arising from H. Obokata et al. Nature 505, 641–647 (2014) doi:10.1038/nature12968; retraction 511, 112 (2014) doi:10.1038/nature13598 ; and H. Obokata et al. Nature 505, 676–680 (2014) doi:10.1038/nature12969; retraction 511, 112 (2014) doi:10.1038/nature13599
Two reports claiming a novel cellular reprogramming phenomenon, stimulus-triggered acquisition of pluripotency (STAP), were published in Nature last year1, 2, but then subsequently retracted3, 4. The identity of STAP cells and STAP-derived stem cells, however, has remained undetermined. Here we report the results of a whole-genome sequencing (WGS) investigation of STAP-related samples kept mainly at the RIKEN Center for Developmental Biology. We show that all purported STAP stem-cell lines were contaminated with embryonic stem (ES) cells, and that chimaeric mice and teratomas supposedly derived from STAP cells instead show ES cell contribution.
The original article1 reported that exposure to low pH can reprogram differentiated cells into unique pluripotent cells (STAP cells), from which two secondary cell lines were established; ES-like STAP stem cells and trophoblast stem-like Fgf4-induced stem cells capable of generating placental cells1, 2. Because STAP cells were not maintained as frozen stocks, we first performed WGS of 15 genomic DNA samples in total, including three representative STAP stem-cell lines with different genetic backgrounds, an Fgf4-induced stem-cell line, and seven ES cell lines established at the Wakayama laboratory before or during the STAP study (Extended Data Table 1). We determined genome-wide patterns of single-nucleotide polymorphisms (SNPs) that distinguish mouse strains 129/Sv (129) and C57BL/6 (B6), as well as green fluorescent protein (GFP) transgene types (Supplementary Methods and Extended Data Fig. 1a). No samples from the Oct4-GFP Fgf4-induced stem cells described in the original letter2 were found (Oct4 is also known as Pou5f1).
The STAP stem-cell line FLS and the Fgf4-induced stem-cell line CTS were reported to carry a homozygous insertion of a single cag-gfp transgene with the genetic background of 129 female × B6 male (Extended Data Table 1). However, these cell lines had co-insertions of two GFP transgenes5, sperm-specific acrosin-promoter-gfp6 and ubiquitously expressed cag-gfp7 (hereafter designated Acr/cag-gfp) at chromosome 3, which originated from an Acr/cag-GFP B6 mouse strain8 not described in the STAP papers1, 2. These STAP cell lines were then compared with four ES cell lines—FES1, FES2, and two nuclear transfer ES lines (ntESG1 and ntESG2) (ref. 9)—established from crossing the Acr/cag-GFP mouse strain with 129 mice in the Wakayama laboratory in 2005 (Extended Data Fig. 1a and Extended Data Table 1). FES1 and FES2 cells shared homologous SNP patterns with these STAP cell lines over the entire genome, including the 129 X chromosome, while ntESG1 and ntESG2 cells bearing B6 X chromosome were excluded from the comparison. Furthermore, these STAP cell lines shared two genomic characteristics with FES1, but not FES2; first, two chromosomal deletions (Fig. 1a) are present only in FES1 and all Acr/cag-GFP STAP stem-cell sublines, but not in the other cell lines examined, in the paternal Acr/cag-GFP mice (frozen stock in 2010), or in potential maternal 129 substrains available in Japan. Second, FES1 and the STAP cell lines with Acr/cag-GFP share large SNP clusters that differ between FES1 and FES2 in three chromosomes (Fig. 1b and Extended Data Fig. 1b, c). These differential SNP clusters probably arose from chromosomal heterogeneity in the parental mouse colonies when FES1 and FES2 were established. It is highly unlikely that the Acr/cag-GFP STAP cell lines and FES1 all independently acquired these two unique deletions and inherited the same three mosaic chromosomes from parental mice. An ES cell stock, 129/GFP ES, was also found to share all these genomic features (Extended Data Table 1).
After the above three SNP clusters reflecting parental heterogeneity are excluded, the remaining 1,290 SNP alleles that distinguish FES1 and FES2 are supposed to have accumulated at or after establishment in 2005. Regarding these SNPs, STAP cell lines FLS3 and CTS1 and 129/GFP ES cells are nearly identical, but differ slightly from FES1 (at 30% of these alleles), suggesting that STAP cell lines FLS and CTS were derived from a sub-stock of FES1 ES cells.
The STAP stem-cell line GLS1–13 was reported as established from STAP cells prepared from genomic Oct4 fragments (GOF) mice (B6 background) carrying the Oct4-gfp transgene10 in 2012. All these cell lines have a large truncation with a terminal inverted repeat in one of two X chromosomes (Extended Data Fig. 2a). An identical X chromosome was found in GOF-ES, an ntES cell line established from GOF mice in 2011, but not in parental GOF mice. It is unlikely that such a peculiar X chromosome abnormality would occur independently, strongly suggesting that the GLS lines were derived from the GOF-ES.
SNP analysis revealed that two independent STAP stem-cell lines, AC129-1 and AC129-2, had a 129B6F1 genetic background, while they were documented in the original article1 as being established from 129 cag-GFP mice. We identified five heterozygous genomic anomalies: four deletions, and a duplication in these STAP stem-cell lines (Extended Data Fig. 2b, d), which were not found in the sequenced parental mouse genomes. We identified that these anomalies and sexual identity were shared by one of six control ES cell lines with cag-gfp, 129B6F1 ES1, established earlier than AC129. This is also the case for the other cag-GFP STAP stem-cell lines, FLS-T1 and T2, established in 2013. The 129B6F1 ES1 also shares a characteristic homozygous B6-SNP cluster in chromosome 6 with these four cag-GFP STAP stem-cell lines (Extended Data Fig. 2c, d). It is unlikely that the 129B6 ES1 line and these cag-GFP STAP stem-cell lines independently inherited all five chromosomal anomalies, the Y chromosome, and the same chromosome 6 from parental mice at establishment.
The article1 describes 2N chimaeric mice generated from STAP cells bearing cag-gfp on the 129B6F1 background and their germ-line transmission (Fig. 4 and Extended Data Fig. 7 in ref. 1) as evidence for pluripotency. We found nine genomic DNA samples for the offspring of STAP cell chimaeric mice (Extended Data Fig. 7c in ref. 1). These contained not only the Acr-gfp insertion but also the two deletions unique to FES1-derived Acr/cag-GFP cell lines described above (Fig. 1c) indicating that the cells transmitted to the germ line in the chimaeric mice were derived from FES1 ES cells.
The article1 also describes teratomas derived from Oct4-GFP STAP cells as evidence for pluripotency (Fig. 2 and Extended Data Fig. 4 in ref. 1). We found a glass slide specimen from which all these teratoma images were taken, and its corresponding paraffin block. Quantitative PCR of genomic DNA extracted from this paraffin block reproducibly indicated that these teratoma tissues formed from FES1-derived cells (Fig. 1d, e). Immunostaining revealed that intestinal epithelium tissue (Fig. 2e, right in ref. 1) and pancreatic tissue (Extended Data Fig. 4c in ref. 1), shown as teratomas from STAP cells1, were GFP-negative and, thus, of host mouse origin (Fig. 1f–h).
Control genomic DNA sequences for STAP cell chromatin immunoprecipitation sequencing (ChIP-seq) experiments (Fig. 4 in ref. 2) had been deposited in the NCBI database2. To gain sufficient sequencing coverage, we re-sequenced the genomic DNA prepared from the STAP cell lysate used for ChIP-seq (Extended Data Fig. 1a). We confirmed that this STAP cell sample shared all the genomic characteristics described above for 129B6F1 ES1 (Extended Data Fig. 2c), indicating that the STAP cell sample used for ChIP-seq was derived from 129B6F1 ES1 cells.
In summary, our investigations based on WGS of STAP-cell-related materials reveal that all of these materials are derived from previously established ES cell lines and refute the evidence shown in the two Nature papers1, 2 that cellular stress can reprogram differentiated cells into pluripotent cells. Data described here were presented to an external investigative committee convened by RIKEN. Raw data are available at the DDBJ sequence read archive (DRA) under accession number DRA002862.
- Stimulus-triggered fate conversion of somatic cells into pluripotency. Nature 505, 641–647 (2014) et al.
- Bidirectional developmental potential in reprogrammed cells with acquired pluripotency. Nature 505, 676–680 (2014) et al.
- Retraction: stimulus-triggered fate conversion of somatic cells into pluripotency. Nature 511, 112 (2014). et al.
- Retraction: bidirectional developmental potential in reprogrammed cells with acquired pluripotency. Nature 511, 112 (2014). et al.
- Quality control method for RNA-seq using single nucleotide polymorphism allele frequency. Genes Cells 19, 821–829 (2014)
- Real-time observation of acrosomal dispersal from mouse sperm using GFP as a marker protein. FEBS Lett. 449, 277–283 (1999) et al.
- ‘Green mice’ as a source of ubiquitous green cells. FEBS Lett. 407, 313–319 (1997) , , , &
- FISH analysis of 142 EGFP transgene integration sites into the mouse genome. Genomics 80, 564–574 (2002) et al.
- Generation of normal progeny by intracytoplasmic sperm injection following grafting of testicular tissue from cloned mice that died postnatally. Biol. Reprod. 73, 390–395 (2005) &
- Identification and characterization of stem cells in prepubertal spermatogenesis in mice. Dev. Biol. 258, 209–225 (2003) et al.
Extended data figures and tables
Extended Data Figures
- Extended Data Figure 1: Genome-wide SNP patterns of STAP-cell-related cells and mice. (786 KB)
a, SNP patterns of STAP stem cells, Fgf4-induced stem cells and related ES cells as revealed by WGS. Chromosomes 1–19 and X are aligned from left to right. All cell lines and mouse strains except for STAP stem cell GLS1, GOF-ES and GOF-mouse are male. 129/GFP ES cells and the re-sequenced control DNA of STAP cells for ChIP-seq (Fig. 4 in ref. 2) are also shown. B6-homozygous, B6/129-heterozygous and 129-homozygous SNPs are shown in magenta, green and blue, respectively. Note that ntESG1 and ntESG2 inherited the B6-type X chromosome from maternal mice. Genomic regions in which FES1 and FES2 ES cells have different SNP clusters in chromosomes (chromosomes 6, 11 and 12) are marked by red rectangles. See b, c and Fig. 1b for a high-resolution map. SNP resolution is 10 Mb. b, c, High-resolution view of chromosomes 6 (b) and 11 (c), which show differential SNP clusters (red rectangles) between FES1 and FES2. In these regions, all SNPs are 129-type in FES1, and B6-type in FES2 on one of the homologous chromosomes. Therefore, in these regions, 129/B6 SNPs (green) and 129/129 SNPs (blue) in FES1 correspond to B6/B6 SNPs (red) and B6/129s (green) in FES2. STAP stem-cell line FLS3, Fgf4-induced stem-cell line CTS1, and 129/GFP ES cells share the same SNP patterns with FES1. SNP resolution is 1 Mb.
- Extended Data Figure 2: STAP stem cells derived from GOF-ES cells and cag-GFP ES cells. (345 KB)
a, Copy number of chromosome X in STAP stem cells GLS1 and GOF-ES cells, both of which have Oct4-gfp transgenes with a B6 background. These lines have one very short X chromosome of ~23 Mb with a terminal inverted repeat and a normal X chromosome. b, PCR detection of chromosomal anomalies and Y chromosome in cag-GFP STAP stem-cell lines and parental mouse strains. Lanes 1–6: control ES cells, 129B6F1 ES1–6; 7: STAP stem cell AC129-1; 8: STAP stem cell AC129-2; 9: STAP stem cell FLS-T1; 10: STAP stem cell FLS-T2; and 11: GOF-ES. Deletions 1–4 and duplication 1 are located on Chr19: 32,857093–32866,121, Chr1:140,698,249–140,702,693, Chr4:123,747,239–123,763,596, Chr10:43,265,147–43,267,270 and Chr1: 180,730,393–180,732,937, respectively. c, Distribution of B6-type and 129-type SNPs along chromosome 6. The B6-homozygous SNP cluster (magenta) in the middle, which probably arose from the inheritance of the parental 129, is heterogeneous in length among six control ES cell lines. The four cag-GFP STAP stem-cell lines share the same length of the B6 SNP cluster with control ES 129B6F1 ES1. Note that the 129/B6-heterozygous SNP region in the 129 cag-GFP mouse is longer than that of AC129-1. d, Table summarizing the chromosomal anomalies and differential types of Chr6 B6-homozogous SNP clusters in the cag-GFP cell lines and parental mice. Control ES cell 129B6F1 ES1 shares all the characteristic features with the four cag-GFP STAP stem-cell lines.
Extended Data Tables
- Supplementary Information (188 KB)
This file contains Supplementary Methods and additional references.