Integration of a Multi-omics Stem Cell Differentiation Dataset Using a Dynamical Model

Abstract

Stem cell differentiation is a highly dynamic process involving pervasive changes in gene expression. The large majority of existing studies has characterized differentiation at the level of individual molecular profiles, such as the transcriptome or the proteome. To obtain a more comprehensive view, we measured protein, mRNA and microRNA abundance during retinoic acid-driven differentiation of mouse embryonic stem cells. We found that mRNA and protein abundance are typically only weakly correlated across time. To understand this finding, we developed a hierarchical dynamical model that allowed us to integrate all data sets. This model was able to explain mRNA-protein discordance for most genes and identified instances of potential microRNA-mediated regulation. Overexpression or depletion of microRNAs identified by the model, followed by RNA sequencing and protein quantification, were used to follow up on the predictions of the model. Overall, our study shows how multi-omics integration by a dynamical model could be used to nominate candidate regulators.

Introduction

Much of the medical potential of pluripotent stem cells is due to their ability to differentiate into all cell types of the adult body [1]. While tremendous progress has been made in guiding cells through successive lineage decisions, the regulatory mechanisms underlying these decisions often remain unknown, especially at the post-transcriptional level. This gap in knowledge hampers the streamlining and acceleration of differentiation protocols.

Materials and method

Experimental methods

Cell culture.

E14 mouse embryonic stem cells were cultured as previously described [2]. Briefly, cells were grown in modified 2i medium [63]: DMEM/F12 (Life technologies) supplemented with 0.5x N2 supplement, 0.5x B27 supplement, 4mM L- glutamine (Gibco), 20 μg/ml human insulin (Sigma-Aldrich), 1x 100U/ml penicillin/streptomycin (Gibco), 1x MEM Non-Essential Amino Acids (Gibco), 7 μl 2-Mercaptoethanol (Sigma-Aldrich), 1 μM MEK inhibitor (PD0325901, Stemgent), 3 μM GSK3 inhibitor (CHIR99021, Stemgent), 1000 U/ml mouse LIF (ESGRO). Cells were passaged every other day with Accutase (Life technologies) and replated on gelatin coated tissue culture plates (Cellstar, Greiner bio-one). During transfections, cells were temporarily cultured in serum+LIF medium (10% ES certified FBS, 1X non-essential amino acids, 0.1mM β-mercaptoethanol, 1X pen/strep, 2mM L-glutamine, 10,000U/ml mLIF, mLIF from Merck, rest from Thermo Fisher Scientific). miR reporter cell line clone selection took place on homegrown mouse embryonic fibroblast feeders.

Results

Pervasive discordance between mRNA and protein in retinoic acid-driven mESC differentiation

We used RA differentiation of mESCs as a generic model for in vitro differentiation. Previously, we characterized this differentiation assay in detail at the transcriptional level by single-cell RNA-seq [2] and showed that RA exposure induces a bifurcation into extraembryonic endoderm-like and ectoderm-like cells. Here, we collected RNA and protein samples during an RA differentiation time course (Fig 1A). For each time point, we quantified poly(A) RNA by RNA-seq and protein expression by tandem mass tag (TMT) labeling followed by tandem mass spectrometry (MS/MS). In total, we obtained RNA and protein abundance estimates for 6271 genes (S1A–S1E Fig) at 8 time points in duplicate. After correction for batch effects due to separate sequencing runs (S1F Fig), we achieved highly similar results for the two biological replicates. 

Discussion

In this study we set out to integrate a multi-omics data set on stem cell differentiation. A range of tools for the integration of multiple omics modalities, both at the bulk and single-cell cell level [41,42], already exist. The most recent approaches have started to aim for biologically meaningful integration by incorporating prior knowledge in various ways. For example, biological knowledge can inform priors of Bayesian models or the topology of networks that represent interactions connecting different modalities [43,44]. A third way to exploit biological relationships between data sets, which we adopted in this study, is the use of dynamical models [17–22]. These models incorporate biophysical relationships between different types of molecules in a quantitative way and can also be used to infer kinetic parameters that have obvious biological interpretations17–22. We speculate that combinations of the mentioned approaches, exemplified by physics or systems-biology informed neural networks [45,46], will become powerful tools for data integration in the future.

Citation: van den Berg PR, Bérenger-Currias NMLP, Budnik B, Slavov N, Semrau S (2023) Integration of a multi-omics stem cell differentiation dataset using a dynamical model. PLoS Genet 19(5): e1010744. https://doi.org/10.1371/journal.pgen.1010744

Editor: Christian Schröter, Max-Planck-Institut fur molekulare Physiologie, GERMANY

Received: August 24, 2022; Accepted: April 14, 2023; Published: May 11, 2023

Copyright: © 2023 van den Berg et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The RNA-seq data has been deposited in GEO (ID: GSE93301). A list of the RNA-seq samples can be found in S3 Table. The raw MS data has been deposited in MassIVE (ID: MSV000080461). The processed data can also be mined and accessed through a web application at www.semraulab.com/multi-omics.

Funding: P. vd B. and S.S. were supported by the Netherlands Organisation for Scientific Research (NWO/OCW), as part of the Frontiers of Nanoscience (NanoFront) program. N.S. was supported by a New Innovator Award from the NIGMS of the NIH under Award number DP2GM123497. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.