Professional Education

  • Doctor of Philosophy, Technische Universitat Munchen (2015)
  • Diplom, Technische Universitat Munchen (2010)
  • Diplom, Ludwig Maximilian Universitat Munchen (2010)

Stanford Advisors

All Publications

  • Transcriptional Dynamics of Hair-Bundle Morphogenesis Revealed with CellTrails. Cell reports Ellwanger, D. C., Scheibinger, M., Dumont, R. A., Barr-Gillespie, P. G., Heller, S. 2018; 23 (10): 2901–14.e14


    Protruding from the apical surface of inner ear sensory cells, hair bundles carry out mechanotransduction. Bundle growth involves sequential and overlapping cellular processes, which are concealed within gene expression profiles of individual cells. To dissect such processes, we developed CellTrails, a tool for uncovering, analyzing, and visualizing single-cell gene-expression dynamics. Utilizing quantitative gene-expression data for key bundle proteins from single cells of the developing chick utricle, we reconstructed de novo a bifurcating trajectory that spanned from progenitor cells to mature striolar and extrastriolar hair cells. Extraction and alignment of developmental trails and association of pseudotime with bundle length measurements linked expression dynamics of individual genes with bundle growth stages. Differential trail analysis revealed high-resolution dynamics of transcripts that control striolar and extrastriolar bundle development, including those that encode proteins that regulate [Ca2+]i or mediate crosslinking and lengthening of actin filaments.

    View details for DOI 10.1016/j.celrep.2018.05.002

    View details for PubMedID 29874578

  • CellTrails: Reconstruction, visualization and analysis of branching trajectories. Ellwanger, D. C. Bioconductor. 2018 ; Release (3.8):


    CellTrails is an unsupervised algorithm for the de novo chronological ordering, visualization and analysis of single-cell expression data. CellTrails makes use of a geometrically motivated concept of lower-dimensional manifold learning, which exhibits a multitude of virtues that counteract intrinsic noise of single cell data caused by drop-outs, technical variance, and redundancy of predictive variables. CellTrails enables the reconstruction of branching trajectories and provides an intuitive graphical representation of expression patterns along all branches simultaneously. It allows the user to define and infer the expression dynamics of individual and multiple pathways towards distinct phenotypes.

    DOI 10.18129/B9.bioc.CellTrails
  • Aminoglycoside Damage and Hair Cell Regeneration in the Chicken Utricle. Journal of the Association for Research in Otolaryngology : JARO Scheibinger, M., Ellwanger, D. C., Corrales, C. E., Stone, J. S., Heller, S. 2017


    In this study, we present a systematic characterization of hair cell loss and regeneration in the chicken utricle in vivo. A single unilateral surgical delivery of streptomycin caused robust decline of hair cell numbers in striolar as well as extrastriolar regions, which in the striola was detected very early, 6 h post-insult. During the initial 12 h of damage response, we observed global repression of DNA replication, in contrast to the natural, mitotic hair cell production in undamaged control utricles. Regeneration of hair cells in striolar and extrastriolar regions occurred via high rates of asymmetric supporting cell divisions, accompanied by delayed replenishment by symmetric division. While asymmetric division of supporting cells is the main regenerative response to aminoglycoside damage, the detection of symmetric divisions supports the concept of direct transdifferentiation where supporting cells need to be replenished after their phenotypic conversion into new hair cells. Supporting cell divisions appear to be well coordinated because total supporting cell numbers throughout the regenerative process were invariant, despite the initial large-scale loss of hair cells. We conclude that a single ototoxic drug application provides an experimental framework to study the precise onset and timing of utricle hair cell regeneration in vivo. Our findings indicate that initial triggers and signaling events occur already within a few hours after aminoglycoside exposure. Direct transdifferentiation and asymmetric division of supporting cells to generate new hair cells subsequently happen largely in parallel and persist for several days.

    View details for DOI 10.1007/s10162-017-0646-4

    View details for PubMedID 29134476

  • Single-cell analysis delineates a trajectory toward the human early otic lineage PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA Ealy, M., Ellwanger, D. C., Kosaric, N., Stapper, A. P., Heller, S. 2016; 113 (30): 8508-8513


    Efficient pluripotent stem cell guidance protocols for the production of human posterior cranial placodes such as the otic placode that gives rise to the inner ear do not exist. Here we use a systematic approach including defined monolayer culture, signaling modulation, and single-cell gene expression analysis to delineate a developmental trajectory for human otic lineage specification in vitro. We found that modulation of bone morphogenetic protein (BMP) and WNT signaling combined with FGF and retinoic acid treatments over the course of 18 days generates cell populations that develop chronological expression of marker genes of non-neural ectoderm, preplacodal ectoderm, and early otic lineage. Gene expression along this differentiation path is distinct from other lineages such as endoderm, mesendoderm, and neural ectoderm. Single-cell analysis exposed the heterogeneity of differentiating cells and allowed discrimination of non-neural ectoderm and otic lineage cells from off-target populations. Pseudotemporal ordering of human embryonic stem cell and induced pluripotent stem cell-derived single-cell gene expression profiles revealed an initially synchronous guidance toward non-neural ectoderm, followed by comparatively asynchronous occurrences of preplacodal and otic marker genes. Positive correlation of marker gene expression between both cell lines and resemblance to mouse embryonic day 10.5 otocyst cells implied reasonable robustness of the guidance protocol. Single-cell trajectory analysis further revealed that otic progenitor cell types are induced in monolayer cultures, but further development appears impeded, likely because of lack of a lineage-stabilizing microenvironment. Our results provide a framework for future exploration of stabilizing microenvironments for efficient differentiation of stem cell-generated human otic cell types.

    View details for DOI 10.1073/pnas.1605537113

    View details for Web of Science ID 000380346200056

    View details for PubMedID 27402757

    View details for PubMedCentralID PMC4968758

  • Large-scale modeling of condition-specific gene regulatory networks by information integration and inference NUCLEIC ACIDS RESEARCH Ellwanger, D. C., Leonhardt, J. F., Mewes, H. 2014; 42 (21)


    Understanding how regulatory networks globally coordinate the response of a cell to changing conditions, such as perturbations by shifting environments, is an elementary challenge in systems biology which has yet to be met. Genome-wide gene expression measurements are high dimensional as these are reflecting the condition-specific interplay of thousands of cellular components. The integration of prior biological knowledge into the modeling process of systems-wide gene regulation enables the large-scale interpretation of gene expression signals in the context of known regulatory relations. We developed COGERE (, a method for the inference of condition-specific gene regulatory networks in human and mouse. We integrated existing knowledge of regulatory interactions from multiple sources to a comprehensive model of prior information. COGERE infers condition-specific regulation by evaluating the mutual dependency between regulator (transcription factor or miRNA) and target gene expression using prior information. This dependency is scored by the non-parametric, nonlinear correlation coefficient η(2) (eta squared) that is derived by a two-way analysis of variance. We show that COGERE significantly outperforms alternative methods in predicting condition-specific gene regulatory networks on simulated data sets. Furthermore, by inferring the cancer-specific gene regulatory network from the NCI-60 expression study, we demonstrate the utility of COGERE to promote hypothesis-driven clinical research.

    View details for DOI 10.1093/nar/gku916

    View details for Web of Science ID 000347914600007

    View details for PubMedID 25294834

  • miR-92a regulates TGF-ß1-induced WISP1 expression in pulmonary fibrosis. international journal of biochemistry & cell biology Berschneider, B., Ellwanger, D. C., Baarsma, H. A., Thiel, C., Shimbori, C., White, E. S., Kolb, M., Neth, P., Königshoff, M. 2014; 53: 432-441


    Idiopathic pulmonary fibrosis (IPF) is the most common and fatal form of idiopathic interstitial pneumonia. MicroRNAs (miRNAs), short, single-stranded RNAs that regulate protein expression in a post-transcriptional manner, have recently been demonstrated to contribute to IPF pathogenesis. We have previously identified WNT1-inducible signaling pathway protein 1 (WISP1) as a highly expressed pro-fibrotic mediator in IPF, but the underlying mechanisms resulting in increased WISP1 expression, remain elusive. Here, we investigated whether WISP1 is a target of miRNA regulation. We applied a novel supervised machine learning approach, which predicted miR-30a/d and miR-92a target sites in regions of the human WISP1 3'UTR preferentially bound by the miRNA ribonucleoprotein complex. Both miRNAs were decreased in IPF samples, whereas WISP1 protein was increased. We demonstrated further that transforming growth factor (TGF)-β1-induced WISP1 expression in primary lung fibroblasts in vitro and lung homogenates in vivo. Notably, miR-30a and miR-92a reversed TGF-β1-induced WISP1 mRNA expression in lung fibroblasts. Moreover, miR-92a inhibition increased WISP1 protein expression in lung fibroblasts. An inverse relationship for WISP1 and miR-92a was found in a TGF-β1 dependent lung fibrosis model in vivo. Finally, we found significantly increased WISP1 expression in primary IPF fibroblasts, which negatively correlated with miR-92a level ex vivo. Altogether, our findings indicate a regulatory role of miR-92a for WISP1 expression in pulmonary fibrosis.

    View details for DOI 10.1016/j.biocel.2014.06.011

    View details for PubMedID 24953558

  • Rare variants in LRRK1 and Parkinson's disease. Neurogenetics Schulte, E. C., Ellwanger, D. C., Dihanich, S., Manzoni, C., Stangl, K., Schormair, B., Graf, E., Eck, S., Mollenhauer, B., Haubenberger, D., Pirker, W., Zimprich, A., Brücke, T., Lichtner, P., Peters, A., Gieger, C., Trenkwalder, C., Mewes, H., Meitinger, T., Lewis, P. A., Klünemann, H. H., Winkelmann, J. 2014; 15 (1): 49-57


    Approximately 20 % of individuals with Parkinson's disease (PD) report a positive family history. Yet, a large portion of causal and disease-modifying variants is still unknown. We used exome sequencing in two affected individuals from a family with late-onset PD to identify 15 potentially causal variants. Segregation analysis and frequency assessment in 862 PD cases and 1,014 ethnically matched controls highlighted variants in EEF1D and LRRK1 as the best candidates. Mutation screening of the coding regions of these genes in 862 cases and 1,014 controls revealed several novel non-synonymous variants in both genes in cases and controls. An in silico multi-model bioinformatics analysis was used to prioritize identified variants in LRRK1 for functional follow-up. However, protein expression, subcellular localization, and cell viability were not affected by the identified variants. Although it has yet to be proven conclusively that variants in LRRK1 are indeed causative of PD, our data strengthen a possible role for LRRK1 in addition to LRRK2 in the genetic underpinnings of PD but, at the same time, highlight the difficulties encountered in the study of rare variants identified by next-generation sequencing in diseases with autosomal dominant or complex patterns of inheritance.

    View details for DOI 10.1007/s10048-013-0383-8

    View details for PubMedID 24241507

  • Rare Variants in PLXNA4 and Parkinson's Disease PLOS ONE Schulte, E. C., Stahl, I., Czamara, D., Ellwanger, D. C., Eck, S., Graf, E., Mollenhauer, B., Zimprich, A., Lichtner, P., Haubenberger, D., Pirker, W., Bruecke, T., Bereznai, B., Molnar, M. J., Peters, A., Gieger, C., Mueller-Myhsok, B., Trenkwalder, C., Winkelmann, J. 2013; 8 (11)
  • Profibrotic Role of miR-154 in Pulmonary Fibrosis AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY Milosevic, J., Pandit, K., Magister, M., Rabinovich, E., Ellwanger, D. C., Yu, G., Vuga, L. J., Weksler, B., Benos, P. V., Gibson, K. F., McMillan, M., Kahn, M., Kaminski, N. 2012; 47 (6): 879-887


    In this study, we explored the regulation and the role of up-regulated microRNAs in idiopathic pulmonary fibrosis (IPF), a progressive interstitial lung disease of unknown origin. We analyzed the expression of microRNAs in IPF lungs and identified 43 significantly up-regulated microRNAs. Twenty-four of the 43 increased microRNAs were localized to the chromosome 14q32 microRNA cluster. We validated the increased expression of miR-154, miR-134, miR-299-5p, miR-410, miR-382, miR-409-3p, miR-487b, miR-31, and miR-127 by quantitative RT-PCR and determined that they were similarly expressed in embryonic lungs. We did not find evidence for differential methylation in this region, but analysis of transcription factor binding sites identified multiple SMAD3-binding elements in the 14q32 microRNA cluster. TGF-β1 stimulation of normal human lung fibroblasts (NHLF) caused up-regulation of microRNAs on chr14q32 that were also increased in IPF lungs. Chromatin immunoprecipitation confirmed binding of SMAD3 to the putative promoter of miR-154. Mir-154 was increased in IPF fibroblasts, and transfection of NHLF with miR-154 caused significant increases in cell proliferation and migration. The increase in proliferation induced by TGF-β was not observed when NHLF or IPF fibroblasts were transfected with a mir-154 inhibitor. Transfection with miR-154 caused activation of the WNT pathway in NHLF. ICG-001 and XAV939, inhibitors of the WNT/β-catenin pathway, reduced the proliferative effect of miR-154. The potential role of miR-154, one of multiple chr14q32 microRNA cluster members up-regulated in IPF and a regulator of fibroblast migration and proliferation, should be further explored in IPF.

    View details for DOI 10.1165/rcmb.2011-0377OC

    View details for Web of Science ID 000314406800019

    View details for PubMedID 23043088

  • Cis-Acting Polymorphisms Affect Complex Traits through Modifications of MicroRNA Regulation Pathways PLOS ONE Arnold, M., Ellwanger, D. C., Hartsperger, M. L., Pfeufer, A., Stuempflen, V. 2012; 7 (5)


    Genome-wide association studies (GWAS) have become an effective tool to map genes and regions contributing to multifactorial human diseases and traits. A comparably small number of variants identified by GWAS are known to have a direct effect on protein structure whereas the majority of variants is thought to exert their moderate influences on the phenotype through regulatory changes in mRNA expression. MicroRNAs (miRNAs) have been identified as powerful posttranscriptional regulators of mRNAs. Binding to their target sites, which are mostly located within the 3'-untranslated region (3'-UTR) of mRNA transcripts, they modulate mRNA expression and stability. Until today almost all human mRNA transcripts are known to harbor at least one miRNA target site with an average of over 20 miRNA target sites per transcript. Among 5,101 GWAS-identified sentinel single nucleotide polymorphisms (SNPs) that correspond to 18,884 SNPs in linkage disequilibrium (LD) with the sentinels (r2 ≥ 0.8) we identified a significant overrepresentation of SNPs that affect the 3'-UTR of genes (OR = 2.33, 95% CI = 2.12-2.57, P < 10(-52)). This effect was even stronger considering all SNPs in one LD bin a single signal (OR = 4.27, 95% CI = 3.84-4.74, P < 10(-114)). Based on crosslinking immunoprecipitation data we identified four mechanisms affecting miRNA regulation by 3'-UTR mutations: (i) deletion or (ii) creation of miRNA recognition elements within validated RNA-induced silencing complex binding sites, (iii) alteration of 3'-UTR splicing leading to a loss of binding sites, and (iv) change of binding affinity due to modifications of 3'-UTR folding. We annotated 53 SNPs of a total of 288 trait-associated 3'-UTR SNPs as mediating at least one of these mechanisms. Using a qualitative systems biology approach, we demonstrate how our findings can be used to support biological interpretation of GWAS results as well as to provide new experimentally testable hypotheses.

    View details for DOI 10.1371/journal.pone.0036694

    View details for Web of Science ID 000305338200038

    View details for PubMedID 22606281

  • The sufficient minimal set of miRNA seed types BIOINFORMATICS Ellwanger, D. C., Buettner, F. A., Mewes, H., Stuempflen, V. 2011; 27 (10): 1346-1350


    Pairing between the target sequence and the 6-8 nt long seed sequence of the miRNA presents the most important feature for miRNA target site prediction. Novel high-throughput technologies such as Argonaute HITS-CLIP afford meanwhile a detailed study of miRNA:mRNA duplices. These interaction maps enable a first discrimination between functional and non-functional target sites in a bulky fashion. Prediction algorithms apply different seed paradigms to identify miRNA target sites. Therefore, a quantitative assessment of miRNA target site prediction is of major interest.We identified a set of canonical seed types based on a transcriptome wide analysis of experimentally verified functional target sites. We confirmed the specificity of long seeds but we found that the majority of functional target sites are formed by less specific seeds of only 6 nt indicating a crucial role of this type. A substantial fraction of genuine target sites arenon-conserved. Moreover, the majority of functional sites remain uncovered by common prediction methods.

    View details for DOI 10.1093/bioinformatics/btr149

    View details for Web of Science ID 000290334500002

    View details for PubMedID 21441577

    View details for PubMedCentralID PMC3087955