(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

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



Identification of cell type specific ACE2 modifiers by CRISPR screening

['Emily J. Sherman', 'Department Of Internal Medicine', 'Division Of Hospital Medicine', 'University Of Michigan', 'Ann Arbor', 'Michigan', 'United States Of America', 'Carmen Mirabelli', 'Department Of Microbiology', 'Immunology']

Date: 2022-05

SARS-CoV-2 infection is initiated by binding of the viral spike protein to its receptor, ACE2, on the surface of host cells. ACE2 expression is heterogeneous both in vivo and in immortalized cell lines, but the molecular pathways that govern ACE2 expression remain unclear. We now report high-throughput CRISPR screens for functional modifiers of ACE2 surface abundance. In liver-derived HuH7 cells, we identified 35 genes whose disruption was associated with a change in the surface abundance of ACE2. Enriched among these ACE2 regulators were established transcription factors, epigenetic regulators, and functional networks. We further characterized individual HuH7 cell lines with disruption of SMAD4, EP300, PIAS1, or BAMBI and found these genes to regulate ACE2 at the mRNA level and to influence cellular susceptibility to SARS-CoV-2 infection. Orthogonal screening of lung-derived Calu-3 cells revealed a distinct set of ACE2 modifiers comprised of ACE2, KDM6A, MOGS, GPAA1, and UGP2. Collectively, our findings clarify the host factors involved in SARS-CoV-2 entry, highlight the cell type specificity of ACE2 regulatory networks, and suggest potential targets for therapeutic development.

The amount of ACE2 on the surface of human cells is an important determinant of SARS-CoV-2 infection, but the molecular pathways that regulate ACE2 remain poorly understood. Identification of these pathways may clarify host factors involved in COVID-19 outcomes and offer targets for therapeutic development. ACE2-targeted therapies may furthermore be less susceptible than viral spike-targeted therapies to evasion by SARS-CoV-2 variants. To systematically identify regulators of human ACE2, we therefore performed high-throughput CRISPR screening for modifiers of ACE2 surface abundance in HuH7 liver-derived and Calu-3 lung-derived cell lines. Unexpectedly, aside from ACE2 itself, we identified distinct sets of ACE2 modifiers in either cell line. For a subset of ACE2 regulators, we validated their functional effect on ACE2, confirmed their relevance to SARS-CoV-2 infection, and clarified their level of regulation. Our findings demonstrate the important influence of cell type on investigations of SARS-CoV-2 infection and nominate candidate pathways for ACE2-targeted therapeutic development.

Importantly, single cell RNA-seq approaches have intrinsic limitations for low abundance transcripts [ 27 ], and most studies of ACE2 mRNA have lacked validation at the protein level. Investigations of ACE2 protein expression have been relatively limited and confounded by uncertain specificity of different commercial ACE2 antibodies. Recently, we engineered ACE2-overexpressing and ACE2-deleted cell lines and performed systematic testing of a panel of commercial antibodies by flow cytometry, finding only 2 of 13 to exhibit specificity and sensitivity for ACE2 surface protein [ 28 ]. Unexpectedly, we found that multiple isogenic cell lines demonstrated heterogeneity of endogenous ACE2 expression, suggesting that they may serve as a simplified model to dissect the molecular pathways that govern ACE2 expression. To this end, we now report the findings of our high-throughput CRISPR screens for modifiers of endogenous ACE2 surface abundance in HuH7 and Calu-3 cells.

Since the onset of the COVID-19 global pandemic, several studies have applied single-cell RNA-seq to examine ACE2 expression in tissues of humans and animal models [ 16 – 19 ]. These studies have consistently found ACE2 mRNA expression to be heterogeneous among cell types within a given tissue. ACE2-expressing cells include alveolar type 2 cells in the lung, goblet cells in the nasopharynx, absorptive enterocytes in the gut, and proximal tubular epithelial cells in the kidney. Even within a given cellular subtype, ACE2 expression is heterogeneous, with only ~1–5% of lung AT2 cells, for example, containing detectable ACE2 mRNA. Prior investigations have identified two distinct promoters for full-length ACE2 which vary across tissues in their relative usage [ 20 ], as well as a cryptic promoter driving expression of an interferon-responsive truncated ACE2 isoform [ 21 , 22 ]. Other studies have identified putative transcription factor binding sites and epigenome signatures associated with the ACE2 locus [ 23 , 24 ] as well as transcriptome profiles associated with ACE2 expression [ 25 , 26 ].

Given its importance in SARS-CoV-2 infection, the interaction between the viral spike protein and ACE2 is an attractive target for therapeutic development. Vaccines against the spike protein are broadly efficacious in reducing the number and severity of COVID-19 infections [ 8 ], but the rapid evolution of SARS-CoV-2 raises concern for the potential of SARS-CoV-2 variants to escape immunity induced by either vaccines or prior infection [ 9 – 12 ]. As an alternative strategy, disruption of host ACE2 may similarly prevent SARS-CoV-2 infection in a manner that is less susceptible to viral evolution. ACE2-targeted therapies may have broader clinical applications, as ACE2 also serves as the cellular receptor for other respiratory viruses such as SARS-CoV-1 [ 13 ] and HCoV-NL63 [ 14 ]. Additionally, ACE2 is an important physiologic regulator of the renin-angiotensin and kallikrein-kinin systems, and its dysregulation has been implicated in pulmonary and systemic hypertension, cardiac fibrosis, atherosclerosis, and acute respiratory distress syndrome [ 15 ]. However, no ACE2-targeted therapies have been clinically approved and their development is limited by uncertainty in the molecular pathways that regulate ACE2.

Overall, we identified only 5 genes in Calu-3 cells whose disruption was associated with a significant change in ACE2 surface abundance (FDR<0.05, absolute value of log 2 fold-change >1). Of these genes, only ACE2 itself had been found to have an effect on ACE2 surface abundance in HuH7 cells. Systematic comparison between the ACE2 screens in HuH7 and Calu-3 cells revealed no significant correlation ( Fig 6C ). To validate the Calu3 ACE2 screen findings, we again performed single gRNA targeting for the 5 identified genes–ACE2, KDM6A, MOGS, GPAA1, and UGP2. We confirmed efficient target protein depletion by immunoblotting ( Fig 6D ). Flow cytometry demonstrated a significant change in ACE2 surface abundance upon each perturbation, with the direction of effect matching that observed in the CRISPR screen ( Fig 6E , S9 Table ). For gene disruptions with the largest effect on ACE2 surface abundance by flow cytometry (ACE2, MOGS, UGP2), immunoblotting of the corresponding cellular lysates for ACE2 abundance was consistent ( Fig 6D ). For gene disruptions with smaller effects by flow cytometry (KDM6A and GPAA1), immunoblotting was equivocal for a similar influence on total cellular ACE2 abundance, potentially due to either the reduced sensitivity of this approach for subtle changes or due to a differential effect on surface-displayed versus total cellular ACE2. Immunoblotting also revealed a change in ACE2 electrophoretic mobility in MOGS and UGP2-targeted cells, suggestive of an effect on ACE2 glycosylation status.

(A) FACS plots of ACE2 surface staining in Calu-3 cells treated with control nontargeting (NT) or SMAD4-targeting gRNA for each of 3 technical replicates. (B) Volcano plots of MAGeCK Robust Rank Aggregation scores and gene-level gRNA log 2 fold-change in ACE2 negative relative to ACE2 positive populations for each gene tested in the focused CRISPR library in Calu-3 cells. Genes with FDR <0.05 and absolute log 2 fold-change >1 are highlighted, with positive regulators in blue and negative regulators in red. (C) Comparison of log 2 fold-change in ACE2 negative relative to ACE2 positive populations in focused CRISPR screens of HuH7 (x-axis) and Calu-3 (y-axis). Only ACE2, highlighted in blue, was identified in both screens with FDR<0.05 and absolute log 2 fold-change >1. Genes identified as significant in only the HuH7 or Calu-3 screens are highlighted in magenta and red, respectively. (D) Immunoblotting of lysates collected from Calu-3 cells targeted by CRISPR with a gRNA against the indicated gene or a nontargeting (NT) control sequence. (E) Percentage of cells with detectable surface ACE2 staining after CRISPR-mediated single gene disruption of the indicated target gene or a nontargeting control in Calu-3 cells. Candidate genes associated with positive or negative regulation of ACE2 in the CRISPR screen are highlighted in blue and red, respectively. ACE2-positive gates were defined on control unstained cells and the proportion of ACE2-positive cells for each population is displayed for each of 3–6 technical replicates. Error bars depict standard deviation. Asterisks indicate p<0.05 by Student’s t-test. Source data and statistical analysis are provided in S9 Table .

To examine the generalizability of our findings in HuH7 cells, we next tested the influence of SMAD4 disruption on ACE2 abundance in Calu-3 cells, a respiratory epithelial cell line that has been shown to recapitulate the SARS-CoV-2 replication kinetics and type I interferon responses of primary airway epithelial cells [ 48 ]. Surprisingly, surface staining of ACE2 revealed no significant change in surface ACE2 abundance upon SMAD4 targeting ( Fig 6A ). To more broadly interrogate the functional regulators of ACE2 in respiratory epithelial cells, we therefore repeated our focused CRISPR screen for ACE2 modifiers in Calu-3 cells ( Fig 6B and S8 Fig , S8 Table ). As was the case in HuH7 cells, we again found ACE2 itself to be readily discriminated as the top hit of the screen. Consistent with our single gene knockout studies, we found no significant enrichment or depletion in either population for gRNAs targeting SMAD4, EP300, PIAS1, or BAMBI.

(A) Percentage of either wild-type or CRISPR-targeted cells with positive staining for SARS-CoV-2 nucleocapsid protein at 2 days post-infection with SARS-CoV-2 at MOI 1. An average of 214 fields of view (range 151–295) were analyzed for each condition over 2 independent biologic replicates. Error bars indicate standard deviation. (B) Infectious titers of cellular supernatants collected at 2 days post-infection with SARS-CoV-2 at MOI 1, plotted as log 10 (TCID50) with error bars indicating standard deviation. For both experiments, asterisks indicate p < 0.05 relative to nontargeting controls by Student’s t-test.

We next analyzed HuH7 cells with CRISPR-mediated disruption of ACE2 modifiers for their sensitivity to SARS-CoV-2 infection. We found that those CRISPR-targeted cells with decreased ACE2 expression (SMAD4, EP300, and PIAS1) also exhibited a decrease in both the proportion of infected cells by SARS-CoV-2 nucleocapsid protein immunofluorescence ( Fig 5A ) and in viral infectivity by TCID50 ( Fig 5B ) at 48 hours post-infection. By contrast, cells with increased ACE2 expression (BAMBI) exhibited an increase in the proportion of SARS-CoV-2-infected cells and viral infectious titers. These findings, together with the correlation of results from our ACE2 screen and the HuH7.5 SARS-CoV-2 screen [ 35 ], support the relevance of our identified ACE2 modifiers to SARS-CoV-2 infection.

(A) Immunoblotting of lysates collected from HuH7 cells targeted by CRISPR with a gRNA against the indicated gene or a nontargeting (NT) control sequence. (B) Proportion of saponin-permeabilized cells exhibiting ACE2 staining above background after CRISPR targeting of the indicated gene in ACE2-enriched HuH7 cells. (C) Immunofluorescence microscopy of ACE2 staining (green) in ACE2-enriched HuH7 cells either untreated (WT) or targeted by CRISPR with a gRNA against the indicated gene or a nontargeting (NT) control sequence. Scale bar = 100 μm. (D) Quantification of relative ACE2 mRNA levels in the indicated cell lines by qRT-PCR of ACE2 mRNA, normalized to the mean Ct for a panel of control transcripts (RPL37, RPL38, and ACTB). Asterisks indicate p<0.05 relative to nontargeting controls by Student’s t-test. Individual data points in panels B and D represent technical replicates.

Among the genes with the largest functional influence on ACE2 abundance in our screen and in our single gene validation experiments were SMAD4 and multiple genes previously associated with SMAD4 signaling. These include EP300, encoding a histone acetyltransferase that is recruited by SMAD complexes to function as a coactivator for target genes [ 44 , 45 ]; PIAS1, encoding an E3 sumo ligase whose substrates include SMAD4 and which has been shown to modulate SMAD4-dependent TGF-β signaling [ 46 ]; and BAMBI, encoding a decoy receptor that negatively regulates TGF-β signaling through SMAD4 [ 47 ]. To establish the level of regulation for ACE2 surface protein by these modifiers, we analyzed HuH7 cells individually targeted for each gene by CRISPR. To distinguish between changes in cell surface ACE2 being due either to a change in total cellular levels or to altered trafficking of ACE2, we quantified total cellular ACE2 protein abundance both by immunoblotting of cell lysates ( Fig 4A ) and by flow cytometry of permeabilized cells ( Fig 4B ). In all cases, changes in total cellular ACE2 abundance were consistent with changes in surface ACE2 abundance. Immunofluorescence microscopy of permeabilized cells likewise demonstrated alterations in total ACE2 abundance with no striking redistribution of ACE2 in CRISPR-targeted cells ( Fig 4C ). Quantitative RT-PCR demonstrated changes in ACE2 mRNA levels in CRISPR-targeted cells that were consistent with the changes in surface-displayed ACE2 protein in these cells ( Fig 4D ). Together, these findings indicate that ACE2 regulation by SMAD4, EP300, PIAS1, and BAMBI in HuH7 cells is mediated at the mRNA level.

Multiple CRISPR screens have implicated cholesterol regulation as an important mediator of host cell interactions with SARS-CoV-2 [ 30 , 31 , 35 , 36 ]. In addition to canonical SREBP regulators, we also noted the identification of several genes in these studies that we had recently identified in a screen for regulators of low-density lipoprotein (LDL) uptake [ 43 ]. These included RAB10 and multiple components of the exocyst complex [ 31 , 35 , 36 ]. We therefore systematically examined the overlap among these CRISPR screens for LDL uptake, viral infection, and ACE2 abundance. We observed that gene disruptions which reduced LDL uptake were more likely than those that did not to also confer cellular resistance to SARS-CoV-2 both at 37°C and at 33°C ( S6A Fig ). By contrast, disruption of these same genes was not associated with a significant change in ACE2 abundance in either of our screens of wild-type or ACE2-enriched HuH7 cells ( S6B Fig ). Further supporting an ACE2-independent effect on viral infection, positive regulators of LDL uptake did not influence cellular sensitivity to HCoV-NL63 ( S6A Fig ). Among these, disruption of the canonical SREBP regulators SCAP, MBTPS1, and MBTPS2 conferred resistance to SARS-CoV-2 infection, with a similar effect for infection with the ACE2-independent HCoV-OC43 and HCoV-229E but not the ACE2-dependent HCoV2-NL63 ( S6C Fig ). These findings suggest that cholesterol regulatory host factors likely influence SARS-CoV-2 infection through an ACE2-independent mechanism.

We cross-referenced the data from our ACE2 CRISPR screen in HuH7 cells to that from a recently reported genome-wide CRISPR screen of HuH7.5 cytopathic effect during infection by SARS-CoV-2 or other coronaviruses [ 35 ]. As expected, ACE2 itself was among the top genes identified both as a positive regulator of ACE2 surface abundance in HuH7 cells in our screen and as a proviral host factor for both SARS-CoV-2 and another ACE2-dependent coronavirus, HCoV-NL63, but not for other coronaviruses, HCoV-229E and HCoV-OC43, that use alternative cellular receptors ( Fig 3C ) [ 41 , 42 ]. Overall, among the genes we identified as ACE2 regulators, those that promoted ACE2 abundance in our screen were more likely to sensitize HuH7.5 cells to SARS-CoV-2 and HCoV-NL63 infection, while those that repressed ACE2 abundance were more likely to confer resistance to SARS-CoV-2 and HCoV-NL63 infection ( Fig 3C and 3D ). A similar correlation for ACE2 modifiers with cytopathic effect was not observed for HCoV-229E and HCoV-OC43 ( Fig 3C and 3D ). These findings support the relevance of our ACE2 screen to viral infection and suggest that even modest changes in ACE2 expression may influence cellular susceptibility to viral cytopathic effect.

(A) Top 10 molecular function ontologies enriched in genes identified as ACE2 modifiers relative to all genes tested in the focused CRISPR library. (B) Network analysis of ACE2 modifiers identified in the focused CRISPR screens of HuH7 WT or ACE2-enriched cells. Significance testing for the number of detected protein-protein interactions relative to a randomly selected gene set was calculated by STRING. Only nodes detected with a significant interaction with another node are displayed. The width of the edge between nodes is weighted by the strength of the protein-protein interaction by STRING. (C) Heat map of log-transformed p-values for each ACE2 modifier identified in this study in comparison to log-transformed p-values for cytopathic effect during infection of HuH7.5 cells with the indicated coronaviruses [ 35 ]. Viruses are grouped by those whose entry is mediated by ACE2 (SARS-CoV-2, HCoV-NL63) or by an alternative receptor (HCoV-229E, HCoV-OC43) (D) Comparison of CPE effect with the indicated coronavirus [ 35 ] for groups of genes identified in this study as ACE2 positive or negative regulators. Positive Z-scores correspond to gene disruption conferring resistance to CPE. Asterisks indicates p<0.001 by Student’s t-test. A significant difference is observed for only those viruses whose cellular entry is ACE2-mediated.

(A-B) Percentage of surface ACE2-positive cells after CRISPR-mediated single gene disruption of the indicated target gene or a nontargeting control on either a HuH7 parental wild-type (A) or serially ACE2-enriched background (B). Candidate genes associated with positive or negative regulation of ACE2 in the CRISPR screen are highlighted in blue and red, respectively. ACE2-positive gates were defined on control unstained HuH7 wild-type cells and the proportion of ACE2-positive cells for each population is displayed for each of 3 independent biologic replicates. Error bars depict standard deviation. Asterisks indicated p-value <0.05 relative to nontargeting controls by Student’s t-test. Source data and statistical analysis are provided in S9 Table .

To validate our CRISPR screen results, we generated single gene CRISPR-targeting lentivirus constructs for 21 of the top-scoring hits from either screen and tested whether single gene disruption affected surface ACE2 levels. We found one gene, UROD, to be a false positive of the screen due to its disruption causing increased cellular autofluorescence in the detection channel of the Alexa Fluor 647-conjugated secondary antibody. UROD-targeted cells exhibited increased fluorescence in this channel even when the conjugated secondary antibody was omitted ( S5A Fig ), and replacement of this secondary antibody with an Alexa Fluor 488-conjugated secondary antibody abrogated the difference in ACE2 signal ( S5B Fig ). These findings are consistent with the known role of UROD in decarboxylating uroporphyrinogen-III, as cells with accumulation of uroporphyrin have been shown to exhibit increased fluorescence between 620–680 nm [ 39 ]. Of the remaining genes, we confirmed a significant effect of single gene CRISPR-targeting on ACE2 surface staining for 17 of 20 genes in either HuH7 wild-type or ACE2-enriched cells ( Fig 2 and S7 Fig , S9 Table ), with none of these genes showing increased autofluorescence and each replicating the change in ACE2 abundance with an Alexa Fluor 488-conjugated secondary antibody. Consistent with the stronger enrichment and depletion of hits in the CRISPR screen of ACE2-enriched cells relative to wild-type cells, validation testing of these ACE2-enriched cells also exhibited increased power to detect significant changes in ACE2 surface abundance ( S9 Table ).

We independently screened both HuH7 wild-type cells, in which ~3–5% of cells express detectable ACE2, and HuH7 cells derived by serial enrichment with 3 rounds of FACS to contain ~60–70% ACE2-positive cells ( Fig 1B and 1C ). Cells were then transduced at >200X coverage with the customized CRISPR library, passaged for 14 days to allow for target gene disruption and turnover of residual protein, and sorted by flow cytometry into selected populations. In the screen of wild-type cells, the ~3–5% of ACE2-positive cells were collected in one population and the median ~10% of ACE2-negative cells were collected in another. In the screen of ACE2-enriched cells, the top ~10% of cells with the greatest ACE2 abundance were collected along with the median ~10% of ACE2-negative cells. The relative abundance of every gRNA in each population was quantified by deep sequencing. Quality control analysis demonstrated improved depth of library coverage and discrimination of control gRNAs in the focused CRISPR screens relative to the primary genome-wide screens ( S2 and S3 Figs). As expected, ACE2 itself was identified among the top positive regulators of ACE2 abundance in both screens, while gRNAs targeting many other genes without a known role in ACE2 regulation exhibited significant enrichment or depletion in sorted populations ( Fig 1D and 1E , S5 and S6 Tables). In total, we identified 19 high-confidence positive regulators and 16 high-confidence negative regulators of ACE2 surface abundance in HuH7 cells (FDR<0.05, absolute value of log 2 fold-change >1). Analysis of gRNA representation in the FACS input samples relative to the plasmid pool revealed no bias in genes influencing cellular viability or proliferation ( S4 Fig and S7 Table ). Supporting the reproducibility of the identified ACE2 modifiers, there was a very high degree of concordance between the results of the independent screens of wild-type and ACE2-enriched cells ( Fig 1F , r = 0.84).

(A) Schematic of ACE2 CRISPR screening strategy. A total of 833 candidate ACE2 modifiers were selected from a variety of sources and targeted with a high-resolution CRISPR library containing 15 gRNAs per gene. This library was then used to screen in parallel both wild-type and ACE2-enriched HuH7 cells. Selection of individual cells was performed based on surface ACE2 abundance by FACS and individual gRNA abundance in each sorted subpopulation was quantified by massively parallel sequencing. (B) Flow cytometry plots of ACE2 surface abundance for wild-type or serially ACE2-enriched HuH7 cells, with or without transduction of a lentiviral CRISPR construct with a ACE2-targeting or control nontargeting (NT) gRNA. ACE2-positive gates were established on unstained wild-type cells. (C) Immunoblot for ACE2 and β-actin in lysates prepared from the same cell populations as in (B). (D-E) Volcano plots of MAGeCK Robust Rank Aggregation scores relative to gene-level gRNA log 2 fold-change for each gene tested in the focused CRISPR library in screens of ACE2 surface abundance in HuH7 wild-type (D) or ACE2-enriched (E) cells. Genes with FDR <0.05 and absolute log 2 fold-change >1 are highlighted, with positive regulators in blue and negative regulators in red. (F) Correlation of gene-level aggregate gRNA log 2 fold-change between the independent secondary screens of HuH7 wild-type and ACE2-enriched cells. Genes identified in both screens with FDR <0.05 and absolute log 2 fold-change >1 are highlighted and annotated.

To identify functional regulators of ACE2 surface abundance, we first defined a list of candidate genes from several different sources ( Fig 1A and S1 Table ). These included modifiers of SARS-CoV-2 cytopathic effect in recently reported CRISPR screens [ 29 – 36 ], genes in proximity to human GWAS loci associated with COVID-19 susceptibility [ 37 ] ( S2 Table ), genes we previously identified by RNA-seq whose correlation was associated with ACE2 expression in sorted HuH7 cells [ 28 ], candidates identified in our own pilot genome-wide CRISPR screens for ACE2 abundance ( S1 Fig and S3 and S4 Tables), and a set of hypothesis-driven manually selected genes. In total, we targeted 833 genes with 15 gRNAs per gene. Amplicons of the pooled gRNA sequences were inserted into the pLentiCRISPRv2 construct [ 38 ] and the diversity and representation of the resulting plasmid pool were verified by deep sequencing.

Discussion

CRISPR screening is a powerful forward genetic tool for the unbiased and high-throughput functional interrogation of the human genome. This approach has been successful in the identification of host factors involved in the molecular pathogenesis for many viruses [49], including SARS-CoV-2 [29–36]. For most identified host factors, however, the molecular basis for interaction with SARS-CoV-2 remains unclear. Consistently, these screens have lent further support to the central importance of ACE2 in SARS-CoV-2 infection, with guide RNAs targeting ACE2 among those conferring the greatest resistance to cytopathic effect. Our current study, enabled by our recently reported identification of a sensitive and specific platform for ACE2 flow cytometry [28], complements the former studies to systematically probe the genetic regulators of ACE2 surface abundance in HuH7 and Calu-3 cells.

An important caveat of our study is the focused nature of our library, limited to high-resolution functional testing of 833 candidate genes. Our initial attempts at genome-wide screening did nominate candidate ACE2 regulators that were included in our focused library, but these pilot studies were limited by an experimental bottleneck at the cell sorting stage with inadequate depth of coverage to support definitive conclusions about each gene tested. Nevertheless, for those genes interrogated by the focused library, several observations support a high degree of confidence in the functional significance of each gene to ACE2 surface abundance. First, the internal control ACE2-targeting gRNAs were clearly depleted in ACE2-positive cells. Second, the genes that were identified as regulating ACE2 abundance exhibited a robust statistical enrichment or depletion for several independent gRNAs. Third, we observed a very high degree of concordance in the degree of enrichment or depletion observed for each gene between the independent screens of wild-type or ACE2-enriched HuH7 cells. Fourth, identified genes were highly enriched for related functional annotations and protein-protein interactions and exhibited significant overlap with previously identified modulators of HuH7.5 cellular susceptibility to ACE2-dependent coronaviruses. Finally, of the screen hits we selected for single gene validation testing, most (17/20 in HuH7 cells, 5/5 in Calu-3 cells) were confirmed to regulate ACE2 surface abundance.

Given the sensitivity of our screens to detect subtle influences on ACE2 abundance, a number of negative findings are noteworthy. Cholesterol regulatory genes in the SREBP pathway have been implicated in SARS-CoV-2 infection, but we did not detect an association with ACE2 abundance either by focused analysis of canonical SREBP regulators or by systematic comparison to all HuH7 LDL uptake regulators identified in our previously reported genome-wide CRISPR screen [43]. These findings are consistent with the association of SREBP regulators with ACE2-independent coronaviruses and the lack of association with ACE2-dependent HCoV-NL63. Angiotensin receptor blockade has been postulated to increase surface ACE2 expression, raising concern for potential increased COVID-19 risk for patients on ACE inhibitors or angiotensin receptor blockers. We did not however detect any association of AGTR1 or AGTR2 with ACE2 abundance. We also did not observe an effect for disruption of the Ang [1–7] receptor MAS1, arguing against feedback regulation of this axis in these cells. Although estrogen regulation of ACE2 expression has been proposed to mediate the sexual dimorphism of COVID-19 susceptibility, and estrogen receptor binding sites have been identified near the ACE2 locus [26], we did not detect an effect of ESR1 or ESR2 disruption on ACE2 abundance. We also did not detect a large effect for any candidate genes in proximity to variants associated with COVID-19 susceptibility by GWAS. Each of these negative results should be interpreted with caution, however, both because of the potential cell type specificity of ACE2 regulation and the intrinsic limitation of CRISPR screens for functionally redundant genes, essential genes, or compensatory mechanisms in mutant cells.

SMAD4, which we identified as a positive regulator of ACE2 gene expression in HuH7 cells, is a common mediator of TGF-β signaling [50]. In the canonical TGF-β signaling pathway, ligand binding signals through receptor-regulated SMADs that complex with SMAD4 to trigger its translocation into the nucleus and binding to specific DNA regulatory elements. Identification of the upstream signals driving SMAD4-dependent ACE2 expression in HuH7 cells is complicated by the diversity of TGF-β superfamily receptors and ligands including bone morphogenic proteins, nodal, and activin. Our screen did identify ACVR1, encoding activin A receptor type 1, as a positive regulator of ACE2, and this gene was similarly identified as proviral for SARS-CoV-2 and HCoV-NL63 in HuH7.5 cells [35]. The limited scope of our screen, however, precludes a comprehensive cataloguing of other potential upstream mediators of SMAD4-dependent ACE2 expression. Supporting the physiologic relevance of these findings, a recent study identified ChIP-seq peaks for SMAD4 in intestinal epithelial cells that overlapped putative enhancer regions near the murine Ace2 locus, and furthermore detected reduced Ace2 intestinal transcript levels upon tissue-specific Smad4 deletion [51].

Our findings are consistent with a growing body of evidence supporting the cell type specificity of genetic interactions relevant to SARS-CoV-2 infection. Although ACE2 itself was the top hit in each of our screens of HuH7 or Calu-3 cells, there was otherwise very little concordance between these screens. Independent CRISPR screens of SARS-CoV-2 cytopathic effect in different cell types have likewise identified limited overlap between studies. Although we observed significant overlap between ACE2 modifiers we identified in HuH7 cells with recently identified SARS-CoV-2 infection modifiers in HuH7.5 cells [35], we observed minimal overlap with SARS-CoV-2 infection modifiers identified in similar genome-scale screens of other cell lines [29,30,32,34,36]. For some of these other studies, this discrepancy may be attributable to their use of cell lines with heterologous overexpression of ACE2 [30,32,36]. In other cases, however, SARS-CoV-2 infection was tested in cell lines with endogenous ACE2 expression [29, 34] and yet still little overlap was observed. For example, we did not detect an effect of HMGB1 disruption on ACE2 abundance in either HuH7 or Calu-3 cells, despite this gene being previously implicated in SARS-CoV-2 cytopathic effect in Vero E6, HuH7.5, and Calu-3 cells and in ACE2 mRNA and protein expression in Vero E6 cells. It is possible that this discrepancy may be related to HGMB1 exerting pleiotropic ACE2-dependent and ACE2-independent effects on SARS-CoV-2 infection in different cell lines. It is also possible that this and other discrepancies across screens may be related to technical differences in culture conditions, CRISPR library design, infection parameters, or functional readout. For example, another SARS-CoV-2 screen of HuH7 cells [34] did not identify ACE2, which was attributed to low levels of ACE2 expression and implies that the conditions of this screen would not enable the detection of ACE2 positive regulators. These technical considerations aside, our findings strongly suggest that ACE2 regulatory networks are highly cell type specific, a conclusion that is further supported by small molecule inhibitor studies [52] and by the tissue-specific patterns of ACE2 promoter usage [20]. The molecular basis for this cell type specificity is not established by our study but may be related to the differential activity of signaling pathways and/or ACE2 cis-regulatory elements in distinct tissues.

In summary, we have applied a functional genomic approach to dissect the regulatory networks of ACE2 protein expression. We have identified many previously unrecognized genetic modifiers of ACE2 expression and a putative mechanism for genes previously implicated in SARS-CoV-2 infection. Our findings highlight the critical influence of cell type on ACE2 regulation and nominate pathways for host-targeted therapeutic development.

[END]

[1] Url: https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1010377

(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/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.