Disclosure of potential conflicts of interest is found at the end of this article.
Author contributions: W.J. and X.S.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; D.Z.: collection and assembly of data, data analysis and interpretation; M.L.: collection and assembly of data; M.D.: data analysis and interpretation; Y.S.: manuscript writing, administrative support, final approval of manuscript; H.D.: manuscript writing, financial support, administrative support, final approval of manuscript. W.J. and X.S. contributed equally to this article.
First published online in STEM CELLSEXPRESS February 3, 2011.
Human ESCs provide a promising cell resource for the treatment of type I diabetes mellitus. Although PDX1-positive pancreatic progenitors can be efficiently generated from human ESCs by stepwise induction, further in vitro differentiation into functional, mature beta cells is not efficient or reproducible. Purification of pancreatic progenitor cells could facilitate the identification of signals that regulate beta cell differentiation and maturation. Here, we report the identification of a novel surface marker for PDX1-positive pancreatic progenitors based on an in vitro human ESC differentiation system. By costaining PDX1 and a panel of cell surface antigens at the pancreatic progenitor stage of human ESC differentiation, we discovered a positive marker, CD24. CD24-positive cells coexpressed most of the key transcription factors of pancreatic progenitors, and the expression of important pancreatic genes was greatly enriched in CD24-positive cells compared with the CD24-negative cells. In addition, CD24-positive cells could differentiate into insulin-producing cells but CD24-negative cells could not. These results indicate that CD24 could be a surface marker for PDX1-positive pancreatic progenitors derived from human ESCs. Enrichment of pancreatic progenitors with this marker will facilitate the investigation of beta cell maturation during the human ESC differentiation. STEM Cells 2011;29:609–617
Human embryonic stem (ES) cells represent a promising source of pancreatic beta cells for treating type I diabetes mellitus and an alternative in vitro model for investigating human pancreatic development . Significant progresses have been made in directing human ESCs to differentiate into pancreatic progenitors and insulin-producing beta cells by stepwise induction [2–8]. However, although pancreatic progenitors can be efficiently generated from human ESCs, efficient in vitro differentiation into mature functional beta cells that are able to reverse diabetes in animal models has not been achieved. The presence of mixed cell populations and cell-cell crosstalk during ESC differentiation makes the investigation of signals involved in the generation of functional insulin-producing beta cells challenging. The purification of pancreatic progenitor cells could facilitate the identification of signals that regulate beta cell differentiation and maturation [1, 9]. Therefore, a surface marker is urgently needed to purify pancreatic progenitors from differentiated human ESCs.
Several reports have identified surface markers of pancreatic progenitor/stem cells [10–13]. Suzuki et al. used c-Met, the receptor for hepatocyte growth factor, to sort progenitor/stem cells that exhibited high proliferative capacity and multiple differentiation potential in adult mouse pancreas . CD133 has been used to isolate embryonic pancreatic progenitors that can generate exocrine, endocrine, and duct cells from neonatal and adult mice . Combined with the negative expression of platelet-derived growth factor receptor beta, CD133 was shown to be a surface marker of pancreatic progenitors in the fetal mice . In another report, CD133+CD49flow marked an endocrine progenitor cell population in the E15.5 mouse fetal pancreas, and this expression pattern was proposed to be conserved in human fetal pancreas . Despite this report, the search for surface markers of human pancreatic progenitors is limited by the unavailability of human fetal material. Because of the potential differences between human and rodent pancreatic development , it is yet to be determined whether these markers can be applied to the isolation of human pancreatic progenitor cells. To our knowledge, no surface marker in human ESC-derived pancreatic cell lineages has previously been reported.
Here, we report that CD24 is a surface marker for PDX1-positive pancreatic progenitors derived from human ESCs. By costaining a panel of surface antigens with PDX1 at the pancreatic progenitor stage of human ESC differentiation, we discovered a positive marker CD24. CD24-positive cells coexpressed most of the key transcription factors of pancreatic progenitor cells, and only CD24-positive cells could differentiate into insulin-producing cells, whereas CD24-negative cells could not.
MATERIALS AND METHODS
Human ESCs Culture and Differentiation
Human ESC lines, H1 and H9, were maintained as reported by WiCell. Briefly, human ESCs were cultured in Dulbecco's modified Eagle's medium/F12 (Invitrogen, Carlsbad, CA, www.invitrogen.com) with 20% KnockOut serum replacement (Invitrogen) containing 10 ng/ml of basic fibroblast growth factor (PeproTech, Rocky Hill, NJ, www.peprotech.com) on mitomycin-C (Roche, Mannheim, Germany, www.roche.com)–treated mouse embryonic fibroblast (MEF) feeder cells.
We performed the pancreatic differentiation according to our previous protocol  with minor modifications. Human ESCs cultured on Matrigel (BD Bioscience, Franklin Lakes, NJ, www.bdbiosciences.com) were initiated to differentiate with 100 ng/ml activin A (PeproTech) and 0.5 μM wortmannin (Sigma-Aldrich, St. Louis, MO, www.sigmaaldrich.com) for 4 days, followed by 2 μM retinoic acid (RA; Sigma-Aldrich), 20 ng/ml fibroblast growth factor-7 (FGF7; PeproTech), and 100 ng/ml NOGGIN (PeproTech) for 6 days, and followed by 50 ng/ml epidermal growth factor (EGF; PeproTech) for 5 days in respective medium described previously .
Cell sorting was performed at days 13–15 of differentiation. The CD24-PE (BD Bioscience)-labeled cells were sorted by a Cytomation MoFlo high-performance cell sorter (Beckman Coulter, Miami, FL, www.coulterflow.com). The sorted cells were plated on Matrigel at a density of 1 × 105 cells per centimeter square, and cultured with 50 ng/ml EGF for 1 week. For further differentiation assay, sorted cells were treated with 10 mM nicotinomide (Sigma-Aldrich) and 50 ng/ml Extendin-4 (Sigma-Aldrich) after EGF treatment for 1 week.
All of the data were generated using the H9 human ESC line if not specifically referred. The differentiation experiments were repeated three or more times with similar results.
The sample cells were fixed in 4% paraformaldehyde and blocked with 5% donkey serum and 0.2% tritonX-100 in phosphate-buffered saline, and then incubated with primary antibody overnight at 4oC, and further incubated with secondary antibody (FITC or TRITC or Cy5-conjugated donkey anti-rabbit or anti-goat or anti-mouse or anti-rat or anti-guinea pig IgG, Jackson ImmunoResearch, West Grove, PA, www.jacksonimmuno.com). No primary antibody added was used as control. The information of used primary antibodies was listed as follows: goat anti-PDX1 (Abcam, Cambridge, MA, www.abcam.com, 1:3,000), rabbit anti-PDX1 (Abcam, 1:1,000), goat anti-SOX17 (R&D Systems, Minneapolis, MN, www.rndsystems.com, 1:500), goat anti-FOXA2 (R&D Systems, 1:400), goat anti-HNF1B (Santa Cruz Biotechnology, Santa Cruz, CA, www.scbt.com, 1:200), rabbit anti-SOX9 (Santa Cruz Biotechnology, 1:200), rabbit anti-HNF6 (Santa Cruz Biotechnology, 1:200), rabbit anti-PROX1 (Abcam, 1:200), mouse anti-CDX2 (Biogenex, Fremont, CA, www.biogenex.com, 1:200), rabbit anti-alpha-fetoprotein (anti-AFP; Zhongshan, Beijing, China, www.zsbio.com, 1:200), rabbit anti-CD24 (FL-80, Santa Cruz Biotechnology, 1:200), mouse anti-CD24 (SN3, Santa Cruz Biotechnology, 1:200), mouse anti-CD24 (ML5, BD Bioscience), goat anti-VE-cadherin (Santa Cruz Biotechnology, 1:200), rabbit anti-CD133 (Santa Cruz Biotechnology, 1:200), mouse anti-E-cadherin (Abcam, 1:200), rabbit anti-PDGFRα (Abcam, 1:100), mouse anti-N-cadherin (BD Bioscience, 1:200), mouse anti-neural cell adhesion molecule (anti-NCAM; Santa Cruz Biotechnology, 1:200), rat anti-homing-associated cell adhesion molecule (anti-HCAM; Santa Cruz Biotechnology, 1:200), mouse anti-NKX6-1 (Developmental Studies Hybridoma Bank, Iowa City, IA, dshb.biology.uiowa.edu, 1:200), guinea pig anti-Insulin (DAKO, Glostrup, Denmark, www.dako.com, 1:200), rabbit anti-OCT4 (Abcam, 1:200), and rabbit anti-NANOG (R&D Systems, 1:50). Images were captured using an Olympus IX-71 phase-contrast fluorescent microscope (Olympus, Tokyo, Japan, www.olympus. com) or LEICA confocal microscope TCS-SP2 (Leica, Wetzlar, Germany, www.leica-microsystems.com) .
Reverse Transcription Polymerase Chain Reaction and Quantitative Polymerase Chain Reaction Analysis of Gene Expression
Reverse-transcription polymerase chain reaction (RT-PCR) was performed as previously described . Before the reverse transcription, the RNA was digested by RNase-free DNase I (Ambion, Austin, TX, www.ambion.com) to remove genomic DNA contamination. The primer sequences and the length of the product were: CD24 (124bp), CTCCATTCCACAATCCCATC and GAAGGAGAGGCAACATCCAA; GAPDH (303bp), GCCAC ATCGCTCAGACACC and GTACTCAGCGGCCAGCATCG; INS (337bp), GAGGCCATCAAGCACCATCAC and GGCT GCGTCTAGTTGCAGTA; NEUROD1 (588bp), GAACGCAG AGGAGGACTCAC and GTGGAAGACATGGGAGCTGT; PDX1 (262bp), CCCATGGATGAAGTCTACC and GTCCT CCTCCTTTTTCCAC. The PCR products were sequenced. Quantitative PCR analysis was performed on ABI PRISM 7300 Sequence Detection System (Applied Biosystems, Carlsbad, CA, www.appliedbiosystems.com) using the SYBR Green PCR Master Mix (TOYOBO, Japan, www.toyobobiologics.com). The expression level of each gene at every checkpoint was normalized. For each sample, at least three independent experiments were performed. The primers sequences and the length of the product were: CD24 (124bp), CTCCATTCCACAATCCCATC and GAAGGAGAGGCAAC ATCCAA; FOXA2 (104bp), CTGAGCGAGATCTACCAGT GGA and CAGTCGTTGAAGGAGAGCGAGT; GAPDH (87bp), TGCACCACCAACTGCTTAGC and GGCATGGAC TGTGGTCATGAG; HES1 (208bp), AGCACACTTGG GTCTGTGC and TGAAGAAAGATAGCTCGCGG; HNF6 (131bp), TGTGGAAGTGGCTGCAGGA and TGTGAAGAC CAACCTGGGCT; OCT4 (103bp), CCGAAAGAGAAAGCG AACCAG and ATGTGGCTGATCTGCTGCAGT; PDX1 (101bp), GGTGGAGCTGGCTGTCATGT and CGCGCTTCTTGT CCTCCTC; PROX1 (102bp), CAATTTCCACACCGC CAAC and TCAGTGGAACTGGCCATCTG; SOX17 (103bp), GCATGACTC CGGTGTGAATCT and TCACACGTCAGGA TAGTTGCAGT.
Flow Cytometric Analysis
Single-cell suspensions of differentiating human ESC cultures were obtained by dissociating cells with 0.25% trypsin. Intracellular antibody staining was performed using Cytofix/Cytoperm and Perm/Wash buffer (BD Bioscience) according to manufacturer's instructions. PE-conjugated mouse anti-CD24 (clone ML5, BD Bioscience) was used in direct labeling. Mouse anti-NCAM (Santa Cruz Biotechnology), goat anti-PDX1 (R&D Systems), rabbit anti-Ki67 (Zhongshan), mouse anti-CD24 (clone SN3, Santa Cruz Biotechnology), and Cy5- or PE-conjugated donkey anti-goat or anti-rabbit or anti-mouse IgG were used for indirect labeling. Isotype added as primary antibody was used as control. Flow cytometry data were acquired on a FACSCalibur (Becton Dickinson, Franklin Lakes, NJ, www.bd.com) and analyzed using CellQuest Software (Becton Dickinson). For each sample, at least three independent experiments were performed.
Screen for a Surface Marker of Human ESC-Derived PDX1-Positive Cells
In the previous work , we characterized a pancreatic progenitor population in differentiated human ESCs that coexpressed a network of key transcription factors, resembling the pancreatic progenitor during in vivo development . Here, we extended this investigation and found that most of PDX1-positive cells also expressed HNF6, SOX9, and PROX1 (supporting information Fig. 1A). We also examined the expression of nonpancreatic markers, including the endodermal marker SOX17, the duodenal and intestinal marker CDX2 and the hepatic marker AFP. Only a few cells expressed SOX17, and the PDX1/HNF6-positive cells were SOX17-negative (supporting information Fig. 1B), suggesting that the obtained pancreatic progenitors were fully differentiated definitive endoderm cells not expressing SOX17 . PDX1-positive cells did not express CDX2 or AFP (supporting information Fig. 1C), which further confirmed the pancreatic lineage of the PDX1-positive cells. Collectively, these results indicate that the human ESC-derived pancreatic progenitors share similar characteristics with bona fide progenitors during in vivo pancreatic development .
To identify a specific surface marker that could be used to purify human ESC-derived pancreatic progenitors, we screened a panel of surface antigens by costaining the commercial antibodies with PDX1 at the pancreatic progenitor stage of differentiation (several representative images are shown in Fig. 1A). We observed that the staining pattern of PDX1 matched that of CD24; in contrast, PDX1 was not observed in CD24-negative cells (data not shown). CD133, which has been reported to be a marker for pancreatic ductal or islet progenitors, and E-cadherin, an epithelial cell marker, exhibited a much broader expression pattern than PDX1. In addition, the expression of NCAM did not overlap with that of PDX1. These data suggest that CD24 is a candidate positive marker of PDX1-positive cells and that NCAM could possibly be used as a negative marker.
Identification of CD24 As a Positive Marker for Human ESC-Derived PDX1-Positive Cells
To confirm the coexpression of CD24 with PDX1, we performed flow cytometric analysis at pancreatic progenitor stage of differentiation. The data showed two major, distinct populations according to the expression of CD24 and PDX1 in the differentiated human ESCs, that is, approximately 47% double-positive cells and 42.6% double-negative cells (Fig. 1B). Next, we sorted the CD24-positive and CD24-negative cells and compared the relative gene expression of PDX1 using quantitative RT-PCR. The CD24-positive population exhibited, on average, a 26-fold and 2.5-fold enrichment of transcriptional expression of PDX1 compared with the CD24-negative population and unsorted cells, respectively (Fig. 1C), which was consistent with the ratios of CD24 and PDX1 in the differentiated culture (Fig. 1B). We also confirmed the exclusive expression of NCAM with PDX1 by flow cytometry (Fig. 1D) and found that CD24-positive cells were NCAM-negative (Fig. 1E and data not shown). Thus, we further characterized the CD24-positive (also CD24-positive NCAM-negative) cells.
We next used additional assays to confirm the coexpression of CD24 with PDX1. Kristiansen et al. reported that different antibodies of CD24 had different staining patterns . Therefore, we tested two other commercial CD24 antibodies, and all of these three antibodies costained with PDX1 (supporting information Fig. 2A). Considering the diversity of different human ESC lines , we examined the coexpression of PDX1 and CD24 in two human ESC lines, H1 and H9, at pancreatic progenitor stage of differentiation. Coexpression of PDX1 and CD24 was observed in both cell lines (supporting information Fig. 2B). We also confirmed the coexpression of PDX1 and CD24 in pancreatic progenitor cells generated in two differentiation conditions, initiated on either Matrigel or MEF feeder cells (supporting information Fig. 2B).
CD24-Positive Cells Express Key Transcription Factors of Pancreatic Progenitor Cells
To determine whether the CD24-positive cell population represented human ESC-derived pancreatic progenitors, we examined the gene expression of CD24-positive cells. Because pancreatic progenitors express a network of several key transcription factors , we performed immunofluorescence assays of CD24 and key transcription factors. CD24-positive cells expressed not only PDX1 but also other key transcription factors of pancreatic progenitors, including SOX9, HNF6, HNF1B, HB9, and NKX6-1 (Fig. 2A). We next sorted CD24-positive and CD24-negative cells and compared their gene expression. Consistent with the immunostaining data, the expression of several key transcription factors, including PDX1, FOXA2, HNF6, HES1, and PROX1, was greatly enriched in the CD24-positive cell population (Fig. 2B). These data show that CD24-positive cells express key transcription factors of pancreatic progenitors. To evaluate the proliferation potential of CD24-positive and CD24-negative cells, we performed flow cytometry of CD24 and the mitosis marker, Ki67. About 35% of the CD24-positive cells were Ki67-positive (Fig. 2C), suggesting a certain proliferation capacity of CD24-positive cells.
CD24-Positive Cells Differentiated into Insulin-Producing Cells, Whereas CD24-Negative Cells Did Not
Based on fluorescence-activated cell sorting (FACS) techniques, the CD24-positive and CD24-negative cells were sorted at the pancreatic progenitor stage of differentiation and cultured with EGF for 1 week. The CD24-positive cells formed clones and maintained the expression of CD24 and PDX1, whereas the CD24-negative cells formed few clones and did not express CD24 or PDX1 (Fig. 3A, 3B). A further differentiation assay was performed to investigate the differentiation potential of the sorted cells. Only the CD24-positive cells differentiated into insulin-producing cells, whereas the CD24-negative cells did not (Fig. 3C, 3D). The differentiated CD24-positive cells also expressed the beta cell transcription factors PDX1, NKX6-1, and NEUROD1 (Fig. 3C, 3D). These data indicate that the CD24-positive cells are pancreatic progenitors that share the same expression pattern of key transcription factors as progenitors during in vivo pancreatic development and can differentiate into insulin-producing cells.
The Expression Pattern of CD24 During Pancreatic Differentiation of Human ESCs
To explore the role of CD24 in the pancreatic progenitors derived from human ESCs, we examined its expression pattern by costaining with several stage-specific transcription factors during the differentiation process. CD24 was broadly expressed in human ESCs (Fig. 4A). Its expression was reduced after the initiation of differentiation. At the definitive endoderm stage, CD24 was weakly expressed in a subset of FOXA2- or SOX17-positive cells and in a subset of FOXA2- or SOX17-negative cells (Fig. 4B). At the early pancreatic specification stage, when HNF1B-positive foregut cells were robustly generated and a few PDX1-positive cells start to appear in HNF1B-positive cells, CD24 was re-expressed in a subset of the HNF1B-positive foregut cells, and the PDX1-positive cells expressed CD24 (Fig. 4C). In addition, when the expression of PDX1 reached its peak, most of the CD24-positive cells were PDX1-positive (Fig. 1A, 1B). The mRNA levels of CD24 and several stage-specific transcription factors during the differentiation process were consistent with the immunostaining pattern (Fig. 4D).
To investigate whether the re-expression of CD24 during differentiation was related to pancreatic induction, we withdrew the pancreatic induction factors RA, NOGGIN, and FGF7 at stage II of differentiation and performed flow cytometric analysis at the pancreatic progenitor stage. Compared with regular induction, few PDX1- or CD24-positive cells were produced without RA, NOGGIN, and FGF7 treatment (Fig. 4E). The expression patterns of CD24 and PDX1 and other stage-specific markers suggest that CD24 may play a role in prepancreatic specification.
Here, we identified a novel candidate surface marker, CD24, for pancreatic progenitors derived from human ESCs. The CD24-positive (also CD24-positive NCAM-negative) cells expressed key transcription factors of pancreatic progenitors and could differentiate into insulin-producing cells, indicating that CD24 is a reliable surface marker for human ESC-derived pancreatic progenitors.
This identification of a surface marker of pancreatic progenitors will facilitate the study of islet differentiation from human ESCs. Kroon et al. showed that human ESC-derived pancreatic endoderm could form mature islet cells 3 months after transplantation into immunodeficient mice; however, the tumorigenicity of the progenitor cells made them impracticable for clinical applications . Thus, mature islet cells, rather than pancreatic progenitor cells, are more suitable therapeutic cell sources. However, the generation of islet cells from pancreatic progenitor cells in vitro has become the bottleneck of ESC therapy for diabetes. Thus far, a reproducible and highly efficient beta cell differentiation protocol has not yet been reported , which can be explained in two ways. First, the considerable genetic variation and epigenetic differences among human ESC lines can affect their propensity to differentiate into certain lineages or cell types . Second, the nonpancreatic cell populations that exist in the differentiation cultures complicate the studies of pancreatic differentiation and limit the practicality of a large-scale screen. In fetal pancreatic development, epithelium-mesenchyme, epithelium-endothelium, and endothelium-mesenchyme interactions occur simultaneously and synergistically determine pancreatic progenitor maintenance, endocrine-exocrine regionalization, islet cell differentiation, and maturation through supportive or inhibitory effects [20, 21]. Signals derived from nonpancreatic lineages are varied, temporal-dependent, and critical for pancreatic lineage differentiation. We also found that sorted pancreatic lineage cells alone did not spontaneously and efficiently differentiate into insulin-producing cells; when the sorted CD24-positive cells were transplanted into immunodeficient mice, only a few insulin-positive cells could be detected in the graft 3 months post-transplantation (data not shown). This suggests that islet cell differentiation requires cell-cell interaction with other cell lineages and/or certain signals from other cell lineages. However, the mixed cell populations in the human ESC differentiation system complicate the elucidation of the inductive or inhibitive roles of other cell types. Based on the CD24 FACS technique, various signals derived from mesenchymal or endothelial cells need to be screened and could be precisely used to promote beta cell differentiation.
CD24 has been widely reported as a stem cell marker in several lineages. Peanut agglutinin and CD24 were used to purify adult neural stem cells that could give rise to neural and muscle cells . Recently, Pruszak et al. reported that the combination of CD15, CD24, and CD29 could define a surface biomarker code for neural lineage differentiation of ESCs . CD24 has also been used as a potential surface marker for mammary gland stem cells . Moreover, CD24 has been broadly used to purify human pancreatic cancer stem cells in combination with two other markers, epithelial specific antigen and HCAM . These reports suggest the potential conservation of CD24 as a stem cell marker. The present work further supports the prevalence of CD24 as a marker of stem/progenitor cells. CD133 is also a widely used stem cell marker and has been reported to be a marker for pancreatic ductal progenitor cells in neonatal or adult mouse pancreas  and a marker for NGN3-positive endocrine progenitors, combined with CD49f, in mouse and human fetal pancreas . Koblas et al. showed that CD133-positive human pancreatic cells could differentiate into insulin-producing cells . However, in our differentiation system, CD133 expression was much broader than that of PDX1 (Fig. 1A) and cannot be used as a suitable surface marker of PDX1 cells. This can be explained in two aspects. First, there are potential differences between rodent and human pancreas development . Many reports about surface markers of pancreatic progenitors were based on mouse model; whether this pattern can be readily translated into human is yet to be determined. Second, the use of markers for enrichment of cells depends critically on the nature of the “nontarget” population. The nontarget population is different between in vivo and in vitro models, that is, in the in vivo development, a certain cell population is regionalized in a specific organ, whereas in the in vitro differentiation model, cells of many other lineages exist. The nontarget population also changes overtime or among differentiation protocols. The broad CD133 expression in our experiment might be due to using an improper differentiation stage or a protocol that generating other cell lineages. In our experiment, we found that CD24 was a more promising marker; using only CD24, we could obtain PDX1-positive pancreatic progenitors from differentiated human ESCs.
Although the functions of CD24 in stem/progenitor cells have not yet been revealed, we note that prior reports have shown a similar enrichment of CD24 in endocrinal progenitors in E15.5 mouse pancreas . In the work identifying CD133+CD49flow cells as pancreatic endocrine markers, Sugiyama et al. also suggested that CD24 could further enrich Ngn3+ pancreatic lineage cells, although the enrichment was not great owing to the high efficiency of CD133+CD49flow. These data suggest that CD24 might also be a marker for pancreatic progenitors. In our in vitro study, we found that although CD24 was expressed in human ESCs, its expression was reduced during endoderm differentiation and reappeared only when pancreatic differentiation was induced (Fig. 4). Moreover, CD24 expression was found in a subset of the primitive gut tube prior to PDX1 expression (Fig. 4C). This expression pattern could provide insight into future study of the function of CD24 in pancreatic development.
We identified a novel surface marker, CD24, to enrich PDX1-positive pancreatic progenitors. To our knowledge, this is the first report of a specific surface marker for pancreatic progenitor lineages differentiated from human ESCs. With this marker, the purification of pancreatic progenitors is feasible, which will facilitate the investigation of the supportive or inhibitory roles of the nonpancreatic cell population during human ESC differentiation.
We thank Yizhe Zhang for technical support on real-time PCR and Liying Du for technical support on FACS analysis. We also thank Chengjun Li, Fangfang Zhu, Song Chen, and other colleagues in our laboratory for technical assistance during experiments. This research was supported by the National Basic Research Program for China (973 Program 2009CB941203), National Natural Science Foundation of China (30830061), National Key Technology R&D Program in 12th Five-Year Plan Period (2011ZX0 9102-010-03), and a 111 Project and Beijing Natural Science Foundation (5100002) to H.D. We apologize to colleagues whose work cannot be cited owing to space constraints.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.