Brief report: miR-290–295 regulate embryonic stem cell differentiation propensities by repressing pax6


  • Author contributions : H.K., E.C., M.L., E.H., and Y.S.: designed the research; H.K.: performed all the work reported in this manuscript, except for establishment of the miR-290–295 null ESC line (E.C., M.L, and G.B.); H.K., E.H., and Y.S.: wrote the manuscript with comments from E.C. H.K. and E.C. contributed equally to this article.


microRNAs of the miR-290–295 family are selectively expressed at high levels in mouse embryonic stem cells (mESCs) and have established roles in regulating self-renewal. However, the potential influence of these microRNAs on cell fate acquisition during differentiation has been overlooked. Here, we show that miR-290–295 regulate the propensity of mESCs to acquire specific fates. We generated a new miR-290–295-null mESC model, which exhibits increased propensity to generate ectoderm, at the expense of endoderm and mesoderm lineages. We further found that in wild-type cells, miR-290–295 repress Pax6 and ectoderm differentiation; accordingly, Pax6 knockdown partially rescues the mESCs differentiation impairment that is caused by loss of miR-290–295. Thus, in addition to regulating self-renewal, the large reservoir of miR-290–295 in undifferentiated mESCs fine-tunes the expression of master transcriptional factors, such as Pax6, thereby regulating the equilibrium of fate acquisition by mESC descendants. Stem Cells 2013;31:2266–2272


Embryonic stem cells (ESCs) possess the unique capacity to either self-renew or differentiate into a diverse array of cell types. Self-renewal and differentiation are regulated by microRNAs (miRNAs; reviewed in [1]). The most abundant family of miRNAs in mouse ESCs (mESCs) are members of the miR-290–295 cluster [2-4]. These miRNAs are transcribed early in the zygote as a single polycistronic unit, exhibit high sequence similarity, and have a common set of targets [5-7]. These properties are shared by the homologous miRNA clusters in other species, including miR-371–3 and miR-302/430/427, suggesting functional conservation in the vertebrate clade [8].

Expression of the miR-290–295 cluster is controlled by the core transcription factor regulatory network, consisting of Oct4, Sox2, and Nanog, which together maintain mESCs in an undifferentiated state [9]. Consistently, miR-290–295 family members have been shown to promote pluripotency [10, 11] and to enhance the reprogramming of somatic cells to pluripotency [12]. Additionally, miR-290–295 regulate the rapid cell cycle of mESCs [13], apoptosis [14], de novo DNA methylation [15, 16], and epithelial to mesechymal transition [17].

Differentiation of mESCs is regulated by key transcription factors, such as Brachyury (T) and Foxa2, which drive mesendoderm specification [18] and Sox1, Nr2f2, and NeuroD1, which promote ectoderm/neuroectoderm differentiation [19-21]. Recent work further identified the paired-box transcription factor, Pax6, as a regulator of neuroectederem specification in humans [22]. Comparable roles for Pax6 homolog have been reported in mouse [19, 23], but these are still debated [22]. Initial studies of miR-290–295 implicated these miRNAs in suppression of mESCs differentiation in general [10] and into mesoderm in particular [24]. Intriguingly, the human homolog, miR-302, was shown to control mesoderm differentiation [8, 25].

Here, we use a new knockout model to show that the miR-290–295 cluster promotes mesendoderm differentiation. We identify Pax6 as a direct target of miR-290–295 that is repressed by miR-290–295 to promote mesendoderm differentiation of mESC. Thus, miR-290–295 activity directs differentiation propensities upon departure from pluripotency, thereby regulating the equilibrium of fate-acquisition by mESCs descendants.

Materials and Methods

Full methods are available as Supporting Information in the online version of the manuscript.

Generation of miR-290–295null Line

To generate miR-290–295 loss of function allele, a gene-trapped mESC clone from the German gene trap consortium (129S2 strain) was microinjected into C57BL/6J host embryonic day (E) 3.5 blastocysts, which gave rise to F1 mice. Heterozygous F2 offspring were intercrossed to produce E3.5 blastocysts, yielding homozygous miR-290–295−/− (miR-290–295null) and miR-290–295+/+ (wild type [WT]) new mESC littermate lines. The new mESC lines were checked for normal karyotype (Supporting Information Fig. S1).

Stem Cell and Embryoid Bodies Culture

Undifferentiated mESCs were cultured on irradiated mouse embryonic fibroblast (MEF) feeders in standard medium. Embryoid bodies (EBs) were generated in nonadherent bacterial Petri dish containing Leukemia inhibitory factor (LIF)-free medium. When mentioned, 0.05 µM of retinoic acid (RA) was added to the medium after 4 days of differentiation, for additional 4 days. For Pax6 knockdown, miR-290–295null mESCs were transfected with either 10 µM of Pax6-targeting small interefering RNAs (siRNAs) or nontargeting control siRNA, prior to EB aggregation.


Loss of miR-290–295 Expression Does Not Compromise the Expression of Pluripotency Markers

To analyze the involvement of miR-290–295 in the differentiation of mESCs, we established by insertional mutagenesis a new miR-290–295null line, in which the expression of miR-290–295 cluster is nullified (Fig. 1A, 1B). miR-290–295null mESCs, that were cultured on MEF feeders under self-renewing conditions exhibited similar morphology to that of WT mESCs, which were derived from a littermate embryo (Fig. 1C). The expression of the pluripotency markers Oct4, Sox2, and Nanog, and the endogenous activity of alkaline phosphatase, were comparable between miR-290–295null and WT mESCs (Figs. 1D–1F, 2C). These results are consistent with previous reports, suggesting that neither miR-290–295 nor Dgcr8 or Dicer1 are required for maintenance of mESC pluripotency [26, 27].

Figure 1.

Nullification of the mmu-miR-290–295 cluster does not compromise pluripotentcy markers in mouse embryonic stem cells (mESCs). (A): A β GAL-Neor fragment (β-Geo—fusion of Lacz and Neor coding sequences) was inserted approximately 200 bp upstream of the miR-290–295 cluster. Pre-miRNAs are depicted by black vertical lines. (B): Expression of miR-290–295 cluster members in undifferentiated mESCs derived from WT (+/+), miR-290–295 heterozygous (+/−), and homozygous miR-290–295null (−/−) embryos. Results are displayed relative to expression in WT cells, normalized to U6 expression, and represented as mean ± SEM, n = 3.*, p <.05; **, p <.001; ***, p <.0005, one-sided t-test. (C): Colonies of WT and miR-290–295null mESCs grown on mouse embryonic fibroblast (MEF) feeder cells in the presence of LIF exhibited similar morphology. Insets depict magnification of the marked area. Scale bar = 200 µm. (D): Quantitative polymerase chain reaction (qPCR) measurements of the pluripotency markers Oct4, Nanog, and Sox2 in WT and miR-290–295null mESCs. Results are displayed relative to expression in WT cells, normalized to β-actin expression, and represents mean ± SEM, n = 3. (E): Western blot analysis of the pluripotency markers OCT4, NANOG, and SOX2 in WT and miR-290–295null mESCs. GAPDH was used as a loading control. (F): Alkaline phosphatase staining of undifferentiated WT and miR-290–295null mESCs. Scale bar = 200 µm. Abbreviations: pA, polyA sequence; SA, splice acceptor site; WT, wild type.

Figure 2.

Undifferentiated miR-290–295null mESCs display elevated expression of ectoderm genes. (A): qPCR measurements of the ectoderm (Sox1, Pax6, and Nestin) and mesendoderm (Sox17, T, and Gsc) genes in undifferentiated WT and miR-290–295null mESCs. Results are displayed relative to expression in WT cells, normalized to β-actin expression, and represents mean ± SEM, n = 3. *, p < .05, two-sided t-test. (B): Western blot of PAX6 protein levels in undifferentiated WT and miR-290–295null mESCs. GAPDH was used as loading control. (C): Immunocytofluorescence of the pluripotent marker OCT4 (green) and the ectoderm marker PAX6 (red) in undifferentiated WT and miR-290–295null mESCs. A merge of OCT4 and PAX6 staining is shown to the right. DAPI nuclei labeling (blue) is shown in the left panel. Insets depict magnification of the defined area, scale bar = 100 µm. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; WT, wild type.

Undifferentiated miR-290–295null mESCs Exhibit Increased Ectodermal Propensity

Despite the lack of change in expression of the pluripotency markers, the expression of the ectoderm genes, Pax6, Sox1 and Nestin, was considerably upregulated in miR–290–295null cells. Mesoendoderm genes, Brachyury (T) and Sox17, were mildly downregulated in miR–290–295null cells, but this change did not approach statistical significance (Fig. 2A). Western blot analysis of Pax6 protein levels revealed marked upregulation in undifferentiated miR–290–295null cells (Fig. 2B). Immunocytofluorescence study showed that Pax6 is up-regulated also in undifferentiated, Oct4-positive miR–290–295null cells (Fig. 2C). This was further confirmed by measuring Pax6 and Oct4 mRNA in miR-290–295null and WT cells sorted based on high levels of the pluripotent marker, SSEA1 (Supplementary Fig. S2). Thus, loss of miR-290–295 expression predisposes undifferentiated mESCs to express higher levels of ectoderm genes.

Loss of miR-290–295 Promotes Differentiation of mESCs Toward Ectoderm at the Expense of Mesendoderm Lineages

To determine if miR-290–295 are involved in regulating the differentiation propensity of mESCs, we studied EBs in LIF-free suspension cultures. Analysis of lineage markers revealed higher levels of the neuroectoderm genes Sox1, Pax6, and Nestin, including PAX6 protein levels, in miR-290–295null EBs relative to WT EBs (Fig. 3A, 3B). The minor upregulation of these neuroectoderm genes in spontaneously differentiating WT EBs is consistent with previous reports [28-33] and might be explained by non-vanishing levels of miR-290–295 in these EBs (Fig. 3C; Supporting Information Fig. S3). The substantial residual levels of miR-290–295 also suggest progressive involvement of this cluster in the transition from pluripotency toward acquisition of lineage-specific fates.

Figure 3.

Differentiation of miR-290–295null mESCs is biased toward ectoderm at the expense of endoderm. (A): Expression of ectoderm (Sox1, Pax6, and Nestin) and endoderm (Sox17, Hnf4a, and Foxa2) markers was measured by qPCR in spontaneously differentiating WT and miR-290–295null (null) embryoid bodies (EBs) relative to WT mESCs at day 0. Following differentiation, ectoderm markers were upregulated in miR-290–295null versus WT EBs, while endoderm markers were downregulated. Data are normalized to β-actin expression and represents mean ± SEM, n = 3. (B): Western blot (top) and densitometry (bottom) of PAX6 protein in spontaneously differentiating WT and miR-290–295null EBs (null). GAPDH was used as a loading control. Densitometry was normalized to GAPDH and displayed relative to WT mESCs at day 0. (C): qPCR measurements of miR-291a-3p in spontaneously differentiating WT EBs relative to expression at day 0. Data are normalized to U6 expression and represents mean ± SEM, n = 2. (D): Same as (A) for the PrE marker Gata6 and the yolk sac marker, Ttr. (E): Representative images of WT and miR-290–295null EBs after 15 days of differentiation in LIF-free medium. WT EBs formed higher numbers of yolk sac-like cysts. Insets depict magnification of the marked area. Scale bar = 1 mm. (F): Relative changes in mRNA levels of ectoderm (Pax6, Nestin) and endoderm (Sox17) genes following treatment with retinoic acid. Shown are ratios between the changes measured for miR-290–295null and WT EBs (null/WT). Data represent mean ± SEM, n = 4. *, p < .05, Wilcoxon rank-sum test. Abbreviation: WT, wild type; PrE, primitive endoderm.

The higher expression of ectoderm genes in miR-290–295null EBs coincided with downregulation of the endoderm regulators Sox17, Foxa2, and Hnf4a (Fig. 3A), and of the transiently expressed early primitive streak genes Brachyury (T), Gsc, and Fgf8 (Supporting Information Fig. S4). The levels of Gata6 were additionally lower in miR-290–295null cells (Fig. 3D), suggesting that differentiation toward primitive endoderm (PrE) was also compromised. Accordingly, there was a significant reduction in the expression of yolk sac marker Transthyretin, Ttr (Fig. 3D) [34], and in the numbers of the PrE-derived yolk sac-like cysts [35, 36] in miR-290–295null EBs, as compared to WT EBs (Fig. 3E). Administration of RA, a known inducer of ectoderm differentiation, increased the expression of Pax6 and Nestin in both WT and miR-290–295null EBs. Noteworthy, the RA-treated miR-290–295null EBs upregulated ectodermal genes and concomitantly downregulated the endoderm regulator Sox17 compared to WT EBs (Fig. 3F; Supporting Information Fig. S5).

We conclude that the intact function of miR-290–295 during differentiation of WT ESCs promotes endoderm and mesoderm fates at the expense of ectoderm lineages. When this intrinsic genetic propensity is experimentally compromised, ESC descendants are biased toward ectoderm. Notably, the levels of the Nodal pathway inhibitors, Lefty1 and Lefty2 were upregulated in undifferentiated but not in differentiating miR-290–295null mESCs (Supporting Information Fig. S6), suggesting that the mesendoderm promoting activity of miR-290–295 might be independent of the Nodal pathway [8, 25].

Pax6 Is Directly Regulated by miR-290–295 Family Members

Since Pax6 has been reported to play a functional role in later stages of neuroectoderm differentiation [19, 22], we hypothesized that the miR-290–295 cluster promotes endoderm and mesoderm lineages by repressing Pax6. Bioinformatic analysis of predicted miR-290–295 targets identified Pax6 as a leading candidate for mediating the repressive effect of this cluster on ectoderm differentiation (Supporting Information Fig. S7). To test if Pax6 is indeed a direct target of miR-290–295, we transfected a dual-luciferase vector harboring the 3′-UTR of murine Pax6 [37] into miR-290–295null and WT mESCs. As expected, the luciferase activity was significantly higher in miR-290–295null cells compared to WT cells (Fig. 4A). To verify that this increase depends in cis on the presence of intact miR-290–295 binding site, we mutated the miR-291a–3p/miR-294/miR-295 “seed-match” site of Pax6 (Fig. 4B). Disruption of the seed binding de-repressed Pax6 reporter activity in WT ESCs, which express normal levels of miR-290–295 (Fig. 4A). These results indicate that Pax6 is a bona fide target of endogenous miR-290–295 cluster in mESCs.

Figure 4.

miR-290–295 promote mesendoderm differentiation by repressing Pax6. (A): Direct repression of Pax6 3′-UTR by miR-290–295 validated by dual luciferase activity assay. WT and miR-290–295null mESCs were harvested 30 hours after transfection with a vector harboring luciferase-coding sequence upstream of a fragment of the Pax6 3′-UTR or 3′-UTR harboring mutation in the predicted miR-291a-3p/294/295 target site. Shown are changes in luciferase activity relative to miR-290–295null cells with WT Pax6 3′-UTR (left). Data represents mean ± SEM, n = 3. *, p < .05, two-sided t-test. (B): The predicted target site (seed match) for miR-291a-3p, miR-294, and miR-295 in Pax6 3′-UTR. The shared seed sequence is marked by the rectangle. Notably, the binding sequence for miR-294-3p or miR-291a-3p is 9 nt long. The mutated sequence is shown in gray and is underlined. (C): Relative mRNA expression of pluripotency (Oct4 and Sox2), ectoderm (Pax6 and Sox1), and early primitive streak markers (Brachyury [T], Goosecoid [Gsc], and Fgf8). Shown are the levels in transfected miR-290–295null embryoid bodies (EBs), 3 days after initiation of differentiation relative to WT mESCs at day 0. Data are normalized to β-actin expression and represents mean ± SEM, n = 3. *, p < .05; **, p < .01, one-sided t test between siPax6 and siControl results. (D): Same as (C) for the endoderm and primitive endoderm lineage markers, Hnf4a and Ttr, 7 days after initiation of miR-290–295null EB differentiation. (E): A model for the regulation of germ layer specification by miR-290–295 and Pax6 in WT and miR-290–295 null (KO) mESCs. Abbreviations: NE, neuroectoderm; PrE, primitive endoderm; WT, wild type.

In order to characterize the genetic interactions of Pax6 and miR-290–295 in differentiation, we knocked down Pax6 by siRNAs in miR-290–295null mESCs. siRNAs against Pax6 (siPax6) led to substantial decrease in Pax6 mRNA levels (Supporting Information Fig. S8) and resulted in downregulation of Sox1 and upregulation of early mesendoderm markers Brachyury (T), Fgf8, and Gsc (Fig. 4C). Furthermore, the endoderm gene Hnf4a and the yolk sac marker Ttr were also upregulated in siPax6-transfected cells (Fig. 4D). These results suggest a model for miR-290–295 activity in facilitating early mesoendoderm lineage specification by repressing proectodermal Pax6 (Fig. 4E).


In this work, we established a genetic model for loss of miR-290–295 activity and showed that this cluster promotes mesendoderm fates via a previously undescribed repression of Pax6. Loss of miR-290–295 activity de-represses Pax6 in undifferentiated and in differentiating mESCs, and drives ectoderm specification at the expense of endoderm and mesoderm lineages. Concomitant knockdown of Pax6 partially recovers the normal balance between the three germ layers, suggesting that Pax6 is a principal target of miR-290–295 in this process. Noteworthy, miR-290–295 is acting to suppress ectoderm specification and to fine-tune Pax6 expression, already at undifferentiated, pluripotent mESCs and its loss results in imbalanced lineage specification during differentiation. These findings portray a molecular mechanism for post-transcriptional tuning of ectodermal Pax6 expression. This adds to later stage regulation of Pax6 by miR-7 during brain and pancreas development [37, 38]. The positive influence of miR-290–295 on endoderm and mesoderm specification may potentially explain the reported impairments in yolk sac and somite development in a related model of miR-290–295-deficient embryos [39] and assists in resolving inconsistencies related to the usage of transient knockdown or overexpression [10, 24].

Undifferentiated ESCs are characterized by a unique signature of globally hyperactive transcription, resulting in low level expression of numerous tissue-specific genes [40, 41]. This “transcriptional priming” is a molecular sign which may be functionally related to the ability to acquire different cell fates upon encountering suitable signals. Such a strategy for maintaining broad low-level transcription of multiple lineage-specific genes may require a stringent post-transcriptional control miRNAs are known for their ability to fine tune gene expression at a post-transcriptional level and therefore may be plausible mediators of this process. Our study demonstrates that miR-290–295 indeed provide a post-transcriptional regulatory mechanism for suppression of the ectoderm fate, while favoring mesoderm and endoderm lineages. These findings therefore support a role for miRNAs in establishing a correct balance between germ layers under conditions of multilineage priming.


We thank Rebecca Haffner-Krausz and Golda Damari for mouse micromanipulation; Dr. Yael Fried and Dr. Elena Ainbinder (Stem Cell Core Facility Unit, Weizmann Institute of Science, Israel) for support in the ESC work; Inbal Zaltsman for assistance with quantitative polymerase chain reaction experiments; Dr. Jacob Hanna for critical reading of the manuscript. This work was supported by grants from the Israel Science Foundation (grant no. 730/12 to E.H. and 1541/09 to Y.S.) and Helen and Martin Kimmel Institute for Stem Cell Research. We also acknowledge support of the Leona M. and Harry B. Helmsley Charitable Trust (to Y.S.) and from Carolito Stiftung, Y. Leon Benoziyo Institute for Molecular Medicine, Charlene Vener New Scientist Fund, Estates of Fannie Sherr, Lola Asseof, Nathan Baltor, Maria Halphen, Julius and Ray Charlestein Foundation and the Fraida Foundation (to E.H.). E.H. is the incumbent of the Dr. Sydney Brenner Chair and the Helen and Milton A. Kimmelman Career Development Chair. Y.S. is the incumbent of Daniel E. Koshland Sr. Career Development Chair at the Weizmann Institute.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.