Choose timezone
Your profile timezone:
Powered by Indico
v3.3-pre
It is now generally accepted that, different classes of regulatory elements generate specific chromatin profiles: Nucleosomes adjacent to promoters are marked by histone modification H3K27ac and H3K4me3, and the promoters themselves are located in accessible regions. Active enhancers also lie in accessible regions but the neighbouring nucleosomes are marked by H3K27ac and H3K4me1. However, identification of these elements based on the epigenetic characteristics is harder than expected. While promoters can be distinguished easily based on experimentally determined transcription start sites, identification of enhancers mainly relies on chromatin profiles.
Leveraging nascent RNA sequencing including CAGE-seq and HiS-NET-Seq, we notice that a proportion of enhancers reside overlapped or close to promoters ("proximal enhancers") and show different epigenetic characteristics than the traditional distal enhancers: Instead of H3K4me1, a higher level of H3K4me3 is usually observed around these enhancers. According to these patterns, we have built a pre-trained Bayesian Network (BN) model specifically to locate promoters, proximal enhancers, and distal enhancers. The BN model is trained on K562 cell line and the AUROC reaches 0.89 and 0.96 for promoters and distal enhancers respectively. Because of the distinct patterns in promoters and distal enhancers, our tool achieves high prediction accuracy. However, for generating the label information for proximal enhancers, an extra step is required using RNA-seq profiles to further distinguish proximal enhancers from promoters.
Whether you simply ask if your gene of interest is expressed in a cell, where the cell is located in 3D, and how to proof hypothesis indicated by sequencing and omics-data: Mic-Service can help you to visualize biological events across scales in space and time. As a central microscopy facility, we offer comprehensive support for a broad range of imaging techniques ranging from wide-field screening and high-resolution confocal imaging to ultrastructural analysis using electron microscopy. Moreover, we support your projects with image processing up to developing novel tools and quantitative pipelines to extract meaningful biological information from complex multidimensional imaging data. On this poster we summarize some of our current projects and ideas, encouraging you to discuss with us future directions in which we should further develop imaging at the MPIMG.
The presence of hair is one of the characterising traits of the mammalian clade, but this structure has also widely diversified across species: the development of spines in hedgehogs serves as an iconic example. Hairs and spines undergo comparable developmental steps, starting with the formation of placodes on the skin surface, which will in time invaginate into the dermis and create the root sheath. However, structural differences between these hair appendages are already present during embryogenesis. These remarkable adaptations are encoded in the genome, yet the precise genetic mechanisms at the base of spine development remain elusive. Here we combine synthetic, computational, and evolution biology to ask how the hedgehog Atelerix albiventris remodelled its hair into spines. Our comparative study focuses on multi-omic read-outs from developing hedgehog and mouse dorsal and ventral skin. We find that Krt1, a structural keratin showing spine-specific expression in postanal hedgehogs, already shows a dorsal expression bias during placode formation in hedgehogs. The dorsal specifying factor Zic4 also shows an expression bias that is specific to hedgehog spine placodes, reflected in the histone modification profile of the locus. Further analysis into cell type specific differences driving spine formation will indicate where these candidates are driving developmental changes. Phenotypic effects of these genomic regions will be validated by replicating hedgehog-specific regulatory landscapes in mouse. Ultimately, we aim to unveil the genetic origin of hedgehog spines, as an example of how animals evolve to repurpose existing structures into novel adaptations.
Long-read transcriptome sequencing (LRTS) offers a direct method for capturing full-length transcripts within a cellular system. Unlike next-generation sequencing (NGS), LRTS circumvents the issue of ambiguous alignment by eliminating the need for fragmentation during library preparation. This makes it particularly adept at accurately detecting isoforms, alleles, and transcripts from gene families, which are often challenging to resolve with short-read sequencing (SRS).
We developed IsoTools, a Python-based LRTS analysis framework. It offers a suite of innovative functions for exploring, analyzing, interpreting, and visualizing LRTS data. Our routes of research include:
In this poster we highlight current results with respect to the above points and discuss potential fields of collaboration.
In mammals, the long non-coding RNA Xist, master regulator of X-chromosome inactivation (XCI), is expressed from one out of two X chromosomes in female cells. The fact that one Xist allele is stably silenced, while the other one is expressed, indicates epigenetic memory. We have shown previously that RE57, a promoter-proximal regulatory DNA element essential for Xist upregulation, gains the repressive histone mark H3K9me3 and DNA methylation (5mC) at the onset of X inactivation only at the active X, where Xist is repressed. We hypothesize that RE57 establishes allelic epigenetic memory at Xist, through the recruitment of repressive chromatin marks.
To test RE57's effect on transcription, we integrated it into synthetic reporter loci. We observed reporter repression in naive (pre-XCI) and differentiated (XCI onset) mESCs, but 5mC recruitment only upon differentiation, suggesting a DNA-methylation-independent mode of repression in naïve cells. We are therefore investigating if RE57 can autonomously recruit H3K9me3 in the synthetic locus as a potential repressor mechanism. Additionally, we are testing whether H3K9me3 is recruited endogenously via antisense transcription of the long non-coding RNA Tsix over Xist.
In parallel, we have developed an inducible dCas9-based epigenetic editing tool for targeted demethylation of H3K9. Using this tool, we are studying whether H3K9me3 is responsible for Xist silencing at the onset of XCI. We will then study epigenetic memory of the Xist locus through transient epigenetic perturbations. In conclusion, the project combines synthetic biology and epigenetic editing to dissect epigenetic memory at the Xist locus.
Neurogenesis in the neocortex follows a precise spatiotemporal sequence. The transcriptional and translational dynamics that control this process remain inadequately understood. Integrating kinetic analysis of gene expression (translation, transcription, and degradation rates) will enhance our ability to distinguish between diverse cell types, offering deeper insights into its developmental mechanisms.
Transposable elements (TEs) are a major constituent of human genes, occupying approximately half of the intronic space. During pre-messenger RNA synthesis, intronic TEs are transcribed along with their host genes but rarely contribute to the final mRNA product because they are spliced out together with the intron and rapidly degraded. Paradoxically, TEs are an abundant source of RNA-processing signals through which they can create new introns, and also functional or non-functional chimeric transcripts. The rarity of these events implies the existence of a resilient splicing code that is able to suppress TE exonization without compromising host pre-mRNA processing. Here we show that SAFB proteins protect genome integrity by preventing retrotransposition of L1 elements while maintaining splicing integrity, via prevention of the exonization of previously integrated TEs. This unique dual role is possible because of L1’s conserved adenosine-rich coding sequences that are bound by SAFB proteins. The suppressive activity of SAFB extends to tissue-specific, giant protein-coding cassette exons, nested genes and Tigger DNA transposons. Moreover, SAFB also suppresses LTR/ERV elements in species in which they are still active, such as mice and flies. A significant subset of splicing events suppressed by SAFB in somatic cells are activated in the testis, coinciding with low SAFB expression in postmeiotic spermatids. Reminiscent of the division of labour between innate and adaptive immune systems that fight external pathogens, our results uncover SAFB proteins as an RNA-based, pattern-guided, non-adaptive defence system against TEs in the soma, complementing the RNA-based, adaptive Piwi-interacting RNA pathway of the germline.
Transcription elongation by RNA polymerase II (Pol II) has emerged as a regulatory hub in gene expression, with TFIIS being one of the oldest and most highly conserved transcription elongation factors. Canonically, TFIIS helps release Pol II from backtracking and prolonged pauses to enable productive RNA synthesis. However, recent evidence points towards a broader role for TFIIS proteins. In this study, we used a functional multi-omics approach in combination with super-resolution imaging to study the role of TFIIS in gene expression related to its non-canonical role as an interactor with RNA processing factors. Our results demonstrate that TFIIS acts as part of a recruitment hub for RNA processing factors associating with RNA PolII to drive proper splicing and transcriptional output. We create stable knockout cell lines to study the effective loss of TFIIS in its two main paralogs in human cells. Our data shows that loss of TFIIS leads to the upregulation of a large group of genes, both at the nascent RNA and stable mRNA levels. We compare this work to TFIIS mutational studies--which inactivate the domain responsible for rescuing backtracked RNA Pol II--and find drastically different transcriptional phenotypes between the mutant and the complete knockout.
How is time encoded in the products of gene expression in the brain? This critical question still needs to be answered despite the intense research on gene expression activity during neuronal development. We hypothesize that different transcription and translation kinetics guide the diversification and specification of neuronal lineages in the brain, within and between species: proteins being synthesized at a specific speed and time in a neuro progenitor will dictate its fate, its tissue position, morphology, connectivity, and function in the circuit. We focus on an evolutionarily recent brain region, the dorsal pallium, in two species with evolutionarily divergent pallium structures: mouse and chicken. Interestingly, neurogenesis in the mouse dorsal pallium structure starts concomitantly with the vascularization of the tissue. We hypothesize that blood vessel development and blood-brain barrier (BBB) maturation, providing the neuronal precursors with amino acids and nucleotides, impacts transcription and translation kinetics and might be a key component of gene expression. We will employ a whole tissue clearing method to visualize and quantify neocortical vascularization throughout neurogenic stages. Furthermore, we will gain an in-depth understanding of BBB development and its relationship with neuronal gene expression by analyzing endothelial cell (EC) metabolism and transport pathways using EC single-cell mRNA sequencing and proteomic studies. We hypothesize that the comparison between mouse and chicken brain vascularization might yield important brain evolution insights.
In the intricate realm of mammalian gene regulation, cis-regulatory elements play a crucial role in orchestrating precise gene expression during development. To exert their regulatory effect, they must come into physical contact with the promoters of their target genes, which is thought to be facilitated by loop extrusion. Here, ring-shaped cohesin complexes enhance the contacts within defined regions of the genome, which are called topologically associating domains (TADs). To explore the intricate relationship between the cis-regulatory landscape, TAD structure, and gene expression, we employ the X inactivation centre (Xic) in mouse embryonic stem cells as a model system. The Xic regulates X-chromosome inactivation (XCI) in female mammals, in which one of the two X chromosomes is randomly chosen and silenced in early embryonic development to compensate for the double dosage of X-linked genes. Previous research has shown that the bipartite TAD structures at the Xic undergoes rewiring throughout development, exhibiting distinct patterns on the active and silenced X chromosomes. By using state-of-the-art techniques (i.e. Tiled-C, Tiled-MCC) to map the chromatin architecture at up to base-pair resolution, we have revealed nano-scale structures with unprecedent detail. Together with novel ChIPmentation data, we are able to show increased cohesin binding at activated enhancers as well as increased CTCF binding at anchor site of rewiring. Interestingly, preliminary data show that CTCF binding at this site is required for proper upregulation of Xist at the onset of XCI. Together, our findings provide insights into the mechanisms of loop extrusion and its importance for gene regulation.
Predicting disease-specific gene sub-networks is a crucial step in network propagation, which operates on the principle that genes associated with a particular trait are likely to interact within molecular net-works. Many bioinformatics methods have been proposed in recent years to reduce the complexity of gene/protein networks. However, current module identification methods have limitations in character-izing these phenotypic changes. The ongoing project aims to develop a robust method for predicting modules from biological networks incorporating multi-omics data. This project builds on pre-existing resources such as the protein-protein interaction data collected and integrated by the ConsensusPathDB system and the network propagation tool NetCore, which employs a robustified random walk with re-start approach. Ongoing collaborations demonstrate applications of NetCore in drug response networks from proteomics and transcriptomics, machine learning interpretability, and infectious response time series, among others. To improve the module identification method of NetCore, we aim to apply node embedding.
X-Chromosome Inactivation (XCI) is a process that has evolved in female mammals to compensate for the different doses of X-chromosomal genes between the sexes. While this process is known to be driven by the expression of the lncRNA Xist from one of the two X chromosomes, it remains to be elucidated what causes Xist expression to occur specifically in females but not in males in the first place. A long-standing hypothesis suggest the presence of X-chromosomally encoded Xist activators (xXA), which are expressed in a double dosage in females compared to males and could thus overcome a threshold necessary for XCI initiation. Although Rnf12 has already been discovered as one of these xXAs, this factor alone cannot explain the female-specificity of Xist expression, as Rnf12 knockouts only delay XCI. In our group we have discovered Zic3 as a novel xXA, which shows a comparable effect on Xist as Rnf12. In this project we aim to investigate the mechanisms behind Xist regulation by Rnf12 and Zic3 and to determine whether the joint activity of these two factors can explain the female-specific upregulation of Xist in mammals.
Mesenchymal stromal cells (MSC) are fibroblast-like non-hematopoietic cells with self-renewal and differentiation capacity, and thereby great potential in regeneration and wound healing. Currently, the cultured MSCs from different tissues and species are used in pre-clinical and clinical studies such as disease and developmental modelling in- and ex vitro, as well as (xeno)transplantation. However, the MSC populations are heterogeneous not only inherently, but also among different model species. In addition, there is a lack information on their internal cell markers for proper characterization of MSC and their differentiated cells. Among the animal models, pigs are of particular importance in MSC research, due to their closeness to human anatomy and physiology. In this study, we are investigating the transcriptome of bone marrow-derived MSC and their differentiated cells in a breed of laboratory minipigs, known as Mini-LEWE, using bulk mRNA sequencing. By doing so, we aim at highlighting several major clusters of genes characterizing the MSCs’ phenotype.
Biomolecular condensates play a critical role in regulating nearly every aspect of cellular functions. Despite the growing understanding of their biological roles, the precise molecular constituents responsible for these functions remain largely unidentified. Current techniques face significant challenges in accurately isolating target condensates from cells and specifically assaying their compositions. To address this, we have adapted and refined the Fluorescence-Activated Non-membrane Condensates Isolation (FANCI) method to develop Nuclear Fluorescence-Activated Condensate Isolation (NuFANCI), which isolates nuclear condensates using the DAPI and FP signal and analyzes their protein, DNA, and RNA compositions. In our proof-of-concept experiments, we successfully isolated endogenous nucleoli averaging 5 µm in diameter, along with several fusion oncoprotein condensates ranging from 2 µm to 300 nm. The identities of these sorted condensates were verified by imaging known condensate markers. Subsequent mass spectrometry (MS) analysis revealed their protein compositions. Our next steps involve leveraging this technique to investigate critical questions, such as the role of chromocenters in genome stability throughout the cell cycle and the decision-making process between NHEJ and HR for DNA repair at 53BP1 DNA-damage loci. By offering a precise method for isolating and characterizing condensates, our work stands to significantly advance the field, potentially uncovering novel therapeutic targets and enhancing our understanding of cellular regulation at the molecular level.
Epigenetic 'readers' are essential for genome regulation, recognizing DNA and histone modifications to influence chromatin organization and gene expression. Dysfunction in these proteins is associated with diseases such as cancer. PHF13, an H3K4me3 reader, modulates chromatin processes and is aberrantly expressed in various cancers. This study investigates how PHF13's expression, chromatin affinity, and functions are regulated. We found that PHF13 oligomerizes via conserved regions in its N- and C-terminus, enhancing chromatin affinity. High expression of PHF13 leads to global chromatin compaction through these dimerizing regions. Unexpectedly, PHF13 also self-associates via intrinsically disordered regions, reducing its chromatin compaction capacity and affinity. Our findings suggest that a balance between PHF13's ordered and disordered regions regulates its chromatin association, potentially linking to its diverse biological functions.
Cell states and their resistance to environmental stimuli are fundamentally governed by chromatin dynamics. Treatment of embryonic stem cells (ESCs) with mTOR inhibitors, such as INK128, induces a dormancy state marked by elevated expression of pluripotency markers alkaline phosphatase 1 and SSEA1, resulting in a more homogeneous pluripotent population as determined by flow-assisted cell sorting. This transition to dormancy is accompanied by significant changes in the chromatin landscape, including DNA methylation, DNA accessibility, transcription factor networks, histone modifications, and transcriptional activity. Although the interplay between environment and cell state is complex and not yet fully understood, it is evident that epigenetics plays a crucial role in maintaining cellular identity and robustness. This study provides an in-depth examination of chromatin modifiers and their specific modifications that facilitate the shift to dormancy in ESCs. Furthermore, we reveal that this state transition induces a transient epigenetic memory, enhancing the cells’ ability to withstand subsequent environmental stresses. Post-transition environmental insults evoke a response reminiscent of the original cell state transition, rather than leading to an exit from pluripotency. Understanding these epigenetic mechanisms and harnessing the ability to induce such states may present new opportunities for regenerating cell identities in aging populations.
The human bromo- and extra-terminal domain (BET) protein family consists of ubiquitously expressed BRD2, BRD3, BRD4, and the testis-specific BRDT. Although the BET proteins have similar structural features such as two bromodomains and an extra-terminal protein-protein interaction domain, the individual roles and potential shared functions have remained unclear. Research by our team recently elaborated on the direct functions of BRD4 in RNA polymerase II (Pol II) transcription control. My project aims to elucidate the specific roles of BRD2 and BRD3 and uncover potential compensatory mechanisms using protein-specific degradation and proximity labelling in human cells. Preliminary findings suggest significant interactions between BRD2 and BRD4, with BRD2 increasing its binding to genomic regions upon BRD4 depletion, indicating compensatory behavior. Early interactome analyses reveal substantial overlap, highlighting the complexity and adaptability of BET protein interactions and underscoring the need for further investigation into their role in gene regulation and cellular processes.
Gene regulatory networks control the transcription of developmental genes. Each connection in the network is formed through intricate interactions between transcription factors (TF) and DNA binding sequences in cis-regulatory elements (RE). However, identifying several of these functional TF-RE interactions at once remains a challenge. Massively parallel reporter assays (MPRA) allow high-throughput screening of thousands of cis-regulatory elements but fall short in pinpointing the trans-acting factors responsible for their activation. Similarly, ChIP-seq and CUT&Tag techniques offer a broad view of TF binding across the genome but lack insight into the functionality of these bindings.
To break through these limitations, we will develop a new technique that combines CRISPR/Cas9-based activation and inhibition with a massively parallel reporter assay. Specifically, we will create a platform that co-expresses a RE-driven reporter and a sgRNA, which can up- or down-regulate a TF. Using amplicon-RNA-seq, we will measure the functional TF-RE interactions. This approach will enable us to screen trans-acting factors against cis-regulatory elements in a highly parallel manner, generating a map of positive interactors.
To optimize this technique we will conduct a comprehensive screen on a small subset of factors with a focus on testing different variations in CRISPR/Cas9 and MPRA systems as well as integration methods. We ultimately aim to apply this technique to study the functional TF-RE interactions at the X-inactivation centre, which regulates Xist.
Morphogenesis is orchestrated by precise spatio-temporal gene expression controlled by regulatory landscapes. It is still unknown to what extent the organisation of enhancers within a locus dictates their overall activity. How does the layout of the regulatory landscape impact gene expression during development?
To address this question, we rearranged the mouse Bmp2 landscape by altering the dosage and position of enhancers within the locus. Firstly, we changed the dosage of regulatory information through deletions or duplications across the locus. We found that the targeted duplication of a 2kb limb enhancer had significantly more effect on Bmp2 expression and limb morphology than large deletions. Secondly, we investigated the position-dependent behaviour of this enhancer duplication by repositioning one copy of this enhancer across the locus. Preliminary results suggest that the synergistic activity of the enhancer duplication depends on their linear proximity: the closer they are, the higher the effect on gene expression and in turn limb malformation.
Indeed, the tissue-specific overexpression of Bmp2 affects several morphological events in limb (notably digit and joint patterning). So finally, we took advantage of the varying levels of Bmp2 misexpression across the different mutants to dissect the role of Bmp signalling on limb patterning through time.
Developmental gene expression patterns result from the combined activities of multiple Cis-Regulatory Modules (CRMs) across regulatory landscapes. Different cooperation modes between CRMs have been reported, including additive, synergistic and repressive, mainly through reporter assays and combinatorial deletion experiments of few CRMs. However, to understand evolutionary traits and genetic diseases it is important to study all CRMs that act together to control individual genes. Hence, we study the morphogen Fgf8, which is regulated by at least 20 putative conserved CRMs embedded in Fgf8´s 170 kb regulatory landscape. Specifically, we focus on the action mode of CRMs controlling Fgf8´s transient expression in the Apical Ectodermal Ridge (AER) at the limb tip during embryonic limb development. To modify multiple CRMs simultaneously, we use a synthetic biology approach that allows us to re-write the locus with flexibility in sequence design. With this, we aim to (1) reconstruct the wildtype Fgf8 landscape at a safe harbor locus and (2) replace CRMs with orthologous sequences from other species, and study their consequences in gene expression and morphological phenotypes.
We started by establishing an experimental readout to carefully quantify the activity of transgenes in vivo. For this, a landing pad (CTCF-LP) which consists of an integrase recognition site flanked by CTCF binding sites was integrated at the safe-harbor locus (H11) in mouse Embryonic Stem Cell (mESC). First, this cell line was used to integrate individual enhancers driving a fluorescent reporter gene (mCherry) all in the exact same genomic context. After tetraploid aggregation, limbs of E10.5 mutant embryos carrying these enhancer integrations were subjected to high-resolution HCR RNA FISH and RT-qPCR. This revealed that the single enhancers drove slightly different transcriptional activities that corresponded only a fraction of the overall transcription driven by the full endogenous landscape.
We then reconstructed a 50 kb genomic region from the endogenous Fgf8 landscape (Fgf8-AER-SynLandscape) which encompasses enhancers driving AER specific Fgf8 expression and integrated it at the H11 locus. We integrated the Fgf8-AER-SynLandscape at the CTCF-LP and generated E10.5 embryos via tetraploid aggregation. HCR RNA-FISH showed that the synthetic 50kb region is able to drive expression at the AER, albeit at lower levels than what is observed at the endogenous locus. Unexpectedly, other predicted CRM activities (e.g. in the brain) were not detected by RNA-FISH, suggesting that other elements elsewhere in the landscape might be required to unleash its full pleiotropic activity.
Currently, we are preparing the integration of the full 170 kb mouse Fgf8 landscape at the CTCF-LP, and will compare the activity driven by this synthetic version with the endogenous Fgf8 expression pattern. Then, we will test mouse Fgf8 landscape variants in which CRMs are added/deleted/shuffled at different positions. In parallel, we are performing a deep genomics-based characterization of wild-type purified AER cells in order to obtain more information about the nature of AER-specific CRM activity, which will guide our choice of variants to build. In addition, we will re-constitute the full orthologous Fgf8 landscape from other species and measure activity changes across the whole mouse embryo. We expect that building and testing custom and inter-species landscape variants will shed light on how CRM activities are integrated to generate morphogen expression patterns in vivo.
X-chromosome inactivation (XCI) is the process by which female mammals inactivate one of their two X chromosomes to compensate the dosage imbalance of X-linked genes. The long non-coding RNA (lncRNA) Xist is the master regulator of XCI and it is upregulated from one of the two X chromosomes promoting the silencing of the whole chromosome in cis. Even though a wide range of Xist regulators has already been identified, it remains unclear how the symmetry between the two X chromosomes is broken to adopt two opposite states within the same cell. To address this question, I aim to quantify the dynamics of Xist upregulation using live cell imaging techniques (in a collaboration) to describe the process with high temporal resolution. To shed light on the possible mechanisms for symmetry-breaking between the two X chromosomes, I am developing a cell sorting system with allelic specificity which will allow the discrimination between the active and inactive X chromosome. Due to the non-coding nature of Xist, it is not feasible to directly tag it with a fluorescent protein. Therefore, I will tag each allele of an X-linked protein-coding gene, which is highly expressed and fast silenced upon Xist upregulation (i.e., Rnf12) with different fluorescent proteins as a proxy for Xist expression. Using this system, I will determine the molecular differences between the Xist-expressing and Xist-silent alleles at the Xist locus at the onset of XCI, such as the presence or absence of DNA-binding proteins, distinctive histone marks, DNA methylation patterns or 3D genome interactions. Finally, from the previous results, I will investigate candidate mechanisms for symmetry-breaking through quantification and perturbation.
Transposable elements (TEs) are mobile viral elements present in virtually all eukaryotic genomes. TE expansion poses a threat to genome integrity and different host defence mechanisms have been described, mainly focusing on prevention of TE insertion. However, until recently, it was unclear how pre-mRNA splicing is robustly accomplished while repressing exonisation of intronic TEs transcribed as part of a host gene transcript. Our lab has recently shown how proteins of the SAFB family bind to intronic LINE-1 TE sequences through a motif-driven, evolutionarily conserved mechanisms, preventing the activation of LINE-1-encoded splice sites. Notably, we found that SAFB proteins also bind to and suppress splicing of a newly described class of giant cassette exons (GCEs) in most somatic tissues. Alternatively spliced GCEs are rare in the human genome and their tissue specific expression despite the presence of SAFB is necessary for normal organ function. While our findings propose a new pathway to resolve host-TE conflicts, it also raises further questions about the fine-tuning of the SAFB pathway in relation to GCE evolution and function and what molecular mechanisms facilitate repression of harmful TE splicing while allowing GCE expression in tissue-specific contexts.
To address this, we have generated a stable HEK293 cell line for rapid inducible SAFB protein degradation. In the SAFB-depleted state, GCE splicing is de-repressed, leading to inclusion in the mature transcript. Such splicing events are expected to be mediated by a yet unknown trans-acting factor. To identify genes involved in GCE splicing, we have designed a high-throughput CRISPRi reverse genetic screen in which knock-down of the target factor combined with induced SAFB depletion would restore the WT splicing phenotype. The resulting transcript isoforms will be detected by RNA-FISH combined with flow cytometry (Flow-FISH). Transduction with a lentivirus library of sgRNAs against candidate genes will allow us to investigate the function of individual candidate genes as well as potential redundancies. Furthermore, screening will be performed in three iterations using isoform-specific probe sets against three tissue-specific GCE genes, namely KIF1B (brain), MAP4 (striated muscle), and CLIP1 (testis) to distinguish between common and tissue-specific regulators of GCE splicing.
In parallel, we interrogated GCE usage in tissue-/cell type-specific contexts. We used a Metazoan Developmental Alternative Splicing database (MeDAS; Zhang et al., 2020) for the expression profile of each exon in different tissues across developmental stages in humans and found 42 alternatively spliced-in GCEs with a tissue-specific usage pattern in brain and heart tissues. In addition, we observed dynamic usage of GCEs, such as ones in NIN and UNC13B through brain development, implying that there are molecular factors defining the splicing outcome. Findings from computational analyses will further inform the design of the CRISPRi screen and shine light on the evolution of the gene regulatory network through alternative splicing to decipher the molecular mechanism of tissue-specific GCE expression.
The aim of our study is to provide a comprehensive and comparative snapshot of the proto-epigenomic landscape in the three major subtypes of Renal Cell Carcinoma (RCC). Here we perform epigenomic and proteomic analyses on 40 patients spanning clear cell (ccRCC), papillary (pRCC) and chromophobe (chRCC) renal cell carcinoma. Our simultaneous profiling of the DNA methylation and hydroxymethylation landscape (oxBS-seq) in tumor and adjacent normal tissue demonstrated a global loss of hydroxymethylation in all three subtypes in tumor vs. normal. Furthermore, we observed that the loss of 5hmC is correlated with slight gain of methylation in the same loci, particularly in regions bearing the H3K36me3 histone modification. Targeted LC-MS/MS profiling of H3 histone modifications revealed a disruption in H3K36me1/2/3 levels, while whole proteome analysis of our patient cohort revealed significant downregulation of several factors that jointly contribute to global loss of 5hmC, such as IDH1/2, FH and SDH. In addition, we observed disruptions in the expression of numerous epigenetic factors, contributing to upregulation of cancer-related pathways. DNA hydroxymethylation has long been viewed as a nonfunctional intermediate, despite its relatively high enrichment in gene bodies which suggests otherwise. Our study points to the biological implications of the loss of DNA hydroxymethylation in kidney cancer and sheds light on its relationship with regulation of gene expression via histone modification.
Noncoding RNAs (ncRNAs) are a large family of RNAs that are not coding for known proteins. About 17 categories of non-coding RNA molecules have been identified so far. Among them, long ncRNAs (lncRNAs) and microRNAs (miRNAs) are the most studied classes of ncRNAs potentially involved in pathological conditions (Ratti et al. 2020).
While most of the efforts in the area have been focused on understanding the consequence of variants on the coding areas of the human genome, there is still a massive ground to cover when it comes to noncoding areas of the genome.
Genomic sequence comparisons between distant species have been extensively used to identify genes and determine their intron-exon boundaries, as well as to identify regulatory elements present in the large noncoding fraction of the genome (Loots et al. 2000; Pennacchio et al. 2001; Pennacchio and Rubin 2001). Cross-species sequence comparison is a powerful approach to identify functional genomic elements, but its sensitivity decreases with increasing phylogenetic distance. We belive that that sequence comparisons of numerous primate species should be sufficient to identify important regions of conservation that encode functional elements, and could have enough information for predicting secondary structures without the noise of more evolutionary separated species.
We use RNAz (Gruber et al, 2010) to predict the secondary structures of the identified lncRNAs. RNAz uses a comparative genomics approach to identify conserved RNA structures and predict their thermodynamic stability. The predicted secondary structures will help us to identify conserved functional domains of lncRNAs and determine the impact of mutations on these domains.
The objective of this project is to understanding the potential pathogenic effects of the mutations on lncRNA by using the predicted secondary structures.
You may think that mass spectrometry, flow cytometry and genomics generate loads of data. But have you ever thought about the pixel? A modern microscope typically records 4 to 12 million individual data points per image, the pixels, and this up to 50 times per second resolved in space and time. An average confocal microscope is capable to count the photons within 120x120 nm. In image analysis and in our brains, these individual measuring points are usually assembled together in more highly structured groups. We then say the droplet, cell, tissue or simply the object. However, in our poster we show you what happens when we leave the subjective decision-driven path and use simple mathematics to draw conclusions and extract findings from the matrix that are available to you as tools. Using pixels, voxels and whatever it is called in higher dimensions we aim towards a more autonomous, objectfied data extraction and interpretation from any imaging-based approach. Enabling you to draw assumptions that are not driven by any monday-friday-, silverback, thesis-, labmeeting, reviewer-request bias.
Identifying differentially methylated regions (DMRs) among samples or multiple groups in an unsupervised manner remains challenging due to the computational complexity. Here, we present metilene3, a fast and sensitive approach to call DMRs without prior information needed and to construct differentially methylated trees (DMTrees) for sample clustering and linkage discovery. Metilene3 detects DMRs by segmenting the genome based on pairwise methylation difference signals, and iteratively groups samples based on the DMR patterns with maximal weight sums. We demonstrate that metilene3 can identify DMRs and infer DMTrees both in human healthy tissue as well as cancer datasets and it is an efficient method to analyze methylation data and discover new biological knowledge.
Paediatric leukaemia remains the most prevalent cancer among children and teens, with acute lymphocytic leukaemia (ALL) accounting for approximately 80% of cases and exhibiting a higher prevalence in males. Among B-ALL subtypes, the TCF3(E2A)-PBX1 fusion t(1;19) is the third most common, present in about 3-5% of B-ALL cases. This translocation results in aberrant protein products with oncogenic potential that affects cell differentiation. Despite advancements in therapies that have improved prognosis and reduced mortality rates, 10-20% of these patients relapse, often with fatal outcomes.
Our interest in this study is to investigate molecular mechanisms associated with relapse TCF3-PBX1 positive paediatric B-ALL, using genomic and transcriptomic data from 36 patients across disease progression stages. The mutation landscape was characterised by frequent alterations in genes associated with RTK-RAS pathways. Integrating multi-omics analyses, we focused on a better understanding of the tumour microenvironment dynamics and niche remodelling. We identified key regulatory factors involved in inflammation and immune escape. We observed downregulation of MHC-class II gene expression and its regulator CIITA in relapse samples, suggesting an immune evasion mechanisms. Further, immune profiling indicated the presence of tumour-associated neutrophils in patients prone to relapse. We observed upregulation of inflammatory signalling, cell adhesion processes, and ATP-related pathways at relapse, suggesting their role in disease progression. Understanding these mechanisms will be crucial for developing targeted therapies to improve treatment outcomes and identifying biomarkers of relapse in these patients.
Xist, the master regulator of X-chromosome inactivation (XCI) in mammals, is upregulated in the epiblast during the formative phase of pluripotency, specifically in females. In the past, its regulation has been tightly linked to repression during naive pluripotency. However the identity of activators at the onset of XCI remains unknown. To fill this gap, we perform a comprehensive CRISPR screen targeting transcription factors during the early differentiation of female mouse embryonic stem cells. We identify a large set of activators, which is transiently expressed during the formative pluripotent stage. Subsequently, we use a series of CRISPR screens targeting individual reporter constructs in order to functionally connect trans-regulators to the cis-regulatory landscape of Xist. We detect a group of factors, including the X-linked TF ZIC3, which activate promoter-proximal elements in a X-dosage sensitive manner. Furthermore, we find a group of factors, including the master regulator of the epiblast OTX2, which interact with distal enhancer elements to control transcript levels following basal Xist activation. With this study, we provide a systems level view of the trans- and cis-regulatory network that links Xist activation to formative pluripotency and ensures female-specificity.
Throughout embryogenesis, Polycomb repressive complex 2 (PRC2) silences transcription via its histone-modifying activity to establish and maintain distinct gene expression that drives cellular plasticity and identity. However, the spatial and temporal dynamics of PRC2 during development and its molecular mechanisms that control proper embryogenesis remain largely unknown. This is especially crucial as aberrant expression of PRC2 has been frequently reported in various diseases, including cancers.
We implemented a mouse embryonic stem cell-based model, termed trunk-like structure (TLS), to recapitulate multiple developmental events of gastrulating embryos, including somitogenesis and neural tube formation. By combining a protein degradation strategy that allows acute depletion of the PRC2 complex in a time-course manner, we showed early onset of PRC2 depletion led to failed symmetry breaking, primitive streak formation, and axial elongation. This suggested that PRC2 activity is instrumental in cellular responses to WNT signaling. Intriguingly, transient restoration of PRC2 during WNT activation could partially rescue these defective phenotypes.
Finally, we are applying low-input Cut&Tag coupled with RNA sequencing to understand the causality between perturbed epigenetic marks and developmental defects. Ultimately, we aim to unveil PRC2-mediated regulation of molecular and cellular processes during development.
During tumorigenesis, normal cells undergo dramatic changes at multiple levels, from morphological to genetic and epigenetic ones. Often, multiple genetic alterations accumulate that can be classified based on their impact on tumor fitness into driver and passenger events. Patterns of somatic mutations are characteristic for different cancer types, e.g. the ratio of point mutations to copy-number alterations and the affected driver genes differ from one type to another. In addition, passenger point mutations accumulate non-uniformly and correlate with epigenetics of the cell of origin. This information can be used to predict the organ, or more specifically the cell type, from which a tumor originated in order to adjust therapeutic strategies for patients. Here, we predict the cell of origin for cancer samples belonging to different cancer types. To do this we utilize mapping from somatic mutations modality to gene expression at the single-cell level.
The development of the micropeptide ‘killswitch’ as a tool to study the intrinsic function of biomolecular condensates on a physiological level has made great progress so far. It makes use of the killswitch solidifying biomolecular condensates, and can be used in a targeted manner. However, the molecular mechanism on how the killswitch acts is not fully understood. The killswitch is composed of a unique 17 amino acids sequence, that harbors hydrophobic residues and a cluster of three aromatic phenylalanines. We could show that the presence of the phenylalanines is crucial and a similarly strong solidification cannot be reproduced by the usage of other aromatic residues. Plus, a clustering of the phenylalanines itself is not necessary for solidification, but enhances the effect. A shuffling of the phenylalanines over the remaining sequence resulted in various phenotypes, highlighting the importance of their position in the sequence. Lastly, in vitro droplet assays revealed a correlation between the capacity to phase separate and the solidifying effect of shuffled variants of the killswitch. These results add up to the standing hypothesis, that the killswitch acts by self-oligomerization mediated by clustered phenylalanines, thereby inhibiting the dynamics inside the condensates.
The maturation of lineage-committed embryonic hepatocytes requires timed activation of gene regulatory networks (GRNs) while silencing embryonic programs to achieve adult hepatic metabolic functions. Our knowledge of the key transcription regulators governing GRN rewiring during late development remains incomplete, hindering the derivation of functional hepatocytes in vitro. To address this, we employed a dCas9 activation screening in combination with single-cell transcriptomics on primary mouse embryonic hepatocytes, allowing for effect ranking among developmentally late-onset transcription regulators. We identify Nr1i3 as a potent inducer of pericentrally expressed metabolic function-related genes and Nfix as a critical suppressor of embryonic signatures. Supplementation of liver zonation patterning signals together with these transcription regulators enhanced the expression of pericentrally zonated Cyp genes, emphasizing the importance of a microenvironment-targeted approach for inducing maturation. Our screening and analysis highlight key regulatory mechanisms underlying organ maturation and offer general insights for improving the functionality of in vitro-derived cells.
Transcriptional regulation, governed by interactions between transcription factors (TFs) and DNA, is crucial for cellular function. We have previously shown that TF activity is sub-maximal and that we can enhance TFs transactivation by changing a sequence feature related to phase separation dynamics at the expense of binding specificity, thus indicating that TF encode a trade-off between activity and specificity termed TF suboptimization. In the current project, we aim to study whether this trade-off is advantageous for the organism fitness, focusing on embryonic development. Our hypothesis is that TFs’ suboptimality ensure ‘right’ developmental trajectories. To test this, we optimized transcription factor T (bra) and we aim to generate embryos using knock-in cell lines. Given T role in guiding mesodermal differentiation of NeuralMesodermal progenitors, we expect that an optimized T will shift the balance toward mesodermal outcome at the expense of a neural one in the developing embryos and drive abnormal development, giving us the basis for the physiological relevance of TF’s suboptimality.
The COVID-19 pandemic provided critical insights into basic immunology, validating longstanding concepts with large-scale observations. Thereby, the rapid development and deployment of effective vaccines, achieved through the synergy of basic and applied research, significantly reduced global morbidity and mortality. However, there was a notable disillusionment as an anticipated sterilizing immunity did not manifest, and COVID-19 vaccination still led to re-infections with often intense symptoms. Also, natural primary upper respiratory tract (URT) SARS-CoV-2 infections did not result in asymptomatic re-infections after a few months. This contrasts sharply with systemic infections or vaccinations, such as those for tetanus or diphtheria, where immunization typically results in symptom-free infection courses for years or even decades.
These observations raise two critical questions: First, how does the topography of a primary infection, i.e., mucosal versus systemic, imprint adaptive peripheral immune memory, given that no long-term local mucosal immunity is established? Second, how does this imprinted immunity evolve when an initial URT infection is followed by repeated parenteral vaccinations against the same disease, and vice versa? Can an initial imprint be overwritten, supplemented, or set anew?
To explore the impact of different exposure sites and antigen nature on human B cell memory complexity and plasticity, I will study the phenotypic diversity, transcriptome profiles, epigenetic landscape, and adaptability of antigen-specific human B cell memory in a context-specific manner. In a preliminary investigation, we isolated SARS-CoV-2 spike RBD- and S2-specific B cells from eight vaccinated individuals with no prior natural SARS-CoV-2 exposure and performed single-cell transcriptome analysis to evaluate their population complexity. We identified naïve and several memory B cell subsets, with distinct reactivation profiles, all recognizing the SARS-CoV-2 spike glycoprotein. One donor experienced a SARS-CoV-2 infection during the study and exhibited an additional activated memory B cell subset with a mucosal homing profile within seven days of symptom onset. These findings warrant further investigations to understand the originating subset(s) of these reactivated memory B cells, as well as their plasticity, dissemination, and protective potential.
Splicing and transcription complexes seem to directly communicate to coordinate their outputs, even though the mechanisms underlying this cross-talk remain unclear.
Two models for the recognition of the splicing unit were proposed: intron- and exon-definition. In the first, the spliceosome recognizes the intron as the splicing unit and places the basal splicing machinery across it. This is probably the dominant mode of splicing in lower eukaryotes such as yeast, which have short introns.
Throughout evolution, though, exons are being separated further by the immensely increasing intron size, due to a burst of transposable elements’ invasion that accompanied the water-to-land transition. Thus, in higher eukaryotes such as vertebrates, the splicing machinery had to adapt its mode of recognition to identify the short exons among the long introns (exon-definition). Meiotic recombination removed part of this excessive transposon content by excising large chunks of DNA, partly reducing intron size but increasing GC content as a by-product. One interesting middle point in the trend of intron expansion/shrinking during evolution is given by the lungfish, that conquered land ∼420 million years ago: its genome size is the longest ever sequenced (43Gb) with the longest intron being 5.8Mb (median intron ~10kb long, compared to ~2kb in humans).
In humans, both long and short introns are present, so it seems that both models (exon- and intron-definition) should co-exist. One possible solution would be that the two models co-exist in different sub-compartments of the nucleus. Indeed, it was recently proposed that genes with long and genes with short introns occupy different sub-nuclear regions which are related to the two splicing unit recognition modes.
Short and GC-rich genes, which usually occupy the A1 chromatin active compartment and have a high RNAPII occupancy, tend to reside in the proximity of Nuclear Speckles (NS). NS are dynamic structures composed of various RNA binding proteins (RBPs) and RNA molecules, while being devoid of DNA.
We found that, upon NS perturbation, the gene expression defects are predominantly post-transcriptional, with exons skipping events on exons that are surrounded by short and GC-rich introns being the most prominent. What is particularly remarkable about these events is that the splice junctions utilized remain canonical; however, they are used in a more chaotic higher-entropy fashion, increasing the complexity of the splicing outcome. During the water-to-land transition, concomitantly with transposable elements invasion of vertebrates’ introns and the consequent increase in GC-content, the core proteins of NS strikingly gained disordered regions, suggesting that NS evolved during this period likely to mediate the orderly matching of consecutive splice-sites in genes with short and GC-rich introns.
During this transition, not only did introns length and NS proteins structure expand, but interestingly also the C-terminal domain (CTD) of RNAPII. The CTD is a low-complexity domain, which is essential for RNAPII regulation during transcription. In yeast it is composed of 26 repeats of 7 amino acids, in humans 52 repeats, while in lungfish it reaches 72 repeats, correlating with the median intron length of these organisms.
In this project yeast, human, and lungfish RNAPII were cloned into cell lines where NS can be dissolved, and the occupancy signal of RNAPII on chromatin as well as the splicing outcome can be studied to gain a better understanding of the importance of the CTD repeats number in the coordination of long as well as short introns splicing.
Static gene expression programs have been extensively characterized in stem cells and mature human cells. However, the dynamics of RNA isoform changes upon cell-state-transitions during cell differentiation, the determinants and functional consequences have largely remained unclear. Here, we established an improved model for human neurogenesis in vitro that is amenable for systems-wide analyses of gene expression. Our multi-omics analysis reveals that the pronounced alterations in cell morphology correlate strongly with widespread changes in RNA isoform expression. Our approach identifies thousands of new RNA isoforms that are expressed at distinct differentiation stages. RNA isoforms mainly arise from exon skipping and the alternative usage of transcription start and polyadenylation sites during human neurogenesis. The transcript isoform changes can remodel the identity and functions of protein isoforms. Finally, our study identifies a set of RNA binding proteins as a potential determinant of differentiation stage-specific global isoform changes.
Cells whose accessibility landscape has been profiled with scATAC-seq cannot readily be annotated to a particular cell type. In fact, annotating cell-types in scATAC-seq data is a challenging task since, unlike in scRNA-seq data, we lack knowledge of "marker regions" which could be used for cell-type annotation. Current annotation methods typically translate accessibility to expression space and rely on gene expression patterns. We propose a novel approach, scATAcat, that leverages characterized bulk ATAC-seq data as prototypes to annotate scATAC-seq data. To mitigate the inherent sparsity of single-cell data, we aggregate cells that belong to the same cluster and create pseudobulk. To demonstrate the feasibility of our approach we collected a number of datasets with respective annotations to quantify the results and evaluate performance for scATAcat. scATAcat is available as a python package at https://github.com/aybugealtay/scATAcat
Liver disease is the only major disease where mortality is rising every year. The Covid-19 pandemic also had a major impact with a 21% death increase from liver disease in 2021 compared to 2019 in the UK. Thus, liver diseases are global health care challenge. The main reason of this situation lies in the absence of curative option for end stage liver disease with the only treatment being organ transplantation. Only a small fraction faction of patients can benefit from this option due to shortage of suitable organs whereas lifelong immunosuppression is necessary and carries associated risks.
To tackle this unmet clinical situation, alternative therapies are urgently needed, and one promising strategy is cell-based therapy using hepatocytes. While there have been some encouraging results with this approach, the limited availability and quality of primary hepatocytes represent a major drawback. Thus, a new and unlimited source of cells is required to make cell therapy in the context of liver disease a clinical reality. Human induced pluripotent stem cell (hiPSCs) derived hepatocytes could provide an advantageous option. However current protocols are time and resource consuming while failing to produce fully functional cells.
Here, we address this challenge by leveraging Forward Programming (FoP), a differentiation strategy, not solely relying on growth factors but, also on the overexpression of cell specific transcription factors (TFs) through the Opti-OX system. By identifying a suitable TFs combination, multiple iterations of protocol optimizations as well as adaptation to GMP compatible reagents in a 3D bioreactor setting, we created a scalable 20-day process with a high yield of ~10 FoP-Hepatocytes per hiPSC. These FoP-Hepatocytes are created at high purity as well as high quality. Indeed, the resulting cells exhibit functional characteristics like albumin, A1AT and clotting factor secretion, Urea and CYP metabolism similar to primary hepatocytes. Finally, Fop-Hepatocytes have been successfully transplanted in mice models for liver failure.
Taken together, these results demonstrate that large scale production of hepatocytes in vitro is feasible and that cells generated by forward programming can be used to counter liver disease symptoms thereby paving the way for future clinical developments in human.
Stem cell-based embryo models provide a means to study the inconspicuous processes of embryonic development at cellular and molecular levels. Various models of early mouse development have been developed over the last years, such as gastruloids and trunk like structures (TLS), that recapitulate symmetry breaking, axial elongation, along with transcriptional and structural features of the developing mouse embryo up to embryonic day E8.5. However, these models at the moment fail in representing the full range of tissues that arise during pre-organogenesis stages. Particularly the anterior neuroectodermal tissues fail to form due to ubiquitous exposure to WNT. Building on our earlier findings of the role of hypoxia in enhancing tissue representation in this developmental window, we now developed a transgene-free 3D embryo model that contains both posterior and anterior tissues. Anterior tissues include brain regions present during early nervous system development such as hindbrain, midbrain and occasionally forebrain, tissues that most of the current mouse embryo models lack. These neural tissues show appropriate anterior-posterior patterning and express cell type-specific markers. Time-resolved RNA-sequencing data identified signaling pathways that contribute to this patterning, in a similar way to their in vivo counterpart. We further show that hypoxia, but not HIF1a, is critical for anterior development. In conclusion, we were able to generate a head-to tail- in vitro model of mouse embryogenesis, which can be used as a platform for studying early mouse development along the anterior-posterior axis in a dish.
Over-expression of GATA6 in pluripotent cells initiates their reprogramming to extraembryonic endoderm (XEN) lineage, which results in a globally distinct and unique DNA methylation pattern that shares similarities with various tumors. Understanding the role of transcriptional regulators and the underlying mechanisms of epigenetic reprogramming can be crucial for elucidating potential oncogenic pathways. Our study takes a broader look at all transcription factors within the GATA family and how they interact with the genome. By specifically comparing early binding dynamics, genomic targets and regulatory consequences of these factors, we seek to uncover key principles that may inform about their normal and possibly pathological regulation.
In animal breeding often one allele is preferred over the other. However, heterozygous animals usually transmit both alleles to an equal proportion of their offspring. Our goal is to achieve preferential inheritance of desired genetic traits.
The t-haplotype, a selfish mouse chromosome is able to promote its own transmission by disabling sperm not carrying it. Using its central component, the Sperm motility kinase in a transgenic approach, we are able to reduce the number of offspring carrying unwanted properties.
Sex ratio distortion is a challenging but particularly interesting application of this principle: Making sperm carrying the preferred sex chromosome more successful in fertilization leads to an - often advantageous - prevalence of one sex in the next generation.
In contrast to alternative genetic approaches towards this goal, the resulting animals having the desired trait are non-transgenic and our method preserves full fertility of the breeding stock.
Understanding organogenesis is essential to reveal the mechanisms resulting into developmental disorders and to develop regenerative therapies. However, early organogenesis remains relatively unexplored especially in human due to ethical and technical reasons. Here, we aim to address this limitation in the context of the Hepato-Pancreato-Biliary system. More precisely, we aim to identify the signaling pathways and transcriptional networks which establish the pancreatic, hepatic and biliary lineages using human induced Pluripotent Stem Cell (hiPSCs) in vitro. We will first generate a human organoid system modeling the induction of the Hepato-Pancreato-Biliary buds with the aim of recapitulating the process leading to the establishment of these three lineages. This differentiation will be analysed using single cell (sc) RNA-sequencing during the formation of the bud organoids to identify master regulators controlling the cell fate choice between the pancreatic and liver. Cell-cell interactions will be also be uncovered using CellPhone DB to identify interplays that could be relevant in the context of pancreas development. Finally, essential factors that will be functionally validated by CRISPR/Cas9 knock out in hiPSCs. In summary, this project will deliver a novel model system to study pancreatic development which could pave the way of producing fully functional cells in vitro for clinical applications.
Male mice heterozygous for a complete t-haplotype preferentially transmit the t-chromosome to their progeny, a phenomenon called transmission ratio distortion (TRD). The t-haplotype is a variant form of mouse chromosome 17. Interactions of various t-haplotype factors affect sperm motility during spermatogenesis and lead to the selective advantage of t-sperm resulting in high transmission rates. Several distorter genes act additively in trans on all cells and impair sperm flagellar function. Only t-sperm carrying the t-haplotype responder gene Smok (sperm motility kinase) TCR regain a normal motility to reach the site of fertilization. The Smok gene family are thought to be components of a signal cascade that regulates sperm flagellar movement by reversible protein phosphorylation. The target protein of SMOK within the axoneme is currently unknown.
In this study we are establishing the SMOK-interaction network within the flagellum of sperm cells by using the BioID – a proximity dependent labeling approach in proteomics study – and the Trimolecular Fluorescence Complementation (TriFC) method. We are focusing on SMOK- interacting proteins that play a significant role within the TRD signaling pathway.
We are interested in the function of PPP2r5e, a regulatory subunit of the PP2A serine/threonine phosphatase and we are analyzing the function of AmmecR1 and of its paralogue AmmecR1-like and the role of Dnali1 – encoding a dynein axonemal light intermediate chain 1 (DNALI1) protein. Dnali1 is predominantly expressed within the axoneme of spermatids and in tissues that exhibit cilia supporting its function as component of the axonemal dynein – a motor protein complex that generates the force for the movement of cilia and flagella. The generation of knockout mouse lines by using the Cre-loxP and CRISPR/Cas9 system, and immunofluorescent stainings will help us to localize and to identify the function of each SMOK-interaction candidate and its contribution to sperm flagellar movement, fertility and to the TRD signaling pathway.
Mechanisms driving Liver regeneration is well-documented in cases of acute liver injury while the regenerative processes occurring in chronic diseases remain controversial. Indeed, animal studies have proposed three potential models: i) hepatocytes and cholangiocytes may revert to a liver cell progenitor state, restoring the corresponding cell compartment; ii) cholangiocytes may transdifferentiate into hepatocytes, and vice versa; iii) hepatic stem cells can be activated, leading to the generation of fresh liver cells. Nevertheless, these mechanisms have not been verified in humans, primarily due to technical and ethical constraints.
To overcome these challenges, we mapped at the single cell level the progression of metabolic dysfunction-associated steatotic liver disease (MASLD) and looked for potential regenerative occurrences. Our investigations revealed that hepatocytes and cholangiocytes can transdifferentiate into one another during chronic injury. This process involves a set of transcription factors, potentially crucial for acquiring the plasticity necessary for such cellular transdifferentiation. To test this hypothesis, we validated the upregulation of these plasticity factors in transdifferentiating cholangiocytes/hepatocytes in vivo by immunostaining of human tissue slides. Additionally, we performed in vitro gain-of-function experiments by overexpressing candidate factors in intrahepatic cholangiocyte organoids obtained from patients with end-stage liver disease and hepatocyte-like cells derived from human induced pluripotent stem cells. Phenotypic assessments confirmed the involvement of these factors in intrahepatic cholangiocyte organoids and induced pluripotent stem cell-generated hepatocyte plasticity.
Collectively, our findings partially elucidate the molecular mechanisms governing regeneration during chronic liver disease, thereby paving the way for the development of novel therapies aimed at promoting tissue repair in cases of chronic injury.
Gene expression is controlled by a multilayered system involving transcription factors and cis-regulatory elements, that together coordinate RNA polymerase II for efficient transcription. Therefore, a deep understanding of gene expression requires detailed knowledge of the location of RNA polymerase II and its regulators. Recent technological advances led to systems-wide methods, such as native elongating transcript sequencing (NET-seq), providing RNA polymerase occupancy profiles with single-nucleotide precision. To allow full use of the high-resolution occupancy data that is obtained by these new methods necessitates corresponding advanced computational tools. In particular, peak callers are required for robust identification of genomic positions of local signal enrichment. Peak calling is particularly challenging in sparse data, a shared feature among high-resolution genome-wide technologies, as current algorithms often do not account for the excess of zeros. State-of-the-art methods for peak calling in sparse data involve non-parametric approaches or ad-hoc workflows, which suffer from long processing times and lack of generalisation, respectively. To overcome these limitations, we are developing a software package for peak calling in sparse high-resolution occupancy data. Our package aims at 1) reducing computation time by a novel parametric strategy, 2) providing a consistent software framework and 3) expanding the user space by delivering an intuitive interface. To validate our approach, we have applied the software to three of the main nascent RNA profiling techniques: PRO-seq, mNET-seq and HiS-NET-seq – all of which are capable of characterising the location of RNA polymerase II at single-nucleotide resolution. Our findings show that computation time can be significantly reduced while increasing the amount of detected peaks in all the data modalities considered. Overall, our results advocate for the viability of an efficient and sensitive peak caller applicable to multiple types of high-resolution data.
Imaging Mass Cytometry (IMC) technology is instrumental for spatial biology investigations, enabling the simultaneous quantification of multiple proteins within different tissue structures and cell compartments. To serve our cancer and autoimmunity-related projects, we have developed MICCRA, a modular end-to-end pipeline dedicated to the automated analysis of IMC data. In comparison to recently published image analysis pipelines, e.g. SIMPLI, MCMICRO, imcRtools, and SPEX, allowing for image batch-processing but not dedicated to IMC data, MICCRA introduced novel modules relevant to IMC data analysis, including automated stitching and normalisation of regions of interest (ROIs), implementation of a distance-based denoising method and performance optimization to FlowSOM clustering. Our work focuses on quantitative data analysis evaluation, which is a challenging and often overlooked aspect in IMC. Devising metrics assessing performance quality of each analysis step, we compared the MICCRA approach to other pipelines. We could show for example that MICCRA’s cell segmentation approach performs well, even for dense tissues, a challenging feature seen with tumor and immune infiltration. In summary, MICCRA is an attractive tool optimizing the automatic analysis of quantitative IMC data in challenging biological contexts.
The rapid advancements in the field of transcriptomics have enabled the study of heterogenous cell populations at a pooled and single-cell level. However, these studies may be affected by several factors, like technology used, laboratory conditions, personnel, and capture times. These factors, better known as batch effects, may induce a qualitative behavior across conditions, that are unrelated to the variables in the study. It adversely affects the interpretation and integration of data and therefore, makes it crucial to eliminate such effects. In the recent past, several approaches for batch correction have been introduced, but majority of them can adjust the data only for the known batch effects. In order to deal with cases, where no prior information about the inherent artifacts is available, a surrogate variable analysis (SVA) - based approach for batch correction has been introduced. The poster highlights the potential of SVA, as a tool for extracting the inherent batch effects in the data. In order to improve the efficiency and scalability of the approach, a metacell based batch correction technique for single-cell RNA-Seq has also been introduced.
Stem cell-based embryo models provide a means to study the inconspicuous processes of embryonic development at cellular and molecular levels. However, most of these models at the moment fail to represent the full range of tissues that arise during pre-organogenesis stages, particularly the anterior neuroectodermal that forms the head. Building on our earlier findings of the role of hypoxia in enhancing tissue representation in this developmental window, we now developed a transgene-free 3D embryo model that contains both posterior and anterior tissues. Anterior tissues include brain regions present during early nervous system development such as hindbrain, midbrain and occasionally forebrain. To investigate the impact of hypoxia on cell fate decisions during development, we performed a combination of bulk and single-cell RNAseq experiments. The analysis revealed hypoxia-dependent differential expression of germ layer activating gene modules and showed the temporal timeframe of the neuroectoderm and the necessity of assembling different aggregates for midbrain formation.
Bats (order Chiroptera) are the only mammals capable of self-powered flight. Unlike in human and mouse, digits II-V are not separated in the forelimb (FL) in bats but connected by the patagia, a type of elastic membrane between the digits forming the wing. The molecular underpinnings of these morphological changes, probably one of the greatest adaptive changes in mammalian evolution, remain obscure. Here, we explore the molecular basis of bat wing formation through a suite of genomic tools and single-cell analyses comparing limb development of FL and HL in mouse and bat. We demonstrate that apoptosis, a process generally thought to be involved in the separation of digits, does occur in the developing bat wing. Our single-cell analyses confirmed the conservation of apoptosis-related gene expression in interdigital cells and revealed in general a striking conservation of cell types between species. To trace down the cellular origin of the FL patagium (chiropatagium), we performed scRNA-seq and label transferring analyses of micro-dissected cells. We find that the chiropatatium cells are primarily composed of fibroblast cells that develop independently from the apoptosis-related RA-signaling cells of the interdigital mesenchyme. Epigenomic and transcriptomic analyses revealed the distal activation of of a gene program characteristic of proximal limb cells with the transcription factors MEIS2 and TBX3 as its most prominent regulators. Transgenic expression of Meis2 and Tbx3 in interdigital cells in mice induced cell proliferation and matrix production and the activation of genes consistent with processes associated with bat wing development. Our results elucidate basic molecular mechanisms of wing formation in bats and illustrate the repurposing of existing developmental programs as an evolutionary molecular mechanism to generate morphological novelties.
DNA methylation profiling of cell-free DNA (cfDNA) is already widely used in clinical diagnostics. Similarly, non-invasive liquid biopsies could be of great utility in dynamic 2D- and 3D in vitro systems. Here, we show that sampling cfDNA from culture medium can be applied for non-disruptive, real-time monitoring of cell cultures. We demonstrate the power and validity of the approach by measuring global DNA methylation dynamics in a mouse model of extraembryonic lineage conversion and focal regulatory changes in a 3D human hepatocyte differentiation system. In addition to accurately reflecting methylation patterns from matched genomic DNA, this information can be used to report on hepatocyte maturation and directly guide adjustments to culture conditions. Our results can be readily extended to other applications, including high-throughput screens and non-disruptive assessment of therapeutically relevant cell populations.
Developmental timing of mouse embryos and human blastoids can be altered in vitro by inducing a reversible dormant state through inhibition of the cellular nutrient sensor mTOR. This state is associated with globally reduced genomic activity, epigenetic rewiring, and species-specific metabolic adaptations. The static nature of the current in vitro dormancy setups restrict adjustments of the embryo and blastoid culture conditions. To create a more in vivo-like system, we are developing a microfluidic-based embryo-on-chip culture system for mouse embryos and human blastoids. This dynamic culture system allows for 1) real-time, species-specific adjustments of the medium composition, 2) adjustment of stiffness via hydrogel embedding, 3) co-culture with endometrial cells, and 4) real-time measurements of culture parameters and embryo live-imaging. Embryo-on-chip systems may enable longer-term preservation of embryos in the dormant state.
The nucleolus, a sub-compartment of the nucleus, is the site of ribosome biogenesis. Additionally, multiple studies suggest that the nucleolus functions as a liquid condensate. Biomolecular condensates consist of proteins and often nucleic acids that undergo multivalent and transient interactions. Intrinsically disordered regions (IDRs) within proteins are crucial biophysical features that promote condensate formation via liquid-liquid phase separation. We investigate the hypothesis that ribosomal proteins, many of which contain elongated IDRs, are capable of undergoing and driving condensate formation. Instead of producing and purifying ribosomal proteins individually, we focused on their collective isolation and purification. To this end, we isolated ribosomes from various species, ranging from bacteria to humans, and purified all their ribosomal proteins (up to ~80 in humans) using biochemical methods. We systematically screened buffer conditions and tested the impact of parameters such as pH, salt concentration, and temperature on the condensate formation of the individual ribosomal protein preparations. In addition, we designed a fluorescence-based high-throughput in vivo screen using peptide fragments that cover the entire sequences of all human and E. coli ribosomal proteins to identify potential nucleolar localization sequences and their evolutionary history.
Precise regulation of gene expression is crucial, especially in cell fate specification during development. Much of this regulatory information is encoded by cis-regulatory DNA elements such as enhancers and silencers as well as epigenetically through three-dimensional genome organisation and chromatin modifications. By applying synthetic genomics at a locus encoding two transcription factors with mutually exclusive expression patterns and CDX2 and PDX1 are lineage-specifying transcription factors expressed mutually exclusive in the developing embryo; while PDX1 is involved in the specification of foregut and foregut-derived organs, CDX2 specifies the mid- and hindgut-derived lineages. Yet, the two genes are encoded in a roughly 50kb regulatory domain containing single-use and putatively pleiotropic regulatory elements. Utilising synthetic genomics, we aim to dissect how lineage choice is encoded at the CDX2/PDX1 locus.
Precise spatiotemporal regulation of gene expression is essential for healthy embryonic development. Long non-coding RNAs (lncRNAs) contribute to the regulation of gene expression. However, our current understanding of lncRNAs derives mostly from in vitro studies, with little known about their function in vivo. The core aim of this project is to dissect mechanisms of gene regulation by lncRNAs in vivo. The lncRNA Maenli is a perfect model to gain insight into lncRNA function as it activates a specific target gene in a tissue-specific manner during mouse embryonic development. This enables a quantitative in vivo read-out without compounding genome-wide effects. Our hypothesis is that Maenli function is restricted to its 3D chromatin domain where it activates its target gene through transcription-associated processes, and not via the mature lncRNA transcript. To test this, I require methods that enable manipulation of the entire ~25kb locus and site-specific insertion in the genome. I will use cutting-edge synthetic biology to re-construct and manipulate Maenli, insert it in specific genomic regions, and examine its effect on target gene expression and phenotype in the mouse embryo. I will use low-input CUT&RUN to assess the functional chromatin state associated with Maenli in the limb bud. This interdisciplinary approach will illuminate specific mechanisms of cis-acting lncRNA function in vivo.
Extraembryonic cell types share an unusual DNA methylation landscape with intermediate global methylation and hypermethylation at hundreds of CpG islands. To better understand the establishment of this DNA methylation pattern, we first investigated the expression dynamics of key epigenetic regulators in an inducible model system that rapidly differentiates ESCs into extraembryonic endoderm cells. We show that key DNA methylation-modifying enzymes are dynamically regulated at both the RNA and protein level during the differentiation. Specifically, the de novo methyltransferases DNMT3A and DNMT3B are initially upregulated upon exit from pluripotency but then show a continuous decrease afterwards. UHRF1 protein levels, crucial for methylation maintenance, are rapidly lost upon induction, pointing to a possible mechanism for the global hypomethylation. Additionally, our data demonstrate that DNA methylation changes are accompanied by a global epigenetic transition that includes loss of H3K36 methylation, which has been shown to recruit de novo methyltransferases to intra- and intergenic regions. We next rescued de novo methyltransferase isoform expression in a DNMT3 double knockout background. We include H3K36me interaction domain mutant isoforms to probe whether loss of this association contributes to a possible redistribution of de novo methyltransferases.
Taken together, our results will help assign specific roles of key enzymes in establishing the extraembryonic methylome.
One of the most extreme examples of limb adaptation in mammals is the bat wing. Bat forelimbs have massively elongated digits (II-V), which are connected to each other by an interdigital membrane. This distinctive type of wing is their most evident adaptation to powered flight, which is also a unique ability among mammals. It has been already suggested that regulatory mutations may be involved in the evolution of this peculiar trait; however, their identification is still challenging. For this reason, we generated a comprehensive dataset of three critical developmental stages for bat (Carollia perspicillata) and mouse (Mus musculus), including genomic, transcriptomic and epigenomic data. This allowed us to identify differences between developing forelimb and hindlimb, and between bat and mouse. Among the candidate loci, we identify several members of the Hoxd cluster showing differential expression, confirming previous findings. Hox genes encode for transcription factors which control the precise animal body plan development, and Hoxd genes have been co-opted to control limb development. Traditionally, Hox genes are expressed in a collinear fashion along the anterior-posterior and the proximal-distal axes. However, we also find that Hoxd genes traditionally expressed in the proximal mouse limb gain a distal expression in the bat forelimb. We also observe several predicted regulatory changes which may be responsible of this, involving alterations of previously characterized enhancers and of CTCF peaks close to the boundary of the cluster, which may explain the ectopic activation of the proximal Hoxd genes. Moreover, we find that a bat accelerated region (i.e., a genomic region that is highly conserved during mammalian evolution, but diverged in bats) interacting with the locus shows a gain of enhancer activity in the distal limb when cloned into a reporter assay, while the syntenic mouse sequence does not drive any specific pattern. To test these hypotheses, we are currently performing further functional studies, involving the replacement of mouse genomic sequences with the syntenic bat sequences in mouse embryonic stem cells. These sequences can be as small as few kb, or as large as over 150 kb, due to the assembly of large constructs using yeast or the isolation of sequences of interest from a bat BAC library. The unique insights coming from the replacement of large regulatory landscapes in transgenic mice will help us to gain a better understanding of limb development and gene regulation across species, both at a molecular and morphological level.
During development, cells with embryonic as well as extraembryonic origin contribute to the gut endoderm in mice. While embryonic cells differentiate to form the adult organs, extraembryonic cells are subjected to programmed cell death and their remnants are cleared via non-professional phagocytosis. Despite their normal elimination around midgestation, loss of p53 enables extraembryonic gut endoderm cells to contribute to later-stage embryonic organs and acquire further differentiated transcriptional programs. This is unexpected, given that these cells maintain their distinct extraembryonic DNA methylation landscape, which, together with resulting transcriptional signatures, clearly distinguishes them from their embryonic counterparts. These observations raise fundamental questions including whether in a p53-mutant context extraembryonic cells can contribute to the adult intestinal organs, which cell identities they can adopt and what consequences could arise from the presence of these cells over the lifespan of an animal. Leveraging mouse models and in vitro culture systems paired with single-cell sequencing and spatial profiling, we are aiming to investigate the potential and challenges of the non-physiological presence of extraembryonic cells during late embryonic development and beyond.
The ever-evolving landscape of genomics has been fueled by advances
in high-throughput sequencing technologies, providing deeper insights
into the intricate mechanisms of gene regulation. High-Sensitive Nascent
Transcript Sequencing (HiS-NET-seq, Bressin et al., Nature Commun,
2023) has emerged as a high-resolution and sensitive technique that
captures nascent RNA transcripts, including lariat intermediates
immediately after RNA splicing, on a genome-wide scale. Despite its
potential, current applications often overlook this valuable feature,
presenting a compelling opportunity to expand our repertoire to study
RNA splicing regulation in cells. We are developing a machine-learning
model that predicts the positions of RNA branch points in the human
genome. RNA branch points play a critical role in the splicing process,
acting as important regulatory elements that dictate the maturation of
RNA transcripts. Since alterations in splicing are a hallmark of human
disease, our results will also have implications for understanding disease
mechanisms in the future.
DNA methylation, an essential epigenetic modification for maintaining genome integrity, involves multi-layered chromatin interactions that require remodeling proteins like the Helicase, Lymphoid-specific (HELLS). However, the precise genomic targets and mechanism of HELLS are not yet fully understood, particularly in human cells. Here, we generated HELLS and DNA methyltransferase (DNMT) knockout human pluripotent stem cells (iPSCs) and assembled telomere-to-telomere maps of whole genome bisulfite sequencing data combined with ATAC-sequencing. Disrupting HELLS induces a significant global loss of DNA methylation, that is distinct from the DNMTs, in particular over centromeric satellite repeats, which is in alignment with its disease-associated phenotype of frequent abnormal mitosis. Using directed differentiation, we show that in iPSCs HELLs is dispensable for local enhancer remodeling and the potential to differentiate into the three germ layers. Taken together, these findings clarify the regulatory interplay between chromatin remodelers and DNA methylation that advance our knowledge regarding the role of HELLS within key structural genomic regions with direct implications for human disease.
Mammals have heteromorphic sex chromosomes (X and Y). Females, with two X chromosomes, employ mechanisms to balance the double dose, such as X-chromosome inactivation. This involves silencing one X chromosome through the expression of Xist, an X-linked lncRNA. Most of our current knowledge about the mechanisms regulating Xist expression has been derived from studies using mice as the primary model, as all placental mammals express Xist to induce X chromosome silencing. However, the dynamics of Xist expression and the strategies for X chromosome silencing appear to vary among species. During early development, XIST is not expressed in mice, while in humans, it is expressed. Thus, we aim to understand where these differences in XIST expression patterns are coming from.We hypothesise that differential expression of transcription factors between humans and mice during early development contributes to the observed differences in XIST expression and regulation. In previous work, we identified the GATA transcription factor family as Xist activators when overexpressed in mouse embryonic stem cells, where these factors are not usually present. By analyzing expression data from human naïve stem cells, we have noted that some members of the GATA family are expressed in humans, qualifying them as one of the main candidates to activate the expression of XIST and account for the differences in XIST expression patterns observed between rodents and hominoids. By testing the role of these factors on XIST regulation in human naïve stem cells, we aim to understand how differential expression of developmental transcription factors can lead to divergent expression patterns and silencing mechanisms of the molecule responsible for X-chromosome inactivation in all placental mammals.
While the genetic code provides a discrete mapping between a coding DNA sequence and its corresponding chain of amino acids, no such direct mapping exists between a regulatory DNA sequence and its effect on target genes. One approach to relating a sequence of nucleotides to regulatory activity are so called sequence-to-function deep learning models. These models have proven effective in learning sequence features associated with various functional readouts, such as chromatin accessibility and enhancer activity. Moreover, they can be employed to engineer regulatory sequences from scratch, aiding in deciphering the gene regulatory grammar. Here, we present an approach to design cell-type-specific enhancer sequences for chondrocytes in the developing mouse limb. Initially, a sequence-to-accessibility model is trained on single-cell ATAC-seq data to predict accessible regions in chondrocytes. By subsequently integrating single-cell ATAC-seq and bulk ChIP-seq data, we circumvent the need for laborious functional enhancer activity assays such as STARR-seq for model fine-tuning. Preliminary evaluation of our model demonstrates its capability to distinguish accessible regions in chondrocytes from GC-content- and repeat-matched negative sequences, achieving an AUC of 0.90, as well as to recover known transcription factor motifs.
Physical forces of the blood stream in our vessels contribute to the health of endothelial cells (ECs). Lining the inner surface of blood vessels, ECs are highly dependent on mechanical cues such as fluid shear stress (FSS) for proper function and a stable identity. FSS induced by the blood flow triggers atheroprotective biochemical signaling and force transduction via the cytoskeleton to the nucleus. Sudden changes in these forces as from high pulsatile FSS to a low turbulent FSS result in phenotypical adaptations as well as changes in gene regulatory responses of ECs. The accessibility, organization, and modification of the chromatin play a key role in this process and must be looked at simultaneously. Therefore, we established a vascular in-vitrosystem for culturing human aortic endothelial cells (HAoECs) under physiologic FSS conditions. In a second set-up, cells were exposed to physical constraints to induce cell elongation and nuclei orientation without the application of FSS. Both systems allow us to dissect how FSS induced morphology changes contribute to FSS-induced changes in gene regulation, independent of the signaling pathways induced by the FSS alone. The differently stimulated HAoECs are inspected morphologically as well as screened with integrative sequencing approaches. We have seen a correlation between cell elongation and differential gene expression, which allows us now to further dissect the molecular mechanism of endothelial atheroprotection. Combining the results from RNA-seq, ATAC-seq, CUT&RUN-seq, and TF footprinting will provide us with first insights into the mechanisms of how ECs process mechanical stimuli on a genomic level. Furthermore, we expect to open up a new possibility to look at the underlying mechanisms of the atheroprotective role of FSS and FSS-induced morphological adaptations in the human vasculature. Identifying key regulators in vessel maintenance contributes to a better understanding of aortopathies and related diseases. On a translational level, it paves the way for preventive targeted (gene) therapy approaches related to vascular mechanotransduction. Future studies will elucidate to which degree mechanogenomic adaptations are cell type-specific or shared between tissues.
Timely resolution of muscle tissue debris upon injury is essential for effective regeneration and requires a temporally regulated regeneration-specific niche supporting muscle stem cells (MuSCs). This niche is provided in part by fibroadipogenic progenitors (FAPs), and in part by immune cells such as macrophages, which activate upon muscle injury. More specifically, transition from pro-inflammatory (M1 macrophage) to anti-inflammatory (M2 macrophage) microenvironment combined with the build-up of a pro-myogenic extracellular matrix (ECM) is required for complete restoration of muscle mass. Previously we identified odd skipped transcription factor 1 (Osr1) as a marker for an embryonic FAP-like cell population, which gives rise to adult FAPs and provides a pro-myogenic niche conductive for muscle formation in development. Moreover, in adult mice, Osr1 is specifically expressed in FAPs upon acute injury and is required for efficient regeneration. Osr1-deficient mice (Osr1cKO) show an imbalanced macrophage polarization (M1:M2 ratio) and a pro-fibrogenic profile of FAPs-secreted ECM. This suggests a connection between aberrant ECM deposition and altered macrophage function, which is so far elusive. The aim of the current study is to identify the exact molecular agents secreted either by macrophages or FAPs, leading to sub-par regeneration observed in Osr1cKO mice upon muscle injury. LC-MS analysis shows that the ECM generated by Osr1cKO FAPs has a unique proteomic profile showing an increase in matrix metalloproteases (MMP9, MMP19) and collagens (Col3a1, Col14a1) at the same time. This is interesting as previous studies pinpoint an effect of matrix metalloproteases and collagens on macrophage recruitment and ultimately tissue regeneration. Bulk RNA sequencing data sets from macrophage subtypes (M1/M2), will be analysed to understand the macrophage response to aberrant ECM and to disentangle the role of the identified ECM regulators in controlling macrophage polarization. Subsequent functional perturbations e.g., in vitro cell co-culture experiments and ECM decellularization-cell reseeding assays will be employed to decipher the exact interactive receptor-ligand pairs. We anticipate our initial results to be a starting point for more detailed quantitative insight into functional communication network between macrophage phenotypes and ECM dynamics during regeneration.
Gene regulation is an inherently quantitative process, where transcription factor levels are translated to transcriptional responses by cis-regulatory DNA. This is particularly important during mammalian development, where subtle changes can result in developmental disorders. Here, we show our scalable approach to measure transcriptome-wide responses to different TF levels in mouse embryonic stem cells. We combined our tunable CRISPRi system CasTuner with single-cell transcriptomics and knocked-down the TF Oct4 to varying levels during LIF-withdrawal differentiation. We found around 200 genes that responded strongly (fold-change >3, p <0.05). To determine each gene’s response-curve to different Oct4 levels, we computationally binned cells by Oct4 expression. We could then group genes by different response behaviors. We currently try to predict which cis-regulatory DNA is required to sense Oct4 levels and investigate which gene regulatory features could explain response behavior. In the future we plan to screen TFs in naive and differentiating stem cells. We expect that such systematic study will improve our understanding of gene regulation and cell fate decisions.
Biomolecular condensates are thought to create subcellular microenvironments that regulate specific biochemical activities, and have been implicated in a wide range of cellular processes. However, directly probing the microenvironment within cellular condensates and testing condensate functions in live cells is a major challenge, because tools to selectively perturb specific condensates in living cells are lacking. Here we developed a non-natural micropeptide (i.e., the “killswitch”) and a Nanobody-based recruitment system as a universal approach to probe the microenvironments and functions of endogenous condensates in live cells. The killswitch is a hydrophobic, aromatic-rich sequence with an ability to self-associate, and no homology to human proteins. When recruited to endogenous and disease-specific condensates in human cells, the killswitch arrested the dynamics of the condensate-forming proteins, which led to predicted and unexpected functional consequences. For example, targeting the killswitch to the nucleolar protein NPM1 inhibited the dynamics of ribosomal proteins within nucleoli. Targeting the killswitch to fusion oncoprotein condensates inhibited the dynamics of effector proteins in the condensates, altered condensate composition, and inhibited proliferation of condensate-driven leukemia cells. In adenoviral nuclear condensates, the killswitch inhibited partitioning of capsid protein into condensates, and suppressed viral particle assembly. The results suggest that the microenvironment within cellular condensates has an essential contribution to non-stochiometric enrichment of effector proteins. The killswitch is a widely applicable tool to alter the material properties of endogenous condensates, and as a consequence, to probe functions of condensates linked to diverse physiological and pathological processes in living systems.
Gene regulatory landscapes represent extended genomic regions containing multiple cis-regulatory elements such as enhancers or promoters, which control gene expression. Our current understanding of gene regulation focuses heavily on the function and interaction of enhancer sequences themselves and often attributes little function to the composition of regulatory landscapes, such as distance between enhancers, enhancer order and the nature of the inter-enhancer sequences. However, exactly these features might harbor unknown layers of regulatory information. Our limited understanding of how enhancers act and interact within regulatory landscapes stems from the lack of efficient ways to manipulate large genomic regions. Here, we overcome these technical challenges by taking advantage of a synthetic regulatory genomics approach that allows us to systematically re-engineer variants of Prdm14 regulatory landscape in mESCs, composed of 5 individual enhancers distributed across 85kb. From short DNA fragments, we assemble BACs carrying wildtype and mutant versions of the entire Prdm14 regulatory landscape in yeast and integrate these into safe harbour loci in mESCs using serine integrases. We created Prdm14 domains with altered inter-enhancer distances as well as sequences and analyse how these alter gene expression and chromatin modifications at the locus using qPCR, single-molecule RNA-FISH and ChIP-seq. By comparing these precisely altered synthetic Prdm14 regulatory landscapes, we aim to understand the mechanisms by which regulatory landscape composition influence gene regulation.
Until recently, ribosomal RNA (rRNA) genes (rDNA) were completely excluded from the human reference genome because of their complex structure. Furthermore, like many other non-coding RNAs, the highly abundant rRNAs are routinely removed from the RNA sequencing data in favour of messenger RNA (mRNA). Despite the growing interest in studying ribosome heterogeneity, research has primarily focused on ribosomal and ribosome-associated proteins, largely neglecting rRNAs due to their high sequence conservation across evolution.
However, the Telomere-to-Telomere (T2T) Consortium has revealed that the numerous rDNA copies in the human genome, distributed among the five acrocentric chromosomes, are not identical and harbor chromosome-specific sequence variations. Most of these variations are found in regions of the 28S rRNA genes known as expansion segments, which are additional blocks of sequences compared to prokaryotic rRNA with an unknown function. Given that rRNA serves as both the structural and functional core of the ribosome, these sequence variations might have functional implications.
To quantify the expression of rRNA sequence variants and unravel their role in ribosome function, we have undertaken the development of a technology known as RiboSTS. A key feature of this technology will be its ability to simultaneously sequence rRNA and mRNA in single cells, while also accurately measuring ribosome abundance within these cells. Importantly, since the majority of cellular energy is dedicated to ribosome biogenesis and mRNA translation, ribosome abundance quickly responds to changes in cellular environment, which makes it an excellent indicator of a cell state.
We will present the first draft of the RiboSTS technology and share our journey in overcoming the various challenges in single-cell RNA sequencing library preparation and data analysis. Notably, our preliminary analysis of rRNA reads from a published single-cell mRNA sequencing dataset of the human brain revealed striking differences in ribosome abundance and the expression of rRNA variants specific to certain neuronal lineages.
The evolution of species or clade-specific phenotypes is expected to be linked to genomic alterations affecting different aspects of gene regulation. While large genomic rearrangement may influence gene regulation via reshuffling the regulatory landscape (TADs, enhancers), smaller scale variations may contribute functional adaptations in proteins. Here we apply cross-species comparison of genomes and proteomes to predict and prioritize alterations strictly specific to species sharing a distinct phenotype with special emphasis on limb development.
Temperature is a crucial environmental factor that influences biochemical processes, including the formation of transcriptional condensates, which are essential for gene regulation. These condensates are formed by intrinsically disordered regions (IDRs) of transcription factors (TFs) through weak, multivalent interactions. This study explores the hypothesis that sequence features in TF IDRs have adapted to different body temperatures in mammals to maintain proper condensate formation.
We propose that genetic adaptations in IDRs, particularly the reduction of alanine-rich regions, enable species with lower body temperatures to form functional condensates. Preliminary analyses suggest a correlation between reduced alanine content in IDRs and lower body temperatures in ectotherms like fish.
Our research aims to investigate this hypothesis through a series of experiments. We plan to conduct in vitro droplet formation assays on purified IDRs from various species at different temperatures. Additionally, we intend to perform transactivation assays in different cell lines to determine if fish IDRs can transactivate in mammalian cells at lower temperatures.
Further planned experiments include characterizing co-factor interactions and studying the role of orthologous TF IDRs in embryonic development in both mouse and zebrafish models. These experiments will help elucidate how temperature-dependent sequence adaptations in IDRs contribute to the regulation of gene expression and developmental processes.
By investigating these mechanisms, we aim to provide insights into how temperature acts as a selective pressure driving the evolution of IDR sequence features, ensuring the maintenance of transcriptional condensates and proper cellular function across diverse thermal environments.
Our group has actively searched for the genetic causes of rare neurodevelopmental disorders, with a focus on X-linked intellectual disability. One of the X-linked disease genes identified by our studies is ZC4H2, which encodes a zinc-finger protein crucial for nervous system development. We and others have associated ZC4H2 with clinically variable phenotypes in affected male and female patients with neurologic and neuromuscular symptoms, such as muscle weakness, spasticity, and seizures. Our understanding of the molecular and cellular mechanisms and the pathophysiology of ZC4H2-associated rare disorders (ZARD) is currently limited. Here we present three complementary approaches to better understand the pathomechanisms of ZARD, the first of which is the generation of knockin mice with patient-specific missense variants. The establishment of mutant embryos and mice could serve as valuable models for studying ZARD, allowing for detailed characterization through various studies. Second, we have generated monoclonal antibodies against ZC4H2 and are currently validating their specificity and sensitivity for detecting endogenous human and mouse ZC4H2 proteins. These antibodies will be essential tools for studies on ZC4H2 expression, localization, and function, including e.g. spatial- and temporal protein expression analysis and biochemical studies. In addition, we are examining the functional effects of ZC4H2 on a novel protein complex partner, the microtubule tethering protein HOOK3. Changes in protein interactions could disrupt normal cellular functions and contribute to the observed clinical phenotypes.
Pluripotency, which is the ability to give rise to all embryonic cell types, is characterized by distinct epigenetic features, including a high abundance of histone modifications associated with active chromatin. Although pluripotency exists only for a few days in pre/peri-implantation embryos, it can be preserved for longer periods in embryonic diapause. Embryonic diapause is a state of reversible dormancy that is utilized by many mammalian species. This dormant state can be enacted in vitro by inhibiting the mTOR pathway in mouse embryos and embryonic stem cells (ESCs). In contrast to normal, proliferative ESCs, dormant embryos, and ESCs show enriched heterochromatin domains and altered nucleolar morphology. However, how dormant cells maintain pluripotency under these repressive conditions is unknown. In my Ph.D. project, I will investigate the role of genome macro-organization and its relationship with locus-specific chromatin regulation in dormancy. I will employ DamID, DNA- RNA FISH, and pharmacological approaches to study the nucleolar morphology, gene expression, and the effects of perturbations on gene regulation and cellular identity. The findings of this project will provide the first comprehensive mechanisms of gene regulation in dormancy and will specifically unravel the link between cellular growth status, nucleolar morphology, and temporal gene expression in maintaining pluripotency.
Pediatric B-cell acute lymphoblastic leukemia (B-ALL) results from a blockage in the differentiation of B-cell progenitors in the bone marrow, causing abnormal blast cell proliferation. Approximately 15% of patients relapse during or after frontline treatment, the testis being the third most common site of relapse after bone marrow and central nervous system. Treatment achieving long-term event-free survival involves orchiectomy, significantly impacting quality of life. The mechanisms underlying the survival of leukemic cells in the testis remain unclear. To better understand B-ALL testicular involvement, we conducted a comprehensive molecular analysis of relapsed pediatric B-ALL, focusing on the comparative transcriptomic landscape of testicular versus bone marrow relapses and the spatial architecture of tumor immune microenvironment (TiME) in the testicular niche.
Bulk RNA-seq data analysis from a cohort of 38 relapsed pediatric ALLs identified gene expression signatures and biomarkers specific to the testicular relapse group. We explored the spatial composition of the TiME at single-cell resolution in the invaded testis by Imaging Mass Cytometry using a 28-plex antibody panel. Our findings indicate that M2 macrophages and different T cell subtypes dominate the microenvironment, albeit with inter-patient heterogeneity with distinct infiltration patterns ranging from immune deserts to tertiary lymphoid-like structures. Immune infiltration also differs in the type of cells involved, sometimes featuring CD56+ NK cells and regulatory T cells. Ongoing spatial transcriptomics and scRNA-seq experiments will allow us to gain a better understanding of the interaction of B-ALL cells with immune and non-immune componens of the testicular niche.
Hepatoblasts are the foetal stem cells of the liver and play a key role in organogenesis. These stem cells have the capacity to differentiate into hepatocytes and cholangiocytes which represent the main functional cell types of the liver. Lineage tracing and genetic studies in the mouse have shown the association of Wnt signalling with proliferation and differentiation of hepatoblasts. However, the exact function of this pathway in hepatic development and cell fate choice is not fully uncovered, especially in human where access to primary tissues is challenging. Here, we took advantage of human hepatoblast organoids (HBOs) to investigate the role of Wnt in self-renewal and cell fate decisions. Notably, HBOs display a transcriptomic profile characteristic of hepatoblasts and they maintain the capacity to differentiate into hepatocytes and biliary cells. Here, we first showed that Wnt plays a key role in hepatoblast self-renewal in vitro by maintaining their proliferative state through regulation of cell cycle-related genes. However, Wnt was not sufficient to block differentiation of HBOs into hepatocytes or cholangiocytes. Finally, using single-cell transcriptomic analyses, we found that Wnt signalling activity correlates with proliferation of hepatoblasts in the human foetal liver, thereby suggesting that the role for Wnt could be conserved in vivo. Taken together, our results support a model where Wnt signalling acts to preserve the proliferative, self-renewal capacity of hepatoblasts without being sufficient to maintain their bipotent state. These results could open new investigations to further understand the role of Wnt signalling in other adult stem cells and developmental progenitors, where Wnt is also known to drive tissue homeostasis and cell fate decisions.
CRISPR knock-out (KO) screens measure the dependency of a cell line to the inactivation of specific genes in a high-throughput manner [1]. In addition, repeating KO screen with drugs highlights potential drug resistance and sensitivity mechanisms. Repeated over multiple cell lines with recurrent alterations, this allows the extraction of biomarkers driving drug dependencies. Those biomarkers can be combined to improve prescription decisions.
CROPseq [2] couples CRISPR knock-out and single cell RNA sequencing to measure the transcriptome of individual gene knock-outs in a pooled setting, allowing the simultaneous measurement of the transcriptomic response to dozens of KO in parallel.
Using public and in-house large knock-out screens combined with a newly generated CROPseq dataset in three B-ALL cell lines, we reconstructed the signaling and dependency landscape of B-cell acute lymphoblastic leukemia. We used CROPseq data to identify the key regulatory modules in each cell line, and expanded the regulatory networks to downstream targets specifically involved in the survival of the B-cell lineage. Those regulatory networks were exploited to identify KOs with high potential therapeutic index for BCP-ALL, alone or in combinations. We then performed drop-out screen with individual drugs targeting each candidate, shedding light on the mechanisms driving resistance or sensitivity to those drugs and refining our understanding of the interactions between candidate genes. We show that a combination of HDAC class I and CDK4/6 inhibitors is a promising candidate to selectively depleted the B-cell lineage at low doses of both drugs.
[1] Hart T., Tong A.H.Y., Van Leeuwen J. et al. Evaluation and design of genome-wide CRISPR SpCas9 knockout screens. G3, issue 8, 2719-2727 (2017). https://doi.org/10.1534/g3.117.041277
[2] Datlinger P., Rendeiro A., Schmidl C. et al. Pooled CRISPR screening with single-cell transcriptome readout. Nat Methods 14, 297–301 (2017). https://doi.org/10.1038/nmeth.4177
Cis-regulatory elements such as promoters and enhancers coordinate the cell-type specific transcription of their target genes. Promoters sit directly upstream of the transcription start site, while often many partially redundant enhancers are distributed in extended regulatory domains surrounding their target gene. Despite their significance, the mechanisms by which enhancers cooperate and interact within regulatory domains, particularly to drive developmental gene expression patterns, remain poorly understood.
For example, how position and spacing between seemingly randomly distributed enhancers affects their functionality remains a poorly understood aspect of a key characteristic of mammalian genomes. Likewise, the contribution of the majority of DNA sequence, all intervening inter-enhancer sequences, remains unknown. The reason for these largely unexplored features lies in our limited ability to systematically alter large genomic regions. Traditional approaches are mostly restricted to investigating individual elements or require repeated targeting of the same locus, hindering a systematic analysis. We recently established a synthetic biology derived workflow that overcomes current limitations, combining methods from microbiology with integrase-based genome engineering for mESCs. This setup allows us to effectively synthesize any DNA sequences of dozens to hundreds of kb and integrate them site-specific into mESCs.
In this project, we employ this approach to investigate the enhancer function of the regulatory domain of Indian hedgehog (Ihh), a gene coding for a key developmental signaling molecule. We created several synthetic regulatory landscapes of varying sizes containing previously described Ihh enhancers with alternate spacing, order, or inter-enhancer sequence. Using multiplex whole-mount fluorescent in situ hybridization, we find that omitting inter-enhancer regions or altering enhancer order does not significantly affect the expression pattern or level of Ihh. However, maintaining the spacing of the wild-type configuration, while changing (reversing) only the inter-enhancer sequences has strong effects on Ihh expression. Our findings reveal a surprising robustness of developmental gene expression to severe pertubations in the arrangement of cis-regulatory elements, prompting new questions about the “mode-of-action” of non-coding sequence composition in mammalian genomes.
Extraembryonic cells share a characteristic DNA methylation landscape that is globally distinct from the somatic methylome found in virtually all embryonic and adult cell types. This non-canonical methylome is characterised by intermediate methylation globally and at hundreds of CpG islands. Why these cells, which form supporting structures for the embryo, utilize an alternate regulatory program is completely unknown.
Here, we investigate fundamental aspects of this regulation including whether it can be transformed into the canonical somatic landscape and what developmental plasticity it confers. Using ectopic transcription factor expression in extra-embryonic endoderm (XEN) cells we probe reprogramming and lineage conversion potential. Our preliminary findings indicate that transcription factor-based reprogramming of XEN cells into induced pluripotent stem cells via OSKM factors is possible albeit with very low efficiency. In contrast, ectopic NGN2 expression appears to result in cell death and failure to move towards a neural identity. Additional factors such as Ascl1, MyoD and Hnf4a are currently being tested to understand the general nature of this finding. Our work will provide crucial insights on the utility and potential evolutionary purpose of this form of epigenome regulation, with major implications for developmental and disease biology.
Most mammalian genes are interrupted with introns that need to be spliced out to produce a functional mRNA following transcription. Splicing is a multi-step, ATP-dependent process conserved in all Eukaryotes, and under defined conditions can be carried out in cell-free lysates. In mammalian nuclei, large, irregularly shaped multi-component membraneless organelles, known as Nuclear Speckles (NS) are associated with splicing, however a direct link connecting splicing to NS has been missing. In this work, by directly targeting the core proteins of NS, SON and SRRM2, we show that NS are essential for the splicing of a distinct subset of introns that are short, GC-rich and found in clusters within short GC-rich genes. Loss of NS integrity leads to thousands of exon skipping and several hundred intron retention events, frequently affecting the very same genes. Mechanistically, SON, together with SRRM2 forms a viscoelastic structure that concentrates splicing factors (SR proteins and the SF3 complex) around highly expressed, GC-rich genes, suppressing irregular splicing events. Evolutionary analysis suggests that NS evolved during the water-to-land transition of Vertebrates, coinciding with genome expansion events caused by elevated transposable element activity. Subsequent genome contraction events driven by meiotic recombination, lead to the creation of highly compartmentalised, compact genes but with elevated GC-content. By facilitating the orderly splicing of these recombination by-products, NS enable streamlined genomes after expansion events, while creating a single point of failure for individuals with mutations impacting the core proteins, SON and SRRM2.