A Genome Scale Transcriptional Regulatory Model of the Human Placenta

Alison G. Paquette, Kylia Ahuna, Yeon Mi Hwang, Jocelynn R. Pearl, Hanna Liao, Paul Shannon, Leena Kadam, Samantha Lapehn, Matthew Bucher, Ryan Roper, Cory C. Funk, James W. MacDonald, Theo K. Bammler, Priyanka Baloni, Heather Brockway, W. Alex Mason, Nicole R. Bush, Kaja Z. LeWinn, Catherine J. Karr, J Stamatoyannopoulos, Louis J. Muglia, Helen Jones, Yoel Sadovsky, Leslie Myatt, Sheela Sathyanarayana, Nathan D. PricebioRxiv 2022 · DOI

TL;DR

Researchers have built the first comprehensive, genome-scale transcriptional regulatory network (TRN) for the human placenta, integrating genomic footprinting and transcriptomic data. This model accurately predicts gene expression and identifies key regulators, offering new insights into placental function and potential targets for addressing preterm birth.

What Problem Does This Paper Address?

The placenta is a vital organ with complex functions, yet our understanding of its gene regulation is incomplete. Disruptions in placental gene regulation are linked to pregnancy pathologies like preterm birth and preeclampsia. While previous studies have focused on individual genes or smaller networks, a genome-wide model of how transcription factors (TFs) control gene expression in the placenta has been lacking. This paper aims to fill that gap by creating a detailed map of these regulatory interactions, which can then be used to understand placental development, function, and disease.

Methods

The authors constructed a transcriptional regulatory network (TRN) using a multi-step approach:

  1. Identifying Open Chromatin Regions (Footprints): They used DNase hypersensitivity sequencing (DNase-seq) data from placental tissue to pinpoint regions of the genome that are accessible to transcription factors.
  2. Curating TF Binding Motifs: They gathered known DNA sequences (motifs) that transcription factors bind to from public databases.
  3. Generating Putative Edges: By overlaying the identified open chromatin regions with TF binding motifs, they predicted potential interactions (edges) between transcription factors (TFs) and their target genes.
  4. Model Parameterization and Pruning: They used transcriptomic data (RNA sequencing) from placental samples to train and refine the model. This involved calculating correlations between TF expression and target gene expression, and then pruning the network to keep the most reliable interactions. They explored various model parameters to optimize accuracy.
  5. Validation: The model's accuracy was assessed using independent datasets and experimental validation (siRNA knockdowns of specific TFs).

Datasets

  • Training and Testing Data: Transcriptomic data from 475 human placental samples (RNA sequencing) and DNase hypersensitivity data from 12 term human placentas. These samples were collected from multiple cohorts, including the Magee Womens Research Institute (MWRI) and the CANDLE study.
  • Independent Validation Data: Not explicitly detailed as a separate dataset for validation, but the model's performance was evaluated on a holdout dataset and through experimental validation.
  • Public Databases: TF binding motif databases (HOCOMOCO, JASPAR, SwissRegulon) and gene annotation databases (Human Protein Atlas) were used.

Key Insights and Ideas

  • Placenta-Specific Regulation: The TRN highlights the existence of transcription factors that are either placenta-specific or placenta-enriched, suggesting specialized regulatory mechanisms in this organ.
  • Sex Differences in Regulation: The study identified a significant number of TF-target gene relationships that differ between male and female placentas, indicating that fetal sex plays a crucial role in placental gene regulation. This could have implications for differential responses to the in utero environment and later-life health outcomes.
  • Preterm Birth Regulators: The TRN successfully identified key TFs that are enriched as regulators for genes previously associated with preterm birth, providing potential targets for further investigation and intervention.
  • "Hub" TFs: The model revealed a small number of TFs with a disproportionately large number of target genes ("hub" TFs), suggesting they play master regulatory roles in the placenta.

Key Results

  • Comprehensive Network: The final model contains 113,158 interactions involving 391 TFs and 7,712 target genes.
  • Predictive Accuracy: The TRN model predicted target gene expression with an out-of-sample R² value greater than 0.25 for 74% of target genes in an independent dataset.
  • Experimental Validation: siRNA knockdowns of four TFs showed concordance with the TRN predictions, achieving an accuracy of >0.7 for three of the four TFs.
  • Sex-Specific Interactions: 5,361 TF-target gene relationships showed significant differences between male and female placentas.
  • Preterm Birth Association: Seven TFs were significantly enriched as regulators of genes previously associated with preterm birth.

Limitations and Future Work

  • Cellular Heterogeneity: The model is based on bulk placental tissue, which is a composite of different cell types. This cellular heterogeneity can create false positive relationships and dilute signal strengths. Future work could focus on cell-type specific TRNs.
  • TF Activity Proxy: TF expression is used as a proxy for TF activity, which doesn't account for post-transcriptional or post-translational modifications.
  • Temporal Resolution: The current model is based on placentas collected at the end of gestation. Future work could incorporate temporal resolution to understand how these regulatory networks change throughout pregnancy.
  • Pathological Conditions: The current model is primarily built on data from non-pathological pregnancies. Further research could explore TRNs in the context of specific placental pathologies.
  • Web Tool Availability: The authors have made the model publicly available as a searchable web tool (PlacentalTRN), facilitating further research and analysis by the scientific community.

Summary updated

Abstract

ABSTRACT Gene regulation is essential to placental function and fetal development. We report a genome-scale transcriptional regulatory network (TRN) of the human placenta built using digital genomic footprinting and transcriptomic data. We integrated 475 transcriptomes and 12 DNase hypersensitivity datasets from placental samples to globally and quantitatively map transcription factor (TF)-target gene interactions. In an independent dataset, the TRN model predicted target gene expression with an out of sample R 2 value greater than 0.25 for 74% of target genes. We performed siRNA knockdowns of 4 TFs and achieved concordance between the predicted gene targets in our TRN and differences in expression of knockdowns with an accuracy of >0.7 for 3 of the 4 TFs. Our final model contained 113,158 interactions across 391 TFs and 7,712 target genes and is publicly available. We identified six TFs which were significantly enriched as regulators for genes previously associated with preterm birth.
BibTeX
@misc{paquette2022genome,
  author = {Alison G. Paquette and Kylia Ahuna and Yeon Mi Hwang and Jocelynn R. Pearl and Hanna Liao and Paul Shannon and Leena Kadam and Samantha Lapehn and Matthew Bucher and Ryan Roper and Cory C. Funk and James W. MacDonald and Theo K. Bammler and Priyanka Baloni and Heather Brockway and W. Alex Mason and Nicole R. Bush and Kaja Z. LeWinn and Catherine J. Karr and J Stamatoyannopoulos and Louis J. Muglia and Helen Jones and Yoel Sadovsky and Leslie Myatt and Sheela Sathyanarayana and Nathan D. Price},
  title = {{A Genome Scale Transcriptional Regulatory Model of the Human Placenta}},
  year = {2022},
  doi = {10.1101/2022.09.27.509541},
  howpublished = {bioRxiv},
}
Figure