|mESC Showdown: 2i vs. Serum - “The Transcriptional and Epigenomic Foundations of Ground State Pluripotency”|
Embryonic stem cells (ESCs) maintained in serum are often morphologically heterogeneous and also express pluripotency associated transcription factors in a heterogeneous manner (Toyooka et al) perhaps suggesting that the pluripotent state in unstable. However, the derivation and maintenance of mESCs in serum-free conditions through the use of two small molecule kinase (Mek and GSK3) inhibitors (“2i” - PD0325901 and CHIR99021) which shield pluripotent cells from differentiation triggers (Ying et al) and eliminates mosaic transcription factor expression (Wray et al), has enabled derivation of germline-competent ESCs from all mouse strains tested and for the first time from rats (Hanna et al, and Nichols et al). 2i conditions are suggested to mimic the environment in the mature mouse inner cell mass (ICM) thereby allowing ESCs to enter a “naïve” ground state (Guo et al). Now, researchers from the laboratories of From the labs of Austin Smith and Hendrik G. Stunnenberg have applied genome wide transcriptional and epigenomic analysis to serum-maintained ESCs and 2i-maintained “naïve” ESCs and have uncovered some interesting differences between ESCs in the two growth conditions (Marks et al).
Six lines in total were studied; three ESC lines derived and maintained in 2i plus LIF (“2i” ESCs) and three ESC lines established and cultured in serum plus LIF (“serum” ESCs), all of which were deemed to be fully pluripotent as demonstrated by their competence to generate high-contribution chimaeras with germline transmission. However, subsequent gene expression analysis found that around 2000 genes were overexpressed 2-fold in 2i ESCs compared to serum ESCs; 160 genes expressed in 2i ESCs were silent in serum ESCs and 461 genes were expressed only in serum ESCs. Most pluripotency-associated genes were transcribed at similar levels, and of the 9 genes in this category that were overexpressed in 2i ESCs, only Tcl1 had been previously implicated as a regulator of self-renewal (Ivanova et al). In serum ESCs, 16 pluripotency-associated genes were over-expressed as compared to 2i ESCs, including Myc and the Id proteins. Functional annotation of 2i over-expressed genes demonstrated enrichment of terms associated with metabolic processes and cell cycle regulation. Genes overexpressed in serum ESCs were enriched for terms associated with developmental processes; particularly ectoderm and mesoderm germ layer specification.
Cellular morphology and homogeneity of pluripotency-associated gene expression differed between the two growth conditions, as expected. 2i ESCs were morphologically uniform and homogeneously expressed pluripotency-associated genes (Nanog, Rex1, Stella, and Klf4), while serum ESCs were heterogeneous for both. Interestingly, when the growth medium conditions were switched, ESCs changed to take on the morphological characteristics of the new condition and previously observed expression differences also switched suggesting that the cell states exhibited in 2i and serum conditions are interconvertible and are dependent on their growth conditions.
Next, the chromatin landscape of these cells were analysed and contrasted to their expression profiles. The 2000 genes most highly expressed under both conditions revealed the expected epigenetic patterns; tri-methylation of Lysine 4 Histone H3 (H3K4me3) associated with active promoters, and tri-methylation of Lysine 36 Histone H3 (H3K36me3) associated with coding body of transcribed genes. Tri-methylation of Lysine 9 H3 (H3K9me3), associated with repressed chromatin, had a similar pattern in 2i and serum ESCs, in that it was prominent at satellites and imprinted genes. Tri-methylation of Lysine 27 H3 (H3K27me3), also associated with repressed chromatin, had a similar promoter pattern under both growth conditions, except for genes with just above basal levels of expression for which in 2i ESCs exhibited less methylation than in serum ESCs. This difference in histone methylation was not associated with an increase in expression. However, when studying H3K27me3 and the levels of Ezh2, the enzyme which mediates this modification, 5 kilobases up and downstream from genes, a major reduction was observed in 2i ESCs. Upon the switching of growth medium, as before for morphology and expression patterns, the epigenomic states also switched with regards to H3K27me3 and Ezh2 occupancy. Further analysis in 2i ESCs of H3K27me3 levels demonstrated a global lack of H3K27me3 at promoter regions, while in serum ESCs 60-65% of H3K27me3 is observed at promoters. Yet, while H3K27me3 is also reduced at long interspersed nuclear element (LINE) repeats in 2i ESCs, much higher levels are observed at satellite sequences. Immunoblotting for H3K27me3 suggested a similar cellular level of this modification between 2i and serum ESCs signifying that H3K27me3 becomes redistributed depending on growth conditions.
When found with H3K4me3 at promoter regions, H3K27me3 marks bivalent genes which are poised for activation in ESCs (Azuara et al, Bernstein et al and Mikkelsen et al) and a reduction in promoter-specific H3K27me3 may alter the levels of bivalently-marked genes. Indeed, serum ESCs contained 3000 promoters which were deemed bivalent, and this dropped to 1000 under 2i conditions, mainly due to the previously observed loss of H3K27me3. Overall, bivalent genes were generally associated with developmental processes under both conditions; 31% of genes were expressed at background levels for both conditions while 14% of bivalent genes were serum ESC-specific and 4% were 2i ESC-specific. Again, these findings were interconvertible upon medium switching. Interestingly, at two representative genes (Hey2 and Meternl), transcription was slightly upregulated in serum ESCs as compared to serum ESCs and this was correlated to a gain in the normally repressive H3K27me3 modification.
Strikingly, RNA polymerase II (Pol II) was evident over transcription start sites at higher levels in 2i ESC than in serum ESCs, suggestive of promoter proximal pausing. The pluripotency-associated transcription factor c-Myc is implicated in Pol II pause release (Rahl et al) and c-Myc mRNA levels were noted to be decreased (40-50 fold) in 2i ESCs. The promoters of c-Myc target genes that are upregulated in serum ESCs promoters exhibited high levels of H3K4me3 and Pol II. In serum ESCs, where c-Myc levels are higher, Pol II was reduced at these promoters and increased over coding bodies, suggesting that c-Myc allows the release of Pol II from promoters. However, direct c-Myc targets represented less than 15% of genes differentially expressed between 2i and serum ESCs and gene ontology classification of c-Myc targets upregulated in serum ESCs did not identify categories associated with developmental processes and suggested that c-Myc is unlikely to be a major determinant of differential gene expression and the heterogeneities observed in serum ESCs.
Histone modification patterns at genes where expression changes more than 2-fold in 2i ESCs compared to serum ESCs were found, showed increased promoter-specific H3K4me3 and coding body specific H3K36me3, while H3K27me3 was reduced. Interestingly, when studying genes upregulated in serum ESCs as compared to 2i ESCs, while H3K36me3 in the coding body increased, no significant changes in H3K4me3 were observed and again, H3K27me3 was increased. Following analysis of Pol II occupancy, genes upregulated in 2i ESCs were linked to increased Pol II at the transcription start site and over the coding body, while genes upregulated in serum ESCs were associated with a reduction in Pol II occupancy of the transcription start site and an increase over the coding body, suggesting that transcriptional elongation at genes already loaded with Pol II is a widespread mechanism of upregulation in serum ESCs.
Lastly and, perhaps most importantly, differentiation under the two growth conditions was analysed. Initial monolayer neural induction with Sox1-GFP ESCs (Ying et al) demonstrated that 2i ESCs differentiated more efficiently, even though serum ESCs already express some neural-associated genes at low levels. Next, Rex1-GPF fractions of serum and 2i ESCs were assayed during differentiation as embryoid bodies (EBs). Both positive fractions initially downregulated Nanog and Rex1 and upregulated Fgf5, then at 3 days, Fgf5 levels decreased and mesodermal and endodermal markers increased (T, Tbx6, Cxcr4, Sox17 and Gata4). The negative populations exhibited accelerated upregulation of T and Tbx6 consistent with their partial differentiation and the loss of self-renewal.
Overall, 2i ESCs exhibit lower expression of lineage-affiliated genes and fewer bivalent domains, with precocious transcription of developmental genes restrained by RNA polymerase II promoter-proximal pausing. Importantly, ESCs transferred between serum and 2i growth conditions switch their transcriptional and epigenomic profiles suggesting that a significant component of ESC signatures reflects an induced serum response. Looking forward, in depth differentiation analysis of these cells may provide further proof of the usefulness of the naïve state of pluripotency. Isolation of human ESCs in the naïve state and subsequent analysis will also be of great interest (Hanna et al) and further, the identification of inhibitors which can induce this state would be of immediate impact.
Azuara, V., et al. (2006).
Bernstein, B.E., et al. (2006).
Guo, G., et al (2010).
Hanna, J., et al (2010).
Hanna, J., et al. (2009).
Ivanova, N., et al. (2006).
Marks, H., et al. (2012).
Mikkelsen, T.S., et al. (2007).
Nichols, J., et al. (2009).
Rahl, P.B., et al. (2010).
Toyooka, et al. (2008).
Wray, J., et al. (2011).
Ying, Q.L., et al (2003).
Ying, Q.L., et al (2008).
STEM CELLS correspondent Stuart P. Atkinson reports on those studies appearing in current journals that are destined to make an impact on stem cell research and clinical studies.
Original study from Cell.