Nuclear Reprogramming of Somatic Cells After Fusion with Human Embryonic Stem Cells

See allHide authors and affiliations

Science  26 Aug 2005:
Vol. 309, Issue 5739, pp. 1369-1373
DOI: 10.1126/science.1116447


We have explored the use of embryonic stem cells as an alternative to oocytes for reprogramming human somatic nuclei. Human embryonic stem (hES) cells were fused with human fibroblasts, resulting in hybrid cells that maintain a stable tetraploid DNA content and have morphology, growth rate, and antigen expression patterns characteristic of hES cells. Differentiation of hybrid cells in vitro and in vivo yielded cell types from each embryonic germ layer. Analysis of genome-wide transcriptional activity, reporter gene activation, allele-specific gene expression, and DNA methylation showed that the somatic genome was reprogrammed to an embryonic state. These results establish that hES cells can reprogram the transcriptional state of somatic nuclei and provide a system for investigating the underlying mechanisms.

The generation of embryonic stem (ES) cell lines and cloned animals by somatic cell nuclear transfer has demonstrated that the cytoplasm of an oocyte can reprogram the genome of a somatic cell to an embryonic state (1, 2). There is considerable interest in how reprogramming occurs, because a mechanistic understanding of the process might allow for the direct conversion of adult somatic cells into human ES (hES) cells and thus the production of genetically tailored cell lines for the study and treatment of human disease (37).

On the basis of previous experiments with murine ES cells (8, 9), we reasoned that hES cells might provide an alternative source of material for the reprogramming of human somatic nuclei. To investigate this, we used polyethylene glycol (PEG) to fuse hES cells with human BJ fibroblasts (Fig. 1A) (Materials and Methods). Because of the inefficient nature of cell-fusion (Materials and Methods), we transfected hES cells and retrovirally transduced BJ fibroblasts with independent drug-resistance markers to select for any rare hybrids that might be generated (Fig. 1A). The resulting hygromycin-resistant, green fluorescent protein (GFP)–positive hES cells (HUES6-GFP) and puromycin-resistant BJ fibroblasts were mixed in the presence of PEG and subjected to dual drug selection under conditions used for maintenance of hES cells (10). After 10 days of selection, 12 (± 3, N = 3) individual colonies of resistant cells were observed. Two typical colonies were picked, expanded, and enzymatically passaged several times before further analysis.

Fig. 1.

Generation of stable hybrid cells through the fusion of hES cells and human somatic cells. Existing somatic and hES cell lines were stably transduced or transfected with independent drug-resistant markers and induced to undergo cell fusion in the presence of PEG. Fused cells were then grown under standard conditions for the maintenance of hES cells in the presence of antibiotics to select for cell hybrids (A). FACS analysis of GFP expression in BJ fibroblasts (B), HUES6-GFP cells (C), and hybrid cells (D). PCR amplification of DNA sequences specific to the retrovirus used to transduce the somatic BJ fibroblasts (E). FACS of HUES6-GFP cells (F) and hybrid cells (G) stained with propidium iodide.

To confirm that these drug-resistant cells arose through cell fusion, we assayed genetic markers carried by the two fusion partners. Fluorescence-activated cell sorting (FACS) analysis demonstrated that, similar to HUES6-GFP cells (Fig. 1C), resistant cells expressed GFP (Fig. 1D). With the use of polymerase chain reaction (PCR), we determined that drug-resistant cells also contained the viral vector introduced into the BJ fibroblasts (Fig. 1E). Consistent with the notion that these cells arose by fusion, they contained twice the relative DNA content of the HUES6-GFP cells (Fig. 1F, 1G). Analysis of chromosome spreads from one of the hybrid cell lines indicated that it consisted predominantly of cells containing 92 chromosomes (fig. S1A). These results demonstrated that, after cell fusion of an hES cell and a somatic cell, stable cell hybrids were produced that contained both the somatic and hES cell chromosomes in a single nucleus (fig. S1B).

In the context of these stable hybrid cells, we addressed the question of whether the somatic genome had been reprogrammed to an embryonic state. If so, then hybrid cells should have a phenotype similar to the parental hES cells. Hybrid cells grew in tight, phase-bright clusters comparable in appearance to hES cells and unlike the spindle-shaped fibroblasts (11) (Fig. 2A). Analysis of the DNA content (Fig. 1G) of hybrid cells suggested that their cell-cycle characteristics were similar to hES cells (Fig. 1F). We passaged the hybrid cell lines more than 50 times (>120 population doublings), demonstrating they also share the immortal growth characteristics of hES cells. Furthermore, hybrid cells expressed several markers characteristic of hES cells, including the OCT4 transcription factor (Fig. 2B), alkaline phosphatase activity (fig. S2A), telomerase activity (fig. S3A), and the embryonic-specific antigens SSEA4 (fig. S2B), TRA1-61 (fig. S2C), and TRA1-80 (fig. S2D) at concentrations similar to those found in HUES6-GFP cells (1016). In contrast, the somatic BJ cells did not exhibit these characteristics.

Fig. 2.

Hybrid cells can assume an hES cell phenotype. Drug-resistant hybrid cells grew in compact, phase-bright cell clusters with a morphology identical to that of hES cells (A). HES cells and hybrid cells expressed the GFP (green) and the transcription factor OCT4 (red), whereas the GFP-negative BJ fibroblasts and mouse embryonic fibroblasts feeder cells did not (B). Immunostaining of teratomas derived from hybrid cells revealed the presence of various cell types, including neurons surrounding an ES-derived hair follicle that expressed a neural-specific tubulin (Tuj1, red) (note autofluorescent hair, green) (C), skeletal muscle expressing myosin heavy chain (MF20, red) (D), and intestinal endoderm expressing the alpha fetal protein (AFP, red) (E). Scale bars, 50 μm.

To determine whether the hybrid cells display the developmental pluripotency of hES cells, we assessed their ability to differentiate in vitro and in vivo (10, 11, 17). When cultured in suspension, both hybrid cell lines formed embryoid bodies (EBs), and they formed teratomas after injection into nude mice. Immunostaining showed that both a teratoma (Fig. 2, C to E) and EBs (fig. S3, B to D) contained cells expressing βIII-tubulin (neurectoderm) (18), muscle-specific myosin (mesoderm) (19), and alpha-fetoprotein (endoderm) (20).

To investigate whether transcription of embryonic genes was reactivated in the somatic chromosomes, we used a transgenic reporter in which GFP expression was dependent on promoter elements from the murine Rex-1 gene (21). Rex-1 is a retinoic acid–regulated zinc-finger protein expressed in ES cells [as well as preimplantation mouse embryos and spermatocytes (21)]. This reporter was active when transfected into HUES6 hES cells (Fig. 3A) but inactive after introduction into BJ fibroblasts (Fig. 3B) (21). When BJ cells carrying the reporter were fused to GFP-negative HUES6 hES cells (Fig. 3C), the resulting hybrids (Fig. 3D) expressed GFP at a concentration similar to that of HUES6 cells transfected with the same reporter.

Fig. 3.

Transcriptional and epigenetic reprogramming of the somatic genome in hybrid cells. To test for reactivation of embryonic genes from the somatic chromosomes, we used a Rex1-GFP reporter that is active when transfected into hES cells (A) but silent when transfected into BJ fibroblasts (B). When transgenic fibroblasts (B) were fused with nontransgenic HUES6 hES cells (C), the resulting cell hybrids expressed GFP (D). Scale bars in (A) to (D), 50 μm. To investigate whether transcription of endogenous embryonic genes could be reactivated, we performed reverse transcription PCR with primers specific to the TDGF1 gene (E). Analysis of DNA methylation at the OCT4 promoter demonstrated that the epigenetic state of the hybrid cells had also been reprogrammed (F). To determine whether the transcription of genes specifically expressed in somatic cells was extinguished after cell fusion, we performed genome-wide transcriptional profiling (G to I). Genes with expression values either twofold higher in BJ cells relative to hES cells (somatic-specific genes) or twofold lower in BJ cells relative to hES cells (hES-specific genes) were noted, and their expression analyzed in the hybrid cells. An Eisengram displaying “heat plots” from representative genes (red, on; blue, off) (G) and Venn diagrams (H and I) depicting the results of pairwise comparisons from transcriptional profiles of the three cell types demonstrate that the somatic program of transcriptional program is silenced in the hybrid cells, whereas the embryonic program predominated.

We demonstrated that endogenous genes regulating developmental pluripotency were transcribed from the somatic genome by carrying out allele-specific expression analysis. The CRIPTO/TDGF1 gene product functions as a component of the NODAL signaling pathway and is expressed in hES cells but not BJ fibroblasts (Fig. 3E and table S4). An expressed single nucleotide polymorphism specific to the BJ cells was identified in the 3′ untranslated region of the CRIPTO/TDFG1 transcript (fig. S3E) and assayed in hybrid cells. Three out of nine cDNAs isolated from hybrid cells contained the BJ-specific polymorphism, indicating that the somatic alleles of this gene were activated after cell fusion (fig. S3F).

To determine whether somatic and embryonic transcriptional states coexisted within the hybrid cells or whether the embryonic program predominated, we performed genome-wide transcriptional profiling. The variance in transcriptional profiles between each replicate and the various cell lines was calculated as a Pearson correlation coefficient (PCC) and assessed over 54,675 probe sets (table S1). The PCCs revealed that the transcriptional profiles for BJ fibroblasts and HUES6-GFP cells were very different [PCC = 0.780, range from 0.768 to 0.789, n = 9], whereas the transcriptional profiles of the HUES6-GFP ES cells and the two independent hybrid cell lines varied minimally (HUES6 versus Hybrid1, PCC = 0.985, range from 0.979 to 0.989, n = 9; HUES6 versus Hybrid2, PCC = 0.984, range from 0.978 to 0.989, n = 9). Indeed, the variance between the HUES6-GFP ES cell line and the two hybrid cell lines was less than that we and others observed between independently derived hES cell lines (HSF-6B versus HSF-1B, PCC = 0.978) (table S1) (22).

Pairwise comparisons of the male BJ cells and the female HUES6-GFP hES cells showed that 3867 transcripts had values of expression at least twofold higher in the BJ fibroblasts. Only 12 of these somatic-specific transcripts retained their higher level of expression in the cell-hybrids (Fig. 3, G and H) (table S2). Five of these 12 were linked to the Y chromosome, suggesting that their expression values reflected the acquisition of a Y chromosome from the BJ cells rather than failures in reprogramming (Fig. 3G and table S2). In fact, the Y-linked genes expressed in hybrid cells were also expressed at similar amounts in male hES cell lines (22, 23). Similarly, only 20 of 2521 hES-specific transcripts (two- or more-fold higher in hES than BJ) were expressed in the hybrids at lower amounts than in the hES cells (Fig. 3, G and I, and table S3), and a majority of these had expression values closer to those found in hES cells than those found in BJ cells. Of particular note, pluripotency genes including OCT4, NANOG, TDGF1, and REX1 (2427) were expressed at remarkably similar concentrations in the hES cells and the hybrid cells (table S4). Overall, these data show that after stable cell hybrid formation, somatic-specific genes were silenced across the entire genome, whereas the embryonic program of transcription predominated. Indeed, by this analysis, one can conclude that >99% of the transcripts analyzed are reprogrammed.

We next investigated whether epigenetic information underlying the transcription of pluripotency genes was reprogrammed by analyzing the status of a differentially methylated region in the promoter of the OCT4 gene (28). Consistent with previous results (28), sequencing of bisulfite-modified DNA showed that CpG dinucleotides in this region were generally methylated in somatic BJ cells and unmethylated in hES cells (Fig. 3F). Importantly, when we analyzed DNA from hybrid cells, this region was demethylated and indistinguishable from the epigenetic state found in hES cells (Fig. 3F).

Lastly, we generated stable cell hybrids by fusing another hES cell line, H9 (10) with TE76.T pelvic bone cells, demonstrating that the ability of hES cells to reprogram the somatic genome is not restricted to a particular hES cell or somatic cell line (See Supplemental Online Material).

In conclusion, these findings show that hES cells have the capacity to reprogram adult somatic cell chromosomes after cell fusion. HES cells may therefore provide a useful complement to human oocytes for biochemical and genetic studies aimed at understanding how to reprogram differentiated cells to an embryonic state and thereby increase their developmental potential. Eventually, this approach might lead to an alternative route for creating genetically tailored hES cell lines for use in the study and treatment of human disease. However, a substantial technical barrier remains before hES cells could be used for therapeutic purposes: specifically, the elimination of the ES cell chromosomes either before or after cell fusion (18). If hES cell enucleation can be performed without the loss of reprogramming activity, and/or if these fusion studies lead to an understanding of the factors needed for reprogramming, these approaches may circumvent some of the logistical and societal concerns surrounding somatic-cell nuclear transfer into human oocytes.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

Tables S1 to S4

References and Notes

View Abstract

Navigate This Article