Synaptic location is a determinant of the detrimental effects of α-synuclein pathology to glutamatergic transmission in the basolateral amygdala

The presynaptic protein α-synuclein (αSyn) has been suggested to be involved in the pathogenesis of Parkinson’s disease (PD). In PD, the amygdala is prone to develop insoluble αSyn aggregates, and it has been suggested that circuit dysfunction involving the amygdala contributes to the psychiatric symptoms. Yet, how αSyn aggregates affect amygdala function is unknown. In this study, we examined αSyn in glutamatergic axon terminals and the impact of its aggregation on glutamatergic transmission in the basolateral amygdala (BLA). We found that αSyn is primarily present in the vesicular glutamate transporter 1-expressing (vGluT1+) terminals in the mouse BLA, which is consistent with higher levels of αSyn expression in vGluT1+ glutamatergic neurons in the cerebral cortex relative to the vGluT2+ glutamatergic neurons in the thalamus. We found that αSyn aggregation selectively decreased the cortico-BLA, but not the thalamo-BLA, transmission; and that cortico-BLA synapses displayed enhanced short-term depression upon repetitive stimulation. In addition, using confocal microscopy, we found that vGluT1+ axon terminals exhibited decreased levels of soluble αSyn, which suggests that lower levels of soluble αSyn might underlie the enhanced short-term depression of cortico-BLA synapses. In agreement with this idea, we found that cortico-BLA synaptic depression was also enhanced in αSyn knockout mice. In conclusion, both basal and dynamic cortico-BLA transmission were disrupted by abnormal aggregation of αSyn and these changes might be relevant to the perturbed cortical control of the amygdala that has been suggested to play a role in psychiatric symptoms in PD.

The amygdala is a key limbic structure for emotion regulation (Janak and Tye, 2015). Compelling clinical evidence indicates that cortical control of the amygdala activity is impaired in PD, leading to an inappropriate encoding of emotion valence or deficits in linking emotion to behavior (Bowers et al., 2006;Hu et al., 2015;Yoshimura et al., 2005). Moreover, the amygdala shows selective vulnerability to Lewy pathology (Harding et al., 2002;Nelson et al., 2018;Sorrentino et al., 2019), thus an impaired amygdala network function has been proposed to underlie the disrupted emotion processing in PD patients (Harding et al., 2002). Still, the normal function of αSyn and how its aggregation can impair amygdala circuit function remain poorly understood. Here, we show that αSyn is primarily present in vesicular glutamate transporter 1-expressing (vGluT1 + ) cortical axon terminals, and, by contrast, is barely detectable in vGluT2 + thalamic axon terminals in the basolateral amygdala (BLA). In an αSyn preformed fibrils (PFFs) model of synucleinopathies, αSyn pathology decreases vGluT1 + cortico-BLA glutamatergic transmission, without affecting the vGluT2 + thalamo-BLA neurotransmission. Furthermore, we demonstrate that a partial (secondary to αSyn aggregation) or complete (Snca KO mice) depletion of soluble αSyn from the axon boutons promotes short-term depression at cortico-BLA synapses in response to prolonged stimulation, leading to an impaired gain control of cortical inputs to the BLA. Therefore, we conclude that both gains of toxic properties and loss of normal function of αSyn contribute to the input-specific disruption of cortico-BLA synaptic connectivity and plasticity in synucleinopathies. Our data support clinical observations of an impaired cortical control of the amygdala activity that could contribute to psychiatric deficits in PD patients.

Results
αSyn localizes preferentially in vGluT1 + axon terminals in mouse brain Compelling evidence from PD patients suggests that the amygdala exhibits an impaired responsiveness to sensory stimuli, arising mainly from the cerebral cortex and the thalamus (Bowers et al., 2006;Hu et al., 2015). Thus, we examined the presence of αSyn in vGluT1 + and vGluT2 + axon boutons in wild type (WT) mouse brains, which mainly come from cortical and thalamic regions, respectively (Fremeau et al., 2001;Kaneko and Fujiyama, 2002;Vigneault et al., 2015). Figure 1 shows that αSyn immunoreactivity colocalizes with vGluT1 + puncta but is absent where there are vGluT2 + puncta, in the BLA, the cerebral cortex, and the striatum.
Earlier in situ hybridization studies showed higher αSyn mRNA expression in the cerebral cortex and the hippocampus than in the thalamus (Abeliovich et al., 2000;Ziolkowska et al., 2005). Next, we used Snca NLS/NLS reporter mice to determine αSyn protein localization in the cortical and thalamic areas. These mice localize endogenous αSyn to the nucleus that allows the visualization of cellular topography (Geertsma et al., 2022), circumventing the diffused αSyn immunoreactivity in WT mice ( Figure 2A). We observed that the αSyn was heavily expressed in cortical layer V/VI neurons, but only moderately or barely expressed in thalamic regions, particularly in the midline thalamus that provides major excitation to the BLA (Ahmed et al., 2021;Amir et al., 2019;Hintiryan et al., 2021;Figure 2B-D). Together, our data show that αSyn is preferentially present at vGluT1 + cerebral cortical neurons and their axon terminals but is absent or expressed at very low levels at vGluT2 + thalamic neurons and their projections.
Glutamate release from vGluT1 + axon terminals is selectively disrupted by αSyn pathology Considering that the endogenous levels of αSyn are determinants of the propensity to form αSyn aggregates, we hypothesize that vGluT1 + neurons and their terminals are more susceptible to αSyn pathology compared to those are vGluT2 + (Erskine et al., 2018;Vasili et al., 2022). To test this hypothesis, we triggered widespread αSyn pathology in the brain using the intrastriatal PFFs seeding model (Luk et al., 2012). One-month post-injection, we detected robust αSyn pathology in vGluT1 + cerebral cortical regions (e.g. the temporal association cortex (TeA), the motor cortex, and the perirhinal cortex, Figure 3 and A1-A) and the BLA ( Figure 3B-C), but we barely observed any cytoplasmic aggregates in vGluT2 + thalamic regions (Figure 3 and A4-A5). This pattern of pathology is supported by differences in endogenous αSyn levels between the cerebral cortex and thalamus  and is consistent with earlier reports (Burtscher et al., 2019;Henderson et al., 2019;Luk et al., 2012;Stoyka et al., 2020). Importantly, neurons in the midline thalamus were retrogradely labeled when retrobeads were injected into the same location as for PFFs in the striatum ( Figure 3A6). We concluded that the absence of cytoplasmic pS129 αSyn pathology in the thalamus was not due to technical issues (i.e. missing thalamostriatal axon terminals for PFFs internalization).
The online version of this article includes the following source data for figure 1: Source data 1. Source data for plot in Figure 1C.
cortico-BLA excitatory postsynaptic currents (EPSCs) was greatly decreased in the slices from PFFsinjected WT mice relative to those from controls ( Figure 3D-E). Consistently, prominent pS129 αSyn pathology can be detected along the external capsule where the cortical afferents enter the BLA ( Figure 3B-C). In contrast, there was no difference in the amplitude of thalamo-BLA EPSCs between the PFFs-injected mice and controls ( Figure 3F-G).
To avoid potential technical issues inherent to the use of electrical stimulation, we employed optogenetic approach to confirm the above results ( Figure 3H-O). One-month post-injection, the amplitude of optogenetically-evoked cortico-BLA EPSCs in slices from PFFs-injected mice was decreased relative to those from PBS-injected controls ( Figure 3H-K). By contrast, we did not detect the difference in the amplitude of optogenetically-evoked thalamo-BLA EPSCs between groups ( Figure 3L-O). Altogether, we demonstrated that the presence of αSyn makes vGluT1 + cortical neurons and their axons more vulnerable to αSyn pathology and that this pathology is associated with detrimental effects on synaptic transmission.
The online version of this article includes the following source data for figure 2: Source data 1. Source data for plots in Figure 2C and D.  by the gained toxic properties of αSyn as it aggregates, instead of intrastriatal PFFs injection per se (Cookson and van der Brug, 2008;Volpicelli-Daley et al., 2011). αSyn pathology decreases the number of functional cortico-BLA inputs Several mechanisms can contribute to the decreased cortico-BLA synaptic strength as αSyn pathology develops, including the loss of synapses, decreased initial release probability, and/or postsynaptic adaptations. To determine the impact of αSyn aggregation on vGluT1 + cortico-BLA innervation, brain sections from control and PFFs-injected mice were processed for immunohistochemical assessment of vGluT1. The density of vGluT1-immunoreactive puncta in the BLA was then determined stereologically (West, 1999). The density of vGluT1-immunoreactive puncta was not altered between PFFs-and PBS-injected mice ( Figure 4A-C). These results indicate that there is no loss of cortical axon terminals or cortico-BLA synapses in PFFs-injected mice at one-month post-injection.
Next, we assessed functional changes in cortico-BLA synaptic transmission using electrophysiology. The initial release probability was estimated by delivering paired-pulses of electric or optogenetic stimulation of cortico-BLA synapses in control and PFFs-injected mice. Using electric stimulation approach, the ratio of EPSC2 to EPSC1 (at a 50 ms inter-pulse interval) was not altered between groups ( Figure 4D-E). Similarly, the ratio of EPSC2/EPSC1 at cortico-BLA synapses was not altered by αSyn pathology when assessed using optogenetics (at a 100 ms inter-pulse interval, Figure 4F-G). Interestingly, in contrast to the short-term facilitation in response to electric stimulation ( Figure 4D-E), paired-pulses of optogenetic stimulation always led to short-term depression of cortico-BLA transmission ( Figure 4F-G), which might be due to the deactivation of ChR2 itself. Together, the above data suggest that the decreased cortico-BLA transmission was not caused by a lower initial release probability.
Further, we estimated the quantal properties of cortico-BLA synapses by measuring the frequency and amplitude of Sr 2+ -induced, optogenetically-evoked asynchronous glutamate release from cortical terminals. The frequency of optogenetically-evoked cortico-BLA asynchronous EPSCs (Sr 2+ -EPSCs) decreased significantly, but the amplitude of cortico-BLA Sr 2+ -EPSCs was not altered in slices from PFFs-injected mice ( Figure 4H-J). Given the unaltered number of vGluT1 + axon terminals and initial release probability of cortico-BLA synapses, fewer readily releasable synaptic vesicles and/or release sites can explain the observed reduction in cortico-BLA transmission.
The online version of this article includes the following source data for figure 4: Figure 4 continued on next page adaptations at cortico-BLA synapses. Surprisingly, we did not detect changes in the AMPA/NMDA ratio (control=1.24 ± 0.11, PFFs=1.5 ± 0.12, n=16 neurons/ 3 mice per group, p=0.15, MWU) or AMPA receptor rectification (control=1.12 ± 0.1, PFFs=1.28 ± 0.08, n=16 neurons/ 3 mice per group, p=0.25, MWU) from thalamo-BLA synapse between groups. Thus, the above data suggest that αSyn pathology induces input-specific decrease of postsynaptic AMPA receptor-mediated cortico-BLA transmission, instead of a global reduction of AMPA receptor function in BLA neurons.
Taken together, the development of αSyn pathology selectively decreases the functional cortico-BLA connectivity by inducing both pre-and post-synaptic adaptations and such functional changes occur prior to overt degeneration of axon terminals.
Pathological aggregation decreases αSyn levels at axon terminals and impairs short-term synaptic plasticity Formation of cytoplasmic aggregates is believed to move soluble αSyn away from the presynaptic boutons, affecting its role in regulating synaptic vesicle pools (Benskey et al., 2016;Luk et al., 2009;Volpicelli-Daley et al., 2011). Consistent with the earlier predictions, the intensity of αSyn immunoreactivity (Syn1 antibody from BD Biosciences) within vGluT1 + puncta decreased dramatically in slices from PFFs-injected mice, leading to an increased proportion of vGluT1 + boutons that lack a detectable level of αSyn ( Figure 5A-D). To avoid technical issues associated with immunostaining (e.g. antigen mask due to αSyn aggregation), we assessed the αSyn immunoreactivity within vGluT1 + terminals using a different monoclonal antibody against mouse αSyn-Syn9027 (Peng et al., 2018). Consistently, both the αSyn immunoreactivity within vGluT1 + terminals and the proportion of vGluT1 + terminals that are αSyn-immunoreactive decreased significantly in slices from PFFs-injected mice relative to controls ( Figure 5E-H). The percentage of αSyn-immunoreactive vGluT1 + terminals detected with Syn9027 in PFFs-injected mice was higher than those detected with Syn1, perhaps reflecting differences in the exposure of their epitopes (Syn1: aa 91-99; Syn9027: aa 130-140). These data suggest that αSyn pathology reduces the amount of soluble αSyn present at the cortical axon terminals in the BLA, and in that way could affect its physiological function.
αSyn modulates the dynamics of synapse vesicle pools (Sulzer and Edwards, 2019), which plays a key role in regulating synaptic plasticity and the computational function of neural circuits (Abbott and Regehr, 2004;Alabi and Tsien, 2012). Thus, we stimulated cortico-BLA synapses repetitively to assess the impact of the observed reduction of αSyn levels on short-term synaptic plasticity. We detected a progressive depression of cortico-BLA EPSCs in control mice ( Figure 6A1 and A2), reflecting mainly a progressive depletion of presynaptic vesicles (Alabi and Tsien, 2012;Cabin et al., 2002). Interestingly, the cortico-BLA EPSCs in the slices from PFFs-injected mice showed a greater depression relative to those from controls ( Figure 6A1 and A2). These results are in line with the key role of αSyn in sustaining and mobilizing synaptic vesicle pools (Cabin et al., 2002;Sulzer and Edwards, 2019;Vargas et al., 2017), and suggest the significant impact of the loss of αSyn function on synaptic plasticity and circuit computation.
Because both gains of toxic properties and loss of normal αSyn function are induced in the PFFs models, we employed the αSyn KO mice to further assess the impact of a loss of αSyn on the shortterm synaptic plasticity of glutamatergic synapses. Consistently, the amplitude of cortico-BLA EPSCs from αSyn KO mice also exhibited greater depression in response to repetitive stimulation relative to littermate WT controls ( Figure 6B1-B2), indicating an impaired mobilization of synaptic vesicle pools. It is worth noting that the earlier onset and greater magnitude of cortico-BLA EPSC depression in αSyn KO mice versus that in PFFs-injected mice ( Figure 6A, B). Thus, it is plausible that the different temporal profiles of short-term synaptic dynamics are linked to the difference in the amount of αSyn present at presynaptic terminals between the PFFs model and KO mice. Last, we did not detect the difference in the temporal profiles of thalamo-BLA EPSCs in slices from αSyn KO mice and those from WT controls (Figure 6-figure supplement 1), which indicates a negligible impact of αSyn depletion on synaptic vesicle dynamics and is consistent with the lack of αSyn presence at vGluT2 + axon terminals ( Figure 1). Together, these results suggest that pathological aggregation decreases the levels Source data 1. Source data for plots in Figure 4.

Figure 4 continued
of soluble αSyn at the axon terminals, leading to a greater synaptic vesicle depletion and impaired short-term plasticity.
The online version of this article includes the following source data for figure 5: Source data 1. Source data for plots in Figure 5. several brain regions, including the BLA and the striatum. Consistent with the observation from the axon terminals, we also detected higher amounts of αSyn in the cell bodies of cortical neurons relative to thalamic neurons using the recently generated Snca NLS/NLS reporter mice (Geertsma et al., 2022). Thus, our results support a cell-and synapse-subtype-selective αSyn protein expression among glutamatergic neurons in the mouse brain.
Endogenous levels of αSyn are determinants of neuronal vulnerability to αSyn pathology. As expected, we observed a decrease in cortical, but not thalamic, inputs to the BLA neurons in a PFFseeding model of synucleinopathies. At a relatively early-stage (i.e. one-month post-injection), αSyn pathology decreases the number of functional cortical inputs to the BLA without loss of synapse or changes in presynaptic release probability. We posit that the disrupted cortico-BLA transmission can be caused by a reduced size of readily-releasable synaptic vesicle pool and/or a number of presynaptic release sites. This hypothesis was built on a wealth of evidence showing that αSyn aggregates decrease the expression of several synaptic vesicle-associated SNARE proteins (e.g. Snap25 and VAMP2) (Volpicelli-Daley et al., 2011). Moreover, our study also suggests a decreased AMPA receptor-mediated response can underlie the impaired cortico-BLA transmission in PFFs-injected mice. On the other hand, an increased relative contribution of GluA2-lacking AMPA receptors can Source data 1. Source data for plots in Figure 6.  be a compensatory mechanism for an overall reduction in AMPA-EPSCs at cortico-BLA synapses-a hypothesis that warrants further investigation. It is worth noting that such a decreased AMPA-EPSC of BLA neurons is input specific, which selectively occurs at cortico-BLA synapses. Thus, the molecular mechanisms that cooperate pre-and post-synaptic changes at the cortical inputs, but not the thalamic inputs, remain to be defined.
Our study further highlights the dependence of regional and cellular vulnerability on the endogenous αSyn, i.e., those neurons or axon terminals that express high levels of αSyn are prone to be functionally impacted by αSyn pathology relative to those that lack of or express low levels of αSyn (Surmeier et al., 2017;Thakur et al., 2019;Vasili et al., 2022). Of particular interest is that vGluT2 + / TH + midbrain dopamine neurons and their axon terminals in the striatum have been shown to be more resilient to neurodegeneration in postmortem PD brains and animal models studies (Buck et al., 2021;Steinkellner et al., 2022). While several other mechanisms have been proposed (Buck et al., 2021), the absence of αSyn at vGluT2 + neurons/terminals could be critical for such resilience.
Loss of normal αSyn function has been thought to be an important but understudied aspect of αSyn pathology. Using an electrophysiological approach, we revealed that a partial or complete removal of αSyn from the axon terminals leads to an enhanced short-term depression in response to repetitive stimulation of cortico-BLA synapses ( Figure 6). Short-term synaptic plasticity is an important 'gain control' mechanism for neurons to properly balance their responsiveness to distinct afferents in an input-specific manner (Abbott et al., 1997). Cortical and thalamic afferents of the BLA exhibit different short-term synaptic plasticity profiles in vivo, which could reflect their different contributions to the formation of emotion-related memory and behavior (Siguròsson et al., 2010). Physiologically, synapse-specific presence of αSyn could underlie such difference in the short-term plasticity profiles of the two inputs (Figures 1 and 6).
One can postulate that once aggregates form, a decreased synaptic strength associated with αSyn toxicity (Figure 3) and an enhanced synaptic depression due to the loss of αSyn normal function at presynaptic boutons (Figures 5 and 6) make the BLA neurons less likely to respond to sustained and repetitive sensory inputs from cortical regions. These circuit changes might explain studies showing a decreased functional connectivity of cortico-amygdala, but not the thalamo-amygdala network in PD patients, which further leads to a failed amygdala responsiveness to the aversive sensory inputs (Hu et al., 2015;Yoshimura et al., 2005).

Animals
Wild type (WT) C57Bl/6J mice (Jax stock#:000664, RRID: IMSR_JAX:000664) of both sexes (3-4 month-old) were obtained from the Van Andel Institute vivarium internal colony and used in the study. αSyn knockout (Snca -/-) mice (RRID: IMSR_JAX:003692) were originally purchased from Jackson laboratories (Bar Harbor, ME) and then were backcrossed on a C57BL/6J background to generate heterozygous Snca +/mice. Experimental Snca -/mice and littermate WT controls were generated from heterozygous Snca +/breeder pairs and were genotyped by Transnetyx (Cordova, TN, USA). Snca NLS/ NLS mice (RRID: IMSR_JAX:036763) were generated and maintained on a C57Bl/6J background as described (Geertsma et al., 2022). Mice were housed up to four animals per cage under a 12/12 hr light/dark cycle with access to food and water ad libitum in accordance with NIH guidelines for care and use of animals. All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee at Van Andel Institute (animal use protocol#: 22-02-007).

Preparation and validation of αSyn preformed fibrils
Purification of recombinant mouse αSyn and generation of αSyn preformed fibrils (PFFs) was conducted as described elsewhere (Luk et al., 2009;Volpicelli-Daley et al., 2014). The pRK172 plasmid (RRID: Addgene_166671) containing the gene of interest was transformed into BL21 (DE3) RIL-competent E. coli (230245, Agilent Technologies). A single colony from this transformation was expanded in Terrific Broth (12 g/L of Bacto-tryptone, 24 g/L of yeast extract 4% (v/v) glycerol, 17 mM KH 2 PO 4 , and 72 mM K 2 HPO 4 ) with ampicillin. Bacterial pellets from the growth were sonicated and the sample was boiled to precipitate undesired proteins. The supernatant was dialyzed with 10 mM Tris, pH 7.6, 50 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride (PMSF) overnight. Protein was filtered with a 0.22 µm filter and concentrated using Vivaspin 15 R 10 K centrifugal filters (VS15RH01, Sartorius). Protein was then loaded onto a Superdex 200 column and 2 mL fractions were collected. Fractions were run on SDS-PAGE and stained with InstaBlue protein stain Fisher Scientific) to select fractions that were highly enriched in αSyn. These fractions were combined and dialyzed in 10 mM Tris, pH 7.6, 50 mM NaCl, 1 mM EDTA, 1 mM PMSF overnight. Dialyzed fractions were applied to a MonoQ column (HiTrap Q HP 17115401, Cytiva) and run using a NaCl linear gradient (0.025-1 M). Collected fractions were run on SDS-PAGE and stained with InstaBlue protein stain. Fractions that were highly enriched in αSyn were collected and dialyzed into DPBS (Gibco). Protein was filtered through a 0.22 µm filter and concentrated to 7 mg/mL (αSyn) with Vivaspin 15 R 10 K centrifugal filters. Monomer was aliquoted and frozen at -80 °C. For preparation of αSyn PFFs, αSyn monomer was shaken at 1000 rpm for 7 days. Conversion to PFFs was validated by sedimentation at 100,000 x g for 60 min and by thioflavin S staining.

Stereotaxic surgery
Mice were placed in a stereotaxic frame (Kopf) under 2% isoflurane anesthesia and were supported by a thermostatic heating pad. Recombinant αSyn fibrils were diluted in phosphate-buffered saline (PBS) to a concentration of 5 mg/ml. Prior to stereotaxic injection, PFFs were pulse sonicated at medium intensity with 30 s on and 30 s off for 10 min using a sonicator (Biorupter Pico). To induce the formation of αSyn aggregation in mouse brain (Luk et al., 2012), sonicated PFFs (2.0 μl) were injected bilaterally into the dorsal striatum (anteroposterior,+0.2 mm from bregma; mediolateral, ± 2.0 mm from the midline; dorsoventral, -2.6 mm from the brain surface) using a 10 μl syringe (Hamilton) driven by a motorized microinjector (Stoelting) at a rate of 0.2 μl/min. Given the neuronal heterogeneity of the BLA, retrobeads (Lumafluor Inc) were diluted 10 × into αSyn fibrils or PBS to label the projection neurons in the BLA for physiological studies. The final injection volume of PFFs with beads mixture was adjusted to keep a constant amount of αSyn injected across animals. Control mice received 2.0 μl PBS or PBS with retrobeads injections in the same location. For optogenetics studies, AAV vectors encoding ChR2(H134R)-eYFP (titer=1 × 10 12 vg/ml, Addgene 26973, RRID: Addgene_127090) were stereotaxically injected into and centered at the TeA (anterioposterior -3.3 mm from the bregma, mediodorsal ± 4.1 mm, dorsoventral -1.5 mm from the brain surface) and the midline thalamus (anterioposterior -1.5 mm from the bregma, mediodorsal 0 mm, dorsoventral -3.3 mm from the brain surface) (Ahmed et al., 2021;Amir et al., 2019). Animals were housed in their home cages before being euthanized for immunohistochemical or physiological studies at one-month post-injection. Details of this protocol can be found at: https://doi.org/10.17504/protocols.io.rm7vzye28l×1/v1.

Tissue collection -perfusion and sectioning
WT and Snca -/mice received overdosage of avertin intraperitoneally (i.p.) and were subsequently subjected to transcardial perfusion with PBS and 4% paraformaldehyde (PFA, pH 7.4). Brains were removed and post-fixed in 4% PFA overnight, before being re-sectioned at 70 μm using a vibratome (VT1000s, Leica Biosystems, Buffalo Grove, IL). Snca NLS/NLS mice were anesthetized with 30 µl of 120 mg/kg Euthanyl (DIN00141704) before being perfused with 10 ml 1 x PBS and 10 ml 4% PFA. Brain tissue was collected and stored for 48 hr in 4% PFA. Brain tissue was then dehydrated in 10, 20, and 30% sucrose solutions for 48 hr each before being flash frozen in -40 °C isopentane for 1 min. Tissues were then cryosectioned at 20 μm and -21°C on the Thermo Scientific HM 525 NX cryostat at the Louise Pelletier Histology Core at the University of Ottawa and stored free floating in 1 × PBS +0.02% NaN3 at 4°C until use. Details of this protocol can be found here.
Details of immunofluorescent staining can be found here.
To assess glutamatergic transmission in the BLA, glass pipettes were filled with a cesium methanesulfonate based internal solution of (in mM): 120 CH 3 O 3 SCs, 2.8 NaCl, 10 HEPES, 0.4 Na 4 -EGTA, 5 QX314-HBr, 5 phosphocreatine, 0.1 spermine, 4 ATP-Mg, and 0.4 GTP-Na (pH 7.3, 290 mOsm). BLA neurons were voltage clamped at -70 mV to assess the EPSCs in response to presynaptic electrical or optogenetic stimulations. Concentric bipolar electrodes (FHC, Bowdoin, ME) or glass pipettes (~2 MΩ) filled with the extracellular solution were used as stimulating electrodes and placed on the external and internal capsules to evoke glutamate release in the BLA from cortical and thalamic axon terminals, respectively (Shin et al., 2010). A range of electrical pulses (intensities=30-120 μA, dura-tion=100 μs) were delivered through a stimulator (Digitimer, UK) to evoke glutamate release at either cortical or thalamic inputs. To study the process of synaptic vesicle pool mobilization during repetitive glutamatergic transmission, we delivered 300 electrical pulses (duration=100 μs) at 12.5 Hz and measured the amplitudes of EPSCs to quantify changes in the presynaptic glutamate release (Cabin et al., 2002). Only one neuron was recorded from each slice in the prolonged repetitive stimulation studies.
Optogenetic stimulation pulses (1 ms duration) were delivered through a 60 × objective lens (Olympus, Japan) using a 470 nm LED light source (CoolLED, UK). To isolate monosynaptic cortico-BLA and thalamo-BLA EPSCs, optogenetically evoked EPSCs were recorded in the presence of TTX (1 μM) and 4-AP (100 μM). Series resistance (Rs <20 MΩ) was regularly monitored throughout the recording to ensure Rs changes were less than 15%. Liquid junction potential (~9 mV) was not corrected. Details of the protocol can be found here.

Additional files
Supplementary files • Transparent reporting form

Data availability
All source data associated with the revised manuscript have been deposited on Open Science Framework: https://doi.org/10.17605/OSF.IO/264SM. All data generated or analyzed during this study are included in the manuscript and source data have been provided for all main and supplementary figures.
The following dataset was generated: