|Home | About | Journals | Submit | Contact Us | Français|
Ectopic expression of the transcription factors Oct4, Sox2, cMyc, and Klf4, as well as variants of this factor combination, are sufficient to confer a pluripotent state upon several differentiated cell types, generating induced pluripotent stem cells (iPSCs) [1-8]. The derivation of iPSCs is a highly inefficient process with the underlying mechanisms largely unknown. This low efficiency argues for the existence of additional cooperative factors, whose identification is critical for understanding the process of reprogramming. Further, the therapeutic use of iPSCs relies on developing efficient non-genetic means of factor delivery, and while a handful of compounds that replace individual factors have been identified, their use yields a further reduction to the already low efficiency of reprogramming [9-11]. Thus, the identification of compounds that enhance rather than solely replace the function of the reprogramming factors will be of great use. Here, we demonstrate that inhibition of the Tgfβ signaling pathway acts as a cooperative factor in the reprogramming of murine fibroblasts, enabling more efficient and faster induction of iPSCs in a dose-dependent manner, while activation of Tgfβ signaling blocks reprogramming. In addition to a strong cooperative effect, use of a Tgfβ receptor inhibitor bypasses the requirement of exogenous cMyc or Sox2, highlighting its dual role as both a cooperative and a replacement factor. The identification of a highly characterized pathway that operates in reprogramming will open up new avenues for mechanistic dissection of the reprogramming process, as well as facilitate the derivation of iPSCs using small molecules.
We identified Tgfβ signaling as a potential cooperative pathway in reprogramming during functional validation of a constructed network that linked retroviral insertion sites across several mouse iPSC lines . Although the biological validation of network targets was largely negative, in agreement with statistical analyses suggesting that the network did not differ from randomly constructed networks, we found that use of a Tgfβ receptor I kinase/activin-like kinase 5 (Alk5) inhibitor could enhance the efficiency of iPSC derivation.
To induce reprogramming, we infected MEFs carrying a reverse tetracycline transactivator (rtTA) transgene with doxycycline-inducible lentiviruses encoding the four reprogramming factors as previously described . Administration of the Alk5 inhibitor during the course of doxycycline (dox) induction elicited a striking increase in the number of iPSC colonies (Figure 1A). These iPSC clones could be expanded in the absence of dox, which is a strong indicator of successful reprogramming [13, 14].
This robust increase in efficiency led us to examine whether the Alk5 inhibitor could also reduce the temporal requirement of factor expression, thus reflecting an increase in the kinetics of reprogramming. To test this, we applied dox for three, four, or five days on four-factor infected MEFs, either in the presence or absence of the Alk5 inhibitor (Figure 1B). While we were unable to obtain iPSCs with three days of dox treatment in either condition, four days of dox treatment was sufficient to give rise to iPSC colonies at a frequency of 0.0013% in the inhibitor-treated condition. These colonies were not immediately apparent and took at least one week after dox withdrawal to emerge. No colonies were observed in the control condition, demonstrating that the inhibitor promoted faster induction of iPSCs. 4-day iPSCs expressed the pluripotency markers Nanog, Oct4, and Sox2 (Figure 1C), which was tested three passages after dox withdrawal to ensure a lack of residual transgene expression . These cells were also competent to form lineages from all three germ layers in the context of a teratoma (Figure 1C), providing a functional test of pluripotency and indicating that use of the Alk5 inhibitor during reprogramming has no adverse effect on the resulting iPSCs.
We next sought to establish a dose-response curve for the Alk5 inhibitor. Using four-factor infected MEFs, we tested concentrations of the Alk5 inhibitor over a 1000-fold range, from 4nM to 4μM. Increasing concentrations of inhibitor led to a progressive increase in the number of iPSC colonies obtained, resulting in a 30-fold enhancement at the highest dose tested (Figure 1D).
To verify that manipulation of Tgfβ signaling affects reprogramming, we investigated whether pathway activation would inhibit iPSC formation. Using four-factor infected MEFs, we applied increasing concentrations of Tgfβ1 or Tgfβ2 during the time of dox induction. Under control conditions, we obtained an average of 14 colonies per well (~0.017%); however, upon addition of 1ng/mL Tgfβ1 or Tgfβ2, the efficiency was reduced to one colony per well (~0.0013%), and at 2.5ng/mL and 5ng/mL, iPSC formation was undetectable (Figure 1E). The opposing effects of pathway activation and inhibition confirm the involvement of Tgfβ signaling in reprogramming.
To gain further insight into the mode of action of the Alk5 inhibitor, we next examined the temporal window during which the inhibitor elicited its strongest effect. In four-factor infected MEFs, the inhibitor was applied in 4-day intervals from days 0-16 of reprogramming. The largest increase in efficiency was observed during the first four days of reprogramming, yielding an increase comparable to full-time application of the inhibitor (Figure 1F). A strong enhancement was also observed with application during the 5-8 day interval, while after dox withdrawal (day 12), the inhibitor appeared to have no effect.
This early-acting effect prompted us to investigate whether the Alk5 inhibitor was acting to ‘prime’ the cells, thus altering or destabilizing the fibroblast state so that it is more amenable to reprogramming. To test this, we applied the inhibitor in 3-day time intervals: before dox administration (days -3 to 0) to test for a priming effect, during the first 3 days of reprogramming (days 0 to +3) as a control for the efficiency increase, during both intervals (days -3 to +3) to test for a synergistic effect, and a no-inhibitor control as a baseline measure (Figure 1G). Administration of the Alk5 inhibitor prior to expression of the reprogramming factors (days -3 to 0) led to a slight reduction in efficiency, thus arguing against a priming effect. In contrast, addition of the inhibitor during the first 3 days of factor expression led to ~5-fold enhancement in efficiency, consistent with our previous findings. The increase in efficiency seen in the -3d/+3d group was comparable to the 0 to +3d group, indicating a lack of a synergistic effect. It therefore appears that Alk5 inhibition must coincide with expression of the reprogramming factors in order to elicit its effect.
One limitation of direct infection is that not every cell receives the full complement of reprogramming factors, and there is a large amount of heterogeneity in expression, as each cell harbors a unique pattern of viral insertions. It is therefore difficult to ascertain whether the increase in efficiency we have observed so far is due solely to a cooperative effect, where the inhibitor enhances efficiency in the presence of all four factors, or a replacement effect, where the inhibitor substitutes for specific reprogramming factors. To this end, we employed a “secondary system” consisting of MEFs isolated from iPSC-derived chimeras. These MEFs contain a homogenous set of inducible lentiviral integrations that were permissive for conversion into primary iPSCs; thus, the levels and stoichiometry of the re-expressed factors are unique to each MEF line [14, 15].
In two different secondary MEF lines, we tested the effect of Alk5 inhibition on both the efficiency and kinetics of reprogramming. In this experiment, we applied dox for 4, 6, 8, or 10 days, either in the presence or absence of the Alk5 inhibitor, then quantified iPSC colony number on day 16. The baseline levels of reprogramming differed between the two lines (Figure 2A), which was not unexpected given that efficiency is dependent upon the levels and stoichiometry of factor expression as previously reported . In both lines, addition of the Alk5 inhibitor mediated an increase in both efficiency and kinetics of reprogramming, consistent with a cooperative effect (Figure 2A). However, the degree to which reprogramming was enhanced was different between the two lines, with one line showing a 2-fold increase and the other a 30-fold increase. We reasoned that a purely cooperative action would produce a consistent fold-increase between the different lines regardless of variation in factor expression levels, which was not the case.
Thus, to further discern between a cooperative and a replacement effect, we employed a polycistronic construct linking all four reprogramming factors on a single transcript (“STEMCCA”) , which in the context of a secondary system enables expression of all factors in >95% of cells (data not shown). In such a system, only a cooperative effect would be revealed, as co-expression of all factors precludes factor replacement. We used these secondary STEMCCA MEFs to establish another dose-response curve for the Alk5 inhibitor. In contrast to primary infected MEFs, which demonstrated a steady increase across the entire dose range tested (Figure 1D), the STEMCCA MEFs reached a maximum efficiency at 0.5μM, representing a ~60-fold increase, and began to decline at higher doses (Figure 2B). A similar pattern was also observed with another inhibitor, SB-431542, that targets the type I Tgfβ receptors, Alk-4, -5, and -7 (Supplementary Figure 1). These data corroborate the notion of a cooperative effect and also introduce the possibility of a factor replacement effect that operates at a different dose range, thus explaining the persistent efficiency increase with higher doses of the Alk5 inhibitor in primary infected MEFs, yet a decline in the secondary STEMCCA MEFs.
We therefore set out to investigate whether the Alk5 inhibitor could function to replace exogenous expression of any of the reprogramming factors. To this end, we infected primary MEFs with different combinations of the four factors in the presence or absence of inhibitor (Supplementary Table 1) and then scored for iPSC formation. In the four-factor control infection (OSMK; O=Oct4, S=Sox2, M=cMyc, K=Klf4), colonies were already apparent after 6 days; we therefore withdrew dox on day 8 to avoid detrimental effects of prolonged factor expression . At this time we also observed colonies in the cMyc replacement condition (OSK+inhibitor), but none without the inhibitor; thus, we withdrew dox from these cultures on day 8 as well. Remarkably, use of the Alk5 inhibitor (OSK+inhibitor) in place of cMyc resulted in a 2.5-fold higher efficiency than with the four factors alone (OSK+M), indicating that Alk5 inhibition was more potent than the action of cMyc in promoting reprogramming (Figure 2C).
For all of the other replacement conditions, we withdrew dox on day 16, which we reasoned would provide sufficient time for reprogramming. While most conditions did not yield any obvious colonies, the Sox2 replacement condition contained numerous colonies morphologically similar to iPSCs. These colonies appeared late in reprogramming, becoming readily apparent after 14 days. Based on morphology, we observed 50 colonies in the inhibitor-treated culture and 4 colonies in the control condition that were identical to iPSCs (data not shown). Withdrawal of dox, however, led to regression of many colonies in both conditions, indicating that the cells were not independent of factor expression. Nonetheless, we picked colonies from both conditions and tested their potential to form dox-independent lines. From the control condition, 0/4 were capable of expansion, indicating that infection with the three factors (OMK) was not sufficient to induce reprogramming, while in the inhibitor-treated culture, 3/22 colonies formed dox-independent iPSC lines, demonstrating successful replacement of Sox2. The efficiency of expansion was low compared to that of four-factor iPSC clones, where all picked colonies (6/6 in this set of experiments) were capable of forming stable dox-independent lines.
Immunostaining of the OMK+inhibitor iPSC lines demonstrated expression of Nanog, Oct4, and Sox2 (tested after three passages without dox) (Figure 2E). To further verify expression of pluripotency genes, quantitative RT-PCR analysis was performed using primer sets that distinguish endogenous and viral transcripts (Supplementary Figure 2A), which revealed pluripotency gene activation and viral gene silencing. To functionally assess pluripotency of the OMK+inhibitor iPSCs, we generated teratomas, which contained lineages from all three germ layers (Supplementary Figure 2B). We also tested the ability of these cells to contribute to chimeric mice; to this end, we labeled iPSCs with a lentivirus constitutively expressing the fluorescent protein, tdTomato, sorted the cells by flow cytometry, and injected them into blastocysts. Mice harvested on embryonic day E16.5 showed varying degrees of contribution (Figure 2F). As a more stringent test of pluripotency, we were also able to generate adult chimeras (Figure 2G), though their potential for germline transmission has not yet been evaluated. To ensure that these iPSCs were free of Sox2 viral integrations, we performed Southern blot analysis using a Sox2 cDNA probe. No extraneous bands were observed in the OMK+inhibitor lines, confirming the absence of exogenous Sox2 (Supplementary Figure 2C).
As a final test to determine whether the Alk5 inhibitor enabled iPSC formation under any of the replacement conditions, we passaged the primary cultures in the absence of dox, which would permit the amplification and selection of dox-independent cells that may not have been visible in the initial culture. We excluded OSK and OSMK conditions since these readily formed iPSCs. Confirming our observations from the primary cultures, only the Sox2 replacement condition (OMK+inhibitor) gave rise to dox-independent colonies (Figure 2D). The control condition (OMK –inhibitor) also contained colonies that expressed alkaline phosphatase (Figure 2D, upper left); however, these colonies appeared fibroblastic and did not grow, indicating that they were not iPSCs. Interestingly, while the Alk5 inhibitor enabled colony formation in the absence of either cMyc or Sox2, we could not obtain any colonies in the absence of both factors (OK; Figure 2D), suggesting that the inhibitor performs distinct functions from Sox2 and cMyc but can preferentially assume their role in the context of reprogramming with the three remaining factors.
The results presented here demonstrate that Tgfβ receptor I kinase inhibition enhances both the efficiency and kinetics of reprogramming in a dose-dependent manner, while activation of the Tgfβ signaling pathway blocks reprogramming. The Alk5 inhibitor exerts its strongest effect during the early stages of iPSC induction and acts in concert with the reprogramming factors to mediate its effect, rather than converting the fibroblasts to a state more amenable to reprogramming. In addition to its cooperative action, the Alk5 inhibitor can replace the individual roles of cMyc or Sox2, although it cannot replace them simultaneously. These results provide the first defined pathway that produces both a strong cooperative effect and can preferentially replace the roles of specific reprogramming factors.
An important question that remains is how Alk5 inhibition acts on a molecular level to enhance reprogramming. We observed that application of the inhibitor prior to factor expression was unable to mediate an increase in efficiency, suggestive of a transient effect and/or one that is context-dependent, requiring expression of the reprogramming factors to carry out its role. The observation that the Alk5 inhibitor acts early in reprogramming raises the question of whether it helps shut down the fibroblast gene expression program. In support of this, it has been shown that activation of Tgfβ signaling can promote an epithelial-to-mesenchymal transition ; fibroblast reprogramming involves a mesenchymal-to-epithelial conversion, which may be enhanced by inhibition of the Tgfβ signaling pathway.
An enhancement in efficiency could also be mediated through increased proliferation. We tested this possibility by examining proliferation rates during reprogramming, but found that the inhibitor led to a decrease in proliferation around day 6 (Supplementary Figure 3A). We noted that the inhibitor also mitigated the decrease in cell number that normally accompanies reprogramming (day 2) (Supplementary Figure 3A); however, the magnitude of this effect was unlikely to explain the large increase in overall reprogramming efficiency. Further, quantification of apoptosis with Annexin V staining showed a negligible difference between inhibitor-treated and untreated cells (Supplementary Figure 3B).
As yet another possibility, we tested the Alk5 inhibitor in the context of reprogramming by cell fusion (embryonic stem cells + MEFs), which led to a negative result (data not shown). This observation raises the question of whether Alk5 inhibition acts on pathways in direct reprogramming that are already operational in embryonic stem cells (ESCs), or whether the mechanisms of reprogramming by direct factors or cell fusion are fundamentally different.
The finding that the Alk5 inhibitor could replace either cMyc or Sox2, but not both together, suggests that its functions are distinct from these reprogramming factors. It is possible that replacement of both factors could occur with different doses of the inhibitor or with a longer period of dox administration; however, the conditions we used that enabled replacement of the individual factors were unable to elicit simultaneous replacement, indicating a unique mode of action. We tested whether the Alk5 inhibitor simply led to induction of Sox2 or cMyc expression; however, analysis by qRT-PCR revealed that only cMyc was significantly induced (1.3-fold) in OSK+inhibitor treated cells (Supplementary Figure 4), and there was no significant induction of Sox2 or cMyc in any of the other cells (OMK or OSMK). A possibility is that the Alk5 inhibitor acts on a pathway that completely bypasses the need for the individual reprogramming factors; for instance, it has been shown that Sox2 is not required for reprogramming by fusion of mESCs and human B lymphocytes .
Several Tgfβ superfamily members represent important contributors to the pluripotent state. For instance, BMP4 signaling helps maintain mouse ESCs in an undifferentiated state , and use of an Alk5 inhibitor in conjunction with GSK3β and MEK inhibitors facilitates rat and human iPSC line propagation and supports a mouse ESC-like phenotype . In contrast to its pluripotency-promoting effect in mouse ESCs, BMP4 induces differentiation of human ESCs towards trophectoderm , while Activin signaling maintains pluripotency of hESCs . The differential requirements of Tgfβ signaling in mouse and human ESCs may explain why we did not observe an increase in the efficiency of human fibroblast reprogramming (data not shown); it will be interesting to see whether activation of other pathways that promote the undifferentiated state also enhance reprogramming.
The highly characterized nature of the Tgfβ signaling pathway makes it an attractive model for examining how interactions between downstream targets and the four factors synergize to enhance reprogramming. The identification of such pathways will greatly facilitate our understanding of reprogramming at the molecular level, potentially leading to the discovery of novel targets that promote reprogramming and are of therapeutic value.
Vectors were constructed as previously described . Viruses were produced and titered by the University of Iowa Gene Transfer Vector Core. For all experiments, cells were infected overnight at an MOI of 10 in the presence of polybrene 6 μg/mL, which was experimentally determined to infect >90% of cells.
Cells were obtained from mice harboring a reverse tetracycline transactivator (rtTA) in the Rosa locus . Fibroblasts were derived from E14.5 mice; all experiments were conducted prior to passage 3. To induce reprogramming, cells were infected with the viral cocktail, then split two days later into ESC media (15% FBS, Invitrogen; 1000U/mL LIF) with doxycycline (1μg/mL). Fibroblast feeder cells were used only to maintain iPSC lines, not during reprogramming. The Alk5 inhibitor (Calbiochem/EMD 616452) was used at a concentration of 1μM, unless otherwise noted. The Alk-4/5/7 inhibitor, SB-431542, was obtained from Sigma; Tgfβ ligands were obtained from Peprotech (Tgfβ1, 100-21; Tgfβ2, 100-35).
Alkaline phosphatase (AP) staining was done using an AP substrate kit according to manufacturer directions (VectorLabs, SK-5100). The following antibodies were used for immunostaining: α-Nanog (1:200, ab21603, Abcam), α-Oct4 (1:100sc-8628, Santa Cruz Biotech), α-Sox2 (1:200, AB5603, Millipore).
RNA was extracted by using a Qiagen RNeasy kit (74104), then converted to cDNA with the Superscript III First-Strand synthesis system (Invitrogen) using oligo-dT primers. qRT-PCRs were carried out using Brilliant II SYBR Green mix (Stratagene) and run on a Stratagene MXPro400. Reactions were carried out in duplicate with –RT controls, and data were analyzed using the delta-delta Ct method.
Annexin V and propidium iodide staining was done using the Annexin V-FITC kit (BD Pharmingen, 556547) according to manufacturer directions. Cells were analyzed by flow cytometry on a FACSCalibur (BD).
A 25cm2 equivalent of confluent iPSCs grown on feeders was harvested and injected subcutaneously into SCID mice. Teratomas were harvested 3 weeks later and stained with hematoxylin/eosin.
BDF1 females were superovulated with PMS and hCG, then mated with BDF1 males. Blastocysts were flushed from uterine horns on day 3.5, and 10-20 labelled iPSCs were injected per blastocyst. Swiss-webster mice were set up with vasectomized males and used as recipients for the injected blastocysts.
Supplementary Table 1. Factor combinations used for replacement experiments.
Supplementary Figure 1. Dose-response curve for the Alk-4/5/7 inhibitor SB-431542. Dox and the inhibitor were applied for 8 days in secondary STEMCCA MEFs, and colonies were quantified on day 12 based on Oct4 immunostaining.
Supplementary Figure 2. Characterization of OMK+inhibitor iPSCs.
A. Quantitative RT-PCR data demonstrating expression of pluripotency genes (left) and silencing of viral genes (right). Expression level was normalized to Gapdh. Three 3-factor lines (OMK+inhibitor), a 4-factor line (OSMK), and a control ESC line (v6.5) were included in the analysis. Left: primers were designed to measure total Oct4 or Nanog; for Sox2, cMyc, and Klf4, primers only amplified endogenous transcripts. Right: primers were designed to only measure viral transcripts.
B. Teratoma derived from an OMK+inhibitor iPSC line demonstrating differentiation into lineages from all three germ layers. Left, keratinized epithelium; middle, cartilage; right, gut-like epithelium.
C. Southern blot analysis for Sox2 integrations. Three 3-factor (OMK +inhibitor) iPSC lines were analyzed (3F), as well as a control four-factor (OSMK) iPSC line, which showed an additional band in both digests, and an ESC line (v6.5), which showed no additional bands. Genomic DNA was digested with either BamHI or XhoI, and the blot was probed with a Sox2 cDNA. Panels on the right show the ethidium-bromide stained gels used for the blots.
Supplementary Figure 3. Cell proliferation and apoptosis during iPSC induction.
A. MEFs were infected with either three (OSK) or four (OSMK) factors, then induced with dox in the presence or absence of the Alk5 inhibitor (1μM). Cells were counted at each timepoint noted, and values represent the fractional change in cell number: (fraction of starting cell number at timepoint b - fraction of starting cell number at timepoint a)/time between b and a.
B. Annexin V and propidium iodide (PI) staining to quantify apoptosis during iPSC induction. Secondary cells carrying the polycistronic STEMCCA construct were treated with dox for 2 days in the presence or absence of the Alk5 inhibitor (1μM), then stained and analyzed by flow cytometry. Annexin V-positive and PI-negative cells represent the fraction of living cells in the process of apoptosis.
Supplementary Figure 4. Induction of Sox2, cMyc, or Nanog with Alk5 inhibitor treatment. MEFs were infected with various factor combinations, then treated with dox for three days in the presence or absence of 1mMeither three (OSK) or four (OSMK) factors, then induced with dox in the presence or absence of Alk5 inhibitor (1μM). qPCR analysis was done using primers that specifically detect endogenous Sox2 or cMyc, as well as total Nanog. Experiments were done with both biological and technical triplicates, and a Student’s T-test (paired, two-tailed) was used to assess statistical significance. The Alk5 inhibitor led to a small (1.3-fold) but significant induction of cMyc in cells infected with OSK (p=0.016); No other genes showed a statistically significant change in expression (Sox2: OMK, p=0.31; OSMK, p=0.43. cMyc: OSMK, p=0.21. Nanog: OMK, p=0.25; OSK, p=0.55; OSMK, p=0.53).
We thank Matthias Stadtfeld for the pilot experiments that led to conception of the study, as well as insightful discussion and critical reading of the manuscript; Jose Polo for insightful discussion. N.M. is supported by a graduate scholarship from the Natural Sciences and Engineering Research Council of Canada and a Sir James Lougheed Award from the Alberta Scholarships Program. K.H. is supported by a National Institutes of Health (NIH) Director’s Innovator Award, the Harvard Stem Cell Institute, the Kimmel Foundation, and the V Foundation.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.