SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. OVERVIEW OF STEM CELL FATE REGULATION
  4. STEM CELL GENETIC PROGRAMS
  5. STEM CELL MICROENVIRONMENTS
  6. CONCLUSION
  7. REFERENCES

Stem cell functions are regulated by a combination of molecular signals that are provided both intrinsically and from the local microenvironment. The complexity of these mechanisms has encouraged the use of interdisciplinary experimental strategies, such as bioengineering methods, to address fundamental questions in stem cell biology. These approaches have primarily aimed to (1) develop tools for the improved control of microenvironmental cues and genetic perturbations, (2) integrate high-throughput technologies to broaden the experimental state space and facilitate systematic examination of combinatorial signals, and (3) construct systems-based models to better define stem cell processes through an understanding of the interdependence of the individual signaling components. Cooperative advancements in these areas will continue to contribute to the evolution of stem cell-based therapeutics. WIREs Syst Biol Med 2012, 4:525–545. doi: 10.1002/wsbm.1189

For further resources related to this article, please visit the WIREs website.


OVERVIEW OF STEM CELL FATE REGULATION

  1. Top of page
  2. Abstract
  3. OVERVIEW OF STEM CELL FATE REGULATION
  4. STEM CELL GENETIC PROGRAMS
  5. STEM CELL MICROENVIRONMENTS
  6. CONCLUSION
  7. REFERENCES

To realize the full potential of cell-based therapies in regenerative medicine, substantial research efforts are aimed at establishing a more complete knowledge of the regulatory mechanisms governing stem and progenitor cell functions. Advances in this area would form the foundation for optimized cell sourcing methods, improved in vitro tissue development and disease models, and the design of stem cell-targeted therapeutic strategies incorporating novel drug candidates. Central to these endeavors are studies focused on the systematic deconvolution of the complex molecular determinants of stem cell fate. In particular, it is increasingly appreciated that stem cell function is governed by the combined influence of (1) intrinsic characteristics, such as epigenetic modifications which impact cell differentiation potential, and (2) the cellular sensing and integration of numerous chemical and physical regulatory signals present within stem cell microenvironments, or niches. In the last decade, the collective application of high-throughput experimental tools and computational modeling has begun to define both the relevant genetic/epigenetic elements and microenvironmental signals as systems of highly interdependent components. Such analyses can provide a framework for relating expression profiles and perturbations with functional properties, and also reveal critical connections between intracellular and extracellular networks underlying stem cell biology. In this review, I will discuss recent work toward the development of systems-based models of stem cell processes with a focus on emerging high-throughput technologies for measuring and manipulating stem cell signaling networks and environmental interactions.

Since the earliest developments in expression profiling and screening technologies, investigators have sought to exploit high-throughput methods to define the characteristics of stem cells which confer their distinguishing properties. These studies have been instrumental in identifying transcription factors and signaling pathways critical for the self-renewal and pluripotency of embryonic stem (ES) cells as well as the multipotent differentiation capacities of various adult stem cell populations.1,2 In parallel, collaborative efforts in cell biology and bioengineering have led to progressively improved in vitro culture strategies, which have complemented advances in transgenic model systems and in vivo analysis, and served to underscore the importance of microenvironmental signals in stem cell processes.3 This review will discuss seminal studies and recent developments in the understanding of both intrinsic and niche-mediated regulation. First, I will focus on stem cell genetic programs (Figure 1), including key insights into transcription factor–DNA interactions and epigenetic modifications, concepts of gene regulatory networks, and recent applications of computational modeling in stem cell biology. In addition, enabling technologies for the high-throughput analysis of stem cell genetics will be highlighted. Next, I will discuss the role of cellular microenvironments in influencing stem cell fate and function, with a primary emphasis on promising microtechnology approaches for the improved throughput and spatiotemporal control of microenvironmental cues within stem cell culture platforms. Finally, throughout the review I will describe recent efforts which leverage the fundamental insights gained toward the engineering of stem cells for translational applications, and conclude with an overview of future directions and challenges.

thumbnail image

Figure 1. Convergent methods for elucidating stem cell genetic programs. An understanding of the intrinsic signaling and gene regulatory mechanisms underlying stem cell function requires the comprehensive analysis and integration of gene expression profiles, protein–protein interactions, transcription factor–DNA binding localizations, miRNA expression and interaction data, and measurements of epigenetic state. Such studies have been enabled by numerous high-throughput experimental methods, coupled with computational tools (light blue boxes). Emerging technologies and approaches (dark blue boxes) should aid further refinements in regulatory models, including insights at the single cell level, and facilitate an improved capacity for dictating stem cell fate through the controlled manipulation of regulatory programs.

Download figure to PowerPoint

STEM CELL GENETIC PROGRAMS

  1. Top of page
  2. Abstract
  3. OVERVIEW OF STEM CELL FATE REGULATION
  4. STEM CELL GENETIC PROGRAMS
  5. STEM CELL MICROENVIRONMENTS
  6. CONCLUSION
  7. REFERENCES

Gene Expression Analysis and Transcription Factor–DNA Interactions

In order to elucidate the unique gene expression signatures governing stem cell behavior, a broad range of approaches have been utilized, incorporating both DNA microarray and sequencing-based technologies. In general, these studies have emphasized comparisons between pluripotent stem cells, adult stem cells, and their differentiated progeny as a means to evaluate the overall transcriptional changes occurring upon differentiation and to identify potential commonalities between stem cell types. Initial investigations comparing the gene expression profiles of ES cells, hematopoietic stem cells (HSCs), and neural stem cells aimed to formulate a set of ‘stemness’ genes shared among stem cell types that could impart self-renewal and multilineage differentiation potential.4,5 Although individual candidate genes were identified in these studies, it is increasingly appreciated that despite some overlapping expression, pluripotent stem cells and various adult stem cells are likely regulated by distinct pathways,6 and a common stem cell phenotype cannot be fully defined by a discrete subset of genes. Furthermore, findings from the initial genome-wide expression analyses of stem cells highlighted the potential confounding influences of heterogeneities in culture environments or in the purities of primary stem cell populations. Subsequent efforts have focused on controlled experiments with individual stem cell types, typically examining transcriptional alterations occurring upon differentiation induction or other perturbations. For example, a number of studies have investigated early gene expression changes in mouse and human ES cells following manipulation of culture conditions or RNA interference (RNAi)-mediated knockdown of previously implicated transcription factors such as Oct4, Sox2, and Nanog, in order to identify additional components/pathways contributing to pluripotency.7–10 In one such study with mouse ES cells, it was demonstrated that early downregulated genes following the removal of LIF or the addition of retinoic acid were enriched for transcription factors, while upregulated genes included transcriptional regulators as well as genes associated with various developmental programs, which differed based on the stimulus.7 Collectively, as expression profiling approaches continue to improve, concurrent developments in statistical methods for enhanced probability scoring and meta-analysis have been shown to facilitate comparisons between independent studies and across species, and to provide better assessments of reproducibility.11–15 In addition, advancements in gene ontology categorization have greatly aided the functional interpretation of gene expression signatures.16–18

As suggested by expression profiling and numerous genetic loss-of-function studies, transcription factors are central to the regulation of stem cell fates. This important role is not only determined by transcription factor expression, but also the functional interaction of transcription factors with their target genomic sequences. Accordingly, significant research efforts have sought to identify binding sites for stem cell transcription factors and establish the genes that are transcriptionally regulated by these factors. Using chromatin immunoprecipitation (ChIP) methods, the genomic DNA bound by specific transcription factors can be enriched and analyzed in a sequence-specific manner to determine the relative representation of transcription factor–gene interactions.19 The three major technologies currently utilized include ChIP-chip, ChIP-PET, and ChIP-seq, and each has been applied toward the analysis of stem cell transcriptional regulation. ChIP-chip uses DNA microarray hybridization to evaluate bound genomic sequences, and studies employing this approach have offered insights into the cooperative roles of ES cell transcription factors. In particular, Nanog was demonstrated to bind to 60% of the Oct4 or Sox2 bound promoters in human ES cells, and approximately 90% of the loci bound by both Oct4 and Sox2 also included Nanog.20 Comparisons to mouse ES cells suggest that there are differences between species,20 although the importance of the co-localization of numerous transcription factors appears to be conserved. In a study that examined the promoter binding of nine transcription factors in mouse ES cells, and correlated this data with expression profiles, it was illustrated that loci which were bound by at least four transcription factors were predominantly active, while promoters with few factors were often inactive.21 Co-localization has also been highlighted in studies employing ChIP-seq, in which high-throughput next generation sequencing is applied to the immunoprecipitated genomic DNA.22 For instance, two major clusters of transcription factor binding sites were identified in mouse ES cells, with the first cluster containing Oct4, Nanog, Sox2, and exhibiting an enriched representation of Smad1, STAT3, and Esrrb.23 The second major cluster of sites contained c-Myc, n-Myc, Zfx, and E2f1. Notably, the authors demonstrate that the majority of the Oct4/Nanog/Sox2 clusters bind outside of promoter regions and can display enhancer activity. This distal regulation of transcription has also been suggested by studies using a parallel strategy, ChIP-PET, which incorporates a paired-end tag (PET) technology for sequencing analysis of the 5′ and 3′ ends of the genomic DNA segments.24 In addition, combinatorial regulation mediated by complexes of key transcription factors has similarly been identified in adult stem/progenitor cell types, including murine multilineage hematopoietic progenitors and neural stem cells in Drosophila.25,26 Overall, the genomic-scale assessment of gene expression and transcription factor–DNA interactions in stem cells has provided important information regarding transcriptional mechanisms and serves as the essential foundation for models of gene regulatory networks.

Gene Regulatory Networks

Systems analysis of stem cells aims to form integrated networks of the relevant regulatory proteins and genomic sequences which govern stem cell self-renewal and differentiation. These gene regulatory networks are predicted based on the computational analysis of transcription factor binding sites and genomic expression profiles, including time course data. In this process, the functional interactions between the components (i.e., nodes) are defined as network motifs, which represent the fundamental building blocks of the network and are determined by algorithms that collectively evaluate the genome-wide binding data and the expression data.27 Such analyses in ES cells has suggested that the regulatory network in pluripotent stem cells is highly interconnected, and exhibits numerous feedforward and autoregulatory feedback loops.28 Specifically, it has been demonstrated that the transcription factors Oct4, Sox2, and Nanog not only bind sequences in each other's genes, which can provide additional co-regulatory control (feedforward), but also bind their own genes (autoregulation). It is hypothesized that the combination of feedforward and autoregulatory loop motifs provides both input responsiveness and stability characteristics for the network. An understanding of the temporal dynamics of gene expression can substantially aid the refinement of the network model and the interpretation of the mechanistic connections between nodes. For example, time course analysis following Nanog knockdown in mouse ES cells enabled the clustering of genes which exhibit similar temporal patterns, and provided confirmatory evidence of the upstream or downstream relationships.10 In addition, thermodynamic models of the interactions between multiple core ES transcription factors, and with RNA polymerase, have been applied to time course gene expression data, and used to predict the contributions of the individual factors in the cooperative regulation of downstream targets.29 Various computational perturbations have also been utilized to evaluate potential dynamic aspects of stem regulatory networks. In particular, it has been suggested that fluctuations in Nanog expression can significantly influence the susceptibility to differentiation induction, which has been confirmed experimentally.30–32 Furthermore, kinetic modeling of Oct4/Sox2/Nanog interactions has indicated that positive feedback between these components coordinately stabilizes the expression of these genes, with the transcriptional network remaining under the control of a bistable switch which is actuated by external stimuli.33 Such interactions between intrinsic control mechanisms and external signals from stem cell microenvironments are fundamental to stem cell regulation and will be discussed in detail later in this review.

As a means to further explore cooperative transcriptional control and the connections with signal transduction pathways, numerous studies have employed proteomics methods to gain an understanding of protein–protein interactions. In this approach, mass spectrometry analysis of affinity purified proteins is utilized to identify interaction partners. An appreciation of the interaction networks surrounding Oct4 and Nanog at the protein level has contributed to the recognition of additional co-regulatory factors,34–36 and protein interaction data has appropriately correlated with ChIP analysis to illustrate a distinct regulatory module centered around c-Myc in pluripotent stem cells.21,37 In addition to protein–protein interaction studies, measurements of signaling protein phosphorylation have also been performed. For example, large-scale phosphorylation analysis has been utilized to assess the dynamic alterations occurring upon differentiation induction in human ES cells.38,39 Computational tools for relating phosphorylation motifs to specific kinases,39,40 and for building logic interaction circuits,41 have been applied to phosphorylation data in ES cells, and have provided further clues into the network dynamics.

Role of miRNAs and Epigenetic Regulatory Mechanisms

It is increasingly appreciated that additional regulatory mechanisms, such as non-coding miRNAs (miRNAs) and epigenetic modifications of DNA and chromatin structure, play significant roles in determining stem cell characteristics. miRNAs are endogenous 20–24 nucleotide transcripts that do not encode proteins, but act to inhibit translation or direct the degradation of specific mRNA sequences, and therefore serve to fine tune protein expression.42 Numerous miRNAs have been demonstrated to be involved in both embryonic and adult stem cell differentiation, and I refer the reader to two comprehensive reviews on this topic.43,44 In ES cells, important transcription factors including Oct4, Nanog, Sox2, and Tcf3 have been shown to occupy the promoters of miRNA genes and function to activate or repress their expression based on the absence or presence of polycomb group proteins, respectively.45 These data suggest that miRNAs are key components of the pluripotent stem cell regulatory network, which is underscored by recent findings demonstrating the capability of a cluster of miRNAs to reprogram somatic cells to a pluripotent state.46,47 Overall, several important characteristics of miRNAs must be considered toward the integration into gene regulatory networks. Specifically, the effect of a miRNA on a target protein's expression level is usually less than a twofold downregulation,48,49 and thus, numerous independent miRNAs often work cooperatively to mediate the effective downregulation of a target. Furthermore, a single family of miRNAs has been estimated to have an average of 300 conserved targets,50 resulting in an extremely wide interaction network. A large number of computational tools have been developed to assist in miRNA target identification, and in the mapping of miRNAs to biological pathways.51,52 In addition, methods utilizing the affinity purification of proteins within the RNA-induced silencing complex (RISC), with subsequent sequence analysis of RISC-associated mRNA-miRNA duplexes, represent promising strategies for the direct assessment of miRNA targets.53–55

Epigenetic regulation is a process by which gene expression is modulated within a cell by the localized methylation of DNA or through chemical modifications of histones affecting chromatin structure. Substantial research efforts have led to an emerging picture of the epigenetic profiles of stem cells. For example, HSCs and committed hematopoietic progenitors have been shown to display different DNA methylation patterns.56 Although the mechanisms continue to be revealed, several studies indicate that DNA methyltransferase activity is critical for HSC function, and distinct methyltransferase enzymes appear to have unique effects on the self-renewal and differentiation of these cells.57–59 Experimental approaches including bisulfite sequencing, in which cytosine but not methylcytosine is converted to uracil, as well as methylated DNA immunoprecipitation methods, have been utilized to gain a global view of DNA methylation.60–63 In general, cytosine–phosphate–guanine (CpG) methylation is associated with transcriptional repression, and promoter analysis in mouse ES cells demonstrated a correlation between repressed differentiation-related genes and the presence of CpG modifications.64 Additionally, a high-throughput sequencing strategy was shown to enable the full genome-wide profiling of DNA methylation in human ES cells, with direct comparisons to human fetal fibroblasts.65 In this study, a significant number of cytosine methylations were identified in non-CG sites, and these were frequently present within gene bodies versus enhancer regions, and in contrast to CpG modifications, were positively associated with transcriptional activity.

In addition to DNA methylation, genome-wide analysis of histone modifications has also been performed in stem cells. Based on such studies with ES cells, an important role for the trimethylations of histone H3 lysine 4 (H3K4me3) and histone H3 lysine 27 (H3K27me3) has been suggested. Specifically, many developmental genes which are repressed in mouse and human ES cells exhibit both H3K4me3 and H3K27me3 modifications.66–68 Independently, H3K4me3 marks are generally associated with active chromatin, while H3K27me3 marks are common at repressed sites. The presence of ‘bivalent domains’ containing both modifications has been postulated as a way to maintain the silencing of developmental genes while keeping them poised for rapid activation. Building on these findings, a recent study performed ChIP-seq analysis of 11 histone modifications, together with DNA methylation measurements, to evaluate genome-wide differences between human ES cells and primary lung fibroblasts.69 In particular, this work demonstrates the capabilities of newly developed computational algorithms for improving the identification of chromatin modification domains, and for testing associations between histone modifications and DNA methylation sites. For the analysis of primary adult stem/progenitor cells present in limited numbers, new methods for increasing the sensitivity of histone mapping are being developed, and are discussed in the following section. As described above, the functional interpretation of genomic and epigenomic signatures, including the identification of novel regulatory regions, is aided by the acquisition of parallel measurements. These include genomic-scale measurements of multiple epigenetic modifications, gene expression, and protein interactions, which can be collectively evaluated with computational methods.10,65 Leveraging this type of approach, a recent study utilized a comparative epigenomic method, based on the genome-wide interspecies (human, mouse, pig) comparisons of pluripotent stem cells, to identify conserved epigenomic marks that display regulatory functions in the early differentiation of ES cells.70 Overall, the genome-wide assessment of miRNA and epigenetic signatures has greatly expanded the collective understanding of stem cell genetic programs.

High-Throughput Genetic Analysis and Perturbation

Advances that promote the increased throughput of immunoprecipitation and expression analysis methods have played a central role in the elucidation of stem cell genomic and epigenomic profiles. For example, an approach commonly referred to as deep sequencing, which significantly increases the number of mappable reads, has shown great utility for identifying rare transcripts (e.g., miRNAs, alternative splicing events) in stem cells.71–74 In order to facilitate studies of rare stem cell populations as well as investigations of stem cell heterogeneity through single cell analysis, a number of strategies for increasing sensitivity and throughput have been developed, which can enable substantial reductions in the required cell numbers for gene expression and ChIP assays. For example, ChIP-seq coupled with PCR amplification with modified random primers was shown to enable whole-genome profiling from a population of 10,000 hematopoietic progenitor cells.75 This procedure exhibited an approximate 80% sensitivity and 90% specificity compared to standard ChIP-seq methods for histone profiling. A modified RNA sequencing strategy has been utilized for the comprehensive sampling of gene expression from single ES cells, and demonstrated key alterations occurring during the adaptation of mouse ES cells to in vitro culture conditions.76 In addition, a methylation-sensitive PCR microreaction approach was recently developed for obtaining DNA methylation profiles of individual cells,77 and could prove useful for future epigenetic studies in stem cells. Microfluidic chips have also emerged as robust platforms for expression profiling. Specifically, microfluidic devices such as the commercially available Dynamic Array™ from Fluidigm, provide the capability to perform thousands of individual nanoliter PCR reactions in a single device, which has been enabling for single cell gene expression studies which require large-scale parallel analysis. Such an approach has been utilized to examine the heterogeneity of gene expression in pluripotent stem cells, and miRNA expression profiles across the differentiation hierarchy of hematopoietic lineages.78,79 Recently, a microfluidic platform was developed which fully integrated single cell trapping and lysis, cDNA synthesis, and PCR amplification into one device for high-throughput single cell expression analysis.80 In addition to expression measurements, microfluidic-based approaches have been utilized for genotyping and immunoprecipitation experiments. For example, a microfluidic device system was demonstrated to facilitate the genotyping of individual chromosomes from a single cell.81 Further, microfluidic platforms have been developed for improving ChIP analysis from extremely limited cell numbers, such as 2000 cells in one strategy,82 and as few as approximately 50 cells in another approach when combined with PCR amplification.83

Together with the increasingly robust analysis methods, numerous strategies for high-throughput genetic perturbation have been developed and applied to investigations of stem cell biology. In particular, gain-of-function screens in ES cells, both genome-scale84 and transcription factor focused,85 have identified components influencing pluripotency maintenance, and further illustrated the interconnectivity of the regulatory network. As highlighted in previous sections, gene expression analysis following RNAi-mediated knockdown of individual core stem cell transcription factors has also been widely utilized in experiments aimed at revealing additional details of the network structure. High-throughput loss-of-function, using RNAi libraries to target numerous genes in parallel, represents a complementary approach toward the identification of novel pathway components. For instance, initial efforts examining the regulation of pluripotency employed focused libraries derived from candidate genes9 or cDNA subtraction,86 to identify factors involved in the self-renewal of mouse ES cells. Subsequently, genome-wide RNAi screens in both mouse and human ES cells have revealed many novel regulators of pluripotent stem cell function.87–89 In addition to standard multiwell methods, recent developments, including pooled RNAi screens, have increasingly expanded the repertoire of experimental designs. In this approach, cells are transduced with a pooled library such that each cell will not receive more than one short hairpin RNA (shRNA) vector. Following a selectable stimulus such as drug exposure or long-term in vitro culture, sequencing or microarray hybridization is performed to evaluate the relative abundance of a particular shRNA, or alternatively, an associated short oligonucleotide barcode. Such an approach has been used to uncover novel targets involved in the proliferation of hematopoietic progenitors and hair follicle stem cells.90,91 Notably, pooled library analysis can additionally facilitate RNAi screens that incorporate in vivo assays, such as studies investigating bone marrow repopulation by hematopoietic stem cells.92 In addition, cellular microarrays, which are discussed in more detail below, have also been explored as a means to miniaturize lentiviral-based RNAi screens for increased throughput.93 Building on these initial efforts, a recent study demonstrated the utility of lentiviral microarrays for large-scale genomic screening in melanoma, through the arrayed overexpression of a kinase library.94 Further, gene expression dynamics in mesenchymal stem cells has been explored using lentiviral microarrays carrying reporter constructs.95 In the future, it is likely that high-throughput RNAi screens based on emerging technologies such as cell microarrays will be broadly applied to a wide range of studies, including the examination of stem cell regulatory mechanisms.

Reprogramming Regulatory Networks

A thorough understanding of the gene regulatory networks in stem cells provides not only a picture of what defines stem cell identity, but can also reveal various entry points for manipulating stem cell function toward the development of translational applications. Perhaps the most significant illustration of this approach is the reprogramming of somatic cells to a population of cells exhibiting pluripotent characteristics, referred to as induced pluripotent stem (iPS) cells. Initial studies in this burgeoning field demonstrated the production of mouse and human iPS cells from fibroblasts with as few as four transcription factors delivered through viral transduction.96–98 Subsequent findings have demonstrated iPS derivation from numerous somatic cell types, refined the transcription factor requirements, and developed novel nonviral methods to induce the reprogramming process.99 In addition, many of the DNA methylation and chromatin analysis methods described in this review have been applied to iPS cells to explore the epigenetic alterations occurring in the reprogramming process.100 Furthermore, building on the precedent set by the generation of iPS cells, and earlier investigations illustrating the influence of MyoD on fibroblast conversion to skeletal muscle,101 direct transcription factor-mediated reprogramming between mature cell states, without a stem cell intermediate stage, has been obtained in numerous contexts and continues to be a highly active area of research.102 An analogous approach utilizing transcription factor delivery has also been shown to increase the efficiency of in vitro stem/progenitor cell differentiation. For example, the introduction of adipogenic transcription factors significantly enhanced the differentiation of white and brown adipocytes from iPS-derived mesenchymal progenitors.103

Substantial efforts have also pursued the manipulation of stem cells with small molecules. High-throughput small molecule screens have served to identify compounds which can support pluripotent stem cell self-renewal and revealed pathways involved in this process.104 Additionally, small molecules are among the many modulating factors which have been shown to enhance the efficiency of iPS reprogramming, and enable the reduction in the number of required transcription factors delivered through viral constructs or other protein/nucleic acid transfection techniques. Further, a small molecule screen served to identify aryl hydrocarbon receptor antagonists that promote the in vitro expansion of HSCs.105 Small molecule treatments have also been shown to aid in the optimization of stem cell differentiation protocols. For example, small molecule library screens were utilized to identify factors promoting endoderm106 and pancreatic progenitor specification,107 as well as the neuronal differentiation of neural progenitor cells.108 In addition, small molecule screening revealed specific inhibitors of the Wnt pathway which can enhance cardiomyocyte differentiation from human ES cell-derived mesoderm.109 Overall, the stimulation/inhibition of pathways with small molecules serves as a parallel strategy to the delivery of protein mediators such as transcription factors, with the goal of shifting the regulatory networks toward a desired cellular fate. Looking forward, the additional integration of developing technologies such as synthetic biology-derived genetic circuits110 could potentially provide unprecedented manipulative control of stem cell phenotypes, in particular, by priming them to accept specific environmental cues.

Interconnections with Microenvironmental Signals

Microenvironmental signals act as inputs to the regulatory networks and are closely associated with the intrinsic components. For instance, mouse ES cells are maintained in an undifferentiated state in vitro by factors including leukemia inhibitory factor (LIF) and bone morphogenetic proteins (BMPs),111,112 which signal through the transcription factors STAT3 and Smad1, respectively. A complex interplay exists between these signaling pathways and the core transcriptional network, and recent analysis strategies have aimed to integrate these elements into a single dynamic system. As mentioned above, STAT3 and Smad1 have been identified in genomic binding sites co-occupied by Oct4, and interestingly, knockdown of Oct4 reduced the binding of STAT3 and Smad1 at these loci.23 These data are suggestive of another feedback loop, in which Oct4 is both regulated by external signals and provides a required role in the transcriptional response to these signals. In addition to LIF and BMP signaling, the Wnt pathway and the downstream effector Tcf3 have been implicated in pluripotency regulation,113–115 and accordingly, Tcf3 has also been shown to co-occupy enhancer sequences bound by Oct4.116,117

The importance of external signaling components has also been highlighted by efforts aimed at establishing molecular induction strategies for the in vitro expansion of HSCs.118 These include several recent studies that have implicated the ligand-activated transcription factor, aryl hydrocarbon receptor (AhR), in HSC function. For example, as mentioned above, AhR antagonists have been identified which promote the self-renewal of HSCs in culture.105 Furthermore, AhR knockout mice exhibit modestly increased numbers of HSC-enriched cells in the bone marrow,119 and treatment with a xenobiotic AhR ligand results in altered HSC self-renewal and a skewing of HSC differentiation.120 Although endogenous AhR ligands involved in HSC function have yet to be identified, AhR-mediated transcriptional regulation of several transcription factors including HES-1, C/EBP, and c-Myc has been demonstrated in various cellular contexts.121–123 Collectively, these findings, together with observed effects of AhR signaling on adhesion and chemokine receptor expression, are suggestive of a role for AhR at the interface of microenvironmental signals and transcriptional networks in HSCs.124 Although clinically relevant HSC proliferation procedures have remained elusive, these studies as well as recent developments based on the activation of the Notch pathway,125 and in bioreactor design through leveraging systems models (discussed below),126 suggest that an improved understanding of HSC regulatory mechanisms could promote the achievement of clinical milestones in the near future. Overall, in order to gain an improved understanding of the interconnected intracellular and extracellular regulatory pathways in stem cells, it is necessary to complement the examination of gene regulatory networks with systematic investigations of microenvironmental regulatory mechanisms. In the following sections I will describe bioengineering approaches for the analysis of stem cell microenvironments.

STEM CELL MICROENVIRONMENTS

  1. Top of page
  2. Abstract
  3. OVERVIEW OF STEM CELL FATE REGULATION
  4. STEM CELL GENETIC PROGRAMS
  5. STEM CELL MICROENVIRONMENTS
  6. CONCLUSION
  7. REFERENCES

Niche-Mediated Regulation of Stem Cell Functions

Fundamental stem cell processes, including self-renewal and differentiation trajectories, are regulated by the integration of chemical and physical signals present within local microenvironments, or niches (Figure 2). These signals are delivered in the form of soluble factors (e.g., growth factors, hormones, metabolites, oxygen), or as insoluble cues (e.g., cell-cell interactions, extracellular matrix, surface-bound signaling molecules), and often exhibit both spatial and temporal dynamics. In vivo, specialized niches direct resident adult stem cell function based on the combinatorial presentation of such microenvironmental stimuli. One of the most characterized adult stem cell niches is the HSC niche in the bone marrow. Numerous cellular components in the bone marrow, including osteoblasts, endothelial cells, and various mesenchymal cell populations have been demonstrated to influence HSC function.127 Substantial research efforts continue to be focused on resolving the cooperative influence of these diverse cell–cell interactions, and toward gaining an improved understanding of the role of HSC anatomical localization within distinct regions of the bone marrow compartment. Furthermore, cells differentiated from HSCs, such as macrophages, can provide feedback signals influencing HSC functions, consistent with the presence of important progeny–stem cell interactions in many stem cell niche contexts.128 In parallel to in vivo investigations of stem cell niche composition and dynamics, the development of engineered culture models (discussed in detail below), represents a promising strategy for elucidating key intercellular signals within stem cell microenvironments.

thumbnail image

Figure 2. Engineering approaches for the study of stem cell microenvironments. Depicted are the prototypical components of a stem cell microenvironment, or niche. A network of interacting microenvironmental signals regulates stem cell functions. Bioengineering approaches, such as microtechnology tools, biomaterial fabrication, and computational methods coordinately enable the deconstruction of these complex microenvironments for systematic investigation, and the construction of engineered microenvironments for translational applications.

Download figure to PowerPoint

Based on the role of the microenvironment in collectively shaping stem cell functions, the development of niche-targeted therapeutic interventions is considered an increasingly viable approach. For example, factors which influence bone marrow adipogenesis,129 or bone marrow osteoblast numbers,130,131 have been shown to affect hematopoiesis, and have been considered as therapeutic options for improving transplantation efficacy. Such strategies are analogous to microenvironment-directed treatments that have been developed for cancer therapy. In particular, several clinical treatments for breast cancer, including inhibitors of aromatase enzymes, angiogenesis inhibitors, and HER family receptor antagonists target stromal interactions.132 Further, the importance of microenvironmental context in cancer stem cell processes has been highlighted by studies analyzing the bidirectional interactions between cancer stem cells and both ‘normal’ and tumorigenic niche components. These interactions have been demonstrated to influence tumor growth and metastasis.133 In addition, the effect of microenvironmental perturbations on the self-renewal and directed differentiation of ES and iPS cells is a highly active field of investigation. An improved understanding of the extracellular signals regulating the differentiation of pluripotent stem cells and stem cell-derived progenitors can provide insights into normal developmental mechanisms and aid in optimizing differentiation protocols for cell sourcing applications. In particular, the application of engineering technologies can enable the deconstruction of the complex spatiotemporal mechanisms regulating stem cell fates (Figure 2), and establish the basis for the development of highly functional in vitro models of tissue development and homeostasis.

Engineering Approaches for the Controlled Presentation of Microenvironmental Signals

Engineered microsystems are designed to present microenvironmental stimuli in a tightly controlled manner, which facilitates the decoupling of signals for systematic examination. Such platforms have found great utility in the in vitro analysis of cell– cell and cell–ECM interactions involved in stem cell regulation. Specifically, in order to control cellular positioning, shape, and exposure to ECM proteins within two-dimensional cultures, a process termed micropatterning can be employed, which was developed based on tools used in the semiconductor industry, and can effectively pattern ECM proteins on a surface with micrometer scale resolution.134,135 This patterning is typically achieved with either photolithography methods, which are based on exposing a photosensitive material to UV irradiation through a patterned mask, or with soft lithography techniques, in which molecules are transferred to a surface using a patterned stamp made from silicone rubber, polydimethylsiloxane (PDMS). Numerous studies have employed micropatterned substrates to investigate the influence of cell shape and cell–ECM interactions on cellular functions due to the capability to independently manipulate the degree of cell spreading and cell–ECM contact.136 In a series of reports examining mesenchymal stem cell (MSC) differentiation, it was demonstrated that individual cell shape and spreading regulate cytoskeletal tension and associated Rho GTPase signaling, which influences the differentiation of these cells toward the osteogenic or adipogenic lineages.137–140 Furthermore, the impact of multicellular organization on determining osteogenic/adipogenic differentiation has been illustrated by studies utilizing ECM domain micropatterning to generate multicellular islands of predetermined geometries.141,142 These findings were similarly suggestive of the important role of cytoskeletal tension in this differentiation process. In addition to surface micropatterning, PDMS stamps can be utilized for the microscale molding of hydrogels such as agarose, polyacrylamide, and poly(ethylene glycol) (PEG) into diverse geometries including microwell structures. Microwells have been widely used for high-throughput analysis (discussed in detail below) and as a tool to investigate cell–cell interactions. For example, a platform incorporating bowtie shaped agarose microwells was developed to examine the effects of homotypic interactions between two adjacent cells, while decoupling the influence of cell-cell contact from the effects of cell spreading.143,144 In order to examine the role of cell–cell interactions in ES cell differentiation, both microwell and micropatterning platforms have been utilized for the formation of ES cell aggregates with tightly controlled diameters. These systems, as well as complementary suspension culture techniques, have collectively demonstrated the role of aggregate size in directing the early lineage commitment of ES cells.145–153

The mechanical properties of the stem cell microenvironment can also significantly influence stem cell function. For instance, various mechanical perturbation regimens have demonstrated the effects of mechanical strain on the self-renewal of human ES cells,154,155 the vascular differentiation of mouse ES cells,156 and the myogenic differentiation of MSCs.157 Furthermore, MSC fate decisions as well as the self-renewal capabilities of skeletal muscle stem cells have been demonstrated to be regulated by substrate rigidity, which is sensed as a mechanical signal.158,159 In order to systematically explore the effects of substrate elasticity characteristics on cellular functions, hydrogel materials including polyacrylamide,160,161 PEG,159 PDMS,162 and hyaluronic acid (HA)163 are commonly utilized as culture surfaces that exhibit tunable elastic moduli upon alterations in the degree of crosslinking. Although polyacrylamide and PEG hydrogels are hydrophilic and generally resist protein adsorption, they can be chemically modified to present either native ECM proteins or adhesive peptides for mediating cell adhesion. In addition to bulk modifications, microtechnology methods have been applied toward the fabrication of micropillar systems which enable the modulation of substrate rigidity without altering the inherent material properties.164–168 In this approach, PDMS is molded into an array of vertical posts on top of which cells can be cultured. The substrate rigidity is then defined by the post dimensions, which are modular and can be adapted for the desired elastic characteristics. Analysis of MSC differentiation on a micropillar array164 demonstrated effects of substrate rigidity on lineage commitment, consistent with polyacrylamide studies,158 and facilitated correlations between traction force (calculated by the deflection of the micropillars), focal adhesions, and cytoskeletal tension.

Substantial efforts have also focused on the development of improved three-dimensional (3D) culture platforms. Highly functional 3D model systems can provide the capability to investigate cellular processes which can be affected by dimensional context, such as growth factor and adhesion receptor signaling, and other complexities in cellular organization and mechanics which are present in 3D tissue architectures.169–173 Many natural/biologically derived and synthetic biomaterials have been extensively examined toward the development of 3D scaffolds that recapitulate the ECM of a relevant tissue or provide specific empirically determined characteristics. For example, a photopolymerizable hyaluronic acid hydrogel platform has been demonstrated to support the self-renewal of human ES cells in 3D, without a requirement for feeder cells.174 In addition, in order to direct or augment the differentiation of both pluripotent and various adult stem cell populations toward specific lineages in a 3D context, a broad range of scaffold materials have been explored, and findings from these studies are summarized in several comprehensive reviews.175–177 Notably, recent advances in biomaterials have significantly improved the capability to present a diverse range of biological signals to cells in 3D. In particular, synthetic hydrogels, such as PEG-based systems, are highly tunable in regards to porosity and mechanical properties, and can be co-encapsulated or conjugated with bioactive factors to add biological functionality in a tightly controlled manner. For instance, the integration of small functional groups has been shown to influence cell-secreted growth factor and ECM presentation,178–181 and various peptide sequences can be incorporated to dictate cellular adhesion and cell-mediated degradation of the hydrogel.182 Releasable factors can also be introduced by the co-encapsulation of microparticles within the 3D network.183–185 Further, novel strategies for the dynamic modulation of hydrogel factor presentation, based on photoactuation chemistries, have been recently demonstrated.186–190 In one study, the temporal control of RGD peptide presentation through a photocleavage process facilitated the chondrogenic differentiation of encapsulated MSCs.190 Collectively, such developments continue to enhance the overall flexibility in the design of scaffold systems, and significantly increase the number of parameters which can be manipulated to influence stem cell function in 3D environments.

High-Throughput Analysis of Stem Cell Microenvironments

As highlighted in previous sections, robust high-throughput analysis methods are central to the elucidation of stem cell genetic regulation at the genomic scale. In an analogous manner, microtechnology approaches have been applied to the development of high-throughput platforms that enable the systematic screening of microenvironmental signals. The fabrication of cellular microarrays represents an example of one of these approaches aimed at exploring, in parallel, a range of combinations of signals that cannot be practically evaluated with standard techniques. Cellular microarrays typically consist of either printed spots of biomolecules, including adhesive components and other signaling/detection elements, onto which cells are seeded, or alternatively, consist of directly printed arrays of live cells encapsulated in hydrogel droplets. Currently, cellular microarrays have been most broadly applied to the assessment of ECM effects, and in particular, the role of combinations of ECM proteins in mediating cell processes.191–196 Together with ECM molecules, microarrays of cell surface ligands and growth factors have been developed, and have provided intriguing insights into how neural197 and mammary progenitor198 differentiation is regulated by combinatorial signals. In addition to natural proteins, microarrays of biomaterials have been utilized to identify synthetic surfaces, with distinct polymer chemistries, which can significantly influence pluripotent and adult stem cell self-renewal and differentiation.178,199–203

Microwell platforms have also been broadly used for the high-throughput analysis of stem cells.204 Microwell arrays are typically fabricated through direct etching of hard materials (e.g. glass, silicon) or through a combination of photopolymerization and soft-lithography-based molding of hydrogels. In particular, this approach has been applied to the analysis of individual stem cells in order to evaluate clonal heterogeneity.159,205 Hydrogel microwells can be functionalized with biomolecules, as highlighted by a recent strategy that paired microwell molding with protein microarraying to analyze neural stem cells and MSCs within microwell arrays presenting a range of combinatorial stimuli.206 Microfluidic-based approaches, integrating microwells or hydrodynamic traps, have also been employed to generate cellular arrays.207–210 In addition, the emerging field of droplet microfluidics211 represents an attractive approach to increase the throughput of 3D fabrication, while maintaining precise control of the environmental components. For example, miniaturized cell-encapsulated hydrogels were fabricated with microfluidic methods in order to examine co-cultures,212 incrementally modulate material stiffness,213 and generate microscale constructs for the assembly of larger patterned structures.214 Furthermore, a recent study described a high-throughput 3D analysis approach, analogous to flow cytometric measurements of individual cells, which could enable the sorting of miniaturized stem cell constructs based on the expression of a differentiation reporter, and was additionally compatible with multiplexing and in vivo assessment.215

Engineered Environments for Optimized Expansion and Differentiation

Building on the mechanistic studies and the increasing capability to precisely control in vitro culture environments, significant research efforts are focused on the development of platforms which could promote the proliferation and directed differentiation of stem and progenitor cells for clinical applications.216 Notably, microfluidic platforms have provided key insights into the influence of flow and nutrient transport on stem cell processes. For example, microfluidic devices, which exhibit spatiotemporal control of perfusion characteristics, were utilized to investigate the effects of hydrodynamic shear and nutrient delivery on ES cell proliferation and differentiation.217–219 Consistent with these microfluidic studies, larger scale perfusion systems have demonstrated that nutrient transport and the retention of autocrine soluble factors, which are collectively determined by the flow characteristics, can significantly influence ES cell proliferation.220,221 Moving forward, a significant challenge in the field is the development of approaches that can adequately translate the combinatorial control of microenvironmental signals achieved at the microscale in laboratory models, to a scale that can attain clinically effective regulation of stem cell proliferation and directed differentiation.

Computational approaches can be employed to assist in the interpretation of combinatorial perturbations. In particular, advanced statistical methods and network models have been applied to pairwise and higher order combinations of signals to evaluate and predict synergistic effects.222–224 Such tools can greatly complement engineered high-throughput platforms for the deconstruction of complex environments. Computational tools, such as learning algorithms, have also been shown to aid in the refinement of experimental conditions, including the iterative identification of a distinct combination of small molecules which support human ES cell self-renewal.225 In addition, toward a more complete understanding of microenvironmental regulation and the optimization of in vitro stem cell platforms, computational models of intercellular communication networks can be formulated. For example, as mentioned previously, HSC expansion has been shown to be regulated by feedback signals from differentiated HSC progeny, and systems models of intercellular interactions have been developed for this process.226,227 By further incorporating models of bioreactor conditions, a recent study computationally predicted parameters that could enhance HSC proliferation, which were confirmed through in vitro expansion experiments.126 Further, intercellular signaling models can be formulated for in vivo developmental processes, such as the recent demonstration of a Wnt-BMP signaling circuit in tooth organogenesis,228 which was predicted based on the integrated analysis of epithelial and adjacent mesenchymal cell gene expression profiles, and validated with genetic mutations in a mouse model. In the future, it is anticipated that systems-based analysis of the multilineage differentiation and patterning of stem cells within in vitro culture models will provide key insights into developmental mechanisms and relevant abnormalities.

CONCLUSION

  1. Top of page
  2. Abstract
  3. OVERVIEW OF STEM CELL FATE REGULATION
  4. STEM CELL GENETIC PROGRAMS
  5. STEM CELL MICROENVIRONMENTS
  6. CONCLUSION
  7. REFERENCES

In recent years, substantial progress in the understanding of stem cell processes has been achieved through studies collectively positioned at the interface of systems biology and bioengineering. Systems models of gene regulatory networks and intercellular interactions have provided an important framework of the complex interconnections that exist between the individual components, and helped to reveal the characteristics that define stem cell identity. Necessarily coupled with these advancements has been the steady development of high-throughput platforms for improved data acquisition with an increasingly broader scope. These include engineering strategies, such as miniaturized cell-based assays, which facilitate the systematic analysis of combinations of cellular perturbations. In addition, evolving approaches for the forward engineering of stem cell genetic programs and the controlled scale-up of in vitro stem cell culture have begun to demonstrate utility for translational applications.

Despite the tremendous advancements in the knowledge of stem cell functions, a number of key issues remain unresolved. For example, due to limited cell numbers, the gene regulatory networks of primary stem and progenitor cell populations remain less established compared to pluripotent stem cells. In order to overcome this limitation, continued progress toward the increased sensitivity and throughput of genomic-scale analysis methods, based on the incorporation of enhanced sequencing approaches and platforms such as microfluidic devices, will be required. Notably, the evolution of increasingly robust single-cell assays should provide unique information regarding the heterogeneity of genomic and epigenomic profiles, and offer insights into stochastic processes regulating stem cell fate.229 In addition, concurrent developments in computational systems will be necessary to support the increased throughput and refinements in the interaction models. In particular, an improved understanding of the quantitative relationships between network components, incorporating relatively subtle effects such as miRNA-mediated modulation and incremental changes in extracellular signal concentration will be an important next step. Furthermore, for many stem cell types it remains unclear how microenvironmental signals interact with gene regulatory networks, and how these systems act to co-regulate stem cell behavior. For instance, although numerous studies have contributed to the understanding of the genetic perturbations required for cellular reprogramming, the role of microenvironmental cues in reprogramming processes has not been extensively explored. Current efforts in the development of proteomics approaches will be necessary to effectively evaluate the complex combinations of ECM and soluble components present within stem cell microenvironments in vivo, and will assist in constructing informative in vitro model systems. Improved methods for investigating bidirectional cell-cell interactions, which build on the engineering strategies described in this review, would also represent important enabling tools for future studies. In particular, the paired application of advanced imaging technologies that facilitate dynamic analysis at multiple length scales (i.e., subcellular, single cell, multicellular) with engineered high-throughput culture platforms could provide unprecedented insights into inductive intercellular interactions. Overall, the coordinated design and development of advanced in vitro platforms that exhibit tight spatiotemporal control of microenvironmental cues, with improved computational methods for analyzing the effects of these combinatorial signals, should provide a clearer understanding of the dynamic pathways defining stem cell fate. By leveraging these complementary disciplines, it is expected that a holistic view of stem cell regulatory mechanisms will continue to emerge, and form the foundation for stem cell therapeutic approaches.

REFERENCES

  1. Top of page
  2. Abstract
  3. OVERVIEW OF STEM CELL FATE REGULATION
  4. STEM CELL GENETIC PROGRAMS
  5. STEM CELL MICROENVIRONMENTS
  6. CONCLUSION
  7. REFERENCES