Hypoxia-inducible factor-1a blocks differentiation of malignant gliomas Huimin Lu 1, *, Yan Li 1,2, *, Minfeng Shu 1 , Jianjun Tang 1 , Yijun Huang 1 , Yuxi Zhou 1 , Yingjie Liang 3 and Guangmei Yan 1 1 Department of Pharmacology, Zhongshan School of Medicine, Sun Yat-Sen University, Guangzhou, China 2 Department of Infectious Diseases, Third Affiliated Hospital of Sun Yat-Sen University, Guangzhou, China 3 Department of Pathology, First Affiliated Hospital of Sun Yat-Sen University, Guangzhou, China Introduction Deviation from the tissue ⁄ lineage-specific differentia- tion program is one of the fundamental aspects of tumorigenesis [1]. The aberrantly differentiated cells show abnormal growth characteristics and distinct Keywords CREB-binding protein (CBP) ⁄ p300; cobalt chloride; differentiation; HIF-1a; malignant gliomas Correspondence G. Yan, Department of Pharmacology, Zhongshan School of Medicine, Sun Yat-Sen University, 74 Zhongshan Road II, Guangzhou, 510080, China Fax: (86) 20-87330578 Tel: (86) 20-87333258 E-mail: ygm@mail.sysu.edu.cn *These authors contributed equally to this work (Received 26 August 2009, revised 2 October 2009, accepted 14 October 2009) doi:10.1111/j.1742-4658.2009.07441.x Aberrant differentiation is a characteristic feature of neoplastic transforma- tion, while hypoxia in solid tumors is believed to be linked to aggressive behavior and poor prognosis. However, the possible relationship between hypoxia and differentiation in malignancies remains poorly defined. Here we show that rat C6 and primary human malignant glioma cells can be induced to differentiate into astrocytes by the well-known adenylate cyclase activator forskolin. However, hypoxia-inducible factor-1a expression stimu- lated by the hypoxia mimetics cobalt chloride or deferoxamine blocks this differentiation and this effectiveness is reversible upon withdrawal of the hypoxia mimetics. Importantly, knockdown of hypoxia inducible factor-1a by RNA interference restores the differentiation capabilities of the cells, even in the presence of cobalt chloride, whereas stabilization of hypoxia- inducible factor-1a through retarded ubiquitination by von Hippel-Lindau tumor suppressor gene silence abrogates the induced differentiation. More- over, targeting of HIF-1 using chetomin, a disrupter of HIF-1 binding to its transcriptional co-activator CREB-binding protein (CBP) ⁄ p300, abol- ishes the differentiation-inhibitory effect of hypoxia-inducible factor-1a. Administration of chetomin in combination with forskolin significantly suppresses malignant glioma growth in an in vivo xenograft model. Analy- sis of 95 human glioma tissues revealed an increase of hypoxia-inducible factor-1a protein expression with progressing tumor grade. Taken together, these findings suggest a key signal transduction pathway involving hypoxia-inducible factor-1a that contributes to a differentiation defect in malignant gliomas and sheds new light on the differentiation therapy of solid tumors by targeting hypoxia-inducible factor-1a. Structured digital abstract l MINT-7292117: CBP (uniprotkb:Q6JHU9) physically interacts (MI:0915) with Hif1a (uni- protkb: O35800)byanti bait coimmunoprecipitation (MI:0006) Abbreviations CBP, CREB-binding protein; CREB, cAMP-responsive element binding protein; DFO, deferoxamine; FBS, fetal bovine serum; GFAP, glial fibrillary acid protein; Glut-1, glucose transporter-1; HIF-1, hypoxia inducible factor-1; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide; PCNA, proliferating cell nuclear antigen; pVHL, von Hippel-Lindau tumor suppressor protein; siRNA, small interfering RNA; VHL, von Hippel-Lindau; VEGF, vascular endothelial growth factor. FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS 7291 invasive and metastatic properties [2]. Treating malig- nant tumors through the induction of cell differentia- tion has been an attractive concept, but clinical development of differentiation-inducing agents, espe- cially for solid tumors, has been limited to date [3]. A successful example of differentiation therapy is the clin- ical use of all-trans retinoic acid for the treatment of acute promyelocytic leukemia [4]. Several cell lines established from solid tumors were also reported to be differentiated by respective differentiation-inducers in vitro [5–8]. However, this differentiation has not been verified in in vivo animal models or clinically, and there exists little convincing explanation for this finding. Solid tumors frequently develop in regions with hypoxia because of an imbalance in oxygen supply and consumption. Recent reports indicate that hypoxic microenvironments contribute to cancer progression by activating adaptive transcriptional programs that pro- mote cell survival, motility and tumor angiogenesis [9,10]. Histopathological analyses frequently reveal the spatial overlap of hypoxia and dedifferentiation within solid tumors, suggesting the role of hypoxia in tumor cell differentiation [11,12]. However, it is unclear whether hypoxia plays a causal role in this relation- ship. Cells within hypoxic regions adapt to this environ- ment by altering their gene-expression program, and thus their phenotype [3]. One of the transcription fac- tors primarily responsible for this change is the hypoxia-inducible factor-1 (HIF-1) [13]. HIF-1 is a heterodimer that consists of a constitutively expressed b subunit (HIF-1b or ARNT) and a catalytic a subunit (HIF-1a) [10,13]. At normoxia, HIF-1a is hydroxyl- ated at specific proline residues by oxygen-dependent prolyl hydroxylases, leading to an interaction with the von Hippel-Lindau tumor suppressor protein (pVHL) ⁄ E3 ligase complex and subsequent ubiquitin- mediated destruction [14–16]. Under hypoxic conditions, HIF-1a escapes from hydroxylation and tranlocates to the nucleus, where it forms a complex with HIF-1b and the cAMP-responsive element binding protein (CREB)-binding protein (CBP) ⁄ p300 co-activator, binds to hypoxia-response elements and transcriptionally modulates target genes [17]. HIF-1 has been shown to play critical roles in tumor angio- genesis, glucose metabolism, invasion ⁄ metastasis, and response to radiation and chemotherapy [18–22]. However, little is known about its possible role in the process of cellular differentiation in solid tumors. Gliomas are the most common and malignant pri- mary brain tumors in humans and are among the most hypoxic tumors known [23,24]. Glioblastoma multi- forme, the highest-grade glioma, is characterized by large necrotic areas within the tumor mass, which cor- relates with enhanced resistance to therapy, increased invasiveness and a poor prognosis for the patient [24]. In addition, malignant glioma cells could be induced to undergo differentiation towards their normal coun- terparts and thus serve as a faithful model to study molecular mechanisms underlying differentiation defects in solid tumors [5,25]. In the present study, cobalt chloride and defer- oxamine (DFO) were used to mimic an intratumoral mild hypoxia condition, and we found that differentia- tion induced by forskolin in rat C6 and primary cultured human malignant glioma cells was reversibly inhibited. Deletion of the endogenous HIF-1a gene restores the differentiation capacities, even in the presence of cobalt chloride. In contrast, stabilization of HIF-1 with small interfering RNA (siRNA) against VHL, which leads to proteosomal degradation of HIF-1a, shows differentia- tion blockage similar to that induced by cobalt chloride. We also demonstrated that inhibition of HIF-1a binding to its transcriptional co-activator, CBP ⁄ p300, abolishes the differentiation-inhibitory effect of HIF-1. Analyses of human glioma tissues have suggested a strong correlation between the expression of HIF-1a and malignancy (World Health Organization grade). Taken together, our results indicate that HIF-1 negatively regulates the differentiation of malignant gliomas and provide new insights into the differentiation therapy by targeting the HIF-1 pathway in solid tumors. Results Forskolin induces differentiation of C6 glioma cells Microscopic observation of C6 glioma cells treated for 24 h with 10 lm forskolin revealed major alterations in morphology. Unlike the mainly polygonal morphology of the vehicle control, the shape of forskolin-treated cells was similar to that of mature astrocytes, with smaller round cell bodies and much longer, fine, taper- ing processes (Fig. 1A). Western blotting analysis further confirmed a significant, dose-dependent, up-regulation of glial fibrillary acid protein (GFAP) protein, an established marker of mature astrocytes [26] (Fig. 1B). Meanwhile, the level of proliferating cell nuclear antigen (PCNA), a well-accepted marker of proliferation that facilitates the rapid processing of DNA [27], was markedly reduced (Fig. 1B). Forskolin also caused a statistically significant sub- dued proliferation, as demonstrated by the 3-(4,5- dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT) assay (Fig. 1C). Additionally, cell cycle analysis HIF-1a modulates malignant glioma differentiation H. Lu et al. 7292 FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS showed an accumulation in the G 1 -phase fraction to 91.2% and a remarkable decrease in the S-phase frac- tion to 4.4% (compared with 82.6% and 13.8% respectively in the vehicle control group) (Fig. 1D). Together, these results agree with those of a previ- ous report [28] and indicate that forskolin has the abil- ity to induce the differentiation of malignant glioma cells into maturate astrocytes and may be an appropri- ate differentiation-inducer in our research. Cobalt chloride and DFO inhibit differentiation induced by forskolin in C6 glioma cells Cobalt chloride has been widely used as a hypoxia- mimicking agent in both in vivo and in vitro studies as a result of its inhibitory effects on HIF-1 degradation, producing biochemical effects similar to those of low (1–3%)-oxygen hypoxia [29,30]. Here we found that forskolin-induced astrocyte-like morphological trans- formation and GFAP increase was blocked by co- incubation with cobalt chloride, while 100 lm cobalt chloride alone caused neither morphological nor GFAP changes (Fig. 2A,B). However, cobalt chloride showed a synergistic effect with forskolin on PCNA levels. In agreement with data on PCNA expression, the forskolin-induced decrease of the S-phase fraction was accentuated by cobalt chloride, while cobalt chloride alone only resulted in a minor reduction in the S-phase fraction. We next clarified whether the blockage in differenti- ation is a cobalt chloride-specific phenomenon. DFO is commonly used as an iron chelator and is completely different from cobalt chloride in its molecular structure and chemical formula [31]. The assays described above were repeated using 100 l m DFO instead of cobalt chloride, and similar results were obtained (Fig. 2A,C,D). Furthermore, both cobalt chloride and DFO caused remarkable accumulation of HIF-1a protein, while HIF-1a expression remained unaltered by expo- sure to forskolin (Fig. 2B,C), suggesting the involve- ment of HIF-1 a in the described differentiation blockage. Therefore, chemical hypoxia is likely to block differ- entiation in C6 glioma cells, causing them to remain in an undifferentiated quiescent state after exiting from the cell cycle. Cobalt chloride reversibly blocks differentiation induced by forskolin Cobalt chloride stabilizes HIF-1a by inhibition of spe- cific prolyl hydroxylase activity, which is reversible upon the withdrawal of cobalt chloride [32]. To eluci- date the role of HIF-1 a more fully, we examined whether withdrawal of cobalt chloride releases C6 cells from differentiation blockage. Significant induction of HIF-1a protein was observed in response to 24 h of exposure to cobalt chloride, while the withdrawal of cobalt chloride at this time point caused a dynamic reduction of HIF-1a expression (Fig. 3A). Concomi- tantly, 24 h after withdrawing cobalt chloride from co-incubation with forskolin, forskolin has resumed its AB C D Fig. 1. Forskolin induces differentiation of C6 glioma cells. C6 cells were incubated with or without forskolin for 24 h or for the time indi- cated. (A) Morphological transformation (original magnification, ·200). (B) Dose-dependent effect of forskolin on GFAP and PCNA expression. (C) Inhibition of cell proliferation. (D) Cell cycle distributions. H. Lu et al. HIF-1a modulates malignant glioma differentiation FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS 7293 capacity to induce morphological changes and altera- tions of GFAP and PCNA expression (Fig. 3B,C). These results suggest the essential role of persistent up-regulation of HIF-1a for the differentiation block- age in glioma cells. HIF-1a is required for the differentiation- inhibitory effect of cobalt chloride in C6 cells We next used siRNA, targeting HIF-1a, to further confirm the role of HIF-1a in differentiation blockage caused by cobalt chloride. As shown in Fig. 4A, cobalt chloride failed to induce HIF-1a accumulation in C6 cells transfected with siHIF-1a. Meanwhile, forskolin- induced morphological transformation and GFAP up-regulation was not inhibited, despite sustained exposure to cobalt chloride (Fig. 4B,C). These data indeed indicate the necessity of HIF-1a accumulation for differentiation blockage induced by cobalt chloride in glioma cells. pVHL is a component of a ubiqitin ligase complex (or E3) that polyubiquitinates HIF-1a in the presence of oxygen. Loss of pVHL function leads to constitu- tive HIF-1a stabilization and activity [14]. To further confirm the role of HIF-1a in the differentiation blockage, gene knockdown of VHL was used to acti- B AC D Fig. 2. Cobalt chloride (CoCl 2 ) and DFO inhibit differentiation induced by forskolin in C6 glioma cells. Cells were pretreated with 100 lM CoCl 2 or 100 lM DFO for 2 h and then treated with 10 lM forskolin for a further 24 h for morphology analyses (A) (original magnification: ·200), western blotting to evaluate HIF-1a, GFAP and PCNA expression (B and C), and cell cycle distributions (D). HIF-1a modulates malignant glioma differentiation H. Lu et al. 7294 FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS vate HIF-1. Thirty-six hours after transfection, the pVHL level was greatly reduced compared with the scrambled control (negative control) group; accord- ingly, HIF-1a was significantly accumulated in C6 cells (Fig. 5A,C). Similarly to chemical hypoxia induced by cobalt chloride or DFO, silence of VHL abrogated the morphological changes and up-regulation of GFAP expression induced by forskolin (Fig. 5B,C). These observations demonstrate that the expression of HIF-1a is necessary and sufficient to abrogate the differentiation capabilities of glioma cells. Chetomin abrogates the differentiation-inhibitory effect of cobalt chloride in vitro Chetomin disrupts the structure of the CH1 domain of CBP ⁄ p300, the transcriptional co-activator, thereby precluding its interaction with HIF-1 and attenuating hypoxia-inducible transcription [33]. We first examined the binding of HIF-1a to CBP ⁄ p300 by immunoprecip- itation. Figure 6A shows that chetomin significantly hampers the binding of HIF-1a to CBP ⁄ p300. We also found that cobalt chloride effectively blocks the mor- phological and GFAP changes induced by forskolin (data not shown), while in the presence of chetomin, this inhibitory effect of cobalt chloride is removed (Fig. 6B,C). However, chetomin alone showed no effects, either on the basal and cobalt chloride-induced HIF-1a protein levels, or on the morphology and A B C a b a b Fig. 3. Cobalt chloride (CoCl 2 ) reversibly blocks differentiation in C6 cells. (A) Cells were incubated with 100 l M CoCl 2 for 24 h and then CoCl 2 was withdrawn. Twenty-four hours later, the HIF-1a protein levels were examined. (B, C) C6 cells were treated with 100 l M CoCl 2 and 10 lM forskolin for 24 h, and then CoCl 2 was withdrawn (b) or not withdrawn (a). Twenty-four hours later, the cell morphol- ogy was analyzed (B) and western blotting was performed to evaluate GFAP and PCNA expression (C). A B C Fig. 4. HIF-1a is required for the differentiation-inhibitory effect of cobalt chloride (CoCl 2 ) in C6 cells. (A) Cells transfected with 30 nM HIF-1a or scrambled siRNAs (Scram) for 24 h were treated with 100 l M CoCl 2 for 2 h. Efficacy of HIF-1a silencing was evaluated using western blotting for HIF-1a accumulation induced by CoCl 2 . (B, C) Cells transfected with 30 n M HIF-1a or Scram siRNAs for 24 h were pretreated with 100 l M CoCl 2 for 2 h and then treated with 10 l M forskolin for a further 24 h before morphology analyses (original magnification: ·200) (B) and western blotting to evaluate GFAP expression (C). H. Lu et al. HIF-1a modulates malignant glioma differentiation FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS 7295 GFAP expression in C6 cells (Fig. 6B,C). The mRNA levels of vascular endothelial growth factor (VEGF) and glucose transporter-1 (Glut-1), two well-known HIF-1 target genes, were further evaluated to measure HIF-1 transcriptional activity. Figure 6D shows that the elevated mRNA levels of VEGF and Glut-1 induced by cobalt chloride were significantly alleviated by exposure to chetomin. Taken together, these data suggest an involvement of HIF-1a transactivation with the co-activator CBP ⁄ p300 in the differentiation-inhibi- tory efficacy of cobalt chloride. Combination of chetomin and forskolin attenuates tumor growth in vivo Besides its in vitro effectiveness, chetomin also affects the HIF-1 pathway and disrupts the interaction of HIF and CBP in vivo [33]. We therefore utilized xeno- graft tumor models to evaluate the effect of chetomin plus forskolin on tumor growth. In comparison to mice treated with vehicle control, neither chetomin nor forskolin alone had any remarkable influences on the growth of tumor xenografts (Fig. 6E,F). However, simultaneous exposure to chetomin and forskolin had significant antitumor activity (Fig. 6E,F) without obvi- ous effects on body weight over the treatment period (data not shown), revealing a synergistic effect between the HIF-1 pathway suppressor and the in vitro differ- entiation agent in malignant glioma xenografts. HIF-1a expression increases with tumor grade and inhibits differentiation induced by forskolin in primary human malignant glioma cells Malignant gliomas are a spectrum of tumors of varying differentiation and malignancy grades. To study HIF-1a expression in human glioma tissues of different grades (World Health Organization I–IV), we performed immunohistochemical staining with paraffin-embedded specimens and analyzed exclusively the non-necrotic region. Expression of HIF-1a protein was found in all 95 samples (representative immuno- staining images are shown in Fig. 7A and the results of HIF-1a immunohistochemistry analyses in patients are summarized in Table S1), and no obvious staining was observed in the five normal brain samples (a representative image is shown in Fig. 7A). Statistical evaluation revealed that the amount of HIF-1a was significantly increased in parallel with increasing glioma grade (Fig. 7B). The percentage of HIF-1a-positive cells in Grade I averaged 19.4%, while those in Grades II, III and IV were 32.5%, 46.1% and 70.5% respectively. Thus, HIF-1a was demonstrated to be broadly accumulated in glioma cells and its over- expression was correlated with glioma malignance grading, in other words, higher levels of expression of HIF-1a suggest a greater degree of differentiation defects. Then, we sought to test the generality of the inhibi- tory effects of HIF-1a in primary cultured human glioma cells. Exposure to the differentiation agent fors- kolin also resulted in differentiated characteristics, of a stellar shape with filamentous processes and increased GFAP expression in the primary glioma cells (Fig. 7C,D). However, co-incubation with cobalt chlo- ride blocked the morphological alterations and increased the amount of GFAP induced by forskolin (Fig. 7C,D). Quantitative analysis indicated that the percentage of GFAP-expressing cells was significantly up-regulated upon treatment with forskolin. Moreover, the up-regulation was reversed by cobalt chloride B C A Fig. 5. VHL knockdown blocks differentiation of C6 cells. (A) Wes- tern blotting was used to estimate the levels of pVHL after trans- fection with 30 n M VHL or scrambled siRNAs (Scram) for 36 h. (B, C) C6 transfected with siVHL for 36 h were treated with 10 l M forskolin for a further 24 h before morphology analyses (B) (original magnification, ·200) and western blotting to evaluate the expres- sion of HIF-1a and GFAP (C). HIF-1a modulates malignant glioma differentiation H. Lu et al. 7296 FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS (Fig. 7E). These results confirm our findings in C6 cells and, moreover, suggest a general correlation of HIF-1a activity with differentiation in malignant glioma cells. Discussion Gliomas derived from astrocytes or astroglial precur- sors are the most common malignant cancers affecting the central nervous system, accounting for > 60% of primary brain tumors [23]. Despite modern treatments with neurosurgical resection, radiotherapy and chemo- therapy, the median life expectancy for patients with malignant gliomas is approximately 12 months [34,35]. Rat C6 glioma cells are one of the well-established gli- oma cell lines with an undifferentiated phenotype and oligodendrocytic, astrocytic and neuronal properties, constituting a useful model in studies of glial-cell A D B E C F Fig. 6. Targeting HIF-1 by chetomin abolishes the differentiation-inhibitory effect of cobalt chloride (CoCl 2 ) in vitro and cooperates with fors- kolin to attenuate glioma growth in vivo. (A) The binding activity of HIF-1a to CBP. Cells stimulated with 1 n M chetomin for 24 h in the pres- ence of 100 l M CoCl 2 were immunoprecipitated with a CBP antibody followed by western blot analysis using an HIF-1a antibody. (B–D) C6 cells were pretreated with 1 n M chetomin for 2 h and then treated with 10 lM forskolin and ⁄ or 100 lM CoCl 2 for 24 h. (B) Morphology analy- ses (original magnification, ·200). (C) HIF-1a and GFAP protein levels were evaluated by western blotting. (D) mRNA levels of VEGF and Glut-1, as measured by RT-PCR. (E–F) The effects of chetomin and forskolin on tumor growth (E) and weight (F) in mice with C6 xenografts. Animals with established tumors after 1 week of growth were divided into groups treated with 5 mgÆkg )1 of forskolin, 0.2 mgÆkg )1 of cheto- min, 5 mgÆkg )1 of forskolin plus 0.2 mgÆkg )1 chetomin, or vehicle control. Data represent means ± standard error of the mean; n = 6 per group; *P < 0.05; **P < 0.01, compared with vehicle. H. Lu et al. HIF-1a modulates malignant glioma differentiation FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS 7297 differentiation [26]. They exhibit reversible defects in differentiation, which upon certain types of stimula- tion, allow them to differentiate normally. GFAP, the 50-kDa type III intermediate filament protein, may serve as a reliable marker of differentiation for normal astrocytes, while PCNA can be considered to be a marker of malignant proliferation and higher expres- sion, which is associated with a higher risk of malig- nancy. Forskolin, a small lipophilic molecule easily to be absorbed and distributed, is a well-known adenylate cyclase activator and a widely reported differentiating agent in various tumors, including glioma cells [36,37]. Our results, showing that some malignant glioma cells exit from the proliferating cell cycle and then may be A C D I II IV B E III Fig. 7. HIF-1a expression increases with tumor grade and blocks differentiation in human gliomas. (A) HIF-1a immunohistochemistry of stained sections from representative tissues of grade I–IV primary gliomas and of normal brains. (B) Statistical analysis of HIF-1a expression indicated that HIF-1a levels are significantly higher in high-grade gliomas than in low-grade gliomas (*, P < 0.01, compare with grade I). (#, HIF-1a is not detected). (C–E) Human malignant glioma cells were treated with 100 l M CoCl 2 for 2 h and then with 10 lM forskolin for a further 24 h. (C) Morphology of cells. (D) Immunocytochemistry of GFAP levels. (E) Quantification of the percentage of GFAP-expressing cells (*, P < 0.01, compared with the control; #, P < 0.01, compared with the forskolin group). HIF-1a modulates malignant glioma differentiation H. Lu et al. 7298 FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS induced to differentiate by forskolin, indicate that this model may be appropriate for the subsequent investi- gation on differentiation. Contrary to the development of differentiation- induction in in vitro cell lines, successful differentia- tion-inducing therapy for in vivo animal models and for patients with malignant gliomas, as well as other solid tumors, has not been reported to date. This sug- gests that the solid tumor microenvironment may counteract the actions of differentiation inducers in ways not present under in vitro culture conditions. This presumption is further strengthened by our data that forskolin alone failed to inhibit the growth of glioma xenografts. One aspect of the microenvironment that differs in tumor tissue versus normal tissue is oxygen tension. Normal oxygen tensions in cortical grey matter gener- ally range from 2.5 to 5.3%, with readings as high as 13% [38]. In contrast, oxygen tensions in solid tumors can range from physiological levels to below 0.1% in necrotic regions [39]. Ample experimental evidence revealed that the growth of malignant cells in vivo requires a hypoxic response and that this occurs pri- marily through the action of HIF-1a [13]. Recent data have also demonstrated that aberrant differentiation typically shows characteristics of abnormal growth and distinct invasion and metastasis, leading to the tumori- genic progression. The data presented here argue that targeting the HIF-1a response for the differentiation defects in solid tumors also has to be taken into account. It is well established that cobalt chloride stabilizes HIF-1a with kinetics similar to that of hypoxia [40]. However, a high concentration of cobalt chloride was previously reported to be antiproliferative, pro-apopto- tic and cytotoxic in diverse established cell lines [41]. Here we found that the use of a non-toxic concentra- tion of cobalt chloride (Fig. S1) provided a mild hypoxic model suitable for investigating the role of HIF-1a in glioma differentiation. HIF-1a is overexpressed in more than 70% of human cancers and regulates multiple steps of tumori- genesis, including tumor formation, progression and response to therapy [18]. However, the precise role of HIF-1a in tumor differentiation is unknown and highly controversial as a result of the conflicting results of several tumor models. While some studies found that HIF-1a mediates differentiation triggered by cobalt chloride or low oxygen tension in acute myeloid leukemic cells and pheochromocytoma PC12 cells [42– 44], other studies have shown that HIF-1a promotes a dedifferentiation phenotype in breast carcinoma and neuroblastoma cells [11,12] and that HIF-1a represses differentiation in lung carcinoma cells and high-grade glioma-derived precursors [45,46]. Here we find that HIF-1a induced by cobalt chloride ⁄ DFO or VHL knockdown abrogates the differentiation-induced potential of C6 malignant glioma cells, while silence or restrained function of HIF-1a restores the susceptibil- ity of C6 cells to forskolin-induced differentiation. These phenomena all indicate the essential role of HIF-1a in the negative regulation of differentiation in malignant gliomas. Interestingly, in contrast to their differentiation- inhibiting effect, cobalt chloride or DFO synergized with forskolin to suppress C6 cell proliferation and retained C6 cells in a neither differentiating nor prolif- erating stage. Although a poorly differentiated pheno- type is usually associated with rapid proliferation, it is well known that the proliferation rate of cells in a hyp- oxic environment is reduced [47], which in tumor lesions could be explained by a shortage of nutrients and growth factors, in addition to low oxygen levels. This suggests that the hypoxic environment as such might have both growth-inhibiting and differentiation- inhibiting efficacy. The coordinated transcriptional response mediated by the HIF-1 pathway requires co-activation by the CBP ⁄ p300 transcriptional co-activators. CBP ⁄ p300 are functional integrators of multiple signal-transduc- tion pathways because diverse transcription factors, among which cAMP-responsive element binding pro- tein (CREB) is a principal one, compete with each other to interact with a limited amount of CBP ⁄ p300 within the cell [48–51]. Competition between different transcription factors for CBP ⁄ p300 has been proposed to play roles in the coordination of gene expression in response to signaling [52]. One of the most remarkable examples of this phenomenon is the reci- procal functional antagonism between p53 and nuclear factor-jB through competition for CBP ⁄ p300 [53,54]. We have previously identified CREB as a req- uisite regulator of cellular differentiation in malignant glioma cells [5]. The present data show that the small molecule chetomin, which disrupts the structure of the CH1 domain of CBP ⁄ p300 and thus precludes its interaction with HIF-1, interferes with the differentia- tion blockage efficacy of HIF-1a. More importantly, in combination with chetomin, forskolin shows remarkable anti-glioma activity in vivo . We thus pro- pose that the differentiation-induction is highly depen- dent on the expression of HIF-1a and that stabilization of HIF-1a may abrogate the differentia- tion-induced potential of malignant glioma cells, at least in part through competition with CREB for binding to CBP ⁄ p300. H. Lu et al. HIF-1a modulates malignant glioma differentiation FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS 7299 It is noteworthy that chetomin has also been previ- ously characterized with potent immunosuppressive activity [55]. However, to date, few studies have reported the potential involvement of this feature in the anti-tumor effect of chetomin. In this context, it will be interesting to establish whether the immunosup- pressive effect participates in the in vivo action of chetomin in malignant gliomas. More than 100 direct HIF target genes have been identified that regulate a number of cellular processes, including glucose metabolism, angiogenesis, erythro- poiesis, survival and invasion [18]. It has also been documented that HIF indirectly regulates proliferation and differentiation through interactions with other sig- naling proteins, such as Notch and Myc. Gustafsson [56] has provided evidence that hypoxia and HIF-1a lead to the inhibition of differentiation in cortical neu- ral stem cells, myogenic satellite cells and C2C12 cells by interacting with and stabilizing the Notch ICD domain. In contrast, Koshiji [57] reported that HIF-1a induces cell cycle arrest in HCT-116 colon cancer cells by functionally counteracting Myc, the inactivation of which results in the differentiation of tumor cells [58– 60]. HIF-1a-induced cell cycle arrest, and thus differ- entiation, in colon carcinomas seems paradoxical to its role in the present study. As discussed above, however, the exact role of HIF-1a in tumor differentiation remains highly controversial as a result of the different tumor models. Undoubtedly, further investigation is warranted for a better understanding of the divergent role and target genes of HIF-1a in different types of cancer. Malignant gliomas constitute a spectrum of brain tumors with varying differentiation and malignancy grades, and with clinical courses that range from indo- lent to highly malignant. Glioblastoma multiforme, the most common and lethal subtype of the malignant gli- omas, is characterized by poorly differentiated cells with intense proliferation and widespread invasion [61]. Our data, showing that an increased amount of HIF-1a protein accompanies progressing malignant gli- oma grade, provide further evidence in support of the correlation between HIF-1a and differentiation defects in solid tumors. In summary, we have shown that HIF-1a protein, inducible by the hypoxia mimicker cobalt chloride, or by DFO, blocks induced differentiation in rat C6 and primary cultured human malignant glioma cells. Loss of HIF-1a abrogates this blockage, whereas forced expression of HIF-1a stimulates this blockage. We also provide evidence that HIF-1a exerts this differentia- tion-inhibitory efficacy by binding to its co-activator CBP ⁄ p300. Collectively, we identify HIF-1a as a nega- tive regulator of the differentiation in malignant glio- mas, suggesting a novel therapeutic strategy by targeting the HIF-1 pathway in the differentiation- inducing therapy in solid tumors. However, the precise mechanism by which HIF-1a competes with CREB for CBP ⁄ p300 and then blocking differentiation is still to be investigated. Experimental procedures Cell culture and drug treatment C6 rat glioma cells were obtained from the American Type Culture Collection (Manassas, VA, USA) and maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Invitro- gen, Grand Island, NY, USA), supplemented with 10% fetal bovine serum (FBS) (Hyclone Laboratories, Logan, UT, USA), in a humidified atmosphere of 5% CO 2 at 37 °C. Human malignant glioma tissues were obtained immediately after surgical removal with approval of the Ethical Committee of Sun Yat-Sen University. Primary cul- tures of human glioma cells were prepared as previously described [5]. Differentiation was induced by 24 h of expo- sure to 10 lm forskolin (Sigma, St Louis, MO, USA) in DMEM containing 1% FBS. For primary human glioma cells, forskolin was added to DMEM containing 10% FBS. The mimicked hypoxia condition was achieved by stimula- tion with 100 lm cobalt chloride or DFO (Sigma) 2 h before treatment with forskolin. Chetomin (Alexis Biochem- icals, San Diego, CA, USA) was dissolved in dimethysulf- oxide and added 2 h before treatment with forskolin at a concentration of 1 nm. For the vehicle control group, 0.1% dimethysulfoxide was used. Morphological evaluation The cell morphologies were studied during the indicated time course using an Olympus (Melville, NY, USA) IX71 inverted microscope and a DP70 CCD camera. MTT assay Cell proliferation was evaluated with the 4-[3-(4-iodophe- nyl)-2-(4-nitrophenyl)-2H-5-tetrazolio]-1,3-benzene disulfo- nate (WST-8) assay using a Cell Counting Kit (CCK-8; Dojindo Molecular Technologies, Gaitherburg, MD, USA). Cell cycle analysis A flow cytometry analysis of the DNA content of cells was performed to assess the cell-cycle phase redistributions, as described previously [62]. In brief, the cells were collected by trypsinization, washed in phosphate-buffered saline HIF-1a modulates malignant glioma differentiation H. Lu et al. 7300 FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS [...]... classification The investigation of these tissues was in accordance with the rules of the Declaration of Helsinki and the Ethical Committee of Sun Yat-Sen University FEBS Journal 276 (2009) 7291–7304 ª 2009 The Authors Journal compilation ª 2009 FEBS 7301 HIF-1a modulates malignant glioma differentiation H Lu et al Immunohistochemistry staining of primary cells and of 4-lm sections of paraffin-embedded samples... using the image pro plus 6.0 software 8 9 Statistical analysis Data are presented as mean ± standard error of the mean of three separate experiments Statistical significance was determined using the Student’s t-test A result with a P-value of less than 0.05 was considered statistically significant Acknowledgements 10 11 12 We thank Professor Ying Guo, Professor Chunkui Shao and Professor Jianyong Shao for... the manufacturer’s protocol Inhibition of protein expression was assessed by immunoblot analysis To assess the efficacy of siHIF-1a, an additional 24 h of treatment with cobalt chloride was used to induce HIF-1a expression Differentiation experiments were carried out 24 or 36 h after siRNA transfection Western blot analysis After lysis of cells and measurement of protein concentration, the cells were... length · width2 [33] After 12 days of treatment, the animals were killed and the tumor weight was determined All animal procedures were performed under the guidelines of the National Institutes of Health Tissue samples and immunohistochemistry Tumor tissues from a total of 95 adult patients, 53 men and 45 women (18 to 73 years of age; average 42 years) diagnosed with gliomas (World Health Organization... Yan G (2007) Cholera toxin induces malignant glioma cell differentiation via the PKA ⁄ CREB pathway Proc Natl Acad Sci U S A 104, 13438–13443 6 Choi EJ, Oh HM, Wee H, Choi CS, Choi SC, Kim KH, Han WC, Oh TY, Kim SH & Jun CD (2009) Eupatilin exhibits a novel anti-tumor activity through the induction of cell cycle arrest and differentiation of gastric carcinoma AGS cells Differentiation 77, 412– 423 7 Chintharlapalli... Krishna TS, Kong XP, Gary S, Burgers PM & Kuriyan J (1994) Crystal structure of the eukaryotic DNA polymerase processivity factor PCNA Cell 79, 1233–1243 28 Chen TC, Hinton DR, Zidovetzki R & Hofman FM (1998) Up-regulation of the cAMP ⁄ PKA pathway inhibits proliferation, induces differentiation, and leads to apoptosis in malignant gliomas Lab Invest 78, 165– 174 29 Badr GA, Zhang JZ, Tang J, Kern TS &... Zhu YS et al (2003) Cobalt chloride and low oxygen tension trigger differentiation of acute myeloid leukemic cells: possible mediation of hypoxia-inducible factor-1 alpha Leukemia 17, 2065– 2073 Pacary E, Tixier E, Coulet F, Roussel S, Petit E & Bernaudin M (2007) Crosstalk between HIF-1 and ROCK pathways in neuronal differentiation of mesenchymal stem cells, neurospheres and in PC12 neurite outgrowth... (2004) cAMP-induced astrocytic differentiation of C6 glioma cells is mediated by autocrine interleukin-6 J Biol Chem 279, 15441–15447 26 Roymans D, Vissenberg K, De Jonghe C, Grobben B, Claes P, Verbelen JP, Van Broeckhoven C & Slegers H (2001) Phosphatidylinositol 3-kinase activity is required for the expression of glial fibrillary acidic protein upon cAMP-dependent induction of differentiation in rat C6... and cancer Pharmacol Ther 73, 51–65 2 Sell S (2004) Stem cell origin of cancer and differentiation therapy Crit Rev Oncol Hematol 51, 1–28 3 Leszczyniecka M, Roberts T, Dent P, Grant S & Fisher PB (2001) Differentiation therapy of human cancer: basic science and clinical applications Pharmacol Ther 90, 105–156 4 Wang ZY & Chen Z (2000) Differentiation and apoptosis induction therapy in acute promyelocytic... EA, Vieira A, Cornell-Kennon S, Lee J, Wang B, Wang J et al (2004) Small molecule blockade of transcriptional coactivation of the hypoxia-inducible factor pathway Cancer Cell 6, 33–43 34 Stupp R, Mason WP, van den Bent MJ, Weller M, Fisher B, Taphoorn MJ, Belanger K, Brandes AA, HIF-1a modulates malignant glioma differentiation 35 36 37 38 39 40 41 42 43 44 45 46 Marosi C, Bogdahn U et al (2005) Radiotherapy . Hypoxia-inducible factor-1a blocks differentiation of malignant gliomas Huimin Lu 1, *, Yan Li 1,2, *, Minfeng. involving hypoxia-inducible factor-1a that contributes to a differentiation defect in malignant gliomas and sheds new light on the differentiation therapy of solid