Naive Human Pluripotent Cells Feature a Methylation Landscape Devoid of Blastocyst or Germline Memory

Abstract

Human embryonic stem cells (hESCs) typically exhibit "primed" pluripotency, analogous to stem cells derived from the mouse post-implantation epiblast. This has led to a search for growth conditions that support self-renewal of hESCs akin to hypomethylated naive epiblast cells in human pre-implantation embryos. We have discovered that reverting primed hESCs to a hypomethylated naive state or deriving a new hESC line under naive conditions results in the establishment of Stage Specific Embryonic Antigen 4 (SSEA4)-negative hESC lines with a transcriptional program resembling the human pre-implantation epiblast. In contrast, we discovered that the methylome of naive hESCs in vitro is distinct from that of the human epiblast in vivo with loss of DNA methylation at primary imprints and a lost "memory" of the methylation state of the human oocyte. This failure to recover the naive epiblast methylation landscape appears to be a consistent feature of self-renewing hypomethylated naive hESCs in vitro.

Figure 1

5iLAF SSEA4 negative subpopulation recapitulates naïve expression pattern. A) Upper: brightfield image of primed UCLA1 hESCs. Lower: Flow cytometry plot of primed UCLA1 hESCs stained for SSEA4 and TRA-1-81 B) Upper: UCLA1 hESCs reverted in 5iLAF. A mixture of round and flat colonies are observed. Lower: Flow cytometry plot of 5iLAF cultured UCLA1 hESCs stained for SSEA4 and TRA-1-81. Scale bar indicates 200μm. C) 5iLAF cells were sorted into SSEA4⁺ and SSEA4⁻ populations. Upon re-plating, the SSEA4⁺ cells formed flat colonies and the SSEA4⁻ cells formed round colonies (n=2 biological replicates). Scale bar= 100μm. D) hESC line called UCLA20n, derived from a 5-day human blastocyst in 5iLAF. Left: Brightfield image. Scale bar indicates 200μm. Right: Flow cytometry plot of TRA-1-85+ (human) UCLA20n hESCs stained for SSEA4 and TRA-1-81. E) Expression of genes identified by others as associating with the naïve and primed states in mice. Expression level is determined by RNA-seq. For 5iLAF SSEA4 negative (neg) and primed hESCs, n=4. For 5iLAF SSEA4 positive (pos), n=2. Other data comes from published RNA-seq or microarray datasets. Methodology, cell type, and citation are indicated. F) A set of “pre-implantation epiblast” and “primed” specific genes were defined based on published data. Expression of these genes is shown for various methodologies, relative to primed controls from the same dataset. UCLA20n was normalized to a primed UCLA1 library generated and sequenced at the same time.

Figure 2

Naïve hESCs fail to recapitulate naïve-specific methylation pattern. A) Average genome wide-CG methylation level in primed and 5iLAF UCLA1 hESCs, shown in comparison with published datasets. For 5iLAF SSEA4 negative (neg) and primed hESCs, n=3. For 5iLAF SSEA4 positive (pos), n=2. B) DNA methylation is shown for a region of chromosome 10. Each bar indicates a single CG, and the height of the bar indicates the percentage of CG methylation. Where multiple CGs are too close to be visually rendered separately, an average value is shown. C) Correlation plots relative to human oocyte using 100kb genome bins. D) DNA methylation over stable primary imprints. The average methylation level of each imprint in a given sample is represented as one point in the box and whisker point. E) DNA methylation over the paternally imprinted H19 locus. Each bar indicates a single CG, and the height of the bar indicates the fraction of CG methylation. Where multiple CGs are too close to be visually rendered separately, an average value is shown. F) Total DNA methylation for three competing approaches for culturing naïve cells. Because the Gafni 2013 data was generated by RRBS, only CGs that had coverage in the Gafni 2013 dataset are included in this analysis to make the data comparable. G) Expression (RPKM) of DNA methyltransferases, DNMT cofactors, and Tet-family oxidases as measured by RNA-seq (n=4). H) RNA and protein levels of DNA methyltransferases in 5iLAF SSEA4 neg UCLA1 hESCs relative to primed. RNA level is determined from RNA-seq data (n=4), protein level from quantitative westerns (UHRF1, n=6 Western blots; DNMT1, DNMT3A, DNMT3B n=2; DNMT3L n=1).