Proteomic profiling of MIS-C patients indicates heterogeneity relating to interferon gamma dysregulation and vascular endothelial dysfunction

Abstract

Multi-system Inflammatory Syndrome in Children (MIS-C) is a major complication of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) infection in pediatric patients. Weeks after an often mild or asymptomatic initial infection with SARS-CoV-2 children may present with a severe shock-like picture and marked inflammation. Children with MIS-C present with varying degrees of cardiovascular and hyperinflammatory symptoms. Here we perform a comprehensive analysis of the plasma proteome of more than 1400 proteins in children with SARS-CoV-2. We hypothesize that the proteome would reflect heterogeneity in hyperinflammation and vascular injury, and further identify pathogenic mediators of disease. We show that protein signatures demonstrate overlap between MIS-C, and the inflammatory syndromes macrophage activation syndrome (MAS) and thrombotic microangiopathy (TMA). We demonstrate that PLA2G2A is an important marker of MIS-C that associates with TMA. We find that IFNγ responses are dysregulated in MIS-C patients, and that IFNγ levels delineate clinical heterogeneity.

Conflict of interest statement

DTT serves on advisory boards for Janssen, Sobi, and BEAM. HB has stock ownership in Kriya Therapeutics. SH serves on the advisory board for Horizon Pharma. MPL is an advisory board member for Octapharma and Shionogi, a consultant for Amgen, Novartis, Shionogi, Dova, Bayer, the United States Department of Justice, Sobi, Principia and Argenx and has received research funding from Sysmex, Novartis, Astra Zeneca Rigel, Principia, Argenx, Janssen, and Dova, and has served as a medical advisor for Rigel, Principia, the Platelet Disorder Support Association, CdLS Foundation and 22q11.2 Society. KES received personal fees from Elsevier, Enzyvant and Immune Deficiency Foundation. HB is a paid consultant for Kriya Therapeutics. EMB receives research funding from AB2Bio. E.J.W. is a founder of Surface Oncology and Arsenal Biosciences. E.J.W. is an inventor on a patent (US patent number 10,370,446) submitted by Emory University that covers the use of PD-1 blockade to treat infections and cancer. The other authors declare no competing interests.

© 2021. The Author(s).

Figures

Fig. 1. Overarching architecture of plasma proteome in patients with MIS-C (N = 22), minimal SARS-CoV-2 (N = 26) infection, and Severe COVID-19 (N = 15) compared to healthy controls (N = 25).

t-distributed Stochastic Neighbor Embedding (tSNE) plots are used to visualize clustering between the four groups of patients with overlay from flow cytometry-based scores of percent of non-naïve CD4+ T cells that are activated (HLA-DR+CD38+), percent non-naïve CD8+ T cells that are activated, percent of non-naive B cells that are plasmablasts, and percent of plasmablasts that are T-bet+. Gray dots indicate that data was not available. Source data are provided as a Source Data file.

Fig. 2. Differentially expressed proteins and ranked pathway analysis for each disease state compared to healthy controls.

Differentially expressed proteins (DEPs), log2 fold change threshold of 2 and FDR threshold of 0.01, between patients with MIS-C (N = 22) and healthy controls (N = 25) are shown in a, along with ranked pathway analysis. Size of dots represents number of enriched genes and intensity of colors represents p-value associated with enrichment analysis. DEPs and ranked pathway analysis are shown for patients with minimal Severe Acute Respiratory Syndrome Corona Virus 2 (SARS-CoV-2; N = 26) infection compared healthy controls (N = 25) (b), and Severe Coronavirus Disease (COVID-19; N = 15) patients compared to healthy controls (N = 25) (c). Network analyses for ranked pathway analysis between patients with MIS-C (N = 22) and healthy controls are shown in d where up-regulated proteins are colored in green and down-regulated proteins are colored in red. Node size of the pathway term represents number of input proteins implicated in the pathway. Source data are provided as a Source Data file.

Fig. 3. IFNγ and IL-10 responses and associated cell types.

Spearman correlations between IFNγ signaling and its canonical response protein (CXCL9), as well as proteins associated with activated T cells (IL2RA), NK cells (NCR1), and macrophages (CD163) for patients with (MIS-C; N = 22), Minimal Severe Acute Respiratory Syndrome Corona Virus 2 infection (SARS-CoV-2; N = 26) infection, and Severe Corona Virus Disease (COVID-19; N = 15) compared to healthy controls (N = 25) are shown in panel a. Correlations between IL-10 and its canonical receptor IL12B, as well as IL2RA, NCR1, and CD163 for each disease category are shown in panel b. Panel c shows correlations between HLADR+CD38+ CD8+ T cells (N = 19) and IFNγ, CXCL9, and IL-10. Dots are colored by each patient’s disease category. Panel d demonstrates correlations between HLADR+CD38+ CD4+ T cells (N = 19) and IFNγ, CXCL9, and IL-10. Dots are colored by disease category. Error bars represent 95% confidence interval. All p-values were calculated with a two-sided test. Source data are provided as a Source Data file.

Fig. 4. Association with MAS and patients with MIS-C and SARS-CoV-2 infection.

a Heatmap of MAS-associated proteins IFNγ, CXCL9, CD163, IL2RA, VSIG4, and HMOX1 with unsupervised hierarchical clustering applied to all patients (N = 88). Color bar represents range of variable NPX for each protein. b Boxplot demonstrating levels of MAS-associated proteins in patients who met (N = 17) and did not meet modified criteria for MAS (N = 23). P-values computed with Wilcoxon test. Horizontal line represents median, with bounds of box representing interquartile range. Whiskers represent 1.5x the interquartile range. Dots represent outliers. c Median log2 transformed maximum ferritin during admission in patients who did (N = 17) and did not (N = 23) meet criteria for MAS. P = 0.0000025, p-value computed with Wilcoxon test. d log2 transformed maximum ferritin during admission for patients with MIS-C (N = 21), Severe COVID-19 (N = 13), and minimal disease (N = 6). P-values computed with pairwise comparisons using Wilcoxon rank sum test following Kruskal–Wallis testing. e Acute and convalescent levels of each MAS-related protein for MIS-C patients on whom both acute and convalescent samples were available (N = 12). Circles represent acute samples with triangles representing convalescent samples. Lines connect matched pairs. Gray boxes and whiskers show median and interquartile range of healthy controls (N = 25). P-values for difference between acute and convalescent samples were computed using a paired samples Wilcoxon test. Horizontal line represents median, with bounds of box representing interquartile range. Whiskers represent 1.5x the interquartile range. Dots represent outliers. f Boxplot of maximum absolute neutrophil count during admission for patients who met (N = 17) and did not meet (N = 23) criteria for MAS. P-value computed with Wilcoxon test. Horizontal line represents median, with bounds of box representing interquartile range. Whiskers represent 1.5x the interquartile range. Dots represent outliers. g Maximum absolute neutrophil count during admission for patients with MIS-C (N = 22), Severe disease (N = 15), and Minimal disease (N = 26). P-values computed with pairwise comparisons using Wilcoxon rank sum test following Kruskal–Wallis testing. Horizontal line represents median, with bounds of box representing interquartile range. Whiskers represent 1.5x the interquartile range. Dots represent outliers. h Maximum absolute neutrophil count during admission correlated with IFNγ, CXCL9, and CD163. R computed using Spearman correlation. Error bands represent 95% confidence interval. All p-values were calculated using two-sided tests. Source data are provided as a Source Data file.

Fig. 5. Evidence of thrombotic microangiopathy and vascular endothelial dysfunction in patients with MIS-C and SARS-CoV-2 infection.

a Correlation matrix of vascular and platelet related proteins and soluble C5B9 (SC5B9) in all patients on whom an SC5B9 was measured (N = 75). Hierarchical clustering was applied to identify surrogate markers associated with SC5B9. Color bar represents range of R correlation between −1 and +1. b Heatmap of candidate surrogate markers associated with SC5B9 including PLA2G2A, PDGFC, SELE, CALCA, NOS3, VWA1, and TYMP. Unsupervised hierarchical clustering was applied to all patients (N = 88) with boxes colored by disease category. Color bar represents range of variable NPX for each protein. c Boxplots of each SC5B9 related protein across disease states (MIS-C N = 22, Severe N = 15, Minimal N = 26, Healthy N = 25). P-values computed by Kruskal–Wallis testing. Horizontal line represents median, with bounds of box representing interquartile range. Whiskers represent 1.5x the interquartile range. Dots represent outliers. d Acute and convalescent levels of each SC5B9-related protein for MIS-C patients on whom both acute and convalescent samples were available (N = 12). Circles represent acute samples with triangles representing convalescent samples. Lines connect matched pairs. Gray boxes and whiskers show median and interquartile range of healthy controls (N = 25). P-values for difference between acute and convalescent samples were computed using a paired samples Wilcoxon test. Horizontal line represents median, with bounds of box representing interquartile range. Whiskers represent 1.5x the interquartile range. Dots represent outliers. e Boxplot demonstrating levels of SC5B9-associated proteins in patients who met (N = 13) and did not meet criteria for thrombotic microangiopathy (TMA; N = 21). P-values computed with Wilcoxon test. Horizontal line represents median, with bounds of box representing interquartile range. Whiskers represent 1.5x the interquartile range. Dots represent outliers. f Correlations between lowest platelet count and lowest hemoglobin during admission and PLA2G2A. Small circles represent individual patients with large central circles representing the mean for each disease category. Dots are colored by disease category as MIS-C (N = 22), Severe (N = 15) or Minimal (N = 26). R computed with Spearman correlation. Error band represents 95% confidence interval. All p-values were calculated using two-sided tests. Source data are provided as a Source Data file.

Fig. 6. Clinical heterogeneity among MIS-C patients defined by their IFNγ signature.

a Overlaps between patients who meet criteria for macrophage activation syndrome (MAS), thrombotic microangiopathy (TMA), and MIS-C. b Differentially expressed proteins between patients with IFNγ-low (N = 13) and IFNγ-high expression (N = 7). Log2fold change threshold of 2 and a nominal p-value cutoff of 0.05 were used. c Unsupervised pathway ranking and d network analysis for differentially expressed proteins between IFNγ-low and IFNγ-high patients. e Maximum ferritin level during admission for patients in the IFNγ-low (N = 12) and IFNγ-high (N = 7) groups. P-value computed using Wilcoxon test. Horizontal line represents median, with bounds of box representing interquartile range. Whiskers represent 1.5x the interquartile range. Dots represent outliers. f Number of patients in IFNγ-low (N = 13) and IFNγ-high (N = 7) groups who did and did not require inotropic support. P-value computed with Fisher’s exact test. NTproBNP expression between IFNγ-low (N = 13) and IFNγ-high groups (N = 7) (g) and in MIS-C patients who did (N = 7) and did not require inotropes (h; N = 15). Horizontal line represents median, with bounds of box representing interquartile range. Whiskers represent 1.5x the interquartile range. Dots represent outliers. i Correlation between IFNγ level and percent DR+CD38+ non-naïve CD8+ CX3CR1+ T cells in MIS-C patients (N = 7). Dots colored by CX3CL1 expression. R value computed using Pearson’s correlation coefficient after normality was demonstrated. Error band represents 95% confidence interval. j CX3CL1 levels between IFNγ-low (N = 13) and IFNγ-high groups (N = 7). P-values computed with Wilcoxon test. Horizontal line represents median, with bounds of box representing interquartile range. Whiskers represent 1.5x the interquartile range. Dots represent outliers. All p-values were calculated using two-sided tests. Source data are provided as a Source Data file.