key: cord-0891604-qte0jp1y
authors: Bernal, Enrique; Gimeno, Lourdes; Alcaraz, María J; Quadeer, Ahmed A; Moreno, Marta; Martínez-Sánchez, María V; Campillo, José A; Gomez, Jose M; Pelaez, Ana; García, Elisa; Herranz, Maite; Hernández-Olivo, Marta; Martínez-Alfaro, Elisa; Alcaraz, Antonia; Muñoz, Ángeles; Cano, Alfredo; McKay, Matthew R; Muro, Manuel; Minguela, Alfredo
title: Activating Killer-cell Immunoglobulin-like Receptors are associated with the severity of COVID-19
date: 2021-04-30
journal: J Infect Dis
DOI: 10.1093/infdis/jiab228
sha: d288eceda1786b9ca09e9de9c51e662dfd843fe1
doc_id: 891604
cord_uid: qte0jp1y

BACKGROUND: Etiopathogenesis of the clinical variability of the coronavirus disease 2019 (COVID-19) remains mostly unknown. Here we investigate the role of Killer-cell Immunoglobulin-like receptor (KIR)/Human Leukocyte Antigen Class-I (HLA-I) interactions in the susceptibility and severity of COVID-19. METHODS: KIR and HLA-I genotyping and NK cell (NKc) receptors immunophenotyping in 201 symptomatic patients and 210 non-infected controls. RESULTS: NKcs with a distinctive immunophenotype, suggestive of recent activation (KIR2DS4 (low) CD16 (low) CD226 (low) CD56 (high) TIGIT (high) NKG2A (high)), expanded in patients with severe COVID-19. This was associated with a higher frequency of the functional A-telomeric activating KIR2DS4 in severe than mild/moderate patients and controls (83.7%, 55.7% and 36.2%, p<7.7x10 (-9)). In mild/moderate patients HLA-B*15:01 was associated with higher frequencies of activating B-telomeric KIR3DS1 compared to patients with other HLA-B*15 subtypes and non-infected controls (90.9%, 42.9% and 47.3%, p<0.002, Pc=0.022). This strongly suggests that HLA-B*15:01 specifically presenting SARS-CoV-2 peptides could form a neo-ligand interacting with KIR3DS1. Similarly, a putative neo-ligand for KIR2DS4 could arise from other HLA-I molecules presenting SARS-CoV-2 peptides expressed on infected/activated lung antigen presenting cells. CONCLUSIONS: Our results support a crucial role of NKcs in the clinical variability of COVID-19 with specific KIR/Ligand interactions associated to disease severity.

The new severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) associated with the coronavirus disease-2019 (COVID-19) emerged in Wuhan (China) and spread worldwide, creating a global pandemic and a historic health and economic crisis [1] . Clinical spectrum of COVID-19 ranges from asymptomatic to pneumonia and the appearance of multiple complications such as acute respiratory distress syndrome, multiple-organ failure and death [2] .

Emerging studies on COVID-19 report low numbers of peripheral blood Natural Killer cells (NKcs) [3] [4] [5] , but increased numbers infiltrating the lung [6] . NKcs are the first line of defense against viral infections. Moderate NKc activation promotes infection control, while stronger activation can be related to immunopathology [7] . NKc activation and function is regulated by the interplay between activating and inhibitory receptors [8] . In humans, natural cytotoxicity receptors (NKp30, NKp44 and NKp46) and DNAX Accessory Molecule-1 (DNAM-1/CD226) are activating receptors that recognize viral-derived/induced products [9] . NKcs can also recognize stress-induced ligands through the natural killer group 2 (NKG2) family, most notably NKG2D, which interacts with the MHC-class-I chain-related protein A/B (MICA/B) and UL16 binding-protein (ULBP). Human NKcs can also express up to 6 different activating Killer-cell Immunoglobulin-like Receptors (aKIR), KIR2DS1-5 and KIR3DS1, that recognize peptide-loaded Human Leukocyte Antigen class-I (HLA-I). Interactions of aKIRs have been reported between KIR2DS1/HLA-C2-allotypes [10] , KIR2DS2/HLA-C1-allotypes [11] , KIR2DS2/HLA-A*11 [12] , KIR2DS4/HLA-A*11/C*02/C*04/C*05/C*16 [13] and KIR3DS1*014/HLA-I alleles with Bw4 epitope [14] .

HLA-F, which is upregulated in local macrophages during viral infections, is a ligand for KIR3DS1 and KIR2DS4 [15, 16] .

To regulate activating signals, a repertoire of inhibitory receptors that survey HLA-I repress NKcs to protect healthy cells from inappropriate killing [17] . Although it may seem contradictory, NKcs become fully competent during a process known as "licensing", where inhibitory receptors interact with A c c e p t e d M a n u s c r i p t 6 (Sequential Organ Failure Assessment) scale was used to assess disease severity [26] . The disease was considered severe when patients required invasive mechanical ventilation (IMV). Serum and EDTA blood samples were obtained to determine biochemical, microbiological, hematological and immunological variables.

HLA-I and KIR genotyping was performed on DNA samples extracted with QIAamp-DNA-Blood-Mini-kit (QIAgen-GmbH, Germany) using Lifecodes HLA-SSO and KIR-SSO kits (Immucor, Stamford, CT) and Luminex®. HLA-A, B and C genotyping allowed for identification of Bw4, C1, and C2 KIR-ligands [27] . Full-exon-5 (fKIR2DS4) and deleted-exon-5 (dKIR2DS4) KIR2DS4 alleles were identified. Centromeric and telomeric AA/Bx genotypes were identified [28] .

The expression of activating and inhibitory receptors on CD3 -CD16 -/+ CD56 ++ (CD56 bright ) and CD3 -CD16 + CD56 + (CD56 dim ) NKc subsets and CD3 + CD4 + and CD3 + CD4 -T cells was evaluated as percentage of positive cells and as mean fluorescence intensity (MFI) using FACSLyric and DIVA-9.0 (Becton-Dickinson, BD) as described in figure 1. Hu-2019 complete proteome reference sequence (GISAID ID: EPI_ISL_402125) to predict strong-binding 9-mer peptides presented for each HLA-B*15 allotypes (percentile rank<0.05 or score >0.8, Supplementary File-1).

Since KIR3DS1 residues interacting with HLA-B*15 presenting peptides have not been described, potential interacting residues (C-alpha8Å) were obtained by using the structure of its homologous KIR3DL1 (with 97% amino acid similarity of extracellular domains [31]) in complex with HLA-B*57:01 presenting peptide (PDB-ID: 5T70), and aligning the protein sequence of KIR3DL1 with that of KIR3DS1 (NCBI accession number: NP_001077008.1). Interaction score between the receptor interacting residues and the SARS-CoV-2 peptide was obtained as the mean affinity alignment score of total interacting residues. Interaction score of KIR3DL1/KIR3DS1 with its Bw4 ligand [32] was calculated to ascertain the affinity of natural occurring KIR/ligand interactions. For this analysis, potential KIR3DS1 residues interacting with the Bw4 epitope (HLA-B positions 77, 80, 81, 82, and 83 [33]) were identified based on the structure of KIR3DL1 in complex with HLA-B*57:01 (PDB ID: 3VH8).

Similar procedure was used to model KIR2DS1/HLA-B*15-peptide interactions. KIR2DS1 interacting residues were obtained using KIR2DL1 structure in complex with HLA-Cw4 presenting a peptide (PDB-ID: 1IM9) [34] and aligning the protein sequence of KIR2DL1 with that of KIR2DS1 (NCBI accession number: XP_011546300.1).

Data were analyzed using SPSS-24.0 (SPSS-Inc., Chicago, IL). Categorical variables were analyzed using frequency tables and Pearson's- 2 or two-tailed Fisher's exact tests. Bonferroni correction (Pc) was applied when needed. Mean and standard deviation (SD) were used for continuous variables and differences were estimated with ANOVA, Student's-t or Mann-Whitney-U tests based on the presence or absence of normal distribution. Association between variables was assessed with A c c e p t e d M a n u s c r i p t 8 binary logistic regression analysis, adjusted for confounding variables. Variables with p<0.05 in the univariate regression analysis were included in the multivariate analysis. The impact of fKIR2DS4 on the mean time for admission to the intensive care unit (ICU) was assessed using Kaplan-Meier and the Mantel-Haenszel log-rank test. The time was censored 30 days after the onset of symptoms. P<0.05 was considered significant. Table 1 summarizes biological, clinical and evolutionary characteristics of patients according to IMV requirements. Although no significant differences in age or sex were observed between patients requiring or not IMV, the former showed worst comorbidity and severity scores, biochemical and hematological parameters, more intense treatments and worst clinical evolution. Higher frequency of Caucasians than Latin-Americans required IMV probably due to their older age (59.0% vs. 53.5%, p=0.019).

First we explored the role of HLA-I allotypes in the susceptibility and severity of SARS-CoV-2 infection.

Comparing controls and patients, we observed a protective role for HLA-A*26 (10.8% vs. 5.5%) and

HLA-A*33 (7.4% vs. 2.0%), and a risk role for HLA-A*23 (4.4% vs. 11.4%), HLA-A*24 (14.3% vs.

22.9%) and HLA-C*01 (3.4% vs. 6.5%). Significant differences disappeared after correction (Supplementary Table E1 ).

A c c e p t e d M a n u s c r i p t No differences were found either for inhibitory or activating KIRs, AA and Bx centromeric and telomeric genotypes, or HLA-ligands (C1, C2, Bw4, or KIR2DS4-ligands), except for fKIR2DS4, which showed higher frequency in COVID-19 patients than in controls (61.7% vs. 36.2%, Pc<9.5x10 -6 ).

Consequently fKIR2DS4 could be considered as a risk factor for symptomatic COVID-19. Besides, fKIR2DS4 was associated to severe diseases with frequencies of 83.7% vs. 56.3% for patients who did or did not require IMV (p<0.001) ( Table 2) . 

To understand the relationship that KIR2DS4 + NKcs may have with disease severity, leukocyte subtypes and KIR + NKc repertoire were analyzed in healthy controls and patients without or with IMV.

To avoid the sequelae of intense treatments applied to patients with severe disease the cellular study was performed at day 75 (IQR61-92) after hospital discharge. However, compared to controls, patients A c c e p t e d M a n u s c r i p t 10 presented reduced numbers of lymphocytes (27.4%, 25.2%, and 20,2%, p<0.01), non-classical monocytes (16.7%, 15.7%, and 12.6%, p<0.01), neutrophils (40.3%, 38.1%, 33.7%, p<0.05), and CD4 + T-lymphocytes (33.7%, 29.4%, and 25.6%, p<0.001), and increased numbers of CD8 + T-lymphocytes (23.9%, 28.5%, 30.0%, p<0.05). These values were accentuated in patients with IMV ( Figure 3A ).

Although the numbers of CD56 bright and CD56 dim NKcs as well as the repertoire of KIR + NKcs did not show significant differences between controls and patients, NKcs expressing KIR2DS4 as the only KIR receptor (single-KIR2DS4 + ) showed higher frequency in patients requiring IMV than in patients without IMV and controls (9.1%, 5.1%, and 4.3%, p<0.05) ( Figure 3B ). When the expression as MFI of each KIR was studied in single-KIR + NKcs, no differences were found between the three study groups for KIR2DL1 + , KIR2DL2/S2 + , KIR2DL3 + , KIR2DS1 + , or KIR3DS1 + NKcs, whereas a slight increase of KIR3DL1 was found in patients requiring IMV compared to those that did not or to controls (7869.5, 6076.1 and 6083.9, MFI, p<0.01). However, the expression of KIR2DS4 showed a prominent downmodulation in both groups of patients compared to controls (5392.5, 6500 and 8721.6 MFI, p<0.001) ( Figure 3C ). Since the expression of KIR receptors is down-modulated when interacting with their specific HLA-I ligand [35], we analyzed the expression of KIRs in presence of their HLA-ligands in COVID-19 patients. As already described, the presence of C2-ligand very significantly induced the down-modulation of KIR2DL1 (1847.5 vs. 1296.8, p<0.001). However, C1-ligands, Bw4-ligands and HLA-A*11/C*2/C*4/C*5/C*16 induced slight non-significant down-modulations of their respective KIR2DL2/S2-L3, KIR3DL1 and KIR2DS4 receptors ( Figure 3D ). Therefore, the strong down-modulation observed for KIR2DS4 in COVID-19 patients did not appear to be associated with the presence of its HLA-ligands.

Additional immunophenotypic characteristics of KIR2DS4 + NKcs are shown in Figure 3E . CD16 expression showed no differences between patients and controls, but, in general, NKcs expressing aKIRs (KIR2DS1, KIR3DS1 or KIR2DS4) showed lower levels of CD16 than NKcs expressing iKIRs (p<0.05). CD56 expression was higher in patients than in controls on NKcs expressing activating A c c e p t e d M a n u s c r i p t 11 KIR2DS1 (p<0.01), KIR3DS1 (p<0.001) and KIR2DS4 (p<0.05), but not on NKcs expressing iKIRs.

CD226 expression was higher on all NKc subsets in patients that did not require IVM (p<0.01, for all KIR + subsets) than in controls and patients with IMV. CD226 showed lower expression on NKcs bearing aKIRs than in those bearing iKIRs (p<0.001). NKG2D expression was lower in patients than in controls for all NKc subsets, with reductions statistically significant on KIR2DL1 + (p<0.05), KIR2DL3 + (p<0.01) or KIR2DS1 + (p<0.001) NKc subsets. In contrast, TIGIT expression was higher in patients (particularly in patients with IMV) than in controls (p<0.05, for all KIR + subsets). Finally, although no significant differences were found between controls and patients in NKG2A expression , in general, KIR2DS4 + NKcs showed higher NKG2A expression than KIR3DS1 + (p<0.01) and the other KIR + NKc subsets (p<0.001). Altogether, KIR2DS4 + NKcs in severe COVID-19 showed a distinctive phenotype different from other NKc subsets consisting in KIR low CD16 low CD226 low CD56 high TIGIT high NKG2A high .

Since the frequency of KIR2DS4-ligands HLA-A*11, C*02, C*04, C*05, and C*16 [36] , all together or individually, did not show differences between controls and patients, we studied other putative HLAligands. Frequency of fKIR2DS4 remained within the ranges described for the global COVID-19 group Bx-telomeric genotype was more frequent in HLA-B*15:01 patients that did not require IMV (84.6% vs. 15.4%, p=0.09), and therefore it was associated with milder COVID-19, which is reasonable since the frequency of the A-telomeric fKIR2DS4 (associated to severe disease) was decreased in these patients. Functionality and number of NKcs are significantly decreased in COVID-19 [3, 4, 37] , and yet the underlying causes have not been explored. During the acute phase of the infection peripheral blood NKcs increase the express of inhibitory (NKG2A), regulatory (TIM-3) and exhaustion (PD-1) molecules and reduce the expression of activating receptors (DNAM-1 and NKG2D) and the secretion of IFNγ [38] . Two months after infection, some of these effects were still observed in patients from our study, particularly in those requiring IMV: reduced number of total lymphocytes, non-classical monocytes, neutrophils and CD4 + T-cell and increased number of CD8 + T-cells and single-KIR2DS4 + NKcs. These single-KIR2DS4 + NKcs showed a distinctive phenotype with a clear down-modulation of its KIR compared to healthy controls, most probably due to active interaction with specific ligands. KIRs are generally down-modulated when interacting with their HLA-ligand [35] . Although in our study KIR/HLAligand-induced down-modulation was observed for most single-KIR + NKc subsets, down-modulation induced by normal-constitutive HLA-I expression was much lower than that observed in single-KIR2DS4 + NKcs, suggesting that KIR2DS4 might interact with unconventional ligands. Like other single-aKIR + NKc subsets (KIR2DS1 + and KIR3DS1 + ), single-KIR2DS4 + NKcs showed reduced expression of CD16, CD226 and NKG2D, but highly expressed CD56 and TIGIT, suggesting that these cells had been recently involved in an activation event [39, 40] . Supporting this idea is the NKG2A overexpression specifically observed in single-KIR2DS4 + NKcs, which is consistent with the strong NKG2A A c c e p t e d M a n u s c r i p t 14 over-expression described during the acute phase [41] and supports a primary role of KIR2DS4 + NKcs in the SARS-CoV-2 response.

In fact, the frequency of fKIR2DS4 gene was significantly increased in all COVID-19 patients, and particularly in those requiring IMV, 83.7% vs. 36.2% in controls. It has been recently suggested that frequency of HLA-C*05 (a KIR2DS4 ligand) was related to mortality variability in different COVID-19 populations [42] . However, in our series no differences between patients and controls were observed in the frequency of HLA-C*05 or KIR2DS4/HLA-C*05 interaction. In fact, fKIR2DS4 frequency did not change in patients bearing its putative ligands (HLA-A*11/C*02/C*04/C*05/C*16 [36] ), which suggests that new ligands could be interacting with KIR2DS4 in COVID-19 patients. In 2013 Goodridge et al. suggested that open-conformer HLA-F could be a ligand for KIR2DS4 [43] ; however, more recent results demonstrate that HLA-F open-conformer is a ligand for KIR3DS1 and KIR3DL2 but not for KIR2DS4 [44] . In contrast, peptide-bound HLA-F is not a ligand for KIR3DS1 because peptides directly hinder KIR3DS1/HLA-F interaction [44] . However, it is possible that a SARS-CoV-2 peptide presented in HLA-F could form a neo-ligand for KIR2DS4, in the same way that HLA-C*05:01 does present bacterial 9-mer peptides carrying a tryptophan at position-8 to become a ligand for KIR2DS4 [45] .

Indeed, HLA-F can present peptides of unconventional length dictated by the R62W mutation that produce an open-ended groove accommodating long peptides [44] . In contrast, this putative neo-ligand formed by HLA-F presenting SARS-CoV-2 peptides will not be recognized by KIR3DS1, as previously indicated [44] and supported by our data, since the frequency of KIR3DS1 in COVID-19 patients did not differ from that in controls.

HLA-F is mainly expressed in activated lymphocytes and monocytes [46] . Alveolar macrophages, the predominant leukocytes in the lung, are strongly activated in COVID-19 patients and include viral particles within their cytoplasm [47] , and therefore, they would be able to present SARS-CoV-2 peptides in HLA-F. Besides, SARS-CoV-2 induces the expression of CXCR3 ligands (CXCL9-11) in lung tissue [48] and macrophages [5] facilitating NKc recruitment from the peripheral blood [37] . A c c e p t e d M a n u s c r i p t 28 

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

Clinical Characteristics of Coronavirus Disease 2019 in China

A single-cell atlas of the peripheral immune response in patients with severe COVID-19

COVID-19 pneumonia: CD8+ T and NK cells are decreased in number but compensatory increased in cytotoxic potential

Single-cell landscape of bronchoalveolar immune cells in patients with COVID-19

COVID-19 severity correlates with airway epitheliumimmune cell interactions identified by single-cell analysis

Natural killer cells act as rheostats modulating antiviral T cells

Up on the tightrope: natural killer cell activation and inhibition

Natural killer cell responses to viral infection

Education of human natural killer cells by activating killer cell immunoglobulin-like receptors

Large Spectrum of HLA-C Recognition by Killer Ig-like Receptor (KIR)2DL2 and KIR2DL3 and Restricted C1 Specificity of KIR2DS2: Dominant Impact of KIR2DL2/KIR2DS2 on KIR2D NK Cell Repertoire Formation

Activating killer cell immunoglobulin-like receptor 2DS2 binds to HLA-A*11

KIR2DS4 is a product of gene conversion with KIR3DL2 that introduced specificity for HLA-A*11 while diminishing avidity for HLA-C

Analysis of binding of KIR3DS1*014 to HLA Suggests Distinct Evolutionary History of KIR3DS1

Open conformers of HLA-F are high-affinity ligands of the activating NK-cell receptor KIR3DS1

Interactions Between KIR3DS1 and HLA-F Activate Natural Killer Cells to Control HCV Replication in Cell Culture

Negative signaling by inhibitory receptors: The NK cell paradigm

NK Cell Education in Tumor Immune Surveillance: DNAM-1/KIR Receptor Ratios as Predictive Biomarkers for Solid Tumor Outcome

Role of Donor Activating KIR-HLA Ligand-Mediated NK Cell Education Status in Control of Malignancy in Hematopoietic Cell Transplant Recipients

Immunology of COVID-19: Current State of the Science

Unique immunological profile in patients with COVID-19

CD56 in the immune system: More than a marker for cytotoxicity? Front Immunol

Expression of the Inhibitory Receptor TIGIT Is Up-Regulated Specifically on NK Cells With CD226 Activating Receptor From HIV-Infected Individuals

Functional exhaustion of antiviral lymphocytes in COVID-19 patients

Population Difference in Allele Frequency of HLA-C*05 and Its Correlation with COVID-19 Mortality

HLA-F and MHC class I open conformers are ligands for NK cell Ig-like receptors

Human Leukocyte Antigen F Presents Peptides and Regulates Immunity through Interactions with NK Cell Receptors

Human NK cell receptor KIR2DS4 detects a conserved bacterial epitope presented by HLA-C

HLA-F is a surface marker on activated lymphocytes

Broncho-alveolar inflammation in COVID-19 patients: a correlation with clinical outcome

Comparative Replication and Immune Activation Profiles of SARS-CoV-2 and SARS-CoV in Human Lungs: An Ex Vivo Study With Implications for the Pathogenesis of COVID-19

Killer cell immunoglobulin-like receptor 3DL1-mediated recognition of human leukocyte antigen B

Drives Development of COVID-19

Abbreviations: KIR: Killer-cell Immunoglobulin-like receptor

KIR2DS4: deleted-exon5-KIR2DS4; HLA: Human Leukocyte Antigen. a Control vs. Covid-19, P<6

P values estimated with Chi-Square. Pc = p-value after Bonferroni correction (x14)

A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t