Pyk2 in dorsal hippocampus plays a selective role in spatial memory and synaptic plasticity

We then examined whether the alteration in NOL observed in Pyk2 KO mice could be mimicked by the targeted deletion of Pyk2 in the dorsal hippocampus. We used Ptk2b^f/f mice and stereotactically injected adeno-associated virus (AAV) expressing either the recombinase Cre fused to GFP (AAV-GFP/Cre) or GFP alone (AAV-GFP) in the dorsal hippocampus at two injection sites. The AAV serotype (AAV1) and the human synapsin-1 (SYN1 gene) promoter controlling Cre expression preferentially target neurons^27,29. Three weeks after the injection, we assessed spatial memory with the NOL test. Twenty-four hours after the first exposure to the objects, Ptk2b^f/f mice injected with AAV-GFP retained their ability to distinguish the moved object (Fig. 2B). In contrast, the mice injected with AAV-GFP/Cre lost this ability (Fig. 2B). The mean difference in time spent exploring NL between AAV-GFP- and AAV-GFP/Cre-injected mice was − 12.9% (95.0% CI − 21.5, − 3.41) with a P value of the two-sided permutation t-test of 0.0092 (Supplementary Table 1). This result, in agreement with a previous report²¹, shows that targeted deletion of Pyk2 in the dorsal hippocampus is sufficient to alter performance in the NOL test. This result rules out developmental effects or indirect consequences of Pyk2 deletion in other brain regions or tissues. In addition, because the AAV serotype and the promoter of the viral vector used display cell type specificity, these results also provide information about the location of Pyk2 critical for the behavioral deficit, presumably dorsal hippocampus pyramidal neurons.

In these experimental conditions, we induced LTP using two different stimulating protocols, which have been previously used to assess the role of Pyk2 in hippocampal synaptic plasticity with contrasting results^21,22. The first protocol was a high-frequency stimulation (HFS) protocol (5 times 1-s stimulation at 100 Hz with intervals of 10 s²¹), and the second protocol was a theta-burst stimulation (TBS) protocol (10 bursts of 4 shocks at 100 Hz with an interburst interval of 200 ms²²). When we induced LTP in slices from wild type animals with the HFS protocol, we elicited a robust and persistent LTP (Fig. 3C,D). In contrast, when we stimulated slices from Pyk2 KO animals with HFS, we found a significantly reduced LTP (Fig. 3C,D, Supplementary Table 2). The mean difference in average EPSP slope at 50–60 min (expressed as % of baseline normalized slope) between Pyk2 WT and KO mice was − 22.6% (95.0% CI − 39.6, − 7.65), with a two-sided permutation t-test p value = 0.0068 (Fig. 3D, Supplementary Table 2). During the same experimental sessions and using different slices, including some from the same animals as for the HFS protocol, we induced LTP using the TBS protocol described above. With TBS we found no difference in the elicited LTP between slices from wild type or Pyk2 KO animals (Fig. 3E,F, Supplementary Table 3). The mean difference in average EPSP slope at 50–60 min between Pyk2 WT and KO mice was − 2.1% (95.0% CI, − 20.5, 16.8), with a two-sided permutation t-test p value = 0.86 (Fig. 3F, Supplementary Table 3). These results obtained with two different types of stimulation in the same mouse line show that the alteration in LTP induced by the absence of Pyk2 depends on the protocol used for inducing synaptic potentiation.

We then induced LTP in different slices with either HFS or TBS, the two protocols described above. We first monitored the fiber volley amplitude over time after HFS and did not observe any major change or difference between AAV-GFP- and AAV-GFP/Cre-injected mice (Fig. 4C), supporting the absence of presynaptic modification. As expected, in slices from Ptk2b^f/f animals injected with AAV-GFP, the HFS protocol induced a robust and persistent LTP (Fig. 4D) at levels comparable to wild type animals (Fig. 3C), showing the lack of effect of the presence of floxed alleles of Ptk2b gene, virus injection or GFP expression on this parameter. In contrast, LTP induced by HFS was markedly reduced in slices from Ptk2b^f/f mice injected with AAV-GFP/Cre (Fig. 4D, Supplementary Table 4). The mean difference in average EPSP slope at 50–60 min (expressed as % of baseline normalized slope) between slices of GFP- and GFP/Cre-expressing mice was − 46.9% (95.0% CI − 74.0, − 22.2), with a two-sided permutation t-test p value = 0.0064 (Fig. 4E, Supplementary Table 4). We also examined the effects of the TBS protocol in slices from Pyk2f.^/f animals injected with AAV-GFP or AAV-GFP/Cre. TBS elicited similar levels of LTP in slices from Ptk2b^f/f animals injected with AAV-GFP/Cre or AAV-GFP (Fig. 4F, Supplementary Table 5). The mean difference in average EPSP slope at 50–60 min between slices of GFP/Cre- and GFP-expressing mice was − 13.1% (95.0% CI − 45.9, 9.07), with a two-sided permutation t-test p value = 0.41 (Fig. 4G, Supplementary Table 5). These results showed that targeted deletion of Pyk2 in dorsal hippocampus CA1 neurons replicates the effects of constitutive knockout, with a selective deficit in LTP induced with a HFS protocol and no alteration with the TBS protocol. They show that the HFS-induced LTP modification is not the consequence of a developmental or indirect alteration. Moreover because the AAV expression detected by GFP fluorescence was expressed in CA1 neurons, mostly in pyramidal layer, and not in CA3, these results indicate that the alteration in HFS-induced LTP very probably results from a post-synaptic defect in these neurons.

All experiments were carried out in male mice. We used Ptk2b^−/− (Pyk2 KO) mice and mice with floxed Pyk2 alleles (Ptk2b^f/f) on a C57Bl/6 background. KO mice were bred by mating Ptk2b^+/− heterozygous mice. The Ptk2b^−/− (KO) and Ptk2b^+/+ (wild type) used for experiments originated from these breeding pairs. Mice were housed under conditions of controlled temperature (23 °C) and illumination (12-h light/12-h dark cycle, ZT0 = 7:00 a.m.). Animal procedures were authorized by and performed in accordance with the animal care committee’s regulations. The study was carried out in compliance with the ARRIVE guidelines (https://arriveguidelines.org↗) and in accordance with guidelines of Declaration of Helsinki and NIH (1985-revised publication no. 85-23, European Community Guidelines), and French Agriculture and Forestry Ministry guidelines for handling animals (decree 87849, license A 75-05-22), with approval of the ethical committee (agreement APAFIS#8861-2016111620082809).

For specific deletion of Pyk2 in the dorsal hippocampus, 6–10 week-old male Pyk2^f/f mice were stereotactically injected either with AAV expressing eGFP fused with the recombinase Cre, including a Woodchuck hepatitis virus posttranscriptional regulatory element (WPRE) to enhance expression, under the control of the human synapsin-1 gene promoter (AAV-GFP/Cre, 105540-AAV1, pENN.AAV1.hSyn.HI.eGFP-Cre.WPRE.SV40, Addgene, Massachusetts, USA) or with AAV expressing only GFP (AAV-GFP, AV-9-PV1917, AAV9.CamKII0.4.eGFP.WPRE.rBG from Perelman) as control. For re-expression of Pyk2 in the dorsal hippocampus of Pyk2 KO mice, the AAV used was AAV1-CamKIIα (0.4)-GFP-2AmPTK2B, polycistronically expressing GFP and Pyk2 separated by a 2A site (AAV-GFP/Pyk2, Vector Biolabs Malvern, PA, USA) or, as control, AAV-GFP, as above. Mice were anesthetized by an intraperitoneal injection of ketamine 15 mg kg⁻¹ and xylazine 3 mg kg⁻¹, and placed in a stereotaxic frame (Kopf Instruments, Sweden). The hippocampus was targeted bilaterally using a 32-gauge needle connected to a 1-ml syringe (Neuro-Syringe, Hamilton). Two injections were done on each side, at different locations in the dorsal hippocampus CA1 region with the following coordinates from the bregma (millimeters): AL − 1.8, ML ± 1.2, DV − 1.5, and AL − 2.3, ML ± 1.5, DV − 1.5. AAVs were delivered at a rate of 0.1 μl min⁻¹ for 5 min (0.5 μl per injection site). After the viral particles were infused, the needle was kept in place for 5 additional minutes before removal. After the procedure, the incision was closed with surgical sutures, and mice were placed in a heated cage until they recovered from anesthesia. After 2 h of monitoring, mice were returned to their home cage for 3 weeks before starting analyses of behavior and electrophysiology. To reduce suffering, mice received a subcutaneous injection of a non-steroidal anti-inflammatory drug (flunixin, 4 mg/kg) just after the surgery and on the next day, when necessary. Behavioral experiments were carried out 3 weeks later in all animals that had been stereotactically injected.

For novel object location (NOL) test we used an open-top arena (45 × 45 × 45 cm) under dim light (60–70-lx intensity). Mice (3–5-month old) were first habituated to the arena for 2 days for 15 min per day. To allow mice to identify their position, 4 different visual cues were placed around the arena: one circle, one square, one rectangle and one with the shape of an hourglass, all solid black on a white background. On day 3, two identical objects (Eiffel tower bronze figurines, painted in gray, about 9 cm-high and 3 cm in diameter) were placed in the arena and mice were left to explore the arena for 10 min. Twenty-four hours later, one of the objects was moved from its original location to the diagonally opposite corner and mice were allowed to explore the arena for 5 min (Fig. 1A). The mice were continuously recorded with a video tracker (Viewpoint, Newcastle Upon Tyne, UK) placed above the arena, and the percentage of time exploring the displaced object (new location, NL) and the unmoved object (old location, OL) was compared. Object exploration was defined as mice actively sniffing or touching the object with their nose or forepaws. Quantification was done manually with a stopwatch.

Male 2–6 month-old mice were anaesthetized by an intraperitoneal injection of ketamine 15 mg kg⁻¹ and xylazine 3 mg kg⁻¹. Following the achievement of full anesthesia and analgesia, mice where immobilized and a sagittal thoracic incision was made in order to expose the heart. An ice-cold low Ca²⁺/high Mg²⁺ artificial cerebrospinal fluid (aCSF) solution (pH 7.4, 310 mOsm, in mM, 92 NaCl, 26 NaHCO₃, 1 NaH₂PO₄, 3 KCl, 12.5 glucose, 20 Hepes, 0.2 CaCl₂, 7 MgCl₂,) was intracardially perfused for 2–3 min until the liver acquired a pale color. The mouse was then swiftly decapitated, the brain was removed and quickly placed in a custom-made 3D-printed brain matrix to rapidly excise the hemispheres with a 45°-angle, necessary for producing hippocampal transversal slices⁴⁴. Each hemisphere was placed on a Petri dish, hold in place with magnets on the vibrating slicer cutting block (Thermo Fisher, USA) and filled with ice-cold low Ca²⁺/high Mg²⁺ aCSF solution with constant carbogen (95% O₂/5% CO₂) bubbling. Slices (350 μm-thick) were cut (0.08 mm s⁻¹, blade vibration, 60 Hz and 1 mm) and individually moved in NMDG-aCSF solution (pH 7.4, 310 mOsm, in mM, 93 N-methyl-d-glucamine [glucamine, NMDG], 30 NaHCO₃, 1.2 NaH₂PO₄, 2.5 KCl, 25 glucose, 20 Hepes, 7 ascorbate, 4 pyruvate, 0.2 CaCl₂, 7 MgCl₂ (adapted from⁴⁵) for 10 min at 32 °C for recovery. Slices were then moved to a custom-made 3D-printed interface chamber⁴⁴ containing a modified aCSF solution (pH 7.4, 310 mOsm, in mM, 92 NaCl, 26 NaHCO₃, 1 NaH₂PO₄, 3 KCl, 12.5 glucose, 20 Hepes, 2 CaCl₂, 1.3 MgCl₂) with constant carbogen (95% O₂/5% CO₂) bubbling and allowed to rest for 2 h at room temperature.

After the 2-h rest, each slice was placed in a custom-made 3D-printed perfusion chamber⁴⁴, with constant flow of recording artificial CSF solution (aCSF, pH 7.4, 310 mOsm, in mM, 124 NaCl, 26 NaHCO₃, 1 NaH₂PO₄, 3 KCl, 12.5 glucose, 5 Hepes, 2.5 CaCl₂, 1 MgCl₂) at 32 °C with constant carbogen (95% O₂/5% CO₂) bubbling and let equilibrate for 30 min. The whole recording was carried out in this aCSF at 32 °C. In the experiments with AAV transduction the expression of GFP was visually verified and only the slices in which GFP expression was clearly visible in and restricted to CA1 were used. The stimulating and recording electrodes were made of silver wire encased in borosilicate glass pulled pipettes (pipette resistance: stimulating electrode 1–1.5 MΩ, recording electrode 5–6 MΩ) filled with aCSF. The electrodes were positioned in the hippocampal CA1 area, in the middle of the stratum radiatum, 300 μm apart from each other. Stimulating currents were delivered using and external ISO-Flexi stimulator (A.M.P.I., Israel) while field excitatory post-synaptic potential (fEPSP) acquisition was performed with a 700B amplifier (Molecular Devices, USA) connected to a 1440a digitizer (Molecular Devices, USA) using the Clampex 10 acquisition software (Molecular Devices, USA). To stimulate the Schaffer collaterals coming from the CA3 area, the stimulating electrode (positioned proximal, towards CA3) delivered a single depolarizing current pulse. A recording electrode was placed distal (towards the subiculum) within the stratum radiatum to record the fEPSPs elicited by Schaffer collaterals stimulation. The current amplitude was set to 50% of the current that elicited the maximal fEPSP amplitude (current injection 0.6–1.2 μA). Stimulating pulses (0.1 ms) were delivered every 20 s (0.033 Hz) and the resulting fEPSPs were recorded. After collecting a 20-min baseline, LTP was induced with either a high-frequency stimulation (HFS) protocol consisting of 5 tetanic bursts of 1 s at 100 Hz separated by 10-s intervals or with a theta-burst stimulation (TBS) consisting of 15 bursts of 4 spikes at 100 Hz separated by 200-ms intervals. After the induction, fEPSPs were monitored during the next 60 min and LTP was quantified by analyzing the slope of the activation phase. Only slices with stable recordings throughout the experiment and, in the case of those injected with AAV, only slices with good GFP expression in CA1, were included in the final analyses.

Tissue samples were sonicated in 10 g l⁻¹ SDS and placed at 100 °C for 5 min. Extracts (15 μg protein) were separated by SDS–PAGE and transferred to nitrocellulose membranes (GE Healthcare, Chicago, IL, USA). Membranes were blocked in TBS (150 mM NaCl, 20 mM Tris–HCl, pH 7.5) with 0.5 ml L⁻¹ Tween 20 TBS-T) with 30 g l⁻¹ BSA. Immunoblots were probed with rabbit polyclonal antibodies for Pyk2 (#P3902, Sigma-Aldrich) diluted 1:1000. Membranes were incubated with the primary antibody overnight at 4 °C by shaking in TBS. After several washes in TBS-T, blots were incubated with secondary anti-rabbit IgG DyLight™ 680 conjugated antibodies (1:10,000; Rockland Immunochemicals, Pottstown, PA, USA). Secondary antibody binding was detected by Odyssey infrared imaging apparatus (Li-Cor Inc., Lincoln, NE, USA).

Data were analyzed with d’Agostino–Pearson and Shapiro–Wilk normality tests. When their distribution did not significantly deviate from normal distribution and the number of replicate was sufficient (> 7) we used Student’s t test for comparing two groups. Comparison of 4 groups were done with 2-way ANOVA and time courses were analyzed with 2-way repeated measures ANOVA, as indicated. Post hoc multiple comparisons after ANOVA were done with Holm-Sidak’s test. All tests were used as two-tailed. Statistical analyses were done with GraphPad Prism 6. Estimation statistics was done as recommended⁴⁶ using the https://www.estimationstats.com/↗ website⁴⁷. Statistical analyses are reported in Supplementary Tables 1–5.

Supplementary Information.