The NSL Chromatin-Modifying Complex Subunit KANSL2 Regulates Cancer Stem–like Properties in Glioblastoma That Contribute to Tumorigenesis

Glioblastoma (GBM) is among the most frequent, aggressive, and lethal tumor types within the central nervous system (CNS; refs. 1, 2) and for which there is no effective treatment. Thus, identification of critical signaling pathways involved in GBM progression may enable the development of new diagnostic and therapeutic strategies. GBM stem cells (GBSC) are small neoplastic cell populations with the ability to self-renew and eventually differentiate to glial and neural cell types (3–5). GBSCs have enhanced tumor initiation capacity and are able to regenerate the heterogeneous cell population observed in GBM, contributing to tumor maintenance and recurrence after treatment (6).

Many of the key signaling pathways involved in embryonic stem cells (ESC) identity, including core pluripotency factors and epigenetic regulators, are also functional in GBSCs (7–10). Expression of POU5F1 (Oct4), an ESC factor, is critical for their stemness in vitro and tumorigenicity in vivo (11–14). In accordance, GBSCs can grow as multipotent clonal spheres, called gliomaspheres, which exhibit most of the biological and pathologic characteristics of cancer stem cells (CSC; refs. 3, 6, 15). We hypothesized that stemness genes are involved in dysregulation of cell plasticity events in GBM. We performed an in silico search to mine publicly available mRNA expression and ChIP-seq data (16–19) using the web tool INSECT (In silico search for co-occurring transcription factors; ref. 20), to identify genes that are commonly expressed both in pluripotent stem cells and CSCs, exhibiting potential binding sites for POU5F1 in their regulatory region. Using this approach, we identified KANSL2 [KAT8 regulatory nonspecific lethal (NSL) complex subunit 2] as a strong candidate gene. The KANSL protein family belongs to the lysine acetyl-transferase KAT8/MOF-NSL complex, and its function has been linked to pluripotency and cellular homeostasis in ESCs (21–23). KANSL2 is a poorly understood member of the NSL complex family, and its role has not been previously explored in glioblastoma.

Here, we characterized the role of KANSL2 in GBM cells. KANSL2 is upregulated in glioma samples and expressed mainly in perivascular regions and in discrete foci within tissues. KANSL2 upregulation was also observed in patient-derived GBM cell lines showing stem cell features with increased POU5F1, NANOG, NESTIN, and CD133 expression. Importantly, we determined that KANSL2 expression is critical for stem cell properties of GBM cells, as KANSL2 depletion reduced neurosphere formation, POU5F1 expression, and tumorigenesis. Moreover, KANSL2 and POU5F1 enforce expression of each other, making a regulatory feedback loop that might control stemness properties in this cancer type.

Unless otherwise stated, reagents were obtained from Life Technologies or Sigma Chemical Co.

Samples obtained from the tumor bank at FLENI Hospital were subjected to histologic diagnosis by experienced neuropathologists and collected with informed consent according to the hospital's Institutional Review Board (in compliance with the October 2013 Helsinki Declaration). Patient and tumor sample information is listed in Table 1. 

Microarray analysis was performed on a cohort containing 52 gliomas specimens as described (Accession Number: E-MTAB-4455; Supplementary Table S1; ref. 24). The Fisher exact test was used for statistical analysis of gene expression, cut-off P value ≤ 0.05 (24).

Biopsies were conducted as previously described (25). For immunohistochemistry staining with Leica Bond Max automated Stainer, the following antibodies were used: anti-KANSL2 (Sigma-Aldrich HPA038497, specificity confirmed by shRNA-mediated knockdown), anti-KAT8 (Santa Cruz Biotechnology INC sc-271691), and anti-AcH4K16 (Abcam ab109463; Table 1; refs. 26, 27). Images were acquired with a NikonDXN1200F digital camera controlled by EclipseNet software (version1.20.0 build 61). Unbiased stereological analysis was also used to quantify anti–KANSL2- and anti–Ki-67- (NCL-Ki67-MM1; Novocastra Laboratories) labeled elements in tumor samples using Stereo Investigator Optical Dissector software (MBF Biosciences; MicroBrightField, Inc.). Necrotic areas were also examined (additional experimental information).

Patient-derived stem cells, G03 and G08, were established from human GBM grade IV biopsies, identified, and characterized previously (25). Briefly, cells with the same genomic alterations as parental tumors were cultured in neural stem cell (NSC) medium plus supplements (25) and plated onto laminin-coated plates. Their cellular hierarchy and plasticity (markers expression, differentiation, and tumorigenic potential) were previously characterized (25). Cell lines U87MG, T98G, LN299, HEK293T, C6, P19, and WA09 (H9) were acquired from the ATCC or WiCell Research Institute, either directly or by colleagues, kept frozen immediately after receipt or used in culture less than 4 months. ATCC cell lines were characterized by short tandem repeat (STR) profiling and WiCell lines by testing established standards for ECSs culture and G-band karyotype. For neurosphere induction, GBM cells were grown to 90% confluence, trypsinized, and plated in NSC medium in ultra-low adhesion multi-well plates (Corning). After 5 days, the number of spheres was quantified using 10x magnifications under a phase contrast microscope (Carl-Zeiss; AxioObserverZ1), an AxioCam(HRm) camera (Carl-Zeiss) and Zen pro2011, and later collected for RNA analysis. For in vitro propagation, spheres were collected by gentle centrifugation, dissociated to single cells, and cultured to produce the next generation of spheres.

cDNA encoding the murine KANSL2 (Gene ID: 69612) was amplified from pluripotent P19 embryonal carcinoma cells with the specific primers: Forward 5-GACCATGAACAGGATTCGGA-3 and Reverse 5-ACCGGTGGACTGATAGAAGTGGG-3, containing EcoRI and Age I sites for cloning into pGEMT vector (Promega). Insert was subcloned into pTagRFP-N (Evrogen) to generate KANSL2-RFP.

Cells were incubated with anti–CD133/1 (AC133)-phycoerythrin (PE) conjugate antibody (130-080-801; Miltenyi Biotec; ref. 25). Data were acquired on a FACSCantoII instrument (BD Biosciences) and analyzed using FlowJo software version10. The isotype control sample was used to establish a gate in the PE channel. Cells showing signal for CD133 above the gate established were deemed to be CD-positive cells. Jazz Cell Sorter (BD Biosciences) was used for the analysis under the settings of "1.5 Drop Pure" from "FACS Software."

Total RNA was extracted by TRIzol following the manufacturer's instructions. cDNA was synthesized using MMLV reverse transcriptase (Promega). Real-time PCR was performed using the Bio-Rad CFX96 Touch Real-Time PCR Detection System and a Real Supermix Kit (Bio-Rad). RPL19 or GAPDH were used as normalization controls. Relative expression was calculated with the 2^−ΔΔCT method (28). Averages of three independent experiments ± SEM are shown. Primers are listed in Supplementary Table S2.

Knockdown cell lines were generated using Sigma Mission shRNA lentiviral plasmids. Lentiviral particles were produced in HEK293T cells by cotransfection of the shRNA vector and lentiviral helper plasmids. To establish stable cell lines, monolayers of different GBM cell lines were subjected to two rounds of infection. Stable control and specific knockdown pools were selected and maintained with puromycin (2.5 μg/mL). Mission pLKO.5-puro Non-Target shRNA plasmid (#SHC202) was used as a control. Knockdown efficiency was confirmed by qRT-PCR and Western blotting. Target sequences are listed in Supplementary Table S3.

HEK293 cells (3 × 10⁵ cells/well) were seeded onto 12-well plates and cotransfected with Lipofectamine 2000 with pKANSL2-RFP, pLM-vexGFP-Oct4 (a gift from Michel Sadelain, Addgene plasmid #22240), β-galactosidase (RSV β-galactosidase), phOCT4 (a gift from Shinya Yamanaka, Addgene plasmid #17221), or pNANOG-Luc (a gift from Ren-he Xu, Addgene plasmid #25900; 500:250:500 ng, respectively). After 24 hours, cells were lysed, and luciferase and β-galactosidase (β-Gal) activities were measured (Promega). To calculate transcriptional activity, each value was normalized to that of β-Gal, and expressed by the mean ± SEM.

Cells were lysed in RIPA with 1% Triton X-100 and a protease inhibitor cocktail (Roche). Protein samples were separated by SDS-PAGE, blotted onto Immobilon-P PVDF membrane (Millipore), and probed with antibody. Primary antibodies specific to KANSL2 (Sigma-Aldrich; HPA038497), KAT8 (Santa Cruz Biotechnology; INC, sc-271691), AcH4K16 (Abcam; ab109463), H3 (Cell Signaling Technology), POU5F1 (Abcam; ab19857), and GAPDH (Abcam; ab8245) were used. Blots were incubated with goat anti-rabbit or anti-mouse secondary antibody (BioRad Life Science) and visualized using ECL (Supersignal; Thermo Fisher Scientific).

All cultures were passaged by mechanical dissociation of spheres and seeded in quadruplicate into 96 wells at a density of 2 × 10³ cells/well. After 72 hours, cell growth was measured by direct counting or with the MTT-based CellTiter96 Aqueous One Solution Cell Proliferation Assay Kit (Promega).

Cells were dissociated and plated at 1, 10, 25, 50, 100, and 200 cells/well in NSC medium into a 96-well plate. Between 5 and 7 days after plating, the number of neurospheres found in each well was quantified under the microscope. Tumor-initiating cell (TIC) frequency and P values were calculated using ELDA software (29).

U87MG cells (1 × 10⁴) were plated in soft agar, each well contained 2 mL of 0.6% agar and NSC medium 2X, and then covered with 2 mL of 0.3% agar and NSC medium. Fresh NSC medium was added twice a week. After 3 weeks, wells were fixed with 4% paraformaldehyde and stained with 0.005% crystal violet.

NODscid mice (The Jackson Laboratory) of 6 to 8 weeks of age according to NIH guidelines were injected subcutaneously on the right dorsum (∼5 mice/group) with 2 × 10⁶ U87MG cells or 1 × 10⁶ LN229 cells. For intracranial tumor development, cells were stereotaxically injected (0.7–1 mm posterior, 2 mm left lateral, 3.5 mm in depth from the dura; ref. 30). Survival was assessed by Kaplan–Meier analysis and long-rank testing.

Data were analyzed using Prism GraphPAD v6.0 and presented as the mean ± SEM from three independent experiments. Two-tailed Student t tests and one-way ANOVA were used to define statistical significance (*, P ≤ 0.05; **, P ≤ 0.005; and ***, P ≤ 0.0005). Progenitor frequencies from limiting dilution assays were determined using the software tool (ELDA; ref. 29).

In order to identify novel genes expressed in stem cells and potentially regulated by POU5F1 in human and mouse, we employed the bioinformatics tool INSECT (20). Among the top scoring genes, we found KANSL2 as a potential candidate (Supplementary Fig. S1A–S1D). KANSL2 (previously known as C12orf41) is a member of the NSL complex expressed in ESCs (31–34), also reported to be upregulated in Ntera-2 human pluripotent embryonic carcinoma cells. This cell line shows ESC features and is considered to be a malignant counterpart of human embryonic cells (18, 19, 35, 36). We first confirmed by qRT-PCR that KANSL2 is indeed expressed in embryonic and carcinoma stem cells and its expression is reduced during differentiation to embryoid bodies (EB; Supplementary Fig. S1E and S1F). It has been reported that GBSCs display neural stem/progenitor cells properties (37, 38). We determined that KANSL2 is expressed in human neuronal progenitors (NP; Supplementary Fig. S1E). Next, we analyzed KANSL2 expression in high-grade glioma samples and normal human cortex by qRT-PCR, observing KANSL2 expression significantly enriched in tumor samples (∼7 fold; Fig. 1A). POU5F1 mRNA was also highly expressed in the same human tissue samples (Fig. 1A; refs. 12, 13).

To further evaluate KANSL2 expression in GBM samples, we analyzed KANSL2 protein distribution and accumulation by IHC staining. All tumor samples were tested in duplicate, and individual sections were scored and quantified (Table 1). We did not observe signal for KANSL2 expression in any of the normal tissues examined (8/8 cases). In contrast, tumor samples displayed consistently high KANSL2 expression (7/7; Fig. 1B; and Table 1). Interestingly, we detected very visible KANSL2 protein signals in tumor blood vessels and perivascular tumor cells (nuclear and cytoplasmic), as well as isolated tumor cells (Fig. 1B). KANSL2 expression followed a gradient from the perivascular zone (described to be enriched for cancer cells with stem cell features; refs. 15, 39–41) toward the main body of the tumor (Fig. 1C). In addition, enhanced KANSL2 expression in the tumor tissue was confirmed in a patient sample by analyzing the tumor area versus the contralateral healthy section from the same individual, observing low KANSL2 signal in a few isolated endothelial-like cells (Fig. 1C).

Next, we assessed whether KANSL2 overexpression was correlated with tumor grade by analyzing KANSL2 expression in 52 glioma tissue samples for which RNA expression profiling was available (24). We found no differences of KANSL2 expression between high (IDH nonmutated) and low (IDH mutated) grade tumors (Fig. 1D).

The KAT8/MOF/MYST acetyltransferase complex, which targets lysine 16 in histone H4 (AcH4K16), was previously reported to be expressed in mouse ESCs and NPs (21, 23). Along with the AcH4K16 marks, we determined the expression of KAT8 in GBM patient samples. KAT8 mRNA levels ranged among tumor samples with no significant differences relative to normal cortex or the tumor grade (Fig. 1E and F). At the protein level, KAT8 expression was detected in 43% of the samples (3/7; Fig. 1G and Table 1). Although KAT8 expression is variable, it seems to have a moderate correlation with AcH4K16 staining (Fig. 1H and Table 1; refs. 42, 43).

The correlation of KANSL2 enrichment by qRT-PCR data, protein staining pattern in high-grade tumors (i.e., tumor grades III and IV vs. normal cortex), and the presence of KANSL2 staining in areas related to the perivascular niches suggested a potential role of KANSL2 in GBM, in particular, in GBSCs. Therefore, we tested the impact of KANSL2 knockdown on the behavior of GBM cells lines.

To gain further insight into the pathophysiologic role of KANSL2 expression in GBM cells, we stably knocked down KANSL2 in human GBM U87MG, T98G, and LN299 cell lines, using two independent shRNAs (referred hereto as KD-K2-1 and KD-K2-2), targeting the coding region and 3′ untranslated region of the KANSL2 mRNA, respectively. A nontargeting hairpin (NT) was used as control, and knockdown efficiency was validated in KD-K2 cells (Fig. 2A and B; Supplementary Fig. S2A and S2B). KD-K2-U87MG cells or NT-U87MG cells were subcutaneously injected into each side flank of NODscid mice, and tumor growth rate was measured. Tumors derived from KD-K2-U87MG cells were significantly smaller compared with controls. The inhibitory effect on tumor growth was observed with both KANSL2 hairpins, minimizing the chances of off-target effects (Fig. 2C–H). To confirm these results, we also evaluated KD-K2 cells derived from the human GBM cell line LN299 (KD-K2 LN299 cells), which displayed a similar inhibition of tumor growth (Supplementary Fig. S2). We also examined the ability of the KD-K2 cells to generate orthotopic tumors. For these studies, we inoculated cells into the forebrain of 7-week-old NODscid mice. Consistently with the in vitro data and the xenotumors at the flank of the mice, depletion of KANSL2 reduced the ability to form tumors in the inoculated brains (Fig. 2I). No changes were observed in the number of Ki-67–positive tumor cells (or in the necrosis percentage); however, the survival of the mice inoculated with KANSL2-depleted cells was significantly prolonged compared with mice injected with control glioma cells (Fig. 2J and K).

Based upon the observation that KANSL2 is overexpressed in patient tumor samples and its expression is higher at the perivascular zones, we hypothesized that KANSL2 regulates stemness properties in GBM. To test this, we evaluated the impact of KANSL2 depletion on stemness-associated markers. We observed reduced POU5F1, NANOG, and NESTIN mRNA expression in adherent monolayers of KD-K2 cells (Fig. 3A). GBM with reduced POU5F1 expression would indicate decreased stemness with the concomitant reduction of NESTIN expression (8, 12, 13, 44), suggesting a potential role of KANSL2 in GBM stemness properties. Accordingly, we observed an increased expression of the glial marker GFAP and the neural marker TUBB3 in KD-K2 cells (Fig. 3B), indicative of cell differentiation (8, 12, 13).

Previous reports described that KAT8 expression is essential for pluripotency in ESCs (21) and that KAT8 overexpression and H4K16 acetylation by KAT8 are hallmarks of embryogenesis and oncogenesis (42, 43). We did not find differences in KAT8 mRNA expression in human GBM samples; however, we observed overall higher staining in KAT8 and AcH4K16 in GBM tissues relative to normal tissue. The KAT8 mRNA–level reduction observed in KD-K2 cells (Fig. 3A) and a robust decrease in H4K16 acetylation could be explained by reduced KAT8 activity (Supplementary Fig. S3B; refs. 21, 42).

To further evaluate the potential link of KANSL2 and the GBM-associated stemness properties, we analyzed GBSCs sphere cultures (Fig. 3C and D). In agreement with the fact that KANSL2 expression is upregulated in GBM cells (Fig. 1), we detected upregulation of KANSL2 expression in sphere cultures, along with increased expression of the stem cell markers POU5F1, NANOG, SOX2, and NESTIN (Fig. 3E and F). Higher expression of KAT8 mRNA was also noticed in these sphere cultures (Fig. 3E and F). To further confirm that U87MG-derived spheres are indeed enriched in GBSCs, we compared their tumorigenic potential. U87MG spheres were more aggressive and lethal in intracranial xenograft assays compared with monolayers after brain inoculation of 10⁵ cells (Fig. 3G). Furthermore, there was a significant delay in the animal survival rate that was comparable with the monolayer cultures, when the spheres culture cell number injected was reduce to 10⁴ cells (Fig. 3E).

To further confirm KANSL2 expression in GBSCs and acknowledging potential limitations in the cellular heterogeneity of commercial cell lines, we sorted U87MG cells and GBM patient–derived cells, named G03 and G08, previously characterized as GBM cells enriched in GBSCs (25), with different levels of CD133 expression, an established GBSCs marker (Fig. 3H–M; refs. 25, 30). Consistently, CD133 expression increased in U87MG spheres compared with monolayers, being KANSL2 expression significantly higher in CD133-positive subpopulation cells (CD133⁺) than in the CD133-negative subpopulation (CD133⁻), in all sorted GBSCs (Fig. 3H–M). In addition, increased expression of NANOG and POU5F1 was detected in CD133⁺ cells concordantly with their increased self-renewal capacity (Supplementary Fig. S4). Altogether, these data indicate that KANSL2 is highly expressed in GBM cells with stem cell–like features, suggesting a common upregulation mechanism during culture conditions requiring “stemness” properties (Fig. 3).

Next, we evaluated the potential role of KANSL2 in the regulation of self-renewal/differentiation properties of GBM cells. The serial neurosphere formation assay tests the self-renewal capacity of neuronal stem cells and brain tumor stem cells (29, 45). Using neurosphere formation and the limiting dilution assays in suspension, we observed that the number of KD-K2–derived spheres was significantly lower compared with NT-derived spheres (Fig. 4 A–C). To confirm our results, clonogenic efficiency of KD-KANSL2 cells was measured using soft-agar media under stemness conditions (30). Indeed, we found that KD-K2 cells have decreased clonogenicity compared with the control (Fig. 4E and F). Furthermore, silencing of KANSL2 in GBSC-enriched cells derived from sphere showed significant reduction of their growth in in vitro (Fig. 4D; ref. 30).

KANSL2 was overexpressed in tumor GBM samples, and it is also enriched in GBSCs. Accordingly, silencing of KANSL2 reduces expression of embryonic stem factors POU5F1 and NANOG and the self-renewal capacity of the cells (Figs. 3 and 4). To confirm the effect of KANSL2 in regulating stem cell factors, we transiently transfected the human GBM and HEK293 cells with either RFP- or the mouse version of Kansl2-RFP (pKansl2-RFP), which shares 95% identity with the human KANSL2 (Fig. 5A–C). KANSL2-forced expression significantly increased POU5F1 and NANOG promoter activities (Fig. 5D and E). In agreement with the reporter assays, transient overexpression of KANSL2 increased the endogenous expression of NANOG and POU5F1 in U87MG and in patient-derived G03 cells.

To further understand the molecular mechanism of KANSL2 function on stemness, we assessed the effect of POU5F1 on KANSL2 expression. POU5F1 expression was stably silenced with two independent shRNAs (referred as KD-POU5F1-1 and KD-POU5F1-2; Fig. 6A and B). As previously reported, downregulation of POU5F1 reduced the ability of human GBM cells to form gliomaspheres and decreased the expression of the neural progenitor NESTIN (Fig. 6C and D; ref. 12). In agreement with our in silico prediction, POU5F1 depletion led to lower KANSL2 expression (Fig. 6D and E). Altogether, these results suggest that POU5F1 and KANSL2 regulate each other in a positive feedback mechanism to control the stemness properties of GBM cells. Therefore, our work reveals a novel role for KANSL2 in the etiology and progression of GBM.

Glioblastoma is enriched in stem cell properties, which has been associated with the aggressiveness, therapy resistance, and recurrence of this tumor type (1–5, 46). Recently, GBM progression was associated with chromatin regulators and a core set of transcription factors that regulate cell fate commitment (47). CSCs, and particularly GBSCs, express many markers found in ESCs during normal development, which may drive tumor initiation and renewal. Using bioinformatic approaches, we identified genes involved in the regulation of SC transcriptional networks (20), particularly in CSCs, which led us to identify KANSL2 as a potential candidate.

In this report, we investigated the role of KANSL2 in GBM. We found that KANSL2 expression is increased in tumor cells regardless of glioma grade. Using loss-of-function approaches in GBM cell lines and GBSC-enriched spheres, we determined that KANSL2 plays a tumorigenic role, promoting the clonogenicity and self-renewal capacity in vitro, and driving tumor growth in vivo, revealing a requirement for KANSL2 expression during GBM tumor growth.

The mechanisms governing KANSL2 expression in gliomas remain to be elucidated. However, we could determine that KANSL2 expression is dependent on POU5F1, and that POU5F1 is regulated by KANSL2 in human GBM cells, and both genes were upregulated under culture conditions requiring stemness. POU5F1 (as well as SOX2 and NANOG; ref. 8) is essential pluripotency factor, is critical for GBM stemness, and regulates gliomagenesis through several pathways (8, 13, 44, 48). Our findings indicate that KANSL2 and POU5F1 are part of a common transcriptional network controlling stemness in this tumor type. In support, KANSL2 and POU5F1 expression profiles are similar among different human glioma tissues, GBSCs cells, and embryonic cells, suggesting that both genes are either controlled by the same transcriptional regulatory program, are functionally related, and/or are members of the same pathway or protein complex. Furthermore, utilizing the luciferase promoter-reporter assays, we revealed a regulatory link (either direct or indirect) between KANSL2 and POU5F1.

KANSL2, as a member of the NSL complex (31–33), was linked to the control of cellular pluripotency, to be required for global acetylation, including H4K16 acetylation in mouse ESCs, and for stem cell proliferation (21, 23). Furthermore, modulation of KAT8 expression was reported during mouse ES cell differentiation into NPs, whereas other KANSL proteins (i.e., KANSL1 and KANSL3) remained unaffected. Despite some discrepancies about the exact contribution of each KAT8-associated complex to stemness in mouse ESCs, their role in human ESCs, neural progenitors and/or during neurogenesis, or cancer has not been previously investigated (43). Similar to pluripotency factors, we observed that KANSL2 depletion led to a reduced expression of stem cell markers while increasing expression of lineage-specific markers such as GFAP and TUBB3, indicating that KANSL2 indeed could function as an inhibitor of cellular differentiation.

It has been proposed that overexpression of members of the KAT8-associated complex could drive tumor formation (42, 43, 49). Therefore, it is possible that KANSL2 overexpression in GBM and in particular in the GBSCs population affects the stoichiometry of these complexes, resulting in a dysfunctional complex. Indeed, we were unable to select glioblastoma cells stably overexpressing KANSL2. However, transient KANSL2 overexpression increased NANOG and POUF51 expression. We posit that KANSL2 expression levels and function could be influenced not only by its interaction partners but also by environmental inputs, having therefore an impact on the dynamic equilibrium of cell heritage identity. In addition, we observed that KANSL2 depletion reduces AcH4K16, which could be explained by reduced catalytic activity of the NSL-complex or by reduced expression of KAT8. KANSL2 has been reported to interact with another member of the KAT8-associated complex, WDR5, a promoter of self-renewal in ES and tumor bladder cells (32, 49), where POU5F1 expression level has been also proposed as a potential prognostic tumor marker (50).

The identification of novel genes involved in cancer progression through de-regulation of cell plasticity events that can regulate CSCs, and in particular GBSCs, is of clinical importance in GBM progression. It would be interesting to investigate whether KANSL2 plays a role as a regulator of stem capacity in other type of tumors where POU5F1 sustains stemness. In addition, in order to expand the generality of our observations, it should be important to conduct future studies using more patient-derived cells, because they better represent the tumor biology and heterogeneity encountered daily in the clinic.

Altogether, our data indicate that KANSL2 gene expression might be considered as a potential marker for GBM cells associated with stem cell properties and could be explored as a prognostic marker in this cancer type.

No potential conflicts of interest were disclosed.

Conception and design: C. Perez-Castro

Development of methodology: N.E. Ferreyra Solari, F.S. Belforte, L. Canedo, G.A. Videla-Richardson, J.M. Espinosa, M.A. Riudavets, C. Perez-Castro

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N.E. Ferreyra Solari, F.S. Belforte, G.A. Videla-Richardson, E. Serna, M.A. Riudavets, H. Martinetto, C. Perez-Castro

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N.E. Ferreyra Solari, F.S. Belforte, L. Canedo, E. Serna, M.A. Riudavets, H. Martinetto, C. Perez-Castro

Writing, review, and/or revision of the manuscript: L. Canedo, J.M. Espinosa, M. Rossi, G. Sevlever, C. Perez-Castro

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.M. Espinosa, C. Perez-Castro

Study supervision: C. Perez-Castro

We thank Ken Kobayashi (AGBT, FBMC-FCEN, IBBEA CONICET-UBA, Buenos Aires, Argentina) for critical reading, and Marcelo Schultz (FLENI, Buenos Aires, Argentina) and IBioBA members (IBioBA-MPSM-CONICET, Buenos Aires, Argentina) for their technical assistance.

This work was supported by grants from Agencia Nacional de Promoción Científica y Técnica, Argentina, Consejo Nacional de Investigaciones Científicas y Técnicas, FOCEM-Mercosur (COF 03/11), NIH grants R01CA117907 and 5P30CA046934, and The Pew Latin American Fellows Program for the repatriation award. N.E. Ferreyra Solari and L. Canedo were financially supported by Bunge & Born Foundation and the Argentinian Instituto Nacional del Cáncer (INC).

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.