sign in sign up
<<
Home , Neurotrophin

Journal of Neuroscience Research 90:1743–1755 (2012)

Hepatocyte Overexpression in the Enhances Learning and Memory Performance in Mice

Takashi Kato,1 Hiroshi Funakoshi,2* Keiichi Kadoyama,3 Satsuki Noma,2,4 Masaaki Kanai,2 Wakana Ohya-Shimada,1,2 Shinya Mizuno,4 Nobutaka Doe,5 Taizo Taniguchi,3 and Toshikazu Nakamura1* 1Kringle Pharma Joint Research Division for Regenerative Drug Discovery, Center for Advanced Science and Innovation, Osaka University, Osaka, Japan 2Center for Advanced Research and Education, Asahikawa Medical University, Asahikawa, Japan 3Department of Pharmaceutical Health Care, Faculty of Pharmaceutical Sciences, Himeji Dokkyo University, Himeji, Japan 4Division of Virology, Department of Microbiology and Immunology, Osaka University Graduate School of Medicine, Osaka, Japan 5Laboratory of and CNS Repair, Institute for Advanced Medical Sciences, Hyogo College of Medicine, Hyogo, Japan

Hepatocyte growth factor (HGF) and its , c- Learning and memory are among the most essential Met, play pivotal roles in the nervous system during activities for intelligent life. A large amount of evidence development and in disease states. However, the suggests that regions such as the physiological roles of HGF in the adult are not and are important for cognition (Sara, well understood. In the present study, to assess its 2000; Frankland and Bontempi, 2005). Compared with role in learning and memory function, we used trans- the accumulating knowledge of the mechanisms underly- genic mice that overexpress HGF in a -specific ing learning and memory impairments under pathophys- manner (HGF-Tg) to deliver HGF into the brain without iological conditions (i.e., neuronal and vascular degener- injury. HGF-Tg mice displayed increased alternation ation/dysfunction in diseases including Alzheimer’s dis- rates in the Y-maze test compared with age-matched ease), the mechanisms regulating physiological learning wild-type (WT) controls. In the Morris water maze and memory performance remain a mystery at the mo- (MWM) test, HGF-Tg mice took less time to find the lecular level. Only limited numbers of molecules have platform on the first day, whereas the latency to been characterized, including CaM kinase II, NMDA re- escape to the hidden platform was decreased over ceptor subunit 2B (NR2B), CREB, and brain-derived training days compared with WT mice. A transfer test revealed that the incidence of arrival at the exact location of the platform was higher for HGF-Tg mice Additional Supporting Information may be found in the online version compared with WT mice. These results demonstrate of this article. that overexpression of HGF leads to an enhancement Contract grant sponsor: Japanese Ministry of Health Labor and Welfare of both short- and long-term memory. Western blot (to H.F.); Contract grant sponsor: Ministry of Education, Science, Sports, analyses revealed that the levels of N-methyl-D- and Culture of Japan (to H.F., T.N.); Contract grant sponsor: Global aspartate (NMDA) receptor subunits NR2A and NR2B, Centers of Excellence (COE) program of the Ministry of Education, Sci- but not NR1, were increased in the hippocampus of ence, and Culture of Japan (to T.N.). HGF-Tg mice compared with WT controls, suggesting W. Ohya-Shimada’s current address is Center for Advanced Research that an upregulation of NR2A and NR2B could repre- and Education, Asahikawa Medical University, Asahikawa, Japan sent one mechanism by which HGF enhances learn- *Correspondence to: Dr. Hiroshi Funakoshi, Center for Advanced ing and memory performance. These results demon- Research and Education, Asahikawa Medical University, Asahikawa 078- strate that modulation of learning and memory per- 8510, Japan. E-mail: [email protected] and Dr. Toshikazu formance is an important physiological function of Nakamura, Kringle Pharma Joint Research Division for Regenerative HGF that contributes to normal CNS plasticity, and Drug Discovery, Center for Advanced Science and Innovation, Osaka we propose HGF as a novel regulator of higher brain University, Osaka Japan, E-mail: [email protected] functions. VC 2012 Wiley Periodicals, Inc. Received 19 September 2011; Revised 1 March 2012; Accepted 13 March 2012 Key words: (HGF); c-Met; Published online 26 April 2012 in Wiley Online Library learning and memory; NMDA receptor; (wileyonlinelibrary.com). DOI: 10.1002/jnr.23065

' 2012 Wiley Periodicals, Inc. 1744 Kato et al. neurotrophic factor (BDNF; Silva et al., 1992, 1998; tests, the Y-maze and the Morris water maze (MWM). Tang et al., 1999; Nakajo et al., 2008). Identification of Our results yield new insights into the regulatory mech- additional molecules affecting these molecular neuroa- anisms of adult learning and memory. daptations in the forebrain is crucial for a better under- standing of the physiological mechanisms that regulate MATERIALS AND METHODS learning and memory performance. Animals Hepatocyte growth factor (HGF) was initially iden- tified as a potent mitogen for primary hepatocytes Transgenic mice expressing rat HGF in the nervous sys- (Nakamura et al., 1984, 1989). HGF exhibits multiple tem (HGF-Tg mice) were prepared, bred, and maintained as biological effects, including mitogenic, motogenic, mor- previously described (Sun et al., 2002b). After back-crossing phogenic, and antiapoptotic activities in a wide variety with WT C57BL/6J mice (SLC, Shizuoka, Japan) for more of cells by binding to the c-Met/HGF receptor tyrosine than 10 generations, homozygous HGF-Tg mice were created kinase (c-Met) (Funakoshi and Nakamura, 2003; Naka- by the mating of heterozygotes and were confirmed either by mura and Mizuno, 2010). HGF plays pivotal roles as an dot blot hybridization or a PCR method, as previously organotrophic factor during development and in the described (Sun et al., 2002b; Kadoyama et al., 2007, 2009; protection and regeneration of a variety of organs under Benkhoucha et al., 2010). Further confirmation was accom- pathophysiological conditions (Funakoshi and Nakamura, plished by progeny tests. Age-matched WT C57BL/6J mice 2003; Nakamura and Mizuno, 2010). In the nervous sys- served as controls. HGF-Tg and WT mice were transferred tem, HGF and c-Met are expressed not only in develop- into experimental rooms (for Y-maze and MWM tasks) and ing but also in diseased brains and play functional maintained for more than 10 days before behavioral assessment. roles during development and in disease processes The acquisition, care, housing, use, and disposition of the ani- (Maina and Klein, 1999; Funakoshi and Nakamura, mals were in compliance with institutional laws and regulations 2011). However, the in vivo role of HGF in the adult of the Osaka University Graduate School of Medicine and Asa- brain is not well understood. hikawa Medical University. The experimental procedures for Notably, brain regions that express HGF and c-Met the MWM tests were conducted according to the National include forebrain regions such as the hippocampus and cer- Institute of Health guidelines. All efforts were made to mini- ebral cortex. Application of HGF in vitro enhances mize animal discomfort and the number of animals used. extension and branching and maturation of the dendritic spine, induces the phosphorylation of Akt and CREB and Measurement of HGF Levels in the translocation of NMDA receptors, and increases the phos- Hippocampus and Cerebral Cortex phorylation of NR2B and neuronal survival in primary After being deeply anesthetized with an overdose of so- hippocampal and/or cerebral cortical (Honda dium pentobarbital, the hippocampus and cerebral cortex of et al., 1995; Sun et al., 2002a; Ishihara et al., 2005; Akita HGF-Tg heterozygous (1/2), HGF-Tg homozygous (1/1), et al., 2008; Lim and Walikonis, 2008). HGF enhances and age-matched WT mice were excised, quickly frozen, and NMDA currents and synaptic plasticity in the hippocam- stored at 2808C until use. Frozen tissue samples were ho- pus and also enhances synaptic long-term potentiation mogenized and centrifuged, and the supernatants were then (LTP) in the CA1 region of the hippocampus in vitro used to quantify HGF protein levels via enzyme-linked im- (Akimoto et al., 2004). Similarly to other neurotrophic fac- munosorbent assay (ELISA; Tokushu Meneki, Tokyo, Japan), tors (Zafra et al., 1990; Funakoshi et al., 1995), HGF par- as previously described (Sun et al., 2002b). ticipates in activity-dependent signaling, including signal- ing in primary hippocampal neurons (Tyndall and Waliko- Histological and Immunohistochemical Analyses nis, 2007). These results led us to hypothesize that Deeply anesthetized animals were transcardially perfused physiological activation of the HGF-c-Met system might with ice-cold phosphate-buffered saline (PBS; pH 7.4) followed underlie the activity-dependent regulation of physiological by ice-cold 4% paraformaldehyde (PFA) in PBS. The extracted learning and memory performance in the adult brain. brains were fixed in 4% PFA in PBS at 48Covernight.For In the present study, to test the possibility that the analyses of brain morphology and MAP2 immunostaining, the HGF-c-Met system plays a role in learning and memory brains were embedded in paraffin. Serial coronal sections (7 performance, we used transgenic mice overexpressing lm) were prepared, deparaffinized, and stained with Nissl or HGF under the control of the neuron-specific enolase were used for MAP2 immunostaining. Primary antibody for (NSE) promoter (HGF-Tg mice) to introduce HGF into MAP2 (Sigma, St. Louis, MO) was applied to the coronal sec- the hippocampus and cerebral cortex (Sun et al., 2002b; tions. The sections were incubated at 48C overnight. After Kadoyama et al., 2007, 2009; Benkhoucha et al., 2010). being washed with PBS, the sections were incubated with the This strategy avoids an insult-dependent, nonessential Envision1 System Labeled Polymer-HRP anti-mouse (Dako upregulation of mRNAs for BDNF and its receptor, Cytomation, Carointeria, CA) antibody as the secondary anti- TrkB (Mudo et al., 1993; Kokaia et al., 1994), the sig- body for 1 hr at room temperature and then stained with DAB naling of which has been implicated in learning and (0.05%), counterstained with Nissl, and mounted with memory performance (Nakajo et al., 2008). Here we EUKITT (O. Kinder, Freiburg im Breisgau, Germany). The show that overexpression of HGF led to enhanced short- sections were observed and data were recorded using an Olym- and long-term memory performance on two behavioral pus Microscope (Olympus, Tokyo, Japan) or an All-in-One

Journal of Neuroscience Research HGF Enhances Learning and Memory in Mice 1745

Fluorescence Microscope (Keyence, Osaka, Japan). For immu- were determined in an aliquot of each preparation using the BCA nostaining of HGF, c-Met, GFAP, and Iba1, the extracted method with bovine serum albumin as the standard. Ten micro- brains were fixed in 4% PFA in PBS at 48Cfor4hrandsubse- grams of individual lysates was diluted in the sample buffer and quently cryoprotected in 10% and 20% sucrose in PBS at 48C heated at 958C for 5 min. The lysates were separated by 5% SDS- overnight. Twenty-micrometer sections were prepared using a PAGE and were electroblotted onto PVDF membranes to quan- cryostat and stored at 2808C. Primary antibodies for rat HGF tify changes in NMDA receptor subunit levels. After blocking in (Yamada et al., 1995; Ohya et al., 2007), c-Met (SP260; Santa 100 mM Tris-buffered saline containing 0.05% Triton X-100 Cruz Biotechnology, Santa Cruz, CA), GFAP (Chemicon, (TBS-T) with 3% BSA at room temperature for at least 1 hr, the Temecula, CA), or Iba1 (Wako, Osaka, Japan) were applied to membranes were incubated in the same solutions with antibodies the coronal sections. The sections were incubated at 48Cover- against NR1 (, Beverly, MA), NR2A (Millipore, night, subsequently washed with PBS, and then incubated with Billerica, MA), NR2B (BD Transduction Laboratories, Franklin secondary antibodies (1:600, Alexa488-conjugated and Lakes, NJ), or tubulin-bIII (Sigma) overnight at a concentration Alexa546-conjugated IgG; Invitrogen, Carlsbad, CA) and the of 1:4,000. Alternatively, lysates were separated by 7.5% SDS- nuclear counterstain reagent TO-PRO-3 iodide (Invitrogen) at PAGE and were electroblotted onto PVDF membranes. After room temperature for 20 min. After washing with PBS, the blocking in TBS-T containing 3% nonfat milk at room tempera- sections were mounted with crystal mount (Biomeda, Foster ture for at least 1 hr, the membranes were subsequently incubated City, CA) and observed under an LSM5 Pascal confocal micro- in the same solutions with antibodies against PSD-95 (Thermo scope (Zeiss, Oberkochen, Germany). Coronal sections of Scientific, Waltham, MA), synapsin I (Millipore), or tubulin-bIII mouse hippocampus and cerebral cortex between 1.70 and 2.06 for 1 hr at a concentration of 1:4,000. After washing three times mm caudal to bregma per Franklin and Paxinos (1997) were with TBS-T, the membranes were incubated with horseradish used for histological analyses, including Nissl staining and peroxidase (HRP)-conjugated goat anti-mouse IgG antibody or MAP2, GFAP, and Iba1 immunostaining. rabbit IgG antibody at a concentration of 1:4,000 at room temper- ature for 1 hr. After three washings, the signals were detected with enhanced chemiluminescence (ECL) and ECL Plus reagents (GE Immunoprecipitation and Western Blotting Healthcare). Densitometric quantification of the scanned band The whole hippocampus and cerebral cortex of mice intensities was performed by digitizing each band via densitometry were dissected under a dissection microscope and used for in Image J software on a Macintosh computer. Western blot analyses. Hippocampal and cerebral cortical lysates were prepared from 22–26-week-old HGF-Tg (1/2), Quantitative Real-Time RT-PCR (1/1), and age-matched WT mice in lysis buffer (50 mM Total RNA was prepared from the hippocampus and cer- Tris-HCl, pH 7.5,, 150 mM NaCl, 1% Triton X-100, 5 mM ebral cortex using Trizol (Invitrogen). One microgram of total EDTA, protease inhibitor cocktail [Roche, Indianapolis, IN], RNA was reverse-transcribed into the first-strand cDNA with and 2 mM Na3VO4) using a Mixer Mill 300 (Qiagen, Ger- an oligo(dT) primer using Superscript III reverse transcriptase mantown, MD) and were subsequently sonicated with a Bio- (Invitrogen) according to the manufacturer’s instructions. For ruptor UCD250 (Cosmo Bio) at 48C. The lysates were preab- quantitative RT-PCR analyses, amplification reactions were sorbed with 20 ll of protein G-Sepharose (GE Healthcare, performed using the ABI Prism 7900 sequence detection system Buckinghamshire, United Kingdom) for 2 hr at 48C, and then (Applied Biosystems, Foster City, CA) with SYBR green equal amounts of supernatants (4 mg protein), as determined qPCR ThunderBird (Toyobo, Osaka, Japan). The protocol using a DC protein assay (Bio-Rad, Hercules, CA), were comprised an initial step (958C for 1 min), 40 DNA amplifica- incubated overnight at 48C with 4 ll of mouse anti-c-Met tion cycles (958C for 15 sec, 608C for 20 sec, and 728Cfor30 (B2; Santa Cruz Biotechnology). The immunocomplexes sec), and a dissociation stage for quality control. Expression was were precipitated with 40 ll protein G-Sepharose. Protein G- normalized to GAPDH levels. The following specific primers sepharose beads were washed three times with lysis buffer and were used for mouse NGF, BDNF, GDNF,andGAPDH: dissolved in sample buffer. Then, the immunoprecipitates NGF forward 50-GCCTCAAGCCAGTGAAATTAGG-30, were separated by 7.5% SDS-PAGE under reducing condi- NGF reverse 50-ACGACCACAGGCCAAAACTC-30;BDNF tions, electroblotted onto PVDF membranes (Bio-Rad), and forward 50-AGAGCTGTTGGATGAGGACCAG-30,BDNF probed with mouse antiphosphotyrosine antibody (pY99; reverse 50-CAAAGGCACTTGACTACTGAGCA-30;GDNF Santa Cruz Biotechnology) and rabbit anti-c-Met antibody forward 50-TGACTCCAATAGCCTGAAGATTAATC-30, (SP260; Santa Cruz Biotechnology). GDNF reverse 50-AATGGTGGCTTGAATAAAATCCA-30; For the quantification of synaptic protein expression, hip- GAPDH forward 50-TGTGTCCGTCGTGGATCTGA-30, pocampal and cerebral cortical lysates of 20-week-old HGF-Tg GAPDH reverse 50-CCTGCTTCACCTTCTTGA-30.For (1/2), (1/1), and WT mice were prepared. Briefly, after amplification of mouse VEGF mRNA levels, universal PCR decapitation and brain extraction, tissue samples were dissected master mix and FAM dye-labeled Taq-Man MGB probe and then homogenized in 200 ll homogenizing buffer containing (Applied Biosystems) were used (Mm00437304_m1). 50 mM Tris-HCl (pH 7.4), 0.5% Triton X-100, 4 mM EGTA, 10 mM EDTA, phosphatase inhibitor cocktail (Nacalai, Kyoto, Behavioral Tests Japan), protease inhibitor cocktail (Roche), 1 nM calyculin A, and Spontaneous alternation test (Y-maze).. The 1 mM dithiothreitol (DTT) using a MixerMill and subsequently maze was made of black, painted wood. Each arm was 40 cm sonicated using a Bioruptor. The concentrations of total lysates long, 12 cm high, and 3 cm wide at the bottom and 10 cm

Journal of Neuroscience Research 1746 Kato et al. wide at the top. The arms converged at an equilateral triangu- Statistical Analysis lar central area that was 4 cm at its longest axis. Each mouse Statistical analysis was carried out in StatView 5.0.1 was placed at the center of the apparatus, and allowed to (SAS Institute, Cary, NC). Escape latencies, swimming speed, move freely through the maze during an 8-min session. The and regional HGF levels in the brain were all analyzed using number of arm entries was recorded visually. Alternation was one-way ANOVA with Scheffe’s and Fisher’s post hoc tests. defined as successive entry into the three arms on overlapping The results are presented as the mean 6 SEM. P < 0.05 was triplet sets. The alternation behavior (%) was calculated as the considered significant. ratio of actual alternations to possible alternations (defined as the number of arm entries minus 2), multiplied by 100. RESULTS Spatial learning and memory test (Morris water maze task).. Spatial learning, comprising cognition/orienta- HGF Protein Levels in the Hippocampus and tion, working memory learning/consolidation, and recall/nav- Cerebral Cortex of HGF-Tg Mice igation, was examined in the MWM task (Morris, 1981). We have previously reported that rat HGF Place task (days 1–5).. Mice were tested in a circular mRNA is specifically overexpressed in the nervous sys- pool (95 cm in diameter, 35 cm in height) that was filled tem (brain and ) of HGF-Tg mice as with water made opaque by adding titanium oxide to a revealed by RNase protection assay and that the total depth of 22 cm, at a temperature of 228C 6 18C. The pool protein levels of mouse and rat HGF in the spinal cord was divided into four virtual quadrants: north, south, east, of HGF-Tg heterozygous (1/2)micearedoubled and west. A white, round platform (10 cm in diameter) was across a time period of 1 week postnatally to adulthood situated at the center of the north quadrant of the pool and (Sun et al., 2002b). HGF-Tg mice have been success- was submerged 0.5 cm below the water surface. White walls fully used for various studies (Sun et al., 2002b; surrounded the pool at a distance of 20–40 cm from the Kadoyama et al., 2007, 2009; Benkhoucha et al., 2010). pool edges. Prominent external cues for orientation were However, the precise levels of HGF in the hippocam- provided by a CCD camera, two posters with high-contrast pus and cerebral cortex have not been measured. black and white patterns, and a black doll. These cues were Therefore, we first determined the levels of HGF pro- attached to the walls at a horizontal distance of 10–30 cm tein in the hippocampus and cerebral cortex in HGF- from the pool and a vertical elevation of 20–50 cm above Tg mice that had been back-crossed to WT (C57BL/ the water surface. 6J) mice for more than 10 generations, compared with Each mouse underwent five escape training trials per WT (C57BL/6J) mice, using ELISA as previously day for 5 consecutive days. In each training trial, the mouse described (Sun et al., 2002b). We have paid particular was placed into the water at the middle point of the circular attention to these brain regions for the following rea- edge in the south, east, or west quadrant, with the head facing sons. First, short-term memory involves the storage and the outer edge of the pool. A training trial terminated when retrieval of information that will be needed within the the mouse reached the platform and remained on it for 10 next several seconds to minutes. This information is sec. If the platform was not found within 60 sec, the mouse stored on a temporary basis and is largely dependent on was guided to the platform by the experimenter and kept regions within the frontal cortex. Long-term memories, there for 10 sec. The order of releasing points varied from day in contrast, are maintained by more stable and perma- to day with pseudorandom sequences for each mouse. The nent changes in neural connections widely spread intertrial interval was 30 sec. The escape latency of each train- throughout the brain. The hippocampus is essential for ing trial was measured to a maximum of 60 sec. integrating new information and for the transfer of in- Transfer (probe) test (day 6).. On day 6, the mice under- formation from short-term to long-term memory. went a transfer test trial, in which the platform was removed. Overall, the hippocampus is believed to orchestrate the Each mouse was released from the south quadrant and was processes that are involved in constructing, integrating, allowed to swim freely for 30 sec. All trials were recorded and storing most kinds of information into long-term with a digital video camera placed above the maze for later memories. ELISA analyses using these mice revealed scoring. that the HGF protein level in the hippocampus of WT Cued task (days 7–8).. On days 7 and 8, the mice were mice was 21.92 6 1.98 ng/g tissue, whereas the level in tested in another pool, in which a visible platform (10 cm in di- HGF-Tg (1/2) mice was 55.41 6 7.58, and the level in ameter) was set. The pool was placed in another area of the HGF-Tg homozygous (1/1) mice was 95.14 6 12.65 same experimental room and enclosed by white featureless (Fig. 1A). These results demonstrate that the level of walls. The platform was elevated 0.5 cm above the water sur- HGF protein in the hippocampus increased in HGF-Tg face and was located in a different quadrant each day. Addition- (1/2)and(1/1) in a dose-dependent manner and that ally, the location of the platform was marked with a centrally HGF protein levels were about fivefold higher in HGF- mounted post (2.5 mm in diameter, 15 cm in height), which Tg (1/1) compared with WT mice. In the cerebral cor- was fitted with a black cube (5 3 5 3 5 cm) on the top. Other tex, the HGF protein level was also higher in HGF-Tg details, including pool size, water condition, and training pro- (1/1) mice compared with WT mice (WT; 44.92 6 cedures were identical to those for the place task. Analyses of 4.06 vs. HGF-Tg (1/1); 158.04 6 8.44;Fig.1B).To- the behavioral tests were performed in Ethovision 2.3.19 or 3.1 gether, these results demonstrate that HGF protein levels software (Noldus, Wageningen, The Netherlands). are increased by about four- to fivefold in the hippocam-

Journal of Neuroscience Research HGF Enhances Learning and Memory in Mice 1747

Fig. 1. Levels of HGF in the hippocampus and cerebral cortex of with antiphosphotyrosine or anti-c-Met antibody in the hippocam- WT and HGF-Tg mice at 22–26 weeks of age. A,B: HGF levels in pus. The right panel shows a quantitative graph. The ratio of phos- the hippocampus (A) and cerebral cortex (B) as analyzed by ELISA photyrosine c-Met (p-c-Met) protein levels per c-Met protein in were increased in HGF-Tg mice (WT, n 5 5; HGF-Tg (1/2), n HGF-Tg mice is expressed relative to ratios of WT mice. IP, immu- 5 4; (1/1), n 5 4). Data are expressed as mean 6 SEM. C: Repre- noprecipitation; IB, immunoblotting. Hippocampus (WT, n 5 4; sentative immunostaining of HGF (green) and counterstaining with HGF-Tg (1/2), n 5 3). Cerebral cortex (WT, n 5 3; HGF-Tg TO-PRO-3 (blue) in CA1 and CA3 regions of the hippocampus (1/2), n 5 3; HGF-Tg (1/1), n 5 3). Data are expressed as the and frontal cortex in wild-type mice and HGF-Tg (1/1) mice. D: mean 6 SEM. *P < 0.05, **P < 0.01, ***P < 0.001. HIP, hippo- Immunoprecipitation by c-Met and subsequent Western blotting campus; CTX, cerebral cortex; pY, phosphotyrosine. pus and cerebral cortex of HGF-Tg (1/1) mice com- was elevated in HGF-Tg (1/2) (1.64 6 0.35) and (1/ pared with WT mice. 1) mice (3.38 6 0.28; Fig. 1D). Thus, HGF-Tg mice We next assessed the localization of HGF protein in displayed increased HGF-c-Met signaling in the hippo- the hippocampus. HGF immunoreactivity (IR) was campus and cerebral cortex in comparison with WT detected predominantly in neurons of the CA3 region of mice. the hippocampus in WT mice, and a higher level of HGF was detected in neurons of CA1 and CA3 regions of the hippocampus and the frontal cortex in HGF-Tg Characterization of HGF-Tg Mice (1/1) mice compared with WT mice (Fig. 1C). In addi- Before behavioral testing of learning and memory tion to the localization in neurons, HGF-IR was detected performance, we briefly examined the morphology of in other areas compared with WT mice, including in the the hippocampus and cerebral cortex to exclude poten- extracellular space surrounding hippocampal and cerebral tial developmental abnormalities in HGF-Tg mice. Con- cortical neurons in HGF-Tg (1/1) mice. sistently with our previous report showing no differences We then examined whether the overexpression of in the muscle weight and spinal number HGF could contribute to the tyrosine phosphorylation between HGF-Tg and WT mice (Sun et al., 2002b), of c-Met (reflecting the activation of c-Met) as a mea- Nissl staining and anti-MAP2 (a neuronal marker) im- sure of intracellular signaling induction downstream of munostaining revealed no obvious differences between c-Met (Funakoshi and Nakamura, 2003). Immunopreci- HGF-Tg (1/1) and WT mice in the gross morphology pitation (IP) of the hippocampal and cerebral cortical tis- of the hippocampus (Fig. 2A,D). To assess further a sues with anti-c-Met antibody and subsequent immuno- potential modulation of glial cells, we performed immu- blotting (IB) with antiphosphotyrosine (pY) antibody nostaining for GFAP (astrocytes; red) and c-Met (HGF revealed that the ratio of tyrosine phosphorylation of c- receptor; green; Fig. 2B,C) as well as Iba1 (microglia; Met (phospho-c-Met) per total c-Met in the hippocam- red) and TO-PRO-3 (blue) in CA1 and CA3 regions of pus increased 2.7-fold in HGF-Tg (1/2) mice com- the hippocampus (Supp. Info. Fig. 1). c-Met-IR was pared with age-matched WT mice (Fig. 1D). In addi- only faintly detected (or was under the detection limit) tion, the phosphorylation of c-Met in the cerebral cortex in the populations of GFAP- and Iba1-positive glial cells.

Journal of Neuroscience Research 1748 Kato et al.

Fig. 2. No difference in gross morphology between WT and HGF-Tg (1/1) mice at 22–24 weeks of age. A,D: Nissl staining and MAP2 immunostaining were performed in the hippocampus (A) and cerebral cortex (D). Top panels: Nissl staining of paraffin-embedded coronal brain sections (7 lm). Bottom panels: MAP2 immunostaining. B,C: Immunostaining of c-Met (green) and GFAP (red), in CA1 (B) and CA3 (C) regions of the hippocampus of wild-type (WT) mice and HGF-Tg (1/1) mice.

Distributions and numbers of these glial cells were not 1.8%) compared with age-matched WT controls (68.2% different between WT mice and HGF-Tg (1/1) mice. 6 1.8%; Fig. 3A), whereas HGF-Tg mice exhibited a These observations demonstrate that HGF was success- decrease in locomotor activity compared with WT mice fully introduced into both the hippocampus (Figs. 1, 2) (1.05 6 0.03 cm/sec vs. 0.73 6 0.04 cm/sec; Fig. 3B). and the cerebral cortex (Figs. 1, 2) of adult HGF-Tg HGF-Tg mice exhibited a reduction in total arm entries (1/1) mice, without apparent or gross developmental in comparison with WT mice (Fig. 3C). These results modification of those tissues. demonstrate that adult HGF-Tg mice were superior to WT mice in terms of short-term spatial memory. Short-Term Spatial Memory in HGF-Tg and WT Mice Water Maze Task for the Analysis of Spatial We then assessed behavioral tests of learning and Learning in HGF-Tg Mice memory in HGF-Tg mice. In the present study, we uti- To assess further whether cognitive ability was lized HGF-Tg (1/1) mice because of their higher enhanced in HGF-Tg mice, we used the MWM proce- expression of HGF. First, the mice were tested for spon- dure, a hippocampus-dependent task that employs spatial taneous alternation behavior in a Y-maze. Spontaneous learning and memory (Morris, 1981). During this task, alternation is generally regarded as a measure of spatial animals learn the location of a hidden platform in a cir- working memory (Sarter and Bruno, 1998; Gerlai, 2001; cular pool by using distal cues. On average, the control Niimi et al., 2008). In the Y-maze, HGF-Tg mice dis- (WT) mice took 52.5 6 2.2 sec to find the platform on played significantly increased alternation rates (74.3% 6 the first day, and this latency was reduced to 24.3 6 2.9

Journal of Neuroscience Research HGF Enhances Learning and Memory in Mice 1749

Fig. 3. Enhancement of short-term learning and memory in a spontaneous alternation task at 16– 18 weeks of age. A: Alternations by HGF-Tg (1/1) mice in the Y-maze task were significantly elevated compared with those of WT mice. The results are expressed as mean 6 SEM (n 5 20). *P < 0.05. B: Locomotor activity in the Y-maze task was decreased in HGF-Tg (1/1) mice. ***P < 0.001. C: Total number of arm entries in the Y-maze task decreased in HGF-Tg (1/1) mice. ***P < 0.001.

Fig. 4. Enhanced performance in a spatial learning and memory task by HGF-Tg (1/1) mice at 20–22 weeks of age. Left: Escape latency in the hidden platform test of the Morris water maze (MWM) task. Data are expressed as mean 6 SEM (n 5 20). *P < 0.05, **P < 0.01. Right: Escape latency in the visible platform test of the MWM task. Data are expressed as mean 6 SEM (n 5 20). on the fifth day (Fig. 4, left). Comparatively, HGF-Tg better escape latency specifically in the hidden test, dem- (1/1) mice took 40.7 6 2.6 sec to find the platform onstrating their superiority in spatial learning. on the first day, and this latency was reduced to 15.4 6 2.4 sec on the fifth day (Fig. 4, left). The latency of escape to the hidden platform in HGF-Tg (1/1) mice Water Maze Task for the Analysis of Spatial was decreased over the training days (P 5 0.013 on the Memory in HGF-Tg Mice first day, P 5 0.025 on the fifth day). To assess whether We next used a transfer test to examine whether the superior behavior of HGF-Tg (1/1) mice in the the mice had formed a spatial memory of the task. The hidden-platform version of the water maze was due to incidence of arrival at the exact location of the platform some sensory-motor or motivational impairment, we was significantly higher in HGF-Tg (1/1) mice com- performed a visible test, in which the platform was easily pared with age-matched WT controls (Fig. 5A,B). recognizable. We found that HGF-Tg (1/1) mice and HGF-Tg (1/1) mice also spent more time in the tar- WT mice reached a similar level of optimal performance geted quadrant compared with WT mice (Fig. 5C; tar- (Fig. 4, right). Thus, HGF-Tg (1/1) mice displayed a geted quadrant 40.5%, P < 0.05), demonstrating that

Journal of Neuroscience Research 1750 Kato et al.

Fig. 5. Enhanced performance in spatial learning and memory tasks showed no preference for the target quadrant at this stage (n 5 20). by HGF-Tg (1/1) mice at 22–24 weeks of age. A,B: Representa- *P < 0.05. Data are expressed as mean 6 SEM. D: Swimming speed tive tracks of the transfer test in WT mice (A) and HGF-Tg (1/1) in the transfer test. There was no difference between WT and HGF- mice (B). C: Location preference in the transfer test conducted the Tg mice at the beginning of testing (0–20 sec; n 5 20). *P < 0.05, day after the fifth training session. HGF-Tg (1/1) mice spent more *P < 0.01. Data are expressed as mean 6 SEM. T, targeted; O, op- time in the target quadrant than other quadrants, whereas WT mice posite; R, right; L, left.

HGF enhanced long-term memory in the MWM task. is correlated with memory formation. In particular, the In addition, swimming velocities during the first 20 sec contribution of NR2B to learning and memory has been in the transfer test did not differ between HGF-Tg well characterized using transgenic mice that overexpress (1/1) and WT mice (Fig. 5D). Because WT mice dis- NR2B (Tang et al., 1999). Thus, we investigated whether play a higher swimming velocity at later time points, it enhancement of learning and memory in HGF-Tg seems unlikely that the shorter stay in the targeted area (1/1) mice correlates with a regulation of NMDARs, of WT mice is due to an impairment of swimming skill including NR2B subunit. In the hippocampus, NR1 lev- or motivation. It is instead postulated that WT mice els did not differ between WT and HGF-Tg (1/1)mice may rely on nonspatial information. Thus, HGF-Tg (Fig. 6A,B), whereas NR2A and NR2B levels were ele- (1/1) mice were more promptly able to develop a nav- vated in HGF-Tg (1/1) mice (Fig. 6A,C,D). In the cer- igational strategy in the MWM compared with WT ebral cortex, NR1 and NR2A levels exhibited no differ- mice, whereas HGF-Tg (1/1) and WT mice showed ence between HGF-Tg (1/1) and WT mice (Fig. the same level of performance in nonspatial learning 6B,C), whereas NR2B levels were increased in HGF-Tg tasks that rely on simple association strategies. Together, mice (Figs. 6D; data not shown). These results suggest our behavioral analyses (Y-maze and MWM) demon- that an upregulation of NR2B and NR2A at the protein strate that HGF-Tg mice display a selective and signifi- level in the hippocampus of HGF-Tg (1/1)miceis,at cant enhancement in both short- and long-term spatial least in part, involved in the modulation of learning and learning and memory. memory performance in HGF-Tg (1/1)mice.

Upregulation of N-Methyl-D-Aspartate Receptors Expression of Synaptic in HGF-Tg Mice in the Hippocampus and Cerebral Cortex of HGF treatment leads to a dose-dependent increase HGF-Tg Mice in total dendritic branch spine numbers and dendritic Activity-dependent neuronal plasticity is critical for complexity in mature cultures of hippocampal neurons memory formation. Activation of NMDA receptors in vitro (Tyndall et al., 2007). PSD-95 is an important (NMDARs) initiates a cascade of molecular events under- scaffold protein abundantly enriched in the postsynaptic lying synaptic plasticity. In the hippocampus, NMDARs density (PSD) of excitatory (Kim and Sheng, are heteromeric complexes composed of obligatory NR1 2004), and HGF is known to increase the number of subunit as well as NR2A and NR2B subunits (Petralia PSD-95 clusters in hippocampal neurons in vitro (Nakano et al., 1994). Many reports suggest that NMDAR activity et al., 2007). Therefore, we investigated whether HGF

Journal of Neuroscience Research HGF Enhances Learning and Memory in Mice 1751

Fig. 6. Effects of HGF overexpression on the levels of NMDA re- bIII-tubulin signal. B–D: Quantification of NMDA receptor expres- ceptor subunits in the hippocampus and cerebral cortex at 20 weeks sion in the hippocampus and cortex normalized to bIII-tubulin (WT, of age. A: Western blot analysis for NR1, NR2A, and NR2B in the n 5 6; HGF-Tg (1/2), n 5 3, HGF-Tg (1/1), n 5 6). Data are hippocampus of HGF-Tg (1/1) and (1/2) mice and WT mice. expressed as mean 6 SEM. *P < 0.05, **P < 0.01. HIP, hippocam- Equal amounts of protein were loaded in each lane, as shown by the pus; CTX, cerebral cortex. could regulate the levels of PSD-95 in vivo. A small levels for NGF, BDNF, GDNF, increase in PSD-95 was detected in the cerebral cortex, and VEGF in the hippocampus and cerebral cortex of but not in the hippocampus, of HGF-Tg (1/1) mice HGF-Tg (1/1) mice to exclude a potential modulation (Fig. 7A,B; data not shown), demonstrating a very lim- of neurotrophic factor mRNA levels. Our results ited effect of HGF on PSD-95 protein levels in the cere- showed no significant modulation of the mRNA levels bral cortex of HGF-Tg (1/1) mice in vivo. We next for these neurotrophic factors (data not shown). examined the levels of synapsin I, a presynaptic protein, in the hippocampus and cerebral cortex. Synapsin I is DISCUSSION the most highly expressed hippocampal presynaptic vesi- An understanding of the regulatory mechanisms cle-associated phosphoprotein (Nayak et al., 1996). underlying higher brain functions, such as learning and Western blot analyses revealed that there were no signif- memory performance, is a basic and critical issue in neu- icant differences in synapsin I levels in the hippocampus roscience. Only limited numbers of molecules have been and cerebral cortex among WT, HGF-Tg (1/2), and associated with such mechanisms. In the current study, (1/1) mice (Fig. 7A,C; data not shown). These results we asked whether HGF could play a role in regulating demonstrate that HGF does not seem to modulate total learning and memory performance under physiological protein levels of PSD-95 and synapsin I in the hippo- conditions, using the Y-maze and MWM as behavioral campus under the present experimental conditions. tests. Here, we provide the first in vivo evidence that HGF functions to enhance spatial learning and memory No Significant Difference in the Levels of performance in the adult brain. HGF-Tg mice displayed Neurotrophic Factor (NGF, BDNF, GDNF, and superior short-term and long-term spatial learning and VEGF) mRNAs in the Hippocampus and Cerebral 1 1 memory compared with age-matched WT mice, as evi- Cortex Between WT and HGF-Tg ( / ) Mice denced by their significantly increased alternation rates Neurotrophic factors have been shown to modulate in the Y-maze test and performance in acquisition and learning and memory performance (Gerlai et al., 2001; transfer MWM tests. Cao et al., 2004; Voikar et al., 2004; Nakajo et al., The molecular mechanisms by which HGF modu- 2008; Pertusa et al., 2008), so we quantified the mRNA lates learning and memory performance are not yet clear.

Journal of Neuroscience Research 1752 Kato et al.

Fig. 7. Expression of synaptic proteins in the hippocampus and cere- served as the loading control (lower lane). B,C: Quantification of bral cortex of HGF-Tg (1/1) and (1/2) mice and WT mice at 20 PSD-95 and synapsin I levels in the hippocampus and cerebral cortex weeks of age. A: Western blot analysis for PSD-95 (upper lane) and normalized to bIII-tubulin (n 5 3). Data are expressed as mean 6 synapsin I (middle lane) in the hippocampus. Equal amounts of pro- SEM. *P < 0.05. HIP, hippocampus; CTX, cerebral cortex. tein were loaded in each lane, and anti-bIII-tubulin Western blotting

However, the present results demonstrate that NR2B term memory lasting for 24–72 hr. Therefore, the func- and NR2A protein levels were increased in the hippo- tions of NR2A and NR2B receptors are differentially campus of HGF-Tg mice, and that NR2B levels were specialized in short-term and long-term memory. More- also increased in the cerebral cortex. This result is con- over, overexpression or knockout of the motor protein sistent with evidence showing that a decrease in the lev- KIF17 modulates both NR2B and NR2A and resulted els of surface NR2A and NR2B subunits following the in a behavioral pattern similar to that observed in HGF- addition of excitotoxic levels of NMDA is prevented by Tg (1/1) mice with regard to MWM results, such as HGF treatment in vitro (Akita et al., 2008). Transgenic the escape latency for the nonvisible test and the time in overexpression of NR2B in forebrain regions (such as quadrants in the transfer test (Figs. 4, 5C; Wong et al., the hippocampus and cerebral cortex) under direction of 2002, and Yin et al., 2011, respectively). In this regard, the CAM-kinase II promoter has been shown to our results suggest that modulation of NR2B and enhance learning and memory performance (Tang et al., NR2A subunits may partially explain the modulation of 1999; Cao et al., 2007), so our results suggest that learning and memory. Increased expression of NR2B increased levels of NR2B in the hippocampus and cere- may play a role in the enhancement of long-term mem- bral cortex might represent a molecular explanation for ory, whereas increased expression of NR2A protein in the enhancement of learning and memory performance the hippocampus may partially explain the enhancement in HGF-Tg mice. In addition to NR2B, the involvment of short-term memory (spatial acquisition) in HGF-Tg of NR2A in long-term potentiation (LTP) at hippocam- mice. pal CA1 synapses and spatial learning and memory (Saki- The postsynaptic protein PSD-95 is known to play mura et al., 1995) has been reported. Furthermore, dif- an important role in architecture and is also ferential roles of NR2A and NR2B in short-term and known to bind NR2B directly, so it is also possible that long-term memory have been reported for rats (Yashiro PSD-95 and presynaptic protein synapsin I levels are and Philpot, 2008; Jung and Suh, 2010). The latter modified in HGF-Tg mice to regulate learning and report indicated that forebrain NR2A expression was memory performance. However, this is unlikely, insofar involved in short-term memory, whereas forebrain as the present results show that levels of these synaptic NR2B expression was positively associated with long- proteins were not markedly altered. HGF has been

Journal of Neuroscience Research HGF Enhances Learning and Memory in Mice 1753 shown to modulate clustering of PSD-95 in the dendri- dall and Walikonis, 2007), we propose that HGF func- tic spine of primary hippocampal neurons (Tyndall and tions actively and locally in the nervous system (e.g., on Walikonis 2006; Nakano et al., 2007), so HGF may the surface of the dendritic spine in the hippocampus) contribute to the clustering of PSD-95, rather than regu- and plays a pivotal role in the modulation of learning lating PSD-95 protein levels, and this mechanism may in and memory performance in an activity-dependent man- turn contribute to the stabilization of NMDARs. ner. Another possible explanation for the enhancement It should be noted that locomotor activity was of learning and memory by HGF is the modulation of slightly altered during the Y-maze test in HGF-Tg mice hippocampal neurogenesis. Neurogenesis occurs in the (Fig. 3B). Similarly, the swimming speed of HGF-Tg postnatal brain in the subventricular zone and the den- mice at the beginning of the MWM task was the same tate gyrus (DG). The rate of neurogenesis within the as that of WT mice but was decreased in HGF-Tg mice DG can be altered under various physiological and path- at later times. The reason for this is not clear. However, ological conditions. Notably, neurogenesis within the our previous study revealed that motor neuron number, DG seems to be linked to hippocampus-dependent muscle weight, and rotorod performance were not dif- learning and memory (Nakazawa et al., 2004; Kee et al., ferent between WT mice and HGF-Tg mice, so HGF 2007; Kitamura et al., 2009). However, the role of neu- seems likely to affect speed that is independent of motor rogenesis in learning and memory is controversial (Jahol- performance ability. kowski et al., 2009). It has not yet been determined In summary, we provide the first in vivo evidence whether HGF contributes to hippocampal neurogenesis that HGF plays an important role in learning and mem- in the adult, but it has been reported that HGF acceler- ory under physiological conditions in the adult, and the ates neurogenesis in the subventricular zone in adults present results may open new avenues for the enhance- (Wang et al., 2011) and that HGF induces neurogenesis ment of learning and memory. Given the notion that in the brain after transient middle cerebral artery occlu- HGF also produces anxiolytic-like activity (Isogawa sion in rats (Shang et al., 2011). Thus the possibility that et al., 2005), we propose HGF as a new contributor to HGF modulates hippocampal neurogenesis is currently the activity-dependent regulation of higher brain func- under investigation. tions in the adult. Several neurotrophic factors, such as BDNF, have also been implicated in learning and memory perform- ACKNOWLEDGMENTS ance, and it is possible that their expression is modified We are grateful to Ms. Higano, Ms. Ikushima, Ms. in HGF-Tg mice. However, this is unlikely, because our Yoneda, and Ms. Ebisawa for secretarial assistance and to preliminary experiments showed that the mRNA levels Profs. K. Takakusaki at Asahikawa Medical University for various neurotrophic factors (including NGF, and S. Matsuyama at Himeji Dokkyo University for BDNF, GDNF, and VEGF) in HGF-Tg mice did not facilitating a collaborative and fruitful environment. differ from levels in age-matched WT mice (data not shown). Furthermore, although certain neurotrophic fac- tors, such as BDNF and VEGF, have been shown to REFERENCES modulate learning and memory performance, their char- Akimoto M, Baba A, Ikeda-Matsuo Y, Yamada MK, Itamura R, Nish- acteristic features are different from those of HGF. For iyama N, Ikegaya Y, Matsuki N. 2004. Hepatocyte growth factor as an example, mice with BDNF and VEGF overexpression enhancer of nmda currents and synaptic plasticity in the hippocampus. do not show superior spatial learning acquisition (Cao Neuroscience 128:155–162. et al., 2004; Nakajo et al., 2008). Given that stimulation Akita H, Takagi N, Ishihara N, Takagi K, Murotomi K, Funakoshi H, of hippocampal neurons with BDNF induces a rapid Matsumoto K, Nakamura T, Takeo S. 2008. Hepatocyte growth factor delivery of NR1- and NR2B-containing NMDA recep- improves synaptic localization of the NMDA receptor and intracellular signaling after excitotoxic injury in cultured hippocampal neurons. Exp tors to the plasma membrane with no increase in NR2A Neurol 210:83–94. observed after short incubation periods (Caldeira et al., Benkhoucha M, Santiago-Raber ML, Schneiter G, Chofflon M, Funa- 2007), our results suggest the possibility that the differ- koshi H, Nakamura T, Lalive PH. 2010. Hepatocyte growth factor ence in the effects of HGF on learning and memory per- inhibits CNS autoimmunity by inducing tolerogenic dendritic cells and 1 1 formance (compared with other neurotrophic factors) CD25 Foxp3 regulatory T cells. Proc Natl Acad Sci U S A may be due to a difference in the modulation pattern of 107:6424–6429. NMDARs by HGF and other neurotrophic factors such Caldeira MV, Melo CV, Pereira DB, Carvalho RF, Carvalho AL, Duarte as BDNF. However, we cannot exclude the possibilities CB. 2007. BDNF regulates the expression and traffic of NMDA recep- that neurotrophic protein levels are modulated or that tors in cultured hippocampal neurons. Mol Cell Neurosci 35:208–219. other neurotrophic factors not investigated here are Cao L, Jiao X, Zuzga DS, Liu Y, Fong DM, Young D, During MJ. 2004. VEGF links hippocampal activity with neurogenesis, learning and regulated in HGF-Tg mice. Further experiments are memory. Nat Genet 36:827–835. required to investigate these mechanisms. Given the Cao X, Cui Z, Feng R, Tang YP, Qin Z, Mei B, Tsien JZ. 2007. notions that HGF is released from hippocampal neurons Maintenance of superior learning and memory function in NR2B in an activity-dependent manner and that HGF induces transgenic mice during ageing. Eur J Neurosci 25:1815–1822. tyrosine phosphorylation and thereby activation of c- Frankland PW, Bontempi B. 2005. The organization of recent and Met localized on the surface of the dendritic spine (Tyn- remote memories. Nat Rev Neurosci 6:119–130.

Journal of Neuroscience Research 1754 Kato et al.

Franklin KBJ, Paxinos G. 1997. The mouse brain in stereotaxic coordi- Morris RGM. 1981. Spatial localization does not require the presence of nates. San Diego: Academic Press. local cues. Learn Motiv 12:239–260. Funakoshi H, Nakamura T. 2003. Hepatocyte growth factor: from diag- Mudo G, Persson H, Timmusk T, Funakoshi H, Bindoni M, Belluardo nosis to clinical applications. Clin Chim Acta 327:1–23. N. 1993. Increased expression of trkB and trkC messenger RNAs in Funakoshi H, Nakamura T. 2011. Hepatocyte growth factor (HGF): the rat forebrain after focal mechanical injury. Neuroscience 57:901– Neurotrophic functions and therapeutic implications for neuronal 912. injury/diseases. Curr Signal Transduction Ther 6:156–167. Nakajo Y, Miyamoto S, Nakano Y, Xue JH, Hori T, Yanamoto H. Funakoshi H, Belluardo N, Arenas E, Yamamoto Y, Casabona A, Pers- 2008. Genetic increase in brain-derived neurotrophic factor levels son H, Ibanez CF. 1995. Muscle-derived neurotrophin-4 as an activity- enhances learning and memory. Brain Res 1241:103–109. dependent trophic signal for adult motor neurons. Science 268:1495– Nakamura T, Mizuno S. 2010. The discovery of hepatocyte growth fac- 1499. tor (HGF) and its significance for cell biology, life sciences and clinical Gerlai R. 2001. Behavioral tests of hippocampal function: simple para- medicine. Proc Jpn Acad Ser B Phys Biol Sci 86:588–610. digms complex problems. Behav Brain Res 125:269–277. Nakamura T, Nawa K, Ichihara A. 1984. Partial purification and charac- Gerlai R, McNamara A, Choi-Lundberg DL, Armanini M, Ross J, terization of hepatocyte growth factor from serum of hepatectomized Powell-Braxton L, Phillips HS. 2001. Impaired water maze learning rats. Biochem Biophys Res Commun 122:1450–1459. performance without altered dopaminergic function in mice heterozy- Nakamura T, Nishizawa T, Hagiya M, Seki T, Shimonishi M, Sugimura gous for the GDNF mutation. Eur J Neurosci 14:1153–1163. A, Tashiro K, Shimizu S. 1989. Molecular cloning and expression of Honda S, Kagoshima M, Wanaka A, Tohyama M, Matsumoto K, Naka- human hepatocyte growth factor. Nature 342:440–443. mura T. 1995. Localization and functional coupling of HGF and c- Met/HGF receptor in rat brain: implication as neurotrophic factor. Nakano M, Takagi N, Takagi K, Funakoshi H, Matsumoto K, Naka- Brain Res Mol Brain Res 32:197–210. mura T, Takeo S. 2007. Hepatocyte growth factor promotes the num- Ishihara N, Takagi N, Niimura M, Takagi K, Nakano M, Tanonaka K, ber of PSD-95 clusters in young hippocampal neurons. Exp Neurol Funakoshi H, Matsumoto K, Nakamura T, Takeo S. 2005. Inhibition 207:195–202. of -inducing factor translocation is involved in protective Nakazawa K, McHugh TJ, Wilson MA, Tonegawa S. 2004. NMDA effects of hepatocyte growth factor against excitotoxic cell death in cul- receptors, place cells and hippocampal spatial memory. Nat Rev Neu- tured hippocampal neurons. J Neurochem 95:1277–1286. rosci 5:361–372. 21 Isogawa K, Akiyoshi J, Kodama K, Matsushita H, Tsutsumi T, Funakoshi Nayak AS, Moore CI, Browning MD. 1996. Ca /calmodulin-depend- H, Nakamura T. 2005. Anxiolytic effect of hepatocyte growth factor ent protein kinase II phosphorylation of the presynaptic protein synap- infused into rat brain. Neuropsychobiology 51:34–38. sin I is persistently increased during long-term potentiation. Proc Natl Jaholkowski P, Kiryk A, Jedynak P, Ben Abdallah NM, Knapska E, Acad Sci U S A 93:15451–15456. Kowalczyk A, Piechal A, Blecharz-Klin K, Figiel I, Lioudyno V, Niimi K, Takahashi E, Itakura C. 2008. Improved short-term memory Widy-Tyszkiewicz E, Wilczynski GM, Lipp HP, Kaczmarek L, Fili- and increased expression of NR2B observed in senescence-accelerated pkowski RK. 2009. New hippocampal neurons are not obligatory for mouse (SAM) P6. Exp Gerontol 43:847–852. memory formation; cyclin D2 knockout mice with no adult brain neu- Ohya W, Funakoshi H, Kurosawa T, Nakamura T. 2007. Hepatocyte rogenesis show learning. Learn Mem 16:439–451. growth factor (HGF) promotes oligodendrocyte progenitor cell prolifer- Jung YH, Suh YH. 2010. Differential functions of NR2A and NR2B in ation and inhibits its differentiation during postnatal development in the short-term and long-term memory in rats. Neuroreport 21:808–811. rat. Brain Res 1147:51–65. Kadoyama K, Funakoshi H, Ohya W, Nakamura T. 2007. Hepatocyte Pertusa M, Garcia-Matas S, Mammeri H, Adell A, Rodrigo T, Mallet J, growth factor (HGF) attenuates gliosis and motoneuronal degeneration Cristofol R, Sarkis C, Sanfeliu C. 2008. Expression of GDNF transgene in the motor nuclei of a transgenic mouse model of ALS. in astrocytes improves cognitive deficits in aged rats. Neurobiol Aging Neurosci Res 59:446–456. 29:1366–1379. Kadoyama K, Funakoshi H, Ohya-Shimada W, Nakamura T, Matsumoto Petralia RS, Wang YX, Wenthold RJ. 1994. The NMDA receptor subu- K, Matsuyama S, Nakamura T. 2009. Disease-dependent reciprocal nits NR2A and NR2B show histological and ultrastructural localization phosphorylation of serine and tyrosine residues of c-Met/HGF receptor patterns similar to those of NR1. J Neurosci 14:6102–6120. contributes disease retardation of a transgenic mouse model of ALS. Sakimura K, Kutsuwada T, Ito I, Manabe T, Takayama C, Kushiya E, Neurosci Res 65:194–200. Yagi T, Aizawa S, Inoue Y, Sugiyama H, et al. 1995. Reduced hippo- Kee N, Teixeira CM, Wang AH, Frankland PW. 2007. Preferential campal LTP and spatial learning in mice lacking NMDA receptor epsi- incorporation of adult-generated granule cells into spatial memory net- lon 1 subunit. Nature 373:151–155. works in the dentate gyrus. Nat Neurosci 10:355–362. Sara SJ. 2000. Retrieval and reconsolidation: toward a neurobiology of Kim E, Sheng M. 2004. PDZ domain proteins of synapses. Nat Rev remembering. Learn Mem 7:73–84. Neurosci 5:771–781. Kitamura T, Saitoh Y, Takashima N, Murayama A, Niibori Y, Ageta H, Sarter M, Bruno JP. 1998. Age-related changes in rodent cortical acetyl- Sekiguchi M, Sugiyama H, Inokuchi K. 2009. choline and cognition: main effects of age versus age as an intervening modulates the hippocampus-dependent period of associative fear mem- variable. Brain Res Brain Res Rev 27:143–156. ory. Cell 139:814–827. Shang J, Deguchi K, Ohta Y, Liu N, Zhang X, Tian F, Yamashita T, Kokaia Z, Metsis M, Kokaia M, Bengzon J, Elmer E, Smith ML, Tim- Ikeda Y, Matsuura T, Funakoshi H, Nakamura T, Abe K. 2011. Strong musk T, Siesjo BK, Persson H, Lindvall O. 1994. Brain insults in rats neurogenesis, angiogenesis, , and antifibrosis of hepato- induce increased expression of the BDNF gene through differential use cyte growth factor in rats brain after transient middle cerebral artery of multiple promoters. Eur J Neurosci 6:587–596. occlusion. J Neurosci Res 89:86–95. Lim CS, Walikonis RS. 2008. Hepatocyte growth factor and c-Met pro- Silva AJ, Paylor R, Wehner JM, Tonegawa S. 1992. Impaired spatial mote dendritic maturation during hippocampal neuron differentiation learning in alpha-calcium-calmodulin kinase II mutant mice. Science via the Akt pathway. Cell Signal 20:825–835. 257:206–211. Maina F, Klein R. 1999. Hepatocyte growth factor, a versatile signal for Silva AJ, Kogan JH, Frankland PW, Kida S. 1998. CREB and memory. developing neurons. Nat Neurosci 2:213–217. Annu Rev Neurosci 21:127–148.

Journal of Neuroscience Research HGF Enhances Learning and Memory in Mice 1755

Sun W, Funakoshi H, Nakamura T. 2002a. Localization and functional Wang TW, Zhang H, Gyetko MR, Parent JM. 2011. Hepatocyte role of hepatocyte growth factor (HGF) and its receptor c-met in the growth factor acts as a mitogen and chemoattractant for postnatal sub- rat developing cerebral cortex. Brain Res Mol Brain Res 103:36–48. ventricular zone-olfactory bulb neurogenesis. Mol Cell Neurosci Sun W, Funakoshi H, Nakamura T. 2002b. Overexpression of HGF 48:38–50. retards disease progression and prolongs life span in a transgenic mouse Wong RW, Setou M, Teng J, Takei Y, Hirokawa N. 2002. Overexpres- model of ALS. J Neurosci 22:6537–6548. sion of motor protein KIF17 enhances spatial and working memory in Tang YP, Shimizu E, Dube GR, Rampon C, Kerchner GA, Zhuo M, transgenic mice. Proc Natl Acad Sci U S A 99:14500–14505. Liu G, Tsien JZ. 1999. Genetic enhancement of learning and memory in mice. Nature 401:63–69. Yamada A, Matsumoto K, Iwanari H, et al. 1995. Rapid and sensitive Tyndall SJ, Walikonis RS. 2006. The receptor Met and enzyme-linked immunosorbent assay for measurement of HGF in rat its hepatocyte growth factor are clustered at excitatory synapses and human tissues. Biomed Res 16:105–114. and can enhance clustering of synaptic proteins. Cell Cycle 5:1560– 1568. Yashiro K, Philpot BD. 2008. Regulation of NMDA receptor subunit Tyndall SJ, Walikonis RS. 2007. Signaling by hepatocyte growth factor expression and its implications for LTD, LTP, and metaplasticity. Neu- in neurons is induced by pharmacological stimulation of synaptic activ- ropharmacology 55:1081–1094. ity. Synapse 61:199–204. Yin X, Takei Y, Kido MA, Hirokawa N. 2011. Molecular motor KIF17 Tyndall SJ, Patel SJ, Walikonis RS. 2007. Hepatocyte growth factor- is fundamental for memory and learning via differential support of syn- induced enhancement of dendritic branching is blocked by inhibitors of aptic NR2A/2B levels. Neuron 70:310–325. N-methyl-D-aspartate receptors and calcium/calmodulin-dependent ki- nases. J Neurosci Res 85:2343–2351. Zafra F, Hengerer B, Leibrock J, Thoenen H, Lindholm D. 1990. Activ- Voikar V, Rossi J, Rauvala H, Airaksinen MS. 2004. Impaired behaviou- ity dependent regulation of BDNF and NGF mRNAs in the rat hippo- ral flexibility and memory in mice lacking GDNF family receptor campus is mediated by non-NMDA glutamate receptors. EMBO J alpha2. Eur J Neurosci 20:308–312. 9:3545–3550.

Journal of Neuroscience Research