(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



A genome-wide CRISPR screen identifies interactors of the autophagy pathway as conserved coronavirus targets

['Annika Kratzel', 'Institute Of Virology', 'Immunology', 'Bern', 'Mittelhäusern', 'Department Of Infectious Diseases', 'Pathobiology', 'Vetsuisse Faculty', 'University Of Bern', 'Graduate School For Biomedical Science']

Date: 2022-01 

Over the past 20 years, 3 highly pathogenic human coronaviruses (HCoVs) have emerged—Severe Acute Respiratory Syndrome Coronavirus (SARS-CoV), Middle East Respiratory Syndrome Coronavirus (MERS-CoV), and, most recently, Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2)—demonstrating that coronaviruses (CoVs) pose a serious threat to human health and highlighting the importance of developing effective therapies against them. Similar to other viruses, CoVs are dependent on host factors for their survival and replication. We hypothesized that evolutionarily distinct CoVs may exploit similar host factors and pathways to support their replication cycles. Herein, we conducted 2 independent genome-wide CRISPR/Cas-9 knockout (KO) screens to identify MERS-CoV and HCoV-229E host dependency factors (HDFs) required for HCoV replication in the human Huh7 cell line. Top scoring genes were further validated and assessed in the context of MERS-CoV and HCoV-229E infection as well as SARS-CoV and SARS-CoV-2 infection. Strikingly, we found that several autophagy-related genes, including TMEM41B, MINAR1, and the immunophilin FKBP8, were common host factors required for pan-CoV replication. Importantly, inhibition of the immunophilin protein family with the compounds cyclosporine A, and the nonimmunosuppressive derivative alisporivir, resulted in dose-dependent inhibition of CoV replication in primary human nasal epithelial cell cultures, which recapitulate the natural site of virus replication. Overall, we identified host factors that are crucial for CoV replication and demonstrated that these factors constitute potential targets for therapeutic intervention by clinically approved drugs.

To identify key HDFs essential for CoV infection, we performed 2 independent genome-wide loss-of-function CRISPR screens with MERS-CoV, a highly pathogenic CoV, and HCoV-229E, an endemic CoV that causes mild respiratory symptoms in humans. We sought to uncover HDFs required for infection by a wide range of CoVs, including highly pathogenic CoVs with pandemic potential. Our results revealed that a number of autophagy-related genes, including FK506 binding protein 8 (FKBP8), transmembrane protein 41B (TMEM41B), vacuole membrane protein 1 (VMP1), and Membrane Integral NOTCH2 Associated Receptor 1 (MINAR1), were among the top hits in both CoV screens, suggesting that host factors involved in autophagy may also be required for CoV replication. Importantly, we found that perturbation of FKBP8 and other members of the immunophilin family by clinically approved and well-tolerated drugs, but not inhibition of late cellular autophagy, inhibited CoV infection in a dose-dependent manner. Overall, the genes and pathways identified in our CoV screens expand the current repertoire of essential HDFs known to be required for CoV replication and can be exploited to identify novel therapeutic targets for host-directed therapies against both existing and future emerging CoVs.

The lack of specific pharmaceutical intervention options and/or prevention measures against HCoVs, as well as ongoing difficulties containing the rapid global spread of SARS-CoV-2, has intensified in the current pandemic, and new therapies are urgently needed. CoVs are obligate intracellular pathogens and thus rely on selected host proteins, termed host dependency factors (HDFs), to achieve virus entry, replication, and release. The identification of HDFs is therefore crucial for understanding essential host–virus interactions required for successful viral replication and can provide a framework to guide the development of new pharmacological strategies for the treatment of CoV infection, including for COVID-19 and future emerging CoVs. CoVs encode a spike surface glycoprotein, which enables specific binding to a cellular receptor to mediate viral entry into the host cell. Known host receptors include dipeptidyl peptidase 4 (DPP4) for MERS-CoV, human aminopeptidase N (ANPEP/APN) for HCoV-229E, and angiotensin converting enzyme 2 (ACE2) for SARS-CoV and SARS-CoV-2 [ 9 – 12 ]. Cleavage of the spike protein by host cell proteases, such as TMPRSS2, cathepsin L, and/or furin, facilitates membrane fusion followed by release of the viral genome into the cellular cytoplasm for replication [ 13 ]. One hallmark that occurs in host cells during replication of positive-stranded RNA viruses is the extensive remodeling of host endomembranes that results in the formation of double-membrane vesicles (DMVs) and convoluted membranes (CMs) to which the viral replication and transcription complexes are targeted [ 14 – 16 ] . Notably, the host factors required for the formation of these structures remain elusive. Newly synthesized viral RNA is assembled to viral particles at the endoplasmic reticulum–Golgi intermediate compartment (ERGIC) and trafficked to the Golgi for posttranslational modifications [ 17 ]. Although little is known about how HCoVs exit from infected cells, recent work found that the β-CoVs MHV and SARS-CoV egress from cells via a lysosome-based pathway [ 18 ].

Coronaviruses (CoVs) are positive-sense single-stranded enveloped RNA viruses with a broad host tropism and, in case of the 3 highly pathogenic zoonotic CoVs, the ability to cross species barriers and infect humans. Since 1960, 7 human coronaviruses (HCoVs) with a suspected zoonotic origin in bats, mice, or domestic animals have been identified, including 4 seasonally circulating well-established human pathogens (HCoV-229E, HCoV-OC43, HCoV-NL63, and HCoV-HKU1) that usually cause mild symptoms like the common cold and diarrhea in immunocompetent patients [ 1 – 4 ]. HCoV infections have therefore generally been considered harmless; however, the relatively recent emergence of 3 highly pathogenic HCoVs, which infect the upper and also lower respiratory tract and cause severe disease in humans, has demonstrated that HCoVs can deleteriously impact human health. Between 2002 and 2003, the highly pathogenic Severe Acute Respiratory Syndrome Coronavirus (SARS-CoV) was responsible for an outbreak of severe viral pneumonia causing disease in over 8,000 patients [ 5 ]. Moreover, the emergence of Middle East Respiratory Syndrome Coronavirus (MERS-CoV) in 2012 marked the second occurrence of a highly pathogenic CoV in humans and has persistently caused endemics in the Middle East via zoonotic transmissions from dromedary camels and nosocomial outbreaks [ 6 – 8 ]. The newly emerged Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), the causative agent of Coronavirus Disease 2019 (COVID-19), continues to create an imminent threat to global health, with almost 250 Mio individuals currently infected in >200 countries and more than 5 Mio fatalities (November 6, 2021) (Johns Hopkins University and Medicine Coronavirus Resource Center).

(A) Inhibitors on WT cell lines (3 graphs): MERS-CoV, SARS-CoV, and SARS-CoV-2 replication in log 10 GE copy numbers/2 μl upon treatment of Huh7 (MERS-CoV) and VeroE6 (SARS-CoV and SARS-CoV-2) cell lines with cyclosporine A and alisporivir at indicated concentrations. Titers are shown at 24 hours postinfection/inhibitor treatment. Raw data can be found in Supporting information S1 Data , tab 12. (B) Inhibitors on KO cell lines (2 graphs): HCoV-229E titers in log 10 TCID 50 /ml in cyclosporine A and alisporivir-treated TMEM41B-KO and WT cells are indicated blue and purple. Titers are shown at 24 hours postinfection/inhibitor treatment. Raw data can be found in Supporting information S1 Data , tab 13. (C) Inhibitors in human nasal epithelial cell cultures (2 graphs): SARS-CoV-2 titers in log 10 TCID 50 /ml in primary human nasal epithelial cells at 48 hours postinfection and in presence of cylosporine A and alisporivir. Raw data can be found in Supporting information S1 Data , tab 14. (D) Immunofluorescence staining of SARS-CoV-2 infected human primary nasal epithelial cells, following DMSO, alisporivir, cyclosporine A, and untreated treatment, at indicated concentrations, 48 hours postinfection. dsRNA is shown in red, tight junctions (ZO-I) are shown in white, and cilia (β-tubulin) are shown in light blue. Images were acquired using an EVOS FL Auto 2 imaging system with a 40× air objective and processed using Fiji. dsRNA (red), cilia (β-tubulin, light blue), and the outline of segmented cells (ZO-I, white) of representative images are shown. Scale bar: 20 μm. Used reagents are listed in detail in Table 1 . CoV, coronavirus; HDF, host dependency factor; hNEC, xxx; KO, knockout; MERS-CoV, Middle East Respiratory Syndrome Coronavirus; SARS-CoV, Severe Acute Respiratory Syndrome Coronavirus; SARS-CoV-2, Severe Acute Respiratory Syndrome Coronavirus 2; TCID 50 , 50% tissue culture infectious dose; WT, wild-type.

TMEM41B, FKBP8, and MINAR1 have all been implicated as interactors of the autophagy pathway ( Fig 4A ). Moreover, FKBP8 is part of a large immunophilin family of proteins. Interestingly, in addition to FKBP8, several cyclophilins (additional members of the immunophilin family) were also significantly enriched in our MERS-CoV and HCoV-229E CRISPR KO screens, including peptidyl-prolyl isomerase (PPI) B, PPIC, PPID, PPIE, PPIF, PPIG, and PPIH. Proteins of this family specifically bind cyclosporine A, an immunosuppressant drug that is usually applied to suppress rejection after internal organ transplantation. Notably, while there are several drugs currently available for SARS-CoV-2 treatment, such as the antiviral drug remdesivir, and neutralizing monoclonal antibodies or antibody cocktails, such as LY-CoV555 and REGN-CoV2, many of these drugs have limitations, including SARS-CoV-2 specificity or use authorization only during certain stages of disease progression. For example, SARS-CoV-2 immune evasion, seen for some emerging variants of concern, render treatment with monoclonal antibodies ineffective if the antibody-specific epitopes are affected. It is thus critically important to not only conduct further research on existing antiviral therapies, but to also identify more drugs that effectively target CoV replication. Given the results from our CRISPR KO screen, we therefore tested whether the preexisting, clinically approved immunophilin targeting drugs cyclosporine A and alisporivir (a nonimmunosuppressant derivative of cyclosporine A currently used for the treatment of hepatitis C virus (HCV) [ 36 ]) as well as bafilomycin A1, which inhibits the autophagic flux by disrupting the lysosomal proton pump V-ATPase, could inhibit HCoV replication in human cells. Importantly, cyclosporin A is also known to inhibit calcineurin (PP3R1, MERS-CoV–specific HDF, Fig 4A ) in its complexed form with the respective immunophilin [ 37 ].

(A) Upon starvation, the mTORC1 complex is blocked and activation of the PI3K complex, as well as the ULK1 complex, leads to the initiation of phagophore formation, as an initial step in the autophagy pathway. MERS-CoV and HCoV-229E top scoring CRISPR KO screen hits FKBP8, MINAR1, TMEM41B, and VMP1 are involved in this early pathway. Furthermore, the ATG8 system containing among others LC3, which is recruited by VPM1 or FBKP8, is necessary for targeting cellular cargo to the autophagosome. PPP3R1 is up-regulated and initiates TFEB translocalization to the nucleus, where it catalyzes transcription of ATGs. MERS-CoV or conserved HDFs are indicated in respective colors. Inhibitor intervention in this pathway is shown in red. (B) Heatmap of canonical autophagy genes showing how host factors, including Beclin-1, ATG5, ATG7, and ATG14 are regulated in this CRISPR KO screen. MERS-CoV–specific host factors are shown in purple, HCoV-229E–specific host factors are indicated in green, and common host factors are marked in orange. Raw data for (B) can be found in Supporting information S1 Data , tab 1. (C) Immunofluorescence staining of LC3-GFP expressing Huh7, TMEM41B-KO, FKBP8-KO, and MINAR1-KO cells upon rapamycin treatment and HCoV-229E infection. LC3-GFP is depicted in green, dsRNA is shown in red and DAPI in blue, and scale bar is 20 μm. Representative images of 1 out of 4 independent replications are shown. Images were acquired using an EVOS FL Auto 2 imaging system with a 20× air objective and processed using Fiji. (D) Quantification of (C) shows LC3 puncta-positive cells upon rapamycin treatment and HCoV-229E infection in native Huh7, as well as TMEM41B-KO, FKBP8-KO, and MINAR1-KO cells. LC3-postive puncta are indicated by arrows in HCoV-229E–infected WT cells. Five images per condition in 3 independent experiments were acquired using an Evos FL Auto 2 imaging system with a 4× air objective, analyzed and quantified in Fiji. Each cell that depicted LC3-positive puncta was counted as a puncta positive cell, independent of the actual number and size of puncta. Statistical significance determined using the Holm–Sidak method, with alpha = 0.05 in GraphPad Prism 8.3.1. Each row was analyzed individually, without assuming a consistent SD. Raw data can be found in Supporting information S1 Data , tab 11. Used reagents are listed in detail in Table 1 . FKBP8, FK506 binding protein 8; HCoV, human coronavirus; HDF, host dependency factor; KO, knockout; MERS-CoV, Middle East Respiratory Syndrome Coronavirus; MINAR1, Membrane Integral NOTCH2 Associated Receptor 1; TMEM41B, transmembrane protein 41B; VMP1, vacuole membrane protein 1; WT, wild-type.

Despite having distinct cellular functions, TMEM41B, FKBP8, and MINAR1 are all involved in the cellular or mitochondrial autophagy pathways, albeit at different stages ( Fig 4A ) [ 28 – 30 , 33 , 34 ]. As autophagy was also identified as one of the main conserved biological clusters in our GO analysis, we next chose to focus on these factors in the context of autophagy for further analysis. To confirm the association of TMEM41B, FKBP8, and MINAR1 with cellular autophagy, we induced autophagy in LC3-GFP transfected KO cells using rapamycin and subsequently infected these cells with HCoV-229E. Under normal physiological conditions, the cytosolic protein LC3 translocates to autophagosomal membrane structures during early autophagy, illustrated in IF via LC3-GFP-positive puncta [ 35 ]. We thus analyzed the ability of LC3-GFP to translocate to such vesicles in TMEM41B-, FKBP8-, and MINAR1-KO cells infected with HCoV-229E and undergoing autophagy as described previously [ 35 ] and analyzed our results using immunofluorescence ( Fig 4C and 4D ). In line with previous reports, we confirmed by visualizing LC3-GFP-positive puncta formation that rapamycin treatment induced specific vesicle formation in native Huh7 cells, but not in TMEM41B-KO, FKBP8-KO, or MINAR1 KO cells, reasserting the necessity of these proteins for autophagosome formation. Similarly, LC3-GFP-positive puncta (indicated by white arrows) accumulated in Huh7 cells during HCoV-229E infection, but less so in TMEM41B-KO, FKBP8-KO, and MINAR1-KO cells ( Fig 4C and 4D ). To further understand the correlation between TMEM41B, FKBP8, and MINAR1 function during HCoV-229E infection and autophagy, we analyzed the protein levels of the autophagy marker LC3II in native, rapamycin-treated, and HCoV-229E–infected KO cells ( S10C Fig ). The accumulation of baseline LC3II in TMEM41B-, FKBP8-, and MINAR1-KO cells confirms a role for all 3 factors during autophagy. However, LC3II levels during HCoV-229E infection and rapamycin treatment of wild-type (WT) and TMEM41-, MINAR1-, and FBKP8-KO cells are not changed, except for a significant accumulation of LC3II in MINAR1-KO cells after rapamycin treatment ( S10B and S10C Fig ). Moreover, protein levels of the autophagy cargo receptor p62 were analyzed in FKBP8-KO cells. Whereas both p62 expression and LC3II expression were significantly decreased in rapamycin-treated cells and reduced in HCoV-229E infected WT cells, similar treatment did not significantly alter p62 or LC3II expression levels in FKBP8-KO cells ( S10B Fig ). Together, these results show that KO of TMEM41B, FKBP8, and MINAR1 impairs membrane remodeling during rapamycin-induced autophagy and compromises LC3-GFP translocation during HCoV-229E infection and that rapamycin-induced autophagy as well as HCoV-229E expression interferes with LC3II and p62 in WT cells, but not in KO cells.

MERS-CoV (A) and HCoV-229E (B) titers upon KO of top scoring HDFs are displayed in Log 10 TCID 50 /ml. Raw data can be found in Supporting information S1 Data , tabs 6 and 7. MERS-CoV (C) , HCoV-229E (D) , SARS-CoV (E) , and SARS-CoV-2 (F) titers upon reconstitution of TMEM41B, FKBP8, and MINAR1 in respective KO cell lines. Titers are shown relative to WT cell line. Results are displayed as a mean of 3 independent experiments with SD represented by error bars. Raw data can be found in Supporting information S1 Data , tab 8. (G) Quantification of western blot analysis of FKBP8 and TMEM41B in Huh7 WT, KO, and reconstituted cells relative to beta actin. Quantification of MINAR-GFP utilizing an anti-GFP antibody in WT, KO, and MINAR1-GFP reconstituted cells relative to beta actin. Raw western blot images can be found in Supporting information S1 Raw Images. Quantification data can be found in Supporting information S1 Data , tab 9. (H) Immunofluorescence analysis of VSVΔG-CoV-spike-GFP and VSVΔG complemented with G in respective CTRL and KO cells. Scale bar is 40 μm. GFP is depicted in green, and DAPI is shown in blue. A total of 20 to 26 images per condition in 3 independent experiments were acquired using an Evos FL Auto 2 imaging system with a 10× air objective, analyzed and quantified in Fiji. Representative images of 1 out of 3 independent replications are shown. (I) Quantification of CoV-mediated entry visualized in (H). Raw data can be found in Supporting information S1 Data , tab 10. In A and B, statistical analysis was determined by ordinary 1-way ANOVA, Dunnett multiple comparison test, using Nev 2020 version 9.0. In C–G, statistical significance was determined in R (version 4.0.2) using a 2-tailed paired t test to compare titers between each KO and KO+ (reconstituted) cell line. A 2-tailed unpaired t test was performed in I, but in this case, each KO was compared to the Huh7-ACE2 control cells. Used reagents are listed in detail in Table 1 . CoV, coronavirus; FKBP8, FK506 binding protein 8; HCoV, human coronavirus; HDF, host dependency factor; KO, knockout; MERS-CoV, Middle East Respiratory Syndrome Coronavirus; MINAR1, Membrane Integral NOTCH2 Associated Receptor 1; SARS-CoV, Severe Acute Respiratory Syndrome Coronavirus; SARS-CoV-2, Severe Acute Respiratory Syndrome Coronavirus 2; TCID 50 , 50% tissue culture infectious dose; TMEM41B, transmembrane protein 41B; VSV, vesicular stomatitis virus; WT, wild-type.

Based on our initial gene enrichment results from the MERS-CoV and HCoV-229E screens, as well as a comparison of the respective results with previously published data [ 28 – 30 ], we selected 21 hits for further experimental validation. Focusing on the highly pathogenic MERS-CoV screen, but also with an interest in examining common hits between both screens, we chose 17 genes that were significantly enriched in the MERS-CoV screen and 4 genes (TMEM41B, ELFN2, NOM1, and KRTAP13-4) that were significantly enriched in both MERS-CoV and HCoV-229E screens. For these 21 hits, stable CRISPR/Cas-9 KO cell lines were generated for each gene and then challenged with either HCoV-229E or MERS-CoV. Specific KO of the MERS-CoV receptor DPP4 and the HCoV-229E receptor APN served as controls. MERS-CoV replication could be significantly reduced in all KO cell lines, except for WNT5A and APN, thus confirming our screen and validating our data analysis ( Fig 3A , S10A Fig ). In contrast to MERS-CoV, HCoV-229E replication was significantly impaired upon deletion of APN as well as CDH7, MINAR1, TMEM41B, and FKBP8. Interestingly, KO of WNT5A significantly reduced HCoV-229E titers ( Fig 3B , S10D Fig ). Importantly, TMEM41B, FKBP8, and MINAR1 KO resulted in impaired titers for both MERS-CoV and HCoV-299E. Strikingly, SARS-CoV and SARS-CoV-2 also replicated to lower titers in respective KO cell lines expressing the specific entry receptor ACE2, confirming a conserved function in the CoV replication cycle for these 3 genes ( Fig 3E and 3F , S10D Fig ). To further validate the effect of the CRISPR/Cas-9–mediated KO of all the 3 host factors, we expressed CRISPR-resistant variants of these host factors and observed a rescue of virus titers for MERS-CoV, HCoV-229E, SARS-CoV, and SARS-CoV-2, thereby confirming the pan-CoV antiviral effect of TMEM41B, FKBP8, and MINAR1 KO ( Fig 3C–3F ). Western blot analysis confirmed stable KO of both FKBP8 and TMEM41B ( Fig 3G ), as well as rescue of protein expression in the transfected condition. As MINAR1 antibodies are not available, reconstituted protein expression is shown using a GFP antibody, detecting the sgRNA-resistant MINAR1-GFP fusion protein. Moreover, CRISPR/Cas-9–mediated genome editing in MINAR1, FKBP8, and TMEM41B KO cell lines were confirmed via Sanger sequencing ( S10F Fig ). To investigate the step of the viral replication cycle for which these factors are required, we employed a recombinant vesicular stomatitis virus (VSV) infection system, which is depleted in its surface protein G and instead stably expresses the spike protein from HCoV-229E, MERS-CoV, or SARS-CoV-2. This system specifically enables the analysis of CoV spike-mediated entry independent of CoV replication [ 31 , 32 ]. We found that the KO of TMEM41B, FKBP8, or MINAR1 did not alter CoV spike-mediated entry, whereas MERS-CoV spike-mediated entry was significantly reduced in DPP4-KO cells and HCoV-229E spike-mediated entry, as well as SARS-CoV-2 spike-mediated entry, was significantly impaired in APN-KO and ACE2-deficient Huh7 cells, respectively ( Fig 3I and 3H ). Collectively, these findings show that there is a conserved requirement for the host factors TMEM41B, FKBP8, and MINAR1 during CoV replication, but not during CoV entry.

To establish which pathways and/or processes may be particularly important for CoV replication, we next focused on conserved representative GO terms that included 1 or more of the 19 genes that were significantly enriched in both of our CoV screens ( Fig 1D and 1E ). The resulting 70 unique GO terms and their relationships to each other are the terms illustrated in Fig 2A . The 7 prominent biological themes these 70 terms clustered into are also shown and include autophagy, immunity, dephosphorylation, Golgi vesicle transport, catabolic processes, homeostatic processes, and developmental processes. To examine each biological cluster in more detail, we constructed cluster-specific heatmaps showing all enriched genes from both CoV screens associated with that cluster ( Fig 2B ). Furthermore, for each cluster, we inspected the network of functionally related GO terms that comprise the cluster ( S3 – S9 Figs). Overall, our results indicate the involvement of diverse BPs in both MERS-CoV and HCoV-229E replication cycle.

(A) Enrichment map summarizing major host biological networks co-opted by CoVs during infection. GO enrichment analysis was performed using hits from both MERS-CoV and HCoV-229E CRISPR screens and filtered to contain conserved representative GO terms and genes. Each node represents an individual GO term and nodes that are functionally related cluster together into a larger network. Node size reflects number of significantly enriched genes in the node, and color indicates the CoV screen for which the node was significant. Raw data for (A) can be found in Supporting information S1 Data , tab 2. A complete list of significant GO terms can be found in S2 Table . (B) Heatmaps of individual biological clusters displayed in (A). Heatmaps contain significantly enriched genes from both CoV screens that were associated with significantly enriched GO terms found within the individual biological clusters in (A). Colored panels on the left-hand side of heatmaps show which CoV screen contained specific enriched genes (purple: MERS-CoV, green: HCoV-229E, and orange: enriched in both CoV screens). Colors in each legend represent the log RRA p-values for each gene in each CoV screen. Heatmap clustering was performed using the complete linkage method and Euclidean distance. Genes associated with significant GO terms are listed in Supporting information S1 Data , tab 2. Raw data for these genes can be found in Supporting information S1 Data , tab 1. CoV, coronavirus; GO, Gene Ontology; HCoV, human coronavirus; MERS-CoV, Middle East Respiratory Syndrome Coronavirus; RRA, robust rank aggregation.

To identify and compare host cell biological processes (BPs) that may be required for CoV replication, we next performed Gene Ontology (GO) enrichment analysis on each screen using the enriched genes identified above. This analysis uncovered multiple BPs that were significantly enriched in both CoV screens, many of which clustered together into 7 overarching biological themes ( Fig 2A ). Next, we calculated the semantic similarity among the 636 unique GO terms (BP) that were identified as significantly enriched in one or both screens (p-value <0.05; S2 Table ). Hierarchical clustering was then used to group similar GO terms together, and a representative term for each group was selected based on scores assigned to each term. The latter analysis led to the identification of 44 conserved representative GO terms and 51 virus-specific representative GO terms ( S2A Fig ). Representative GO terms found in both MERS-CoV and HCoV-229E screens included a number of immune-related terms as well as terms related to the regulation of phosphorylation, kinase activity, autophagy, and lipid transport. Several specific GO terms were also significantly enriched in both screens, including neutrophil-mediated immunity, regulation of protein dephosphorylation, and regulation of the c-Jun N-terminal kinase (JNK) cascade ( S2B Fig ). GO terms specific to our MERS-CoV screen included regulation of exit from mitosis, protein glycosylation, and syncytium formation via plasma membrane fusion. By contrast, GO terms specific to HCoV-229E included regulation of coagulation and nitric oxide biosynthesis ( S2A Fig ).

Using the MAGeCK pipeline [ 23 ], we performed paired analyses on uninfected and infected samples from each screen and computed gene-level scores using sgRNA log fold changes (LFCs) to identify KO genes that were significantly enriched in our MERS-CoV and HCoV-229E infected samples. Overall, we identified 1,149 significantly enriched genes in the MERS-CoV screen and 517 significantly enriched genes in the HCoV-229E screen using the robust rank aggregation (RRA) algorithm implemented in the MAGeCK pipeline and the “alpha median” method to calculate gene-level LFCs between samples. RRA analysis using the second-best LFC method identified 989 significantly enriched genes in the MERS-CoV screen and 332 significantly enriched genes in the HCoV-229E screen ( S1 Table ). To prioritize genes and generate a more robust data set, we focused on genes identified as significantly enriched by both LFC methods (i.e., RRA p-value ≤ 0.05 and LFC ≥ 2; S1C and S1D Fig ). In total, 944 genes from the MERS-CoV screen and 332 genes from the HCoV-229E screen met these criteria, including 19 genes that were identified by both methods in both screens ( Fig 1B and 1D , S1D Fig ). Top scoring genes from both screens are shown in Fig 1E , including several virus-specific genes as well as the 19 aforementioned common genes. Importantly, in the MERS-CoV screen, the DPP4/CD26 host cell receptor was identified as the top scoring gene, whereas in the HCoV-229E screen, the top scoring gene was ANPEP/CD13. Moreover, the known DPP4 transcription factor HNFA1 was ranked second in the MERS-CoV screen, demonstrating the robustness of our CoV screens. We did not observe significant enrichment of the cellular proteases TMPRSS2 (p-value = 0.32526 and rank = 6,626 for HCoV-229E; p-value = 0.90798 and rank = 19,707 for MERS-CoV), TMPRSS4 (p-value = 0.60519 and rank = 10,916 for HCoV-229E; p-value = 0.60285 and rank = 12,909 for MERS-CoV), furin (p-value = 0.60519 and rank = 13,518 for HCoV-229E; p-value = 0.17977 and rank = 6,118 for MERS-CoV), or cathepsin L (p-value 0.29553 and rank = 6,102 for HCoV-229E and p-value = 0.34682 and rank = 9,302 for MERS-CoV) in our CoV screens, suggesting that despite the importance of these factors for CoV infection, they might not be essential, or their specific function may be carried out by other functionally similar proteins [ 24 ]. Indeed, one limitation of genome-wide CRISPR screens is that genes with redundant functions that may normally contribute to a particular cellular process are often missed. Interestingly, we did find that TMPRSS9 and cathepsin H were significantly enriched in our HCoV-229E and MERS-CoV screens, respectively. TMPRSS9 is highly expressed in lungs [ 25 ], and a possible role of TMPRSS9 in biological pathways leading to respiratory symptoms has been suggested [ 26 ]. Its role in HCoV infection needs to be further elucidated. Moreover, cathepsin H plays a role in surfactant processing in alveolar type II cells, a major CoV target cell type in the lung [ 27 ].

(A) Native Huh7 cells were transduced with the GeCKOv2 lentiviral genome-wide CRISPR library, ensuring a coverage of approximately 500 cells per sgRNA. Transduced cells were selected and then infected with either MERS-CoV or HCoV-229E at indicated MOIs and temperatures. Surviving cells were harvested and prepared for deep sequencing. Deconvolution identified both virus-specific and pan-coronavirus HDFs. (B) Volcano plot showing the LFC (Log 2 FC) and log 10 p-value for each gene in the MERS-CoV CRISPR screen. Genes with a FC ≥2 and p-value <0.05 are highlighted in red. Selected top genes are annotated in the plot, including the MERS-CoV receptor (DPP4) and the 5 most highly ranked genes in the MERS-CoV screen. (C) Volcano plot showing the Log 2 FC and log 10 p-value for each gene in the HCoV-229E CRISPR screen. Genes with a FC ≥ 2 and p-value <0.05 are highlighted in red. Selected top genes are annotated, including the HCoV-229E receptor (ANPEP) and the 5 most highly ranked genes in the HCoV-229E screen. (D) Pairwise comparison of enriched genes in the HCoV-229E and MERS-CoV CRISPR screens. Dotted lines indicate a Log 2 FC ≥ 2. Genes with a Log 2 FC ≥ 2 and p-value <0.05 in both screens are highlighted in red and annotated. (E) Heatmap comparing the log RRA p-values for selected top virus-specific and common hits in both CoV screens. CoV receptors (DPP4 and ANPEP) are demarcated by the blue boxes, MERS-CoV specific genes by the purple boxes, and HCoV-229E specific genes by the green boxes. Common significantly enriched genes, which are also annotated in Fig 2D , are demarcated by the orange boxes. Heatmap clustering was performed using the complete linkage method and Euclidean distance. Raw data for B–E can be found in Supporting information S1 Data , tab 1. CoV, coronavirus; DPP4, dipeptidyl peptidase 4; HCoV, human coronavirus; HDF, host dependency factor; KO, knockout; LFC, log fold change; MOI, multiplicity of infection; MERS-CoV, Middle East Respiratory Syndrome Coronavirus; RRA, robust rank aggregation; sgRNA, single guide RNA.

Discussion

The identification of HDFs essential for HCoV infection offers great potential to reveal novel therapeutic targets and enhance our understanding of HCoV infection and pathogenesis (e.g., COVID-19). Here, we have performed 2 independent genome-wide CRISPR/Cas-9 KO screens in Huh7 cells with HCoV-229E and MERS-CoV to identify functionally important genes during HCoV infection. Using MERS-CoV as a representative emerging virus and HCoV-229E as a representative endemic virus, we identified multiple virus-specific and conserved HDFs, including several that are required for the replication of the novel pandemic CoV SARS-CoV-2. GO enrichment analysis revealed that the conserved HDFs were involved in diverse BPs that clustered into 7 major categories. Interestingly, we found that MERS-CoV and HCoV-229E seemed to exploit different components of the same BPs, as the majority of genes involved in each biological cluster were virus specific, but the overall BPs were similar. This may be due to evolutionary differences between the viruses, as MERS-CoV is part of the betacoronavirus genus, whereas HCoV-229E is a member of alphacoronavirus genus. Furthermore, many commonly enriched genes were involved in Golgi vesicle transport, or more specifically in vesicle coating and budding from membranes, as well as regulation of endocytosis and exocytosis, which are known to be associated with virus entry and exit [40]. Moreover, Golgi vesicle markers have been found in close proximity to CoV replication compartments, suggesting another potential function for genes in this cluster during CoV replication, e.g., membrane reorganization for membranous replication compartments [41]. A second prominent category was the immune system cluster, which may be associated with direct exploitation of immunological host responses against CoVs and thus offer potential intervention strategies. These strategies may also have antiviral efficacy and work to lower dysfunctional immune responses, which is a known driver of disease progression and severe lung pathology [42]. Another major category containing enriched genes in both HCoV screens was dephosphorylation. Genes involved in phosphorylation and kinase activities were strongly enriched in our screens, suggesting that these processes are required for HCoV replication and that other CoVs also exploit the host’s phosphorylation machinery for their benefit. Importantly, recent work observed striking changes in phosphorylation on host and viral proteins during SARS-CoV-2 infection, including many changes related to dephosphorylation and altered kinase activity [43,44]. For example, the JNK signaling cascade, but also the regulation of tau-protein kinase activity, were highly enriched in our MERS-CoV screen. JNKs belong to the mitogen-activated protein kinase (MAPK) family, and SARS-CoV-2 infection was recently shown to promote p38 MAPK signaling activity [43]. Of note, the FKBP8 gene clustered into the dephosphorylation category, and the MINAR1 gene was included in regulation of tau-protein kinase activity, suggesting that these 2 genes may influence CoV replication via other BPs in addition to autophagy. Along this line, therapeutical intervention targeting AP2M1 (part of the clathrin-dependent endocytic pathway) phosphorylation using a kinase inhibitor resulted in reduced SARS-CoV, MERS-CoV, and SARS-CoV-2 infection, exemplifying the antiviral potential of targeting specific phosphorylation sites during viral infection [45]. Our analysis also found that genes involved in catabolic and homeostatic processes were significantly enriched in both CoV screens. Interestingly, a similar cluster linked to cholesterol metabolism was identified in previous studies, including SARS-CoV-2, HCoV-229E, and HCoV-OC43 genome-wide CRISPR/Cas-9–mediated KO screens and SARS-CoV-2 interactome studies [46,47] and has been linked to CoV entry and membrane fusion [48]. Furthermore, one gene that is worth mentioning, as its KO had the highest effect on MERS-CoV and HCoV-229E infectious titer reduction, is cadherin 7 (CDH7). CDH7 belongs to a family of calcium-dependent cell adhesion proteins and is expressed on the plasma membrane [49]. Interestingly, viruses such as Epstein–Barr virus (EBV), hepatitis B virus (HBV), HCV, and human herpesvirus 8 (HHV8) have previously been shown to degrade cadherins for their benefit [50]. A possible interaction between CoVs and CDH7 is therefore definitely plausible. As it is located at the plasma membrane, it might be important for HCoV entry or could possibly play a role in transmission, as has been observed for measles virus [50] and HBV [51], which use formation of adherent junctions for efficient virus transmission.

For our downstream experimental analysis, we focused on the autophagy cluster. Autophagy is a cellular stress response (e.g., to starvation or infection by pathogens) that involves the recycling of proteins and cell organelles to maintain cellular homeostasis [52]. These processes exploit a very wide-ranging group of cellular trafficking pathways required for transportation of cytoplasmic material to the lysosome for destruction. Interestingly, several autophagy-related genes were identified as top hits in our screen, including TMEM41B, FKBP8, and MINAR1. Further, we demonstrated that these host factors are required for replication of several CoVs. A similar recently published genome-wide CRISPR screen for MERS-CoV HDFs [53] identified TMEM41B as a proviral gene specific for MERS-CoV replication in African Green Monkey VeroE6-Cas-9 cells, underlining the importance of this host factor for MERS-CoV replication among different model systems. The same study identified genes involved in diverse biological pathways such as chromatin remodeling, histone modification, cellular signaling, and RNA regulation as essential for MERS-CoV replication. Another recent genome-wide CRISPR-mediated KO screen for HCoV-229E and SARS-CoV-2 host factors in Huh7 cells identified TMEM41B, as well as PIK3C3 (endocytic trafficking and autophagy), as top-ranked genes for both viruses, with a higher dependency of HCoV-229E on TMEM41B, which is in line with our results. Notably, this study also shows that disruption of the autophagy genes ATG5 and ATG7, which are required for phagophore expansion, do not block SARS-CoV-2 and HCoV-229E infection, suggesting that later stages of autophagy are not required for CoV replication [54]. These results are reflected in our study since ATG5 and ATG7 were not identified as hits in our CoV screens (Fig 4B). Finally, TMEM41B was also identified as a HCoV-229E host factor in Huh7.5 cells, along with VMP1, which ranked second in our HCoV-229E host factor screen. Interestingly, SARS-CoV-2 and HCoV-OC43, as well as HCoV-NL63, are also dependent on TMEM41B for efficient replication [20].

The ER localized TMEM41B was recently identified as a gene required for early autophagosome formation and lipid mobilization in 3 independent genome-wide CRISPR KO screens that aimed to identify host factors essential for autophagy. They also observed that TMEM41B and the well-characterized early-stage autophagy protein VMP1 (top scoring HDF in our HCoV-229E screen with p-value = 3.1525e-08 and rank = 2; p-value = 3.4752e-02 and rank = 1919 for MERS-CoV) implement related functions [28–30]. Furthermore, interaction of TMEM41B with Beclin1 (PI3K complex) underscores the importance of this protein in the induction of autophagy [55]. Interestingly, the FK506-binding protein 8 (gene: FKBP8, protein: FKBP38), a member of the immunophilin protein family, is located in the outer mitochondrial membrane and plays a key role in mitophagy by inhibiting the mTORC1 complex during nutrient deprivation [56]. Moreover, FKBP8 targets Beclin-1 to ER–mitochondria membranes during mitophagy and recruits LC3A to damaged mitochondria, thereby actively inducing the removal of excess mitochondria by autophagy [33]. FKBP8 itself avoids degradation by escaping from mitochondria and is translocated to the ER [57]. MINAR1 (also known as Ubtor or KIAA1024) was the third MERS-CoV HDF identified with a possible indirect involvement in autophagy regulation. The otherwise very rudimentarily characterized protein plays a role in regulating cell growth and mTOR signaling, as MINAR1 depletion resulted in higher mTOR activity [34] (Fig 4A). In addition, the phosphatase PPP3R1, commonly referred to as calcineurin, is up-regulated during cell starvation and controls the activity of the TFEB transcriptional regulator of lysosomal biogenesis and autophagy [58]. Importantly, the interaction between autophagy components and CoVs, but also other positive-stranded RNA viruses, during viral replication has been under discussion for a long time, as parts of the autophagy process show similarities to the process of DMV formation [41,59,60]. CoVs rely on the formation of replication complexes at DMVs, the presumed site of viral genome replication and transcription. Due to a lack of conventional ER or Golgi protein markers, the exact origin of DMVs remains unclear, and studies investigating the possible involvement of the early autophagy machinery in the conversion of host membranes into DMVs reached conflicting conclusions [61,62]. Another possibility is that single components of the autophagic machinery may be hijacked by CoVs independently of their activity in autophagic processing. The nonlipidated autophagy marker LC3 has been observed to localize to DMVs, and the down-regulation of LC3, but not inactivation of host cell autophagy, protects cells from CoV infection [59,63–65]. We show that TMEM41B, MINAR1, and FKBP8 are involved in regulating LC3-positive puncta formation following chemical induction of autophagy and HCoV-229E infection and that KO of each gene distinctly impairs HCoV replication. Further, the accumulation of LC3II protein expression in TMEM41B-, FKBP8-, and MINAR1-KO cells compared to WT cells confirmed an impaired autophagic flux in all 3 KO cells, validating these host factors as components of the autophagy pathway. When analyzing LC3II and p62 protein expression during HCoV-229E infection or rapamycin treatment in WT cells, we observed a significant decrease of both LC3II and p62 expression being in line with an up-regulated autophagic flux in the rapamycin-treated condition. As neither LC3II nor p62 protein expression was changed in the analyzed rapamycin-treated or HCoV-229E infected KO-cell lines (except for an upregulation of LC3II in rapamycin-treated MINAR1-KO cells, which is possibly still connected to a decrease in autophagic flux), further studies are required to disentangle the role of these 2 proteins during CoV infection and to clarify whether the autophagy pathway is influenced by CoV infection. Our results clearly show that with TMEM41B, FKBP8, and MINAR1, components of the autophagy pathway are involved in HCoV infection. Whether these components act via their native function which they exhibit during autophagy or whether HCoVs recruit these factors to act in an autophagy-independent way remains elusive. Further roles of the 3 identified host factors have been suggested. Both TMEM41B and FKBP8 are thought to interact with Beclin-1, which is a core subunit of the PI3K complex that drives autophagy [55,66]. Captivatingly, inhibition of SKP2, another Beclin-1 interactor, reduced MERS-CoV infection [67]. Recent work suggested a putative autophagy-independent role for TMEM41B as a pan-coronavirus and flavivirus replication factor, which is recruited to flavivirus RNA replication complexes to facilitate membrane curvature and create a protected environment for viral genome replication [68,69]. Furthermore, MINAR1 serves as a regulator of mTOR signaling, which regulates numerous cellular processes including the cap-dependent mRNA translation and synthesis machinery required during viral replication. These observations add further potential layers of modulation by TMEM41B, FKBP8, and MINAR1 during CoV replication.

Independently of the exact underlying mechanism, our results suggest that the HDFs FKBP8, TMEM41B, and MINAR1 herein represent potential targets for host-directed therapeutics. Its immunomodulating component makes FKBP8 a very interesting HDF for CoV replication. Importantly, FKBP8 has further been shown to be involved in virus replication in a completely autophagy-independent way. FKBP8 interacts with retinoic acid inducible protein 1 (RIG-I), virus-induced signaling adaptor signaling (VISA), and IFN regulatory factor 3 (IRF3) during Sendai virus infection. Knockdown of FKBP8 promotes the activation of IFN-beta and the antiviral response during Sendai virus infection in HEK293T cells, suggesting a possible immunomodulatory component for its role in CoV infection [70]. In addition to FKBP8, several cyclophilins were up-regulated in both of our HCoV screens. Cyclophilins express PPI activity, which catalyzes the isomerization of peptide bonds in proline residues from trans to cis, thereby facilitating protein folding. Proteins of this family specifically bind cyclosporine A, an immunosuppressant drug that is usually applied to suppress rejection after internal organ transplantation. Moreover, FKBPs and cyclophilins have been the focus of several CoV studies showing impaired HCoV-229E, HCoV-NL63, SARS-CoV, and MERS-CoV replication upon FKBP and cyclophilin inhibitor treatment [71–76]. Given the lack of specific treatment options during the ongoing SARS-CoV-2 pandemic, we tested cyclosporine A, as well as alisporivir, a nonimmunosuppressant derivative of cyclosporine A, and showed that antiviral intervention using these clinically approved drugs inhibited the replication of the highly pathogenic CoVs MERS-CoV, SARS-CoV, and SARS-CoV-2 in a dose-dependent manner. Moreover, inhibitor treatment on top of TMEM41B-KO could further inhibit HCoV-229E. While Huh7 and VeroE6 cells are valuable model cell lines for highly pathogenic CoVs, they likely do not capture important aspects of infection compared to primary human airway epithelial cells nor fully recapitulate the complex cellular milieu present in human patients. To address these limitations, we also tested cyclosporine A and alisporivir on primary human nasal epithelial cell cultures and found that these compounds potently inhibited SARS-CoV-2 replication at concentrations known to be achievable and efficacious in patients. Together, these findings depict a promising path toward the repurposing of cyclosporine A and alisporivir as COVID-19 treatment options. Infection with highly pathogenic CoVs is frequently accompanied by inflammatory immunopathogenesis, including the virus-induced destruction of lung tissue and subsequent triggering of a host immune response. Importantly, in certain cases, a dysregulated immune response is associated with severe lung pathology and systemic pathogenesis [42]. The latter highlights the need for dual-acting antiviral drugs that also target inflammation and/or cell death. Of interest, alisporivir also blocks mitochondrial cyclophilin-D, a key regulator of mitochondrial permeability transition pore (mPTP) opening, which is a mechanism involved in triggering cell death. Hence, besides its antiviral properties, it is possible that alisporivir also reduces CoV-induced lung tissue damage [77]. Trials using either cyclosporine A in patients with moderate COVID-19 (ClinicalTrials.gov identifier: NCT04412785 and NCT04540926) or alisporivir (ClinicalTrials.gov identifier: NCT04608214) for the treatment of hospitalized COVID-19 patients have been registered.

The identification of MINAR1, TMEM41B, and FKBP8 as conserved HCoV HDFs in our MERS-CoV and HCoV-229E screens extends our knowledge of host–virus dynamics during HCoV infection. Furthermore, the involvement of FKBP8 and other members of the cyclophilin family in HCoV replication provides more information on how cyclosporin A and alisporivir are able reduce CoV replication by interfering with essential HCoV HDFs. We confirm the potential of both inhibitors as treatment against MERS-CoV and HCoV-229E infection and additionally observed similar reduction in SARS-CoV-2 replication. Altogether, our findings highlight the potential of using genome-wide CRISPR/Cas-9 KO screens to identify novel HDFs essential for HCoV infection, which can, in turn, be used in combination with clinically available drugs to identify and evaluate host-directed therapies against existing and future pandemic CoVs.

[END]

[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001490

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/