LBH589 Inhibits Glioblastoma Growth and Angiogenesis Through Suppression of HIF-1a Expression

Glioblastoma (GBM) is an angiogenic malignancy with a highly unfavorable prognosis. Angiogenesis in GBM represents an adap- tation to a hypoxic microenvironment and is correlated with tumor growth, invasion, clinical recurrence, and lethality. LBH589 (also called panobinostat) is a histone deacetylase (HDAC) inhibitor with potent antitumor activity. In the current study, we investigated the mechanism and effects of LBH589 on GBM growth and hypoxia-induced angiogenesis in vitro and in vivo. To determine the antitumor and angiogenesis activity and mechanism of LBH589, we used cell proliferations in vitro and GBM xenografts in vivo. To clarify mechanisms of LBH589 on angiogenesis, HDAC assay, RT-PCR, Western blot, and co-immunoprecipitation assays were performed. We found LBH589 displayed significant antitumor effects on GBM as demonstrated by inhibited cell prolif- eration, slower tumor growth, and decreased microvessel density of subcutaneous xenografts. These actions of LBH589 resulted from the disruption of heat shock protein 90/HDAC6 complex, in- creased HIF-1a instability and degradation, and decreased VEGF expression. Our results indicate the potential antiangiogenic activ- ity of LBH589 in human GBM and provide some preclinical data to warrant further exploration of HDAC inhibitors for the treatment of advanced glioma. Moreover, our study supports the role of HDAC inhibitors as a therapeutic strategy to target tumor angiogenesis.

Glioblastoma (GBM) is the most common and aggres- sive primary brain tumor, with a median overall survival of only 10–14 months (1). Despite decades of research into itstreatment, prognosis remains poor, with only 3%–5% of patients surviving >3 years (2). Treatment of GBM consists of maximal surgical resection, radiotherapy, and adjuvant che-motherapy with temozolomide (3). Regardless of multimodal- ity treatment, recurrence or progression of disease is inevitable and uniformly fatal. Therefore, it is necessary to develop ef- fective, novel therapies for GBM.Angiogenesis is one of the key features of GBM. Neoan- giogenesis of GBM is a crucial factor in their growth, aggres- sive behavior, clinical recurrence, and prognosis of patients(4). Hypoxia-inducible factor 1 (HIF-1) is an important tran- scription factor that responses to hypoxia to regulate angio- genesis, cellular survival, and tumor invasion (5). As a subunit of HIF-1, HIF-1a is responsible to its oxygen sensitivity, de- termining the availability of HIF-1. HIF-1 activates a large battery of genes, including vascular endothelial growth factor (VEGF), a highly potent cytokine known to promote angio- genesis and increase oxygen delivery to hypoxic regions (6). Recently, HIF-1a and its signaling pathway have become tar- gets for GBM chemotherapy aimed at inhibiting angiogenesis(7). For example, it has been demonstrated that histone deace- tylase (HDAC) 6 contributed to HIF-1a stability indirectly through deacetylation of heat shock protein (Hsp) 90, which enhanced the Hsp90 chaperone function toward its client pro- teins, including HIF-1a (8, 9). Therefore, we speculated HDAC inhibitors might inhibit the deacetylase activity of HDAC6 and indirectly decrease HIF-1a expression through Hsp90/HDAC6-HIF-1a pathway.LBH589 is a hydroxamic acid and shows its potent in- hibitory activity against all class I (including HDAC1, 2, 3, 8), II (including HDAC4, 5, 6, 7, 9, 10), and IV (HDAC11) puri- fied recombinant HDAC enzymes at low nanomolar concen- trations (10).

LBH589 was also at least 10-fold more potent as a HDAC inhibitory compared with vorinostat, which showed antitumor activity in GBM as a pan-HDAC inhibitor (10, 11). So far, cotreatments or single-drug treatments with LBH589 exhibited significantly in vitro and in vivo antitumor effects across a variety of hematologic malignancies and solid tumors, including diffuse large B-cell lymphoma, Hodgkin lymphoma, multiple myeloma, pancreatic cancer, and nonsmall cell lung cancer (12–14). However, the effect of LBH589 on GBM has not been reported in preclinical studies. In this study, we detected the antitumor activity of LBH589 in vitro and in vivo against human GBM U87MG and U251MG cells using either cell proliferation assay or nude mouse xenograft model. The correlation between deace- tylation activity of LBH589 and its effects on tumor growthand angiogenesis was also evaluated in vitro and in vivo.The U87MG and U251MG human GBM cell lines were purchased from R&S Biotechnology Co., Ltd. (Shanghai, China) and cultured in DMEM medium (Gibco BRL, Gai-thersburg, MD) with 10% fetal bovine serum. For normoxic conditions, cells were incubated in an atmosphere of 5% CO2 and 20% O2. For hypoxic treatment, the cells were incubated in an atmosphere of 1% O2, 5% CO2, and 94% N2 under inter- mittent flushing with nitrogen. LBH589 was obtained fromSelleck Chemicals (Shanghai, China) and dissolved in di- methyl sulfoxide (DMSO) for the preparation of stock solu- tions (50 mM).Cell proliferation after treatment with LBH589 at differ- ent concentrations (10, 20, and 40 nM) was measured by CCK-8 assay (Dojindo, Shanghai, China). The U87MG and U251MG cells were seeded at a density of 5 × 104 cells/ 100 mL DMEM/well in 96-well plates and were allowed an overnight period for attachment.

Cells were incubated in a hu- midified incubator under hypoxic conditions at 37 ◦C for 24, 48, and 72 hours. After the designated time, CCK-8 solution was added to each well containing 100 mL of the culture me- dium and was incubated for another 4–5 hours at 37 ◦C. The amount of formazan dye was measured at 450 nm absorbance using Multiskan MK3 microplate reader (Thermo Scientific, Rockford, IL). All experiments were performed in triplicate and repeated 3 times.Ten 6-week-old male athymic nude C57/BL6J mice (Wanleibio, Shenyang, China) were housed under pathogen- free conditions. All animal studies were done according to the protocol approved by the Animal Care and Use Committee at the Shandong University. U87MG tumor cells (1 × 107 cells in 200 lL medium) were injected subcutaneous into the right flank. The tumor volume was measured using a vernier caliper every 3 days and tumor volumes were calculated using the el- lipsoid formula: length × width × height × 0.52 (15). At 13 days after implantation of tumor cells (tumor volume 100– 300 mm3), 5 mice were injected intraperitoneally with LBH589 (20 mg/kg/day). Likewise, the other 5 mice were injected intraperitoneally with DMSO. After injection for 4 weeks, animals were killed and tumor volume and weightwere recorded. The tumor tissues were fixed with 10% buffered formalin and embedded in paraffin, then cut into 5-mm-thick sections on glass slides. Paraffin sections were dewaxed and conventional hematoxylin and eosin (H&E) staining was carried out using standard histologic procedures.

For immunohistochemistry, dewaxed sections were incubated with primary antibodies against CD34 (1:200, ZSGB-Bio, Beijing, China) overnight at 4 ◦C. Detection of bound antibod- ies was done using the EnVision System (Dako, Denmark) following the manufacturers’ instructions. The microvessel density (MVD) was determined by counting the number of blood vessels stained with the antiCD34 antibody. The MVD for each tumor was expressed as the average count of the 10 most densely stained fields identified in the ×200 field (16). Each positive endothelial cell cluster of immunoreactivity in contact with the selected field was counted as an individual vessel in addition to the morphologically identifiable vessels with a lumen.HDAC activity assays were performed using the colori- metric HDAC activity assay from BioVision (BioVision, San Francisco, CA) according to manufacturer’s instructions and previous described (17). Briefly, 50 mg of cellular extracts from LBH589 treated hypoxic U87MG and U251MG cells were diluted in 85 mL ddH2O; then, 10 mL of 10× HDAC as- say buffer were added followed by additional of 5 mL of the colorimetric substrate; samples were incubated at 37 ◦C for 1 hour. Subsequently, the reaction was stopped by adding 10 mL of lysine developer and left for additional 30 minutes at 37 ◦C. Samples were then quantitated using Multiskan MK3 microplate reader (Thermo Fisher Scientific) at 400 nm. HDAC activity was expressed as relative OD values per mg of protein sample. The kit contains negative and positive controls that consist of nuclear extract of HeLa treated or not with Tri- chotatin A, respectively.Quantitative Real-Time Reverse Transcription PCRTotal RNA was extracted from LBH589 treated hypoxic U87MG cells and quantified using RNA Isolation Reagent kit (BioTeke, Beijing, China).

cDNA was obtained from 1 mg of total RNA from each sample (Super M-MLV reverse tran- scriptase, BioTeke) and 1 mL of cDNA sample was used as template for PCR with SYBR Green One Step RT-qPCR Kit (Solarbio, Beijing, China) following the manufacturer’s instructions. PCR were carried out in an Exicycler 96 (Bion- eer, Dajeon, Korea). For each sample and experiment, tripli- cates were made and normalized by b-actin mRNA levels. PCR amplification was performed with the following primers: HIF-a forward, 50-GCCACATCATCACCATATAGAG-30, reverse, 50-TCAAAGCGACAGATAACACG-30; VEGF for- ward, 50-ACACACCCACCCACATACATA-30, reverse, 50- ACTCAAGTCCACAGCAGTCAA-30; HDAC6 forward, 50-TGGCTATTGCATGTTCAACCA-30, reverse, 50-GTCG AAGGTGAACTGTGTTCCT-30; b-actin forward, 50-CTTAG TTGCGTTACACCCTTTCTTG-30, reverse, 50-CTGTCACC TTCACCGTTCCAGTTT-30.Total proteins were extracted from LBH589 treated hyp- oxic U87MG cells using lysis buffer (Beyotime, Wuhan, China). Western blot analyses were carried out according to methods as previously described (18). The monoclonal or polyclonal antibodies were purchased from Abcam (Cam- bridge, UK; antibodies against HIF-1a, acetylated lysine, HDAC6, and acetylated H3/H3), Santa Cruz Biotechnology (Santa Cruz, Santa Cruz, CA; antibody against Hsp90).

Co- immunoprecipitation was performed as previously described(18). In brief, cell lysates (500 mg) were incubated with HIF- 1a, Hsp90, or normal rabbit IgG, or normal mouse IgG mono- clonal antibodies (Santa Cruz Biotechnology) for 1 hour, then immunoprecipitated out with Protein A-agarose beads (Santa Cruz Biotechnology) at 4 ◦C overnight. The immunoprecipi- tates were washed 3 times in the lysis buffer, and proteins were eluted with the SDS sample loading buffer prior to West- ern blot analyses with specific antibodies against acetylated lysine (Ace-lysine) and HDAC6.Data were analyzed using SPSS software (SPSS Inc., Chicago, IL). For 2 groups in the animal experiment, the Stu- dent’s t test was used with equal/unequal variance version depending on variance ratio test. For 3 or more groups, Levene tests were performed to assess variance homogeneity. Stan- dard 1-way ANOVA followed with Student–Newman–Keuls multiple comparison procedures were performed with homo- geneous variances, and the Welch’s ANOVA followed by Games–Howell multiple comparisons test was performed withheterogeneous variances. A final value of p < 0.05 was con- sidered significant. GraphPad Prism 5 (GraphPad SoftwareInc., La Jolla, CA) was used to generate histograms and line chart. RESULTS LBH589 Inhibited Growth of U87MG and U251MG Cells In Vitro and In VivoAfter treatment with serial dilutions (10, 20, and 40 nM) at different times (24, 48, and 72 hours), the antitumor activity of LBH589 was assessed in human GBM U87MG and U251MG cell lines. Under phase-contrast microscope, the number of LBH589 treated U87MG cells was significantly de-creased at 48 and 72 hours (Fig. 1A). The CCK-8 assay showed the relative concentration of DMSO (≤0.1%) has little influence on cell proliferation compared with the control group (p > 0.05). However, treatment with LBH589 resulted in significant inhibition of U87MG cell proliferation at con-centrations of 20 and 40 nM at both 48 and 72 hours comparedwith DMSO treatment (Fig. 1B). In detail, after LBH589 treat- ment for 40 nM, the OD values were decreased by 57.6% (p ¼ 0.000) and 52.2% (p ¼ 0.000) at 48 and 72 hours, respec-tively. On the basis of the growth inhibition assay, the IC50 ofLBH589 treatment at 48 and 72 hours was calculated to be 35 nM and 50 nM in U87MG cells. However, no significant inhibition of U87MG cells proliferation was observed after1002 24 hours treatment of LBH589 (p > 0.05). These results showed LBH589 inhibited U87MG cells proliferation.We also detected the effect of LBH589 on the prolif- eration of U251MG cells. The densities of cells were also significantly decreased at 48 and 72 hours (Fig. 1A). After 24 hours, the OD value were significantly decreased by30.0% for 40 nM group compared with DMSO treatment (p ¼ 0.000, Fig. 1C). After treatment with LBH589 for 48and 72 hours at 3 concentrations (10, 20, and 40 nM), U251MG cell proliferations were significantly inhibited compared with DMSO treatment (p < 0.01, Fig. 1C). Forexample, the OD values at 40 nM groups were decreased by 29.5% at 48 (p ¼ 0.000) and 72 hours (p ¼ 0.000). Our results also displayed LBH589 inhibited U251MG cellsproliferation.Nude mice bearing U87MG xenografts were treated in- traperitoneally with LBH589 (20 mg/kg/day) beginning on day 13 after tumor implantation. The tumor volume was sig- nificantly smaller from the day 25 than controls that had re- ceived DMSO (p < 0.01, Fig. 2A, B). On day 42, LBH589suppressed tumor growth by 53.1% (p ¼ 0.001) in volume.These data indicate that LBH589 inhibited tumor growthin vivo.We detected the MVD in the tumor tissues from the U87MG xenograft mice. Immunohistochemical staining usingantibody against CD34 showed the MVD in the solid areas of the tumor was significantly reduced by 63.0% in LBH589 treatment group (p ¼ 0.001, Fig. 3A, C). H&E staining showedsignificant tumor necrosis in both groups. Moreover, H&E staining presented numerous dilating blood vessels surround- ing necrosis foci in DMSO treated tumors, while only a few vessel fissures were found in LBH589 treated tumors (Fig. 3B). Immunostaining using antiCD34 antibody showed the MVD surrounding necrosis foci was significantly reducedby 86.1% compared with the DMSO treatment controls (p ¼ 0.000, Fig. 3C). These results indicated that LBH589inhibited angiogenesis in U87MG xenografts.LBH589 Inhibited HDACs Activity in U87MG and U251MG CellsTo investigate whether a decrease in deacetylase activity could be achieved by LBH589 treatment in U87MG and U251MG cells, total HDACs activity was evaluated in cell extracts by colorimetric commercial HDACs activity assay. As shown in Figure 4A, after LBH589 treatment for 48 hoursin U87MG cells, HDACs activity was significantly decreased by 83.2% (p ¼ 0.022), 90.2% (p ¼ 0.019), and 98.1%(p ¼ 0.016) in the corresponding treatment groups (10, 20, and40 nM) compared with DMSO treatment group. Additionally, HDACs activity was also significantly decreased by 76.5% (p ¼ 0.023), 92.6% (p ¼ 0.024), and 93.0% (p ¼ 0.033) in 10,20, and 40 nM groups after treatment for 48 hours in U251MGcells. Our results indicated LBH589 inhibited HDAC activity in U87MG and U251MG cells. were significantly increased by 45.9% (p ¼ 0.004), 68.3%(p ¼ 0.003), and 122.3% (p ¼ 0.000) in the 10, 20, and 40 nM of LBH589 treatment groups compared with the DMSOgroup, respectively. LBH589 also increased the acetylation level of tubulin by 87% (p ¼ 0.03) and 142% (p ¼ 0.004) in the 20 nM and 40 nM treatment cells, respectively. Theseresults suggested LBH589 inhibited deacetylation of nuclear and cytoplasmic proteins in U87MG cells.Mechanisms of LBH589 Inhibited Angiogenesis in U87MG CellsConsidering the important role of HIF-1a and VEGF in angiogenesis of GBM (7), we first detected HIF-1a and VEGF expressions at mRNA and protein levels after LBH589 treat- ment for 48 hours under hypoxic conditions. Compared with DMSO treatment group, VEGF mRNA expression was re-duced by 50.7% in LBH589 treatment group (p ¼ 0.026, Fig. 5A). However, no significant difference of HIF-1amRNA level was observed between DMSO and LBH589 treatment groups (p ¼ 0.035). At protein levels, LBH589 treat- ment significantly reduced HIF-1a and VEGF expressions by93.8% (p ¼ 0.000) and 75.7% (p ¼ 0.001), respectively(Fig. 5B). These results suggested HIF-1a may be regulatedby LBH589 at the posttranslational level.Previous studies have demonstrated that HIF-1a degra- dative pathway was regulated by Hsp90 (19). Accordingly, co-immunoprecipitation with immunoblotting analysis showed HIF-1a coprecipitated the decreased level of Hsp90 in U87MG cells after LBH589 treatment for 48 hours under hyp- oxia compared with the DMSO group (Fig. 5C). However, there was no significant difference of HDAC6 expression be- tween the 2 groups. Second, HDAC6 has been demonstrated to interact with Hsp90 and regulate its acetylation (8).1003 Therefore, we next determined whether HDAC activity inhibi- tion caused an increase in Hsp90 acetylation. Co- immunoprecipitation with immunoblotting analysis showed Hsp90 coprecipitated with the increased level of acetylated lysine residues (Ace-lysine) and the decreased level of HDAC6 compared with the DMSO group (Fig. 5D), indicating that Hsp90/HDAC6 complex formation was disrupted by LBH589. However, there was no difference of HDAC6 mRNA or protein expression level between both groups (data not shown). These results showed that LBH589 inhibited HDAC6-mediated deacetylation of Hsp90, resulting in HIF- 1a instability and degradation. DISCUSSION In this study, we found LBH589 significantly inhibited the proliferations of U87MG and U251MG cells in vivo. In addition, LBH589 also significantly suppressed growth of U87MG xenografts in vitro. These results suggest antitumor potency of LBH589 on GBM. GBM is also one of the most highly vascularized human tumors, and its growth and survival is dependent on angiogenesis (4). HIF-1a is one of the best de- scribed regulatory gene and transcription factor activated un- der hypoxia, leading to VEGF transcription and angiogenesis in GBM (20). VEGF increases vascular endothelial cell prolif- eration in GBM. Multiple strategies have been developed to target VEGF-mediated angiogenesis (21). In our study, we found LBH589 inhibited HIF-1a at the protein level rather than the mRNA level under hypoxic conditions. LBH589 inhibited VEGF at both protein and mRNA levels. Therefore, the anti-angiogenesis effect of LBH589 was confirmed by the inhibition of HIF-1a activity and decreased VEGF secretion in our study. Furthermore, Zhong et al found the HIF-1a immu- nostaining was especially intense in pseudopalisading tumor cells surrounding areas of necrosis in human GBM specimens (22), suggesting the increased proangiogenic factors surround- ing the necrosis foci. Actually, microvascular hyperplasia in the GBM is a form of angiogenesis that is induced by hypoxic pseudopalisading cells and is usually present in regions adja- cent to necrosis (23). Hence, the significantly decreased MVD detected in implanted U87MG xenografts, especially sur- rounding necrosis foci, provided further evidence that LBH589 inhibits hypoxia-induced angiogenesis in GBM.It has been demonstrated HDAC6 contributed to HIF-1a stability indirectly through deacetylation of Hsp90, which en- hanced the Hsp90 chaperone function toward its client pro- teins, including HIF-1a (8, 9). On the other hand, in our study, although there was no effect on HDAC6 expression, LBH5891005 significantly inhibited HDAC activity in U87MG and U251MG cells and increased protein acetylation in U87MG cells, suggesting LBH589 may exert its inhibitory effects on GBM through decreased HDAC6 activity. Accordingly, the loss of deacetylation activity of HDAC6 induced by LBH589 increased the level of acetylated Hsp90 detected using co- immunoprecipitation. The hyperacetylation of Hsp90 re- pressed its chaperone function, leading to the accumulation of immature HIF-1a and degradation mediated by the von Hippel–Lindau (VHL) protein, a tumor suppressor that acted as an ubiquitin ligase (E3) to enhance HIF-1 ubiquitination and proteasomal degradation (24). Thus, we speculated LBH589 may inhibit angiogenesis through targeting HDAC6 activity in the Hsp90/HDAC6 complex and facilitating HIF-1 degradation (Fig. 6). In addition, it has been reported HDAC1, 3, and 4 could directly bound to HIF-1a and induced deacety- lation of lysine residues and thus enhanced HIF-1a stability (25, 26). HDAC7 also cotranslocated to the nucleus with HIF- 1a under hypoxic conditions and increased transcriptional activity of HIF-1a through the formation of a complex with HIF-1a and p300 (27). Jeong et al reported that HIF-1a protein was acetylated at the K532 residue by ARD1 (Arrest- Defective-1), promoting its interaction with VHL and thus en- hancing its ubiquitination and degradation (27). These findings suggest multiple Class I and Class II HDACs regulate HIF-1a stability as demonstrated in Figure 6. Thus, it cannot be excluded1006 that LBH589, as a pan-HDAC inhibitor, may also decrease HIF- 1a levels through simultaneously targeting other HDACs. On the other hand, it has been demonstrated LBH589 inhibited VEGF-induced endothelial cell motility and tube formation detected using human umbilical vein endothelial cells (29). These results provide evidence that LBH589 might inhibit an- giogenesis in GBM through targeting both the tumor and stromal endothelial cell compartments. Thus, further research is needed to understand other pathways which act simultaneously to en- hance antiangiogenic effects of LBH589 in GBM. In conclusion, we found LBH589 inhibited proliferation and angiogenesis of GBM both in vivo and in vitro through in- creased HIF-1a instability and degradation mediated by Hsp90/HDAC6 complex under hypoxic conditions. Although a recent phase II study reported the addition of LBH589 to bevacizumab did not significantly improve the 6-month progression-fee survival rate of patients with recurrent GBM compared with historical controls of bevacizumab monotherapy (30), we think our study will provide some pre- clinical data exploring the activity of the LBH589 to poten- tially improve the therapeutic response for GBM in the LBH589 future.