All human primary fibroblast cells were generated in vitro after written informed consent using protocols approved by the Royal Free research ethics committee, Royal Free Hospital, London, UK. Fibroblasts were derived from a skin biopsy and cultured as described previously (3). We used PD1 (70 year old suffering from Parkinson disease) and RDP1 (49 year old with rapid onset Parkinsonism dystonia) as described previously (3). We also used BJ fibroblast cell line (a normal neonatal fibroblast cell line from ATCC).

We employed episomal plasmid method of reprogramming as described previously (2, 3) and used the following plasmids: pCXLE-hOct3/4-shp53 (Addgene number 27077), pCXLE-hSK (Addgene number 27078), pCXLE-hUL (Addgene number 27080), (the latter three referred to as reprogramming factors-(RF)), and pCXLE-EGFP (Addgene number 27082). For inhibition of Mbd3, we used pSMP-MBD3_1 (Addgene number 36371) (9). The Mbd3 isoform that we used for over-expression experiments was Mbd3b (8) which shares >80% homology to human Mbd3 (10) and is the most abundant isoform in embryonic stem cells (11). The outline of experiments is summarized in Fig. 1A. Briefly, cells were transfected using program U-023 on a Nucleofector (I) 2b device (Amaxa Nucleofector kit R) or Pulse code – EO150 (solution P2) on AMAXA 4D Nucleofector (LONZA). GFP transfection efficiency was measured using flow cytometry after a minimum of 24 hours of transfection. Established iPSC cell lines were maintained as described previously (3) and were passaged at 1 : 1~1 : 12 using gentle cell dissociation reagent (12). At day 25~30 post transfection, cells were either fixed for whole plate alkaline phosphatase staining as per manufacturer’s instructions or iPSC colonies were picked for expansion.

Immunoblotting was carried out as described previously (13). The blots were probed either with Mbd3 (ABCAM), α-tubulin (ABCAM), or GAPDH (Cell Signalling). Densitometry was performed using NIH image J software and a sampling window was used to measure the intensity of all bands on the blot and normalized against the band intensity of the internal control.

Immunofluorescence was carried on established iPSC cell lines as described previously (3). Primary antibodies namely TRA-1-60 and TRA-1-80 (Santa Cruz Biotechnologies) were used at a 1 : 100 dilution followed by goat anti-mouse IgG/IgM Alexa 488 as a secondary antibody at a concentration of 1 : 400. The cells were visualised using Axiovision microscope equipped with 10× objective. For established iPSC cell lines, RNA was harvested with TRIzol and 1μg of RNA was used for subsequent reverse transcriptase reactions with Superscript III first strand synthesis system. For qRT-PCR cDNA was diluted in 1 in 20 and were set up in duplicates using SYBR Green Master Mix. Delta Ct values with GAPDH were calculated and brought to power −2. Error bars represent±SEM of technical duplicates. Oligonucleotide primers details have been described elsewhere (3). To demonstrate spontaneous in vitro differentiation, iPSC were grown to confluency and maintained in suspension in EB media. The expression of trilineage marker genes were assayed after 14 days of differentiation (3). Statistical significance was determined by non-parametric Mann-Whitney U-test and in all cases p≤0.05 was considered to be significant (Supplementary Methods).

In order to assess the role of Mbd3 in establishment of pluripotency of human somatic cells, a plasmid construct expressing short-hairpin (shp) RNA targeted to human Mbd3 transcript (shpMbd3) was transfected together with Yamanaka episomal plasmids expressing human Oct3/4, shRNA against p53, Sox2, Klf4 and L-Myc/Lin28 (referred to as RF) in BJ foetal fibroblast cell line and two adult patient derived fibroblast lines; PD1 (70 years old with Parkinson’s disease) and RDP1 (49 years old with rapid onset of Parkinsonism dystonia) (Fig. 1A). GFP transfection efficiency of all three lines was comparable with mean (±SD) expression levels of 32.8±1.68%. Immunoblotting after 24~48 hours of transfection showed 50% knock-down of Mbd3 compared to sham (p<0.05) (Fig. 1B), but did not result in a difference in the proliferation rate of cells transfected with RF+shpMbd3 when compared with cells transfected with RF alone as seen morphologically. This suggests that partial knock-down of Mbd3 does not cause cell cycle arrest (8) of human fibroblasts, however, cell cycle analysis are required to further confirm this. iPSC colonies emerged at approximately the same time (days 12~20 post transfection) in the cells transfected with RF+shpMbd3 and RF alone cultures suggesting that Mbd3 does not alter the kinetics of reprogramming as reported previously (8). We used alkaline phosphatase activity as an early indicator of pluripotency prior to performing an extended assessment of pluripotent stem cell markers (14). Our results showed that the number of alkaline phosphatase positive colonies in cells transfected with RF+shpMbd3 was significantly reduced (p<0.05) by 5~33 fold in established BJ fibroblast cell line and in PD1 and RPD1 primary fibroblast cells compared to the numbers observed with cells transfected with RF alone (BJ mean (±SD): RF 200±52.9 and RF+shpMbd3 35±6.8; RPD1 mean (±SD): RF 24±4.2 and RF+shpMbd3 0±0; PD1 mean (±SD): RF 65±1.5 and RF+shpMbd3 2±1; (n=3) Fig. 1C, D). Interestingly, this effect was more profound in patient derived fibroblasts than in neonatal BJ fibroblast cell-line. The reason for this difference is unclear but could reflect the biological age of these lines and disease status of the patients. It was however, possible to expand iPSC colonies from cells treated with RF+ shpMbd3 although less efficiently and with delayed kinetics compared to colonies expanded from cells treated with RF only. Phenotypic characterisation of expanded iPSC colonies showed that they expressed TRA-1-60 and TRA-1-80 pluripotent markers with no difference in level of expression between cells treated with RF+shpMbd3 and RF only (Fig. 2A). Additionally, qRT-PCR analysis showed that iPSCs expanded from RF and RF+shpMbd3 expressed pluripotent genes such as Oct-4 and Nanog (Fig. 2B) at comparable levels. To investigate the role of NuRD in differentiation, iPSCs were maintained in suspension in EB media and under these conditions iPSCs formed aggregates called embryoid bodies which consists of a heterogeneous mix of derivatives of the three germ layers. iPSCs derived from RF+shpMbd3 failed to form EBs as previous reports (4). In contrast, EBs emerged from RF iPSCs with evidence of expression of trilineage markers such as Cdx2 (mesoderm), Pax6 (ectoderm) and MixI (endoderm) by qRT-PCR analysis (Fig. 2C).

We next looked at over-expression of Mbd3b, a major isoform expressed in pluripotent stem cells (8, 11) and confirmed by immunoblotting an approximately 50% significant increase in the levels of Mbd3b (Fig. 3A). An almost two-fold significant (p<0.05) increase in the number of alkaline phosphatase colonies was observed in cultures transfected with RF+Mbd3b when compared with RF alone (Fig. 3B) (BJ: RF 155±5 and RF+Mbd3b 290±10; RPD1: RF 30±1.5 and RF+Mbd3b 64±1.3; PD1: RF 70±2.0 and RF+Mbd3b 150±1.0). iPSC colonies emerged at approximately the same time in cells treated with RF+Mbd3b and in RF alone cultures and the overall proliferation kinetics of iPSC between these two conditions were similar (data not shown). iPSC clones from both these conditions expressed pluripotent markers (TRA-1-81 and TRA-1-60) with evidence of endogenous expression of Nanog, Oct 4 and Sox 2 by qRT-PCR (Fig. 3C, D) with no difference in level of expression. Importantly, iPSCs derived from fibroblasts transfected with RF+Mbd3b formed EBs with comparable kinetics to that observed with cells treated RF alone (Fig. 3E) and expressed markers indicative of the three germ layers (Fig. 3F, G).