Inhibition of the V-ATPase by Archazolid A: A New Strategy to Inhibit EMT

This article is featured in Highlights of This Issue, p. 2327

Tumor initiating cells (TIC) have tumor-initiating ability and show increased resistance to chemotherapeutics and, therefore, account for metastasis and tumor recurrence. Therefore, TICs limit therapeutic success and represent major problems in cancer therapy (1–3). TIC properties such as self-renewal and invasiveness have been closely associated with epithelial–mesenchymal transition (EMT), which is a biological process that confers a mesenchymal stem-like phenotype to cells: cancer cells become highly malignant and invasive, acquire self-renewal capacity, and develop elevated resistance toward therapeutics (4–6). Thus, EMT plays a pivotal role in cancer progression, relapse, and metastasis (7).

A hallmark of EMT is the change of structural proteins that maintain the cytoskeleton and cell–cell adhesions. Especially the loss of the transmembrane glycoprotein E-cadherin represents a crucial event during EMT as it results in breakdown of cell–cell contacts and accounts for increased motility of mesenchymal cells (8–10).

E-cadherin is controlled both by transcriptional processes and by endolysosomal internalization and degradation. Transcriptional repression of E-cadherin is mediated by transcription factors such as Snail1/2, ZEB1/2, Slug, and Twist1 (11–16). Internalization of E-cadherin from the cell surface into endosomes and its recycling back to the surface or its degradation in lysosomes is a dynamic process and ensures the formation of adherens junctions and thus cell adhesion (17, 18). The loss of E-cadherin is balanced by the increased expression of mesenchymal proteins such as N-cadherin, vimentin, and fibronectin.

Despite or maybe just because of the tremendous contribution of cancer stem cells (CSC) and EMT to metastasis, tumor recurrence, and therapeutic resistance, strategies for interfering with the EMT process are rare. Currently, only compounds that interfere with TGFβ such as galunisertib and fresolimumab get evaluated in clinical trials for glioblastoma, hepatocellular carcinoma, pancreatic cancer, and melanoma (19, 20). However, the actual evidence for EMT in clinical specimens is limited and there are no solid studies linking EMT to treatment outcomes or patient survival. Thus, finding novel strategies for targeting the EMT is essential and might contribute to find new strategies that can help to improve cancer therapy.

We hypothesized that Vacuolar H⁺-ATPase (V-ATPase) is a promising target to address the EMT. V-ATPase is heavily involved in the regulation of trafficking by acidifying endosomes and lysosomes. By regulating endolysosomal processes, the V-ATPase affects tumor cell hallmarks such as proliferation, cell death, and motility and pharmacologic inhibition of the V-ATPase has shown potent anti-cancer and anti-metastatic effects in vitro and in vivo (21–26).

Our goal was to characterize a potential function of the V-ATPase in EMT and to analyze the underlying mechanism in order to judge the inhibition of V-ATPase as a promising new strategy to address EMT.

Isopropyl-β-D-thiogalactopyranosid (IPTG), 4-hydroxytamoxifen (4-OH-TX), and poly-(2-hydroxyethyl methacrylate, poly-HEMA) were purchased from Sigma-Aldrich. TGFβ was purchased from PeproTech GmbH. Archazolid A and Concanamycin (27) are myxobacterial natural compound provided by Prof. Rolf Müller (Helmholtz Centre for Infection Research and Department of Pharmaceutical Biotechnology, Saarland University, Saarbrücken, Germany).

Immortalized human mammary epithelial (HMLE) cells transduced with Twist1-ER were kindly provided by Dr. Christina Scheel (Helmholtz Centre Munich, 2013) and described by Casas and colleagues (14). HMLE Twist1-ER cells were cultivated in Mammary Epithelial Cell Growth Medium (MECGM, Ready-to-use, PromoCell GmbH) supplemented with penicillin/streptomycin (PAA Laboratories), and 10 μg/mL blasticidin (Gibco) at 37°C and 5% CO₂.

The fusion protein of Twist1 and the modified hormone-binding domain of the estrogen receptor (Twist1-ER) was activated by treatment with 4-hydroxytamoxifen (4-OH-TX, 20 nmol/L) for 10 days. Before EMT induction, epithelial HMLE Twist1-ER cells were pretreated with Archazolid A (1 nmol/L or 10 nmol/L) for 24 hours and during EMT Archazolid A (0.1 nmol/L) was added. Mesenchymal HMLE Twist1-ER cells were treated with Archazolid A (1 or 10 nmol/L) for 24 hours.

Epithelial HMLE Twist1-ER cells were treated with TGFβ (5 ng/mL) for 12 days. Before EMT induction, epithelial HMLE Twist1-ER cells were pretreated with Archazolid A (1 nmol/L or 10 nmol/L) for 24 hours and during EMT Archazolid A (0.1 nmol/L) was added.

For the lentiviral transduction of HMLE Twist1-ER cells with V-ATPase shRNA MISSION Lentiviral Transduction Particles [Vector: pLKO.1-puro-IPTG 3xLacO; SHC332V-1EA; Clone ID: (1) TRCN0000029559, (3) TRCN0000029561, Sigma-Aldrich] and MISSION 3xLacO Inducible Non-Target shRNA Control Transduction Particles (SHC332V, Sigma-Aldrich) as a control were used according to the manufacturer's protocol. HMLE Twist1-ER cells were transduced with a multiplicity of infection (MOI) of one. Successfully transduced cells were selected by adding 0.5 μg/mL puromycin to the medium and puromycin was also added to the medium during cultivation. For shRNA expression 1 mmol/L IPTG was added for 4 days. In order to ensure V-ATPase downregulation, 1 mmol/L IPTG was constantly added.

Cell viability was measured according to Nicoletti and colleagues (28) by which the percentage of apoptotic nuclei after propidium iodide (PI; Sigma-Aldrich) staining is measured. Briefly, cells were treated as indicated, harvested, and PI staining was performed. Subdiploid DNA content was determined by flow cytometry (Becton Dickinson) and considered as apoptotic.

Assays were performed as previously described by Dontu and colleagues with modifications (29). In brief, cell viability was measured by Vi-CELL XR (Beckman Coulter) and 20,000 viable cells/well were seeded in ultra-low 12-well attachment plates, coated with poly-(2-hydroxyethyl methacrylate; poly-HEMA) in MEGM Mammary Epithelial Cell Growth Medium Bullet Kit (Lonza) containing B27 (Gibco), 20 ng/mL EGF (PeproTech), 10 ng/mL bFGF (Peprotech) and 1 % methylcellulose (Sigma-Aldrich). For secondary mammosphere formation, primary spheres were dissociated by Trypsin/EDTA and reseeded at 20,000 viable cells/well. After 7 days, each well was imaged by confocal microscopy (LSM 510 Meta, Zeiss), and quantity and size of mammospheres were analyzed by using ImageJ.

A total of 1 × 10⁵ cells were suspended in MECGM without FCS and added on top of the Transwell permeable supports (Corning Incorporated). MECGM (negative control, NC) or MECGM with 10% FCS (positive control, PC) was added to the bottom of the membrane. Boyden chambers were incubated at 37°C, 5% CO₂ for 6 hours. Migrated cells were stained with crystal violet. Cells on the upper side were removed with cotton swabs. Migrated cells were photographed using Zeiss Axiovert 25 microscope (Zeiss) and Canon 450D camera (Canon), counted and normalized to control cells.

Immunoblotting was performed as previously described (30). The following antibodies were used: E-cadherin, vimentin, claudin-1, ATP6V0C (NBP1-59654, Novus), fibronectin (Santa Cruz Biotechnology), HRP-goat anti-rabbit (Bio-Rad Laboratories GmbH), HRP-goat anti-mouse (Santa Cruz Biotechnology). Proteins were detected by using the ChemiDoc Touch Imaging System (Bio-Rad Laboratories GmbH).

Cells were treated as described in figure legends and seeded on μ-slides from ibidi (ibidi GmbH), stained for 60 minutes with 100 nmol/L LysoTracker dye (Moleculare Probes) and Hoechst 33342 (5 μg/mL; Sigma-Aldrich) at 37°C and imaged by confocal microscopy without fixation.

Total mRNA was isolated from cell culture samples using Qiagen RNeasy Mini Kit (Qiagen). Reverse transcription was performed with the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer's instructions. Real-time PCR was performed with the 7300 Real-Time PCR System (Applied Biosystems). SYBR Green Mix I (Applied Biosystems) was used for: E-cadherin (forward: 5′-CAG CAC GTA CAC AGC CCT AA-3′, reverse: 5′-AAG ATA CCG GGG GAC ACT CA-3′), vimentin (forward: 5′-CGG CGG GAC AGC AGG-3′, reverse: 5′-TCG TTG GTT AGC TGG TCC AC-3′), N-cadherin (forward: 5′-ACA GTG GCC ACC TAC AAA GG-3′, reverse: 5′-CCG AGA TGG GGT TGA TAA TG-3′), fibronectin (5′-GCT GAC AGA GAA GAT TCC CGA-3′, reverse: 5′-CCA GGG TGA TGC TTG GAG AA-3′) primer (Metabion) and TaqMan Universal PCR Mastermix (Life Technologies Corporation) was used for Taqman gene expression assay for V-ATPase Hs00798308_sH (Applied Biosystems). GAPDH or Actin were used as housekeeper. Obtained average CT values of target genes were normalized to control as ΔCT. Changes in expression levels were shown as fold expression (2^ΔΔCT) calculated by the ΔΔCT method (31).

Cells were treated as indicated, fixed with 4% PFA for 10 minutes, permeabilized with 0.2% Triton X-100, and blocked with 0.2% BSA in PBS. Primary antibodies were diluted in 0.2% BSA and incubated overnight at 4°C. Secondary antibodies were also diluted in 0.2% BSA and incubated for 45 minutes at room temperature. Subsequently, cells were mounted with FluorSave Reagent and analyzed with Leica-SP8 confocal microscope (Leica Microsystems). The following antibodies or dyes were used: E-cadherin, vimentin, N-cadherin (Cell Signaling), E-cadherin (HECD1, Invitrogen), Lamp-1 (Developmental Studies Hybridoma Bank), Hoechst 33342 (5 μg/mL; Sigma-Aldrich), Alexa Fluor 488 goat anti-rabbit, Alexa Fluor 546 (Invitrogen). Images were either taken using Leica-SP8 confocal microscope or Zeiss LSM 510 Meta confocal laser microscope.

HMLE Twist1-ER cells were cultivated as described for 10 days. On day 10, cells were reseeded into ibidi μ-slides, washed with ice-cold PBS, and slides were kept at 4°C for 10 minutes. In the following, the samples were incubated with anti-E-cadherin antibody (2 μg/mL; HECD1, Invitrogen) for 45 minutes at 4°C. After washing steps with PBS, MECGM was added and incubated for indicated time points (0, 5, 10, and 15 minutes) at 37°C. Cells were fixed with 4% paraformaldehyde (PFA) for 15 minutes at room temperature. After incubation with secondary antibody (1:400) and Hoechst 33342 (5 μg/mL) for 30 minutes, cells were washed, mounted with FluorSave Reagent (Merck) and confocal microscopy was performed.

Non-targeting and V-ATPase shRNA cells were treated with IPTG as described above. Cells were transfected with E-cadherin-eGFP (addgene 28009) and Rab5-RFP (addgene 14437) by using Dharmafect1 (Dharmacon) according to the manufacturer's instructions. Life-cell imaging was performed using a Leica-SP8 confocal microscope. Cells at 24 to 48hours posttransfection were measured for 500 seconds. Single vesicles were detected in each frame individually and then connected between frames with a minimum pair-wise distance algorithm. Mean squared displacements (MSD) were calculated for each track, fitted, and averaged with an active transport model (Eq. A). For the fit, the first half of the data points were used.

All experiments were performed at least three times unless otherwise indicated in the figure legend. Statistical analysis was performed using GraphPad Prism software version 5.04. Graph data represent means |\pm $| SEM. One-way ANOVA/Tukey multiple comparison test and individual Student t tests were conducted. P values less than 0.05 were considered statistically significant.

HMLEs overexpressing Twist1 fused to the ligand binding domain of a mutated estrogen receptor (ER) served as a model to study EMT (14, 32). Tamoxifen-mediated Twist activation in HMLE cells induced a mesenchymal phenotype and EMT (Supplementary Fig. S1A and S1B) and resulted in the formation of mammospheres (Supplementary Fig. S1C; ref. 32). Besides that, as other way to induce EMT, TGFβ treatment of HMLE cells was applied.

To investigate the function of V-ATPase in mammosphere formation, we generated inducible V-ATPase knockdown HMLE cells by stable transduction of HMLE cells with an IPTG inducible V-ATPase V₀ subunit c shRNA vector. IPTG treatment of V-ATPase shRNA HMLE cells induced V-ATPase downregulation as shown by RT-PCR (Fig. 1A). In addition, LysoTracker staining showed decreased acidification of the lysosomes of V-ATPase shRNA HMLE cells upon IPTG treatment (Fig. 1B). Of note, the mammosphere forming potential which was induced by Tamoxifen-mediated Twist activation was inhibited in the IPTG-treated V-ATPase shRNA knockdown HMLE cells compared with IPTG-treated HMLE cells transduced with non-targeting shRNA (Fig. 1C).

To judge the V-ATPase as a pharmacologically accessible target in mammosphere formation, epithelial HMLE cells were pretreated with Archazolid A (1 nmol/L and 10 nmol/L, 24 hours) before EMT was induced by Tamoxifen-driven Twist activation or TGFβ. During EMT induction (10 days), Archazolid A was present at a very low concentration (0.1 nmol/L) which did not affect cell viability or proliferation (25, 33, 34) and Supplementary Fig. S2). During the assays subsequent to EMT induction, Archazolid was removed, in order to ensure that specifically its influence on EMT was analyzed. Diminished acidification of lysosomes confirmed V-ATPase inhibition by Archazolid A (Supplementary Fig. S3). Of note, Archazolid A treatment of HMLE cells in their epithelial state decreased the formation mammospheres induced by tamoxifen-driven Twist activation as well as TGFβ (Fig. 2A and B). Moreover, Archazolid A treatment of epithelial cells before EMT induction reduced cell migration (Fig. 2C). This was not due to increased cell death as shown by Nicoletti assays (Supplementary Fig. S2A) as well as by staining of mammospheres with propidium iodide (PI) that indicates dead single cells but living mammosphere-forming cells regardless from Archazolid treatment (Fig. 2D). Importantly, as shown by repeated mammosphere assays, the effect of Archazolid A was sustained (Fig. 2E). This set of data shows that Archazolid A inhibits two crucial functional characteristics of tumor-initiating cells, namely migration and self-renewal. As Archazolid A was applied on epithelial cells before EMT induction, a function of V-ATPase in the EMT process is suggested.

So far, our data showed that mammosphere formation and migration are reduced when inhibition of V-ATPase is applied in epithelial cells before EMT. Besides interfering with EMT, it is of utmost clinical importance, to target already existing, poorly accessible mesenchymal cells. Therefore, Archazolid A was applied to fully transitioned mesenchymal HMLE cells that had already undergone EMT. In fact, Archazolid A inhibited both migration (Fig. 3A) and mammosphere formation (Fig. 3B) of fully transitioned mesenchymal HMLE cells. An apoptotic effect could be excluded (Supplementary Fig. S2B). Moreover, in order to verify that the effect of V-ATPase inhibition is not cell type specific, we used basal triple-negative MDA-MB-231 breast cancer cells. Archazolid was shown previously to reduce MDA-MB-231 lysosome acidification (24). In fact, V-ATPase inhibition both by Archazolid A (Fig. 3C) as well as by Concanamycin (Fig. 3D) reduced mammosphere formation of MDA-MB-231 cells, verifying that V-ATPase inhibition even is effective in cells with a high EMT/CSC phenotype.

To verify our hypothesis that V-ATPase inhibition interferes with EMT, expression of classical EMT markers was analyzed. Changes of self-renewal markers Sox2 and OCT-4—that only might be detectable after cell sorting—could not be observed in whole-cell populations (Supplementary Fig. S4). In fact, EMT induction by Tamoxifen-driven Twist activation or TGFβ decreased the expression of epithelial markers like E-cadherin, β-catenin, or claudin-1 and increased the expression of mesenchymal markers like vimentin, fibronectin, and N-cadherin (Fig. 4; Supplementary Fig. S4). Whereas mRNA expression was not modified by V-ATPase inhibition as shown by RT-PCR (Supplementary Fig. S5), Archazolid A strongly influenced the protein level of EMT markers: levels of the epithelial markers E-cadherin, β-catenin and claudin-1 were increased by Archazolid A whereas the expression of the mesenchymal markers fibronectin and vimentin was reduced as shown by Western blot analysis (Fig. 4A and B). These results were confirmed by immunostainings: the EMT-driven loss of the epithelial markers E-cadherin and β-catenin and the increase of the mesenchymal markers vimentin and N-cadherin were prevented both by Archazolid A (Fig. 4C and D) and V-ATPase knockdown (Fig. 4E).

As V-ATPase inhibition interfered with mammosphere formation of fully transitioned mesenchymal cells with EMT/CSC phenotype as well, we checked its effect on E-cadherin as a central EMT marker in MDA-MB-231 cells. Cytoplasmic E-cadherin localization was not changed by Archazolid in these cells (Fig. 4F), suggesting that different targets of V-ATPase are addressed by Archazolid in these cells.

In summary, this set of data suggests that V-ATPase inhibition preserved an epithelial phenotype and interfered with EMT.

In order to elucidate the underlying mechanism of V-ATPase in EMT, we focused on E-cadherin as the endosomal trafficking and recycling of E-cadherin represents a crucial process during EMT (35). We investigated whether V-ATPase inhibition influenced the localization of E-cadherin to endosomes by using Rab5 as an endosomal marker. In epithelial cells, E-cadherin was localized mainly to the membrane (Fig. 5A, left). In cells that had undergone EMT due to Tamoxifen-driven Twist activation, E-cadherin was diminished at cell–cell contacts and localized intracellularly (Fig. 5A, middle). Prevention of EMT by IPTG-induced V-ATPase knockdown resulted in localization of E-cadherin at the membrane and in enlarged Rab5-positive endosomes (Fig. 5A, right), suggesting that Archazolid A inhibits E-cadherin internalization and degradation. To get a more detailed insight into the impact of V-ATPase in E-cadherin recycling during EMT, which represents a highly dynamic process, surface-E-cadherin was analyzed by internalization assays. In epithelial HMLE Twist1-ER cells, E-cadherin was internalized and recycled back to the membrane (Fig. 5B, left). In mesenchymal cells that had undergone EMT, surface E-cadherin was internalized and the total concentration of the protein was decreased (Fig. 5B, middle). Archazolid A treatment before EMT induction inhibited E-cadherin internalization and resulted in accumulation of E-cadherin at the cell surface (Fig. 5B, right). Thus, cells treated with Archazolid A maintained epithelial characteristics. This was confirmed by live-cell–imaging experiments and subsequent single-particle tracking of E-cadherin containing vesicles. The average mean squared displacement (MSD) analysis of tracked E-Cadherin containing vesicles shows active endosomal transport processes. The velocities of E-cadherin–containing Rab5-positive vesicles were highly increased in cells that had undergone EMT (Fig. 6B and D; Supplementary Fig. S6) compared with epithelial cells without EMT induction (Fig. 6A and D; Supplementary Fig. S6). The velocities of E-cadherin containing vesicles were decreased in V-ATPase knockdown cells (Fig. 6C and D; Supplementary Fig. S6). In summary, our data suggest that Archazolid A-mediated inhibition of EMT is linked with a disturbed internalization and increased accumulation of E-cadherin at the cell surface.

This study introduces V-ATPase inhibition as a promising new strategy to interfere with EMT in cancer. During recent years, important functions of V-ATPase in cancer have been elucidated, like an implication of V-ATPase in tumor cell death, invasion, and metastasis (23–25,34,36). As a consequence, V-ATPase has emerged as a promising target for cancer therapy. However, only few reports point to a function of V-ATPase in TICs and the mode of action of V-ATPase to contribute to TIC regulation has not been elucidated until now. Inhibition of V-ATPase has been shown to eradicate TICs in rhabdomyosarcoma (37). Furthermore, knockdown as well as pharmacologic V-ATPase inhibition by bafilomycin A1, a first-generation V-ATPase inhibitor, diminished the neurosphere-forming ability of glioblastoma and suppressed the expression of stem cell markers (38). In line with this study, our results revealed a requirement of V-ATPase for mammosphere formation of mesenchymal HMLE cells, which was blocked by Archazolid A-mediated pharmacologic inhibition of V-ATPase. Moreover, investigating the underlying mechanism of V-ATPase in mammosphere formation, our study reveals that the inhibition of V-ATPase blocks the EMT process. So far, little is known about the function of V-ATPase in EMT: V-ATPase has been shown to promote EMT in kidney proximal tubular cells, suggesting an implication of V-ATPase in the pathogenesis of kidney tubulointerstitial fibrosis (39). In breast cancer, V-ATPase has been shown to contribute to DMXL-2–driven EMT via Notch signaling activation (40). However, the detailed function of V-ATPase, its mode of action, and the therapeutic effects of V-ATPase inhibition in EMT in cancer have not been investigated so far.

According to an increasing number of studies, EMT can lead to the generation of TICs and plays a central role in cancer metastasis. The induction of EMT in nontumorigenic mammary epithelial cells promotes a mesenchymal phenotype and induces TIC formation (6). EMT is associated with a poor clinical outcome in cancers such as bladder cancer, oral squamous cell carcinoma, ovarian cancer, cervical cancer, as well as breast cancer (41–45). Moreover, EMT is associated with therapy resistance in some cancers, including lung, pancreatic, and breast cancer (46–48). Thus, pharmacologic targeting of EMT represents an effective strategy to inhibit metastasis and TIC formation and might increase the vulnerability of cancer cells to chemotherapeutics. As a consequence, an increasing number of EMT- and TIC-targeting therapies have been developed during recent years but, unfortunately, only very few showed promising results in clinic. In detail, several Notch pathway inhibitors were clinically tested, but revealed cytotoxic effects. Only compounds which block TGFβ–induced EMT such as galunisertib and fresolimumab revealed successful anti-tumor activity in clinical trials (49, 50). According to our present study, inhibition of V-ATPase by Archazolid A might be judged as a possible new option to target EMT and the formation of tumor-initiating cells.

In search for the mechanism of how V-ATPase is involved in the EMT process, we focused on the cell-adhesion protein E-cadherin. Reduced cell–cell adhesion and changes in the cytoskeleton are essential features in EMT. Particularly the loss of E-cadherin represents a hallmark of EMT as it results in dissociation of cell–cell contacts in mesenchymal cells and accounts for metastasis (51). Besides genetic and epigenetic downregulation, endocytosis and recycling represent crucial posttranscriptional processes in the regulation of E-cadherin dynamics. Because of its ability to acidify endolysosomal compartments, V-ATPase has been associated with trafficking and recycling of receptors like the epidermal growth factor receptor (EGFR), or the transferrin receptor (23, 25). Furthermore, V-ATPase has been associated with endosomal trafficking of cell-adhesion proteins, including E-cadherin (39, 52, 53). In our study, we showed that V-ATPase inhibition could prevent the EMT-induced loss of E-cadherin, β-catenin, and claudin-1 and diminished the upregulation of mesenchymal markers like vimentin, fibronectin, and N-cadherin. In detail, V-ATPase inhibition interfered with E-cadherin internalization, recycling, and lysosomal degradation, which resulted in maintenance of E-cadherin at the cell surface, opposing a mesenchymal and preventing an epithelial phenotype.

Importantly, in line with previous studies, we show that besides interfering with the EMT process by disturbing E-cadherin, V-ATPase inhibition as well interferes with cells that already have an established EMT/CSC phenotype. Interestingly, this was not based on E-cadherin. By disturbing the endolysosomal system, V-ATPase inhibition addresses various targets including the Rho-GTPase Rac1 (25), integrin activation (24), iron metabolism (23), Notch (54), amongst others. Most probably, these targets contribute to the effects of Archazolid on cells with EMT/CSC phenotype.

Taken together, our finding that V-ATPase inhibition prevents E-cadherin loss upon EMT induction and thereby inhibits malignant TIC formation evokes new options for interfering with EMT in cancer therapy. To conclude, our study provides evidence for V-ATPase inhibition by Archazolid A as a promising new strategy to block EMT and suggests investigating V-ATPase inhibition as a potential option to interfere with TICs in cancer therapy.

No potential conflicts of interest were disclosed.

Conception and design: H. Merk, A.M. Vollmar, J. Pachmayr

Development of methodology: H. Merk, S. Zahler, A.M. Vollmar, J. Pachmayr

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.K. Messer, M.A. Ardelt, S. Zahler

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Merk, P.K. Messer, M.A. Ardelt, D. Lamb, S. Zahler, J. Pachmayr

Writing, review, and/or revision of the manuscript: H. Merk, P.K. Messer, R. Müller, A.M. Vollmar, J. Pachmayr

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Müller, J. Pachmayr

Study supervision: J. Pachmayr

We thank Dr. Christina Scheel (Helmholtz-Center Munich, Institute of Stem Cell Research) for providing the HMLE cells. We thank Silvia Schnegg and Julia Blenninger for their kind help with the experiments.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.