Maternal Obesity is Associated with Phenotypic Alterations in Fetal Immune Cells by Single-Cell Mass Cytometry

Abstract

Problem:

Prenatal exposure to metabolic dysregulation arising from maternal obesity can have negative health consequences in postnatal life. To date, the specific effects of maternal obesity on fetal immunity at a cellular level have not been well characterized.

Method of Study:

Using cord blood mononuclear cells (CBMCs) and cord plasma (n=9/group) isolated from infants born to women with a high BMI (>25kg/m²) compared to women with a normal BMI (18–25kg/m²), we evaluated differences in immune cell populations using single-cell mass cytometry (CyTOF). CBMCs were matched according to potentially confounding variables, such as maternal and gestational age, ethnicity, smoking status, and gravidity. Statistical results were adjusted for fetal sex. Data was analyzed by viSNE and FlowSOM softwares in Cytobank™.

Results:

In newborn CBMCs from women with high BMI, we observed changes in frequency and phenotype of immune cell populations, including significant increases in CD4⁺ T cells and decreases in myeloid cell populations. IL-12p40 and MDC concentrations were significantly elevated in the high BMI group compared to control.

Conclusion:

This study demonstrates an association between maternal obesity and fetal immunity. Our results warrant following long-term immunologic outcomes and associated clinical risks in children born to women with a high pre-pregnancy BMI.

Introduction

The Developmental Origins of Health and Disease (DOHaD) hypothesis suggests that adult diseases originate from exposure to environmental influcences during in utero development¹. One such factor is maternal obesity, which not only has detrimental health implications for the mother, but can also have direct consequences on offspring during gestation². It has been well documented that obesity increases the risk of developing gestational diabetes and pre-eclampsia during pregnancy, which further increases the risk of maternal metabolic syndrome and of cardiovascular disease later in life³. For the fetus, maternal obesity is associated with the risk of congenital anomalies, miscarriage, autism and intrauterine demise^4–6. In addition, children born to obese mothers are more likely to become obese themselves during adulthood⁷.

Despite these associations, the specific molecular mechanisms of how body mass index (BMI) impacts fetal immunity during pregnany are weakly understood. Yet, it has been well documented that increased adipose tissue in non-pregnant adults leads to low grade systemic inflammation through NF-κB signaling that initiates insulin resistance and type II diabetes^8–10. Adipose tissue includes adipocytes and immune cells (macrophages and leukocytes), which have roles in controlling metabolic and inflammatory pathways through the release of hormones and cytokines, also known as adipokines¹¹. During gestation, increased maternal BMI is positively correlated with the detection of adipokines, tumor necrosis factor (TNF)-α, MCP-1, insulin, and leptin, in maternal blood¹². While the placenta deploys numerous mechanisms to protect the fetus from inflammation, placental immune regulation may be disrupted with maternal obesity. For example, placenta from obese mothers show a 2–3 fold increase in Hofbauer cells (fetal macrophages) compared to those from lean women¹³. Additionally, an increased number of inflammatory lesions are noted by histology in placenta from obese women (BMI ≥30 kg/m²) compared to those with a normal pregravid BMI¹⁴. Activation of pro-inflammatory pathways p38-MAPK and STAT3, as well as secretion of interleukin (IL)-6, were noted in placenta from women with high BMI¹², suggesting that obesity-induced inflammation impacts placental function.

While studying in utero development remains challenging, new technologies such as single cell (sc)RNAseq and multi-parameter cytometry provide a uniqe opportunity into characterizing multi-dimesional cellular interactions that occur at the maternal-fetal interface. For example, scRNAseq has been utilized to identify cell cross-talk between the maternal decidua and fetal trophoblast cells during the first trimester¹⁵. Others have used scRNAseq to define novel cell types and transcriptional signatures unique to parturition at term compared to preterm¹⁶. Cord blood collected from the umbilical cord after birth also provides a unique opportunity to characterize the signals being passed from mother to baby via the placenta to understand changes that occur in the fetus. Mass cytometry (CyTOF) is a powerful single-cell immunophenotyping technique and in combination with unbiased computational clustering tools, can be used to characterize immune cell populations and their dynamics^17,18. This technique is especially useful when limited by sample volume and for identifying rare cell populations^19,20.

In this study, we hypothesized that newborns from mothers with high BMI exhibited distinct T cell phenotypes and up-regulated inflammatory cytokines compared to newborns from women within a normal BMI range, reflecting chronic exposure to inflammatory signals in cord blood. Using mass cytometry-derived immunophenotyping and cytokine anlaysis from cord blood, we show how maternal BMI impacts neonatal immunity on immune cell populations in cord blood. Our results identify significant differences in the abundance of CD4⁺ T cells, CD56⁺ natural killer (NK) cells and CD68⁺ myeloid cells between the two groups, underscoring the impact of maternal obesity on fetal immunity.

Methods and Materials

Subject Selection

This study was approved by the Mayo Clinic institutional review board (#19–000312), utilizing stored cord blood mononuclear cells (CBMCs) and cord blood plasma collected through the Mayo Clinic Placenta and Cord Blood Biobank. Cord blood plasma was isolated and stored at −80 °C until their use. CBMCs were isolated by Ficoll density gradient centrifugation, resuspended in freezing medium (fetal bovine serum with 8% DMSO) and were stored in liquid nitrogen until their use. Samples were selected based on the mother’s pre-pregnancy BMI measured 6 months before pregnancy (to ensure total pregnancy exposure) and delineated into one of two groups: Normal BMI (18–25kg/ m²) or High BMI (>25kg/m²). In addition, samples in both groups were matched according to age, race, smoking status, use of assisted reproductive technology, maternal pregnancy complications, gestational age at delivery, and fetal complications. Nine samples were selected from women with normal BMI and nine with high BMI.

Mass Cytometry

Cell-labeling was performed as per manufacturer recommendations (Fluidigm Sciences, San Francisco, CA). Briefly, CBMCs were resuspended in 0.5μM Cell-ID cisplatin solution (Fluidigm Sciences) to stain dead cells, and then washed twice with Maxpar Cell Staining Buffer (MCSB, Fluidigm Sciences). In 50μL MCSB, the antibody cocktail consisting of 33 metal-conjugated antibodies was added, and samples were incubated at room temperature for 45 min with agitation. The antibodies are detailed in Supplementary Table 1. Following staining, cells were washed, fixed in 1.6% PFA, and incubated at room temperature for 20 min with agitation. Fixed cells were stored overnight at 4°C. The next day, cells were placed in 1mL intercalation solution [62.5nM Cell-ID Intercalator-Ir (Fluidigm Sciences)] and in 50μL of diluted barcoding solution prior to overnight incubation at 4°C. Barcoded samples were washed and resuspended in Cell Acquisition Solution-EQ Bead mixture (Fluidigm Sciences) to a concentration of 5×10⁵ cells/mL for normalization between samples and batches. Samples were then loaded onto a Helios CyTOF system (Fluidigm Sciences), acquired at a rate of 200–400 events per second, and cytometry data were collected as .fcs files. Clean-up of samples entailed removing dead cells, debris and doublets prior to data analysis using Cytobank™ cloud-computing software (Supplemental Figure 1).

Multiplex Cytokine Analysis

Plasma from the same cord blood samples above were thawed, and cytokines measured by MagPlex^® cytokine/chemokine multiplex assays (MilliporeSigma, Burlington, MA)²¹. Briefly, 20μL of each sample was added to a 96-well plate in duplicate. Antibody-conjugated beads specific for 38-targets were added to the wells and incubated overnight. Next morning, samples were incubated with detection antibodies, followed by Streptavidin prior to being run on a Luminex xPONENT technology (Luminex Corporation, Austin, TX). Concentrations were calculated using MILLIPLEX Analyst 5.1 software (MilliporeSigma).

Data Analysis

Patient demographics and cytokine concentrations were compared between groups by Mann-Whitney U tests, including a Bonferroni correction for multiple comparrisons. CyTOF data analysis was performed using viSNE and FlowSOM softwares in Cytobank™ (Cytobank, Inc.). viSNE plots were generated with the following parameters: desired total events (proportional sampling): 50,000 per group; channels: all 33 antibody-metal channels; compensation: file-internal compensation; iterations: 2000; perplexity: 70; and theta: 0.5. Advanced clustering analyses to identify cell subtypes were performed using FlowSOM with the following parameters: event sampling method: equal; desired events per file: 90,000; total events actually sampled: 265,183; clustering method: hierarchical consensus; number of clusters: 250; number of metaclusters: 20; iterations:10; and seed: 2114214004. Immune cell subtypes displaying a minimum fold-change (in means) of 2 between groups with P-value < 0.05 (multivariate linear regression model adjusting for maternal BMI and fetal sex) were considered as differentially abundant.

Results

Patient Demographics

Nine cord blood samples from the high- and nine from the normal BMI group were analyzed. The median BMI for the normal BMI cohort was 21.9 kg/m² compared to 36.3 kg/m² in the high BMI group. Cohort demographics are presented in Table 1. All participants were Caucasian, non-smokers, presented for induction of labor for term pregnancy and had healthy term infants. One mother from each group was diagnosed with hypertension just prior to delivery and was induced for that reason. Two of the participants in the normal BMI group had a cesarean section, while three in the high BMI group underwent a C-section. All cesarean sections were for non-reassuring fetal heart tracing (NRFHT). The majority of infants in the normal and high BMI groups were male (5/9 and 7/9, respectively).

Table 1.

Patient demographics.

	Normal BMI (18–25kg/m2, n = 9)	High BMI (>25kg/m2, n = 9)	P-value
Pre-pregnancy BMI (kg/m2)	21.9 (19.5–24.5)	36.3 (26.7–47.3)	0.0003
Maternal age (years)	32 (30–40)	34 (29–41)	0.2402
Gravida	2 (1–5)	2.5 (2–7)	0.207
Gestational age (weeks.days)	39 (37.4–41.1)	39.3 (34.4–41.1)	0.9993
Fetal weight (grams)	3,550 (3,080–4660)	3,850 (2,830–4,410)	0.1245

Data presented as medians with ranges. P-values were identified using the Mann-Whitney U test.

Visualization of Immune Cell Densities Reveals Clusters Associated with Maternal BMI Group

Following raw data acquisition from the mass cytometer, normalization, debarcoding, and cell clean-up, mean cell viability was 69% and the mean number of events (cells) analyzed per sample was 117,796. We validated the presence of major cell populations by 2D plots (Supplemental Figure 2). These plots demonstrated that the CD14+ population was lost either during the freezing process or due to poor antibody specificity; however, CD68 staining, another marker of monocytes/macrophages was strong. viSNE dimensionality reduction plots were constructed to identify global differences in immune cell type densities between the normal and high BMI groups. We identified many immune cell populations common to both groups, including CD4⁺ T helper cells, CD8⁺ cytotoxic T cells, CD20⁺ B cells, CD16⁺CD56⁺ natural killer cells, and CD68⁺ myeloid cells (Figure 1). However, within these clusters were clear differences in cell densities as demonstrated by the red (high cell abundance) regions within commonly-observed immune cell populations. Comparing distinct spatial regions in the viSNE plots revealed differences in abundance of immune cell types, including a higher population of activated myeloid cells (CD68⁺CD4⁺CD16⁺CD86⁺CD95⁺HLA-DR/ABC⁺ and CD68⁺HLA-DR/ABC⁺) in infants born to women with normal BMI compared to infants born to women with high BMI. In contrast, infants born to women with high BMI had distinct populations of activated T (CD3⁺CD8⁺CD27⁺CD38⁺CD45RA⁺ and CD3⁺CD4⁺CD27⁺CD38⁺) and activated B cells (CD20⁺CD40⁺HLA-ABC⁺CD45RA⁺CCR6⁺CD38⁺), which were not expressed in infants born to mothers with normal BMI.

Figure 1.

viSNE visualization displaying clusters of CBMC immune cell subtypes from infants with normal BMI mothers (left) and those from infants with high BMI mothers (right).

T Cell and Myeloid Cell Subtypes are Significantly Associated with Maternal BMI

To investigate which CBMC subtypes are associated with high maternal BMI, we utilized FlowSOM for unsupervised clustering and subsequent subpopulation detection within the immune cell profiles. In brief, FlowSOM utilizes cell abundance data to divide cell types into clusters and meta-clusters (i.e., aggregates of clusters), based upon the similarity of cell-specific marker expression profiles. After correcting for fetal sex, our statistical analyses identified differences in 13 clusters between the high and normal BMI groups (Table 2). FlowSOM results showed that maternal BMI had a significant impact on the abundance of myeloid cell and T cell subtypes in fetal CBMCs. Specifically, CD68⁺ cells, which include populations of monocytes and macrophages, were decreased in number in the high BMI compared to the normal BMI group (range of fold-changes: 3.22–10.17). Natural killer cells (CD56⁺CD16^-) were also found to be decreased in CBMCs from infants with high BMI mothers. Subtypes of T helper cells increased 2- to 4-fold in CBMCs from mothers with high BMI compared to those from mothers with normal BMI. Many of these T cell populations were CD45RA⁺, indicating a naïve phenotype. The abundance of CD3⁺CD4⁺CD27^midCD45RA^lowCCR7^mid cell subtype significantly correlated with BMI (Fig. 2; Spearman’s ρ = 0.56, P = 0.02), which may be reflective of maternal metabolic health. Taken together, our data indicate that maternal BMI is significantly associated with T cell and myeloid cell subtype numbers in fetal CBMCs.

Table 2.

Immune cell subtypes that are differentially abundant between high and normal BMI groups.

Cell Subtype	Marker Expression Phenotype‡	High BMI		Normal BMI		Fold-change†	P-value
		Mean (%)	SD (%)	Mean (%)	SD (%)
Myeloid cell	CD68midCD16lowFaslow	0.09	0.09	0.29	0.19	3.22(−)	0.022
Myeloid cell	CD68+CD16+CD86+Fas+HLA-DR+CCR2+CCR5+	0.10	0.10	0.37	0.32	3.70(−)	0.044
Myeloid cell	CD68+CD45RA+HLA-DR+CCR2+CCR5+	0.07	0.08	0.59	0.47	8.43(−)	0.011
Myeloid cell	CD68hiCD16−	0.06	0.04	0.61	0.73	10.17(−)	0.024
Myeloid cell	CD68lowCD16−	0.20	0.21	0.69	0.67	3.45(−)	0.046
Natural killer cell	CD56+CD16lowCD45RA+	0.22	0.30	0.74	0.56	3.36(−)	0.028
T helper cell	CD3+CD4+CD25hiCD27hi	0.44	0.26	0.22	0.10	2.00(+)	0.028
T helper cell	CD3+CD4+CD25midCD27hi	0.42	0.26	0.21	0.11	2.00(+)	0.027
T helper cell	CD3+CD4+CD27midCD45RAlowCCR7mid	0.73	0.88	0.20	0.12	3.65(+)	0.02
T helper cell	CD3+CD4+CD27hiCD45RAlow	0.26	0.18	0.10	0.06	2.60(+)	0.013
T helper cell	CD3+CD4+CD27midCCR7low	0.23	0.24	0.05	0.05	4.60(+)	0.005
T helper cell	CD3+CD4+CD27midCCR7mid	0.70	0.82	0.21	0.27	3.33(+)	0.036
T helper cell	CD3+CD4+CD16midCD27midCD68midFasmidCCR7hi	0.29	0.41	0.11	0.08	2.63(+)	0.043

^‡All clusters are CD45⁺CD38⁺HLA-ABC⁺ in addition to the listed markers. P-values from multivariate linear regression model adjusting for maternal BMI and fetal sex.

^†Positive and negative signs indicate higher or lower mean in the High BMI group, respectively.

Figure 2.

Abundance of immune cell subtype CD3+CD4+CD27midCD45RAlowCCR7mid in cord blood shows significant correlation with pre-pregnancy maternal BMI (ρ = 0.56, P = 0.02). Cord blood cell abundances and BMI were assessed in 18 study participants partitioned into two BMI groups (Normal BMI, n = 9; High BMI, n = 9). Shaded region corresponds to 95% confidence interval

Cytokine Profiles in Cord Blood Based on BMI

Using cord blood plasma, cytokines were measured and compared between the normal BMI and high BMI groups. We observed a signficant increase in IL-12p40 (43 vs. 21pg/mL, P = 0.04) and macrophage-derived chemokine (MDC; 353 vs. 200pg/mL, P = 0.04) in the high BMI group compared to normal BMI (Figure 3A and 3B). Interestingly, no differences were observed in the concentrations of known adipokine TNF-α (Figure 3C) and monocyte chemoattractant protein (MCP)-1 (Figure 3D) between the two groups.. Other measured cytokines and chemokines were either not significant or below the assay limit of detection (Supplemental Table 2). Lastly, we analyzed cytokines based on cesearan section vs. vaginal delivery. As labor was induced in all participants and some underwent cesearn section for NRFHT, we did not observe any differences in cord blood inflammatory cytokines based on mode of delivery.

Figure 3.

Expression of cytokine and chemokines in cord blood based on maternal BMI. Plasma concentrations of (A) IL-12p40, (B) MDC, (C) TNF-α, and (D) MCP-1 in the High BMI versus Normal BMI groups. Data shown as median with interquartile ranges (n = 9/group). *P-value ≤ 0.05 by Mann-Whitney U test.

Discussion

This study utilized immunophenotyping by mass cytometry of CBMCs as well as cytokine profiling to characterize the effects of high maternal BMI on fetal immunity. Mass cytometry enables an unbiased approach for characterizing immune targets, even those not expected to be associated with specific cell types. We observed that cellular abundances in CD4⁺ T cell and myeloid cell subtypes in CBMCs at birth were altered based on maternal BMI. We also observed that IL-12p40 and MDC was increased in cord blood plasma of infants born to high BMI women compared to normal BMI. It is unclear whether these immunologic differences persist in postnatal life, and whether they have a role in the chronic health outcomes observed in children born to obese mothers.

Obesity commonly occurs during pregnancy and is associated with adverse short- and long-term clinical outcomes for both the mother and exposed offspring⁵. Much of this effect is hypothesized to be due to obesity-associated inflammation¹². In a previous study, obese mothers were found to give birth to infants with reduced CBMC populations of eosinophils and CD4⁺ T helper cells, along with higher plasma concentrations of proinflammatory IL-6 and interferon (IFN)-α2²². Additionally, umbilical cord monocytes of infants born to obese mothers generated only weak inflammatory responses after stimulation by lipopolysaccharide (LPS) compared to lean mothers²³. Inflammatory TNF-α was also higher in cord blood plasma from obese/overweight mothers; however, at 9-month follow-up, TNF-α levels did not correlate with infant body weight²⁴.

In contrast to previous reports, our data identified an increase in T cell populations in CBMCs from infants with mothers with high BMI. Expression of CD27 on CD4+ T helper cells can distinguish naive cells in peripheral blood²⁵. Coexpression of CD25⁺ and CD27⁺ on CD4⁺ T helper cells distinguishes a T regulatory cell phenotype^26,27. An increase in T regulatory cells and skewing towards a Th2-dominant phenotype was observed in the peripheral blood of obese compared to lean adults²⁸.

Compared to adults, the majority of T cells from cord blood are naïve and, upon stimulation, cannot release cytokines at levels comparable to adults²⁹. Therefore, the previously reported decrease in CD4⁺ helper cells by flow cytometry²² may have missed the increase in this subtype of T cells we have identified by mass cytometry. We also observed a decrease of CD68⁺ myeloid cells in infant CBMCs whose mothers have a high BMI using mass cytometry. While CD68⁺ expression most often defines phagocytic macrophage cells, it can also be expressed on dentritic cells and monocytes³⁰. Macrophages are known to increase in adipose tissue and secrete inflammatory cytokines TNF-α and IL-6³¹. NK cells were also decreased in our obese cohort. This innate cell type makes up 30% of the cells in CBMC and, unlike peripheral blood cells, CBMC NK cells demonstrate an immature phenotype with reduced cytotoxic functions ³². A study using obese mice also found that NK cytotoxic effects are lost due to lipid accumulation in the cell ³³. Therefore, an increase in T cells and a decrease in CD68⁺ myeloid cells and CD56⁺ NK cells in infant CBMCs from mother’s with high BMI may be an adaptation maintain homeostasis by controlling obesity-associated inflammatory responses while possibly imparing the ability of the neonate to fight infection.

We did not observe an increase in cytokines TNF-α, IL-6, IFN-2α or chemokine MCP-1 as others have reported ^12,22–24. This may be due to the fact that we only included women presenting for induction of labor, which may be unique to spontaneous labor or elective c-section. However, we did observe a increase in IL-12p40 and MDC in cord blood from the high BMI group. IL-12p40 is produced by activated myeloid cells and acts as an agonist for IL-12 receptor, blocking cellular proinflammatory signaling ³⁴. In non-pregnant adult women, IL-12p40 is found at elevated levels in obese and overweight individuals compared to lean, which positively correlated with fat mass³⁵. MDC (CCL22) is also secreted by myeloid cells and this chemokine attracts cells with a chronic inflammatory phenotype, as seen in allergies ^36,37. While not specifically studied in the context of obesity, glutamine induces production of MDC by macrophages³⁸. Therefore, expression of these two mediators may play an important role in responding to obesity-induced inflammation.

This study was performed on a small cohort (n=9/group) making it impossible to be conclusive among subjects with high BMI based on CDC definitions (i.e. overweight, class 1), However, it does demonstrate the power and utility of multi-parameter cytometry. In flow cytometry, fluorescently tagged antibodies can be captured by different lasers and data analyzed using the same tools described. Using 20-color flow cytometry, Vazquez et al show that immune cells in the decidua have a unique profile compared to circulating immune cells³⁹. More recently, through the use of metal conjugated antibodies, the complexity of the immune system can be investigatedby mass cytometry, which has the benefit of minimal spectral overlap that limits flow cytometry.. This technology has further demonstrated just how diverse the human immune system is and how one’s environment influences this diversity through twin studies⁴⁰. Specific to reproductive biology, mass cytometry has been utilized to profile differences in maternal and fetal immune cells and their respective kinase signaling cascades with and without stimulation⁴¹. Others have utilized this technology to longitudinally study changes in immune cell populations in infants born preterm compared to term¹⁹. The researchers observed significant differences in immune phenotypes immediately after birth in preterm compared to term infants, but these changes converged after 3 months¹⁹. Thus, mass cytometry is a versatile technology which has the power to advance our knowledge and understanding of the intricacies of immunity. Combining advanced technologies that generate complementary datasets measuring cellular identify by cytometry^19,39 and transcriptional activation, as with scRNAseq^15,16, should be a priority to move the field of fetomaternal immunity forward in the next decade.

In conclusion, the data indicate maternal BMI can impact fetal immunity, but whether these changes in immune and inflammatory profiles persist after birth remains an important topic of study. Obesity-induced inflammation during pregnancy is critical as metabolic programming during gestation may lead to an increased risk of cardiovascular and metabolic morbidity (i.e., diabetes and heart disease) during postnatal life. Given the increasing global prevalence of obesity, further research is urgently needed to investigate the short- and long-term effects of maternal obesity on fetal, infant, and childhood immunity in order for efficacious and safe clinical interventions to be implemented.