Thrombospondin 4 deficiency in mouse impairs neuronal migration in the early postnatal and adult brain

Molecular and Cellular Neuroscience 61 (2014) 176–186

Contents lists available at ScienceDirect

Molecular and Cellular Neuroscience journal homepage: www.elsevier.com/locate/ymcne

Thrombospondin 4 deficiency in mouse impairs neuronal migration in the early postnatal and adult brain F. Girard ⁎, S. Eichenberger, M.R. Celio Anatomy Unit and Program in Neuroscience, Department of Medicine, Faculty of Science, University of Fribourg, Route A. Gockel 1, CH1700 Fribourg, Switzerland

a r t i c l e

i n f o

Article history: Received 24 September 2013 Revised 24 April 2014 Accepted 20 June 2014 Available online 28 June 2014 Keywords: Rostral-migratory stream Neurogenesis Astrocytes Extra-cellular matrix Thrombospondin 1

a b s t r a c t In the post-natal rodent brain, neuronal precursors originating from the sub-ventricular zone (SVZ) migrate over a long distance along the rostral migratory stream (RMS) to eventually integrate the olfactory bulb neuronal circuitry. In order to identify new genes specifically expressed in the RMS, we have screened the Allen Brain Atlas Database. We focused our attention on Thrombospondin 4 (Thbs4), one of the 5 members of the Thrombospondin family of large, multidomain, extracellular matrix proteins. In post-natal and adult brain Thbs4 mRNA and protein are specifically expressed in the neurogenic regions, including the SVZ and along the entire RMS. RMS cells expressing Thbs4 are GFAP (Glial Fibrillary Acidic Protein) positive astrocytes. Histological analysis in both wild-type and Thbs4 knock-out mice revealed no major abnormality in the general morphology of these neurogenic regions. Nevertheless, immunostaining for doublecortin demonstrates that in Thbs4-KO, migration of newly formed neurons along the RMS is somehow impaired, with several neurons migrating out of the RMS. This is further supported by a Bromodeoxyuridine-based in vivo approach showing a decrease in the number of newly born neuronal precursors reaching the olfactory bulb, while proliferation in the SVZ is not affected compared to wild-type, both in young animals (P15) and in adults (8 to 12 weeks of age). Corroborating this observation, the number of Parvalbumin- and Calbindin-immunoreactive interneurons in the olfactory bulb is also reduced in Thbs4-KO. Together, these observations support a role for the astrocyte-secreted protein Thbs4 in the migration of newly form neurons within the RMS to the olfactory bulb. © 2014 Elsevier Inc. All rights reserved.

Introduction In the post-natal rodent brain, neurogenesis persists throughout adulthood, in mainly two neurogenic regions: the subventricular zone (SVZ) of the lateral ventricles, and the subgranular zone (SGZ) of the dentate gyrus. While in the latter, newly generated neuronal cells migrate over a short distance and differentiate into a single neuronal type (excitatory glutamatergic granule neurons), immature neurons generated within the SVZ migrate over a long distance to reach the olfactory bulb (OB) and differentiate into distinct neuronal types (the majority becoming GABA- or dopaminergic, and few being glutamatergic). In the SVZ, the neural stem cells (NSCs) are specialized astrocytes (Type B cells). These cells give rise to rapidly dividing transit-amplifying cells (Type C cells), which in turn generate neuronal precursors (Type A cells). These later migrate along the so-called rostral migratory stream (RMS) through a process called tangential migration involving homophilic interaction of neuronal precursors, surrounded by astrocytes. Once they reach the OB, immature neurons switch to radial migration, differentiate into granule and periglomerular neurons, and eventually integrate existing neuronal circuitry. Neurogenesis in the adult brain has been the subject of ⁎ Corresponding author. E-mail address: [email protected] (F. Girard).

http://dx.doi.org/10.1016/j.mcn.2014.06.010 1044-7431/© 2014 Elsevier Inc. All rights reserved.

numerous reviews in the past decade (for recent reviews, see Basak and Taylor, 2009; Cayre et al., 2009; Faigle and Song, 2013; Ihrie and Alvarez-Buylla, 2011; Kriegstein and Alvarez-Buylla, 2009; Ming and Song, 2011; Whitman and Greer, 2009; Zhao et al., 2008). Numerous studies have led to the identification of intrinsic and extrinsic factors, such as growth and trophic factors, transcription factors and cell adhesion molecules, involved in various aspects of adult neurogenesis, including neural stem cells, proliferation of immature neurons, their migration through the RMS and their differentiation in the OB (see references aforementioned). In particular, the importance of the extracellular environment surrounding the migrating cells has been emphasized. Indeed, several molecules of the extracellular matrix (ECM) are expressed along the RMS, and are crucial for migration. These include tenascin C, heparan- and chondroitin sulfate proteoglycans, laminins, matrix metalloproteases, thrombospondin 1, netrin, and reelin (reviewed in Dityatev et al., 2010; Kazanis and ffrench-Constant, 2011). In this study, we have screened the Allen Brain Atlas Database to identify genes specifically expressed in the RMS. We report here the implication of the ECM protein Thrombospondin-4 (Thbs4) in early postnatal and adult mouse neurogenesis. Because of their ability to bind various ligands, proteins of the thrombospondin family have been involved in the regulation of diverse cellular processes, including cell proliferation and migration, ECM remodeling, and various pathologies

F. Girard et al. / Molecular and Cellular Neuroscience 61 (2014) 176–186

(vascular and heart pathologies, cancers, skeletal pathologies) (reviewed in Adams, 2001; Mosher and Adams, 2012; Risher and Eroglu, 2012). We show here that Thbs4 is expressed in the post-natal mouse brain in RMS astrocytes. In Thbs4 knock-out mice, although proliferation of neuronal precursors in the SVZ is not affected, a decrease in the number of early born neurons reaching the olfactory bulb implicates the ECM protein Thbs4 in neuron migration along the RMS. Results Searching for genes expressed within the rostral migratory stream We, and others, have previously shown that the Allen Brain Atlas Database of in situ hybridization data (ABA) can be efficiently screened to identify regionally enriched expression of genes in discrete brain areas

177

(Girard et al., 2011, and references therein). The gene finder search facility of AGEA at ABA enables users to search a local region of interest for the top 200 genes within the ABA database, that exhibit localized enrichment in this particular region. We performed such a screen of the ABA in situ hybridization data, focusing on two particular regions: the descending RMS and the straight arm of the RMS abutting the olfactory bulb entry (See Fig. 1). From each screen, the top 200 genes were further visually examined for their RMS expression. We eliminated those genes with ubiquitous brain expression, or too low – thus questionable – expression in the RMS. A list of 93 genes showing enriched expression in the RMS was eventually obtained (Table 1 and Fig. 1). According to Gene Ontology, these genes can be classified as follows: Transcriptional regulators: Arx, Dlx1, Dlx2, Dlx6os1, Id3, Id4, Mrg1, Myt1, Nfia, Nfib, Nfix, Rai17, Sall3, Sox2, Sox11, Sp8, TcfE2a, Tcf4, Tle1, Tbtb20, Ybx1, Zfhx1b, and Zfp57; translational regulators: Arbp, Rps5, Rps12, and

Fig. 1. Examples of genes specifically expressed in the rostral migratory stream. The first panel on the left panel shows Nissl staining of a mouse brain sagittal section, depicting the rostral migratory stream (RMS), the sub-ventricular zone (SVZ) and the main olfactory bulb (MOB). Arrows mark the positions chosen for the ABA data screen using the AGEA tool: the descending RMS and the straight arm of the RMS. Scale bar is 1 mm. All the other panels are ISH images taken from the ABA, and show SVZ/RMS specific expression of some genes identified through the ABA screening. This compound image completes the list presented in Table 1: only those genes for which no data exist concerning involvement in neurogenesis/neuronal migration are shown here.

178

F. Girard et al. / Molecular and Cellular Neuroscience 61 (2014) 176–186

Table 1 Results of the ABA screen for genes enriched in the RMS. Gene symbol

Complete name

Gene Ontology

Arbp Arx Btg1 Carhsp1 Ccnd2 Cd9 Cd24a Cd63 Ckap2 Clu Cspg2 C4b Dbi Dkk3 Dlx1 Dlx2 Dlx6os1 D0H4S114 Dpysl3 Dpysl5 Edg1 Gfap Gja1 Gldc Gnai2 Gpc2 Grid2ip Hepn1 Hist1h1a Hist1h1b Hist1h2bg Id3 Id4 Igfbp5 Lrp8 Marcksl1 Mdk Mki67 Mmp14 Mrg1 Msi2 Mtss1 Myt1 Neil3 Nfia Nfib Nfix Nnat Ntrk2 Plxnb2 Ppap2b Ppp1r14b Prokr2 Ptprz1 Rai17 Rhobtb3 Rps5 Rps12 Rps15a Rrm2 Sall3 Samd14 Scara3 Sdk2 Sema6c Sfrp1 Slit1 Smoc1 Sox2 Sox11 Sp8 Spag5 Stmn1 Syncrip

Acidic ribosomal phosphoprotein P0 (=Rplp0) Aristaless related homeobox gene B-cell translocation gene 1 Calcium regulated heat stable protein 1 Cyclin D2 CD9 antigen CD24a antigen Cd63 antigen Cytoskeleton associated protein 2 Clusterin Chondroitin sulfate proteoglycan 2 (=NG2 = versican) Complement component 4B Diazepam binding inhibitor Dickkopf homolog 3 Distal-less homeobox 1 Distal-less homeobox 2 Dlx6 opposite strand transcript 1 (=Evf2) Human D4S114 (=P311) Dihydropyrimidinase-like 3 (=TUC4 = CRMP4) Dihydropyrimidinase-like 5 (=CRMP5) Endothelial differentiation sphingolipid G-protein-coupled receptor 1(=S1pr1) Glial fibrillary acidic protein Gap junction membrane channel protein alpha 1 (=connexin43) Glycine decarboxylase Guanine nucleotide binding protein, alpha inhibiting 2 Glypican 2 (cerebroglycan) Glutamate receptor, ionotropic, delta 2 (Grid2) interacting protein 1 (=delphilin) Hepacam, hepatocyte cell adhesion molecule Histone 1, H1a Histone 1, H1b Histone 1, H2bg Inhibitor of DNA binding 3 Inhibitor of DNA binding 4 Insulin-like growth factor binding protein 5 Low density lipoprotein receptor-related protein 8 (=ApoER2) MARCKS-like 1 Midkine Antigen identified by monoclonal antibody Ki 67 Matrix metallopeptidase 14 Myeloid ecotropic viral integration site-related gene 1 (=Meis2) Musashi homolog 2 Metastasis suppressor 1 Myelin transcription factor 1 Nei like 3 Nuclear factor I/A Nuclear factor I/B Nuclear factor I/X Neuronatin Neurotrophic tyrosine kinase, receptor, type 2 Plexin B2 Phosphatidic acid phosphatase type 2B Protein phosphatase 1, regulatory (inhibitor) subunit 14B Prokineticin receptor 2 Protein tyrosine phosphatase, receptor type Z, polypeptide 1 Retinoic acid induced 17 (=Zmiz1) Rho-related BTB domain containing 3 Ribosomal protein S5 Ribosomal protein S12 Ribosomal protein S15a Ribonucleotide reductase M2 Sal-like 3 Sterile alpha motif domain containing 14 Scavenger receptor class A, member 3 Sidekick homolog 2 Semaphorin 6C Secreted frizzled-related sequence protein 1 Slit homolog 1 SPARC related modular calcium binding 1 SRY-box containing gene 2 SRY-box containing gene 11 trans-acting transcription factor 8 Sperm associated antigen 5 Stathmin 1 Synaptotagmin binding, cytoplasmic RNA interacting protein

Translational regulation Transcriptional regulator (homeobox) (*) Cell cycle/proliferation/apoptosis RNA binding Cell cycle/proliferation/apoptosis Cell adhesion/migration Cell adhesion/migration (*) Lysosome/endosome function Cytoskeletal dynamics/chromosome segregation Chaperone activity ECM organization/cell migration (*) Immune/inflammatory response Neuropeptide Wnt signaling pathway regulation Transcriptional regulator (homeobox) (*) Transcriptional regulator (homeobox) (*) Non-coding RNA, transcriptional regulator Unknown Enzyme (neuronal progenitor marker) (*) Enzyme, axon guidance/neurite outgrowth/neurogenesis (*) Gpcr (*) Cytoskeletal dynamics (NSC/astrocyte marker) (*) Gap junction channel (*) Enzyme, glycine metabolism GTPase activity/GPCR signaling Cell adhesion/ECM Modulation of glutamate receptor signaling Cell adhesion/migration Cell cycle/proliferation/apoptosis Cell cycle/proliferation/apoptosis Cell cycle/proliferation/apoptosis Transcriptional regulator (*) Transcriptional regulator (*) Negative regulation of insulin-like growth factor receptor signaling pathway Receptor; Reelin pathway (*) Cell cycle/proliferation/apoptosis Growth factor Cell cycle/proliferation/apoptosis (proliferating cell marker) Enzyme, ECM organization/remodeling Transcriptional regulator (homeobox) (*) RNA binding (*) Cytoskeletal dynamics Transcriptional regulator (zinc finger) DNA repair (*) Transcriptional regulator (*) Transcriptional regulator (*) Transcriptional regulator (*) Unknown Neurotrophin receptor (*) Cell adhesion/migration (*) Enzyme, lipid phosphatase Phosphatase inhibitor GPCR (*) Midkine receptor Transcriptional regulator (zinc finger) Unknown Translational regulation Translational regulation Translational regulation Enzyme, nucleotide metabolism/DNA replication and repair Transcriptional regulator (zinc finger) (*) Unknown Unknown Cell adhesion/migration Cell adhesion/migration Wnt signaling pathway regulation Cell adhesion/migration (*) ECM organization Transcriptional regulator (HMG box) (NSC marker) (*) Transcriptional regulator (HMG box) (*) Transcriptional regulator (zinc finger) (*) Cytoskeletal dynamics Cytoskeletal dynamics RNA binding

F. Girard et al. / Molecular and Cellular Neuroscience 61 (2014) 176–186

179

Table 1 (continued) Gene symbol

Complete name

Gene Ontology

Syne2 S100a16 Tcfe2a Tcf4 Thbs4 Tiam2 Tle1 Tnfrsf19 Tnrc9 Top2a Unc84b Vim Ybx1 Zbtb20 Zfhx1b Zfhx2as Zfp57 2610109H07Rik 4833414E09Rik

Synaptic nuclear envelope 2 S100 calcium binding protein A16 Transcription factor E2a (=Tcf3) Transcription factor 4 Thrombospondin 4 T-cell lymphoma invasion and metastasis 2 Transducin-like enhancer of split 1 Tumor necrosis factor receptor superfamily, member 19 (=Troy) Trinucleotide repeat containing 9 (=Tox3) Topoisomerase (DNA) II alpha Unc-84 homolog B (=Sun2) Vimentin Y box protein 1 Zinc finger and BTB domain containing 20 Zinc finger homeobox 1b Zinc finger homeobox 2, antisense Zinc finger protein 57 RIKEN cDNA 2610109H07 gene (=Draxin/neucrin) RIKEN cDNA 4833414E09 gene

Cytoskeletal/nuclear dynamics (*) Calcium-binding Transcriptional regulator (*)/Wnt signaling pathway Transcriptional regulator (*)/Wnt signaling pathway ECM organization GTPase activity (*) Transcriptional regulator (*) Cytokine receptor (*) Unknown Enzyme, cell cycle/proliferation/apoptosis Cytoskeletal/nuclear dynamics (*) Cytoskeletal dynamics Transcriptional regulator Transcriptional regulator (zinc finger) Transcriptional regulator (zinc finger) (*) Unknown Transcriptional regulator (zinc finger) Wnt signaling pathway regulation (*) Unknown

The abbreviated gene name, the complete name, and the classification according to Gene Ontology are given. Asterisks highlight genes for which an involvement in neurogenesis/neuronal migration is documented in the literature. The expression of the genes without asterisk is documented in Fig. 1.

Rps15a; cell cycle/DNA replication and repair/proliferation/apoptosis: Btg1, Ccnd2, Hist1h1a, Hist1h1b, Hist1h2bg, Marcksl1, Mki67, Neil3, Rrm2, and Top2a; cell adhesion/migration: Cd9, Cd24a, Hepn1, Plxnb2, Sdk2, Sema6c, and Slit1; extracellular matrix organization: Cspg2, Gpc2, Mmp14, Smoc1, and Thbs4; receptors: Edg1, Lrp8, Ntrk2, Prokr2, Ptprz1, and Tnfrsf19; cytosleletal dynamics: Ckap2, Gfap, Mtss1, Spag5, Stmn1, Syne2, Unc84b, and Vim; RNA binding: Carhsp1, Msi2, and Syncrip; miscellaneous: Cd63, Clu, C4b, Dbi, Dkk3, Dpysl5, Gja1, Gldc, Gnai2, Igfbp5, Mdk, Ppap2b, Ppp1r14b, Sfrp1, S100a16, Tiam2, and 2610109H07Rik; and unknown: D0H4S114, Dpysl3, Grid2ip, Nnat, Rhobtb3, Samd14, Scara3, Tnrc9, Zfhx2as, and 4833414E09Rik. From these results we can draw several conclusions. A significant proportion of these genes are expressed, apart from the RMS and SVZ, in very few brain areas. Most of them also display expression in the subgranular zone of the dentate gyrus, the other main neurogenic region of the adult rodent brain (see the ABA ISH images for SGZ expression). Finally, for some of these genes, expression in the brain is only detected in the SVZ/RMS/SGZ areas. Collectively, these observations suggest rather specific roles of these genes in the global process of neurogenesis. For 35 of these genes, data in the literature support roles in proliferation of neuronal precursors, neuronal migration, neuronal differentiation and specification of neuroblast identity, while for the others (most of them being presented in Fig. 1), there is no published data concerning their expression or possible involvement in neurogenesis. Numerous transcription factors were identified, including some with known functions in neurogenesis (Arx, Dlx1, Dlx2, Id4, Id4, Sall3, Sox2, Sox11 and Sp8, Zfhx1b). These transcription factors are of different types: HMG domain, zinc finger, and homeobox (Table 1). Another over-represented class comprises genes encoding proteins involved in cell cycle/proliferation/DNA replication and repair/apoptosis, which is not surprising since SVZ and RMS are zones of active cellular proliferation. In addition, the screen revealed several genes encoding proteins involved in regulating the Wnt signaling pathway (Dkk3, Sfrp1, Draxin/neucrin, Tcf2a, Tcf4, Syne2), further highlighting the importance of the Wnt pathway in neurogenesis. Several genes encoding proteins of the ECM were also identified, including the chondroitin sulfate proteoglycan family protein Cspg2 (also called versican or NG2), glypican 2 (Gpc2), the matrix metalloproteinase MMP14, the EF-hand calcium binding protein Smoc1, and the matricellular calcium binding protein thrombospondin 4. Because of the general interest of our group in calcium-binding proteins, the importance of the ECM in neurogenic niche organization and function, and since two recent studies reported the involvement of thrombospondin 1 in adult neurogenesis

(Blake et al., 2008; Lu and Kipnis, 2010), we decided to focus our attention on the secreted ECM protein thrombospondin 4. Thrombospondin 4 is expressed along the RMS In situ hybridization experiments show a very specific expression of Thbs4 mRNA in the neurogenic regions of adult mouse brain, including the SVZ and along the entire RMS, and an absence of expression in the OB (Fig. 2, and ABA data set for Thbs4). Polyclonal antibody to Thbs4 detects the protein mostly extracellularly in a similar pattern (Fig. 3E). The number of Thbs4 expressing cells is higher in younger animals (P14), when neurogenesis occurs at a higher rate (Fig. 2C). Expression is also detected in the corpus callosum, in the hippocampal formation, both in the SGL of the dentate gyrus and in the radiatum layer (Fig. 2C–G and ABA data set for Thbs4). In a recent report, Benner et al. (2013) have shown that SVZ contains cells expressing Thbs4 which are GFAP positive (astrocytic marker), Dcx negative (a marker for early born neurons), NG2/Olig2 negative (a marker for oligodendrocyte precursors) and Mash1 negative (marker for NSCs and TAPs). Double ISH/immunostaining experiment reveals that all Thbs4 expressing cells in the RMS are astrocytes, i.e. immunoreactive for GFAP (Fig. 3A–D). Thbs4 mRNA is detected only in a subset of GFAP positive astrocytes present in the RMS, as some GFAPN0 cells were Thbs4b0. Similar observations were made in the corpus callosum (not shown). These results are in agreement with previous studies showing that thrombospondins 1, 2 and 4 are astrocyte-secreted proteins (Christopherson et al., 2005; Kim et al., 2012; Lu and Kipnis, 2010). Histological analysis in Thbs4 KO mice Thbs4−/− mice appear normal at birth and displayed no obvious phenotype during early and adult development (Frolova et al., 2010). We controlled that Thbs4 immunoreactivity in the neurogenic regions and corpus callosum was completely abolished in Thbs4 KO brains (Fig. 3F). We first compared by Nissl staining the general morphology of the neurogenic regions and OB in wild-type and Thbs4−/− mouse brains, both in young adults (P14 to P21) (Fig. 4) and older animals (8 to 12 weeks of age) (not shown). At the histological level, we found no obvious differences between both genotypes, in the morphology of the RMS, the size of the OB, or the cell density within the RMS (Fig. 4A–F). This was further confirmed by immunostaining for GFAP and PSA-NCAM (a marker for proliferating neuroblasts), which showed that in Thbs4−/−, chains of migrating neurons form correctly and are ensheathed in “tunnel-like” structures formed by GFAP-positive

180

F. Girard et al. / Molecular and Cellular Neuroscience 61 (2014) 176–186

Fig. 2. Thbs4 gene expression in the adult neurogenic regions. A: Thbs4 mRNA detection in a 14 μm sagittal brain cryosection from a P60 mouse. The image focuses on Thbs4 expression in the RMS. B: schematic representation of sagittal brain section. Arrows point to the area presented at high magnification in panels D to F, respectively RMS, SVZ, hippocampus (hylus) and corpus callosum. C to G: Thbs4 mRNA detection in a 14 μm sagittal brain cryosection taken from a P14 mouse, using DIG-labeled Thbs4 full-length antisense probe. Scale bar in panel C is 500 μm. D to G are higher magnifications from different brain areas (as schematized in panel B) of the composite image presented in panel C. Abbreviations: MOB, main olfactory bulb; RMS, rostral migratory stream; SVZ, sub-ventricular zone; CTX, cortex; HPF, hippocampal formation; VL, lateral ventricle; cc, corpus callosum.

astrocytes similar to wild-type (Fig. 4G–H). Nevertheless, when examining doublecortin (Dcx) immunostaining (a marker for new, yet immature, neurons), we observed subtle differences between wild-type and Thbs4 KO mice. In wild-type mice, most neurons are correctly arranged tangentially, with their processes oriented parallel to the RMS, and occasionally, few neurons are seen outside of the RMS (Fig. 5A, B). In contrast, in Thbs4−/− mice, chains of Dcx-positive neuronal precursors at the edge of the RMS are disorganized, with numerous Dcx-positive neurons showing processes oriented perpendicular to the RMS (Fig. 5C, D). In addition, several Dcx positive cells are seen outside the RMS. These observations suggest that migration in Thbs4 KO is perturbed, with

some neurons possibly disassembling from the RMS, and adopting an erratic migration. Neuronal precursor proliferation and migration in Thbs4 deficient mouse Since Thbs4 expression is seen both in the SVZ (a place of active cell proliferation) and RMS (where newly form neurons migrate actively), we monitored both cell proliferation and cell migration using a classical Brdu based in vivo approach. Wild-type and Thbs4−/− mice, young animals (P14/15) and adults (8 to 12 weeks of age), received a single pulse of Brdu (2 h), and cell proliferation in the SVZ was analyzed by

F. Girard et al. / Molecular and Cellular Neuroscience 61 (2014) 176–186

181

Fig. 3. Thbs4 expressing cells are GFAP positive astrocytes. A–C: Double detection of Thbs4 mRNA by ISH (A) and GFAP by immunostaining (B) in a 14 μm sagittal brain cryosection from P60 mouse, at the level of the RMS. Panel C is a merged image. Note that all Thbs4 expressing cells are GFAP positive, but that several GFAP positive are Thbs4 negative. D: merged image of Thbs4 mRNA and GFAP immunostaining obtained by confocal microscopy, highlighting GFAP immunoreactive filaments associated with the two cells expressing Thbs4. E–F: Thbs4 immunostaining in 30 μm sagittal sections from Thbs4+/+ (E) and Thbs4−/− (F) P60 brains. Scale bars: A–C: 50 μm; D: 25 μm; E–F: 1 mm.

monitoring Brdu incorporation in proliferating cells through Brdu immunostaining. We found no significant difference in the number of Brdu positive cells in the SVZ between both genotypes, in young animals and in adults (Fig. 6A and B respectively), showing that Thbs4 deficiency does not alter cell proliferation in the SVZ. The migration of newly formed neuronal precursors along the RMS to the OB was also examined, especially in light of the phenotype observed with Dcx immunostaining. Young mice (P17/18) received a single injection of Brdu (100 mg/kg), and were euthanized 5 days later (P22/23). Positive Brdu positive neuronal precursors that have reached the OB were counted in all the different cell layers of the OB (with the exception of the sub-ependymal zone) (Fig. 7A). We noticed a statistically significant difference in the number of Brdu positive cells between both genotypes, with a 17% decrease when comparing to Thbs4+/+. The same observation was made in adult mice, with Thbs4−/− mice showing a 28% decrease in the number of young neurons reaching the OB (Fig. 7B). This observation suggests that neuronal migration within the RMS is slightly impaired in Thbs4 knockout, resulting in a lower number of neurons reaching the OB. Since Thbs1 was previously shown to influence the migration of neuronal precursors within the RMS in young animals (Blake et al., 2008) and proliferation in the SVZ of adult mice (Lu and Kipnis, 2010), we wanted to examine to what extent cell migration to the OB would be affected in double KO mice for Thbs1 and 4. In a similar in vivo Brdu assay, single Thbs1 deficiency alone results in lower Brdu positive cells reaching the OB in young animals (25% less compared to wild-type) (Fig. 7A), as already described (Blake et al., 2008), but not in adults (Fig. 7B). In young double KO animals, the decrease of Brdu positive cells in the OB was higher than both simple KO (40% less compared to wild-type) (Fig. 7A). This effect was not observed in the adult mice, as both Thbs4 KO and Thbs1Thbs4 double KO showed a similar decrease in Brdu positive cells (29–25% respectively) when compared to wild-type (Fig. 7B). As an alternative method to Brdu incorporation, we examined Parvalbumin (Pvalb) and Calbindin (CalbD-28K) immunoreactivity in neurons in the OB from P22 brains. Pvalb is specific for the outer plexiform layer, while CalbD-28K marks the glomerular layer (Celio, 1990, and Fig. 7D). In Thbs4 KO, we observed a significant decrease in the number of neurons stained for Pvalb and CalbD-28K, when compared to wild-type, respectively 20 and 25% (Fig. 7E), corroborating the observations made with the Brdu based approach.

Discussion Members of the Thrombospondin family were recently shown to have important functions in early post-natal and adult rodent brain, both during neurogenesis for Thbs1 (Blake et al., 2008; Lu and Kipnis, 2010) and astrogenesis for Thbs4 (Benner et al., 2013). Here, we present evidence that Thbs4 is specifically expressed by astrocytes along the entire RMS, and plays a role in the migration of newly formed neurons within the RMS, both in young animals and adults. Thbs4 is expressed in a restricted pool of GFAP positive astrocytes in the SVZ, the RMS, the corpus callosum, and the hippocampus. No other expression is detected outside of these areas, suggesting a very specific role for Thbs4 in neurogenesis. By contrast, Arber and Caroni (1995) described a neuronal localization of Thbs4 immunoreactivity in the cerebellum of adult rats, and using ISH to Thbs4 mRNA they observed staining in the cerebellum and in the hippocampus (CA1, CA3, and DG). Expression in the neurogenic regions was not mentioned in their report. Thus these results are difficult to reconcile with the data from the ABA data and our own results. We used a commercially available anti-Thbs4, which shows exactly the same pattern of expression as Thbs4 mRNA, and which gives no detectable staining in Thbs4 KO mice, thus confirming the specificity of these experiments. Our results agree with a recent report by Benner et al. (2013) showing that Thbs4 is expressed by SVZ astrocytes (GFAPN0, DcxN0, NG2/Olig2N0 and Mash1N0), and is up-regulated in SVZ astrocytes upon injury in mouse brain. Although Thbs4 shows SVZ expression, cell proliferation assay in vivo in young and adult animals clearly reveal no effect of Thbs4 in the proliferation of neuronal precursors in the SVZ (Fig. 6). Two recent studies examined the importance of Thbs1 in post-natal neurogenesis (Blake et al., 2008; Lu and Kipnis, 2010), and argued for a role of Thbs1 in the proliferation of neuronal precursors, both in the SVZ and the SGZ, through in vivo Brdu incorporation assay in Thbs1−/− mice. Interestingly, this effect on proliferation was observed only in adults (8–10 weeks of age), and not in younger animals (P7 to P21). As already observed for Thbs1 (Blake et al., 2008), the general morphology of the RMS (as analyzed by Nissl staining) is not severely altered by Thbs4 deficiency. Nevertheless, Dcx immunostaining reveals some defects in neuronal RMS migration (Fig. 5). In addition, in Thbs4−/− background, the number of Brdu positive neuronal precursors reaching the OB is

182

F. Girard et al. / Molecular and Cellular Neuroscience 61 (2014) 176–186

Fig. 4. Histological/immunological analysis in Thbs4 KO mice. A to F: Nissl staining of 30 μm sagittal sections from Thbs4−/− (A, C, E) and Thbs4+/+ (B, D, F) brains at post-natal day P21. Panels C to F show higher magnifications, focusing on two portions of the RMS (highlighted with arrows in panel A). Scale bar is 100 μm for panels C to F. G, H: double immunostaining for GFAP (green) and PSA-NCAM (red) in 30 μm sagittal sections from Thbs4+/+ (G) and Thbs4−/− (H) P21 brains. The pictures are taken at the position shown in panels C and D (descending RMS).

decreased, both in young animals and adults. This was correlated in the Thbs4 KO with a decrease in the number of Parvalbumin and CalbindinD-28K immunoreactive neurons, in the olfactory bulb (Fig. 7). Together, these observations support a role for Thbs4 in neuronal migration along the RMS. Lack of Thbs4 could decrease the adhesive effect, resulting in some Dcx positive neurons migrating outside the RMS. In addition, the speed of migration could also be affected in Thbs4−/−, eventually resulting in less neurons reaching the OB during a defined period of time. Thbs4 has been involved in cell proliferation/ adhesion/migration events in different cell types, including neutrophils and smooth muscle cells (Congote et al., 2004; Pluskota et al., 2005; Stenina et al., 2003). In the developing retina, Thbs4 was proposed to act as an organizer of adhesive- and axon outgrowth-promoting molecules in the ECM (Dunkle et al., 2007). The structural basis for the versatility of functions of thrombospondins relies on their ability to interact with different partners, including α2δ1 and neuroligin in synaptogenesis (Eroglu et al., 2009; Xu et al., 2010), VLDR/ApoER2 in the RMS (Blake et al., 2008) and integrin (reviewed in Risher and Eroglu, 2012). Although these studies were focused on Thbs1, it was reported that Thbs4 has the ability, at least in vitro, to bind several of the Thbs1 interacting partners. Thbs1 was shown to stabilize chains of migratory neuronal precursors from SVZ explants, and in Brdu assay, a significant reduction (almost 50%) of Brdu positive cells reaching the OB in young

animals was noticed (Blake et al., 2008). Interestingly, we found here that this effect was not observed in adults (Fig. 7). From these results, it was proposed that, through direct binding of Thbs1 to ApoER2 and VLDLR receptors, Thbs1 might stabilize the chain of migrating neuroblasts within the RMS, and that after reaching the OB entry, Reelin could displace Thbs1 from ApoER2/VLDLR, thus inducing a detachment of migrating neurons from the chains and a switch to radial migration in the OB (Blake et al., 2008). Thus, based on Blake et al. (2008) and our study, this raises the question whether both thrombospondins might have partially redundant roles in the stabilization of migrating neuroblast chains in the RMS. Here we found that in double KO for Thbs1 and Thbs4, the number of Brdu positive cells was more severely affected than in the simple KO, but that the decrease observed (40%) was the addition of the decrease seen in both simple KO (17% for Thbs4 and 25% for Thbs1). Thus, it is perhaps more plausible that both proteins act on different pathways. Thbs1 was proposed to act in the RMS as an alternative physiological ligand to Reelin for binding to VLDR/ApoER2, without activation of the canonical Reelin signaling pathway (Blake et al., 2008). One interesting candidate for Thbs4 is the Notch pathway, given the recently described ability of Thbs4 to directly interact with the Notch1 receptor to increase astrogenesis after injury (Benner et al., 2013). ABA images for Notch1 reveals a strong expression in the SVZ, RMS and OB.

F. Girard et al. / Molecular and Cellular Neuroscience 61 (2014) 176–186

183

Fig. 5. Dcx immunostaining reveals defects in the RMS in Thbs4 KO mice. Dcx immunostaining in the RMS, in Thbs4+/+ (A, B) and Thbs4−/− (C, D) P21 mice. Shown are low and high magnification images, highlighting the disorganization of Dcx positive migrating neurons in the RMS in Thbs4 KO. Scale bars: 250 μm (A–B) and 125 μm (C–D).

In conclusion, Thbs4 has to be added to the growing list of ECM proteins involved in proliferation/migration of neuronal precursors within the neurogenic niches in the early post-natal and adult brain. While Thbs1 (Blake et al., 2008; Lu and Kipnis, 2010), but not Thbs4 (this study), is involved in cell proliferation in the adult SVZ, both thrombospondins appear to be important for migration of newly born neurons within the RMS and their eventual integration in the OB, with the difference that Thbs1 involvement is observed only in young animals, while Thbs4 function in neuron migration also extends to adult. Experimental methods Animals C57BL/6 mice were used for in situ hybridization experiments. The mouse strain B6.129P2-Thbs4tm1Dgen/J (Thbs4 KO in C57BL/6 background) was obtained from The Jackson Laboratory (Bar Harbor, Maine, USA). In this strain, the Thbs4 gene was inactivated by replacing 14 nucleotides in the first intron with a LacZ-neomycin resistance cassette (Lac0-SA-IRES-lacZ-Neo555G/Kan (for additional information, see Frolova et al., 2010, and http://jaxmice.jax.org/strain/005845. html). Heterozygous Thbs4+/− mice were crossed with each other in order to generate homozygous Thbs4+/+ (referred here as wild-type) and Thbs4−/− mice. The mouse strain B6.129S2-Thbs1tm1Hyn/J (Thbs1 KO in C57BL/6 background) was obtained from The Jackson laboratory (http://jaxmice.jax.org/strain/006141.html). Double KO mice for both thrombospondins, 1 and 4, were generated by crossing these two strains. Double heterozygous animals were crossed to each other to

generate double homozygous animals. Genotyping was made by PCR following manufacturer's recommendations. The veterinary committee for animal research of the Canton of Fribourg, Switzerland, approved all experiments performed on animals. In situ hybridization In situ hybridization (ISH) experiments were performed essentially as previously described (Girard et al., 2011), using 12–14 μm brain cryosections. Digoxygenine (DIG) labeled Thbs4 RNA probes were prepared as follows. The complete full-length mouse Thbs4 cDNA was purchased from Imagenes (Berlin, Germany). In vitro transcription was performed following manufacturer recommendations (Roche Diagnostics, Rotkreuz, Switzerland), using DIG-UTP and T3 or T7 RNA polymerases (respectively for the sense and antisense RNA probe) and full-length Thbs4 cDNA as template (since the Thbs4 sequence is flanked by T3 and T7 core promoter sequences). The sense probe was used as a negative control. A shorter probe was also prepared by RT-PCR using the Thbs4 primer sequences designed by the Allen Brain Atlas Project (http://www.brain-map.org). Both antisense probes gave essentially similar results, but the full-length probe was preferred because of a stronger signal. Histology, Brdu administration and immunostaining The following primary antibodies were used: polyclonal rabbit antiGFAP (diluted 1:500; Dako Schweiz AG, Baar, Switzerland); monoclonal mouse anti-PSA NCAM (diluted 1:250; Millipore AG, Zug, Switzerland);

184

F. Girard et al. / Molecular and Cellular Neuroscience 61 (2014) 176–186

Fig. 6. Quantitative analysis of cell proliferation in the SVZ. Brdu-based in-vivo assay measuring cellular proliferation in the subventricular zone, performed in young animals (P14/15) (A) and adults (8–12 weeks of age) (B). A: Brdu positive cells, counted on 10 consecutive 30 μm sections, in a 200 μm2 area of the SVZ. Counting was done on one hemisphere, n = 6 animals for Thbs4−/−, n = 4 animals for Thbs4+/+. B: Brdu positive cells in the SVZ counted on 10 sections (30 μm) (20 consecutive 30 μm sections were collected, and counting was performed every 2 sections). Counting was done on one hemisphere, n = 4 animals for both genotypes. C: The areas taken into consideration for counting are highlighted with dashed line (since proliferation occurs at a higher rate in younger animals, the area for cell counting was chosen smaller compared to older animals). In all cases, plots show the average + s.e.m (black bar: Thbs4+/+; empty bar: Thbs4−/−). ns: the difference is not statistically significant (Student's t test: p N 0.1).

monoclonal rat anti-Brdu (diluted 1:200; Abcam, Cambridge, UK); affinity-purified goat anti-human Thbs4 (diluted 1:500; R&D Systems Europe, Abingdon, UK); polyclonal goat anti-Dcx C-18 (diluted 1:200; Santa Cruz Biotechnology, Heidelberg, Gernany); rabbit antiparvalbunin PV25 (diluted 1:2500; Swant, Bellinzona, Switzerland); mouse anti-Calbindin CB300 (diluted 1:2500; Swant, Bellinzona, Switzerland). The following secondary antibodies were employed, depending on the experiment: goat anti-rabbit Alexa 488, goat antimouse Alexa 568, donkey anti-mouse Cy3 (Jackson Immunoresearch Laboratory, Rheinfelden, Switzerland), biotinylated goat anti-rat and

donkey anti-goat (Vector Laboratories, Servion, Switzerland), avidin/ HRP detection kit (Vector Laboratories). Mice at the appropriate age were anesthetized with pentobarbital (100 mg/kg body weight), trans-cardially perfused with ice-cold 0.9% NaCl, and fixed by perfusion with 4% paraformaldehyde (PFA) in PBS. Brains were then removed, post-fixed overnight at 4 °C in 4% PFA, and embedded in 20% sucrose/0.1 M TBS pH 7.3 containing 0.02% sodium azide. For Nissl staining, 30 μm thick sagittal cryosections were prepared and collected on Super Frost Plus slides. Double immunostaining for GFAP and PSA-NCAM (presented in Fig. 4), and immunostaining for

F. Girard et al. / Molecular and Cellular Neuroscience 61 (2014) 176–186

185

Fig. 7. Quantitative analysis of cell migration to the olfactory bulb. A–B: Brdu-based in-vivo assay examining cell migration to the olfactory bulb, performed in young animals (P14/15) (A) and adults (8–12 weeks of age) (B). A: Brdu positive cells in the OB were counted on 10 sections (40 consecutive 30 μm sections were collected, and counting was done every 4 sections). n = 4 animals for all genotypes analyzed. B: Brdu positive cells in the OB were counted on 20 sections (40 consecutive 30 μm sections were collected, and counting was done every 2 sections). n = 5 animals for Thbs4−/−, n = 4 animals for Thbs4+/+, n = 3 animals for Thbs1−/−Thbs4−/− and Thbs1−/−. Brdu positive cells were counted in all the main olfactory bulb cell layers, with the exception of the central subependymal zone (see schematic representation of an OB section in panel C). C–E: Quantification of Pvalb and CalbD-28K immunoreactive neurons in the opl and gl layers of the main olfactory bulb, respectively. C: Schematic representation of an OB section, highlighting the two areas (red rectangles) selected for quantification (legend: gl: glomerular layer; gr: granular layer; mi: mitral layer; ipl: inner plexiform layer; opl: outer plexiform layer; SEZ: sub-ependymal zone). D: representative example of immunostaining for Parvalbumin (green) and CalbindinD-28K (red) in an OB section from Thbs4+/+ brain. E: Pvalb and CalbD-28K neurons were counted on 10 sections (20 consecutive 30 μm sections were collected, and immunostaining was performed every 2 sections). n = 3 animals, aged P22, for both genotypes. In all cases, plots show the average + s.e.m (black bar: Thbs4+/+; empty bar: Thbs4−/−, light gray: Thbs1−/−Thbs4−/−; dark gray: Thbs1−/−). Statistical significance was determined with Student's t test: * p b 0.01; ** p b 0.001; *** p b 0.0001; ns: not statistically significant p N 0.1.

186

F. Girard et al. / Molecular and Cellular Neuroscience 61 (2014) 176–186

Thbs4 (presented in Fig. 3), were performed on 4 μm paraffin sections from Thbs4+/+ or Thbs4 KO mouse brain, following standard procedures. Dcx immunostaining (presented in Fig. 5) was performed under standard conditions, using 30 μm thick sagittal brain cryosections of both genotypes. Primary antibodies were applied overnight, or for a longer time, at 4 °C. For double ISH/immunostaining experiments (presented in Fig. 3), brain cryosections were processed first for ISH with DIG-labeled Thbs4 probe, followed by 2 h incubation with anti-DIG (Roche Diagnostics, Rotkreuz, Switzerland) and anti-GFAP. Subsequent steps included incubation with Alexa Fluor488 donkey anti-rabbit to detect GFAP, followed by incubation with NBT-BCIP (Roche Diagnostics, Rotkreuz, Switzerland) to reveal Thbs4 mRNA. For quantification of Parvalbumin and CalbindinD-28K neurons in the olfactory bulb (presented in Fig. 7), 3 animals of both genotypes (Thbs4+/+ and Thbs4−/−) were PFA-perfused at age P22. Brains were dissected and sectioned as described before. 20 consecutive sections were collected, and immunostaining was performed every 2 sections. Free-floating sections were incubated with primary antibodies overnight at 4 °C, followed by 2 h incubation with secondary antibodies under standard conditions. Different protocols of Brdu administration were used. For the analysis of cell proliferation in the SVZ of young (P14/15) and older (8 to 12 weeks) animals, mice received a single intra-peritoneal injection of Brdu (Roche Diagnostics, Rotkreuz, Switzerland) (100 mg/kg body weight) dissolved in 0.9% NaCl, 2 h before sacrifice. For analyzing cell migration to the olfactory bulb, animals received a single injection of Brdu (100 mg/kg) at P17/18, and were sacrificed 5 days later (age P22/23). For adult mice (8 to 12 weeks), the protocol was identical (i.e. sacrificed after 5 days). In all cases, mice were anesthetized and PFA-perfused, and their brains were carefully removed as described before. 30 μm thick coronal cryosections sections were prepared under standard conditions. Sections were maintained in TBS containing 0.02% sodium azide until use. Sections were collected on Super Frost Plus slides, treated for 30 min at 37 °C in 2 N HCl, rinsed in 0.1 M Borate buffer pH 8.5 followed by wash in 0.1 M TBS pH 7.3, and processed for immunohistochemistry with anti-Brdu using a standard protocol. For all histological/immunological experiments, extreme care was taken when preparing the brain sections in order to compare sections exactly at the same position and orientation. For statistical analysis, a sufficient number of animals and brain sections were taken into consideration (see captions to Figs. 6 and 7). Results are given as average + s.e.m. Statistical significance was determined with Student's t test: not significant p N 0.1; * p b 0.01; ** p b 0.001; *** p b 0.0001. Image analysis All image analyses were performed with a Nikon Eclipse E400 bright field microscope, Zeiss Axiophot fluorescence microscope, Leica TCS SP5 confocal laser microscope, and Hamamatsu Nanozoomer. Image postprocessing and contrast adjustments were performed with Adobe Photoshop and Nanozoomer slide processing software. Acknowledgements We thank the technical assistance of C. Marti and M. Sanchez for help in cryo-sectionning, and paraffin sections, respectively. This study has received financial support from the Canton of Fribourg, the Swiss National Science Foundation (Grant Nr. 3100A0-11352) and the Novartis Foundation.

References Adams, J., 2001. Thrombospondins: multifunctional regulators of cell interactions. Annu. Rev. Cell Dev. Biol. 17, 25–51. Arber, S., Caroni, P., 1995. Thrombospondin-4, an extracellular matrix protein expressed in the developing and adult nervous system, promotes neurite outgrowth. J. Cell Biol. 131, 1083–1094. Basak, O., Taylor, V., 2009. Stem cells of the adult mammalian brain and their niche. Cell. Mol. Life Sci. 66, 1057–1072. Benner, E., Luciano, D., Jo, R., Abdi, K., Paez-Gonzalez, P., Sheng, H., Warner, D., Liu, C., Eroglu, C., Kuo, C., 2013. Protective astrogenesis from the SVZ niche after injury is controlled by Notch modulator Thbs4. Nature 497, 369–373. Blake, S., Strasser, V., Andrade, N., Duit, S., Hofbauer, R., Schneider, W., Nimpf, J., 2008. Thrombospondin-1 binds to ApoER2 and VLDL receptors and functions in postnatal neuronal migration. EMBO J. 27, 3069–3080. Cayre, M., Canoll, P., Godlman, J., 2009. Cell migration in the normal and pathological mammalian brain. Prog. Neurobiol. 88, 41–63. Celio, M., 1990. Calbindin D-28K and parvalbumin in the rat nervous system. Neuroscience 35, 375–475. Christopherson, K., Ullian, E., Stokes, C., Mullowney, C., Hell, J., Agah, A., Lawler, J., Mosher, D., Bornstein, P., Barres, B., 2005. Thrombospondins are astrocyte-secreted proteins that promote CNS synaptogenesis. Cell 120, 421–433. Congote, L., Difalco, M., Gibbs, B., 2004. The C-terminal peptide of thrombospondin 4 stimulates erythroid cell proliferation. Biochem. Biophys. Res. Commun. 324, 673–678. Dityatev, A., Seidenbecher, C., Schachner, M., 2010. Compartmentalization from the outside: the extracellular matrix and functional microdomains in the brain. Trends Neurosci. 33, 503–512. Dunkle, E., Zaucke, F., Clegg, D., 2007. Thrombospondin 4 and matrix threedimensionality axon outgrowth and adhesion in the developing retina. Exp. Eye Res. 84, 707–717. Eroglu, C., Allen, N., Susman, M., O'Rourke, N., Park, C., Ozkan, E., Chakraborty, C., Mulinyawe, S., Annis, D., Huberman, A., Green, E., Lawler, J., Dolmetsch, R., Garcia, K., Smith, S., Luo, Z., Rosental, A., Mosher, D., Barres, B., 2009. Gabapentin receptor alpha 2 beta 1 is a neuronal thrombospondin receptor responsible for excitatory CNS synaptogenesis. Cell 139, 380–392. Faigle, R., Song, H., 2013. Signaling mechanisms regulating adult neural stem cells and neurogenesis. Biochim. Biophys. Acta 1830, 2435–2448. Frolova, E., Pluskota, E., Krukovets, I., Burke, T., Drumm, C., Smith, J., Blech, L., Febbraio, M., Bornstein, P., Plow, E., Stenina, D., 2010. Thrombospondin-4 regulates vascular inflammation and artherogenesis. Circ. Res. 107, 1313–1325. Girard, F., Meszar, Z., Marti, C., Davis, F., Celio, M., 2011. Gene expression analysis in the parvalbumin immunoreactive PV1 nucleus of the mouse lateral hypothalamus. Eur. J. Neurosci. 34, 1934–1943. Ihrie, R., Alvarez-Buylla, A., 2011. Lake-front property: a unique germinal niche by the lateral ventricles of the adult brain. Neuron 70, 674–686. Kazanis, I., ffrench-Constant, C., 2011. Extracellular matrix and the neural stem cell niche. Dev. Neurobiol. 71, 1006–1017. Kim, D., Li, K., Boroujerdi, A., Peter, Y., Zhou, C., Deng, P., Park, J., Zhang, X., Lee, J., Corpe, M., Sharp, K., Steward, O., Eroglu, C., Barres, B., Zaucke, F., Xu, Z., Luo, Z., 2012. Thrombospondin-4 contributes to spinal sensitization and neuropathic pain states. J. Neurosci. 32, 8977–8987. Kriegstein, A., Alvarez-Buylla, A., 2009. The glial nature of embryonic and adult neural stem cells. Annu. Rev. Neurosci. 32, 149–184. Lu, Z., Kipnis, J., 2010. Thrombospondin-1, a key astrocyte-derived neurogenic factor. FASEB J. 24, 1925–1934. Ming, G., Song, H., 2011. Adult neurogenesis in the mammalian brain: significant questions and significant answers. Neuron 70, 687–702. Mosher, D., Adams, J., 2012. Adhesion-modulating/matricellular ECM protein families: structural, functional and evolutionary appraisal. Matrix Biol. 31, 155–161. Pluskota, E., Stenina, O., Krukovets, I., Szpak, D., Topol, E., Plow, E., 2005. Mechanism and effect of thrombospondin-4 polymorphisms on neutrophil function. Blood 106, 3970–3978. Risher, C., Eroglu, C., 2012. Thrombospondins as key regulators of synaptogenesis in the central nervous system. Matrix Biol. 31, 170–177. Stenina, D., Desai, S., Krukovets, I., Kight, K., Janigro, D., Topol, E., Plow, E., 2003. Thrombospondin 4 and its variants: expression and differential effects on endothelial cells. Circulation 108, 1514–1519. Whitman, M., Greer, C., 2009. Adult neurogenesis and the olfactory system. Prog. Neurobiol. 89, 162–175. Xu, J., Xiao, N., Xia, J., 2010. Thrombospondin 1 accelerates synaptogenesis in hippocampal neurons through neuriglin 1. Nat. Neurosci. 13, 22–24. Zhao, C., Deng, W., Gage, F., 2008. Mechanisms and functional implications of adult neurogenesis. Cell 132, 645–660.