key: cord-0315636-026cf7v9 authors: Al Abed, A.S.; Sellami, A.; Marighetto, A.; Desmedt, A.; Dehorter, N. title: Autism Spectrum Disorder is a Risk Factor for PTSD-like Memory Formation date: 2021-01-19 journal: bioRxiv DOI: 10.1101/2021.01.18.427217 sha: eb269589033188907f7aaa96c6792ed6c3afc2c1 doc_id: 315636 cord_uid: 026cf7v9 The predisposition of Autism Spectrum Disorder (ASD) to Post-Traumatic Stress Disorder (PTSD) remained to be established. Here, we show that exposure to a single mild stressful event induces maladaptive memory, which recapitulates all features of PTSD, and is associated with the broad dysfunction of the prefrontal-hippocampo-amygdalar network. Using optogenetics, we demonstrate that prefrontal cortex hyperactivation triggers this PTSD-like memory. Finally, we show that recontextualization of the traumatic event normalizes maladaptive memory in ASD conditions. Overall, this study provides the first direct demonstration that ASD represents a risk factor for PTSD and uncovers new mechanisms that underlie pathological memory formation. to the electric foot-shock or the tone presentation between WT and Cntnap2 KO mice (Figure S1A), we concluded that the altered memory did not stem from different sensory processing (3) . Additionally, the Cntnap2 KO mice showed significantly higher freezing levels when presented to a similar tone (3kHz) to the conditioning tone (1kHz), than compared to a white noise exposure (i.e. highly dissimilar from the conditioning tone; Figure S1B ).These results indicate partial generalization of fear(11) when exposed to similar salient element of the traumatic event (1) . We therefore demonstrate the increased susceptibility of the Cntnap2 KO mice to develop a maladaptive memory that recapitulates all the characteristics of PTSD. To confirm that this profile is a feature of the autistic condition and not due to specific features of the Cntnap2 KO mouse model, we submitted the phosphatase and tensin homolog conditional knockout (Pten cKO) mouse model of autism (14) to the restraint stress and unpaired fear conditioning paradigm. The Pten cKO mice also displayed maladaptive memory (Figure 1E ), overall proving that the autistic condition is a risk factor for developing PTSD-like memory. Maladaptive memory is underpinned by imbalanced neuronal activity within the prefrontal-hippocampoamygdalar network, in both mice and humans (15, 16) . Interestingly, the medial prefrontal cortex (mPFC), hippocampus, and amygdala have also been implicated in the pathophysiology of ASD (17) (18) (19) . To uncover potential divergent mechanisms in PTSD and ASD, we mapped the brain activations induced by re-exposure to the conditioning context (Figure 2A) . We found a widespread imbalance in recall-induced c-Fos expression in the stressed Cntnap2 KO mice compared to the stressed WT mice ( Figure 2B ). In addition, analyses of pair-wise between-structure correlations of c-fos expression ( Figure 2C ) demonstrated a drastic alteration in the functional connectivity of the autistic mice. In particular, we observed a hypoactivation of the hippocampus and amygdala, as well as a hyperactivation of the mPFC. These different levels of activation are consistent with those previously reported in ASD (17, 18, 20) but are opposite to the patterns observed in PTSD (i.e. mPFC hypoactivation and amygdala hyperactivation (11)). Such a discrepancy between the behavioral impairments and the expected alterations in brain activations has been previously observed in autism. For instance, while increased amygdala activity underlies excessive anxiety (21) , it has been shown that ASD patients display anxiety traits in spite of amygdala hypofunction (17) . With this in mind, our results suggest that rather than the activation levels of a particular structure (i.e hypo-or hyper-activation), the alterations of the default state network in ASD could be responsible for the behavioral defects. Within the prefrontal-hippocampo-amygdalar network, the mPFC underpins a key function in the pathophysiology of both PTSD and ASD (16, 18, 22) . In addition, modulating the activity of the mPFC normalizes the core symptoms of ASD (20) . We investigated the causal role of the mPFC in the overlap between ASD and PTSD. Using optogenetic manipulation during the conditioning session, we determined the consequences of counteracting mPFC hyper-activation on memorization ( Figure 2D ). We found that, unlike the control GFP-infected group, the ArChT-expressing stressed Cntnap2 KO mice displayed normal memory of the stressful event (i.e. no fear responses to the irrelevant tone, and high fear responses to the predictive context; Figure 2E ). We concluded that mPFC inhibition in stressed ASD mice prevents PTSDlike memory formation. To then assess whether combining the acute mild stress with mPFC hyperactivation could elicit the development of pathological memory, we employed optogenetic stimulation during conditioning in stressed WT mice. Unlike GFP-injected mice, ChR2-expressing WT mice displayed a strong fear response to the tone, combined with a decreased response to the context ( Figure 2F ). Together, these results demonstrate for the first time that mPFC hyperactivation during a trauma, following an acute mild stress, leads to PTSD in both autistic and WT mice. However, previous studies reported mPFC hypoactivation in patients suffering from PTSD (16) . We propose that activation levels could differ according to the stage of memorization. Indeed, while hypoactivation of the mPFC could be associated with the recall of a traumatic event, our study manipulates the activity of the mPFC during the encoding of the trauma. Future studies should investigate whether mPFC hyperactivity during memory formation could result in circuit changes, ultimately leading to mPFC hypofunction during long-term memory recall. In addition, the maintenance of the excitation/inhibition balance in the mPFC, controlled by parvalbumin interneurons (23) , is critical for adapted fear memorization (22) and altered in ASD (20) . Deciphering the molecular mechanisms underlying the finetuning of cortical interneuron activity(24) will provide critical insights into the pathophysiology of both ASD and PTSD. Despite the divergence observed in the activation levels of the amygdala and the mPFC, we uncovered a common pattern of activation, shared between the ASD and PTSD conditions: a hypoactivation of the hippocampus ( Figure 2B) . The key function of the hippocampus in the generation of PTSD has been recently demonstrated and the lack of hippocampal activation has been shown to prevent the encoding of the context surrounding a stressful event, hence shifting the fear towards a salient, irrelevant cue associated with the trauma (12, 13) . Our results confirm the pivotal role of the hippocampus-dependent contextualization capacity in determining the nature of the memory formed (i.e. normal vs. pathological memory) during a stressful event, in both ASD and PTSD. We next examined the malleability of the pathological memory developed in ASD conditions. We used recontextualization(12), a behavior-based rehabilitation strategy, which has been suggested as a therapeutic strategy in PTSD patients (13) . The protocol consists of re-exposing the Cntnap2 KO mice displaying longterm PTSD-like memory to the original tone in the conditioning context with no footshock ( Figure 3A ). During the recontextualization session, the stressed Cntnap2 KO mice replicated the amnesia to the conditioning context, followed by a strong and high fear response to the tone ( Figure 3B ). 24h later, they exhibited normal, contextualized fear memory ( Figure 3C ). Together, these results demonstrate that pathological memory can be reshaped into adapted fear memory. By reactivating the traumatic memory in the original environment, recontextualization allows the re-allocation of trauma representation into specific context, thereby suppressing abnormal hypermnesia (12) . While this is promising, one caveat for the implementation of this procedure in patients will arise from being able to identify the origin of the trauma. Our results show that, contrary to a control population, in which PTSD-related memory is triggered by an extreme stress(1), a single mild stress is sufficient to produce pathological memory of an event in ASD. In line with the well-demonstrated impairments in coping with stress (3), we suggest that everyday life situations could ultimately become traumatizing in ASD. Yet, the current prevalence of PTSD in autism does not differ from the normal population (7). This low prevalence is most likely underestimated and could stem from the challenges of detecting PTSD in ASD (8) . Finally, this work encourages awareness of the potential consequences of stress in other psychiatric disorders, in which declarative memory formation is similarly impaired (25) . Overall, our study provides a new tool to further dissect the overlap between PTSD and ASD and the underlying alterations in emotional memory. It encourages future research to enhance detection and implementation of therapeutic strategies to pave the way towards alleviating the uncontrollable reactivation of traumatic memory in autism. This is particularly timely in the context of environmental pressure such as the COVID-19 pandemic(26), in which lockdowns and curfews disrupt routines and may produce long-lasting cognitive impairments in vulnerable populations. Mice. 3 to 5-month-old naive mice were individually housed in standard Makrolon cages, in a temperatureand humidity-controlled room under a 12-h light/dark cycle (lights on at 07:00), with ad libitum access to food and water. We used both males and females to minimize the number of mice produced in this study. WT and KO mice (Cntnap2 KO, Jackson ID: #017482(10)) were non-littermates, derived from breeders of the same colony. Pten control (Pten F/F ) and cKO mice (Pten F/F ; Emx-Cre) were male littermates, kindly provided by Dr Andreas Frick (Neurocentre Magendie, Bordeaux University, France). All experiments took place during the light phase. Every effort was made to minimize the number of animals used and their suffering. All procedures were conducted in accordance with the European Directive for the care and use of laboratory animals (2010-63-EU) and the animals care guidelines issued by the animal experimental committee of Bordeaux University (CCEA50, agreement number A33-063-099; authorization N°21248), and from the Australian National University Animal Experimentation Ethics Committee (protocol numbers A2018/66 and A2020/26). To assess the quality of the fear memory in the Cntnap2 WT and KO mice, we used the only model available that allows the assessment of both memory components of PTSD (emotional hypermnesia and contextual amnesia) in mice (11) . Habituation (Day 0): The day before fear conditioning, mice were individually placed for 4 min into a white squared chamber (30x15cm, Imetronic, France) with an opaque PVC floor, in a brightness of 40 lux. The box was cleaned with 1% acetic acid before each trial. This pre-exposure allowed the mice to acclimate and become familiar with the chamber later used for the tone re-exposure test. Acute mild stress: To elicit stress response, we performed a 30min restraint stress under bright light (100Lux). Stressed mice were taken to a neutral room and placed into a perforated 50mL Falcon® tube allowing air circulation. Non-stressed control mice were taken to the same room for 30min but were kept in their home cage. Conditioning (Day 1): Acquisition of fear conditioning was performed in a different context, a transparent squared conditioning chamber (30x15 cm) in a brightness of 100 lux, given access to the different visual-spatial cues of the experimental room. The floor of the chamber consisted of 30 stainless-steel rods (5 mm diameter), spaced 5 mm apart and connected to the shock generator. The box was cleaned with 70% ethanol before each trial. All animals were trained with a tone-shock un-pairing procedure, meaning that the tone was non-predictive of the footshock occurrence. This training procedure, fully described in previous studies (11) , promotes the processing of contextual cues in the foreground. Briefly, each animal was placed in the conditioning chamber for 4 min during which it received two tone cues (65 dB, 1 kHz, 15 s) and two foot-shocks (squared signal: 0.4 mA, 50 Hz, 1 s), following a pseudo-randomly distribution. Specifically, animals are placed in the conditioning chamber and receive a shock 100 s later, followed by a tone after a 30s interval. After a 20s delay, the same tone and shock spaced by a 30s interval were presented. Finally, after 20s, mice were returned to their home cage. As the tone was not paired to the footshock, mice selected the conditioning context (i.e. set of static background contextual cues and odor that constitutes the environment in which the conditioning takes place) and not the tone as the correct predictor of the shock (Figure 1A , Day1). Memory Tests (Day 2): 24 hours after conditioning, mice were submitted to two memory retention tests and continuously recorded for off-line second-by-second scoring of freezing by an observer blind to experimental groups. Mouse freezing behavior, defined as a lack of movement (except for respiratory-related movements), was used as an index of conditioned fear response (27) . Mice were first submitted to the tone re-exposure test in the safe, familiar chamber during which three successive recording sessions of the behavioral responses were previously performed: one before (first 2 min), one during (next 2 min), and one after (2 last min) tone presentation ( Figure 1A , Day 2) . Conditioned response to the tone was expressed by the percentage of freezing during tone presentation, compared to the levels of freezing expressed before and after tone presentation (repeated measures on 3 blocks of freezing). The strength and specificity of this conditioned fear was attested by a tone ratio that considers the percentage of freezing increase to the tone with respect to a baseline freezing level (i.e., pre-and post-tone periods mean). The tone ratio was calculated as follows: [% freezing during tone presentation -(% pre-tone period freezing + % post-tone period freezing)/2] / [% freezing during tone presentation + (% pre-tone period freezing + % posttone period freezing)/2]. Two hours later, mice were submitted to the context re-exposure test. They were placed for 6 min in the conditioning chamber. Freezing to the context was calculated as the percentage of the total time spent freezing during the successive three blocks of 2-min periods of the test. To demonstrate that the PTSD-like memory was long-lasting, mice were again submitted to the two memory tests (tone-test and context-test spaced out of 2h, as in day 2), 20 days after fear conditioning. Optogenetic manipulation of the mPFC. Surgery: Mice were injected bilaterally 4 weeks before behavioral experiments with an Adeno-Associated Virus (AAV) to inhibit (AAV5-CaMKIIα-ArchT-GFP, UNC Vector Core) or activate glutamatergic neurons (AAV5-CaMKIIα-ChR2 (H134R)-EYFP, UNC Vector Core). Control mice were injected with an AAV expressing GFP only (AAV5-CaMKIIα-GFP). We used glass pipettes (tip diameter 25-35 µm) connected to a picospritzer (Parker Hannifin Corporation) into the mPFC (0.1 µl/site; AP +1.9mm; L ±0.35mm; DV -1.3mm). Mice were then implanted with bilateral optic fiber implants, 10 days before behavior (diameter: 200µm; numerical aperture: 0.39; flat tip; Thorlabs) directed to the mPFC (AP: +1.8, L: ± 1.0, DV: -1.3, θ: 10°). Implants were fixed to the skull with Super-Bond dental cement (Sun Medical, Shiga, Japan). Mice were perfused after experiments to confirm correct placements of fibers ( Figure S2) . For optogenetic manipulations, we used a LED (Plexon®) at 465nm with a large spectrum to allow the activation of both ArChT and ChR2. Light was continuously delivered to inactivate the mPFC, and was delivered at 5Hz (5ms ON,195ms OFF) for mPFC activation. Mice were submitted to the fear conditioning procedure described above and pyramidal cells of the mPFC were either inhibited or activated during the whole conditioning session. Mice were then immediately placed in the conditioning chamber and submitted to the fear conditioning paradigm. The next day, fear memory was tested as described above. Recontextualization was performed as previously described (12) . Briefly, two days after long-term testing (Day 23), mice were re-exposed to the tone cue in the conditioning context, without electric shock (Figure 3 , Day 23). The first 2 min (pre-tone) allowed us to assess the level of conditioned fear to the conditioning context alone, while the conditioned response to the tone was assessed during the next 2 min, both by the percentage of freezing during the tone presentation and by the tone ratio described above. 24h later, fear expression was assessed by re-exposing the mice to the regular (i.e. separated from each other) tone and context tests (same tests as in day 2). c-Fos immunohistochemistry. 90 minutes after the context test on Day 2, animals were perfused transcardially with 0.01M phosphate buffered saline (PBS) to eliminate blood and extraneous material, followed by 4% paraformaldehyde (PFA). Brains were postfixed for 36 hours in PFA. Tissues were sectioned at 40µm using a Leica 1000S vibratome and kept in a cryoprotective ethylene glycol solution at -20˚C until processed for immunofluorescence. Sections were first washed and permeabilized, then non-specific binding sites were blocked by immersing the tissue in 10% normal donkey serum, 2% BSA in PBS-Triton during 2h. Tissues were then stained using the primary antibodies overnight: mouse anti-c-Fos (1:1000; Santa Cruz). After 3x 15' washes, we added anti-rabbit, anti-chicken, anti-mouse, Alexa 488 or 555 (1:200; Life Technologies) secondary antibodies for 2h. After 3x 15' washes slices were stain during 10' with DAPI (5µM; Sigma), mounted on Livingstone slides then covered with Mowiol (Sigma) and coverslip (Thermofisher). c-Fos staining was imaged using a Zeiss Axio Observer fluorescent microscope (20x objective). Stained sections of control and mutant mice were imaged during the same imaging. Immunofluorescence signals were quantified using the ImageJ (FIJI) software with routine particle analysis procedures, to obtain nuclear masks, divided by the area to obtain cell density per mm 2 . Statistics. Data are presented as the mean ± SEM error bar. Statistical analyses were performed using 2-sided ANOVA, followed by Fisher's PLSD post-hoc test when appropriate. Analyses were performed using the StatView software. Normality of the data was confirmed using the Kolmogorov-Smirnov test. Statistical significance was considered at P<0.05. Figure 1C ). Data are presented as mean ±SEM. ***: p<0.001; **: p<0.01; *: p<0.05; Bracket shows interaction significance between freezing and genotype. Diagnostic and Statistical Manual of Mental Disorders : DSM-5 What we can learn from a genetic rodent model about autism Inattention and Hyperactivity in Association With Autism Spectrum Disorders in the CHARGE Study Cognitive control and episodic memory in adolescents with autism spectrum disorders Abnormal Fear Memory as a Model for Posttraumatic Stress Disorder Autism Spectrum Disorder and Post-Traumatic Stress Disorder: An unexplored co-occurrence of conditions Neglected causes of post-traumatic stress disorder A Systematic Review of the Assessment and Treatment of Posttraumatic Stress Disorder in Individuals with Autism Spectrum Disorders Loss of Cntnap2 Causes Axonal Excitability Deficits, Developmental Delay in Cortical Myelination, and Abnormal Stereotyped Motor Behavior Absence of CNTNAP2 leads to epilepsy, neuronal migration abnormalities, and core autism-related deficits Glucocorticoids can induce PTSD-like memory impairments in mice Preventing and treating PTSD-like memory by trauma contextualization Intrusive Images in Psychological Disorders: Characteristics, Neural Mechanisms, and Treatment Implications PTEN in Autism and Neurodevelopmental Disorders Adaptive emotional memory: the key hippocampal-amygdalar interaction Brain circuit dysfunction in traumatic stress disorder : from mouse to man Anxiety and social deficits have distinct relationships with amygdala function in autism spectrum disorder Reduced Prefrontal Synaptic Connectivity and Disturbed Oscillatory Population Dynamics in the CNTNAP2 Model of Autism Hippocampal deficits in neurodevelopmental disorders Modulation of prefrontal cortex excitation/inhibition balance rescues social behavior in CNTNAP2-deficient mice Contributions of the amygdala to emotion processing: From animal models to human behavior Prefrontal parvalbumin interneurons shape neuronal activity to drive fear expression Neocortical excitation/inhibition balance in information processing and social dysfunction Tuning of fast-spiking interneuron properties by an activity-dependent transcriptional switch The Cognitive Neuroscience of Memory Function and Dysfunction in Schizophrenia COVID-19 in children and young people Conditioned and unconditional components of post-shock freezing We would like to thank Dr Andreas Frick for the Pten cKO mice. We thank all the personnel of The Australian Phenomics Facility and of The Neurocentre Magendie involved in mouse care. Competing interest: The authors declare no competing interests. Data and material availability: All data is available in the main text or the supplementary materials.