Maternal T cells in the Human Placental Villi support an Allograft Response during Non-Infectious Villitis

Abstract

During human pregnancy, proinflammatory responses in the placenta can cause severe fetal complications, including growth restriction, preterm birth, and stillbirth. Villitis of unknown etiology (VUE), an inflammatory condition characterized by the non-infectious infiltration of maternal CD8+ T cells into the placenta, is hypothesized to be secondary to either a tissue rejection response to the haploidentical fetus or from an undiagnosed infection. Here we characterized the global T cell receptor (TCR) beta chain profile in human placental T cells diagnosed with VUE compared to control and infectious villitis diagnosed placentae by immunoSEQ. Deep sequencing demonstrated that VUE is driven predominantly by maternal T cell infiltration, which is significantly different from controls and infectious cases; however, these T cell clones show very little overlap between subjects. Mapping TCR clones to common viral epitopes (cytomegalovirus [CMV], Epstein-Barr virus and influenza A) demonstrated that antigen specificity in VUE was equal to controls and significantly lower than CMV-specific clones in infectious villitis. Our data indicate VUE is an allograft response, not an undetected infection. These observations support the development of screening methods to predict those at risk of VUE and the use of specific immunomodulatory therapies during gestation to improve outcomes in affected fetuses.

Introduction

Villitis of unknown etiology (VUE) is a destructive inflammatory process in placental tissue initially described in 1975 (1). Lesions have since been characterized at a cellular level by infiltration of maternally-derived CD8+ T cells and fetal macrophages (Hofbauer cells) into the chorionic villi (2, 3). VUE of various grade is diagnosed in 6-34% of all third trimester placentae (4, 5); however, pathological evaluation of the placenta is not routinely preformed in all cases, and thoroughness of histologic sampling has not historically been standardized (6), so a true incidence is difficult to ascertain. Clinical outcomes of VUE most often result in the delivery of a healthy infant, but VUE is also associated with intrauterine growth restriction (IUGR) (7, 8), preterm birth (9, 10), fetal/neonatal demise (7, 11) and neonatal neurocognitive impairment (12, 13). As the risk for recurrence is reported to be between 10-37% (2, 14), it is important to establish a diagnosis, and counsel affected women on the risks of a VUE diagnosis in a subsequent pregnancy. Specific management including treatment guidelines for pregnancies following a diagnosis of VUE are lacking, and reflect limited understanding of the pathophysiology of this poorly-characterized condition.

Villitis of unknown etiology may be secondary to an undetected infection during gestation or reflect an aberrant maternal immune response, resembling tissue rejection (2, 4, 5). While some studies have reported infectious villitis cases that were misdiagnosed as VUE (15), there are many features unique to each pathology. For example, infectious villitis typically develops earlier in pregnancy, is rarely recurrent, leads to nearly all villi being abnormal, and can show a mixed inflammatory infiltrate including lymphocytes, neutrophils, and/or plasma cells (the latter especially with CMV infection), whereas VUE typically develops near term, presents as focal/patchy infiltration of lymphocytes into the terminal and stem villi, and is often recurrent (2). As the fetus is a semi-allograft, maternal immunity must be tightly regulated (16, 17). Potential breakdown in tolerance of the fetal allograft and immune rejection is supported by the fact that VUE is also associated with maternal autoimmunity (18, 19), neonatal alloimmune thrombocytopenia (20), and in vitro fertilization with a donor oocyte (21, 22).

Maternal T cells, specifically cytotoxic CD8+ cells, are not typically found in placental villi but are present in VUE (2, 3). These cells play an important role in host defense against viruses by recognizing non-self antigen presented to their T cell receptor (TCR) by major histocompatibility complex (MHC) class I molecules. Killing of target cells expressing specific antigen occurs by release of proinflammatory cytokines and cytotoxic perforin and granzyme B. The TCR is composed of two protein chains, with 90% of the T cell population expressing an α and β chain, and a small subset having a γ and δ chain. The diversity of TCRs is mostly found in the third complementarity-determining region (CDR3). This variability is due to somatic variable (V), diversity (D) and joining (J) recombination in the α/β chain, giving T cells a diverse antigen repertoire to initiate immune responses. Profiling the T cell repertoire using single-molecule DNA sequencing and multiplex PCR methods have provided valuable insight into many diseases (23, 24). In transplant biology, TCR spectratyping has been used to monitor patients after haematopoietic stem cell transplantation and helps predict the efficacy and occurrence of graft-versus-host disease (GVHD) (25–27). In cancer, TCR profiling demonstrated an expansion of T cell clones in patients who responded to immunotherapy, compared to those who progressed (28).

Sequencing the T cell repertoire has not been utilized for characterization of VUE but may provide valuable new insights into the origin and pathogenesis of this poorly understood condition. Therefore, this study defined TCR diversity in 10 placentae with VUE compared to 10 gestational aged-matched unaffected control placentae and 5 placentae with infectious villitis. We hypothesized that T cells from placentae with VUE would demonstrate distinct clonal expansion, independent of known infectious causes. We noted that VUE expanded clones were unique to each subject reflecting a broad repertoire that was not pathogen-specific unlike the infectious villitis cases. Increased T cell clonality in placentae diagnosed with VUE, with minimal overlap between patients and common viral epitopes, suggests that T cell expansion is not linked to a single initiating antigen. These results further support an allograft response by maternally-derived T cells in gestation during VUE.

Materials and Methods

Patient selection

This study is approved by the Mayo Clinic institutional review board (IRB# 16-006099) in accordance to the Declaration of Helsinki principles. Twenty five residual placental tissue samples (formalin-fixed paraffin embedded - FFPE) collected between 2014 and 2019 were utilized for this study. Ten tissues with a pathological diagnosis of high grade VUE were selected by Amsterdam Criteria (29). These samples were matched by gestational age with 10 control placentae. An additional five placentae diagnosed with infectious villitis (4 cytomegalovirus [CMV] and 1 toxoplasma gondii confirmed cases), were included but were not matched to VUE or control cases.

Fluorescence in situ hybridization

Six placental samples from a male fetus were used to identify infiltrating cell origin by fluorescence in situ hybridization (FISH). Slides were placed in a 90°C oven for 15 minutes. Slides were then deparaffinized with xylene (2 times, 15 minutes each) at room temperature (RT), dehydrated in 100% ethanol for 5 minutes at RT, and placed in 10mM Citric Acid (pH 6.0) and microwaved for 10 minutes. The slides were then dehydrated using an ethanol series (70, 85, 100%) 2 minutes each at RT. Working solution of DXZ1/DYZ3 (Abbott Laboratories, Des Plaines, IL, USA) was made by mixing 1 µL of concentrated DXZ1/DYZ3 probe with 9 µL of LSI/WCP hybridization buffer (Abbott Laboratories). The working solution was applied to the target areas, cover slipped, co-denatured with a ThermoBrite (Abbott Laboratories) at 83°C for 5 minutes, and hybridized overnight in a 37°C humidified oven. Following hybridization, slides were soaked in RT 2xSSC/0.1% NP-40 to remove coverslips, placed in 2xSSC/0.1% NP-40 at 74°C for 2 minutes and then placed into RT 2xSSC/0.1% NP-40 for 2 min. The slides were stained with 4’−6,-diamidino-2-phenylindole (DAPI) (Vector Laboratories) and cover slipped. Tissue samples were scanned in their entirety and the qualitative result was determined based on observed signal patterns by CytoVision (Leica Biosystems, Wetzlar, Germany).

Deep sequencing

Each FFPE tissue was cut into five, 10 micron scrolls and DNA isolated using the AllPrep DNA/RNA FFPE Tissue kit per manufacturer’s instruction (Qiagen, Valencia, CA). Concentration and purity of isolated DNA was determined by NanoDrop, while RNA quality was assessed using a Bioanalyzer (Agilent Technologies, Santa Clara, CA). The extracted DNA was used for deep sequencing of the TCR-β chain of all T cells in placental villi using the ImmunoSEQ platform v4 (Adaptive Biotechnologies, Seattle, WA). Investigators running this assay were blinded to the group allocation of the samples. Briefly, the assay amplifies rearranged complementarity-determining region 3 (CDR3) of TCRs using a bias-controlled, multiplex-PCR method (23, 24). Samples were sequenced on an Illumina MiSeq platform using a 100 cycle paired end protocol (Illumina, San Diego, CA). The quantity of nucleated cells and T cells was determined during sequencing through the addition of reference gene primers and TCR-β specific primers (24). Raw sequencing data were uploaded and are freely accessible through immuneACCESS (clients.adaptivebiotech.com/pub/enninga-2020-ji; DOI 10.21417/EALE2020JI).

Calculations of richness and diversity

Diversity and richness were calculated using the Adaptive Biotechnologies ImmunoSEQ Analyzer Software. Heterogeneity, defined here as T cell richness in our sample, was estimated by the iChao1 method (30). iChao1 is a nonparametric estimator of the lower bound, or minimum number of unique templates predicted within a subject’s repertoire at 95% confidence. Diversity scores were used to analyze highly abundant TCR clones in each sample (31). Clonality is based on Pielous evenness (J) and Shannon diversity (H) (32), which calculates a number between 0 and 1, with the higher value indicating lower diversity, or increased expansion of a single clone, in the sample. Formulae are given below, where p_i is the proportion of clonotype i, b is the base of the logarithm, and S is the number of clonotypes.

$$Shannon\left(H\right)=-\sum p_i{log}_b{p}_i$$

$$Pielou\left(J\right)=\frac{H}{logS}$$

$$Clonality=1-J$$

Statistical Analysis

Sequencing data was analyzed using the ImmunoSEQ Analyzer from Adaptive Biotechnologies (http://www.adaptivebiotech.com/immunoseq). For patient demographics and continuous variables, medians and interquartile ranges were utilized. Kruskal-Wallis tests were used to compare groups, and the false discovery rate was used for multiple comparisons testing using the Benjamini, Krieger and Yekutieli method (33). For VJ and epitope mapping, means and standard deviations were reported. Group comparisons were made by ANOVA using the Benjamini Yekutieli test to correct for multiple comparisons. Amino acid sequences and their corresponding VDJ alleles determined during sequencing were uploaded into the VDJdb browser (https://vdjdb.cdr3.net/), a curated database of T cell receptor sequences with known antigen specificity (34). T cell clones specific for CMV, EBV, and influenza A were identified in each group using simple scoring with a confidence of 3 and substitution of 2. All analyses were two-sided and significance was defined as a p-value less than 0.05. Graphing and analyses were performed using GraphPad Prism 8 (GraphPad Software, San Diego, CA).

Results

Subject characteristics

Placentae were collected following delivery and underwent gross and histological evaluation. For this study, 10 VUE diagnosed placentae were chosen and matched with 10 control placentae by gestational age at delivery. An additional 5 confirmed infectious villitis cases (4 CMV and 1 Toxoplasma gondii) were also included but were not matched to the control or VUE cases. Two control placentae were delivered preterm, but lacked significant pathology upon examination. No differences were observed between all three groups related to BMI, APGAR scores and fetal length. Women in the cohort with infection were younger, had smaller fetuses, and delivered preterm or suffered an intrauterine demise (n=2). Women in the VUE cohort had significantly higher gravidity and parity than the control and infectious villitis cohorts. Women in the VUE cohort also had more prior miscarriages compared to the other cohorts and 50% underwent a Cesarean delivery for the current pregnancy similar to the infection cohort. These data are summarized in Table 1.

Table 1:

Clinical characteristics of fetal and maternal subjects

Maternal Demographics	Control (n=10)	VUE (n=10)	Infection (n=5)	Kruskal-Wallis p-value
Age (years)	30 (27-38)	33 (22-40) ‡	24 (24-32) Ø	0.042
Gravidity	2 (1-4)	3.5 (2-6) #,‡	2 (2-3)	0.008
Parity	1 (1-3)	2 (1-5) #	2 (1-2)	0.018
Pre-pregnancy BMI (kg/m2)	23.4 (18.8-45.2)	28.8 (19.8-45.2)	26.2 (21.5-35.8)	0.187
Smoking	2/10	1/10	1/5
Prior miscarriage	0/10	6/10	1/5
Vaginal Delivery	9/10	5/10	2/5
Autoimmune Disease	0/10	3/10	0/5
Fertility Treatment	0/10	1/10	0/5
Fetal Demographics	Control (n=10)	VUE (n=10)	Infection (n=5)	Kruskal-Wallis p-value
Gestational Age at Birth	38 2/7 (34 2/7-40 1/7)	38 4/7 (34 3/7-39 5/7) ‡	33 3/7 (27 1/7-36 1/7) Ø	0.006
APGAR 5 minute	8 (6-9)	8 (4-9)	6 (0-9)	0.579
Weight (kg)	3.4 (2.3-4.5)	3.0 (2.6-3.7) ‡	1.4 (1.0-3.0) Ø	0.014
Length (cm)	50 (44.5-53.5)	50.5 (44.5-54.0)	38 (33.0-48.3)	0.114
Female Fetus	4/10	8/10	2/5

^*Data are presented as medians with ranges. Significant post hoc comparisons indicated by

^#control vs. VUE,

^Øcontrol vs. infection

^‡VUE vs. infection.

Total T cell composition in the placenta

By definition, histological evaluation demonstrated increased lymphocytic infiltration of chorionic villi, with resulting fetal vascular obliteration and tissue necrosis in placentae diagnosed with VUE compared to unaffected controls by H&E staining at 100X magnification (Figure 1A). Conversely, infectious villitis histology demonstrated at least focal plasma cell infiltration and CMV inclusions, which were absent in VUE and controls. The one case of toxoplasmosis demonstrated focal lymphoplasmacytic deciduitis and granulomas, along with Toxoplasma tachyzoites and pseudocysts confirmed by immunohistochemistry in sections of the umbilical cord. FISH probes against the X (green) and Y (red) chromosomes were consistent with previous reports that the infiltrating cells in diagnosed VUE from male placentae are of maternal origin (XX positive; Figure 1B). While the infectious villitis cases demonstrated XX staining, these maternal cells were observed at low levels and in small pockets indicating that, unlike VUE, this pathology is predominantly driven by cells of fetal origin. These small pockets of maternal cells in infectious villitis may coincide with the small pockets of plasma cells observed by H&E. Following nucleic acid extraction, the median amount of DNA was not significantly different between the three groups (p=0.093; data not shown). Deep sequencing demonstrated a similar number of all nucleated cells in both control and VUE tissues (69,540 versus 56,214, respectively, p=0.1738) but significantly more nucleated cells were present in infected placentae (295,562, p=0.0208; Figure 1C). As expected from histological review, the median number of T cells in placentae identified by deep sequencing was higher in the VUE group compared to the control group (1,340 versus 292, p=0.0012); however, the infectious villitis cases had the highest number of T cells identified (9,194) which was significantly different from control (p<0.0001) and VUE (p=0.0289; Figure 1D) subjects. Lastly, the ratio of T cells to all nucleated cells isolated from placental villi demonstrated a 6-fold increase in T cells in VUE compared to controls (p=0.0011) and a 150 fold increase compared to infectious villitis (p<0.0001; Figure 1E). These results indicate that unlike VUE, which is driven predominately by maternal T cells in the placenta, infectious villitis is characterized by massive fetal immune cell infiltration into the villi involving fewer T cells.

Figure 1: Basic characteristics of VUE, infectious villitis and control placentae.

A) Representative H&E images of VUE, infectious villitis and control placentae (100X magnification). B) FISH-based detection of male fetal and maternal cells in the chorionic villi (X chromosome in green, Y chromosome in red). C) Sampling depth between groups. D) Total T cell counts in each group. E) Ratio of T cells to all nucleated cells. Bars designate median and interquartile range (n=5-10/group). Boxes indicate 25th and 75th percentiles; lines indicate median and whiskers mark the range. Data were compared by Kruskal-Wallis testing, with post hoc analysis using the Benjamini, Krieger and Yekutieli method. Significant comparisons (p≤0.05) are indicated by # control vs. VUE, Ø control vs. infection and ‡ VUE vs. infection.

T cell receptor diversity in VUE

TCR composition in the placenta was measured and compared between groups. The average total number of productive TCR rearrangements in VUE diagnosed placentae was significantly higher compared to controls (2060 versus 489, p=0.002), but lower when compared to infectious villitis (11,528, p=0.0496; Figure 2A). This result was also observed for unique TCR sequences, or sequences with a defined TCR-β CDR3 combination in VUE compared to controls (1293 versus 473, p=0.0387) and VUE compared to infectious villitis (1293 versus 3653, p=0.0415; Figure 2B). Next the unique/total CDR3 reads ratio was measured to understand TCR diversity, which is an estimate the number of clones in the entire repertoire. A lower ratio means less diversity, or an expansion of certain clones. The median ratio of VUE TCRs was 0.68, which was significantly lower than control TCRs (0.96; p=0.001; Figure 2C), suggesting that T cell diversity is decreased in VUE diagnosed placentae. At the same time, the median ratio of infectious villitis TCRs was 0.50, which was not different from VUE (p=0.06). iChao1 was used to estimate TCR richness of individual repertoires in a sample by calculating a lower bound of richness. The data demonstrated decreased richness in VUE clones compared to infection (7,724 versus 14,988, p=0.0257); however, VUE richness was unchanged compared to control cases (7,692, p=0.7829; Figure 2D). Finally, clonality was measured with numbers closest to 0 indicating the sample is highly polyclonal whereas numbers closest to 1 are monoclonal. Significant expansion of T cell clones was observed in VUE cases (0.057) and infectious villitis (0.133) compared to controls (0.006, p=0.0025 and p=0.0274, respectively; Figure 2E). These data indicate that CDR3 T cell clones in VUE and in infectious villitis undergo clonal expansion in placental tissue compared to controls; however, the TCR repertoire of VUE T cells is less diverse compared to infectious villitis.

Figure 2: TCR diversity and clonality are related to VUE diagnosis and not infectious etiology.

A) Sum of productive templates, B) unique TCR reads, C) unique/total reads ratio of the TCR repertoire in VUE and control. D) richness as determined by iChao1. E) Clonality of VUE compared to infection and control (values closer to 1 represent samples with few predominant rearrangements dominating the repertoire, whereas clonality values near 0 represent polyclonal samples). Box graphs indicate 25^th and 75^th percentiles (n=5-10/group); lines indicate median and whiskers mark the range. Data were compared by Kruskal-Wallis testing, with post hoc analysis using the Benjamini, Krieger and Yekutieli method. Significant comparisons (p≤0.05) are indicated by # control vs. VUE, Ø control vs. infection and ‡ VUE vs. infection.

Distribution of clones in VUE

Identifying dominant and shared T cell clones may assist with recognizing which epitopes are driving TCR expansion in VUE and infectious villitis. Therefore, the most frequently occurring rearrangements within a sample were identified, along with how much of the sample is made up of these rearrangements considering the top 100 clones in each group. Across all placentae diagnosed with VUE and infectious villitis, there was a larger distribution of top T cell clones making up the entire sample compared to controls (Figure 3A). This suggests the TCR-β repertoire is driven largely by expanded clonotypes, which is not observed when compared to controls. Further considering only the top 15 most abundant clones, these made up 15% of the entire repertoire in VUE placentae, and 20% of the infectious T cell repertoire compared to controls (9%, p=0.007 and p=0.002; Figure 3B). There was no difference in abundant clones when comparing VUE to infectious villitis (p=0.085). Together, this indicates that expanded T cell clones make up a larger population of the total T cell pool in VUE and infectious villitis compared to controls. The length of the CDR3 in nucleotides was compared between control, infectious villitis, and VUE placentae to identify possible differences in rearrangement length which could be used to distinguish disease versus control. Although the frequency of these clones is highest in VUE, the median length of the CDR3 in each group was 42 nucleotides (Figure 3C). Lastly, clonotype sharing between all three groups was measured. There was little overlap between TCR clones in VUE samples compared to the control group (Figure 4). However, increased overlap was observed between infectious villitis cases and shared similarly at low abundance across controls and VUE. Looking specifically at sequences, the most abundant TCR rearrangements in VUE were only present in one subject, which was similar for infectious villitis but sequence abundance was much higher (Table 2). For shared rearrangements, the top amino acid sequences were present in 40% of VUE samples at low abundance (Table 3). Conversely in infectious villitis, there were more shared clones but at a much lower frequency, suggesting T cell responses to infection (i.e. CMV) are not driven by the same viral antigen. In sum, these data suggest that although expansion of certain T cells is occurring in VUE, no single antigen is responsible for this result as clonal overlap is limited; therefore, each case is unique to an individual pregnancy.

Figure 3: Characteristics of T cell clones in VUE compared to controls.

A) Proportion of the top 100 clones in VUE (V), infectious villitis (I) and control (C) placentae. B) Quantification of the top 15 clones in each group. Bars designate median and interquartile range. Data were compared by Kruskal-Wallis testing, with post hoc analysis using the Benjamini, Krieger and Yekutieli method. Significant comparisons (p≤0.05) are indicated by # control vs. VUE, Ø control vs. infection and ‡ VUE vs. infection. C) Distribution patterns of unique CDR3 sequences with different nucleotide lengths in control, infection and VUE samples. Data were fitted by non-linear curve; vertical line indicates median length (n=5-10/group).

Figure 4: Clonotype sharing between VUE, infected and control placentae.

TCR-β amino acid clonotypes that overlap between VUE (V), infectious villitis (I) and control (C) repertoires (n=5-10/group). Assessed using the Morisita-Horn index and illustrated as a heat map.

Table 2:

Most abundant TCR rearrangements in each cohort

Control	Amino Acid (AA) Sequence	Productive Frequency	# of Subjects AA Sequence is Present In
1	CASSIDRGKDEQFF	3.67%	1
2	CASSQTGPEQFF	3.18%	1
3	CASSPDTVNTEAFF	2.49%	1
4	CASSSDRAQETQYF	2.36%	1
5	CASSTNRDTQYF	2.09%	2
VUE	Amino Acid (AA) Sequence	Productive Frequency	# of Subjects AA Sequence is Present In
1	GDRGRNTIYF	4.95%	1
2	CASSLGGGPDEQFF	4.94%	1
3	ASTSGGLKETQYF	3.30%	1
4	CASSLIRWAYSYNEQFF	2.84%	1
5	CASSAVGNQPQHF	2.44%	1
Infection	Amino Acid (AA) Sequence	Productive Frequency	# of Subjects AA Sequence is Present In
1	CASSVGDRGTEAFF	18.83%	1
2	CASSQRATQYF	16.27%	1
3	CASSLTGNTGELFF	13.89%	2
4	CASSLIGVSSYNEQFF	12.01%	1
5	CASSISGGSTGYTEAFF	11.82%	1

Table 3:

Most commonly shared TCR rearrangements in each cohort

Control	Amino Acid (AA) Sequence	Productive Frequency	# of Subjects AA Sequence is Present In
1	CASSLQNTEAFF	0.57%	3
2	CASSLAGGPDTQYF	0.38%	3
3	CAWSVGYEQYF	0.44%	2
4	CASSTNRDTQYF	2.09%	2
5	CASSQDQETQYF	0.44%	2
VUE	Amino Acid (AA) Sequence	Productive Frequency	# of Subjects AA Sequence is Present In
1	CASSLGETQYF	0.71%	4
2	CASSLAGGYEQYF	0.31%	4
3	CASSLASTDTQYF	0.69%	3
4	CASSLAETQYF	0.18%	3
5	CASSLQGNYGYTF	0.15%	3
Infection	Amino Acid (AA) Sequence	Productive Frequency	# of Subjects AA Sequence is Present In
1	CASSLGQNTEAFF	0.054%	4
2	CASSFTDTQYF	0.044%	4
3	CASSVGGNTEAFF	0.037%	4
4	CASSLTANYGYTF	0.036%	4
5	CAWTGGYGYTF	0.042%	3

V-J gene segments in VUE

The composition of the T receptor-β V and J segments were profiled in VUE, infectious and control placentae to identify areas of high variability which are disproportionally represented in our groups. For the V gene, 59 different alleles and 13 J gene alleles were measured. Using a Benjamini Yekutieli correction for multiple testing, we observed differences in V gene families V2, V4, V5, V6, V7, V9 and V10 between our three groups (Figure 5A). Interestingly, the J1 family showed increased representation in infectious villitis, while the J2 family was observed at a significantly lower frequency in VUE and infectious villitis than in controls (Figure 5B). Using these segments, VDJdb browser was used to query CDR3 data against antigens specific to CMV, EBV and influenza A. T cell clones specific to these common viruses were identified in every sample in our cohort; however, the bulk of TCR clonotypes detected in VUE did not recognize known viral epitopes. As expected, Figure 5C showed infectious CMV cases having higher median frequencies of clones recognizing CMV epitopes (0.011) compared to VUE (0.001; p<0.0001). No differences were observed between CMV specific frequencies in VUE compared to control placentae (0.0011 vs. 0.0018; p=0.282). Interestingly, a significant increase in clonotype frequency was only observed when comparing control to both placental pathologies for EBV (p<0.0001; Figure 5D) and influenza A specific epitopes (p<0.0001; Figure 5E). Thus, the frequency of T cells recognizing viral antigens in VUE was equal to the control placentae and not to what is observed in infectious villitis. Together, these data indicate that the maternally-derived T cells infiltrating placental villi during VUE do not recognize common viral antigens like their infectious counterparts, providing more evidence that VUE represents an allograft response and not an undetected infection.

Figure 5: Gene usage and antigen specificity in VUE compared to infectious villitis and control T cells.

A) TCR-β V gene usage in different groups. B) TCR-β J gene usage in each group. Antigen specificity determined by VDJdb database of clones per group per antigen in C) cytomegalovirus (CMV), D) Epstein-barr virus (EBV) and E) influenza A virus. Graphs represents mean and standard deviation of hundreds of clones from each group (n=5-10/group). Data were compared by ANOVA, and multiple comparison corrections were applied by Benjamini Yekutieli test. Significant comparisons (p≤0.05) are indicated by # control vs. VUE, Ø control vs. infection and ‡ VUE vs. infection.

Discussion

In normal pregnancies, T cells do not accumulate inside the fetal chorion of the placenta but are found in relatively small numbers in the maternal decidua (35). VUE is a non-infectious placental pathology characterized by the presence of high levels of maternally-derived CD8+ cells in the chorionic villi; however, little is known about these T cells. This study profiled the total TCR in VUE, villitis with diagnosed infection (mainly CMV) and control placentae in order to characterize the specificity of these cells. As expected, T cells make up a majority of the cells infiltrating during VUE but not during infectious villitis. Although specific T cells clones were expanded in both VUE and infectious villitis, abundant clones were not shared between the cases profiled suggesting a unique immune response is generated in each pregnancy. Lastly, antigen mapping indicated that infiltrating T cells in VUE did not recognize common viral pathogens compared to infectious villitis. This observation supports the hypothesis that VUE represents an individually distinct allograft response, potentially resulting in fetal rejection in the third trimester or earlier, as opposed to a viral-specific response to a common epitope or group of epitopes.

How a fetus escapes maternal immune rejection is of great interest in those studying prematurity and also transplantation biology. Tissue rejection by T cell priming may occur in three main ways (36). First, host T cells can be presented with foreign MHC-peptide complexes by donor antigen presenting cells (APCs), known as direct recognition. Second, donor peptides can be processed and presented by host APCs to T cells, known as indirect recognition. Lastly, semi-direct recognition occurs when donor MHC-peptide complexes are captured by recipient APCs. Direct recognition is believed to be the main mechanism of acute rejection after solid organ transplantation (36); therefore, the inability for maternal T cells to directly interact with fetal/placental APCs may support survival of the latter. Transplant rejection can also occur indirectly through cross-primed CD8+ T cells as demonstrated by allo-skin grafts in SCID mice (37). Work using adoptive transfer of T cells into pregnant transgenic mice demonstrated that fetal recognition by maternal T cells occurs exclusively through indirect recognition (38). Organ rejection after transplantation demonstrated expansion of the TCR repertoire in blood and tissue of patients who developed T cell-mediated rejection compared to those who did not (39). T cell profiling has therefore been used to monitor patients following transplantation, who are at risk of graft rejection. This method may be useful in diagnosing VUE prior to delivery.

Erlebacher et al showed T cells that recognize fetal antigen have no cytotoxic function and are clonally deleted (38). Peripheral deletion is a (de)selection step outside the thymus that removes lymphocytes that either recognize self-antigen or are weakly primed (i.e. inhibitory co-stimulatory molecules, lack of inflammatory response) (40–42). During pregnancy, numerous mechanisms at the fetal-maternal interface contribute to immune tolerance towards fetal antigens (43). One could hypothesize that during VUE, maternal T cells are indirectly presented with fetal/placental peptides by maternal APCs. Alternatively, maternally-derived CD8+T cells migrating to the chorionic villi may be presented with fetal antigens by placental macrophages (Hofbauer cells). In both scenarios if there are pro-inflammatory signals present and an absence of negative regulatory receptors, an allograft response may be generated resulting in fetal rejection. Our data demonstrated low frequency of T cells in VUE and control placentae recognizing antigens specific to CMV, which was increased in patients with a diagnosed CMV infection during pregnancy. In a study examining cord blood of HIV-exposed uninfected infants, HIV-specific CD8+ T cells were identified in 86% (6/7) of the samples (44). Clonotype frequency was 4-fold higher in this study compared to the VUE results described here. Additionally, we were surprised that more clonotype sharing between our CMV cases was not observed; however, data from over 600 cases of CMV affected subjects reported that few CMV-specific clones were shared, indicating that T cell responses show high levels of specificity in each person (45, 46).

Perturbations in immunity have been described during VUE, which may regulate whether a T cell response is generated. For example, the placental and blood cell transcriptome of women with a VUE diagnosis showed a significant increase in expression of chemokines CXCL9, CXCL10, CXCL11 and CXCL13; this pattern was not observed in cases of acute chorioamnionitis (47). Soluble CXCR3, the receptor for CXCL9 and CXCL10, was found at high levels in amniotic fluid of patients with preterm labor and placental villitis (48). High levels of these chemokines may result in the trafficking of maternal T cells into placental villi. Others have profiled C4d, a component of the complement pathway associated with antibody-mediated rejection after organ transplantation (49). Cd4 staining was identified in over 80% of VUE cases, in syncytiotrophoblast cells, compared to control placentae (5%) (50). Higher levels of C4d deposition were also found in VUE placentae compared to placentae infected with CMV (51). In pregnancy, paternal human leukocyte antigen (HLA) is expected to be diverse compared to the maternal HLA-type, which may have important roles in VUE and TCR diversity. We have observed significant increases in HLA class I/II expression and HLA mismatch between mother and child in VUE compared to control placentae (52). In a cohort of term versus preterm birth subjects, HLA-reactive antibodies in maternal serum were identified at a higher rate in preterm birth, which corresponded with placental pathology (53). While no data exist on TCR diversity in pregnancy, transplant biologists aim to use spectratyping to predict graft rejection of allogeneic HLA (25–27). As VUE is often recurrent, one could hypothesize that maternal T cells which recognize paternal HLA in the first pregnancy may more effectively mount an anti-fetal response in a subsequent pregnancy due to the generation of memory T cells.

There are some limitations to interpreting our study results. First, as the paraffin blocks utilized contained small amounts of decidual tissue, some of the captured T cells, especially in control placentae, may have been decidual and not infiltrating T cells. Alternatively, fetal T cells may also be included in our analysis, especially in infectious villitis; however, these numbers in the villous capillary vessels of VUE placentae likely would be small as lymphocyte populations are predominately of maternal origin (54). Second, the current method cannot determine whether the TCR sequences are specific to helper, regulatory, or cytotoxic T cells. In the decidua, T cells make up between 10-20% of all the leukocytes; 30-45% are CD4+ T cells, 45-75% CD8+ T cells and 5% are regulatory T cells (reviewed in (55)). For VUE placentae, we assumed the majority of the cells profiled are CD8+ cytotoxic T cells, as the presence of CD4+ helper T cells in VUE is reported to be low (2). Lastly, γδ-T cells have also been described at low levels in the decidua, especially during early pregnancy (56). The data presented here did not include γδ-T cell receptors.

Despite these limitations, the current study does provide new details on the abundance and extent of the T cell compartment in both VUE and normal human placentae. The lack of shared TCR clones between sequenced VUE placentae provides further support for an alloimmune reaction; whether this reaction is maternal or fetal in origin is important to determine. Although VUE occurs commonly, the pathogenesis of this condition is unknown. Clear guidelines for management in initial and subsequent pregnancies to improve outcomes are lacking. Better characterization of the T cell repertoire and stimuli prompting potential migration into placental villi during VUE is fundamental not only to reproductive biologists, but may also provide insights into tissue rejection, important for the field of transplant biology. Further research in this field may not only support the development of specific immunomodulatory therapies during gestation to improve outcomes in affected fetuses, but could also have broad implications for bone marrow and solid organ transplantation, as well as immuno-oncology.

Key Points.

• Maternal T cells drive VUE responses; fetal immune cells drive infectious etiology.

• Clonal overlap between VUE subjects is minimal but common in placental infection.

• Contrary to placental infection, VUE T cells don’t recognize common viral epitopes.