Generation and characterization of human iPSC via integration-free minicircle vector
Although somatic cell nuclear transfer (SCNT) mediated generation of ESCs and animal cloning have achieved remarkable progress in animals and human, ethical issues remain as an obstacle in clinical applications. Patient-specific iPSCs have similar potential as ESCs for disease modeling, drug screening and regenerative medicine without much ethical concerns, providing a better source for personalized stem cell applications. However, majority of the strategies in iPSC derivation are based on viral integration which may lead to oncogenic mutations. To address the safety concerns, we employed a minicircle vector reported previously to generate human iPSC from primary cells (Additional file 1: Fig. S1a). Dermal skin from a healthy donor was used for fibroblast preparation. To minimize the impact of cellular senescence on reprogramming [27], we started the nucleofection with MIP 247 vector in fibroblasts within passage 10. The process of reprogramming is illustrated in (Additional file 1: Fig. S1a). At least six independent iPSCs clones were collected and expanded on either autologous feeder-layer or Matrigel (Fig. 1a). Alternatively, we also used renal epithelial cells collected from urine to derive iPSCs to avoid the invasive approach in biopsy (Additional file 1: Fig. S1c).
To characterize these iPSCs, genomic DNA were extracted to confirm the absence of minicircle plasmid insertion with four pairs of primers (Additional file 1: Fig. S1b). As shown in (Additional file 1: Fig. S1d), neither iPSC clones nor the H9 hESCs were positive for vector insertion. The iPSCs expressed pluripotency markers, such as GDF3, OCT3/4, NANOG, REX1, SOX2, and DNMT3b, at levels comparable to that in human ESC H9 (Fig. 1b). Bisulfite sequencing at OCT3/4 locus in the genomic DNA revealed a significant reduction in DNA methylation in iPSCs (Fig. 1c), indicating epigenetic memory has been erased during reprogramming. The pluripotency was also confirmed by immunostaining of SSEA4, OCT3/4, NANOG, TRA1-60, and Alkaline Phosphatase (AP) activities, as well as the capability of in vitro embryoid body (EB) formation (Fig. 1d). Genome integrity was shown by normal chromosome pairing (44 + XX) in karyotyping examination (Fig. 1e). In addition, when injected into immunocompromised mice, these iPSCs gave rise to three germ layers demonstrated by the hematoxylin and eosin (H&E) staining in the teratoma formation assay (Fig. 1f). Taken together, iPSCs generated from human primary cells are footprint free and clinically compliant, given that no animal components were used in the primary cell culture and reprogramming process.
Generation of Mesenchymal stem cells from human PSCs by a two-stage induction approach
MSCs can be induced from hPSCs through an intermediate stage, the NCCs [24,25,26, 28,29,30]. However, the efficiency of NCCs induction from hPSCs is relatively low. The use of flow cytometric selection for HNK1/p75++ to purify NCCs is not practical for large-scale production of MSCs with a possibility of introducing undesired contamination [25, 31]. Recently, Menendez et al. reported that over 85% pure NCCs can be generated directly from hPSCs without FACS sorting. However, the in-house made culture media for hPSCs and for NCCs induction are complicated. The entire process of NCCs induction takes more than 20 days [26] and two extra weeks are then required for transition from NCCs to MSCs. Given that 1) NCCs from neural ectoderm are easily transdifferentiated into MSCs with high efficiency; 2) non-NCC cells derived from neural ectoderm, such as neurons and astrocytes, are more hypersensitive to stress than MSCs, it is plausible that MSCs can be directly and rapidly induced from neural ectoderm cells without NCC purification.
To verify the hypothesis, we firstly induced hPSCs to neural ectoderm using modified medium composed of 48.5% of Neurobasal medium, 48.5% of DMEM/F12 basal medium, 1% of N2 and 2% of B27, 10 μM of SB431542 (TGF-β pathway inhibitor) and 3 μM of ChIR99021 (WNT signal activator). After induction for 5 days, cells exhibited neural-like morphology, spreading out and separating from each other, though small patches of hPSCs-like clusters may still be observed (Fig. 2a). The qPCR analysis revealed that OCT3/4, NANOG, and REX1 were dramatically decreased while neural ectoderm-associated genes including PAX3, ZIC1, SOX10, SOX9, AP2a, FOXD3 and p75, were upregulated (Fig. 2b). In line with this, immunostaining showed loss of pluripotent markers OCT3/4, NANOG and AP2α and expression of PAX7 and typical neural progenitor/stem cell marker NESTIN (Additional file 1: Fig. S2c). The induced neural ectoderm cells were capable of suspension culture (Additional file 1: Fig. S2a). p75 has been identified as one of the MSCs markers in vivo. [32] Ninety-nine percent of neural ectoderm cells were found positive for p75 (Additional file 1: Fig. S2b). However, the percentage of p75high was lower compared with previous study. [26] Upon culture for two weeks, these neural ectoderm cells were able to be differentiated into adipocytes, chondrocytes, and osteocytes in corresponding induction media (Additional file 1: Fig. S2d). Collectively, these data suggest that these neural ectoderm cells are not pure NCCs, rather populations manifesting MSCs-like properties.
To induce MSCs from neural ectoderm cells, we switched the culture medium directly to MSC medium containing 10% of serum supplemented with 2 ng/ml of bFGF and 2 ng/ml of EGF. To generate clinical compliant MSCs, human umbilical cord blood serum (hUCBS) was used in the medium instead of FBS during the induction of MSCs. Spindle-like morphology was initially observed on day 4 after medium switch. Cells were split at a ratio 1:1 or 1:2 when reached full confluency. Moderate cell death was observed at the first two passages. After three passages, the cells became homogenous with enhanced proliferation rate. Flow cytometry analyses showed over 90% of the cells were positive for typical MSC surface markers (CD73 99.9%, CD90 99.5%, CD105 91.6% and CD44 95.8%) and negative for CD45, CD34, CD11b, CD19 and HDL-DR (< 1%) (Fig. 2c). To examine if these MSCs retain full differentiation potentials, MSCs at passage 5 were induced for adipogenesis, chondrogenesis and osteogenesis. As shown in Fig. 2e, differentiated cells were stained positively by Oil red O, Alcian blue, and Alizarin red (Fig. 2d). Colony formation assay showed a strong self-renewal capacity of the neural ectoderm cells-derived MSCs (Fig. 2e).
hPSCs-derived MSCs exhibit similar properties and gene expression profile as bone marrow-derived MSCs
Differences have been reported in the efficiency of MSC induction between human ESCs and iPSC. It is in general easier for ESCs than iPSCs to differentiate into MSCs as the later require longer time and specific stress adaptation [21, 22]. To investigate whether the reduced efficiency of differentiation in iPSCs is a result of the clonal variation of iPSC quality, we derived MSCs from the second human iPSCs clone (WT-MSCs) and human ESCs H1 (H1-MSCs). Interestingly, the iPSCs and ESCs showed comparable differentiation efficiencies in our hands and similar MSCs surface markers including CD73, CD90, and CD44, with an exception for CD105 exhibiting higher percentage in iPSC-derived WT-MSCs (84%) than in H1-MSCs (43%) (Fig. 3a). The variable percentage of CD105 is in line with the previously reported NCCs-induced MSCs [33, 34]. Overall, these results demonstrated that MSCs could be robustly and efficiently generated within a substantially shorter time from both human ESCs and iPSCs by our protocol.
To explore the similarity and difference between hPSCs-derived MSCs and mature MSCs directly isolated from tissues, we took primary bone marrow MSCs (BM-MSCs) for comparison. BM-MSCs exhibited high percentage of surface markers CD73, CD90, CD105 and CD44. However, a sub-population of these BM-MSCs were also positive for CD34, CD11b, CD19, CD45 and HLA-DR cocktails (Fig. 3a). When differentiation capabilities were examined, adipocyte, chondrocytes and osteoblasts were all successfully induced by BM-MSC, ESCs-MSCs and iPSCs-MSCs verified by staining of Oil Red O, Alcian Blue and Alizarin Red S (Fig. 3b). Of note, the oil droplets in BM-MSCs was found significantly larger than that observed in H1-MSCs or WT-MSCs (Fig. 3b), which is consistent with the previous reports [22,23,24]. To further compare these MSCs generated from different sources, gene expression profiles were analyzed. As shown in Fig. 3c, a high correlation co-efficient (R2) was observed between BM-MSCs and either WT-MSCs (90.6%) or H1-MSCs (89.0%), suggesting that both hPSCs-derived MSCs and ESCs-derived MSCs are highly similar to primary BM-MSCs. The hPSCs-derived MSCs appeared to be more homogenous with 91.6% similarity (H1-MSCs vs WT-MSCs) (Fig. 3c). While BM-MSCs went cell cycle arrest after passage 5, hESCs derived MSCs did not show cellular senescence before passage 20.
Generation and characterization of iPSCs from a HGPS patient
The human progeroid syndrome Hutchinson-Gilford progeria syndrome (HGPS) is predominantly caused by an autosomal dominant mutation (p.G608G, c.1824C. > T) in exon 11 of LMNA gene, resulting in the exposure of a cryptic splicing site and generation of a truncated mutant lamin A isoform, termed progerin. Progerin is an immature truncated lamin A missing 50 amino acids covering the second cleavage site. It is permanently farnesylated, therefore tightly attaches to the nuclear envelope, leading to the deformation of nuclear architecture, genome instability and epigenetic alternations [35]. Progerin expression is also observed in the elderlies [36] and ectopic expression of progerin in normal cells recapitulates accelerated cellular senescence in vitro and results in the collapse of tissue homeostasis in vivo. [37,38,39] Interestingly, neural cells of HGPS patients are unaffected, whereas mesoderm linages, especially mesenchymal cells, are severely affected [40,41,42], implicating HGPS-MSCs as an ideal cell model for stem cell aging and drug screening.
We firstly generated iPSCs using dermal fibroblasts derived from a HGPS patient. The expression of progerin were confirmed in HGPS fibroblasts by Western blotting using antibodies against lamin A/C (Fig. 4d). To generate iPSC cell lines, HGSP fibroblasts at passage 5 were reprogrammed with a minicircle vector expressing 4 Yamanaka factors. Valproic acid and ascorbic acid were supplemented in the culture medium to improve the reprogramming efficiency. Among six HGPS-iPSC clones, two were chosen for further characterization. No detectable integration of vector fragments into iPSC genome was observed in both clones (Additional file 1: Fig. S3a).
Similar to H9 ESCs, both HGPS-iPSCs clones expressed pluripotency associated markers, such as GDF3, NANOG, OCT3/4, SOX2, hTERT and DNMT3b at mRNA level (Additional file 1: Fig. S3b, c). Immunostaining and immunochemistry confirmed the expression of NANOG, SSEA4, OCT3/4, TRA1-60, and AP activity in HGPS-iPSCs. In addition, reprogrammed HGPS-iPSCs can form EB (Fig. 4a). Consistent with previous reports, the transcription of lamin A, lamin C or progerin were silenced whereas lamin B1 mRNA was upregulated significantly in HGPS-iPSCs, compared with the parental fibroblasts (Additional file 1: Fig. S3d). Bisulfite sequencing revealed that the endogenous OCT3/4 promoter region was in an open status in HGPS-iPSCs (Fig. 4b). Karyotyping and in vivo teratoma formation assay showed that HGPS-iPSCs retained the genomic integrity (44 + XY) with full differentiation potential (Additional file 1: Fig. S3e, f). HGPS cells, including fibroblasts, and smooth muscle cells (SMCs) and MSCs manifested disrupted nuclear architecture and abnormal epigenetic modifications. The absence of abnormal nuclear blabbing (Fig. 4c) and lamin A/C expression as well as the increased expression of lamin B1 and HDAC1 indicated that defects in the nucleus and epigenetics were reset in HGPS-iPSCs (Fig. 4d) and reprogramming rejuvenated the premature aging in HGPS cells.
Recapitulation of premature aging in HGPS-iPSCs derived MSCs (HGPS-MSCs)
HGPS-iPSCs were further induced to generate MSCs using our aforementioned protocol. The differentiation of both WT-iPSCs and HGPS-iPSCs were performed in parallel. At passage 3, the MSCs differentiated from both WT-iPSCs and HGPS-iPSCs exhibited typical MSC surface markers with over 95% of cells positive for CD73, CD90, CD105 and CD44. Meanwhile, the MSCs negative markers were extremely low, with less than 0.1% of cells positive for CD11b, CD20, CD34, CD45 and HLA-DR (Additional file 1: Fig. S4a). These MSCs were able to further differentiate into adipocytes, osteocytes and chondrocytes (Additional file 1: Fig. S4b).
Further characterization of HGPS-MSCs by double staining of lamin B2 and progerin revealed the presence of progerin and nuclear abnormality. About half of the HGPS-MSCs (47.8 ± 10.3%) exhibited detectable progerin expression. Significant higher percentage of cells manifested nuclear blabbing (78.1 ± 12.7% in HGPS-MSCs Vs 21.1 ± 6.8% in WT-MSCs) (Fig. 4e). HGPS-MSCs exhibited significantly reduced proliferation (57.3 ± 5.8% in WT-MSCs Vs 7.9 ± 1.6% in HGPS-MSCs) (Fig. 4f) and remarkably increased senescence (86.7 ± 5.6% in HGPS-MSCs Vs 7.5 ± 4.3% WT-MSCs) and increased DNA damage (Additional file 1: Fig. S4c). In line with these observations and our previous findings (Liu et al. 2005), the DNA-damage checkpoint response in HGPS-MSCs was defective upon 10 Gy of γ-irradiation, as indicated by the delayed recruitment of 53BP1 (Additional file 1: Fig. S5). As expected, HGPS-MSCs manifested senescence associated secretory phenotypes (SASP) with dramatic upregulation of IL1a, IL1b, IL6, IL8 and PAI (Fig. 4h). In addition, significant reduction in heterochromatin markers, such as HP1a, H3K9me3 and H3K27me3 were observed in HGPS-MSCs (Additional file 1: Fig. S6). Collectively, these results demonstrated that HGPS-MSCs derived by our protocol recapitulate the phenotypes observed in HGPS cells and can serve as a cell model for laminopathy-based premature aging.
Correction of pathogenic mutation in HGPS cells by a CRISPR/Cas9-based double selection system
As MSCs transplantation can ameliorate aging associated disorders and extend lifespan [43], it is plausible MSCs will serve as a novel therapeutic strategy for HGPS in addition to other approaches like targeting lamin A post translational process, modulating progeria or prelamin A level, activation of autophagy and reprogramming [44,45,46]. To fulfil this purpose and avoid potential undesired immune response, we generated genetically rectified autologous MSCs. The homologous recombination (HR) strategy was adopted to correct HGPS mutation in iPSCs derived from the patient. Earlier HR methods mainly employ a single antibiotic resistant gene cassette to enrich positive cells, which requires an additional round of screening to remove DNA fragment. To accelerate this process, we designed a double selection donor vector containing antibiotic resistant gene and fluorescent protein mCherry gene cassette flanking with a 3.5 kb and 1.7 kb homologous arm, respectively (Fig. 5a). To correct HGPS-iPSC, the guide RNA out of gene body region was designed and a specificity-enhanced Cas9 was used to minimize the undesired side-effects caused by gene editing. After nucleofection and puromycin selection, colonies were collected and expanded to screen for HGPS-iPSC clones with LMNA gene correction by Sanger sequencing (Fig. 5b). The mutation-corrected HGPS-iPSCs clone (HGPS-CiPSCs) was further examined for its pluripotency. As shown in Fig. 5c, the HGPS-CiPSCs were positive for SSEA4, OCT3/4 and TRA1-60. Karyotyping assay confirmed the genome integrity in HGPS-CiPSCs (Fig. 5d). We then generated MSCs from HGPS-CiPSCs. As shown in Fig. 5e, surface markers confirmed HGPS-CiPSCs-derived MSCs met the criteria of MSCs (Fig. 5e). HGPS-CiPSCs-derived MSCs showed a significant downregulation of p16 and a dramatic decline in SASP-associated genes, compared to their counterpart HGPS-MSCs (Fig. 5f). To examine the fate of these MSCs in vivo, we labelled both HGPS-MSCs-derived and HGPS-CiPSC-derived MSCs with luciferase via lentiviral infection. Equal number (106 cells) of HGPS-MSCs-derived and HGPS-CiPSC- derived MSCs were inoculated into immunocompromised mice at the middle parts of left and right tibialis anterior (TA) muscle, respectively. Luminescence was measured every other day. At day 7, accelerated decay of luminescence was observed in the left leg at the engrafted site (Fig. 5g), indicating HGPS-CiPSC derived MSCs have significant survival capability in vivo. These data collectively demonstrated rejuvenation of premature senescence after mutation correction.