Mechanisms Of Cell Fate Decisions Research Articles

Uncovering underlying physical principles and driving forces of cell differentiation and reprogramming from single-cell transcriptomics

Recent advances in single-cell sequencing technology have revolutionized our ability to acquire whole transcriptome data. However, uncovering the underlying transcriptional drivers and nonequilibrium driving forces of cell function directly from these data remains challenging. We address this by learning cell state vector fields from discrete single-cell RNA velocity to quantify the single-cell global nonequilibrium driving forces as landscape and flux. From single-cell data, we quantified the Waddington landscape, showing that optimal paths for differentiation and reprogramming deviate from the naively expected landscape gradient paths and may not pass through landscape saddles at finite fluctuations, challenging conventional transition state estimation of kinetic rate for cell fate decisions due to the presence of the flux. A key insight from our study is that stem/progenitor cells necessitate greater energy dissipation for rapid cell cycles and self-renewal, maintaining pluripotency. We predict optimal developmental pathways and elucidate the nucleation mechanism of cell fate decisions, with transition states as nucleation sites and pioneer genes as nucleation seeds. The concept of loop flux quantifies the contributions of each cycle flux to cell state transitions, facilitating the understanding of cell dynamics and thermodynamic cost, and providing insights into optimizing biological functions. We also infer cell-cell interactions and cell-type-specific gene regulatory networks, encompassing feedback mechanisms and interaction intensities, predicting genetic perturbation effects on cell fate decisions from single-cell omics data. Essentially, our methodology validates the landscape and flux theory, along with its associated quantifications, offering a framework for exploring the physical principles underlying cellular differentiation and reprogramming and broader biological processes through high-throughput single-cell sequencing experiments.

Deciphering driver regulators of cell fate decisions from single-cell transcriptomics data with CEFCON

Single-cell technologies enable the dynamic analyses of cell fate mapping. However, capturing the gene regulatory relationships and identifying the driver factors that control cell fate decisions are still challenging. We present CEFCON, a network-based framework that first uses a graph neural network with attention mechanism to infer a cell-lineage-specific gene regulatory network (GRN) from single-cell RNA-sequencing data, and then models cell fate dynamics through network control theory to identify driver regulators and the associated gene modules, revealing their critical biological processes related to cell states. Extensive benchmarking tests consistently demonstrated the superiority of CEFCON in GRN construction, driver regulator identification, and gene module identification over baseline methods. When applied to the mouse hematopoietic stem cell differentiation data, CEFCON successfully identified driver regulators for three developmental lineages, which offered useful insights into their differentiation from a network control perspective. Overall, CEFCON provides a valuable tool for studying the underlying mechanisms of cell fate decisions from single-cell RNA-seq data.

Single-cell multi-omics profiling links dynamic DNA methylation to cell fate decisions during mouse early organogenesis

BackgroundPerturbation of DNA methyltransferases (DNMTs) and of the active DNA demethylation pathway via ten-eleven translocation (TET) methylcytosine dioxygenases results in severe developmental defects and embryonic lethality. Dynamic control of DNA methylation is therefore vital for embryogenesis, yet the underlying mechanisms remain poorly understood.ResultsHere we report a single-cell transcriptomic atlas from Dnmt and Tet mutant mouse embryos during early organogenesis. We show that both the maintenance and de novo methyltransferase enzymes are dispensable for the formation of all major cell types at E8.5. However, DNA methyltransferases are required for silencing of prior or alternative cell fates such as pluripotency and extraembryonic programmes. Deletion of all three TET enzymes produces substantial lineage biases, in particular, a failure to generate primitive erythrocytes. Single-cell multi-omics profiling moreover reveals that this is linked to a failure to demethylate distal regulatory elements in Tet triple-knockout embryos.ConclusionsThis study provides a detailed analysis of the effects of perturbing DNA methylation on mouse organogenesis at a whole organism scale and affords new insights into the regulatory mechanisms of cell fate decisions.

An ECHO of Cartilage: In Silico Prediction of Combinatorial Treatments to Switch Between Transient and Permanent Cartilage Phenotypes With Ex Vivo Validation.

A fundamental question in cartilage biology is: what determines the switch between permanent cartilage found in the articular joints and transient hypertrophic cartilage that functions as a template for bone? This switch is observed both in a subset of OA patients that develop osteophytes, as well as in cell-based tissue engineering strategies for joint repair. A thorough understanding of the mechanisms regulating cell fate provides opportunities for treatment of cartilage disease and tissue engineering strategies. The objective of this study was to understand the mechanisms that regulate the switch between permanent and transient cartilage using a computational model of chondrocytes, ECHO. To investigate large signaling networks that regulate cell fate decisions, we developed the software tool ANIMO, Analysis of Networks with interactive Modeling. In ANIMO, we generated an activity network integrating 7 signal transduction pathways resulting in a network containing over 50 proteins with 200 interactions. We called this model ECHO, for executable chondrocyte. Previously, we showed that ECHO could be used to characterize mechanisms of cell fate decisions. ECHO was first developed based on a Boolean model of growth plate. Here, we show how the growth plate Boolean model was translated to ANIMO and how we adapted the topology and parameters to generate an articular cartilage model. In ANIMO, many combinations of overactivation/knockout were tested that result in a switch between permanent cartilage (SOX9+) and transient, hypertrophic cartilage (RUNX2+). We used model checking to prioritize combination treatments for wet-lab validation. Three combinatorial treatments were chosen and tested on metatarsals from 1-day old rat pups that were treated for 6 days. We found that a combination of IGF1 with inhibition of ERK1/2 had a positive effect on cartilage formation and growth, whereas activation of DLX5 combined with inhibition of PKA had a negative effect on cartilage formation and growth and resulted in increased cartilage hypertrophy. We show that our model describes cartilage formation, and that model checking can aid in choosing and prioritizing combinatorial treatments that interfere with normal cartilage development. Here we show that combinatorial treatments induce changes in the zonal distribution of cartilage, indication possible switches in cell fate. This indicates that simulations in ECHO aid in describing pathologies in which switches between cell fates are observed, such as OA.

Transcription Factors: The Fulcrum Between Cell Development and Carcinogenesis.

Higher eukaryotic development is a complex and tightly regulated process, whereby transcription factors (TFs) play a key role in controlling the gene regulatory networks. Dysregulation of these regulatory networks has also been associated with carcinogenesis. Transcription factors are key enablers of cancer stemness, which support the maintenance and function of cancer stem cells that are believed to act as seeds for cancer initiation, progression and metastasis, and treatment resistance. One key area of research is to understand how these factors interact and collaborate to define cellular fate during embryogenesis as well as during tumor development. This review focuses on understanding the role of TFs in cell development and cancer. The molecular mechanisms of cell fate decision are of key importance in efforts towards developing better protocols for directed differentiation of cells in research and medicine. We also discuss the dysregulation of TFs and their role in cancer progression and metastasis, exploring TF networks as direct or indirect targets for therapeutic intervention, as well as specific TFs’ potential as biomarkers for predicting and monitoring treatment responses.

Cell-to-cell diversification in ERBB-RAS-MAPK signal transduction that produces cell-type specific growth factor responses

Growth factors regulate cell fates, including their proliferation, differentiation, survival, and death, according to the cell type. Even when the response to a specific growth factor is deterministic for collective cell behavior, significant levels of fluctuation are often observed between single cells. Statistical analyses of single-cell responses provide insights into the mechanism of cell fate decisions but very little is known about the distributions of the internal states of cells responding to growth factors. Using multi-color immunofluorescent staining, we have here detected the phosphorylation of seven elements in the early response of the ERBB–RAS–MAPK system to two growth factors. Among these seven elements, five were analyzed simultaneously in distinct combinations in the same single cells. Although principle component analysis suggested cell-type and input specific phosphorylation patterns, cell-to-cell fluctuation was large. Mutual information analysis suggested that each cell type uses multitrack (bush-like) signal transduction pathways under conditions in which clear fate changes have been reported. The clustering of single-cell response patterns indicated that the fate change in a cell population correlates with the large entropy of the response, suggesting a bet-hedging strategy is used in decision making. A comparison of true and randomized datasets further indicated that this large variation is not produced by simple reaction noise, but is defined by the properties of the signal-processing network.

In Vivo Pre-Instructed HSCs Robustly Execute Asymmetric Cell Divisions In Vitro.

Hematopoietic stem cells (HSCs) are responsible for life-long production of all mature blood cells. Under homeostasis, HSCs in their native bone marrow niches are believed to undergo asymmetric cell divisions (ACDs), with one daughter cell maintaining HSC identity and the other committing to differentiate into various mature blood cell types. Due to the lack of key niche signals, in vitro HSCs differentiate rapidly, making it challenging to capture and study ACD. To overcome this bottleneck, in this study, we used interferon alpha (IFNα) treatment to ”pre-instruct” HSC fate directly in their native niche, and then systematically studied the fate of dividing HSCs in vitro at the single cell level via time-lapse analysis, as well as multigene and protein expression analysis. Triggering HSCs’ exit from dormancy via IFNα was found to significantly increase the frequency of asynchronous divisions in paired daughter cells (PDCs). Using single-cell gene expression analyses, we identified 12 asymmetrically expressed genes in PDCs. Subsequent immunocytochemistry analysis showed that at least three of the candidates, i.e., Glut1, JAM3 and HK2, were asymmetrically distributed in PDCs. Functional validation of these observations by colony formation assays highlighted the implication of asymmetric distribution of these markers as hallmarks of HSCs, for example, to reliably discriminate committed and self-renewing daughter cells in dividing HSCs. Our data provided evidence for the importance of in vivo instructions in guiding HSC fate, especially ACD, and shed light on putative molecular players involved in this process. Understanding the mechanisms of cell fate decision making should enable the development of improved HSC expansion protocols for therapeutic applications.

Ferroptosis and endoplasmic reticulum stress

Ferroptosis is a newly identified form of regulated cell death (RCD) in 2012, which is characterized by the iron-dependent accumulation of lipid hydroperoxides to lethal levels. The initiation of ferroptosis is tightly linked with numerous essential biological processes, including iron, lipid, amino acid dysregulation. The endoplasmic reticulum (ER) is a dynamic intracellular organelle, which plays an essential role in synthesis, folding, modification and transport of proteins. Under stress conditions, increased levels of unfolded or misfolded proteins within the ER lumen trigger a condition referred to as “ER stress”. The initial ER stress can protect cell for its survival, but prolonged or severe stress can lead to cell death. Recent studies reveal that ER stress is closely related to the process of ferroptosis. In cancer cells, the ferroptotic agents can induce ER stress and subsequently the ER stress pathways negatively regulate the activation of ferroptosis and result in drug resistance. However, under pathological process, ER stress promotes the activation of ferroptosis. The interconnections between ferroptosis and ER stress have become an important scientific topic in understanding the mechanisms of cell fate decision. This review summarized the regulatory networks of ferroptosis and the relationship between ferroptosis and ER stress. We expected to provide new insights and references for ferroptosis research and the therapeutic strategy of ferroptosis associated diseases.

Inclusion of temporal information in single cell transcriptomics.

Single cell transcriptomics has emerged as a powerful method for dissecting cell type diversity and for understanding mechanisms of cell fate decisions. However, inclusion of temporal information remains challenging, since each cell can be measured only once by sequencing analysis. Here, we discuss recent progress and current efforts towards inclusion of temporal information in single cell transcriptomics. Even from snapshot data, temporal dynamics can be computationally inferred via pseudo-temporal ordering of single cell transcriptomes. Temporal information can also come from analysis of intronic reads or from RNA metabolic labeling, which can provide additional evidence for pseudo-time trajectories and enable more fine-grained analysis of gene regulatory interactions. These approaches measure dynamics on short timescales of hours. Emerging methods for high-throughput lineage tracing now enable information storage over long timescales by using CRISPR/Cas9 to record information in the genome, which can later be read out by sequencing.

Fate decisions mediated by crosstalk of autophagy and apoptosis in mammalian cells.

Autophagy is an important cell activity which is the process of formation of autophagosomes, docking with lysosomes and degradation. The intrinsic pathway of apoptosis involves mitochondrial outer membrane permeabilization (MOMP) and cytochrome c release followed by caspase activation. Many molecules, e.g., Ca2+ and mTOR, and different stresses such as endoplasmic reticulum (ER) stress and nutritional stress take part in these two processes. However, the mechanism of how they work together so as to determine cell fate decisions remains to be clarified. Here, we present a computational model for cell fate decisions based on intertwined dynamics with autophagy and apoptosis involving Ca2+, mTOR, and both ER stress and nutritional stress. In agreement with experimental observations, the model predicts that both Ca2+ and the stresses play critical roles in regulating the choice between autophagy and apoptosis in a combinatorial way. The model presented here might be a good candidate for providing the qualitative mechanism of cell fate decisions mediated by Ca2+, mTOR, and two kinds of stress.

ECHO, the executable CHOndrocyte: A computational model to study articular chondrocytes in health and disease

Computational modeling can be used to investigate complex signaling networks in biology. However, most modeling tools are not suitable for molecular cell biologists with little background in mathematics. We have built a visual-based modeling tool for the investigation of dynamic networks. Here, we describe the development of computational models of cartilage development and osteoarthritis, in which a panel of relevant signaling pathways are integrated. In silico experiments give insight in the role of each of the pathway components and reveal which perturbations may deregulate the basal healthy state of cells and tissues.We used a previously developed computational modeling tool Analysis of Networks with Interactive Modeling (ANIMO) to generate an activity network integrating 7 signal transduction pathways resulting in a network containing over 50 nodes and 200 interactions. We performed in silico experiments to characterize molecular mechanisms of cell fate decisions. The model was used to mimic biological scenarios during cell differentiation using RNA-sequencing data of a variety of stem cell sources as input. In a case-study, we wet-lab-tested the model-derived hypothesis that expression of DKK1 (Dickkopf-1) and FRZB (Frizzled related protein, WNT antagonists) and GREM1 (gremlin 1, BMP antagonist) prevents IL1β (Interleukin 1 beta)-induced MMP (matrix metalloproteinase) expression, thereby preventing cartilage degeneration, at least in the short term. We found that a combination of DKK1, FRZB and GREM1 may play a role in modulating the effects of IL1β induced inflammation in human primary chondrocytes.

Modeling the regulation of p53 activation by HIF-1 upon hypoxia.

As a famous tumor suppressor, p53 is also activated under hypoxic conditions. Hypoxia-inducuble factor1, HIF-1, is involved in the activation of p53 upon hypoxia. However, how p53 is modulated by the HIF-1 pathway to decide cell fate is less understood. In this work, we developed a network model including p53 and HIF-1 pathways to clarify the mechanism of cell fate decision in response to hypoxia. We found that HIF-1α and p53 are activated under different conditions. Under moderate hypoxia, HIF-1α is activated to induce glycolysis or angiogenesis, and promotes partial accumulation of p53 by inducing PNUTS. Under severe hypoxia, p53 rises to high levels due to ATR-dependent stabilization and promotes Mdm2-dependent HIF-1α degradation. As a result, fully activated p53 triggers apoptosis. Of note, competition for p300 between HIF-1α and p53 plays a key role in regulating their transcriptional activities. This work may advance the understanding of the mechanism for p53 regulation by HIF-1 in thehypoxic response.

InVitro Modeling of Human Germ Cell Development Using Pluripotent Stem Cells.

SummaryDue to differences across species, the mechanisms of cell fate decisions determined in mice cannot be readily extrapolated to humans. In this study, we developed a feeder- and xeno-free culture protocol that efficiently induced human pluripotent stem cells (iPSCs) into PLZF+/GPR125+/CD90+ spermatogonium-like cells (SLCs). These SLCs were enriched with key genes in germ cell development such as MVH, DAZL, GFRα1, NANOS3, and DMRT1. In addition, a small fraction of SLCs went through meiosis in vitro to develop into haploid cells. We further demonstrated that this chemically defined induction protocol faithfully recapitulated the features of compromised germ cell development of PSCs with NANOS3 deficiency or iPSC lines established from patients with non-obstructive azoospermia. Taken together, we established a powerful experimental platform to investigate human germ cell development and pathology related to male infertility.

Intra-lineage Fate Decisions Involve Activation of Notch Receptors Basal to the Midbody in Drosophila Sensory Organ Precursor Cells

Notch receptors regulate cell fate decisions during embryogenesis and throughout adult life. In many cell lineages, binary fate decisions are mediated by directional Notch signaling between the two sister cells produced by cell division. How Notch signaling is restricted to sister cells after division to regulate intra-lineage decision is poorly understood. More generally, where ligand-dependent activation of Notch occurs at the cell surface is not known, as methods to detect receptor activation invivo are lacking. In Drosophila pupae, Notch signals during cytokinesis to regulate the intra-lineage pIIa/pIIb decision in the sensory organ lineage. Here, we identify two pools of Notch along the pIIa-pIIb interface, apical and basal to the midbody. Analysis of the dynamics of Notch, Delta, and Neuralized distribution in living pupae suggests that ligand endocytosis and receptor activation occur basal to the midbody. Using selective photo-bleaching of GFP-tagged Notch and photo-tracking of photo-convertible Notch, we show that nuclear Notch is indeed produced by receptors located basal to the midbody. Thus, only a specific subset of receptors, located basal to the midbody, contributes to signaling in pIIa. This is the first invivo characterization of the pool of Notch contributing to signaling. We propose a simple mechanism of cell fate decision based onintra-lineage signaling: ligands and receptors localize during cytokinesis to the new cell-cell interface, thereby ensuring signaling between sister cells, hence intra-lineage fate decision.

Factor graph analysis of live cell-imaging data reveals mechanisms of cell fate decisions.

Cell fate decisions have a strong stochastic component. The identification of the underlying mechanisms therefore requires a rigorous statistical analysis of large ensembles of single cells that were tracked and phenotyped over time. We introduce a probabilistic framework for testing elementary hypotheses on dynamic cell behavior using time-lapse cell-imaging data. Factor graphs, probabilistic graphical models, are used to properly account for cell lineage and cell phenotype information. Our model is applied to time-lapse movies of murine granulocyte-macrophage progenitor (GMP) cells. It decides between competing hypotheses on the mechanisms of their differentiation. Our results theoretically substantiate previous experimental observations that lineage instruction, not selection is the cause for the differentiation of GMP cells into mature monocytes or neutrophil granulocytes. The Matlab source code is available at http://treschgroup.de/Genealogies.html.

The linear interplay of intrinsic and extrinsic noises ensures a high accuracy of cell fate selection in budding yeast.

To gain insights into the mechanisms of cell fate decision in a noisy environment, the effects of intrinsic and extrinsic noises on cell fate are explored at the single cell level. Specifically, we theoretically define the impulse of Cln1/2 as an indication of cell fates. The strong dependence between the impulse of Cln1/2 and cell fates is exhibited. Based on the simulation results, we illustrate that increasing intrinsic fluctuations causes the parallel shift of the separation ratio of Whi5P but that increasing extrinsic fluctuations leads to the mixture of different cell fates. Our quantitative study also suggests that the strengths of intrinsic and extrinsic noises around an approximate linear model can ensure a high accuracy of cell fate selection. Furthermore, this study demonstrates that the selection of cell fates is an entropy-decreasing process. In addition, we reveal that cell fates are significantly correlated with the range of entropy decreases.

Dual Functions of Hypoxia-Inducible Factor 1 Alpha for the Commitment of Mouse Embryonic Stem Cells Toward a Neural Lineage

Embryonic stem (ES) cells are useful for elucidating the molecular mechanisms of cell fate decision in the early development of mammals. It has been shown that aggregate culture of ES cells efficiently induces neuroectoderm differentiation. However, the molecular mechanism that leads to selective neural differentiation in aggregate culture is not fully understood. Here, we demonstrate that the oxygen-sensitive hypoxia-inducible transcription factor, Hif-1α, is an essential regulator for neural commitment of ES cells. We found that a hypoxic environment is spontaneously established in differentiating ES cell aggregates within 3 days, and that this time window coincides with Hif-1α activation. In ES cells in adherent culture under hypoxic conditions, Hif-1α activation was correlated with significantly greater expression of neural progenitor-specific gene Sox1 compared with ES cells in adherent culture under normoxic conditions. In contrast, Hif-1α-depleted ES cell aggregates showed severe reduction in Sox1 expression and maintained high expression of undifferentiated ES cell marker genes and epiblast marker gene Fgf5 on day 4. Notably, chromatin immune precipitation assay and luciferase assay showed that Hif-1α might directly activate Sox1 expression. Of additional importance is our finding that attenuation of Hif-1α resulted in an increase of BMP4, a potent inhibitor of neural differentiation, and led to a high level of phosphorylated Smad1. Thus, our results indicate that Hif-1α acts as a positive regulator of neural commitment by promoting the transition of ES cell differentiation from the epiblast into the neuroectoderm state via direct activation of Sox1 expression and suppressing endogenous BMP signaling.

Transcriptional mechanisms of cell fate decisions revealed by single cell expression profiling

Transcriptional networks regulate cell fate decisions, which occur at the level of individual cells. However, much of what we know about their structure and function comes from studies averaging measurements over large populations of cells, many of which are functionally heterogeneous. Such studies conceal the variability between cells and so prevent us from determining the nature of heterogeneity at the molecular level. In recent years, many protocols and platforms have been developed that allow the high throughput analysis of gene expression in single cells, opening the door to a new era of biology. Here, we discuss the need for single cell gene expression analysis to gain deeper insights into the transcriptional control of cell fate decisions, and consider the insights it has provided so far into transcriptional regulatory networks in development.

Model composition through model reduction: a combined model of CD95 and NF-κB signaling pathways

BackgroundMany mathematical models characterizing mechanisms of cell fate decisions have been constructed recently. Their further study may be impossible without development of methods of model composition, which is complicated by the fact that several models describing the same processes could use different reaction chains or incomparable sets of parameters. Detailed models not supported by sufficient volume of experimental data suffer from non-unique choice of parameter values, non-reproducible results, and difficulty of analysis. Thus, it is necessary to reduce existing models to identify key elements determining their dynamics, and it is also required to design the methods allowing us to combine them.ResultsHere we propose a new approach to model composition, based on reducing several models to the same level of complexity and subsequent combining them together. Firstly, we suggest a set of model reduction tools that can be systematically applied to a given model. Secondly, we suggest a notion of a minimal complexity model. This model is the simplest one that can be obtained from the original model using these tools and still able to approximate experimental data. Thirdly, we propose a strategy for composing the reduced models together. Connection with the detailed model is preserved, which can be advantageous in some applications. A toolbox for model reduction and composition has been implemented as part of the BioUML software and tested on the example of integrating two previously published models of the CD95 (APO-1/Fas) signaling pathways. We show that the reduced models lead to the same dynamical behavior of observable species and the same predictions as in the precursor models. The composite model is able to recapitulate several experimental datasets which were used by the authors of the original models to calibrate them separately, but also has new dynamical properties.ConclusionModel complexity should be comparable to the complexity of the data used to train the model. Systematic application of model reduction methods allows implementing this modeling principle and finding models of minimal complexity compatible with the data. Combining such models is much easier than of precursor models and leads to new model properties and predictions.

Skeletal and Cardiac Myogenesis Accompany Adipogenesis in P19 Embryonal Stem Cells

P19 embryonic carcinoma cells resemble normal embryonic stem (ES) cells. They generate cardiac and skeletal myocytes in response to retinoic acid (RA) or oxytocin (OT). RA treatment followed by exposure to triiodothyronine (T3) and insulin induces ES cells differentiation into adipocytes and skeletomyocytes. On the other hand, OT (10(-7) M) was reported to inhibit 3T3 preadipocyte maturation. The present work was undertaken to determine whether P19 cells have an adipogenic potential that could be affected by OT. Cells were treated with RA (10(-6) M)/T3+insulin (adipogenic protocol) or 10(-7) M OT (cardiomyogenic protocol), and analyzed by polymerase chain reaction, immunotechniques, and cytochemistry. Oil-Red-O staining and expression of peroxisome proliferator-activated receptor-gamma (PPARgamma) and aP2 indicated the generation of adipocytes in cultures submitted to the adipogenic protocol. Contracting cells were also generated. Cells positive for sarcomeric actinin and negative for cardiac troponin inhibitor (cTpnI) indicated generation of skeletomyocytes, and cTpnI positive cells revealed generation of cardiomyocytes. Levels of cTpnI and of the skeletal marker MyoD were almost similar in both protocols, whereas no Oil-Red-O staining was associated with the cardiomyogenic protocol. Addition of 10(-7) M OT to the adipogenic protocol did not affect Oil-Red-O staining and PPARgamma expression. Interestingly, Oct3/4 pluripotency marker disappeared in the adipogenic protocol but remained expressed in the cardiomyogenic one. P19 cells thus have an adipogenic potential non affected by 10(-7) M OT. RA/T3+insulin combination generates a larger spectrum of mesodermal cell derivatives and is a more potent morphogenic treatment than OT. P19 cells could help investigating mechanisms of cell fate decision during development.