Conceived and designed the experiments: JC GYW JHK. Performed the experiments: JC GYW XDG WJH. Analyzed the data: GYW CQL. Contributed reagents/materials/analysis tools: CQL JMW JHK. Wrote the paper: JC GYW CQL JHK.
The authors have declared that no competing interests exist.
Induced pluripotent stem (iPS) cells were first generated by forced expression of transcription factors (TFs) in fibroblasts. Recently, iPS cells have been generated more rapidly and efficiently using miRNAs with or without other transcription factors. However, the specific and collaborative roles of miRNAs and transcription factors in pluripotency acquisition and maintenance remain to be further investigated. Here, based on the miRNA profiling in mouse embryonic fibroblasts (MEFs), MEFs infected with Oct3/4, Sox2, Klf4 and c-Myc (OSKM) for 1, 2, 4, or 8 day, two iPS cell lines and ES cells, representing iPS activation and maintenance steps, we found that two unique miRNA sets are responsible for different steps of iPS generation, and the miRNA expression profiles of iPS cells are very similar to that of ES cells. Furthermore, we searched for transcription factors binding sites at the promoter regions of up-regulated miRNAs, and found that up-regulated miRNAs such as the miR-429-200 and miR-17 clusters are directly activated by exogenous TFs. The GO and pathway enrichment for candidate target gene sets of miRNAs or OSKM provided a clear picture of division and collaboration between miRNAs and OSKM during completion of the iPS process. Compared with the pathways regulated by OSKM, we found that miRNAs play critical roles in regulating iPS-specific pathways, such as the adherens junction and Wnt signaling pathways. Furthermore, we blocked miRNA expression using Dicer knockdown, and found that the level of miRNAs was decreased following this treatment, and the efficiency of iPS generation was significantly repressed. By combining high-throughput analysis, biostatistical analysis and functional experiments, this study provides new ideas for investigating the important roles of miRNAs, the mechanisms of miRNAs and related signaling pathways, and the potential for many more applications of miRNAs in somatic cell reprogramming.
Mouse embryonic fibroblasts (MEFs) can be successfully reprogrammed to a pluripotent state using four transcription factors (TFs): Oct3/4, Sox2, Klf4 and c-Myc (OSKM), which are identified as reprogramming factors
However, reprogramming that results from the induction of defined factors is slow (needs 2 or 3 weeks) and inefficient (less than 1%), suggesting that the four transcription factors are capable, but somewhat insufficient, for cell reprogramming. The slowness and inefficiency of reprogramming may lead to defective reprogramming and not only prevent the clinical applications, but also lead to misunderstandings regarding the mechanisms underlying reprogramming. In addition, the use of proto-oncogenes, such as Klf4 or c-Myc, would increase the risk of tumor formation when integrated into the iPS cell genome. Therefore, many researchers have investigated novel reprogramming factors and/or combinations of these factors, such as L-Myc
In 2008, Marson and his colleagues carried out a systematic analysis of miRNAs and the transcription factors Oct3/4, Sox2, Nanog, and Tcf3 and connected miRNA genes to the core transcriptional regulatory circuitry of embryonic stem cells
Presently, the complex process of iPS generation has been classified into three phases, initiation, maturation and stabilization, based on gene expression profiling and cell morphology changes
Unsupervised hierarchical clustering of miRNA expression data across ES cells, early-infected series, iPS and MEF cells reveals interesting patterns of miRNA expression among these cell types (
(
According to a slight change in the early-infected series, a significant expression signature for activating the iPS reprogramming process is that it requires at least a 1.5-fold change in expression between the early-infected series and MEF cells. In total, 114 miRNAs are significantly changed and can be classified into four groups (
(
The significant expression signature for maintenance of the iPS state is that it requires at least a two-fold change in expression between iPS and MEF cells. We found that 191 miRNAs that are present at significantly different levels in iPS cells (iPSC1 and iPSC2) compared to MEF cells. The miRNA expression profiles are more similar within iPS cells than between MEF and ES cells (
(
Many miRNAs may play important roles in the whole reprogramming process, including the activation and maintenance of pluripotency. We therefore checked the overlapping miRNA expression signature for the activation and maintenance of iPS and found 73 miRNAs which are termed common significant miRNAs. Among these, 34 tend to be down-regulated during the 8 days after infection and in iPS cells, whereas 21 tend to be up-regulated during the 8 days after infection and in iPS cells. The remaining 18 miRNAs exhibit different trends among early-infected MEFs and iPS cells, and these may change significantly from the activation step to the maintenance step.
We then used TAM analyses to determine the enrichment family or cluster of these groups. Among 82 pre-miRNAs for these 73 mature miRNAs, which are commonly involved for activation and maintenance, we found all 7/8 members of the miR-17 family, all 3/3 members of the miR-19 family, 6/6 miRNAs in the cluster of miR-17-92 and 3/3 miRNAs in the cluster of miR-106-93. All of these miRNAs are up-regulated in iPS cells, indicating the importance of miR-17 and miR-19 in the activation and maintenance of iPS pluripotency (
Group | Cluster | HP | P | HS | S | Exact |
Bonferroni | FDR | Members |
Common | mir-17-92 | 48 | 247 | 6 | 6 | 4.14E-05 | 1.08E-03 | 7.90E-04 | mir-17;mir-18a;mir-19a;mir-19b-1; mir-20a;mir-92a-1 |
mir-25-93-106 | 48 | 247 | 3 | 3 | 0.0070 | 0.18 | 0.030 | mir-106b;mir-93;mir-25 | |
Unique maintenance | mir-290 | 64 | 247 | 7 | 7 | 6.08E-05 | 1.58E-03 | 7.90E-04 | mir-290;mir-291a;mir-291b;mir-292; mir-293;mir-294;mir-295 |
let-7 | 64 | 247 | 3 | 3 | 0.017 | 0.44 | 0.055 | let-7a-1;let-7d;let-7f-1 | |
mir-429-200 | 64 | 247 | 3 | 3 | 0.017 | 0.44 | 0.055 | mir-200a;mir-200b;mir-429 | |
Unique activation | mir-497-195 | 20 | 247 | 2 | 2 | 0.0063 | 0.16 | 0.030 | mir-195;mir-497 |
mir-29 | 20 | 247 | 2 | 2 | 0.0063 | 0.16 | 0.030 | mir-29a;mir-29b-1 | |
mir-30 | 20 | 247 | 2 | 2 | 0.0063 | 0.16 | 0.030 | mir-30c-1;mir-30e |
Group | Family | HP | P | HS | S | Exact |
Bonferroni | FDR | Members |
Common | mir-25 | 76 | 422 | 3 | 4 | 0.020 | 0.73 | 0.081 | mir-25;mir-92a-2;mir-92a-1 |
mir-19 | 76 | 422 | 3 | 3 | 0.0057 | 0.21 | 0.030 | mir-19a;mir-19b-2;mir-19b-1 | |
mir-17 | 76 | 422 | 7 | 8 | 3.32E-05 | 1.23E-03 | 4.08E-04 | mir-106a;mir-106b;mir-17;mir-18a; mir-20a;mir-20b;mir-93 | |
Unique maintenance | mir-290 | 89 | 422 | 7 | 7 | 1.53E-05 | 5.66E-04 | 2.83E-04 | mir-290;mir-291a;mir-291b; mir-292;mir-293;mir-294; mir-295 |
mir-8 | 89 | 422 | 5 | 5 | 0.00038 | 0.014 | 0.0035 | mir-141;mir-200a;mir-200b;mir-200c; mir-429 | |
let-7 | 89 | 422 | 10 | 12 | 5.16E-06 | 1.91E-04 | 1.91E-04 | let-7a-1;let-7a-2;let-7c-1;let-7c-2;let-7d;let-7e;let-7f-1;let-7f-2;let-7g;let-7i | |
Unique activation | mir-15 | 44 | 422 | 3 | 5 | 0.0091 | 0.34 | 0.042 | mir-195;mir-15a;mir-15b |
mir-29 | 44 | 422 | 3 | 4 | 0.0039 | 0.15 | 0.024 | mir-29a;mir-29b-1;mir-29b-2 | |
mir-30 | 44 | 422 | 4 | 6 | 0.0013 | 0.049 | 0.0099 | mir-30b;mir-30c-1;mir-30c-2;mir-30e |
P represents the number of miRNAs included in all miRNA sets, S represents the number of miRNAs included in miRNA set A, HP represents the number of input miRNAs included in P, and HS represents the number of miRNAs that are of interest included in S;
miRNAs with a higher expression level in iPS cells (the infected series) than MEFs are marked in red, whereas miRNAs with a lower expression level in iPS cells are marked in blue.
Based on the miRNA promoter information and TF binding data from Marson et al. 2008, we summarized the transcription factors that bind at the promoter regions of up-regulated miRNAs (
Groups | miRNAs | TSS position | Oct4 | Sox2 | Nanog | Tcf3 |
Common | mir-17 cluster | chr14:113921300-113927025 | Oct4 | |||
mmir-96-183 | chr6:30114875-30130825 | Oct4 | Sox2 | Nanog | Tcf3 | |
mir-20b-10b/a | chrX:48985925-48991150 | Oct4 | Sox2 | Nanog | Tcf3 | |
Unique activation | mir-31 | chr4:88399300-88401975 | Oct4 | Sox2 | ||
mir-27a | chr8:87086300-87095525 | Oct4 | Sox2 | |||
mir-27b | chr13:63284054-63284254 | Oct4 | Sox2 | Nanog | ||
mir-29a-b | chr6:31006975-31008175 | Sox2 | ||||
mir-101b | chr19:29167271-29167471 | Sox2 | ||||
Unique maintenance | mir-290 cluster | chr7:3218001-3219675 [mm9] | Oct4 | Sox2 | Nanog | |
mir-429-200a/b | chr4:154903109-154903309 | Sox2 | ||||
mir-141-200c | chr6:124683151-124685075 | Oct4 | ||||
mir-124-1 | chr14:63540450-63546275 | Oct4 | Nanog | |||
mir-150 | chr7:44988600-44989794 | Oct4 | Sox2 | |||
mir-205 | chr1:195208350-195211700 | Oct4 | Sox2 | Nanog | ||
mir-181d | chr8:87069525-87071900 | Oct4 | Sox2 | Nanog | ||
mir-135b | chr1:134018876-134019076 | Oct4 | Sox2 | Nanog | Tcf3 | |
mir-124-2 | chr3:17986635-17986835 | Oct4 | Sox2 | Nanog | Tcf3 | |
mir-302 | chr3:127537000-127537941 | Oct4 | Sox2 | Nanog | Tcf3 | |
mir-367 | chr3:127537000-127537941 | Oct4 | Sox2 | Nanog | Tcf3 | |
mir-182 | chr6:30114875-30130825 | Oct4 | Sox2 | Nanog | Tcf3 | |
mir-363 | chrX:48985925-48991150 | Oct4 | Sox2 | Nanog | Tcf3 | |
mir-672 | chrX:100403736-100403936 | Oct4 | Sox2 | Nanog | Tcf3 |
miRNA promoter information and TFs binding data were obtained from Table S6 of Marson et al. 2008.
We defined direct targets for all significant miRNAs as requiring the agreement of all three popular prediction methods (TargetScan, Pictar and miRnada) and exhibiting a significant opposite tendency (at least a two-fold expression change) in the mRNA expression profiles of iPS cells and MEFs. We found that 527 target genes of 201 miRNAs met these criteria. To classify transcription factors as direct targets, we required a validated TF binding signal and higher expression in iPS cells (at least a two-fold increase in expression relative to MEFs). The mRNA expression data and TF binding information were collected from a previous report
GO enrichment analyses of the target genes for miRNAs and transcription factors showed that miRNAs and OSKM act both independently and in collaboration to complete the iPS induction process. These two sets of targets focus on metabolic and developmental processes, which are critical for most biological processes, including the iPS process. The direct targets of OSKM are enriched in cell cycle, cell division, cell proliferation, response to stress and stem cell maintenance, whereas the direct targets of miRNAs focus on the regulation of cellular and biological processes, cell motion and cellular component morphogenesis (
(
Furthermore, pathway enrichment analyses of the two target sets provide a clear picture of division and collaboration in the iPS process between miRNAs and OSKM in regulating signaling pathways (
(
Transcription factors are marked in red squares. Significant miRNAs are labeled using diamond, and candidate target genes of miRNAs and TFs are marked using circles. Pink and bright green colors represent high- and low- expression levels, respectively. Purple octagons represent the related signaling pathways in pluripotency-associated pathways. Green lines represent activation, red lines represent inhibition, and blue lines represent candidate target genes that belong to the corresponding signaling pathways.
To confirm the synergetic cooperation of miRNAs and transcription factors in iPS generation, we constructed a retroviral vector to knock down the expression of Dicer, a key molecule for the generation of miRNAs. Due to the decreased expression level of Dicer, the expression levels of miRNAs were down-regulated (
(
At present, iPS generation has been classified into three phases, initiation, maturation and stabilization, based on gene profiling during OSKM-induced MEF reprogramming
In the activation step of iPS generation, increased expression of the miR-29 family and decreased expression of the miR-30 family are essential. The combined miRNA expression, miRNA target and signaling pathway assays revealed that the members of the miR-30 family may negatively regulate genes involved in MAPK signaling and adherens junctions
OSKM transcription factors, in particular Oct3/4 and Sox2 are recognized as the most important reprogramming factors, and to date, viral transduction of OSKM factors remains a robust method to induce iPS cell generation and is often used to fully investigate the mechanisms of pluripotency generation. In our study, we have combined miRNA expression, miRNA targets, TFs binding information, and mRNA expression to study the iPS process, including during the first 8 days post-infection and final iPS cell generation. We found that synergic cooperation between miRNAs and OSKM plays a critical role in completion of the reprogramming process. KEGG pathway enrichment analysis of the candidate targets of miRNAs and OSKM demonstrated that both focus on the cell cycle and pathways involved in cancer. A sub-network based on miRNAs, TFs, candidate target genes, and signaling pathways demonstrated that transcription factors might function through directly binding to the promoter region of specific miRNAs. Interestingly, our study indicated that miRNAs are more specific than OSKM in pluripotency acquirement and maintenance, as indicated by the closer regulation of critical developmental signaling, such as in the adherens junction and Wnt signaling pathways
Somatic cell reprogramming by viral transduction is a double-edged sword. Despite the robustness of the protocol, the host cell viral response acts as a roadblock to efficient reprogramming by initiating a damaging and repressive chain of events, which includes ROS production, DNA damage, the activation of p53 and senescence. Enrichment analysis showed that the direct targets of OSKM are enriched in targets involved in these correlated process or pathways, whereas the direct targets of miRNAs are related to processes and pathways correlated with pluripotency. Rather than replacing one or more transcriptional factors using miRNAs in iPS induction
Aided by high-throughput analyses, such as miRNA profiling analysis, experimental studies can discover not only the phenomena and specific mechanisms behind biological processes but also reveal the entire process, including miRNAs, transcription factors, signaling pathways and the regulatory networks between them. On the basis of high-throughput assays of differently treated samples, we indentified significant miRNA and gene signals using biostatistical analysis. Based on the existing extensive miRNA target database and another study of TFs binding sites, we discovered candidate regulation information for the entire system. The corresponding expression trend in iPS cells and MEFs improved the correctness of these regulatory relationships. We then developed a clear picture of the entire iPS reprogramming process, which involves specificity and collaboration between miRNAs and TFs. This comprehensive analysis of the relationships among candidate molecules, miRNAs, transcription factors, and signaling pathways, may provide more effective experimental strategies and facilitate future research in related fields.
In conclusion, on the basis of the miRNA expression profile during the iPS activation and maintenance stages in MEFs, MEFs infected with OSKM for the first 8 days, iPS and ES cells, we found that unique sets of miRNAs are responsible for various stages of iPS cell generation by combining experimental and biostatistical analysis. Furthermore, GO and pathway enrichment assays provide a clear picture of the synergetic cooperation between miRNAs and TFs. Furthermore, miRNAs play critical roles in regulating iPS-specific pathways, such as the adherens junction and Wnt signalling pathways. These findings may provide new evidence for clarifying the roles and mechanisms of miRNAs and signaling pathways during somatic cell reprogramming.
The vectors pMX-Oct4, pMX-Sox2, pMX-Klf4 and pMX-c-Myc were used in this study for four transcription factors
To generate retroviruses for four transcription factors (OSKM), Plat-E cells were seeded at a density of 8×106 cells per 100 mm dish on the day before transfection. On the next day, pMX-Oct4, Sox2, Klf4 and c-Myc vectors were introduced into the Plat-E cells using Fugene HD transfection reagent (Roche), according to the manufacturer’s instructions. Fugene HD transfection reagent (20 µl) was diluted in 500 µl of Opti-MEM and incubated for 5 min at room temperature (RT). Plasmid DNA (8 µg) was added to the mixture, which was then incubated for 15 min at RT. After incubation, the DNA/Fugene HD mixture was added drop wise onto the Plat-E cells. The cells were then incubated overnight at 37°C under 5% CO2. After transfection for 8∼10 h, the medium was replaced with 8 ml of fresh medium, and virus-containing supernatant was harvested at 48 h after transfection. Virus-containing supernatants derived from these Plat-E cultures were filtered through 0.45 µm Millex-HV (Millipore) filters and 4 µg/ml of polybrene (Sigma) was added at a final concentration.
A series of infected-MEF cells was collected after infection with OSKM for 1, 2, 4, and 8 d. For iPS cells induction, OG-MEFs were seeded before infection for 13 h. Viral supernatant was added to the OG-MEFs and centrifuge at 800×g; 90 min later, virus-containing supernatant was replaced with 1 ml of high-glucose DMEM containing 10% FBS (Hyclone) and 1×P/S (Thermo). Two days later, the medium was replaced with KOSR medium containing LIF, β-mercaptoethanol, L-glutamine and NEAA, whereupon the cells were maintained until GFP-positive colonies appeared.
Mouse miRNA expression data were extracted using an Agilent Mouse miRNA Array. For the analyses, the array data for MEFs, the MEF-infected series, E14 and iPS cells were normalized by using Robust Multichip Analysis (RMA) in R (Bioconductor 2.8). The resulting mouse miRNA data sets contained 325 miRNAs (at least expressed in one sample). Global array clustering was performed using Cluster 3.0 and presented using Java Treeview 1.1.6; miRNA expression values were presented as a log2 ratio compared to MEFs.
A hypergeometric test was used to determine significant overrepresentation of the miRNA sets among a list of miRNAs of interest. Assuming that P represents the number of miRNAs included in all miRNA sets, S represents the number of miRNAs included in miRNA set A, HP represents the number of input miRNAs included in set P, and HS represents the number of miRNAs that are of interest included in set S, and the probability of HS miRNAs of interest being in miRNA set A is calculated using the TAM method
We searched miRNA target information using three popular prediction methods (TargetScan, Pictar, and miRnada) and the miRGen v3 database (
GO and pathway enrichment analysis of the candidate target sets, which were regulated by miRNAs and TFs are completed using the database DAVID v6.7 for annotation, visualization and integrated discovery
(TIF)
(TIF)
(XLS)
(DOC)
(XLS)
(DOC)