|
|
Post by Admin on Nov 24, 2020 20:30:14 GMT
Validation of Enriched Genes Using CRISPR Knockout, RNA Interference, and Small Molecule Inhibitors To test the ability of top-ranked genes to block SARS-CoV-2 viral infection, we picked 30 genes ranked among the top 200 genes in our RRA analysis for independent validation. Each gene was targeted with 3 guide RNAs distinct from the guides present in the GeCKOv2 genome-wide CRISPR library (Table S3). The guides were synthesized individually and cloned into the pCC1 lentiviral vector, a modified version of lentiCRISPRv2 with the F+E optimized guide RNA scaffold (Legut et al., 2020). For a subset of genes, we validated Cas9-mediated loss-of-function by western blot for protein expression (Figure S3A).  Figure S3 Validation of Top-Ranked Genes Using CRISPR Perturbations and RNA Interference, Related to Figure 4 Show full caption Cas9-perturbed A549ACE2 lines were infected with SARS-CoV-2 at an MOI of 0.1 and the percentage of infected cells was determined by immunofluorescence against the SARS-CoV-2 N protein at 36 h post-infection (hpi). For all of the Cas9-perturbed cell lines, we observed a reduced percentage of infected cells with an up to 10-fold reduction in SARS-CoV-2 infection, compared to the cell lines with non-targeting sgRNAs (Figures 4A and 4B ). Among the genes where loss provided the greatest protection against SARS-CoV-2 infection were vesicular trafficking genes like RAB7A, CCDC22, and VPS35, and other genes with well-established roles such as the ACE2 receptor and the protease cathepsin L (CTSL) (Figure 4B). We found a significant negative correlation between the percent infection in the arrayed CRISPR validation and the median fold-change from the genome-wide CRISPR screen (rs = −0.6, p = 5 × 10−4) (Figure 4C).  Figure 4 Arrayed Validation of Genome-Scale SARS-CoV-2 Screen and Identification of Druggable Gene Targets Show full caption For these genes, we performed a full multi-step viral replication growth curve and found that the CRISPR perturbations decrease viral load across all time points (5, 10, 24, and 48 h) compared to the non-targeting control (Figure S3B). To test if the identified hits block viral infection in other cell lines, for 8 genes we generated polyclonal CRISPR knockouts in a human liver cell line (Huh7.5ACE2), which were then infected with SARS-CoV-2 at an MOI of 0.1. Compared to non-targeting guide RNA controls, we found reduced infection with all 8 genes tested in the Huh7.5ACE2 cells (Figure S3C). As an orthogonal gene perturbation method, we also validated a more extensive list of top-ranked genes via siRNA knockdown for 48 h, followed by infection with SARS-CoV-2 at MOI of 0.1 (Table S3). Quantification of N-protein immunofluorescent images revealed a substantial reduction of the percent infected cells (Figure S3D). By cross-referencing highly ranked genes from the CRISPR screen with the Drug Gene Interaction database (DGIdb), we identified a set of 69 druggable genes (Figure 4D; Table S4; Cotto et al., 2018). We selected 9 genes that were a primary or a secondary target of one of the 26 small molecule inhibitors (Figure 4E). Among the 26 inhibitors, 9 are FDA approved and 7 are in Phase 2 or Phase 3 clinical trials for diverse diseases (Table S4). We pre-treated A549ACE2 cells for 2 h with 10 μM of each inhibitor and then infected with SARS-CoV-2 and analyzed the cells at 36 hpi. As a positive control, we also included remdesivir, which inhibits the viral RNA polymerase and is the only currently approved treatment for COVID-19 in the United States (Beigel et al., 2020). We assessed the efficacy of each inhibitor on blocking viral infection using immunofluorescence and quantitative PCR (qPCR). Seven of the 26 inhibitors that we tested (PIK-III, Compound-19, SAR405, autophinib, ALLN, tamoxifen, and ilomastat) resulted in >100-fold reduction of viral load as measured by qPCR (Figure 4E). Among the best performing inhibitors, 4 of them target the same gene PIK3C3 (also known as VPS34) and 2 inhibitors (autophinib and ALLN) reduced the viral load more than 1000-fold. Similar results were obtained by immunofluorescent imaging of SARS-CoV-2 N protein (Figure S4A). By testing the top four PIK3C3 inhibitors in combination with CRISPR targeting of PIK3C3, we found that Compound-19, PIK-III and autophinib were specific while SAR405 resulted in greater viral inhibition in PIK3C3 CRISPR-perturbed cells, suggesting some potential off-target activity for SAR405 (Figure S4B).  Figure S4 Perturbations of Enriched CRISPR Screen Genes with Small-Molecule Inhibitors, Related to Figure 4 Combinations of some of the top performing inhibitors overall showed an additive effect and further protected the cells from SARS-CoV-2 infection (Figure S4C). Next we measured cell viability at 36 h post-inhibitor treatment using flow cytometry. We observed more than 50% reduction of A549ACE2 cell viability with two pan-HDAC inhibitors, panobinostat and pracinostat (Figure 4F). |
|
|
|
Post by Admin on Nov 25, 2020 5:50:13 GMT
Single-Cell Sequencing Identifies Cholesterol Biosynthesis as a Common Mechanism Underlying Multiple Enriched Genes from the CRISPR Screen Next, we sought to understand the mechanisms underlying how individual genes identified in our loss-of-function screen prevent SARS-CoV-2 infection and if host gene loss alters cell transcriptional programs. For this, we utilized the expanded CRISPR-compatible cellular indexing of transcriptomes and epitopes by sequencing (ECCITE-seq) method to couple pooled CRISPR perturbations of our top hit genes with a single-cell transcriptomic and proteomic readout (Mimitou et al., 2019; Figure 5A; Table S5). ECCITE-seq is a high-throughput approach to identify the molecular mechanisms and cellular pathways that drive infection resistance; the pooled format also provides a more controlled experiment that may be less susceptible to batch variation. Importantly, cells were infected at low MOI to maximize the fraction of cells that express a single guide RNA, and therefore can be assigned a specific gene perturbation. For this, we pooled all individual guide RNA plasmids (3 per target gene) used to validate our genome-scale screen and 9 non-targeting (NT) sgRNAs (Figure 4B; Table S3).  Figure 5 Single-Cell Transcriptomics (ECCITE-Seq) Identifies Shared Target Gene Signatures for Lipid and Cholesterol Regulation In an initial ECCITE-seq experiment, we identified a median of 152 single cells per target gene. We observed specific reduction of target gene expression in cells grouped by target gene, indicating nonsense-mediated decay of transcripts with frameshift indel mutations after CRISPR modification (Figure S5A). This effect was more pronounced for genes with higher expression. Using differential gene expression analysis between cells with non-targeting guide RNAs and cells with targeting guide RNAs, we identified for 11 of the 30 target genes more than 5 differentially expressed genes with a minimal log fold change (see STAR Methods), implying that loss of these genes results in a detectable transcriptomic shift. It is likely that loss of the other 19 genes results in more subtle changes in only a few genes; however, even for “non-perturbed” genes with sufficient basal expression, we detected clear changes in the CRISPR target gene. We repeated the ECCITE-seq experiment focusing on the 11 genes with detectable transcriptomic shifts upon target gene perturbation. Combining both replicate experiments, we obtained 18,853 cells that expressed only one guide RNA with a median of 1,388 cells per target gene.  Figure S 5ECCITE-Seq Identifies Cholesterol Gene Signature Shared across Multiple Top-Ranked Genes, Related to Figure 5 We found that loss of 6 of the “perturbed” genes—ATP6AP1, ATP6V1A, CCDC22, NPC1, PIK3C3, and RAB7A, which are part of the endosomal entry pathway—yielded similar gene expression signatures among upregulated differentially expressed genes (Figures 5B, 5C, and S5B). These 6 target gene perturbations all led to upregulation of pathways affecting lipid and cholesterol homeostasis (Figure 5C). Recently, we performed a large survey of >20,000 potential pharmacological treatments for COVID-19, and for compounds effective at preventing viral infection, we identified induction of the cholesterol biosynthesis pathway as a potential mechanism of viral inhibition (Hoagland et al., 2020). Loss-of-function mutations in these 6 genes may function through a similar mechanism (induction of cholesterol synthesis) that combats the virus-mediated suppression of cholesterol synthesis. Among the significant differentially expressed genes, we also found 61 genes from the enriched CRISPR screen genes (n = 20 genes upregulated; n = 41 genes downregulated) (Table S5). For example, NPC1, ATP6V1F, and ATP6V1G1 are upregulated in most of the 6 endosomal entry pathway gene-perturbed cells (Figure 5C), suggesting compensatory upregulation of related genes to mitigate target gene loss. To understand how these changes impact lipid production, we measured cholesterol levels in cells after CRISPR perturbation and found that loss of these genes increases cholesterol by between 10% and 50%, depending on the perturbation (Figure 5D). To show that increases in cholesterol leads to increased SARS-CoV-2 resistance, we treated A549ACE2 cells with amlodipine, a calcium-channel antagonist that increases intracellular cholesterol (Mori et al., 1988; Ranganathan et al., 1982). We verified that amlodipine increases cholesterol levels in A549ACE2 cells (Figure S5C) and found that pre-treatment with amlodipine results in reduced SARS-CoV-2 viral infection, as measured by qPCR for nucleocapsid RNA, plaque formation, and number of viral RNA reads from RNA-sequencing, with only a modest impact on cell viability (Figures S5D–S5G). RNA-sequencing of cells treated with amlodipine shows a similar differential gene expression profile as seen in our ECCITE-seq with CRISPR perturbations with the most significant upregulated pathway as cholesterol biosynthesis (Figures S5H and S5I). |
|
|
|
Post by Admin on Nov 25, 2020 19:31:56 GMT
RAB7A Knockout Results in a Reduced Cell Surface Expression and an Increased Endosomal Accumulation of ACE2 Next, we sought to determine if any of the top-ranked genes regulate cell surface expression of ACE2, because surface ACE2 is required for SARS-CoV-2 infection (Hoffmann et al., 2020a). To measure the cell surface expression of ACE2, we performed flow cytometry on A549 wild-type cells and a panel of A549ACE2 CRISPR-perturbed cells. ACE2 expression was detected in A549ACE2, but not in A549 wild-type cells, validating the antibody specificity (Figures 6A and S6). Flow cytometry analysis of the A549ACE2 CRISPR-perturbed lines revealed that RAB7A knockout cells have a significantly reduced cell surface expression of ACE2 compared to cells transduced with a non-targeting guide RNA (Figure 6B). Efficient Rab7a protein depletion across the RAB7A CRISPR-perturbed lines used was confirmed by western blot (Figure S6C).  Figure 6 RAB7A Loss Results in a Reduced Cell Surface Expression and an Increased Endosomal Accumulation of ACE2  Figure S6 Flow Cytometry for Cell Surface ACE2 Expression and Protein Analysis of RAB7A Protein after CRISPR Targeting, Related to Figure 6 Rab7a is a small GTPase that is involved in regulating cellular processes such as vesicular transport and membrane trafficking (Guerra and Bucci, 2016). We hypothesized that the observed reduction of ACE2 at the cell surface in RAB7A knockout cells may be due to impaired vesicular trafficking and accumulation of ACE2 in the cytoplasm. Immunofluorescence in the A549ACE2 polyclonal cell line revealed that ACE2 is detected in most cells. We also found that RAB7A knockout leads to an increased accumulation of ACE2 in the cytoplasm and in vesicle-like hollow structures reminiscent of endo-lysosomes (Figure 6C). We detected these accumulations in about 35% of RAB7A knockout cells with ACE2 staining, compared to the control where ACE2 was primarily localized at the plasma membrane and small cytoplasmic puncta (Figure 6D). Finally, we investigated which cellular compartments accumulate ACE2 in RAB7A knockout cells. Co-immunofluorescence images showed that in RAB7A knockout cells ACE2-containing vesicles often colocalize with EEA1, an early endosomal marker and less frequently with LysoTracker, a lysosome marker (Figure 6E). Because ACE2 was overexpressed in our A549ACE2 cells, we wondered whether Rab7a loss would lead to similar cytoplasmic sequestration in cells with endogenous ACE2 expression. Flow cytometry analysis in Caco-2 colon and Calu-3 lung cells revealed that RAB7A knockout cells have significantly reduced cell surface expression of ACE2 compared to cells transduced with a non-targeting guide RNA (Figures 6F and 6G). We also found that RAB7A knockout in Caco-2 cells results in larger cytoplasmic ACE2 puncta compared to the control cells, further supporting cytoplasmic accumulation of ACE2 in the absence of Rab7a (Figures 6H and 6I). |
|
|
|
Post by Admin on Nov 26, 2020 4:26:07 GMT
Discussion
Given the current COVID-19 global pandemic, there is an urgent need to better understand the complex relationships between host and virus genetic dependencies. We report a genome-wide loss-of-function screen in human lung cells that identified host genes required for SARS-CoV-2 viral infection. We selected and validated 30 genes that were ranked among the top 200 genes. To support the ability of the screen to identify key dependencies, some of the well-known host genes involved in SARS-CoV-2 Spike protein binding and entry such as the ACE2 receptor and Cathepsin L were among the top-scoring genes (Hoffmann et al., 2020a). One of the validated genes (SIGMAR) encodes the Sigma-1 receptor that was recently identified to be modulated by drugs effective against SARS-CoV-2 in vitro (Gordon et al., 2020). Overall, the top-ranked genes clustered within several protein complexes including vacuolar ATPases, Retromer and endosome, Commander, ARP2/3, PI3K, and others, highlighting both the critical importance of multiple genes within each pathway to viral pathogenesis and the diversity of molecular pathways involved in SARS-CoV-2 infection.
Using a “minipool” CRISPR library of perturbations targeting top-ranked genes from the genome-scale CRISPR screen and single-cell transcriptomics, we identified a group of 6 genes (RAB7A, PIK3C3, NPC1, CCDC22, ATP6V1A, and ATP6AP1) that had a similar transcriptional signature—upregulation of the cholesterol synthesis pathway. By measuring the cholesterol levels, we found that CRISPR-driven loss of those 6 genes result in increased cellular cholesterol. Some of the 6 genes have previously been implicated in regulating low-density lipoprotein (LDL) cholesterol. For example, depletion of Rab7a leads to LDL accumulation in endosomes and NPC1 knockout cells show a reduction of cholesterol at the plasma membrane and an accumulation in the late endosome/lysosome compartments (Chang et al., 2005; Girard et al., 2014; Millard et al., 2000; Neufeld et al., 1996). We have recently, in an independent study, identified that SARS-CoV-2 infection negatively downregulates the cholesterol synthesis pathway, and viral infection can be counteracted by drug treatments that upregulate the same pathway (Hoagland et al., 2020). It is possible that changes in lipid composition directly impacts SARS-CoV-2 virion maturation and infectivity, as has been previously shown for hepatitis C and influenza A (Aizaki et al., 2008; Bajimaya et al., 2017). In this study, we showed that amlodipine, a calcium-channel antagonist, upregulates cholesterol levels and blocks SARS-CoV-2 infection. In addition, recent clinical studies have suggested that patients taking amlodipine or similar dihydropyridine calcium channel inhibitors have a reduced COVID-19 case fatality rate (Solaimanzadeh, 2020; Zhang et al., 2020). An important future research direction will be to further understand the relationship between cholesterol synthesis pathways and SARS-CoV-2.
Furthermore, we screened a panel of the top-ranked genes and identified that Rab7a regulates cell surface expression of ACE2, likely by sequestering ACE2 in endosomal vesicles. Rab7a is involved in vesicular trafficking and its depletion has been shown to sequester other cell receptors in endosomes (Rush and Ceresa, 2013). Interestingly, RAB7A knockout cell lines showed both altered cholesterol biosynthesis and sequestration of ACE2 receptor. Previous proteomics work showed that Rab7a has a strong interaction with viral protein nsp7 (Gordon et al., 2020). However, there is no nsp7 in the incoming virion, implying a post-entry/post-translational role for Rab7a. Thus, it is possible that loss of Rab7a blocks SARS-CoV-2 pathogenesis via multiple separate pathways, which is supported by the observation that it is the top-performing gene in our arrayed validation.
While this study was under review, a few other groups released preprints with loss-of-function CRISPR screens to identify host factors required for SARS-CoV-2 infection (Heaton et al., 2020; Wei et al., 2020; Zhu et al., 2020). Notably, only two studies (ours and Zhu et al. [2020]) have substantial overlap (11 and 14 genes among the top 36 genes from our MOI 0.3 and 0.01 screens, respectively, as shown in Figure S1D and nearly all of the top-ranked gene categories shown in Figure 2A). Given that these independent screens utilized different CRISPR libraries, the corroboration by Zhu et al. (2020) provides further support for our conclusions. The overlap between either of these studies with another screen performed using African green monkey cells (Wei et al., 2020) was limited to ACE2 and CTSL, two genes with established roles in viral entry. The differences in the overlap among the top-ranked genes might be due to technical aspects (such as different CRISPR libraries or variations in guide representation) or biological differences (such as different cell types or different host species). An exciting avenue for future study would be to investigate if SARS-CoV-2 perhaps utilizes multiple cell-type-specific genetic circuits.
A key element in our study was harnessing genome-scale loss-of-function to develop more refined therapeutic hypothesis. Our study suggests that PIK3C3 is a promising drug target: four out of the seven PIK3C3 inhibitors resulted in more than 100-fold reduction of SARS-CoV-2 viral load (SAR405, Compound-19, PIK-III, and Autophinib). Using a PIK3C3 polyclonal knockout A549ACE2 cell line, we found that among the top four PIK3C3 inhibitors, SAR405 may have some off-target effects (Figure S4B). Considering that our polyclonal knockout line likely has some residual PIK3C3, future work will be required to test the inhibitor specificity using a PIK3C3 monoclonal knockout cell line. Another drug that shows a substantial reduction in SARS-CoV-2 viral load is tamoxifen. Tamoxifen is an FDA-approved drug given as prophylaxis to patients at risk of breast cancer and works via modulation of the estrogen receptor. Tamoxifen was included in our study as it targets protein kinase C as a secondary target (O’Brian et al., 1985). This mechanism is further supported by the observation that A549 cells have undetectable transcript levels of estrogen receptor 1 (Human Protein Atlas) (Uhlen et al., 2010). Considering that tamoxifen is typically given to patients for years as a cancer therapy and prophylactic (Marchant, 1976), it would be interesting to investigate if patients taking tamoxifen have a reduced risk of SARS-CoV-2 infection and/or display less severe symptoms post-infection.
Finally, many approaches for therapeutic discovery have focused on large-scale screens of compound libraries. Even when promising therapeutic candidates are identified, it can be challenging to understand the mechanisms responsible for reducing viral pathogenesis. Our forward-genetics approach allows us to first identify key host genes, which can then be targeted through a diversity of methods such as small-molecule inhibitors, blocking antibodies, or gene knockdown. A key advantage of this approach is that the mechanism of action for any therapeutic is well-established from the outset.
Taken together, our integrative study identifies essential host genes in SARS-CoV-2 viral pathogenesis and, through a broad range of analytic and experimental approaches, validates their central role in infection. We also identify potential mechanisms underlying top-ranked genes, including cholesterol synthesis and endosomal function. In addition to guiding new therapeutic targets to help end this pandemic, our study provides a framework for harnessing massively parallel genome editing to understand disease genetics and mechanisms.
Limitations of Study
Our study identified host factors required for SARS-CoV-2 infection in human A549 cells that overexpress ACE2. Future work will be needed to explore limitations of our study. (1) Considering the ACE2 overexpression in our screen, it will be interesting to screen human cells expressing endogenous ACE2, which may potentially identify transcriptional regulators of ACE2. (2) The A549 cell line used in our CRISPR screen is a lung adenocarcinoma cell line. Given that various organs are affected by SARS-CoV-2, it will be helpful to understand whether there are tissue-specific host factors. (3) Although we show through multiple, distinct genetic perturbations that upregulation of the cholesterol biosynthesis pathway and increase in cellular cholesterol blocks SARS-CoV-2, the precise mechanisms of how changes in cholesterol disrupt viral infection remain to be elucidated. (4) Recent genome-wide association studies have uncovered human genetic variants associated with COVID-19 risk and severity. Since the majority of such variants are in noncoding regions, integrative analysis of genome-wide CRISPR screens may help pinpoint the causal genes through which these variants function.
|
|
|
|
Post by Admin on Jan 12, 2021 20:50:16 GMT
A shadow over the promising inhaled interferon beta COVID-19 therapy has been cleared with the discovery that although it appears to increase levels of ACE2 protein - coronavirus' key entry point into nose and lung cells - it predominantly increases levels of a short version of that protein, which the virus cannot bind to.
The virus that causes COVID-19, known as SARS-CoV-2, enters nose and lung cells through binding of its spike protein to the cell surface protein angiotensin-converting enzyme 2 (ACE2).
Now a new, short, form of ACE2 has been identified by Professor Jane Lucas, Professor Donna Davies, Dr. Gabrielle Wheway, and Dr. Vito Mennella at the University of Southampton and University Hospital Southampton NHS Foundation Trust.
The study, published in Nature Genetics, shows that as well as the long-form of ACE2 used by SARS-CoV-2, a shorter form of ACE2 exists that lacks the SARS-CoV-2 binding site.
Supporting interferon treatment
Naturally occurring antiviral proteins called interferons have shown promise in treating COVID-19. However, previous studies have shown that interferons increase levels of ACE2 - casting doubts over the potential for such treatments, with the possibility that increased ACE2 could see these drugs actually worsen COVID-19 impacts.
But this latest research shows that it is predominantly the short ACE2, which lacks the viral binding site, that is increased in response to interferons. Since levels of the longer form of ACE2 remain unchanged, interferons do not appear to boost entry points for the virus, supporting their use in treating COVID-19 patients.
This helps to explain the hugely promising results for a trial of an inhaled interferon-beta treatment for COVID-19 patients, developed in Southampton by a team led by Professor Tom Wilkinson of the University of Southampton.
This research gives a new insight into this short form of ACE2 and shows how it has a very different role to the longer form of ACE2 that acts as an entry point for SARS-CoV-2.
Short ACE2 lacks the binding site for SARS-CoV-2, so it cannot be used as an entry point for the virus. Instead, its regulation by interferons suggests it may be involved in the body's antiviral response.
As the researchers showed that short ACE2 does not increase in response to SARS-CoV-2 infection, it is unlikely to be involved in the body's immune response to COVID-19, but it did increase in response to another common respiratory virus.
These results will enable researchers to distinguish between these two forms of ACE2, knowledge which could prove invaluable for developing more sophisticated treatments for COVID-19 patients.
We were excited to discover a new form of ACE2 and became even more interested when we realised that may be protective against SARS-CoV-2 in the airways rather than an entry site for infection. We believe this may have important implications for managing COVID-19 infection and we are starting further studies to investigate this further."
Jane Lucas, Professor and Study Lead Author, Professor of Paediatric Respiratory Medicine, University of Southampton and Honorary Consultant Paediatric Respiratory Medicine, University Hospital Southampton
Source:
University of Southampton
Journal reference:
Blume, C., et al. (2021) A novel ACE2 isoform is expressed in human respiratory epithelia and is upregulated in response to interferons and RNA respiratory virus infection. Nature Genetics. doi.org/10.1038/s41588-020-00759-x.
|
|
