Siglec-9 Restrains Antibody-Dependent Natural Killer Cell Cytotoxicity against SARS-CoV-2

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection alters the immunological profiles of natural killer (NK) cells. However, whether NK antiviral functions are impaired during severe coronavirus disease 2019 (COVID-19) and what host factors modulate these functions remain unclear. We found that NK cells from hospitalized COVID-19 patients degranulate less against SARS-CoV-2 antigen-expressing cells (in direct cytolytic and antibody-dependent cell cytotoxicity [ADCC] assays) than NK cells from mild COVID-19 patients or negative controls. The lower NK degranulation was associated with higher plasma levels of SARS-CoV-2 nucleocapsid antigen. Phenotypic and functional analyses showed that NK cells expressing the glyco-immune checkpoint Siglec-9 elicited higher ADCC than Siglec-9^- NK cells. Consistently, Siglec-9⁺ NK cells exhibit an activated and mature phenotype with higher expression of CD16 (FcγRIII; mediator of ADCC), CD57 (maturation marker), and NKG2C (activating receptor), along with lower expression of the inhibitory receptor NKG2A, than Siglec-9^- CD56^dim NK cells. These data are consistent with the concept that the NK cell subpopulation expressing Siglec-9 is highly activated and cytotoxic. However, the Siglec-9 molecule itself is an inhibitory receptor that restrains NK cytotoxicity during cancer and other viral infections. Indeed, blocking Siglec-9 significantly enhanced the ADCC-mediated NK degranulation and lysis of SARS-CoV-2-antigen-positive target cells. These data support a model in which the Siglec-9⁺ CD56^dim NK subpopulation is cytotoxic even while it is restrained by the inhibitory effects of Siglec-9. Alleviating the Siglec-9-mediated restriction on NK cytotoxicity may further improve NK immune surveillance and presents an opportunity to develop novel immunotherapeutic tools against SARS-CoV-2 infected cells. IMPORTANCE One mechanism that cancer cells use to evade natural killer cell immune surveillance is by expressing high levels of sialoglycans, which bind to Siglec-9, a glyco-immune checkpoint molecule on NK cells. This binding inhibits NK cell cytotoxicity. Several viruses, such as hepatitis B virus (HBV) and HIV, also use a similar mechanism to evade NK surveillance. We found that NK cells from SARS-CoV-2-hospitalized patients are less able to function against cells expressing SARS-CoV-2 Spike protein than NK cells from SARS-CoV-2 mild patients or uninfected controls. We also found that the cytotoxicity of the Siglec-9⁺ NK subpopulation is indeed restrained by the inhibitory nature of the Siglec-9 molecule and that blocking Siglec-9 can enhance the ability of NK cells to target cells expressing SARS-CoV-2 antigens. Our results suggest that a targetable glyco-immune checkpoint mechanism, Siglec-9/sialoglycan interaction, may contribute to the ability of SARS-CoV-2 to evade NK immune surveillance.

Keywords: COVID-19; SARS-CoV-2; Siglec-7; Siglec-9; antibody-dependent cell cytotoxicity; natural killer cells.

FIG 1

Hospitalized coronavirus disease 2019 (COVID-19) is associated with reduced CD56^dim natural killer (NK) cell degranulation against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Spike-expressing target cells. (A) Schematic overview of the experiments to evaluate the direct cytotoxicity- and antibody-dependent cell cytotoxicity (ADCC)-mediated degranulation of CD56^dim NK cells during different severities of COVID-19. To examine direct cytotoxicity-mediated degranulation, peripheral blood mononuclear cells (PBMCs) from each donor, from three COVID-19 status groups (n = 8 SARS-CoV-2 negative, n = 12 mild COVID-19, and n = 21 hospitalized COVID-19), were cocultured (2) or not (1) with SARS-CoV-2 Spike-expressing 293T target cells. To examine ADCC-mediated degranulation, identical cocultures were performed in the presence of the negative (3) or positive (4) antibody pool. Direct cytotoxicity was assessed by subtracting (1) from (2). 10:1 effector:target [E:T] ratio. ADCC was assessed by subtracting the results of subtracting (1) from (3) from the results of subtracting (1) from (4). (B to E) Direct cytotoxicity-mediated degranulation and cytokine production of the CD56^dim NK population was assessed as the percentage of (B) CD107a⁺ interferon γ (IFN-γ)⁺, (C) CD107a⁺ tumor necrosis factor α (TNF-α)⁺, (D) IFN-γ⁺, and (E) IFN-γ⁺ TNF-α⁺ cells. Medians and interquartile ranges (IQR) are displayed. Kruskal-Wallis tests with Dunn’s multiple comparisons correction were used for statistical analyses. (F to I) ADCC-mediated degranulation and cytokine production of the CD56^dim NK population was assessed as the percentage of (F) CD107a⁺ IFN-γ⁺, (G) CD107a⁺ TNF-α⁺, (H) IFN-γ⁺, or (I) IFN-γ⁺ TNF-α⁺ cells. Median and IQR are shown. Kruskal-Wallis tests with Dunn’s multiple-comparisons correction were used for statistical analyses. (J to M) Spearman’s rank-order correlations between plasma N-antigen load and the percentage of CD56^dim NK cells expressing (J) CD107a⁺, (K) IFN-γ⁺, (L) CD107a⁺ IFN-γ⁺, and (M) IFN-γ⁺ TNF-α⁺ during direct cytotoxicity assays; only samples from COVID-19-positive donors were used (n = 33).

FIG 2

The Siglec-9⁺ CD56^dim NK subpopulation exhibits a higher SARS-CoV-2 specific ADCC than the Siglec-9^– CD56^dim NK subpopulation. (A to I) ADCC-mediated degranulation and/or cytokine production of the Siglec-9⁺ and Siglec-9^– CD56^dim NK cells within each COVID-19 status group as assessed by the percentage of cells expressing (A) CD107a, (B) IFN-γ, (C) TNF-α, (D) CD107a and IFN-γ, (E) CD107a and TNF-α, and (F) IFN-γ and TNF-α. The mean fluorescence intensity (MFI) was also evaluated for (G) CD107a, (H) IFN-γ, and (I) TNF-α. Wilcoxon’s signed-rank tests were used to compare the Siglec-9⁺ and Siglec-9^– CD56^dim NK cells within each COVID-19 status group. (J) Spearman’s correlation heat-map showing associations between direct cytotoxicity and ADCC-mediated NK degranulation of the Siglec-9⁺ and Siglec-9^– CD56^dim NK cells and plasma N-antigen load. Square color represents correlation strength, with blue representing negative correlations and red representing positive correlations; only samples from COVID-19-positive donors were used (n = 33).

FIG 3

The Siglec-7⁺ CD56^dim NK subpopulation exhibits higher SARS-CoV-2-specific direct cytolytic and ADCC activity than does the Siglec-7^– CD56^dim NK subpopulation. (A to C) Direct cytotoxicity-mediated degranulation and/or cytokine production of the Siglec-7⁺ and Siglec-7^– CD56^dim NK cells within each COVID-19 status group as assessed by the percentage of cells expressing (A) CD107a, (B) TNF-α, and (C) CD107a and TNF-α. (D to I) ADCC-mediated degranulation and/or cytokine production of the Siglec-7⁺ and Siglec-7^– CD56^dim NK cells within each COVID-19 status group as assessed by the percentage of cells expressing (D) CD107a, (E) IFN-γ, (F) TNF-α, (G) CD107a and IFN-γ, (H) CD107a and TNF-α, or (I) IFN-γ and TNF-α. Wilcoxon’s signed-rank tests were used to compare the Siglec-7⁺ and Siglec-7^– CD56^dim NK cells within each disease group. (J) Spearman’s correlation heat-map showing associations between direct cytotoxicity- and ADCC-mediated NK degranulation of the Siglec-7⁺ and Siglec-7^– CD56^dim NK cells and plasma N-antigen load. Square color represents correlation strength, with blue representing negative correlations and red representing positive correlations; only samples from COVID-19-positive donors were used (n = 33).

FIG 4

Siglec-9 marks activated and mature CD56^dim NK subpopulation in vivo. (A) Comparisons of the expression of CD16 (FcγRIII; mediator of ADCC), CD57 (maturation marker), NKG2C (activating receptor), and NKG2A (inhibitory receptor) on Siglec-9⁺ and Siglec-9^– CD56^dim NK cells obtained from 79 individuals with three COVID-19 disease states (negative; n = 12, mild; n = 26, and hospitalized; n = 41). Siglec-9⁺ cells exhibit higher levels of CD16, CD57, NKG2C, and lower levels of NKG2A compared to Siglec-9^– cells. Wilcoxon’s signed-rank tests were used to compare the Siglec-9⁺ and Siglec-9^– CD56^dim NK cells within each COVID-19 disease state group. Mann-Whitney U tests were used to compare disease state groups. (B) Comparisons of CD16, CD57, NKG2C, and NKG2A expression on Siglec-7⁺ and Siglec-7^– CD56^dim NK cells. Siglec-7⁺ cells exhibited higher levels of CD16, NKG2C, and NKG2A than Siglec-7^– cells. No differences were observed in CD57 expression between the Siglec-7⁺ and Siglec-7^– cells. As in panel A, Wilcoxon’s signed-rank tests were used to compare the Siglec-7⁺ and Siglec-7^– CD56^dim NK cells within each COVID-19 status group, and Mann-Whitney U tests were used to compare the status groups.

FIG 5

Siglec-9, but not Siglec-7, marks CD56^dim NK cells with high ADCC activity against SARS-CoV-2. (A) Percentage of each Siglec-expressing cell subpopulation (Siglec-7^– Siglec-9^–, Siglec-7⁺ Siglec-9^–, Siglec-7^– Siglec-9⁺, or Siglec-7⁺ Siglec-9⁺) within the CD56^dim NK cells from all donors (n = 79). (B to E) In vivo expression of (B) CD16, (C) CD57, (D) NKG2C, and (E) NKG2A on the indicated cell subpopulations (n = 79). Median and IQR are shown. Friedman tests with Dunn’s multiple-comparisons correction were used for statistical analyses. (F to H) The ADCC-mediated degranulation of the indicated NK cell subpopulations (n = 41) as assessed by (F) the percentage of cells expressing CD107a, (G) the MFI of CD107a, or (H) the percentage of cells expressing CD107a and IFN-γ. Median and IQR are displayed. Friedman tests with Dunn’s multiple-comparisons correction were used for statistical analyses.

FIG 6

Blocking Siglec-9 interactions, using a Siglec-9-blocking antibody, enhances the anti-SARS-CoV-2 ADCC of CD56^dim NK cells. (A to E) The impact of the Siglec-9-blocking antibody, compared to an isotype control, on ADCC-mediated NK degranulation and/or cytokine production against SARS-CoV-2. PBMCs from six healthy controls were used as effector cells, and SARS-CoV-2 Spike-expressing 293T cells were used as target cells. Cells were cocultured at a 10:1 (E:T) ratio for 12 h. NK degranulation and/or cytokine production was assessed as the percentage of cells expressing (A) CD107a, (B) CD107a and IFN-γ, (C) CD107a and TNF-α, (D) TNF-α, (E) and IFN-γ and TNF-α. Paired t tests were used for statistical analysis. (F) The impact of the Siglec-9-blocking antibody, compared to an isotype control, on the ADCC-mediated lysis of SARS-CoV-2 target cells. Purified NK cells isolated from the PBMCs of five healthy donors were used as effector cells, and SARS-CoV-2 S CHO-K1 cells were used as target cells. Cells were cocultured at a 5:1 E:T ratio for 5 h. The SARS-CoV-2 S CHO-K1 cells stably express the SARS-CoV-2 Spike (S) protein and a HaloTag-HiBiT protein. When the target cells are lysed by ADCC, the intracellular HaloTag-HiBiT protein interacts with the extracellular detection reagent to generate a luminescence signal that can quantitatively measure the degree of target cell lysis. A paired t test was used for statistical analysis.

FIG 7

Model of how a Siglec-9-blocking antibody increases the cytotoxicity of Siglec-9⁺ NK cells. Left panel: Siglec-9^– cells have low cytotoxicity. Middle panel: the Siglec-9⁺ CD56^dim NK subset has high ADCC activity, possibly due to elevated expression of CD16 (FcγRIII; a mediator of ADCC activity), CD57 (maturation marker), and NKG2C (activating receptor), and reduced expression of the inhibitory receptor NKG2A, compared to Siglec-9^– CD56^dim NK cells. However, the Siglec-9 molecule itself is an inhibitory receptor which restrains the cytolytic ability of these highly cytotoxic Siglec-9⁺ CD56^dim NK cells by binding to Sialic acid on the surface of target cells. Right panel: blocking the inhibitory receptor, Siglec-9, using a blocking antibody can unleash a higher ADCC potential of the Siglec-9⁺ CD56^dim subpopulation.