key: cord-0313306-imte5ong
authors: Joyce, Jonathan D.; Moore, Greyson A.; Goswami, Poorna; Leslie, Emma H.; Thompson, Christopher K.; Bertke, Andrea S.
title: SARS-CoV-2 Infects Peripheral and Central Neurons of Mice Before Viremia, Facilitated by Neuropilin-1
date: 2022-05-20
journal: bioRxiv
DOI: 10.1101/2022.05.20.492834
sha: dd14c16df5dd35d284fcc66e866eb6b3a5ecf24e
doc_id: 313306
cord_uid: imte5ong

Neurological symptoms are increasingly associated with COVID-19, suggesting that SARS-CoV-2 is neuroinvasive. Although studies have focused on neuroinvasion through infection of olfactory neurons and supporting cells or hematogenous spread, little attention has been paid to the susceptibility of the peripheral nervous system to infection or to alternative routes of neural invasion. We show that neurons in the central and peripheral nervous systems are susceptible to productive infection with SARS-CoV-2. Infection of K18-hACE2 mice, wild-type mice, and primary neuronal cultures demonstrates viral RNA, protein, and infectious virus in peripheral nervous system neurons, spinal cord, specific brain regions, and satellite glial cells. Moreover, we found that SARS-CoV-2 infects neurons at least in part via neuropilin-1. Our data show that SARS-CoV-2 rapidly invades and establishes productive infection in previously unassessed sites in the nervous system via direct invasion of neurons before viremia, which may underlie some cognitive and sensory symptoms associated with COVID-19.

Up to 80% of people infected with SARS-CoV-2, the virus responsible for coronavirus disease 32 2019 (COVID-19), report neurological symptoms. These symptoms span the central nervous 33 system (CNS; dizziness, headache, cognitive and memory deficits) and peripheral nervous 34 system (PNS-somatic and autonomic systems; impaired taste, smell, sensation, orthostatic 35

intolerance, syncope) 1-3 . Fatigue, memory issues, "brain fog," and autonomic dysfunction can 36 persist as part of post-acute sequalae of SARS-CoV-2 infection ("long COVID") 4 . Detection of 37

virus, viral RNA, and antigen, in cerebrospinal fluid (CSF) and brains of COVID-19 patients 38

indicates SARS-CoV-2 is neuroinvasive, which has been documented for common-cold 39 coronaviruses (HCoV-OC43, HCoV-229E) and epidemic coronaviruses (MERS, SARS) 5-13 . 40

As anosmia is a primary symptom of COVID-19, studies were previously conducted to assess the 41 possibility of CNS invasion via olfactory sensory neurons (OSNs), using transgenic mice 42 expressing human angiotensin converting enzyme 2 (hACE2), the receptor for SARS-CoV-2 14-16 . 43

Since RNA, protein, and virus were detected in OSNs, sustentacular cells, and brain 44

homogenates, little attention has since been paid to alternative routes of neural invasion or to the 45 role of PNS infection. Further, different regions of the brain have only recently come under 46 scrutiny. Given that PNS symptoms are more commonly reported than CNS symptoms among 47 non-hospitalized COVID-19 patients, who constitute the bulk of those infected, study of the 48 susceptibility of the PNS to infection is needed 17 . 49 Therefore, we assessed the susceptibility of PNS sensory (dorsal root ganglia-DRG, trigeminal 50 ganglia-TG) and autonomic (superior cervical ganglia-SCG) neurons to infection with SARS-CoV-51 2 following intranasal inoculation of K18-hACE2 transgenic mice (hACE2 mice), wild-type 52 C57BL/6J mice (WT), and guinea pigs (GPs). We also assessed neuroinvasion of the spinal cord 53

and specific brain regions (olfactory bulb, cortex, hippocampus, brainstem, cerebellum), 54 characterized viral growth kinetics in primary neuronal cultures, and investigated the contribution 55 of neuropilin-1 (NRP-1) to neuronal entry. 56

We show that PNS sensory neurons are susceptible and permissive to productive infection with 57 SARS-CoV-2 but that autonomic neurons, while susceptible to infection, sustain significant 58 cytopathology and do not release infectious virus. Recovery of infectious virus from the TG, which 59

innervates oronasal mucosa and extends central projections into the brainstem, suggests an 60 alternative route of neuroinvasion independent of OSNs. We also show that infectious virus can 61 be recovered from spinal cord and lumbosacral DRGs, which may underlie some sensory 62 disturbances experienced in COVID-19. We further show that SARS-CoV-2 is capable of 63

replicating in specific brain regions, with highest RNA concentrations and viral titers found in the 64 hippocampus, which may contribute to memory disturbances associated with COVID-19.

respectively, could serve as neural routes for neuroinvasion. The trigeminal nerve provides 86 sensory innervation to the nasal septum, as well as oronasal mucosa, and projects into the 87 brainstem. The SCG provides sympathetic innervation to salivary glands and vasculature of the 88 head and brain, with preganglionic neurons residing in the spinal cord. To assess susceptibility of 89 these peripheral neurons to infection with SARS-CoV-2, TGs and SCGs were assessed for viral 90 RNA, protein, and infectious virus. We detected viral RNA at 3-and 6-dpi in TGs and SCGs in 91 both inoculum groups in hACE2 and WT mice. Viral RNA concentrations were lower in WT than 92 hACE2 mice ( Fig. 1a-b ). RNA concentrations increased over time in TGs and SCGs of hACE2 93

and WT mice, suggesting genome replication. Immunostaining detected SARS-N in the majority 94

of TG neurons of hACE2 mice, as well as in ≈30% of TG neurons of WT mice (Fig. 1c , Extended 95

Data Fig. 4 ). All SCG neurons from hACE2 mice were SARS-N-positive, showing substantial 96 pathology with vacuolated neurons and loss of ganglionic architecture. Although SCGs from WT 97 mice remained intact, SARS-N was evident in all neurons and some vacuolization was observed. 98

Infectious virus was detected in TGs collected 3 dpi (0.5 log PFU/mg homogenate) and 6 dpi (2 99 log PFU/mg homogenate) from hACE2 mice. Infectious virus was not detected from SCGs. 100

Considering the pathology of the ganglia, the virus may have produced such significant 101 cytotoxicity that production of viral progeny was impossible. The use of multiple complementary 102 assays indicate that viral RNA and protein can be isolated from both TGs and SCGs of both 103

hACE2 and WT mice, and that infectious virus can be recovered from TGs. These data indicate 104

that TGs and SCGs are susceptible to infection with SARS-CoV-2, that the TG may serve as a 105

route of CNS invasion, and that neuroinvasion can occur independent of hACE2. All GPs were 106

negative for viral RNA. was detected at increasing concentration in SCGs of hACE2 and WT mice in both inoculum groups from 3 110 to 6 DPI. The SCG provides sympathetic innervation to the salivary glands and blood vessels of the head, 111 neck, and brain. Three-way ANOVA detected a significant difference (F(7, 40) = 3.902, P = 0.0025) in RNA 112 genome copy number. Tukey's honestly significant difference (HSD) post hoc tests detected significant 113 differences between the hACE2 and WT groups inoculated with 10 3 PFU assessed at 3 dpi (p= 0.0464) as 114 well as the WT groups inoculated with 10 3 PFU assessed at 3-and 6-dpi (p=0.029). b, SARS-CoV-2 RNA 115 was detected at increasing concentration in TGs of hACE2 and WT mice in both inoculum groups from 3 116 to 6 DPI. The TG provides sensory innervation to the face, including the nasal septum, and sends 117 projections to the brain stem, thereby providing an alternative entry point for SARS-CoV-2. No statistically 118 significant differences were detected between the groups (F(7, 40) = 1.405, P = 0.2305). c,

Immunofluorescence for SARS-CoV-2 nucleocapsid (SARS-N, grey) and NeuN (red) labeled neurons in 120 TG and SCG sections. SARS-N is more prevalent in hACE2 than in WT but observable in both. No 

studies have assessed viral RNA and infectious virus in the brain, brain homogenates are typically 178 tested, which doesn't allow for analysis of spatial differences in the presence of virus in discreet 179 brain regions and could obscure detection of low levels of RNA or infectious virus in specific 180 regions 14, 16, 20-22 . To determine if SARS-CoV-2 is present in specific brain regions, we assessed 181 the olfactory bulb, hippocampus, cortex, brainstem, and cerebellum for SARS-CoV-2 RNA, 182

protein, and infectious virus. We detected viral RNA in all brain regions at 3 dpi, which increased 183 by 6 dpi (Fig. 3a-e) , in both hACE2 and WT mice. Higher concentrations of viral RNA were found 184 in hACE2 mice compared to WT mice in all regions. The highest RNA concentration was detected 185

in the hippocampus at 6 dpi ( Fig. 3b) , followed by brainstem, cortex, olfactory bulb, and 186 cerebellum in hACE2 mice. In WT mice, all regions contained similar quantities of SARS-CoV-2 187

RNA, suggesting that ACE2-independent spread or replication through the nervous system may 188 differ compared to hACE2 mice. Infectious virus was recovered from the hippocampus and 189

brainstem (3 log PFU/mg homogenate) and cortex as early as 3 dpi in hACE2 mice and from all 190 regions by 6 dpi, with highest concentrations in the hippocampus and cortex (5 log PFU/mg 191 homogenate) followed by olfactory bulb and brainstem (4 log PFU/mg homogenate) and 192 cerebellum (2 log PFU/mg homogenate) (Fig. 3f ). Unexpectedly, low levels of infectious virus 193

were recovered from the hippocampi and brainstem of WT mice. These results indicate that viral 194 invasion and replication is not consistent across the brain and the highest concentrations of virus 195

are found in areas functionally connected to the olfactory and limbic systems, two systems 196 strongly impacted during COVID-19. All GPs were negative for viral RNA. in the olfactory bulb. Tukey's HSD detected differences in the olfactory bulb between hACE2 and WT 204 groups inoculated with 10 5 PFU assessed at 3 dpi (p<0.0001) as well as between those groups assessed 205 at 6 dpi (p=0.001). A significant difference (F(7, 41) = 5.106, P = 0.0003) was also detected in the 206 hippocampi of hACE2 mice inoculated with 10 5 PFU assessed at 3-vs 6-dpi (p=0.0017). A significant 207 difference (F(7, 41) = 6.917, P = <0.0001) was also detected in the brainstem of the hACE2 and WT groups 208 inoculated with 10 5 PFU assessed at 3 dpi (p= 0.0103). While differences were detected in the cortex (F(7, 209 41) = 6.098, P = <0.0001) and the cerebellum (F(7, 41) = 6.652, P = <0.0001) none were between relevant 

To further assess localization of SARS-CoV-2 in brain regions, we immunostained sagittal 219

sections of hACE2 brains for SARS-N and neuronal marker NeuN (Fig. 4 , Extended Data Fig. 5 ). 220

At 6 dpi, minimal SARS-N was found in the striatum, geniculate nucleus, superior colliculus, 221

superior olive or cerebellum. In other regions, including frontal cortex, lateral preoptic area and 222 visual cortex, thalamus and nucleus accumbens, the majority of neurons were positive. Some of 223 these regions also showed diffuse SARS-N in tissue (cortex, reticular formation), suggesting 224 differences in pathology within various brain regions. As the reticular formation facilitates motor 225 activity associated with the vagus nerve, pathology in this region would correlate with vagus nerve 226 dysfunction, which is associated with even mild COVID-19 cases 23 . Despite finding >10 4 copies 227 of viral RNA in the olfactory bulb of all six hACE2 mice, infectious virus was recovered from only 228 one of three mice and SARS-N was absent in the caudal olfactory bulb in assessed mice. 229

However, axons of neurons with cytoplasmic N protein in the olfactory tubercle stained positive 230

for SARS-N, suggesting axonal spread from the olfactory bulb into the olfactory tubercle, where 231 productive infection ensues. vivo. Neurons were harvested from 8-10-week-old hACE2 and WT mice to establish primary 256 neuronal cultures, which were then infected. Media and cells were analyzed separately for viral 257 RNA (RT-qPCR) and infectious virus (plaque assay) to differentiate between intracellular 258

replication and release of infectious virus. Viral RNA levels increased in SCG neurons between 259 2-and 3-dpi, although no increase in viral RNA was detected in media (Fig 6a) . Infectious virus 260

wasn't detected in the cellular or media fractions (Fig 6b) . These data show that while viral 261 genome replication occurs in SCGs, infectious virus is not released, suggesting abortive infection 262

in sympathetic autonomic neurons. Considering the pathology of the SCGs in vivo, SARS-CoV-2 263 appears to be cytotoxic to SCG neurons prior to production of viral progeny. Viral RNA levels 264

increased in TG neurons between 1-and 2-dpi, subsequently falling at 3 dpi (Fig 6a) . Infectious 265 virus was recovered from neurons and media from 2-to 4-dpi, declining at 5 dpi in both (Fig 6b) , 266

indicating that in TG neurons, genome replication peaks at ~48 hpi, after which infectious virus is 267 released. While DRGs follow a similar pattern, viral RNA levels exhibited a cyclical pattern with 268 peaks occurring ~48 h intervals at 2-and 4-dpi (Fig 6a) . Infectious virus was recovered from 269 neurons and media in a similar fashion (Fig 6b) . fixed at 1, 2, and 3-dpi for immunostaining, which showed few DRG neurons (<5%,) in any culture 277 became productively infected (Fig 6d) , which is reflected in the modest increase in viral RNA and 278

infectious virus (Fig 6a,b, Extended Data Fig. 6 ). In addition to genome replication, SCG neurons 279

were permissive for N protein expression, shown by a positive immunofluorescence signal 1 dpi, 280

while activated satellite glial cells accompanied dying neurons by 3 dpi (Fig 6c) . In TG and DRG 281

neurons, several phenotypes were observed, including perinuclear SARS-N staining at 1 dpi and 282

punctate staining in the cytoplasm of enlarged neurons at 3 dpi, likely representing replication 283 compartments (Fig 6c, Supplementary Video 3-4) . Infected sensory neurons showed a variety of 284

phenotypes, including loss of membrane integrity, cytoplasmic puncta, and seemingly healthy 285

neurons strongly expressing SARS-N ( Fig. 7 inset 1,2) . Infected satellite glial cells, which 286 appeared to be activated, were also present in some primary neuronal cultures (Fig. 7 inset 3) . 287

Infection of WT neurons (Fig. 6a,b) showed viral RNA concentrations increased from 1-2 dpi in 288

SCGs and increases in media at 2-and 4-dpi, similar to hACE2 neurons. Low levels of infectious 289

virus were recovered in neurons and media at 1 dpi. In TGs, viral RNA concentrations cyclically 290 peaked at 2-and 5-dpi with minimal increase in media, similar to hACE2 neurons. Low levels of 291 infectious virus were recovered in neurons and media 1-2 dpi. Viral RNA concentrations in DRGs 292

steadily increased from 1-4 dpi reaching similar levels to viral RNA concentrations in hACE2 293

DRGs at 4 dpi, with a concomitant increase in media. Low levels of infectious virus were recovered 294 in media from 1-2 dpi but a second release, as from hACE2 neurons, was not apparent. These 295 data show invasion and replication in neurons lacking hACE2, although reduced compared to 296

hACE2-expressing neurons. 

(2), and seemingly healthy neurons with extensive neurites with strong SARS-N+ staining (arrow in 3).

Infected satellite glial cells were also observed (arrowheads in 3); many appeared to be activated, noted by 326 the presence of extended cellular processes. These findings are similar to immunostaining of DRGs in vivo, 327 which also contained numerous infected satellite glial cells.

Neuroinvasion of the PNS and CNS occurs before viremia and involves neuropilin-1. To 329 determine if neuroinvasion is driven by hematogenous entry or direct neuronal entry, PNS and 330 CNS tissues were assessed 18 hpi and 42 hpi after intranasal inoculation of hACE2 and WT mice 331 (Fig. 8a,b) . Although no viral RNA was detected in blood at 18 hpi, viral RNA was detected in all 332

PNS and CNS tissues and salivary glands (innervated by the SCG), with the exception of spinal 333 cord and cerebellum of WT mice. By 42 hpi, viral RNA was detected in blood in a single hACE2 334 mouse and had increased in all hACE2 PNS and CNS tissues except SCG. In WT mice, viral 335

RNA was no longer detected in salivary glands, SCG, or DRG but had increased in brainstem 336

and hippocampus. Immunostaining did not detect SARS-N in any tissues, indicating that the virus 337 was transiting through PNS tissues when collected but had not yet begun replication, which 338

occurs later during infection. These data demonstrate that neuroinvasion occurs rapidly after 339 infection, is mediated by invasion of and transport along neurons, and can occur independent of 340 hACE2. 341

Since our WT mice were infected despite absence of hACE2, we investigated the contribution of 342

neuropilin-1 (NRP-1) to neuronal entry in sensory neurons. NRP-1 has been shown to interact 343

with SARS-CoV-2 spike, thereby enhancing viral binding and entry in non-neuronal cells 24-26 . 344

Presence of NRP-1 was confirmed on DRG neurons and satellite glial cells via immunostaining 345 (Fig. 8c) . Primary cultured DRG neurons from hACE2 and WT mice were pretreated with 346 EG00229, a selective NRP-1 antagonist, infected with SARS-CoV-2, and viral RNA 347

concentrations were assessed 2 dpi. Viral RNA concentrations were significantly reduced by 348 99.8% in hACE2 neurons (p=0.0081) and 86.7% in WT neurons (p=0. 0141) (Fig. 8d) , indicating 349 that NRP-1 is a SARS-CoV-2 co-receptor in neurons irrespective of hACE2 expression. 

Discussion 374

Neurotropic viruses can enter the nervous system hematogenously or through neural pathways. 375

From blood, viruses can infect endothelial cells, gaining access to underlying tissues through lytic 376 destruction of vasculature, or can be transported across the vasculature inside extravasating 377

leukocytes. Viruses can enter neural pathways through peripheral sensory, autonomic, and/or 378 motor axon terminals and transport retrograde toward the CNS, often moving trans-synaptically 379 along functionally connected pathways. SARS-CoV-2 likely uses both mechanisms. SARS-CoV-380

2 CNS invasion has been proposed via infection of the nasal neuroepithelium with invasion of 381

OSNs, the olfactory bulb, and its cortical projections 14-16, 18, 20-22 . Organoid, stem cell, microfluidic, 382

and mouse models, correlating with human autopsy findings, demonstrate disruption of 383 endothelial barriers and choroid plexus integrity, as well as transcytosis of SARS-CoV-2, 384 supporting hematogenous CNS entry 5, 18, 27-31 . While OSNs are a key constituent of the nasal 385 neuroepithelium, the oronasopharynx is innervated by other sensory and autonomic pathways 386 through which SARS-CoV-2 may enter the nervous system. Utilizing hACE2 mice, WT mice, and 387

primary neuronal cultures, we show susceptibility of peripheral neurons to SARS-CoV-2 infection, 388

demonstrating differential replication kinetics and cytopathic outcomes following infection of 389 sensory, autonomic, and central neurons. We also show evidence supporting axonal transport of 390 SARS-CoV-2 and CNS entry, preceding viremia, through neural pathways that functionally 391 connect to brain regions responsible for memory and cognition, which are affected in COVID-19. 392

Furthermore, we show that SARS-CoV-2 can use NRP-1 for neuronal entry in the absence of 393 hACE2 expression. As COVID-19 neurological symptoms are often related to peripheral neuron 394 dysfunction rather than exclusive to CNS symptoms, focusing solely on CNS neuroinvasion takes 395 a myopic view of the potential impacts of SARS-CoV-2 in the PNS.

Our detection of SARS-CoV-2 in the TG and SCG makes anatomical sense given their innervation 398 of the oronasal mucosa and glands. and provide a rationale for a deeper investigation of the DRG as a site of productive viral infection 451

in COVID-19, as well as the possibility and direction of axonal transport, since infection of the 452

DRGs may contribute to some sensory disturbances suffered by COVID-19 patients. 453

Our detection of viral RNA and infectious virus in specific brain regions is more granular than what 454

has been previously reported and shows the hippocampus, cortex, and brainstem are all rapidly 455

invaded by SARS-CoV-2 through neural pathways before viremia. TG neurons, with axonal 456

projections to both oronasal epithelium and brainstem, or SCG neurons, with synaptic connectivity 457

to salivary glands and brainstem, could deliver virus directly to the CNS. Viral RNA and protein 458

have been detected in the olfactory bulb, brainstem, cerebellum, and cortex of COVID-19 459 patients 6, 39, 40 . Our results indicate that viral penetration and replication within the brain is 460 dependent on the region assessed, which suggests the presence of factors (cell types, synaptic 461

connections, vascularization level) that favor invasion and replication in some regions over others.

Our detection of infectious virus in the hippocampi and brainstems of both WT and hACE2 mice 463

further underscore the importance of these structures in COVID-19 pathology. Interestingly, there 464

were substantially fewer infected cells in some aspects of the nervous system, which might be 465 due to the particular time point assessed or to the types of neurons present in these areas. 466

Notably, SARS-CoV-2 was detected in CNS and PNS of both hACE2 and WT mice, indicating 467 that hACE2 expression is not a requirement for neuronal infection. We demonstrate that NRP-1, 468

which was shown previously to mediate entry into non-neuronal cells, also mediates entry into 469

neurons. Our inhibition of NRP-1 in cultured DRG neurons from hACE2 mice reduced infection to 470 a greater extent than in neurons from WT mice, indicating that NRP-1 can serve as a co-receptor 471

to enhance infection in the presence of hACE2 expression or an alternative receptor independent 472 of hACE2. 473

Since GPs were used as an animal model for SARS-CoV-1, we also assessed their potential as 474 a model for SARS-CoV-2 41, 42 . Contrary to previous studies using intraperitoneal inoculation, we 475

show that GPs are resistant to intranasal infection, having no detectable viral RNA in any tissues 476 tested and no SARS-N in lungs or brains. A previous study showed that SARS-CoV-1 did not 477 efficiently infect transfected cells expressing gpACE2 43 and recent in silico modeling of SARS-478

CoV-2 spike protein and gpACE2 binding showed altered kinetics, which was believed to limit 479 infectability 44, 45 . Our results provide direct in vivo confirmation, resolving the question of the 480 susceptibility of GPs to infection with SARS-CoV-2 following intranasal inoculation. 481

The extent of both acute and long-term neurological impacts of SARS-CoV-2 are only beginning 482

to be realized. The existence of sensory and autonomic disorders, lasting months beyond initial 483 infection, necessitates a better understanding of the impact of SARS-CoV-2 on the entirety of the susceptible to, and in most cases permissive to, productive infection with SARS-CoV-2 via direct 491 neural invasion rather than hematogenous spread. Presence of infectious virus in these tissues 492

shows that routes of neuroinvasion exist beyond OSNs and that invasion can occur independent 493 of ACE2 using NRP-1 as a co-receptor. Our findings support the need to investigate these sites 494 of neuroinvasion to a greater depth than currently exists. 495

Online content statement 497

Any methods, additional references, Nature Research reporting summaries, source data, 498 extended data, supplementary information, acknowledgements, peer review information; details 499 of author contributions and competing interests; and statements of data and code availability are 500 available at ___. in January 2020 who was diagnosed with COVID-19 after returning from visiting family in Wuhan, 611

China. Viral stocks were titrated in duplicate using a standard plaque assay on Vero E6 cells with 612 agarose overlay 46 . Vero E6 cells were maintained following standard cell culture protocols. 613

Mouse infections. Eight to ten-week-old male and female B6.Cg-Tg(K18-ACE2)2Prlmn/J mice 614 (Stock # 034860; Jackson Laboratory; n=12, 2 groups of 6 mice), and their wild-type C57BL/6J 615 counterparts (n=12, 2 groups of 6 mice) were inoculated intranasally with SARS-CoV-2 isolate 616 USA-WA1/2020 (Extended Data Fig 1a) . Inoculations were carried out under ketamine/xylazine 617 anesthesia in the on campus ABSL-3 facility after a one-day acclimation period. Mice received 20 618 µL of either 10 3 PFU (2 groups per mouse type) or 10 5 PFU (2 groups per mouse type) of SARS-619

CoV-2 in 1X PBS. The inoculum was split between the nares for each mouse. Uninfected K18-620 hACE2 control mice (n=2) and C57BL/6J wild-type control mice (n=2) were housed in a separate 621 on campus ABSL-1 facility. Aliquots of the inocula and viral stock were saved for back titration 622 using plaque assay for infectious viral titer and RT-qPCR for RNA copy number. All mice were 623 genotyped following Jackson Laboratory protocol # 38170 V2. Mice were assessed daily for signs 624 of disease and changes in weight and temperature. Mice from each group (K18-hACE2, WT) and 625 inoculum dose (10 5 PFU, 10 3 PFU) were euthanized 3 days post infection (dpi) (n=3) and at 6 dpi 626 (n=3). Tissues collected included blood, CNS tissues (olfactory bulb, hippocampus, brainstem, 627 cerebellum, cortex, spinal cord), PNS tissues (autonomic ganglia: superior cervical ganglia-SCG; 628 sensory ganglia: dorsal root ganglia-DRG, trigeminal ganglia-TG), viscera (lung, spleen, liver, 629 kidney, pancreas). Half of the tissues were collected in TRI Reagent for RNA extraction and RT-630 qPCR and the other half collected in 10% formalin for immunostaining. Brains were split into 631 hemispheres maintaining attachment with the olfactory bulb. One hemisphere was fixed in 632

formalin for immunostaining and the other dissected out into individual brain regions with each 633 placed in TRI Reagent for RT-qPCR. This experiment was repeated as above for reproducibility. 634

Tissue collection from the second experiment was split between TRI Reagent for RT-qPCR as 635

above or flash frozen on dry ice for plaque assay. Blood, spleen, liver, kidney, and pancreas 636 samples from the initial infection study were assessed via RT-qPCR to assess for disseminated 637 infection at 3-and 6-dpi. As these tissues were not the main focus of the investigations, they 638

were not assessed in the replicate study at 3-and 6-dpi. Lungs from the initial and replicate 639

infection studies were assessed to verify infection at 3-and 6-dpi. 640

To assess viral spread through nervous tissues at earlier timepoints during infection, and to 641 determine the role, if any that viremia plays verses direct neuronal invasion, K18-hACE2 (n=10) 642

and WT mice (n=10) were infected with 10 5 PFU SARS-CoV-2 as described above and were 643 euthanized 1-and 2-dpi (n=5 of each group/day). Blood, PNS tissues, CNS tissues (with addition 644 of pituitary gland), and salivary glands were collected as described above for RT-qPCR, plaque 645 assays, and immunostaining. 646 Guinea pig infections. After a two-day acclimation period, three-week-old female Hartley guinea 647 pigs (Hilltop Lab Animals) were infected with either 10 3 PFU (n=12) or 10 5 PFU (n=12) of SARS-648

CoV-2 as described for K18-hACE2 mice in the ABSL-3 facility (Extended Data Fig 1a. ). 649

Uninfected controls (n=2) were housed in the ABSL-2 facility. Guinea pigs were monitored daily 650

as described for mice. Tissue types collected, methods of collection, and downstream assays 651 were the same as described for the mice. 652 RNA extraction and SARS-CoV-2 specific RT-qPCR. RNA was extracted and RT-qPCR 653 performed as previously described 47 . Briefly, tissues were homogenized in 200 µL TRI Reagent 654 (Fisher Scientific) using a handheld tissue homogenizer with sterile pestles (Cole-Parmer). RNA 655 was extracted using a standard guanidinium thiocyanate-phenol-chloroform extraction. RNA 656 purity and quantity was assessed using a NanoDrop 2000 spectrophotometer (ThermoFisher). Immunofluorescence. Tissues were prepared for immunostaining as previously described 48 . 665

Briefly, viscera were fixed in 10% formalin and ganglia were fixed in 4% paraformaldehyde 666 overnight, moved to 30% sucrose overnight, and subsequently embedded in optimal cutting 667 temperature media (ThermoFisher). A Leica CM3050-S cryostat (Leica Biosystems) was used to 668 prepare 7 µm sections from each tissue block. Slides were rinsed in 1X PBS then blocked in 3% 669 normal donkey serum, 0.1% Triton-100X, and 1X PBS for 30 min at room temperature. SARS-670

CoV-2 N protein was visualized using an Alexa Fluor® 488 conjugated rabbit monoclonal anti-671 SARS-CoV-2 nucleocapsid antibody at a 1:1000 concentration (NBP2-90988AF488; Novus 672 Biologicals). hACE2 was visualized using an Alexa Fluor® 594 conjugated mouse monoclonal 673 anti-ACE2 antibody at a 1:1000 concentration (sc-390851 AF594; Santa Cruz Biotechnology). 674

NeuN was visualized using an Alexa Fluor® 647 conjugated rabbit monoclonal anti-NeuN 675 antibody at a 1:1000 concentration (ab190565; Abcam). α-d-galactose carbohydrate residues on 676 sensory neurons was visualized using the Bandeiraea simplicifolia isolectin B4 (IB4) conjugated 677

to rhodamine at a 1:250 concentration (RL-1102; Vector Laboratories). Tyrosine hydroxylase was 678 visualized using an Alexa Fluor® 594 conjugated mouse monoclonal anti-TH antibody at a 1:500 679 concentration (818004; Biolegend). Glutamine synthetase was visualized using a mouse 680 monoclonal anti-GS antibody at a 1:100 concentration (MA5-27750; Invitrogen) followed by an 681

Alexa Fluor® 594 conjugated goat anti-mouse monoclonal antibody at a 1:1000 concentration 682 (A11005; Invitrogen). Neuropilin-1 was visualized using a goat polyclonal anti-NRP1 antibody at 683 a 15 µg/mL concentration (AF566; R&D Systems) followed by an Alexa Fluor® 647 conjugated 684 donkey anti-goat monoclonal antibody at a 1:1000 concentration (ab150135; Abcam). Nuclei were 685 visualized with 4′,6-diamidino-2-phenylindole (DAPI) in SlowFade Diamond antifade mounting 686 medium (ThermoFisher). Primary antibodies were incubated with tissues overnight at 4°C in 1% 687 normal donkey serum, 0.1% Triton-100X, and 1X PBS. Secondary antibodies were incubated with 688 tissues for 1 hour at room temperature. 689

Confocal microscopy and image analysis. Imaging was performed using a Leica SP8 scanning 

Neuronal infection studies, 773 both immunostaining as well as plaque assay and RT-qPCR studies, were repeated in three 774 separate experiments, with duplicate samples for each ganglion and timepoint in K18-hACE2 and 775 in duplicate in WT mice. Neuropilin-1 inhibition studies were repeated twice, with duplicate 776 samples for each timepoint in K18-hACE2 and WT mice. Mouse infection studies were repeated 777 as described. Guinea pig infection studies were conducted as described. All statistical analyses 778 were performed in JMP Pro 16 (SAS Institute) and confirmed in GraphPad Prism version 8 during 779 figure creation. For statistical analysis

RT-qPCR data was log transformed before analysis to correct for normality of distribution. RT-781 qPCR data was analyzed using a multifactorial ANOVA. If significance was found, pairwise 782 analysis was performed using Tukey's honestly significant difference (HSD) post hoc test

Further information on research design is available in the

Severe Acute Respiratory Syndrome Coronavirus 2 from Patient with 791 Coronavirus Disease, United States

SARS-CoV-2 Remains 793 Infectious on Refrigerated Deli Food, Meats, and Fresh Produce for up to 21 Days

Assessment of Two Novel Live-Attenuated Vaccine Candidates for 796

Guinea Pigs. Vaccines (Basel)

A5-positive primary sensory neurons are nonpermissive for productive 798 infection with herpes simplex virus 1 in vitro

Response to Herpes Simplex Virus 1 Infection in Primary Adult Murine Hippocampal 801 Neurons

LAT region factors mediating differential neuronal tropism of HSV-1 803 and HSV-2 do not act in trans

Acknowledgments 805 This research was funded by internal COVID-19 rapid response seed funding from the Fralin Life 806 Sciences Institute at Virginia Tech. The following reagent was deposited by the Centers for 807 Disease Control and Prevention and obtained through BEI Resources

Will Stone, and Addie Hayes for various forms of assistance

Author contributions 811

writing-original draft preparation, JDJ; writing-review 814 and editing

WT mice, and 822 guinea pigs. a, Graphical abstract outlining the experimental approach used in both mouse and guinea pig 823 infections highlighting intranasal infection of groups with either 3 log PFU or 5 log PFU SARS-CoV-2, clinical 824 evaluation (temperature, weight, survival), collection of tissues from half of each group at 3-and 6-dpi, and 825 downstream analysis for SARS-CoV-2 RNA copies (RT-qPCR), virus and host antigen (immunostaining), 826 and infectious virus (plaque assay). b Clinical data by inoculum group for mice and guinea pigs including 827 weight, temperature, and survival. Weight (grams) for each animal was recorded daily. Weight is reported 828 as the mean percentage increase/decrease for each inoculum group relative to the mean starting weight 829 for that group

Both inoculum groups of guinea pigs gained weight but at a 831 slower rate than uninfected controls. Temperature (°C) for each animal was recorded daily. Temperature is 832 reported as the mean percentage increase/decrease for each inoculum group relative to the mean starting 833 temperature for that group. While transient increases in temperature in the 5 log PFU inoculum groups in 834 both mice and guinea pigs were noted, they were not significant. Kaplan-Meier survival plots were created 835 for each inoculum group

Mortality was noted at 6 dpi (14%, n= 2 of 14)

While differences were detected in the lungs (F(7, 41) = 2.745, p = 841 0.0197) none were between relevant groups. b, Low concentrations of SARS-CoV-2 RNA were detected 842 in the blood of hACE2 in both inoculum groups at 3 dpi and both hACE2 and WT mice in both inoculum 843 groups at 6 dpi. A significant difference (F(7, 40) = 3.417, P = <0.006) was detected in the WT group 844 inoculated with 10 3 PFU assessed at 3-vs 6-dpi (p= 0.0321). Low concentrations of SARS-CoV-2 RNA 845 were also detected in the spleen c, liver d, kidney e, and pancreas f, of both inoculum groups in hACE2 846 and WT mice, mostly appearing at 6 dpi. Data are the mean ± s.e.m. Log transformed RNA genome copy 847 numbers were statistically compared by three-way ANOVA

Extended Data Fig 3. | Immunofluorescence for SARS-N protein and NeuN in sections of brain and 853 lung from infected guinea pigs. All images were acquired using a Leica SP8 confocal microscope, using 854 identical image acquisition and ImageJ settings used

All images were acquired using a Leica SP8 confocal microscope, using identical image 859 acquisition settings (laser power and gain) across all sections shown. All images were colorized, z-860 projected, and prepared using identical contrast and brightness parameters in ImageJ. SARS-N is present 861 in neurons in the DRG, TG and SCG in infected mice. Minimal background immunofluorescence is 862 observed in sections from uninfected mice

confocal microscope. Sections of ganglia, brain, and spinal cord were imaged with identical laser 691 power and gain settings within tissue type to account for background immunofluorescence. Cells 692 from in vitro studies were imaged with varying laser power and/or gain due to the wide range of 693 immunofluorescence observed within given experiments. Images were imported into ImageJ and 694contrast and brightness was adjusted identically across all images within tissue types. 3D models 695were made using ImageJ and SyGlass VR imaging software. (SCG) were collected from mature mice and cultured as previously described 49-51 . Briefly, ganglia 709were harvested from 8-10-week-old K18-hACE2 and WT mice and enzymatically digested with 710 a sequential incubation of papain (Worthington) and collagenase/dispase (Sigma-Aldrich) 711 followed by washes in Neurobasal A (Invitrogen) after each digestion. Ganglia were triturated into 712 single cell suspensions via pipette and further washed. SCGs and DRGs were brought to volume 713in "complete media" containing Neurobasal A with 2% B27 (Invitrogen), 1% penicillin-streptomycin 714 (Thermo Fisher), L-glutamine (Thermo Fisher), fluorodeoxyuridine (Sigma-Aldrich), and 715neurotrophic factors (PeproTech). Neurons were plated at a concentration of 3,000 neurons/well 716 in Matrigel coated Lab-Tek II 8-well chamber slides or 24-well plates (Thermo Fisher). TGs were 717 collected after separation from debris via density gradient centrifugation using OptiPrep (Sigma-718Aldrich) with subsequent washes in Neurobasal A. TGs were plated as described for SCGs and 719DRGs. 720Neuronal infection. Neurons from K18-hACE2 and WT mice were inoculated with SARS-CoV-2 721isolate USA-WA1/2020 at 30 MOI in 100 µL Neurobasal A (Invitrogen) for 8-well chamber slides 722and 200 µL for 24-well plates for 1 h. Following the 1 h adsorption the inoculum was removed, 723fresh complete media was added (without fluorodeoxyuridine), and neurons were incubated at 72437°C with 5% CO2. Aliquots of the inocula and viral stock were saved for back titration using 725 plaque and RT-qPCR. 726

To 727 quantify the number of autonomic (SCG) and sensory (DRG) neurons infected per ganglia, 8-well 728 chamber slides were fixed with paraformaldehyde at 1-, 2-, and 3-dpi and stained as described 729above for the detection of SARS-CoV-2 nucleocapsid. The number of infected neurons from each 730 ganglion were counted for each day, averaged, and reported as the percentage of infected 731neurons per 500 neurons counted. DRG neurons were chosen as the representative sensory 732 neuron as they had the more dynamic replication kinetics with successive rounds of replication. RNA extraction and viral genome copy number quantitation via RT-qPCR as described above. 741Samples were stored at 4°C until processing. For quantification of viral titer in neurons vs that 742 released into the media, neurons and media were collected separately in duplicate (TGs, DRGs) 743or singularly (SCGs). To correct for evaporation of media throughout the time course the final 744 volume of collected media was brought up 500 µL by adding DMEM prior to plaque assay. 745Neurons were collected in 500 µL DMEM after scraping with a pipette tip. Samples were 746immediately stored at -80°C until processing for plaque assay as described above. Following 747 collection of the media but prior to collection of the neurons in TRI reagent or DMEM, the neurons 748were gently washed with 500 µL DMEM which was then discarded, to remove any residual media 749 containing RNA or virus. A similar rinse was performed immediately after the 1 h inoculation to 750 remove any residual inoculum. 751

culture. Primary neuronal cultures of DRGs from K18-hACE2 and WT mice were established as 753 described above. Neurons were pretreated with 100 µM of the NRP-1 antagonist EG00229 (6986; 754Tocris) dissolved in DMSO prior to infection as described for Caco-2 cells 24 . EG00229 putatively 755blocks binding between the carboxyl-terminal sequence of SARS-CoV-2 S1 which has a C-end 756 rule (CendR) motif and the extracellular b1b2 CendR binding pocket of NRP-1, which has been 757 suggested as an alternative co-receptor for SARS-CoV-2 in non-neuronal cells 24-26 . Neurons were 758 infected as described above. To determine if NRP-1 blockade impacted SARS-CoV-2 entry and 759 therefore subsequent replication in neurons, neurons and media were collected together in LS-760 TRI Reagent (Fisher Scientific). RNA was isolated and virus replication assessed via RT-qPCR 761 as described above. Samples were collected at initial peak replication times as determined 762 through our previous neuronal growth kinetics studies (DRG; 3 dpi) to assess if theses peaks 763were blunted or completely inhibited. Infected neurons from K18-hACE2 and WT mice not treated 764with EG00229 but with an equivalent amount of DMSO, the solvent for EG00229, served as 765 controls. 766Statistics and reproducibility. Sample sizes were not statistically calculated as they are similar 767to sample sizes used in other SARS-CoV-2 studies using K18-hACE2 mice 14, 15, 18, 20 . Animals 768were randomly assigned to either inoculum group or control group ensuring the groups were age-769matched. Measurements were taken from distinct samples. RT-qPCR and plaque assays were 770 performed in duplicate for each sample when assessing both in-vivo and ex-vivo infections. RT-771

The authors declare no competing interests.