Main

Despite the ongoing coronavirus disease 2019 (COVID-19) pandemic, the effect of maternal SARS-CoV-2 infection on the immunology of the developing neonate is unclear. SARS-CoV-2 infection in pregnancy has been reported to lead to variable outcomes for the mother. The majority of infected pregnant women are asymptomatic or experience mild symptoms1. Nonetheless, pregnant women with SARS-CoV-2, particularly in the third trimester, are at increased risk for hospitalization and intensive care2, and existing evidence demonstrates that rates of maternal infection increased in the second wave3. Although the rates of preterm birth did not seem to be elevated in initial reports, data suggest pregnant women are at a higher risk of subsequent preterm birth, although this effect is mostly due to clinical intervention based on maternal illness rather than the spontaneous onset of preterm labor4.

With respect to the neonate, a national UK surveillance study suggested neonatal SARS-CoV-2 infection is uncommon even in neonates born to mothers with perinatal infection5. Similarly, a study from the United States of 116 mothers with confirmed perinatal SARS-CoV-2 infection did not identify any neonatal cases6. There are now clear criteria to define mother-to-child transmission of SARS-CoV-2 (https://www.who.int/publications/i/item/WHO-2019-nCoV-mother-to-child-transmission-2021.1/), and a small number of individual case reports have documented evidence of vertical transmission7,8,9. The results of these and other studies suggest that vertical transmission is possible, but infrequent.

Although vertical transmission of SARS-CoV-2 is infrequent, the potential immunological perturbations induced in the pregnant mother10 might leave an immunological legacy on the newborn with far-reaching consequences. Indeed, evidence exists that perinatal inflammation can modulate the developing immune system of neonates born prematurely11. The immune system of the fetus can be altered by the presence of human immunodeficiency virus (HIV) or hepatitis C virus (HCV) in mothers, either with or without vertical transmission12,13; metabolites derived from the maternal intestinal microbiota can shape the immune system of the offspring14; and that modulation of the neonatal immune system is associated with diseases in later life15. In the few studies assessing the immune status of neonates born to SARS-CoV-2-infected mothers, there has been little evidence of effect in the cellular and humoral immunity of the neonate16. Similarly, a small study of SARS-CoV-2 infection during pregnancy has been associated with a cytokine response in the fetal circulation (that is, umbilical cord blood) with no effect on the cellular immune repertoire17. However, to our knowledge, none of these studies have included a comprehensive analysis of the cellular immune profile of neonates born to SARS-CoV-2-exposed mothers (at any point in their pregnancy) compared to profiles of neonates born to unexposed mothers. Indeed, it is increasingly challenging to include an appropriate control group in a global pandemic. Additionally, neonatal T cell responses specific to SARS-CoV-2 peptides are yet to be investigated. Moreover, simultaneous analysis of cytokines and antibody titers in neonates and their paired mothers is lacking.

Here, we assess the immunological status of neonates born to mothers with SARS-CoV-2 who tested positive either during the 2 weeks directly before birth or earlier in pregnancy, compared with those born to mothers who were never exposed to SARS-CoV-2, to identify if there is a legacy of maternal infection and in utero priming of the neonatal immune system.

Results

Reduced antibody transfer in neonates born to mothers with recent or ongoing infection

Maternal and neonatal characteristics are shown in Table 1, and the group descriptions are available in Fig. 1a. Neonates born to mothers with SARS-CoV-2 exposure (SARS-CoV-2 exposed (SE), n = 30) were grouped into those born to mothers with recent or ongoing infection (R/O, identified by positive PCR test within 2 weeks of birth, n = 16), or those born to mothers testing positive earlier in gestation (recovered group (R), determined positive at least 2 weeks before birth, n = 14, median positive swab 48.5 d before birth).

Table 1 Maternal and neonatal clinical characteristics
Fig. 1: Neonates born to SE mothers are IgG+IgM and reduced antibody transfer was observed in the R/O group.
figure 1

a, Study outline illustrating recruitment of neonates and mothers in the SE group (n = 30), comprising two groups (R, n = 14; R/O, n = 16), and neonates in the NSE group (n = 15). Figure created with BioRender.com. b, Heat map displaying paired maternal and neonatal IgG and IgM levels against S, RBD and N SARS-CoV-2 epitopes measured in plasma from maternal blood and cord blood taken at the time of delivery. Peak IgG/IgM levels (fold change over background) are displayed within the neonate–mother dyads (each row; R, n = 14; R/O, n = 15), ordered by the number of days between maternal positive nasopharyngeal swab and birth. Fourfold change (over background) was defined as the IgG seropositivity threshold. c, Box plot of the mean IgG (average of IgG S, N and RBD) transfer ratio between the neonates and their paired mothers within the R (n = 14) and R/O (n = 15) groups. d, Maternal and neonatal paired peak IgG levels (fold change over background) in dyads with seropositive mothers only (fourfold change over background is indicated by the dashed gray line) within the R (IgG S, n = 11; IgG RBD, n = 9; IgG N, n = 9) and R/O (IgG S, n = 10; IgG RBD, n = 9; IgG N, n = 12) groups. Each line joins a neonate and their paired mother. Box plots follow standard Tukey representations; the central line denotes the median, upper and lower lines represent the 75th and 25th percentiles, respectively, and whiskers represent 1.5 times the 75th and 25th percentiles. P values were assessed by unpaired Wilcoxon rank-sum tests (c) and paired Wilcoxon tests (d). All statistical tests were two-sided and all P values (*P < 0.05; **P < 0.01; ****P < 0.0001) were unadjusted.

Source data

In both the recent/ongoing and recovered group, regardless of the time between the positive swab and birth or maternal IgM levels, SARS-CoV-2-specific IgM was not detected in cord plasma, suggestive of a lack of vertical transmission (Fig. 1b). This was true for IgM directed against the spike (S) protein, the receptor-binding domain (RBD) within the spike protein as well as the nucleoprotein (N) for which background IgM reactivity has been shown to be higher18. By contrast, SARS-CoV-2-specific IgG against all three antigens was detected in neonates born to SARS-CoV-2-exposed mothers (Fig. 1b). While there was no significant difference in SARS-CoV-2-specific IgG levels in the mother–neonate dyads in the recovered group, there were significantly lower levels of SARS-CoV-2-specific IgG in neonates born to mothers with recent/ongoing infection (Extended Data Fig. 1a). Thus, when the ratio of neonatal immunoglobulin to their paired maternal immunoglobulin (transfer ratio) was calculated for each antigen, the mean transfer ratio of all three antigens was significantly lower in the R/O group (0.79) compared to the R group (1.04; Fig. 1c). This was despite the presence of high levels of maternal IgG in at least some mothers in the R/O group (Extended Data Fig. 1a). Indeed, when only comparing IgG levels in neonates born to seropositive mothers, transfer of SARS-CoV-2-specific IgG to the neonate was significantly lower in the R/O group (Fig. 1d), whereas again the antibody levels were comparable between mother and neonate in the R group. Antibody transfer did not appear to differ with the sex of the newborn, with both male and female neonates in the R/O group having significantly less SARS-CoV-2-specific IgG compared to their seropositive mothers (Extended Data Fig. 1b).

Elevated plasma cytokine levels in mothers with recent or ongoing infection and their neonates

SARS-CoV-2 infection in adults is known to be associated with marked elevation of several plasma cytokines including interferon gamma-induced protein 10 (IP-10), interleukin (IL)-1β, CXCL8, IL-6 and IL-10 (refs. 19,20,21). Plasma cytokine concentrations were assessed both in maternal and cord blood using a multiplex assay. Maternal plasma cytokine analysis revealed significantly elevated IP-10 and IL-1β levels in mothers with recent or ongoing SARS-CoV-2 infection compared to recovered mothers, while IL-10, CXCL8 and IL-6 levels in maternal plasma were similar between the two SE groups (Fig. 2a). When assessing neonatal cytokine levels, IL-10 was significantly elevated in the cord plasma from neonates born to mothers with recent or ongoing infection compared to those born to recovered mothers, where no IL-10 could be detected. CXCL8 levels were also numerically higher in the R/O group (median, 93 pg ml−1) compared to the recovered group (median, 28 pg ml−1), although this did not reach significance, largely driven by three neonates with undetectable levels (Fig. 2b). However, concentrations of this chemokine were significantly higher in neonates than their paired mothers in both SE groups (Fig. 2c), which was not seen with any of the other cytokines tested. Conversely, neonatal IP-10 was equivalent between mothers and neonates in the recovered group and actually significantly lower in neonates than their paired mothers in the R/O group (Extended Data Fig. 2a). The majority of neonates born to recovered mothers who showed elevated levels of CXCL8 were born by vaginal delivery (Fig. 2d), known to elevate several cytokines22. However, there were still notable, albeit insignificant (P = 0.07, unpaired two-tailed Kolmogorov–Smirnov test), increases of CXCL8 in neonates born via cesarean section (CS) in the R/O group, compared to the R group, suggesting this was possibly a neonatal response to maternal SARS-CoV-2 infection.

Fig. 2: Elevated plasma cytokine levels in neonates and paired mothers within the R/O group.
figure 2

a,b, IP-10, IL-1β, IL-10, CXCL8 and IL-6 levels in mothers (a) and their neonates (b) within the R group (mothers—all cytokines, n = 13; neonates—CXCL8 & IL-6, n = 14, IP-10, n = 13; IL-1β, n = 12; IL-10, n = 11) and the R/O group (mothers—IL-6, n = 15; IL-10 & IL-1β, n = 14; CXCL8 & IP-10, n = 13; neonates—IP-10, CXCL8, IL-10 and IL-6, n = 16; IL-1β, n = 15) measured in plasma from maternal blood and cord blood taken at the time of delivery. c, Paired maternal and neonatal plasma CXCL8 levels within the R (n = 13) and the R/O (n = 14) groups. Each line joins a neonate and their paired mother. d, Neonatal CXCL8 levels in babies born via CS (left, n = 18) or vaginal delivery (VD; right, n = 12). Horizontal dotted lines represent the minimum (and maximum, for IL-6) detectable concentrations. e, Spearman correlation matrices of all significant (P < 0.05) correlations of neonatal and maternal cytokines in the R (left, n = 14) and R/O (right, n = 16) groups. fi, Spearman correlation plots between neonatal IP-10 and CXCL8 (f), maternal IL-1β and neonatal CXCL8 (g), and maternal IL-1β and neonatal IP-10 (h) in the R/O group (f: n = 15, g and h: n = 14) and between maternal IL-1β and days between maternal positive swab in the SE group (i; n = 27). Plots indicate a generalized linear regression line with shaded 95% confidence intervals. Associated correlation coefficients and P values are in orange (fh) and dark red (i). In the cytokine scatterplots (a, b and d), the central line represents the median value. P values were assessed by unpaired Kolmogorov–Smirnov tests (a, b and d) and paired Wilcoxon tests (c). All statistical tests were two-sided and all P values (*P < 0.05; **P < 0.01) were unadjusted.

Source data

Spearman rank test identified significant correlations in cytokine levels both within and between maternal and cord blood in both R and R/O groups, although this was much more evident in in the R/O group, suggestive of a greater degree of immune co-regulation in recent/ongoing SARS-CoV-2 infection (Fig. 2e). For example, within the R/O group, there were significant correlations between neonatal CXCL8 and IP-10 (Fig. 2f), and maternal IL-1β levels correlated with both neonatal CXCL8 and IP-10 (Fig. 2g,h). Interestingly, in all the SE neonates, maternal IL-1β levels negatively correlated (R = −0.49, P = 0.016, two-tailed Spearman correlation test) with days between a positive COVID swab and birth (Fig. 2i), suggesting this cytokine was indicative of recent infection in the mothers. No differences were seen in plasma IL-12p70, granulocyte–macrophage colony-stimulating factor (GM-CSF), interferon (IFN)-α2, IFN-λ1, IFN-λ2/3, IFN-β, tumor necrosis factor (TNF) and IFN-γ in mothers or neonates from either group (Extended Data Fig. 2b).

Recent or ongoing maternal infection affects neonatal immune cells

To investigate the potential impact of maternal SARS-CoV-2 infection on the cellular immune compartment of the neonate at birth, we used multiparametric flow cytometry to phenotype cord blood leukocytes and assess their in vitro functional capacity upon mitogen stimulation (Figs. 36). Gating strategies are shown in Extended Data Fig. 3. To establish if maternal SARS-CoV-2 infection altered the developing immune system of the neonate, cellular immune profiles in the combined (R and R/O) SARS-CoV-2-exposed group (SE; n = 30) were compared to those from term neonates born to healthy mothers collected before the pandemic (non-SARS-CoV-2-exposed (NSE), n = 15) but measured simultaneously. We performed t-distributed stochastic neighbor embedding (t-SNE) dimensionality reduction (Fig. 3a) on 91 individual flow cytometry immune parameters and observed that the immune profiles in the SE group clustered away from the immune profiles of the NSE group. Significant correlations (as assessed by two-tailed Spearman correlation tests) between immune parameters appeared different in neonates born to SE mothers compared to NSE mothers (Fig. 3b,c). For example, the significant correlation observed between IFN-γ-producing natural killer (NK) cells and CD8+ T cells in the NSE group was completely absent in the SE group, and a significant correlation appeared between the important effector populations, CD4+ terminal effector memory RA-positive (TEMRA) cells and CD8TEMRA cells in the SE group, where there had been no such correlation in the NSE group (Extended Data Fig. 4a,b). Indeed, when focusing on the immune parameters that drive the biggest differences, we identified that neonates born to mothers in the R/O group separated furthest away from the NSE group, using three-dimensional principal-component analysis (3D PCA; Fig. 4a), and that all three groups tended to segregate based on their maternal SARS-CoV-2 status upon unbiased hierarchal clustering analysis (Fig. 4b). Moreover, there was no clear clustering based on alternative confounding factors such as sex, ethnicity, mode of delivery and other maternal characteristics such as chorioamnionitis and gestational diabetes (Extended Data Fig. 5). Despite the observation that the neonates did segregate based on their mother’s SARS-CoV-2 status, many parameters, including those known to be perturbed in the adult response to SARS-CoV-2, were not significantly different between any of the groups. T cell lymphopenia was not observed and, moreover, the relative frequencies of major adaptive lymphocyte subsets in the neonate (for example, CD4+ and CD8+ αβ T cells, γδ T cells and B cells) were unaffected by maternal exposure to SARS-CoV-2 and were similar to those seen in the NSE group (Fig. 4c). Whereas these neonatal immune cell populations were not affected in totality by maternal SARS-CoV-2 exposure, more specific changes in adaptive cell populations were still evident. Thus, while the percentage of CD161-expressing CD8+ T cells was equivalent between NSE and R groups, there was a significant increase in this population in neonates born to mothers with recent or ongoing infection (Fig. 4d). Furthermore, the percentage of CD25+FOXP3+ regulatory T (Treg) cells was significantly increased in neonates from the R/O group compared to the NSE and R groups (Fig. 4e). When assessing all the neonates born to SARS-CoV-2-exposed mothers, the percentage of Treg cells significantly correlated with the percentage of Vδ2+ T cells (Fig. 4f).

Fig. 3: Immune differences between neonates born to mothers exposed to SARS-CoV-2 and unexposed mothers.
figure 3

a, t-SNE dimensionality reduction of neonatal immune profiles measured from CBMCs in neonates born to SE mothers (n = 28) and NSE mothers (n = 13). b,c, Spearman correlation matrix of significant (P < 0.05) correlations of neonatal cellular populations in the NSE (n = 14; b) the SE (n = 30; c) groups. Spearman correlation tests were two-sided and P values were unadjusted. TEM cell, effector memory T cell; TCM cell, central memory T cell; TN cell, naive T cell. MC, monocyte.

Source data

Fig. 4: Recent or ongoing maternal SARS-CoV-2 infection affects neonatal immune cells.
figure 4

a, 3D PCA dimensionality reduction of neonatal immune profiles in the R/O (n = 15), R (n = 13) and NSE (n = 13) neonatal groups. PC1, PC2 and PC3 explain 20.6%, 11.8% and 8.4% of the variance, respectively. b, Dendrogram cluster heat map of 45 flow cytometry immune populations in neonates within the NSE (n = 14), R (n = 14) and R/O (n = 15) groups. c, Box plots displaying the proportions of αβ T cells, CD4+ T cells, CD8+ T cells and γδ T cells (NSE, n = 15; R, n = 14; R/O, n = 16), and the proportion of B cells (NSE, n = 14; R, n = 14; R/O, n = 16). d,e, Box plots displaying the proportions of CD161+CD8+ T cells and FOXP3+CD25+ Treg cells (NSE, n = 15; R, n = 14; R/O, n = 16). f, Spearman correlation plot in the SE group (n = 30) of neonatal proportions of FOXP3+CD25+ Treg cells and Vδ2+ γδ T cells. Data indicate a generalized linear regression line with shaded 95% confidence intervals. Associated correlation coefficients and P values are in dark red. All box plots follow standard Tukey representations; the central line denotes the median, upper and lower lines represent the 75th and 25th percentiles, respectively, and whiskers represent 1.5 times the 75th and 25th percentiles. P values were assessed by unpaired Wilcoxon rank-sum tests (ce). All statistical tests were two-sided and all P values (*P < 0.05) were unadjusted.

Source data

Fig. 5: Increased proportions of innate immune cells in neonates born to mothers with recent or ongoing infection.
figure 5

a, Representative flow cytometry plots of the percentage of CD56+ NK cells (of CD3 live cells) in an NSE and R/O neonate, measured from CBMCs, and the box plots for the total neonates (NSE, n = 15; R, n = 14; R/O, n = 16). b,c, Box plots displaying the proportions of NK T cells and Vδ2+ γδ T cells (NSE, n = 15; R, n = 14; R/O, n = 16). d,e, Spearman correlation plots in the SE group (n = 30) between neonatal CXCL8 and the proportions of NK T cells (d) or CD69+ NK cells (e). Plots indicate a generalized linear regression line with shaded 95% confidence intervals. Associated correlation coefficients and P values are in dark red. All box plots follow standard Tukey representations; the central line denotes the median, upper and lower lines represent the 75th and 25th percentiles, respectively, and whiskers represent 1.5 times the 75th and 25th percentiles. P values were assessed by unpaired Wilcoxon rank-sum tests (ac). All statistical tests were two-sided and all P values (*P < 0.05; ***P < 0.001) were unadjusted.

Source data

Fig. 6: Increased percentages of cytokine-producing cells in neonates born to mothers exposed to SARS-CoV-2 after polyclonal stimulation.
figure 6

a,b, Representative flow cytometry plots of the percentage of TNF+ CD4+ T cells (a) and the percentage of IFN-γ+ CD8+ T cells (b) in an NSE, R and R/O neonate, measured from CBMCs. Box plots displaying the proportions of TNF+ (a) and IFN-γ+ (b) CD4+ T cells, CD8+ T cells, γδ T cells and NK cells (NSE, n = 15; R, n = 14; R/O, n = 16), as well as NK T cells (NSE, n = 15; R, n = 13; R/O, n = 15), following polyclonal stimulation (phorbol 12-myristate 13-acetate (PMA; 10 ng ml−1), ionomycin (1 μg ml−1), brefeldin A (20 ng ml−1) and monensin solution (2 μM) at 37 °C for 4 h). c,d, Spearman correlation plots between neonatal CD4+ TEM cells/TNF+CD4+ T cells (c), CD4+ TEM cells/IFN-γ+CD4+ T cells (d) in the SE group (n = 30). Plots indicate a generalized linear regression line with shaded 95% confidence intervals. Associated correlation coefficients and P values are in dark red. All box plots follow standard Tukey representations; the central line represents the median, the upper and lower lines represent the 75th and 25th percentiles, respectively, and whiskers represent 1.5 times the 75th and 25th percentiles. P values were assessed by unpaired Wilcoxon rank-sum tests (a and b). All tests were two-sided and all P values (*P < 0.05; **P < 0.01; ***P < 0.001) were unadjusted.

Source data

Increased percentages of innate immune cells in neonates born to mothers with recent or ongoing infection

In contrast to the relatively unaffected adaptive immune cell compartment, changes in innate cell subsets were more prominent following maternal SARS-CoV-2 exposure. NK, NK T and innate-like Vδ2+ γδ T cells were all significantly elevated in neonates born to mothers with recent or ongoing infection when compared to the NSE group (Fig. 5a–c). NK and Vδ2+ γδ T cells were also significantly elevated in neonates from the R/O compared to R group. There was also a change in monocyte populations with significantly enhanced percentages of alternative monocytes and subsequently significantly reduced CD38+ classical monocytes in neonates born to mothers with recent or ongoing infection compared to the NSE group (Extended Data Fig. 6a). These cells were, for the most part, not significantly elevated in neonates born to mothers with previous SARS-CoV-2 exposure (R) compared to the NSE counterparts, consistent with reactive neonatal responses to recent or ongoing SARS-CoV-2 infection. Indeed, when looking at all the neonates born to SE mothers, raised percentages of NK cells showed a negative correlation (R = −0.39, P = 0.04, two-tailed Spearman correlation test) with days from positive SARS-CoV-2 swab result to birth, further suggesting this was a neonatal response to maternal infection (Extended Data Fig. 6b). Interestingly, within the SE group, there was a significant correlation between the levels of cord blood CXCL8 and the percentage of NK T cells (Fig. 5d) and the concentrations of cord blood CXCL8 and NK cell activation (as assessed by CD69 expression; Fig. 5e) suggesting these are key immune markers associated with the neonatal response to maternal infection.

Increased cytokine functionality in neonates born to mothers exposed to SARS-CoV-2

When assessing functionality by intracellular cytokine staining (ICS) after polyclonal activation, the ability of immune cells to produce cytokines upon stimulation was significantly elevated in neonates born to SE mothers. Consistent with changes observed in plasma cytokine concentrations and cellular immune composition, enhanced cytokine production by neonatal immune cells was associated with maternal SARS-CoV-2 infection. Indeed, cytokine potential was significantly enhanced in several different cell types in neonates born to mothers previously exposed to SARS-CoV-2 (at any time point). Thus, the percentages of CD4+, CD8+, γδ, NK and NK T cells producing TNF were significantly increased in neonates in the R/O group compared to the NSE group, but also significantly increased in all these cell types in the R group compared to the NSE counterparts (Fig. 6a). The percentage of cells expressing IFN-γ was also significantly increased in CD4+, CD8+, γδ and NK T cells from the R/O group compared to the NSE group (Fig. 6b), and significant increased percentages of IFN-γ-producing CD8+ and NK T cells were seen in the R group compared to the NSE group (Fig. 6b). IL-17-expressing cells were also increased albeit to a lesser extent, with significant increases in CD4+ cells expressing IL-17 in the R/O group compared to the NSE group, and the percentages of IL-17-producing γδ T cells were significantly increased in R/O and R groups compared to the NSE group (Extended Data Fig. 6c). The percentages of CD4+, CD8+ or γδ T cells making CXCL8 was equivalent in all the groups (Extended Data Fig. 6d). The observed enhanced cytokine functionality (TNF and IFN-γ) in CD4+ T cells from neonates born to mothers exposed to SARS-CoV-2 significantly correlated with effector memory CD4+ T cells (Fig. 6c,d). Indeed, CD38 expression, known to decrease during maturation11, negatively correlated with both TNF and IFN-γ-producing CD4+ and CD8+ cells in the SE group (R = −0.45 to −0.59, P ≤ 0.01, two-tailed Spearman correlation tests; Extended Data Fig. 6e). Owing to limited numbers of mothers with severe disease, we were unable to establish if the extent of immune imprinting was related to maternal infection status.

Viral peptide responses detected in mothers exposed to SARS-CoV-2 but rarely in neonates

To assess whether the increased cytokine functionality observed in cells from SE neonates was due to the presence of SARS-CoV-2-specific T cells, we stimulated cord or peripheral blood mononuclear cells (PBMCs) from neonates and their seropositive mothers (n = 8), respectively, with SARS-CoV-2-specific peptides and assessed responses via IFN-γ FluoroSpot assay. All, except one, of the SARS-CoV-2-exposed mothers had positive responses against all the SARS-CoV-2 peptide pools with a combined median to the matrix and nucleocapsid, spike 1 (N terminus) and spike 2 (C terminus) peptide pools of 36 IFN-γ+ spots per 106 cells (range, 15–100) in R/O mothers (n = 4) and a combined median to the three peptide pools of 15 IFN-γ+ spots per 106 cells (range, 1–70) in R mothers (n = 4). However, no responses were detected to any SARS-CoV-2 peptides in the R/O, R or NSE (n = 5) neonatal groups with the exception of one neonate in the R/O group where a weak IFN-γ response to SARS-CoV-2 spike (N terminus) and matrix/nucleocapsid peptides was observed (Fig. 7).

Fig. 7: SARS-CoV-2-specific IFN-γ responses in mothers and neonates.
figure 7

a, IFN-γ+ spots per 106 cells detected by FluoroSpot assay following mononuclear cell stimulation with matrix and nucleocapsid, N terminus spike and C terminus spike peptide pools for 24 h at 37 °C in NSE neonates (n = 5; diamonds), SE mothers and their paired neonates (n = 8 pairs) grouped based on the maternal SARS-CoV-2 status (R/O, filled circles; R, open circles). Individual colors are paired to show the mother–neonate dyads. b, Representative FluoroSpot images of IFN-γ (FITC channel) in a mother with positive responses and their paired neonate with negative responses (top) and in the mother–neonate dyad where both individuals had positive responses. White arrows indicate additional IFN-γ+ spots in the neonate that have not been enlarged. Horizontal dotted lines represent the threshold level for positivity. In the scatterplots, the central line represents the median value. P values were assessed by paired Wilcoxon tests between paired SE mothers and their neonates (a). All statistical tests were two-sided and all P values (*P < 0.05; **P < 0.01) were unadjusted.

Source data

Discussion

Our study investigated the immune profiles of neonates born to mothers with SARS-CoV-2 exposure. While we did not observe SARS-CoV-2-specific IgM in the neonates and hence, by definition, saw limited vertical transmission of SARS-CoV-2, we did find multiple immunological perturbations within the neonate associated with maternal SARS-CoV-2 exposure during pregnancy, many of which were associated with recent or ongoing infections. Taken together, our findings are suggestive of an immunological legacy imprinted on the neonate following maternal SARS-CoV-2 exposure, with potential far-reaching consequences.

Although we did not directly assess neonates for the presence of SARS-CoV-2, we did assess SARS-CoV-2-specific IgM levels, which were negligible in all 30 neonates born to SARS-CoV-2-exposed mothers, suggestive of no vertical transmission. SARS-CoV-2-specific IgG was, however, transferred to the neonates from their mothers, suggestive of the transfer of protective immunity. There was a correlation between maternal and neonatal SARS-CoV-2 IgG levels in mother–neonate dyads in both groups as previously suggested23. However, even though for many pathogens, umbilical cord titers of IgG at normal-term delivery are higher than those in maternal blood24,25, there were reduced levels of SARS-CoV-2-specific IgG in neonates born to mothers with recent or ongoing infection compared to their paired mothers. This did not appear to be a threshold issue, as many mothers exhibited high levels of SARS-CoV-2-specific IgG, which was not transferred efficiently to their newborn. Reduced placental transfer of SARS-CoV-2-specific immunoglobulin has been observed previously26, possibly due to altered glycosylation27. Furthermore, reduced maternal SARS-CoV-2-specific antibody titers and impaired placental antibody transfer were also noted in pregnancies with a male fetus28, although there did not appear to be any sex bias in our dataset. It is currently unclear whether antibodies induced via vaccination, as opposed to natural infection, differ in terms of their glycosylation status and subsequent placental transfer, which is an important area of research. Vaccination to SARS-CoV-2 in the second and third trimester did elicit placental transfer of antibodies, with a reduced transfer ratio observed in the last trimester29. Our study adds further evidence suggesting that the second trimester may represent more opportune vaccination timing, at least with respect to the transfer of passive immunity to the neonate.

In utero exposure to environmental factors, infection and/or maternal inflammation is increasingly recognized to affect the developing immune system and subsequent responses both to infection30, immune-mediated diseases31 and neurodevelopmental problems32. Indeed, long-term effects cannot be ruled out as observed in survivors after in utero exposure to the 1918 (Spanish) influenza pandemic33. Perhaps unsurprisingly, cord plasma of neonates born to mothers with recent or ongoing infection, expressed elevated concentrations of some cytokines known to be associated with inflammation and COVID-19, consistent with placental immune activation19. Elevated levels of IP-10, IL-6, IL-10, CXCL8 and IL-1β have been associated with adult infection20,21, with increased IL-6 and IL-10 also associated with severity in SARS-CoV-2 infection in children34. Interestingly, this conventional COVID-19 signature was skewed more toward IL-10 (and to a lesser extent, CXCL8) in the neonates born to mothers with recent or ongoing infection. Indeed, increased IL-10 was observed in the R/O group, suggesting a potential imbalance between anti-inflammatory and pro-inflammatory cytokines. Theoretically, the increased concentrations of cytokines in the neonate could be explained by transfer of maternal cytokines through the placental tissues. Indeed, neonates with raised IP-10 concentrations were born to mothers who had high plasma IP-10 measurements, although in general concentrations were significantly lower in neonates compared to mothers. However, in the case of cord plasma CXCL8, a chemokine that correlated strongly with maternal IL-1β (previously associated with maternal SARS-CoV-2 infection35), levels were significantly higher than those observed in their mothers, suggesting that at least some of these elevated cytokines were neonatally derived, and a direct response to maternal infection. Elevated cord plasma cytokines, including CXCL8, have been previously observed in some neonates born to SARS-CoV-2-positive mothers17. Similarly, increased cord plasma CXCL8 has been observed after in utero exposure to infection36,37, suggesting that maternal infection can directly influence fetal CXCL8. Raised neonatal inflammatory markers do appear to be related to the maternal infection status at the time of birth, as such profiles are less obvious in neonates born to mothers with recovered infection. Indeed, where we did see slight elevations in CXCL8 in the recovered group, this could be explained by the mode of delivery, as labor is known to drive elevation of these cytokines22.

In contrast to COVID-19 infection in adults where reductions in peripheral NK, NK T, Vδ2+ and mucosal-associated invariant T cells have been observed20,38,39, neonates born to mothers exposed to SARS-CoV-2 actually exhibited elevated percentages of NK, NK T and Vδ2+ γδ T cells and also CD161-expressing CD8+ T cells (the majority of which are likely to be mucosal-associated invariant T cells). These innate-like cells may be responding to the inflammatory cytokine milieu in the context of maternal infection potentially as a protective response in the neonate vis-à-vis their likely beneficial role in severe adult COVID-19 (ref. 38). Immune activation of these cells has been seen at the maternal–fetal interface19, and although a previous report suggested there was no elevation of NK cells in neonates born to mothers with COVID-19 (ref. 16), this was only compared to reported reference levels and there was no direct comparator group in their study.

As well as alterations in these cell populations, we also identified enhanced cytokine potential upon in vitro stimulation. This was observed not only in neonates born to mothers with recent or ongoing infection but also in those born to recovered mothers, which suggests potential in utero priming of the immune response. Whether the immune profiles identified would be similar in neonates directly infected with SARS-CoV-2 after birth is unclear and this could not be assessed in our study. The percentages of CD4+, CD8+, NK, NK T or γδ T cells that produced TNF (or IFN-γ and IL-17 to a lesser extent) were significantly greater in neonates born to mothers who were exposed to SARS-CoV-2. At birth, neonatal T cells predominantly produce CXCL8 with a limited capacity for IFN-γ and IL-17 production, which increases with age40. Thus, these findings may reflect some accelerated maturation of the neonatal immune system induced in utero by maternal SARS-CoV-2 infection. Much of this enhanced cytokine functionality correlated with other markers of immune maturation such as increased percentages of memory T cells and decreased T cell expression of CD38. IFN-γ expression is controlled by epigenetic mechanisms in neonates41, so it is tempting to speculate that maternal SARS-CoV-2 infection may have induced some epigenetic changes in these loci. Indeed, maternal exposure to polycyclic aromatic hydrocarbons directly altered this locus in cord blood mononuclear cells (CBMCs)42, and other data suggest that epigenetic modifications during gestation can shape the future development of diseases like obesity, type 2 diabetes, allergy, asthma and infections43. Increased proportions of cytokine-producing T cells have also been observed following in utero exposure to infection44,45. Certain antigens are known to cross the placenta, and fetal priming of the immune system has been observed in response to bacteria detected in fetal organs46. Our observation that there was no specific reactivity to SARS-CoV-2 peptides in the majority of SE neonates tested (despite maternal positivity) suggests that the enhanced cytokine responsiveness detected is due to indirect exposure to inflammation/cytokines priming the immune response during fetal development. Nevertheless, SARS-CoV-2-reactive cells were detected in one neonate. This may corroborate that vertical transmission, albeit infrequent, can occur. Indeed, viremia is associated with severity47, and this neonate expressed the highest plasma levels of CXCL8, IL-6 and IL-10 of any of the neonates and was born to a pyrexic mother with placental abruption, which may have aided any viral transmission. By the same token, potential reactivity to viruses without obvious direct transmission has also been suggested by reduction in T cell receptor diversity observed in HIV-exposed uninfected neonates12.

Taken together, these data strongly suggest that maternal SARS-CoV-2 infection shapes the immune profile of neonates to different extents, depending on the time of exposure, and that this may be associated both with and without vertical transmission of SARS-CoV-2. We identified a transient response to maternal inflammation in the form of increased cytokines in cord plasma but also altered immune cell functionality in neonates exposed to SARS-CoV-2 at any point during gestation, suggesting some immune imprinting. While the etiology of the observed immune perturbations in the neonate remains unclear, the consequences could be far reaching. A hyper-inflammatory response similar to multisystem inflammatory syndrome in children has been observed in a neonate following in utero exposure to SARS-CoV-2 (mother infected 9 weeks before birth) with no evidence of direct neonatal infection48.

Long-term follow-up of the newborns in our study would establish if maternal exposure to SARS-CoV-2 has a long-lasting impact on the child. These data may also have implications regarding the vaccination regimen for pregnant women. Indeed, the reduced transfer of protective antibodies to the neonate we observed in those born to mothers with recent infection may suggest second trimester or early third trimester vaccination is preferable.

Methods

Study design and human material

This study complies with all ethical regulations and was approved by the relevant research ethics committee (REC) as subsequently described. Umbilical cord blood (and paired maternal peripheral blood) was collected over the COVID-19 pandemic (28 May 2020–1 March 2021) at the time of birth from neonates born to mothers who were exposed to SARS-CoV-2 (SE) attending the maternity unit at Guy’s and St Thomas’ NHS Foundation Trust (GSTT), London, in heparinized blood tubes at the time of birth to investigate the immune status (REC approval no. 19/SC/0232, Infectious Diseases Bio-Bank—Guys & St Thomas’s Hospital, South Central, Hampshire B REC). Written informed consent was obtained from all participants before study entry. Mother–neonate dyads were categorized into recent or ongoing (R/O) and recovered (R) groups, according to the number of days before birth that the mother received a positive status from a SARS-CoV-2 nasopharyngeal swab (R/O, <14 d; R, >14 d). Maternal and neonatal SE plasma samples were used for antibody and cytokine testing. Neonatal cell samples were used in flow cytometry experiments to measure neonatal cell populations and some (n = 8) were used in SARS-CoV-2-reactive cell testing experiments along with their paired mother cell samples (n = 8). Umbilical cord blood was also collected at the time of birth from neonates born to healthy mothers at GSTT before the COVID-19 pandemic (until 1 Jan 2020), hence their mothers were not infected with SARS-CoV-2 at any time during their pregnancy, termed the non-SARS-CoV-2 exposed (NSE) group (REC approval no. 17/LO/0641, London–Camberwell St. Giles REC). NSE samples were used as a control group for the flow cytometry. An additional group of NSE samples (n = 5) were used as a control group for the SARS-CoV-2-reactive cell testing experiments. The clinical details of the groups can be found in Table 1.

Isolation of cord blood mononuclear cells and plasma

CBMCs were isolated from all neonate groups. Peripheral blood mononuclear cells (PBMCs) were also isolated from eight SE mothers for use in SARS-CoV-2-specific FluoroSpot assays. Neat plasma was also isolated from the SE neonates and their paired mothers. More specifically, for the paired maternal peripheral blood samples, whole blood was centrifuged at 2,000g for 10 min at room temperature, and plasma was collected from the upper layer and stored at −80 °C in polypropylene tubes. For the neonatal samples, cord blood was layered onto Ficoll (GE Healthcare), within a 15-ml polypropylene conical tube, and centrifuged at 800g for 15 min (break OFF) at room temperature. Plasma was then collected from the top fraction and stored at −80 °C in polypropylene tubes. The CBMC layer was isolated and subsequently washed twice with pre-warmed base medium (RPMI-1640 + l-glutamine, Gibco), and then complete medium (CM; RPMI-1640 + 330 l-glutamine; 10% heat-inactivated FBS; 1% penicillin–streptomycin, Gibco), under centrifugation at 300g for 5 min at room temperature. For maternal PBMC isolation, the blood was diluted with half volume in PBS and layered onto Leucosep (Greiner) tubes containing 15 ml Ficoll. The PBMC layer was then isolated and washed twice with PBS under centrifugation at 300g for 5 min at room temperature. The cell pellets were then frozen in Cryostor CS10 (Sigma) within polypropylene cryovials before proceeding with flow cytometry.

Polyclonal stimulation

Before staining the CBMCs for flow cytometry immunophenotyping with panel 4 (Supplementary Table 1), cells were thawed and plated in 96-well round-bottom plates (Corning) within 200 μl CM containing PMA (10 ng ml−1; Sigma), ionomycin (1 μg ml−1; Sigma), brefeldin A (20 ng ml−1; Sigma) and monensin solution (1×; BioLegend). A brefeldin A-only and monensin-only control was also plated for each neonate. Cells were incubated at 37 °C for 4 h in the Biosafety level 3 (BSL-3) containment laboratory, in accordance with the King’s College London safety rules before proceeding with flow cytometry staining. Cells were removed from BSL-3 once they were fixed with Cell Fix (1×; BD) for a minimum of 10 min.

Flow cytometry staining and acquisition

CBMCs were thawed and plated in 96-well round-bottom plates before staining in one of four panels (Supplementary Table 1) assessing the following immune cell populations (panel 1, T cell naive/memory status; panel 2, myeloid and B cells; panel 3, T and NK cell activation status; panel 4, T and NK cytokine potential). All four panels contained surface-marker staining, and panel 1 and 4 also contained intracellular staining. All the following wash steps were performed under 2,000g, for 1 min at room temperature. For each panel, cells were washed with 100 μl Dulbecco’s PBS (1×; Gibco) and resuspended in 100 μl PBS, containing Zombie NIR Fixable Viability dye (1:1,000 dilution; BioLegend), with the addition of TCR Vδ1-FITC (TS8.2; Thermo Fisher) in panel 1, for 15 min in the dark at 4 °C. Cells were then washed with 150 μl eBioscience FOXP3 fixation/permeabilization buffer (Invitrogen), for panel 1, or FACS buffer (PBS, 0.5% heat-inactivated FBS, 2 mM EDTA, Invitrogen), for panels 2–4. The wash step was repeated with 200 μl volume. Cells were then resuspended in 100 μl eBioscience FOXP3 permeabilization buffer (FPB; Invitrogen), for panel 1, or 50 μl surface antibody cocktail within FACS buffer, for panels 2–4, for 30 min (panels 2–3) or 20 min (panel 4), at 4 °C in the dark. Cells were subsequently washed in 100 μl FPB (panel 1) or 150 μl FACS buffer (panels 2–4), and again with 200 μl volume before resuspending in 50 μl antibody cocktail in FPB (panel 1), or 100 μl Cell Fix (1×; panels 2–4) for 30 min (panels 1–3) or until ICS (panel 4) at 4 °C in the dark. Panels 1–3 were then washed twice in 100 μl FPB (panel 1) or FACS buffer (panels 2–3) and resuspended in 200 μl FACS buffer until acquisition. For the panel 4 ICS, cells were centrifuged at 2,000g, for 1 min at room temperature, and resuspended in 50 μl ICS antibody cocktail within Permeabilization Wash Buffer (1×; PWB; BioLegend) for 30 min in the dark at room temperature. Cells were washed twice in 150 μl PWB, followed by 200 μl FACS buffer before resuspending the cells in FACS buffer for acquisition in a high-throughput sampler on a four-laser LSR Fortessa (BD) using FACSDiva (v8, BD), at a flow rate of 1 μl s−1.

Flow cytometry data analysis

Raw FCS files were analyzed using FlowJo (v10.6.2, BD), and the gating strategies are included in Extended Data Fig. 3. As an internal control, the same adult sample was run alongside each flow cytometry experiment for consistency and to aid setting gates. Data were cleaned up by gating on the Time parameter to ensure that only cells going through a constant flow stream were analyzed, and cell populations were excluded from downstream analysis if the event count in the parent population was <30.

SARS-CoV-2-specific antibody testing

ELISA assays were conducted as previously described18,49. All plasma samples were heat inactivated at 56 °C for 30 min before use. High-binding ELISA plates (Corning, 3690) were coated with antigen (nuclear protein (N), spike glycoprotein (S) or the RBD at 3 µg ml−1 (25 µl per well) in PBS, either overnight at 4 °C or for 2 h at 37 °C. Wells were washed with PBS-T (PBS with 0.05% Tween-20) and then blocked with 100 µl of 5% milk in PBS-T for 1 h at room temperature. Wells were emptied and serial dilutions of plasma (starting at 1:25, fivefold dilution) were added and incubated for 2 h at room temperature. Control reagents included CR3009 (2 µg ml−1; N-specific monoclonal antibody), CR3022 (0.2 µg ml−1; S-specific monoclonal antibody), negative-control plasma (1:25 dilution), positive-control plasma (1:50 dilution) and blank wells. Wells were washed with PBS-T. Secondary antibody was added and incubated for 1 h at room temperature. IgG was detected using goat anti-human Fc-AP (1:1,000 dilution; Jackson, 109-055-098), and wells were washed with PBS-T. AP substrate (Sigma) was added and plates read at 405 nm. IgM was detected using goat anti-human IgM-HRP (1:1,000 dilution; Sigma, A6907) and wells were washed with PBS-T. One-step 3,3′,5,5′-tetramethylbenzidine substrate (Thermo Fisher Scientific) was added and quenched with 2 M H2SO4 before reading at 450 nm (HRP). Samples were run in duplicate and peak fold-change values above background were plotted for each patient. IgG transfer ratios were calculated to display the difference in paired neonate and maternal levels (neonatal IgG peak fold change over background/maternal IgG peak fold change over background).

Quantification of plasma cytokines

Paired maternal and neonatal plasma were thawed and tested in the 13-plex LegendPlex Human Anti-Virus Response Panel kit (BioLegend), to quantify levels of IL-1β, IL-6, TNF, IP-10, CXCL8, IL-12p70, IFN-α2, IFN-λ1, IFN-λ2/3, GM-CSF, IFN-β, IL-10 and IFN-γ. The assay was performed according to the manufacturer’s instructions and was modified by diluting the kit reagents 2× (beads, detection antibodies and streptavidin-PE). All plasma samples were diluted 2× with assay buffer, and resulting sample concentrations were calculated according to the dilution factor. In brief, 25 μl diluted plasma or standard, and mixed beads (1:1 ratio) were added to each well (in V-bottom 96-well plates) and incubated for 1.5 h. The samples were washed twice with wash buffer, incubated with 25 μl detection antibodies for 1 h and then 25 μl streptavidin-PE was added for a further 30 min. The samples were then washed once with wash buffer, resuspended in 200 μl wash buffer and acquired on a high-throughput sampler with a three-laser FACSCanto (BD). All incubation steps were performed under 600 r.p.m. on an orbital shaker at room temperature and protected from light. Data were cleaned up by excluding cytokines if the bead event count was <90, to ensure accurate analyses, performed using the Windows LegendPlex (v8.0, BioLegend) software.

IFN-γ+ SARS-CoV-2-positive cell detection by FluoroSpot assay

Cryopreserved CBMCs and PBMCs from SE neonates and their pared mothers, respectively, were thawed and stimulated with SARS-CoV-2 peptide pools to test for IFN-γ+-positive cells in FluoroSpot assays (Mabtech), adapted from Monin et al.50. Specifically, cells were stimulated with peptide pools spanning the SARS-CoV-2 Matrix and Nucleocapsid (M&N; Miltenyi), N-terminal domain of the SARS-CoV-2 spike protein (S1; JPT Peptide Technologies) or C-terminal domain of the SARS-CoV-2 spike protein (S2; JPT Peptide Technologies). CBMCs from NSE neonates were also tested. All FluoroSpot reagents were from Mabtech unless otherwise specified and used according to the manufacturer’s instructions. In brief, 96-well PVDF membrane plates were coated with anti-IFN-γ monoclonal capture antibody (1-D1K, FSX-01A-1; 15 μg ml−1) at 4 °C. Plates were washed 4× with 200 μl cold PBS and blocked with 200 μl medium for 1 h (RPMI-GlutaMAX (Gibco), 2% penicillin, streptomycin and amphotericin B (Gibco) and 10% human AB serum (Sigma). After thawing, cells were rested at room temperature for 1 h and plated at 1 × 105 or 2 × 105 cells per well in the coated plate at 100 μl final volume of medium containing M&N (1:200 dilution), S1 (1:2,000 dilution) or S2 (1:2,000 dilution) for 14 h at 37 °C in duplicate conditions. Dimethylsulfoxide was used as a negative control, and each individual sample was plated at an equal cell number per condition. The incubation was performed in the BSL-3 containment laboratory, in accordance with the King’s College London safety rules before fixing and reading the plates. After incubation, plates were washed 4× with 200 μl PBS before adding 100 μl anti-IFN-γ BAM monoclonal detection antibody (7-B6-1-BAM, FSX-01A-1; 1:200 dilution) in PBS-0.1% BSA for 2 h at room temperature in the dark. Plates were washed as above, and 100 μl anti-BAM-490 (FSX-01A-1; 1:200 dilution) in PBS-0.1% BSA was added for 1 h at room temperature in the dark. The wash step was repeated and 50 μl fluorescence enhancer was added for 10 min in the dark before removing the liquid and the plastic tray and drying the plates overnight. A total of 100 μl 2% paraformaldehyde (Alfa Aesar) in PBS was then added for 30 min. After fixing, the plates were dried and the number of IFN-γ+ spots per 106 cells was detected and analyzed using AID iSpot Spectrum reader and EliSpot software (v8.0 Autoimmun Diagnostika). A threshold level of positivity was set at >1 IFN-γ+ cells per well (equivalent to 2.5 cytokine-secreting cells per 106 cells).

Statistical analyses

Analyzed flow cytometry populations, plasma cytokines and antibodies were imported into an Excel spreadsheet and analyzed in R (v4.0.3) to generate box plots, dimensionality reduction plots, Spearman correlation plots and heat maps. Clustered heat maps were performed on scaled and centered data using the heatmap.2 function (gplots package), and clustered according to the Euclidean method. The corrplot package was used to generate Spearman correlation matrices, and only significant values (P < 0.05) are displayed. GraphPad Prism (v9.0) was also utilized to generate scatterplots for cytokines and the antibody heat map. All statistical tests were used to measure the differences between biologically distinct samples. Unadjusted P values (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001) were assessed by the Kolmogorov–Smirnov test (to compare cytokine concentrations between the groups), two-sided Wilcoxon rank-sum tests (for immune cell populations between the groups) and two-sided paired Wilcoxon tests (between paired maternal and neonatal antibody/cytokine levels).

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.