key: cord-0933190-rtus1qyw
authors: Crunfli, F.; Corasolla Carregari, V.; Veras, F. P.; Vendramini, P. H.; Valenca, A. G. F.; Antunes, A. S. L. M.; Brandao-Teles, C.; Zuccoli, G. d. S.; Reis-de-Oliveira, G.; Silva-Costa, L. C.; Saia-Cereda, V. M.; Codo, A. C.; Parise, P. L.; Toledo-Teixeira, D. A.; de Souza, G. F.; Muraro, S. P.; Melo, B. M. S.; Almeida, G. M.; Firmino, E. M. S.; Ludwig, R. G.; Palermo Ruiz, G.; Knittel, T. L.; Davanzo, G. G.; Gerhardt, J. A.; Rodrigues, P. B.; Forato, J.; Amorim, M. R.; Brunetti Silva, N.; Martini, M. C.; Benatti, M. N.; Batah, S.; Siyuan, L.; Pereira Silva, R. E. M.; Joao, R. B.; Silva, Scardua
title: SARS-CoV-2 infects brain astrocytes of COVID-19 patients and impairs neuronal viability
date: 2020-10-13
journal: nan
DOI: 10.1101/2020.10.09.20207464
sha: f38d64dd4011b7531634ae73e88b975ebfb230d8
doc_id: 933190
cord_uid: rtus1qyw

COVID-19 patients may exhibit neuropsychiatric and/or neurological symptoms. We found that anxiety and cognitive impairment are manifested by 28-56% of SARS-CoV-2-infected individuals with mild or no respiratory symptoms and are associated with altered cerebral cortical thickness. Using an independent cohort, we found histopathological signs of brain damage in 19% of individuals who died of COVID-19. All of the affected brain tissues exhibited foci of SARS-CoV-2 infection, particularly in astrocytes. Infection of neural stem cell-derived astrocytes changed energy metabolism, altered key proteins and metabolites used to fuel neurons and for biogenesis of neurotransmitters, and elicited a secretory phenotype that reduces neuronal viability. Our data support the model where SARS-CoV-2 reaches the brain, infects astrocytes and triggers neuropathological changes that contribute to the structural and functional alterations in the brain of COVID-19 patients.

COVID-19 patients may exhibit neuropsychiatric and/or neurological symptoms. We found that anxiety and cognitive impairment are manifested by 28-56% of SARS-CoV-2-infected individuals with mild or no respiratory symptoms and are associated with altered cerebral cortical thickness. Using an independent cohort, we found histopathological signs of brain damage in 19% of individuals who died of COVID-19. All of the affected brain tissues exhibited foci of SARS-CoV-2 infection, particularly in astrocytes. Infection of neural stem cell-derived astrocytes changed energy metabolism, altered key proteins and metabolites used to fuel neurons and for biogenesis of neurotransmitters, and elicited a secretory phenotype that reduces neuronal viability. Our data support the model where SARS-CoV-2 reaches the brain, infects astrocytes and triggers neuropathological changes that contribute to the structural and functional alterations in the brain of COVID-19 patients.

COVID-19 is a disease caused by infection with the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Although the most commonly observed symptoms of COVID-19 are respiratory and related to pulmonary infection, a growing body of evidence has demonstrated that SARS-CoV-2 may have extrapulmonary effects 1 . Notably, over 30% of COVID-19 patients manifest neurological and even neuropsychiatric symptoms 2,3 , eventually presenting some degree of encephalitis 4 . More than half of these patients continue to exhibit neurological symptoms even three months after the onset of the disease, when the virus is no longer detected 5 . This is consistent with substantial damage to the nervous system 6 .

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020. 10.09.20207464 doi: medRxiv preprint To explore the extent of brain damage in COVID-19 patients, we performed cortical surface-based morphometry analysis using high-resolution 3T MRI of 81 subjects diagnosed with COVID-19 who had mild respiratory symptoms and did not require hospitalization or oxygen support. The analysis was performed within a median interval of 54 days after SARS-CoV-2 detection by RT-qPCR and the subjects were compared to 145 healthy volunteers (Supplementary Table 1 and 2). The analysis revealed areas of reduced cortical thickness in the left lingual gyrus, calcarine sulcus -including the cuneus -and olfactory sulcus -including the rectus gyrus (Fig. 1a) . In contrast, increased thickness was detected in the central sulcusincluding the precentral and postcentral gyrus -and superior occipital gyrus (Fig. 1a) , which can be associated with vasogenic edemas 7 . A subgroup of these individuals (n = 61) were subjected to neuropsychological evaluation for anxiety (Beck Anxiety Inventory, BAI), depression (Beck Depression Inventory, BDI), logical memory (Wechsler Memory Scale), cognitive functions (TRAIL Making Test) and fatigue (Chalder Fatigue Questionnaire, CFQ). These tests were performed between 21 and 120 days after diagnosis (median of 59 days). Symptoms of anxiety were identified in approximately 28% of the subjects, and 20% of individuals presented symptoms of depression ( Supplementary Fig. 1a ). Abnormal performances were observed in nearly 28% of participants on logical memory and approximately 34% and 56% on TRAIL A and B, respectively (Supplementary Table 3 and Supplementary Fig. 1b) . We also correlated the changes in cortex thickness with the neuropsychological evaluation. We identified a negative correlation between BAI and cortical thickness of orbitofrontal regions (adjusted for CFQ) ( Table 4 ) and a positive correlation between TRAIL B and cortical thickness of the right gyrus rectus ( Fig. 1c and Supplementary Table 5 ). We also found partial correlations between logical memory (immediate recall, adjusted for BAI, BDI and CFQ) and . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020. 10 .09.20207464 doi: medRxiv preprint cortical thickness of regions associated with language (Supplementary Table 6 ). These results suggest that a thinner cortex in these areas is associated with poor performance on this verbal memory task. Ov erall, our findings evidence major alterations in brain cortex structure associated with neuropsychiatric symptoms as a consequence of COVID-19.

Brain alterations in COVID-19 patients could be a consequence of inflammatory or hemodynamic changes secondary to peripheral infection or could be caused by the ability of SARS-CoV-2 to invade the central nervous system (CNS) and compromise cell viability and brain function. Although exacerbated inflammation and cardiovascular dysfunction have been well characterized in COVID-19 patients 8 , infection of the CNS by SARS-CoV-2 remains under debate. We performed a minimally invasive autopsy via endonasal trans-ethmoidal access of brain samples from 26 individuals who died of COVID-19 and analyzed histopathological features. We found alterations consistent with necrosis and inflammation in 19% of the brain tissues from these individuos (5/20) (Fig. 2a and Supplementary Fig. 2a) . Notably, SARS-CoV-2 genetic material and spike protein was detected in all of these five samples (Fig. 2b,c) . SARS-CoV-2 spike protein was found in about one third of the cells in one of the slices of brain tissue analyzed (Fig. 2d) , the majority of these cells being astrocytes (Fig. 2e ). We also found the virus in neurons ( Supplementary Fig. 2b ), but not in microglia ( Supplementary Fig. 2c ). The presence of SARS-CoV-2 spike protein correlated with the presence of double-stranded RNA (dsRNA) in the infected cells (Fig. 2f) , indicating replicative virus in the brain tissue.

We have also conducted liquid chromatography-mass spectrometry (LC/MS)-based shotgun proteomics with a different set of samples, consisting of 12 postmortem brain samples from COVID-19 patients vs. 8 SARS-CoV-2-negative controls. We identified 119 differentially expressed proteins with the most highly enriched pathways being associated with . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint neurodegenerative diseases and carbon metabolism (Fig. 2g) . Moreover, astrocyte proteins were enriched among the differentially expressed proteins, consistent with the higher frequency of infected astrocytes (Fig. 2h) .

In order to investigate the consequences of SARS-CoV-2 infection on astrocytes, we generated human neural stem cell-derived astrocytes, exposed them to the virus for 1h, and analyzed cell response after 24h. We confirmed that SARS-CoV-2 is able to infect human astrocytes ( Fig. 3a-c) . Notably, we also found the presence of dsRNA in SARS-CoV-2-infected astrocytes i n vitro , but not in mock control cells (Fig. 3a-c) . Altogether these results indicate that astrocytes are permissive cells for SARS-CoV-2 infection and represent a site for virus replication in the central nervous system.

In an attempt to identify downstream mechanisms triggered by SARS-CoV2 infection and possibly involved in the pathophysiology of COVID-19, we employed an unbiased proteomic analysis of infected astrocytes. LC/MS-based proteomics revealed 233 differentially expressed proteins in SARS-CoV-2-infected astrocytes compared to mock control cells. A group of approximately 50 proteins formed a molecular signature which distinguished infected astrocytes from controls (Fig. 4a) . Pathway enrichment analysis revealed that these proteins are involved in a wide range of biological processes (Fig. 4b) . When we compared these pathways to those normally enriched in brain proteomics, we found carbon metabolism, glycolysis/gluconeogenesis, biosynthesis of amino acids, pentose phosphate pathway and necroptosis in common (Fig. 4c) . In agreement, LC/MS-based metabolomic analysis of SARS-CoV-2-infected astrocytes showed marked changes in metabolic intermediates of glycolysis, the TCA cycle and anaplerotic reactions, indicating extensive remodeling of astrocyte metabolism ( Supplementary Fig. 3 ). This phenomenon was marked by the decrease in pyruvate . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint and lactate, which are downstream metabolites of the glycolytic pathway, as well as a reduction in glutamine and intermediates of glutamine metabolism such as glutamate, GABA and alpha-ketoglutarate (Fig. 4d) . Astrocyte bioenergetics were further characterized by Seahorse Extracellular Flux analysis, showing increased respiration in infected cells ( Fig. 4e and Supplementary Fig. 4 ). Together, these results demonstrate increased metabolic activity in SARS-CoV-2-infected astrocytes and a reduction of metabolites used by these cells to support neuronal metabolism.

These observations prompted us to investigate whether neuronal viability could be indirectly affected by infection of astrocytes with SARS-CoV-2. To test that, we cultured differentiated SH-SY5Y neurons with the conditioned medium of S ARS-CoV-2-infected astrocytes. SH-SY5Y cells are more closely related to adrenergic neurons, but they also express dopaminergic markers 9 . As observed in Fig. 4f , conditioned medium of S ARS-CoV2-infected astrocytes induced apoptosis in SH-SY5Y neurons. These results suggest that the infection of astrocytes by SARS-CoV-2 might be a trigger event to reduce neuronal viability.

This study and other reports showing alterations in brain structure and the manifestation of neurological symptoms in COVID-19 patients 10, 11 have raised a debate on whether these clinical features are consequence of peripheral changes or rather the potential ability of the virus to invade the CNS. Our findings support the latter, at least in part, as we found SARS-CoV-2 in the brain tissue collected from patients who died of COVID-19. The potential of SARS-CoV-2 to infect brain cells has been demonstrated using in vitro models such as stem cell-derived neural cells and cerebral organoids 12 . Viral particles have also been found in the brain 13 , localized in the microvasculature and in neurons 12 , as well as in the choroid plexus 14 and meninges 15 .

However, the magnitude of this infection and its distribution in the brain tissue had not been . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint demonstrated. We show that SARS-CoV-2 may infect almost one third of the cells in the brain and the majority of these cells are astrocytes.

In astrocytes, SARS-CoV-2 infection results in marked changes in cellular metabolism.

Since astrocyte metabolism is key to support neuronal function, we hypothesized that these changes could indirectly impact neurons. Astrocytes support neurons metabolically by exporting lactate 16 . One of the most critical alterations caused by SARS-CoV-2 infection in astrocytes is the decrease in pyruvate and lactate levels. In spite of these intracellular changes, lactate is not altered in the conditioned medium of SARS-CoV-2-infected astrocytes ( Supplementary Fig. 5 ), suggesting that infected astrocytes divert pyruvate to lactate production and export it to preserve neuronal metabolism; however, in order to sustain its own metabolism, carbon sources other than the ones derived from glucose would have to be used to fuel the TCA cycle. Supporting this notion, intermediates of glutamine metabolism such as glutamate and GABA are decreased in SARS-CoV-2-infected astrocytes, suggesting that glutaminolysis is being used as an alternative source of carbons to fuel astrocyte oxidative metabolism. Importantly, astrocyte-derived glutamine is required for neuronal synthesis of the neurotransmitters glutamate and GABA 17 .

Astrocytes play a vital role in neurotransmitter recycling, a crucial process for the maintenance of synaptic transmission and neuronal excitability. This is especially important for glutamatergic synapses since proper glutamate uptake by astroglia prevents the occurrence of excitotoxicity 18 .

Upon this uptake, glutamine synthetase converts glutamate to glutamine, which can then be transferred back to neurons, thus closing the glutamate-glutamine cycle. This is also true for GABAergic synapses, where the neurotransmitter GABA is taken up by astrocytes and metabolized first to glutamate and then to glutamine 19 . Moreover, astrocytes are responsible for maintaining glutamate levels in the brain. Hence, given the importance of the coupling between . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint astrocytes and neurons, astrocytic alterations in glutamine metabolism are expected to compromise neuronal function, affecting synaptic function and plasticity 20 .

We also found that SARS-CoV-2 infection elicits a secretory phenotype in astrocytes that results in increased neuronal apoptosis. Neuronal death may explain, at least partially, the alterations in cortical thickness found in COVID-19 patients. A recent study with 60 recovered patients and 39 healthy controls also identified gray matter abnormalities 97 days after the onset of the disease, with increased volume in some areas of the brain 5 . While that study analyzed hospitalized patients, we evaluated individuals that did not have to be hospitalized (i.e. had mild respiratory symptoms), and nevertheless, we observed notable alterations of cortical thickness.

Importantly, some of these alterations correlated with symptoms of anxiety and impaired cognition, which is consistent with previous literature 21, 22 . Since one of the hypotheses for the neuroinvasive mechanism of SARS-CoV-2 is via the olfactory nerves 23 , we speculate that the associations between BAI and TRAIL B scores and structural alterations in the orbitofrontal region may be a result of the action of the virus in this cortical area, closely related to the olfactory nerves.

Our findings are consistent with a model in which SARS-CoV-2 is able to reach the central nervous system of COVID-19 patients, infects astrocytes and secondarily impairs neuronal function and viability. These changes are likely to contribute to the alterations of brain structure as observed here and elsewhere, thereby resulting in the neurological and neuropsychiatric symptoms manifested by some COVID-19 patients. Our study comes as a cautionary note that interventions directed to treat COVID-19 should also envision ways to prevent SARS-CoV-2 invasion of the CNS and/or replication in astrocytes.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint

Eighty-one patients (60 women, median 37 years of age) previously infected with SARS-CoV-2 were enrolled prospectively for this study after signing an informed consent form approved by the local Ethics Committee. These individuals presented mild symptoms during the acute phase and did not require hospitalization or oxygen therapy. They had a median interval of 54 days (range 16-120 days) between their RT-PCR exam and the day of MRI scanning and interview. For cortical thickness analysis, we included one hundred and forty-five controls (103 women, median 38 years of age) 24 from our Neuroimaging databank, given the difficulties and risk of recruiting healthy volunteers during the pandemic. The outpatients and healthy controls were balanced for age (p=0.45) and sex (p=0.65). The neuropsychological evaluations and neuroimaging analyses were approved by the Research Ethics Committee of the University of Campinas (CAAE: 31556920.0.0000.5404) and all subjects signed a consent form to participate.

We performed neuropsychological evaluations of sixty-one of these patients. They were tested for symptoms of anxiety using the Beck Anxiety Inventory (BAI) and symptoms of depression using the Beck Depression Inventory (BDI) 25 . Symptoms of anxiety were confirmed for those with a BAI higher than 10 points, and depression symptoms defined for those with minimum of 14 points on the BDI. In terms of depression, subjects were categorized with mild . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. 26 , in which the examiner verbally presents two stories, and each story includes 25 pertinent pieces of information. Subjects are required to recall details of each story immediately after its presentation and again after 20 minutes. To evaluate other cognitive functions, we applied the TRAIL Making Test (TMT) , which is subdivided into two steps.

Step A assesses processing speed and visual search in a task that requires ascending connection order of 25 numbers, randomly arranged.

Step B evaluates alternating attention and cognitive flexibility in a task associated with shifting rules in an ascending sequence of 25 numbers. A training stage is applied to both steps. We calculated z-scores for the results of these tests based on Brazilian normative data 26, 27 . For each test, the function was categorized as:

"preserved" if the z-score was higher than -0.99; "below average" when the z-score was between -1 and -1.49; "minor impairment" if the z-score was between -1.5 and -1.99; and "major impairment" for z-score values equal to or lower than -2. The Chalder Fatigue Questionnaire 28 was used to evaluate fatigue in these subjects; they were instructed to answer 11 questions (measured on a Likert scale 0-3), which yields a global score out of 33.

We obtained the structural, 3D, T1-weighted images from a 3T Achieva-Philips MRI is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint Thickness (CT) maps, according to the default parameters. The T1 images were initially spatially registered and segmented into gray matter, white matter and cerebrospinal fluid. Finally, we calculated the cortical thickness using the projection method described by Dahnke et al. 29 . For voxelwise analysis of extracted maps, we used CAT12/SPM12 tools for an independent T-test (comparing COVID-19 patients and healthy controls), including age and sex as covariates. The results displayed were corrected for multiple comparisons using False Discovery Rate (FDR) 30 correction (p<0.05). For anatomical identification, we used the "Destrieux Atlas 2009" 31 .

Cortical parcellation was performed with standard CAT12 tools to extract the cortical thickness of regions of interest for correlations with neuropsychological scores. We used SPSS 22 for statistical analysis of clinical and neuropsychological variables. The FDR procedure was applied to adjust p-values for multiple comparisons (when necessary) with R software 32 .

Twenty-six individuals who died from complications related to COVID-19 were autopsied with an ultrasound-guided, minimally invasive approach using endonasal trans-ethmoidal access . Brain tissue samples were collected and fixed using a 10% neutral buffered formalin solution. After fixation, the tissue was embedded in a paraffin block and sectioned into slices with a thickness of 3 μm. The sections were stained by H&E and immunofluorescence. For proteomic analysis, twelve COVID-19 patients were autopsied using the same approach. Brain tissue samples were collected and macerated in lysis buffer (100 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, 0.5% Triton X-100) prior to trypsin digestion.

The autopsy studies were approved by the National Commission for Research Ethics (CAAE: 32475220.5.0000.5440 and CAAE: 38071420.0.1001.5404).

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. 

The HIAE-02-SARS-CoV-2/SP02/human/2020/BRA (GenBank accession number is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint (ICB-USP). Viral stock was propagated in Vero CCL-81 cells (ATCC) cultivated in DMEM supplemented with 10% heat-inactivated FBS and 1% of penicillin and streptomycin, and incubated at 37°C with 5% CO 2 atmosphere. Viral titer was determined by the plaque-forming assay using Vero cells. 

Astrocytes infected with SARS-CoV-2 and a mock control were collected in biological triplicates. Cells were chemically lysed with Lysis Buffer: (100 mM Tris-HCl, 1 mM EDTA, 150 mM NaCl, 1% Triton-X, protease and phosphatase inhibitors) and mechanically lysed with an ultrasonication probe in 3 cycles of 20s each with 90% of frequency. The total protein extract was quantified by BCA, according to the manufacturer's instructions (Thermo Fisher Scientific, MA, USA). 30 µg of total protein extract from each sample was transferred to a Microcon-10 Centrifugal Filter, with 10 kDa cutoff, for FASP protein digestion 35 . Proteins were reduced (10 mM DTT), alkylated (50 mM IAA) and digested overnight by trypsin at 37°C in 50 mM . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint ammonium bicarbonate (AmBic), pH 8.0. One the following day, trypsin activity was quenched by adding formic acid (FA) to a final concentration of 1% (v/v) and the peptides were recovered from the filter in 50 mM AmBic, concentrated in a SpeedVac and stored at -80°C until use. The raw data from each experiment were processed in Progenesis QI for proteomics (Waters Corporation, Milford, MA). Tandem mass spectra were searched against the Homo sapiens proteome database (UNIPROT Protein reviewed release 2020-04), using tolerance parameters of 20 ppm for precursor ions and 10 ppm for product ions. For peptide identification, carbamidomethylation of cysteines was set as a fixed modification, oxidation of methionines as a . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint variable modification, 2 missed cleavages, and a false discovery rate (FDR) < 1%. Protein identification was performed using a minimum of 1 fragment ion matched per peptide, a minimum of 3 fragment ions per protein and a minimum of 1 peptide per protein.

Label-free quantitative analysis was carried out using the relative abundance intensity normalized by all peptides identified. The expression analysis was performed considering the technical replicates for each experimental condition, following the hypothesis that each group is independent. Proteins with ANOVA (p) ≤ 0.05 between the groups were considered differentially expressed.

The medium was washed twice with PBS at physiologic pH, then the cells (10 6 is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint gradient from 99% to 1% buffer A over 7 min. The column was returned to 99% buffer A for 2 min for re-equilibration before the next injection for a total run time of 10 min. Data acquisition was performed in negative mode and the instrument was operated in MS e mode in the m/z range of 50-800 Da, with an acquisition time of 0.1 s per scan.

The raw files were preprocessed by Progenesis QI software by Waters®, and identification was executed using multiple data banks. Identification of the metabolites of interest was carried out manually by spectral features, and the identification level 3 was obtained according to Schrimpe-Rutledge et al. 36 using 5ppm as the error cutoff. The integration area of each peak was used to calculate the violin plot graph and an unpaired t-test with Welch's correction was used for statistical comparison. All analysis were performed using GraphPad Prism 8.0 software (San Diego, CA, USA) and a significance level of p ≤ 0.05 was adopted. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint specificity, a melting curve analysis was performed by heating samples from 65°C to 99°C (1°C increment changes at 5s intervals). All sample measurements were performed in duplicate.

Primers were designed with PrimerBlast and used at the concentration of 200 nM. Data were normalized to the expression of 18S (Fwd 5' CCCAACTTCTTAGAGGGACAAG 3'; Rev 5' CATCTAAGGGCATCACAGACC 3') and the relative quantification value of each target gene was determined using a comparative CT method 37 . For virus detection, SARS-CoV-2 nucleocapsid N1 primers were used as previously described (Fwd 5' CAATGCTGCAATCGTGCTAC 3'; Rev 5' GTTGCGACTACGTGATGAGG 3') 38,39 . A serial dilution of SARS-CoV-2 was used as a standard curve. Data were expressed as mean ± SEM.

Statistical significance analysis was calculated by two-tailed unpaired Student's t-test. All analysis were performed using GraphPad Prism 8.0 (San Diego, CA, USA) and a significance level of p ≤ 0.05 was adopted.

Astrocytes were plated on Seahorse XF-24 plates at a density of 1.5x10 4 cells per well and incubated in complete culture medium for two days at 37°C in 5% CO 2 . 24 hours before the experiment, cells were either infected by SARS-CoV-2 (MOI 0.1) or not infected (MOCK). One day post-infection, the culture medium was changed to Seahorse Base medium (supplemented with 1 mM pyruvate, 2 mM glutamine, and 10 mM glucose) and cells were incubated at 37°C in a non-CO 2 incubator for 1 hour. OCR (Oxygen Consumption Rate) was measured over the course of the experiment under basal conditions and after the injections of oligomycin (1 µM), FCCP (5 µM) and antimycin A/rotenone (100 nM/1 µM). Protein concentration was determined for each well to normalize the data. Data were expressed as mean ± SEM of at least two independent experiments performed in quintuplicate. Statistical significance analysis was . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint calculated by two-tailed unpaired Student's t-test. All analysis were performed using GraphPad Prism 8.0 software (San Diego, CA, USA) and a significance level of p ≤ 0.05 was adopted.

Astrocytes and SH-SY5Y cells were cultured separately in standard conditions until complete differentiation. Next, astrocytes were infected either with MOCK or SARS-CoV-2 (MOI 0.1) and after 24 hours, the medium was removed and cells were washed with PBS and cultured for 24 hours. The SH-SY5Y medium was removed and replaced by astrocyte-conditioned medium and cells were incubated for 24 hours. After incubation, cells and the medium were collected for Flow Cytometer Analysis, following the described procedure below. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint

Astrocyte supernatants from infected or non-infected cells were collected and deproteinized using 1 M perchloric acid (PCA) for 1 hour. Protein precipitation was performed by centrifugation at 18,000 xg for 5 minutes. pH was neutralized using KOH and lactate concentration was determined using a colorimetric L-lactate assay kit (cat. no. 138; Labtest).

Data were expressed as mean ± SEM of at least two independent experiments performed in triplicate. Statistical significance was calculated by two-tailed unpaired Student's t-test. All analysis were performed using GraphPad Prism 8.0 software (San Diego, CA, USA) and a significance level of p ≤ 0.05 was adopted.

The mass spectrometry proteomic data have been deposited to the ProteomeXchange . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint c) KEGG-enrichment analysis of differentially expressed proteins in SARS-CoV-2 infected astrocytes vs. mock as compared to postmortem brain tissue from COVID-19 patients vs.

controls. Dot size represents the number of proteins related to the respective cell type and the p value adjusted by the false discovery rate (FDR). d) High-resolution mass spectrometry quantification of pyruvate, lactate, glutamine, glutamate, GABA, and a-ketoglutarate in . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint SARS-CoV-2 infected astrocytes vs. mock. The integration area of each peak was used to calculate the violin plot graph and unpaired t-test with Welch's correction was used for statistical comparison. e) Oxygen consumption rate (OCR) of SARS-CoV-2 infected astrocytes vs. mock.

SeaHorse Flux Analysis using the MitoStress test where basal respiration was measured followed by determination of oligomycin-, FCCP-, and rotenone/antimycin-induced respiration. f) Human neuronal cell line SH-SY5Y was cultured for 24 h in the presence of astrocyte-conditioned media (ACM) from mock or SARS-CoV-2 (CoV-2) infected cells. Representative histograms of cell apoptosis as measured by ApotrackerGreen staining. Data are representative of at least two independent experiments performed in triplicate (flow cytometry and metabolomics analysis) or quintuplicate (SeaHorse Flux Analysis), and shown as mean ± SEM. P values were determined by two-tailed unpaired with Welch's correction (c and d) and one-way ANOVA followed by Tukey's post hoc test (e). *P < 0.05; **P < 0.01; **** P < 0.0001 compared to mock, and #P < 0.05 compared to CoV-2 (ACM+). . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted October 13, 2020. . https://doi.org/10.1101/2020.10.09.20207464 doi: medRxiv preprint

Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study

COVID-19: ICU delirium management during SARS-CoV-2 pandemic

Neurologic Manifestations of Hospitalized Patients With Coronavirus Disease

Neurological and neuropsychiatric complications of COVID-19 in 153 patients: a UK-wide surveillance study

Cerebral Micro-Structural Changes in COVID-19 Patients -An MRI-based 3-month Follow-up Study

Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) and the Central Nervous System

Differential cortical thickness across the central sulcus: a method for identifying the central sulcus in the presence of mass effect and vasogenic edema

Prevalence and impact of cardiovascular metabolic diseases on COVID-19 in China

Considerations for the use of SH-SY5Y neuroblastoma cells in neurobiology

Are we facing a crashing wave of neuropsychiatric sequelae of COVID-19? Neuropsychiatric symptoms and potential immunologic mechanisms

Psychiatric and neuropsychiatric presentations associated with severe coronavirus infections: a systematic review and meta-analysis with comparison to the COVID-19 pandemic

Neuroinvasion of SARS-CoV-2 in human and mouse brain

Neuropathology of patients with COVID-19 in Germany: a post-mortem case series

Human Pluripotent Stem Cell-Derived Neural Cells and Brain Organoids Reveal SARS-CoV-2 Neurotropism

Update on neurological manifestations of COVID-19

Neuronal-astrocyte metabolic interactions: understanding the transition into abnormal astrocytoma metabolism

Knockout of GAD65 has major impact on synaptic GABA synthesized from astrocyte-derived glutamine

Histopathological alterations revealed by H&E images of postmortem brain tissue from individuals who died of COVID-19. Samples from 26 individuals were analyzed and 5 showed alterations. Case 1: intraparenchymal cerebral vessel with margination of inflammatory cells through endotelium; Case 2: focal infiltration of inflammatory cells -diapedesis; Case 3: intraparenchymal vascular damage

The image depicts staining for: b) nuclei (DAPI, blue), NeuN (red, neuron marker), dsRNA (magenta), and SARS-CoV-2-S (green); and c) nuclei (DAPI, blue), ionized calcium-binding adaptor molecule 1 (Iba1, red, microglia marker), dsRNA (magenta), and SARS-CoV-2-S (green)

Supplementary Fig. 3: Metabolomic analysis of SARS-CoV-2-infected astrocytes

High-resolution mass spectrometry quantification of citrate, palmitate, acetate, fumarate, succinate, oxaloacetate, and malate in SARS-CoV-2 infected astrocytes vs. mock. The integration area of each peak was used to calculate the violin plot graph and unpaired t-test with Welch's correction was used for statistical comparison

Human neural stem cell-derived astrocytes were infected in vitro with SARS-CoV-2 (MOI 0.1) for 1 h, washed thoroughly and harvested after 24 h. Mock was used as a control

We thank Edison Luiz Durigon for providing the SARS-CoV-2 299. We thank Gabriela Lopes Vitória, Elzira E. Saviani and Paulo Baldasso for technical support. The authors would