key: cord-0281140-ofavwj52
authors: Loo, Lipin; Waller, Matthew A.; Cole, Alexander J.; Stella, Alberto Ospina; Moreno, Cesar L.; Denes, Christopher E.; Hamoudi, Zina; Chung, Felicity; Aggarwal, Anupriya; Low, Jason K. K.; Patel, Karishma; Siddique, Rezwan; Mackay, Joel; Turville, Stuart; Hesselson, Daniel; Neely, G. Gregory
title: LRRC15 suppresses SARS-CoV-2 infection and controls collagen production
date: 2021-11-10
journal: bioRxiv
DOI: 10.1101/2021.11.09.467981
sha: 6a947de2c985bfd8fad1fdea8a1f051072962336
doc_id: 281140
cord_uid: ofavwj52

Although ACE2 is the primary receptor for SARS-CoV-2 infection, a systematic assessment of factors controlling SARS-CoV-2 host interactions has not been described. Here we used whole genome CRISPR activation to identify host factors controlling SARS-CoV-2 Spike binding. The top hit was a Toll-like receptor-related cell surface receptor called leucine-rich repeat-containing protein 15 (LRRC15). LRRC15 expression was sufficient to promote SARS-CoV-2 Spike binding where it forms a cell surface complex with LRRC15 but does not support infection. Instead, LRRC15 functioned as a negative receptor suppressing both pseudotyped and live SARS-CoV-2 infection. LRRC15 is expressed in collagen-producing lung myofibroblasts where it can sequester virus and reduce infection in trans. Mechanistically LRRC15 is regulated by TGF-β, where moderate LRRC15 expression drives collagen production but high levels suppress it, revealing a novel lung fibrosis feedback circuit. Overall, LRRC15 is a master regulator of SARS-CoV-2, suppressing infection and controlling collagen production associated with “long-haul” COVID-19. In Brief Using pooled whole genome CRISPR activation screening, we identify the TLR relative LRRC15 as a novel SARS-CoV-2 Spike interacting protein. LRRC15 is not a SARS-CoV-2 entry receptor, but instead can suppress SARS-CoV-2 infection. LRRC15 is expressed by lung fibroblasts and regulates both collagen production and infection of ACE2-expressing target cells. This may provide a direct link between SARS-CoV-2 particles and lung fibrosis seen in “long-haul” COVID-19 patients. Highlights Whole genome CRISPR activation screening implicates the TLR relative LRRC15 in SARS-CoV-2 Spike binding LRRC15 suppresses live SARS-CoV-2 virus infection LRRC15 is expressed in lung fibroblasts and sequesters virus while controlling collagen production LRRC15 can act as a master regulator of infection and fibrosis, potentially controlling SARS-CoV-2 infection outcomes and “long-haul” COVID-19

In Brief 27 28

Using pooled whole genome CRISPR activation screening, we identify the TLR relative 29 LRRC15 as a novel SARS-CoV-2 Spike interacting protein. LRRC15 is not a SARS-CoV-2 30 entry receptor, but instead can suppress SARS-CoV-2 infection. LRRC15 is expressed by lung 31 fibroblasts and regulates both collagen production and infection of ACE2-expressing target cells. 32

This may provide a direct link between SARS-CoV-2 particles and lung fibrosis seen in "long-33 haul" COVID-19 patients. Although ACE2 is the primary receptor for SARS-CoV-2 infection, a systematic assessment of 48 factors controlling SARS-CoV-2 host interactions has not been described. Here we used whole 49 genome CRISPR activation to identify host factors controlling SARS-CoV-2 Spike binding. The 50

top hit was a Toll-like receptor-related cell surface receptor called leucine-rich repeat-containing 51

protein 15 (LRRC15). LRRC15 expression was sufficient to promote SARS-CoV-2 Spike 52 binding where it forms a cell surface complex with LRRC15 but does not support infection. 53

Instead, LRRC15 functioned as a negative receptor suppressing both pseudotyped and live 54 SARS-CoV-2 infection. LRRC15 is expressed in collagen-producing lung myofibroblasts where 55 it can sequester virus and reduce infection in trans. Mechanistically LRRC15 is regulated by 56 TGF-β, where moderate LRRC15 expression drives collagen production but high levels suppress 57 it, revealing a novel lung fibrosis feedback circuit. Overall, LRRC15 is a master regulator of 58 SARS-CoV-2, suppressing infection and controlling collagen production associated with "long-59

haul" COVID-19. 60 61

Keywords 62 LRRC15, SARS-CoV-2, COVID-19, Spike, CRISPR activation screen, gain of function, long-63 haul COVID-19 64 65

Introduction 66 67

The Coronavirus 2019 (COVID-19) pandemic, caused by SARS-CoV-2, represents the greatest 68 public health challenge of our time. As of November 2021, there have been over 250,000,000 69

reported cases of COVID-19 globally and in excess of 5,000,000 subsequent deaths (WHO). 70

SARS-CoV-2 shows high sequence similarity (79.6%) with severe acute respiratory syndrome 71 coronavirus (SARS-CoV-1), and because of this similarity, angiotensin-converting enzyme 2 72 (ACE2), the primary entry receptor for SARS-CoV-1, was quickly identified as the SARS-CoV- To identify novel host factors that can influence cellular interactions with the SARS-CoV-2 78

Spike protein, we used a whole genome CRISPR activation approach. Using the Calabrese 79

Human CRISPR Activation Pooled Library (Sanson et al., 2018) , we identified a TLR-related 80 cell surface receptor named leucine-rich repeat-containing protein 15 (LRRC15) as a novel 81 SARS-CoV-2 Spike binding protein in three independent whole genome screens. LRRC15 was 82 confirmed to promote Spike binding via flow cytometry, immunoprecipitation and confocal 83 microscopy. Mechanistically, LRRC15 is not a SARS-CoV-2 entry receptor, instead ectopic 84 LRRC15 expression was sufficient to inhibit SARS-CoV-2 pseudovirus infection and can also 85 suppress live SARS-CoV-2 infection. LRRC15 is primarily expressed in innate immune barriers 86 Table 1) . Moreover, we conducted 2 additional 133 screens under slightly different conditions, and in all screens our top hit was LRRC15 134

(Supplementary Figure 2A-F) . 135 136 We expressed the LRRC15 sgRNAs that were hits in our screens in HEK293T-CRISPRa cells 137 and confirmed that they induce expression of LRRC15 (~approximately 2000 fold induction, 138

Supplementary Figure 2G) . Moreover, LRRC15-overexpressing cells dramatically increased 139 SARS-CoV-2 Spike488 binding, with LRRC15 sgRNA 1 inducing binding to levels comparable 140 to cells overexpressing ACE2 sgRNA3 ( Figure 2E ). LRRC15 overexpression did not itself 141 upregulate ACE2 transcription, suggesting the increased Spike binding in LRRC15-expressing 142 cells is independent of ACE2 upregulation (Supplementary Figure 2H) . LRRC15 in SARS-CoV-2 Spike binding and ensure the interaction was not an artifact of our 160

CRISPRa strategy, we transfected LRRC15-GFP cDNA into HEK293T cells and observed 161

Spike647 binding by flow cytometry. There are two reported isoforms of LRRC15 (LRRC15_1 162 and LRRC15_2), with LRRC15_1 having 6 additional amino acids at the N-terminus. Although 163 cells transfected with GFP alone showed no binding to Spike647, cells expressing LRRC15 164 isoform 1 or 2 both showed strong Spike binding ( Figure 3D ). While LRRC15-dependent Spike 165 binding was higher than cells stably expressing ACE2 (62.1% and 64.5% vs. 48.8%), co-166 expression of LRRC15 with ACE2 was additive resulting in 86.3% positive (LRRC15_1) or 167 83.8% positive (LRRC15_2) cells ( Figure 3E) . Interestingly, all cells (100%) stably expressing 168 ACE2 and TMPRSS2 bound Spike647 regardless of LRRC15 expression ( Figure 3F ). However, 169

LRRC15 expression in HEK293T-ACE2-TMPRSS2 cells still enhanced the amount of cell 170 surface Spike647 bound by each cell as measured by mean fluorescence intensity ( Figure 3G ). 171

Moreover, both LRRC15 isoforms colocalized with Spike647 ( Figure 3H ). To independently 172

confirm an interaction between LRRC15 and SARS-CoV-2 Spike protein, we added Spike to 173 LRRC15-expressing cells and then immunoprecipitated LRRC15. While control GFP transfected 174

HEK293T cells did not show any signal at the size predicted for Spike (~200 kDa, (Hsieh et al., 175 2020)) (Supplementary Figure 3B-C) , when we expressed and then pulled down either 176 LRRC15_1 or LRRC15_2, in both cases we co-immunoprecipitated Spike protein in the eluate 177 ( Figure 3I ). Taken together, these data show that LRRC15 expression is sufficient to confer SARS- CoV-2 Spike binding to HEK293T cells, and LRRC15 can further enhance Spike  179  interactions in the presence of ACE2 and TMPRSS2.  180  181  LRRC15 is not a SARS-CoV-2 entry receptor but can suppress Spike-mediated entry and live  182  virus infection  183  184 We next asked if LRRC15 can act as a receptor for SARS-CoV-2 and mediate viral entry. For 185 this we used a SARS-CoV-2 pseudotyped lentivirus system (SARS-CoV-2 pseudovirus) that 186 displays the SARS-CoV-2 Spike protein and carries a luciferase reporter ( Figure 4A , 187

Supplementary Figure 4A) . Surprisingly, LRRC15 did not confer SARS-CoV-2 pseudovirus 188 tropism in minimally infectable HEK293T cells ( Figure 4B ). We then tested if LRRC15 189 expression impacted infection of HEK293T cells expressing ACE2 and TMPRSS2 ( Figure 4C , 190 HEK293T-ACE2 cells shown in Supplementary Figure 4B- to 24% suppression at the highest viral dose (5x10 particles) ( Figure 4C ). Next we tested if 195 LRRC15 expression can also suppress viral replication and cytopathic effect in a live CoV-2 infection system. HEK293T-ACE2-TMPRSS2 cells were infected with increasing doses 197 of SARS-CoV-2 (D614G and Delta variants, Figure 4D ) and cell death was assessed 48 h later. 198

Ectopic expression of LRRC15 significantly inhibited D614G infection (two-way ANOVA, 199 p<0.05) but not the Delta variant ( Figure 4E and 4F At the tissue level, LRRC15 RNA is most abundant in the placenta, with expression also found in 206 skin, tongue, tonsils, and lung (Uhlén et al., 2015) . At the single cell level, we used the COVID-207

19 Cell Atlas data set to confirm LRRC15 expression in placenta decidua stromal cells (Vento-208  Tormo et Figure 5C) . Together, these data support our in vitro observations that 218 LRRC15 does not mediate SARS-CoV-2 infection but may instead act as an innate immune 219 barrier. In contrast, ACE2 was detected primarily in uninfected type I (AT1) and (AT2) alveolar 220 epithelium (Figure 5D) , and SARS-CoV-2-infected alveolar epithelium ("Other epithelial cells") 221 that lost AT1/2 markers and upregulated ribosomal transcripts consistent with viral infection and 222 cell death. 223 224

As single cell data showed an absence of SARS-CoV-2 mRNA in LRRC15 + fibroblasts, we next 225 tested infectivity of lung fibroblasts (IMR90) with SARS-CoV-2 pseudovirus. IMR90 fibroblasts 226 express LRRC15 endogenously (Supplementary Figure 5G) and possess a low level of intrinsic 227 SARS-CoV-2 Spike binding activity (Supplementary Figure 5H) . Endogenous LRRC15 228 expression was confirmed via Western blot (Supplementary Figure 6A-B) . Transfection of 229

LRRC15-GFP cDNA in these fibroblasts further enhanced Spike binding capacity ( Figure 5E ). 230

However, similar to WT HEK293T cells, ectopic expression of LRRC15 was not sufficient to 231

confer SARS-CoV-2 pseudovirus tropism ( Figure 5F ), confirming that LRRC15 is not a SARS-232

CoV-2 entry receptor. Since LRRC15 and ACE2 expression are mutually exclusive in the lung, 233 we next investigated whether LRRC15 + fibroblasts could act in trans to sequester SARS-CoV-2 234

pseudovirus and suppress infection of the highly permissive HEK293T-ACE2-TMPRSS2 line. 235

Indeed, co-incubating permissive HEK293T-ACE2-TMPRSS2 cells with LRRC15 + fibroblasts 236 could suppress SARS-CoV-2 pseudovirus transduction ( Figure 5G) and participate in tissue repair and fibrosis (Buechler et al., 2021) . We also observed lung 251 LRRC15 + myofibroblasts in multiple COVID-19 patient data sets, and these cells express 252 collagen ( Figure 6B ). LRRC15 is upregulated in response to proinflammatory cytokines like 253

IL1β, IL6, and TNF (Satoh et al., 2002) , and TGFβ also upregulates LRRC15 ( Figure 6C ) and 254 COL1A1 transcripts ( Figure 6D ). Together, LRRC15 is expressed on collagen producing 255 fibroblasts both in vitro and in the lung of COVID-19 patients and may regulate lung fibrosis. 256 257

To directly investigate the relationship between LRRC15 and collagen, we expressed low (Lo) or 258 high (Hi) levels of LRRC15 in fibroblasts (or provided GFP as a transfection control) and then 259 evaluated COL1A1 expression ( Figure 6E) . Surprisingly, Lo LRRC15 promoted COL1A1 260 expression while Hi LRRC15 did not ( Figure 6F ). This bimodal regulation was confirmed with 261

Western blotting ( Figure 6G , quantified in H-I, full blots in Supplementary Figure 6 ). When 262 taken together, our working model is that LRRC15 expression is induced by inflammatory 263 cytokines in COVID-19 lung fibroblasts, where it acts as an innate antiviral barrier that can 264 sequester SARS-CoV-2 and decrease infection. As infection resolves, and the proinflammatory 265

context of the lung changes, LRRC15 expression would reduce, and this would then switch 266 LRRC15 + fibroblasts from antiviral role to instead promote lung repair ( Figure 6J) Data and code availability 362 CRISPR screen raw read counts have been deposited at GSE186475 and are publicly available as 363 of the date of publication. CRISPR screen analysis is shown in Figure 2 and Supplementary 364 Figure S2 . CRISPR screen output is deposited in Supplementary Table S1 . This paper also 365 analyzes existing publicly available single cell RNA-sequencing data. The accession numbers for 366 these datasets are listed in the Key Resources Table. All data reported in this paper will be shared 367 by the lead contact upon request. This paper does not report original code. Any additional 368 information required to reanalyze the data reported in this paper is available from the lead 369 contact upon request. 370 371

Experimental SARS-CoV-2 Spike protein production 438

The expression construct for recombinant soluble trimeric SARS-CoV-2 spike protein (residues 439 1-1208, complete ectodomain) was generously provided by Dr Florian Krammer (Icahn School  440 of Medicine, Mt Sinai). This protein was used for the initial setup of the screen (shown in Figure  441 1) and in one CRISPRa screen (Screen 2). This construct includes the SARS-CoV-2 spike native 442 signal peptide (residues 1-14) to target the recombinant protein for secretion, stabilising proline 443 substitutions at residues 986 and 987, substitution of the furin cleavage site (residues 682-685) 444

with an inert GSAS sequence, and a C-terminal His6-tag to enable affinity purification. For cell mixing experiments, increasing proportions of HEK293T-ACE2 cells (0%, 1%, 20%, 520 50%, 80% and 100%) were combined with decreasing proportions of wildtype (WT) HEK293T 521 cells (100%, 99%, 80%, 50%, 20%, 0%) to a total of 10 6 cells per sample. These samples were 522 incubated with 50 ߤ g/mL Spike488 as described above and analyzed using the Cytek Aurora 523 (Cytek Biosciences). 524

To confirm the validity of this assay in detecting binding in cells expressing CRISPRa 525 machinery, a clonal line of HEK293T with stable expression of a plasmid encoding dCas9-VP64 526

and SAM system helper proteins (pPB-R1R2_EF1aVP64dCas9VP64_T2A_MS2p65HSF1-527

IRESbsdpA) (HEK293T-CRISPRa) was transduced with lentivirus carrying ACE2 sgRNA 1 or 528 non-targeting control sgRNA. These cells were then incubated with Spike488 as previously 529 described and analyzed on the Cytek Aurora (Cytek Biosciences). 530 531 CRISPR activation screening 532

HEK293T-CRISPRa cells were transduced with concentrated Human CRISPR activation pooled 533 library set A (Addgene #92379)-carrying lentivirus at a multiplicity of infection (MOI) of 534 approximately 0.5. Cells were selected on puromycin dihydrochloride (Gibco) at a concentration 535 of 1.6 µg/mL for 3 days (screen 1 and 2). 3x10 7 cells (>500 cells/guide) were incubated with 536

Spike488 for 30 min at 4°C, washed to remove excess spike protein, and sorted for increased 537

Alexa Fluor 488 intensity using the BD FACSMelody Cell Sorter (BD Biosciences). Gates for 538 flow assisted cytometric sorting were set using non-targeting control (NTC) sgRNA-transduced 539 cells as a negative control and ACE2 sgRNA-transduced cells as a positive control, both of which 540 had been incubated with Spike488 under the same conditions as stated previously. Unsorted cells 541

were maintained separately so as to be used as a diversity control. Cells were expanded and 542 2x10 6 cells were then collected for genomic DNA (gDNA) extraction for sorted samples and 543 3x10 7 for the unsorted diversity control. Remaining diversity control cells were re-seeded and 544 once again incubated with Spike488 under the same conditions as stated previously (screen 3). 545

These Spike-incubated cells were sorted again but selected on puromycin for eight days prior to 546 expansion and collection of 1x10 7 cells from both the sorted cell population and the unsorted 547 diversity control population for gDNA extraction. Gating strategy is shown in Supplementary 548 Figure 1B qPCR. These cells were then transfected with empty TurboGFP control and LRRC15-TurboGFP 712 (Lipofectamine LTX with plus reagent (ThermoScientific)). Cells were checked for Spike 713 binding activity by incubation with Spike647 and detection via flow cytometry 24 h post-714 transduction. Then, these fibroblasts were infected with SARS-CoV-2 pseudovirus as described 715

above and luciferase luminescence were compared to HEK293T-ACE2-TMPRSS2 cells. 716 717

For viral immobilization assay, 6,000 HEK293T-ACE2-TMPRSS2 cells were incubated with 718 12,000 fibroblasts expressing GFP or LRRC15-GFP and SARS-CoV-2 pseudovirus (5x10 8 719 particles in polybrene, as described above) for an hour at 37°C before seeding in a 96-well plate. 720 Transduction was quantified as described above and luminescence was normalized to GFP 721 controls. 722 723

Quantification of collagen production in fibroblast 724 5 ng/mL of TGF-ߚ (R&D Systems) was added to fibroblasts and incubated for 24 h before 725 collection for LRRC15 and COL1A1 RT-qPCR. 

Mobilization of HIV spread by diaphanous 2 dependent filopodia in 1002 infected dendritic cells

Genome-wide CRISPR screening identifies 1005 TMEM106B as a proviral host factor for SARS-CoV-2

Lung transplantation for patients with 1008 severe COVID-19

EnhancedVolcano (Bioconductor)

Cross-tissue organization of 1012 the fibroblast lineage

The telomere deprotection 1014 response is functionally distinct from the genomic DNA damage response

Pooled extracellular receptor-1017 ligand interaction screening using CRISPR activation

Identification of Required Host Factors for 1020 SARS-CoV-2 Infection in Human Cells

COVID-19 tissue 1023 atlases reveal SARS-CoV-2 pathology and cellular targets

The extracellular leucine-rich repeat superfamily; a comparative 1026 survey and analysis of evolutionary relationships and expression patterns

Targeting lentiviral vector expression to hepatocytes limits transgene-specific immune response 1030 and establishes long-term expression of human antihemophilic factor IX in mice

Pulmonary fibrosis and COVID-19: the 1033 potential role for antifibrotic therapy

Bidirectional genome-wide 1036 CRISPR screens reveal host factors regulating SARS-CoV-2, MERS-CoV and seasonal HCoVs

Integrated analysis of multimodal single-cell data

The innate immune response to bacterial flagellin is 1043 mediated by Toll-like receptor 5

A Cellular Anatomy of the Normal Adult 1046 Human Prostate and Prostatic Urethra

A Multibasic Cleavage Site in the 1048 Spike Protein of SARS-CoV-2 Is Essential for Infection of Human Lung Cells

Compact and highly active next-generation 1052 libraries for CRISPR-mediated gene repression and activation

Structure-based design of prefusion-stabilized 1055 SARS-CoV-2 spikes

SARS-CoV-2 infection of the oral cavity and saliva

ACE2 receptor expression and severe acute 1061 respiratory syndrome coronavirus infection depend on differentiation of human airway epithelia

Highly accurate protein structure prediction 1065 with AlphaFold

Optimisation of 1067 a multipartite human immunodeficiency virus based vector system; control of virus infectivity 1068 and large-scale production

A crucial role of angiotensin converting enzyme 2 (ACE2) in SARS 1071 coronavirus-induced lung injury

Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 1074 receptor

Angiotensin-converting enzyme 2 is 1077 a functional receptor for the SARS coronavirus

MAGeCK enables robust identification of essential genes from genome-1080 scale CRISPR/Cas9 knockout screens

Genomic characterisation and epidemiology of 2019 novel coronavirus: implications 1083 for virus origins and receptor binding

scRNA-seq assessment of the human lung, spleen, and esophagus tissue stability after cold 1087 preservation

Single-Cell Analysis of Crohn's Disease Lesions 1090 Identifies a Pathogenic Cellular Module Associated with Resistance to Anti-TNF Therapy

A molecular single-cell lung atlas of lethal COVID-19

Tetherin inhibits retrovirus release and is 1096 antagonized by HIV-1 Vpu

Tumor antigen LRRC15 impedes adenoviral 1098 infection: implications for virus-based cancer therapy

A cell atlas of human thymic 1101 development defines T cell repertoire formation

The spatial landscape of lung pathology during COVID-1104 19 progression

The Rag GTPases bind raptor and mediate amino acid signaling to 1107 mTORC1

Optimized libraries for CRISPR-Cas9 1110 genetic screens with multiple modalities

A novel member of the leucine-rich repeat 1112 superfamily induced in rat astrocytes by beta-amyloid

Genome-Scale 1116 Identification of SARS-CoV-2 and Pan-coronavirus Host Factor Networks

LRRC15 1120 mediates an accessory interaction with the SARS-CoV-2 spike protein

Fast, scalable generation of high-quality protein multiple 1124 sequence alignments using Clustal Omega

PrimerBank: a resource of human and 1126 mouse PCR primer pairs for gene expression detection and quantification

Single-cell transcriptomes of the human skin reveal age-1130 related loss of fibroblast priming

Molecular Evolutionary Genetics Analysis 1132 (MEGA) for macOS

SARS-CoV-2 neutralizing antibodies: 1135 Longevity, breadth, and evasion by emerging viral variants

Proteomics. Tissue-based map 1138 of the human proteome

UniProt: the universal protein knowledgebase in 2021. Nucleic 1140 Acids Res

Single-cell reconstruction 1143 of the early maternal-fetal interface in humans

A cellular census of human lungs identifies novel 1146 cell states in health and in asthma

Integrative analysis of pooled CRISPR genetic screens using 1149

Structural and Functional Basis of SARS-CoV-2 Entry by Using Human ACE2

Genetic Screens 1155 Identify Host Factors for SARS-CoV-2 and Common Cold Coronaviruses

LRRC15 1158 promotes osteogenic differentiation of mesenchymal stem cells by modulating p65

cytoplasmic/nuclear translocation

Genome-wide CRISPR 1162 Screens Reveal Host Factors Critical for SARS-CoV-2 Infection

Cryo-EM structure of the 2019-nCoV spike in the prefusion 1165 conformation

A pneumonia outbreak associated with a new coronavirus of 1168 probable bat origin

Genome-wide CRISPR activation screen identifies candidate receptors for SARS-1171

CoV-2 entry