All mice were housed at the Faculty of Veterinary Medicine, Hokkaido University, in an AAALAC-accredited facility, following guidelines from the Institutional Animal Care and Use Committee (approval No. 23-0041). Jcl:ICR mice were purchased from CLEA Japan (Tokyo, Japan) and acclimated for a week. They had unrestricted access to food and water under a 12-h light/dark cycle.

Mouse-adapted prion 22L, Chandler, and Obihiro strains were used. Brains from terminal-stage diseased and age-matched mock-infected mice were used to prepare 10% (w/v) brain homogenates in phosphate-buffered saline (PBS) as stock solutions.

Four-week-old female Jcl:ICR mice were anesthetized with sevoflurane (Maruishi Pharmaceutical, Osaka, Japan) and inoculated either intracerebrally (i.c.) with 20 μL of 2.5% brain homogenates or intraperitoneally (i.p.) with 200 μL of 1.0% brain homogenates from prion-infected or age-matched mock-infected mice. In intracerebral inoculation, brain homogenates (20 μL) were inoculated into the left hemisphere with two-stage needle of 27-gauge under anesthesia with sevoflurane.

For stereotaxic inoculation, eight-week-old female ICR mice were anesthetized with an intramuscular injection of xylazine (10 mg/kg) and ketamine (100 mg/kg), then placed on a stereotaxic apparatus (Narishige, Tokyo, Japan). A 1 cm midline incision was made on the dorsal skull surface to expose the skull and position the drill over the Bregma point. Coordinates for targeting the right thalamus were anteroposterior (AP) −1.7 mm, medial-lateral (ML), −1.5 mm from Bregma, and 4 mm depth; for the left striatum, AP 1 mm, ML 1.6 mm, and 4 mm depth. Coronal sections for the thalamus and striatum were at Plates 45–46 (Bregma between −1.67 and −1.79 mm) and 22–23 (Bregma between 0.97 and 1.09 mm), respectively (Paxinos and Franklin, 2013). These coordinates avoided ventricles. Brain homogenates (0.5 μL, 0.1% in PBS) from prion 22L strain-infected or age-matched mock-infected mice were stereotaxically inoculated into the thalamus or striatum. Injections were performed with Hamilton syringes (#80301, Hamilton Company, Reno, NV, USA) and 31-gauge steel bevel needles (#7750–22 Hamilton Company) at 0.062 μL/min using a microinjection pump (#78–8130, KD Scientific, Holliston, MA, USA). The needle remained in place for 2 min post-injection to prevent backflow. The skin incision was closed with a synthetic absorbable suture (#D6284, Ethicon, Raritan, NJ, USA). Needle patency was confirmed before and after injections.

Mice were euthanized under anesthesia with sevoflurane and their brains were taken and rinsed with PBS, embedded in Tissue-Tek O.C.T. compound (#4583, Sakura Finetek, Torrance, CA, USA) for frozen sections, and stored at −80°C. Ten μm sagittal plane frozen sections were cut using a cryostat (#Leica CM1950, Leica, Heidelberger Straße, Germany) and mounted on slide glasses (#APS-02, Matsunami, Osaka, Japan). For formalin-fixed paraffin-embedded (FFPE) sections, brains were fixed in 10% neutral buffered formalin (#062–01661, Wako, Osaka, Japan) for 5 days. Whole brains were then cut at the coronal plane to include the entire striatum or thalamus, dehydrated, and embedded in paraffin. FFPE sections were cut into 4 μm slices using a microtome and mounted on slide glasses (#APS-02, Matsunami). Sections were stained with hematoxylin (#30022, Carazzi’s Hematoxylin, Muto Pure Chemicals, Tokyo, Japan) and eosin (#32002, 1% Eosin Y Solution, Muto Pure Chemicals) or subjected to immunohistochemistry (IHC) for PrP^Sc detection.

The sections were deparaffinized, rehydrated, and autoclaved at 135°C for 30 min for PrP^Sc detection (Song et al., 2008). They were then treated with 3% H₂O₂ in methanol for 5 min, blocked with 5% fetal bovine serum (FBS) in PBS for 30 min at room temperature (r.t.) and incubated overnight at 4°C with anti-PrP monoclonal antibody (mAb) 132 (1 μg/mL). After washing with PBS, the sections were incubated for 1 h at r.t. with anti-mouse IgG F (ab)₂ -Peroxidase antibody (1:2,000, #A3682, Sigma-Aldrich, St. Louis, MS, USA). The sections were washed with PBS and developed using the DAB Substrate Kit, Peroxidase (HRP), with Nickel (#SK-4100, Vector Laboratories, Newark, CA, USA), followed by counterstaining with hematoxylin. Finally, the sections were mounted with Mount-Quick (#Mount-Quick, Daido Sangyo, Tokyo, Japan). Observations and image acquisition were performed using the Keyence BZ-X810 Fluorescence Microscope and BZ-X800 Viewer software (Keyence, Itasca, IL, USA).

The frozen brain sections were fixed with 4% paraformaldehyde (PFA) (#163–20,145, Wako) for 10 min at r.t. After washing with PBS containing 0.01% Tween 20, the sections were permeabilized with 0.1 M glycine and 0.1% Triton X-100 for 20 min at r.t., followed by incubation with 2.5 M guanidine hydrochloride for 15 min at r.t. The sections were then blocked with 5% FBS in PBS for 1 h at r.t. and incubated overnight at 4°C with Alexa Fluor 488-conjugated anti-PrP mAb 132 and anti-NeuN rabbit monoclonal antibody (1:2,000, #ab177487, Abcam, Cambridge, UK) in 1% FBS in PBS. For the secondary antibody reaction, the sections were incubated with goat anti-rabbit IgG F(ab’)₂ Alexa Fluor 555 antibody (1:1,000, #A-21430, Thermo Fisher Scientific, Waltham, MA, USA) and counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (#D3571, Thermo Fisher Scientific, 5 μg/mL in 1% FBS in PBS) for nuclei at r.t. for 1 h. Finally, the sections were mounted with Prolong Gold antifade reagent (#P36934, Thermo Fisher Scientific). Tile scan fluorescent images were acquired using a laser scanning confocal microscope LSM700 (Zeiss, Oberkochen, Germany) with a 20× objective lens.

In this study, thalamic (ThN), striatal (StN), and cerebral cortex neuronal cultures (CxN) were prepared from ICR mouse embryos at embryonic day 14. Pregnant mice were euthanized with sevoflurane, and the uterus containing the fetuses was removed. The brains of the fetuses were dissected in ice-cold PBS under a stereomicroscope. The medulla oblongata, hippocampi, and meninges were carefully removed, and the thalami, striata, and cerebral cortices were separately collected. Tissue dissociation, cell seeding, and inhibition of astrocyte proliferation were performed as previously described (Tanaka et al., 2016). At 7 days in vitro (div), primary neuronal cultures were exposed to microsomal fractions from prion 22 L-infected mouse brains, containing 5 ng rPrP equivalent proteinase K-resistant PrP (PrP-res) per 1.0 × 10⁵ cells, by replacing half the medium in each well. At 4 days post-inoculation (dpi), the medium was completely replaced with fresh neuronal medium.

Neurons cultured on 6-well plastic plates were lysed with 300 μL/well of lysis buffer (0.5% Triton X-100, 0.5% sodium deoxycholate, 150 mM NaCl, 5 mM EDTA, and 10 mM Tris–HCl [pH 7.5]). The cells underwent one freeze–thaw cycle followed by pipetting. After centrifugation, protein concentrations of the lysates were measured using a DC protein assay kit (#5000116, Bio-Rad-Life Science, Hercules, CA, USA) and adjusted to 0.5–0.8 mg/mL. SDS-PAGE and immunoblotting for PrP-res detection were performed as previously described (Tanaka et al., 2016; Tanaka et al., 2020).

PrP^Sc-specific immunofluorescence staining of prion-infected primary neuronal cultures using anti-PrP mAb 8D5 was performed as previously described (Tanaka et al., 2016; Tanaka et al., 2020). For staining Vglut1, Vglut2, and Vgat, cells were fixed with 4% PFA in PBS at r.t. for 10 min and permeabilized with PBS containing 0.1 M glycine and 0.1% Triton X-100 at r.t. for 10 min. After blocking with 5% FBS in PBS, the cells were incubated with primary antibodies: anti-Vgat rabbit polyclonal antibody (1:4,000, #131002, Synaptic System, Göttingen, Germany), anti-Vglut1 guinea pig polyclonal antibody (1:4,000, #135304, Synaptic System), and anti-Vglut2 guinea pig polyclonal antibody (1:2,000, #135404, Synaptic System). Anti-MAP2 chicken polyclonal antibody (1:2,000, #ab5392, Abcam) was used for staining neurons. Secondary antibodies used were goat anti-chicken IgY conjugated with Alexa Fluor 647 (1:4,000, #ab150173, Abcam) for detecting MAP2, goat anti-rabbit IgG Alexa Fluor 555 (1:4,000, #A21245, Thermo Fisher Scientific) for Vgat, and goat anti-guinea pig IgG Alexa Fluor 488 (1,4,000, #A11073, Thermo Fisher Scientific) for Vglut1 and Vglut2. Nuclei were counterstained with DAPI.

The STAIN Perfect Immunostaining Kit A (#SP-A-1000, ImmuSmol, Bordeaux, France) was used to stain L-glutamate and GABA in glutamatergic and GABAergic neurons in primary neuronal cultures. Cells were fixed, permeabilized, stabilized, and saturated with reagents from the kit according to the manufacturer’s instructions. They were then incubated overnight at 4°C with primary antibodies: anti-L-glutamate antibody (1:1,000, #IS1001, ImmuSmol), anti-GABA antibody (1:1,000, #IS1036, ImmuSmol), and mAb 8D5 (1 μg/mL) in the antibody diluent. After three washes with Wash 2 solution, cells were incubated for 60 min at r.t. with secondary antibodies: goat anti-rabbit IgG conjugated with Alexa Fluor 555 antibodies (1:4,000, #A21245, Thermo Fisher Scientific) for anti-L-glutamate, goat anti-chicken IgY conjugated with Alexa Fluor 647 antibodies (1:4,000, #ab150173, Abcam) for anti-GABA, and F(ab’)₂-goat anti-mouse IgG conjugated with Alexa Fluor 488 antibodies (1:4,000, #A-11017, Invitrogen) for PrP^Sc in 1% FBS-PBS. Nuclei were counterstained with DAPI.

Stained cells were observed using a Zeiss LSM800 laser scanning microscope (Zeiss) with a 20x objective lens. For higher resolution, Airyscan using highly sensitive GaAsP detectors on the Zeiss LSM800 with a 63x objective lens. Acquired images were analyzed with ZEN 2.6 software (blue edition).

The RNAscope^® 2.5 HD Detection Reagent-RED kit (#322360, ACDbio, Newark, CA, USA) was used for RNAscope in situ hybridization following the manufacturer’s instructions. FFPE tissue sections were baked at 60°C for 1 h and then deparaffinized. The sections were treated with Hydrogen Peroxide (#322381, ACDbio) for 10 min at r.t. After washing twice with deionized water (DW) for 30 s at r.t., target retrieval was performed by heating the sections at 94°C for 15 min in Target Retrieval Reagent (#322000, ACDbio). After another 30 s wash with DW at r.t., the sections were dehydrated in 100% ethanol for 2 min at r.t. and dried at 60°C for 5 min. Next, the sections were treated with Protease Plus (#322381, ACDbio) for 30 min at 40°C in a moisture chamber within a hybridization incubator and washed twice with DW for 1 min at r.t. The RNAscope probes, including positive control (Ppib, #313911), negative control (Dapb, #310043), Vglut1 (Slc17a7, #416631), Vglut2 (Slc17a6, #319171), and Vgat (Slc32a1, #319171) (ACDbio), were pre-warmed at 40°C for 10 min and then hybridized to the tissue sections for 2 h at 40°C in the moisture chamber. After hybridization, the sections were washed with Wash Buffer (#310091, ACDbio) for 2 min at r.t. and then proceeded to signal amplification. For this step, the sections were sequentially incubated with AMP1 for 30 min, AMP2 for 15 min, AMP3 for 30 min, AMP4 for 15 min, AMP5 (containing preamplifiers and amplifiers) for 30 min, and AMP6 for 15 min at 40°C in a moisture chamber, following the manufacturer’s instructions. Color development was performed using a mixture of RED-A and RED-B solutions at a 60:1 ratio for 10 min at r.t. in the dark. The development was stopped by washing with tap water for 2 min. For counterstaining, the sections were stained with hematoxylin. To enhance contrast between the reddish color developed by the kit and the reddish color of hematoxylin staining, the color of hematoxylin was converted to blue by soaking the sections in 0.02% ammonia for 10 s at r.t. Finally, the tissue sections were dehydrated at 60°C for 15 min, soaked twice in xylene for 3 s each, and mounted with Mount-Quick.

RNAscope staining bright field images were captured using a Nano Zoomer 2.0RS digital slide scanner (Hamamatsu Photonics, Hamamatsu, Japan). The virtual slide images were read and cropped with QuPath, an open-source image analysis software. Four brain regions were used as indicators for cropping thalamus images: the hippocampus (dorsal border), the third ventricle (medial border), the reticular nucleus (RT) (lateral border), and the zona incerta (ventral border). The cropped images were further divided into 500 × 500 pixel (112.7 × 112.7 μm) parcels for annotation and analysis. The number of parcels covered the entire thalamus: 15–18 parcels for the mediolateral axis and 17 parcels for the dorsoventral axis.

To develop a deep learning model for classifying RNAscope-positive and negative cells, Microsoft VoTT (Visual Object Tagging Tool, Microsoft, Redmond, WA, USA) was used for annotation. Sections counterstained with hematoxylin allowed identification of RNAscope-positive cells by red punctate signals around hematoxylin-labeled nuclei. A total of 366 images were annotated, identifying 2,929 RNAscope-positive cells and 14,115 negative cells, resulting in 17,044 cells. The annotated dataset was exported to Roboflow (ver1.1.44, Roboflow Inc., Iowa, USA) to create training, validation, and test splits: 70% for training, 20% for validation, and 10% for testing.

The computer-assisted object detection and classification algorithm, You Only Look Once version 5 (YOLOv5) (Jocher et al., 2021; Redmon et al., 2016) was used for precise classification of stained cells. YOLOv5x served as the trained model. The optimal model was selected with a learning rate between 0.01 and 0.1. Model fitness was evaluated using mean average precision (mAP), calculated by averaging the precision (AP) over two classes and/or various intersection over union (IoU) thresholds. IoU thresholds were determined by dividing the area of overlap by the area of union. Two mAP values, using IoU thresholds of 0.5 and 0.5–0.9, were employed and weighted at 0.1 and 0.9, respectively.

Statistical analyses were performed using JMP Pro version 17 (JMP, Cary, NC, USA).

To investigate PrP^Sc distribution in prion 22L strain-infected mice brains, we analyzed PrP^Sc deposition using sagittal cryosections from mice inoculated i.c. or i.p. with prions, at 120 and 186 dpi, respectively, around the clinical onset (Figure 1). PrP^Sc signals were absent in age-matched mock-infected mice (Figures 1A,E), while intense PrP^Sc signals were observed in the cerebral cortex and thalamus of i.c. inoculated mice (Figure 1I). Conversely, PrP^Sc signals in the striatum were notably weaker than in the cerebral cortex and thalamus (Figure 1I). The PrP^Sc distribution in i.p. inoculated mice was similar to that in i.c. inoculated mice; intense signals were seen in the cerebral cortex and thalamus, but weaker in the striatum (Figure 1M). This weak PrP^Sc staining in the striatum, regardless of inoculation route, suggests inefficient 22 L prion propagation in the striatum. A similar weaker PrP^Sc signal in the striatum compared to the cerebral cortex and thalamus was also observed in mice i.c. inoculated with Chandler and Obihiro strains (Supplementary Figure 1).

The weak PrP^Sc deposition in the striatum suggests that prions propagate more efficiently in the cerebral cortex and thalamus. To further analyze this, we stereotaxically injected 22L prions into either the thalamus or striatum and analyzed PrP^Sc deposition by IHC (Figure 2). No visible PrP^Sc-positive stains were observed in the thalamus and striatum after inoculating age-matched mock-infected brain homogenates (Figures 2A–C,G–I). In the thalamus, positive signals were observed along the needle track at 1 dpi, likely from the PrP^Sc in the inoculum (Figure 2D). By 14 dpi, diffuse PrP^Sc stains were observed both along and away from the needle track (Figure 2F). Conversely, in the striatum, PrP^Sc stains were observed mainly along the needle tracks up to 14 dpi (Figures 2J–L). These results suggest that 22L prions propagate more in the thalamus than in the striatum. We did not analyze prion propagation in the cerebral cortex due to its insufficient thickness for stereotaxic injection.

Reports indicate that the cerebral cortex has a high proportion of Vglut1-positive excitatory neurons, while Vglut1 or Vglut2-positive neurons are present in the thalamus, and Vgat-positive inhibitory neurons predominate in the striatum (Fremeau et al., 2004b; Rudy et al., 2011). Given the inverse relationship between GABAergic neuron distribution and PrP^Sc deposition in the striatum (Figure 1), we first confirmed the gene expression of Vgat (a marker for GABAergic inhibitory neurons) and Vglut1 and Vglut2 (markers for glutamatergic excitatory neurons) in mouse brains using RNAscope in situ hybridization (Figure 3). Vglut1-positive cells were mainly observed in the cerebral cortex, pyramidal cell layers, dentate gyrus of the hippocampus, thalamus, and granular cell layer of the cerebellum. Vglut2-positive cells were primarily found in the thalamus and midbrain. On the contrary, Vgat-positive cells were predominantly located in the striatum and reticular nucleus of the thalamus, with additional presence in the midbrain and sparse distribution in the cerebral cortex (Figure 3).

The predominance of GABAergic neurons (Figure 3) and the inefficient 22 L prion propagation in the striatum (Figures 1, 2) suggest a neuronal cell-type-dependent propagation efficiency of the 22 L strain. To investigate this, we prepared primary neuronal cultures from cortices (CxN), thalami (ThN), and striata (StN) to analyze prion propagation in vitro based on neuronal cell type.

First, we examined the expression of Vglut1 and Vglut2, synaptic vesicle membrane transporters specific to glutamatergic excitatory neurons, and Vgat, a synaptic GABA transporter of GABAergic and glycinergic neurons, in each primary neuronal culture using immunofluorescence assay to assess if the proportion of Vglut1-, Vglut2-, and Vgat-positive neurons in the cultures reflects their proportion in the brain regions from which the tissues were collected (Figure 4). Vglut1-positive neurons were predominantly observed in CxN, with a few Vglut2- and Vgat-positive neurons also present. In ThN, Vglut2-positive neurons were mainly observed, with some Vgat-positive neurons. Conversely, Vgat-positive neurons were the most abundant in StN (Figures 4A,B). Since the proportion of Vglut1-, Vglut2-, and Vgat-positive neurons in each neuronal culture from mouse embryos appears roughly consistent with their proportions in adult mouse brains (as shown by RNAscope analysis in Figure 3), we used CtN, ThN, and StN to analyze whether prion propagation differs depending on neuronal cell type.

We previously reported that anti-PrP mAb 8D5 detects PrP^Sc as string-like stains in prion-infected primary neurons under non-denatured conditions (Tanaka et al., 2016; Tanaka et al., 2020). Figure 5A shows immunofluorescence detection of PrP^Sc using mAb 8D5 in neurons exposed to microsomal fractions from 22L-infected mouse brains. String-like stains were observed in all three neuronal cultures infected with 22L prions, while no PrP^Sc stains were seen in mock-infected cultures (Figure 5A). Many of the string-like PrP^Sc stains appeared to be outside of primary neurons. Most of PrP^Sc in prion-infected primary neurons are located on the cell surface (Tanaka et al., 2020), whereas MAP2 is an intracellular protein interacting microtubules of the cellular cytoskeleton. Thus, the string-like PrP^Sc stains appeared apart from MAP2 positive regions is due the difference in the cellular localization of PrP^Sc and MAP2. Immunoblotting revealed time-dependent increases in PrP-res from 7 to 28 dpi in CxN, ThN, and StN exposed to 22 L prions, indicating successful prion propagation (Figure 5B). Additionally, PrP-res levels in CxN appeared higher than in ThN and StN (Figure 5C). Although neurons with string-like PrP^Sc stains were present in all three cultures, the frequency of such neurons appeared to vary. Therefore, we classified PrP^Sc-positive neurons based on the number of string-like PrP^Sc stains (Tanaka et al., 2020) to compare prion propagation efficiency among neuronal cultures.

Figure 6A illustrates the classification of PrP^Sc-positive neurons based on the number of string-like PrP^Sc signals on the surface of the soma. Neurons with five or more, fewer than five, and no apparent string-like PrP^Sc signals surrounding the soma were categorized as PrP^Sc ++, PrP^Sc +, and PrP^Sc low neurons, respectively (Figure 6A). We conducted three independent experiments, classifying a total of 131, 205, and 176 neurons from CxN, ThN, and StN, respectively, into these categories (Figure 6B). The frequency of PrP^Sc ++, PrP^Sc +, and PrP^Sc low neurons varied significantly among CxN, ThN, and StN (χ² test, p < 0.05). The frequency of PrP^Sc ++ neurons was significantly higher than the expected count in CxN, while PrP^Sc + and PrP^Sc low neurons were significantly lower than the expected count. In contrast, StN had significantly lower frequencies of PrP^Sc ++ and higher frequencies of PrP^Sc low neurons (residual analysis following χ² test). These results suggest that the 22L strain propagates more efficiently in CxN than in StN.

The predominance of Vglut1-positive neurons in CxN aligns with the abundance of Vglut1-positive neurons in the cerebral cortex, while the predominance of Vgat-positive neurons in StN corresponds with the abundance of GABAergic neurons in the striatum. These observations suggest that the propagation efficiency of the 22L strain differs between glutamatergic excitatory and GABAergic inhibitory neurons. To explore this possibility, we first distinguished glutamatergic neurons from GABAergic neurons using the STAIN Perfect Immunostaining Kit A (Supplementary Figure 2). The kit identified glutamatergic neurons as L-glutamate-positive but GABA-negative, and GABAergic neurons as both L-glutamate- and GABA-positive. Most neurons in the CNS use L-glutamate as an excitatory neurotransmitter, while GABA is used as an inhibitory neurotransmitter in inhibitory neurons (Amara and Kuhar, 1993). As expected, most neurons in primary cultures were divided into two populations: L-glutamate-positive but GABA-negative neurons (Supplementary Figure 2, upper panel) and L-glutamate- and GABA-positive neurons (Supplementary Figure 2, lower panel). CxN comprised approximately 70% glutamatergic and 23% GABAergic neurons; ThN had about 62% glutamatergic and 29% GABAergic neurons, while StN consisted of roughly 71% GABAergic and 24% glutamatergic neurons (Supplementary Figure 2).

We then performed multiple immunofluorescence staining using the STAIN Perfect Immunostaining Kit A to detect L-glutamate and GABA, along with mAb 8D5 for PrP^Sc. Figure 7A shows representative images for classifying PrP^Sc ++, PrP^Sc +, and PrP^Sc low neurons in either glutamatergic (upper panel) or GABAergic neurons (lower panel). Neurons were categorized based on the string-like PrP^Sc stains around the soma, as in Figure 6. A total of 145, 209, and 185 neurons in CxN, ThN, and StN, respectively, were analyzed (Figure 7B). The frequencies of PrP^Sc ++, PrP^Sc +, and PrP^Sc low neurons in GABA-negative and GABA-positive neurons in CxN differed significantly (p < 0.05, Fisher’s exact test). Significant differences were also observed in ThN (p < 0.05, χ² test), with residual analysis showing a significantly lower frequency of PrP^Sc low GABA-negative neurons and a significantly higher frequency of PrP^Sc low GABA-positive neurons (p < 0.05). Significant differences were also observed in StN (p < 0.01, χ² test); the frequency of PrP^Sc ++ GABA-negative neurons was significantly higher, while that of PrP^Sc ++ GABA-positive neurons was significantly lower than the corresponding expected counts. Given the varying proportions of GABA-negative and GABA-positive neurons among the three neuronal cultures, we combined the results and compared the frequencies of PrP^Sc ++, PrP^Sc +, and PrP^Sc low in GABA-negative and GABA-positive neurons (Figure 7B). The frequencies of PrP^Sc ++ neurons were significantly higher than the expected count in GABA-negative neurons and lower than the expected count in GABA-positive neurons (p < 0.01). Conversely, the frequencies of PrP^Sc low neurons were significantly lower in GABA-negative neurons and higher in GABA-positive neurons (p < 0.01). These findings indicate that the 22 L prion strain propagates more efficiently in GABA-negative glutamatergic neurons than in GABA-positive GABAergic neurons in primary neuronal cultures.

Our results indicate that the 22L strain propagates more efficiently in glutamatergic neurons than in GABAergic neurons. We have observed massive neuronal loss in the lateral thalamus of Jcl:ICR mice infected with the 22L, Chandler, and Obihiro strains (unpublished observation), similar to previous reports (Cunningham et al., 2005; Reis et al., 2015). Thus, to identify the types of neurons lost in the lateral thalamus of 22L strain-infected mice, we analyzed Vglut1, Vglut2, and Vgat-expressing cells using RNAscope in situ hybridization (Figure 8). Vglut1 and Vglut2 expression was observed in both the lateral and medial thalamus (Figure 8A). Compared to age-matched mock-infected mice, the number of Vglut1- and Vglut2-expressing neurons significantly decreased in the ventral posterolateral nucleus (VPL). In in the ventral posteromedial nucleus (VPM), Vglut1-expressing neurons tended to decrease but no apparent decrease in Vglut2-expressing neurons was observed in 22L strain-infected mice (Figure 8B). No decrease in Vgat-expressing neurons was observed in the reticular nucleus of the thalamus (RT), which is adjacent to VPL and VPM and primarily composed of Vgat-expressing interneurons.