Abstract
Major cell entry factors of SARS-CoV-2 are present in neurons; however, the neurotropism of SARS-CoV-2 and the phenotypes of infected neurons are still unclear. Acute neurological disorders occur in many patients, and one-third of COVID-19 survivors suffer from “brain diseases”. Here, we show that SARS-CoV-2 invades the brains of five patients with COVID-19 and Alzheimer’s, autism, frontotemporal dementia or no underlying condition by infecting neurons and other cells in the cortex. SARS-CoV-2 induces or enhances Alzheimer’s-like neuropathology with manifestations of β-amyloid aggregation and plaque formation, tauopathy, neuroinflammation and cell death. SARS-CoV-2 infects mature but not immature neurons derived from inducible pluripotent stem cells from healthy and Alzheimer’s individuals through its receptor ACE2 and facilitator neuropilin-1. SARS-CoV-2 triggers Alzheimer’s-like gene programs in healthy neurons and exacerbates Alzheimer’s neuropathology. A gene signature defined as an Alzheimer’s infectious etiology is identified through SARS-CoV-2 infection, and silencing the top three downregulated genes in human primary neurons recapitulates the neurodegenerative phenotypes of SARS-CoV-2. Thus, SARS-CoV-2 invades the brain and activates an Alzheimer’s-like program.
Introduction
The COVID-19 pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has infected at least 213 million people worldwide and 37.8 million Americans to date. SARS-CoV-2 not only causes respiratory syndromes but also leads to neurological abnormalities, with an 85% occurrence rate in patients with Alzheimer’s disease [1, 2]. In fact, neurological symptoms, including hypogeusia, headache and anosmia, precede the onset of respiratory symptoms in the majority of patients with COVID-19. The major chronic sequelae of COVID-19 are expected to be cognitive decline and dementia [3]. A recent study has shown that one-third of COVID-19 survivors exhibit substantial neurological and psychiatric morbidity in the 6 months after SARS-CoV-2 infection [4]. Furthermore, the “brain disease” risk is not limited to patients who have severe COVID-19 [4]. Thus, the COVID-19 pandemic provides a unique but unwelcomed opportunity to study the contribution of SARS-CoV-2 to neurological disorders, including Alzheimer’s disease.
It has been proposed that β-coronaviruses, including SARS-CoV-2, can invade the central nervous system [5]. The two coronaviruses closely related to SARS-CoV-2, Middle Eastern respiratory syndrome coronavirus (MERS-CoV) and severe acute respiratory syndrome coronavirus 1 (SARS-CoV-1), can infect the central nervous system [6, 7]. A recent study shows the presence of SARS-CoV-2 in a patient’s olfactory mucosa and its neuronal projections [8]. SARS-CoV-2 RNA is present in 36.4% of brain biopsies of fatal COVID-19 cases [9]. A recent study observes the presence of SARS-CoV-2 spike protein in some brain regions in all three COVID-19 cases studied [10]. In transgenic hACE2 mice, SARS-CoV-2 is massively present in the brain at post infection day 5 [11]. In addition to olfactory transmucosal entry of SARS-CoV-2 into the central nervous system (CNS) [8], other potential routes of SARS-CoV-2 brain entry, including blood–brain barrier (BBB) passage, especially under conditions of a compromised BBB, such as in cases of Alzheimer’s disease and autism [12, 13], and the infiltration of infected immune cells have been proposed [14], though evidence for infected immune cells remains scarce. However, neurological disorders are not limited to severe COVID-19 cases, suggesting that multiple brain entries may account for SARS-CoV-2 neurotropism. Although SARS-CoV-2 can enter the brain, experimental evidence of its presence in key brain regions involved in cognitive functions is still lacking, and little is known about the functional impact of SARS-CoV-2 CNS invasion on neurons.
SARS-CoV-2 entry into host cells is facilitated by its cell surface receptors. The spike proteins of both SARS-CoV-2 and the earlier SARS-CoV-1 bind to angiotensin-converting enzyme 2 (ACE2) as the first step of cellular entry [15]. The presence of ACE2 in neurons of the brain has been demonstrated [16]. However, ACE2 is expressed at relatively low levels in most tissues [17]. SARS-CoV-2 exhibits much higher infectivity than SARS-CoV-1. This evidence implicates either the existence of other cell surface receptors in SARS-CoV-2 cell entry or the facilitation of cell entry by other factors. Indeed, neuropilin-1 (NRP1) has been identified as a facilitator of SARS-CoV-2 entry when ACE2 is expressed, leading to high rates of infection [17]. NRP1 effectively binds to the protease furin-cleaved spike protein of SARS-CoV-2 [17], a process that does not occur for SARS-CoV-1. NRP1 is abundantly expressed in cells of many tissues including neurons [18]. The presence of ACE2 and NRP1 in neurons suggests the high possibility of neurotropism. It is of interest to determine in which cell types in the CNS and at which stage of neural development SARS-CoV-2 can exert its infectibility.
There is a profound link between SARS-CoV-2 infection and dementia/Alzheimer’s disease. The 2002 and 2012 SARS and MERS epidemics caused memory impairment in many recovered patients [19]. Neurological syndromes related to Alzheimer’s disease, including neuroinflammatory syndromes [20, 21], seizures [22], delirium [23, 24], and alterations in personality, behavior, and cognitive deficits [20, 25], frequently occur in patients with COVID-19 who recover from COVID-19. Thus, SARS-CoV-2 infection of the brain may increase the risk of Alzheimer’s disease. The direct effects of SARS-CoV-2 on neuronal function and survival, the inflammatory cytokine response, and hypoxia may lead to an Alzheimer’s-like manifestation [26, 27]. Additionally, Alzheimer’s patients have twice the risk of contracting COVID-19, which deteriorates Alzheimer’s symptoms and increases mortality [28]. The cellular mechanism underlying the possible neurotropism of SARS-CoV-2 needs to be revealed for the development of possible treatments for a large number of COVID-19 survivors suffering from neurological disorders.
In this study, we sought to test whether SARS-CoV-2 is neurotropic and infects neural cells in the cognitive center in five patients with Alzheimer’s disease, autism, frontotemporal dementia and no underlying condition, respectively. SARS-CoV-2 infection alters transcriptomic landscapes in favor of the development of Alzheimer’s-like neuropathology in non-Alzheimer’s individuals and exacerbates Alzheimer’s neuropathology in patients with Alzheimer’s. We found that SARS-CoV-2 invades the cognitive centers of all five COVID-19 patients, leading to Alzheimer’s-like neuropathology or Alzheimer’s neuropathology exacerbation. SARS-CoV-2 infects human inducible pluripotent stem cell (iPSC)-derived mature neurons from healthy individuals, leading to amyloid beta (Aβ) deposition, increased inflammation, neuronal death and increased expression of Alzheimer’s mediators. Strikingly, we found that SARS-CoV-2-infects neurons from healthy individuals through a shared gene expression program with Alzheimer’s neurons, leading to activation of the infectious pathways and supporting the infectious etiology of Alzheimer’s disease.
Results
SARS-CoV-2 invades cognitive centers of the brain
Because ACE2, the SARS-CoV-2 cellular receptor, and NRP1, a facilitator of SARS-CoV-2 entry, are expressed in CNS neurons, we hypothesized that SARS-CoV-2 can infect neural cells in the brain, especially under conditions such as Alzheimer’s disease and autism with BBB compromise. SARS-CoV-2 has been observed in cells of olfactory bulbs. We focused on cognitive centers of the brain. The SARS-CoV-2 spike protein and nucleocapsid protein were detected in cells of the inferior frontal cortexes of five COVID-19 cases: two autism cases, one Alzheimer’s case, one frontotemporal dementia (FTD) case and one case without any underlying health conditions (apparently healthy case) (Fig. 1, Fig. S1). In contrast, there was no positive staining of spike protein or nucleocapsid protein in the cortexes of the non-COVID-19 autism brains (Fig. 1A, C, Fig. S1A). Spike protein- and nucleocapsid protein-positive viral particles were robustly present in the cytoplasm and cellular projections of cortical cells (Fig. 1E, F). RNAscope in situ hybridization detected the abundant presence of SARS-CoV-2 genomic RNA in cortical cells of the five COVID-19 cases (Fig. 1G, H; Fig. S1F). PCR analysis using the CDC method [29] confirmed the presence of SARS-CoV-2 RNA in the three cases that had frozen brain tissues (Fig. 1I). In the 31-year-old COVID-19 autism case, the amounts of SARS-CoV-2 RNA in the inferior frontal cortex and the dorsolateral prefrontal cortex were comparable to those in the lungs (Fig. 1I). In the COVID-19 FTD case and the COVID-19 apparently healthy case, SARS-CoV-2 RNA was also detected, whereas there was no SARS-CoV-2 RNA in the age-matched non-COVID-19 FTD individual or in the age-matched non-COVID-19 apparently healthy individual (Fig. 1I).
Additional analyses showed that cells of the three cortical regions, the entorhinal cortex, the inferior frontal cortex and the dorsolateral prefrontal cortex, in the two COVID-19 autism cases and the COVID-19 Alzheimer’s case possessed abundant spike protein staining signals (Fig. 1J, K, L). The numbers of SARS-CoV-2 spike protein-positive cells in these three regions were different in these three COVID-19 cases (Fig. 1J, K, L). These findings demonstrated that SARS-CoV-2 was neurotropic in this cohort of COVID-19 patients (Table S1).
SARS-CoV-2 infects CNS cells expressing ACE2 and NRP1
Because SARS-CoV-2 exists in cortical cells in COVID-19 patients, these cells must express ACE2 and NRP1. Indeed, the SARS-CoV-2 spike protein staining signal was colocalized with both ACE2 and NRP1 signals (Fig. 2A, B), indicating that SARS-CoV-2 infects cortical cells using these proteins. ACE2 was colocalized with the neuron marker neurofilament light chain (NFL) (Fig. 2C, D), the pan-neuron marker class III beta-tubulin (Tuj1) (Fig. 2E), the oligodendrocyte marker 2’,3’-cyclic nucleotide-3’-phosphodiesterase (CNPase) (Fig. 2F), and the microglia marker ionized calcium-binding adapter molecule 1 (Iba1) (Fig. 2G). Quantitative analysis of double ACE2 and cell marker positive cells indicated that all neurons expressed ACE2 (Fig. 2H). Next, we determined which cell types in the cortexes were infected by SARS-CoV-2. We used dual immunohistological labeling with antibodies against cell type-specific markers and the SARS-CoV-2 spike protein. Both the mature neuron marker neuronal nuclear protein (NeuN) and the pan-neuron marker class III beta-tubulin (Tuj1) were colocalized with the spike protein in cells of the inferior frontal cortex (Fig. 2I, J). In the same region, both GABAergic inhibitory neurons (GAD65-positive) and glutamatergic excitatory neurons (glutamine synthetase-positive) contained SARS-CoV-2 spike protein staining signals (Fig. 2K, L). The SARS-CoV-2 spike protein staining signal was also present in Iba1-positive microglia and CNPase-positive oligodendrocytes in the same region (Fig. 2M, N). However, there was no detectable SARS-CoV-2 spike protein staining signal in glial fibrillary acidic protein (GFAP)-positive astrocytes (Fig. 2O). Quantification of double SARS-CoV-2 spike protein- and different cell type marker-positive cells indicated that spike protein was present in all neurons (Fig. 2P). Thus, SARS-CoV-2 infects various CNS cell types. Because infected peripheral immune cells may enter the brain, T and B cells were examined in the inferior frontal cortex. There were no T, B cell, or macrophage marker staining signals in this region of the COVID-19 autism cases (Fig. S2). However, a significant number of these immune cells were detected in the COVID-19 Alzheimer’s brain (Fig. S2).
SARS-CoV-2 causes cell death via multiple pathways
To determine whether SARS-CoV-2-infected cells in the cortical regions undergo cell death, we used immunohistological dual labeling with antibodies against the spike protein and cleaved caspase-3. There was no detectable cleaved caspase 3 staining signal in cells of the inferior frontal cortexes of non-COVID-19 autism controls (Fig. 3A). Over 90% of cleaved caspase 3-positive cells were SARS-CoV-2 spike protein positive in the cortexes of two COVID-19 autism cases (Fig. 3A). In the COVID-19 Alzheimer’s disease case, over 20% of cleaved caspase 3-positive cells were SARS-CoV-2 spike protein positive (Fig. 3B). Cleaved caspase 3 was induced in two COVID-19 autism cases and the COVID-19 case without underlying health conditions and was enhanced in the COVID-19 Alzheimer’s case and the COVID-19 FTD case (Fig. S3A, B, C).
To further determine the type of programmed cell death, we examined the presence of necroptotic, ferroptotic, and senescent cells. The cell necroptosis markers phospho-MLKL (mixed lineage kinase domain-like) and phospho-RIPK3 coexisted in spike protein-positive cells (Fig. 3C, Fig. S3D). Similarly, the two cell ferroptosis markers TfR1 (transferrin receptor) and ASCL4 (long-chain fatty acyl-CoA synthetase 4) were expressed in cells with positive spike protein signals (Fig. 3D, Fig. S3E). Because the cytokine associated with SARS-CoV-2 may trigger the cellular senescence program, we detected the senescence marker DPP4 (dipeptidyl-peptidase 4) in a subset of spike protein-positive cells (Fig. 3E). There were negligible double spike protein- and TfR1- or p-RIPK3-positive cells in the COVID-19 Alzheimer’s case (Fig. 3D, Fig. S3D). These findings suggest that SARS-CoV-2 leads to cell death and senescence through multiple pathways.
SARS-CoV-2 induces neuroinflammation
It has been suggested that SARS-CoV-2 infection leads to neuroinflammation. However, direct evidence on the link between SARS-CoV-2 and neuroinflammation is still lacking. The protein expression of two cytokines, IL-1β and IL-6 (interleukin 6), was significantly increased in the cortical cells of COVID-19 autism patients compared to the cortical cells of non-COVID-19 autism patients (Fig. S4A, B, C), providing the direct evidence of neuroinflammation induced by SARS-CoV-2.
SARS-CoV-2 induces Alzheimer’s-like phenotype development and exacerbation
Cellular Aβ (amyloid beta) aggregates were observed in the cortexes of the two COVID-19 autism cases (Fig. 4A) and the one COVID-19 case without underlying health conditions (Fig. 4B), whereas it was not present in the cortexes of age-matched non-COVID-19 autism cases and an apparently healthy individual (Fig. 4A, B). Extracellular Aβ plaques were present in one of the COVID-19 autism cases (Fig. 4A). Immunofluorescence analysis with thioflavin-T confirmed that the cytoplasmic deposition of Aβ in the COVID-19 autism cases consisted of aggregated Aβ (Fig. 4A). There was more Aβ plaque deposition per measured area in the cortexes of the COVID-19 Alzheimer’s and FTD brains than in the cortexes of age-matched non-COVID-19 Alzheimer’s and FTD brains (Fig. 4C, D). One of the neuropathological hallmarks of Alzheimer’s disease is the development of intracellular neurofibrillary tangles (NFTs) composed of hyperphosphorylated Tau (microtubule-associated protein tau). p-Tau-containing NFTs are associated with neuronal dysfunction, cognitive deficits and neuronal death [30, 31]. p-Tau-containing NFTs were present in the inferior cortexes of the two COVID-19 autism cases (Fig. 4E). Cellular p-Tau deposition was induced in these two COVID-19 autism cases and the COVID-19 case without underlying health conditions (Fig. 4E, F), whereas there were no signals or negligible p-Tau staining signals in the cortexes of non-COVID-19 control brains (Fig. 4E, F). There were significantly higher numbers of Pick bodies in the COVID-19 FTD case than in age-matched non-COVID-19 FTD cases (Fig. 4G, H). Thus, SARS-CoV-2 infection is linked to Alzheimer’s neuropathology.
SARS-CoV-2 infects iPSC-derived mature neurons
Based on the above findings in neurons of COVID-19 patients’ brain cortexes, we propose that SARS-CoV-2 can effectively infect neurons. To establish an in vitro platform to study this process, we obtained iPSCs derived from age-matched healthy individuals and Alzheimer’s patients and differentiated it into neurons, followed by SARS-CoV-2 infection. SARS-CoV-2-GFP (in which GFP replaced the viral open reading frame ORF7a [32]) at a multiplicity of infection (MOI) of 0.1 or 0.2 did not infect any cells at iPSC neuron differentiation day 35 (Fig. 5A, Fig. S5A) and some of these cells expressed the pan-neuron marker Tuj1 but did not express the mature neuron marker NeuN (Fig. 5B, Fig. S5B). At iPSC differentiation day 50, SARS-CoV-2-GFP at an MOI of 0.05, 0.1 or 0.2 effectively infected cells differentiated from iPSCs from healthy individuals and Alzheimer’s patients (Fig. 5C, Fig. S5C, D), and SARS-CoV-2-GFP could be detected in cell culture media 72 hours post SARS-CoV-2-GFP infection (Fig. 5D), indicating that the virus not only infects cells but also replicates intracellularly. SARS-CoV-2- infected cells were essentially all Tuj1-positive cells at iPSC differentiation day 50 (Fig. 5E), and no GFAT-positive astrocytes were detected in mock- and SARS-CoV-2-infected cells (Fig. 5F). At iPSC differentiation day 35, there was no expression of ACE2 or NRP1 proteins (Fig. 5G, H). At iPSC differentiation day 50, robust ACE2 and NRP1 protein expression existed in Tuj1-positive neurons (Fig. 5I, J). Over 40% of Tuj1-positive neurons were ACE2- and/or NRP1-positive (Fig. 5I, J). Subsequent experiments were conducted on iPSC differentiation day 50. These results suggest that SARS-CoV-2 infects mature neurons via ACE2 with the facilitation of NRP1.
SARS-CoV-2 induces Alzheimer’s phenotypes in iPSC-derived cells
Because SARS-CoV-2 induces Aβ cellular aggregates and extracellular plaques, p-Tau cellular deposition and NFTs and neuroinflammation in COVID-19 patients, we hypothesized that SARS-CoV-2 infection can turn neurons derived from iPSCs from healthy individuals into Alzheimer’s-phenotype neurons. Neurons differentiated from iPSCs of healthy individuals and Alzheimer’s patients were infected with wild-type SARS-CoV-2 (WA-1 strain [33]) at an MOI of 0.1 for 48 hours (Fig. 6, Fig. S6). SARS-CoV-2 induced cellular Aβ aggregates in healthy neurons and increased cellular Aβ aggregates in Alzheimer’s neurons (Fig. 6A). Cellular p-Tau deposition was induced in healthy neurons after 72 hours of SARS-CoV-2 infection (Fig. 6B), and the virus further increased cellular p-Tau deposition in Alzheimer’s neurons (Fig. 6B). Compared to their mock-infected counterparts, both healthy neurons and Alzheimer’s neurons had higher levels of major inflammatory cytokines including IL-1β, IL-6, IFNγ and TNFα after SARS-CoV-2 infection (Fig. 6C). Among the critical Alzheimer’s mediators, amyloid precursor protein (APP), enzyme β-secretase 1 (BACE1), and presenilin 1/2 (PSEN1/2), SARS-CoV-2 significantly increased BACE1 expression in healthy neurons and Alzheimer’s neurons but did not affect the expression of the other Alzheimer’s mediators (Fig. 6D, E). Likewise, SARS-CoV-2 significantly increased the number of cleaved caspase 3-positive cells differentiated from iPSCs of healthy individuals and Alzheimer’s patients (Fig. 6F). Thus, SARS-CoV-2 triggers an Alzheimer’s-like cellular program in neurons derived from iPSCs of healthy individuals and enhances Alzheimer’s phenotypes in cells derived from Alzheimer’s iPSCs.
Alzheimer’s infectious etiology genes identified via SARS-CoV-2
Over 95% of Alzheimer’s cases are sporadic and their causes are still unclear. Studies of DNA viruses have shown that Alzheimer’s etiology has an infectious component [34, 35]. Based on the above observation that SARS-CoV-2 induces Alzheimer’s phenotypes in neurons derived from iPSCs of healthy non-Alzheimer’s individuals, we aimed to utilize SARS-CoV-2 infection to reveal genes responsible for the Alzheimer’s infectious etiology. The transcriptomes of neurons differentiated from iPSCs of healthy individuals and Alzheimer’s patients were determined by RNA sequencing (Fig. 7A-E). Under mock infection conditions, 553 genes were significantly upregulated, while 71 genes were significantly downregulated, in Alzheimer’s neurons compared to neurons from iPSCs of healthy individuals (designated healthy neurons) (Fig. 7A). SARS-CoV-2 upregulated 75 genes and downregulated 19 genes in healthy neurons (Fig. 7B). To extract the genes responsible for Alzheimer’s infectious etiology, 24 overlapping genes were identified between the Alzheimer’s neuron-mock-infected group and the healthy neuron-SARS-CoV-2-infected group (Fig. 7D). Pathway analysis revealed that the changes in these 24 genes activated infection pathways elicited by bacteria and viruses (Fig. 7D).
Compared to healthy neurons without viral infection, Alzheimer’s neurons infected with SARS-CoV-2 had 517 upregulated genes and 256 downregulated genes (Fig. 7C). This number of downregulated genes (256) was higher than the number of downregulated genes in Alzheimer’s neurons (71) (Fig. 7A, C). In Alzheimer’s neurons, SARS-CoV-2 further increased the expression of 25 upregulated genes and decreased the expression of 34 downregulated genes by several-fold (Fig. S6A), indicating that the virus deteriorates Alzheimer’s conditions, and pathway analysis pointed to the further activation of neuroinflammation and other processes in Alzheimer’s neurons (Fig. S6B).
Top genes in the Alzheimer’s infectious etiology transform neurons
The 24 overlapping genes between Alzheimer’s neurons without SARS-CoV-2 infection and healthy neurons infected by SARS-CoV-2 are potential genes involved in the Alzheimer’s infectious etiology. To evaluate whether the top upregulated genes among these 24 genes can turn healthy neurons into Alzheimer’s-like neurons, we overexpressed the top three genes individually, FCGR3, LILRB5 and OTOR, in human primary neurons from a heathy individual. Overexpression of these genes in healthy neurons did not trigger cellular Aβ aggregation or cellular p-Tau deposition (Fig. S7D). In contrast, when the top three downregulated genes in the 24-gene list, GJA8, CryAA2 and PSG6, were individually silenced in healthy human neurons, cellular expression of Aβ42 and p-Tau, cellular Aβ aggregation and cellular p-Tau deposition were induced (Fig. 7F, Fig. S7E). Furthermore, silencing these top three downregulated genes simultaneously in healthy human neurons robustly triggered cellular Aβ aggregation and cellular p-Tau deposition (Fig. 7G). Thus, silencing the top three downregulated genes reprogrammed healthy human neurons into Alzheimer’s-like neurons.
Discussion
Although it is still controversial whether SARS-CoV-2 invades patients’ brains [10, 36], we provide strong evidence that SARS-CoV-2 can invade the cognitive centers of the brain, leading to Alzheimer’s- like phenotypes and exacerbation of Alzheimer’s neuropathology. Examination revealed that iPSC- derived neurons infected by SARS-CoV-2 recapitulated the key aspects related to clinically observed neurological disorders in COVID-19 patients and survivors: neurotropism and Alzheimer’s induction or enhancement by SARS-CoV-2.
Our observations from both COVID-19 autism and Alzheimer’s brains support the effective infection of neurons, oligodendrocytes and other brain cells by SARS-CoV-2. Thus, the present study demonstrates the neurotropism of SARS-CoV-2. Such neurotropism has long been known to occur for other types of human respiratory coronaviruses [37]. Consistent with the current findings, in human iPSC- derived brain sphere neurons, SARS-CoV-2 at a dose equivalent to that in the present study has been found to not only infect neurons but also replicate itself in these neurons [38]. SARS-CoV-2 viral particles are present in both neuronal cell bodies and neurites [38]. SARS-CoV-2 can effectively infect cortical-like neurons in iPSC-derived brain organoids and these neurons express the SARS-CoV-2 receptor ACE2 and key coronavirus entry-associated proteases [39]. Regarding the infectibility of human brains by SARS-CoV-2, viral RNA has been detected in 36.4% of brain biopsies of fatal COVID-19 cases [9], suggesting that SARS-CoV-2 invades some COVID-19 patients’ brains but may not invade all patients’ brains. The current study examined cortical neurons of the brains of COVID-19 patients with either autism or Alzheimer’s disease and observed full infectibility by SARS-CoV-2 in all cases. Similarly, a recent study using spike protein staining demonstrated that SARS-CoV-2 was present in some brain regions in all three COVID-19 cases studied [10]. Using multiple approaches, we showed the presence of spike and nucleocapsid proteins and SARS-CoV-2 viral particles in the cortical regions of all three COVID-19 cases with either autism or Alzheimer’s.
An early study showed no evidence of SARS-CoV-2 infection of neurons [40]. In iPSC-derived brain organoids, SARS-CoV-2 infects only mature choroid plexus cells, not neurons or glial cells [40]. This discrepancy may be due to the different developmental stages of neurons. We show that iPSC-derived immature neurons do not express ACE2 and NPR1 and thus are not able to be infected by SARS-CoV-2, whereas mature neurons do express these two SARS-CoV-2 receptors and are infected by this virus. Pellegrini et al. [40] did not observe ACE2 expression in neurons. In COVID-19 patients, not all neurons are infected by SARS-CoV-2 [10], suggesting that different types of neurons have differential expression of ACE2 and other SARS-CoV-2 entry factors and variable SARS-CoV-2 infectability. We and others [10] have shown that cortical neurons including excitatory and inhibitory neurons, are able to be infected by SARS-CoV-2.
SARS-CoV-2 enters hosts through their noses, mouths and eyes; thus, the anatomical proximity between the nasal cavity/nasopharynx and the olfactory mucosa may enable olfactory transmucosal entry of SARS-CoV-2 into the CNS [8]. The other major CNS entry route may depend on BBB leakage in conditions of autism and Alzheimer’s disease [12, 13]. This may have been the case in our study, in which three COVID-19 patients had either autism or Alzheimer’s disease. The BBB entry route is supported by a direct observation of BBB damage in COVID-19 patients [41]. Furthermore, the SARS-CoV-2 spike protein can disrupt human BBB integrity in 3D microfluidic in-vitro models [42]. The infiltration of infected immune cells into the brain has been proposed as a potential CNS entry route [14]. In agreement with a prior report, no sign of infected immune cell infiltration into COVID-19 autism patients’ brains was found in the present study. Immune cells were present in the cortex in the COVID-19 Alzheimer’s case. Thus, BBB leakage may be the major CNS entry route for SARS-CoV-2 into autism brains.
Apoptotic cell death is observed in the cells of brain organoids infected with SARS-CoV-2 [10, 39, 43, 44]. SARS-CoV-2-infected cells may not be very susceptible to apoptosis because the majority of cells undergoing apoptosis are noninfected but adjacent to SARS-CoV-2-positive cells [10]. The causes of apoptosis may be SARS-CoV-2-induced cellular inflammation and hypoxia, which occur in brain organoids infected by this virus [10, 43]. In COVID-19 patients, we found that apoptosis mostly occurred in cells infected by SARS-CoV-2. We found that cellular inflammation was induced in both iPSC-derived neurons and COVID-19 patient cortexes, suggesting that cellular inflammation is the primary factor leading to cell death. Other factors such as p-Tau, as detected by our group and others [44], may also play important roles in defining cell fates. SARS-CoV-2 infection induces not only apoptosis but also other nonapoptotic cell death programs, such as necroptosis and ferroptosis, as well as senescence [45], as observed in the current study. Thus, SARS-CoV-2 infection alters cellular programs, leading to critical cell mass loss and cellular dysfunction.
COVID-19 patients and survivors experience Alzheimer’s-like neural syndrome including memory loss, delirium and cognitive deficits [25], suggesting that there is a cellular mechanism underlying these phenomena. A prior study showed altered distribution of Tau from axons to soma and Tau hyperphosphorylation in neurons of brain organoids exposed to SARS-CoV-2 [44]. We observed p-Tau tangles in the cortexes of two young autism COVID-19 patients, suggesting that neurodegenerative characteristics manifest in COVID-19 patients’ brains. Aβ deposition and neuroinflammation were also present in the cognitive centers of these two young autism COVID-19 patients and in iPSC-derived neurons infected by SARS-CoV-2. These observations potentially explain the Alzheimer’s-like brain disease observed in individuals exposed to SARS-CoV-2.
The pathogenesis of Alzheimer’s disease contains an element of infectious disease pathogenesis. Pathogen infections can lead to the onset and progression of Alzheimer’s disease. Insertions of viral DNA genomes into spontaneous late-onset Alzheimer’s patient genomes have been determined [34]. Viral DNA insertions in the host genome correlate with the induction of critical Alzheimer’s mediators, such as enzymes involved in Aβ species production, aggregation and plaque formation [34]. Direct evidence for the involvement of viral DNA in Alzheimer’s pathogenesis comes from the demonstration that herpes simplex virus type I (HSV-1) induces multicellular amyloid plaque–like structure formation, gliosis, and neuroinflammation in iPSC-derived neural cells and a 3D human brain-like model [35]. We provide direct evidence that the COVID-19 pandemic-causing virus SARS-CoV-2 triggers an Alzheimer’s-like molecular program involving a group of 24 genes related to infectious disease pathway activation. The β- secretase BACE1, which produces all monomeric forms of amyloid-β (Aβ), including Aβ42, was induced by SARS-CoV-2 in iPSC-derived neurons from healthy individuals; thus, BACE1 may be a primary driver of the Alzheimer’s-like phenotypes induced by SARS-CoV-2. SARS-CoV-2-induced hypoxia [10] may be responsible for BACE1 induction because hypoxia facilitates Alzheimer’s pathogenesis by inducing BACE1 expression [46]. Because 26 of the 29 SARS-CoV-2 proteins physically associate with many proteins in human cells [47], it is also possible that one or some of the 29 SARS-CoV-2 proteins interact with transcriptional regulators in host cells, leading to BACE1 upregulation. The mechanism underlying SARS-CoV-2-induced Alzheimer’s-like phenotypes needs to be further investigated.
In summary, we found that SARS-CoV-2 neurotropism exhibited full penetrance in the cortexes of COVID-19 autism and Alzheimer’s patients. SARS-CoV-2 infection induced Alzheimer’s-like phenotypes in autism patients and exacerbated neuropathology in Alzheimer’s patients. SARS-CoV-2 infection triggered cellular and molecular Alzheimer’s pathogenesis programs in iPSC-derived neurons from healthy individuals and enhanced neuropathological phenotypes in iPSC-derived neurons from Alzheimer’s patients. We reveal a list of 24 genes that potentially mediate the infectious etiology of Alzheimer’s disease under the condition of SARS-CoV-2 infection.
Author Contributions
Shen WB, Penghua Yang, Montasir M, Xu C, Logue J, and Baracco L researched the data. Frieman M, Reece EA, Blanchard T, Li L, Han Z, and Rissman R analyzed the data and revised the manuscript. Peixin Yang conceived the project, designed the experiments, and wrote the manuscript. All authors approved the final version of the paper.
Declaration of Interests
The authors declare no competing interests.
Supplemental figure titles and legends
Methods
Mammalian iPSCs and culture conditions
Human iPSCs (1 cases of familial AD, 2 cases of sporadic AD, and 3 cases of apparently healthy controls) purchased from the Coriell Institute for Medical Research were used in our study. iPSCs were maintained on irradiated mouse CF-1 feeder layer (ATCC, Manassas, VA) at 37 °C and 5% CO2 in Knockout DMEM medium (Invitrogen, Waltham, MA) supplemented with 20% Knockout Serum Replacer (KSR) (Invitrogen, Waltham, MA), 0.1 mM nonessential amino acids (Invitrogen, Waltham, MA), 2 mM GlutaMAX (Invitrogen, Waltham, MA), 0.1 mM β-Mercaptoethanol (Sigma-Aldrich, St. Louis, MO), 10 ng/ml recombinant human basic fibroblast growth factor (Invitrogen, Waltham, MA) (hiPSC medium).
Viral strains
Vero E6 cells were maintained in EMEM (ATCC) media with 10% Serum Plus II Medium Supplement (Sigma-Aldrich). SARS-CoV-2 virus were obtained from the CDC following isolation from a patient in Washington State (WA-1 strain - BEI #NR-52281). SARS-CoV-2 GFP was generously provided by Dr. Ralph S. Baric. Stocks were prepared by infection of Vero E6 cells.
Brain Tissues
The major source for postmortem tissues is the Brain and Tissue Bank at University of Maryland (UMB). The Brain and Tissue Bank is a brain and tissue repository of the NIH NeuroBioBank. The Brain Tissue Bank is a national resource for investigators utilizing human post-mortem brain tissues and related biospecimens for research in understanding the conditions of the nervous system. All brain tissue is procured, stored, and distributed according to applicable State and Federal guidelines and regulations involving consent, protection of human subjects and donor anonymity. All brain tissues we obtained from the Brain Tissue Bank are de-identified.
Following formalin-fixed brain tissues were obtained from the Brain and Tissue Bank at UMB: COVID-19 autism (ASD, ID #6436, 38 year-old, #6437, 30 year), COVID-19 Alzheimer’s disease (AD, #6535, 77 year), age-matched non-COVID-19 controls (6 cases), non-COVID-19 ASD (6 cases), non-COVID-19 FTD (frontotemporal dementia, 4 cases), and apparently healthy subjects (9 cases). We also obtained 4 cases of non-COVID-19 AD from the NIH NeuroBioBank. Samples of three brain regions were obtained from individual: dorsolateral prefrontal cortex (Broca area 8), inferior frontal cortex (Broca area 44), entorhinal cortex.
Following frozen brain tissues were also obtained from the Brain and Tissue Bank. The frozen tissues were used for RNA extraction followed by reverse transcription and PCR detection of SARS-CoV-2 with CDC specific primers. The frozen tissues include SARS-CoV-2 infected lung and brain tissues (ASD, #6436), age-matched non-COVID-19 ASD controls and FTD controls.
In addition, paraffin sections and frozen brain tissues of the subjects of COVID-19 FTD (1 case) and COVID-19 individual without underlying condition, and non-COVID-19 AD (3 cases) were obtained from the Biomarker Laboratory and Biorepository at University of Southern California Alzheimer’s Therapeutic Research Institute (USC ATRI) at University of California San Diego (UCSD).
Neuronal Cells
Primary Human Neurons were obtained from Neuromics (CA3 Bioscience, MN, USA). The cells were maintained in neuron growth medium supplemented (CA3 Bioscience, MN, USA) with 10% FBS at 37 °C in a humidified atmosphere of 5% CO2 in culture flask.
Immunohistology staining of spike protein and markers of neural cells and programmed cell death
The formalin-fixed brain tissue blocks were immersed in 30% sucrose, then embed in OCT cryostat sectioning medium. The tissues were sectioned on a cryostat at 10 μm. After rehydrated, the sections were treated with 2% H2O2 for 15 minutes to quench the endogenous peroxidase. The staining was performed as previously described [1, 2]. Briefly, the slices were incubated with blocking solution consisting of 4% normal donkey serum (NDS), 0.2% triton (TX)-100 in phosphate-buffered saline (PBS) for 60 min, then with primary antibody diluted in blocking solution at 4 °C overnight. The slices were washed with PBS and incubated for 1 h in a biotinylated secondary antibody anti-mouse or rabbit (depending on the host species producing the primary antibodies). After three times of PBS washes, the sections were incubated in ABC solution (1:500; Vectastain Elite Kit) for 1h. The ABC solution was prepared and placed on ice for 30 minutes before use. Antibody labeling was then visualized via precipitation with a diaminobenzidine (DAB, 10mg/50ml) chromogen solution in PBS + 0.003% H2O2 (brown products) or with DAB in 0.175 M sodium acetate + 0.5% nickel ammonium sulfate + 0.003% H2O2 (for Ni-DAB staining, black products). If necessary, the sections were counterstained with hematoxylin before coverslip.
For immunohistology dual labeling, the sections were subject Ni-DAB staining with first primary antibody as described above. After PBS wash, the sections were re-blocked for 1 hours and following incubation with the second primary antibody at 4°C overnight. Next day, the sections were washed with PBS and incubated with alkaline phosphatase (AP)-conjugated goat secondary antibody (1:200) for 1 h, followed by incubation with AP red substrate (1:100) for 20-30 minutes.
DAB-stained sections were counterstained with hematoxylin. Sections were immersed in hematoxylin solution for 3 min and rinsed in running tap water until rinse water is colorless. After differentiation by dipping slides 10 times in acid rinse solution (1% hydrochloric acid (HCl) in 70% ethanol), the slides were incubated in the bluing solution (1.5% lithium carbonate) for 30 s. There is no hematoxylin counterstaining performed for Ni-DAB or Ni-DAB/AP dual labeling before dehydrate and coverslip.
Dehydrate and coverslip sequence: 70% ethanol, 3 minutes; 95% ethanol I and II, 3 minutes for each; 100% ethanol I and II, 3 minutes for each; xylene I and II, 5 minutes for each. Mounting medium: Permount Mounting Medium.
Aβ and p-Tau Immunohistochemistry
β-amyloid (Aβ) and phospho-tau (pTau) Immunohistochemistry was carried out as previously described [3]. Briefly, the coronal sections were rehydrated, and endogenous peroxidase was quenched by treating with 0.3% H2O2. For Aβ staining, the antigen retrieval was carried by using 70% formic acid for 20 minutes. Antigen retrieval for the other antibodies was by autoclaving at 120°C for 5 min. The primary antibody was applied on the sections at 4°C overnight, followed by 60-min incubation with specific secondary antibody coupled with HRP (Histofine simple stain MaxPo M/R, Nichirei Bioscience Inc., Japan) at room temperature. DAB reaction was performed to visualize the color. All antibody dilutions and washing steps were performed in phosphate buffer, pH 7.2. HRP intensities and cell counts in 3 different regions (Frontal Cortex, Entorhinal Cortex) in three sections/slices from each group were measured using the ImageJ Fiji platform. Data were represented as staining intensity per square millimeter. Because the MaxPo M/R antibodies are Fab fragments, the blocking is not necessary.
Thioflavin staining
The Aβ staining Thioflavin-T were performed following the similar antigen protocol. After blocking with 5% goat serum for 30 minutes overnight incubation with anti Aβ antibody 6E10 were performed. In the following day the brain sections were probed with Alexa-fluor594 labeled secondary antibody. After an extensive washing, Thioflavin-T (0.1% in 50% DMSO) were applied over the brain sections for 5 minutes. An extensive washing was performed using PBS containing 0.05% tween20. Furthermore, the fluorescent background was quenched with Tureblack (Biotium, USA). Thioflavin-T and Alexa-fluor594 intensity were measured using a fluorescence laser microscope (LSM780, Zeiss, Germany).
Immunofluorescence staining
For immunofluorescence staining, cells were cultured on a collagen-coated coverslip. After specific treatments, cells were fixed with 4% paraformaldehyde-PBS. The cell membrane was permeabilized with 0.25% Triton-X100 and/or 5 mg/ml digitonin followed by blocking with 5% BSA. After probing with primary antibodies, specific Alexa-fluor labeled secondary antibody were used. 4’,6-Diamidino-2-phenylindole dihydrochloride (DAPI) staining was used to visualize the nuclei (Thermo Fisher Scientific). Fluorescence was assessed using a fluorescence laser microscope (LSM780, Zeiss, Germany).
Detection of SARS-CoV-2 RNA using RNAscope
RNA In situ hybridization in FFPE slides was performed by using the RNAScope 2.5 HD Detection (RED) Kit (ACD, CA), based on the protocol provided by the manufacturer. Briefly, after deparaffinization and antigen retrieval, the slides were hybridized with the 40-ZZ positive-sense RNA probe, V-SARS-CoV-2-S (Ref#: 854841, ACD) in the oven at 40 °C for 2 hours. We then washed the slides with 1 X Wash Buffer for 2 min twice. The remaining hybridization procedure at 40 °C include: 1) Amp 1, 30 min; 2) Amp 2, 15 min; 3) Amp, 30 min; 4) Amp 4, 15 min; 5) Amp 5, 30 min, and 6) Amp 6, 15 min with an interval of repeated 1X Wash buffer for 2 min. Subsequently, the signals were detected by incubating with a mixture of RED-B and RED-A at ratio of 1:60 for 10 min at room temperature followed by counterstaining with 50% hematoxylin for 2 min.
Generation of human induced pluripotent stem cells (hiPSCs)
Human fibroblast from Coriell Institute were induced into pluripotent stem cell with CMV promoter Thomson factors lentivirus set which contains Lenti-virus harboring Oct4, Sox2, Nanog, and Lin28 (Cat#: G353, ABM Inc., Canada). Briefly, human fibroblasts were seeded on 6 well plate with 8 x 104 per well at day -2 and transduced with Lenti-virus particles (MOI of 4:3:3:3). Cells were cultured for 7 days and replated at 2 x 104 cells per well of 6 well plate on irradiated mouse embryonic fibroblasts (MEF) feeder layer. From day 8, medium was replaced with Knockout DMEM (Invitrogen) supplemented with 20% Knockout Serum Replacer (KSR, Invitrogen), 0.1 mM nonessential amino acids (Invitrogen), 2 mM GlutaMAX (Invitrogen), 0.1 mM β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO), 10 ng/ml recombinant human basic fibroblast growth factor (Invitrogen) (hiPSC medium). On week 3-4, iPSCs were identified from morphology change and picked out into 48-well plates with MEF feeder.
Neuron differentiation from hiPSCs
hiPSCs were differentiated into neurons as described with modification (Nat Med. 2018 May;24(5):647-657). hiPSCs dissociated with collagenase IV (Stem Cell Technologies) were cultured in suspension to form embryoid bodies (EB) in hES medium without bFGF for 5 days followed by maintenance in NIM medium containing Dulbecco’s modified Eagle’s medium/F12 and Neurobasal Medium (1:1, Thermo Fisher), 1% N2 Supplement (Life Technologies), 1% B27 Supplement (Life Technologies), nonessential amino acids, and 0.5% penicillin/streptomycin (Life Technologies) supplemented with inhibitors of the TGF-β receptor (SB431542, Stemgent; 5 µM) and the bone morphogenetic protein receptor (LDN-193189, Stemgent; 0.25 µM). On day 7, spheres were transferred to wells coated with Matrigel (BD Biosciences) and grown in NPM medium which is same to NIM medium but replace SB431542 and LDN-193189 with 10 ng/ml bFGF (PeproTech), 10 ng/mL epidermal growth factor (EGF) (PeproTech), and 2 µg/ml heparin (Sigma). On day 15, medium was changed to NDM medium which is same to NPM medium but replace bFGF, EGF and heparin in NPM with brain-derived neurotrophic factor (10 ng/ml; PeproTech), and glial cell–derived growth factor (10 ng/ml; PeproTech). Around day 20, neurons were observed and were further differentiated for 30 days.
Immunofluorescence staining in iPSCs-derived neurons
Neurons derived from AD or Control (from healthy individuals) hiPSCs with or without SARS-CoV-2 infection were fixed in 4% paraformaldehyde (PFA) for 10 minutes followed by blocking in 5% bovine serum albumin in PBST (0.1% Triton X-100 in PBS) for 10 minutes. The following antibodies were used as primary antibodies: Spike (1:200), Tuj1 (1:500), ACE2 (1:200), NRP1 (1:200), NeuN (1:200), GFAP (1:200), c-Cas3 (1:200). Normal rabbit or mouse IgG using the same dilutions as primary antibodies were used as controls. After washing with PBS, neurons were incubated with secondary antibodies. Then, neurons were counterstained with DAPI and mounted with aqueous mounting medium (Sigma, St Louis, MO). Images were captured under a microscope (Keyence BZ X700, Osaka, Japan).
SARS-CoV-2 production and infection of hiPSCs derived neurons
The stock of SARS-CoV-2 virus (CDC, WA-1 strain - BEI #NR-52281, and SARS-CoV-2 GFP generously provided by Dr. Ralph S. Baric, at the Department of Epidemiology and Department of Microbiology and Immunology, University of North Carolina at Chapel Hill), were prepared by infection of Vero E6 cells for two days when CPE (cytopathic effects) was starting to become visible. Media were collected and clarified by centrifugation prior to being aliquoted for storage at −80 °C. Neurons derived from hiPSCs were infected with SARS-CoV-2 at 5 x 103, 1 x 104, or 2 x 104 plaque forming units (pfu) per well of 6 well plate for 48 h in NDM medium. All work with infectious virus was performed in a Biosafety Level 3 laboratory and approved by our Institutional Biosafety Committee.
SARS-CoV-2 Titering by semi-solid plaque assay
VeroE6 cells were plated in 12 well plates with 2×105 cells per well one day prior to processing. On the day of processing, samples were serially diluted 1:10 and 200uL of each sample dilution was added to each well in singlet and incubated for 1 hour at 37°C (5% CO2) with rocking every 15 minutes. Following incubation, 2 mL of a semi-solid agar overlay, DMEM (gibco) containing 4% fetal bovine sera (gibco) and 2% agarose, was added to each well. Plates are incubated for 3 days at 37°C (5% CO2) before plates were fixed with 4% paraformaldehyde, stained with crystal violet stain, and plaques counted.
Neural maturation, siRNA Transfection, and Lenti virus transduction
Primary Human Neurons obtained from Neuromics (Edina, MN) grow for four weeks in the Neuromics growth medium supplemented with brain derived neurotropic factor (BDNF). Medium were changed every 2 days. For further experiment, cells were plated on a 6-well plate in neuron growth medium supplemented with 5% FBS. siRNA transfection was carried out using X-tremeGENE siRNA transfection reagent (Roche, Basel, Switzerland) at 60% cell confluency. Medium was replaced with fresh medium 24 hours post-transfection and maintained for additional 48 hours. Lentivirus mediated overexpression of the FCGR, LILRB5 and OTOR carried out using virus particles obtained from OriGene (Rockville, MD). Virus transduction were carried out at 5.0 MOI using polybrene. GFP expressing lenti-particles were used as the control.
RNA sequencing
mRNAs were extracted from neurons derived from AD or Control hiPSCs with or without SARS-CoV-2 infection and sequenced at BGI on DNBSeq platform. Briefly, mRNAs were extracted with Trizol reagent. First-strand cDNA was generated using random hexamer-primed reverse transcription, followed by a second-strand cDNA synthesis. The synthesized cDNA was subjected to end-repair and then was 3’ adenylated. Adapters were ligated to the ends of these 3’ adenylated cDNA fragments, followed by PCR. PCR products were purified with Ampure XP Beads and dissolved in EB solution. Library was validated on the Agilent Technologies 2100 bioanalyzer. The double stranded PCR products were heat denatured and circularized by the splint oligo sequence. The single strand circle DNA were formatted as the final library. The library was amplified with phi29 to make DNA nanoball (DNB) which had more than 300 copies of one molecular. The DNBs were load into the patterned nanoarray and single end 50 (pair end 100) bases reads were generated in the way of sequenced by synthesis.
Quantitative reverse transcription PCR
mRNAs were extracted from neurons derived from AD or Control hiPSCs with or without SARS-CoV-2 infection in Trizol reagent (Thermofisher, Waltham, MA) and followed by cDNA synthesis with SuperScript™ III Reverse Transcriptase kit (ThermoFisher). RT-PCR were performed with PowerUp™ SYBR™ Green Master Mix (ThermoFisher) with 45 cycles and PCR products were loaded into 2% agarose gel and run at 75 v for 5.5 h. Gel Image were recorded with Bio-Rad ChemiDoc MP imaging system (Hercules, CA).
QUANTIFICATION AND STATISTICAL ANALYSIS
In the experiment assessing the impact of SARS-CoV-2 virus on the brain pathology, we analyzed (a) 5 Covid19 samples consisting of AD-Covid19 (1 case), FTD-Covid19 (1 case), ASD-Covid19 (2 cases), and Covid-19 without underlying condition (1 case); (b) 6 cases of sporadic AD as the controls; and (c) 8 cases of age-matched, non-Covid-19 healthy controls. Labeled cells from at least 3 random fields (n ≥ 3) were counted. Data are presented as the means ± standard errors (SEs). Student’s t test was used for two group comparisons. One-way ANOVA was performed for comparisons of more than two group using graphPad prism software v7. In ANOVA, a Tukey test was used to estimate the significance between groups. Differences were considered statistically significant when P < 0.05.
In the experiment with cell culture studies, we differentiated neurons from 3 iPSC lines derived from AD fibroblast cells and 3 lines from healthy subjects as the controls. The experiments were repeated in triplicate as we previously described [4, 5]. Data are presented as the means ± standard errors (SEs). Student’s t test was used for two group comparisons. One-way ANOVA plus Tukey test was performed for comparisons of more than two group. Differences were considered statistically significant when P < 0.05.
Resource Tables
Acknowledgments
This work was supported by NIH grants R01HD100195, R01HD102206, R01HD099843, R01DK083243, R01DK101972, R01HL131737, R01HL134368, R01HL139060, R01DK103024 and P30-AG062429. No potential conflicts of interest relevant to this article are reported. The authors thank Jeffrey Metcalf, Sara Shuldberg and Sophia Perrott from the Rissman lab and UCSD Shiley-Marcos ADRC for technical assistance with this project.
Footnotes
Conflict of interest statement: The authors have declared that no conflict of interest exists
References
Reference
- [1].
- [2].
- [3].
- [4].
- [5].