Baby’s First MФ: Temporal Regulation of Hofbauer Cell Phenotype Influences Ligand-Mediated Innate Immune Responses Across Gestation

Abstract

The importance of fetal placental macrophages (Hofbauer cell (HCs)) is underscored by their appearance eighteen days post-conception and maintenance through term; however, how human HCs evolve during healthy pregnancy, and how microenvironment and ontogeny impact phenotype and function remains unknown. Here, we comprehensively classify human HCs ex vivo, interrogate phenotypic plasticity, and characterize antiviral immune responses through gestation. Activated HCs were abundant in early pregnancy and decreased by term; molecular signatures emphasize inflammatory phenotypes early in gestation. Frequency of HCs with regulatory phenotypes remained high through term. Further, term HCs exhibited blunted responses to stimulation, indicating reduced plasticity. IFN-λ1 is a key placental interferon that appeared less protective than IFN-α, suggesting a potential weakness in antiviral immunity. Ligand-specific responses were temporally-regulated: we noted an absence of inflammatory mediators and reduced antiviral gene transcription following RIG-I activation at term despite all HCs producing inflammatory mediators following IFN-γ + LPS stimulation. Collectively, we demonstrate sequential, evolving immunity as part of the natural history of HCs through gestation.

Introduction

Macrophages are a heterogenous population of immune cells whose functions include antigen presentation, phagocytosis, cytokine secretion, and coordination of downstream innate and adaptive immunity. These cells orchestrate immune response generation and tissue homeostasis and repair by constantly changing their functional state in response to the local tissue microenvironment(1). Broadly classified as M1 or M2, macrophage differentiation follows T-helper cell paradigms(2, 3). This simplification only partially reflects characteristics of macrophage phenotypic intermediates(4). Classically-activated M1 macrophages secrete proinflammatory cytokines, mediate pathogen resistance, and contribute to tissue destruction. Alternatively-activated macrophages are subdivided by distinct surface molecule expression, cytokine secretion, and effector function(5, 6). M2a macrophages are activated by IL-4 + IL-13 and involved in tissue repair and immunoregulation; M2b responses support humoral and allergic reactions and are induced by immune complexes; M2c macrophages, which arise from IL-10 or glucocorticoid stimulation, suppress inflammation and remodel extracellular environments(5).

Placental macrophages, Hofbauer cells (HCs), were first described towards the end of the nineteenth century(7) as large, pleomorphic, highly vacuolated cells of fetal origin(8, 9, 10, 11). Found in chorionic villi as early as 18 days post-conception(9), they are maintained through birth, albeit with reduced numbers(12). Although their ontogeny has not been fully characterized, studies suggest that the earliest HCs derive from mesenchymal progenitor cells, followed by progressive recruitment of in situ differentiated yolk-sac-, fetal liver-, and bone-marrow-derived-monocytes (Fig1)(13). They preferentially localize near fetal vessels and trophoblasts, suggesting roles in placental development, immunity, and homeostasis(14). Term HCs are a mixture of M2 subtypes, with heterogenous surface marker expression, cytokine secretion, and function(14, 15, 16, 17). They express the three FcγRs(18), the pan-macrophage marker CD68(19), and can be stimulated by glucocorticoids(20) and IL-10(21) to express CD163, CD206, and CD209(21). Unstimulated HCs constitutively express IL-10 and TGF-β (22). Variations in shape, vesiculation(13), surface molecule expression, effector function(23), and motility(12), are all consistent with plasticity and a wide range of roles in placental homeostasis. Furthermore, term HCs are capable of robust proinflammatory responses to bacterial lipopolysaccharide (LPS)(24) and viruses(25, 26, 27, 28). However, there are few studies describing HC phenotype and function through the natural time-course of a healthy pregnancy.

FIGURE 1.

HC phenotypic heterogeneity likely stems from a combination of different origins and changing microenvironment at the fetal-maternal interface.

In eutherian mammals, the placenta alters local and systemic maternal immunity to limit infection and sustain a healthy pregnancy(29, 30, 31, 32). Immunologic homeostasis at the maternal-fetal interface can be characterized by three distinct phases(33, 34): Implantation and placentation during the first and early second trimester resembles an “open wound,” which requires a strong inflammatory response(35, 36). Next, the mother, placenta, and fetus reach symbiosis, with induction of an anti-inflammatory/tolerogenic state(37). Parturition necessitates recrudescence of the inflammatory process, delivery, and placental rejection (38, 39).

It is unknown how human HCs evolve during gestation, and how the combined influences of microenvironment and ontogeny impact phenotype and function. The only multi-dimensional analysis of HC diversity through gestation was performed by Reyes et al. (17). In this study of rhesus macaques, HCs exhibited plasticity and pleomorphism through gestation. Here we undertook a comprehensive classification of human HCs ex vivo, interrogated plasticity following in vitro stimulation, and characterized antiviral immune responses through gestation.

Materials and Methods:

ETHICS STATEMENT:

Second trimester human placentae were obtained from a free-standing clinic in Atlanta, GA from consented donors who elected to terminate pregnancies prior to 21 weeks and 6 days of gestation. Human term placentae (>37 weeks gestation) were collected from hepatitis B, HIV-1 seronegative women (>18 years of age) immediately after elective caesarean section without labor from Emory Midtown Hospital, Atlanta, GA. This study was approved by the Emory University Institutional Review Board (IRB 000217715). Written informed consent was acquired from all donors before sample collection. Samples were de-identified before primary HC isolation.

PLACENTAL DISSECTION AND HOFBAUER ISOLATION:

HCs were isolated from membrane-free villous placenta as previously described (22). On average, the purity was >95%. After isolation, HCs were cultured in complete RPMI medium consisting of 1x RPMI (Corning Cellgro), 10% FBS (Optima, Atlanta Biologics), 2mM L-glutamine (Corning Cellgro), 1mM sodium pyruvate (Corning Cellgro), 1x Non-essential Amino Acids (Corning Cellgro), 1x antibiotics (penicillin, streptomycin, amphotericin B; Corning Cellgro) at 37°C and 5% CO2. HCs were treated with the following as indicated, following resuspension per the manufacturer’s instructions: 50ng/mL LPS (00–4976) (Thermo); 100IU/mL IFN-α A/D (PHC4044), 20ng/mL IFN-γ (PHC4031), 20ng/mL IL-4 (PHC0044), 20ng/mL IL-13 (PHC0134) (Life Technologies); 100ng/mL IL-29 (ab50032), 20ng/mL IL-10 (ab9613) (Abcam); 50μg/mL heat-agglutinated-IgG (HAGG) produced from IgG from human serum (L4506) (Sigma) as previously described (40); 20ng/mL IL-1β (78034.1) (Stemcell); 5’ppp-dsRNA (tlrl-3prna), control for 5’ppp-dsRNA (tlrl-3prnac) (InvivoGen).

FACS:

Hoffbauer cells (250,000 per sample) were blocked for 10min on ice with 0.25μl/sample Human TruStain FcX (BioLegend) in FACS buffer (1x PBS, 0.1% BSA, 1mM EDTA) and live/dead stained for 10 min on ice with Calcein Violet 450AM (Life Technologies). HCs were stained for surface markers for 20 min on ice using 0.25μl/sample of the following anti-human antibodies in FACS buffer CD209 [E9A8] HLA-DR [L243] CD163 [GHI/61] CD68 [Y1/82A] (BioLegend), CD14 [61D3] (Invitrogen), R&D Systems: hMMR [FAB25342T] (R&D Systems), CD86 [2331(FUN-1)] CD80 [L307.4] (BD Biosciences). FACS samples resuspended in FACS buffer were run on a CytoFLEX flow cytometer following calibration using 6-peak Rainbow Calibration Particles (BioLegend) and analyzed using FlowJo software. Compensation values were calculated using UltraComp eBeads (Life Technologies) and gating strategy as in SFig1.

RNA ISOLATION AND RT-PCR:

Directly isolated, treated, and control HCs (100,000 cells per condition) were lysed in RNA Lysis Buffer (Zymo Research). Total RNA was isolated from cells using the Quick-RNA MiniPrep Kit (Zymo Research). Purified RNA was reverse transcribed using random primers with the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). HC gene expression was quantified by RT-PCR using QuantiTect SYBR Green PCR Kits (Qiagen) using specific primers for host genes. CT values were normalized to the reference gene B-actin and represented as indicated. HC gene expression was quantified by RT-PCR using QuantiTect SYBR Green PCR Kits (Qiagen) using specific primers as follows: B-actin (FWD 5′-GGC CCA GTC CTC CCC AAG TCC AC-3′, REV 5′-GGT AAG CCC TGG CTG CCT CCA CC −3′) IFIT1 (FWD 5′-GCC ACA AAA AAT CAC AAG CCA-3′, REV 5′-CCA TTG TCT GGA TTT AAG CGG-3′) IFITM1 (FWD 5′-GGC TTC ATA GCA TTC GCC TAC TC-3′, REV 5′-AGA TGT TCA GGC ACT TGG CGG T-3′) OAS1 (FWD 5′-CAA GCT CAA GAG CCT CAT CC-3′, REV 5′- TGG GCT GTG TTG AAA TGT GT-3′) iNOS (FWD 5′-AGC GGG ATG ACT TTC CAA GA-3′, REV 5′- GGA CCC CAG GCA AGA TTT G-3′) ARG2 (FWD 5′-CGC GAG TGC ATT CCA TCC T-3′, REV 5′-TCC AAA GTC TTT TAG GTG GCA G-3′) ARG1 (FWD 5′-TCA TCT GGG TGG ATG CTC ACA C-3′, REV 5′- GAG AAT CCT GGC ACA TCG GGA A-3′) STAT6 (FWD 5′- CGA GTA GGG GAG ATC CAC CTT-3′, REV 5′- GCA GGA GTT TCT ATC AAG CTG TG-3′) STAT5A (FWD 5′- CGA CGG GAC CTT CTT GTT G-3′, REV 5′- GTT CCG GGG AGT CAA ACT TCC-3′) STAT3 (FWD 5′- ACC AGC AGT ATA GCC GCT TC-3′, REV 5′- GCC ACA ATC CGG GCA ATC T-3′) STAT1 (FWD 5′- CGG CTG ATT TTC GGC ACC T-3′, REV 5′- CAG TAA CGA TGA GAG GAC CCT-3′) IFNA (FWD 5′- GAC TCC ATC TTG GCT GTG A-3′, REV 5′- TGA TTT CTG CTC TGA CAA CCT-3′) IFN-B (FWD 5′- GTC TCC TCC AAA TTG CTC TC-3′, REV 5′- ACA GGA GCT TCT GAC ACT GA-3′) MDA5 (FWD 5′-GGC ATG GAG AAT AAC TCA TCA G-3′, REV 5′- CTC TTC ATC TGA ATC ACT TCC C-3′) MX1 (FWD 5′- CAA TCA GCC TGC TGA CAT TG-3′, REV 5′- TGT CTC CTG CCT GTG GAT G-3′) RIG-I (FWD 5′-ATC CCA GTG TAT GAA CAG CAG-3′, REV 5′- GCC TGT AAC TCT ATA CCC ATG TC-3′) VIPERIN (FWD 5′-CCA GTG CAA CTA CAA ATG CGG C-3′, REV 5′- CGG TCT TGA AGA AAT GGC TCT CC-3′). RT-PCR was performed in 96-well plates and run on a Thermo Fisher QuantStudio 3.

LUMINEX:

Cytokine concentrations in the supernatants of treated HCs (500,000 cells per condition) and accompanying controls were assessed using a human cytokine 25-plex panel (Invitrogen) per the manufacturers’ instructions. Plates were read on a Luminex 100 Analyzer.

STATISTICAL ANALYSIS:

Baseline expression of phenotypic markers was analyzed using linear regression and the resulting graph showing 95% confidence bands of the best-fit-line. R² (goodness-of-fit) is reported in the figure body, and deviation of the slope from zero in the results. Differences in canonical polarization at baseline were analyzed using 2-way ANOVA, with significance set at p<0.05. Phenotypic markers, cytokine production, and gene expression were analyzed using multiple one-sample t-tests comparing untreated and treated values. Analyses were corrected for multiple comparisons by controlling the false discovery rate. Discovery was determined using the two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli, with Q = 5%, without assuming consistent SD (41). Changes in canonical polarization after treatments were analyzed with 2-way ANOVA with correction for multiple comparisons by controlling the false discovery rate, as above. Outlier identification was performed using the ROUT method, with Q=5%.

All statistical analysis was performed using GraphPad Prism 8.1.2 software. In each of the main and supplemental figure legends, “n” represents the number of placental donors from which HCs were derived. Further experimental statistical details are described in the Figure legends.

Results

Early/mid-gestation is associated with an abundance of activated HCs, while term cells are less galvanized:

To determine if HC phenotype changes with gestational age, we employed fluorescence-activated cell sorting (FACS) to characterize isolated, CD14⁺ early/mid-gestation (12–24 weeks gestational age) and term (>37 weeks) HCs ex vivo using established macrophage surface markers(5, 6). The percentage of HCs expressing each marker is visualized on a per donor basis. We noted a higher frequency of activated CD68⁺, CD80⁺, or CD86⁺ HCs in early/mid-gestation (Fig2Ai-iii, STableI). An average of 75% of HCs isolated from placentae between 12 – 22 weeks gestation were CD68⁺, compared to 40% at term (Fig1Ai, STableI); linear regression revealed 63% of variation in CD68 expression can be explained by gestational age, with a significantly (p=0.0001) non-zero slope of 2% fewer CD68⁺ cells per week of advanced gestation. Further, 96% of early/mid-gestation HCs were CD80⁺; at term, CD80⁺ HC frequency varied, with groups of high- and low-expressing donors (Fig2Aii, STableI). As such, only 50% of variation in CD80 can be attributed to advancing gestational age; however, advancing gestational age is associated with a 1.8% loss in CD80⁺ cell per week (p=0.0014). CD86 expression was limited to 13% of early/mid-gestation HCs, and decreased 4-fold by term to an average of 3% (Fig2Aiii, STableI); 56% of variation in CD86 can be attributed to advancing gestational age, with a loss of 0.5% of CD86⁺ cells per week (p=0.0005). HLA-DR⁺ HC frequency showed donor variability and no association with gestational age (Fig2Aiv, STableI). Median fluorescence intensity (MFI) was analyzed as a measurement of molecular density and was not associated with gestational age for CD68, CD80, CD86, or HLA-DR (STableI). Therefore, while there were fewer activated HCs at term, these expressed a phenotype comparable to activated HCs in early/mid-gestation.

FIGURE 2.

Temporal kinetics of human HCs reveal that early/mid-gestation has an abundance of activated M2A and M2C macrophages, while term HCs are less galvanized and predominantly of the M2C-phenotype. Human placental macrophages were isolated from freshly obtained tissue samples. (A) Individual donors are distinguished by color. CD14+ live macrophages populations were co-stained for activation markers (i) CD68, (ii) CD80, (iii) CD86, MHC class II cell surface receptor (iv) HLA-DR, and regulatory markers (v) CD163, (vi) CD206, and (vii) CD209 followed by representative flow plots of average of samples <16 weeks, 16–24 weeks, and at term. Linear regression was performed to analyze the natural progression of phenotypes occurring through gestation. (B) Relative expression of ArgI, ArgII, and iNOS was determined by qRT-PCR of HC’s using the ΔCt method. (C) HC’s were grouped into canonical macrophage subtypes via FACS as in Supplementary Figure 1. Population changes were analyzed with 2-way ANOVA and corrected for multiple comparisons using the Sidak method. * = q<0.05; n.d.=not detected

The frequency of CD209⁺ HCs was reduced from 99% in early/mid-gestation to 79% at term, with 50% of variation dependent on advancing gestational age, and a loss of 1% of CD209⁺ cell per week of advancing age (p=0.0013) (Fig2Avii, STableI). CD163⁺ and CD206⁺ cells were constant in frequency throughout gestation; linear regression analysis of these markers suggests that expression of CD163 (Fig2Av, STableI) and CD206 (Fig2Avi, STableI) is not dependent on gestational age. Lastly, regulatory molecule MFI did not associate with gestational age (STableI), highlighting that regulatory HCs are consistent in phenotype, but not frequency, throughout gestation.

To define HCs within prototypic M1 or M2 subtypes, we used qRT-PCR to determine expression levels of arginase (ArgI, ArgII) and inducible nitric oxide synthase (iNOS). M1 macrophages express iNOS, which metabolizes arginine to nitric oxide, making it available for downstream production of reactive nitrogen species. M2 macrophages express arginase for downstream synthesis of molecules important for cellular proliferation and tissue repair(5, 6). ArgI and ArgII transcripts were detected in all HCs; ArgI and ArgII expression was not affected by advancing gestational age (p>0.05 for both) (Fig2Bi). iNOS expression was not detected (n.d.) in most samples (6/11) (Fig2Bii). Three iNOS+ donors were isolated from <18 weeks. Additionally, we observed two term iNOS+ donors, potentially suggesting these placentae were preparing for parturition as a similar increase in iNOS has been reported in placental trophoblasts following labor(42).

Using canonical macrophage surface markers, we characterized HC subtypes throughout gestation (SFig1)(5, 6). M1 (CD206⁻CD209⁻HLA-DR⁺CD86⁺CD80⁺)-type macrophages were not observed throughout gestation (Fig2C). M2A (CD206⁺CD209⁺HLA-DR⁺CD163⁺)-type macrophages constituted 40% of early/mid-gestation HCs, and 12.4% of term HCs (p=0.02). M2B (CD206⁻CD209⁻HLA-DR⁺CD86⁺CD80⁻)-type macrophages were rarely identified (3.2%), and only at term. M2C (CD206⁺CD209⁺HLA-DR⁻CD163⁺) macrophages comprised 72.8% of HCs at term, and 46.6% at early/mid-gestation (p=0.03). 13% of HCs at early/mid-gestation and 11.4% at term had a non-M1-M2 macrophage subtype.

Lastly, protein analysis was performed on supernatants collected from HCs after 48 hours of culture to determine if phenotypic variations potentiated functional diversity (STableIII). We observed negligible production of monocyte growth factor GM-CSF, eosinophil growth factor IL-5, and T-cell growth factors IL-2, or IL-15; inflammatory cytokines IFN-γ, IL-12, and IL-17A; anti-inflammatory cytokine IL-13; and chemokines eotaxin and MIG. One donor (15.4 weeks) was excluded from evaluation following outlier analysis due to suspected abnormal clinical characteristics. Although the data is preliminary, we observed some trends of gestationally-dependent cytokine production when comparing average cytokine concentrations produced by HCs isolated at early/mid-gestation and term. Early/mid-gestation HCs produced 1.7-fold more IFN-α than term HCs (q=0.01). Additionally, inflammatory IL-1β and IL-6 were 2.5-fold higher at early/mid-gestation. Anti-inflammatory IL-1RA was 3.2-fold higher in early/mid-gestation (q=0.0008), and IL-4 was 2-fold higher (q=0.04). The T-cell chemokine RANTES was 6.8-fold higher at early/mid-gestation (q=0.0005) and IP-10 trended higher as well. Production of monocyte chemoattractant MCP-1, and proinflammatory granulocyte chemoattractants MIP-1α and MIP-1β was 6.2-fold (q=0.0006), 3.5-fold (q=0.03), and 3-fold (q=0.008) higher, respectively, at early/mid-gestation. Lastly, neutrophil chemoattractant IL-8 was produced in increasing quantities as gestation progressed, with term cells producing 20% more than early/mid-gestation HCs (q=0.02).

Early/mid-gestation HCs exhibited greater plasticity compared to term HCs:

Functional plasticity is an essential feature of macrophages which allows tailored stimuli-dependent responses, as well as education of downstream adaptive immune responses. To investigate whether gestational age impacts plasticity, we treated HCs in vitro with stimuli chosen to recapitulate macrophage polarization into the canonical macrophage subtypes: classically-activated M1, which are functionally pro-inflammatory and antimicrobial, as well as subclasses of alternatively-activated M2 macrophages, which are broadly anti-inflammatory. M2A macrophages enhance endocytic activity, promote cell growth and tissue repair. M2B macrophages regulate the breadth and depth of immune responses and inflammatory reactions with production of both pro- and anti-inflammatory cytokines. Lastly, M2C macrophages play crucial roles in the phagocytosis of apoptotic cells. Although these treatments were not designed to recapitulate conditions observed during normal or pathogenic pregnancies, evaluation of HC responses to conventional macrophage stimuli provides valuable insight into HC macrophage biology.

We modeled M1 polarization using IFN-γ + LPS(43). In early/mid-gestation, HLA-DR⁺ HCs was 2.4-fold more common 24 hours post-treatment (HPT) (q=0.01) and continued rising through 48 HPT. CD163⁺ HCs decreased 47%, (q=0.0002) (Fig3Ai, STableII). iNOS and ArgI relative expression was analyzed 24 HPT to evaluate transcriptional regulation of differentiation. iNOS transcription rose 200-fold in early/mid-gestation (Fig3Bi), suggesting potential for antimicrobial activity; ArgI was unaltered. Ex vivo HC M2-preference was expected based on previous reports(15, 16, 17); IFN-γ + LPS induced an M1-skew in some donors (Fig3Bii) (44). Subtype analysis revealed a decrease in M2A-type macrophages at 24 HPT, while M2C-type and non-M1/M2 subtypes modestly increased (Fig3C). As different STAT-analogues are upregulated during macrophage subtype differentiation and cytokine-induced activation, we evaluated STAT1, STAT3, STAT5A, and STAT6 transcription(45). Following IFN-γ + LPS, <1% of early/mid-gestation HCs were M1-like, despite a 10-fold increase in STAT1, a factor downstream of IFN-γ signaling (Fig3D). M2A-associated STAT6, inflammatory-signaling-associated STAT5A, and M2C-associated STAT3 remained unchanged. IFN-γ + LPS influenced the secretome of early/mid-gestation HCs compared to untreated control: growth factors GM-CSF, IL-7, and IL-15 increased (all q=0.01) (Fig3Ei, STableIII). Secretion of inflammatory IL-1β expanded by 69-fold (q=0.01), IL-12 by 2,735-fold (q=0.08), TNF-α by 14,538-fold (q= 0.004), and IL-6 beyond the range of our assay (Fig3Eii, STableIII). IL-1RA, IL-4, IL-10, and IL-13 were boosted by 1.8-fold (q=0.05), 2.6-fold (q=0.01), 576-fold (q=0.05), and 3.9-fold (0.01), respectively (Fig3Eiii, STableIII). Proinflammatory chemokines MIP-1α and MIP-1β rose 5-fold and 2.1-fold, respectively (both q=0.01). T-cell chemoattractant RANTES increased 2.3-fold (q=0.04). Absence of increases in neutrophil chemoattractant IL-8 suggest a protective mechanism against deleterious inflammatory effects at the maternal-fetal interface (Fig3Ev, STableIII).

FIGURE 3.

IFN-γ + LPS co-stimulation reveals early/mid-gestation HCs demonstrate greater plasticity than term macrophages. HCs were stimulated with IFN-γ + LPS simultaneously in order to induce a M1 phenotype. (A) Mean log₁₀ (Fold Change) ± SEM of (i) percent positive and (ii) MFI of positive population with values meeting thresholds q<0.05 and fold change of +/− 1.5, followed by representative flow plots therein (B) (i) Relative expression of ArgI and iNOS was determined by qRT-PCR using the ΔΔCt method (ii) M1:M2 ratio as ΔCt-ARGI/ΔCt-iNOS visualized as median and quartiles. (C) HC’s were grouped into canonical macrophage subtypes via FACS as in Supplementary Figure 1. (D) Relative expression of STAT1, STAT3, STAT5A, and STAT6 ex vivo and after stimulation was determined by qRT-PCR using the ΔCt method. (E) Quantification of (i) growth factors (ii) inflammatory cytokines (iii) anti-inflammatory cytokines (iv) IL-2R and (v) chemokines following 48 hours of in vitro culture. Open circles indicate control untreated cells and closed circles IFN-γ + LPS stimulated cells, visualized as mean ±SEM. Changes over untreated were analyzed via multiple one-sample T-tests, with discovery determined using the Two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli. * = q<0.05, ** = q<0.01. n.d.=not detected

HLA-DR⁺ HCs at term were 2.5-fold more frequent after 48 hours of IFN-γ + LPS stimulation (Fig3Ai, STableII); a 2.7-fold CD80 upregulation was similarly delayed (q=0.005) (Fig3Aii, STableII). We noted a 2.4-fold reduction in CD163⁺ HCs also 48 HPT(q=0.00002). Both ArgI and iNOS transcription expanded 10-fold (Fig3Bi), maintaining a neutral M1:M2 ratio (Fig3Bii). M2C-type macrophages at baseline constituted 72.8% of term HCs, but were 13% 48 HPT with IFN-γ + LPS (Fig3C). M2A-type cell frequency increased 4.8-fold by 48 HPT (q=0.0007). Two percent of HCs were categorized as M1 after 48 HPT. Freshly isolated term HCs had little transcription of STAT1, but 24 HPT with IFN-γ + LPS increased STAT1 100-fold; STAT5A increased 3-fold; STAT3 and STAT6 remained unchanged (Fig3D). IFN-γ + LPS-stimulated term HCs secreted similar cytokines to early/mid-gestation HCs: growth factors GM-CSF and IL-7 were increased 19-fold (q=0.02) and 7.3-fold (q=0.02), respectively, over untreated controls (Fig3Ei, STableIII). Inflammatory cytokines increased but did not reach statistical significance (Fig3Eii, STableIII). IL-4 expanded 5.7-fold (q=0.017) (Fig3Eiii, STableIII). Proinflammatory chemokines MIP-1α and MIP-1β were elevated 34-fold and 5.3-fold, respectively over untreated controls (Fig3Ev, STableIII). Early/mid-gestation HCs appeared primed for a fast and robust response, as evidenced by the timing of phenotypic changes, and changes in secreted proteins. Term HCs produced some of the same cytokines after extended stimulation, equally supporting the recruitment of monocytes and T-cells, while maintaining a baseline M1:M2 ratio.

We used IL-4 + IL-13 to model M2A-type polarization(43). In early/mid-gestation, we noted a 1.9-fold loss of CD68⁺ cells (q=0.00003) and a 4.6-fold expansion of CD86⁺ cells 48 HPT (Fig4Ai, STableII). We detected a 2.8-fold increase in HLA-DR density 24 HPT (q= 0.004). Expression of CD80, CD206, and CD86 increased 48 HPT (Fig4Aii, STableII). ArgI transcription was reduced 24 HPT, while iNOS increased 10-fold (Fig4Bi); increased iNOS transcription reduced M2-dominance (Fig4Bii). IL-4 + IL-13 treatment reduced M2A-like cells; at baseline, these cells made up 40.4% of HCs, but after 24 HPT, this decreased to 10.4% (q<0.0001) (Fig4C); we noted a greater frequency of M2C-type HCs. STAT1 transcription was upregulated 10-fold, STAT6 was also elevated, and no changes were noted in STAT3 or STAT5A transcription (Fig 4D).

FIGURE 4.

HCs are highly responsive to IL-4 + IL-13 co-stimulation. HCs were stimulated with IL-4 + IL-13 simultaneously in order to induce a M2A phenotype. (A) Mean log₁₀ (Fold Change) ± SEM of (i) percent positive and (ii) MFI of positive population with values meeting thresholds q<0.05 and fold change of +/− 1.5, followed by representative flow plots therein (B) (i) Relative expression of ArgI and iNOS was determined by qRT-PCR using the ΔΔCt method (ii) M1:M2 ratio calculated as ΔCt-ARGI/ΔCt-iNOS visualized as median and quartiles. (C) HC’s were grouped into canonical macrophage subtypes via FACS as in Supplementary Figure 1 (D) Relative expression of STAT1, STAT3, STAT5A, and STAT6 ex vivo and after stimulation was determined by qRT-PCR using the ΔCt method. * = q<0.05, ** = q<0.01 *** = q<0.001 **** = q<0.0001; n.d.=not detected

At term, we observed a 1.7-fold loss of CD68⁺ HCs (q=0.002) and CD86⁺ HC frequency increased 9.7-fold 48 HPT (q=0.0005) (Fig4Ai, STableII). Further, we noted a 1.9-fold increase in CD209⁺ HCs (q=0.05) (Fig4Aii, STableII). CD206 and CD209 expression increased 24 HPT. At 48 HPT, CD80 expression increased 2.6-fold (q=0.006) and CD86 rose 1.6-fold (q=0.0001). We observed no changes in ArgI or iNOS transcription (Fig4Bi), and therefore no change in M1:M2 ratio (Fig4Bii). As expected following IL-4 + IL-13, the M2A subtype increased 3.2-fold from baseline 24 HPT (Fig4C). M2C-type HCs decreased to 42.5% 24 HPT and rebounded 48 HPT, suggesting a strong M2C preference at term. STAT1, STAT3, and STAT5A transcription were increased; confoundingly, STAT6 transcription, which is associated with IL-4 + IL-13 signaling, was reduced (Fig4D). IL-4 + IL-13-induced phenotypic changes occurred faster and more robustly at term than in early/mid-gestation HCs. Additionally, IL-4 + IL-13 treatment predictable increased M2A-like macrophages at term, not in early/mid-gestation.

We modeled M2B-type polarization using IL-1β treatment and heat-agglutinated human IgG (HAGG)(40). At early/mid-gestation, no changes were observed at the population level or with surface molecule density (STableII). No changes were observed in ArgI or iNOS transcription (Fig5Bi), and M1:M2 ratio was maintained (Fig5Bii). We noted a 4-fold reduction (q=0.005) in M2A-like cells through 48 HPT, with a 1.6-fold increase (q=0.006) in M2C-type cells (Fig5C). STAT1 and STAT6 transcription was minimally increased, while STAT3 and STAT5A remained constant (Fig5D).

FIGURE 5.

Co-stimulation of HCs with IL-1β + HAGG suggests mechanism to limit antibody-mediated activation at the placenta. HCs were stimulated with IL-1β + HAGG simultaneously in order to induce a M2B phenotype (A) Mean log₁₀(Fold Change)±SEM of percent positive with values meeting thresholds q<0.05 and fold change of +/− 1.5, followed by representative flow plots therein (B) (i) Relative expression of ArgI and iNOS was determined by qRT-PCR using the ΔΔCt method (ii) M1:M2 ratio calculated as ΔCt-ARGI/ΔCt-iNOS visualized as median and quartiles. (C) HC’s were grouped into canonical macrophage subtypes via FACS as in Supplementary Figure 1 (D) Relative expression of STAT1, STAT3, STAT5A, and STAT6 ex vivo and after stimulation was determined by qRT-PCR using the ΔCt method. * = q<0.05

In term HCs, we observed a 1.5-fold decrease in CD163⁺ HC frequency 48 HPT (Fig5Ai, STableII) and no changes in surface molecule density (STableII). No significant changes in ArgI or iNOS transcription (Fig5Bi), or M1:M2 ratio were detected (Fig5Bii). We observed opposite polarization patterns at term compared to early/mid-gestation HCs; M2C-like cells reduced to 32.5% 24 HPT (Fig5C) and M2A-like cells expanded 4-fold. We found 100-fold increase in STAT1 and STAT5A transcription, 10-fold STAT3 increase and maintenance in STAT6 (Fig5D). In sum, both early/mid-gestation and term HCs exhibited resistance to immune complex-mediated-activation.

Finally, M2C-polarization studies were conducted using IL-10 stimulation(43). CD86⁺ early/mid-gestation HCs decreased 4.7-fold by 48 HPT (q<0.000001) (Fig6Ai, STableII). We observed a delayed 1.9-fold increase in CD163⁺ cell frequency (q=0.009), as well as 1.7-fold increase of CD163 MFI (q=0.00004) (Fig6Aii, STableII). HLA-DR density reduced 2.6-fold 24 HPT (q=0.00002). ArgI and iNOS transcription was unaffected (Fig6Bi) and the baseline M1:M2 ratio was maintained (Fig6Bii). The proportion of M2C-like cells following IL-10 treatment reached 76.8% 24 HPT (q=0.001) (Fig6C). STAT6 transcription expanded 10-fold; transcription of the remaining STATs was unchanged (Fig6D).

FIGURE 6.

Early/mid-gestation HCs treated with IL-10 progress towards a term HC phenotype. HCs were stimulated with IL-10 in order to induce an M2C phenotype. (A) Mean log₁₀ (Fold Change) ± SEM of (i) percent positive and (ii) MFI of positive population with values meeting thresholds q<0.05 and fold change of +/− 1.5, followed by representative flow plots therein (B) (i) Relative expression of ArgI and iNOS was determined by qRT-PCR using the ΔΔCt method (ii) M1:M2 ratio calculated as ΔCt-ARGI/ΔCt-iNOS visualized as median and quartiles. (C) HC’s were grouped into canonical macrophage subtypes via FACS as in Supplementary Figure 1 (D) Relative expression of STAT1, STAT3, STAT5A, and STAT6 ex vivo and after stimulation was determined by qRT-PCR using the ΔCt method. * = q<0.05, ** = q<0.01 *** = q<0.001 **** = q<0.0001.

In term HCs, we observed a delayed 2.8-fold increase in CD163⁺ cell frequency (q=0.03), and a 2.6-fold reduction in CD86⁺ HCs (q<0.00001) (Fig6Ai, STableII). No additional phenotypic changes were observed (STableII). Decreased ArgI and decreased iNOS transcription (Fig6Bi) maintained M1:M2-neutrality (Fig6Bii). Confoundingly, M2C HCs were reduced from 72.8% to 29% 24 HPT (q<0.0001) (Fig6C). This was accompanied by a 4.0-fold increase in M2A-like cells (q=0.0006). STAT1, STAT3, and STAT5A transcription increased ~10-fold, and STAT6 slightly decreased (Fig6D). Both early/mid-gestation and term HCs progressed deeper into a regulatory phenotype following IL-10 exposure.

Placental IFN-λ1 may be less protective than IFN-α:

Our data suggests HCs differentially respond to polarizing stimuli with advancing gestational age. Therefore, we investigated whether this translated to differential responses to antiviral interferons (IFNs). We treated HCs with IFN-α or IFN-λ1, which is prevalent at the placenta and implicated in local viral resistance(46). Following IFN-α stimulation of early/mid-gestation HCs, CD80 expression increased 1.7-fold 48 HPT (q=0.003) (Fig7Ai, STableII). IFN-λ1 induced no significant marker changes (STableII). M2C-like cells increased 1.7-fold (q<0.0001) 24 HPT and M2A-like cells were lost (Fig7B,C) following both IFN- α and IFN-λ1. At term, neither IFN-α nor IFN-λ1 influenced HC phenotype (STableII). Following stimulation of term HCs with IFN-α or IFN-λ1, M2C-like cells were lost and M2A-like cells increased 4-fold (q=0.005) (Fig7C). Non-M1/M2 subtype increased among term but not in early/mid gestation HCs.

FIGURE 7.

Placental IFN-λ1 has reduced antiviral effects, as compared to IFN-α. HCs were stimulated with (A,B, D) IFN-α or (C,D) IFN-λ1. (A) Mean log₁₀ (Fold Change) ± SEM of MFI of positive population with values meeting thresholds q<0.05 and fold change of +/− 1.5, followed by representative flow plots therein (B, C) HCs were grouped into canonical macrophage subtypes via FACS as in Supplementary Figure 1 (D) Relative expression of ISGs IFIT1, IFITM1, OAS1, and VIPERIN at early (3 HPT) and late (24HPT) timepoints was determined by qRT-PCR using the ΔΔCt method visualized as mean ± SEM. * = q<0.05, ** = q<0.01

Key interferon-stimulated-gene (ISG) expression is pivotal to protection from viral infection. To determine antiviral mediator activation following IFN stimulation, we evaluated ISG transcription via qRT-PCR at short and prolonged exposures (Fig7E). IFN-α rapidly induced vigorous transcription of IFIT1, IFITM1, OAS1, and Viperin in both early/mid-gestation and term HCs. Importantly, IFN-λ1 did not induce ISG transcription as quickly or robustly at all timepoint; in both early/mid-gestation and term HCs, a longer IFN-λ1 exposure period was required to reach the ISG induction levels noted 3 HPT with IFN-α. Early/mid-gestation and term HCs responded similarly to IFN-α and IFN-λ1 stimulation. Both populations remained phenotypically stable through 48 HPT and both upregulated key ISGs to similar levels.

RIG-I agonism reveals that viral pattern-recognition-receptor (PRR) responses may be temporally regulated:

We observed differential responses to macrophage polarizing stimuli across gestation; however, we noted similar responses to antiviral IFNs throughout gestation. We further evaluated these functional discrepancies by interrogating the RIG-I pattern recognition receptor (PRR) pathway. Early/mid-gestation HCs increased frequency of CD86⁺ HCs 48 HPT (Fig 8A, STableII). iNOS transcription increased 2.2-fold 24 HPT (Fig8Bi), which shifted the M1:M2 ratio away from M2-dominance (Fig8Bii). We noted a loss of M2A-like HCs 24 HPT, with expansion of M2C-like HCs (Fig8C). STAT1 transcription increased 58-fold; STAT3 and STAT5A transcription remained unchanged, and STAT6 rose 7-fold, suggesting a role for alternative macrophage activation in response to RNA viruses (Fig8D). IFN-α and IFN-β transcription increased 43-fold (q=0.09) and 75-fold (q=0.09), respectively (Fig8Ei). MDA5 and RIG-I transcription rose 7- and 8-fold, respectively (Fig8Eii). Together, these suggest a capacity for antiviral responses following both active infection and replication of virus. Directly antiviral MX1 and Viperin were 9- and 49-fold more abundant after RIG-I agonism (Fig8Eiii). Although these findings are preliminary due to limited sample size, we observed increases in IFN-α (Fig 8Fi), IFN-γ, IL-1β, IL-12, and TNF-α over untreated controls in early/mid-gestation (STableIII). Lastly, we observed induction of T-cell chemoattractants RANTES and IP-10 (Fig8Fiv, STableIII).

FIGURE 8.

Agonism of the antiviral intracellular innate immune PRR RIG-I reveals temporally-controlled inflammatory responses, with early/mid-gestation HCs, but not term HCs, adopting activated signatures. HCs were stimulated with 5’ppp-dsRNA. (A) Mean log₁₀ (Fold Change) ±SEM of percent positive with values meeting thresholds q<0.05 and fold change of +/− 1.5, followed by representative flow plots therein (B) (i) Relative expression of ArgI and iNOS were determined by qRT-PCR using the ΔΔCt method (ii) M1:M2 ratio calculated as ΔCt-ARGI/ΔCt-iNOS visualized as median and quartiles.. (C) HC’s were grouped into canonical macrophage subtypes via FACS as in Supplementary Figure 1 (D) Relative expression of STAT1, STAT3, STAT5A, and STAT6 ex vivo and after stimulation were determined by qRT-PCR using the ΔCt method. (E) Relative expression of (i) interferons, (ii) intracellular pattern recognition receptors, and (iii) interferon-stimulated-genes was determined by qRT-PCR using the ΔΔCt method visualized as mean ± SEM. (F) Quantification of (i) inflammatory cytokines (ii) anti-inflammatory cytokines (iii) IL-2R and (iv) chemokines following 48 hours of in vitro culture. Open circles indicate control untreated cells and closed circles RIG-I agonist stimulated cells, visualized as mean ± SEM. Changes over untreated were analyzed via multiple one-sample T-tests, with discovery determined using the Two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli. * = q<0.05, ** = q<0.01.

Term HCs were phenotypically stable after RIG-I activation (STableII) but increased iNOS 3.3-fold 24 HPT (Fig8B). We observed a loss of M2C-like cells and a 4-fold M2A-like cell increase (q=0.005) (Fig8C). STAT1 rose 300-fold, STAT3 30-fold and STAT5A 22-fold (Fig8D). STAT6 transcription was unchanged. IFN-α transcription was unchanged, and IFN-β only 10-fold higher (Fig8Ei). MDA5 and RIG-I both increased 3-fold, but less than at early/mid-gestation (Fig8Eii). Finally, ISG MX1 was upregulated 6-fold and Viperin 8-fold (Fig8Eiii), similarly to early/mid-gestation. Protein analysis of term HC supernatants collected after 48 hours of RIG-I agonism showed minimal changes over untreated controls. No changes in growth factors, most inflammatory (Fig8Fi, STableIII), or regulatory cytokines (Fig8Fii, STableIII) were observed. IFN-α increased 2.6-fold (Fig8Fi, STableIII) similarly to early/mid-gestation HCs, and IP-10 511-fold (Fig8Fiv, STableIII).

The RIG-I pathway exemplified the greatest difference between early/mid-gestation and term HCs. Early/mid-gestation HCs upregulated activation molecules, antimicrobial-associated enzymes, and inflammatory transcription factors; this all cumulated in production of inflammatory mediators which was not seen in term HCs. Enhanced IFN, PRR-associated, and VIPERIN gene transcription at early/mid-gestation underscores the robustness of the RIG-I induced response.

Discussion

HCs are an abundant immune cell population in the human placenta and present throughout gestation. They may be the “first” fetal immune cell. However, there are very few studies identifying changes to phenotype and function during healthy pregnancy. We address this gap demonstrating dynamic immunophenotypes ex vivo across gestation. Activated HCs were present from early pregnancy in high numbers. Although their activated cell frequency decreased by term, phenotype essentially remained unchanged. Cytokine quantification and enzyme expression underscored inflammatory phenotypes early in gestation and suggest recruitment of labor-promoting neutrophils at term(47). HCs expressing tolerogenic markers were highest at mid-gestation, accompanied by an M2A- to M2C-like shift. Chemokine levels may reflect monocytic recruitment mid-gestation, the time when fetal-liver and bone-marrow-derived-monocytes populate the placenta. Our data suggest the existence of two subgroups at term: CD80^hi, iNOSi⁺, M2B-like HCs and a subset echoing regulatory-phenotypes observed mid-gestation. Phenotypic discrepancies may prepare for parturition and placental rejection, which has been previously reported to occur with a resurgence of inflammation and increases in iNOS in labored trophoblasts(42). The presence of M2B-like macrophages in some donors at term suggests a role for this subtype in initiating or regulating labor-associated inflammation; M2B macrophages have been reported to regulate the depth and breadth of inflammatory reactions(47). In summary, HCs appear to be a heterogenous population of M2-like macrophages, with regulatory and anti-inflammatory functions critical to maternal-fetal homeostasis and fetal development. Their differentiation may be etiologically determined and only further influenced by currently undefined local immune signals(13). Our study suggests that HC phenotypes are temporally-regulated through gestation and these changes impact susceptibility to vertically-transmissible infections.

We interrogated HC functional plasticity through gestation via in vitro macrophage-polarizing stimuli. HC plasticity was reduced at term; phenotypic changes were less marked and delayed. Following stimulation with IFN-γ + LPS, which may occur during maternal bacterial infection, early/mid-gestation HCs demonstrated elevated iNOS and STAT1 transcription, while term HCs maintained their M1:M2 ratio. Phenotypic changes and pyrogen production(24) were comparable but delayed at term. This difference in response may be due to no ex vivo STAT1 transcription at term; in contrast, early/mid-gestation HCs exhibited transcription of all measured STATs, including STAT1. Importantly, we did not observe any neutrophil-chemotactic IL-8; recruitment of these strongly inflammatory cells to such an immunoregulated organ may result in pre-term labor or miscarriage. All HCs responded similarly to IL-4 + IL-13 stimulation, consistent with M2 immmunodominance-like phenotypes (14, 15, 16, 17). It follows that this stimulation would most readily produce phenotypic changes in a way that primes for activation without excessive inflammation. No responses were observed following HAGG + IL-1β stimulation, suggesting HCs are resilient to phenotypic changes induced by immune complexes in the presence of inflammatory cytokines and that changes in STAT gene transcription alone may not be sufficient to produce phenotypic changes in our experimental timeline. Since transcytosis of maternal IgG is a key function of the placenta, our data suggests there may be mechanisms to prevent aberrant immune complex activation of HCs during this process, potentially through FcγRIIb-mediated inhibition(48). HCs have previously been reported to express FcγRIIb and endocytosed IgG co-localizes with FcγRIIb in endothelial cell endosomes. Early/mid-gestation HC stimulation with IL-10 promoted a term-HC-like phenotype, with increased expression of regulatory markers and loss of activation.

We evaluated how IFN responses differ from viral PRR activation. Term HC phenotype was stable following IFN stimulation; IFN-α activated early/mid-gestation HCs. Sensitivity to IFN-α was greater than to IFN-λ1, as IFN-α induced transcription of ISGs faster and more robustly. IFN-λ1 is a key IFN produced by placental syncytiotrophoblasts, and the reduced antiviral state following stimulation with this IFN may highlight a gap in placental antiviral immunity. RIG-I initiates antiviral responses to viral RNA(49), and stimulation with RIG-I agonist mimics activation of this PRR following select viral infections. RIG-I-induced ISG transcription was stronger at early/mid-gestation. Early/mid-gestation HCs quickly adopted classically-activated signatures and some donors produced inflammatory cytokines whereas term HCs were unaffected despite transcriptional adaptations. Kim et al. demonstrated pro-M1 gene hypermethylation(50) and Blumenstein et al(51, 52) suggested suppression of cytokine signaling (SOCS) protein is differentially regulated with the onset of labor at term. As such, genomic modifications or inhibitory proteins may contribute to plasticity variations. Inflammatory mediators can damage the villous cell barrier or induce preterm labor(53) therefore this response may be protective or pathogenic. Additional studies are needed to define HC function in healthy pregnancies, and those confounded by infectious and noninfectious complications.

Tissue-specific immune responses among privileged sites are well documented(54). We demonstrate temporal control of immune responses as part of the natural order of placental maturation. We demonstrate changing HC phenotype ex vivo and evolving responses to select stimuli. Understanding how HCs maintain pregnancy is of utmost importance to guide development of therapies to offset aberrant HC phenotype-mediated complications during pregnancy.

A deficiency in the study is lack of labored tissues at term and preterm, which would allow for analysis of mechanisms responsible for placental rejection, guidance on pathogenic changes, and potential therapeutic targets. Conclusions here and in similar studies are limited by incomplete patient histories: placentae from elective terminations are considered “normal and healthy;” however, no information was available to investigators regarding why pregnancy was terminated. This study would have benefited from a greater sample size with a greater range of gestational ages, especially in the first trimester; limited donors may not reflect the “true” situation of a specific time internal.

Future studies should prioritize collecting a greater samples number from a breadth of gestational ages and consider donor demographics. Studies detailing HCs isolated from pathogenic premature deliveries could identify important changes leading to parturition. Analyzing HCs from labored placentae may elucidate the M2B-like HCs identified here. Analysis of non-M1/M2 populations and HCs as a whole would benefit from high-dimensional techniques using unsupervised clustering algorithms and functional analyses performed on sorted populations. Consistent changes observed here over a breadth of stimulating conditions suggests a strong plating effect, and future studies should be mindful of the effect of such phenotypic changes on their conclusions.

Key points:

• HC exhibit a more inflammatory phenotype early in gestation

• HC responses to some pathogenic stimuli reduce with advancing gestational age