Author contributions: W.L. and K.J.: manuscript writing, S.D.: manuscript writing and conception and design.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS October 25, 2011.
Stem cells are essential for maintaining tissue homeostasis and mediating physiological and pathological regeneration. Recent breakthroughs in stem cell biology have generated tremendous enthusiasm and hope for the therapeutic potential of stem cells in regenerative medicine. However, this research is still in an early development stage. An improved understanding of stem cell biology is required to precisely manipulate stem cell fate and to harness these cells for regenerative medicine. Small molecules, targeting specific signaling pathways and mechanisms, are powerful tools for manipulating stem cells for desired outcomes. Those small molecules are increasingly important in probing the fundamental mechanisms of stem cell biology and facilitating the development of therapeutic approaches for regenerative medicine. These could involve cell replacement therapies with homogenous functional cells produced under chemically defined conditions in vitro and the development of small-molecule drugs that modulate patient's endogenous cells for therapeutic benefit. STEM CELLS2012;30:61–68
Stem cells possess two fundamental characteristics: they can self-renew themselves and can differentiate into an array of specific cell types. They have essential roles in generating the hierarchical cellular lineages during development, maintaining tissue homeostasis, and mediating physiological/pathological regeneration in adults. These properties and functions make stem cells excellent model systems to study the basic biology of human development and tissue homeostasis and also offer significant promise for developing treatments for devastating human diseases and injuries.
However, before we can realize the promise, several obstacles must be overcome. For example, renewable sources of stem cells must be developed for any therapeutic applications. Although significant progress has been made in maintaining embryonic stem cells (ESCs) when compared to the past decades, many substantial challenges remain in isolating and expanding most tissue-specific adult stem cells. To fully harness their clinical potential, functional expansion of these therapeutically valuable adult stem cells is needed. In addition, although ESCs can self-renew infinitely and generate any cell types under appropriate conditions, they are prone to cause teratomas and cannot be directly used to repopulate host tissues in vivo before they differentiate into tissue-specific cells. Great efforts are still required to improve our ability to coax stem cells, especially the ESCs, into the desired developmental stages (e.g., linage-specific stem/progenitor cells) or functional cells for disease therapy.
Small molecules, modulating specific target(s) in the signaling pathways or epigenetic mechanisms, are emerging as valuable tools with distinct advantages for manipulating stem cell fates [1, 2]. For example, regulating protein functions is much easier with small molecules than by genetic manipulation. Importantly, the effects of small molecules are typically rapid and reversible and can be fine tuned by varying concentrations and combinations of small molecules. These characteristics provide temporal and flexible regulation of complex signaling networks. In addition, virtually unlimited structure and functionality diversity endowed by synthetic chemistry provide small molecules with theoretically unlimited potential for precisely controlling cell phenotypes, which could be extensively explored by phenotype-based high-throughput screening.
As a nascent field, stem cell research will continue to benefit from its crossover with chemistry. In this review, we discuss the new developments of chemical approaches to stem cell biology and regenerative medicine. The examples are not intended to be comprehensive. Rather, we want to emphasize the conceptual points, current challenges, and potential opportunities for this emerging research field.
STEM CELL SELF-RENEWAL
The derivation of ESCs from mice and subsequently from human and other species represents one of the major milestones in genetics, developmental biology, and human biomedical research [3–5]. Extensive efforts have been made to develop better ways to maintain self-renewal of these versatile cells.
Supporting ESC self-renewal with small molecules under chemically defined conditions has particular advantages. By using green fluorescent protein (GFP) expression under control of the Oct4 promoter as a primary indicator of pluripotency, we screened synthetic small-molecule libraries under chemically defined conditions in the absence of feeder cells, serum, and leukemia inhibitory factor (LIF) . A novel compound, pluripotin/SC1 (Table 1; S1), was identified that maintains long-term self-renewal and germline competence of mouse ESCs (mESCs) in vitro by dual inhibition of two endogenously expressed differentiation-inducing proteins, RasGTPase activating protein and extracellular signal-regulated kinase-1 (ERK1) . This proof-of-concept study demonstrated that modulators of stem cell fate can be identified by carefully designed phenotypic screens. More importantly, the fact that pluripotin maintains ESC self-renewal, independent of the exogenous activation of conventional self-renewal pathways, by simply inhibiting the activity of endogenous differentiation-inducing proteins has provided a fundamental new view on the mechanism of ESC self-renewal. Thus, ESCs have an intrinsic ability to maintain pluripotency and do not require exogenous stimulation. A more recent study supports this conceptual advance. A combination of specific chemical inhibitors (CHIR99021 and PD0325901, Table 1; S2 and S3) of glycogen synthase kinase-3 (GSK3) and mitogen-activated protein kinase/ERK kinase (MEK) similarly supported the derivation and long-term self-renewal/germline competence of mESCs in the absence of exogenous cytokines . Those small molecules provide a platform for generating pluripotent cell lines from refractory mouse strains or other species, for example, pluripotent cell lines from nonobese diabetes/severe combined immunodeficiency (NOD-SCID) and SCID beige mice [8, 9], and rats [10–13]. Notably, these small molecules are also used to capture the naïve, mESC-like human pluripotent stem cells (hPSCs). Conventional human ESCs (hESCs) correspond very closely to epiblast stem cells, which are derived from the postimplantation egg cylinder-stage epiblasts of mouse [14, 15], and display very different gene expression and signaling dependency for self-renewal/differentiation from mESCs, which are derived from inner cell mass of preimplantation blastocysts. For example, both LIF and bone morphogenetic protein 4 (BMP4) are typically used for maintaining the pluripotency of mESCs [16, 17]. Also, inhibition of the MEK-ERK pathway promotes mESC self-renewal . In contrast, hESCs typically depend on basic fibroblast growth factor (bFGF) and activin A for long-term self-renewal, and LIF does not promote hESC self-renewal , and BMP4 induces differentiation of hESCs . By combining genetic reprogramming and cell signaling modulation by small molecules that favor the naïve pluripotent state, mESC-like human induced PSCs (m-hiPSCs) were generated from human fibroblasts by expressing reprogramming factors in culture medium that contains human LIF . m-hiPSCs form small domed colonies and display stable long-term self-renewal when cultured in the presence of three chemicals, PD0325901, A-83-01, and CHIR99021. A-83-01 is a small-molecule inhibitor of the transforming growth factor β (TGFβ)/activin receptors (Table 1; S4). Recently, another study reported that hESCs could be stably maintained under the combination of bFGF, CHIR99021 and PD0325901 . However, the cells cultured under this condition seemed to resemble the conventional hESCs. This study also showed that undifferentiated hESCs were maintained only under low concentration of CHIR99021, and hESCs would differentiate at higher concentrations of CHIR99021. This could be due to specificity of CHIR99021 and/or dosage effect of signaling pathway modulation. Such differential dose-dependent effects are not uncommon for both small molecules and growth factors/cytokines. Nevertheless, considerations on small molecule's specificity must be taken when interpreting their affected biological phenotype and mechanism. For a more thorough overview of small molecules that maintain ESC self-renewal, readers are encouraged to examine comprehensive reviews on the topic [1, 22-24].
Table 1. Structures of small-molecular modulators of stem cell fate
In contrast to robust ESC self-renewal conditions, long-term self-renewal of tissue-specific stem cells remains challenging. Here, we discuss new developments and possible strategies for expanding tissue-specific stem cells, which are directly applicable to regenerative medicine. Although tissue-specific stem cells exist in many adult tissues, and many of them have considerable self-renewing capacity under physiologically or pathologically regenerative conditions, it is technically challenging to expand most types of tissue-specific stem cells ex vivo. These challenges might reflect currently limited understanding of the extremely complex in vivo stem cell microenvironment. Before thorough dissection of the mechanisms for stem cell microenvironment, which is essential to rationally devise appropriate conditions for stem cell self-renewal, phenotypic screening (e.g., using the expression of stem cell markers as readout) of small-molecule libraries represents a fertile approach to identify the conditions that expand tissue-specific stem cells.
Using expression of CD34 as a readout to screen small-molecule libraries, Boitano et al.  identified a purine derivative, SR1 (Table 1; S5), that promotes ex vivo expansion of primary CD34-positive hematopoietic stem cells (HSCs) from human cord blood. Treatment with SR1 led to a 50-fold expansion of CD34-positive cells and a 17-fold increase in cells that functionally repopulated the hematopoietic system of NOD/SCID mice. Follow-up studies showed that SR1 promotes HSC expansion by directly antagonizing the aryl hydrocarbon receptor, representing a new mechanism to modulate HSC expansion. This study again reinforces the notion that unbiased phenotypic screen is useful to probe novel mechanisms for controlling stem cell fate.
Because of limited donor cell sources and often invasive nature of cell isolation from adults, an alternative approach to obtain tissue-specific stem cells is to differentiate PSCs (e.g., ESCs) that have unlimited supplies. Similarly, capturing and stably expanding hESC-derived tissue-specific stem/progenitor cell types remain a significant challenge for translating hESCs toward various in vitro and therapeutic applications. Recently, we identified novel combinations of small molecules for either inducing or expanding primitive neural stem cells (pNSCs) from hESCs in culture . We found that synergistic inhibition of GSK3, TGFβ, and Notch signaling pathways by small molecules efficiently converted monolayer-cultured hESCs into homogenous primitive neuroepithelium within 1 week under chemically defined conditions. Importantly, these pNSCs represent the prerosette stage neuroepithelia and stably self-renew in the presence of LIF, GSK3 inhibitor (CHIR99021) and TGFβ receptor inhibitor (SB431542, Table 1; S6), which are distinct from previously identified neural precursor cells that typically depend on bFGF and epidermal growth factor (EGF) as mitogens. Most remarkably, after long-term passages under the small molecule condition, these pNSCs maintain highly neurogenic differentiation propensity, remain plastic to instructive regional patterning cues toward midbrain and hindbrain neuronal subtypes, and exhibit in vivo functions. This study provided a “check-point” strategy to get around the issues that the typical hESC differentiation is a nonstop process and impurities of differentiated cells from each step of differentiation are carried over leading to low efficiency and significant heterogeneity in terminally differentiated cells.
STEM CELL DIFFERENTIATION
Typically, applications based on hPSCs (e.g., ESCs or iPSCs) require their in vitro differentiation into a desirable cell population. Although significant progress has been made over the years on ESC differentiation into a wide variety of cell types , we focus here on some of the existing challenges and newly developed strategies by applying small molecules. In addition to the strategy of capturing and maintaining the intermediate stem/progenitor cells during hPSC differentiation discussed above, substantial efforts are highly desirable to more efficiently induce hPSC differentiation in a homogenous manner under chemically defined conditions.
Chemical approaches have been particularly useful for accelerating differentiation process, increasing differentiation efficiency, and normalizing different differentiation propensity of diverse hPSC lines . Based on known mechanisms of neural development and hESC differentiation, Chambers et al.  developed an efficient neural induction method for hPSCs that bypasses the conventional embryoid body formation. They found that combination of Noggin (a natural BMP antagonist) and SB431542 (TGFβ receptor inhibitor) promotes rapid neural induction of more than 80% of hESCs in a monolayer fashion. Those two signaling pathway inhibitors appear to function synergistically to destabilize self-renewal of hESCs (e.g., TGFβ signaling is essential for self-renewal of hESCs), promote neural induction, and prevent cells from differentiating into trophectoderm, mesoderm, and endoderm lineages (for which BMP and TGFβ signaling have an inductive effect). This study suggested that directed PSC differentiation toward a specific lineage can be achieved by deliberately combining the inductive signals for the desired cell lineage and the inhibitory signals blocking PSC self-renewal and their differentiation toward undesired lineages.
Recent efforts have also focused on discovery approaches to identify small molecules for certain steps during ESC differentiation toward specific lineage. Using mESCs stably transfected with the dTomato reporter gene under the control of the Sox17 promoter, Borowiak et al. screened a collection of 4,000 compounds for small molecules that could induce definitive endoderm (DE) induction in the absence of activin A (a typically used DE inducer). Two structurally similar small molecules, IDE1 (Table 1; S7) and IDE2, were found to induce DE differentiation in up to 80% of mESCs (or 50% of hESCs) in the absence of activin A . Similar to activin A, both IDE1 and IDE2 induce Smad2 phosphorylation in mESCs, while their targets remain unknown. However, IDE1 and IDE2 seem to share some properties with 1m (Table 1; S8), a GSK3 inhibitor that can transiently upregulate NODAL expression and induce DE from hESCs under chemically defined condition . The endoderm-like cells induced by IDE1 and IDE2 were shown to have the ability to differentiate into pancreatic lineage when they were subsequently treated with another small molecule, Indolactam V (Table 1; S9), which was identified in a separate screen for small molecules that can induce Pdx1 expression from hESC-derived DE cells . Indolactam V is an activator of protein kinase C (PKC), revealing a potential role of PKC during pancreatic development.
iPSCs generated from somatic cells by overexpression of defined transcription factors have attracted enormous interest [33, 34]. The simplicity of such genetic reprogramming approach has opened up unprecedented opportunities to generate patient-specific cells for disease modeling and potential therapeutic applications without the controversies associated with hESCs. However, there are critical concerns that the genetic technique initially used to generate iPSCs might result in genome modifications by oncogenes and potentially harmful genetic and epigenetic alterations in target cells. Some key advances toward overcoming these safety concerns have been achieved with nonintegrating gene delivery methods [35–37], using cell penetrating recombinant proteins or repeated transfection of synthetic reprogramming mRNAs [38–40]. Nevertheless, new methods for generating iPSCs with better qualities (e.g., as identical to ESC as possible) through improved efficiency and specificity in the process are highly desirable.
An alternative method to using transcription factors is to activate endogenous reprogramming mechanisms through small molecules that not only can provide a better nongenetic reprogramming approach but also ultimately will fundamentally change the reprogramming (toward a directed and specific process). We and others have identified small molecules with various mechanisms of action that can exert powerful effects on enhancing reprogramming and replacing transcriptional factors . Using formation of compact colonies that express GFP under the control of Oct4 promoter as a readout, we first screened chemical collections in neural progenitor cells for reprogramming small molecules and identified a small-molecule inhibitor of G9a histone methyltransferase, BIX-01294 (Table 1; S10), that can substitute Oct4 and significantly improve reprogramming efficiency . It was further demonstrated that BIX-01294 can also enable the reprogramming of mouse embryonic fibroblasts (MEFs) into iPSCs in the absence of Sox2 expression by only two exogenous factors Oct4 and Klf4 . A subsequent chemical screen in fibroblasts with BIX-01294 identified a DNA methyltransferase inhibitor, RG108 (Table 1; S11), and a L-type calcium channel agonist, BayK8644 (Table 1; S12), that can work synergistically with BIX-01294 to increase reprogramming efficiency . Consistent with epigenetic mechanisms in reprogramming, several studies also showed other commonly used, small-molecule inhibitors of epigenetic enzymes, including histone deacetylase inhibitors (e.g., valproic acid, Table 1; S13) could improve mouse and human somatic cell reprogramming [43–45]. In particular, valproic acid enabled reprogramming of human fibroblasts into iPSCs with only two factors (Oct4 and Sox2)  and MEF reprogramming with recombinant cell-penetrating reprogramming proteins . In addition to these direct epigenetic modulators, we found in another study that GSK3 inhibitor CHIR99021 can facilitate reprogramming of MEFs by only Oct4 and Klf4. When combined with Parnate (Table 1; S14), a lysine-specific demethylase 1 inhibitor, CHIR99021 could enable reprogramming of human primary keratinocytes by only Oct4 and Klf4 . Lyssiotis et al.  identified another GSK3 inhibitor, kenpaullone (Table 1; S15), that can replace Klf4 in reprogramming MEFs transduced with Oct4, Sox2, and c-Myc. However, because kenpaullone was shown to inhibit various other kinases, its mechanism in facilitating reprogramming remains elusive. Through a hypothesis-driven study, we found that the dual inhibition of MEK and TGFβ by PD0325901 and SB431542 dramatically improved (>100-fold) the generation of iPSCs from human fibroblasts within 7 days of treatment (14 days of transfection) with an efficiency of >1% by enhancing mesenchymal–epithelial transition . Other concurrent studies showed that TGFβ receptor inhibitors can replace Sox2 in MEF reprogramming [49, 50].
Notably, we recently identified a small-molecule activator of 3′-phosphoinositide-dependent kinase-1 PS48 (Table 1; S16) that enabled the reprogramming of human primary cells transduced with only Oct4 when combined with sodium butyrate (a histone deacetylase inhibitor), A-83-01, and PD0325901 . In-depth mechanistic studies revealed that PS48 acts at the early phase of reprogramming at least in part by inducing a metabolic switch from mitochondrial oxidation (differentially used by adult somatic cells) to glycolysis (almost exclusively used by PSCs) during the reprogramming process. It was further demonstrated that additional small molecules that promote glycolytic metabolism also enhance reprogramming, including fructose 2,6-bisphosphate (Table 1; S17) (an activator of phosphofructokinase 1, a key rate-limiting enzyme of glycolysis), and N-oxaloylglycine and Quercetin (Table 1; S18) (both stimulate glycolytic genes by activating hypoxia-inducible factor-1). In contrast, a specific glycolysis inhibitor (2-deoxy-D-glucose) inhibits reprogramming without altering cell proliferation, which can be potentially used for eliminating undifferentiated PSCs from their differentiation culture. This study suggested that metabolism modulation represents another fundamental mechanism in somatic cell reprogramming, in addition to other direct epigenetic and signaling mechanisms.
Induced pluripotency is established in a stepwise and stochastic fashion [52, 53]. Only a rare subset among various intermediate cells finally becomes pluripotent under extended expression of reprogramming factors and favorable culture conditions. We reasoned that it might be possible to guide those initial epigenetically unstable cells (induced by the iPSC-reprogramming factors) into lineage-specific cell types under favorable condition without traversing pluripotency (Fig. 1). We found that through temporally restricting ectopic overexpression of iPSC factors in fibroblasts, epigenetically “activated” cells could be generated rapidly, which can then be coaxed to “relax” back into certain differentiated state by each specific culture conditions (that favor lineage-specific cell types and simultaneously inhibit the establishment of pluripotency), ultimately giving rise to somatic cells entirely distinct from the starting population. For example, we found that with as little as 4 days of the iPSC-factor expression (far shorter than what is required for induction of pluripotency), MEFs can be directly reprogrammed to spontaneously contracting cardiomyocytes over a period of 11–12 days under the treatments with a small molecule Janus Kinase (JAK) inhibitor for the first 9 days (that blocks establishment of pluripotency by inhibiting the LIF signaling) , and BMP4 from day 9 onward (that mediates cardiac mesoderm induction). Interestingly, extending JAK inhibitor treatment beyond 9 days to overlap with BMP4 treatment was detrimental for the induction of cardiomyocytes. This observation is consistent with previously reported requirement for JAK/signal transducer and activator of transcription signaling in cardiomyogenesis [55, 56]. Applying the same concept and approach, neural and definitive endodermal cells were directly reprogrammed from fibroblasts rapidly and efficiently using transient expression of iPSC factors and treatments with FGFs/EGF (toward neural cells)  or activin A (toward definitive endodermal cells) (Fig. 1). In comparison to transdifferentiation using overexpression of tissue-specific transcription factors [58, 59], our iPSC-factor-based transdifferentiation paradigm has a number of advantages: it is a single combination of transcription factors that is applicable to induce reprogramming toward various lineage-specific cell types; its transient expression could be more easily replaced by nonintegrating or nongenetic methods; and most significantly, progenitor populations belonging to these lineages are generated in the process, which can be isolated and expanded for various applications . Such direct reprogramming to proliferating progenitors will dramatically increase the utility of this transdifferentiation paradigm.
It is worthy to note that the functions/effects of many small molecules discussed could be highly dependent on the specific culture conditions. Various elements, including the presence of undefined supplements (such as serum) or even the protein concentration in culture media, could have impact on the effectiveness of small molecules in a specific context.
SMALL MOLECULE THERAPEUTICS MODULATING STEM CELLS
Chemical approaches could facilitate translation of stem cell research into clinical applications in at least two ways. First, as mentioned above, chemical approaches could provide robust tools to precisely manipulate stem cell fate and function in vitro to generate sufficient number of safe, homogenous, and functional cells for cell therapy. However, development and manufacture of cell-based therapy typically are more complex and such therapy also costs more for patients than conventional small-molecule and protein therapeutics. Many issues in cell-based therapy even with transplantable cells, including immune-related ones, cell homing, engraftment, and long-term maintenance of transplanted cells' functions in the target tissue remain challenging. Alternatively, chemical approaches also offer a complementary strategy by directly modulating endogenous tissue-specific stem/progenitor (or even more differentiated) cells in vivo for therapeutic benefits.
Recently, Zaruba et al.  described a small-molecule-based regenerative strategy for myocardial infarction by enhancing the recruitment of endogenous bone marrow stem/progenitor cells to the heart through inhibition of CD26/dipeptidylpeptidase IV in vivo via a chemical compound, ultimately increasing the formation of new blood vessels and improving heart functions. In ischemic heart tissue, stromal cell-derived factor 1α (SDF-1α) is the major chemokine attracting endogenous endothelial progenitors expressing SDF-1α receptor (C-X-C chemokine receptor type 4, CXCR4) homing to the heart. However, SDF-1α is sensitive to a number of protease (including CD26) cleavages. The authors demonstrated that combined administration of granulocyte-colony stimulating factor (functions to mobilize stem/progenitor cells, including endothelial progenitors, from bone marrow) and a CD26 inhibitor Diprotin A (Table 1; S19) intraperitoneally enhanced recruitment of CXCR4-positive stem/progenitor cells to myocardium and improved myocardial function by increasing neovascularization, leading to increased animal survival. This study represents an excellent example of using small molecule in vivo to modulate endogenous stem/progenitor cells behavior (i.e., homing to injury site) for tissue repair. Similar strategies might entail modulation of endogenous stem/progenitor cell fate (e.g., survival, expansion, differentiation, and reprogramming), behavior (e.g., migration and niche interactions), and state/function (e.g., quiescence and polarization) by small-molecule and/or protein therapeutics to achieve tissue repair and regeneration.
To avoid in vivo systemic exposure of small-molecule drugs that may have side effects on other tissues/organs given their ability to modulate key developmental signaling pathways, another strategy is to modulate stem/progenitor cells ex vivo to enhance their functions for transplantation. North et al. screened a collection of biologically active compounds using zebrafishes to identify the modulators of HSC induction in the zebrafish aorta–gonad–mesonephros region, where the first definitive HSCs primarily arise. They found a number of small molecules that enhance prostaglandin E2 (PGE2) synthesis, and PGE2 itself, increased HSC numbers in zebrafish [61, 62]. Ex vivo temporal treatment of murine and human HSCs with 16,16-dimethyl-PGE2 (dmPGE2) (Table 1; S20), a more stable analog of PGE2, enhanced their engraftment in vivo possibly through induction of genes involved in HSC homing, including CXCR4. These findings led to rapid clinical studies of short-term ex vivo-treated human cord blood cells by dmPGE2 for improved transplantation in adult patients with hematologic malignancies . Similar clinical studies with human cord blood cells temporally treated with a CD26 inhibitor are also ongoing based on the HSC homing mechanism on the SDF-1α–CXCR4 axis.
Although stem cell research and regenerative medicine are still in an early development stage, they have had substantial growth in recent years. In particular, the iPSC technology has generated tremendous enthusiasm and efforts to explore their various applications. Now, chemical approaches are becoming increasingly accessible and valuable in discovery biology and have already played an essential role in stem cell research and regenerative medicine. It is clear that chemical approaches in precisely controlling cell fate, behavior, and state/function will continue to open up new opportunities for the field of stem cell biology and regenerative medicine.
Sheng Ding is supported by funding from NICHD, NHLBI, NEI, and NIMH/NIH, California Institute for Regenerative Medicine, Prostate Cancer Foundation, and the Gladstone Institute. We thank Gary Howard for editing of this manuscript. We apologize to all scientists whose research could not be properly discussed and cited in this review owing to space limitations.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.