Extracellular matrix protein anosmin‐1 modulates olfactory ensheathing cell maturation in chick olfactory bulb development

Olfactory ensheathing cells (OECs) are a specialized class of glia, wrapping around olfactory sensory axons that target the olfactory bulb (OB) and cross the peripheral nervous system/central nervous system boundary during development and continue to do so post‐natally. OEC subpopulations perform distinct subtype‐specific functions dependent on their maturity status. Disrupted OEC development is thought to be associated with abnormal OB morphogenesis, leading to anosmia, a defining characteristic of Kallmann syndrome. Hence, anosmin‐1 encoded by Kallmann syndrome gene (KAL‐1) might modulate OEC differentiation/maturation in the OB. We performed in ovo electroporation of shRNA in the olfactory placode to knock‐down kal in chick embryos, resulting in abnormal OB morphogenesis and loss of olfactory sensory axonal innervation into OB. BLBP‐expressing OECs appeared to form a thinner and poorly organized outmost OB layer where SOX10 expressing OECs were completely absent with emergence of GFAP‐expressing OECs. Furthermore, in embryonic day 10 chick OB explant cultures, GFAP expression in OECs accumulating along the OB nerve layers was dramatically reduced by recombinant anosmin‐1. We then purified immature OECs from embryonic day 10 chick OB. These cells express GFAP after 7 days in vitro, exhibiting a multipolar morphology. Overexpression of chick anosmin, exogenous anosmin‐1 or FGF2 could inhibit GFAP expression with cells presenting elongated morphology, which was blocked by the FGF receptor inhibitor Su5402. These data demonstrate that anosmin‐1 functions via FGF signalling in regulating OEC maturation, thereby providing a permissive glial environment for axonal innervation into the OB during development.


| INTRODUCTION
Olfactory sensory neurons, whether after damage or as part of normal cell turnover, continually extend axons from the olfactory epithelium, through the lamina propria to enter the outer layer of the olfactory bulb (OB) (Farbman, 1990;Graziadei & Graziadei, 1979). In the OB, they form synapses with the dendrites of mitral/tufted cells establishing glomeruli-like structures. This property depends on the specialized glial cells in the olfactory system, so-called olfactory ensheathing cells (OECs) (Schwob, 2002). OECs originate from the neural crest (Barraud et al., 2010) and wrap around olfactory sensory axons migrating towards the anterior forebrain where they, along with olfactory axons, form the olfactory nerve layer (ONL) of the OB. It has been proposed that OECs may exhibit the unique property of maintaining continuous open channels, permitting ingress of olfactory nerve fibres into the bulb (Li, Field, & Raisman, 2005;Williams, Franklin, & Barnett, 2004). Moreover, cross-talk between astrocytes and OECs are believed to underpin the mechanism of continuous neuronal re-innervation, as OECs interweave with astrocytes within the deeper ONL (Doucette, 1991;Li et al., 2005;Raisman, 1985).
In the developing ONL of the OB, two OEC subpopulations, from different stages of maturation and evincing different antigenic profiles, exert distinct functions during olfactory axonal outgrowth and targeting to specific glomeruli. Thus, p75 and GFAP OEC expression in the outer region of ONL are thought to characterize a more mature OEC population for olfactory axon outgrowth and OB targeting, whereas NPY and Runx1 positive OEC subpopulations from the inner ONL are thought to function as more immature OECs influencing axonal sorting and specific synapse formation (Astic, Pellier-Monnin, & Godinot, 1998;Murthy, Bocking, Verginelli, & Stifani, 2014). Furthermore, transcription factor SOX10 positive OEC subpopulations play a role in fasciculation and targeting of the olfactory nerve fibres into the OB for ONL organization (Pingault et al., 2013).
In humans, spontaneous mutations of the neural crest OEC marker SOX10 have recently been described in Kallmann syndrome (KS) patients, the development disorder characterized by abnormal OB morphogenesis and delayed puberty due to disrupted migration of gonadotropin-releasing hormone (GnRH) neurons along with the olfactory, vomeronasal and terminal nerves (Barraud, St John, Stolt, Wegner, & Baker, 2013;Pingault et al., 2013). Recent study demonstrated that OEC maturation is involved in neurite outgrowth of GnRH neurons during migration (Geller et al., 2017). These findings possibly link the OEC development with the aetiology of KS.
Loss of function of extracellular matrix protein anosmin-1, encoded by the X-linked KAL-1 gene, is almost invariably associated with a highly penetrant phenotype with abnormal sense of smell, frequently resulting from OB dysgenesis. Immunohistochemical study of two KAL-1 deleted or mutated human foetuses (aged 19 and 25 weeks) revealed that axons of olfactory sensory neurons could not penetrate into the forebrain, failing to establish synaptic connection within the OB (Schwanzel-Fukuda, Bick, & Pfaff, 1989;Teixeira et al., 2010). During human embryogenesis, anosmin-1 is expressed in the interneurons and ONL of the developing OB, and has been shown to demonstrate chemoattractive and stimulative properties on axonal branching, neurite outgrowth and neuronal migration Hu et al., 2009;Legouis, Lievre, Leibovici, Lapointe, & Petit, 1993;Lutz et al., 1994;Soussi-Yanicostas et al., 1998). More recent evidence, however, suggests that anosmin-1 may also modulate glial cell development. Thus, anosmin-1 actions within the neural crest, the site of origin of OECs, may result from its effect in enhancing FGF8 activity while inhibiting BMP5 and Wnt3a signalling (Endo, Ishiwata-Endo, & Yamada, 2012). Furthermore, as the migration of OEC from the olfactory placode towards the telencephalon occurs before KAL-1 expression in the presumptive OB, suggesting that anosmin-1 does not appear essential for initial OEC migration. Based on these observations, we hypothesized that anosmin-1 might modulate OEC maturation, enabling the establishment of stable synaptic connections between olfactory sensory axons and the OB anlage during embryogenesis.
In the present study, we have performed in ovo electroporation of shRNA in the olfactory placode to knock-down kal function in chick embryos to further elucidate the mode of action of anosmin-1 in OB morphogenesis, and demonstrate its function in the development and maturation of OECs. Moreover, in OB explants and in vitro OB dissociated cell cultures, exogenous anosmin-1 down-regulated GFAP expression in OEC via an interaction with FGF signalling. Our data provide a novel mechanistic link between loss of function mutations in anosmin-1 in humans and the resulting severe anosmia.

| Plasmids used for in ovo electroporation
Three 21-nucleotide long sequences, within the chicken kal cDNA (NM_205424) were selected to generate shRNA and designed by GenScript (GenScript USA Inc.) to knockdown chicken anosmin. Selected shRNAs were cloned into the miRNA site of the pRFP-C-RS vector (kindly provided by Professor Uwe Drescher, King's College London). To evaluate their efficiency, CHO cells were transfected with a chicken 6xHis-tagged anosmin expression construct (kal in pCAGGS, kindly provided by Ken Yamada, NIH) and each shRNA, including the empty vector or scrambled shRNA were used as controls. Cells were cultured for 2 days, and processed for Western blot analysis using antibodies to chicken anosmin (kindly provided by Ken Yamada, NIH) and β-actin (Sigma). The optical density of the specific band was quantified and silencing, relative to β-actin for each individual sample, calculated. Finally, general silencing was calculated relative to the control (pRFP-C-RF vector alone). Optimal silencing was obtained with kal shRNA-2 (>95%, sequence: CGGACTGGTAGATCCTTACCT). Percentage silencing was calculated from three independent experiments. This construct was chosen for in ovo electroporation to knock-down kal in chick. No effect was seen when cells were transfected with scrambled control shRNA.

| In ovo electroporation
The stage of chick embryos used in this study is within half the gestation period [embryonic day (E) 11] and is approved under the Animals (Scientific Procedures) Act 1986. Fertilized hens' eggs (Henry Stewart Farm) were incubated at 38°C to Hamburger-Hamilton (HH) stage 11-12 (E2) and processed for electroporation. kal shRNA constructs or scrambled shRNA at concentrations of 0.8 or 2 mg/ml were microinjected into the midbrain to avoid artificial damage on the anterior forebrain and then current was applied across the midbrain and one side of olfactory placode, using physical positioning of electrodes. Upon application of the electric current, kal shRNA construct was electroporated into cells on the same side of the olfactory placode as the anode, since DNA is negatively charged, and generates an internal negative, unmodified control on the non-electroporated side. After applying five 4.0-4.2 V pulses, 50 ms each, at 500 ms intervals with the electroporator (CUY21SC, NEPA GENE CO., LTD), eggs were sealed and incubated to E10. Embryos were harvested and the whole brain was dissected in 0.1 M PBS, fixed overnight in 4% (w/v) paraformaldehyde in PBS (PFA) for immunostaining and in situ hybridization as described below.

| Primary cultures of OB glial cells
E10 OBs were digested with 0.05% (v/v) trypsin for 25 min at 37°C and rinsed twice with DMEM/F12 medium (ABM). ABM with 10% (v/v) FBS (AGM) and 10% (w/v) BSA were added and trituration repeated to dissociate cells. 4% (w/v) BSA was gently layered under the cell suspension to create a distinct interface, spun for 15 min at 500 g in a cooled centrifuge at 4°C, and the cell pellet re-suspended in AGM. Approximately 7 × 10 6 cells were plated in AGM for 2 weeks until the cells had grown to complete confluence. After shaking the flask at 250 rpm to remove neurons and microglia, the media was removed and cells were rinsed twice with 10 ml ABM, followed by incubation with 0.1% (v/v) trypsin for 3-4 min. Cells were re-suspended with AGM and centrifuged at 400 g for 15 min. The cells were washed three times with ABM. OB glial cells were then plated at a density of 2 × 10 5 cells in ADM (AGM + 0.5xG5 supplement, Gibco) into 6-well plates containing 0.05 mg/ml poly-D-Lys coated coverslips. In one experiment, cells were treated with 5 nM anosmin-1, 2 nM FGF2 and 5 nM anosmin-1 plus 20 μM Su5402 (FGFR inhibitor) or 2 nM FGF2 for 7 days in culture, fixed in PFA. In another experiment for expression of anosmin, primary OB cells were transfected with 1 μg kal-pCAGGS construct or empty vector with FuGene6 in serumfree medium. After incubation for 48 hr, cells were fixed for immunostaining or lysed for Western blotting to detect GFAP expression. The GFAP band and fluorescence intensities were measured using the Software Image J (National Institutes of Health, Bethesda, MD). The relative GFAP band to actin ratio was calculated and all relative GFAP intensities were then compared with the control sample (vector only or no treatment group, 100%). The statistical significance was calculated using paired Student's t-test (n = 3).

RNA probes
RNA probe for in situ hybridisation was generated by PCR amplification using specific primers for chick kal (forward: AACACGCTGGGGTCAGATAC and reverse: TGGAGGAAGGTCAGGTGTTC) and cloned into pGEM-T Easy vector (Promega). RNA probes were transcribed using either the T7 or SP6 RNA polymerase on linearized DNA template and then purified using Microspin G-50 columns (GE Healthcare), according to the manufacturer's instructions and stored at −20°C until required.
Embryos were then washed and equilibrated in NTMT (100 mM NaCl, 100 mM Tris-HCl (pH9.5), 50 mM MgCl2 and 1% (v/v) Tween-20) for 10 min each. The colour reaction was carried out in NTMT, using NBT/BCIP at 5 μl/ml as the chromogenic substrate. The colour reaction was permanently terminated by washing embryos in PBST and then embryos were re-fixed in PFA.

| Rt-pcr
Total mRNA was extracted with Trizol from purified OB immature glial cells and reverse transcribed with Superscript II RT kit (Invitrogen). Specific primers were designed to amplify genes including kal, fgfr1-3 b and c isoforms (Appendix S1).

| In vivo knock-down of kal results in abnormal OB development
To gain insights into the potential role of chick anosmin in OB development, we electroporated chicken embryos with RFP fluorescently tagged shRNA constructs to knock-down kal expression, using scrambled shRNA as control. We first evaluated the efficiency of shRNA to silence kal by co-transfecting CHO cells with the chick kal expression vector and scrambled shRNA, or with the three designed kal shRNAs ( Figure 1a). All of the three kal shRNAs showed significant inhibition of kal gene expression, with optimal silencing obtained for kal shRNA-2; this was subsequently used for in ovo electroporation experiments.
We further examined whether kal shRNA-2 could effectively knock down endogenous kal by in situ hybridization. E2 chick embryos were electroporated with vector only, scrambled shRNA or 2 mg/ml shRNAs respectively (Figure 1b). After further incubation to E10, embryos were processed for in situ hybridization with a specific kal riboprobe. Kal transcripts were specifically detected in OBs electroporated with vector only and confined to a distinct band that coincides with the mitral cell layer (Figure 1b I), which is consistent with the previously described kal expression profile in chicken (Legouis et al., 1993;Lutz et al., 1994). Control hybridizations with sense riboprobes gave no signal (

| Dose-relevant effects of kal knockdown on OB morphogenesis
To further investigate the specificity and dose-relevant effect of kal shRNA-2, chick embryos were electroporated at E2 with shRNA constructs at different concentrations and then processed to visualize RFP at E10 (Figure 2). The transfected forebrains usually displayed much higher RFP expression, in one OB up to E10 stage. It was observed that scrambled construct and low (0.8 mg/ml) kal siRNA-2 concentrations resulted in normal OB morphogenesis, even on the OB side with stronger RFP expression (Figure 2a,b). However, kal knock-down, effected by increasing kal shRNA-2 to 2 mg/ml resulted in striking and reproducible olfactory system defects, including OB hypoplasia and primary failure of olfactory sensory axons to reach the forebrain. Such hypoplastic OBs generally had much higher RFP expression, indicating greater kal transcription inhibition; by contrast, the contralateral side without RFP expression had morphologically normal OB (Figure 2c). The general morphology of the whole forebrain was similar on both sides, indicating a specific function of kal on OB differentiation and evagination from the anterior forebrain. These morphological defects induced by chicken kal knock-down closely resemble the histopathological observations of human individuals affected by the loss of function of KAL-1 in X-KS patients. Furthermore, human anosmin-1 when mutated in X-KS causes an extremely high prevalence of complete anosmia due to severely hypoplastic or (more usually) absent OBs. Therefore, loss of OB following kal shRNA-2 electroporation observed in in situ hybridisation and dose-relevant experiments validates F I G U R E 1 Silencing efficiency of kal shRNA on kal gene expression. (a) CHO cells were co-transfected with a chick 6xHis-tagged anosmin-1 expression construct (kal in pCAGGS) without (−) or with empty vector, scrambled shRNA and three kal shRNAs (1-3), cultured for 2 days and then processed for immunoblotting using anosmin-1 antibody, using β-actin antibody as loading control. kal shRNA-2 generated optimal silencing on anosmin-1 expression and was subsequently used for in ovo electroporation on chick embryos. (b) Chick E2 embryos were electroporated with 2 mg/ml shRNAs including vector only (−) (n = 6), scrambled shRNA (n = 6) or kal siRNA-2 (n = 3). After incubation to E10 stage, the embryos were dissected, fixed and processed for in situ hybridization with anti-sense kal probe; sense kal probe was used as negative control. In vector only (−) embryo forebrains (I), the kal expression was present in the normal OBs with anti-sense kal probe, but absent with sense kal probe (II). Low magnification images are on the left side and high magnification images on the right side. (III) In scrambled shRNA embryo forebrains, the kal expression was present in both normal OBs with anti-sense kal probe. Low magnification images are on the left side and high magnification images on the right side. (IV) For kal shRNA-2 electroporation, left side image is low magnification showing normal and abnormal OB from one electroporated embryo. The normal OB with kal expression developed well on one side of forebrain, while on the contralateral side of forebrain without kal expression, OB was absent. Their higher magnification images are on the middle and right side, respectively. Representative images from embryo forebrains in each group are shown. Scale bar: low magnification image, 100 μm; high magnification image, 20 μm. F: forebrain; single arrow: normal OB; double arrow: absent OB the specificity and efficiency of kal shRNA-2. We then consider loss of OB with stronger RFP expression as the indicator for kal knock-down and performed the following histochemistry and immunofluorescence on the sections of both OB-present/absent forebrains from each embryo.

| OEC defects in chick embryos with kal gene knock-down
In chick, olfactory sensory neurons extend their axons towards the forebrain reaching the presumptive bulb region by stage 21 (E3.5). By stage 30 (E7) the bulb region begins to differentiate, acquiring its characteristic structure by stage 36 (E10) (Gomez & Celii, 2008). The H&E stained specimen of one side of the 2 mg/ml kal shRNA-2 electroporated OB, revealed a distinct laminar organization at E10, including olfactory nerve, external plexiform, mitral cell, internal plexiform and granule cell layers (Figure 3a,b). The connection of olfactory axons with this side of the OB was evident by βIII-tubulin immunostained olfactory axons in the outmost ONL (Figure 3c). On the contralateral presumptive OB with knock-down kal, there was no distinct laminar structure formation and there was a lack of olfactory axon connections with the forebrain. Notably, the bulk of the cell bodies of F I G U R E 2 Dose-relevant effect of kal shRNA-2 on OB development of E10 chick embryos. E2 chick embryos were electroporated with 2 mg/ml scrambled shRNA (a) and kal shRNA-2 at 0.8 (b), 2 mg/ml (c) respectively, incubated for 8 days to E10 stage, and dissected for visualization under epifluorescence microscopy. (a) and (b) Both OBs developed normally with scrambled shRNA and lower kal shRNA-2 concentration even in the OBs with higher RFP expression (n = 8 embryos, for each group from two independent experiments). (c) Higher concentration of kal shRNA-2 at 2 mg/ml resulted in very small or no OBs and the lack of penetration of olfactory sensory axon bundles in one side of forebrain with much higher RFP expression. The contralateral side with less RFP showed normal OB formation and normal connection of axon bundles to the OB. OECs are specialized glial cells involved in the growth and guidance of olfactory axons. They originate from the neural crest in mice and chicks and are heterogeneous in the pattern of markers they express in the PNS and CNS (Barraud et al., 2010;Forni, Taylor-Burds, Melvin, Williams, & Wray, 2011;Katoh et al., 2011). SOX10, a marker of neural crest-derived OECs, is co-expressed with another OEC marker, BLBP, along the olfactory, vomeronasal and terminal nerve pathways, in the frontonasal mesenchyme, and in the migratory mass and the OB in mouse (Pingault et al., 2013).
To examine the possibility that abnormal OB morphogenesis in kal knock-down embryos might result from the OEC developmental defects, we immunostained sections of E10 embryos electroporated with scrambled and kal shRNA-2 with antibodies for SOX10 and BLBP. Consistent with previous reports (Barraud et al., 2013;Pingault et al., 2013), SOX10 and BLBP-expressing cells in scrambled control embryos were localized to the outermost layer of the OB, corresponding to the ONL. BLBP-positive cells were rarely observed around the invading olfactory axons, being more densely expressed in ONL forming a much thicker layer (Figure 4a). In contrast, SOX10-expressing cells were abundant in olfactory axon bundles and formed a much thinner layer around the superficial surface of OB (Figure 4b). We then determined the expression patterns of OEC markers in the presence of kal knock-down. On the side of the morphologically normal OB, BLBP-and SOX10-positive cells showed a similar expression profile F I G U R E 3 Morphological defects of E10 chick OBs electroporated with 2 mg/ml kal shRNA-2. E2 chick embryos were electroporated with 2 mg/ml kal shRNA-2, incubated to E10, fixed and paraffin sectioned for H&E (a,b) and βIII-tubulin immunostaining (c). (a) H&E staining revealed normal OB morphology with distinct lamination from outside to inside: olfactory nerve layer (ONL), external plexiform layer (EPL), mitral cell layer (ML), internal plexiform layer (IPL) and granule cell layer (GL). By contrast, on the kal knock-down side of forebrain, there was no obvious typical laminar structure formation. (b) H&E staining on a deeper section. Similar morphological defects of OBs were observed on different sections, demonstrating that such defects are not due to the orientation during sectioning. (a1) and (a2): the higher magnification of the boxed regions in (a). (c) Section was immunostained with βIII-tubulin antibody (red) and counterstained with Hoechst (blue). Olfactory sensory axons were present in the ONL of OB (white arrows) but absent on the kal knock-down side of the forebrain. The higher magnification of the boxed region is shown in (c1). Representative images show staining on the sections of three OBs (n = 3) under each of the stated conditions. Scale bar: (a) and (b), 100 μm; other images, 20 μm. F: forebrain; black arrow: normal OB; asterisk: abnormal OB. [Colour figure can be viewed at wileyonlinelibrary.com] to control embryos. By contrast, in the kal knock-down OB, BLBP expression appeared to incompletely encircle the surface of anterior forebrain, at the site of the OB anlage, forming a thinner and poorly organized layer (Figure 4a). Strikingly, SOX10-expressing cells were completely absent in the anterior forebrain where there was no OB formation and lack of olfactory axonal connection (Figure 4b). These observations imply that the chick kal product, anosmin, may be a determinant of differential OECs expression of SOX10 and BLBP. It is also possible that the survival and apoptosis of SOX10-and BLBP-positive OECs might be directly or indirectly modulated by anosmin, especially for SOX10-expressing OECs.
GFAP, an astrocyte marker, is also expressed in more developmentally mature OECs in the middle-to-outer region of the ONL (Astic et al., 1998;Murthy et al., 2014;Raucci, Tiong, & Wray, 2013). To examine the role of anosmin-1 in OEC maturation, we further performed GFAP immunostaining on E10 forebrains electroporated with kal shRNA-2. GFAP expression was found to be weakly positive along the peripheral edge of the kal knock-down forebrain but negative on the contralateral side with normal OB (Figure 4c). Taken together, these results support a potential role of anosmin-1 in determining the fate of OECs by preventing OECs from differentiating towards maturity.

| Anosmin-1 inhibits OECs maturation in chick E10 OB explants
As GFAP expression in the ONL is weak during early development stage (Astic et al., 1998), we further investigated the inhibitory capability of anosmin-1 on the maturation of immature OB glial cells in ex vivo explant cultures. We dissected E10 OB from the forebrain and embedded them in collagen gel. As seen in Figure 5a,b, when OB explants (n = 9) were placed in culture for 2 hr, no obvious GFAP-positive cells could be detected in all OB explants. After 3 days of culture, in approximately 90% of explants GFAP-positive cells were observed accumulating along the outermost ONL of the OB (n = 15). Such a response is predominant in the anterior tip of OB at a site where incoming axons of olfactory sensory neurons normally penetrate, which is highly likely due to enhanced differentiation of immature OECs into mature GFAP+ OECs. Following 3 day anosmin-1 treatment, GFAP expression in ONL was dramatically reduced in 80% OB explants (n = 20) (Figure 5a,b, and Figure S1), demonstrating that anosmin-1 is able to inhibit OECs maturation.

| Anosmin-1 inhibits GFAP expression in purified E10 OB immature glial cells
To further investigate the action of anosmin-1 on OEC maturation, we isolated OB glial cells from E10 chick embryos by differential attachment to a tissue culture flask. Enzymatically dissociated cells were shaken to remove neurons and microglia and then glial differentiation induced by addition of medium with G5 supplement. As seen in Figure 6a, purified OB cells expressed transitin (the avian homologue of nestin), vimentin and BLBP, but not other neuronal, oligodendrocyte and radial glia markers including Pax6, TuJ-1 and O4 (data not shown), indicating that they are immature OB glial cells, most likely a mixture of immature OEC subpopulations. With respect to three FGF receptors, the c isoforms of fgfr 1-3 and fgfr 1b, 3b isoforms were found to be expressed. These putative immature glial cells expressed kal (Figure 6b), as shown by RT-PCR, but with very low or even no anosmin expression, which cannot be detected by immunocytochemistry and Western blotting using an antibody specific to chicken anosmin.
After 7 DIV, these OB immature glial cells began to express GFAP. GFAP-positive cells exhibited flat and multipolar morphology (Figure 6c). Overexpression of chicken anosmin by the transfection of an anosmin expression construct demonstrated a more elongated morphology with less GFAP expression as compared with the empty vector control (Figure 7a I). Moreover, Western blotting further showed that overexpression of chicken anosmin significantly reduced GFAP expression by approximately 50% (Figure 7a II and III). Taken together, these data demonstrated that anosmin prevents OB immature glial cells from differentiating into a mature glial state.
We further examined the involvement of FGF signalling in mediating the actions of recombinant anosmin-1 in blocking OEC cell maturation. In Figure 7b I and II, exogenous F I G U R E 5 Anosmin-1 inhibits OECs maturation in chick E10 OB explants. Intact E10 OBs in collagen gel were cultured for 2 hr or 3 DIV with or without 5 nM recombinant anosmin-1, either fixed directly (a) or cryostat sectioned (b), followed by immunohistochemical staining with GFAP antibody. anosmin-1 maintained cells at 7 DIV with reduced GFAP expression by approximately 70%, which was reversed by the FGFR inhibitor Su5402 with cells presenting an elongated morphology. We also count the cells and found no effect of anosmin-1 on the cell numbers (data not shown). A similar effect to down-regulate GFAP expression by approximately 70% was seen with 2 nM FGF2 treatment. Taken together, these data demonstrated a role of anosmin-1 on OB immature glial cell differentiation via FGF signalling, inhibiting differentiation into the mature glial cells by promoting FGF signalling.

| DISCUSSION
Since KAL-1 is absent in the genomes of mouse and rat, other vertebrate species have been employed to investigate the biology of anosmin-1 during development. As previously revealed in zebrafish, anosmin-1 influences fasciculation and terminal targeting of olfactory axons into the OB (Yanicostas, Herbomel, Dipietromaria, & Soussi-Yanicostas, 2009). This might be the direct effect of anosmin-1 on axonal outgrowth and targeting or the secondary effect on OEC differentiation and maturation for defasciculation and synaptogenesis inside the OB. Our chick model using in ovo electroporation to knock-down kal shows that loss of function of anosmin results in abnormal OEC maturation, suggesting a novel mechanism whereby anosmin-1 modulates OEC development during olfactory system ontogeny.
Loss of function mutations in SOX10 has been found in KS patients with hearing impairment (Barnett et al., 2009;Finzsch et al., 2010;Pingault et al., 2013). SOX10, a member of high mobility group domain-containing transcription factors, has recently been found to be associated with olfactory development. In SOX10-deleted mice, expression of the early glial differentiation marker BLBP was significantly decreased in the outmost ONL (Pingault et al., 2013), consistent with our observations that kal depleted chick embryos showed thinner BLBP layer with complete absence of SOX10 in the ONL. We also observed GFAP expression in the ONL at E10, at a much earlier stage than normal GFAP expression when chick anosmin was down regulated. This sort of GFAP upregulation might be accompanied by decreased expression of SOX10 as evidenced in p75-deleted mice that these two OEC markers are expressed in an inverse manner (Raucci et al., 2013). Furthermore, it has been reported that mouse OECs are GFAP positive at E15 in the olfactory axon fascicles, but negative in the ONL until E18.5 with persistent weak expression until P20 (Astic et al., 1998). However, significantly higher kal expression occurs in chick OB from E10 to newly hatched chick (21 days) followed by progressively reduced expression in adulthood (Rugarli et al., 1993); such an expression profile is inversely correlated with GFAP, further reinforcing the fundamental regulatory role of anosmin-1 in delaying OECs maturation from immature SOX10 and BLBP-expressing cells.
Disrupted balance between SOX10/BLBP+ immature OEC and GFAP+ mature OEC populations in the absence of anosmin-1 is thought to be associated with abnormal OB development. In SOX10 null mice, OBs were morphologically well formed and innervated by incoming olfactory axons, in contrast to the situation in the current study in which the OBs were completely lost in the absence of anosmin action. One explanation is that anosmin-1 might regulate other SOX family proteins in additional to SOX10, affecting multifunctional activities between SOX proteins as reported for SOX10 on vagal neural crest cells and various other neural crest cell derivatives (Paratore, Goerich, Suter, Wegner, & Sommer, 2001;Southard-Smith, Kos, & Pavan, 1998). It is also possible that anosmin-1 might determine the differentiation of multiple subpopulations of immature OECs other than SOX10+ OECs only. The fact that BLBP+ OEC layer in anosmin-depleted OBs turned to be dramatically thinner supports this possibility. Another possibility is an unexpected upregulation of GFAP in OECs between E8 and E10 when chick olfactory axons connect to the OBs. It has been shown F I G U R E 7 Anosmin-1 prevents OB immature glial cells from differentiating into a mature state via FGF signalling. (a) OB immature glial cells were transfected with chicken anosmin expression vector and empty vector. OB immature glial cells with chicken anosmin overexpression demonstrated a more elongated morphology shown by immunocytochemistry (I) and diminished GFAP expression by western blot (II). Representative blots from three independent experiments are shown. (III) The relative band intensities were calculated and expressed as the relative GFAP to actin ratio and then compared with the control sample (vector only, 100%). *p < 0.05. (b) (I) OB immature glial cells were cultured in differentiation medium and treated with 5 nM anosmin-1, 2 nm FGF2, 5 nM anosmin-1 plus 20 μM FGFR1 inhibitor Su5402 for 7 DIV. Anosmin-1 and FGF2 reduced GFAP expression. FGFR1 inhibitor, Su5402, reversed anosmin-1 induced GFAP expression. Representative figures are from three independent experiments. Scale bar: 20 μm. (II) The fluorescence intensities of GFAP expression were calculated and compared with the control sample (100%). **p < 0.01. [Colour figure can be viewed at wileyonlinelibrary.com] that the magnitude of GFAP expression in astrocytes is associated with the stiffness of intermediate filaments (Duffy, Huang, & Rapport, 1982). Premature maturation of OECs might lead to loss of immature OEC phenotype of shape, motility and flexibility which facilitate olfactory axonal sorting and specific synapse formation in OBs; this would result in lack of OB cell differentiation required for normal OB morphogenesis. Finally, H&E stained sections showed that there is an irregular mass of mitral cells in the kal knock-down E10 OBs without the typical formation of the stratified structure (Figure 3a1 and a2), possibly altering the environment for OEC maturation, innervation and targeting in ONL.
In chick, transient expression of kal was first observed at the early HH5+ to HH9 developmental stages, when the cranial neural crest is formed; the expression level progressively decreases during or after the time of neural crest cell migration at HH9 (Endo et al., 2012). As SOX10-positive OECs originate from neural crest, it is possible that anosmin-1 plays a role in OEC specification during gastrulation. In the present study, we electroporated chick embryos at HH 11-12 (E2), prior to the developmental window for initial OEC fate specification in the neural crest and then focused on OEC development in the OB at E10; we are therefore not able to give a clear answer to this question as to whether anosmin-1 is involved in OEC fate determination inside neural crest.
OECs stimulate axon growth during development and promote plasticity of spared fibres when implanted into injured CNS (Raisman & Li, 2007;Tennent & Chuah, 1996); these properties make them an attractive cell type for cell-based treatment of neuronal regeneration after CNS injury (Gómez et al., 2018). Different subpopulations of OECs are believed to perform distinct subtype-specific functions (Honore et al., 2012;Ubink & Hokfelt, 2000). One of the interesting properties of OECs is their interaction with astrocytes, preventing reactive astrocytosis and glial scar formation after their implantation at a CNS lesion site (Lakatos, Barnett, & Franklin, 2003;Lakatos, Franklin, & Barnett, 2000;Santos-Silva et al., 2007). These actions are likely to be mediated by immature OECs. It has been suggested that FGF signalling via an FGF2/ FGFR1/HS pathway underlies the molecular regulation of OECs/astrocyte interaction (Higginson et al., 2012;Santos-Silva et al., 2007). We have previously proposed a mechanism whereby anosmin-1 has a dual role on the assembly and activation of the FGF2/FGFR1/HS complex (Hu et al., 2009). Our present results might support a potential role of anosmin-1 as a candidate molecule in preserving OECs in an immature state to minimize astrocyte activation for the incoming or regenerative axons navigating through an astroglial barrier.
Anosmin-1 is not sufficient by itself to induce proper OB formation as evidenced by the fact that ablation of chick olfactory placode resulted in histologically abnormal OB where anosmin-1 was still expressed at high levels (Lutz et al., 1994). In anosmin-1 mutated human KS foetuses and FGFR1-deficient mice, neurogenesis of olfactory neurons appears to be preserved (Hebert, Lin, Partanen, Rossant, & McConnell, 2003;Schwanzel-Fukuda et al., 1989;Teixeira et al., 2010). This implies that the main consequences of anosmin-1/FGF signalling lie within the domain of glia development/interaction, especially modifying OECs maturation. Our studies enable us to propose a model whereby, during F I G U R E 8 A model proposed for anosmin-1 regulation on OEC maturation via FGF signalling for OB development. (a) During OB development, SOX10+ and BLBP+ OECs, after birth in the neural crest, wrap olfactory sensory axons migrating towards the anterior forebrain where they, along with olfactory axons, form the ONL of OB. Anosmin-1 keeps SOX10+ and BLBP+ OECs in an immature state to assist olfactory axon targeting, penetration, defasciculation and synapse formation inside the OB, subsequently inducing OB morphogenesis. (b) Loss of function of anosmin-1 observed in X-KS embryos on FGF signalling regulation appears to cause OEC maturation/differentiation into GFAP+ cells, thereby preventing olfactory axons forming stable connections with the forebrain. CP: cribriform plate; F: forebrain; OB: olfactory bulb; OE: olfactory epithelium; ORN: olfactory receptor neuron OB development, loss of function of anosmin-1 on FGF signalling regulation disrupts OEC maturation/differentiation, thereby interfering with olfactory axons from targeting and penetrating into the forebrain (Figure 8).