|Home | About | Journals | Submit | Contact Us | Français|
The miRNA pathway, as a fundamental mechanism of gene regulation, plays a key role in controlling the establishment, self-renewal, and differentiation of stem cells. Such regulation is manifested as fine-tuning the temporal- and tissue-specificity of gene expression. This fine-tuning function is achieved by (1) miRNAs form positive and negative feedback loops with transcription factors and epigenetic factors to exert concerted control of given biological processes; and/or (2) different miRNAs converge to control one or more mRNA targets in a signaling pathway. These regulatory mechanisms are found in embryonic stem cells, iPS cells, and adult tissue stem cells. The distinct expression profiles of miRNAs and their regulatory roles in various types of stem cells render these RNAs potentially effective tools for clinical diagnosis and therapy.
How gene regulation defines the fate, self-renewal, and differentiation of stem cells represents a central question in stem cell biology. Towards this question, much of the current effort has been devoted to niche-signaling, epigenetic, and transcriptional regulation of gene activities in stem cells. In contrast, gene regulation at post-transcriptional levels, such as translational and post-translational regulation, remains largely unexplored. This situation, however, is being improved by the recent studies that reveal the key regulatory role of miRNAs in controlling stem cell function and animal development through modulating gene regulation network. The microRNA (miRNA), first discovered in C. elegans, is a family of small non-coding RNAs existing in diverse organisms that bind to the 3′ untranslated region (3′ UTR) of mRNAs via incomplete sequence complementarity to regulate their translation and stability 1–3. In the human genome, miRNA genes constitute ~4% of genes and regulate the expression of more than a third of the protein-coding genes in a target specific manner at the post-transcriptional level 4. In this review, we summarize the latest knowledge of the regulatory function of miRNAs in the self-renewal and differentiation of embryonic stem cells and tissue stem cells, starting with a brief review of miRNA biogenesis and regulatory mechanism as background.
miRNAs are produced in distinct steps in nuclear and cytoplasmic compartments. In the canonical miRNA biogenesis pathway (Fig. 1, left panel), primary miRNA transcripts (pri-miRNAs) are generated by RNA polymerase II and then cleaved into ~70-nucleotide precursor-miRNAs (pre-miRNAs) in the nucleus by the microprocessor complex, comprised of a RNAse III enzyme called Drosha and a dsRNA binding protein DiGeorge syndrome Critical Region Gene 8 (DGCR8)5, 6. Pre-miRNAs are subsequently trafficked by Exportin 5 to the cytoplasm, where they are further cleaved by another RNAse III enzyme called Dicer at a site 22nt away from the Drosha cleavage site into mature miRNA duplexes 5, 7. Dicer is in a complex that contains the dsRNA binding protein (Transactivating Region Binding Protein) TRBP, PACT (Protein Activator of PRK), and Loquious (Loqs). This complex further separates the two RNA strands of the miRNA/miRNA* duplex, incorporates one RNA strand (miRNA) into Ago and releases the other strand (miRNA*), which is typically subject to degradation 8–10. The miRNA-containing Ago is then assembled into miRNA-induced Silencing Complex (miRISC), which recognizes the target mRNA via incomplete sequence complementarity between the miRNA and mRNA targets. Nucleotides 2–8 of miRNAs are critical for the target recognition through perfect complementarity with the target, and are thus called the “seed motif” 11, 12. A single miRNA can recognize multiple targets, regulating up to several hundred different species of mRNAs. In addition, multiple miRNAs can act simultaneously on one mRNA target. Such combinatorial modes of action allow several hundreds of miRNAs to modulate the expression of thousands of mRNA species at the posttranslational levels. Such regulation is conceivably achieved via decreased translational efficiency and/or reduced mRNA stability 13. Recently, Bartel and colleagues found that decreased mRNA levels constitutes most (~80%) of the reduced protein production14.
In addition to the canonical pathway, there are microprocessor-independent and Dicer-independent pathways (Fig. 1, middle panel). In the microprocessor-independent pathway (Fig. 1, middle panel), the cropping function by Drosha is replaced by three different mechanisms: (1) by spliceosome for mirtron pathway; (2) by Dicer for snoRNA-, tRNA-, and shRNA-derived precursor; and (3) by tRNase Z for tRNA-derived pathway 15–20. Recently, Dicer-independent miRNAs were discovered in zebrafish and mammals 21, 22. In this pathway (Fig. 1, right panel), the RNA cleavage activity of Ago2 mediates the maturation of pre-miR-451. miR-451 is unique in that its 5′-end was defined by Drosha cut, yet its 3′-end is flexible and extends over the loop region of the hairpin with a length range of 20–30 nt. Moreover, 1–5 non-templated uridine residues were also found at the 3′-end of longer reads, implying that the Ago2 cleavage products undergo uridylation and subsequent trimming by a nuclease. These alternative pathways illustrate the ability of cells to exploit a wide variety of mechanisms to generate miRNAs and to execute gene regulation.
Embryonic stem cells (ESCs) are derived from the inner cell mass (ICM) of mammalian embryos at the blastocyst-stage. They possess two unique properties: the ability to differentiate into any fetal or adult cell type (pluripotency) and the capacity to replicate for many generations (self-renewal). Such properties are sustained by delicate regulatory mechanisms including extracellular signaling, epigenetic programming, and epigenetic regulation, The chromatin state in ESC is largely open, rendering it accessible for transcriptional activation, while specific epigenetic regulators control the chromatin state locally to block the activation of differentiation-specific genes. In this chromatin context, ESC core transcriptional factors (Oct4, Sox2, Nanog etc.) together with a variety of signaling pathways govern ESC pluripotency. 23, 24. In addition, the molecular program underlying the ESC pluripotency and self-renewal has increasingly been linked to miRNAs, which will be reviewed in this section.
miRNA expression profile is distinct in ESC, with most of the ESC-enriched miRNAs sharing a 5′-proximal AAGUGC sequence signature 19, 25–28. In humans, the miR-371 family (homologous to the miR-290 family) and the miR-302 family represent the majority of ESC-specific miRNAs. Additionally, these miRNAs tend to be cotranscribed as polycistronic transcripts, and share a common upstream transcriptional regulation by a set of core stem cell transcription factors 29. The levels of these miRNAs decrease as the ESCs differentiate, which indicates the reducing function in differentiated cells 30.
Intrigued by the distinct expression pattern, researchers have investigated the roles of miRNAs in regulating ESC self-renewal and differentiation. These investigations revealed the following three mechanisms of miRNAs in establishing and maintaining ESC stemness (Fig. 2), while directing ESC differentiation into different lineages.
In order for ESCs to differentiate properly, miRNAs suppress the expression of ESC core transcription factors (Fig. 2, left branch). For example, miR-145 represses c-Myc, Sox2, Oct4 and Klf4 31, 32. In mouse ESCs (mESCs), Oct4, Sox2 and Nanog are regulated by miR-134, miR-296 and miR-470, whereas Sox2 and Klf4 are cooperatively modulated by miR-200c, miR-203 and miR-18333–35. One common theme in miRNAs and the ESC pluripotency transcription machinery is double-negative feedback. On one hand, the aforementioned miRNAs repress the expression of pluripotency transcription factors at the posttranscriptional level. On the other hand, pluripotency transcription factors in return silence the promoter regions of those miRNAs through co-occupation of Polycomb group proteins and the resultant H3K27me3 29.
ESCs exhibit a significantly shorter cell cycle. For example, cell cycle for mESCs is 12 hours instead of approximately 24 hours for somatic cells. The key underlying this short cell cycle is a shortened G1 phase, where G1-to-S transition is primarily unchecked, via a constitutively active cyclin E-Cdk2 complex 36. The global depletion of miRNAs in Dgcr8 and Dicer1 mutant mESCs show proliferation defects 37–39. Especially, DGCR8-deficient mESCs show accumulated cells arrested at G1 phase, indicating the function of miRNAs in mediating the constitutive activation of cyclin E-Cdk2 complex, thus promoting G1-to-S transition. This activation of cyclin E-Cdk2 by miRNAs was intensively studied, which revealed that miR-290 and miR-302 suppress cyclin E-Cdk2 upstream inhibitors (including Cdkn1a, Rbl2, and Lats2) in mESCs, therefore enabling the unrestricted G1-to-S transition and short cell cycle (Fig. 2, middle branch) 36, 40.
Such function in regulating cell cycle is also conserved in hESCs, where G1-to-S transition checkpoint genes Cdkn1a and Cdkn1c are suppressed by miR-372 andmiR-92a, respectively 41, 42. miR-372 belongs to the miR-371 family, which is homologous to mouse the miR-290 family, indicating such suppression of cell cycle checkpoint is highly conserved. Moreover, G2-to-M transition is regulated by miRNAs as well. miR-195 represses WEE1, an inhibitory kinase in G2-to-M transition that blocks the G2 cyclin B-Cdk complex 42.
One interesting mode of action is convergence of multiple miRNAs on one pathway. It is noteworthy that miR-290 and miR-302 possess the same target recognition motif 36. Such a shared motif allows multiple miRNAs to target multiple cyclin E-Cdk2 upstream inhibitors, thus promoting mitosis progression beyond G1-to-S transition check point.
ESCs harbor a relatively open chromatin, possibly as a unique property underlying their pluripotency and self-renewal. During differentiation, distinctive regions on the chromatin undergo de novo DNA methylation and mediate proper differentiation. In Dicer-null ESCs with global miRNA depletion, the levels of DNA methyltransferases are reduced, which compromises de novo DNA methylation. This implicates the miRNA function in epigenetic regulation (Fig. 2, right branch) 40, 43. Notably, Oct4 is incompletely and reversibly silenced due to DNA methylation defects. Such compromised silencing of Oct4 may well explain the failure of differentiation in Dicer-null ESCs. Transfection of miR-290 family was shown to partially rescue this DNA methylation phenotype by restoring the levels of Dnmt3a and Dnmt3b, but not that of Dnmt1. Such restoration is mediated by miR-290 family miRNAs suppressing their direct targets, retinoblastoma-like 2 protein (Rbl2), a transcriptional repressor of Dnmts. Specifically, upon transfection of miR-290 cluster, Oct4 promoter methylation was restored 40, 43. One less explored question is whether miRNAs also shape the histone code to regulate ESC self-renewal and differentiation.
Five years ago, Yamanaka’s group discovered that fibroblasts could be reprogrammed into induced pluripotent stem cells (iPSCs) by expressing Sox2, Oct4, Klf4 and c-Myc 44. The combination of Sox2, Oct4, Lin28, and Nanog also achieved the iPSC reprogramming 45. As we are gaining more and more insights into the mechanisms underlying somatic reprogramming, the roles of miRNAs in regulating this process have begun to be explored. miRNAs regulate somatic reprogramming at two different levels, cell proliferation and epigenetic programming. At the cell proliferation level, ES cell cycle-specific miRNAs miR-291-3p, miR-294 and miR-295 can replace c-Myc to facilitate Sox2-, Oct4- and Klf4-induced somatic reprogramming with a similar efficiency 46. Given the fact that these miRNAs are under the control of c-myc and other reprogramming transcription factors and that these miRNAs are not able to further increase somatic reprogramming efficiency in the presence of c-myc, it is likely that these miRNAs are downstream effectors of c-myc-mediated cell proliferation pathway 13, 29.
In addition to replacing c-Myc, miRNAs can promote reprogramming through epigenetic regulation. It was already known that the decreased Dnmt activity leads to elevated efficiency of iPS generation 47, 48. miR-302, abundantly expressed in ES cells, was recently shown to regulate Dnmt and enhance iPS generation 49. In hair follicle cells, overexpression of miR-302 to levels comparable to those of hESCs leads to suppression of AOF2, a stabilizer for Dnmt1. Therefore, induced expression of miR-302 promotes global DNA demethylation and in turn fosters iPS generation. Consistent with this, miR-302b and miR-372 were recently also shown to promote reprogramming of human fibroblast to iPS, through targeting multiple targets and regulating cell cycle, epithelial-mesenchymal transition, epigenetic regulation and vesicular transport50. In addition, as previously discussed, the miR-290 family suppresses the repressor of Dnmt, therefore promoting Dnmt activity. Therefore, it would be of interest to examine whether miR-290 family antagomirs could enhance the somatic reprogramming efficiency.
Tissue stem cells reside in diverse types of adult tissues and are self-renewing progenitor cells responsible for the establishment and/or maintenance of their resident tissues. Current studies on miRNAs have revealed their roles in ectodermal and mesodermal tissues. In this section, we will first use germline stem cell system to illustrate the miRNA regulation of tissue stem cell in Drosophila and C. elegans, sine the germline represents most of the studies on miRNAs in stem cells in lower model systems. We will then highlight one mammalian ectodermal tissue (neural system) and three mammalian umesodermal tissues (blood, muscle, and bone) in which the role of miRNAs is best known. It is only a matter of time before the role of miRNAs in endodermal tissues will be revealed..
Gametogenesis is a highly active process driven by germline stem cells in the gonad, with the exception of oogenesis in mammals. At present, little is known about the role of miRNAs in the C. elegans and mammalian germline. However, the Drosophila ovary provides a powerful platform to dissect the role of miRNA in regulating germline stem cells. During Drosophila oogenesis, germline stem cell (GSC) resides at the anterior tip of the ovary in direct contact with their niche cells called cap cells, and divides asymmetrically to produce a daughter GSC and a differentiating cell called the cystoblast. The daughter GSC remains anchored to cap cells; whereas the cystoblast undergoes further oogennic differentiation 51. miRNA was found to play a role in regulating GSC division and maintenance when Loquacious, Dicer-1 and Ago-1 were demonstrated to be important in Drosophila oogenesis52–55. More specifically, bantam miRNA was shown to be essential for GSC maintenance, where bantam represses primordial germ cell differentiation and regulates GSCs as an extrinsic factor56–58. In addition, miR-7 and miR-278 were found to regulate the cell cycle of GSCs. miR-278 depletion causes GSCs to divide slower whereas miR-7 depletion results in abnormal cell cycle progression. Such two microRNAs targets the 3′ UTR of dacapo mRNA, which encodes a cyclin-dependent kinase inhibitor, that governs the G1/S transition59–61. These results illustrate a common theme where multiple miRNAs converge to regulate the same pathway to fine tune the developmental process (Fig. 4B).
Neurogenesis, starting from neural stem cells and neural progenitor cells, yields new neurons and supporting cells during both embryonic development and adult neural system maintenance (Fig. 3). In the nervous system, recent progress has identified several miRNAs important for neural development in multiple model organisms. In zebrafish, depletion of maternal and zygotic Dicer causes severe morphogenesis defect including incomplete neural tube closure, indicating that miRNAs regulate brain morphogenesis 62. In Drosophila, miR-9a inhibits excess sensory organ precursors production by targeting Sens 63, 64, whereas miR-124a promotes dendritic branching of dendritic arborization sensory neurons by regulating unknown targets65.
The study of miRNAs in mammalian neurogenesis is currently focused on miRNAs that show abundant or exclusive expression in the brain, or rapidly increased expression upon differentiation of ESCs to neural stem cells. The latter include miR-9 and miR-124. The respective nucleotide sequences of these two miRNAs are highly conserved 66.
A miRNA can regulate different mRNA targets at different stages of neurogenesis. For example, miR-9 stimulates the division and limits migration of ESC-derived neural progenitors by regulating its target Stathmin in hESCs 67. During the later differentiation of multipotent neural stem cells, miR-9 promotes the neuronal differentiation and suppresses neuronal stem cell self-renewal by down-regulating the expression level of its target, TLX, a highly conserved nuclear receptor (Fig. 3) 68. Conversely, miR-9 is also subject to the regulation of TLX, as the promoter region of miR-9 is occupied by TLX and the corepressor HDAC5. Such a feedback regulatory loop is exploited to ensure the delicate regulation of key players in developmental processes. Such feedback loop raises new questions about the sequence of actions for miR-9 and transcription factor TLX. Since both miR-9 and TLX coexists during the neuronal differentiation, it would be interesting if one could examine the temporal order of miR-9 and TLX expression to determine whether miR-9 regulate TLX first or vice versa. In addition to promoting neuronal differentiation, miR-9 also inhibits the astrocytic fate 69. In addition, miR-9 and miR-124 act in a concerted manner to control neurogenesis in ESCs 67, 69. Both miRNAs contribute to the decreased STAT3 phosphorylation level, thereby inhibiting the astrocytic fate during neural differentiation.
Recently, miR-9* and miR-124 were found to guide the directly convertion of human fibroblasts to neurons through SWI/SNF-like BAF chromatin-remodelling complexes. This indicates that miRNA could control cell lineage through epigenetic regulation70.
From the above review, three common principles underlying miRNA-mediated regulation have emerged (Fig. 4): (1) miRNAs act as an fine-tuning developmental switch in a stage-and tissue-specific manner (Fig. 4A); (2) different miRNAs converge to control the same signaling pathways by regulating one or more component simultaneously (Fig. 4B); and (3) miRNAs and key developmental regulators form feedback loops to ensure delicate control of fundamental biological processes (Fig. 4C). These modes of miRNA regulation repeatedly appear as common themes in other tissues, as further reviewed below. However, these are not parallel or exclusive modes in action. Often, the fine-tuning function of miRNAs is achieved via either their targeting of signaling pathways and/or forming feedback loops with other types of developmental regulators such as epigenetic factors or transcriptional factors.
Hematopoiesis, initiated by multipotent hematopoietic stem cell (HSC) and giving rise to all types of blood cells, is a multi-stage and highly regulated proliferation and differentiation process. Since little is known about the function of miRNA in hematopoiesis in Drosophila, this section is dedicated to mammalian hematopoiesis. During hematopoiesis in mammals, transcriptional factors have differential expression patterns in hematopoietic stem cells and different progenitors, where they control gene expression, which in turn regulate lineage specification and differentiation of those stem cells and progenitors. More and more recent studies show that miRNAs also exhibit lineage specific expression profile and control stem cell self-renewal and differentiation, yet through a different mechanism--posttranscriptional regulation. The aforementioned features of miRNA-mediated regulation make them prime candidates as key regulators of hematopoiesis.
A common feature of miRNA regulation in hematopoiesis is that miRNAs display distinct stage- and lineage-specific expression profiles and functions (Fig. 3) 71. Based on bioinformatic, miRNA profiling, and reverse genetic analyses, the role of miRNAs in regulating hematopoiesis has been increasingly revealed. For example, miR-181 and miR-128 are highly enriched in early hematopoietic stem cells (HSCs), and maintain the stem cell identity by limiting the differentiation of HSCs into all hematopoietic lineages. On the other hand, miR-223 promotes the differentiation of HSCs 72. Therefore, miRNAs with opposite effect act together on the same HSC differentiation pathway to exert tight regulation. As another example of such antagonistic regulation, miR-155, miR-24a, and miR-17 inhibit multipotent progenitor differentiation into myeloid lineage; whereas miR-146 and miR-181 inhibit and promote the lymphoid lineage differentiation, respectively. Such regulation also occurs during the terminal differentiation. miR-181a, highly enriched in the CD4+CD8+ double positive cells, promotes the T-cell positive selection 73, whereas miR-150 acts as a block to limit pro-B to pre-B transition 74.
The second common feature of miRNA regulation of hematopoiesis is the feedback loop between miRNAs and transcription factors, whereby miRNAs modulate the expression level of transcription factors or the inbihitors and stabilizers of transcription factors. Conversely, transcription factors bind to miRNA genes and regulate their transcription. For example, during granulopoiesis, the interplay among miR-223, transcription factors NFI-A, and C/EBPα reveals the fine and concerted control of granulocytic differentiation 75. NFI-A binds to the promoter region of pre-miR-223 and represses miR-223 expression, while upon retinoic acid (RA)-induced differentiation, C/EBPα outcompetes NFI-A for the same promoter region and positively regulates miR-223 expression. The replacement of NFI-A by C/EBPα and the resulting granulocytic differentiation is also strengthened by the feedback loop regulation of miR-223, which suppresses NFI-A expression at the posttranscriptional level.
The third common feature is that one miRNA could regulate several different lineages. For instance, miR-155 controls both T cell and B cell differentiation. On one hand, miR-155 regulates B cell polyclonal expansion, as evidenced by the pre-B cell over proliferation in Eμ-miR-155 76. On the other hand, miR-155 is important for T cell differentiation into Treg and Th1, as miR-155 knockout mice show reduced Treg cells, immunodeficiency and compromised inflammation77, 78. This one-to-many relationship between miRNAs and biological processes is accompanied by the combinatorial regulation of miRNAs to ensure accurate and tight control over stem cell proliferation and differentiation.
Bone formation (osteogenesis) consists of a series of differentiation events starting from multipotent mesenchymal stem cell and resulting in the development of the bone. The commitment of mesenchymal stem cell to the bone lineage is mediated by transcription factors Dlx5, Runx2, and growth factor BMP. Recent studies showed that the interaction between miRNAs and these key regulators are also very important. For example, miR-125b inhibits BMP4-mediated mesenchymal stem cell differentiation to osteoblasts in mice79, while miR-141 and miR-200a suppress pre-osteoblast differentiation through targeting transcription factor Dlx5 80. As opposed to these miRNAs, miR-2861 shows a stimulatory effect on BMP2-induced osteoblast differentiation, in part by maintaining the expression level of Runx2 81. Interestingly, this role is mediated through miR-2861 targeting an epigenetic factor, HDAC5, which accelerates the degradation of Runx2.
miRNAs also play a role in chondrogenesis, which is initiated by the differentiation of mesenchymal stem cells into chondrogenic progenitors. A BMP-2 early responsive target miRNA, miR-199a*, inhibits early stage of chondrogenesis and the production of cartilage oligomeric matrix protein and collagen, primarily through inhibiting Smad1 posttranscriptionally 82.
Like the bone, skeletal and cardiac muscles are derived from mesoderm progenitors. Skeletal muscle development is subject to the stimulatory miRNAs (miR-1, -26a, -206 and -214) and inhibitory miRNAs (miR-133, -221 and -222). Notably, miR-1 and miR-26a execute such regulation through targeting epigenetic factors, Hdac4 and Ezh2, respectively83–85. During the development of cardiac muscle development, positive and negative feedback regulatory loops co-exist. In the positive feedback loop, transcription factor MEF2 activates the expression of miR-1, which then down-regulates the epigenetic repression of HDAC towards MEF2, which further elevates the expression of MEF2. This positive feedback loop could ensure a rapid and lasting control of cardiac muscle development. On the other hand, a negative feedback loop could allow precise and fine regulation, where miR-133 suppresses the transcription factor SRF that activates the expression of miR-133. Therefore, the recurring theme of feedback loop mechanism enables precise control of key biological processes. Unlike cardiac and skeletal muscles, smooth muscle is produced from the ectoderm-derived neural crest stem cell. Recently, miR-145 was shown to be necessary for neural crest stem cell differentiation into smooth muscle cells, through targeting transcription factors Klf4, Myocardin, and Elk1 86.
Given the complex and fine-tuned nature of miRNA regulation of gene expression, its deregulation may lead to a variety of diseases including cancer. Indeed, miRNA expression profiles are often skewed in tumor cells 87, 88. Moreover, certain miRNAs display oncogenic or tumor suppressive properties 89. Therefore, miRNAs show promising potential as tools for cancer diagnosis, prognosis and treatment. Mounting evidence indicates that miRNA expression profile can be used to distinguish between normal tissues and tumor tissues, and among different cancer subtypes 90–92. More specifically, in a breast cancer study, nine miRNAs provide the discriminating power to classify basal vs. luminal cancer subtypes 93. In addition to diagnosis, miRNAs also have prognostic value. For example, miR-21 has the prediction power for breast cancer aggression, whereas high miR-155 and low let-7a levels are associated with poor survival 94, 95. Based on this, it would be of interest to examine whether these miRNAs also could predict the patient’s response to treatments. Moreover, miRNAs have also been used in treating cancer. Viral delivery of let-7 and miR-26a to mice suffering from lung cancer and liver cancer exhibited decreased tumor growth, through targeting oncogenes or cell cycle regulators, respectively 96–98. As seen in the success of these studies, miRNAs showed unique therapeutic advantages in cancer, probably because of the low risk of off-target effect and the broader impact on their target genes. However, a viral miRNA delivery system also has its limitations. First, the viral delivery system imposes the risk of adverse effect including immune response, viral integration into genome, and cytotoxicity. Second, such delivery has minimal, if any, cellular targeting specificity, thus affecting both cancer cell and normal cells. One possible approach to address these drawbacks is to develop tumor cell-targeting liposome, which employs Transferrin (Tf) to target Tf-Receptor enriched tumor cells 99. With rapid new development in miRNA-based therapy, we are entering an age when the fundamental research on miRNAs will soon be applied to clinical practice.
We are grateful to Robert Ross, Jianquan Wang, and anonymous reviewers for critical reading of this review. The research in the Lin lab is supported by the NIH (DP1OD006825, R01 HD33760, and R01HD42012), the G. Harold and Leila Y. Mathers Foundation Award, and the Ellison Medical Foundation Senior Scholar Award, and the Connecticut Stem cell Research Funds (08-SCD-Yale-004).