Pregnancy-specific responses to COVID-19 revealed by high-throughput proteomics of human plasma

Abstract

Background: Pregnant women are at greater risk of adverse outcomes, including mortality, as well as obstetrical complications resulting from COVID-19. However, pregnancy-specific changes that underlie such worsened outcomes remain unclear.

Methods: Plasma samples were collected from pregnant women and non-pregnant individuals (male and female) with (n = 72 pregnant, 52 non-pregnant) and without (n = 29 pregnant, 41 non-pregnant) COVID-19. COVID-19 patients were grouped as asymptomatic, mild, moderate, severe, or critically ill according to NIH classifications. Proteomic profiling of 7,288 analytes corresponding to 6,596 unique protein targets was performed using the SOMAmer platform.

Results: Herein, we profile the plasma proteome of pregnant and non-pregnant COVID-19 patients and controls and show alterations that display a dose-response relationship with disease severity; yet, such proteomic perturbations are dampened during pregnancy. In both pregnant and non-pregnant state, the proteome response induced by COVID-19 shows enrichment of mediators implicated in cytokine storm, endothelial dysfunction, and angiogenesis. Shared and pregnancy-specific proteomic changes are identified: pregnant women display a tailored response that may protect the conceptus from heightened inflammation, while non-pregnant individuals display a stronger response to repel infection. Furthermore, the plasma proteome can accurately identify COVID-19 patients, even when asymptomatic or with mild symptoms.

Conclusion: This study represents the most comprehensive characterization of the plasma proteome of pregnant and non-pregnant COVID-19 patients. Our findings emphasize the distinct immune modulation between the non-pregnant and pregnant states, providing insight into the pathogenesis of COVID-19 as well as a potential explanation for the more severe outcomes observed in pregnant women.

Fig. 1. The plasma proteome of COVID-19 patients differs according to disease severity and pregnancy status.

a Illustration of the study design showing the number of non-pregnant controls (n = 41; 22 male, 19 female), non-pregnant COVID-19 cases (n = 52; 22 male, 30 female) pregnant controls (n = 29), and pregnant COVID-19 cases (n = 72) from whom peripheral plasma samples were profiled. b Gestational age at sampling (gray circle) and at delivery (green triangle) for each pregnant control (upper panel) and case (lower panel). c Principal component (PC) plot of the plasma proteome of all study samples according to PC1 and PC2. Black = control, red = case. Circle = non-pregnant, triangle = pregnant. Increasing shape size corresponds to increasing COVID-19 severity. d PC plot representing the relationship between the plasma proteome of all study samples according to PC1 and PC3. e Violin plot representing the relationship between PC3 and COVID-19 severity among all study samples.

Fig. 2. The plasma proteome shows increasing perturbation with COVID-19 severity in pregnancy.

a Graphical representation showing the comparison of plasma proteomes between each classified subset of pregnant COVID-19 cases and controls. b Volcano plot showing the proteins modulated in asymptomatic COVID-19 cases compared to controls. Red = proteins with q < 0.1 and fold change > 1.25, green = proteins with q ≥ 0.1 and fold change >1.25, gray = proteins with q ≥ 0.1 and fold change ≤1.25, blue = proteins with q < 0.1 and fold change ≤1.25. c Volcano plot showing the proteins modulated in mild COVID-19 cases compared to controls. d Volcano plot showing the proteins modulated in moderate COVID-19 cases compared to controls. e Volcano plot showing the proteins modulated in severe COVID-19 cases compared to controls. f Volcano plot showing the proteins modulated in critical COVID-19 cases compared to controls. g Comparison of the magnitude of proteomic changes among pregnant COVID-19 case subsets, using the comparison between critical cases vs. controls as the reference. Spearman’s correlation and p-value are provided for the asymptomatic vs. control, mild vs. control, moderate vs. control, and severe vs. control contrasts compared to the reference. The proteins included in this analysis (gray dots) are those 1,072 identified as differentially abundant in the comparison between pregnant critically ill cases vs. controls.

Fig. 3. The plasma proteome shows increasing perturbation with COVID-19 severity in non-pregnant individuals.

a Graphical representation showing the comparison of plasma proteomes between each classified subset of non-pregnant COVID-19 cases and controls. b Volcano plot showing the proteins modulated in moderate COVID-19 cases compared to controls. Red = proteins with q < 0.1 and fold change >1.25, green = proteins with q ≥ 0.1 and fold change >1.25, gray = proteins with q ≥ 0.1 and fold change ≤1.25, blue = proteins with q < 0.1 and fold change ≤1.25. c Volcano plot showing the proteins modulated in severe COVID-19 cases compared to controls. d Volcano plot showing the proteins modulated in critical COVID-19 cases compared to controls. e Comparison of the magnitude of proteomic changes among non-pregnant COVID-19 case subsets, using the comparison between critical cases vs. controls as the reference. Spearman’s correlation and p-value are provided for the moderate vs. control and severe vs. control contrasts compared to the reference. The proteins included in this analysis (gray dots) are those 2966 identified as differentially abundant in the comparison between non-pregnant critically ill cases vs. controls.

Fig. 4. The protein response to COVID-19 is dampened in pregnancy regardless of disease severity.

a Graphical representation showing the comparison of 486 plasma proteins that are modulated in both pregnant COVID-19 cases vs. controls and in non-pregnant COVID-19 cases vs. controls. b Heatmap representation of the 486 proteins with shared dysregulation between pregnant and non-pregnant COVID-19 patients. Clusters 1 and 2 include proteins with increased abundance while clusters 1 and 3 include proteins with decreased abundance in cases compared to controls. c Correlation between the magnitude of proteomic changes in pregnant moderate cases vs. controls and that in non-pregnant moderate cases vs. controls. Slope of the regression line (red line), Spearman’s correlation, and p-value are provided. Dotted blue line represents the parity line. d Correlation between the magnitude of proteomic changes in pregnant severe cases vs. controls and that in non-pregnant severe cases vs. controls. e Correlation between the magnitude of proteomic changes in pregnant critical cases vs. controls and that in non-pregnant critical cases vs. controls.

Fig. 5. The biological processes and pathways perturbed after COVID-19 differ between pregnant and non-pregnant patients.

a Volcano plot showing the proteins modulated in all pregnant COVID-19 cases compared to controls. Red = proteins with q < 0.1 and fold change >1.25, green = proteins with q ≥ 0.1 and fold change >1.25, gray = proteins with q ≥ 0.1 and fold change ≤1.25, blue = proteins with q < 0.1 and fold change ≤1.25. b Volcano plot showing the proteins modulated in all non-pregnant COVID-19 cases compared to controls. c Venn diagram showing the overlap of biological processes enriched among proteins modulated by COVID-19 between pregnant and non-pregnant cases compared to controls. d Bar plot showing the odds ratios for top biological processes enriched among proteins modulated by COVID-19 in pregnant cases compared to controls. Asterisk indicates the odds ratio calculated as infinite. e Bar plot showing the odds ratios for top biological processes enriched among proteins modulated by COVID-19 in non-pregnant cases compared to controls. f Bar plot showing the odds ratios for top biological processes enriched among proteins modulated by COVID-19 in both pregnant and non-pregnant cases compared to controls. g Venn diagram showing the overlap of C2 pathways enriched among proteins modulated by COVID-19 in pregnant and non-pregnant cases compared to controls. h Bar plot showing the odds ratios for top C2 pathways enriched among proteins modulated by COVID-19 in pregnant cases compared to controls. i Bar plot showing the odds ratios for top C2 pathways enriched among proteins modulated by COVID-19 in non-pregnant cases compared to controls. j Bar plot showing the odds ratios for top C2 pathways enriched among proteins modulated by COVID-19 in both pregnant and non-pregnant cases compared to controls.

Fig. 6. COVID-19 drives distinct and shared angiogenic and inflammatory profiles in pregnant and non-pregnant individuals.

a Representative diagram illustrating the comparison between pregnant and non-pregnant COVID-19 cases and controls for specific proteins associated with angiogenesis, endothelial dysfunction, and intravascular inflammation. A core set of 33 proteins that are significantly modulated with COVID-19 in opposite directions between pregnant and non-pregnant patients. Note the negative slope and correlation coefficient. b–g Violin plots showing the modulation of b sFLT-1, c AGT, d TNFRSF1B, e VWF, f ELANE, and g H3C1 levels with COVID-19 severity in non-pregnant and pregnant cases and controls. Black = control, gray = asymptomatic, blue = mild, yellow = moderate, red = severe, and brown = critical. RFU = relative fluorescence units.

Fig. 7. Pregnant women with COVID-19 display a dampened systemic cytokine response.

a Representative diagram illustrating the evaluation and comparison of specific cytokines in the circulation of pregnant and non-pregnant COVID-19 cases and controls. b–n Violin plots showing the modulation of b IL-6, c IL-1β, d IL-18, e TNF, f IL-17A, g IL-1α, h IFNγ, i IL-10, j TGFβ1, k CCL1, l CCL22, m CXCL13, and n CXCL10 levels with COVID-19 severity in non-pregnant and pregnant cases and controls. Black = control, gray = asymptomatic, blue = mild, yellow = moderate, red = severe, and brown = critical. RFU = relative fluorescence units.

Fig. 8. The plasma proteome allows for the identification of COVID-19 patients and can distinguish mild and severe diseases.

a Receiver operating characteristic (ROC) curves for discrimination of all COVID-19 cases from all controls (black curve), only pregnant COVID-19 cases from pregnant controls (red curve), and only non-pregnant COVID-19 cases from non-pregnant controls (blue curve). b Bar plot displaying the relative importance of the top 50 proteomic predictors for identifying all COVID-19 cases. c ROC curves for discrimination of severe/critical cases from controls (red curve), moderate cases from controls (yellow curve), and asymptomatic/mild cases from controls (blue curve). Data from both pregnant and non-pregnant cases are included in this analysis. d Bar plot displaying the relative importance of the top 50 proteomic predictors for distinguishing severe/critical COVID-19 cases from controls.