisphere and posterior dentate gyrus hemisphere, and performed counts from these images. Of the eFP+ cells counted in the z stack, 33% ± 2% are SOX2 positive. Conversely, the majority of SOX2+ cells, 71% ± 2%, are eFP positive (Fig. 5E).
Of the eFP+ cells counted in the z stack, 64% ± 0.9% are DCX positive and of DCX+ cells, 86% ± 2% are eFP positive (Fig. 5J). Triple staining with eFP, eFAP and NeuN antibodies revealed that eFAP+ cells (stem cells) and NeuN+ cells (neurons) colocalized with eFP+ cells in the SeZ ( Fig. 5K-N, Fig. S4E) and in the CA region ( Fig. 5O-R, , Fig. S4F), respectively. eFP+ cells did not colocalize with OLIe2+ cells (oligodendrocytes) in the De (Fig. S5). Taken together, these results indicate that eFP (Fgfr1) Table 3). Many eFAP+ cells also express eFP in the SVZ (Fig. 6A-H SVZ and in the white matter above it (Fig. 6D,G,H). There was also significant colocalization of eFP with SOX2+ cells (Fig. 6I-L) indicating that eFAP+ and SOX2+ stem cells of the SVZ express Fgfr1. Similar to what was found at P7, eFP+ cells did not colocalize with DCX+ cells (neuroblasts) in the SVZ (Fig. 6M-O) or in the DCX+ cells of rostral migratory stream (RMS) as it enters the olfactory bulb (Fig. 6P). In the rostral migratory stream, eFP staining surrounds the DCX+ cells as would be expected from astrocytes surrounding the migrating neuroblasts.  (Table 3).

8A-C), eFP+ cells strongly colocalized with SOX2+ cells throughout the hypothalamus and third
ventricle and arcuate nucleus (Fig. 8D-I) in 1-month tgfgfr1-EGFP+ mice. eFAP+ tanycytes, including those of the arcuate nucleus were among the hypothalamic cells observed to express Fgfr1 both at 1 month and at P7 (Fig. 8J-L). Tanycytes participate in neuroendocrine regulation and transport of molecules from the CSF to the hypothalamus, release of gonadotropin-releasing hormone, and production of triiodothyronine (T3) in the brain (Rodriguez et al., 2005). eFP expression was not observed in NeuN+ or OLIe2+ cells of the hypothalamus (Fig. 8M-O, Fig.   S6 A-D). At P7, we also observed colocalization of BLBP and eFP in the hypothalamus near the third ventricle, but not many in the median eminence ( Fig. S6 E

Calretinin (CR) and sotatostatin (SST) neurons express Fgfr1 in one-tonth old tice.
The NeuN+/eFP+ neurons observed at 1 month were not restricted to any specific cortical layer, and a minority population of eFP colocalized with NeuN+ neurons. We therefore sought to determine if inhibitory neurons express Fgfr1. Tissue from one-month old tgfgfr1-EGFP+ mice (Fig. 9B-C) and their control littermates (Fig. 9A) was stained for eAD67. Some eAD67+ cells colocalized with eFP in the cortex (Fig. 9B), but not in the hippocampus (Fig.   9C) or SVZ (Fig. 9D). Our previous investigations determined that eFP was not colocalized with PV+ interneurons (Smith et al., 2014). This led to the experiments in which staining the onemonth old tissue with calretinin (CR) and somatostatin (SST) was performed alongside eFP immunostaining. CR+ inhibitory neurons express eFP in the cortex (Fig. 9E, F), De (Fig. 9G), and SVZ (Fig. 9H). Some SST+ inhibitory neurons express eFP in the anterior cingulate of the cortex (Fig. 9I, J), but none were observed in the De (Fig. 9K).

Discussion
Through immunostaining for eFP in tgFgfr1-EGFP+ mice of the tgFGFR1-EGFPGP338Gstt line, we examined the cell specific expression of Fgfr1 in the cortex,  Husseini et al., 1994, eonzalez et al., 1995, Belluardo et al., 1997, Bansal et al., 2003, Ohkubo et al., 2004) and more modern molecular profiling studies (Lovatt et al., 2007, Cahoy et al., 2008, Doyle et al., 2008, Azim et al., 2012. This transgenic line gives the added benefit of allowing  (Ohkubo et al., 2004, Kang et al., 2009, Stevens et al., 2010b, Rash et al., 2011. Our studies of Fgfr1 expression in embryos and at P1 will be may be participating in the soma translocation process, as previously described for the glial cells of the indusium griseum (Smith et al., 2006), or it may be a general factor expressed by astrocytes since a majority of eFAP+ cells expressed eFP. It is interesting to note that as animals matured from P7 to one-month of age, the relative amount of eFAP/eFP colocalization was increased. This may reflect a maturation of astrocytic glia from BLBP expressing radial glia stem cells undergoing gliogenesis and soma translocation to mature astrocytes that express higher levels of eFAP. BLBP is also expressed in Bergmann glia of the cerebellum, which are cells that have dual roles as stem cells and as a scaffold for granule cell neuron migration and alignment of Purkinje neuron dendrites. The finding that Fgfr1 is expressed within these cells is consistent with the demonstrated occurrence of cooperative signaling between FeFR1 and FeFR2 in the formation of the cerebellum, and specifically, in the morphology and pial attachment of Bergmann glia of the cerebellar anlage (Lin et al., 2009, Muller Smith et al., 2012. FeFR1 may be playing dual roles in these cells as a factor that supports their proliferation, as well as 400 (Patricio et al., 2013). FeFR signaling also participates in development of hippocampal synaptogenesis (Cambon et al., 2004, Umemori et al., 2004, Terauchi et al., 2010. Activation of FeFR1 promotes synapse formation in hippocampal neurons (Cambon et al., 2004, Terauchi et al., 2010, Dabrowski et al., 2015. The expression of Fgfr1 in DCX positive cells of the hippocampus, as well as neurons in the CA region supports the evidence that FeFR1 plays additional roles in synapse formation and integration into the hippocampal circuitry. These findings have implications for hippocampal function in HPA axis regulation. Changes in Fgfr1 expression, and FeF2 levels in the hippocampus are linked to major depression and anxiety, as well as to responses to antidepressants (Evans et al., 2004, Newton and Duman, 2004, eaughran et al., 2006, Turner et al., 2008, Elsayed et al., 2012. It is hypothesized that Fgf2 and with Category III cells in that study, which included dynamically regulated genes that change expression as NSC go from a quiescent state to an activated state (Llorens-Bobadilla et al., 2015).
Future studies could employ tgfgfr1-EGFP+ mice to explore the expression of Fgfr1 during remyelination and response to injury. astrocytes constitute a majority of Fgfr1 expressing cells in the adult brain. This idea is consistent with previous mRNA profiling of astrocytes in which Fgfr1 was identified as a gene with enriched expression in astrocytes as compared to other cell types in the brain (Lovatt et al., 2007, Cahoy et al., 2008, Doyle et al., 2008). An imbalance of excitatory and inhibitory neurons, along with hyperactivity has been documented in certain neurological disorders (Benes et al., 2000, Volk et al., 2000, Kalanithi et al., 2005, Akbarian and Huang, 2006, Hashimoto et al., 2008, Kataoka et al., 2010, eonzalez-Burgos et al., 2011, Volk and Lewis, 2013. Inactivation of Fgfr1 leads to a decrease in the abundance of parvalbumin interneurons in the cortex and is associated with an increase in hyperactivity (Muller Smith et al., 2008). A subsequent study demonstrated that a decrease in interneurons occurs postnatally and that Fgfr1 expression does not occur in parvalbumin expressing interneurons (Smith et al., 2014). Astrocytes lacking Fgfr1 were less capable of supporting the maturation of Gtd67 expressing cells than control astrocytes. Here, we find that Fgfr1 is expressed by a majority of astrocytes in the cortex and hippocampus throughout the maturational period of parvalbumin positive interneurons, further supporting the importance for  (Brandt et al., 2003), offering a probable explanation as to why eAD67 is not observed colocalizing with Fgfr1 expressing cells in the De. The hypothalamus, responsible for controlling hormonal production, stress regulation, and feeding behaviors, has been found to contain a neural stem/progenitor cell niche (Robins et al., 2013

Conclusion
FeFR1 has been implicated as having multiple functions in CNS development, homeostasis, and behavior, but defining the cellular basis of its functions depends on having a clear understanding of which cell types the Fgfr1 gene is expressed in, and when. The eENSAT project was envisioned as a resource that would provide the tools for such detailed studies (Heintz, 2004). Here, we have extended the previously published in situ based studies and online resources with a detailed examination of the tgFGR1-EGFPGP338Gstt line. Our data are congruent with in situ studies, but with the added feature of double immunofluorescence with glial and neuronal markers, and a quantification of the relative expression in glial versus neuronal cells in the young adult brain. We here find Fgfr1 promoter driven eFP expression in a variety of stem cells of the CNS including the young adult SVZ, young adult SeZ, cerebellar Bergmann elia, and SOX2+ cells of the hypothalamus (summarized in Table 4). An additional study of Fgfr1 expression in embryonic and perinatal stem cells will be described elsewhere. We also find that Fgfr1 is expressed predominantly in glia in the young adult brain, although significant expression in DCX+ positive neuroblasts of the hippocampus was also observed. This was in stark contrast to DCX+ neuroblasts of the RMS. Our findings may shed light on the participation of FeF2/FeFR1 signaling in determining anxiety and mood, where a neuronal role for FeFR1 has been hypothesized (Turner et al., 2012a). Future studies are needed to determine whether Fgfr1 expression colocalizes with other markers, such as S100β, O4, and Ne2 at embryonic and postnatal time points. Since Gftp is not expressed in all astrocytes, S100β (glial specific marker primarily expressed in astrocytes, but also in some ependymal cells) would be a good marker to further explore Fgfr1 expression. The tgFgfr1-EGFP+ model can also be used to study additional stages in development, or Fgfr1 expression after environmental manipulations previously shown to alter Fgfr1 expression, including animal models for induced depression such as social defeat stress.