Antibody mediated activation of natural killer cells in malaria exposed pregnant women

Immune effector responses against Plasmodium falciparum include antibody-mediated activation of innate immune cells, which can induce Fc effector functions, including antibody-dependent cellular cytotoxicity, and the secretion of cytokines and chemokines. These effector functions are regulated by the composition of immunoglobulin G (IgG) Fc N-linked glycans. However, a role for antibody-mediated natural killer (NK) cells activation or Fc N-linked glycans in pregnant women with malaria has not yet been established. Herein, we studied the capacity of IgG antibodies from pregnant women, with placental malaria or non-placental malaria, to induce NK cell activation in response to placental malaria-associated antigens DBL2 and DBL3. Antibody-mediated NK cell activation was observed in pregnant women with malaria, but no differences were associated with susceptibility to placental malaria. Elevated anti-inflammatory glycosylation patterns of IgG antibodies were observed in pregnant women with or without malaria infection, which were not seen in healthy non-pregnant controls. This suggests that pregnancy-associated anti-inflammatory Fc N-linked glycans may dampen the antibody-mediated activation of NK cells in pregnant women with malaria infection. Overall, although anti-inflammatory glycans and antibody-dependent NK cell activation were detected in pregnant women with malaria, a definitive role for these antibody features in protecting against placental malaria remains to be proven.

www.nature.com/scientificreports/ Antigen-specific Ab engagement with FcγRIIIa on NK cells was recently identified as a key vaccine-induced functional immune responses linked to protection by RTS,S/AS01, the only licensed P. falciparum vaccine 30 . In addition, in vitro assays demonstrated the ability of NK cells to kill IEs via ADCC, and IgG Abs to PfEMP1 were sufficient to promote NK-dependent growth inhibition of P. falciparum in IEs 31 . This study also showed that naturally acquired IgG of multigravid women specific for VAR2CSA promotes NK-dependent lysis of IEs 31 . The ability of IgG Abs against the DBL2 and its flanking ID regions of VAR2CSA to induce ADCC is still unexplored 32 , but is of special interest, since the two leading placental malaria vaccine candidates PRIMVAC (Institut National de la Santé et de la Recherche Médicale, France) and PAMVAC (University Hospital Tuebingen, Germany) both include DBL2 domains 33,34 .
Fc effector functions such as ADCC are regulated through multiple structural and genetic components of the Ab, FcγR, and effector cell 35 , including post-translational modifications of glycans on the Fc domain of Abs, specifically at asparagine 297 on IgG 36 . Multiple factors can influence glycosylation patterns of IgG Abs including age, sex 37 , epigenetics 38 , disease state 39,40 , infection [41][42][43] , or vaccination 44 . Glycosylation patterns of IgG Abs can also undergo temporary changes during pregnancy, when galactosylation and sialylation of IgG Abs increase 45,46 . This has been associated with a less inflammatory profile 47 , which may contribute to acceptance of the placenta by the maternal immune system during pregnancy 48,49 . Changes in the composition of the asparagine 297 glycan can also influence the binding affinity of IgG Abs to FcγRs, and thereby change the magnitude of effector functions initiated, including ADCC and Ab-dependent cellular phagocytosis 50 . Human NK cells primarily express one Fc gamma receptor (FcγRIIIa), and responses through FcγRIIIa are highly regulated by IgG N-linked glycosylation, more so than any other human FcγR [51][52][53] . Some studies suggest that the presence/absence of key glycoforms can modulate FcγR affinity and ADCC activity by up to 20-fold 36,51,54,55 .
Here, we investigate the ability of IgG Abs of pregnant women from a malaria-endemic area specific to DBL2 and DBL3 (both VAR2CSA domains) to activate human primary NK cells from malaria-naïve donors to secrete IFNγ and TNFα cytokines, and upregulate CD107a expression, which is a surrogate for granzyme B degranulation and ADCC activity 56 . In addition, we evaluated pregnancy-associated glycosylation patterns of IgG Abs and their effect on NK-mediated effector functions in the context of P. falciparum infection during pregnancy.

Results
Primary human NK cells are activated by DBL2 or DBL3-specific IgG Abs from pregnant women with malaria. NK cells are major innate immune mediators of cytotoxicity. To evaluate the capacity of DBL2 and DBL3-specific IgG Abs to induce NK-mediated effector functions, we used purified IgG from two groups of pregnant women at mid pregnancy with peripheral P. falciparum parasitemia at delivery, and who were either positive (N = 50) or negative for P. falciparum IEs in the placenta (N = 27) (Fig. 1b).
We modified previously described Ab-dependent NK cell activation assays that have been utilized to assess responses to influenza, human immunodeficiency virus (HIV) and Mycobacterium tuberculosis proteins [24][25][26]57 for the use with VAR2CSA domain antigens (Fig. 1a). DBL2 was chosen because of its relevance in the development of placental malaria vaccines 33,34 . DBL3 is another domain of the VAR2CSA protein, which can be recognized by IgG Abs generated by pregnant women with malaria 58 . We characterized the ability of Abs against these domains to activate primary human NK cells, isolated from the blood of three malaria-naïve healthy donors. NK cells were identified via flow cytometry (Fig. 2a) and the levels of Ab-mediated NK cell activation in response to DBL2 and DBL3 were measured as indicated by intracellular cytokine production of IFNγ and TNFα, and the upregulation of cell surface degranulation marker CD107a (Fig. 2b-c). In order to optimize the Ab-dependent NK cell activation assay for malaria antigens, DBL2 (50-300 ng/well), DBL3 (50-300 ng/well) and IgG Ab (0.0625-0.25 mg/ml) concentrations were first titrated using four individual control Ab samples from pregnant women with malaria and a malaria-naïve individual ( Fig. S1a-d).
We evaluated purified IgG from pregnant women at mid pregnancy with peripheral P. falciparum parasitemia at delivery, and who were either positive or negative for P. falciparum IEs in the placenta. Their purified IgG was assessed in the presence of DBL2 or DBL3 for induction of Ab-mediated NK cell activation (Fig. 3). For both antigens, we observed upregulation of NK cell degranulation (CD107a; Fig. 3a,d, DBL2: p-value = 0.0198, DBL3: p-value = 0.0006) mid pregnancy in pregnant women who have non-placental malaria at delivery compared to non-pregnant malaria-naïve healthy individuals. In addition, DBL3-specific IgG mid pregnancy from pregnant women with non-placental malaria at delivery induced significantly higher IFNγ and TNFα production (Fig. 3e,f, IFNγ: p-value = 0.0322, TNFα: p-value = 0.0184) compared to IgG from non-pregnant malaria-naïve healthy individuals. Relative to IgG from non-pregnant malaria-naïve healthy individuals, IgG from pregnant women with placental malaria were associated with significantly higher NK cell degranulation (CD107a upregulation; Fig. 3a, p-value = 0.0342) in response to DBL2 and TNFα production (Fig. 3f, p-value = 0.0077) in response to DBL3 antigen. Differences in NK cell degranulation or cytokine production between pregnant women with non-placental malaria and women with placental malaria were only observed in CD107a expression (Fig. 3d, p-value = 0.0393) in response to DBL3.
Ab-dependent NK cell activation assays were validated with the use of two control antigens. In the presence of IgG Abs from malaria-naïve healthy individuals, influenza H3 (positive control) induced NK cell activation 26 , whereas the negative control SIV gp120 did not (Fig. S2a-   www.nature.com/scientificreports/ In addition, DBL2-and DBL3-binding capacities of IgG1-4 subclasses from pregnant women with placental (N = 50) and non-placental malaria (N = 27) were investigated via multiplex assays and correlated to the expression of CD107a, IFNγ and TNFα by Ab-activated NK cells (Figs. S3 and S4). The majority of non-placental malaria Ab-mediated NK cell activation was driven by IgG1; however, these correlations, if present, were weak to moderate (max Spearman ρ = 0.553, p-value = 0.0051). No significant correlations between IgG subclasses and Ab-mediated NK cell activation were observed for the placental malaria cohort, suggesting that other Ab features in addition to IgG subclasses may contribute to the modulation of Ab-mediated NK cell activation.
These results show that IgG to DBL2 and DBL3 from malaria-exposed pregnant women can activate NK cells, but that DBL2-or DBL3-specific Ab-mediated NK cell activation does not appear to predict subsequent placental malaria.
Polyfunctional NK cell activation profiles in pregnant women with malaria. Although levels of individual DBL2-or DBL3-specific NK cell activation markers did not differ the groups of pregnant women with and without placental malaria, it is possible that the proportion of activated NK cells expressing multiple activation markers ("polyfunctional NK cell activation") differed between the two groups of pregnant women. We therefore assessed the polyfunctional ability of NK cells to secrete TNFα, IFNγ and/or to express CD107a in different combinations. Activated NK cells were selected based on their CD56 and CD16 expression (CD56 dim CD16 bright and CD56 bright CD16 neg/dim ) (Fig. S5). The levels of CD56 expression have been associated with NK effector function 59 . CD56 bright NK cell subsets have been combined here due to low cell numbers, but are mainly characterized by their poor cytotoxic capacity and their high capacity to secrete several types of postactivation cytokines 60 . The CD56 dim CD16 bright NK cell population represents around 90% of peripheral blood NK cells and exhibit potent cytotoxic activity 60,61 .
The polyfunctional ability of IgG Abs from pregnant women with non-placental malaria or placental malaria to activate NK cells was not significantly different ( Figs. 4b and 5b). These findings indicate that IgG Abs against www.nature.com/scientificreports/ DBL2 and DBL3 from malaria-exposed pregnant women can induce NK cells, with the main response being an upregulation of CD107a expression, but also with a small subset inducing polyfunctional NK cell responses.   www.nature.com/scientificreports/ N-acetylglucosamine (GlcNAc) with variable additions of sugars such as fucose, galactose, sialic acid and bisecting GlcNAc (Fig. 6a) 36 . These post-translational modifications tune the affinity of IgG Abs for FcγRs, such as FcγRIIIa on NK cells, and regulate effector function 25,62,63 . We evaluated N-linked glycosylation patterns of IgG Abs in pregnant women using plasma collected at 14-26 weeks' gestation. Samples from pregnant women with P. falciparum infection (N = 11) and uninfected pregnant women (N = 41) at enrolment were analyzed, together with malaria-naïve healthy pregnant women (N = 10) and non-pregnant women controls (N = 13). N-linked glycosylation profiles were analyzed via microchip capillary electrophoresis-laser-induced fluorescence. No statistically significant differences were observed between the two groups of pregnant women (Fig. 6b-l), suggesting that malaria infection in second trimester of pregnancy does not change the total IgG N-linked glycan profile. However, comparing N-linked glycosylation profiles of IgG from pregnant women, regardless of infection or malaria exposure status, with the profiles of uninfected non-pregnant women, the total IgG of pregnant women exhibited a higher degree of total galactosylation (Fig. 6b,  Examining the distribution of specific glycan structures, significantly decreased proportions of G0 and G1F glycan structures were observed in pregnant women compared to non-pregnant women (Fig. 6e,h), whereas elevated proportions of G2 and G2S1, structures were observed in pregnant women compared to non-infected non-pregnant women (Fig. 6i-l). However, differences between malaria-naïve healthy pregnant women and malaria-exposed pregnant women were observed for G1 and G2F glycan structures (Fig. 6g,j), suggesting that malaria-naïve healthy pregnant women have slightly higher anti-inflammatory glycan structures (more galactose and fucose) in comparison to malaria-exposed pregnant women. Placental malaria infections may induce slightly more inflammatory glycan structures within pregnant women. However, no significant differences were observed for total galactose or fucose glycan structures were compared healthy pregnant control (Fig. 6b,d).
N-linked glycosylation profiles of purified total IgG from pregnant women at 14-26 weeks of gestation were also assessed to determine if they could predict future clinical outcome (placental malaria status), with no significant differences observed (Fig. S6a-k). We did note that similar differences in N-linked glycosylation profiles were maintained between infected pregnant and non-infected non-pregnant women. Furthermore, DBL2-and DBL3-specific Ab-mediated expression of activation markers by NK cells and IgG N-linked glycosylation patterns of malaria-exposed pregnant women did not correlate (Fig. S6), except for DBL2-mediated IFNγ production and total fucose (ρ = − 0.4255, p-value = 0.0108) along with a trend for DBL2-mediated CD107a expression and total fucose (ρ = − 0.3744, p-value = 0.0543), and DBL2-mediated TNFα production and total galactose (ρ = 0.3625, p-value = 0.0323). This suggests that the N-linked glycosylation profiles of total IgG in pregnant women may be more influenced by pregnancy than by malaria infection.

Discussion
Naturally acquired immunity to malaria is complex, and likely requires a combination of cell-mediated and humoral immune responses, including the secretion of cytokines, cellular cytotoxicity, and production of functional Abs in order to efficiently clear parasites 18,64 . It has previously been shown that antigen-specific Ab-mediated phagocytosis and engagement with FcγRIIIa on NK cells are linked to protection by the sporozoite-based malaria vaccine RTS,S/AS01 30 . In addition, Ab-dependent NK cell cytotoxicity towards IEs in malaria-exposed individuals can inhibit P. falciparum growth 31 . Furthermore, adaptive NK cells, a sub-population of differentiated specialized NK cells, were associated with lower parasitemia and protection against malaria infection through enhanced ADCC response to IEs in the presence of naturally acquired Abs from malaria-resistant individuals 32,65 . The potential of NK cell-mediated ADCC to protect individuals against placental malaria is still to be determined. A limited number of studies have investigated NK cells in the placenta and in the blood at various timepoints during malaria infections [66][67][68][69] , however none have considered the implications of ADCC. Here, we demonstrated that Abs generated by pregnant women with malaria from a malaria endemic area in their second trimester against the VAR2CSA subdomains DBL2 or DBL3 were able to induce NK cell activation, but no significant differences in responses were associated with susceptibility to subsequent placental malaria. We observed that the majority of Ab-mediated NK cell activation in women with placental malaria was driven by IgG1, even though a recent study identified malaria-specific IgG1 and IgG3, and engagement with FcγRIIIa (linked to Ab-mediated NK cell activity), as key prediction parameters for protection in malaria RTS,S/AS01 vaccinees 30 . However, even though IgG3 shows high affinity to FcγRs and especially to FcγRIIIa, within this study, it did not correlate with NK cell activation in pregnant women. This suggests that IgG subclass distribution may not be the only factor that modulates NK cell activation during pregnancy, and that other IgG features, such as N-linked glycans may impact Ab-mediated NK cell activation. Currently, the importance of VAR2CSA-specific Ab-mediated responses in protection from placental malaria is unclear, as Ab responses to recombinant VAR2CSA antigens at delivery are associated with the presence of placental infection, and may represent markers of infection, rather than correlates of protection 14 . In addition, there is no malaria-specific antigen, which can be used as a universal antigenic control to assess blood stage parasite Ab-mediated NK cell responses.
Within our study, Ab-mediated activation of NK cells was largely associated with the upregulation of CD107a expression, a surrogate marker of ADCC activity 56,70 , with smaller fractions of activated NK cells producing only IFNγ and TNFα or in combination with CD107a, suggesting that the majority of Ab-activated NK cells were potentially cytotoxic in the absence of inflammation. We speculate that this balance of Ab-mediated activation may be beneficial, as excessive secretion of pro-inflammatory cytokines, such as IFNγ and TNFα, in the placenta of malaria-infected women, especially in primigravidae, has been associated with placental pathology and adverse www.nature.com/scientificreports/  www.nature.com/scientificreports/ clinical outcomes 27,71 . Placental malaria is associated with activation of pro-inflammatory host cells, such as monocytes and macrophages causing inflammation of the placenta 8,72 . Our findings suggest that Ab-mediated activation of NK cells, potentially does not contribute to the overproduction of pro-inflammatory cytokines and resulting pathologies. Several studies suggest that progesterone, estrogen and cortisol dampen NK cell cytotoxic activities during pregnancy [73][74][75] . However, the increased concentration of cortisol during pregnancy could also inhibit NK cell activity against P. falciparum IEs 76 . Here, NK cells purified from whole blood of malaria-naïve healthy donors were used instead of pregnant women with malaria, and effects of pregnancy-associated hormones were not represented by non-pregnant malaria-naïve healthy donor blood NK cells.
We would like to acknowledge limitations of our study. NK cells from healthy malaria-naïve donors were used instead of pregnant women with malaria. This could have skewed for specific NK cell subsets, which may be underrepresented during pregnancy. Our work studied peripheral NK cells which may have substantially different responses to uterine NK cells and adaptive NK cells. Uterine NK cells are functionally different, do not circulate outside the uterus, and are more difficult to access for functional studies 77 . The analysis of peripheral NK cells is however relevant in that they can access the site of infection, the syncytiotrophoblast cell surface lining of the maternal peripheral blood 8 . Nevertheless, in future studies, parasite loads in the placenta at birth, or other clinical markers of disease severity, could also be considered, but this information was not collected for the majority of individuals in this cohort.
One limitation regarding the plate-bound Ab-mediated NK cell assay is that it does not mimic the interaction between NK cell and IEs as shown before 31,78 . However, our study is complementary to current placental malaria vaccine studies, which also only use DBL2 antigens 33,34 , and in vitro NK cell ADCC assays could be used for high-throughput screens of serum samples 32 . Furthermore, additional activation markers of NK cell subpopulations, such as CD57, CD25, CD69 or the inhibitory receptor programmed death-1 (PD-1) could be considered for future studies 23,[78][79][80] .
IgG N-linked glycosylation profiles can influence the engagement of IgG Abs with FcγRIIIa on NK cells 36 . Surprisingly, we did not observe any differences in Fc N-linked glycan profiles between pregnant women infected with P. falciparum and their non-infected counterparts. Consistent with previous studies 45, 46 , we observed a higher degree of galactosylation and sialylation of IgG Abs from pregnant women, regardless of malaria infection, compared to non-pregnant women. Overall, these results suggest that anti-inflammatory Fc N-linked glycans are elevated in both healthy and malaria-exposed pregnant women, which may dampen the Ab-mediated activation of NK cells in pregnant women with malaria infection. These changes have been associated with a less inflammatory profile during pregnancy. Fc N-linked glycan patterns of IgG Abs can be globally modulated during the course of inflammation, autoimmune disease or pregnancy 46,81 . For example, in patients with lupus erythematosus or rheumatoid arthritis (RA), reduced galactosylation and sialylation of IgG Abs correlates with proinflammatory immune responses and disease severity 82 . Intriguingly, the majority of pregnant women with RA undergo pregnancy-induced remission, which occurs simultaneously with the upregulation of IgG galactosylation and sialylation, such that inflammatory RA-associated glycosylation patterns are masked by pregnancy [83][84][85] . We observe a similar increase in galactosylation and sialylation of the IgG Abs in pregnant women, regardless of malaria infection status 46,81 . We acknowledge that the malaria-naïve healthy pregnant women in our study are more progressed in their pregnancy, which could affect the glycosylation profiles. However, these Fc N-linked glycans may explain why secretion of pro-inflammatory cytokines was suppressed from NK cells within our assays.
In healthy pregnancy, highly galactosylated Abs may be more effectively transferred across the placenta and may be able to mediate CD107a degranulation of both maternal and cord NK cells 86 . The transfer of maternal Abs across the placenta is mediated by binding to the neonatal Fc receptor (FcRn), which is a key process for neonatal immunity, as neonates cannot sufficiently generate IgG Abs 87 . However, contradictory roles for IgG glycosylation on FcRn binding have been reported 86,88,89 , including studies which show significant Ab galactosylation-driven changes in FcRn affinity and NK cell-activating Abs are selectively transferred across the placenta 86,88 , while another study showed that placental IgG transport is not Fc glycosylation selective 89 . In addition, Jennewein et al. considered transfer of maternal Abs across the placenta via binding to FcRn and FcγRIIIa, while other more recent studies suggest that FcγRIIIa do not play a role in maternal-fetal Ab exchange 87 . Defining the mechanisms of placental transfer, including the role of Fc glycosylation, may offer novel insights for the rational development of maternal vaccines to enhance transfer of protective Abs to fetuses and reduce their vulnerability 86,90 , and should be considered in the further development of vaccine candidates.
A limitation of capillary electrophoresis-laser-induced fluorescence is that not all Fc N-linked glycan profiles are clearly detectable. However, these additional patterns make up only a small fraction of human IgG Fc N-linked glycans 91 , and more sensitive techniques such as liquid chromatography mass spectrometry require extensive protein clean up and in-solution digestion, in-depth proteome and glycoform analysis 92 . Furthermore, the evaluation of DBL2-and DBL3-specific IgG glycosylation profiles would allow us to more accurately assess the contribution of N-linked glycans to Ab-mediated NK cell activation, unfortunately these assays require large volumes of plasma samples 25,93 , which were not available for this cohort. Previous studies examining antigen-specific N-linked glycosylation of IgG from HIV-infected pregnant women have observed significantly different profiles between HIV, tetanus and pertussis toxin specific-IgG 94 , thus future studies where adequate sample is available should assess for malaria-specific IgG glycan patterns. Vaccine studies assessing healthy non-pregnant volunteers have demonstrated that antigen-specific IgG glycosylation profiles can be modulated by vaccination 44 . It is still unclear if antigen-specific IgG glycosylation profiles can be modulated in pregnant women, or if pregnancy-associated global glycan changes will mask any antigen-specific glycosylation effects as observed in RA [83][84][85] . Determining if antigen-specific Ab glycosylation patterns are associated with clinically relevant outcomes of placental malaria could inform the design of the next generation of maternal vaccines 95 . Overall, our study highlights the necessity to better understand Ab effector functions, such as Ab-mediated NK cell activation, and the potential effect of In this cross-sectional study, parasitemia status was determined at delivery by light microscopy of Giemsastained peripheral blood smears, as well as by polymerase chain reaction of peripheral blood at delivery 97 . Samples were categorized based on the presence of P. falciparum parasites in peripheral blood at delivery. Groups included women who were positive for P. falciparum IEs in the placenta (placental malaria, N = 50) or women who were positive for P. falciparum IEs in peripheral blood but did not show any sequestering of IEs in the placenta (non-placental malaria, N = 27) (Fig. 1). The groups were frequency matched for primigravidae, age, bed net use, rural residency and type of malaria preventive treatment received (Table 1). Ethical approval was obtained from the PNG Institute of Medical Research Institutional Review Board, the PNG Medical Research Advisory Council, and the Melbourne Health Human Research Ethics Committee.
Plasma samples from malaria-naïve healthy Melbourne donors (N = 8) were chosen as negative controls, because many matched women from PNG would have been exposed to malaria and skewed negative responses. For the Fc N-linked IgG glycan profiling, samples of malaria-naïve healthy non-pregnant women (N = 13) were used (age: 34.3 ± 7.7 years). Plasma samples from individual healthy Melbourne donors were obtained in accordance with the University of Melbourne Human ethics approval (#1443420) and the Australian National Health and Medical Research Council Statement on Ethical Conduct in Human Research. Samples from malaria-naïve healthy pregnant women at the end of their second/beginning of their third trimester (N = 10) were obtained to compare pregnancy-specific Fc N-linked IgG glycan profiles (age: 31.3 ± 2.8 years; mean gestational age: 196 ± 4 days). Ethical approval was granted by the Mercy Health Board Human Research Ethics Committee (R10/16). All participants provided written informed consent.
IgG antibody purification. IgG Abs were purified from plasma of pregnant women at enrolment according to manufacturer's protocol via Melon Gel chromatography (Melon Gel IgG Purification Kit, Thermo Fisher Scientific, USA) 98 . IgG Ab samples were centrifugated through 100 kDa Amicon Ultra filters (Merck & Co, USA) at 14,000 × g for 10 min to remove excess albumin proteins and buffer exchanged into phosphate buffered saline (PBS). The IgG concentration and purity were quantitated using a human IgG ELISA development kit (Mabtech AB, Sweden). The IgG Ab samples were diluted in PBS to adjust Ab concentration to 0.25 mg/ml for Ab-dependent NK cell activation assays and 2 mg/ml for N-linked glycan profiling. The samples were stored at − 20 °C until further use.  100 (200 ng/well) at 4 °C for 12 h. Bovine serum albumin (BSA; Sigma-Aldrich, USA) was used to control for unspecific binding. Simian immunodeficiency virus (SIV) envelope protein gp120 (Sino Biological Inc., China) and influenza hemagglutinin (H)3 (A/Switzerland/9715293/2013; Immune Technology Corp., USA) were used as negative and positive antigen controls, respectively. H3 was selected as a universal technical control, as all individuals have been previously exposed to influenza, and Abs to H3 are highly cross reactive and strong inducers of NK cell activation 24 .

Multiplex assays of antibodies binding to Duffy binding-like domains. DBL2 and DBL3 domains
were coupled to Bio-Plex magnetic carboxylated microspheres (Bio-Rad, Hercules, USA) as per manufacturer's instructions. The antigen-coupled microspheres were resuspended in storage buffer (PBS, 0.05% sodium azide), and stored in the dark at 4 °C for immediate use. Their concentration was determined using a hemocytometer.
The DBL-coupled microspheres were mixed, resuspended in 1% PBS-BSA and added to wells of a 96-well round bottom plate (Greiner Bio-One, Kremsmünster, Austria) containing plasma in a 1:100 dilution in PBS. The sealed plates were incubated on a plate shaker overnight at 4 °C. After incubation, the plates were centrifuged and washed with PBS-0.1% Tween using a magnetic plate-washer (Bio-Plex Pro wash station, Bio-Rad). The antihuman Ab (total IgG, IgG1, IgG2, IgG3, IgG4) detectors conjugated with phycoerythrin (PE; all SouthernBiotech, Birmingham, USA) were added and the mixture was incubated for 2 h on a plate shaker. After washing with PBS and resuspending in xMAP drive fluid (Life Technologies, Carlsbad, USA), the plates were read on a Bio-Plex MAGPIX multiplex reader (Bio-Rad), and analysed using Bio-Plex Manager software (Bio-Rad). The median fluorescence intensity is directly proportional to the amount of Ab bound to the antigens 101 .
IgG N-linked glycan profiling. N-linked glycan profiles of purified IgG Abs (2 mg/ml) were measured on the LabChip GXII Touch instrument (PerkinElmer, USA) according to the ProfilerPro glycan profiling LabChip GXII Touch protocol. Microchip capillary electrophoresis-laser-induced fluorescence analysis of digested and labelled N-linked glycans was performed. The relative prevalences of several glycan profiles of IgG Abs were analyzed using the LabChip GX Reviewer (PerkinElmer) software. Peaks were assigned based on migration of known standards and glycan digests 91 . Peak area and therefore the relative prevalence of each glycan pattern was calculated.
Statistical methods. Statistical analyses were performed in Prism version 8 (GraphPad, USA). Statistical comparison of NK cell activation markers between groups was performed using Kruskal-Wallis test with Dunn's multiple comparison method. Statistical comparison between groups for the analysis of activated NK cells polyfunctionality was performed using multiple t tests corrected for multiple comparisons using the Holm-Šídák method. Spearman's rank correlation coefficients of antigen binding and NK cell activation were calculated. Kruskal-Wallis tests with Dunn's multiple comparison method were conducted to determine the significance of differences observed in glycan prevalence between pregnant women and their non-pregnant counterparts. Statistical significance was considered when p-values were less than 0.05. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.