Mechanistic insights into TLR7-mediated clinical outcome in COVID-19 and the potential modulatory role of N-acetylcysteine

Men with loss-of-function (LOF) variants of toll-like receptor 7 (TLR7) are more susceptible to critical COVID-19. We aimed to determine the pathogenicity and binding a�nity changes of TLR7 variants towards SARS-CoV2 RNA, MyD88, imiquimod, and N-acetylcysteine (NAC). TLR7 deleterious mutations increase the risk of critical COVID-19 by 16-folds (95% CI: 2.40–106.73, p = 0.005). LOF-TLR7 variants had impaired SARS COV-2 viral RNA sensing (-292.09 ± 11.86 Kcal/mol) compared to hypofunctional (-304.65 +/- 8.84 Kcal/mol), and neutral (-310.01 ± 5.29 Kcal/mol) TLR7 variants (p < 0.00001). The hypofunctional TLR7 variants had the lowest binding a�nity (-270.31 ± 8.35 Kcal/mol) to MyD88, while the LOF variants had the highest binding a�nity (-284.87 ± 13.20 Kcal/mol, F:4.66, p = 0.02). The TLR7 variants in critical COVID-19 had a higher binding a�nity to N-acetylcysteine (NAC) than those in severe COVID-19 (-7.68 ± 0.68 vs.-7.18 ± 0.35, p = 0.03). The binding a�nity of imiquimod to TLR7 variants is not altered by the TLR7 functionality (F = 1.09, p = 0.36). To conclude, men with deleterious TLR7 mutations are highly prone to critical COVID-19 due to loss of viral RNA sensing ability of TLR7, which impairs anti-viral response. NAC can circumvent the impact of these TLR gene variants and elicit anti-viral responses by relaying TLR7-MyD88 signaling while imiquimod was not effective.

Sixteen men with deleterious TLR7 variants exhibited critical COVID-19 due to the inability of plasmacytoid dendritic cells to produce type 1 interferons (IFNs) [5].Four men from two families with severe COVID-19 requiring mechanical ventilation had loss-of-function (LOF) variants of TLR7 [6], which reduced expression of IRF7, IFNB1, and ISG15 upon stimulation with the TLR7 agonist imiquimod leading to impaired type I IFN signaling [6].Two out of fourteen COVID-19 patients requiring high-ow nasal oxygen had pathogenic TLR7 missense variants that impaired type I and II IFN responses in peripheral mononuclear cells upon stimulation with the TLR7 agonist [7].In a cohort of 79 severe COVID-19 cases, 2.1% of men exhibited TLR7 deleterious variants resulting in reduced expression of TLR7 [8].An Egyptian study demonstrated a high frequency of TLR7 rs3853839 in COVID-19 patients correlating well with disease severity and biochemical variables such as ferritin, C-reactive protein (CRP), interleukin (IL)-6, and D dimer [9].A pan ancestry exome-wide association study demonstrated a 4.53-fold increased risk of severe COVID-19 with TLR7 LOF missense variants [10].RNA sequencing analysis revealed profound impairment of the TLR7 pathway in LOF variants [11].The IFN-gamma upregulation is impaired even in hypofunctional and hypomorphic TLR7 variants [11].
The serine protease 2 transmembrane protein (TMPRSS2) of the host cell activates the spike protein of SARS-CoV-2 to facilitate the internalization of viral particles [12,13].The viral RNA either produces new virions through the host ribosomes or enters the endosome.The activation of innate immune response depends on the capability of various cellular sensors to recognize SARS-CoV-2 ssRNA and other intermediates of viral replication and induction of downstream signaling.Cytoplasmic RIG-I and MDA5 are the primary sensors located at the lung epithelium [14,15,16].LR7 is the sensor at the endosomal membrane of plasmacytoid dendritic cells (pDCs) and B cells.It senses the viral genome and triggers the recruitment of the adaptor protein MyD88 to carry a cascade of downstream signaling [17].
The current study aimed to elucidate the structural and mechanistic aspects of reported TLR7 variants that can distinguish critically ill COVID-19 patients from severe and moderately ill.

Data collection
We searched PubMed, Medline, and Google Scholar databases to retrieve all the original studies that have reported an association of TLR7 variants with COVID-19 severity.We have included case series with clinical and biochemical data and categorized the disease severity into mild, moderate, severe, and critical.In silico modeling We have used Phyre 2 module to develop three-dimensional structures of all the TLR7 variants based on the cryo-electron microscopic (EM) structure of human TLR7 in complex with UNC93B1 (PDB ID: 7CYN) as a template (http://www.sbg.bio.ic.ac.uk/phyre2).We have retrieved the sequence from UniProtKB (ID: Q9NYK1, TLR7_Human).

Position-Speci c Evolutionary Preservation (PSEP) time
The Panther module (http://www.pantherdb.org/tools/csnpScoreForm.jsp) calculated the evolutionary preservation time of a given amino acid in the lineage.The functional impact of the non-synonymous mutation is proportional to PSEP time.

Thermal Stability assessment
The CUPSAT (http://cupsat.tu-bs.de/)module determined the thermal stability with the crystal structure of TLR7 (PubMed ID: 7CYN) as the reference by using amino acid atomic potentials and torsion angles.

Stability changes upon mutations
PremPs (https://lilab.jysw.suda.edu.cn/research/PremPS/)module evaluated the effect of missense mutations on protein stability by calculating the change in unfolding Gibbs free energy.

Nucleic acid-binding ability
We predicted the nucleic acid binding sites of TLR7 based on electrostatic potentials.The HDock server (http://hdock.phys.hust.edu.cn/)used a hybrid algorithm of template-based and ab initio docking for TLR7 ssRNA [SARS CoV-2 Chain T, RNA (5'-R(P*UP*UP*CP*AP*UP*AP*AP*CP*UP*UP*AP*A)-3')] docking and assessed the binding a nity change with each mutation.

Molecular Docking
We have docked imiquimod and N-acetylcysteine to different TLR7 variants using Swiss Dock.

Protein-protein interactions and network analysis:
We have used the Inbio-discover module to evaluate protein-protein interactions and network analysis (https://inbio-discover.com/)

Statistical analysis
The student t-test evaluated differences in the distribution of continuous variables between two groups.
Analysis of variance (ANOVA) assessed the changes in binding a nities of various molecules towards the TLR7 variants segregated based on functionality.The Fisher's exact test calculated the association of different pathogenic scores with COVID-19 severity in terms of odds ratio (OR) and 95% con dence interval (95% CI) based on the data computed in the 2 ⋅ 2 contingency table.All the statistical associations are signi cant at p < 0.05.

Discussion
This study provides structural and mechanistic insights into 25 TLR7 variants reported contributing to severe or critical COVID-19.Patients with pathogenic TLR mutations got severe or critical COVID-19 at younger ages.
TLR7 mutations in highly conserved regions with higher SNAP2 and lower SIFT scores caused critical illness.
The TLR7 extracellular domain is the hot spot for the reported variants in severe or critical COVID-19.Nucleic acids can bind to the largest positive patches in the C and D chains of TLR7.TLR7 variants impaired the ability of TLR7 to sense the SARS CoV-2 viral RNA, thus resulting in reduced antiviral response and loss of immunity.
Impaired sensing is prominent in LOF variants followed by hypofunctional variants.The agonist binding patterns remained unaltered irrespective of the type of mutation.No differences were observed in MyD88 binding to TLR7 variants reported in critical or severe COVID-19.However, MyD88 binding a nity was higher in LOF variants than in hypofunctional or neutral TLR7 variants.N-acetylcysteine (NAC) can bind effectively to TLR7 variants reported in critical COVID cases more than to those variants in severe COVID.This supports the possible bene cial effect of NAC supplementation in critical COVID cases with loss of function variants of TLR7.NAC prevents the activation of NF-κB by scavenging ROS, inhibits IKKb and NF-κB nuclear translocation, and impairs the synthesis of proin ammatory cytokines hence substantiating this observation [18].
The cryo-EM structure of TLR7 in complex with molecular chaperone Unc-93 homolog B1, TLR signaling regulator (UNC93B1) provided the mechanistic basis of tra cking of TLR from endoplasmic reticulum to the endosome [19].The transmembrane and juxta membrane regions of TLRs interact with the amino-terminal sixhelix bundle of UNC93B1 [19].This cryo-EM structure served as a template for simulating the different variants of TLR7 in the current study.
SARS-CoV-2 infection activates the RNA sensor pathways of the TLR3 and TLR7 via IRF3 and NFB transduction pathways, respectively, which induces pro-in ammatory cytokines [21].SARS-CoV-2 singlestranded RNA (ssRNA) fragments serve as disease-associated molecular patterns (DAMPs) resulting in the activation of endosomal TLR7/8 and MyD88 pathway, which in turn lead to the production of IFN and cytokines; and Th1 polarization [22].We report that Hypofunctional or LOF variants of TLR7 exhibit a lack of such antiviral response due to decreased sensing of viral RNA.
A transient decrease in the number of circulating plasmacytoid dendritic cells (pDCs) during the early stages of the onset of COVID-19 symptoms contributes to disease severity as pDCs are essential for sensing the virus and producing an antiviral response mediated through TLR7-MyD88 signaling [22].
The current study summarises the ndings of the four published studies [5][6][7][8] by demonstrating the loss of viral RNA sensing ability of TLR7 in virtue of the reported mutations correlating with the reported mRNA expression pro ling.The imiquimod binding is not affected by the TLR7 mutations.MyD88 showed more binding a nity with LOF variants than hypofunctional variants.NAC showed more binding a nity towards

TLR7 variants reported to
critical COVID than those in severe COVID.Loss of viral sensing ability due to TLR7 mutations contribute to the severity of COVID-19 by impairing TLR7-MyD88 signaling-mediated antiviral response.NAC might restore the anti-viral response, thus helpful in reducing mortality and morbidity associated with COVID-19 in TLR7 loss-of-function variants.Puskar K, Ta 'asan S, Schwartz R, LeDuc PR (2006) Evaluating spatial constraints in cellular assembly processes using a monte carlo approach.Cell Biochem Biophys.2006.45, 195-201.

Figures
Figures

B)
HDock server was used to dock protein-protein interactions between TLR7 and MyD88 as a measure of the downstream signaling cascade.

Table 2
Impact of TLR7 pathogenic mutations on the clinical outcome in COVID-19 The largest positive patches in Chain A, Chain B, Chain C, and Chain D contained 67, 87, 168, and 150 amino acid residues, respectively.Hence, Chain A and Chain B are classi ed as non-nucleic acid-binding, while Chain C and Chain D are classi ed as nucleic acid binding.LOF: loss-of-function COVID-19 disease severity: moderate (2), severe (3), critical (4)