Research Groups

Projects in Computational RNA Biology

  • Area 1: Learning the Molecular Code of RNA Regulation
  • Area 2: Regulatory Mechanisms in Human (Patho)Physiology
  • Area 3: Computational Toolbox

 

Area 1: Learning the Molecular Code of RNA Regulation

RNA regulation plays a major role in cellular function, and alterations such as alternative splicing are frequently implicated in human disease including cancer. Key players are RBPs that recognise cis-regulatory RNA sequence elements, such as short sequence motifs or structured RNA folds. However, the position and function of most cis-regulatory elements remain unknown, hindering the interpretation of disease-associated mutations and the resulting splicing changes.

We recently established a high-throughput screening approach to decode the cis-regulatory landscape of alternative splicing decisions in collaboration with Dr. Julian König and Dr. Stefan Legewie at IMB in Mainz. As a prototype example, we dissected the cancer-relevant alternative exon 11 in the proto-oncogene MST1R (RON) encoding for a receptor tyrosine kinase (Braun et al., 2018). Skipping of RON exon 11 results in the pathological isoform RONΔ165 that promotes tumour invasiveness and is frequently upregulated in tumours. Mathematical modeling of splicing kinetics in the Legewie group enabled us to identify more than 1,000 mutations affecting RON exon 11 skipping. Importantly, the measured effects correlated with RON alternative splicing in cancer patients bearing the same mutations. Moreover, we highlighted the RNA-binding protein heterogeneous nuclear ribonucleoprotein H (HNRNPH) as a key regulator of RON splicing in healthy tissues and cancer. Our results offer insights into splicing regulation and the hidden layer of splicing-effective mutations in cancer.

 

Figure2_RON_mutagenesis
Figure 2: A high-throughput screen to map splice-regulatory mutations for alternative exon 11 in the proto-oncogene RON. (A) High-throughput detection of splicing-effective mutations. Mutagenic PCR creates mutated minigene library (left) that gives rise to alternatively spliced transcripts (middle). Mutations and corresponding splicing products are characterised by DNA and RNA sequencing, respectively, and linked by unique 15-nt barcode sequence in each minigene (coloured boxes). Black and grey boxes depict constitutive and alternative exons, respectively. (B) Model-inferred landscapes of 1,747 single mutation effects on three different RON transcript isoforms with alternative exon (AE) inclusion, AE skipping and full intron retention (IR) in human HEK293T cells. Each mutation effect is indicated as a coloured dot according to the inserted nucleobase (see legend). Red lines indicate median (dashed) ± 2 standard deviations (SD; solid) for unmutated wild-type (wt) minigenes. [Images taken from Braun et al., 2018]

 

In collaboration with JunProf. Müller-McNicoll at GU Frankfurt, we investigate how SR proteins connect pre-mRNA splicing with alternative polyadenylation and nuclear export of specific mRNA isoforms. In this project within the DFG Collaborative Research Consortium SFB902 "Molecular Principles of RNA-based Regulation", we discovered that SRSF3 and SRSF7 regulate alternative polyadenlyation in opposing directions, thereby changing the mRNA’s fate from export and translation in the cytoplasm to retention in the nucleus (Müller-McNicoll et al., 2016). In a parallel line of research, we reported that SRSF5 effects an export switch of its target mRNAs during neuronal differentiation through a reversible inhibition of nucleo-cytoplasmic shuttling (Botti et al., 2017).

Intracellular mRNA transport and local translation are evolutionarily conserved concepts that play essential roles in cellular compartmentalisation from yeast to human. In collaboration with several groups of the DFG Research Unit FOR2333 "Macromolecular complexes in mRNA localization", we employ our expertise on protein-RNA interactions to study the molecular determinants of RBP binding in mRNA translocation. By studying similar processes in different model organisms, we decipher the global rules of zipcode recognition and pinpoint analogies as well as differences across evolution. For instance, we recently described how the RNA-binding protein Rrm4 recognises its target mRNAs in infectious hyphae of the corn pathogen Ustilago maydis ( Olgeiser et al, 2019). In budding yeast, we further study the specific targeting of mRNAs that enables a spatio-temporal protein expression in the bud tip. The most prominent example is the targeted ASH1 mRNA whose protein product regulates mating-type switching in the daughter cells. Recognition by the mRNA transport machinery requires zipcodes that are encoded in the secondary structure of the RNA molecules.

 

References:

Olgeiser, L, Haag, C, Boerner, S, Ule, J, Busch, A, Koepke, J, König, J, Feldbrügge, M§, Zarnack, K§ (2019) The key protein of endosomal mRNP transport Rrm4 binds translational landmark sites of cargo mRNAs. EMBO Rep 20, e46588. doi: 10.15252/embr.201846588. (§, shared correspondence)

Braun, S*, Enculescu, M*, Setty, ST*, Cortés-López, M, de Almeida, BP, Sutandy, FXR, Schulz, L, Busch, A, Seiler, M, Ebersberger, S, Barbosa-Morais, NL, Legewie, S§, König, J§, Zarnack, K§ (2018) Decoding a cancer-relevant splicing decision in the RON proto-oncogene using high-throughput mutagenesis. Nat Commun 9, 3315. doi: 10.1101/gr.229757.117. (*, equal contribution; §, shared correspondence)

Botti, V, McNicoll, F, Steiner, MC, Richter, FM, Solovyeva, A, Wegener, M, Poser, I, Brandl, H, Zarnack, K, Wittig, I, Neugebauer, KM, Müller-McNicoll, M (2017) Cellular differentiation state modulates the mRNA export activity of SR proteins. J Cell Biol, pii: jcb.201610051. doi: 10.1083/jcb.201610051.

Müller-McNicoll, M, Botti, V, de Jesus Domingues, AM, Brandl, H, Schwich, OD, Steiner, MC, Curk, T, Poser, I, Zarnack, K, Neugebauer, KM (2016) SR proteins are NXF1 adaptors that link alternative RNA processing to mRNA export. Genes Dev 30, 553-566.

 

Area 2: Regulatory Mechanisms in Human (Patho)Physiology

RNA metabolism is often disturbed in human disease, including neurodegeneration and cancer. In collaboration with molecular biologists and pre-clinical researchers, we examine transcriptome changes under different pathophysiological conditions.

Cancer-associated fibroblasts (CAFs) in the tumour microenvironment show a remarkable degree of heterogeneity. We investigate the plasticity of CAFs during tumorigenesis based on transcriptional profiling of fibroblasts isolated from murine mammary carcinoma in collaboration with Dr. Andreas Weigert at the Institute of Biochemistry I, GU Frankfurt. Using in-depth computational analysis, we recently reported a shift in fibroblast populations over time and trained a machine learning model that distinguishes tumour stages (Elwakeel et al, 2019). The underlying gene signature correlates with survival in human breast cancer patients, supporting a clinical relevance of our findings.

 

Figure3_CafFigWebPage02
Figure 3: Gene expression changes of cancer-associated fibroblasts (CAFs) over time. (A) Gene expression profiles of CAFs from early-stage (EC) and late-stage (LC) carcinoma from a mouse breast cancer model (PyMT mice). Normalised RNA-Seq data are shown for 906 genes that discriminate EC and LC CAFs according to differential gene expression analysis. Marker genes from distinct CAF subtypes (mCAFs, matrix CAFs; vCAFs, vascular CAFS; cCAFs, cycling CAFs; dCAFs, developmental CAFs) are indicated on the right. (B) Histological validation of an enriched gene signature, in early-stage (upper panel) and late-stage (lower panel) PyMT tumors. Tumors were analyzed with PhenOptics for the expression of marker genes. Images show the combined expression of all markers. [Images taken from Elwakeel et al., 2019]

 

Human cells maintain elaborate surveillance mechanisms to detect and degrade faulty mRNAs from premature polyadenylation, which can disrupt the open reading frame and result in non-functional protein products. The ribosome-associated quality control (RQC) machinery detects ribosomes that have been stalled when running into poly(A) sequences and promotes their release and recycling. We recently identified the RNA-binding E3 ubiquitin ligase Makorin Ring Finger Protein 1 (MKRN1) as a novel sensor of poly(A) sequences in RQC ( Hildebrandt et al., 2019). Using a multi-omics approach in collaboration with Dr. Petra Beli and Dr. Julian König at IMB Mainz, we found that MKRN1 specifically binds at poly(A) sequences, interacts with poly(A)-binding proteins and translational regulators, and directly ubiquitylates ribosomal proteins. MKRN1 thereby bridges between specific RNA recognition and ubiquitin-mediated protein quality control. Reporter assays directly demonstrate that ribosomes no longer stall at poly(A) sequences when MKRN1 is depleted. Our results cumulate in a functional model that MKRN1 acts as a roadblock on poly(A) sequences that prevents ribosomes from translating by ubiquitylating ribosomal proteins. The project is funded within the DFG Priority Programme SPP1935 "Deciphering the mRNP code: RNA-bound determinants of post-transcriptional gene regulation".

Hypoxia, i.e. the lack of adequate supplies of oxygen, is a common characteristic of the cellular microenvironment of many tumours, and a key factor to induce the growth of new blood vessels. RNA-regulatory processes have been implicated in the control tissue homeostasis and pathophysiological processes in the cardiovascular system as well as during tumorigenesis. We therefore investigate changes in RNA processing during hypoxic stress, with a particular focus on the formation and function of circular RNAs (circRNAs). Using a consolidated pipeline, we detected thousands of circRNAs in total RNA-seq data from in different human cancer cell lines, including dozens that respond to hypoxic stress ( Di Liddo et al, 2019).

Moving beyond cancer, we recently initiated a project to study the role of RBP-RNA interactions in circRNA biogenesis and function in the cardiovascular system in collaboration with Prof. Dr. Stefanie Dimmeler at the Institute for Cardiovascular Regeneration, GU Frankfurt. The project is funded as part of the DFG Collaborative Research Consortium TRR267 "Non-coding RNAs in the cardiovascular system" between GU Frankfurt and TU Munich.

 

References:

Di Liddo, A, de Oliveira Freitas Machado, C, Fischer, S, Ebersberger, S, Heumüller, A, Weigand, JE, Müller-McNicoll, M, Zarnack, K (2019). A combined computational pipeline to detect circular RNAs in human cancer cells under hypoxic stress. J Mol Cell Biol 11, 829-844. doi: 10.1093/jmcb/mjz094.

Hildebrandt, A, Brüggemann, M, Rücklé, C, Boerner, S, Heidelberger, JB, Busch, A, Hänel, H, Voigt, A, Möckel, MM, Ebersberger, S, Scholz, A, Dold, A, Schmid, T, Ebersberger, I, Roignant, JY, Zarnack, K§, König, J§, Beli, P§ (2019) The RNA-binding ubiquitin ligase MKRN1 functions in ribosome-associated quality control of poly(A) translation. Genome Biol 20, 216. doi: 10.1186/s13059-019-1814-0 (§, shared correspondence)

Elwakeel, E*, Brüggemann, M*, Fink, AF, Schulz, MH, Schmid, T, Savai, R, Brüne, B, Zarnack, K§, Weigert, A§ (2019) Phenotypic plasticity of fibroblasts during mammary carcinoma development. Int J Mol Sci 20, pii: E4438. doi: 10.3390/ijms20184438. (*, equal contribution; §, shared correspondence)

 

Area 3: Computational Toolbox

Our work builds on custom code and pipelines for the fast and efficient analysis of omics data. Our primary focus lies on the characterisation of protein-RNA interactions and their reliable quantification from high-throughput datasets, such as (i)CLIP, complemented by functional studies based on RNA-seq and ribosome profiling (RiboSeq), among others.

Exemplary applications include:

  • We benchmarked CLIP data analysis strategies across datasets for different RBPs and experimental protocols (Haberman et al., 2017), and compiled a computational pipeline for extracting RBP binding sites from iCLIP data ( Busch et al., 2019).
  • We built a consolidated pipeline to detect and quantify circRNAs from deep RNA-Seq data, allowing to profile the expression, regulation and molecular features of circRNAs (Di Liddo et al., 2019).
  • We contributed to a pipeline to detect upstream open reading frames (uORFs) from ribosome profiling data and to assess their translation-regulatory potential on the associated ORFs (Scholz et al., 2019).

 

Figure4_circRNA_pipeline
Figure 4: Computational pipeline to detect and quantify circRNAs from rRNA-depleted total RNA-Seq data. Initial circRNA predictions by find_circ (Memczak et al., 2015) and CIRCexplorer (Zhang et al., 2014a) are combined and rigorously filtered to obtain a comprehensive catalogue of circRNAs. In parallel, linearly mapped reads are used for the analyses of differential gene expression and alternative splicing. [Image taken from Di Liddo et al., 2019]

 

References:

Busch, A, Brüggemann, M, Ebersberger, S§, Zarnack, K§ (2019). iCLIP data analysis: A complete pipeline from sequencing reads to RBP binding sites. Methods, pii: S1046-2023(18)30494-8. doi: 10.1016/j.ymeth.2019.11.008. (§, shared correspondence)

Di Liddo, A, de Oliveira Freitas Machado, C, Fischer, S, Ebersberger, S, Heumüller, A, Weigand, JE, Müller-McNicoll, M, Zarnack, K (2019). A combined computational pipeline to detect circular RNAs in human cancer cells under hypoxic stress. J Mol Cell Biol 11, 829-844. doi: 10.1093/jmcb/mjz094.

Scholz, A, Eggenhofer, F, Gelhausen, R, Grüning, B, Zarnack, K, Brüne, B, Backofen, R, Schmid, T (2019) uORF-Tools – Workflow for the determination of translation-regulatory upstream open reading frames. PLOS One 14, e0222459. doi: 10.1371/journal.pone.0222459.

Haberman, N, Huppertz, I, Attig, J, König, J, Wang, Z, Hauer, C, Hentze, MW, Kulozik, AE, Le Hir, H, Curk, T, Sibley, CR, Zarnack, K§, Ule, J§ (2017) Insights into the design and interpretation of iCLIP experiments. Genom Biol 18, 7. doi: 10.1186/s13059-016-1130-x. (§, shared correspondence)