brain research 1632 (2016) 141–155

Available online at www.sciencedirect.com

www.elsevier.com/locate/brainres

Research Report Mice null for NEDD9 (HEF1) display extensive hippocampal dendritic spine loss and cognitive impairment

n D.C. Knutsona, A.M. Mitzeya, L.E. Taltonb, M. Clagett-Damea,c, aDepartment of Biochemistry, University of Wisconsin–Madison, 433 Babcock Drive, Madison, WI 53706, USA bBehavioral Testing Core Facility, University of California, Los Angeles, CA 90095, USA cPharmaceutical Sciences Division, School of Pharmacy, University of Wisconsin–Madison, 777 Highland Avenue, Madison, WI 53705, USA article info abstract

Article history: NEDD9 (neural precursor cell expressed, developmentally down-regulated 9) is a member Accepted 1 December 2015 of the CAS (Crk-associated substrate) family of scaffolding that regulate cell Available online 10 December 2015 adhesion and migration. A Nedd9 knock-out/lacZ knock-in mouse (Nedd9 / ) was devel- oped in order to study Nedd9 expression and function in the nervous system. Herein we Keywords: show that NEDD9 is expressed in the adult brain and is prominently expressed in the NEDD9 hippocampus. Behavioral testing uncovered functional deficits in Nedd9 / mice. In the Retinoic acid-induced Morris water maze test, Nedd9 / mice showed deficits in both the ability to learn the task differentiation as well as in their ability to recall the platform location. There was no change in the gross Hippocampus morphology of the hippocampus, and stereological analysis of BrdU-labeled newly formed Dendritic spine density hippocampal cells suggested that this defect is not secondary to altered neurogenesis. CAS However, analysis of the hippocampus revealed extensive loss of dendritic spine density in Null mutant both the dentate gyrus (DG) and CA1 regions. Spine loss occurred equally across all branch orders and regions of the dendrite. Analysis of spine density in Nedd9 / mice at 1.5, 6 and 10 months revealed an age-dependent spine loss. This work shows that NEDD9 is required for the maintenance of dendritic spines in the hippocampus, and suggests it could play a role in learning and memory. & 2015 Elsevier B.V. All rights reserved.

Abbreviations: ABL, Abelson murine leukemia; ANOVA, analysis of variance; ATN, anterior group of the dorsal thalamus; AAALAC, Association for the Assessment and Accreditation of Laboratory Animal Care International; BLA, basolateral amygdalar nucleus; BST, bed nuclei of the stria terminalis; BCAR1, breast cancer anti- resistance 1; BrdU, bromodeoxyuridine; CASS4, CAS scaffolding family member 4; CP, caudoputamen; CBN, cerebellar nuclei; CN, cochlear nuclei; C.E., coefficient of error; CA1, cornu ammonis 1; CA1sp, cornu ammonis 1 pyramidal area; CA3, cornu ammonis 3; CA3sp, cornu ammonis 3 pyramidal area; CA1sr, cornu ammonis 1 stratum radiatum; COA, cortical amygdalar area; Cre, Cre recombinase; CAS, Crk- associated substrate; DG, dentate gyrus; DGsg, dentate gyrus granule cell layer; DR, dorsal nucleus raphe; EFS/Sin, embryonal Fyn substrate/Src-interacting; ES, embryonic stem; EP, endopiriform nucleus; VII, facial motor nucleus; FAK, kinase; þ GCL, granule cell layer; HET, Nedd9 / , heterozygous; H, hilus; HEF1, human enhancer of filamentation 1; HY, hypothalamus; isl, islands of Calleja; CTX, isocortex; KO, knock-out, Nedd9 / ; LSN, lateral septal nucleus; LOAD, late-onset Alzheimer’s disease; http://dx.doi.org/10.1016/j.brainres.2015.12.005 0006-8993/& 2015 Elsevier B.V. All rights reserved. 142 brain research 1632 (2016) 141–155

treatment with a Rho kinase inhibitor (Bargon et al., 2005). 1. Introduction Knockdown of Nedd9 reduces neural crest cell migration in the chick embryo (Aquino et al., 2009). More recently, single NEDD9 (HEF1/CAS-L/CASS2) is a member of the CAS family of nucleotide polymorphisms (SNPs) in Nedd9 have been linked multidomain docking or scaffolding proteins that includes to pathological conditions including late-onset Alzheimer’s CAS BCAR1/p130 , EFS/SIN and CASS4/HEPL (Guerrero et al., (LOAD) and Parkinson’s disease (Beck et al., 2014; Fu et al., 2012; Nikonova et al., 2014; Tikhmyanova et al., 2010). Nedd9 2012; Li et al., 2008; Tedde et al., 2010; Xing et al., 2011). fi fi was rst identi ed as a developmentally down-regulated However, where Nedd9 is found in the adult brain and in mouse brain (Kumar et al., 1992), and was later whether it plays a role in normal nervous system function fi independently identi ed as a human protein that enhanced is not well understood. pseudohyphal budding in yeast (HEF1; Law et al., 1996) and as In this report, we generate a mouse null for Nedd9 that also a protein expressed in lymphocytes (CAS-L; Minegishi et al., contains a knock-in of lacZ in order to examine the expres- fi fi 1996) with similarity to the rst identi ed CAS family mem- sion and function of NEDD9 in the adult nervous system. CAS ber, p130 (Reynolds et al., 1989; Sakai et al., 1994). EFS/Sin NEDD9/LacZ is highly expressed in the adult hippocampus, was initially described as a Fyn SH3-domain-binding protein and Nedd9 / null mutant mice show a deficit in learning and (Alexandropoulos and Baltimore, 1996; Ishino et al., 1995), memory along with a dramatic loss of dendritic spines, and the most recently described CAS family member, CASS4, providing evidence of a new function for NEDD9 in nervous has many similar activities to NEDD9 in vitro (Singh et al., system maintenance. 2008). CAS family members act downstream of cell surface receptors and cytoplasmic protein tyrosine kinases (FAK, SFK, and ABL) and function as platforms for the assembly of 2. Results multiprotein complexes to regulate cytoskeletal dynamics. Early studies described NEDD9 as a signaling intermediate in 2.1. Generation of the Nedd9 knock-out (KO)/lacZ knock-in pathways affecting cellular attachment to the extracellular (Nedd9 / ) mouse matrix and cellular motility in lymphoid and epithelial cells (Fashena et al., 2002; Law et al., 1998; Ohashi et al., 1999; van To generate Nedd9 null mutant mice (Nedd9 / ), the Nedd9 Seventer et al., 2001). Our lab identified Nedd9 in a screen for genomic region containing 3 of the NEDD9 protein was all-trans retinoic acid responsive in a neuroblastoma replaced with lacZ and a Neo resistance cassette by homo- cell line that extends neurites and shows changes in adhesive logous recombination in embryonic stem (ES) cells. The Nedd9 properties in response to retinoic acid (Merrill et al., 2004a, gene has two promoters and two translational start sites 2004b). (found in 2A and 2B) that produce two nearly identical Although much of what is currently known about Nedd9 isoforms of the protein differing only in the first three amino comes from studies of its role in cancer cell metastasis and acids (Sasaki et al., 2005). LacZ was introduced into exon 3, the lymphocyte migration, evidence is emerging that it also first exon common to both transcripts, putting lacZ in frame functions in the nervous system. Nedd9 mRNA is expressed with either of the two Nedd9 translational start sites to in mouse brain, and is also found in the developing nervous generate a fusion protein composed of the first 5 amino acids system of the rat embryo (Kumar et al., 1992; Merrill et al., of the NEDD9 protein and 1 foreign amino acid followed by 2004b). Mouse embryos transgenic for a 5.2 kb region of the lacZ sequence (Fig. 1A). This disrupted the protein coding Nedd9 promoter express high levels of Nedd9 in the caudal sequence of Nedd9 and inserted a stop codon and a poly- hindbrain neuroepithelium, spinal cord, dorsal root ganglia adenylation signal into exon 3 of the gene in order to and migrating neural crest, and expression is dependent terminate transcription prior to reaching the remaining Nedd9 upon an intact retinoic acid response element (Knutson and coding exons (exons 4–8) and allows for lacZ expression Clagett-Dame, 2008, 2015). Nedd9 is expressed in neural driven by the Nedd9 promotor. Correct integration of the progenitor and progenitor-like cells. It is induced in neuro- targeting vector was identified by PCR and Southern blot blastoma cells undergoing differentiation by retinoic acid analysis (Fig. 1B). The mutant loci were confirmed by PCR (Merrill et al., 2004a, 2004b), and neurite length and number screening of tail-derived genomic DNA (Fig. 1C), and two is enhanced in PC12 cells overexpressing Nedd9 (Sasaki et al., females and one male heterozygous (HET) mouse from the 2005). The formation of neurites in a non-neuronal cell type same ES line were generated. Western blot analysis showed a has also been reported with Nedd9 overexpression and complete loss of NEDD9 protein in the null mutant (Fig. 1D).

(footnote continued) MB, midbrain; ML, molecular layer; V, motor nucleus of the trigeminal; NEDD9, neural precursor cell expressed, developmentally down-regulated 9; Neo, neomycin; ACB, nucleus accumbens; PB, parabrachial nucleus; PFA, paraformaldehyde; PGK, phosphoglycerine kinase; PG, pontine gray; PCL, Purkinje cell layer region; SGZ, subgranular zone; PSD, postsynaptic density; RHP, retrohippocampal region; RSP, retrosplenial area; SNPs, single nucleotide polymorphisms; SPV, spinal nucleus of the trigeminal; SFK, ; SOC, superior olivary complex; TRN, tegmental reticular nucleus; TH, thalamus; þ þ VNC, vestibular nuclei; WT, wild-type, Nedd9 / n Corresponding author at: Department of Biochemistry, University of Wisconsin–Madison, 433 Babcock Drive, Madison, WI 53706-1544, USA. Fax: þ1 608 262 7122. E-mail address: [email protected] (M. Clagett-Dame). brain research 1632 (2016) 141–155 143

2.2. Deletion of Nedd9 does not affect growth and 2.3. Nedd9 is highly expressed in the adult hippocampus reproduction β-galactosidase (LacZ) staining was examined in thick coronal Nedd9 KO/lacZ knock-in animals grew normally, were repro- (Fig. 2A–F, right half) and sagittal (Supplementary Fig. 1A–E) ductively active, survived up to 2.5 years of age, and were sections of Nedd9 KO/lacZ knock-ins to determine the location of indistinguishable from wild-type littermates in agreement NEDD9 expression in the brain. β-galactosidase staining was / with the results of Seo et al. (2005) in CAS-L-deficient mice. At observed in several regions of the adult brain in Nedd9 necropsy, there was no evidence of any abnormality in the mutants with particularly pronounced expression in multiple orientation/placement of major organs in Nedd9 / mice layers of the cerebral cortex and hippocampal formation. NEDD9/LacZ staining was observed in the hippocampus proper, (data not shown). with most prominent expression in the pyramidal area of the 144 brain research 1632 (2016) 141–155

CA1 and CA3 fields (CA1sp and CA3sp, respectively), and in the were examined as a control, and no LacZ staining was observed granule cell layer of the dentate gyrus (DGsg) (Fig. 2CandD; in these sections (Fig. 2A–F, left half). Supplementary Fig. 1A, B and D). The retrohippocampus (RHP), and the retrosplenial cortex (RSP) were also stained (Fig. 2Cand 2.4. Nedd9 / mice show deficits in a spatial learning D; Supplementary Fig. 1 A and D), and expression was observed task throughout multiple layers of the isocortex (CTX or neocortex) (Fig. 2A–E; Supplementary Fig. 1A and D). Two additional sites of The expression of NEDD9 in the brain prompted us to test expression included the anterior group of the dorsal thalamus whether deletion of Nedd9 led to any behavioral alterations. fi (ATN; Supplementary Fig. 1A)andthelateralseptalnucleus Mice rst underwent a primary SHIRPA screen designed to (LSN; Fig. 2B). Other regions of lesser staining included the test muscle, cerebellar, sensory and neuropsychiatric func- / endopiriform nucleus (EP; Fig. 2A–D), bed nuclei of the stria tion (Irwin, 1968; Rogers et al., 1997). The Nedd9 mutants fi þ/þ terminalis (BST, Supplementary Fig. 1A), cortical amygdalar area did not differ signi cantly from the Nedd9 animals in any (COA; Fig. 2B–D), nucleus accumbens (ACB; Fig. 2A; Supplemen- of the 42 SHIRPA tasks and the performance of both groups was within the range considered normal for the C57BL/6 tary Fig. 1A), islands of Calleja of the olfactory tubercle (isl; strain (data not shown). Thus, KO of the Nedd9 gene had no Fig. 2A) the basolateral amygdalar nucleus (BLA; Fig. 2D) and the gross effect on sensory modalities of vision, audition, whisker caudoputamen (CP; Fig. 2B; Supplementary Fig. 1A and D). There or vestibular function or other attributes such as aggression, was also light or diffuse staining in the hypothalamus (HY; gait, heart rate, body or limb tone, muscle strength, respira- Fig. 2D; Supplementary Fig. 1A) and midbrain (MB; Supplemen- tion, or pain sensitivity. tary Fig. 1A), including the dorsal nucleus raphe (DR, Fig. 2E). In Because the NEDD9–LacZ fusion protein was highly the hindbrain, staining in the pons occurred in the parabrachial expressed in the hippocampus, mice were trained and tested nucleus (PB; Supplementary Fig. 1C), tegmental reticular nucleus in the hidden platform version of the Morris water maze, a (TRN; Fig. 2E), pontine gray (PG; Supplementary Fig. 1AandC), test in which the hippocampus is known to support spatial superior olivary complex (SOC; Supplementary Fig. 1AandC), processing demands. In this test, the animal learns the and the motor nucleus of the trigeminal (V; Supplementary location of the escape platform in relation to the cues in Fig. 1C). Medullary expression included the vestibular nuclei the periphery of the room. The training (T) and probe (VNC; Supplementary Fig. 1A), facial motor nucleus (VII; Supple- (P) testing schedule is depicted in Fig. 3A. In the training mentary Fig. 1A and C), lateral reticular nucleus (LRN; Supple- portion, wild-type (WT) mice showed more rapid learning of mentary Fig. 1A and C), cochlear nuclei (CN; Supplementary the task compared to mutants, with a significantly shorter Fig. 1D), and the spinal nucleus of the trigeminal (SPV; Fig. 2F; latency to find the platform in bin 2. The latency to find the Supplementary Fig. 1D). In the cerebellum, expression was seen platform improved with training in WT, and this group in the cerebellar nuclei (CBN; Fig. 2F; Supplementary Fig. 1Aand showed significantly shorter latencies in bins 6 and 7 of the D), as well as between the granule and molecular cell layers of test compared to Nedd9 / mice (Fig. 3B). the folia, a region in which the Purkinje cells are located (PCL; After training sessions 10, 16, 22, and 28, spatial learning was þ þ Fig. 2F; Supplementary Fig. 1AandE).BrainsfromNedd9 / mice assessed using probe trials (P1, P2, P3, and P4) in which the

Fig. 1 – Generation of Nedd9 KO/lacZ knock-in mice. Schematic of the Nedd9 genomic (A) and the location of the 50 (pink) and 30 (purple) homology arms in the targeting vector. The targeted KO/knock-in (lacZ, light blue) allele before and after Cre recombination is also depicted. Dotted lines indicate the intronic regions that flank exon 3, these comprise the homology arms used in the targeting vector. The approximate size and locations of exons 3–6 are depicted by black rectangles labeled beneath with the exon number. Restriction sites are listed. The 50 (50, red box), 30 (30, green box) and Neo (Neo, dark blue) probes are indicated, as well as location of genotyping PCR primers (P1–P3, gray arrows). Expected sizes of restriction fragments detected by probes used are shown below the appropriate sequence (red, green and dark blue arrow-headed lines). LoxP sites flank the Neo selection gene (PGK Neo) allowing for later deletion by mating with Cre expressing mice. Analysis of target gene integration by Southern blotting, PCR and Western blotting to verify loss of both Nedd9 mRNA and protein in the null mutant mice (B–D). Southern blots (B) show the expected integration of the targeting construct. DNA was isolated from expanded targeted ES cells and from WT cells (positive probe control was included but is not shown). Top panel: DNA digested with ScaI and hybridized with the 50-probe shows expected WT band (27.7kb) and targeted KO/knock-in allele (KO band, 15.3 kb) bands in HET ES cells. Bottom panel: DNA digested with XbaI and hybridized with the 30-probe shows expected WT band (10.0 kb) and targeted KO/knock-in allele (KO band, 8.2 kb) bands in HET ES cells. PCR analysis of tail snip DNA þ þ þ (C) from Nedd9 / (WT), Nedd9 / (HET), and Nedd9 / (KO) adult male mice. Primers P1, P2, and P3 (shown in A) were combined. WT band: 753 bp and KO/knock-in band (KO band): 334 bp. Western blot analysis (D) indicates the complete loss of þ þ endogenous NEDD9 protein expression in Nedd9 / animals. Protein-containing lysates prepared from Nedd9 / (WT), þ Nedd9 / (HET), and Nedd9 / (KO) adult lungs were resolved on a SDS polyacrylamide gel. Lung was chosen for analysis as it was enriched in the NEDD9 protein enabling definitive detection on the Western blot. The blot was cut in half and the upper half was probed with a rabbit anti-NEDD9 antibody that detected the expected 105 kDa (phosphorylated) and 115 kDa þ þ þ (hyperphosphorylated) bands (Law et al., 1998) in the Nedd9 / and Nedd9 / but not the Nedd9 / sample. As a loading control, the bottom half was probed with mouse anti-β-actin clone AC-15 antibody that detected the expected 42 kDa band in all samples. brain research 1632 (2016) 141–155 145

Fig. 2 – NEDD9 is highly expressed in the hippocampus of the adult mouse brain. NEDD9–LacZ fusion protein expression was mapped by β-galactosidase staining of coronal brain slices. Images are arranged from rostral (anterior) to caudal (posterior), with the cartoon at top (modified from Sunkin et al., (2013)) showing the relative position of each section. For each section (A– þ þ F), the left hemisphere from a Nedd9 / mouse is shown as a control and the right hemisphere is a representative Nedd9 / section from a mouse with lacZ knocked into both alleles. Sections were counterstained with nuclear fast red. Labeled areas indicate regions positive for β-galactosidase stain (blue). Abbreviations: BLA, basolateral amygdalar nucleus; CA1, cornu ammonis 1; CA3, cornu ammonis 3; CP, caudoputamen; CBN, cerebellar nuclei; CN, cochlear nucleus; COA, cortical amygdalar area; DG, dentate gyrus; DR, dorsal nucleus raphe; EP, endopiriform nucleus; VII, facial motor nucleus; HPF, hippocampal formation; HY, hypothalamus; isl, islands of Calleja; CTX, isocortex; LSN, lateral septal nucleus; ACB, nucleus accumbens; PCL, Purkinje cell layer region; RHP, retrohippocampal region; RSP, retrosplenial area; TRN, tegmental reticular nucleus; SPV, spinal nucleus of the trigeminal; and VNC, vestibular nuclei. Scale bar: 500 μm. 146 brain research 1632 (2016) 141–155

Fig. 3 – Nedd9 / mice show deficits in the Morris water maze, a test of spatial learning and memory. Morris water maze training and probe testing scheme (A). Animals were trained over a 14-day period; each of the 14 gray arrows indicates a single day. Animals received two 2 independent training sessions/day (numbered T1–T28) to find a hidden platform in relation to cues in the periphery of the room, and the latency to escape was recorded. Data is pooled in 2-day bins (numbered at top 1–7). Mice were placed on the platform for 30 s (X) before the first training trial. Following training sessions 10, 16, 22 and 28 a probe trial (P1–P4) was performed. Following the final probe trial, a visual platform trial (V) was repeated 5 times. Escape latency (time to find hidden platform) from a variable start position was significantly longer in Nedd9 / mice in bins 2, 6 and 7 (B). In the 4 probe tests, Nedd9 / mutants showed significant deficits in recalling the platform location as measured by the time spent in the target quadrant (C) and proximity to the original platform location (average distance from the platform þ þ across the trial) (D). Significant differences between Nedd9 / and Nedd9 / mice are marked above the pair of data points, *po0.05; **po0.005. brain research 1632 (2016) 141–155 147

þ þ Fig. 4 – The morphology and size of the hippocampus in the Nedd9 / mutants is similar to that of the Nedd9 / mice. þ þ Representative images of H&E stained coronal brain sections from Nedd9 / (A) and Nedd9 / (B) mice. Lineal measurements are depicted in images by the yellow line (C–E), as is the manner in which circumference of the hippocampus was determined (F). The group means7SEM are shown in the bar graphs below; no significant group differences were observed for any þ þ measurement (C–F). Representative images from DAPI stained coronal brain sections from Nedd9 / (G) and Nedd9 / (H) mice. There was no significant difference between genotypes in the total number of cells (I) in a uniform area along the length of the lower blade of the DG; graph shows the group means7SEM. Abbreviations: CA1, cornu ammonis 1; CA3, cornu ammonis 3; DG, dentate gyrus; SGZ, subgranular zone; and GCL, granule cell layer. Scale bar: A and B, 100 μm; G and H, 10 μm.

platform was removed from the pool, and the mice were differences in these measures were observed (data not shown). allowed to search for 60 s. By the second and third trials, Thus, deletion of Nedd9 ledtoimpairedperformanceinthe þ þ Nedd9 / mice spent significantly more time in the target Morris water maze test. quadrant than did littermate Nedd9 / mice (Fig. 3C). The Nedd9 null mutants did not improve until the fourth probe 2.5. Gross hippocampal morphology is unaltered in adult trial, and never improved above chance (25%). In a measure of Nedd9 / mutant mice proximity to the platform location (average distance from the þ þ platform across the trial), Nedd9 / mice were also significantly We next evaluated whether the overall morphology or size of closer to the platform in probe trials 2 and 3 (Fig. 3D). To control the hippocampus was altered in Nedd9 / mice at 3–4 month for the possibility that either a motor deficit or decreased (mo) or 12–14 mo of age. Gross hippocampal organization þ þ motivation to escape could account for the differences in appeared similar for the Nedd9 / and Nedd9 / groups performance, the average latency to find a raised visible plat- (Fig. 4A vs B). No difference in the distance between the end form as well as the average swim speed and total distance of CA1 and the end of the lower blade of the DG, the ends of traveled were measured in probe trials. No significant the upper and lower blade of the DG, the end of CA1 and the 148 brain research 1632 (2016) 141–155

tip of crest of the DG, or the overall circumference of the 2.6. Cell proliferation and survival of newly born cells in þ þ hippocampus was noted between Nedd9 / and Nedd9 / the DG is not affected in adult Nedd9 / mice mice at either 3–4mo(Fig. 4C–F) or 12–14 mo (data not shown) of age. In order to determine whether the number of cells in The subgranular zone (SGZ) of the DG is one of two brain the DG was altered in the null mutants, cells within a defined regions in which neurogenesis occurs in adults, and hippo- region were counted in confocal stacks. No significant differ- campal neurogenesis in the adult is thought to play a role in ence in cell number between genotypes was observed learning and memory (Deng et al., 2010). It has also been (Fig. 4G–I). Thus the overall gross structure and size of the suggested that Nedd9 plays a role in rendering cells compe- hippocampus was unchanged in the Nedd9 / mice. tent to differentiate into neurons (Vogel et al., 2010). For these reasons, BrdUþ cells were evaluated by stereological quanti- tation to determine whether the number of newly born cells in the DG of adult mice or their survival was affected by Nedd9 gene deletion. BrdUþ cells were counted in three regions of the DG; the SGZ, H, and dentate ML at two time points. A group of animals was evaluated 24 h following BrdU admin- istration to evaluate cell proliferation, and a second group of animals was evaluated 3 weeks (wk) following BrdU admin- istration to measure differences in survival of newly born cells (Fig. 5A). Cell proliferation at 24 h was not significantly different between genotypes (Fig. 5D) nor was there a sig- nificant difference in the survival of newly born cells at 3 wk (Fig. 5E) for any of the hippocampal regions counted. Although we did not phenotype the BrdUþ cells, the majority of newborn cells in the SGZ differentiate into neurons (Abrous et al., 2005) and thus the present results suggest that the impaired spatial learning in Nedd9 / mice cannot be explained by altered neurogenesis.

2.7. Nedd9 / mice show a significant reduction in hippocampal dendritic spine density

A change in the number and morphology of dendritic spines is associated with long-term activity dependent synaptic plasticity and may also be important in cognition (Levy et al., 2014). When spine density for entire dendritic trees was assessed by live microscopy in individual hippocampal granule cells of the DG at 6-mo, a dramatic reduction in overall spine density was observed in the Nedd9 / group

Fig. 5 – There is no difference in cell proliferation or survival þ þ of newly born cells in the DG of the Nedd9 / versus the Nedd9 / hippocampus. Experimental timeline (A): at 90 days of age all mice were given a single BrdU injection. Brains from half the mice were collected 24 h post-injection to analyze cell proliferation and half were collected 3 wk post-injection to analyze cell survival. A representative anti- BrdU-stained image 24 h (B) and 3 wk (C) after the injection with BrdU. The majority of BrdUþ cells are observed in the SGZ; arrows indicate example cells or aggregates of BrdUþ cells. Stereological counting of BrdUþ cells at 24 h (D) and 3 wk (E) in the “total” hippocampus (comprised of the sum of individual cell counts from the SGZ, H and ML of the DG) and in the individual subfields. No significant differences were þ þ observed between Nedd9 / and Nedd9 / mice in the number of BrdUþ cells in any subfield at either time point. Scale bar: B and C, 100 μm. Abbreviations: SGZ, subgranular zone; GCL, granule cell layer; H, hilus; and ML, molecular layer. brain research 1632 (2016) 141–155 149

Fig. 6 – Spine density is reduced in Nedd9 / mice at 6 mo of age. Representative images (A) of Golgi-impregnated DG cells. þ þ There was a significant reduction in overall spine density (B) in the Nedd9 / DG as compared to Nedd9 / but no difference in the total dendritic length (C). The reduction in spine density in the Nedd9 / group was significant across all branch orders (D) as well as generally uniform and significant across the entire length of the dendrite (F) in Sholl analysis. Cartoon of a Sholl analysis (E) with concentric circles at 30 μm steps (modified from Magarinos et al., 2006). Sholl analysis showed no significant difference in the frequency of intersections (G) assessed at 30 lm intervals radiating from the soma. Significant differences þ þ between the Nedd9 / and Nedd9 / groups are shown by asterisks; **po0.005; ***po0.0001. Scale bar: 5 μm.

(Fig. 6A and B), whereas there was no difference in the total 2.8. Nedd9 / mice show an age-dependent reduction in dendritic length (Fig. 6C). The reduction in spine density was dendritic spines not specific to any particular branch order and was relatively uniform across branches of different orders (Fig. 6D). Sholl During development and adulthood, changes in dendritic analysis (Sholl, 1956), a method of placing regularly spaced spine number and morphology accompany synapse forma- concentric circles centered on the soma, is depicted in Fig. 6E. tion, maintenance and elimination. During early postnatal life spines are very dynamic while during adulthood spines Sholl analysis indicated that similar to branch order, the are generally stable for months or years (Koleske, 2013). Spine reduction in spine density was relatively uniformly across the / density in the DG of Nedd9 mice and their WT counter- entire dendritic tree of the Nedd9 / group (Fig. 6F). The parts was evaluated at 1.5, 6, and 10 mo of age; times ranging numbers of dendritic intersections crossing each progres- from early to late adulthood. Representative images of den- sively distal circle were also counted, and no difference was þ þ drites and spines in the DG of a Nedd9 / and Nedd9 / þ/þ observed in dendritic complexity between the Nedd9 and mouse are shown in Fig. 7A. Tracing and spine counts done / Nedd9 mice (Fig. 6G). A similar reduction in dendritic spine by live microscopy for the entire dendritic tree show that density was obtained in both the basal and apical CA1 regions spine density remained relatively constant over time in þ þ of the hippocampus, and there was also a reduction in Nedd9 / mice. In contrast, spine density in the KO was total apical but not basal dendritic length in this region reduced by 25%, 53% and 66% from that of WT at 1.5, 6 and (Supplementary Fig. 2). 10 mo, (Fig. 7B). The reduction in density in the mutants was 150 brain research 1632 (2016) 141–155

We and others have shown that mice mutant for Nedd9 are viable and fertile (Seo et al., 2005). Although some tissues such as the developing retinal ganglion cell layer exhibit functional redundancy in the action of CAS family members (Riccomagno et al., 2014), the requirement for NEDD9 in the maintenance of dendritic spines on hippocampal neurons in the DG and CA1 regions cannot be compensated by other CAS family members. Therefore, in addition to the previously described physiologic role for NEDD9 in immune function (Seo et al., 2005), our work shows a unique requirement for NEDD9 in the maintenance of synaptic connectivity in the adult hippocampus. Dendritic spines protrude from the shaft or branch of the dendrite at excitatory synapses, and serve as post-synaptic terminals that receive and process synaptic information (von Bohlen Und Halbach, 2009). During late adolescence and early adulthood spine-based synaptic connections are fine-tuned, and the dynamic turnover of spines is reduced (Holtmaat et al., 2005; Koleske, 2013; Yang et al., 2009; Zuo et al., 2005). In the present work, spine density in the Nedd9 / group was below that of the WT at all times. As expected the density of þ þ spines in Nedd9 / mice did not change from 1.5 to 10 months of age, whereas spine density decreased progres- sively with age in the mutants, suggesting that Nedd9 is needed for spine maintenance. Because spines are believed to Fig. 7 – The reduction in hippocampal spine density in underlie memory acquisition and memory storage (Bourne Nedd9 / mice becomes more pronounced with age. and Harris, 2007; Yang et al., 2009) it is intriguing that the / Representative images of a uniform length of DG cell performance of Nedd9 mice in a hippocampus-based dendrite (A) showing dendritic spines at various ages. learning and memory task is also impaired. Analysis of spine density (B) revealed a significant reduction Dendritic spines consist of a head that contains the in overall spine density in the KO DG at 1.5, 6, and 10 mo of postsynaptic density (PSD), connected by a thin neck to the age compared to WT. Significant differences between the dendritic shaft. Structural support for the dendritic spine is þ þ Nedd9 / and Nedd9 / groups are shown by asterisks; provided by an actin , which also organizes *po0.05; ***po0.0001. There was no difference in spine signaling machinery at the PSD (Koleske, 2013; Korobova fi density across time for WT mice. In contrast spine density and Svitkina, 2010; Levy et al., 2014). The lamentous actin decreased from 1.5 to 6 mo, and was further decreased at in dendritic spines is highly dynamic (Star et al., 2002), and 10 mo for the Nedd9 / mice (B). Significant differences with thus, the long-term stability of dendritic spine structure in time are indicated by similar letters: (a) po0.005; (b) po0.05; the nervous system depends upon the actin control machin- and (c) po0.0001. Scale bar: 3 μm. ery. In neurons, the maintenance of dendritic spines involves cell surface receptor signaling to cytoplasmic non-receptor tyrosine kinase families resulting in modifications in phos- relatively uniform across all branch orders and across all phorylation of downstream scaffolding proteins, changes in regions of the dendrite for all ages examined, whereas there Rho-family kinase activity and ultimately stabilization of the þ þ was no difference between the Nedd9 / and Nedd9 / groups spinoskeleton core. In non-neuronal cells, NEDD9 is known to in the total dendritic length at any age (data not shown). provide a scaffolding function linking cytoplasmic non- Thus, Nedd9 null mutant mice showed a reduction in den- receptor tyrosine kinase signaling to the downstream mod- dritic spines in the DG compared to WT and this difference ulation of actin dynamics. Many of the signaling molecules became more pronounced with age. involved in NEDD9-mediated changes in cell adhesion and migration (Guerrero et al., 2012; Nikonova et al., 2014; Tikhmyanova et al., 2010) are also important in dendritic 3. Discussion spine maintenance (Koleske, 2013). Thus it is possible that NEDD9 plays a scaffolding function to regulate actin In the present study, NEDD9 is identified as a new factor dynamics within the spinoskeleton core. However, we cannot involved in dendritic spine maintenance. Using a Nedd9 rule out that it could also affect input to the PSD to alter spine KO/lacZ knock-in mouse we show that NEDD9 is expressed stability. in the adult hippocampus. Deletion of Nedd9 results in a The loss of dendritic spines is a contributing factor to the reduction of dendritic spines at 1.5, 6 and 10 mo, of age. In pathology of a number of psychiatric illnesses as well as addition, adult Nedd9 null mutant mice show a deficit in the neurodegenerative diseases such as Alzheimer’s disease Morris water maze, a hippocampal-dependent learning and (Kulkarni and Firestein, 2012; Levy et al., 2014; Lin and memory task. Koleske, 2010; Penzes et al., 2011; Pozueta et al., 2013). Spine brain research 1632 (2016) 141–155 151

loss leads to changes in synaptic connectivity that are Southern blot hybridization using multiple probes (Fig. 1A). The thought to contribute to the impaired perception, cognition, 50-probe (50, 457 bp) was generated with primers 50-CCC TGG memory, mood and decision-making that accompany these CAG TGG AAG TAA TAA GC-30 and 50-CCT GGT CAT TCA GTC disorders (Koleske, 2013). The dramatic reduction in dendritic CCG ACA-30 was used for hybridization against ScaI-digested spines in Nedd9 / mice is intriguing in light of recent studies genomic DNA. The 30-probe (30, 478 bp) was generated with that associate SNPs in two of the CAS family members, primers 50-ATG TTG CTG CGG AAG GCT CC-30 and 50-TTA GAG 0 NEDD9 and CASS4, as well as in a CAS interacting partner, GGA ATC CAG TTT CAG TGG-3 was used for hybridization PTK2B, with susceptibility to LOAD (Beck et al., 2014; Fu et al., against XbaI-digested genomic DNA. A Neo probe (632 bp) was 0 0 2012; Li et al., 2008; Tedde et al., 2010; Xing et al., 2011), a generated with primers 5 -GAC TGG GCA CAA CAG ACA ATC-3 0 0 disease in which synaptic signaling is disrupted due to a and 5 -CCA AGC TCT TCA GCA ATA TCA C-3 and hybridized massive loss of synapses leading to progressive loss of against XbaI-digested genomic DNA to check for randomly cognition (DeKosky and Scheff, 1990; Penzes et al., 2011; integrated targeting vector. Two correctly targeted clones were Terry et al., 1991). identified and injected into Balb/C blastocysts. Four chimeras, In summary, we show that the scaffolding protein NEDD9 two from each targeted ES line, survived to reproductive age is essential for dendritic spine maintenance in the hippo- and were mated to C57BL/6 Cre deleter mice (to remove the campus. A large reduction in spine density occurs throughout PGK-Neo region), and the offspring were screened for germline adulthood in the Nedd9 / mouse, whereas dendritic branch transmission by coat color followed by Southern blot hybridiza- 0 0 length and overall branch structure is largely unaffected. tion (5 and 3 probes). The presence or absence of a Neo cassette NEDD9 null mutant mice also have deficits in spatial learning was also determined by Southern blot (Neo probe). þ þ þ and memory. Analyses of these genetically modified mice WT (Nedd9 / ), HET (Nedd9 / ), and KO (Nedd9 / ) mice support the idea that NEDD9 participates in the maintenance were distinguished by PCR genotyping (Fig. 1C) of genomic of neuronal synaptic connectivity, and suggest NEDD9 loss DNA obtained by tail biopsy (DirectPCR Lysis Reagent for may alter cognitive ability through reduction of excitatory mouse tail; Viagen Biotech, Los Angeles, CA). Genotyping synapses. primer pairs for the WT Nedd9 allele amplified from intron 2 into exon 3; upstream primer (P1) 50-CCA GCA CTG GAG GGA AAT CT-30 and the downstream primer (P2) 50-ATT AGT 4. Experimental procedure GCC GCC AAT GTT CC-30 produced a 753 bp band. Primers for the Nedd9 KO/lacZ knock-in allele amplified from intron 2 into 4.1. Nedd9 KO/lacZ knock-in targeting construct lacZ; upstream primer (P1) 50-CCA GCA CTG GAG GGA AAT CT-30 and the downstream primer (P3) 50-GAT GTG CTG CAA A Nedd9 targeting vector (Fig. 1A) was generated from GGC GAT TA-30 produced a 334 bp band. The presence of a 753 pPCR_LPacINeo (Ozgene Pty. Ltd., Murdoch, Australia), which bp product indicated the presence of at least one WT Nedd9 was further modified to contain a lacZ (β-galactosidase) allele. The appearance of a 334 bp product indicated the reporter upstream of a phosphoglycerine kinase (PGK) pro- presence of at least one Nedd9 KO/lacZ knock-in allele. moter driving a neomycin (Neo) selection marker, all inserted into the unique PacI site in the multiple region of the 4.3. Western blot vector. The lacZ arm contained an XbaI site at the 30 end for þ þ þ genomic screening. The PGK-Neo coding sequence was Tissue from 7- to 9-mo-old female Nedd9 / , Nedd9 / and flanked by loxP sites so that it could be deleted later by Cre. Nedd9 / mice was homogenized in buffer (1:1, g:ml in RIPA 0 The Nedd9 5 homology arm (5867 bp) was generated by PCR Lysis and Extraction Buffer, Thermo Fisher Scientific, Grand 0 0 with the primers (for 5 and 3 arm primers the Nedd9 Island, NY) containing phosphatase and protease inhibitors 0 homology sequence is in bold type); forward 5 -TGG CGC (PhosSTOP and Complete Ultra Tablet Mini, Roche). Lysates 0 GCC GCG GCC GCA GAT GGA TTT TGT AAA GAC GGT T-3 were diluted 1:1 with 2 Laemmli Buffer (Bio-Rad, Hercules, 0 0 and reverse 5 -GCC ATA AGA TTC TGG GAA GGG AAG-3 from CA) and boiled for 10 min prior to loading on SDS 129Sv/j genomic DNA and ligated into the XhoI site of polyacrylamide gels. 0 pPCR_LPacINeo. The Nedd9 3 homology arm (5624 bp) was For Western blot analysis, 20 μg of total protein (based on 0 generated by PCR with the primers; forward 5 -TGG CGC GCC BCA assay) was resolved on a 10% denaturing acrylamide gel 0 AGT ACT CAA ACC CTG TGA GTG AGT CTT CG-3 and reverse (Bio-Rad, Hercules, CA) and transferred overnight to a nitro- 0 0 5 -CAA CAC GCG GCC GCG GCT GGC AAT GGC ACC TAT G-3 cellulose membrane (Bio-Rad, Hercules, CA). Blots were then inserted into the AscI site. The arm contained a ScaI site blocked in PBS containing 5% skim milk powder and 0.1% 0 at the 5 end for genomic screening. The final targeting Tween-20 (rt, 1 h). Blots were then incubated in primary construct was characterized by DNA sequencing and restric- antibody diluted in block (rt, 1 h). Endogenous NEDD9 was tion mapping then linearized with AclI, and transfected into detected with a 1:5000 dilution of rabbit anti-NEDD9 (Abcam, Bruce4 ES cells (derived from C57BL/6) by electroporation. Cambridge, MA). Mouse anti-β-actin clone AC-15 (1:2000, Sigma-Aldrich), served as a loading control. Membranes were 4.2. Generation of Nedd9 KO/lacZ knock-in mice incubated with secondary antibodies conjugated to HRP (Nedd9 / ) (Southern Biotech, Birmingham, AL) in blocking buffer for 1 h. Antibody–antigen conjugates were visualized by a Super- ES cells that survived G418/ganciclovir positive–negative selec- signal West Pico Chemiluminescent kit (Pierce/Thermo Fisher tion were screened for successful homologous recombinants by Scientific, Grand Island, NY) and exposed to X-ray film. Bio- 152 brain research 1632 (2016) 141–155

Rad Precision Plus Protein Dual Color Standard was used to Following training sessions 10, 16, 22 and 28, mice were determine molecular size. given a single probe trial in which the platform was removed from the maze and the mice were allowed to swim freely for 4.4. Animal husbandry 60 s before being returned to their home cages. Main mea- sures taken during the probe trial included platform proxi- Procedures involving animals were reviewed and approved by mity, speed, total path length and percent time in target the Animal Care and Use Committee at the University of quadrant. Wisconsin–Madison, and were conducted in a facility accre- Following completion of the training and probe trials, mice dited by the AAALAC. Mice were housed in shoebox cages were subjected to a visible platform test where a distinct local fi with lids in pathogen-free facilities and maintained on a 12 h- cue was xed to the center of the platform. Mice were given light/12 h-dark cycle in the Department of Biochemistry. Mice 5 consecutive trials, with pseudorandom start locations. were provided with standard laboratory chow (LabDiet 5008, fi PMI Nutrition International, Richmond, Indiana) and reverse 4.6. Tissue collection and xation of MultiBrain sections osmosis filtered water ad libitum. Germline Nedd9 knockout/ lacZ knock-in mice were generated by Ozgene (Perth, Austra- Animals were anesthetized and perfused transcardially with lia). All other animals were obtained from in-house breeding cold PBS (pH 7.4), followed by perfusion with cold 4% PFA and programs. Behavioral studies were conducted at the UCLA 4% sucrose in phosphate buffer (0.1 M, pH 7.4). The brains fi 1 Behavioral Testing Core Facility (Los Angeles, CA). were then dissected from the skull and xed overnight at 4 C in the same fixative. Tissues were washed in PBS and shipped to NeuroScience Associates (Knoxville, TN) for their proprie- 4.5. Behavioral tests tary embedding and MultiBrain section production.

4.5.1. SHIRPA 4.7. Histology and Immunohistochemistry Mice 5- to 6-mo-old (14 KO, 7M/7F and 14 WT, 7M/7F) were individually placed in a large Plexiglass cylinder (30 cm 4.7.1. β-galactosidase and nuclear fast red staining diameter 45 cm height) standing on an elevated metal grid MultiBrain floating coronal sections (35 μm) from 10- to 15- and they were then videotaped in a quiet room under dim mo-old mice, and sagittal sections (30 μm) from 7-mo-old lighting conditions for a period of 10 min. Videotapes were mice were washed twice in PBS with shaking for 5 min at rt scored by a blinded observer (every 2 s; 300 samples/mouse) and were then transferred to 100 ml β-galactosidase staining for a variety of behaviors including: walking, grooming, solution (100 mM potassium phosphate pH 7.4, 5 mM EGTA, leaning, counter-clockwise turning, clockwise turning, climb- 2 mM MgCl2, 0.02% Igepal CA630, 0.01% deoxycholic acid, ing, rearing and standing still. 5 mM potassium ferricyanide (III), and 5 mM potassium hex- acyanoferrate (II) trihydrate) with 1 ml of 50 mg/ml X-gal 4.5.2. Morris water maze added just prior to use, followed by shaking at rt overnight fi Mice 6- to 7-mo-old (9 KO, 5M/4F and 7 WT, 2M/5F) were rst protected from light. Sections were then transferred to slides, trained in a hidden platform water maze task to investigate allowed to air dry followed by treatment (3–5 min each) with: spatial learning. Mice were trained in a custom water maze 95% EtOH, 37% formalin in EtOH (1:9) ratio, 95% EtOH, 70% 1 pool (122 cm diameter) maintained at 24 C with tracking EtOH, 50% EtOH, tap water, and nuclear fast red solution, and information (swim speed, trajectory length, average distance were then placed in slowly running tap water until it rinsed from the platform, relative time in different areas of the pool, clear, followed by dehydration in a series of EtOH, xylene, and etc.) calculated by Water2020 Software (HVS Image, Mountain coverslipped with cytoseal and imaged on a Nikon SMZ-U View, CA). The trials were videotaped, and the latency and Zoom 1:10 scope outfitted with a QImaging Retiga 2000R Fast time measures were analyzed from the tape. The escape 1394 camera (QImaging, Surrey, BC, Canada) and MetaMorph platform remained in the same location for all training trials Meta Imaging 7.8 series software (Molecular Devices, and was located 1 cm below the level of the water, which had Sunnyvale, CA). been made opaque with non-toxic white tempera paint. The training and testing scheme is depicted in Fig. 3A. On 4.7.2. Hematoxylin and Eosin (H&E) and DAPI staining the first day of training, mice were placed on the hidden Coronal MultiBrain sections (30 μm) from 3- to 4-mo-old mice platform for 30 s before the first trial. On each of the fourteen were stained with either H&E or DAPI by NeuroScience training days, mice received two training trials. In each Associates (Knoxville, TN). H&E sections were imaged as training trial, mice were placed in the pool 5 cm from the described above. DAPI sections were imaged on a Nikon center edge of one of the three pool quadrants not containing A1R Confocal System microscope with NIS Elements Viewer the escape platform. Start location varied from trial-to-trial in Version 3.2. a pseudorandom order. Mice were allowed to swim freely until they found the hidden platform or until a maximum of 4.7.3. Golgi-Cox stain 60 s had elapsed. At the end of each trial, mice were allowed Anesthetized animals were sacrificed and brains quickly to remain on the platform or were placed on the platform for dissected from the skull, rinsed briefly in double distilled 5 s before returning to their home cages. The main measure water and prepared using the NeuroTechnologies (Columbia, taken during the training trials was latency to find the MD) FD Rapid GolgiStain Kit as directed for analysis by platform. Neurodigitech (San Diego, CA). brain research 1632 (2016) 141–155 153

4.8. Hippocampal morphology first contoured (10 objective lens), followed by serial count- ing (40 objective lens) along the anterior–posterior axis of 4.8.1. Hippocampal measurements the hippocampus. Before analysis, imaging parameters such Using a mouse brain atlas (Sunkin et al., 2013) and H&E as the scan grid size, counting frame and z-depth size were stained sections corresponding to the region 1.555 mm to validated and consistently kept under control of C.E.o0.1 1.655 mm relative to bregma, hippocampi from 6 WT (5M/ during analysis. During analysis the investigators were kept 1F) and 6 KO (3M/3F) 3- to 4-mo-old mice were analyzed. blinded to the genotypes. Brain IDs were unblinded after Using Meta Imaging 7.8 series software, the distance between completion of the analysis. 3 sets of points and the circumference of the hippocampus was measured (Krieger et al., 2012) in pixels and converted to 4.10. Spine density analysis microns. Two blinded individuals independently measured hippocampi from both hemispheres; these measurements Slides with serial coronal sections (120 μm thickness) that were in close agreement. covered the rostro-caudal axis of DG of the hippocampus were generated from Golgi-Cox stained tissue. Sections were 4.8.2. DG DAPI cell counts analyzed using a stereology-based software (NeuroLucida, v9, Using a mouse brain atlas (Sunkin et al., 2013) and DAPI Microbrightfield, VT), installed on a Dell PC workstation that stained sections corresponding to the region 1.555 mm to controlled a Zeiss Axioplan 2 image microscope with Optro- 1.655 mm relative to bregma, DG from 6 WT (4M/2F) and nics MicroFire CCD camera (1600 1200) digital camera, 6 KO (2M/4F) 3- to 4-mo-old mice were analyzed. Two μm motorized X, Y, and Z-focus for high-resolution image acqui- thick z stack confocal sections (60 lens, 2 digital zoom sition and digital quantitation. Spine analysis was based on adjustment) were taken of each DG using a Nikon A1R the method of Magarinos et al. (2006) and facilitates a direct confocal system microscope and camera. A uniform counting comparison of groups when sample preparation and analysis window was set (using NIS elements software) from the is the same. lateral edge of the crest of the DG and extending along the The sampling process was conducted as follows: based on inferior arm. Using Meta Imaging 7.8 series software, DAPI- the digital Allen Brain Atlas (Sunkin et al., 2013), the entire stained cells were counted in the most medial section of rostro-caudal axis of the region of interest was previewed, each stack. under low-and high-mag Zeiss objectives (10 and 20 ); cells with the least truncations of distal dendrites as possible 4.9. Imaging and stereological analysis of BrdU-positive were located and compared under higher-mag Zeiss objec- (BrdUþ) cells tives (40 and 63 ); low magnification images were cap- tured for archiving purposes, and lastly a Zeiss 100 Mice for BrdU stereology were administered a single intra- objective with immersion oil (N.A¼1.25) was used to perform peritoneal injection of a 200 mg/kg solution of BrdU in PBS (10 3D dendritic reconstruction, followed by continuous counting WT, 10 KO) at 90 days of age. Half (5 WT, 3M/2F and 5 KO, 4M/ of spines throughout the entire dendritic trees. The criteria 1F) of the animals were euthanized 24 h later in order to for selecting candidate neurons for analysis was based on assess cell proliferation, the other half (5 WT, 4M/1F and 5 KO, (i) visualization of a completely filled soma with no overlap of 3M/2F) were euthanized 3 wk later to analyze the number of neighboring soma and completely filled dendrites, (ii) the BrdU labeled cells that survived. Animals were anesthetized tapering of most distal dendrites; and (iii) the visualization of and perfused transcardially with cold PBS (pH 7.4), followed the complete 3-D profile of dendritic trees using the 3-D by perfusion with cold 4% PFA and 4% sucrose in PBS (0.1 M, display of imaging software. Neurons with incomplete pH 7.4). The brains were then dissected from the skull and impregnation and/or neurons with truncations due to the fixed 24 h at 4 1C in the same fix. Tissues were washed in PBS plane of sectioning were not collected. followed by sectioning (50 μm, coronal) and MultiBrain pro- The sample population was composed of at least 24 cessing for BrdU immunochemistry (NSA). Every 6th section granule cells from a minimum of at least 5 mice per group in the region of the hippocampus was used for BrdU analysis randomly chosen rostro-caudally from the region of interest for a total of 20 sections. Sterological analysis of BrdUþ cells (DG or CA1), which was located in the anterior two-thirds of was performed by Neurodigitech. the hippocampus. After completion, the digital profile of Serial slides of 5 brains per group were imaged on a Zeiss neuron morphology including the dendrograms, spine Axioplan 2 digital microscopy system, equipped with X–Y counts, and Sholl analysis was analyzed for statistical motorized stage, Z stepper, Optronics/MBF CX9000 RGB CCD significance. (1600 1200 pixels) and LUDL controller for navigations The sampling site covered the entire ML of the DG. For all across the A–P axis of the hippocampal subfields of the DG, analyses, the investigators were kept blind as to genotypes. including SGZ (defined for counting purposes as a two- Total and subtype spine densities were calculated by dividing nucleus-wide band of the SGZ and inner one-third of the the total number of spines per cell by the total dendritic granule cell layer, GCL), hilus (H) and molecular layer (ML). length of the cell. Group data are reported as mean7SEM. Quantitative analyses were performed on blinded-coded slides using commercial software, StereoInvestigator (Micro- 4.11. Statistical analysis brightfield, VT) for stereology-based counting of BrdUþ cells in the SGZ and other subfields, based on the mouse brain Statistical analysis was performed in GraphPad Prism 6.0 atlas (Paxinos and Franklin, 2004). The target regions were (GraphPad Software, Inc., La Jolla, CA). Statistical comparisons 154 brain research 1632 (2016) 141–155

between the two groups were made using Student's t-test or Guerrero, M.S., Parsons, J.T., Bouton, A.H., 2012. Cas and NEDD9 Multiple t-tests when appropriate. For multigroup comparisons, contribute to tumor progression through dynamic regulation a two-way ANOVA was used with Tukey's post-hoc test to of the cytoskeleton. Genes Cancer 3, 371–381. Holtmaat, A.J., Trachtenberg, J.T., Wilbrecht, L., Shepherd, G.M., analyze data from Sholl analysis, branch order and the spine Zhang, X., Knott, G.W., Svoboda, K., 2005. Transient and density time course. Descriptive statistical results are presented persistent dendritic spines in the neocortex in vivo. Neuron 7 fi fi r as mean SEM. Statistical signi cance was de ned as p 0.05. 45, 279–291. Ishino, M., Ohba, T., Sasaki, H., Sasaki, T., 1995. Molecular cloning of a cDNA encoding a phosphoprotein, Efs, which contains a Acknowledgments Src homology 3 domain and associates with Fyn. Oncogene 11, 2331–2338. Irwin, S., 1968. Comprehensive observational assessment: Ia. A We thank Laura Vanderploeg in the Biochemistry Media systematic, quantitative procedure for assessing the beha- Group for her assistance in preparing the artwork, and the vioral and physiologic state of the mouse. Psychopharmaco- College of Agriculture and Life Sciences Statistical Consulting logia 13, 222–257. Lab for statistical support. We thank Clagett-Dame lab mem- Knutson, D.C., Clagett-Dame, M., 2008. atRA Regulation of NEDD9, ber James Jones for his assistance in hippocampal measure- a gene involved in neurite outgrowth and cell adhesion. Arch. Biochem. Biophys. 477, 163–174. ments. This work was supported by the Steenbock Research Knutson, D.C., Clagett-Dame, M., 2015. A complex RARE is – Fund, University of Wisconsin Madison. required for the majority of Nedd9 embryonic expression. Transgenic Res. 24, 123–134. Koleske, A.J., 2013. Molecular mechanisms of dendrite stability. Appendix A. Suplementary information Nat. Rev. Neurosci. 14, 536–550. Korobova, F., Svitkina, T., 2010. Molecular architecture of synaptic actin cytoskeleton in hippocampal neurons reveals a Supplementary data associated with this article can be found mechanism of dendritic spine morphogenesis. Mol. Biol. Cell. in the online version at: http://dx.doi.org/1016/j.brainres.2015. 21, 165–176. 12.005. Krieger, M., Both, M., Kranig, S.A., Pitzer, C., Klugmann, M., Vogt, G., Draguhn, A., Schneider, A., 2012. The hematopoietic cytokine granulocyte–macrophage colony stimulating factor is important for cognitive functions. Sci. Rep. 2, 697. references Kulkarni, V.A., Firestein, B.L., 2012. The dendritic tree and brain disorders. Mol. Cell. Neurosci. 50, 10–20. Kumar, S., Tomooka, Y., Noda, M., 1992. Identification of a set of genes with developmentally down-regulated expression in the Abrous, D.N., Koehl, M., Le Moal, M., 2005. Adult neurogenesis: mouse brain. Biochem. Biophys. Res. Commun. 185, from precursors to network and physiology. Physiol. Rev. 85, 1155–1161. 523–569. Law, S.F., Estojak, J., Wang, B., Mysliwiec, T., Kruh, G., Golemis, Alexandropoulos, K., Baltimore, D., 1996. Coordinate activation of Cas E.A., 1996. Human enhancer of filamentation 1, a novel c-Src by SH3- and SH2-binding sites on a novel p130 -related p130cas-like docking protein, asociates with focal adhesion protein, Sin. Genes Dev. 10, 1341–1355. Aquino, J.B., Lallemend, F., Marmigere, F., Adameyko, I.I., Golemis, kinase and induces pseudohyphal growth in Saccharomyces E.A., Ernfors, P., 2009. The retinoic acid inducible Cas-family cervisiae. Mol. Cell. Biol. 16, 3327–3337. signaling protein Nedd9 regulates neural crest cell migration Law, S.F., Zhang, Y.Z., Klein-Szanto, A.J.P., Golemis, E.A., 1998. Cell by modulating adhesion and actin dynamics. Neuroscience cycle-regulated processing of HEF1 to multiple protein forms 162, 1106–1119. differentially targeted to multiple subcellular compartments. Bargon, S.D., Gunning, P.W., O’Neill, G.M., 2005. The Cas family Mol. Cell. Biol. 18, 3540–3551. docking protein, HEF1, promotes the formation of neurite-like Levy, A.D., Omar, M.H., Koleske, A.J., 2014. Extracellular matrix membrane extensions. Biochim. Biophys. Acta 1746, 143–154. control of dendritic spine and synapse structure and plasticity Beck, T.N., Nicolas, E., Kopp, M.C., Golemis, E.A., 2014. Adaptors in adulthood. Front. Neuroanat. 8, 116. for disorders of the brain? The cancer signaling proteins Li, Y., Grupe, A., Rowland, C., Holmans, P., Segurado, R., Abraham, NEDD9, CASS4, and PTK2B in Alzheimer’s disease. R., Jones, L., Catanese, J., Ross, D., Mayo, K., Martinez, M., Oncoscience 1, 486–503. Hollingworth, P., Goate, A., Cairns, N.J., Racette, B.A., Perl- Bourne, J., Harris, K.M., 2007. Do thin spines learn to be mush- mutter, J.S., O’Donovan, M.C., Morris, J.C., Brayne, C., room spines that remember? Curr. Opin. Neurobiol. 17, Rubinsztein, D.C., Lovestone, S., Thal, L.J., Owen, M.J., 381–386. Williams, J., 2008. Evidence that common variation in NEDD9 DeKosky, S.T., Scheff, S.W., 1990. Synapse loss in frontal cortex is associated with susceptibility to late-onset Alzheimer’s and biopsies in Alzheimer’s disease: correlation with cognitive Parkinson’s disease. Hum. Mol. Genet. 17, 759–767. severity. Ann. Neurol. 27, 457–464. Lin, Y.-C., Koleske, A.J., 2010. Mechanisms of synapse and den- Deng, W., Aimone, J.B., Gage, F.H., 2010. New neurons and new drite maintenance and their disruption in psychiatric and memories: how does adult hippocampal neurogenesis affect neurodegenerative disorders. Annu. Rev. Neurosci. 33, learning and memory? Nat. Rev. Neurosci. 11, 339–350. 349–378. Fashena, S.J., Einarson, M.B., O’Neill, G.M., Patriotis, C., Golemis, Magarinos, A.M., McEwen, B.S., Saboureau, M., Pevet, P., 2006. E.A., 2002. Dissection of HEF1-dependent functions in motility Rapid and reversible changes in intrahippocampal connec- and transcriptional regulation. J. Cell Sci. 115, 99–111. tivity during the course of hibernation in European hamsters. Fu, Y., He, F., Tang, N.L., Tam, C.W., Lui, V.W., Chiu, H.F., Lam, L.C., Proc. Natl. Acad. Sci. USA 103, 18775–18780. 2012. NEDD9 gene polymorphism influences the risk of Alz- Merrill, R.A., Ahrens, J.M., Kaiser, M.E., Federhart, K.S., Poon, V.Y., heimer disease and cognitive function in Chinese older Clagett-Dame, M., 2004a. All-trans retinoic acid-responsive persons. Alzheimer Dis. Assoc. Disord. 26, 88–90. genes identified in the human SH-SY5Y neuroblastoma cell brain research 1632 (2016) 141–155 155

line and their regulated expression in the nervous system of lymphocyte trafficking and marginal zone B cell maintenance. early embryos. Biol. Chem. 385, 605–614. J. Immunol. 175, 3492–3501. Merrill, R.A., See, A.W., Wertheim, M.L., Clagett-Dame, M., 2004b. Sholl, D.A., 1956. The measurable parameters of the cerebral Crk-associated substrate (Cas) family member, NEDD9, is cortex and their significance in its organization. Prog. Neuro- regulated in human neuroblastoma cells and in the embryo- biol., 324–333. nic hindbrain by all-trans retinoic acid. Dev. Dyn. 231, 564–575. Singh, M.K., Dadke, D., Nicolas, E., Serebriiskii, I.G., Apostolou, S., Minegishi, M., Tachibana, K., Sato, T., Iwata, S., Nojima, Y., Canutescu, A., Egelston, B.L., Golemis, E.A., 2008. A novel Cas Morimoto, C., 1996. Structure and function of Cas-L, a 105-kD family member, HEPL, regulates FAK and cell spreading. Mol. Crk-associated substrate-related protein that is involved in β1 Biol. Cell 19, 1627–1636. -mediated signaling in lymphocytes. J. Exp. Med. 184, Star, E.N., Kwiatkowski, D.J., Murthy, V.N., 2002. Rapid turnover of 1365–1375. actin in dendritic spines and its regulation by activity. Nat. Nikonova, A.S., Gaponova, A.V., Kudinov, A.E., Golemis, E.A., 2014. Neurosci. 5, 239–246. CAS proteins in health and disease: an update. IUBMB Life 66, Sunkin, S.M., Ng, L., Lau, C., Dolbeare, T., Gilbert, T.L., Thompson, 387–395. C.L., Hawrylycz, M., Dang, C., 2013. Allen Brain Atlas: an Ohashi, Y., Iwata, S., Kamiguchi, K., Morimoto, C., 1999. Tyrosine integrated spatio-temporal portal for exploring the central of Crk-associated substrate lymphocyte-type nervous system. Nucleic Acids Res. 41, D996–D1008. is a critical element in TCR- and beta 1 integrin-induced T Tedde, A., Bagnoli, S., Piaceri, I., Lucenteforte, E., Bessi, V., Bracco, lymphocyte migration. J. Immunol. 163, 3727–3734. L., Mugelli, A., Sorbi, S., Nacmias, B., 2010. Different implica- Paxinos, G., Franklin, K.B.J., 2004. The Mouse Brain in Stereotaxic tion of NEDD9 genetic variant in early and late-onset Alzhei- Coordinates. Elsevier Science (USA), Academic Press, San mer’s disease. Neurosci. Lett. 477, 121–123. Diego, California. Terry, R.D., Masliah, E., Salmon, D.P., Butters, N., DeTeresa, R., Hill, Penzes, P., Cahill, M.E., Jones, K.A., VanLeeuwen, J.-E., Woolfrey, K. R., Hansen, L.A., Katzman, R., 1991. Physical basis of cognitive M., 2011. Dendritic spine pathology in neuropsychiatric dis- alterations in Alzheimer’s disease: synapse loss is the major orders. Nat. Neurosci. 14, 285–293. correlate of cognitive impairment. Ann. Neurol. 30, 572–580. Pozueta, J., Lefort, R., Shelanski, M.L., 2013. Synaptic changes in Tikhmyanova, N., Little, J.L., Golemis, E.A., 2010. CAS proteins in Alzheimer’s disease and its models. Neuroscience 251, 51–65. normal and pathological cell growth control. Cell. Mol. Life Reynolds, A.B., Kanner, S.B., Wang, H.-C.R., Parsons, J.T., 1989. Sci. 67, 1025–1048. Stable association of activated pp60src with two tyrosine- van Seventer, G.A., Salmen, H.J., Law, S.F., O’Neill, G.M., Mullen, phosphorylated cellular proteins. Mol. Cell. Biol. 9, 3951–3958. M.M., Franz, A.M., Kanner, S.B., Golemis, E.A., van Seventer, J. Riccomagno, M.M., Sun, L.O., Brady, C.M., Alexandropoulos, K., M., 2001. Focal adhesion kinase regulates beta1 integrin- Seo, S., Kurokawa, M., Kolodkin, A.L., 2014. Cas adaptor dependent migration through an HEF1 effector pathway. proteins organize the retinal ganglion cell layer downstream Eur. J. Immunol. 31, 1417–1427. of integrin signaling. Neuron 81, 779–786. Vogel, T., Ahrens, S., Buttner, N., Krieglstein, K., 2010. Trans- Rogers, D.C., Fisher, E.M., Brown, S.D., Peters, J., Hunter, A.J., forming growth factor beta promotes neuronal cell fate of Martin, J.E., 1997. Behavioral and functional analysis of mouse mouse cortical and hippocampal progenitors in vitro and phenotype: SHIRPA, a proposed protocol for comprehensive in vivo: identification of Nedd9 as an essential signaling phenotype assessment. Mamm. Genome. 8, 711–713. component. Cereb. Cortex 20, 661–671. Sakai, R., Iwamatsu, A., Hirano, N., Ogawa, S., Tanaka, T., Mano, von Bohlen Und Halbach, O., 2009. Structure and function of H., Yazaki, Y., Hirai, H., 1994. A novel signaling molecule, p130, dendritic spines within the hippocampus. Ann. Anat. 191, forms stable complexes in vivo with v-Crk and v-Src in a 518–531. tyrosine phosphorylation-dependent manner. EMBO J. 13, Xing, Y.-Y., Yu, J.-T., Yan, W.-J., Chen, W., Zhong, X.-L., Jiang, H., 3748–3756. Wang, P., Tan, L., 2011. NEDD9 is genetically associated with Sasaki, T., Iwata, S., Okano, H.J., Urasaki, Y., Hamada, J., Tanaka, Alzheimer’s disease in a Han Chinese population. Brain Res. H., Dang, N.H., Okano, H., Morimoto, C., 2005. Nedd9 protein, a 1369, 230–234. Cas-L homologue, is upregulated after transient global ische- Yang, G., Pan, F., Gan, W.-B., 2009. Stably maintained dendritic mia in rats: possible involvement of Nedd9 in the differen- spines are associated with lifelong memories. Nature 462, tiation of neurons after ischemia. Stroke 36, 2457–2462. 920–924. Seo, S., Asai, T., Saito, T., Suzuki, T., Morishita, Y., Nakamoto, T., Zuo, Y., Lin, A., Chang, P., Gan, W.-B., 2005. Development of long- Ichikawa, M., Yamamoto, G., Kawazu, M., Yamagata, T., Sakai, term dendritic spine stability in diverse regions of cerebral R., Mitani, K., Ogawa, S., Kurokawa, M., Chiba, S., Hirai, H., cortex. Neuron 46, 181–189. 2005. Crk-associated substrate lymphocyte type is required for