Florian Buettner

Computational analysis of cell-to-cell heterogeneity in single-cell RNA-sequencing data reveals hidden subpopulations of cells

Recent technical developments have enabled the transcriptomes of hundreds of cells to be assayed in an unbiased manner, opening up the possibility that new subpopulations of cells can be found. However, the effects of potential confounding factors, such as the cell cycle, on the heterogeneity of gene expression and therefore on the ability to robustly identify subpopulations remain unclear. We present and validate a computational approach that uses latent variable models to account for such hidden factors. We show that our single-cell latent variable model (scLVM) allows the identification of otherwise undetectable subpopulations of cells that correspond to different stages during the differentiation of naive T cells into T helper 2 cells. Our approach can be used not only to identify cellular subpopulations but also to tease apart different sources of gene expression heterogeneity in single-cell transcriptomes.

Characterization of transcriptional networks in blood stem and progenitor cells using high-throughput single-cell gene expression analysis

Cellular decision-making is mediated by a complex interplay of external stimuli with the intracellular environment, in particular transcription factor regulatory networks. Here we have determined the expression of a network of 18 key haematopoietic transcription factors in 597 single primary blood stem and progenitor cells isolated from mouse bone marrow. We demonstrate that different stem/progenitor populations are characterized by distinctive transcription factor expression states, and through comprehensive bioinformatic analysis reveal positively and negatively correlated transcription factor pairings, including previously unrecognized relationships between Gata2, Gfi1 and Gfi1b. Validation using transcriptional and transgenic assays confirmed direct regulatory interactions consistent with a regulatory triad in immature blood stem cells, where Gata2 may function to modulate cross-inhibition between Gfi1 and Gfi1b. Single-cell expression profiling therefore identifies network states and allows reconstruction of network hierarchies involved in controlling stem cell fate choices, and provides a blueprint for studying both normal development and human disease.

Decoding the regulatory network of early blood development from single-cell gene expression measurements

Reconstruction of the molecular pathways controlling organ development has been hampered by a lack of methods to resolve embryonic progenitor cells. Here we describe a strategy to address this problem that combines gene expression profiling of large numbers of single cells with data analysis based on diffusion maps for dimensionality reduction and network synthesis from state transition graphs. Applying the approach to hematopoietic development in the mouse embryo, we map the progression of mesoderm toward blood using single-cell gene expression analysis of 3,934 cells with blood-forming potential captured at four time points between E7.0 and E8.5. Transitions between individual cellular states are then used as input to develop a single-cell network synthesis toolkit to generate a computationally executable transcriptional regulatory network model of blood development. Several model predictions concerning the roles of Sox and Hox factors are validated experimentally. Our results demonstrate that single-cell analysis of a developing organ coupled with computational approaches can reveal the transcriptional programs that underpin organogenesis.

Combined Single-Cell Functional and Gene Expression Analysis Resolves Heterogeneity within Stem Cell Populations

Summary Heterogeneity within the self-renewal durability of adult hematopoietic stem cells (HSCs) challenges our understanding of the molecular framework underlying HSC function. Gene expression studies have been hampered by the presence of multiple HSC subtypes and contaminating non-HSCs in bulk HSC populations. To gain deeper insight into the gene expression program of murine HSCs, we combined single-cell functional assays with flow cytometric index sorting and single-cell gene expression assays. Through bioinformatic integration of these datasets, we designed an unbiased sorting strategy that separates non-HSCs away from HSCs, and single-cell transplantation experiments using the enriched population were combined with RNA-seq data to identify key molecules that associate with long-term durable self-renewal, producing a single-cell molecular dataset that is linked to functional stem cell activity. Finally, we demonstrated the broader applicability of this approach for linking key molecules with defined cellular functions in another stem cell system.

Diffusion maps for high-dimensional single-cell analysis of differentiation data

MOTIVATION Single-cell technologies have recently gained popularity in cellular differentiation studies regarding their ability to resolve potential heterogeneities in cell populations. Analyzing such high-dimensional single-cell data has its own statistical and computational challenges. Popular multivariate approaches are based on data normalization, followed by dimension reduction and clustering to identify subgroups. However, in the case of cellular differentiation, we would not expect clear clusters to be present but instead expect the cells to follow continuous branching lineages. RESULTS Here, we propose the use of diffusion maps to deal with the problem of defining differentiation trajectories. We adapt this method to single-cell data by adequate choice of kernel width and inclusion of uncertainties or missing measurement values, which enables the establishment of a pseudotemporal ordering of single cells in a high-dimensional gene expression space. We expect this output to reflect cell differentiation trajectories, where the data originates from intrinsic diffusion-like dynamics. Starting from a pluripotent stage, cells move smoothly within the transcriptional landscape towards more differentiated states with some stochasticity along their path. We demonstrate the robustness of our method with respect to extrinsic noise (e.g. measurement noise) and sampling density heterogeneities on simulated toy data as well as two single-cell quantitative polymerase chain reaction datasets (i.e. mouse haematopoietic stem cells and mouse embryonic stem cells) and an RNA-Seq data of human pre-implantation embryos. We show that diffusion maps perform considerably better than Principal Component Analysis and are advantageous over other techniques for non-linear dimension reduction such as t-distributed Stochastic Neighbour Embedding for preserving the global structures and pseudotemporal ordering of cells. AVAILABILITY AND IMPLEMENTATION The Matlab implementation of diffusion maps for single-cell data is available at https://www.helmholtz-muenchen.de/icb/single-cell-diffusion-map. CONTACT fbuettner.phys@gmail.com, fabian.theis@helmholtz-muenchen.de SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.

Diffusion pseudotime robustly reconstructs lineage branching

The temporal order of differentiating cells is intrinsically encoded in their single-cell expression profiles. We describe an efficient way to robustly estimate this order according to diffusion pseudotime (DPT), which measures transitions between cells using diffusion-like random walks. Our DPT software implementations make it possible to reconstruct the developmental progression of cells and identify transient or metastable states, branching decisions and differentiation endpoints.

Computational assignment of cell-cycle stage from single-cell transcriptome data.

The transcriptome of single cells can reveal important information about cellular states and heterogeneity within populations of cells. Recently, single-cell RNA-sequencing has facilitated expression profiling of large numbers of single cells in parallel. To fully exploit these data, it is critical that suitable computational approaches are developed. One key challenge, especially pertinent when considering dividing populations of cells, is to understand the cell-cycle stage of each captured cell. Here we describe and compare five established supervised machine learning methods and a custom-built predictor for allocating cells to their cell-cycle stage on the basis of their transcriptome. In particular, we assess the impact of different normalisation strategies and the usage of prior knowledge on the predictive power of the classifiers. We tested the methods on previously published datasets and found that a PCA-based approach and the custom predictor performed best. Moreover, our analysis shows that the performance depends strongly on normalisation and the usage of prior knowledge. Only by leveraging prior knowledge in form of cell-cycle annotated genes and by preprocessing the data using a rank-based normalisation, is it possible to robustly capture the transcriptional cell-cycle signature across different cell types, organisms and experimental protocols.

Effects of High-Dose Oral Insulin on Immune Responses in Children at High Risk for Type 1 Diabetes: The Pre-POINT Randomized Clinical Trial

IMPORTANCE Exposing the oral mucosa to antigen may stimulate immune tolerance. It is unknown whether treatment with oral insulin can induce a tolerogenic immune response in children genetically susceptible to type 1 diabetes. OBJECTIVE To assess the immune responses and adverse events associated with orally administered insulin in autoantibody-negative, genetically at-risk children. DESIGN, SETTING, AND PARTICIPANTS The Pre-POINT study, a double-blind, placebo-controlled, dose-escalation, phase 1/2 clinical pilot study performed between 2009 and 2013 in Germany, Austria, the United States, and the United Kingdom and enrolling 25 islet autoantibody-negative children aged 2 to 7 years with a family history of type 1 diabetes and susceptible human leukocyte antigen class II genotypes. Follow-up was completed in August 2013. INTERVENTIONS Children were randomized to receive oral insulin (n = 15) or placebo (n = 10) once daily for 3 to 18 months. Nine children received insulin with dose escalations from 2.5 to 7.5 mg (n = 3), 2.5 to 22.5 mg (n = 3), or 7.5 to 67.5 mg (n = 3) after 6 months; 6 children only received doses of 22.5 mg (n = 3) or 67.5 mg (n = 3). MAIN OUTCOMES AND MEASURES An immune response to insulin, measured as serum IgG and saliva IgA binding to insulin, and CD4+ T-cell proliferative responses to insulin. RESULTS Increases in IgG binding to insulin, saliva IgA binding to insulin, or CD4+ T-cell proliferative responses to insulin were observed in 2 of 10 (20% [95% CI, 0.1%-45%]) placebo-treated children and in 1 of 6 (16.7% [95% CI, 0.1%-46%]) children treated with 2.5 mg of insulin, 1 of 6 (16.7%[ 95% CI, 0.1%-46%]) treated with 7.5 mg, 2 of 6 (33.3% [95% CI, 0.1%-71%]) treated with 22.5 mg, and 5 of 6 (83.3% [ 95% CI, 53%-99.9%]) treated with 67.5 mg (P = .02). Insulin-responsive T cells displayed regulatory T-cell features after oral insulin treatment. No hypoglycemia, IgE responses to insulin, autoantibodies to glutamic acid decarboxylase or insulinoma-associated antigen 2, or diabetes were observed. Adverse events were reported in 12 insulin-treated children (67 events) and 10 placebo-treated children (35 events). CONCLUSIONS AND RELEVANCE In this pilot study of children at high risk for type 1 diabetes, daily oral administration of 67.5 mg of insulin, compared with placebo, resulted in an immune response without hypoglycemia. These findings support the need for a phase 3 trial to determine whether oral insulin can prevent islet autoimmunity and diabetes in such children. TRIAL REGISTRATION isrctn.org Identifier: ISRCTN76104595.

Single-cell transcriptomic reconstruction reveals cell cycle and multi-lineage differentiation defects in Bcl11a-deficient hematopoietic stem cells

BackgroundHematopoietic stem cells (HSCs) are a rare cell type with the ability of long-term self-renewal and multipotency to reconstitute all blood lineages. HSCs are typically purified from the bone marrow using cell surface markers. Recent studies have identified significant cellular heterogeneities in the HSC compartment with subsets of HSCs displaying lineage bias. We previously discovered that the transcription factor Bcl11a has critical functions in the lymphoid development of the HSC compartment.ResultsIn this report, we employ single-cell transcriptomic analysis to dissect the molecular heterogeneities in HSCs. We profile the transcriptomes of 180 highly purified HSCs (Bcl11a+/+ and Bcl11a−/−). Detailed analysis of the RNA-seq data identifies cell cycle activity as the major source of transcriptomic variation in the HSC compartment, which allows reconstruction of HSC cell cycle progression in silico. Single-cell RNA-seq profiling of Bcl11a−/− HSCs reveals abnormal proliferative phenotypes. Analysis of lineage gene expression suggests that the Bcl11a−/− HSCs are constituted of two distinct myeloerythroid-restricted subpopulations. Remarkably, similar myeloid-restricted cells could also be detected in the wild-type HSC compartment, suggesting selective elimination of lymphoid-competent HSCs after Bcl11a deletion. These defects are experimentally validated in serial transplantation experiments where Bcl11a−/− HSCs are myeloerythroid-restricted and defective in self-renewal.ConclusionsOur study demonstrates the power of single-cell transcriptomics in dissecting cellular process and lineage heterogeneities in stem cell compartments, and further reveals the molecular and cellular defects in the Bcl11a-deficient HSC compartment.

Feature ranking of type 1 diabetes susceptibility genes improves prediction of type 1 diabetes

Aims/hypothesisMore than 40 regions of the human genome confer susceptibility for type 1 diabetes and could be used to establish population screening strategies. The aim of our study was to identify weighted sets of SNP combinations for type 1 diabetes prediction.MethodsWe applied multivariable logistic regression and Bayesian feature selection to the Type 1 Diabetes Genetics Consortium (T1DGC) dataset with genotyping of HLA plus 40 SNPs within other type 1 diabetes-associated gene regions in 4,574 cases and 1,207 controls. We tested the weighted models in an independent validation set (765 cases, 423 controls), and assessed their performance in 1,772 prospectively followed children.ResultsThe inclusion of 40 non-HLA gene SNPs significantly improved the prediction of type 1 diabetes over that provided by HLA alone (p = 3.1 × 10−25), with a receiver operating characteristic AUC of 0.87 in the T1DGC set, and 0.84 in the validation set. Feature selection identified HLA plus nine SNPs from the PTPN22, INS, IL2RA, ERBB3, ORMDL3, BACH2, IL27, GLIS3 and RNLS genes that could achieve similar prediction accuracy as the total SNP set. Application of this ten SNP model to prospectively followed children was able to improve risk stratification over that achieved by HLA genotype alone.ConclusionsWe provided a weighted risk model with selected SNPs that could be considered for recruitment of infants into studies of early type 1 diabetes natural history or appropriately safe prevention.