key: cord-0281307-w2kq9l9u
authors: Zhang, Yachun; Xing, Xudong; Long, Ben; Cao, Yandi; Hu, Simeng; Li, Xiangning; Yu, Yalan; Tian, Dayong; Sui, Baokun; Luo, Zhaochen; Liu, Wei; Lv, Lei; Wu, Qiong; Dai, Jinxia; Zhou, Ming; Han, Heyou; Fu, Zhen F.; Gong, Hui; Bai, Fan; Zhao, Ling
title: A spatial and cellular distribution of neurotropic virus infection in the mouse brain revealed by fMOST and single cell RNA-seq
date: 2021-03-26
journal: bioRxiv
DOI: 10.1101/2021.03.26.436691
sha: bb1e102463119377d4cf7b0a93a2c2d10831e8d3
doc_id: 281307
cord_uid: w2kq9l9u

Neurotropic virus infection can cause serious damage to the central nervous system (CNS) in both human and animals. The complexity of the CNS poses unique challenges to investigate the infection of these viruses in the brain using traditional techniques. In this study, we explore the use of fluorescence micro-optical sectioning tomography (fMOST) and single cell RNA sequencing (scRNA-seq) to map the spatial and cellular distribution of a representative neurotropic virus, rabies virus (RABV), in the whole brain. Mice were inoculated with a lethal dose of recombinant RABV expressing enhanced green fluorescent protein (EGFP) under different infection routes, and a three-dimensional view of the distribution of RABV in the whole mouse brain was obtained using fMOST. Meanwhile, we pinpointed the cellular distribution of RABV by utilizing scRNA-seq. Our fMOST data provide the first evidence that RABV can infect multiple nuclei related to fear independent of different infection routes. More surprisingly, scRNA-seq data indicate that besides neurons RABV can infect macrophages and NK cells in vivo. Collectively, this study draws a comprehensively spatial and cellular map of RABV infection in the mouse brain, providing a novel and insightful strategy to investigate the pathogenesis of neurotropic viruses.

Neurotropic virus infection can cause serious damage to the central nervous system (CNS) in both 27 human and animals. The complexity of the CNS poses unique challenges to investigate the 28 infection of these viruses in the brain using traditional techniques. In this study, we explore the use 29 of fluorescence micro-optical sectioning tomography (fMOST) and single cell RNA sequencing 30 (scRNA-seq) to map the spatial and cellular distribution of a representative neurotropic virus, 31 rabies virus (RABV), in the whole brain. Mice were inoculated with a lethal dose of recombinant 32 RABV expressing enhanced green fluorescent protein (EGFP) under different infection routes, 33 and a three-dimensional view of the distribution of RABV in the whole mouse brain was obtained 34 using fMOST. Meanwhile, we pinpointed the cellular distribution of RABV by utilizing 35 scRNA-seq. Our fMOST data provide the first evidence that RABV can infect multiple nuclei 36 related to fear independent of different infection routes. More surprisingly, scRNA-seq data 37 indicate that besides neurons RABV can infect macrophages and NK cells in vivo. Collectively, 38 this study draws a comprehensively spatial and cellular map of RABV infection in the mouse 39 brain, providing a novel and insightful strategy to investigate the pathogenesis of neurotropic 40 viruses. 41

Neurotropic virus infection can cause an array of immediate and delayed neuropathology in 43 human and animals, and nearly half of the emerging viruses can invade the CNS 1 . These 44 infections pose a major challenge to human and animal healthcare worldwide. The complex 45 structures and functions of the CNS and the diversity of neurotropic viruses make it difficult to 46 find an effective treatment of these diseases. By combining human pathological data with 47 experimental animal models, virologists have much advanced our understanding of the 48 mechanisms underlying how viruses enter the CNS and cause neurological disease, but more 49 in-depth studies are still in urgent need to facilitate the development of novel vaccines and 50 antiviral therapeutics 2 . Rabies virus (RABV) is a typical neurotropic virus belonging to 51

Rhabdoviridae family. RABV hijacks the cellular transport machinery and moves along 52 microtubules by retrograde axonal transport to the nearest sensory neurons in the dorsal root 53 ganglion or anterior horn of the spinal cord 3 . After replication, RABV travels along the 54 corticospinal tract to the brain where the virus can efficiently replicate and infect in most regions, 55 resulting in fatal encephalitis 4 . However, due to the complexity of the CNS, the exact regions and 56 cell types infected by RABV are still kept elusive, impeding the endeavor of researchers to find 57 the therapeutic target for RABV. 58 For many decades, light microscopy was the key tool for investigating the invasion and spread 59 of neurotropic viruses in the host's brain. However, to observe a whole animal brain, which is 60 centimeter-sized, is beyond the view field of modern light microscopy techniques including 61 confocal and multiphoton microscopy. Histological sections were usually prepared for observing 62 the internal microstructures of large specimens, but it is difficult to align the serial figures of 63 continuous ultrathin sectioning on large specimens by a traditional light microscopy. In 2010, a 64 micro-optical sectioning tomography (MOST) system was developed, which allowed the mapping 65 of a whole mouse brain to the single neuron level 5 . In 2013, a fluorescence MOST (fMOST) was 66 developed by using a resin-embedding method for maintaining fluorescence and an automated 67 fluorescence MOST system for long-term stable imaging 6 . Recently, a modified fMOST, 68 benefiting from simple sample preparation and high-throughput imaging, was applied to determine 69 LS and SC etc. 14 . Interestingly, under all three infection routes, we identified the extensive RABV 154 infection in these nuclei ( Figure S5B ) and the anatomical localization of these nuclei was shown 155 in Figure S5A . Amygdala is known to play the central role in processing fear 15, 16 , and is divided 156 into cortical division and striatal division, the later one including medial amygdala (MEA) and 157 central amygdala (CEA). Interestingly, we found that there is extensive distribution of RABV in 158 CEA. Moreover, a newly found fear-related area, BST, as well as other fear-associated regions 159 including HPF, PAG, PVH etc. were also extensively infected by RABV. Quantification of 160 RABV-infected cells suggested that the infection routes had no significant impact on RABV 161 distribution in the nuclei related to fear ( Figure S5C ), indicating that these nuclei are tightly 162 associated with the pathogenesis of rabies. 163

Emerging evidences have shown that RABV is not a strict neurotropic virus and besides neurons it 165 can infect other cell types 17, 18 . To comprehensively investigate the infection susceptibility of 166 diverse cell populations and their contributions to RABV pathogenesis, we performed 167 droplet-based scRNA-seq (10×Genomics) on a total of six mouse brain samples including two 168 from uninfected mice (healthy: n=2), two from mice with paralysis (paralyzed: n=2) and two from 169 mice in the moribund stage post RABV infection (moribund: n=2) ( Figure 4A ). Following 170 euthanasia, mouse brain tissue was obtained and rapidly digested to a single cell suspension. The 171 single cell suspension from the uninfected brains was directly subjected to scRNA-seq. While the 172 single cell suspension from RABV-infected brains was firstly enriched for cells containing EGFP 173 signals (with active RABV infection) by FACS, and then analyzed by scRNA-seq (Figures 4A, 174 S6A and S6B) . With the unified single-cell analysis pipeline,~0.77 billion unique transcripts were 175 obtained from 54,452 cells. Among these cells, 16,199 cells (29.75%) were from the healthy mice, 176 13,417 cells (24.64%) were from the paralyzed mice and 24,836 cells (45.61%) were from the 177 moribund mice ( Figure S5C ). All high-quality cells were integrated into an unbatched and 178 comparable dataset and subjected to principal component analysis after correction for read depth 179 and mitochondrial read counts ( Figure S6D ). 180

Using graph-based clustering, uniform manifold approximation and projection (UMAP), we 181 captured the transcriptomes of 21 high confidence cell clusters ( Figure 4B ) according to the 182 expression of canonical markers ( Figure 4C ). Overall, the landscape contained the following cell 183 lineages: T cells (Cd3d + ), natural killing (NK) cells (Klrk1 + ), B/plasma cells (Cd79a + ), monocytes 184 (Plac8 + ), macrophages (Ms4a7 + Lyz2 + ), granulocytes (S100a8 + ), monocytes derived dendritic cells 185 (H2-Aa + ), microglia (P2ry12 + ), neurons (Meis2 + ), oligodendrocytes (Ermn + ) and its precursors 186 (Neu4 + ), astrocytes (Aldoc + ), ependymal cells (Tmem212 + ), schwann cells (Ttr + ), endothelial cells 187 (Cldn5 + ), fibroblasts (Dcn + ), meningeal cells (Slc4a10 + ) and mural cells (Vtn + ). 188

Consistent with previous studies 19, 20 , the uninfected mouse brain consisted of microglia, 189 neurons, endothelial, oligodendrocytes, and etc. ( Figure 4D and Figure Macro-C1 and Macro-C2 were predominantly found in the paralyzed and moribund mouse brains, 205

whereas Macro-C4 was predominantly observed in the healthy mouse brains. Macro-C3 could be 206 found in both uninfected and infected brains, while its majority resided in the paralyzed and 207 moribund mouse brains ( Figures 5B and S7A) . 208

To characterize each cluster, we first identified differentially expressed genes (DEGs) and then 209 performed functional enrichment analyses ( Figures 5C and 5D ). Interferon response related DEGs, 210 such as Isg15, Ifi27l2a, Ifit3, Oasl1 and Ifi209 were exclusively expressed in Macro-C1. In line 211 with this, functional enrichment analyses revealed that Macro-C1 was an interferon 212 response-related cluster. Macro-C2 was found to be related to cellular detoxification, metabolic 213 and catabolic processes due to highly expressed proteases genes, such as Psmb, Psmd and Gsr. 214 Macro-C3 highly expressed MHC-II genes, and functional enrichment analyses suggested that 215

Macro-C3 might take part in antigen processing and presentation. Macro-C4 exclusively 216 expressed Mrc1, Cd163, Gas6, Clec10a, Folr2 and related to myeloid leukocyte migration, similar 217 to the recently reported border-associated macrophages (BAMs), a brain resident macrophage 218 residing in the dura mater, subdural meninges and choroid plexus 22 . 219

Macrophages are conventionally classified into canonical M1 and M2 classes, the 220 pro-inflammatory and anti-inflammatory macrophages, respectively 23 . We found that no 221 macrophage cluster exhibited only M1 or M2-like phenotype, whereas Macro-C4 exhibited a more 222 M2-dominant gene signature, such as Cd163, Mrc1 and Ccl24 ( Figure 5E ). These data indicated 223 that macrophage activation during RABV infection did not agree with the polarization model, 224 either as discrete states or along a spectrum of alternative polarization trajectories. 225

Our data probably captured infiltrating macrophages asynchronously transitioning from one 226 transcriptomic state to the next, we thus employed Monocle2 algorithm 24 to perform the 227 pseudotime analysis. The inferred dynamic trajectory progressing exhibited a typical branched 228 structure: with Macro-C1 as the root, Macro-C2 and Macro-C3 as the ending clusters (Figures 6A 229 and S7C) . To confirm that the ordering was correct, three marker genes were selected and plotted. 230

As shown in Figures 6B and 6C , the expression level of Isg15 decreased along the pseudotime; the 231 expression level of Psmb5 peaked in the middle of the pseudotime and the expression level of 232 H2-DMb1 was the highest at the end of the pseudotime, demonstrating a reasonable ordering. 233

To understand the biological processes driving pseudotime components, we wondered which 234 genes covary in expression with pseudotime. We clustered the representative genes identified as 235 significantly covarying with pseudotime and identified three groups of genes expressed early, 236 mid/mid-late and late ( Figure 6D ), consistent with the clusters identified above. Taken together, 237 our results suggest that there are at least three different roles for the infiltrating macrophages, 238 which we termed interferon-responsive macrophages, proteasome-active macrophages, and 239 antigen processing and presentation macrophages during RABV infection in the brain. 240

NK cells are considered to be an important player of the innate immunity by controlling microbial 242 infections 25 . Intriguingly, we found that there was an obvious enrichment of NK cells in the 243 mouse brain after RABV infection ( Figure 4E ). Re-clustering of the total of 719 NK cells revealed 244 that there were at least three sub-clusters (NK-C1 to C3) ( Figures Figures 7E and 7F ). These results indicate that NK cells were impaired by 259 RABV infection with higher exhausted and apoptotic degree than that with healthy condition. 260

The inferred dynamic trajectory progressing of three NK cell types also exhibited a typical 261 branched structure with NK-C1, NK-C2 and NK-C3 distributed as the root, middle and ending 262 clusters, respectively. We found that the pseudotime was also consistent with developmental 263 conditions where the start corresponded to the healthy condition and the end to the moribund 264 condition ( Figures 7G and S8C ). The expression level of Eomes decreased along the pseudotime, 265 the expression level of Gzmb peaked at the middle and of Ifi205 was the highest at the end of the 266 pseudotime ( Figures 7H and 7I ). To understand the biological processes driving pseudotime 267 components, we further clustered the representative genes identified as significantly covarying 268 with pseudotime and found the marker genes clustered by pseudotemporal expression pattern also 269 revealed the same three clusters following similar kinetic trends ( Figure 7J) . 270

Due to the complexity of the brain, and lacking appropriate tools, researchers are facing great 272 difficulty to investigate the invasion and distribution of neurotropic viruses in the brain, which 273 impedes the invention of effective therapies for neurotropic viruses. The CLARITY, approach 274 creates a tissue-hydrogel hybrid in which tissue components are replaced with exogenous elements. 275

And then light-microscopy techniques can be used to access the entire mouse brain 27 . Recently, 276 three-dimensional light sheet and confocal laser scanning microscopy was also applied to 277 investigate the distribution of RABV in the brain and peripheral nerves 28 . fMOST technology 278 makes it possible to map the spread of neurotropic viruses in the whole brain. It rapidly acquires a 279 full-volume dataset in a whole mouse brain with resolution to a single neuron. The recently 280 emerging scRNA-seq provides an efficient technique to investigate the transcriptomic 281 characteristics during infection at single cell resolution. In this study, we investigated the spatial 282 and cellular distribution of neurotropic virus infection in the mouse brain by jointly using fMOST 283 and scRNA-seq. 284

We identified specific nuclei in which viral spread or the viral load depended on the route of 285 infection. Considering the role infection route has on the distribution of RABV in the brain, those 286 regions commonly infected are likely critical for deciphering RABV pathogenesis and the 287 potential targets for future therapeutics. Considering the influence of infection route on viral 288 distribution, knowledge of those regions that are commonly infected is especially important for 289 understanding RABV pathogenesis. Previous studies have suggested that aggressive behavior is 290 critical for RABV pathogenesis, because it leads to the efficient transmission of RABV to other 291 hosts by the rabid animals. In RABV-infected skunks, the red nucleus (RN) and midbrain raphe 292 nuclei were found to be associated with aggressive behavior 29 . Notably, our data indicate that 293 RABV is distributed to the RN only post i.m. inoculation (Figure 3) . 294

Fear is another cardinal characteristic of rabies, which is an emotion that has powerful influence 295 on behavior and physiology. Fears, depending on the type of threat stimuli, are processed in 296 independent neural circuits that involve the amygdala and downstream hypothalamic and 297 brainstem circuits 16 . Rabies patients often have abnormal symptoms like hydrophobia, aerophobia 298 and phonophobia indicating that there is a disorder in the processing of fear. In this study, we 299 found that several nuclei related to fear, such as CEA, BNST, PAG, and BLA, were extensively 300 infected by RABV ( Figure S4) . Notably, the route of infection made no difference in the viral load 301 in these areas, indicating that RABV infection in fear-related areas is independent of the infection 302 route. 303

Besides fear, inspiratory muscle spasms are another cardinal feature of rabies infection. In the 304 late stages, periodic and ataxic breathing is often observed 30 . Previous studies have shown that the 305 midbrain and medulla, especially the pontine tegmentum, are the regions most affected by 306 RABV-induced inflammation 30 . Accumulating evidence suggest that pontine nuclei, including 307

Kölliker-Fuse (K-F) and parabrachial (PB) nuclei comprise the pontine respiratory group (PRG), 308 which regulates the inspiratory-expiratory phase transition 31, 32 . Other studies report that neurons 309 in the superior colliculus (SC) 33 , RN 33 , and intermediate reticular nucleus (IRN) 34 also influence 310 the respiratory cycle. In our study we showed that the PB, SC, RN and IRN were extensively 311 infected by RABV regardless of the infection route (Figure 3) , suggesting that the abnormal 312 breathing may be related to the neuronal damage caused by RABV infection in these nuclei. 313

The fMOST results provide the spacious distribution of RABV in the brain, but could not 314 determine the exact infected cell types. To further investigate the cellular distribution of RABV, 315 scRNA-seq analysis was implemented and the results revealed that cell types in the brain from 316 healthy mice consisted of microglia, neurons, endothelial, oligodendrocytes, and etc. which was 317 consistent with the previous reports 19, 20 . Different from fMOST data which showed that most of 318 the RABV-infected cells were neurons, the vast majority of RABV-positive cells isolated from the 319 brains of paralyzed and moribund mice were immune cells, such as macrophages and NK cells. In summary, we utilized fMOST technology to reveal the susceptible nuclei infected by RABV 368 in the mouse brain, and take advantage of scRNA-seq technology to analyze the RABV-infected 369 cells and illustrate the roles of some immune cells. The joint use of these two technologies 370 allowed us to portray an integrated map of RABV infection in the mouse brain. Our results shed a 371 light for future investigation of the pathogenesis and clinical therapy of rabies and other 372 neurotropic viruses such as ZIKA and dengue viruses. 373

The experiments involving mice in this study were performed in accordance with the 376 recommendations in the Guide for the Care and Use of Laboratory Animals of the Ministry of 377 Science and Technology of China and were approved by the Scientific Ethics Committee of 378

Huazhong Agricultural University (permit number HZAUMO-2016-052). 379

Groups of six-week-old female C57BL/6 mice were inoculated with 10×LD50 RABV-EGFP by 381 intramuscular (i.m.), otic subcutaneous (o.s.), or intranasal (i.n.) route (n=3). In severe paralysis, 382 mice were anesthetized with ketamine/xylazine and then perfused by intra-cardiac injection of 383 PBS followed by 10% neutral-buffered formalin. Brains were removed and fixed with 10% 384 neutral-buffered formalin for 24 hours at 4 ℃. After fixation, each brain was rinsed overnight at 385 4 ℃with PBS and embedded with oxidized agarose before imaging. 386

The agarose-embedded brains were imaged by fMOST system 8 . During the process of imaging, 388 the brain was immersed in water and a water immersion objective (1.0 NA, 20×) was used. The 389 fluorescent signals were detected via a scientific complementary metaloxide semiconductor 390 (sCMOS) camera, a modern scientific camera with highly sensitive and high-speed. A 3D stage 391 accurately moved the brain mosaic-by-mosaic to extend the field of view in the imaging part, and 392 then the stage moved the brain towards the oscillatory blade to remove the imaged tissue in the 393 vibrating sectioning part. The full-volumetric imaging was performed with the cycle of imaging 394 and sectioning until the whole-brain dataset was finished. For a single mouse brain, the dataset 395 included approximately 3000 coronal sections was collected in three days at a voxel resolution of 396 0.32 µm × 0.32 µm ×5 µm. 397

Image preprocessing for mosaic stitching and uneven lateral illumination correction was 399 performed as reported previously 7 . We visualized the datasets using Amira software (v 5.2.2, FEI, 400 France) to generate the figures of maximum intensity projection, volume, surface rendering, and 401 for the movies. The preprocessed dataset was imported into Amira software using a desktop 402 graphical workstation (T7600, Dell Inc., USA). 403

To quantify the infected cells, the EGFP-labeled cells were automatically segmented and 405 coregistered to Mouse Reference Atlas as previously described 44, 45 . Briefly, each coronal section 406 was background subtracted, Gaussian filtered, and threshold segmented to binary image, and the 407 infected cells were segmented with individually adjusted binary thresholds according to varying 408 fluorescent intensities. And then the soma coordinates were warped and coregistered to the 409 corresponding Allen Mouse Reference Atlas (Allen Institute for Brain Science) coordinate using 410 non-rigid registration with free-form deformation. The anatomical sub-regions of which infected 411 cells belonged to were then mapped to the Allen Mouse Reference Atlas. We calculated the total 412 number and the density of EGFP-labeled infected cells for each anatomical sub-region with 413 different infection routes (n = 3 per infection route). 414

Groups of six-week-old female C57BL/6 mice were i.m. inoculated with 10×LD50 RABV-EGFP. 416

At the stage of paralysis and moribund, mice were anesthetized with ketamine/xylazine and then 417 brains were collected. Single-cell suspensions were obtained with adult brain dissociation kit 418 (Miltenyi Biotec, . EGFP positive cells were sorted using the GFP channels of the 419 Bio-Rad S3e instrument. cDNA libraries were prepared from single-cell suspensions following the 420 instruction of 10×Genomics 3' V3. Cells obtained from the whole brain of the healthy mice with 421 the same method and enriched by flow cytometry were used as control. RNA-sequencing was 422 performed by Novogene (Nanjing, China). All details regarding the Seurat analyses performed in this work can be found in the website 439 tutorial (https://satijalab.org/seurat/v3.0/pbmc3k_tutorial.html). 440

To compare cell types and proportions across three conditions, we employed the integration 442 methods described at https://satijalab.org/seurat/v3.0/integration.html 47 . The Seurat package 443 (Version 3.0.0) was used to assemble multiple distinct scRNA-seq datasets into an integrated and 444 unbatched dataset. In brief, we identified 2000 features with high cell-to-cell variation as 445 described above. Secondly, we identified "anchors" between individual datasets with the 446 FindIntegrationAnchors function and inputted these "anchors" into the IntegrateData function to 447 create a "batch-corrected" expression matrix of all cells, which allowed cells from different 448 datasets to be integrated and analyzed together. 449

First, macrophages were extracted from the overview integrated dataset. Next, the major cell types 451 were integrated for further sub-clustering. After integration, genes were scaled to unit variance. 452

Scaling, PCA and clustering were performed as described above. NK cells were also extracted and 453 sub-clustered using the procedure used for macrophages. 454

After non-linear dimensional reduction and projection of all cells into two-dimensional space by 456 UMAP, cells clustered together according to common features. The FindAllMarkers function in 457

Seurat was used to find markers for each of the identified clusters with "wilcox" method and obey 458 "min.pct = 0.2" and "logfc.threshold = 0.25". Clusters were then classified and annotated based on 459 expressions of canonical markers of particular cell types. Clusters expressing two or more 460 canonical cell-type markers were classified as doublet cells and excluded from further analysis. 461

Differential gene expression testing was performed using the FindMarkers function in Seurat, and 463 the Benjamini-Hochberg method was used to estimate the false discovery rate (FDR). DEGs were 464 filtered using a minimum fold change of 1.5 and a maximum FDR value of 0.01. Enrichment 465 analysis for the functions of the DEGs was conducted using clusterProfiler 48 in default parameters 466 and the Benjamini-Hochberg method was used to estimate FDR. Gene sets were derived from the 467 Biological Process of GO Ontology. 468

We used cell scores in order to evaluate the degree to which individual cells expressed a certain 470 pre-defined expressed gene set. The cell scores were initially based on the average expression of 471 the genes from the pre-defined gene set in the respective cell 49 . The AddModuleScore function in 472

Seurat was used to implement the method with default settings. We used APOPTOTIC 473 SIGNALING PATHWAY (GO: 0097190) to define the apoptosis score. 474

We applied the Monocle 24 (version 2) algorithm to determine the potential lineage differentiation 476 between diverse cell populations refer to the tutorial here 477 (http://cole-trapnell-lab.github.io/monocle-release/docs/). First, store data in newCellDataSet 478 object with the parameter "expressionFamily= negbinomial.size()" and "lowerDetectionLimit = 479 0.5". Before construct single cell trajectories, size factors and dispersions were estimated and 480 filtered low-quality cells with default settings. Then the trajectory was inferred with the default 481 parameters of Monocle after dimension reduction and cell ordering based on top 500 genes 482 differing between clusters. Finally, the results of inferred pseudotime trajectory were presented 483 and shown with the first two components. 484

The statistical tools, methods and thresholds for each analysis are explicitly described with the 486 results or detailed in the Figure Legends 

Cytotoxic Lymphocyte Responses

Detrimental Contribution of the Immuno-Inhibitor B7-H1 to 603

Rabies Virus Encephalitis

A whole-brain atlas of inputs to serotonergic neurons of the 605 dorsal and median raphe nuclei

Visualization and Quantification of Post-stroke Neural 607

Connectivity and Neuroinflammation Using Serial Two-Photon Tomography in the Whole Mouse 608

Integrating single-cell transcriptomic data 610 across different conditions, technologies, and species

Comprehensive Integration of Single-Cell Data

clusterProfiler: an R Package for Comparing Biological Themes 614

Single-Cell Transcriptomic Analysis of Primary and Metastatic 616 Tumor Ecosystems in Head and Neck Cancer

The three-dimension (3D) distribution of RABV in the whole mouse brain

Scheme of the experimental workflow for mapping the 3D distribution of RABV in the whole 622 mouse brain using the fMOST technique. (B) Groups of C57BL/6 mice were inoculated with 623 10×LD50 of the recombinant RABV expressing EGFP

the otic subcutaneous (o.s.), or the intranasal (i.n.) route. At the moribund stage, the brains were 625 harvested and prepared for fMOST processing

RABV in a whole mouse brain by vertical and sagittal views are shown. The overall distribution 627 of RABV in the mouse brain can be viewed in supplementary video

Anatomical classification of RABV distribution by different infection routes

The anatomical localization of the selected coronal sections shown in (C-E). Scale bar, 1 mm

Enlarged views, indicated by a yellow, red, and purple box, respectively, of the motor cortex

bed nuclei of the stria terminalis (BNST), and the superior colliculus (SC) from panel C. A single 633

C-E) The distribution of RABV by infection 634 route, i.m., o.s

The raw sequence data reported in this paper have been deposited in the Genome Sequence 491Archive at the BIG Data Center, Beijing Institute of Genomics (BIG), Chinese Academy of 492 

The authors declare no competing interests.