Activity-dependent neuroprotective protein (ADNP; ∼124 kDa) (Bassan et al., 1999; Zamostiano et al., 2001) is essential for brain development (Pinhasov et al., 2003). The highly conserved ADNP gene (Zamostiano et al., 2001) is abundantly expressed in the hippocampus (Bassan et al., 1999), cerebral cortex (Bassan et al., 1999; Zamostiano et al., 2001), and cerebellum (Zamostiano et al., 2001) and is modulated by injury (Zaltzman et al., 2004; Gozes et al., 2005b) and hormonal activity (Furman et al., 2005). Inhibition of ADNP expression results in cancer cell death (Zamostiano et al., 2001) and recombinant ADNP exhibits protection against severe oxidative stress in a neuronal cell model (Steingart and Gozes, 2006).

The deduced protein structure of ADNP contains nine zinc fingers, a proline-rich region, a nuclear bipartite localization signal, and a homeobox domain profile, implying a transcription factor function (Zamostiano et al., 2001). Affymetrix microarrays (Affymetrix, Santa Clara, CA) were used on ADNP knockout and control mouse embryos to reveal marked differences in expression profiles. A group of dramatically up-regulated gene transcripts in the ADNP-deficient embryos clustered into a family encoding for proteins associated with metabolism such as apolipoproteins, cathepsins, and metallothioneins, and a down-regulated gene cluster was associated with neurogenesis (Ngfr, neurogenin1, neurod1) and heart development (Myl2) (Mandel et al., 2007).

Within its homeobox profile, ADNP contains a nuclear export signal and an import signal, signifying possible extracellular functions. Furthermore, a potential interaction with cytoplasmic microtubules has been observed (Furman et al., 2004). Structure-activity studies identified a short peptide sequence in ADNP, NAP (NAPVSIPQ), that mimics the neuroprotective activity of the parent protein and crosses the blood brain barrier after systemic or intranasal administration (Gozes et al., 2005a). Importantly, NAP interacts with tubulin and enhances microtubule assembly (Divinski et al., 2004) to increase neurite outgrowth and to specifically protect neurons and glial cells against severe toxicities (Divinski et al., 2006). In vitro, NAP was found to reduce tau hyperphosphorylation (Gozes and Divinski, 2004) that has been associated with neurodegeneration/tauopathy and cognitive decline in vivo (Modrego, 2006). Two different lines of investigation were taken to ascertain specificity for NAP activity. Firstly, a scrambled NAP peptide was used in neurite outgrowth experiments in hippocampal cell in vitro showing essentially loss of activity (Smith-Swintosky et al., 2005). Secondly, NAP administration to rabbits (Gozes et al., 2005a) and repeated administration to rodents (Allon Therapeutics, personal communication) did not elicit any immune response. Based on these results and a clean toxicology profile, NAP is currently in phase II clinical trials targeting cognitive impairments (Allon Therapeutics Inc., Vancouver, BC, Canada). Because the homozygous ADNP knockout phenotype was found to be embryonic lethal (Pinhasov et al., 2003), the goal was set to assess the heterozygous phenotype to obtain a better understanding of neuroprotection in association with reduced expression of ADNP and to evaluate the potential ameliorative value of NAP therapy in an in vivo ADNP-deficient model.

Materials and Methods

Animals. All procedures involving animals have been approved by the Animal Care and Use Committee of Tel Aviv University and the National Institutes of Health (Bethesda, MD). ADNP heterozygous mice (Pinhasov et al., 2003) were housed in a 12-h light/12-h dark cycle facility, and free access to rodent chow and water was available.

ADNP^+/–Mice Generation Procedure. The procedure to generate ADNP^+/– animals was described previously (Pinhasov et al., 2003). In short, a bacterial artificial chromosome (pBeloBAC11, 7.3 kb) 129Sv mouse library was screened (Genome Systems Inc. St. Louis, MO), and a bacterial artificial chromosome clone containing the mouse ADNP gene was isolated. The X-pPNT vector (8231 bp) was used to generate a targeting vector for the ADNP gene. To construct the targeting vector, exons III, IV, and V of the ADNP gene and their adjacent introns were replaced with the neomycin resistance gene. Flanking regions of the ADNP gene, 3.3 kb from the 5′ end and 3 kb from the 3′ end, were preserved. A herpes simplex virus thymidine kinase gene was cloned 3′ of the targeting insert as a negative selection marker. The final targeting vector was 14 kb long. The linearized targeting vector was transfected into embryonic stem cells derived from 129/sv mouse for homologous recombination. Targeted positive clones were double selected with G418 and ganciclovir and screened by Southern blot hybridization outside the flanking homologous sequences. Positive clones were microinjected into C57BL/6-derived blastocysts and gave rise to chimeric offspring that carried the mutation into the germ line. Chimera males transmitted the mutation on breeding with C57BL/6 female mice. Generated heterozygous mice were subsequently crossed to produce ADNP^+/+, ADNP^+/–, and knockout (ADNP^–/–) progeny. The knockout progeny (ADNP^–/–) is embryonic lethal (Pinhasov et al., 2003).

Genotyping. Genomic DNA was extracted from mouse tails. Lysis buffer (100 mM Tris HCl, pH 8.5, 5 mM EDTA, 0.2% SDS, 200 mM NaCl, 0.35 mg/ml proteinase K) was added to each sample followed by a 16-h incubation period at 37°C and vigorous mixing. Genomic DNA was then precipitated by isopropanol, washed by 70% ethanol, dissolved in TE buffer (10 mM Tris-HCl, pH 8, 100 μM EDTA), and subjected to polymerase chain reaction (PCR) for genotyping.

NAP Treatment. NAP (NAPVSIPQ; Allon Therapeutics Inc.) was injected to newborn mice over a 2-week period. NAP was diluted to a final concentration of 25 μg/ml saline before injection (Bassan et al., 1999). Daily s.c. injections included 20 μl, 0.5 μg/mouse/day (days 1–4); 40 μl, 1 μg/mouse/day (days 5–10); and 80 μl, 2 μg/mouse/day (days 11–14), and exact saline quantities were injected to control mice. At the age of 2 months, male mice were subjected to behavioral testing (see below). In a separate experiment, NAP treatment included daily intranasal administrations over a 2-week period to 2- and 9-month-old male mice (0.5 μg/5 μl/mouse/day). For intranasal administration, the peptide was dissolved in a vehicle solution, in which each milliliter included 7.5 mg of NaCl, 1.7 mg of citric acid monohydrate, 3 mg of disodium phosphate dihydrate, and 0.2 mg of benzalkonium chloride solution (50%) (Alcalay et al., 2004). NAP or vehicle solution (DD) was administered to mice hand-held in a semisupine position with nostrils facing the investigator. A pipette tip was used to administer 2.5 μl/each nostril. The mouse was hand-held until the solution was totally absorbed (∼10 s). Nasal NAP application was performed daily, once a day, for 2 weeks (5 days a week). In the 2nd week, NAP was applied 1 h before the daily Morris water maze test (see below), which was conducted for 5 consecutive days.

Gene Array. All experiments were performed using Affymetrix MOE430A oligonucleotide arrays, as described in http://www.affymetrix.com/products/arrays/specific/mouse430a_2.affx. The precise experimental details for the ADNP^+/– embryos are outlined in a previous study (Mandel et al., 2007). In short, total RNA was extracted from four ADNP knockout embryos, four normal embryos, and six heterozygous embryos completely devoid of extra embryonic tissue. A pool of two genotype-identical litter mates was used on seven different arrays (pooling of two embryos was necessary to obtain sufficient RNA for the microarray analysis). Gene expression results were analyzed using the novel Expander software as before. From the 22,690 probes on the array, only 13,814 showed a significant signal (i.e., were present on the array and showed expression intensity > 20 arbitrary units), indicating that these genes were expressed at the E9 stage of embryonic development.

Tissue Culture Experiments. Experiments were conducted with glial cultures used obtained from newborn mouse cerebral cortex (ADNP^+/+ and ADNP^+/–) that were used as support cells to neuronal cells obtained from newborn rat brains as before (Bassan et al., 1999; Divinski et al., 2006). Neuronal survival was evaluated by direct neuronal counting (Bassan et al., 1999; Divinski et al., 2006). All experiments described below were performed on male mice.

RNA Isolation. Total RNA was isolated from frozen brains using the TriPure reagent (Roche Diagnostics, Basel, Switzerland) and/or the QIAGEN RNeasy mini kit (QIAGEN, Hilden, Germany).

Reverse Transcription and Quantitative Real-Time RT-PCR. Total RNA was treated by DNaseI (Ambion, Austin, TX), and 1 μgof RNA/sample was reverse-transcribed by SuperScript III reverse transcriptase (100 U; Invitrogen, Carlsbad, CA) using oligo(dT)18 primers (1 h at 50°C, 10 min at 75°C). Real-time RT-PCR was performed using FastStart DNA Master SYBR Green 1 dye-base detection kit (Roche Diagnostics). A Light Cycler instrument was used with its internal relative quantification software (Roche Diagnostics), which utilizes melting point analysis to assess the specificity of the amplified genes. All reactions were performed with a magnesium chloride concentration of 2.5 mM, primer concentrations of 0.05 μM, and 2.5 μl of the reverse transcription product in a 10-μl reaction mixture. Annealing temperature was 60°C for the ADNP transcript and for the internal mRNA standard, and encoding the ribosomal protein L19 Primer sequences for PCR analyses was according to previous publications (Mandel et al., 2007). The ADNP mRNA primers were: sense, 5′-CGAAAAATCAGGACTATCGG-3′; and antisense, 5′-TGAAAGTGCTGAGGCTGCTA-3′. The Pax6 primers used (a kind gift from Dr. Ruth Asheri-Padan, Tel Aviv, Israel) were: sense, 5′-ACTCCACCCGGCAGAAGATC-3′; and antisense, 5′-CCAGTCTCGTAATACCTGCCC-3′.

Protein Isolation. Mice were sacrificed by cervical dislocation. The cerebral cortex was isolated from each brain, weighed, and homogenized in 10 volumes of ice-cold lysis buffer (10 mM Tris-HCl, 150 mM NaCl, 20 mM NaF, 1 mM Na₃VO4, 2 mM EGTA, 0.5% Triton X-100, and 0.1% SDS) containing a protease inhibitor cocktail (Sigma, Saint Louis, MO). The suspended tissue was then sonicated using an ultrasonic cell disrupter (Microson; Misonix Inc., Farmingdale, NY). The resulting homogenate was centrifuged (3000g, 20 min, 4°C), the supernatant was then collected, and protein concentrations were determined by BCA-200 protein assay kit (Pierce, Rockford, IL).

Western Blot Analysis. Proteins (15 μg/lane) of the resulting supernatant (above) were separated by electrophoresis on a 10% polyacrylamide gel and electrotransferred to nitrocellulose filters followed by Western blot analysis (Zamostiano et al., 2001). Nonspecific antigen sites were blocked using a solution containing 5% (w/v) nonfat dried milk in 10 mM Tris, pH 8, 150 mM NaCl, and 0.05% Tween 20. For ADNP detection, the rabbit polyclonal ADNP antibody (BL1034; Bethyl Laboratories, Inc., Montgomery, TX) was used. The epitope recognized by this antibody maps to a region between residues 1050 and 1102 (C terminus) of ADNP (Zamostiano et al., 2001). Other antigens were detected using the following antibodies: 1) monoclonal (PHF-tau, AT-8; Innogenetics, Alpharetta, GA); 2) polyclonal (Anti-Tau [pT²³¹], Biosource, Camarillo, CA; AT180), both are mouse antibodies recognizing phosphorylated tau; 3) mouse monoclonal TAU-5 antibody that recognizes total tau (reacts with the nonphosphorylated as well as the phosphorylated forms of tau; Biosource); 4) polyclonal antibody to phosphor-GSK3α/β (pTyr279/216); 5) anti-GSK-3β [pS⁹], purchased from Biosource; and 6) monoclonal mouse actin antibodies (ImmunO, Aurora, OH). Peroxidase-conjugated affinity pure goat anti-mouse (Jackson Immuno Research Laboratories. Inc., West Grove, PA) and anti-rabbit IgG secondary antibodies (Sigma, Rehovot, Israel) were visualized by ECL Plus detection system (Amersham Biosciences, Buckinghamshire, UK). The ADNP antibody was diluted at 1:3000, and the AT-8 antibody was used at a dilution of 1:500, whereas other primary antibodies were diluted at 1:1000. The secondary anti-mouse antibody was used at a dilution of 1:10,000, and the secondary anti-rabbit antibody was used at a dilution of 1:25,000.

Assessment of Short-Term Spatial Memory in a Water Maze. Male mice were subjected to two daily tests in a water maze, including a hidden platform (Gozes et al., 2000). Water temperature was maintained at 23 to 24°C, bath water was changed every day, and water was made opaque by the addition of dry milk powder. Maximum duration of each daily test for each mouse (two daily trials) was 180 s. Every day for the first test, both the platform and the animal were situated in a new location with regard to the pool (with the pool being immobile). The experiment was performed as follows: the animal was positioned on the platform for 0.5 min, then placed in the water. The time required to reach the platform (indicative of learning and intact reference memory) was measured (first test). After 0.5 min on the platform, the animal was placed back in the water (in the previous position) for a second test and to search for the hidden platform (retained in the previous position). The time required to reach the platform in the second trial was recorded, indicative of short-term (working) memory. A maximal exploration time of 90 s/trial was allocated to each test. To avoid bias related to changes in motor activity in the various treatment groups, the last test included a visible platform trial that was performed 2 h after the previous test. Measurements were performed with the HVS video tracking system (HVS Image Ltd., Hampton, UK).

Assessment of Motor Behavior. Tests were performed in an 80-cm-diameter open field. Each male mouse (2 or 9 months old) was placed in the middle of the field, and the path traveled was measured the imaging system (above) for 3 min or 1 h.

Social Recognition. The social recognition test, based on the social memory test of Thor and Holloway as reproduced by Hill et al. (2007) assessed the time a male mouse spent sniffing and following an unfamiliar female mouse during repeated presentations at 30-min intervals. Four 3-month-old male control mice (ADNP^+/+) and three 3-month-old ADNP^+/– mice were tested individually by introducing them into a clean mouse cage containing an unfamiliar aged (18 months old) female mouse (female 1). Over a 5-min period, the experimenter recorded the amount of time the male mouse sniffed and followed the female mouse. The mice were then separated for 30 min followed by a second pairing of 5 min, during which the amount of time the male mouse sniffed and followed the female mouse was recorded. This was repeated three more times at 30-min intervals. After five pairings with the same female, the male was then paired with an unfamiliar aged (18 months old) female mouse (female 2), and the amount of time spent sniffing and following the unfamiliar female was recorded.

Tissue Preparation for Histology and Immunohistochemistry. Male mice 5 to 11 months of age were perfused transcardially under deep anesthesia with 4% paraformaldehyde in 0.1 M phosphate-buffered saline, pH 7.4. The brains were removed and further cut along the midline. The two brain hemispheres were dissociated and placed in the same fixative at 4°C for 4 h and then immersed in 30% sucrose in phosphate-buffered saline, pH 7.4, for cryoprotection. Brain hemispheres, in pairs, were frozen on block, and serial sagittal cryostat sections, 8 μm thick, were cut at five levels, 200 μm apart, initiating from the inner interface of each hemisphere and mounted on glass slides (Superfrost Plus).

Histology-Immunohistochemistry. A number of sections from each brain level were subjected to hematoxylin-eosin staining for the identification of morphological changes. Adjacent sections were stained with Fluoro-Jade B, a fluorescent chromofluor that selectively labels degenerating neurons (Anderson et al., 2005). In brief, the slides were first immersed in a solution containing 1% sodium hydroxide in 80% alcohol for 5 min. This was followed by 2 min in 70% alcohol and 2 min in distilled water. The slides were then transferred to a solution of 0.06% potassium permanganate for 10 min on a shaker table to ensure consistent background suppression between sections and further rinsed in distilled water for 2 min. The staining solution was prepared from a 0.01% stock solution of FluoroJade B (Chemicon, Temecula, CA) in 0.1% acetic acid vehicle in a concentration of 0.0004%, just before use. After 20 min in the staining solution, the slides were rinsed for 1 min in each of three distilled water washes. The slides were then placed on a slide warmer, set at approximately 50°C, until they were fully dry. The dry slides were cleared by immersion in xylene for at least 1 min before coverslipping with DPX (Fluka, Milwaukee, WI; or Sigma), a nonaqueous, nonfluorescent plastic mounting media, followed by thorough examination.

Additional sections were immunostained for the detection of phosphorylated tau-related pathology and astrogliosis. In brief, sections were rinsed in Tris-buffered solution, and following incubation with a blocking buffer, they were treated with primary antibodies against phosphorylated tau (AT-8, AT180; Innogenetics) or glial fibrillary acidic protein (GFAP; DakoCytomation Denmark A/S, Copenhagen, Denmark). AT8 and AT180 immunoreactivities were then visualized with a modified labeled streptavidin (LSAB) technique (Dako LSAB2 System Peroxidase), whereas for GFAP, incubation with goat anti-rabbit (Vector, Burlingame, CA) secondary antibody followed by avidin-biotin complex treatment were performed. The peroxidase reaction was visualized with 0.05% 3.3′-diaminobenzidine and 0.02% hydrogen peroxide. Sections were then counterstained with hematoxylin, dehydrated in ethanol, and cleared in xylene. In a number of sections, double immunohistochemistry for either AT8 or AT180 and GFAP was performed. In these cases, the first and second primary antibody reactions were visualized with 3.3′-diaminobenzidine and VIP (Vector Laboratories) as chromogens. Omission of the primary antibody was used for negative controls.

Quantitative Assessment of Neurodegeneration and Astrogliosis. In each group (ADNP^+/+ and ADNP^+/– male mice), the numbers of Fluoro-Jade-stained neurons in cortex and hippocampus, of both hemispheres, were determined as described previously (Sato et al., 2001) with some modifications. Analysis was performed on three adjacent sagittal sections obtained at the five brain levels. Fluoro-Jade-positive neurons were quantified by capturing the images from a number of visual fields in each section onto an Intel Pentium computer (IBM, Armonk, NY) using a digital camera (Nikon DS-5Mc-L1; Nikon, Tokyo, Japan) mounted on a microscope. In the cortex, 8 to 10 randomly selected visual fields were captured and the final magnification of the captured images was ×240. A square, 610 × 610 μm with 100 square subdivisions (6.1 × 6.1 μm), was centered in each captured image. In the hippocampus, images of CA1, CA3, CA4, and hillus areas were captured along the whole hippocampal area with a final magnification of 480×. Depending on the size of each area, two to four images were captured. In each image, a rectangle, 305 × 122 μm with 40 square subdivisions (3.05 × 3.05 μm), was positioned over the captured hippocampal area.

For all captured images, each square subdivision was counted if it contained at least one labeled element, defined as a labeled cell body and/or its processes. The tissue section, which exhibited the greatest number of Fluoro-Jade-positive structures, was selected for statistical analysis (Sato et al., 2001). The scores of the areas studied represented subjective assessments of fluorescence intensity and numbers of labeled cells, as follows: in the cortex, 0, no labeling observed in any square subdivision; 1 (+), 1 to 10 square subdivisions positive; 2 (++), 11 to 30 square subdivisions positive; 3 (+++), 31 to 60 square subdivisions positive; and 4 (++++), >60 square subdivisions positive; in the hippocampus, 0, no labeling observed in any square subdivision; 1 (+), 1 to 10 square subdivisions positive; 2 (++), 11 to 20 square subdivisions positive; and 3 (+++), >20 square subdivisions positive.

The number of AT8 and AT180 positive glial cells were assessed semiquantitatively using light microscopy in three serial sections adjacent to those selected for Fluoro-Jade B evaluation at all five preselected brain levels. A total of eight randomly selected microscopic fields per section, each of 22,500 μm² as defined by an ocular morphometric grid adjusted at the prefrontal lens, were evaluated. The scores of the areas studied represented subjective assessments of staining intensity and numbers of labeled cells, as follows: 0, no labeling observed; 1 (+), small number of cells that are only weakly labeled; 2 (++), moderate number of cells that are clearly labeled; and 3 (+++), large number of intensely labeled cells present (Simic et al., 2000).

All quantitative assessments were performed by two independent observers. In cases where significant discrepancies were obvious between the two observers, the evaluation was repeated by a third individual.

Statistical Analyses. Statistical tests used one-way analysis with pair-wise multiple comparison procedures (Student-Newman-Keuls method). Histological and immunohistochemical data used semiquantitative and nonparametric statistics.

Results

ADNP Partial Deficiency Results in Multiple Gene Expression Changes: Measurements during Pregnancy. To identify downstream target genes of ADNP, global gene expression profiles of whole E9 knockout embryos (knockout ADNP^–/–), their wild type (wild-type ADNP^+/+), and their heterozygous (heterozygous ADNP^+/–) littermates, were examined using the Affymetrix microarrays (Mandel et al., 2007). E9 embryos were chosen because it is the period by which embryos lacking ADNP exhibit distinct morphological and developmental changes just before degeneration and in utero absorption (Mandel et al., 2007). In contrast to ADNP^–/– embryos, the heterozygous embryos undergo normal embryogenesis, albeit with slight developmental delays, exhibiting viable phenotype (Pinhasov et al., 2003). Likewise, complete ADNP deficiency results in major changes in gene expression (Mandel et al., 2007). Here, the heterozygous phenotype exhibited minor changes in gene expression including up-regulation in a number of central gene products (Fig. 1A). These up-regulated genes are involved in key cellular functions including transcription, nervous system development, and control of DNA and dopamine metabolism as well as signaling pathways (Fig. 1B). These results suggested potential global effects for ADNP on mature brain activity, which may be deciphered with ADNP^+/– mice.

To further validate the array data and to identify potential changes in adult ADNP^+/– mice, we chose to follow up the expression of Pax6, a gene that showed increased expression in the ADNP^+/– embryos (Fig. 1, A and B) and that was found to be expressed in the adult subcortical domains (Stoykova and Gruss, 1994). Quantitative reverse transcription and real-time PCR was performed on individual RNA samples prepared from subcortical domains of 2-month-old male mice (ADNP^+/+, n = 4; ADNP^+/–; n = 4, each in triplicate). Results showed a significant 2.104 ± 0.267-fold increase in Pax6 expression in the ADNP^+/– mice (p < 0.004).

ADNP Partial Deficiency Results in Reduced Neuroprotective Potential: Measurements in Newborn Brains. Given the changes in gene expression that were observed above, the finding of ADNP as a glial protein (Bassan et al., 1999; Furman et al., 2004) and the neuroprotective activity of recombinant ADNP (Steingart and Gozes, 2006), it was of interest to evaluate the degree of protective support given by an astrocyte feeder layer derived from ADNP^+/– mice to neuronal cultures. Astrocytes were derived from newborn cerebral cortical tissues. Results showed a significant reduction in neuronal survival from 410 ± 22 neurons in the ADNP^+/+ cultures to 334 ± 20 in ADNP^+/– supported cultures (n = 4 independent experiments, each in triplicate; p < 0.01).

Reduced ADNP Expression in the Brains of 2-Month-Old ADNP^+/–Mice. Quantitative reverse transcription real-time PCR was utilized to analyze ADNP expression (Fig. 2A). Quantitative reverse transcription-PCR was performed on mRNA isolated from 2-month-old male mice tissues. A reduction of 42% in the cortex, of 38% in cerebellum, and of 50% in hippocampus was observed in the levels of ADNP mRNA in ADNP^+/– male mice compared with ADNP^+/+ mice (Fig. 2A). Northern blot hybridization (Bassan et al., 1999; Zamostiano et al., 2001) indicated similar results (data not shown).

At the protein level, ADNP expression was assessed by Western analysis (Furman et al., 2004) using actin as a control. As expected, results showed reduced expression of ADNP at the protein level in ADNP^+/– mice at all ages tested. An example for 9-month-old mice (cerebral cortex tissue) is shown for both control (densitometric scanning results in comparison with actin standardized at 100%, 100 ± 24, n = 3) and ADNP^+/– (34.7 ± 3, n = 3, p < 0.01; Fig. 2B).

Phosphorylated Tau Is Increased in the Brains of ADNP^+/–Mice and Is Reduced by NAP Treatment. As shown above, ADNP deficiency was associated with multiple changes in gene expression and reduced neuronal survival. Thus, tau phosphorylation (a major marker for neurodegeneration) was investigated in the cerebral cortex of the ADNP-deficient male mice in comparison with control mice (ADNP^+/+). Two antibodies (AT-8 and AT180) that recognize different sites of phosphorylation on tau, Ser202/Thr205 and Thr 231, respectively, were used, and results were calibrated against actin immunoreactivity as an internal standard. Two age groups were tested, 2- and 9-month-old mice. The 2-month-old mice were further divided into two groups, one injected daily with saline (s.c.) for the first 2 weeks of life and one similarly injected with NAP. Densitometric analysis of the Western blots at 2 months of age (Fig. 3A) showed an ∼40% increase in tau phosphorylation (Ser202/Thr205) in the ADNP^+/– mice compared with the ADNP^+/+ mice (p < 0.05). Daily injection of NAP for the first 2 weeks of life resulted in >2-fold reduction in tau phosphorylation (Ser202/Thr205) in the ADNP^+/– mice (p < 0.001) (Fig. 3A). The increase in tau phosphorylation in ADNP^+/– mice compared with ADNP^+/+ mice and the decrease after NAP treatment were also apparent when the results were calibrated against total tau immunoreactivity (tau5 antibody; p < 0.01) (Fig. 3A).

Because phosphorylated tau increased in the cortex of ADNP^+/– mice, the levels of the active [phosphor-GSK3α/β (pTyr279/216)] and the inactive [GSK3β (pSer9)] forms of GSK3, which regulate phosphorylation at specific sites of tau, were measured in the protein extracted from the cerebral cortex of 2-month-old male mice. Although there was only a small, statistically insignificant increase in the ratio of active/inactive GSK3β in ADNP^+/– mice compared with ADNP^+/+ mice, NAP treatment resulted in a significant (p < 0.05) ∼2-fold reduction in the ratio of active/inactive GSK3β in both ADNP^+/– and ADNP^+/+ mice (Fig. 3A).

The increased tau phosphorylation in ADNP^+/– mice as compared with ADNP^+/+ mice was maintained as the mice matured, as seen at 9 months of age (Fig. 3B). Similar results were obtained for both phosphorylation at (Ser202/Thr205) and (Thr 231) on tau. In comparison with the insignificant increase in the ratio of active/inactive GSK3β in 2-month-old ADNP^+/– mice, the older, 9-month-old mice showed a significant 50% increase in active GSK3β, as shown in the ratio of active/inactive enzyme (Fig. 3B, p < 0.01).

Neurodegeneration and the Appearance of Astrocytic Tau-Like Pathology in ADNP^+/–Mice. Male mice (5–11 months old) were analyzed for potential tau pathology at the anatomical level. Double labeling with phosphorylated tau antibodies and antibodies to GFAP for specific astrocyte staining were conducted. Results showed that among the ADNP^+/– male mice, there was a high prevalence of “thorn-shaped” astrocytes. Tau-related pathology was evident both in the soma and in the processes in a number of swollen astrocytes as well as in numerous isolated thread-like processes (Fig. 4, A–D; A and C, ADNP^+/+; B and D, ADNP^+/–) in the 11-month-old ADNP^+/– mice. In addition, the number of swollen astrocytes, regardless of whether they presented tau-related pathology or not, was higher in 10- to 11-month-old ADNP^+/– male mice compared with age-matched ADNP^+/+ animals. GFAP/AT8 or AT180-labeled astrocytes did not show significant differences (data not shown). Accordingly, GFAP/AT8 or AT180-labeled astrocytes with varying morphologies were counted, and results indicated that dystrophic astrocytes exhibiting tau-related pathology were found at significantly higher percentage in the ADNP^+/– animals as compared with ADNP^+/+ mice. Comparative analysis, performed according to the scale described under Materials and Methods, showed that the vast majority (percentage of visual fields) with normal-appearing astrocytes (grade 0) were higher among the ADNP^+/+ compared with ADNP^+/– male mice. In contrast, a higher number of grades 1 to 3 dystrophic astrocytes were found in the brain of ADNP^+/– animals as compared with ADNP^+/– mice (Fig. 4E, p < 0.001). The distribution of astrocytes was not specific; thorn- or tuft-like astrocytes were located at the parahippocampal area, thalamus, cerebellum, and brainstem. The age of onset for such an astroglial reaction was >8 months since at this age, a less intense staining either for AT8 or AT180 was detected in the ADNP^+/– mice compared with 11-month-old ADNP^+/– male mice.

The tau immunohistochemistry results that indicated neurodegeneration and astrocyte damage were corroborated by Fluoro-Jade B staining in both the hippocampus and the cortex of the ADNP^+/– mice (Fig. 5, A–H; A, C, E, and G, ADNP^+/+; B, D, F, and H, ADNP^+/–). Neurodegeneration increased with age and was most apparent at 11 months of age. Comparative analysis showed that the vast majority (percentage of visual fields) with normal-appearing neurons (grade 0) were higher among the ADNP^+/+ compared with ADNP^+/– male mice. In contrast, a higher number of grades 1 to 3 dystrophic neurons were found in the cortex of ADNP^+/– animals as compared with ADNP^+/– mice (p < 0.001, Fig. 5I). Similar results were found in the hippocampus (Fig. 5J, p < 0.001). Degenerative neurons were not evident before the 8th month of age.

ADNP^+/–Mice Exhibit Significant Spatial Learning Deficits. Three different studies were performed to assess potential spatial learning and memory deficits under the influence of the ADNP-deficient phenotype and possible reversal by the ADNP-derived neuroprotective peptide, NAP (Gozes et al., 2005a). In the first study, 2-month-old ADNP^+/– male mice were treated by intranasal administration of either vehicle or NAP (daily treatments for 2 weeks), and those were compared with vehicle-treated ADNP^+/+ male mice. In the second study, newborn mice were injected (s.c.) daily for 2 weeks with increasing concentrations of NAP or vehicle (Bassan et al., 1999). The third study included mice treated as in the second studies that were further subjected to NAP or vehicle treatment (daily intranasal administration) at the age of 9 months for 2 weeks. The first two groups were subjected to a Morris water maze at the age of 9 weeks. The 9-month-old mice were subjected to the Morris water maze at 9 months plus 1 week. Groups 1 and 3 continued to receive NAP during the 5 testing days of the water maze.

Behavioral assessments were performed in a water maze by measurements of the time required to find a hidden platform. Two daily tests were performed over 5 testing days. The platform location and the animal's starting point were held constant within each pair of daily trials, but the location of the platform and the animal's starting point were changed every day (Gozes et al., 2000). In the first daily test, indicative of reference memory, ADNP^+/– male mice were impaired compared with control animals (*, p < 0.05; **, p < 0.01; ***, p < 0.001; Fig. 6, A, C, and E). Furthermore, although the 2-month-old ADNP^+/+ mice learned the task after 3 testing days (#, p < 0.05; ##, p < 0.01; ###, p < 0.001), ADNP^+/– male mice did not learn the task (Fig. 6, A and B). At 9 months of age, the ADNP^+/+ did not show significant improvement in the latency to find the hidden platform in a water maze over a 5-day testing period (Fig. 6E). However, significant differences were apparent between the ADNP^+/+ mice and the ADNP^+/– mice, with the ADNP^+/– mice exhibiting increased latencies to find the hidden platform (Fig. 6E). NAP treatment significantly improved water maze performance as evidenced in studies 1 and 3 on the 5th testing day (Fig. 6, A and E). In study 2, NAP treatment significantly improved learning on the 3rd day, which was not apparent in the vehicle-treated ADNP^+/– mice (Fig. 6C).

In the second daily test, that evaluates short-term working memory, significant differences between ADNP^+/– and ADNP^+/+ male mice were apparent between the 2nd and 5th days of testing, depending on the study group (p < 0.05, Fig. 6, B, D, and F). In contrast to the ADNP^+/+ mice that learned the task, no significant learning curve was observed in the ADNP^+/– male mice at 9 months of age (**, p < 0.01, Fig. 6F). Similar results in the Morris water maze were obtained with 2-month-old mice that showed learning impairment (Fig. 6, B and D). NAP treatment significantly improved water maze performance as evidenced in studies 1 and 3 on the 5th testing day (Fig. 6, B and F). NAP treatment was also assessed in ADNP^+/+ mice showing minor changes at 2 months of age and significant improvement at 9 months of age as shown before (data not shown; Alcalay et al., 2004). On the last testing day, the mice were subjected to a visible platform test to assess their motor abilities. All tested mice found the visible platform over a 60-s test period. The mean platform finding time measured in seconds was 11.52 ± 1.65 (n = 21, 2-month-old ADNP^+/+ males, receiving saline for the first 2 weeks of life), 22.42 ± 5.39 (n = 14; 2-month-old ADNP^+/– males, receiving saline for the first 2 weeks of life), and 26.3 ± 8.3 s (n = 10; 2-month-old ADNP^+/– males, receiving NAP for the first 2 weeks of life), suggesting a significant difference between the ADNP^+/– and the ADNP^+/+ group at 2 months of age (p < 0.05), which was not ameliorated by NAP treatment. However, at 9 months of age with 2 weeks of NAP or vehicle nasal treatment, no significant differences were observed, and the mean platform finding time was 24.00 ± 9.58 (n = 8, ADNP^+/+ males), 26.6 ± 10.47 (n = 5; ADNP^+/– males), and 11.17 ± 4.6 s (n = 6; NAP-treated ADNP^+/– males). An additional test for motor performance was the open field test. Here also, no differences were observed between ADNP^+/– and ADNP^+/+ male mice (even when the test period was extended from 3 min to 1 h, data not shown); however, a statistically significant difference was found between 2- and 9-month-old mice, suggesting an age-dependent reduction in motor activity (Fig. 6G).

To strengthen the results regarding potentially reduced memory capabilities in the ADNP^+/– mice, a social recognition test was implemented. Olfactory cues are the predominant mechanism through which social familiarity develops (Hill et al., 2007), and results depicted in Fig. 6H showed that: 1) ADNP^+/– mice significantly differed from control, ADNP^+/– mice in their recognition of novel females (both females 1 and 2; p < 0.05). 2) Although both ADNP^+/– and ADNP^+/+ mice habituated to the novel mouse and sniffed and followed the female less with succeeding introductions through trials 1 through 5, the habituation was more rapid for the ADNP^+/+ mice showing a significant reduction already during the second trial with female 1 (p < 0.01), whereas the ADNP^+/– mice were slower and showed significant reduction in sniffing and following female 1 only on the third trial (p < 0.05 as compared with the first trial with female 1). 3) Furthermore, the ADNP^+/– mice exhibited an abnormal social memory response because, unlike control mice, they did not show renewed interest with the introduction of an unfamiliar female in trial 6. Although the ADNP^+/+ mice sniffed and followed the novel female significantly longer in trial 6 as compared with trial 5 (p < 0.01), there was no significant increase in the ADNP^+/– mice.

Discussion

The current study showed that ADNP^+/– mice exhibited reduced expression (∼50%) of brain ADNP and an increase in markers of neuronal degeneration that are associated, in part, with tauopathy. Furthermore, a decreased ability of the ADNP-deficient astrocytes to support neuronal survival was observed. Motor activity was similar in the ADNP^+/– male mice as compared with ADNP^+/+ male mice. In contrast, learning deficits in the Morris water maze were apparent at 2 months and were sustained at 9 months of age in ADNP^+/– male mice. These results were corroborated in a social interaction working memory test with 3-month-old male mice. Treatment with the ADNP-derived neuroprotective peptide NAP resulted in reduced tau hyperphosphorylation and increased performance in the Morris water maze that evaluates spatial learning and memory.

Studies in other laboratories indicated that deletions in the human ADNP chromosomal region (20q12-13.2; Zamostiano et al., 2001) may be associated with mental retardation in man (Borozdin et al., 2007). Most recently, a reduction in the ADNP mRNA levels was observed in the cerebral cortex and astrocytes from prenatal ethanol-exposed rat fetuses. Furthermore, cocultures of prenatal ethanol-exposed astrocytes with control neurons caused a marked decrease in neuronal growth, differentiation, and synaptic connections relative to the cocultures with control astrocytes, effects that were reverted by the addition of NAP (Pascual and Guerri, 2007). Other studies showed that prenatal NAP administration, in combination with ADNF-9 (SALLRSIPA), prevented the alcohol-induced spatial learning deficits and attenuated alcohol-induced proinflammatory cytokine increase in a model of fetal alcohol syndrome (Vink et al., 2005).

Decreased ADNP content was associated with changes in the expression of several gene products (Fig. 1, A and B), including Pax6. Our results indicate a complex association between Pax6 and ADNP expression because complete ADNP knockout resulted in a significant reduction in Pax6 expression in the primordial brain area (Pinhasov et al., 2003), whereas here, a partial deficiency in ADNP resulted in increases in Pax6 in the developing embryos as well as in subcortical domains of the adult brain. Pax6 has been associated in the specification of the subcortical domains of the evolutionary old limbic system (Stoykova and Gruss, 1994) and in neurogenesis in general. Increases in Pax6 expression may suggest aberrant developmental processes and potential compensatory mechanisms associated with ADNP deficiencies.

Recent findings have shown that ADNP may colocalize with microtubules in the cytoplasm of astrocytes (Furman et al., 2004). The tubulin cytoskeleton is intimately associated with neuronal structure and function (e.g., Gozes and Sweadner, 1981), and the pathology of tau, a microtubule-associated protein that loses proper functionality in the hyperphosphorylated state, has been implicated in multiple neurodegenerative diseases related to axonal dysfunction (Mandelkow et al., 2003). Intracellular accumulation of hyperphosphorylated tau as neurofibrillary tangles parallels memory disturbances in Alzheimer's disease (Tolnay and Probst, 1999). Disorders sharing similar pathological insoluble tau deposits, collectively called tauopathies, also include Down's syndrome, corticobasal degeneration, progressive supranuclear palsy, Pick's disease, Guam amyotrophic lateral sclerosis/Parkinson's dementia complex, and frontotemporal dementia with Parkinsonism chromosome 17-type. Furthermore, tau hyperphosphorylation has been observed in models of diabetes and suggested to contribute to the increased susceptibility to Alzheimer's disease that is observed in diabetic patients (Clodfelder-Miller et al., 2006).

Tau aggregates may be intrinsically toxic, killing neurons directly. Comparative protein analysis of the cerebral cortex of adult ADNP^+/– versus ADNP^+/+ mice indicated a significant increase in phosphorylated tau. Furthermore, NAP treatment protected against this increased tau phosphorylation. The target for NAP activity has been suggested to be tubulin, where NAP stimulates microtubule assembly and protects astrocytes (Divinski et al., 2004) and neurons (Divinski et al., 2006) against microtubule dysfunction and tau hyperphosphorylation in cell cultures (Gozes and Divinski, 2004). Recent studies have extended these findings to show that chronic intranasal NAP treatment reduces tau hyperphosphorylation in a mouse model of Alzheimer's disease, the triple transgenic model, that overexpresses both the amyloid β peptide and hyperphosphorylated tau (Matsuoka et al., 2007).

In the mammalian brain, different kinases, their regulators, and phosphatases form multimeric complexes with cytoskeletal proteins and regulate multisite phosphorylation from synthesis in the cell body to transport and assembly in the axon (Veeranna et al., 2000). Neurodegeneration interferes with the delicate homeostasis resulting in activation of hyperphosphorylation events (Wang et al., 2007). Overactivation of proline-directed kinases, such as cyclin-dependent kinase 5 and glycogen synthase kinase 3 (GSK3), has been implicated in the aberrant phosphorylation of tau at proline-directed sites (Plattner et al., 2006). Although cyclin-dependent kinase 5 was not implicated in ADNP-deficient pathology (data not shown), the ratio between the active and the inactive forms of GSK3β, was significantly increased in the cerebral cortex of the ADNP^+/– male mice (9 months old). It is thus hypothesized that deficits in ADNP will result in decreased astrocyte and neuronal function, leading to increases in active GSK3β toward tau hyperphosphorylation, neurodegeneration, and impaired cognition. NAP treatment that drives microtubules to polymerization and stabilization (Divinski et al., 2004, 2006) resulted here in normalization of GSK expression.

Morphological evaluations in adult male mice indicated that the ADNP-deficient male mice (>8 months) differ from the corresponding controls, exhibiting increased number of cortical and hippocampal neurons with degenerative features. These morphological changes support the significant spatial learning deficits displayed by the ADNP^+/– mice. In vitro, astrocytes derived from ADNP^+/– mice were already deficient in providing neuronal support at birth. These deficiencies translated into morphological changes at ∼8 months of age and were more pronounced among the older ADNP^+/– animals. ADNP^+/– mice exhibited a remarkable astroglial tau-related pathology. AT8- and AT180-positive material was found to be deposited in a cuff-like manner around astrocytes. In ADNP^+/– mice, a number of astrocytes appeared to have either a thorn- or tuft-like shape that may be caused by deposition of intensively labeled AT8- and/or AT180-positive material in the soma and proximal processes. Similar changes have been described in the aged human medial temporal lobe (Grundman et al., 2004). In addition, both the distribution throughout the brain as well as the shape of the affected astrocytes in ADNP^+/– mice share some similarities with the astroglial tau pathology noticed in a number of disorders. In particular, tuft-like astrocytes with tau-related pathology have been reported in progressive supranuclear palsy and tau-positive astrocytic plaques are considered pathognomonic for corticobasal degeneration (Forman et al., 2002; Fulga et al., 2007). Astrocytes play a dynamic role in CNS function, including maintenance of the blood-brain barrier, immune modulation, neurogenesis, synaptogenesis, and modulation of synapse function (Ransom et al., 2003). The astrogliosis observed in tauopathies may be associated with neurofibrillary tangle formation and not directly astrocytic tau pathology as tau expression in astrocytes is normally low (Togo and Dickson, 2002; Dabir et al., 2006).

Bioinformatic studies map the ADNP gene locus to chromosome 20q13.13-13.2 (Zamostiano et al., 2001) that is flanked by diabetes type II-linked genes (Klupa et al., 2000) (http://ensembl.rzpd.de/Homo_sapiens/contigview?c=20:48584983;w=50000.5). For example, the ADNP gene is separated by 3438 base pairs from DPM1 (D20S196; http://www.ncbi.nlm.nih.gov/genome/sts/sts.cgi?uid=57739) that has been linked to diabetes type II (p = 0.00010). Diabetes type II has been associated with poor cognitive performance and dementia, particularly in elderly patients. Based on the current results, the ADNP linkage to diabetes might be associated with cognitive outcome in the patients.

In conclusion, this paper describes that functional ADNP is important for neuroprotection in vivo through the tau phosphorylation cascade. The ADNP^+/– mouse offers a novel paradigm of cognitive deficits associated with tau pathology that may model neurodegeneration related to type II diabetes. Treatment with the drug candidate NAP enhanced cognition and significantly ameliorated deficiencies associated with ADNP knockdown. The significant preclinical promise in NAP is now being tested by Allon Therapeutics with two formulations in phase II clinical trials in Alzheimer's disease/mild cognitive impairment (intranasal), mild cognitive impairment associated with coronary artery bypass graft surgery (i.v.), and cognitive impairment associated with schizophrenia (i.v.).