Facilitating clinically relevant skin tumor diagnostics with spectroscopy-driven machine learning.

In the dawning era of artificial intelligence (AI), health care stands to undergo a significant transformation with the increasing digitalization of patient data. Digital imaging, in particular, will serve as an important platform for AI to aid decision making and diagnostics. A growing number of studies demonstrate the potential of automatic pre-surgical skin tumor delineation, which could have tremendous impact on clinical practice. However, current methods rely on having ground truth images in which tumor borders are already identified, which is not clinically possible. We report a novel approach where hyperspectral images provide spectra from small regions representing healthy tissue and tumor, which are used to generate prediction maps using artificial neural networks (ANNs), after which a segmentation algorithm automatically identifies the tumor borders. This circumvents the need for ground truth images, since an ANN model is trained with data from each individual patient, representing a more clinically relevant approach.

T-cell commitment inheritance-an agent-based multi-scale model.

T-cell development provides an excellent model system for studying lineage commitment from a multipotent progenitor. The intrathymic development process has been thoroughly studied. The molecular circuitry controlling it has been dissected and the necessary steps like programmed shut off of progenitor genes and T-cell genes upregulation have been revealed. However, the exact timing between decision-making and commitment stage remains unexplored. To this end, we implemented an agent-based multi-scale model to investigate inheritance in early T-cell development. Treating each cell as an agent provides a powerful tool as it tracks each individual cell of a simulated T-cell colony, enabling the construction of lineage trees. Based on the lineage trees, we introduce the concept of the last common ancestors (LCA) of committed cells and analyse their relations, both at single-cell level and population level. In addition to simulating wild-type development, we also conduct knockdown analysis. Our simulations predicted that the commitment is a three-step process that occurs on average over several cell generations once a cell is first prepared by a transcriptional switch. This is followed by the loss of the Bcl11b-opposing function approximately two to three generations later. This is when our LCA analysis indicates that the decision to commit is taken even though in general another one to two generations elapse before the cell actually becomes committed by transitioning to the DN2b state. Our results showed that there is decision inheritance in the commitment mechanism.

T-cell commitment inheritance - an agent-based multi-scale model.

T-cell development provides an excellent model system for studying lineage commitment from a multipotent progenitor. The intrathymic development process has been thoroughly studied. The molecular circuitry controlling it has been dissected and the necessary steps like programmed shut off of progenitor genes and T-cell genes upregulation have been revealed. However, the exact timing between decision-making and commitment stage remains unexplored. To this end, we implemented an agent-based multi-scale model to investigate inheritance in early T-cell development. Treating each cell as an agent provides a powerful tool as it tracks each individual cell of a simulated T-cell colony, enabling the construction of lineage trees. Based on the lineage trees, we introduce the concept of the last common ancestors (LCA) of committed cells and analyse their relations, both at single-cell level and population level. In addition to simulating wild-type development, we also conduct knockdown analysis. Our simulations showed that the commitment is a three-step process over several cell generations where a cell is first prepared by a transcriptional switch. This is followed by the loss of the Bcl11b-opposing function two to three generations later which is when the decision to commit is taken. Finally, after another one to two generations, the cell becomes committed by transitioning to the DN2b state. Our results showed that there is inheritance in the commitment mechanism.

B1 SINE-binding ZFP266 impedes mouse iPSC generation through suppression of chromatin opening mediated by reprogramming factors.

Induced pluripotent stem cell (iPSC) reprogramming is inefficient and understanding the molecular mechanisms underlying this inefficiency holds the key to successfully control cellular identity. Here, we report 24 reprogramming roadblock genes identified by CRISPR/Cas9-mediated genome-wide knockout (KO) screening. Of these, depletion of the predicted KRAB zinc finger protein (KRAB-ZFP) Zfp266 strongly and consistently enhances murine iPSC generation in several reprogramming settings, emerging as the most robust roadblock. We show that ZFP266 binds Short Interspersed Nuclear Elements (SINEs) adjacent to binding sites of pioneering factors, OCT4 (POU5F1), SOX2, and KLF4, and impedes chromatin opening. Replacing the KRAB co-suppressor with co-activator domains converts ZFP266 from an inhibitor to a potent facilitator of iPSC reprogramming. We propose that the SINE-KRAB-ZFP interaction is a critical regulator of chromatin accessibility at regulatory elements required for efficient cellular identity changes. In addition, this work serves as a resource to further illuminate molecular mechanisms hindering reprogramming.

CELLoGeNe - An energy landscape framework for logical networks controlling cell decisions.

Experimental and computational efforts are constantly made to elucidate mechanisms controlling cell fate decisions during development and reprogramming. One powerful computational method is to consider cell commitment and reprogramming as movements in an energy landscape. Here, we develop Computation of Energy Landscapes of Logical Gene Networks (CELLoGeNe), which maps Boolean implementation of gene regulatory networks (GRNs) into energy landscapes. CELLoGeNe removes inadvertent symmetries in the energy landscapes normally arising from standard Boolean operators. Furthermore, CELLoGeNe provides tools to visualize and stochastically analyze the shapes of multi-dimensional energy landscapes corresponding to epigenetic landscapes for development and reprogramming. We demonstrate CELLoGeNe on two GRNs governing different aspects of induced pluripotent stem cells, identifying experimentally validated attractors and revealing potential reprogramming roadblocks. CELLoGeNe is a general framework that can be applied to various biological systems offering a broad picture of intracellular dynamics otherwise inaccessible with existing methods.

Neural tube patterning: from a minimal model for rostrocaudal patterning toward an integrated 3D model.

Rostrocaudal patterning of the neural tube is a defining event in vertebrate brain development. This process is driven by morphogen gradients which specify the fate of neural progenitor cells, leading to the partitioning of the tube. Although this is extensively studied experimentally, an integrated view of the genetic circuitry is lacking. Here, we present a minimal gene regulatory model for rostrocaudal patterning, whose tristable topology was determined in a data-driven way. Using this model, we identified the repression of hindbrain fate as promising strategy for the improvement of current protocols for the generation of dopaminergic neurons. Furthermore, we combined our model with an established minimal model for dorsoventral patterning on a realistic 3D neural tube and found that key features of neural tube patterning could be recapitulated. Doing so, we demonstrate how data and models from different sources can be combined to simulate complex in vivo processes.

Multi-scale Dynamical Modeling of T Cell Development from an Early Thymic Progenitor State to Lineage Commitment.

Intrathymic development of committed progenitor (pro)-T cells from multipotent hematopoietic precursors offers an opportunity to dissect the molecular circuitry establishing cell identity in response to environmental signals. This transition encompasses programmed shutoff of stem/progenitor genes, upregulation of T cell specification genes, proliferation, and ultimately commitment. To explain these features in light of reported cis-acting chromatin effects and experimental kinetic data, we develop a three-level dynamic model of commitment based upon regulation of the commitment-linked gene Bcl11b. The levels are (1) a core gene regulatory network (GRN) architecture from transcription factor (TF) perturbation data, (2) a stochastically controlled chromatin-state gate, and (3) a single-cell proliferation model validated by experimental clonal growth and commitment kinetic assays. Using RNA fluorescence in situ hybridization (FISH) measurements of genes encoding key TFs and measured bulk population dynamics, this single-cell model predicts state-switching kinetics validated by measured clonal proliferation and commitment times. The resulting multi-scale model provides a mechanistic framework for dissecting commitment dynamics.

A quantitative model of cellular decision making in direct neuronal reprogramming.

The direct reprogramming of adult skin fibroblasts to neurons is thought to be controlled by a small set of interacting gene regulators. Here, we investigate how the interaction dynamics between these regulating factors coordinate cellular decision making in direct neuronal reprogramming. We put forward a quantitative model of the governing gene regulatory system, supported by measurements of mRNA expression. We found that nPTB needs to feed back into the direct neural conversion network most likely via PTB in order to accurately capture quantitative gene interaction dynamics and correctly predict the outcome of various overexpression and knockdown experiments. This was experimentally validated by nPTB knockdown leading to successful neural conversion. We also proposed a novel analytical technique to dissect system behaviour and reveal the influence of individual factors on resulting gene expression. Overall, we demonstrate that computational analysis is a powerful tool for understanding the mechanisms of direct (neuronal) reprogramming, paving the way for future models that can help improve cell conversion strategies.

Kinetic models of hematopoietic differentiation.

As cell and molecular biology is becoming increasingly quantitative, there is an upsurge of interest in mechanistic modeling at different levels of resolution. Such models mostly concern kinetics and include gene and protein interactions as well as cell population dynamics. The final goal of these models is to provide experimental predictions, which is now taking on. However, even without matured predictions, kinetic models serve the purpose of compressing a plurality of experimental results into something that can empower the data interpretation, and importantly, suggesting new experiments by turning "knobs" in silico. Once formulated, kinetic models can be executed in terms of molecular rate equations for concentrations or by stochastic simulations when only a limited number of copies are involved. Developmental processes, in particular those of stem and progenitor cell commitments, are not only topical but also particularly suitable for kinetic modeling due to the finite number of key genes involved in cellular decisions. Stem and progenitor cell commitment processes have been subject to intense experimental studies over the last decade with some emphasis on embryonic and hematopoietic stem cells. Gene and protein interactions governing these processes can be modeled by binary Boolean rules or by continuous-valued models with interactions set by binding strengths. Conceptual insights along with tested predictions have emerged from such kinetic models. Here we review kinetic modeling efforts applied to stem cell developmental systems with focus on hematopoiesis. We highlight the future challenges including multi-scale models integrating cell dynamical and transcriptional models. This article is categorized under: Models of Systems Properties and Processes > Mechanistic Models Developmental Biology > Stem Cell Biology and Regeneration.

A deterministic method for estimating free energy genetic network landscapes with applications to cell commitment and reprogramming paths.

Depicting developmental processes as movements in free energy genetic landscapes is an illustrative tool. However, exploring such landscapes to obtain quantitative or even qualitative predictions is hampered by the lack of free energy functions corresponding to the biochemical Michaelis-Menten or Hill rate equations for the dynamics. Being armed with energy landscapes defined by a network and its interactions would open up the possibility of swiftly identifying cell states and computing optimal paths, including those of cell reprogramming, thereby avoiding exhaustive trial-and-error simulations with rate equations for different parameter sets. It turns out that sigmoidal rate equations do have approximate free energy associations. With this replacement of rate equations, we develop a deterministic method for estimating the free energy surfaces of systems of interacting genes at different noise levels or temperatures. Once such free energy landscape estimates have been established, we adapt a shortest path algorithm to determine optimal routes in the landscapes. We explore the method on three circuits for haematopoiesis and embryonic stem cell development for commitment and reprogramming scenarios and illustrate how the method can be used to determine sequential steps for onsets of external factors, essential for efficient reprogramming.

Different reprogramming propensities in plants and mammals: Are small variations in the core network wirings responsible?

Although the plant and animal kingdoms were separated more than 1,6 billion years ago, multicellular development is for both guided by similar transcriptional, epigenetic and posttranscriptional machinery. One may ask to what extent there are similarities and differences in the gene regulation circuits and their dynamics when it comes to important processes like stem cell regulation. The key players in mouse embryonic stem cells governing pluripotency versus differentiation are Oct4, Sox2 and Nanog. Correspondingly, the WUSCHEL and CLAVATA3 genes represent a core in the Shoot Apical Meristem regulation for plants. In addition, both systems have designated genes that turn on differentiation. There is very little molecular homology between mammals and plants for these core regulators. Here, we focus on functional homologies by performing a comparison between the circuitry connecting these players in plants and animals and find striking similarities, suggesting that comparable regulatory logics have been evolved for stem cell regulation in both kingdoms. From in silico simulations we find similar differentiation dynamics. Further when in the differentiated state, the cells are capable of regaining the stem cell state. We find that the propensity for this is higher for plants as compared to mammalians. Our investigation suggests that, despite similarity in core regulatory networks, the dynamics of these can contribute to plant cells being more plastic than mammalian cells, i.e. capable to reorganize from single differentiated cells to whole plants-reprogramming. The presence of an incoherent feed-forward loop in the mammalian core circuitry could be the origin of the different reprogramming behaviour.

Nanog, Oct4 and Tet1 interplay in establishing pluripotency.

A few central transcription factors inside mouse embryonic stem (ES) cells and induced pluripotent stem (iPS) cells are believed to control the cells' pluripotency. Characterizations of pluripotent state were put forward on both transcription factor and epigenetic levels. Whereas core players have been identified, it is desirable to map out gene regulatory networks which govern the reprogramming of somatic cells as well as the early developmental decisions. Here we propose a multiple level model where the regulatory network of Oct4, Nanog and Tet1 includes positive feedback loops involving DNA-demethylation around the promoters of Oct4 and Tet1. We put forward a mechanistic understanding of the regulatory dynamics which account for i) Oct4 overexpression is sufficient to induce pluripotency in somatic cell types expressing the other Yamanaka reprogramming factors endogenously; ii) Tet1 can replace Oct4 in reprogramming cocktail; iii) Nanog is not necessary for reprogramming however its over-expression leads to enhanced self-renewal; iv) DNA methylation is the key to the regulation of pluripotency genes; v) Lif withdrawal leads to loss of pluripotency. Overall, our paper proposes a novel framework combining transcription regulation with DNA methylation modifications which, takes into account the multi-layer nature of regulatory mechanisms governing pluripotency acquisition through reprogramming.

A somatic mutation of GFI1B identified in leukemia alters cell fate via a SPI1 (PU.1) centered genetic regulatory network.

We identify a mutation (D262N) in the erythroid-affiliated transcriptional repressor GFI1B, in an acute myeloid leukemia (AML) patient with antecedent myelodysplastic syndrome (MDS). The GFI1B-D262N mutant functionally antagonizes the transcriptional activity of wild-type GFI1B. GFI1B-D262N promoted myelomonocytic versus erythroid output from primary human hematopoietic precursors and enhanced cell survival of both normal and MDS derived precursors. Re-analysis of AML transcriptome data identifies a distinct group of patients in whom expression of wild-type GFI1B and SPI1 (PU.1) have an inverse pattern. In delineating this GFI1B-SPI1 relationship we show that (i) SPI1 is a direct target of GFI1B, (ii) expression of GFI1B-D262N produces elevated expression of SPI1, and (iii) SPI1-knockdown restores balanced lineage output from GFI1B-D262N-expressing precursors. These results table the SPI1-GFI1B transcriptional network as an important regulatory axis in AML as well as in the development of erythroid versus myelomonocytic cell fate.

Dynamic analysis of gene expression and genome-wide transcription factor binding during lineage specification of multipotent progenitors.

We used the paradigmatic GATA-PU.1 axis to explore, at the systems level, dynamic relationships between transcription factor (TF) binding and global gene expression programs as multipotent cells differentiate. We combined global ChIP-seq of GATA1, GATA2, and PU.1 with expression profiling during differentiation to erythroid and neutrophil lineages. Our analysis reveals (1) differential complexity of sequence motifs bound by GATA1, GATA2, and PU.1; (2) the scope and interplay of GATA1 and GATA2 programs within, and during transitions between, different cell compartments, and the extent of their hard-wiring by DNA motifs; (3) the potential to predict gene expression trajectories based on global associations between TF-binding data and target gene expression; and (4) how dynamic modeling of DNA-binding and gene expression data can be used to infer regulatory logic of TF circuitry. This rubric exemplifies the utility of this cross-platform resource for deconvoluting the complexity of transcriptional programs controlling stem/progenitor cell fate in hematopoiesis.

Probing the role of stochasticity in a model of the embryonic stem cell: heterogeneous gene expression and reprogramming efficiency.

BACKGROUND: Embryonic stem cells (ESC) have the capacity to self-renew and remain pluripotent, while continuously providing a source of a variety of differentiated cell types. Understanding what governs these properties at the molecular level is crucial for stem cell biology and its application to regenerative medicine. Of particular relevance is to elucidate those molecular interactions which govern the reprogramming of somatic cells into ESC. A computational approach can be used as a framework to explore the dynamics of a simplified network of the ESC with the aim to understand how stem cells differentiate and also how they can be reprogrammed from somatic cells. RESULTS: We propose a computational model of the embryonic stem cell network, in which a core set of transcription factors (TFs) interact with each other and are induced by external factors. A stochastic treatment of the network dynamics suggests that NANOG heterogeneity is the deciding factor for the stem cell fate. In particular, our results show that the decision of staying in the ground state or commitment to a differentiated state is fundamentally stochastic, and can be modulated by the addition of external factors (2i/3i media), which have the effect of reducing fluctuations in NANOG expression. Our model also hosts reprogramming of a committed cell into an ESC by over-expressing OCT4. In this context, we recapitulate the important experimental result that reprogramming efficiency peaks when OCT4 is over-expressed within a specific range of values. CONCLUSIONS: We have demonstrated how a stochastic computational model based upon a simplified network of TFs in ESCs can elucidate several key observed dynamical features. It accounts for (i) the observed heterogeneity of key regulators, (ii) characterizes the ESC under certain external stimuli conditions and (iii) describes the occurrence of transitions from the ESC to the differentiated state. Furthermore, the model (iv) provides a framework for reprogramming from somatic cells and conveys an understanding of reprogramming efficiency as a function of OCT4 over-expression.

Prepatterning in the stem cell compartment.

The mechanism by which an apparently uniform population of cells can generate a heterogeneous population of differentiated derivatives is a fundamental aspect of pluripotent and multipotent stem cell behaviour. One possibility is that the environment and the differentiation cues to which the cells are exposed are not uniform. An alternative, but not mutually exclusive possibility is that the observed heterogeneity arises from the stem cells themselves through the existence of different interconvertible substates that pre-exist before the cells commit to differentiate. We have tested this hypothesis in the case of apparently homogeneous pluripotent human embryonal carcinoma (EC) stem cells, which do not follow a uniform pattern of differentiation when exposed to retinoic acid. Instead, they produce differentiated progeny that include both neuronal and non-neural phenotypes. Our results suggest that pluripotent NTERA2 stem cells oscillate between functionally distinct substates that are primed to select distinct lineages when differentiation is induced.

Modeling the evolution of culture-adapted human embryonic stem cells.

The long-term culture of human embryonic stem (ES) cells is inevitably subject to evolution, since any mutant that arises with a growth advantage will be selectively amplified. However, the evolutionary influences of population size, mutation rate, and selection pressure are frequently overlooked. We have constructed a Monte Carlo simulation model to predict how changes in these factors can influence the appearance and spread of mutant ES cells, and verified its applicability by comparison with in vitro data. This simulation provides an estimate for the expected rate of generation of culture-adapted ES cells under different assumptions for the key parameters. In particular, it highlights the effect of population size, suggesting that the maintenance of cells in small populations reduces the likelihood that abnormal cultures will develop.

Modified variational Bayes EM estimation of hidden Markov tree model of cell lineages.

MOTIVATION: Human pluripotent stem cell lines persist in culture as a heterogeneous population of SSEA3 positive and SSEA3 negative cells. Tracking individual stem cells in real time can elucidate the kinetics of cells switching between the SSEA3 positive and negative substates. However, identifying a cell's substate at all time points within a cell lineage tree is technically difficult. RESULTS: A variational Bayesian Expectation Maximization (EM) with smoothed probabilities (VBEMS) algorithm for hidden Markov trees (HMT) is proposed for incomplete tree structured data. The full posterior of the HMT parameters is determined and the underflow problems associated with previous algorithms are eliminated. Example results for the prediction of the types of cells in synthetic and real stem cell lineage trees are presented. AVAILABILITY: The Matlab code for the VBEMS algorithm is freely available at http://www.acse.dept.shef.ac.uk/repository/vbems_lineage_tree/VBEMS.ZIP CONTACT: visakan@sheffield.ac.uk SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.