Projects | Pascal GP Martin

Bioinformatics and Biostatistics

Sun, 30 Aug 2020 22:57:36 -0400

While developing transcriptomic approaches during my PhD, I quickly felt the need to learn more tools from biostatistics in order to perform thorough analyses of my data.

I quickly initiated a sustained collaboration with mathematicians from Toulouse Math Institute, in particular:

We first made an inventory of methods relevant to analyze transcriptomic data ( Baccini et al., 2005).
In order to integrate transcriptomic and lipidomic data acquired during my PhD, Igniacio Gonzalez developed the regularized version of Canonical Correlation Analysis ( Gonzalez et al., 2008).
This initial work was then taken in multiple directions by Kim-Anh Lê Cao who developed the first version of the mixomics package during her PhD (see for example Lê Cao et al., 2009) and still actively leads the development of this package and of data integration methods.

With Sébastien Déjean, we also developed methods to analyze transcriptomic data acquired during time-series experiments ( Déjean et al., 2010).

This experience in biostatistics was extensively used in my research projects. Notably I developed a pipeline for the analysis of microarray data that is still the foundation of the statistical analyses performed by the GeT-TRiX transcriptomic facility.

Since 2015, I have gained more experience in bioinformatics for the analysis of next-generation sequencing (NGS) data, in particular using R and Bioconductor.
I have developed my first R package named GeneNeighborhood. It allows to explore the orientation and proximity of the direct upstream and downstream neighbors of a predefined set of genes.
I have also developed the NanoBAC R package and a snakemake pipeline to assemble the reads obtained from BAC sequencing on the Oxford Nanopore long read sequencers.

Chromatin and Transcription

Sun, 30 Aug 2020 22:57:36 -0400

RNA polymerase II (RNAPII) is 12-subunit enzyme responsible for the production of all protein-coding RNA. It also transcribes a number of small RNAs (snRNA and miRNAs) and a majority of long non coding RNAs (lncRNA). Transcription by RNA polymerase II involves several steps and a number of proteins:

Transcription by RNAPII

Transcription occurs in the context of chromatin, which is defined by the association of DNA with histone proteins. DNA (~147bp) is wrapped around the nucleosome, an octamer that consists of two copies of each of the four core histone proteins (H2A, H2B, H3 and H4). Further chromatin compaction involves histone H1 that binds between nucleosomes.

From DNA to chromatin

" 0321_DNA_Macrostructure.jpg" By OpenStax CC BY 4.0, via Wikimedia Commons

Transcription factors, architectural proteins (e.g. insulators) and RNAPII interact with a number of proteins that are able to modify histones (e.g. HAT or histone acetyltransferases) and to move nucleosomes (e.g. ATP-dependent chromatin remodeling complexes).

I started to become more interested in transcription and chromatin organization in 2013-2015, when I joined the group of Olivier Cuvier at the CBI / LBME (CNRS / Univ. Toulouse), in order to develop my skills in genomics and bioinformatics.

During these 2 years, I participated in 2 published studies:

A study on Drosophila insulator binding proteins (IBPs) in which we found that ChIP-seq peaks representing indirect protein-DNA interactions could be used to decipher the long-range targets of IBPs and their interaction with promoter-proximal pausing of RNAPII ( Liang et al., Mol Cell, 2014)
A study in collaboration with the group of Monsef Benkirane showing that subunits of the Integrator complex interact with NELF and play a role genome-wide in NELF-mediated RNAPII pause/release ( Stadelmayer et al., Nat Commun, 2014).

During this period I also collaborated closely with the group of Jérôme Cavaillé to uncover the molecular mechanisms underlying the neonatal mortality in a murine model deficient for a cluster of miRNA ( Labialle et al., EMBO J, 2014).

In 2015-2018, I obtained an Agreenskills+ fellowship and joined the group of Scott Michaels. I worked in close collaboration with Xuhong Yu to unravel the function of a new family of negative transcription elongation factors on RNAPII transcription and chromatin looping. We named these proteins Border (BDR). The analysis of our NGS (Next-generation Sequencing) data allowed me to propose a mechanism by which the BDR proteins modulate gene expression by slowing down the progression of RNAPII at the 3' end of genes and thus limit transcriptional interferences in closely spaced tandem gene pairs. A first publication ( Yu*, Martin*, Michaels ; *co-first authors, Nat Comm, 2019) describes the role of these factors in limiting transcriptional interferences in the genome of Arabidopsis thaliana. I also developed the GeneNeighborhood R package to allow others to explore the orientation and proximity of the direct upstream and downstream neighbors of their genes of interest. In a second publication ( Yu*, Martin*, et al., Michaels ; *co-first authors, Curr Biol, 2021), we described the interaction of BDR with the FPA protein and their role in regulating flowering-time in Arabidopsis thaliana.

Endoreduplication in tomato fruit

Sun, 30 Aug 2020 22:57:36 -0400

Endoreduplication is a process by which cells replicate their DNA but do not divide. Successive rounds of this modified cell cycle, called the endocycle, generate polyploid cells with increasing numbers of copies of their genome.

Cell cycle and endocycle

CC BY 4.0

Endoreduplication is a phenomenon that occurs in a wide range of animal species including mammals, insects and other metazoans and in different tissues. It also occurs in fungi but it is most widespread in plants, occurring in up to 90% of Angiosperms.
Early during tomato fruit development, cell divisions are replaced by endoreduplication, giving rise to progressively larger cells with ploidy levels up to 512C, that contribute to the growth and structure of the tissue.

Endoreduplication in tomato fruit development

Picture by Ulysse Tuquoi ( CC BY 4.0)

Endoreduplication contributes to cell growth and cell differentiation pathways and is sometimes activated in response to environmental cues in plants. Much work has been devoted to identify the pathways that control the endocycle. However, little is known about the gene regulations that are associated with endoreduplication and how chromatin structure and epigenetic marks are affected in highly polyploid cells.

Since 2021, I joined the FDFE team that studies endoreduplication in tomato fruit.

Using a cell sorter, it is possible to separate the nuclei of different ploidy levels (2C, 4C, 8C, etc.) from the pericarp (the fleshy part of the fruit) of tomato fruits. We used this technique to analyze by RNA-seq the transcriptome of cells from different ploidy levels and at different developmental stages. This allowed us to show that ploidy in itself plays an important role in shaping the transcriptome of cells in the pericarp Tourdot et al., 2024.

Based on this observation, we have proposed to investigate the epigenome of endoreduplicated cells and to evaluate if and how it is involved in regulating gene expression changes observed in polyploid cells. We are developing ATAC-seq and CUT&RUN / CUT&TAG protocols to investigate chromatin in polyploid nuclei from tomato fruit.

This work is funded by projects from INRAE pertament BAP, from Université de Bordeaux and from the ANR (TomEndCODE project).

Food toxicology

Sun, 30 Aug 2020 22:57:36 -0400

In the context of the ToxAlim INRA research unit, I contributed to several research projects aiming at understanding the mechanisms and impacts of different food contaminants.

I managed with Thierry Pineau (2008-2009) and then H Guillou (2010-2011) the team that became in 2010 the group of Integrative Toxicology & Metabolism and then I developed and led the GeT-TRiX transcriptomic facility (2011-2015). During these years I supervised the work of 3 PhD students: Alexandre Eveillard (with T. Pineau), Béatrice Roques (with C. Viguié) and Nabila Moussaoui (with H. Houdeau) and contributed to significant projects, including:

Using combined transcriptomics and metabolomics approaches, the work of A. Eveillard evidenced new pathways targeted by DEHP, the main phthalate plasticizer that can leak from food containers and contaminate food products at their contact ( Eveillard et al., Toxicol Appl Pharmacol, 2009). In particular, we found that the nuclear receptor CAR was activated upon DEHP exposure ( Eveillard et al., Biochem Pharmacol, 2009).

DEHP activates CAR and PPARalpha

With Béatrice Roques, we studied the pathways by which fipronil, a very common phenylpyrazole insecticide, alters thyroid hormone homeostasis ( Roques et al., Tox Sci, 2012). We found that liver metabolic pathways involved in both xenobiotic and thyroid hormone metabolism were activated upon fipronil exposure and that the nuclear receptors CAR and PXR were involved in regulating the expression of the corresponding genes ( Roques et al., Biochem Pharmacol, 2013). We also generated preliminary evidences that this is associated with compensatory gene regulations in the thyroid gland itself ( Martin et al., Tox Appl Pharmacol, 2020

Fipronil and TH metabolism

Largely based on transcriptomic approaches, the work of A. Marmugi (advisor L Lakhal) allowed us to reveal non-monotonic effects of low doses of bisphenol A on liver gene expression and lipid storage ( Marmugi et al., Hepatology, 2013).

Effect of low BPA doses on the liver

With Nabila Moussaoui and Eric Houdeau, we studied how maternal separation and stress hormones alter intestinal permeability specifically in young rat pups ( Moussaoui et al., PLoS One, 2014).

These studies, as well as those listed below, provided important insights into the mechanisms and effects of several food contaminants, sometimes contributing directly to the assessement of the toxicity of these substances by regulatory agencies.

Nuclear receptors

Sun, 30 Aug 2020 22:57:36 -0400

Nuclear receptors (NR) are ligand-activated transcription factors that regulate numerous biological processes in metazoans. The human genome encodes 48 NRs (49 in mouse, 21 in Drosophila and >270 in Caenorabditis).

Nuclear receptors share a common structure and are organized in independent functional domains.

Nuclear Receptor Structure

The DNA binding domain (DBD) is the most conserved domain. The structure of the ligand binding domain (LBD), which defines a ligand binding pocket buried in a 3 layer “alpha-helical sandwich” is also well conserved among the nuclear receptors. The conformational changes induced upon ligand binding modifies the interactions with various proteins, including cofactors involved in chromatin remodeling, nucleosome modification and interactions with the basal transcription machinery.

Nuclear receptors regulate gene expression as monomers, homodimers or heterodimers, which contributes to define the type of DNA response element they can bind to in the promoters of their target genes. Traditionnally 4 classes of nuclear receptors have been defined:

Classes of Nuclear Receptors

Steroid hormone receptors (androgen receptor AR, estrogen receptors ER, glucocorticoid receptor GR, mineralocorticoid receptor MR and progesterone receptor PR) comprise the class I nuclear receptors. These receptors function as homodimers, are activated by low doses of steroid hormones, bind IR3-like response elements (inverted repeats separated by 3 nucleotides) and regulate numerous physiological processes, including reproduction, metabolism or immunity.
Class II is composed of the nuclear receptors that function as heterodimers with the RXR (Retinoïd X Receptors). They include the PPARs, LXRs, FXR, RAR, PXR, CAR, VDR and TR. They generally bind to DR-like response elements (direct repeats of the consensus motif AGGTCA separated by a variable number of nucleotides) but other types of response elements have also been described.
Class III members are the non-steroid hormone nuclear receptors that also function as homodimers such as RXRs, HNF4 or COUP-TF and typically bind DR-like motifs.
Nuclear receptors of Class IV function as monomers and bind a single consensus motif. This class includes NGFI_B, Nur1, SF-1, LRH-1 and RORs.

Note that some receptors might function as monomer or as dimers, such as rev-erb NRs.

During the first ~12 years of my career, I have mostly focused on better understanding the role played by the class II nuclear receptors PPARalpha annd RXR on the regulation of gene expression in the liver.

The liver plays central roles in the detoxification of xenobiotics, including drugs and toxicants, but also in energy metabolism, notably via its roles in lipid and lipoprotein metabolism. As such, it represents both a relevant drug target and a tissue which dysfunction can have broad, systemic consequences.

During my PhD, I studied the activation of the PPARalpha-RXR heterodimer through nutritional (dietary fatty acids; Martin et al., Hepatology, 2007), pharmacological (RXR and PPARalpha activating-drugs; Martin et al., Gene Expr, 2006) and physio-pathological cues (fasting; Déjean et al., 2010, Blavy et al., 2009) and the corresponding impacts on gene expression and lipid storage in the liver (hepatic steatosis). See my PhD manuscript for further details.

Later, I shared my experience in the analysis of the transcriptome by qPCR and microarrays by developping a transcriptomic facility in my research department: the GeT-TRiX facility. With the development of the ToxAlim INRA unit, I went on to address research questions in molecular food toxicology.