Abstract
Preterm birth (PTB) is leading contributor to infant death in the United States and globally, yet the underlying mechanistic causes are not well understood. Previous studies have suggested a role for advanced villous maturity in both spontaneous and iatrogenic preterm birth. To better understand pathological and molecular basis of idiopathic spontaneous preterm birth (isPTB), we compared placental morphology and transcriptomic analysis in carefully phenotyped cohorts of PTB due to intraamniotic infection, isPTB, and healthy term placentae. Characteristic features of precocious placental villous maturation were uniquely demonstrated in isPTB placentae. Transcriptomic analyses revealed isPTB candidate genes. These include an upregulation of three IGF binding proteins (IGFBP1, IGFBP2, and IGFBP6), supporting a role for IGF signaling in isPTB. Additional Gene Ontology analyses identified alterations in biological processes such as immunological activation and programmed cell death. Our data suggest that premature placental aging may contribute to the pathogenesis of isPTB and provide a molecular basis of this subset of cases of preterm birth.
Introduction
Every year, 1 million infants die from complications resulting from their birth before 37 completed weeks of gestation. In 2017, the incidence of prematurity in the United States was 9.6%1 and worldwide incidences reached approaching 20% in some regions2. While preterm birth is a multifactorial syndrome, there are two primary classifications: spontaneous preterm birth (sPTB) and iatrogenic preterm birth as a result of fetal or maternal complications. Although risk factors have been identified the underlying molecular mechanisms of truly idiopathic spontaneous preterm birth (isPTB) remain unclear3.
The role, if any, of the placenta in isPTB is not clearly defined and remains under investigated. The placenta is a transient two-sided organ, providing a maternal/fetal interface, and its proper development and function is essential to a successful pregnancy outcome4. Placentation is the result of a highly complex web of molecular mechanisms originating from both the mother and the fetus, many of which are not yet fully understood even under healthy, normal conditions. Recent advances in placental transcriptomics have identified changes in gene expression and regulatory mechanisms across normal gestation5–8. Yet, transcriptomics of isPTB are currently limited.
During the third trimester of pregnancy, specific hallmarks of placental maturity are observed including an increased number of terminal villi, syncytial nuclear aggregates, and vasculosyncytial membranes 9–11. Syncytial nuclear aggregates (SNA) are multi-layered aggregate of at least 10 syncytial nuclei extending out from the villous surface, that do not come in contact with other villi 9. Terminal villi were defined as branched villi <80μm in diameter. Vasculosyncytial membranes are defined as regions in the villi where the fetal capillaries are immediately adjacent to areas of the syncytiotrophoblast free from syncytial nuclei, thus are areas of direct diffusion 11. These hallmarks are utilized in histological assessments to determine the maturity of the placenta, with the term placentas possessing the highest amounts of these hallmarks9–11. Previous histological studies of the villous trophoblast in isPTB (<37 weeks GA) without intra-amniotic infection identified at least two distinct morphological phenotypes, with the significant majority of placental samples demonstrating advanced villous maturation (AVM). The AVM samples reflected all the hallmarks of a term placenta; the remaining samples had no hallmarks of AVM12,13. These data suggest there are multiple subclasses of isPTB: those with or without infection and those with or without AVM. Thus, identifying and assessing placental maturity through histological and transcriptomic studies is necessary to define aberrant molecular mechanisms underlying the placental pathophysiology associated with subtypes of isPTB. The objective of this study was to identify transcriptomic signatures of placental maturity in a histologically defined cohort of idiopathic spontaneous preterm births.
Results
Study Characteristics
Maternal and fetal characteristics for the three different pregnancy outcomes included in this study are presented in Table 1. Significant differences were observed in gestational age and fetal weights between intra-amniotic infection (IAI) and isPTB samples compared to the term samples (P<0.001). Among the isPTB samples, two of neonates were small for gestational age (SGA) with a fetal weight less than the 10th percentile. Within the IAI samples, only one neonate was SGA. All term births for which there was fetal weight available (n=9) were appropriate for gestational age.
Assessment of advanced maturity in isPTB villi
Stereological assessment identified no significant differences between the isPTB and term samples in number of syncytial nuclear aggregates (SNAs) or terminal villi per high powered field (villi<80µm in diameter) (Table 2; Figure 1A-C, F-G). One isPTB case could not be classified as there was no sample left to assess morphology. There was a trend towards no significance between the IAI and term samples in the numbers of terminal villi observed (P=0.078 Table 2) with no differences observed in the number of SNAs between IAI and term samples. Vasculosyncytial membrane counts were significantly reduced in IAI samples compared to isPTB and term samples (P=0.001 Table 2 and Figure 1). Perivillous fibrin deposits were observed in each of the sample types (Figure 1).
RNA Sequence results
All fastq files passed initial quality control assessment in FASTQC. A total of 1,246,073,145 unpaired reads were generated for the 31 samples and 761,182,139 reads (61.08%) were successfully aligned once to the human genome GRCh38 (Supplemental Tables S1 and S2). As there are multiple paralogous gene families expressed in the placenta, we opted not to include reads that mapped multiple times in our final analyses. Further quality control assessments were completed to examine sequence quality per sample (Supplemental Figure 2 A-C).
Identification of differentially expressed genes
Due to the origin of the samples and inclusion of a pre-existing dataset, we performed multiple quality control assessments, including principle component analyses (PCA) to identify potential batch effects (Supplemental Figure S1). We did not observe any significant batch effects and thus, did not remove or control for them in subsequent tests for differential gene expression. The IAI samples were entirely female in origin and had a small sample size; therefore, so we did not control for any covariate factors such as fetal sex as it could be potentially biased. We performed differential gene expression testing in three pairwise comparisons: IAI compared to term births, isPTB compared to IAI birth, and isPTB compared to term births. Genes were considered significant with an adjusted P-value of <0.1 and absolute log2 fold-change >1.5 (Figure 2A, all significant genes labeled red). We identified 160 significant genes in IAI verses term, with 117 upregulated and 43 downregulated. In the isPTB verses IAI comparison, we identified a total of 94 significant genes with 62 upregulated and 32 downregulated. Lastly, in the isPTB verses term comparison, we identified 158 significant genes with 157 upregulated and only 2 genes downregulated (Supplemental Tables S3,S4 and S5)
Categorization of maturation, gestation, and isPTB candidate genes
Differently expressed genes alone are not enough to identify transcriptomic signatures due to advanced maturation, gestational age, infection, or isPTB pathophysiology. Therefore, to identify candidate genes in each of these categories, we intersected the significant genes, both upregulated and downregulated, from each of the differential gene expression comparisons (Figure 2B). Genes categorized as infection (n=37) are represented in the intersection of IAI vs term and isPTB vs term. A previous study by Ackerman et al14 using a subset of these data profiled genes and pathways involved in intra-amniotic infection and we did not further explore these results. We did confirm that several of the genes they identified in their study (ACTA1, FUT9, MPO, S100A12) were present in this category in our results. Gestational age genes (total n= 123) are represented by the intersection of IAI vs term and isPTB vs term (n=11) and the genes exclusive to IAI vs term (n=112) (Figure 2B). Maturation genes (total n= 186) are represented by the intersection of isPTB vs IAI (n=18) and the genes exclusive to isPTB v IAI (n=39) and isPTB vs term (n=129) (Figure 2B).
To further refine our analysis, we compared the expression pattern of each candidate genes in each group across all three differential expression datasets (Figure 2C). To detect a maturation signal, we examined the 186 maturation candidate genes and identified those differentially expressed in the isPTB v IAI data with an absolute log2 fold change of >1.5. We then compared their expression across the other two differential expression datasets to identify genes that a similar pattern of expression in term v IAI, but were not differentially expressed in isPTB v term. 21 genes met these criteria and are represented in a heatmap (Figure 2; Table 3). Of these genes, 10 were upregulated and 11 were downregulated. We also were able to identify 13 of maturation genes which demonstrate differences in expression in isPTB vs IAI and term vs IAI similar to the maturation signal candidates; however, they are also differentially expressed in the isPTB vs term data (Figure 2; Table 3). Within this subset of maturation genes, 10 genes were upregulated and 4 were downregulated. Lastly, we were able to identify isPTB specific genes from the remaining maturation candidates by identifying genes in the isPTB vs term with an absolute log2 fold change of 1.5 and with a similar expression pattern in isPTB v IAI and the opposite expression or no difference in expression in term vs IAI (n=141) (Figure 3 and Table 4). Of these genes, only 2 were downregulated with the remaining 139 genes being upregulated.
We identified gestational age candidates using the same approach, first identifying genes in the IAI vs term data with an absolute log2 fold change of > 1.5 as this represents the greatest difference in gestational ages between our samples. We then compared the expression changes in isPTB v term and isPTB v IAI (Figure 2). While we did observe differences in expression in the isPTB vs IAI comparison, this is likely due to the differences in gestational ages of these samples (isPTB 29-36 wks v IAI 25-31wks). Using this approach, we identified 32 candidate genes with 29 of the genes upregulated and 3 downregulated (Figure 2; Supplemental Table S6).
Functional classification and enrichment analyses for candidate genes
We assessed enrichment in each of the candidate gene categories. There was no enrichment for Reactome pathways or GO terms in either of the maturity categories. However, there was enrichment in the gestational age candidates for cellular components related to the extracellular region (GO:0005576). These included 19 genes including receptors and ligands for the WNT signaling pathway, cell proliferation, and inflammatory response. The candidate genes associated with isPTB physiopathology had significant pathway enrichment including the IGF signaling pathway including three IGF binding proteins (Table 7). Furthermore, the isPTB candidate genes were enriched for GO terms associated with immunity, signaling, and regulation of blood flow (Table 5 and Supplemental Tables S7 and S8).
In addition to the enrichment analyses, we also performed functional classifications on all the candidate genes to identify additional pathways and functions (Supplemental Figures S3 and S4). The maturation candidate genes were divided into two subsets: those that showed no differences in expression (signal) between isPTB and term, and a subset that were expressed more in isPTB than term (drivers). While the candidate genes shared many biological processes, notable differences occur between the groups specifically within the metabolic processes (GO:0008152), immune system processes(GO:0002376), and locomotion (GO:0040011) (Supplemental Figure S3).
Discussion
The numerous subclassifications of preterm birth coupled with a limited understanding of the placental role in birth timing, have made identifying the etiological underpinnings of this devastating pregnancy outcome exceedingly difficult. We previously attempted to identify molecular signatures of spontaneous PTB using publicly available microarray data15 and identified several placental genes and pathways associated with spontaneous preterm birth. However, those analyses lacked complete covariate information, appropriate gestational age controls, and were on mixed array platforms. In this study, we overcame those limitations and combined transcriptome analyses with histological assessment of matched placental samples to assess maturity in order to elucidate the placental role in isPTB, and to identify a placental molecular signature associated with isPTB.
Morgan et al. hypothesized that placental maturity, as assessed by markers of villous maturation, and not infection may be the leading cause of both idiopathic and iatrogenic PTB12,13. However, only one other study has linked placental maturity to the molecular etiology of PTB. A recent study by Leavey et al. of placental maturity in pre-eclampsia (a common reason for iatrogenic preterm delivery) identified morphological and molecular similarities between PE placentas with advanced villous maturity (AVM) and normal term placentas16. Furthermore, they demonstrated that AVM placentas had a shift in their molecular signature and thus appeared older than their actual gestational age at delivery. These data along with Morgan et al.13 suggest a role for the placenta in PTB etiology, a role where AVM is potentially affecting placental output through increased terminal villi, more syncytiotrophoblast, and thus an increase in placental output in terms of secreted proteins and exosomes earlier in gestation. If placental output has a role in modulating birth timing, the AVM placentas regardless of whether they are idiopathic or iatrogenic may lead to PTB. These data also indicate that placental maturity and its associated molecular signatures may have a profound impact on how we utilize placental output to assess adverse pregnancy outcomes such as PTB and its subclassifications in the clinical setting.
In our current study, all isPTB placentas demonstrated changes in villous structure consistent with AVM along with peri-villous fibrin deposition with no significant differences observed compared to term births despite being delivered on average 5.8 weeks before term. This is consistent with previous histological observations12,13 which defined AVM as an increase in syncytial nuclear aggregates and terminal villi. Our assessment of vasculosyncytial membranes and fibrin deposition further strengthens the interpretation of advanced placental maturity in isPTB. In contrast, the IAI samples demonstrated structural hallmarks appropriate for their gestational maturity and age, as previous analysis of those samples concluded acute inflammation was the likely cause of their preterm delivery14. Given that control tissues from 20-36 weeks of gestation are not available due to the ethics of mid-late-gestation sampling or termination, placentas from IAI were the most appropriate controls that we could utilize for our transcriptomic analyses. Placentas in both the isPTB and IAI groups demonstrated perivillous fibrin deposition which may impact the amount placental surface area available for nutrient and oxygen exchange, leading to reduced fetal growth and SGA neonates. However, only three neonates were observed to be SGA, two were isPTB and one IAI; therefore, we do not believe the fibrin deposition is problematic or affecting placental function.
One of the primary obstacles in utilizing placental samples for transcriptomics has always been the lack of appropriate gestational aged (GA) control placentas. In most studies, GA matched placentas are not normal and thus are not suitable controls. As Ackerman et al14 have shown, the most parsimonious cause for the preterm birth in the IAI samples is infection, not adverse placental maturity or even placental insufficiency, making them the most appropriate controls. As such, we were able to then use them to conduct three pairwise comparisons to identify candidate genes that represented differences in maturity, gestation, and isPTB specific to their expression patterns.
The primary goal of this study was to determine if a molecular signature of maturity and isPTB could be identified from placentas demonstrating AVM. In our preliminary analyses, we classified significantly differentially expressed genes into gestational age, maturity, and isPTB candidates. Two maturity candidate genes, keratin 24 (KRT24) and shisa like 1 (SHISAL1) were significantly upregulated between the isPTB and term samples compared to IAI samples. Keratins are intermediate filament proteins expressed in epithelial cells that have a variety of roles in the cell including providing structural support to the cell and cellular mechanics17. While KRT24 is similar to other type I keratins, is found in epithelial cells, and is believed to be a terminal differentiation marker. Its increased expression induces senescence, autophagy, and apoptosis18. Given that KRT24 localizes to the villous trophoblasts19, the increase in terminal villi observed in isPTB likely accounts for increase expression. SHISAL1 function is currently unknown, but the shisa family of genes encode transmembrane proteins with roles in development as wingless and INT1 (WNT) and fibroblast growth factor (FGF) signaling antagonists20 and myocyte fusion21.
The two most downregulated candidates genes for maturity were cellular retinoic acid binding protein 1 (CRABP1) and lysosomal associated membranes protein family member 5 (LAMP5). CRABP1 is an epigenetically modulated tumor suppressor gene with a known role in retinoic acid signaling in the placenta22. Its downregulation in the isPTB and term samples could indicate altered retinoic acid transport and metabolism in fetal tissues22,23. LAMP5 is a recently identified member of the lysosomal associated membrane protein gene family24. Unlike the other broadly expressed family members, LAMP5 expression has only been observed in the brain, dendritic cells, and placental trophoblasts19. LAMP proteins are involved in autophagy and lysosomal formation and transport25. While we did not identify any specific enrichment for pathways in the maturity candidate gene list, functional analyses revealed WNT signaling and transforming growth factor beta (TGFβ) signaling as potential pathways of interest. We also examined the biological processes represented by the maturation genes although none were significantly enriched. Given that various bioprocesses these genes represent are active in placental trophoblasts, further investigations to refine their roles in placental development and maturation longitudinally are warranted. For the gestational age candidates, growth regulating estrogen receptor binding 1 (GREB1) and dickkof WNT signaling pathway inhibitor (DKK1) are upregulated in isPTB and IAI compared to term while carboxypeptidase X, M14 family member (CPXM2) is most downregulated gene. GREB1 appears to be localized to the fetal endothelial cells19 and is known to localize to maternal endometrial epithelial cells26. GREB1 function has not been studied in fetal villous physiology; however, in endometriosis26 and more recently in decidualization18. DKK1, a WNT signaling antagonist, was also identified as a highly upregulated gene in our array study; however, it was not significant. While previous studies have focused on its role in the decidua, we show that DKK1 localized to the syncytiotrophoblast in isPTB with a reduction in expression in term samples as expected from the expression data. It is known that WNT signaling, is essential to placental development acting through trophoblast proliferation and inhibition of apoptosis27,28. It has been theorized that aberrant DKK1 expression, especially upregulation in iatrogenic PTB samples, could be associated with etiology or pathophysiology28,29. However, these studies as well as our previous transcriptomics study lack appropriate gestational age controls. Given our data in this study, we posit that the difference in DKK1 expression between PTB samples is due to gestational age rather than pathology. Yet, as it does have such a strong expression pattern between gestational ages, it is worth further study to determine its precise role in normal placental maturation.
Within the isPTB candidate genes, we observed significant enrichment for genes within the IGF (insulin like growth factor) signaling pathway, specifically IGF binding proteins: IGFBP1, IGFBP2, IGFBP6 (Tables 4 and 5). IGFBPs bind to IGF1 and IGF2 modulating their bioavailability to activate the IGF signaling pathway30. IGFBP1 was thought to be primarily expressed in the decidua, but we and others31 have shown that it is present in the syncytiotrophoblast (Figure 4). IGFBP2 has been shown to be expressed in the placental villi via qPCR and western blotting31 but its specific localization is not known. IGFBP6 is expressed in the villous trophoblasts19. Upregulation of each of these genes indicates potential changes in the IGF signaling that could alter the development of the placenta. The IGF signaling pathway is associated with a plethora of biological processes essential to placental growth including trophoblast migration, nutrient sensing and transport, metabolism, and proliferation through the activation of the mitogen activated protein kinase (MAPK), extracellular signal regulated (ERK), and phosphoinositide 3-kinase/mammalian target of rapamycin (PI3K/mTOR) pathways which are downstream of the IGF1R32-34. Previous studies have demonstrated IGF1 and 2 regulate trophoblast physiology in the developing placenta; therefore, the aberrant upregulation of these specific IGFBPs could alter IGF signaling and growth of the placenta via trophoblast differentiation and metabolism33,34. While much of the study of IGF signaling has been focused on fetal growth, we only observed two cases of SGA in our isPTB samples, with the remaining cases at or above the 20th percentile for their gestational age, suggesting the alteration in IGF signaling modulating gestational length is not affecting fetal growth. Furthermore, it is important to note that both IGFBP2 and IGFBP6 have roles independent of the IGF signaling including integrin binding modulation through nuclear factor kappa light change enhancer of activated B cells (NFκB) signaling35, inhibition of angiogenesis36 and could also be modulating additional biological processes outside the IGF signaling pathway.
In our previous PTB transcriptomics study15, we identified IGFBP1 as an upregulated but non-significant gene of interest and PI3K signaling pathway as a significant network potentially predictive of PTB. These data along with the current findings support a role for IGF signaling in isPTB independent of aberrant fetal growth. IGFBP1 has been further linked to preterm birth as a marker of cervical ripening through analysis of vaginal fluids37,38. While these studies focused on the role of IGFBP1 in infection with premature rupture of membranes (PROM) or fetal growth syndromes suggesting a role for circulating decidual derived IGFBP1, the apparent upregulation of IGFBP1 in the syncytiotrophoblast as observed in this study (Figure 4) suggests origin of circulating IGFBPs are not limited to the decidua and their role in isPTB requires further investigation.
Further Gene Ontology (GO) analyses of the isPTB candidate genes revealed enrichment for cellular components including the T-cell receptor complex, immunological synapse and various membrane associated complexes. GO enrichment for biological processes included programmed cell death, T-cell activation, and cellular defense response among others. Taken together these data suggest an enrichment for an immunological trigger of apoptosis or autophagy; however, it is unclear if this is trophoblastic in origin or a result of alterations in the villous stroma. We did not identify any enrichment for GO terms or pathways in the maturity, we did identify cellular components of interest in the gestational age candidates including several members of the WTN signaling subpathways along with cell proliferation and inflammatory response signals all of which require further analyses for their role in birth timing. We also performed functional classifications for biological processes and pathways to determine differences and similarities between the different candidate classifications. Interestingly, we observed the candidates shared numerous biological pathways, but differed in three specific areas, locomotion, metabolic, and immune processes.
In summary, this is the first study to associate placental morphological phenotypes and genome-wide transcriptome signatures in isPTB. Our work has shown not only a role for the placenta in isPTB, but that accelerated placental maturity directly affects pregnancy outcomes. The similarities of the isPTB and term placentas on both the morphological and molecular levels suggests a precocious maturation phenotype in isPTB, further demonstrating the need to precisely identify the subclassifications of spontaneous preterm birth. Identifying the molecular signatures related to the pathophysiological differences and similarities between isPTB and term placentas will allow us to gain insight birth timing and potentially develop meaningful clinical therapeutic interventions.
Materials and Methods
Study Population
This study was approved by the Cincinnati Children’s Hospital Medical Center institutional review board (#IRB 2013-2243, 2015-8030, 2016-2033). De-identified term (n=9) and idiopathic spontaneous preterm (n=8) placental villous samples and patient information were obtained from the following sources: The Global Alliance to Prevent Prematurity and Stillbirth (GAPPS) in Seattle Washington USA, the Research Centre for Women’s and Infant’s Health (RCWIH) at Mt Sinai Hospital Toronto Canada, and the University of Cincinnati Medical Center (UCMC). Inclusion criteria included: maternal age 18 years or older, singleton pregnancies with either normal term delivery (38-42 weeks gestation) or preterm delivery (29-36 weeks gestation) without additional complications other than idiopathic spontaneous preterm. Utilizing an RNA sequencing power calculation from39, it was determined that 10-15 transcriptomes for this particular study. Thus, previously published RNA sequence based, placental villous transcriptomes (GEO GSE73714 term=5, isPTB=5, inter-amniotic infection=5) were also utilized14. Birth weight percentiles were estimated using the WHO weight percentiles calculator40 with an US average birth weight of 3400gm41.
Transcriptome Generation
All placental samples from GAPPS, RCWIH, and UCMC which were collected within 60 minutes of delivery and snap frozen prior to biobanking. Total RNA was prepared from placental villous samples thawed in RNAIce Later (Invitrogen) as per manufacturer instructions. Total RNA was isolated using the RNAeasy Mini Kit (Qiagen). 50-100 µg of total RNA was submitted to the University of Cincinnati Genomics, Epigenomics and Sequencing Core for RNA quality assessment and sequencing. Long RNA total libraries were generated using a RiboZero Kit (Illumina) and sequencing was run on a Illumina High Seq 2100 system to generate single end 50bp reads at a depth of 50 million reads. Details on the collection of placental samples and generation of transcriptomes from GSE73714 can be found here14.
RNA-sequence Analyses
To facilitate RNA sequence analyses, a local instance of the Galaxy42 was utilized with the following tools: FASTQC (Galaxy v0.71)43, TrimGalore! (Galaxy v0.4.3.1)44, Bowtie2 (Galaxy v2.3.4.1)45, and FeatureCounts (Galaxy v1.6.0.3)46. The quality of the raw fastq files was assessed with FASTQC and with adapters subsequently trimmed with TrimGalore. Trimmed sequences were then aligned to the University of California Santa Cruz (UCSC) human genome hg38 using Bowtie2. FeatureCounts was used to generate total read counts per gene and to generate a count matrix file to be used in differential gene expression analyses.
Differential Expression Analyses
After annotation, all non-coding transcripts were removed from the count matrix. Counts per gene were calculated and genes with less than 70 counts total across all samples were removed. The count data was then normalized using the counts per million (CPM) method to allow for various quality control analyses to ensure the data was ready for differential expression testing (Supplemental Figure S1). Differential expression tests were conducted in EdgeR (Emperical Analyses of Digital Gene Expression in R)47 on only protein coding genes. Within EdgeR, data were normalized using TMM (trimmed means of m values)48 to account for differences in library sizes. Comparisons for differential expression testing were as follows: IAI compared to term births, isPTB compared to term births, and isPTB compared to IAI births. Multiple corrections testing was performed using the Benjamini Hochberg method with a Q value of <0.05. Venny v2.049 was utilized to generate Venn diagrams and identify candidate genes for maturation, gestational age, and isPTB. Heatmaps were generated in Prism v7 (GraphPad).
Pathway and Gene ontology Analyses
Significant genes were divided into upregulated and downregulated categories then entered into the Panther Pathway DB50 for statistical overrepresentation analyses for Reactome Pathways and gene ontology (GO). Fisher’s Exact tests were used to determine significance and Bonferroni correction for multiple comparisons. Pathways were considered significant if they had an adjusted p-value <0.05 and enrichment score of >4.
Histology and Immunohistochemistry (IHC)
Serial sections were stained with Hematoxylin and Eosin (H&E) and assessed for placental maturity as described below. Immunohistochemistry was performed as previously published9. Briefly, all slides were incubated 95°C target retrieval solution for 30 minutes then washed in deionized water. Slides were then incubated in 3% hydrogen peroxide for 10 minutes followed by blocking in 10% normal goat serum +1% bovine albumin for 60 minutes. Primary antibodies generated in rabbit sera were diluted in phosphate buffered saline (PBS): DKK1 (1:20, GeneTex GTX40056) and IGFBP1 (1: 50, GeneTex GTX31149). Slides were incubated overnight at 4°C washed and incubated with biotinylated secondary antibody (anti-rabbit) for 60 minutes. Antibody binding was detected using DAB and slides were counterstained with hematoxylin. All slides were imaged on a Nikon Eclipse 80i microscope.
Clinical definitions
Gestational age was established based on last menstrual period confirmed by an ultrasonographic examination prior to 20 weeks 51. IAI was established based on analysis of the amniotic fluid retrieved in sterile conditions by trans-abdominal amniocentesis. Amniotic fluid infection was established by a positive Gram stain or a positive microbial culture result52. Preterm birth was defined as delivery of the neonate <37 weeks GA53. Idiopathic preterm birth was established absent IAI and histologic inflammation of the placenta and fetal membranes as assessed by a clinical pathologist54.
Morphological Assessment of Placental Maturity
Syncytial nuclear aggregates (SNA) were defined as a multi-layered aggregate of at least 10 syncytial nuclei extending out from the villous surface but not in contact with other villi 9. Terminal villi were defined as branched villi <80μm in diameter. Vasculosyncytial membranes are defined in11. Fibrin deposition was quantified by a score of 0-3 with where 0 was no fibrin observed and where 3 was the majority of the field containing fibrin. SNAs, terminal villi, and vasculo-syncytial membranes were counted manually and fibrin deposition was scored in 20 high powered fields (hpf) by two blinded reviewers.
Statistical Analyses
Data were analyzed in Prism7.0 (GraphPad). Data were evaluated for normality and non-parametric tests applied as appropriate. Non-parametric data are expressed as median and range and were analyzed by Kruskal-Wallis Test ANOVA with Dunn’s Multiple Comparisons. Categorical data were analyzed using Fisher’s Exact Test.
Author Contributions
HMB-Designed and preformed research, analyzed data, and wrote the paper HNJ-Assisted in research design and advised data analyses, edited paper WEA-provided RNA sequence data, advised on research design, edited paper IAB - provided RNA sequence data, advised on research design, edited paper CSB - provided RNA sequence data, advised on research design, edited paper SGK-provided preterm birth samples, edited paper LJM-Assisted in research design and advised data analyses, edited paper
Data Availability
Data will be made available in GEO as per funding requirements upon manuscript acceptance for publication.
Additional information
The authors declare they have no competing interests.
Acknowledgements
The authors would like to express their gratitude to the patients who donated their placentas for research and to Pietro Presicce, Paranthaman Senthamarai Kannan, and Manuel Alvarez (Kallapur lab), GAPPS, and RCWIH for assisting us in obtaining the placental samples. Additionally, the authors thank the staff at the UC Genomics Core for their help generating the bulk of the transcriptomics data for this project. This work was supported by grants to LM from the March of Dimes Prematurity Research Center Ohio Collaborative Funding 07/13-6/18, The Bill and Melinda Gates Foundation: Systems Biology Approaches to Birth Timing and Prematurity OPP1113966 Funding 11/14-10/17, and the Eunice Kennedy Shriver National Institute of Child and Health & Human Development of the National Institutes of Health Award Number R01HD091527.