B16 - Simultaneous profiling of DNA methylation, DNA accessibility, and the transcriptome in single cells

Introduction: : Neural tube defects (NTDs) are defects that affect the brain, spine, or spinal cord and are one of the most common congenital malformations, affecting over 300,000 births annually worldwide, with significant implications for neonatal health [1]. These defects may result in paralysis or death, depending on the severity of the defect, and are corrected with surgery when possible. The neural tube is formed during the third and fourth weeks of pregnancy and first involves a thickened layer of ectoderm cells, the neural plate, bending and folding to form a groove and then fusing to create the neural tube. While the number of births with NTDs has decreased by about 30% since folate fortification of grain products has been implemented, the precise molecular mechanisms underlying these defects, and their formation in vivo, remain poorly understood [2]. Neural tube development occurs after implantation, beyond the ethical limits of human embryo studies, rendering direct measurement impossible. Animal models do not capture important human-specific genetic aspects of NTDs [3]. Additionally, folate plays a crucial role in DNA methylation; however, it remains unclear how epigenetic changes, such as DNA accessibility and DNA methylation (5mCpG), evolve with the emergence of different cell types during human development. To address this gap, we have designed a novel method that enables the simultaneous measurement of DNA methylation, DNA accessibility, and transcriptomic profiles within the same cell and implemented an in vitro model developed by the Streichan lab [4].

Materials and

Methods: : Unlike recently developed methods, this method does not require the separation of DNA and RNA before amplification, leading to higher-resolution sequencing [5]. To achieve this, mRNA is reverse transcribed with cell-specific barcodes, and simultaneously the methyltransferase M.CviPI is used to methylate cytosines in a GpC context within open chromatin (Fig. 1A). Later, MspJI is added, which recognizes methylated cytosines (CpG and GpC) in the genome and creates double-stranded DNA breaks that are ligated to adapters for downstream sequencing. This innovative approach is the first method to capture methylation information at individual CpG dyads on both strands, which is critical for highly dynamic systems such as development [6]. To accomplish this, the top strand methylation status is known through the MspJI cut site, and the bottom strand methylation status is interrogated through the enzymatic conversion of unmodified cytosines to uracil using an APOBEC deaminase. Illumina libraries for the mRNA and DNA are then prepared. DNA methylation is called using Bismark, and the reads are then assigned to the methylome (CpG) or DNA accessibility (GpC) for each cell [7]. The method represents a significant advancement in the field, as it allows for the comprehensive characterization of individual cells, which will reveal insights that were previously unattainable using traditional techniques. To understand how NTDs arise, we are implementing key gene knockouts relevant to neural tube development (ZIC2 and HES1) using CRISPR-Cas9.

Results, Conclusions, and Discussions:: This method has been validated by assessing DNA accessibility and methylation at DNaseI hypersensitivity sites (DHSs) in HEK 293T cells. DHSs are regions of DNA that are easily cleaved by the DNaseI enzyme and are a hallmark of active or regulatory regions of the genome. We investigated the averaged single-cell DNA accessibility and DNA methylation profiles at DHSs, split by previously reported signal strength. As expected, more accessible sites had larger peaks and less DNA methylation (Fig. 1B). We have also optimized the amount of GC methyltransferase used for the method.

Next, we applied this method to Karzbrun et al.’s model of neural tube morphogenesis [4]. In this system, cells are seeded on micropatterns, Matrigel is applied to generate a 3D pluripotent monolayer epithelium, and morphogenesis is driven by using a neural induction medium containing TGF-β inhibitor, followed by exposure to bone morphogenetic protein 4 (BMP4) [4]. This system results in a tube-shaped neural tissue covered with surface ectoderm (Fig. 2). We sorted cells 72 and 120 hours after BMP4 exposure. Single-cell transcriptome sequencing revealed 3 clusters: surface ectoderm, neural ectoderm, and neural crest cells (Fig. 3A-B). The surface ectoderm population expressed transcription factors associated with non-neural ectoderm and markers for epidermal differentiation. The neural ectoderm cells expressed genes related to forebrain markers. The neural crest cluster expressed genes associated with neural plate border, neural crest specification, migration, and diversification. When separating the neural crest population by the two time points, the 120-hour population showed greater upregulation of migration and diversification genes (Fig. 3C). We expect to see differences in DNA accessibility and methylation between the major cell types present in this system.

This study is the first to use a multi-omics approach to investigate the epigenetic landscape in human neural tube morphogenesis. The epigenome plays a crucial role in cell fate decisions, and the single-cell method provides a high-throughput technique to understand epigenetic regulation in complex systems. We plan to use this method to identify key epigenetic and transcriptional changes occurring with ZIC2 and HES1 gene knockouts. This work will illuminate the involvement of the epigenome in NTDs.

Acknowledgements and/or References (Optional)::

References:
[1] Flores AL, Vellozzi C, Valencia D, Sniezek J. Indian J Community Health. 2014 Nov;26(Suppl 1):3-5.
[2] Williams J, Mai CT, Mulinare J, et al. Morb Mortal Weekly Rep MMWR. 2015;64(01):1-5.
[3] Lee, S. & Gleeson, J. G. Trends Neurosci. 2020 43, 519–532.
[4] Karzbrun, E. et al. Nature 2021, 599, 268–272.
[5] Chialastri, A; Karzbrun, E.; Khankhel, A.H.; Radeke,M.J.; Streichan, S.J; Dey, S.S. bioRxiv 2022.
[6] Chialastri, A.; Sarkar, S.; Schauer, E.E.; Lamba, S.; Dey, S.S. Nat. Struct. Mol. Biol. 2024 31, 8, 1296-1308.
[7] Krueger F, Andrews SR. Bioinformatics. 2011 Jun 1;27(11):1571-2.