Wnt Signaling Orchestration with a Small Molecule DYRK Inhibitor Provides Long-Term Xeno-Free Human Pluripotent Cell Expansion

Cells and Cell Culture

The hESC lines HES2 [2], HES3 [2], H1 [1], and H9 [1] and the human fetal dermal fibroblast-derived iPSC line [25] were maintained using standard cell culture methodology. For enzymatic bulk expansion, the cells were cultured on mitotically arrested mouse embryonic fibroblast (MEF) feeder cell layers in Dulbecco's modified Eagle's medium/Ham's F-12 medium (DMEM/F-12; Sigma D6421; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) supplemented with 20% knockout serum replacement (KSR; Invitrogen, Carlsbad, CA, http://www.invitrogen.com), l-glutamine, nonessential amino acid, and 4 ng/ml recombinant bFGF2 (fibroblast growth factor 2 [FGF2]; Peprotech, Rocky Hill, NJ, http://www.peprotech.com), as previously described. For feeder-free culture, the MEF feeder layer was removed via separation from embryonic stem cell clumps by sedimentation under gravity during passage, and cells were cultured in MEF-conditioned medium (MEF-CM) on 30-fold diluted Matrigel (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com)-coated culture dishes.

The replating assay was previously described [26]. Briefly, the hESCs in feeder-free culture were dissociated completely with 0.05% trypsin-EDTA (Invitrogen) and seeded at 10⁴ cells per well in Matrigel-coated 6-well culture plates and cultured in MEF-CM. Various concentrations of Wnt3a (purchased from Peprotech or prepared in our laboratory on the basis of a previous report [27]; the activity of the purified Wnt3a was confirmed using the super TOPFLASH assay [28]), IQ-1, ID-8, and/or ICG-001 (all synthesized in our laboratory; identity and purity confirmed by proton nuclear magnetic resonance and liquid chromatography/mass spectrometry [LC/MS]) were supplemented into the culture media at the onset of seeding and then continuously until the end of culturing. For all assays, the cell and colony morphology were examined under a microscope, and the replating efficiency was examined by counting the number of colonies after 7 days of culture.

For xeno-free, feeder-free culture, the cells were passaged onto 2 μg/cm² human fibronectin (from human foreskin fibroblasts; Sigma-Aldrich F2518), 10–20 μg/cm² human laminin (from human fibroblasts; Sigma-Aldrich L4544), and 10 μg/cm² human vitronectin (Recombinant Vitoronectin; Invitrogen PHE0011)-coated culture dishes and cultured in DMEM/F-12 (Sigma D6421) supplemented with l-glutamine, nonessential amino acids, insulin-transferrin-selenite (ITS) (Sigma-Aldrich I1884; dosage twice as high as manufacturer's instructions), 1 μg/ml l-ascorbic acid, 20–25 ng/ml Wnt3a, and 500 nM ID-8, with or without 4 ng/ml bFGF. The culture medium was changed daily. Some of the colonies detached from the edge of the dish after 5–7 days of culture. At this time or when colonies became confluent, cells were treated with 0.5 mg/ml GRGDTP peptide (AnaSpec, Fremont, CA, http://www.anaspec.com) or 1 μg/ml echistatin (Sigma-Aldrich) and incubated at 37°C for 15–45 minutes until all colonies started to detach at the edge region. The colonies were detached and partially dissociated into clumps by gentle pipetting, and then the clumps were washed with culture medium, collected by centrifugation, and seeded onto new dishes at a 1:2 or 3 split ratio.

For examination of cell attachment, subconfluent cells were dissociated as clumps, and a part of the solution containing cell clumps was dissociated into single cells with 0.25% trypsin-EDTA for cell counting. The clumps were seeded into 24-well plates with an estimated 5,000-cell clumps per well. After overnight culture, nonattached cells were washed away with phosphate-buffered saline (PBS), attached cells were dissociated with 0.25% trypsin-EDTA, and then the attached cell number was examined. For examination of cell population growth, the cells were seeded into 24-well plates with an estimated 2,000- or 5,000-cell clumps per well, and the cells were dissociated and counted every 24 hours.

Experiments using hESCs and hiPSCs in this study have been approved by University of Southern California Stem Cell Research Oversight Committee (No. 2008-6-1).

Quantitative Polymerase Chain Reaction Analysis

Total RNA was isolated from hESCs cultured in feeder-free conditions with or without Wnt3a and ID-8 using an RNeasy micro kit (Qiagen, Hilden, Germany, http://www1.qiagen.com). Reverse transcription was performed with Omniscript (Qiagen), and quantitative reverse transcription–polymerase chain reaction (PCR) was then performed using each gene-specific primer/probe mix (TaqMan Gene Expression Assays), TaqMan 2 × Master Mix, and the ABI Prism 7900 Sequence Detection system (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com) according to the manufacturer's protocol. The PCR data were analyzed by the Δ/Δ C_T method and normalized with Cycrophilin A (PPIA) expression with RQ Manager software (Applied Biosystems).

Identification of ID-8 Target

The cells were washed twice with 5 ml of ice-old PBS, harvested by scraping into 1 ml of PBS containing protease inhibitor cocktail (Calbiochem, San Diego, CA, http://www.emdbiosciences.com) and collected by centrifugation. The cell pellet was frozen and stored in liquid nitrogen. A total of 2 g of cell pellets were collected from 20 dishes each of 15-cm diameter. The pellets were thawed on ice and then suspended in 20 ml of chilled M-Per buffer (Pierce, Rockford, IL, http://www.piercenet.com) containing 1 mM dithiothreitol (DTT) and the protease inhibitor cocktail. The cell suspension was mixed and lysed on a nutating mixer for 30 minutes at 4°C. The lysate was then centrifuged at 10,000 rpm for 30 minutes at 4°C, and the supernatant was divided into two equal aliquots in 50-ml centrifuge tubes. Dimethyl sulfoxide (DMSO) control was added to one tube, and 100 μM ID-8 was added to the other tube. The tubes were incubated for 1 hour on a nutator at 4°C. Four hundred microliters of a 50% slurry of streptavidin-Sepharose (GE Healthcare Life Sciences, Piscataway, NJ, http://www.gelifesciences.com) was pre-equilibrated in M-Per buffer and then added to 10 ml of 50 μM biotinylated ID-8 and incubated for 1 hour at 4°C. The Sepharose beads were then washed extensively with M-Per to remove unbound biotinylated ID-8. The streptavidin-Sepharose bound with biotinylated ID-8 was mixed with the DMSO- or free ID-8-treated cell lysates described above. The mixture was then incubated on a nutating mixer for 4 hours at 4°C. The Sepharose beads, containing proteins bound to biotinylated ID-8, were washed three times with 1 ml of M-Per buffer and then mixed with 2× Laemmli Sample Buffer and boiled for 5 minutes. The proteins and Sepharose were then separated using Illustra MicroSpin Columns (GE Healthcare Life Sciences). Bound proteins eluted in Laemmli Buffer were subjected to SDS-polyacrylamide gel electrophoresis (PAGE) and silver stained (Invitrogen). The specific bands decreased by free competitor ID-8 were analyzed by LC/MS (LTQ-XL; Thermo Scientific). The elution was also analyzed by Western blotting with specific antibodies against dual-specificity tyrosine phosphorylation-regulated kinase 1a (DYRK1a) (2771; Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com), DYRK2 (ab37912; Abcam, Cambridge, MA, http://www.abcam.com), DYRK3 (sc-66868; Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com), DYRK4 (ab37911; Abcam), or HIP2K (sc-100383; Santa Cruz Biotechnology).

Knockdown Assay

Several DYRK microRNAs (miRNAs) were designed and lentiviral vector plasmids were constructed using the Block-iT Pol II miR RNAi kit (Invitrogen) according to the manufacturer's protocol. The knockdown efficiencies of the miRNA expression plasmids were evaluated by transient transfection and Western blotting. The most efficient miRNA sequence, antisense, 5′-tcaaggagtcaatttcgtaacg-3′, and sense, 5′-gttacgaatgacctccttg-3′, was used for further lentiviral production.

To produce the miRNA expression lentiviral vectors, the plasmid was cotransfected with packaging plasmids for vesicular stomatitis virus G protein pseudotyped virus into 293T cells with Fugene6 transfection reagent (Roche Applied Science, Indianapolis, IN, http://www.roche-applied-science.com). Viral supernatants were collected and concentrated by ultracentrifugation for 2 hours. Feeder-free H9 cells were infected with the virus in MEF-CM supplemented with 10 μg/ml polybrene. The infected cells were selected and isolated as pools using 0.5–1 μg/ml Blasticidin over 1 week. The infected H9 cell pools were then used for examination of replating efficiency.

Nuclear Purification and Immunoprecipitation

The cells were washed twice with ice-cold PBS, scraped, and harvested following treatment with DMSO control, 500 nM ID-8, or 20 μM ICG-001 for 24 hours. The nuclear fraction was purified using an NE-PER nuclear extraction kit (Pierce, Thermo Scientific). For prevention of protein degradation, 1 mM DTT and a protease inhibitor cocktail (Calbiochem) were supplemented during harvest and processing. A total 5 μg of nuclear protein from each treated sample was then reacted with 1 μg of anti-CREB-binding protein (CBP) antibody (sc-369; Santa Cruz Biotechnology), anti-p300 antibody (sc-584; Santa Cruz Biotechnology), or normal rabbit IgG in 1 ml of immunoprecipitation buffer (25 mM Tris, HCl, pH 7.5, 150 mM NaCl, 5% glycerol, 0.5% Nonidet P40, 1 mM EDTA, 1 mM DTT, and protease inhibitor cocktail) at 4°C overnight. The protein-antibody complex was then reacted with 50 μl of 50% slurry of protein A-agarose beads (Roche Applied Science) on a nutator for 1 hour at 4°C. The bead complex was washed three times with immunoprecipitation buffer, resuspended with 2× Laemmli buffer, and boiled for 5 minutes. The protein and Sepharose were then separated on an Illustra MicroSpin Column (GE Healthcare Life Sciences). Bound proteins eluted in the Laemmli buffer were subjected to SDS-PAGE and transferred onto polyvinylidene difluoride membrane overnight. The coimmunoprecipitated β-catenin was detected by Western blotting with an anti-β-catenin antibody (610153; BD Biosciences). The intensity of the β-catenin Western blotting signal was analyzed using UN-SCAN-IT software (version 5.1, Silk Scientific, Orem, UT, http://www.silkscientific.com) and normalized with the signals in the total nuclear lysate.

Characterization, Karyotype Analysis, and Differentiation Assay

The hESC colonies were fixed with 4% paraformaldehyde (PFA), and alkaline phosphatase (ALP) activity was determined with a Vector Blue Alkaline Phosphatase Substrate Kit (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). Immunostaining was carried out with the following primary antibodies: TRA-1–60 and TRA-1–80 (Santa Cruz Biotechnology), anti-stage-specific embryonic antigen (SSEA)-3 and -4 (Developmental Studies Hybridoma Bank, Iowa City, IA, http://www.uiowa.edu/∼dshbwww), GCTM-2 (our laboratory), anti-OCT-3 (clone C-10; Santa Cruz Biotechnology), rabbit anti-SOX2 (Millipore, Billerica, MA, http://www.millipore.com), or goat anti-Nanog (R&D Systems Inc., Minneapolis, http://www.rndsystems.com). Then, samples were incubated with Alexa Fluor 488- or 594-conjugateed antibodies (Molecular Probes, Eugene, OR, http://probes.invitrogen.com), and staining was detected by indirect immunofluorescence microscopy.

For fluorescence-activated cell sorting (FACS) analysis, cells were dissociated with TrypLE (Invitrogen), fixed with 2% PFA, and reacted with TRA-1–60 or GCTM-2 antibody followed by Alexa Fluor 488-conjugated anti-mouse IgM antibodies. The stained cells were then analyzed using an LSRII flow cytometer (BD Biosciences).

For karyotyping, hESCs were treated with 100 ng/ml colcemid (KaryoMax; Invitrogen), dispersed, and fixed. Twenty to 50 cells at metaphase in each sample were analyzed for G-banding at the 300–500 band levels following trypsin digestion and Giemsa staining.

For differentiation assays, embryoid bodies (EBs) were formed for 2 weeks in low-attachment culture plates and then plated onto glass culture slides as described previously [26]. After the cells grew out from attached EBs, they were fixed with 4% PFA, incubated with anti-βIII-tubulin (clone TU-20; Chemicon, Temecula, CA, http://www.chemicon.com), anti-glial fibrillary acidic protein (anti-GFAP; clone 4A11; BD Biosciences), anti-α-smooth muscle actin (αSMA) (clone 1A4; DakoCytomation, Glostrup, Denmark, http://www.dakocytomation.com), anti-Desmin (clone Ab-1; Thermo Scientific), anti-SOX17 (clone 245013; R&D Systems), anti-SOX7 (goat polyclonal; R&D Systems), or anti-α-fetoprotein (AFP) antibody (clone C3; Sigma-Aldrich) as the primary antibodies and then detected using Alexa Fluor 488- or 594-conjugated secondary antibodies (Molecular Probes) and fluorescence microscopy.

For teratoma formation, hESCs were dissociated as clumps, and a 200-μl suspension containing approximately 10⁷ cells was subcutaneously injected into SCID mice (Jackson Laboratory, Bar Harbor, ME, http://www.jax.org). After 2–3 months, the resulting teratomas were dissected out and fixed with Bouin's fixative solution. Samples were embedded in paraffin, sectioned at 5-μm thickness, and subsequently stained with hematoxylin and eosin. This study has been approved by the University of Southern California Institutional Animal Care and Use Committee (No. 11273).

Small Molecule ID-8 Supports hESC Maintenance in Combination with Wnt-3a

The effects of the canonical Wnt ligand Wnt3a and several candidate small molecule Wnt signaling modulators were first evaluated briefly in a replating assay [26] (Fig. 1A). In the replating assay, colonies were dissociated into single cells and seeded on Matrigel-coated plates in MEF-CM [29]. We evaluated cell survival by colony number (replating efficiency: colony number/seeded cell number), proliferation by colony size, and the undifferentiated state by morphology and ALP positivity. When HES2 cells [2] were dissociated and seeded in control MEF-CM, the replating efficiency was ∼0.6%. Supplementation with Wnt3a enhanced colony formation slightly (∼1% at 100 ng/ml). The colonies were larger and consisted of more cells. All the colonies were still ALP positive. However, the cells in the Wnt-treated colonies appeared flatter than with the control non-Wnt conditions (Fig. 1B, 1C). These results suggested that Wnt3a supplementation enhances hESC survival and proliferation but also partially induces differentiation. As anticipated, treatment with the small molecule ICG-001, which directly modulates Wnt/β-catenin signaling by antagonizing the interaction of β-catenin with transcriptional cofactor CBP and enhancing its interaction with transcriptional cofactor p300 [21], strongly induced cell differentiation in the presence or absence of Wnt3a, and almost all colonies were ALP negative. The other modulator, IQ-1, which is sufficient to maintain mESC proliferation and pluripotency for extended periods of time in the absence of serum and LIF in conjunction with Wnt3a by enhancing the CBP/β-catenin association [22], did not significantly affect hESC survival and proliferation by itself even at the highest dose in the nontoxic range (10 μM), although it partially prevented Wnt-induced differentiation. By contrast, the third modulator, ID-8, which was identified in the same screen as IQ-1 [24], increased hESC survival (∼1.1% at 0.5 μM). The combination of Wnt3a and ID-8 further increased survival (∼1.7%) and completely prevented Wnt-induced differentiation morphologically without disrupting proliferation. Cells treated with this combination expressed pluripotency markers, including ALP, OCT4, NANOG, and SOX2 at essentially the same levels as control hESCs (Fig. 1D, 1F). Wnt-induced differentiation marker gene expression (i.e., GATA6, SOX17, T [Brachyury], GSC [goosecoid] and CDX2) was dramatically reduced by ID-8 at a dose that maintained the undifferentiated state in this assay (Fig. 1G, 1H). To examine the generality of this effect in hESCs, replating assays were performed with the additional hESC lines HES3 [2] and H9 [1] (Fig. 1E). In both, the combination of Wnt and ID-8 enhanced hESC replating efficiency, and colonies expressed ALP and displayed undifferentiated morphology.

ID-8 Targets DYRK in hESCs

We next sought to investigate the molecular targets and mechanism(s) of action of ID-8 in Wnt/β-catenin signaling. To identify the molecular target(s) of ID-8 (Fig. 2A), we used affinity chromatography (Fig. 2B). We identified DYRK2 as a binding partner of ID-8 in cell lysates; its binding to the ID-8 affinity column was competed away by free ID-8. Immunoblotting of the bound proteins identified not only DYRK2 but also DYRK4 (Fig. 2C). These data suggested that these or other members of the DYRK family are critical molecular targets of ID-8 in hESCs. To confirm this, we designed miRNAs to target the DYRK family and evaluated their effects on protein expression after transfection by immunoblotting (Fig. 2D, 2E). The DYRK miRNA-transfected hESCs maintained their undifferentiated state, as did the untransfected or scrambled control miRNA-transfected cells in MEF-CM or on feeders (supplemental information Fig. 1), showing that there was no adverse effect of the miRNA under conditions that support hESC renewal. Next, we performed replating assays with the stably transfected hESCs. Survival and self-renewal were not different between DYRK-knockdown cells and control cells in the absence of Wnt. However, both parameters were enhanced in the DYRK-knockdown cells in the presence of Wnt ligand (Fig. 2F). The effect of DYRK-knockdown was thus very similar to ID-8 treatment in the replating assay. Taken together, these data indicate that members of the DYRK family are direct targets of ID-8 and that ID-8 enhances Wnt-mediated hESC survival and proliferation via inhibition of DYRKs.

Based upon our model and previous investigations in mESCs [22], we anticipated that ID-8 would modulate Wnt/β-catenin signaling through the enhancement of CBP/β-catenin association at the expense of the p300/β-catenin association in hESCs. To test this hypothesis, CBP and p300 were immunoprecipitated from purified nuclear fractions and then coprecipitated β-catenin amount was examined by immunoblotting (Fig. 2G, 2H). As expected, levels of the CBP/β-catenin complex were significantly enhanced in ID-8-treated cells compared with control and in particular to ICG-001-treated cells. A similar result was obtained in DYRK-knockdown cells (Fig. 2I, 2J). We conclude that ID-8, through inhibition of DYRK family members, supports maintenance of the undifferentiated state in the presence of Wnt by modulating Wnt/β-catenin signaling and enhancing the CBP/β-catenin association in hESCs, similar to the mechanism observed in mESCs treated with IQ-1 and Wnt.

ID-8 and Wnt-Based Xeno- and Feeder-Free Culture System for hESCs and Human iPSCs

Because Wnt and IQ-1 could maintain mESCs without serum and LIF [22], we next investigated the development of a xeno-free, feeder-free hESC culture system using ID-8 with Wnt activation. We first examined the maintenance of hESCs in conventional hESC medium, supplemented with Wnt3a and ID-8, and stepwise elimination of bFGF and KSR components, such as albumin, ITS, and ascorbic acid. After extensive investigation, we determined that the minimal medium required to maintain hESC proliferation and pluripotency consisted of DMEM/F-12 with Wnt3a (25 ng/ml), ID-8 (0.5 μM), and ITS (supplemental information Fig. 2A). In the absence of ID-8, Wnt3a, or ITS, no colonies were obtained. l-Ascorbic acid did not affect the maintenance of hESCs in this assay, apart from causing growth inhibition at doses greater than 25 μg/ml. In the minimal effective condition, hESCs maintained their undifferentiated morphology in the absence of bFGF; however, proliferation was significantly enhanced as in MEF-CM by supplementation of 4 ng/ml bFGF (supplemental information Fig. 2A). This concentration of bFGF is much lower than that reported previously for xeno- and feeder-free medium supplemented with bFGF (100 ng/ml) [10].

Initially, we maintained cells on Matrigel, a complex animal-derived extracellular matrix (ECM). However, we investigated various ECM combinations and found that a combination of fibronectin, laminin, and vitronectin (FLV) was best for supporting hESC attachment (supplemental information Fig. 2B) and self-renewal and for maintenance of undifferentiated colony morphology in the medium. Using this ECM combination, the majority of the colonies remained intact up to 7 days in this culture condition, although some colonies tended to detach after 5 days. Therefore, it proved optimal to passage at around 5 days after seeding.

Because of the extremely low level of protein in these cultures (i.e., no albumin or Matrigel), the hESCs are extremely sensitive to damage caused by the dissociation enzymes normally used for passaging, such as collagenase IV, dispase, or trypsin/EDTA. ROCK inhibitor, which has been reported to enhance hESC survival [30], was not very effective under these conditions and also induced differentiation, even with manual mechanical passaging. To develop a passaging process compatible with the very low protein content of our defined medium, we investigated several integrin antagonists. We found that both the synthetic peptide disintegrin GRGDTP and echistatin effectively induced colony detachment and allowed for the passage of hESCs in our medium.

Combining these discoveries, a simple xeno-free and feeder-free culture system was developed incorporating ITS/Wnt3a/ID-8/DMEM/F-12 media, FLV-coated dishes, and disintegrin passaging. Ascorbic acid could be omitted in short-term initial screening; however, we decided to use 1 μg/ml ascorbic acid for additional long-term experiments because it was reported that ascorbate supports hESCs epigenetic status [31]. hESCs transferred into this culture system from normal serum replacement-containing medium were stable and required no culture adaptation. All three hESC lines examined (HES2, HES3, and H9) maintained undifferentiated morphologies for >24 passages (Fig. 3A, passage 17) in this system without bFGF supplementation. The cell proliferation and population doubling rate in these culture conditions without bFGF was slightly lower than that with bFGF (data not shown) or the control MEF-CM (supplemental information Fig. 3A, 3C). All pluripotent stem cells markers examined, including OCT4, SOX2, NANOG, SSEA3, SSEA4, TRA-1–60, TRA-1–81, GCTM-2, and ALP, were expressed in the cells (Fig. 3B). FACS analysis showed that the populations of TRA-1–60- or GCTM-2-positive cells were similar to those maintained under MEF-CM culture conditions (Fig. 3C). Gene expression levels of pluripotent stem cell markers, including OCT4, NANOG, SOX2, and DNMT3B, were not significantly different from those of cells maintained on feeder layers and/or in MEF-CM (Fig. 3D). Normal karyotypes were also maintained in these cells after 22 passages (Fig. 3E). After 17–24 passages, EB-mediated in vitro differentiation and immunostaining for the three germ layer cell markers were performed to evaluate the differentiation capacity of the cultured hESCs (Fig. 4A). These cells readily differentiated into derivatives expressing markers characteristic of all three germ layers, with ectodermal cells expressing βIII-tubulin and GFAP, mesodermal cells expressing αSMA and Desmin, endodermal cells expressing AFP, definitive endodermal cells expressing SOX17 but not expressing SOX7, and extraembryonic endodermal cells expressing SOX7 strongly and SOX17 weakly. In addition, after 22 passages, the cells formed teratomas containing derivatives of all three germ layers when injected subcutaneously into immunodeficient mice (Fig. 4B). In addition to the female hESC cell lines used in the experiments above, a male hESC line H1 [1] and a human dermal fibroblast-derived iPSC line [25] were examined in our culture system. Both cell line cells were maintained in the undifferentiated state for at least 15 passages (Fig. 3A; supplemental information Fig. 3A, 3C). These data suggest that our xeno-free, feeder-free culture system can be used to maintain diploid human pluripotent stem cells long-term.