To identify additional growth factors for optimizing propagation of human embryonic stem cells (hESCs), we mined publicly available data sets for the transcriptomes of murine and human ESCs and feeder cells, thereby generating a list of growth factors and complementary receptors. We identified the major pathways previously reported to be important, as well as several new ones. One pathway is the Pleiotrophin (PTN)-Pleiotrophin receptor (PTPRZ1) axis. Murine fibroblasts secrete Ptn, whereas hESCs expressed PTPRZ1, which is downregulated upon differentiation. Depletion of PTPRZ1 resulted in decreased colony formation and lower recovery of hESCs. Supplementation of chemically defined medium for feeder-free propagation of hESCs with PTN allowed higher recovery of hESCs without loss of pluripotency. PTN-PTPRZ1 functions here predominantly via an antiapoptotic effect mediated in part by the activation of Akt. These findings reveal the underlying importance of PTN in hESC survival and its usefulness in the clonal manipulation and large-scale propagation of hESCs.
Disclosure of potential conflicts of interest is found at the end of this article.
More than 150 human ESC (hESC) lines have been derived and maintained indefinitely in culture when propagated on mouse embryonic fibroblasts (MEFs). Propagation of mESCs and hESCs as undifferentiated colonies is dependent partly on cell density of the culture, with low density severely reducing the replating efficiency . Extensive studies have shown that mESCs are sustained by activation of the gp130/STAT3 pathway, generally achieved by the addition of leukemia inhibitory factor (LIF) or one of its family members to the culture. However, LIF-independent methods for maintaining mESCs have also been described, indicating that multiple paths to self-renewal may exist in mESCs . Single-cell propagation of hESCs has proven difficult, raising the possibility that gap junctional communication or autocrine factors are critical for self-renewal, similar to what was found for human embryonal carcinoma cells [3, 4] and possibly in hESCs also . Similar to mESCs, hESCs can be propagated in culture with MEFs or human embryonic fibroblasts (HEFs), although the compositions of growth factors that support them are likely to be different. Several laboratories have reported that the addition of LIF to hESCs does not maintain self-renewal [6, , , , –11]. However, although the signaling pathways that maintain hESCs in the undifferentiated state have not been fully delineated, different defined-media preparations capable of supporting hESCs have been reported (supplemental online Table 1), implicating a number of pathways (supplemental online Table 2) that include inhibition of cell death, enhanced proliferation, or inhibition of differentiation.
Although HEFs and MEFs appear capable of supporting most hESC populations, this property appears to be optimal at a specific stage of the embryo from which the fibroblast is obtained . Several components of the culture that are important for maintaining hESCs in MEF conditioned medium (MEF-CM) include factors present in the substrate that supports hESC growth, factors secreted by the MEFs , factors generated by the ESCs themselves (as suggested by density and cloning efficiency studies ), and the receptors that are found on the hESCs [15, 16]. For example, hESCs can be maintained in an undifferentiated state with high FGF concentrations in the presence of Matrigel (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com) but not laminin or tissue culture plastic .
The large data sets from whole genome-array studies of gene expression in MEFs and hESCs provide an opportunity to computationally tease out key components that are responsible for the unique ability of MEFs to maintain ESC lines. Here, we adopted a comparative bioinformatics approach to identify candidate signaling pathways that may be active when ESCs are propagated on MEFs. In our study, we identified the major pathways previously reported to be important, as well as several new ones. One such pathway is the Pleiotrophin (PTN)-Pleiotrophin receptor (PTPRZ1) axis, the function of which in hESC culture is being validated.
Materials and Methods
The biocomputational approach adopted in our study is summarized in supplemental online Fig. 1. We hypothesized that in any hESC cultures, there are likely three sources of growth factors present: factors that are associated with the substrate used, factors that are secreted by MEFs, and autocrine growth factors produced by hESCs. Since MEF-CM for propagating hESCs could include the largest variety of factors, we attempted to identify all growth factors present at detectable levels in MEFs. Total RNAs from seven different pools of MEFs (four C57BL6 and three CBI mouse strains) were probed using the Illumina array beads (Illumina Inc., San Diego, http://www.illumina.com) to generate information for MEF growth factors. Data for growth factors in human feeders and in hESCs were generated by analyzing previously published data sets (supplemental online Table 3).
The potential growth factors (GFs) and growth factor receptors (GFRs) essential for propagating hESCs and maintaining hESC pluripotency were obtained via queries into National Center for Biotechnology Information (NCBI) Gene using various keywords and Gene Ontology terms. The lists obtained were then supplemented through manual curation and filtered to include only those genes whose expression levels were detectable in any of the data sets (supplemental online Table 3). The lists were further refined by retaining only plausible GF-GFR binding interactions using either the Biomolecular Interaction Network Database (BIND) or the Human Protein Reference Database (HPRD). As interaction records typically do not span organisms, the murine growth factors were mapped to their human homologs using HomoloGene to determine candidate interactions between murine GFs and human GFRs. Finally, various gene expression data sets were used to check whether the expression of GFs was present to evaluate their importance for sustaining self-renewal. Likewise, expression data sets were used to identify GFRs enriched in hESCs compared with differentiated human embryoid bodies (hEBs). The data sets used included (a) six hESC and five hEB expression data points from four experiments [11, 18, –20] and (b) seven MEF cell lines. All data sets were normalized using the cross-correlation method . The criteria used to generate various tables of data derived from this analysis are explained in the respective supplemental online Tables 4–7.
Cell Culture, Transfection of hESCs, and Colony-Forming Assay
The human embryonic stem (ES) cell lines H1 and H9 (WiCell Research Institute, Madison, WI, http://www.wicell.org) were cultured in feeder-free conditions on Matrigel (Becton Dickinson) or fetal bovine serum (FBS)-coated plates. The cells were maintained in either MEF-CM or chemically defined medium (CDM) containing 10 ng/ml Activin and 12 ng/ml FGF at 37°C with 5% CO2. The conditioned medium was prepared according to Xu et al. . Briefly, it involves the conditioning of human embryonic stem (hES) medium (80% Dulbecco's modified Eagle's medium (DMEM)-Ham's F-12 medium, 20% Knockout serum replacement (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), 1 mM l-glutamine, 0.1 mM β-mercaptoethanol, 1% nonessential amino acids, and 4 ng/ml human basic fibroblast growth factor (bFGF) on MEFs overnight with an additional 4 ng/ml of bFGF before feeding hESCs. The preparation of CDM was done in accordance to Vallier et al. . For MEF culture, a total of seven different types of MEFs were used in our study. The MEFs were cultured in medium consisting of 90% DMEM high glucose, 10% FBS, 2 mM l-glutamine, and 50 μg/ml Pen-Strep (all from Invitrogen) to various passage numbers (P2, P3, and P4), and their growth were arrested by treating with either 10 μg/ml mitomycin-C or irradiation. Their RNAs were subsequently extracted for Illumina microarray analysis. ON-TARGETplus SMARTpool human PTPRZ1, PTN (Dharmacon, Lafayette, CO, http://www.dharmacon.com), and control short interfering RNA (siRNA) (Ambion, Austin, TX, http://www.ambion.com) was used for transfection into hESCs using Dharmafect 2 (Dharmacon) according to the manufacturer's recommendations. Briefly, 60 μM of siRNA was used for each transfection into 1.0 × 105 cells in suspension and subsequently plated onto a 12-well tissue culture plate. For colony-forming assay, single hESC suspension was transfected with PTPRZ1-RNA interference (RNAi) and seeded onto MEFs. Colonies were stained for alkaline phosphatase using the Alkaline Phosphatase Detection Kit (Chemicon, Temecula, CA, http://www.chemicon.com) according to the manufacturer's instructions.
RNA Extraction and cRNA Synthesis
For each MEF sample, 2 × 106 cells were harvested and lysed in TRIzol Reagent (Invitrogen) and purified with the RNeasy Mini Kit (Qiagen, Hilden, Germany, http://www.qiagen.com) according to the manufacturer's recommendations. Five hundred nanograms of total RNA from each sample was used to generate cRNA using the Illumina Totalprep RNA Amplification Kit (Ambion) according to the manufacturer's recommendations.
Reverse Transcription and Quantitative Real-Time Polymerase Chain Reaction
RNA samples (500 ng) were reverse-transcribed to obtain cDNA using the High Capacity cDNA Archive Kit (Applied Biosystems) according to manufacturer's recommendations. Quantitative polymerase chain reaction (PCR) analyses were performed using TaqMan TAMRATH probes (OCT4, SOX2, NANOG, PTPRZ1, PTN, SOX1, T-BRACHYURY, SOX17, and β-ACTIN) and analyzed using an ABI PRISM 7900HT Fast Real-Time PCR System. The threshold cycle (Ct) was determined to be ≥35. Each experiment was repeated at least twice with different batches of hESCs. Error bars were calculated from technical replicates based on duplicates real-time PCR measurements of DNA.
Immunocytochemical analysis was performed using mouse anti-human RPTP (receptor protein tyrosine phosphatase) β antibody (specific against PTPRZ1) and goat anti-human PTN antibodies (R&D Systems Inc., Minneapolis, http://www.rndsystems.com) to demonstrate the presence of the pleiotrophin receptor and pleiotrophin in hESCs. Cells were first harvested and washed once with phosphate-buffered saline (PBS). Fixation of cells was achieved with 4% formaldehyde for 30 minutes. The cells were then blocked with PBS containing 5% FBS and 1% bovine serum albumin for 30 minutes at room temperature. Primary antibodies were added at 1:1,000 dilutions and incubated for 1 hour at room temperature. After washing, the cells were incubated for another 45 minutes at room temperature without light exposure with either 1:500 diluted Alexa Fluor 488 or fluorescein isothiocyanate (FITC)-labeled secondary antibodies. Nuclei were counterstained with 4,6-diamidino-2-phenylindole. The cells were observed under a confocal microscope.
MEF-CM for Western blot analysis was prepared by incubating MEFs with hES medium in the absence of knockout serum replacement. The conditioned medium was harvested after 24 hours and concentrated using Amicon (Millipore, Billercia, MA, http://www.millipore.com) Ultra-15 Centrifugal Filter Units. For the preparation of cell lysates, cell pellets were incubated in RIPA lysis buffer (1% Nonidet P-40, 0.5% deoxycholate, 5 M NaCl, and 1 M Tris, pH 7.4) for 10 minutes at 4°C and then centrifuged at 11,400g for 15 minutes at 4°C. Quantification of protein extract was carried out using the Protein Assay (Bio-Rad, Hercules, CA, http://www.bio-rad.com) according to the manufacturer's instructions. Electrophoretic analysis was performed using 10%–15% SDS–polyacrylamide gel electrophoresis gel (Bio-Rad). Gels were blotted onto nitrocellulose membrane (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com), which was then probed with either goat anti-human PTN antibody (R&D Systems) or mouse anti-human RPTPβ antibody (BD Biosciences, San Diego, http://www.bdbiosciences.com). For the Western blot analysis on AKT and ERK, primary antibodies raised from rabbit (anti-human AKT, anti-phospho-AKT [Ser473], anti-p44/42 mitogen-activated protein [MAP] kinase, and anti-phospho-p44/42 MAP kinase [Thr202/Tyr204]; Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com) were used. Primary antibodies were detected with species-specific horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com) and the enhanced chemiluminescence Western blotting detection system (Amersham Biosciences) according to the manufacturer's instructions.
Apoptosis and cell cycling assays on hESCs were analyzed by flow cytometry using the Annexin-V-FITC Apoptosis Detection Kit II (BD Biosciences) and the FITC BrdU Flow Kit (BD Pharmingen, San Diego, http://www.bdbiosciences.com/pharmingen), respectively, according to the manufacturer's instructions. Samples were analyzed on a FACSCalibur flow cytometry (Becton Dickinson).
Using the criterion of signal intensity >10 from the Illumina array data (background signal subtracted), a total of 224 growth factors were found in MEFs (supplemental online Table 4). An abbreviated list of selected factors, expressed at high to low levels, is shown in Table 1. To generate a database for membrane receptors expressed in hESCs, we interrogated existing hESC gene-expression data sets (supplemental online Table 3), NCBI Gene, and Gene Ontology (combined with a curation of gene probes present in the Illumina bead array). Using a similar criterion of signal intensity >10 as a cutoff, 469 membrane receptors in hESCs were identified (supplemental online Table 5). To identify receptors in hESCs that may be involved in ES cell maintenance, we further imposed stricter criteria of signal intensity >50 and ES/EB ratio ≥1.2 to select for receptors that are significantly downregulated when hESC differentiate. A total of 52 membrane receptors were identified (supplemental online Table 6), a shortened list of which is shown in Table 2. By further pairing up the 52 receptors with their corresponding growth factors expressed in MEFs at signal intensities >10 and using interaction data from BIND and HPRD, we obtained 27 GF-GFR pairs (supplemental online Table 7). These can be clustered into at least eight pathways (Fig. 1A) (PTN, WNT, BMP, FGF, ERB, ACTIVIN, FLT4, and NOTCH). Of these, FGF, ACTIVIN, BMPs, and NOTCH have already been verified as essential or playing significant role in maintaining pluripotency in hESCs or modulating decisions between self-renewal and differentiation. More interestingly, we also uncovered several novel GF-GFR pairings whose expression patterns suggest that they may also play an important role in hESCs. One such receptor, with the highest ES/EB ratio and expressed at high levels in hESCs, is the Pleiotrophin receptor (PTPRZ1). Most interestingly, the corresponding growth factor in MEFs, pleiotrophin (ptn), is expressed at very high levels in MEFs. Murine and human PTN are 98% identical over 168 amino acids and share 82% homology in nucleotide sequences.
Table Table 1.. Selected list of growth factors from supplemental online Table 4 (list of growth factors expressed in MEF, human fibroblast and hESCs)
Table Table 2.. Selected list of human membrane receptors expressed in hESCs and hEBs arranged in descending order of signal intensity ratio of ESC/EB (Illumina platform)
Expression of Pleiotrophin in MEFs and Pleiotrophin Receptor in hESCs
To determine whether Ptn was secreted by MEFs into the supernatant, we performed Western blot analysis of MEF-CM (hES medium cultured with MEFs for 24 hours; described in Materials and Methods). An 18-kDa protein similar to recombinant human PTN was detected (Fig. 1B). We estimated that a typical MEF-CM contained Ptn in the range of 50–80 ng/ml. We next demonstrated the presence of pleiotrophin receptor (PTPRZ1) in hESCs by Western blot analysis and immunostaining. Western blots revealed a 250-kDa protein band corresponding to PTPRZ1, and the same antibody showed a speckled pattern observed on hESC membrane (Fig. 1D, 1E). Although there were generally high levels of PTPRZ1 in hESCs, we observed this to vary between different hESC lines.
PTPRZ1-siRNA Knockdown of Pleiotrophin Receptor Affects hESC Growth
The presence of Ptn in MEF-CM coupled with high levels of PTPRZ1 in hESCs suggests that Ptn may be one of the growth factors in MEF-CM supporting hESC growth and survival. To test this, we disrupted the expression of PTPRZ1 in hESCs using RNAi under two conditions. First, hESCs maintained under feeder-free conditions in the presence of MEF-CM were transfected with PTPRZ1-RNAi, and the total cells recovered were assessed after 7 days. Compared with control RNAi, the number of cells recovered after PTPRZ1-RNAi treatment was approximately 60% lower (Fig. 2A). Real-time PCR results confirmed that PTPRZ1-RNAi treatment reduced the receptor transcripts in hESCs by 70% (Fig. 2B). Significantly, the knockdown of PTPRZ1 led to reduction of OCT4 and SOX2 transcripts by approximately 40%. OCT4 knockdown using OCT4-RNAi also resulted in 90% decrease of PTPRZ1 transcripts (Fig. 2B), consistent with the marked decrease of PTPRZ1 in differentiating ESCs (Fig. 1D). These results showed clearly that MEF-CM cannot maintain hESCs without PTPRZ1, demonstrating for the first time, the importance of Pleiotrophin pathway in hESCs.
PTN-PTPRZ1 Axis Mediates Clonal Propagation of hESCs
We assessed the effect of knockdown of PTPRZ1 on the clonal propagation of hESCs. Single-cell suspensions of hESCs after PTPRZ1-RNAi treatment were reseeded onto MEF feeders. Single colonies arise after 5–10 days in culture, and the frequency of colony-forming cells (clonal efficiency) was determined (supplemental online Fig. 2). Compared with hESCs treated with control RNAi, PTPRZ1-RNAi treated hESCs yielded 30% fewer colonies. (Fig. 2C, 2D). Together, these experiments demonstrated that PTPRZ1 mediates important signals in hESCs for survival, proliferation, and maintenance of pluripotency state.
Presence of Autologous PTN-PTPRZ1 Signaling Loop in hESCs
We next validated the transcriptome data of various hESC lines to show that hESCs also express PTN. Immunostaining of hES cells confirmed a positive expression for the growth factor (Fig. 1E), and Western blot analysis showed the presence of PTN in hESC-conditioned medium in the range of 20 ng/ml (Fig. 1C). This suggests that there may be an autologous PTN-PTPRZ1 signaling loop in hESCs. To test this, we knocked down PTPRZ1 in hESCs maintained both feeder- and PTN-free medium. Surprisingly, the total hESCs recovered after 7 days was 20%–30% less compared with control RNAi (Fig. 2E), indicating that even in culture medium with no exogenous pleiotrophin, PTPRZ1 is also engaged in maintaining hESCs. To test whether endogenous PTN contributed to this effect, RNAi knockdown of endogenous PTN in hESCs was performed. We were able to suppress PTN transcripts by 60% (Fig. 2F), and as a result, recovery of hESC after 7 days was suppressed by approximately 40%–45% compared with control RNAi (Fig. 2E). Our data therefore indicate that an active autocrine PTN-PTPRZ1 loop is engaged by hESCs, further emphasizing the importance of signals from the pleiotrophin receptor.
Pleiotrophin Enhances Cell Growth Primarily by Preventing Apoptosis
To determine the cellular function supported by PTN, we examined the effect of PTN on apoptosis and cell cycle. Using Annexin V as a marker for apoptotic cells, hESCs maintained on MEF-CM were assessed for degree of apoptosis 24 hours after RNAi knockdown of the receptor. Flow cytometric analysis showed that the percentage of cells undergoing apoptosis was approximately 38% higher compared with control treated cells (Fig. 3, Figure 3.A). Similarly, Ptprz1-RNAi treatment of hESCs maintained feeder-free in CDM containing Activin and FGF resulted in an increase of apoptotic cells by 17%, demonstrating the subtle but detectable impact of endogenous PTN.
We next performed bromodeoxyuridine (BrdU) incorporation to assess the effect of Ptn on cell cycle. We compared BrdU labeling of hESCs maintained in MEF-CM, before and after RNAi knockdown of PTPRZ1. There was no significant difference in BrdU incorporation between control cells (66.2%) and RNAi treated cells (63%) (Fig. 3, Figure 3.A). In CDM supplemented with Activin and FGF, approximately 53% of the cells were BrdU-labeled. Upon addition of 100 ng/ml of PTN, BrdU-labeled cells increased slightly, to 60.5% (Fig. 3, Figure 3.A). We concluded from these studies that the major effect from the engagement of PTPRZ1 is to enhance cell survival by providing antiapoptotic signals and, to a lesser extent, some increase in cell cycle.
PTN belongs to the heparin-binding cytokine family that includes Midkine (MK). MK and PTN were shown to use PTPRZ1 to activate phosphatidylinositol 3-kinase and Erk pathways in osteoblasts and neuronal cells. PTN was also shown to inactivate the phosphatase activity of PTPRZ1, resulting in an increased tyrosine phosphorylation of β-catenin [24, 25]. On the basis of these studies, we tested whether PTPRZ1 in hESCs could act through the Erk or Akt pathways. We observed that in hESCs maintained and passaged in Activin and FGF, addition of PTN resulted in increased phosphorylation of Akt detectable within 24–48 hours, whereas Erk phosphorylation was not altered (Fig. 3, Figure 3.B, 3, Figure 3.C). Likewise, in hESCs maintained in Activin and FGF medium supplemented with PTN, removal of PTN led to decreased levels of phosphorylated Akt (data not shown).
Pleiotrophin Enhances Growth of hESCs in Long-Term Culture
To investigate whether PTN may be gainfully used to support the propagation of hESCs, we asked whether hES medium (basic medium used to prepare MEF-CM) supplemented with PTN alone may be sufficient to maintain growth and self-renewal of hESCs. Using hES medium alone, hESCs rapidly differentiate. By sharp contrast, in medium supplemented with various concentrations of PTN (10, 50, and 100 ng/ml), hESC colonies were larger, while maintaining a tight undifferentiated morphology for at least 3 days (supplemental online Fig. 3). However, PTN alone was not capable of sustaining hESCs in an undifferentiated state. By day 7, transcript levels of pluripotency-associated genes such as Oct4 and Nanog were reduced (Fig. 3, Figure 3.D).
Finally, we tested whether PTN could be used to enhance feeder-free expansion of undifferentiated hESCs. Two growth factors commonly used in chemically defined medium are Activin and FGF. In short-term 7-day cultures, supplementation of PTN to Activin + FGF medium yielded approximately 50% higher total number of hESCs recovered at various days (Fig. 3, Figure 3.E). When extended to long-term culture, the recovery of the total accumulative number of hESCs after 30 days of continuous passaging was reproducibly 150%–200% (1.5–2-fold) higher in Activin/FGF/PTN-containing medium compared with cells maintained in Activin/FGF medium only (Fig. 3, Figure 3.F). Importantly, the addition of PTN did not alter the pluripotency of hESCs as measured by Tra-1–60 when cells were passaged in 100 or 200 ng/ml PTN, but some decrease was observed with a higher dosage of PTN (500 ng/ml) (Fig. 3, Figure 3.G). Altogether, these results indicated that PTN can be used to supplement other hESC culture medium to increase recovery of undifferentiated hESCs.
We have described here a process of data mining to identify candidate growth factors that may be critical for ESC self-renewal. We identified many candidate receptor-growth factor pairings that include most of those previously known to be important for ES cells. Testing one novel pathway here, we showed that the PTPRZ1 is actively engaged by hESCs as revealed by a reduction of hESC colony-forming efficiency and increased apoptosis when the receptor was depleted. Our evidence for its action via Akt is consistent with Akt's known potent antiapoptotic effect by inactivating proapoptotic proteins [26, 27]. Others have also shown that activation of Akt signals can maintain pluripotency of murine and primate ES cells . Here, we further demonstrated that PTN is secreted by hESCs and MEFs, both contributing to the enhanced propagation of hESCs. Recently it has been reported that other factors, such as neurotrophin, may be used to enhance clonal growth of hESCs . Our genomic data showed that neurotrophin is expressed at very low to nondetectable levels by MEFs and hESCs. Here, we illustrated that PTN-PTPRZ1 pathway functions significantly as an autologous loop in supporting hESC survival and their clonal growth. Most importantly, we can conclude that for the purpose of growing hESCs in feeder-free defined medium, the addition of PTN to enhance expansion of hESCs should be considered. Human fibroblast feeders express much lower levels of PTN compared with murine MEFs. For the practical application of hESCs in regenerative medicine, a means to propagate hESCs in completely human material is necessary. Human recombinant PTN would therefore provide a useful supplementary growth factor in medium for large-scale propagation of hESCs in human feeders or under feeder-free conditions. The data-mining method described here can be applied to screen for other putative protein-protein interactions for other classes of molecules, such as transcription factors, thereby exploiting existing databases to obtain insights into the process of stem cell self-renewal.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.
This work is supported by the Agency for Science, Technology and Research (Singapore). B.S.S. is a recipient of the A*STAR graduate scholarship. This work is also partially supported by NIH Grants DK04763 and AI54973 (to B.L.) and by Biomedical Research Council Grant BMRC/05/1/21/19/391 (to E.H.L.).