How microRNAs facilitate reprogramming to pluripotency

Summary

The ability to generate pluripotent stem cells from a variety of cell and tissue sources through the ectopic expression of a specific set of transcription factors has revolutionized regenerative biology. The development of this reprogramming technology not only makes it possible to perform basic research on human stem cells that do not have to be derived from embryos, but also allows patient-specific cells and tissues to be generated for therapeutic use. Optimizing this process will probably lead to a better and more efficient means of generating pluripotent stem cells. Here, we discuss recent findings that show that, in addition to transcription factors, microRNAs can promote pluripotent reprogramming and can even substitute for these pluripotency transcription factors in some cases. Taking into consideration that microRNAs have the potential to be used as small-molecule therapeutics, such findings open new possibilities for both pluripotent stem cell reprogramming and the reprogramming of cells into other cell lineages.

Introduction

MicroRNAs (miRNAs) are small non-coding RNAs with an average length of 22 nucleotides that act by binding to complementary sequences within mRNA molecules, usually in the 3′ untranslated (3′UTR) region. They promote either degradation of the mRNA molecules or inhibit the translation of these transcripts. The interaction between miRNA and mRNAs is mediated by what is known as the ‘seed sequence’, a 6–8-nucleotide region of the miRNA that directs sequence-specific binding to the mRNA through imperfect Watson–Crick base pairing. More than 900 miRNAs are known to be expressed in mammals, and many of these can be grouped into families on the basis of their seed sequence.

Whereas it has become clear that miRNAs have important roles in regulating protein levels inside cells, it has remained unclear whether they can induce dramatic changes in gene expression along with changes in cellular phenotype. However, recent studies have demonstrated that certain miRNAs can promote cellular reprogramming and trans-differentiation in a way that is comparable to reprogramming through expressing lineage-specific transcription factors. Certain miRNA clusters, including miR-290-295 and miR-302-367, can dramatically promote reprogramming into induced pluripotent stem cells (iPSCs) and, in the case of miR-302-367, can even completely substitute for the standard transcription factors OCT4 (also known as POU5F1), SOX2 [for SRY (sex determining region Y)-box 2], KLF4 (for Krüppel-like factor 4) and MYC, that are known to induce reprogramming (Anokye-Danso et al., 2011; Lin et al., 2011; Miyoshi et al., 2011). Such findings offer new approaches for increasing reprogramming efficiency into iPSCs, but little is understood about the molecular mechanisms underlying the promotion of cellular reprogramming by miRNAs. Here, we highlight recent studies that have investigated miRNA-based cellular reprogramming and have shed light on the specific cellular programs, such as proliferation, senescence, and the mesenchymal–epithelial transition (MET), that underlie the ability of certain miRNA families to promote dramatic alterations in cellular phenotype.

Interplay between transcription factors and miRNAs in the maintenance of pluripotency and self-renewal

Embryonic stem cells (ESCs) express a core set of cell-type-specific transcription factors (OCT4, SOX2 and NANOG) and miRNAs (miR-302-367 and miR-290-295). These factors are crucial in determining the identity of pluripotent cells in the sense that changes in their expression levels direct pluripotent cells into a differentiated state, thereby abrogating self-renewal. In order to preserve the identity of pluripotent cells, it is crucial that the expression levels of these factors are carefully balanced. This is facilitated by the autoregulatory capability of the core transcription factors (Fig. 1). Lack of NANOG expression fails to establish ground-state pluripotency (Silva et al., 2008). A reduction in OCT4 expression by one-half in ESCs leads to their differentiation into trophoblasts, whereas overexpression of OCT4 induces primitive endoderm and mesoderm (Niwa et al., 2000).

Fig. 1.

Connection between miRNAs and the core transcription factors regulating pluripotency. Diagrammatic representation of the interconnectivity between miRNAs and genes known to affect pluripotency, including some of the direct targets of miRNAs. The core transcription factors promote the expression of ESC-specific genes and miRNA expression and, at the same time, repress developmental genes. The combined effect of OCT4, SOX2, NANOG and ESC-specific miRNA repression upregulates expression of ESC-specific genes. OCT4, SOX2 and NANOG form an autoregulatory loop that leads to a sustained positive feed-forward pathway. The miRNA let-7 is downregulated by high levels of LIN28 in pluripotent cells.

To avoid alterations in ESC identity, ESC-specific miRNAs that are associated with pluripotency form a tightly intertwined connection with OCT4, SOX2 and NANOG (Fig. 1). The expression levels of miR-302-367 are directly linked to the levels of the three transcription factors (Card et al., 2008; Marson et al., 2008; Barroso-delJesus et al., 2008). There are conserved binding sites for OCT4, SOX2 and NANOG within the miR-302-367 promoter (Barroso-delJesus et al., 2008; Card et al., 2008), and all three of these core pluripotency transcription factors have been demonstrated to bind to the miR-302-367 promoter and activate gene expression, thus linking miR-302-367 expression directly to the central pluripotency network (Fig. 1) (Barroso-delJesus et al., 2008; Barroso-del Jesus et al., 2009; Card et al., 2008; Marson et al., 2008). The promoters of two other miRNAs that are expressed in ESCs, miR-290-295 and miR-106-363, are also co-occupied by OCT4, SOX2 and NANOG (Marson et al., 2008). Shutting down OCT4 expression dramatically reduces the level of miR-290-295 expression (Marson et al., 2008). Conversely, miR-302-367 indirectly induces expression of POU5F1 (which encodes OCT4), SOX2 and NANOG by reducing the expression of developmental genes (Lin et al., 2011; Anokye-Danso et al., 2011). Other ESC-specific miRNAs that have seed sequences that are similar to that of miR-302-367 could thus have a redundant function. Currently, there are no reports that show induction of OCT4, SOX2 and NANOG by other ESC-specific miRNAs. However, taken together, the studies described above suggest that miR-302-367 is part of a positive feed-forward loop that includes OCT4, SOX2 and NANOG in pluripotent cells. Within this autoregulatory network, miR-302-367 inhibits a set of factors that might promote cell differentiation and the loss of ESC identity. Alternatively, miR-302-367 can inhibit repressors of the pluripotent transcription factors (Fig. 1).

In addition to functioning as activators of pluripotency and self-renewal in ESCs, OCT4, SOX2 and NANOG repress developmental genes and the miRNAs associated with lineage commitment. Developmental genes, such as HOXB1, PAX6, ISL1 and NEUROG1, are silenced by co-occupation of their promoters by OCT4, SOX2 and NANOG (Boyer et al., 2005). These three transcription factors preferentially activate pluripotent genes, while repressing developmental genes as a result of additional interactions with the polycomb group of epigenetic regulators (Marson et al., 2008). The polycomb protein, SUZ12 together with OCT4, SOX2 and NANOG represses the expression of miR-708, miR-124, miR-155, miR-615, miR-9 and miR-375 during development by co-occupying their promoters. Decreasing the levels of OCT4 and NANOG during differentiation disrupts the repression of developmental genes and the genes encoding miRNAs and leads to the transcription of lineage-committed genes. This is exemplified by OCT4 and miR-145, which antagonize each other: knockdown of miR-145 expression impairs lineage-committed differentiation as a result of OCT4 and SOX2 elevation (Xu et al., 2009).

The core ESC transcription factors OCT4, SOX2 and NANOG, in addition to the miRNA-binding protein LIN28, can reprogram somatic cells into a pluripotent state (Yu et al., 2007). Moreover, as LIN28 regulates the biogenesis of let-7, its ability to enhance reprogramming suggests that repression of the miRNA let-7 is important in this process. The let-7 precursor is present in very low levels in ESCs (Viswanathan et al., 2008), and Blelloch and colleagues have shown that let-7 inhibits self-renewal in ESCs by opposing mediators of the cell cycle (Melton et al., 2010). It is well known that LIN28 opposes the maturation of Let-7 transcripts (Viswanathan et al., 2008). Although the core pluripotent transcription factors, in collaboration with the polycomb group, keep developmental genes and miRNAs silent, other ESC-enriched factors, such as LIN28, could block expression of undesirable miRNAs (Fig. 1). Thus, the combined repression of lineage-commitment programs, together with promotion of pluripotent programs, sustain pluripotency and self-renewal in ESCs through the interplay between the core transcription factors and miRNAs.

How miRNAs promote reprogramming

The use of miRNAs to promote cellular reprogramming originally stemmed from parallel discoveries describing their roles in regulating pluripotency and the search to replace transcription factors in the reprogramming cocktail with alternative factors. Many small-molecule inhibitors have been found to improve reprogramming efficiency, and these small molecules function by inhibiting specific enzymes or signaling pathways. A screen for small molecules that could replace the oncogene MYC in the OCT4–SOX2–KLF4–MYC cocktail led to the discovery of the histone deacetylase (HDAC) inhibitor valproic acid (VPA) could act as a inducer of pluripotency (Huangfu et al., 2008). Similar screens have subsequently identified small molecules that can replace SOX2 (Shi et al., 2008), KLF4 (Lyssiotis et al., 2009) or both (Zhu et al., 2010). In addition, other small-molecule inhibitors, including inhibitors of mitogen-activated protein kinase (MAPK), glycogen synthase kinase 3 beta (GSK3β), transforming growth factor beta (TGF-β), DNA (cytosine-5-)-methyltransferase (DNMT), and many more, can enhance reprogramming efficiency despite being unable to replace the pluripotency transcription factors (Mikkelsen et al., 2008; Silva et al., 2008; Ichida et al., 2009). Enhancing reprogramming efficiency with small-molecule inhibitors has demonstrated that the negative regulation of genes that are involved in lineage specification and the alteration of the epigenome have a large role in reprogramming. Using RNA interference (RNAi) to negatively regulate the expression of epigenetic factors and tumor suppressors also promotes reprogramming (Mikkelsen et al., 2008; Hanna et al., 2009). Studies have shown that using small interfering RNA (siRNA) against differentiated lineage markers together with the DNMT1 inhibitor 5-azacytidine helps to achieve full reprogramming of partially reprogrammed cells (Mikkelsen et al., 2008). Moreover, the RNAi-mediated knockdown of the tumor suppressors p53 and p21 (encoded by CDKN1A) accelerates the reprogramming process (Hanna et al., 2009). Together, these studies demonstrate the importance of inhibiting the differentiated state in cellular reprogramming, and that this can be achieved by inhibiting specific proteins with chemicals or by decreasing their expression levels through RNAi.

RNAi in the form of miRNAs was first implicated in promoting pluripotency following the observation that distinct miRNAs are differentially expressed in pluripotent and lineage-committed cells (as described above) (Houbaviy et al., 2003). The global loss of miRNA biogenesis in Dicer-null embryos results in a loss of OCT4 expression (Bernstein et al., 2003). In addition, Dicer-null ESCs have impaired differentiation capacity and are not pluripotent, which supports the importance of miRNAs in maintaining the pluripotent state (Kanellopoulou et al., 2005). Mouse ESCs (mESCs) lacking the miRNA biogenesis microprocessor component DGCR8 (for DiGeorge syndrome critical region gene 8) lack microprocessor-dependent miRNAs and display an altered cell cycle and a decreased level of proliferation. This phenotype can be rescued by ectopically transfecting these mESCs with miRNAs that are highly expressed in wild-type mESCs, which suggests that miRNAs have active roles in maintaining the pluripotent state (Wang et al., 2008).

Intriguingly, these ESC-associated miRNAs oppose the let-7 family of miRNAs (Melton et al., 2010). Transfecting miRNA-deficient Dgcr8-null mESCs with miR-294 rescues the loss of pluripotency markers that is caused by transfecting the same cells with let-7. Let-7 is highly expressed in mouse embryonic fibroblasts (MEFs) suggesting that it is associated with a more-differentiated cell phenotype (Marson et al., 2008), whereas miR-294 enhances reprogramming of these cells (Melton et al., 2010). In addition, miR-294 induces the expression of LIN28 in Dgcr8-null mESCs (Melton et al., 2010), which promotes pluripotency by inhibiting let-7 (Viswanathan et al., 2008; Yu et al., 2007). Mitigating the effects of let-7 could partially explain how miR-294 and related miRNAs promote reprogramming.

miR-291, -294 and -295 and miR-302 share a similar seed sequence and enhance reprogramming efficiency in a seed-sequence-dependent manner (Judson et al., 2009). Additional reports have found that other miRNAs containing this seed sequence also increase reprogramming efficiency (Table 1; Li et al., 2011; Subramanyam et al., 2011). Together, these observations indicate that the common targets of these miRNAs are possible ‘roadblocks’ to reprogramming.

Table 1.

ESC-specific miRNAs and their targets

miRNA	Validated targets	References
miR-302 and -367	AOF1, AOF2, AKT1, ANK2, CDKN1A, CHD9, LATS2, LEFTY1, LEFTY2, MBD2, MBP, MECP1-p66, MECP2, MMD, PTEN, RAB11FIP5, RBL2, RHOC, SMAD2, SMARCC2, STAT3, TGFBR2	(Wang et al., 2008; Rosa et al., 2009; Lin et al., 2011; Subramanyam et al., 2011; Lipchina et al., 2011)
miR-290 and -295	CDKN1A, DKK1, FBXL5, RBL2, LATS2, TGFBR2, RASA2, RHOC	(Zovoilis et al., 2009; Lichner et al., 2011)
miR-200c	BMI1, ETS1, SUZ12, ZEB1, ZEB2	(Park et al., 2008; Iliopoulos et al., 2010; Chan et al., 2011)

The important role of specific miRNAs in reprogramming suggests that combinations of miRNAs might act synergistically to promote reprogramming. Liao et al. have speculated that if an individual miRNA is important for reprogramming, its entire cluster of co-expressed miRNAs might enhance this process (Liao et al., 2011). This group found that both the miR-302-367 and the miR-106a-363 clusters substantially enhance induced pluripotent stem cell (iPSC) reprogramming, because they share two seed sequences (Fig. 2). Furthermore, both seed sequences are required for this enhancement, and this finding has been supported by findings made in our own group. We have shown that the miR-302-367 cluster is sufficient to reprogram human foreskin fibroblasts into iPSCs, without any additional factors (Anokye-Danso et al., 2011). miR-302-367 also reprograms MEFs into iPSCs at high efficiency in the presence of the HDAC inhibitor VPA. Together, these studies indicate that clusters of miRNAs probably promote pluripotency through the coordinated expression of multiple members within a multi-cistronic cluster. The ability of naturally occurring miRNA clusters to reprogram cells to iPSCs implies that engineered clusters of reprogramming miRNAs could also be created.

Fig. 2.

Comparison of ESC-expressed miRNAs and their seed sequences. miRNAs that are expressed in ESCs can be grouped according to similarity within the seed sequences. Shared seed sequences are indicated in color. Underlined seed sequences are not shared. miRNA sequences were obtained from the Targetscan and miRBase databases.

The exact combination of miRNAs that might prove optimal for iPSC reprogramming is still a work in progress. Miyoshi et al. screened miRNA mimics for a combination that would reprogram somatic cells to an iPSC state when transiently transfected into cells (Miyoshi et al., 2011). The combination of miR-200c, miR-302s and miR-369 has been demonstrated to reprogram both human and mouse adipose stromal cells to iPSCs (Miyoshi et al., 2011). However, this combination does not efficiently reprogram fibroblasts, suggesting that optimal combinations of miRNAs and small molecules might yet have to be discovered or that these need to be individually determined for different cell types.

Whereas the above studies show that combinations of miRNAs promote reprogramming more efficiently than single miRNAs, some reports have demonstrated that miR-302 alone can reprogram a somatic cell to pluripotency. Expressing miR-302 in human hair follicle cells at levels that are 1.3-fold higher than the expression level in human ESCs (hESCs) can reprogram these cells to iPSCs (Lin et al., 2011). These findings could reflect that hair-follicle-derived keratinocytes are particularly amenable to reprogramming (Aasen et al., 2008), possibly owing to endogenous expression of the reprogramming factors KLF4 and MYC (Maherali et al., 2008) or of miRNAs that might promote reprogramming. Alternatively, miR-302 might be able to reprogram any cell type when it is expressed at a certain level. Additional expression of other miRNAs, such as miR-367, miR-200c and/or miR-369, might increase the efficiency and spectrum of cell types that can be reprogrammed by miR-302.

These studies have shown that certain miRNAs, alone or in combination with each other, improve reprogramming efficiency, and can also reprogram cells without a requirement for exogenously expressed transcription factors. The exact combination of miRNAs varies among different reports, and the different combinations could have varying reprogramming efficiencies or the combination required could be cell-type specific. Different cell types, which have specific transcriptomes, probably need to have different genes repressed by miRNAs to allow initiation of reprogramming.

How similar these miRNA-reprogrammed iPSCs are to transcription-factor-reprogrammed iPSCs or ESCs has not been extensively studied. In the same way that there is controversy regarding the similarity between ESCs and iPSCs reprogrammed with transcription factors (Marchetto et al., 2009; Chin et al., 2010; Ghosh et al., 2010; Guenther et al., 2010; Gutierrez-Aranda et al., 2010; Newman and Cooper, 2010; Bock et al., 2011; Munoz et al., 2011), there might be unknown differences between ESCs and iPSCs reprogrammed with miRNAs. If there are substantial differences between iPSCs that have been reprogrammed using miRNAs and other pluripotent cells, perhaps these differences could be mitigated by optimizing the combinations of miRNAs used in reprogramming or combining miRNAs with transcription factors.

Potential reprogramming mechanisms that are induced by miRNAs

Studies of transcription factor-mediated reprogramming have indicated that a number of steps are involved in the transformation of a somatic cell into a pluripotent cell. As discussed above, several miRNAs have been demonstrated to promote reprogramming. The targets of these miRNAs are involved in many processes that are required for iPSC reprogramming or pluripotent cell maintenance, including cell proliferation, the inhibition of chromatin remodeling factors, MET, inhibiting senescence and apoptosis, and, potentially, the regulation of alternative splicing factors (Fig. 3 and Table 1). Here, we discuss the current evidence for the role of miRNAs in these processes and how this contributes to the generation of pluripotent cells. We draw on many examples from ESCs, because the role of miRNAs in the pathways highlighted in these studies leads us to hypothesize that these could be key mechanisms in miRNA-mediated reprogramming of cells into iPSCs.

Fig. 3.

Potential miR-302-367 targets that affect cellular reprogramming. miR302–367, and other related miRNAs, target multiple cellular processes as shown. The combined repression of these targets affects a global change in cell proliferation, epigenetic state, MET and suppression of developmental factors, which leads to reprogramming of the cell phenotype. It is likely that the combined action of most, if not all, of these processes is required for efficient cellular reprogramming. This diagram shows how the miR-302-367 cluster coordinates multiple cellular processes that are important for reprogramming of somatic cells into pluripotent stem cells as well as maintaining the pluripotent stem cell phenotype. Some of the targets known to be affected in each process are shown.

Proliferation

Pluripotent stem cells have a rapid cell cycle and lack the G1-S barrier that is found in somatic cells. mESCs have been shown to have constitutively high cyclin-A–cyclin-dependent-kinase (CDK) 2 and cyclin-E–CDK2 complex activity, which induces S phase and DNA replication. Furthermore, they lack detectable expression of the CDK inhibitor p21 (Stead et al., 2002). hESCs and human iPSCs also have a shortened G1 phase relative to human fibroblasts (Becker et al., 2006). Therefore, during reprogramming, somatic cells lose their characteristic G1-S barrier and adopt the pluripotent cell cycle phenotype. Cell proliferation has been shown to be rate limiting in reprogramming, and decreased expression of p53 or p21 following treatment with specific siRNAs speeds up reprogramming by increasing the cell division rate (Hanna et al., 2009).

The G1-S barrier in ESCs is regulated by miR-302 and miR-291, -294 and -295 (which contains the same seed sequence as miR-302). mESCs that lack DGCR8, and thus are deficient in miRNAs, have an extended G1 phase in comparison with that of wild-type mESCs, and ectopically expressing the ESC-associated miRNAs miR-302d and miR-291, -294 and -295 restores the percentage of cells in G1 to the level of normal mESCs (Table 1; Wang et al., 2008). Inhibition of miR-302 causes hESCs to accumulate in G1, whereas expression of miR-302 in human fibroblasts increases the percentage of cells in S phase (Card et al., 2008). miR-302 has been shown to increase the speed of reprogramming by targeting p21 and the retinoblastoma family member retinoblastoma-like 2 (RBL2), which promotes cell cycle quiescence by forming a repressive complex inhibiting transcription of cell cycle genes (Litovchick et al., 2007). These targets were validated by reintroducing the miRNAs associated with ESC cell cycle progression, miR-291, -294 and -295, and miR-302d, into miRNA-deficient mESCs, thereby reducing the expression of these two proteins to the level that is normally observed in wild-type mESCs (Wang et al., 2008). Further evidence for the role of miR-291, -294 and -295, and miR-302 has been obtained from luciferase assays where these miRNAs were used to target the 3′ UTRs of CDKN1A and RBL2 mRNAs attached to the luciferase transcript (Wang et al., 2008; Tian et al., 2011). In vivo experiments in the developing lung have shown that overexpressing the miR-302-367 cluster results in decreased transcription and expression of CDKN1A and RBL2 (Tian et al., 2011). Conversely, cell proliferation in the developing lung can be reduced following the expression of a miRNA sponge transcript, which contains binding sites for miR-302-367 and therefore sequesters and depletes the mature miRNAs (Tian et al., 2011). These studies have shown that the ESC-associated miRNA clusters miR-302-367 and miR-290-295 promote cellular proliferation, which speeds up reprogramming (Hanna et al., 2009). Increased proliferation could facilitate reprogramming by helping to reset the epigenetic landscape through synthesis of naïve DNA, thereby removing epigenetic marks that repress the pluripotency network in differentiated cells. This could allow pluripotency-associated genes to become de-repressed and, in turn, promote pluripotent reprogramming.

Chromatin remodeling

Reprogramming causes widespread epigenetic changes that correspond to changes in gene expression (Koche et al., 2011). Unlike somatic cells, the chromatin in pluripotent cells is not condensed and there is a highly dynamic exchange of chromatin-associated proteins (Meshorer et al., 2006). The promiscuous basal transcription that accompanies this dynamic chromatin state (Efroni et al., 2008) might reflect the transcriptional plasticity of pluripotent cells and their ability to acquire diverse differentiated fates, suggesting that the chromatin state of pluripotent cells is a key component of pluripotency.

Small-molecule inhibitors of chromatin-modifying HDACs or DNMTs increase reprogramming efficiency (Huangfu et al., 2008; Mikkelsen et al., 2008), presumably by de-repressing genes that are silenced during development. De-repressing endogenous pluripotency genes might be accomplished by miRNAs alone, although the epigenetic state of a cell might dictate which miRNAs are sufficient for this process and whether further inhibition of chromatin-modifying complexes by small molecules is helpful.

We have recently shown that repression of HDAC2 expression or activity is required for miR-302-367-mediated reprogramming of cells into iPSCs (Anokye-Danso et al., 2011). miR-302-367 is unable to fully reprogram normal, untreated MEFs. However, treatment with VPA or deletion of the Hdac2 gene allows MEFs to be reprogrammed at high efficiency. Interestingly, human foreskin fibroblasts, which normally express only low levels of HDAC2, can be reprogrammed without treatment by VPA (Anokye-Danso et al., 2011). These data support the idea that miRNA pathways and epigenetics collaborate in promoting iPSC reprogramming.

miR-302-367 and miR-302, -369 and -200c might also assist in epigenetically reactivating pluripotency genes by targeting chromatin-modifying genes that promote lineage restriction. miR-302 targets the mRNA encoding several chromatin-modifying enzymes, including the lysine-specific histone demethylases AOF1 and AOF2 (also known as KDM1B and KDM1A, respectively), and the methyl-CpG-binding proteins GATAD2B (also known asp66 beta) and MECP2 (Lin et al., 2011).

AOF1 demethylates mono- and di-methylated H3K4, and evidence has shown that AOF1 can also repress transcription independently of the amine oxidase domain that is responsible for its demethylase activity (Yang et al., 2010). Loss of AOF2 in mESCs leads to the loss of DNMT1 and genome-wide DNA hypomethylation, and AOF2-null mouse embryos fail to gastrulate (Wang et al., 2009), which suggests that AOF2 has a role in exit from pluripotency in the early embryo. Downregulating AOF2 in cells to be reprogrammed could promote rapid erasing of genome methylation and also prevent re-differentiation of partially reprogrammed cells.

GATAD2B and MECP2 bind methylated DNA and recruit histone deacetylase complexes that typically repress transcription (Nan et al., 1998). Inhibition of the genes encoding these proteins, therefore, might be able to de-repress genes that are important for iPSC reprogramming. A recent study has shown that miR-302 targets GATAD2B and MECP2 and suppresses their expression (Lin et al., 2011). This suppression correlates with increased expression of the pluripotency transcription factors NANOG, SOX2 and OCT4 in reprogramming human hair follicle cells. Suppression of GATAD2B and MECP2 also correlates with enhanced global genomic demethylation and specific demethylation of the NANOG and POU5F1 promoters in human iPSCs. miR-302 might affect these changes through direct targeting of the GATAD2B and MECP2 genes. Downregulation of genes encoding chromatin-modifying components, including MECP2, MBD2 (for methyl-CpG binding domain protein 2) and SMARCC2 (for SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily c, member 2) is observed when miR-302 or miR-372 is expressed together with the OCT4, SOX2, KLF4 and MYC reprogramming cocktail (Subramanyam et al., 2011). The effect is not as strong in fibroblasts that are not being reprogrammed by these factors, which suggests that the transcriptome of cells undergoing reprogramming is different from that of normal cells. Consequently, the set of targets for miR-302 and miR-372 changes during reprogramming.

These studies suggest that miR-302 promotes iPSC reprogramming, in part, by targeting chromatin remodeling enzymes and modifying the epigenetic landscape of reprogramming cells. Differential expression of chromatin modifying factors might account for some of the variable requirements for additional miRNAs or HDAC inhibitors among different cell types.

MET

Fibroblasts undergo MET during reprogramming (i.e. they adopt the epithelial state of pluripotent cells) (Li et al., 2010). Experimentally knocking down genes that are associated with MET during reprogramming suppresses the ability of cells to form colonies that express alkaline phosphatase, which is an early sign of reprogramming (Samavarchi-Tehrani et al., 2010). By contrast, suppression of the opposite transition, the epithelial–mesenchymal transition (EMT), by inhibiting TGF-β signaling improves reprogramming efficiency (Ichida et al., 2009). The TGF-β signaling pathway promotes EMT, and miR-302 promotes reprogramming in part by decreasing the expression of TGF-β receptor 2, thereby decreasing SMAD2 and/or SMAD3 phosphorylation and activation (Subramanyam et al., 2011). Paradoxically, miR-302 is known to downregulate the TGF-β superfamily ligands left–right determination factors (LEFTY) 1 and 2, which negatively regulate TGF-β signaling (Barroso-delJesus et al., 2011). However, these signaling components might not be expressed in cells undergoing reprogramming during the stage when TGF-β inhibition is crucial, in which case there would be no effect from being targeted by reprogramming miRNAs. Such context-dependent expression patterns might explain why miRNAs are able to target both activators and inhibitors of the same pathway depending on the cell-type-specific context.

Bone morphogenetic protein (BMP) signaling also promotes MET and reprogramming, and the inhibition of BMP signaling suppresses MET and reduces reprogramming efficiency. These effects are mediated, at least in part, by the miR-200 family, whose expression is dependent on BMP signaling (Samavarchi-Tehrani et al., 2010). Furthermore, it has been shown that miR-302 targets inhibitors of BMP signaling in hESCs (Lipchina et al., 2011). Whereas these studies have also suggested that miR-17, -20a and -106a target the same genes, miR-302 has specifically been shown to have a role in the regulation of BMP signaling during reprogramming. Inhibition of miR-302-367 by miRNA antagonists (known as antagomirs) results in the decreased expression of downstream BMP targets, including ID1 (inhibitor of DNA binding 1) and TLX2 (T-cell leukemia homeobox 2). Genes that were identified in this study as being associated with BMP signaling or that are highly regulated by inhibiting miR-302-367 were examined further using siRNA knockdown strategies in hESCs (Lipchina et al., 2011). Three of these genes, DAZAP2, SLAIN1 and TOB2, have been shown to be BMP inhibitors, and their knockdown with siRNA in hESCs inhibits the differentiation of these cells into PAX6-expressing neural lineages (Lipchina et al., 2011). Together, these studies show that reprogramming miRNAs promote MET, which is a crucial step for iPSC reprogramming, by inhibiting the TGF-β pathway and inducing BMP signaling.

Senescence and apoptosis

Primary cells are prone to senescence during reprogramming, because tumor suppressors are activated following an increased number of division cycles (Utikal et al., 2009) and because the reprogramming transcription factors induce p21 expression (Hong et al., 2009), which can lead to growth arrest and senescence (Vogt et al., 1998). miRNAs that have a role in reprogramming might be involved in directly inhibiting senescence and apoptosis as part of the reprogramming mechanism.

Genetic deletion studies show that miR-290-295 inhibits apoptosis. mESCs that contain DNA damage as a result of γ-irradiation or doxorubicin treatment undergo apoptosis, but mESCs lacking miR-290-295 are more sensitive to DNA-damage-induced apoptosis than are wild-type mESCs (Zheng et al., 2011). Caspase 2 and EI24 (for etoposide induced 2.4 mRNA) are apoptosis-promoting targets of miR-290-295, and their suppression with siRNA partially rescues the apoptotic phenotype that is observed in mESCs lacking miR-290-295. A similar phenotype rescue is also observed following the reintroduction of miR-290-3p or miR-295 into these cells (Zheng et al., 2011). Thus, miR-290-295 and related miRNAs might protect cells from stress-induced apoptosis during reprogramming.

miRNAs that have the same seed sequence and predicted targets as miR-302 and miR-291, -294 and -295, such as miR-17-5p and miR-20a, can also prevent apoptosis (Matsubara et al., 2007). Lung cancer cell lines that express high levels of the miR-17-92 cluster are normally resistant to senescence and apoptosis. However, these cells undergo senescence and apoptosis when treated with antisense oligonucleotides targeting miR-17-5p and miR-20a to reduce their expression (Matsubara et al., 2007). This study confirms that miRNAs sharing the same seed sequence as miR-291, -294 and -295, and miR-302 have potent anti-apoptotic effects, suggesting that miR-302 might also have these properties.

Anti-senescence and anti-apoptotic effects could account for some of the improved efficiency of reprogramming by miR-302-367 compared with that observed upon transcription-factor-mediated reprogramming. However, protection from mitotic arrest and apoptosis, along with inhibition of tumor suppressor expression, could potentially make iPSCs prone to DNA damage accumulation.

Alternative splicing

Pluripotent stem cells express many alternatively spliced mRNAs, including alternatively spliced isoforms of pivotal pluripotency genes such as SALL4 (Rao et al., 2010), POU5F1 (Takeda et al., 1992), NANOG (Das et al., 2011), NR2F2 (Rosa and Brivanlou, 2011), and FGF4 (Mayshar et al., 2008). Alternatively spliced isoforms of these and other genes characterize cell fate transitions between pluripotent and various differentiated lineages. A more recent example of the importance of alternative splicing in pluripotency has been reported for the gene encoding the transcription factor forkhead box P1 (FOXP1). Gabut et al. have identified an ESC-specific isoform of FOXP1 that regulates the pluripotent state (Gabut et al., 2011). The ESC-specific FOXP1 isoform allows mESCs to undergo self-renewal in the absence of the cytokine leukaemia inhibitory factor (LIF), which is normally crucial for preventing differentiation (Gabut et al., 2011). Depleting mESCs of Foxp1 by using shRNA results in spontaneous differentiation. Thus, alternative splicing of a single gene, in this case FOXP1, controls the switch between pluripotency and differentiation. Such cell fate changes that appear dependent on alternative splicing suggests that it might be a key mechanism controlling the pluripotent state.

The targets of miR-302 that have been predicted by using bioinformatics approaches include a large number of genes that are associated with alternative splicing as determined by gene ontology analysis (Huang et al., 2009; Dweep et al., 2011). The control of alternative splicing by miRNAs that are enriched in ESCs, such as miR-302, could allow a cell to rapidly switch from expressing differentiation-associated alternative splice variants to pluripotency-associated isoforms without the delay that would be involved in de-repressing epigenetically silenced genes. Additional research in this area is likely to prove fruitful in understanding post-transcriptional mechanisms that promote pluripotency.

miRNAs as a new tool for cellular reprogramming

Despite the advantages that are associated with iPSCs and the flurry in research following the initial discovery of transcription factors that can induce pluripotency, reprogramming technology is still somewhat in its infancy. Viral delivery systems, insertion mutations, the use and reactivation of oncogenic transcription factors, low efficiency and inconsistencies between cell lines from the same parental cells remain hurdles in using iPSCs to study human disease. There have been several important technical advances that have helped to increase efficiency and usability of iPSCs. The use of recombinant OCT4, SOX2, KLF4 and MYC that can penetrate the cell membrane means the genomic modification associated with an integrative viral delivery system can be avoided (Zhou et al., 2009). Although non-integrative vector delivery systems require the use of exogenous DNA, in most cases these result in apparent stable integration of exogenous DNA (Seki et al., 2010; Okita et al., 2008; Stadtfeld et al., 2008; Kaji et al., 2009; Woltjen et al., 2009). Synthetic mature mRNAs for the four reprogramming transcription factors have also been successfully used to reprogram cells to pluripotency (Warren et al., 2010).

Ideally, reprogramming by means of pharmacologically active compounds would eliminate most of these challenges. Thousands of pharmacological compounds have been screened for their effect on reprogramming. This has led to the identification of some compounds that either improve reprogramming efficiency in addition to the overexpression of OCT4, SOX2, KLF4 and MYC or replace the need for exogenous expression of the transcription factors with the exception of OCT4 (Shi et al., 2008; Li et al., 2009; Lin et al., 2009).

Although reprogramming using integrative miRNA expression vectors is highly efficient and does not require the use of oncogenes, some of the other problems encountered with the overexpression of transcription factors are not eliminated (Anokye-Danso et al., 2011). Recent studies that have shown that miRNA mimics can reprogram somatic cells to pluripotency have opened up the possibility of reprogramming somatic cells with the simple transduction of small RNAs (Miyoshi et al., 2011). Synthetic miRNA mimics could also be used to convert cells into iPSCs, but the conversion rate is low (Miyoshi et al., 2011). Low efficiency in reprogramming using mimics can be attributed to their half-life and/or the mode of mimic delivery into cells. Bhatia and colleagues have developed dendrimer-conjugated magnetic nanoworms (referred to as ‘dendriworms’) that deliver siRNA into tissues of living animals (Agrawal et al., 2009). Dendriworms have been found to be 2.5-fold more efficient than cationic-lipid-based siRNA carriers. siRNAs introduced into the cell in this way reduce the expression of their targets at efficiencies that are comparable to those from integrative vector shRNA. Generally, the levels of transfected miRNA mimics decline rapidly within a few days (Judson et al., 2009). Designing synthetic mimics with a half-life that is long enough to rival continuous production of exogenous miRNAs by viral integrants would increase reprogramming efficiency. Moreover, a combination of synthetic miRNA mimics and chemical compounds could be employed to improve, for example, the efficiency with which miR-302, -200 and -369 leads to iPSC formation. A similar approach could potentially also be used to convert cells directly into a number of different adult cell types.

Direct differentiation or transdifferentiation has gained popularity in recent years. Transforming one differentiated cell type into another bypasses the pluripotent stage of reprogramming. Forced expression of transcription factors has been used to transdifferentiate fibroblasts into neurons (Vierbuchen et al., 2010), exocrine cells into pancreatic islet cells (Zhou et al., 2008), and fibroblasts into cardiomyocytes (Ieda et al., 2010). Crabtree and colleagues have reported that the addition of miRNAs to a cocktail of transcription factors enhances transdifferentiation of human fibroblasts into functional neurons (Yoo et al., 2011). The use of transcription factors could even be dispensable if known repressors of neurogenic factors are repressed by miRNA. Screening for compounds that could replace the neurogenic factors would move this technology a step closer to the generation of patient-specific neurons that are free from transgenes and genomic modification and are therefore suitable for use in cell therapy. Moreover, such an approach could be used in vivo to promote neural regeneration from endogenous neural progenitor cells.

A viable approach in the search for safer, easier and more-efficient ways to obtain iPSCs could be a screen for compounds that activate transcription from native miRNA promoters. It is known that OCT4, SOX2 and NANOG co-occupy the promoter region and regulate transcription of miR-302-367 (Barroso-delJesus et al., 2008; Card et al., 2008; Marson et al., 2008). Thus, chemicals that could, for example, replace exogenous OCT4 and SOX2 and activate endogenous miR-302-367 expression sufficiently to induce reprogramming would aid the generation of integration-free iPSCs. Alternatively, systematic screening of miRNAs that could indirectly activate transcription factors would be beneficial. This approach is particularly relevant in terms of achieving direct differentiation of one type of cell to another. Different cells types have been transdifferentiated in vivo into clinically important cells by injection of viruses encoding transcription factors (Zhou et al., 2008; Qian et al., 2012). Use of viruses and transcription factors could be replaced with use of miRNAs. With a better technique of delivery and extended half-life, miRNAs have the potential to be a new tool in tissue engineering and regenerative medicine.

Summary and future directions

The finding that certain miRNAs can promote pluripotent reprogramming provides a new set of tools for both the iPSC field as well as for studying direct reprogramming into differentiated cells lineages. Additional uses of miRNAs to promote tissue regeneration might also prove useful. The nature of miRNAs allows for them to be used in a similar manner to other small-molecule approaches for in vivo therapy. These characteristics include the ability to modify both miRNA mimics and antagomirs with compounds, such as cholesterol, that enhance cellular uptake, the overall ease of generating such small RNA molecules as compared with other small molecule compounds, and the transient and non-integrative aspect of miRNA mimics and antagomirs. The field of miRNA-mediated reprogramming is nascent and much is still unclear about the underlying mechanisms and whether this approach will be widely applicable. Whereas a recent spate of reports have shown important proof of concepts, future studies will be required to determine how applicable such approaches will be to the wide variety of cells that are used in cellular reprogramming.