SARS-CoV-2 infects human neural progenitor cells and brain organoids

Dear Editor, Coronavirus disease 2019 (COVID-19) caused by the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has resulted in over 13 million confirmed cases and more than 580,045 deaths across 218 countries and geographical regions as of July 16, 2020. This novel coronavirus primarily causes respiratory illness with clinical manifestations largely resembling those of SARS. However, neurological symptoms including headache, anosmia, ageusia, confusion, seizure, and encephalopathy have also been frequently reported in COVID-19 patients. In a study of 214 hospitalized COVID-19 patients in Wuhan, China, neurologic findings were reported in 36.4% of patients, and were more commonly observed in patients with severe infections (45.5%). Similarly, a study from France reported neurologic findings in 84.5% (49/58) of COVID-19 patients admitted to hospital. Importantly, a recent study in Germany demonstrated that SARS-CoV-2 RNA could be detected in brain biopsies in 36.4% (8/22) of fatal COVID-19 cases, which highlights the potential for viral infections in the human brain. To date, there has been no direct experimental evidence of SARS-CoV-2 infection in the human central nervous system (CNS). We recently demonstrated that SARS-CoV-2 could infect and replicate in cells of neuronal origin. In line with this finding, we showed that SARS-CoV-2 could infect and damage the olfactory sensory neurons of hamsters. In addition, angiotensin-converting enzyme 2 (ACE2), the entry receptor of SARS-CoV-2, is widely detected in the brain and is highly concentrated in a number of locations including substantia nigra, middle temporal gyrus, and posterior cingulate cortex. Together, these findings suggest that the human brain might be an extra-pulmonary target of SARS-CoV-2 infection. To explore the direct involvement of SARS-CoV-2 in the CNS in physiologically relevant models, we assessed SARS-CoV-2 infection in induced pluripotent stem cells (iPSCs)-derived human neural progenitor cells (hNPCs), neurospheres, and brain organoids. We first evaluated the expression of ACE2 and key coronavirus entry-associated proteases in hNPCs. Our data suggested that ACE2, TMPRSS2, cathepsin L, and furin were readily detected in the hNPCs (Supplementary information, Fig. S1). Next, we challenged iPSC-derived hNPCs with SARSCoV-2 at 10 multiplicity of infection (MOI) and with SARS-CoV as a control. Supernatant was harvested at 0, 24, and 48 h post infection (hpi) for virus replication assessment. Interestingly, our data suggested that SARS-CoV-2, but not SARS-CoV, could replicate in hNPCs (Fig. 1a; Supplementary information, Fig. S2). In addition, we quantified the cell viability of SARS-CoV-2infected hNPCs. Importantly, SARS-CoV-2 infection significantly reduced the viability of hNPCs to 4.7% (P < 0.0001) and 2.5% (P < 0.0001) of that of the mock-infected hNPCs at 72 and 120 hpi, respectively (Fig. 1a). In contrast to the substantial cytotoxicity induced by SARS-CoV-2 in the infected hNPCs, SARS-CoV-2 infection did not significantly upregulate interferon (Supplementary information, Fig. S3) and pro-inflammatory (Supplementary information, Fig. S4) response in the infected hNPCs. Next, we challenged 3D neurospheres with SARS-CoV-2 and harvested supernatant samples from the infected neurospheres at 0, 24, 48, and 72 hpi for virus replication assessment. We found the SARS-CoV-2 RNA-dependent RNA polymerase (RdRp) copy number significantly increased in a time-dependent manner (Fig. 1b, left). In addition, a significant amount of infectious virus particles were released from the infected neurospheres as determined by plaque assays (Fig. 1b, right). In parallel, SARS-CoV-2-infected neurospheres were cryosectioned and immunostained for viral antigen assessment. Importantly, SARS-CoV-2 nucleocapsid (N) protein was readily detected across the infected neurospheres but no positive signals were detected in the mock-infected neurospheres (Fig. 1c). Furthermore, electron microscopy detected extensive viral particles in vacuoles within the double-membrane structures, which may represent sites of viral particle formation (Fig. 1d). These findings indicate that neurospheres were permissive to SARS-CoV-2 infection and supported productive virus replication. Next, we examined whether SARS-CoV-2 could infect 3D human brain organoids. We generated iPSC-derived human brain organoids using previously described protocols. The 35-day-old brain organoids showed self-organizing internal morphology with fluid-filled ventricular-like structures resembling that of developing cerebral cortex (Fig. 1e). Cryosectioning and immunostaining were performed to determine the expression and distribution of neuronal markers in 35-day-old brain organoids. Pan-neurons, early forebrain, and hNPCs markers were identified by TUJ1, PAX6, and NESTIN staining, respectively. The TUJ1 staining identified a primitive cortical plate with early neurons (Fig. 1e), whereas PAX6 staining represented the radial glia in the cerebral cortex (Fig. 1e). In addition, NESTIN staining identified active proliferating NPCs in the brain organoids (Fig. 1e). Overall, these results indicated that telencephalon development and cerebral neurogenesis could be modelled by our organoid system. To investigate whether the brain organoids were permissive to SARS-CoV-2 infection, human iPSC-derived 35-day-old brain organoids were challenged with SARS-CoV-2. Importantly, extensive SARS-CoV-2 antigen was detected in the infected samples at 72 hpi (Fig. 1f), indicating that SARS-CoV-2 directly infected the brain organoids. Immunofluorescence staining and confocal microscopy revealed SARS-CoV-2-N signals in the peripheral regions (Fig. 1f, arrows) and in deeper regions of the brain organoids (Fig. 1f, white arrowheads). In addition, cell-cell fusion was readily detected in regions with robust SARS-CoV-2 infection (Fig. 1f, yellow arrowheads). No SARS-CoV-2-N signals were detected in the mock-infected brain organoids (Fig. 1f). We next analyzed supernatant samples from infected brain organoids to evaluate SARS-CoV-2 virus particle release. The results demonstrated SARS-CoV-2 RdRp gene copy number increased in a timedependent manner, suggesting active release of progeny virus particles from infected brain organoids (Fig. 1g, left). Specifically,

Cells were plated on low attachment 96-well plates to form embryoid body (EB).
After 7 days, the EBs were plated to form rosettes expressing neural progenitors using 1 defined medium DMEM/F-12 supplemented with 20 ng/mL FGF2 and Gem21 NeuroPlex (Gemini Bio-Products). For neurosphere generation, 4,000 neural progenitor cells were seeded on low attachment plates under rotation without FGF2.

Formation of brain organoids
Human brain organoids were generated from iPSCs as previously described 2,3 .
After a further week in Media III, the organoids were transferred to Media II and incubated until day 35 with the media refreshed every 3 days.

Viruses and biosafety
The SARS-CoV-2 HKU-001a (GenBank accession number: MT230904) and SARS-CoV GZ50 (GenBank accession number: AY304495) were propagated as previously described 4 . Both viruses were titered in Vero E6 cells by plaque assay. All experiments involving live SARS-CoV-2 and SARS-CoV followed the approved standard operating procedures in our Biosafety Level 3 facility 4 .

Infection of neurospheres or organoids with SARS-CoV-2
To evaluate whether neurospheres or organoids were permissive to SARS-CoV-2, neurospheres or brain organoids were inoculated with 6 × 10 6 PFU/mL SARS-CoV-2 in organoid culture medium and incubated at 37 o C for 24 hours. The inoculum was aspirated at 24 hours post virus challenge. Organoids were washed with culture medium three times and then further incubated until harvest at the indicated time points.

Infection of hNPCs with SARS-CoV-2 or SARS-CoV
To evaluate whether hNPCs were permissive to SARS-CoV-2 infection, differentiated hNPCs were challenged with 10 MOI SARS-CoV-2 or SARS-CoV. Supernatant samples from cells were harvested at 0, 24, and 48 hpi. Samples were lyzed with AVL buffer (Qiagen) and virus replication was determined by qRT-PCR.

Immunostaining and confocal microscopy
Immunofluorescence staining and confocal microscopy were performed as previously described 5 with slight modification. Briefly, organoids were fixed in fresh paraformaldehyde (4% PFA, Sigma-Aldrich) at pH 7.4 and 4℃ overnight. Organoids were then transferred to 30% (wt/vol) sucrose solution in PBS at 4℃ overnight.

Plaque assays
Supernatants from infected organoids and neurospheres were harvested at 0, 24, 48, and 72 hpi and titrated on Vero E6 cells. After incubation at 37°C for 72 hours, cells were fixed with 10% neutral-buffered formalin. For forming unit (PFU) visualization, fixed samples were stained with 0.5% crystal violet in 25% ethanol/distilled water for 10 minutes for plaque.

Cytotoxicity assay
Cell viability was determined by the CellTiter-Glo luminescent cell viability assay (Promega). The kit detects adenosine triphosphate (ATP) levels as a function of cell viability, and was used according to manufacturer's specifications.