Selective expression of the neurexin substrate for presenilin in the adult forebrain causes deficits in associative memory and presynaptic plasticity

Experimental Neurology 347 (2022) 113896 Available online 15 October 2021 0014-4886/© 2021 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Research paper Selective expression of the neurexin substrate for presenilin in the adult forebrain causes deficits in associative memory and presynaptic plasticity Ana C. S´ anchez-Hidalgo a , b , 1 , Francisco Arias-Arag´ on a , b , M. Teresa Romero-Barrag´ an c , Celia Martín-Cuevas a , b , 1 , Jos´ e M. Delgado-García c , Amalia Martinez-Mir a , Francisco G. Scholl a , b , * a Instituto de Biomedicina de Sevilla (IBiS), Hospital Universitario Virgen del Rocío/CSIC/Universidad de Sevilla, Avda. Manuel Siurot s/n, Sevilla 41013, Spain b Departamento de Fisiología M´ edica y Biofísica, Facultad de Medicina, Universidad de Sevilla, Avda. S´ anchez Pizju´ an, 4, Sevilla 41009, Spain c Division of Neurosciences, Pablo de Olavide University, Seville 41013, Spain ARTICLE INFO Keywords: Neurexin Presenilin Alzheimer's disease Synapse Synaptic plasticity Memory ABSTRACT Presenilins (PS) form the active subunit of the gamma-secretase complex, which mediates the proteolytic clearance of a broad variety of type-I plasma membrane proteins. Loss-of-function mutations in PSEN1/2 genes are the leading cause of familial Alzheimer's disease (fAD). However, the PS/gamma-secretase substrates relevant for the neuronal deficits associated with a loss of PS function are not completely known. The members of the neurexin (Nrxn) family of presynaptic plasma membrane proteins are candidates to mediate aspects of the synaptic and memory deficits associated with a loss of PS function. Previous work has shown that fAD-linked PS mutants or inactivation of PS by genetic and pharmacological approaches failed to clear Nrxn C-terminal fragments (NrxnCTF), leading to its abnormal accumulation at presynaptic terminals. Here, we generated transgenic mice that selectively recreate the presynaptic accumulation of NrxnCTF in adult forebrain neurons, leaving unaltered the function of PS/gamma-secretase complex towards other substrates. Behavioral characterization identified selective impairments in NrxnCTF mice, including decreased fear-conditioning memory. Electrophysiological recordings in medial prefrontal cortex-basolateral amygdala (mPFC-BLA) of behaving mice showed normal synaptic transmission and uncovered specific defects in synaptic facilitation. These data functionally link the accumulation of NrxnCTF with defects in associative memory and short-term synaptic plasticity, pointing at impaired clearance of NrxnCTF as a new mediator in AD. 1. Introduction Neurexins (Nrxns) form a large family of presynaptic adhesion proteins that couple neurotransmitter release and trans-synaptic interactions with postsynaptic ligands (Dean et al., 2003; Südhof, 2017). Genetic studies and analysis of protein levels in peripheral samples from patients suggest a role for altered Nrxn levels in preclinical and clinical stages of Alzheimer's disease (AD) (Duits et al., 2018; Goetzl et al., 2018; Lle´ o et al., 2019; Martinez-Mir et al., 2013). Nrxns are subject to regulation at several levels. At the transcriptional level, the use of alternative promoters together with extensive alternative splicing generates hundreds of isoforms that differ at the extracellular domain (Gomez et al., 2021; Schreiner et al., 2014; Treutlein et al., 2014). Proteolytic cleavage is emerging as a new post-translational mechanism that regulates Nrxn function (Saura et al., 2011; Servi´ an-Morilla et al., 2018). Proteolytic regulation of Nrxns proceeds in two sequential steps. First, a cleavage in the juxtamembrane region releases the specific ectodomain of the processed isoform and generates a common membrane-bound Nrxn C-terminal fragment (NrxnCTF) (Saura et al., 2011). Subsequently, NrxnCTF is cleared by Presenilins (PS), the active subunit of the gamma-secretase complex, which mediates the intramembrane cleavage of a number of substrates (Bot et al., 2011; Saura et al., 2011; Servi´ an-Morilla et al., 2018). Importantly, mutations in PSEN1/2 genes are the main cause of familial AD (fAD) explaining about 90% of fAD cases (Cacace et al., 2016). Several findings support that fAD-linked PSEN mutations act trough * Corresponding author at: Instituto de Biomedicina de Sevilla (IBiS), Campus del Hospital Universitario Virgen del Rocío, Avda. Manuel Siurot s/n, Sevilla 41013, Spain. E-mail address: [email protected] (F.G. Scholl). 1 Present address: Spanish Network for Research in Mental Health (CIBERSAM), Monforte de Lemos AV, 3–5, Madrid 28029, Spain. Contents lists available at ScienceDirect Experimental Neurology journal homepage: www.elsevier.com/locate/yexnr https://doi.org/10.1016/j.expneurol.2021.113896 Received 2 July 2021; Received in revised form 27 September 2021; Accepted 10 October 2021 Experimental Neurology 347 (2022) 113896 2 a loss-of-function mechanism (De Strooper, 2007; Shen and Kelleher 3rd, 2007; Wolfe, 2007). In a seminal study, Shen and collaborators showed that conditional mice lacking PS1/2 genes in adult forebrain neurons (PScKO mice) develop key symptoms resembling AD, including early and progressive deficits in synaptic plasticity and memory, followed by late synapse loss, neurodegeneration and hyperphosphorylation of Tau proteins (Saura et al., 2004). Biochemical evaluation of fAD-associated gamma-secretase complexes and the generation of mouse models recapitulating fAD-linked PSEN1 mutations have further supported the loss-of-function hypothesis for PS in AD (Sun et al., 2017; Xia et al., 2015; Zhou et al., 2017). As such, loss of PS function results in decreased gamma-secretase activity inducing the accumulation of its proteolytic substrates in a context-dependent manner. Thus, the accumulation of a particular substrate coinciding with the onset of symptoms helps to define potential candidates associated with impaired PS function. However, the high number of substrates for PS/gamma-secretase challenges the identification of single candidates with a functional relevance in the loss of PS function (Güner and Lichtenthaler, 2020; Haapasalo and Kovacs, 2011). Therefore, the PS/gamma-secretase substrates that can produce key synaptic and memory defects associated with a loss of PS function are not completely known. Nrxns are candidates to mediate aspects of the synaptic and behavioral deficits caused by loss of PS/gamma-secretase function. It has been previously shown that chemical or genetic inhibition of PS/gammasecretase activity results in the accumulation of NrxnCTF in vitro and in vivo (Bot et al., 2011; Saura et al., 2011). Moreover, the clearance of NrxnCTF is inhibited in cells expressing fAD-associated PSEN1 mutations, suggesting that NrxnCTF accumulates in fAD (Bot et al., 2011; Saura et al., 2011). In vivo, NrxnCTF accumulates at presynaptic terminals of PScKO mice at a time coinciding with the onset of synaptic and behavioral deficits but prior to neurodegeneration (Saura et al., 2011). Importantly, the sole expression of NrxnCTF in cultured hippocampal neurons while maintaining PS function mimics the deficits in neurotransmitter release and calcium influx caused by the absence of PS1/2 genes or gamma-secretase activity (Servi´ an-Morilla et al., 2018). Although these findings suggested a role for the impaired synaptic clearance of NrxnCTF in the loss of PS function, the deficits produced by the brain accumulation of NrxnCTF are currently unknown. In this study, we generated transgenic mice that selectively accumulate NrxnCTF in adult forebrain neurons. Biochemical characterization showed that NrxnCTF mice recapitulate the distribution of NrxnCTF found in conditional mice lacking PS1/2 function in the forebrain. However, PS/gamma-secretase function towards other substrates is not affected in NrxnCTF mice. Importantly, we found that the selective accumulation of NrxnCTF produces behavioral deficits, including decreased fear-conditioning memory. Electrophysiological recordings in alert behaving mice revealed impaired short-term synaptic plasticity at mPFC-BLA synapses of NrxnCTF mice, in line with the presence of a putative presynaptic mechanism. These data indicate that the impaired synaptic clearance of NrxnCTF caused by a loss of PS function leads to synaptic plasticity and behavioral deficits in mature brain. 2. Materials and methods 2.1. Generation of transgenic NrxnCTF mice The HA-NrxnCTF construct contains a signal peptide followed by a HA tag and the last 85C-terminal residues of human Nrxn1. A DNA fragment containing the TRE promoter, the coding sequence of HANrxnCTF and a WPRE fragment (woodchuck hepatitis virus posttranscriptional regulatory element) was injected into the pronucleus of FVB/N zygote for transgenic mouse production. TRE-NrxnCTF mouse lines were mated with CaMKII α -tTA mice (Mayford et al., 1996) in a C57BL/6 J background. Doxycycline (DOX) was provided in the diet (SAFE, 40 mg/kg). PScKO tam mice were generated by crossing fPS1/ fPS1;PS2−/−with CaMKII α -CreERT2 mice (Erdmann et al., 2007). Generation and characterization of fPS1/fPS1;PS2−/−mice have been previously described (Saura et al., 2004; Shen et al., 1997). Animals were kept at 22 ◦C on a 12 h dark/light cycle and food and water were provided ad libitum. Mice were used according to animal care standards and all protocols were approved by the Committee of Animal Use for Research at the University of Seville (Spain). 2.2. Biochemical analysis Forebrain tissues were homogenized in lysis buffer (50 mM Tris-HCl pH 7.4; 100 mM NaCl; 5 mM MgCl2; 1% Triton X-100 and 0.1% SDS) containing a protease inhibitor cocktail (Roche). Synaptosome fractions were isolated as previously described (Carlin et al., 1980; Saura et al., 2011). In brief, tissues from 5-6-month-old mice (three animals per preparation) were homogenized with a Teflon-glass homogenizer in buffer A (5 mM HEPES pH 7.4, 0.32 M sucrose, 1 mM NaHCO3, 1 mM MgCl2, 0.5 mM CaCl2, 0.1% NP-40) supplemented with protease and phosphatase inhibitors cocktail (Sigma). The resulting homogenate (L fraction) was centrifuged at 1400g for 10 min. The supernatant was saved (S1) and the pellet was further homogenized in buffer A and centrifuged at 710g for 10 min. The supernatants (S1 and S2) were combined and centrifuged (13,800 g for 10 min). The pellet was resuspended in buffer B (6 mM Tris HCl pH 8.0, 0.32 M sucrose, 1 mM NaHCO3, 0.1% NP40, with phosphatase and protease inhibitors), homogenized (M fraction) and loaded on top of an ice-cold discontinuous sucrose gradient (1.2 M, 1 M and 0.85 M) and centrifuged at 82,500 g for 2 h. Synaptosome fraction (S fraction) was collected from the interface between 1 M and 1.2 M layers, diluted in buffer C (12 mM Tris pH 8.0, 1% Triton X-100, with phosphatase and protease inhibitors) and incubated on ice for 15 min. The suspension was centrifuged at 32,800 g for 1 h. The supernatant was collected (presynaptic fraction) and the pellet was resuspended and sonicated in buffer D (40 mM Tris pH 8.0, 1% NP40, with phosphatase and protease inhibitors) to obtain the postsynaptic fraction. All steps were performed at 4 ◦C. Fractions were aliquoted, frozen in liquid N 2 and stored at −80 ◦C. Protein concentration was determined with the BCA protein assay kit (Pierce). Western blot experiments of lysates containing equal-protein loading (15-40 μ g per lysate) were performed using the following primary antibodies: rat anti-HA (1:500, Roche), rabbit anti-Neurexin 1,2,3 (1:500, Synaptic Systems), mouse anti-PSD95 (1:1000, ThermoScientific), mouse anti-SNAP25 (1:2000, Sigma Aldrich) and mouse anti-β actin (1:5000, Sigma Aldrich). Immunoreactivity was detected with appropriate secondary antibodies conjugated with horseradish peroxidase (1:5000, Jackson ImmunoResearch). Chemiluminescence was developed using Clarity ECL Substrate (Bio-Rad) or Clarity Max ECL Substrate (Bio-Rad) on an ImageQuant LAS4000 Mini (GE Healthcare Life Sciences). 2.3. Behavioral analysis Behavioral studies were performed in two independent cohorts of male NrxnCTF and littermate control (wild-type, TRE-NrxnCTF and CaMKII α -tTA) mice of the same age (5–8 months). To elude early transgene expression, all mice were fed with diet with DOX from gestation to 30 days after birth. Mice were caged in the behavioral room for 30 min before the behavioral examination. Tests were carried out from 9 a.m. to 6 p.m. All behavioral tasks and quantification analysis were performed by researchers blind to the genotype of the mice. 2.4. Open field test The tested mouse was allowed to freely explore a square open field arena (45 ×45 cm, Harvard Apparatus) for 15 min. Automatic detection of the mouse, total travelled distance and time spent in the central zone (20 cm apart from the walls) were recorded with the Smart software A.C. S´ anchez-Hidalgo et al. Experimental Neurology 347 (2022) 113896 3 (PanLab). 2.5. Self-grooming Stereotyped behavior was studied with the self-grooming test. Mice were placed in a clean cage without bedding. After 10 min of habituation, behavior was video-recorded for 10 min. Time spent in selfgrooming and numbers of bouts were manually quantified. 2.6. Three-chamber test The test was performed as described previously (Silverman et al., 2010) with minor modifications. A social interaction box (Harvard Apparatus) divided in 3 compartments was used. The social arena was made of a transparent box (42 ×60 cm) with two transparent sliding doors that divide left, right and center chambers (42 ×20 cm). In the first 10-min session, the tested mouse was placed in the central chamber with the sliding doors closed. Then, a 10-min habituation session proceeded where the tested mouse had access to the sided chambers, containing two empty circular cages (8 cm diameter). After habituation, circular cages housing an unfamiliar C57BL/6 J mouse of the same sex and age or an inanimate object (plastic cube) was located in the corners of the left or right chambers. A cylindrical bottle filled with water was placed on top of the enclosures to prevent the tested mouse from climbing. The tested mouse was located in the central chamber and allowed to explore the arena for 10 min. The location of the cages was alternated between tests. Tests were video-recorded and the number of transitions between compartments was automatically registered with the Smart software. Time spent in close contact with the cages was manually measured. 2.7. Novel object recognition and Barnes maze tests The tested mouse was placed in the center of the open field arena and allowed to freely explore the area for 15 min. Following habituation to the arena, two identical objects were placed in opposite corners, 15 cm apart from the walls. The tested mouse was placed back in the center of the field and allowed to explore the area for 10 min. After 24 h, one of the familiar objects was replaced for a new object of different size, colour and shape. Then, the tested mouse was placed in the arena and allowed to explore the area for 10 min. The location of the new object was alternated between assays. The test was video-recorded and the time spent interacting with the objects was manually measured. The Barnes maze assay was performed as previously described (Andrade-Talavera et al., 2015). Briefly, mice were located on top of an elevated round platform (Harvard Apparatus) delimited by 20 evenly spaced holes in the periphery. To assess spatial memory, visuo-spatial cues were located hanging on panels in the proximity surrounding the platform. During learning, all holes but one, the escape hole, were covered and mice were exposed to the platform. Bright light and standing fans serve as motivating factors to induce scape behavior. The escape hole is maintained at a fixed location for the duration of training, which involves four daily trials (maximum time to find the escape hole: 180 s per trial). In unsuccessful trials, mice were gently guided to the escape hole at the end of the trial. Twenty-four hours and eight days after training, mice were tested in a single session of 90 s in the same setup with all holes covered. The time in reaching the escape hole and spent in each quadrant was automatically quantified (SMART software). 2.8. Fear conditioning test The contextual and auditory fear-conditioning test was evaluated in a test box (Harvard Apparatus). Mice were allowed to freely explore the test box for 2 min before they were exposed to conditioning phase consisting of three repetitions of a 30 s tone co-terminating for 2 s with a mild electric shock (0.2 mA). After 24 h, mice were placed in the same conditioning room for 5 min (contextual phase). Then, mice were exposed to three repetitions of the tone but in a different shaped chamber and without a shock. Freezing response during the tests was detected with sensors located in the floor. Packwin software was used to automatically analyze the freezing response. 2.9. Surgery For electrode implantation, mice were anaesthetized with 1–2% isoflurane, supplied from a calibrated Fluotec 5 (Fluotec-Ohmeda, Tewksbury, MA, USA) vaporizer, at a flow rate of 1–2 L/min oxygen (AstraZeneca, Madrid, Spain) and placed into a stereotaxic frame. Additional anesthesia was delivered by a special mouse mask (David Kopf Instruments, Tujunga, CA). Animals were implanted with a bipolar stimulating electrode in the medial prefrontal cortex (mPFC) (1.8 mm rostral to Bregma; 0.3 mm lateral; and 2 mm deep from the brain's surface) (Franklin and Paxinos, 2007) and a bipolar recording electrode in the basolateral amygdala (BLA) (1.8 mm posterior to Bregma; 3 mm lateral; and 3.9 mm deep from the brain's surface). Electrodes were made of 50 μ m, Teflon-coated, tungsten wires (Advent Research, Eynsham, UK) and their tip were bared of their isolating cover for 0.5 mm. Two screws, each one connected to a bare silver wire, were affixed to the skull as a ground. The six wires were soldered to a 6-pin socket (RS Amidata, Madrid, Spain), which was covered and fixed to the brain with dental cement. Further information about this experimental procedure is detailed elsewhere (Gruart et al., 2006). After surgery, animals recovered for a week before the start of the recording sessions. 2.10. Stimulation and recording procedures Recording sessions were carried out in 6 mice at a time. Animals were placed in individual small (5 cm ×5 cm ×10 cm) plastic chambers located inside a larger Faraday box (30 cm ×30 cm ×20 cm). Field excitatory postsynaptic potentials (fEPSPs) were evoked with the help of Cibertec CS20 stimulators (Cibertec, Madrid, Spain) and recorded with Dagan Corporation EX4-400 Quad Differential amplifiers (Dagan Corporation, Minneapolis MN USA) at a bandwidth of 0.1 Hz-10 kHz, through a high-impedance probe (2 ×10 12 Ω, 10 pF). 2.11. Paired-pulse facilitation and input/output curves Input/output curves were tested with paired pulses (40 ms of interstimulus interval) of increasing intensities (0.02–0.4 mA, in steps of 0.02 mA). Animals received ten pulses for each intensity, which were averaged in order to measure the evoked fEPSP amplitude. For the paired-pulse facilitation test, 17 mice were stimulated with paired pulses of different inter-stimulus intervals (10, 20, 40, 100, 200 and 500 ms; ten pulses each) with intensities corresponding to 40% of the intensity necessary to evoke a saturating fEPSP response. 2.12. Long-term potentiation Baseline recordings were carried out for 15 min prior to the LTP induction. Field EPSPs were evoked with paired (40 ms inter-stimulus interval) 100 μ s squared, biphasic pulses at a rate of 3/min (0.05 Hz). Pulse intensity was set at 40% of the amount necessary to evoke a maximum fEPSP response (0.05–8.0 mA) (Gruart et al., 2006; Gureviciene et al., 2004). LTP was induced by applying a high-frequency stimulation (HFS) protocol, which consisted of five 100 ms trains of pulses at a frequency of 200 Hz presented at a rate of one per second. This protocol was presented 6 times (1/min). In order to avoid evoking large population spikes and/or the appearance of EEG seizures, the stimulus intensity during the HFS was set at the same value as that used for generating baseline recordings. None of the animals used in this study presented any after-discharge or motor seizure following the HFS protocol, as checked by on-line EEG recordings and visual observation of A.C. S´ anchez-Hidalgo et al. Experimental Neurology 347 (2022) 113896 4 the stimulated mouse (Madro˜ nal et al., 2009). After the induction, fEPSPs were recorded with the same procedure of the baseline along four sessions (1 h in day 1 and 30 min in days 2–4). Field EPSP amplitude was measured in the recordings before and after evoking long-term potentiation (LTP) and represented as a percentage of the base line (taken as a 100%), calculating the average of 15 pulses for each 5 min interval. 2.13. Histology At the end of the recording sessions, mice were deeply anaesthetized (sodium pentobarbital, 50 mg/kg) and perfused transcardially with 0.9% saline followed by fixation with 4% paraformaldehyde in phosphate-buffer (PFA). Brains were removed from the skull and maintained in PFA overnight at 4 ◦C and cryoprotected with an increasing gradient (5–30%) of sucrose in PBS. 50 μ m brain slices was obtained with a cryotome (Leica, Wetzlar, Germany). Brain sections with the structures of interest were mounted on gelatinized glass slides and stained using a Nissl technique (0.1% Toluidine Blue). Micrographs of the electrode scar were taken with a light microscope. 2.14. Data collection and analysis Behavioral tests were video-recorded and manually or automatically quantified and data were analyzed using one-way ANOVA with Tukey post-hoc test. Field EPSPs were stored digitally on a computer through the analogue/digital converter CED 1401 Plus (CED, Cambridge, England), at a sampling frequency of 5 kHz with an amplitude resolution of 16 bits. Data were analyzed off-line for fEPSP amplitude with the help of CED Spike 2 and Signal (Systat Software, San Jose, CA, USA) programs. For this, 15 successive fEPSPs were averaged every 5 min to quantify EPSP amplitudes. Results were represented with Microsoft Excel (Microsoft, Redmon, WA, USA) and CorelDraw (Corel Corporation, Ottawa, Canada) programs and computed and processed for statistical analysis using the Sigma Stat for Windows package. Regression analyses were used to study the relationship between LTP and paired-pulse variables (Student's t-test). Unless otherwise indicated, data are represented as the mean ±SEM. Acquired data were analyzed using a two-way ANOVA, with sessions as repeated measure. Contrast analysis was added for a further study of significant differences. 3. Results 3.1. Generation of mice expressing NrxnCTF in the forebrain The accumulation of NrxnCTF in mice lacking PS1/2 genes in the adult forebrain suggests a contribution of the impaired processing of Nrxns in the neuronal defects caused by a loss of gamma-secretase function. However, the broad accumulation of substrates following inhibition of PS/gamma-secretase activity avoids the identification of single substrates with a functional impact. To face this limitation, we have developed transgenic mice that selectively accumulate NrxnCTF in Fig. 1. Generation of NrxnCTF transgenic mice. A) Schematic drawings showing the proteolytic processing of Nrxns by metalloproteases and PS (left) and the HAtagged NrxnCTF transgene used in this study (right). The regions recognized by the HA and the cyto-Nrxn antibodies are indicated. LNS, Laminin/Neurexin/Sex hormone binding-globulin domain; TM, transmembrane region. B) Representation showing the genetic approach for the expression of NrxnCTF in the forebrain. C) Expression of NrxnCTF in forebrain tissues of NrxnCTF mouse lines. NrxnCTF expression was analyzed in cortex, hippocampus and striatum by Western-blot experiments with HA and cyto-Neurexin antibodies, as indicated. Lysates from single transgenic CaMKII α -tTA and TRE-NrxnCTF mouse lines were analyzed as additional controls. A.C. S´ anchez-Hidalgo et al. Experimental Neurology 347 (2022) 113896 5 the adult forebrain (Fig. 1). Mutagenesis analysis has previously showed that the last 85C-terminal residues of Nrxn1 represent the NrxnCTF substrate for PS/gamma-secretase (Servi´ an-Morilla et al., 2018). Therefore, we generated a cDNA encoding NrxnCTF tagged at the Nterminus with an HA epitope to allow detection (Fig. 1A). The NrxnCTF construct was cloned under the control of the tetracycline-responsive element (TRE) (Fig. 1B). Two transgenic TRE-NrxnCTF mouse lines were obtained by nuclear injection, lines B and C. Transgene expression from the TRE promoter requires the co-expression in the same cells of the tetracycline transactivator (tTA). In PScKO mice, conditional deletion of PS genes in forebrain is achieved by the activity of a CaMKII α promoter (Saura et al., 2004). Similarly, we chose a CaMKII α -tTA allele to drive NrxnCTF expression in forebrain neurons (Mayford et al., 1996). Thus, we aimed at accumulating NrxnCTF with the same distribution than a loss of PS function would produce in forebrain neurons, but leaving PS/gamma-secretase activity unaltered (Fig. 1B). Double transgenic CaMKII α -tTA;TRE-NrxnCTF mice (termed NrxnCTF mouse lines for simplicity) were born at the expected ratio. Expression of NrxnCTF was analyzed in lysates of forebrain regions by Western-blot experiments. The HA antibody recognized a major band of 14 kDa in lysates from hippocampus, cortex and striatum of NrxnCTF mice of line B (Fig. 1C). However, NrxnCTF was undetectable in the hippocampus and cortex of line C, and low expression levels were found at the striatum of the same line (Fig. 1C). A similar pattern of NrxnCTF expression was found in both mouse lines with an antibody that recognizes the common cytoplasmic tail of Nrxns, although a minor band of a lower mobility was more clearly detected in the transgenic mouse lines (Fig. 1C). In contrast, the HA antibody did not detect ectopic NrxnCTF expression in single transgenic mice (Fig. 1C), indicating the absolute requirement of the CaMKII α -tTA driver for transgene expression. Notably, the expression pattern of NrxnCTF in line B recapitulates the accumulation of endogenous NrxnCTF in mice lacking PS function in the forebrain (Saura et al., 2011) (see below). Based on these data, we selected NrxnCTF mice from line B for further experiments. 3.2. Synaptic expression of NrxnCTF in adult forebrain It has been reported that the CaMKII α -tTA driver can promote transgene expression during preor early postnatal development (Nicholls et al., 2008; Roberts et al., 2009). This early expression might result in confounding effects when the impact of the transgene in the mature brain is upon evaluation (Rodgers et al., 2012), such in this study. We took advantage of the temporal-regulation provided by the Tet-off system to strictly limit the expression of NrxnCTF to adult forebrain. With this aim, we mated single transgenic CaMKII α -tTA and TRENrxnCTF mice in the presence of diet with DOX and maintained the parental mice and their progeny in this diet until weaning, at postnatal day 30 (P30) (Fig. 2A). This time point matches the time at which PS gene deletion was reported in PScKO mice (Saura et al., 2004). As shown in Fig. 2B, this strategy results in the adult onset expression (>P30) of NrxnCTF at similar levels than untreated NrxnCTF mice. Then, we interrogated if transgenic NrxnCTF recapitulates the expression at presynaptic terminals of synaptosome preparations previously described for the endogenously produced NrxnCTF in PScKO mice (Saura et al., 2011). In Western blot experiments of synaptosome preparations, transgenic NrxnCTF was found at synaptic fractions, where it concentrates at presynaptic, but not postsynaptic, terminals of cortex and hippocampus (Fig. 2C). The distribution of the synaptic markers PSD95 and SNAP25 within the postand presynaptic fractions, respectively, confirmed the isolation protocol (Fig. 2C). These data indicate successful accumulation of NrxnCTF at presynaptic terminals of adult forebrain neurons once the formation of the synaptic circuitry is completed. Fig. 2. Adult onset expression and distribution in synaptosome fractions from forebrain regions of NrxnCTF. A) Temporal control of NrxnCTF expression with DOX. B) Western-blot experiments showing NrxnCTF expression in adult forebrain of transgenic mice at different times after DOX removal at P30 (on/off). For comparison, NrxnCTF expression of untreated mice (No DOX) was analyzed for comparison. C) Synaptosome preparations from cortex and hippocampus of NrxnCTF mice were incubated with HA, SNAP25 and PSD95 antibodies, as shown. L, lysate; SN, supernatant; Mb, membrane; S, synaptosome; Pre, presynaptic fraction and Post, postsynaptic fraction. HA antibody detects the expression of NrxnCTF in presynaptic fractions of forebrain tissues. A.C. S´ anchez-Hidalgo et al. Experimental Neurology 347 (2022) 113896 6 3.3. Normal PS/gamma-secretase activity in NrxnCTF mouse Our approach based on the selective accumulation of NrxnCTF aims at preserving PS levels and gamma/secretase function towards other substrates. To obtain experimental support for this hypothesis, we analyzed PS1 levels and the expression of representative PS substrates in lysates from NrxnCTF mice. To establish suitable comparisons for these experiments, we obtained mice with loss of PS function in adult forebrain neurons. For that, we mated PS1f/f;PS2−/−mice with transgenic CaMKII α -CreERT2 mice to obtain PS1f/f;PS2−/−;CaMKII α -CreERT2 mice. Then, Cre activity was induced with Tamoxifen at P30 to generate PScKO tam mice. As a surrogate for the loss of PS function, we first analyzed the production of APP-CTF in PScKO tam mice at different times after treatment with Tamoxifen. As shown in Fig. 3A, APP-CTF levels clearly increased in PScKO tam mice from one to four months after treatment, indicating effective loss of PS function. Then, we selected for comparison PScKO tam and NrxnCTF mice at 4.5 months, allowing for the same time after PS gene deletion and NrxnCTF expression, respectively. The levels of PS1 were reduced in the forebrain of PScKO tam mice compared with controls, as expected (Fig. 3B, C). The remaining PS1 levels detected in PScKO tam mice likely represent PS1 expression in CaMKII α -promoter inactive neurons and in non-neuronal cells (Saura et al., 2004). By contrast, the expression of PS1 was not altered in NrxnCTF mice (Fig. 3B, C). Moreover, whereas N-Cadherin-CTF and APP-CTF accumulated in PScKO tam mice, neither of these PS/gammasecretase substrates showed accumulation in NrxnCTF mice. These data suggest that the levels and the proteolytic activity of the PS/ gamma-secretase complex towards its substrates are not broadly affected by the expression of NrxnCTF. As a further characterization of the mouse model, we decided to compare the expression of transgenic NrxnCTF with NrxnCTF produced in PScKO tam mice. For that, Western blot experiments were performed with an antibody that recognizes the common cytoplasmic tail of Nrxns (Fig. 1). As expected, loss of PS function in PScKO tam mice produced the accumulation of NrxnCTF in cortex and hippocampus (Fig. 3B, C). Importantly, the cyto-Nrxn antibody detected NrxnCTF proteins in Fig. 3. Normal PS function in NrxnCTF mice. A) Accumulation of APP-CTF in hippocampal and cortical lysates of PScKO tam mice at different times after PS deletion. NrxnCTF mice do not accumulate APP-CTF, as shown. B) Expression of PS1 and of selected PS/gamma-secretase substrates in PScKO tam and NrxnCTF mice. Cortical and hippocampal lysates were analyzed by Western-blot with PS1, APP, N-Cadherin and cyto-Nrxn antibodies, as indicated. The cyto-Nrxn antibody recognizes the cytoplasmic tail of both endogenously produced NrxnCTF in PScKO tam mice and NrxnCTF expressed in transgenic mice. Note that expression of PS1 or accumulation of gamma-secretase substrates are not affected in NrxnCTF mice. A.C. S´ anchez-Hidalgo et al. Experimental Neurology 347 (2022) 113896 7 transgenic mice with a similar pattern and tissue distribution than NrxnCTF accumulated in PScKO tam mice (Fig. 3B, C). The level of NrxnCTF expression in transgenic mice was 4-5-fold higher than those of NrxnCTF in PScKO tam mice, indicating moderate over-expression. A recent work failed to uncover accumulation of NrxnCTF in neurons lacking PS function (Barthet et al., 2018). While the reasons for these results are not known, detection of Nrxn proteins have been proved difficult and require proper immunological tools (Dean et al., 2003). Our experiments using different brain regions of mice lacking PS further confirm the accumulation of NrxnCTF. Collectively, the biochemical characterization indicate that NrxnCTF mice reproduce the impaired clearance of the Nrxn substrate for PS with the same tissue distribution and subcellular localization than a loss of PS function would generate in adult forebrain neurons, but leaving intact PS/gamma-secretase function towards other substrates. 3.4. Decreased stereotyped behavior and subtle social interaction deficits in NrxnCTF mice NrxnCTF mice represent a suitable animal model to study the effects produced by a selective loss of PS/gamma-secretase function towards Nrxns. First, we conducted a battery of tests to uncover potential behavioral deficits. Within each test, NrxnCTF mice were compared with two types of controls, one control group comprised by wild-type and TRE-NrxnCTF mice and a second control group formed by CaMKII α -tTA mice. These comparisons rule out any potential contribution that the CaMKII α -tTA allele might cause in the phenotypes under study (Han et al., 2012). Locomotion was analyzed in the Open Field assay. Total distance as well as the distance travelled at the periphery and at the center were found similar in NrxnCTF mice and control groups (Fig. 4A), indicating that locomotion is not affected in NrxnCTF mice. Besides memory problems, neuropsychiatric features are often A B Control CaMKIID-tTA NrxnCTF 0 2000 4000 6000 8000 Distance (cm) Periphery Center 0 2000 4000 6000 8000 Distance (cm) Control CaMKIID-tTA NrxnCTF C Control CaMKIID-tTA NrxnCTF 0 50 100 150 200 Total interaction time (s) Control CaMKIID-tTA NrxnCTF Control CaMKIID-tTA NrxnCTF 0 20 40 60 80 100 Self Grooming time (s) ** ** Control CaMKIID-tTA NrxnCTF 0 10 20 30 Social index Object Mouse 0 50 100 150 200 Interaction time (s) ** *** *** Fig. 4. Locomotion, repetitive behavior and social interaction in NrxnCTF mice. A) Open field assay. Total distance and distance travelled per area are shown. Total distance: control 5893 ±273.7; CaMKII α -tTA 5008 ±450.4; NrxnCTF 6509 ±705.3. Distance at the periphery: control 4892 ±218.2; CaMKII α -tTA 4374 ±406.9; NrxnCTF 5569 ±625.2. Distance at the center: control 1001 ±86.29; CaMKII α -tTA 634.1 ±83.31; NrxnCTF 940.5 ±120. Distance in cm. Control, n =29; CaMKII α - tTA, n =15; NrxnCTF, n =24. B) Reduced self-grooming time in NrxnCTF mice compared with control groups. Control 67.10 ±10.09; CaMKII α -tTA 72.10 ±13.29; NrxnCTF 28.31 ±5.19. Time in s. Control, n =28; CaMKII α -tTA, n =14; NrxnCTF, n =24. C) Social interaction in the three-chamber test. Graphs show total time spent in interaction with the mouse and the object (left), time interacting with the mouse or the object (center) and social index (left). Total interaction time: control 156.0 ±10.77; CaMKII α -tTA 140.5 ±11.86; NrxnCTF 152,7 ±12,13. Time interacting with the object: control 36.19 ±5.20; CaMKII α -tTA 35.51 ±4.92; NrxnCTF 53.61 ±9.08; time interacting with the mouse: control 119.8 ±9.26; CaMKII α -tTA 105.0 ±12.42; NrxnCTF 99.05 ±12.85. Time in s. Social index: control 26.18 ± 2.99; CaMKII α -tTA 23.43 ±4.15; NrxnCTF 14.46 ±4.93. Control, n =17; CaMKII α -tTA, n =12; NrxnCTF, n =12. ** p <0.01, *** p <0.001. A.C. S´ anchez-Hidalgo et al. Experimental Neurology 347 (2022) 113896 8 associated with AD, including altered stereotypic behavior and social apathy (Bozeat et al., 2000; Cerejeira et al., 2012; Prioni et al., 2012). Repetitive behavior was studied in the self-grooming assay in which the time dedicated to perform a stereotyped sequence of movements is quantified (Fig. 4B). Interestingly, self-grooming time in NrxnCTF mice was decreased by over 50% as compared with control and CaMKII α -tTA mice (Fig. 4B). The number of self-grooming bouts was similar in the three groups (control: 3.71 ±0.38; CaMKII α -tTA: 4.42 ±0.74; NrxnCTF: 3.41 ±0.32; p >0,05), but NrxnCTF mice showed shorter duration of self-grooming events (control: 19.64 ±3.09; CaMKII α -tTA: 18.76 ± 3.70; NrxnCTF: 9.02 ±1.71; p <0.01; time in s). Social interaction was evaluated in the three-chamber test (Fig. 4C). In this assay, the tested mouse is placed in a middle chamber and the time interacting with an inanimate stimulus or with a mouse located in side chambers is analyzed. Total interaction time with either of the two stimuli was found similar in the three animal groups (Fig. 4C). Furthermore, NrxnCTF mice showed preferential interaction with the mouse compared with the object, although the average time interacting with the object slightly increased in NrxnCTF mice as compared with control groups, without reaching statistical significance (Fig. 4C). This trend towards decreased social interaction in NrxnCTF mice was detected with the social index, which measures the percentage of time that each animal differentially interacts with the mouse (Fig. 4C). These findings indicate that expression of NrxnCTF in adult forebrain produces decreased stereotyped behavior and subtle deficits in social interaction. 3.5. Normal recognition memory and spatial memory in NrxnCTF mice Memory deficit is a core symptom in AD and mouse models lacking PS expression. Therefore, we analyzed the performance of NrxnCTF mice in several memory assays. The novel object recognition test evaluates the innate preference of a mouse to interact with a novel object 0 20 40 60 80 Interaction time (%) *** ** ** Control CaMKIID-tTA NrxnCTF Familiar New AB 1234 0 25 50 75 100 125 Latency to target hole (s) * Training session (day) CD 0 10 20 30 40 50 60 70 Adjacent - Target Adjacent + Opposite Time (%) 0 10 20 30 40 50 60 70 Adjacent - Target Adjacent + Opposite Time (%) Control CaMKIID-tTA NrxnCTF Control CaMKIID-tTA NrxnCTF Control CaMKIID-tTA NrxnCTF Fig. 5. Normal spatial and recognition memories in NrxnCTF mice. A) Novel object recognition. Similar average time interacting with the familiar or novel object in NrxnCTF mice compared with control groups. Control, familiar object 39.86 ±3.75%, novel object 60.14 ±3.75%; CaMKII α -tTA, familiar object 33.42 ±4.22%, novel object 66.58 ±4.22%; NrxnCTF, familiar object 36.79 ±2.74%, novel object 63.21 ±2.74%. Control, n =17; CaMKII α -tTA, n =5; NrxnCTF, n =11. B-D) Barnes maze test. B) Time latency to find the escape hole during the four training sessions. Control 64.72 ±6,35, 57.11 ±6.73, 51.19 ±8.71, 38.91 ±5.54; CaMKII α - tTA 93.00 ±14.65, 80.9 ±15.41, 59.97 ±8.16, 38.53 ±6.15; NrxnCTF 68.94 ±11.47, 42.12 ±6.07, 30.79 ±5.06, 33.28 ±6.28. Time in s. C) Percentage of time spent in quadrant one day after training. Adjacent -, control 20.28 ±3.23%, CaMKII α -tTA 26.69 ±10.34%, NrxnCTF 16.54 ±3.12%; target, control 52.22 ±5.22%, CaMKII α -tTA 48.99 ±11.06%, NrxnCTF 42.39 ±6.23%; adjacent +, control 19.13 ±3.68%, CaMKII α -tTA 17.96 ±7.96%, NrxnCTF 27.76 ±6.46%; opposite, control: 8.36 ±1.47%, CaMKII α -tTA: 6.36 ±1.74%, NrxnCTF 13.32 ±2.55%. D) Percentage of time spent in quadrant eight days after training. Adjacent-, control 10.74 ±1.97%, CaMKII α -tTA 4.08 ±1.35%, NrxnCTF 8.60 ±2.19%; target, control 44.74 ±5.89%, CaMKII α -tTA 55.90 ±11.20%, NrxnCTF 51.57 ±8.16%; adjacent +, control 19.29 ±2.89%, CaMKII α -tTA 24.60 ±10.19%, NrxnCTF 18.91 ±5.82%; opposite, control 14.99 ±3.13%, CaMKII α -tTA 7.31 ±3.14%, NrxnCTF 11.02 ±2.32%. Control, n =18; CaMKII α -tTA, n =12; NrxnCTF, n =12. * p <0.05, ** p <0.01, *** p <0.001. A.C. S´ anchez-Hidalgo et al. Experimental Neurology 347 (2022) 113896 9 presented along a familiar one. NrxnCTF mice showed no differences with control groups in the preferential interaction with the novel object (Fig. 5A). Spatial memory was studied in the Barnes Maze. In this test, mice are exposed to a round open space delimited by holes in the periphery, where only one of them is an escape hole. After repeated sessions, mice learn to find shelter in the escape hole using spatial cues placed in the periphery. During learning, NrxnCTF mice and control groups spent similar time to find the escape hole in the initial and the final fourth session, although the escape time in the second and third sessions was reduced in NrxnCTF mice (Fig. 5B). After learning, the escape hole is closed and spatial memory is evaluated one and eight days later. At both time points, NrxnCTF mice showed no differences with control groups in the time spent in the target quadrant (Fig. 5C, D). These data indicate that recognition and spatial memories are not affected in NrxnCTF mice. 3.6. Decreased fear-conditioning memory in NrxnCTF mice We evaluated fear-memory, a form of associative learning depending on the amygdala and hippocampus (Kim and Fanselow, 1992; Phillips and LeDoux, 1992). In the fear-conditioning assay, mice learn by repetition to associate a neutral stimulus, a tone (the conditioned stimulus, CS), with a behavioral significant stimulus co-terminating with the CS, a mild electric shock (the unconditioned stimulus, US). NrxnCTF mice showed no significant differences with controls during conditioning evaluated as the freezing time during CS-US pairing (Fig. 6A). Contextualand auditory-fear memory was evaluated one day after CS-US pairing. NrxnCTF mice showed similar freezing time than controls when exposed to the arena where CS-US stimuli were paired, indicating that NrxnCTF mice can evoke a freezing response (Fig. 6B). However, NrxnCTF mice showed reduced freezing time to the tone presented in a new arena (Fig. 6C). The reduced CS-dependent memory of NrxnCTF mice was not due to an effect of the CaMKII α -tTA allele (Fig. 6C). These data indicate that expression of NrxnCTF in mature forebrain decreases associative fear memory. 3.7. Altered short term synaptic plasticity at mPFC-BLA synapses of alert behaving NrxnCTF mice Cortical inputs from the mPFC regulate the expression of fear memories stored in the amygdala (Burgos-Robles et al., 2009; Corcoran and Quirk, 2007; Sotres-Bayon and Quirk, 2010). The presynaptic expression of NrxnCTF by cortical neurons along with the decrease in fear memory in NrxnCTF mice, pointed at mPFC-BLA synapses as an appropriate pathway to uncover potential synaptic effects caused by NrxnCTF. Control and NrxnCTF mice (n =10 mice/group) were 123 0 5 10 15 Freezing time (s) Control CaMKIID-tTA NrxnCTF A Tones BC Control CaMKIID-tTA NrxnCTF 0 5 10 15 20 25 Freezing time (s) Control CaMKIID-tTA NrxnCTF 0 20 40 60 80 100 120 Freezing time (s) Control CaMKIID-tTA NrxnCTF 0 10 20 30 40 50 Freezing time (s) * * Fig. 6. Decreased fear memory in NrxnCTF mice. A) Freezing time during the conditioning session. Control 14.20 ±3.37; CaMKII α -tTA 20.17 ±2.63; NrxnCTF 14.87 ±2.45. B) Freezing time in the context. Control 67.35 ±14.98; CaMKII α - tTA 90.30 ±15.66; NrxnCTF 84.85 ± 16.47. C) Decreased freezing response during tone presentation in NrxnCTF mice. Control 38.56 ±4.99; CaMKII α -tTA: 41.46 ±7.35; NrxnCTF 19.04 ±4.30. Control, n =19; CaMKII α -tTA, n =13; NrxnCTF, n = 14. Time in s. * p <0.05. A.C. S´ anchez-Hidalgo et al.