Full Text (PDF)

Published OnlineFirst October 27, 2009; DOI: 10.1158/1541-7786.MCR-09-0225

z-Leucinyl-Leucinyl-Norleucinal Induces Apoptosis of Human Glioblastoma Tumor–Initiating Cells by Proteasome Inhibition and Mitotic Arrest Response

Massimiliano Monticone,1 Emanuela Biollo,2 Andrea Fabiano,4 Marina Fabbi,3 Antonio Daga,3 Francesco Romeo,3 Massimo Maffei,3 Alice Melotti,3,4 Walter Giaretti,3 Giorgio Corte,3,4 and Patrizio Castagnola3

1Centro Biotecnologie Avanzate; 2Dip. Chimica e Tecnologie Farmaceutiche ed Alimentare, Università di Genova; 3Istituto Nazionale per la Ricerca sul Cancro; and 4Dip. Oncologia Biologia e Genetica, Università di Genova, Italy

Abstract may have a potential therapeutic value that deserves γ-secretase inhibitors have been proposed as drugs further investigation in animal models. (Mol Cancer Res able to kill cancer cells by targeting the NOTCH pathway. 2009;7(11):1822–34) Here, we investigated two of such inhibitors, the Benzyloxicarbonyl-Leu-Leu-Nle-CHO (LLNle) and the N-[N-(3,5-difluorophenacetyl)-L-alanyl]-S-phenylglycine Introduction t-butyl ester (DAPT), to assess whether they were effective Glioblastomas (GBM) are poorly differentiated astrocytic in killing human glioblastoma tumor–initiating cells tumors arising in the central nervous system, which despite re- (GBM TIC) in vitro. We found that only LLNle was able at cent improved treatment modalities are still characterized by the micromolar range to induce the death of GBM TICs very poor prognosis. Several studies have shown the existence by apoptosis. To determine the cellular processes that of a subpopulation of cells within glioma tumors displaying were activated in GBM TICs by treatment with LLNle, we cancer stem cells properties (1-3). The term “tumor initiating analyzed the amount of the NOTCH intracellular domain cells” (TIC) is frequently used to describe such cells (4), and and the gene expression profiles following treatment with we shall use it throughout this article to indicate cells with can- LLNle, DAPT, and DMSO (vehicle). We found that LLNIe, cer stem cell capacity. Due to the fact that TICs promote the beside inhibiting the generation of the NOTCH intracellular tumor chemoresistance (5, 6), radioresistance (3, 7), and angio- domain, also induces proteasome inhibition, proteolytic genesis (5, 8), it is conceivable that finding a manner to kill stress, and mitotic arrest in these cells by repressing them would improve GBM therapy (9, 10). genes required for DNA synthesis and mitotic progression The NOTCH pathway plays important roles during the cen- and by activating genes acting as mitotic inhibitors. DNA tral nervous system development, contributing to the mainte- content flow cytometry clearly showed that cells treated nance of neural stem/progenitor cell pool (11), to promote the with LLNle undergo arrest in the G2-M phases of the cell neural lineage entry of embryonic stem cells (12) and in the cycle. We also found that DAPT and L-685,458, another differentiation of astroglia from the adult hippocampus-derived selective Notch inhibitor, were unable to kill GBM TICs, multipotent progenitors in rat (13). NOTCH signaling activa- whereas lactacystin, a pure proteasome inhibitor, was tion requires the proteolytic processing of this type I integral effective although at a much less extent than LLNle. These membrane protein by a two-step cleavage process catalyzed data show that LLNle kills GBM TIC cells by inhibiting first by a metalloprotease and then by the γ-secretase complex the proteasome activity. We suggest that LLNle, being able composed of the integral membrane proteins presenilin, nicas- to target two relevant pathways for GBM TIC survival, trin, Aph1, and Pen-2 (14-17). Increased activation of the NOTCH signaling has been reported in several tumor types (18). Recent studies showed that this pathway induces the survival and/or proliferation in GBM and glioma cells (19-22), and Received 5/21/09; revised 8/21/09; accepted 9/9/09; published OnlineFirst 10/27/09. the expression of stem cell markers in glioma cells (21, 22). Grant support: CIPE 2007-Regione Liguria (Stem Cells) and by Regione Liguria- According to these findings, the inhibition of this pathway “Accordo collaborazione scientifica tra Liguria e Piemonte, per l'anno 2008.” The costs of publication of this article were defrayed in part by the payment of leads to depletion of stem-like cells and to the block of the page charges. This article must therefore be hereby marked advertisement in engraftment in embryonal brain tumors (23). Furthermore, accordance with 18 U.S.C. Section 1734 solely to indicate this fact. enhanced NOTCH signaling may lead to the tumor radiore- Authors' contributions: M. Monticone designed research, performed research, analyzed data, wrote the article; M. Fabbi performed research, analyzed data, sistance mechanisms deployed by GBM (24). Targeting the wrote the article; E. Biollo, A. Fabiano, M. Maffei, A. Melotti, and A. Daga NOTCH pathway specifically in GBM TICs seems therefore performed research; W. Giaretti and G. Corte, analyzed data; P. Castagnola a rational approach for exploring novel and hopefully more designed research, performed research, analyzed data, wrote the article. Requests for reprints: Patrizio Castagnola, Istituto Nazionale per la Ricerca sul effective therapeutic strategies for the management of this Cancro, Largo R. Benzi, 10, 16132 Genova, Italy. Phone: 39-0105737500; Fax: malignancy. 39-0105737505. E-mail: [email protected] Copyright © 2009 American Association for Cancer Research. Several molecular tools are available for targeting the doi:10.1158/1541-7786.MCR-09-0225 Notch pathway such as specific siRNAs, shRNAs, or drugs

1822 Mol Cancer Res 2009;7(11). November 2009 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst October 27, 2009; DOI: 10.1158/1541-7786.MCR-09-0225

LLNle Kills Human Glioblastoma Cells 1823

such as γ-secretase inhibitors. The latter are small cell per- 24 hours of treatment with DAPT or LLNle, TICs did not show meant molecules able to inhibit the γ-secretase by distinct me- major changes in the amount of NICD in the samples when chanisms (25). Many researchers are carefully studying these compared with the control sample. At 48 hours, instead, a dra- molecules because they can be potentially effective in cancer matic decrease of the signal corresponding to the NICD in the and in the Alzheimer disease (18, 26). Recently, the results of LLNle-treated TICs with respect to the DMSO and DAPT- a phase I trial and of phase II trials (in Alzheimer disease) treated cells was observed (Fig. 2). No detectable protein deg- with the γ-secretase inhibitor LY450139 have been published radation was observed by Ponceau Red staining performed (27-29). before challenging with the primary antibody (Fig. 2). Similar We presently investigated, by evaluating by flow cytometry results were obtained with TICs from PT2 and PT3 (see Supple- and gene expression profiling, two drugs, N-[N-(3,5-difluoro- mentary Data S1 and S2, respectively). phenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester (DAPT) and Benzyloxicarbonyl-Leu-Leu-Nle-CHO (LLNle), known LLNle Induces Apoptosis in GBM TICs to be active as γ-secretase inhibitors, for their ability to inter- To investigate the killing mechanism underling the activity fere in vitro specifically with survival properties of GBM TICs of LLNle in GBM TICs, we took advantage of a multicolor la- previously obtained in our laboratory (30). Here, we clearly beling system able to detect apoptosis and necrosis at the same show that LLNle is effective in killing these cells by inhibiting, time. This system utilizes the use of allophycocyanin-conjugat- in addition to the NOTCH, also the ubiquitin-proteasome path- ed Annexin V, C12-resazurine, and Sytoxgreen to discriminate way. These data suggest that preclinical studies should definite- metabolically active cells, early apoptotic, late apoptotic, and ly be carried on to evaluate whether LLNle is able to necrotic cells. The nonfluorescent C12-resazurine is reduced significantly improve the survival of immunodeficient mouse to orange-fluorescent resorufin by live cells; the far-red fluores- xenotransplanted with human GBM TICs. cent allophycocyanin-Annexin V binds to exposed phosphati- dylserine on the membrane of apoptotic cells; and Sytoxgreen, Results being cell impermeant, is able to bind to the DNA of cells with The γ-Secretase Inhibitor LLNleu Kills GBM TICs in compromised membranes such as late apoptotic and necrotic Culture cells. These cell subpopulations can be discriminated by flow To determine whether the γ-secretase inhibitors LLNleu and cytometry. After 48 hours of treatment with LLNle, PT1 GBM DAPT have any effect on GBM TICs in vitro, cells obtained TICs showed an increase in the far-red and green-labeled cell from patient (PT)1 (PT1) were cultured with medium alone subpopulation (late apoptotic) and in the far-red and orange-la- or supplemented with different amount of DMSO (vehicle), beled cell subpopulations (early apoptotic) when compared LLNleu, and DAPT. After 48 hours, the cell viability was de- with the DMSO treatment (R3 and R7, respectively, in Fig. termined by using the MTT compound. PT1 GBM TICs treated 3). It should be pointed that most cells in the region R1, after with LLNle displayed a dose-dependent reduction of cell via- 48 hours of treatment with LLNle, are still resorufine positive bility compared with untreated cells (Fig. 1). Cells treated with and hence metabolically active (Fig. 3). The flow cytometric DMSO or with DAPT displayed a reduction of viability ≥50% results obtained after 48 hours of exposure of TICs to DAPT of the untreated controls only at the higher end of the concen- were very similar to those showed by the TICs exposed to trations tested, corresponding to 3.6% and 270 μmol/L, respec- DMSO (data not shown). Similar results were obtained also tively (Fig. 1). These experiments also showed that the viability for PT2 and PT3 GBM TICs (see Supplementary Data S3). obtained with DMSO and DAPT followed a similar trend (Fig. 1). Consistent results were also obtained from PT2 and LLNle Affects the Expression of Genes That Have a Role PT3 (data not shown). We used the 7.5 μmol/L concentration in Proteasome Activation and Cell Cycle Arrest in subsequent experiments with LLNle because it was the con- To investigate the molecular basis underling the LLNle-in- centration yielding ∼50% killing activity in the GBM TICs. We duced apoptosis in GBM TICs, we determined the gene expres- arbitrarily decided to use the 7.5 μmol/L concentration also for sion profile of these cells after exposure to LLNle, DAPT, and DAPT for the following reasons: to use an equimolar concen- DMSO by using GeneChip microarrays. Analysis of GBM TIC tration with respect to LLNle, because the highest concentra- gene expression profile by Principal Component Analysis tion tested (270 μmol/L) often led to the formation of (PCA) revealed a distinct partition between those treated with crystals of the compound and because the 45 μmol/L concen- LLNle and those treated with DAPT and DMSO (Fig. 4). Each tration did not display a higher killing activity than the 7.5 dot in the two-dimensional plot represents one sample. The dis- μmol/L concentration (Fig. 1). tance between any pair of dots was related to the similarity between the two observations in high-dimensional space. Samples LLNle Inhibits the Generation of the NOTCH Intracellular that were near each other in the plot were similar in a large Domain number of variables, i.e., expression level of individual genes. To verify whether LLNle and DAPT actually decrease the Conversely, samples that were far apart in the plot were differ- extent of the Notch protein processing by γ-secretase in PT1 ent in a large number of variables. PCA showed that for each GBM TICs and therefore the generation of the transcriptionally patient, GBM TIC samples were clustered closer to each other active intracellular domain (Notch1 intracellular domain, at 24 hours than at 48 hours (Fig. 4). Furthermore, the 24-hour NICD), we subjected both floating and adherent cells to cell ly- time point samples treated with DMSO and DAPT were clus- sis and Western blot analysis using an antibody able to recog- tered closer to each other than to those treated with LLNle. This nize uniquely this portion of the Notch1 protein (Fig. 2). After distribution in the two-dimensional space was less evident at

Mol Cancer Res 2009;7(11). November 2009 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst October 27, 2009; DOI: 10.1158/1541-7786.MCR-09-0225

1824 Monticone et al.

LLNle and DAPT Effects on NOTCH Target Genes To assess the effects of LLNle and DAPT on the NOTCH pathway in GBM TICs, we examined the expression of NICD transcriptional target genes JAG1, HES1, HEY1, and HEY2 (32-34) as evaluated by the GeneChip microarray studies. The values of mRNA expression for these genes showed that the effects of the two drugs were not homoge- neous across the three patients (see Supplementary Data S5). In particular, we observed, in some cases, only negligible changes in gene expression due to the treatment with DAPT and LLNle, and downregulation of gene expression for some genes in treated cells of some patients at 24 hours but upregulation for the same genes at 48 hours in treated cells of the same patients (see Supplementary Data S5). The only consistent change was observed for the HES1 transcript, which was downregulated by both DAPT and LLNle in FIGURE 1. The γ-secretase inhibitor LLNleu kills GBM stem cells in the cells of all three patients examined at the 24-hours time culture. Survival curves of GBM TIC cells derived from PT1 are based on the results of MTT assays after continuous exposure to drugs for point (see Supplementary Data S5). However, even for HES1 48 h. Average values from at least three independent experiments are at 24 hours, the changes in the expression levels were <2- shown; points, mean; bars, SD. Concentration (v/v) for DMSO (vehicle) was fold in the DAPT and LLNle TIC samples with respect to (1)0.0005%,(2)0.0028%,(3)0.0167%,(4) 0.1002%, (5) 0.6017%, and (6) 3.61%. Concentration for DAPT and LLNle was (1) 0.034 μmol/L, the DMSO-treated cells (see Supplementary Data S5). (2) 0.021 μmol/L, (3)1.25μmol/L, (4)7.5μmol/L, (5)45μmol/L, and (6) 270 μmol/L. LLNle Induces Hsp70 Expression and the Accumulation of Poly-Ubiquitinylated Proteins in a Dose-Dependent 48 hours (Fig. 4). To gain a more mechanistic understanding of Fashion the processes affected by LLNle, the Expression Analysis Sys- To investigate in more detail the effects of LLNle on the pro- temic Explore score (31) was used to identify Gene Ontology teasome pathway in GBM TICs, we studied the expression of (GO) functional categories, which were significantly over- the Hsp70, a known marker of proteasome inhibition, by West- represented. After filtering the results to avoid redundant and/ ern blot analysis (35, 36). Our results show that Hsp70 expres- or generic categories, statistically significant GO terms associ- sion is higher in PT1 cells treated with LLNle for 24 hours than ated with LLNle-regulated genes were found (Table 1). The results clearly show that the genes whose expression changed by a factor 2, after treatment with LLNle in the cells from the three patients examined, are associated to catabolic processes and proteasome activity at 24 hours (Table 1) and to cell cycle, cell proliferation, chromosome segregation, response to stress, and p53 signaling pathway at the 48-hours time point (Table 1). In particular, after 24 hours, LLNle induced the upregulation of transcript coding for several proteasome subunits, whereas after 48 hours, it downregulated several genes required for cell cycle progression, cell proliferation, chromosome segregation, and upregulated several genes having a role as inhibitors of these processes (Table 1). Dif- ferential gene expression by the GBM TICs of the three patients subjected to the three different treatments (LLNle, DAPT, and DMSO) was visualized by a heat map visualization obtained by hierarchical clustering (HCL), which gener- ates a tree (dendrogram) to group similar objects together (Fig. 5). Again, the DAPT gene expression profile resulted very similar to the one determined in the DMSO-treated cells (Table 1; Fig. 5), whereas the one determined for LLNLe was clearly distinct from those determined in the DAPT and DMSO-treated cells. Expression changes were confirmed by Real-time PCR FIGURE 2. LLNle inhibits the generation of the NOTCH intracellular do- analysis of arbitrarily selected genes resulting as deregulated main and enhances the expression of Hsp70. Western blot analysis performed withanti NICD and anti Hsp70 of protein extract from cultures of from the gene array experiments (Fig. 6). A complete list of GBM TIC cells derived from PT1. Cells were treated for the indicated time significantly regulated genes in the comparison between the period with0.1% (v/v) DMSO (vehicle)or with7.5 μmol/L DAPT or LLNle. Equal amount of total proteins were loaded in eachlane, and after blotting, LLNle-treated samples versus DAPT-treated samples or control the membrane was stained using Ponceau Red to show integrity of total samples (DMSO) is provided (see Supplementary Data S4). protein pattern and loading before challenging it with specific antibodies.

LLNle Kills Human Glioblastoma Cells 1825

FIGURE 3. LLNle induces apoptosis in GBM TICs. Flow cytometric analysis of GBM TICs derived from PT1 after 48 hof treatment with0.1% (v/v) DMSO (vehicle) or with 7.5 μmol/L LLNle. Cells were labeled withSytox Green, Resorufin, and Allophycocyanin( APC)-conjugated Annexin V. in PT1 cells treated with DAPT or with DMSO (Fig. 2). This cent emission dramatically with respect to the unbound state, increase was maintained after 48 hours of exposure to LLNle DNA content can be derived by the intensity of the fluorescent (Fig. 2). Similar results were obtained with TICs from PT2 and signal examined by flow cytometry. In Fig. 8, the DNA content PT3 (see Supplementary Data S1 and S3, respectively). To in- of untreated nuclei in the G1 phase peaks at ∼200 units of fluo- vestigate whether proteasome inhibition was dose dependent, rescent light emission, whereas G2+M nuclei have a DNA con- we subjected PT1 GBM TIC cells to a concentration series of tent corresponding to the double value of 400 units and S phase LLNLe and analyzed by immunoblot analysis the expression of nuclei have fluorescence values comprised between 200 and Hsp70 along with that of poly-ubiquitinylated proteins and that 400 units. After 24 hours of treatment with LLNle, PT1 of the NICD (Fig. 7). This experiment clearly shows that a GBM TICs showed an increase of the peak corresponding to dose-dependent increase in Hsp70 exists and that it can be de- G2+M nuclei with respect to the control samples and an in- tected after 48 hours of exposure even at the 1.25 μmol/L crease of nuclei in the S phase with respect to controls (Fig. LLNle concentration (Fig. 7). This concentration of LLNle is 8). In addition, one could observe the clear appearance of par- the lowest at which a decrease of the NICD is detectable ticles related to fragmented (apoptotic) nuclear bodies having (Fig. 6). A clear-cut increasing of poly-ubiquitinylated protein lower fluorescent emission, hence lower DNA content, than un- accumulation was also observed at 1.25 μmol/L LLNle with treated G1 nuclei (Fig. 8). After 48 hours of exposure to LLNle, respect to lower doses. This accumulation peaked at 7.5 a dramatic increase of fragmented nuclei having less than G1 μmol/L (Fig. 7). DNA content was observed along with a still higher peak of nuclei corresponding to the G2+M phase DNA content, and LLNle Induces Cell Cycle Arrest and Mitotic Catastrophe higher fraction of nuclei in S phase (Fig. 8). in GBM TICs On the basis of the results obtained by the DAVID analysis, Proteasome but not Notch-Specific Inhibitors Kill GBM we hypothesized that GBM TICs would undergo cell cycle ar- TICs rest and mitotic catastrophe following LLNle treatment. To test Although DAPT is one of the most selective Notch inhi- this hypothesis, we analyzed the nuclear DNA content by 4′,6- bitors (37), to establish whether another selective Notch in- diamidino-2-phenylindole (DAPI) staining of TICs after treat- hibitor with high solubility and reported specificity (nmol/L ment with LLNle, DAPT, and DMSO. Since this dye binds range) would be more effective in killing GBM TICs, the with a stoechiometric ratio to DNA, and increases its fluores- compound L-685,458 was used. The experiments performed,

1826 Monticone et al.

however, showed that the viability obtained with DMSO and and found that this compound was indeed active with a IC50 L-685,458 followed a similar trend and no inhibition of PT1 of 145.8 μmol/L (Fig. 9). Western blot analysis performed GBM TIC viability was observed up to the 78.1 μmol/L with GBM TICs treated for 48 hours with lactacystin showed concentration (see Supplementary Data S7). The effects ob- a steady increase in the amount of protein poly-ubiquitination tained at the highest concentration of L-685,458 used, 390.5 up to the 145.8 μmol/L concentration (see Supplementary μmol/L, were due to the solvent toxicity (see Supplementary Data S10). However, on molar basis, the effects of lactacystin Data S7). on both inhibition of cell viability and protein poly-ubiquiti- To assess whether L-685,458, although ineffective in kill- nation were achieved at least at one order of magnitude higher ing GBM TICs, was able to inhibit the generation of NICD, than LLNle (compare Fig. 1 with Supplementary Fig. S9 and we analyzed by Western blot PT1 GBM TICs after 48 hours Fig. 7 with Supplementary Data S10). of treatment with this compound. This analysis showed no detectable changes in the amount of NICD at concentrations up to 15.63 μmol/L of L-685,458 (see Supplementary Discussion Data S8). In this study, we investigated the potential killing activity on As we proved that LLNle inhibits the poly-ubiquitinine GBM TICs of two well-known γ-secretase inhibitors with dis- proteasome pathway and kills GBM TICs, we expected that tinct chemical structure, LLNle and DAPT. The hypothesis that a pure proteasome inhibitor would yield similar results. There- γ-secretase inhibitors would have caused GBM TIC cell death fore, we treated PT1 GBM TICs with lactacystin for 48 hours was mainly based on recently published literature, which

FIGURE 4. Microarray analysis performed withThe Institute for Genomic Re- searchmultiple experimental viewer program: PCA. Micro- array analysis of GBM TICs treated with0.1% (v/v) DMSO (vehicle) or with 7.5 μmol/L DAPT or LLNle for the indicated time period. Probe sets associated to dysregulation of gene expression levels among the nine samples for each time point were identified using SAM (see Materials and Meth- ods). PCA is shown to provide the two-dimensional projec- tions onto the plane spanned by the two principal compo- nents for the three different data sets for eachpatient for eachtime point.

LLNle Kills Human Glioblastoma Cells 1827

Table 1. GO (at level 2) and KEGG Pathway Analyses of the Comparison between the LLNle-Treated and the Control (DMSO) GBM TICs from PT1-3

System Gene Category-Term Count % P Most Regulated Genes

7.5 μmol/L LLNle vs 0.1% DMSO 24-h exposure GO biological process Catabolic process 14 13.2 2.80E-04 HSPA1B, MMP3, UFD1L, PSMC4, DEDD2, CYLD, PSMC6, PSMC2, PSMB7, PSMC1, UBR1, SOD1, VCP, PGM1 KEGG pathway Proteasome 5 4.7 1.80E10-5 PSMD6, PSMD12, PSMC2, PSMB7, PSMD2 7.5 μmol/L LLNle vs 0.1% DMSO 48-h exposure GO biological process Cell cycle 65 25.2 1.8E-30 DUSP1, RGS2, PPP1R15A, DDIT3, CITED2, GADD45B, TSPYL2, CCNB1, PLK1, SPC25, MKI67, CCNA2, NUF2, PBK, DLG7 Chromosome segregation 16 6.2 2.7E-15 RIOK3, ESPL1, NCAPD2, NCAPG2, NCAPG, SGOL1, NDC80, NUSAP1, NCAPH, CENPF, SGOL2, CENPE, NUF2, TOP2A, DLG7 Response to stress 48 18.6 1.8E-13 HSPA6, HSPA1B, DUSP1, PPP1R15A, IL11, SGK1, DNAJB1, DDIT3, HSPB8, CITED2, DUSP10, SQSTM1, SERPINE1, HSPA5, TOP2A Cell proliferation 28 10.9 6.10E-06 IL11, ADAMTS1, HBEGF, LIF, CLU, TPX2, TIMELESS, CENPF, BUB1B, BUB1, KIF2C, PLK1, CDCA7, MKI67, DLG7 KEGG pathway Cell cycle 15 5.8 3.30E-09 GADD45B, GADD45A, CHEK2, CCNE2, CHEK1, ESPL1, CDC20, MCM6, SKP2, BUB1B, BUB1, CCNB2, CCNB1, PLK1, CCNA2 p53 signaling pathway 9 3.5 1.4E-05 SERPINE1, GADD45B, GADD45A, SESN2, CHEK2, CCNE2, CHEK1, CCNB2, CCNB1

NOTE: Nonredundant functional categories, number of genes contained within each category and percentages ranked by the degree of over-representation in the category as determined by Expression Analysis Systemic Explore (P) are shown. Redundant categories with similar gene members were removed to yield a single representative category. Up to 15 of the most regulated gene members, ordered from the most upregulated gene (bold) to the most downregulated one, are shown in the last column. Only categories associated with a P < 5.0E-4 are shown. Gene name symbols used are those approved by the Human Genome Organisation Gene Nomenclature Committee (http://www.genenames.org/).

reported that γ-secretase inhibitors kill Kaposi's tumor cell proliferation, and chromosome segregation, together with an in- lines (38), enhance chemotherapy-induced apoptosis in colon creased expression of genes able to actively inhibit these pro- cancer cells (39) and in myeloma cells (40), and reduce the processes and those belonging to the p53 signaling pathway. liferation of lung cancer cells (41). More in detail, after 24 hours of exposure to LLNle, we Our results indicate that LLNle is indeed able to induce found upregulation of genes coding for several proteasomal death by apoptosis in human GBM TICs. We also show that subunits, for proteins involved in ubiquitinylated protein deg- after challenge with LLNle, these cells undergo a G2-M cell radation (i.e., UFD1L, CYLD, UBR1, VCP), and in cell death cycle arrest that is already detectable after 24 hours of treatment signaling (DEDD2). After 48 hours, we found upregulation of a and more evident after 48 hours. Such an arrest is concomitant number of genes implicated in the negative regulation of cell to the appearance of sub-G0-G1 nuclei, or nuclear fragments, cycle and cellular proliferation (i.e., DUSP, RGS2, PPP1R15A), and Annexin V binding, both indicative of apoptosis. in response to stresses (i.e., heat shock proteins, DUSP1, On the other hand, treatment with DAPT, one of the most PPP1R15A, IL11, SGK1, DNAJB1, DDIT3), and in the TP53- selective Notch inhibitors, in our experimental conditions mediated arrest of the cell cycle (SERPINE1, GADD45B, GAD- yields little or no effective GBM stem cell killing and inhibition D45A, SESN2). Among the genes downregulated after the 48-hour of the Notch1 processing. Furthermore, here, we show that an- time point, we found those involved in stem cell survival as other selective Notch inhibitor, L-685,458, is equally ineffec- DLG7 (47), in maintaining cell proliferation (i.e., cyclins, tive in impairing both GBM TIC viability and Notch1 MKI67), and in centrosome maturation and chromosome segre- processing, even when used at concentrations higher than the gation (i.e., PLK1, CENPE, BUB1). ones previously reported as effective (42-45). Interestingly, The upregulation of the expression of some genes coding the potency in growth suppression of DAPT was reported as for proteasomal proteins, as a consequence of the inhibition mild compared with LLNle also in a different cellular system of proteasome activity by both RNA interference and small (46). Some of the possible explanations for these results could molecules such as GM132, is a well-established phenomenon be the requirement for a much higher concentration, an ex- observed in several organisms including mammals (48-51). tremely short half-life of the molecules, or the presence of a Therefore, the data reported here, which show the ability of molecular sink in the cells used for the study. LLNle to inhibit the proteasome activity and concomitantly to Genome-wide expression analysis of the effects of LLNle on upregulate the expression of genes coding for several protea- GBM TICs was a further aim of the study. The results indicate some subunits, are in line with these previous studies. The that the apoptotic process induced by LLNle in the low micro- central role of proteasome inhibition in the activity of LLNle molar range is most likely related to the inhibition of proteaso- on GBM TICs is further supported by the observation that mal degradation of cellular proteins followed by a reduced LLNle induces accumulation of poly-ubiquitinylated proteins expression of a number of genes required for cell cycle, cell that cannot be processed by the inhibited proteasome and by

1828 Monticone et al.

the induction of genes related to the unfolded protein re- as a cystein-proteinase inhibitor (56). The subsequent replace- sponse and endoplasmic reticulum stress (including HSPA1B). ment of the small and hydrophilic acetyl group with a bulkier Similar results have been previously described in different and hydrophobic benzyloxycarbonyl group (z group) at the cell types with proteasome inhibitors (35, 36, 52-54). Further- amino end of the tripeptide led to the z-LLNle molecule, more, the close similarity of the chemical structure of LLNle used in the present study. To the best of our knowledge, there with those of GM132 and N-acetyl-leucyl-leucyl-norleucinal, are no reports specifically addressing the equivalence or dif- well-known proteasome inhibitors, should be underlined (55). ference in proteinases specificity of the parent and daughter In particular, it should be pointed that the latter was reported molecules. However, as a matter of fact, z-LLNle has been

FIGURE 5. Microarray analysis performed withTheIn- stitute for Genomic Research multiple experimental viewer program: HCL. Microarray analysis of GBM TICs treated with0.1% (v/v) DMSO (vehicle) or with7.5 μmol/L DAPT or LLNle for the indicated time period. Heat map visualization obtained by HCL. Ratios for eachprobe relative to the mean value were used to rear- range the gene list on the basis of their expression pattern. Probes corresponding to genes withsimilar regulation trend were placed close to eachoth- er. The color-ratio bar indicates intensity of gene upregulation (red), downregulation (green), and no change (black).

LLNle Kills Human Glioblastoma Cells 1829

FIGURE 6. Real-time RT- PCR validation of microarray data. Real-time RT-PCR analysis performed on GBM TICs treated with0.1% (v/v) DMSO (vehicle) or with 7.5 μmol/L DAPT or LLNle for the indicated time period. DMSO (white columns), DAPT (gray columns), and LLNLe (black columns). Validation of microarray data was accom- plished on randomly selected genes from Table 1 and showed, in arbitrary units, LLNle-dependent regulation of gene expression. Expression levels are relative to the expression of the housekeeping low-density lipoprotein receptor–related protein-associated protein 1 (LRPAP1). Columns, mean; bars, SD. Gene name symbols used are those approved by the Human Genome Organisation Gene Nomencla- ture Committee (http://www. genenames.org/) widely used as γ-secretase and hence as NOTCH inhibitor in With respect to the LLNle effect upon the NOTCH pathway, an increasing number of articles (38, 57-66). Here, we also the following issues should be stressed: first, we showed that show that lactacystin, a pure proteasome inhibitor, inhibits LLNle downregulates in GBM TICs the generation of the GBM TIC viability as expected from the results obtained NICD, which is responsible for the activation of the nuclear with LLNle. However, we show that in our experimental con- signal cascade of the NOTCH pathway. This effect was slightly text, its potency is at least one order of magnitude lower than more evident after 48 hours of treatment. On the contrary, sec- that of LLNle. ond, we did not observe a consistent change in the expression Consistently with our data, two studies published during the levels of several known NOTCH direct target gene transcripts revision of this article show that in breast cancer cells, LLNle in- in TICs of the three patients examined. hibits the proteasome, induces G2-M arrest, and triggers apoptosis Notwithstanding the lack of clear-cut downregulation of (67), and that among LLNle, DAPT, and L685,458, only LLNle NOTCH target genes after treatment of GBM TICs with LLNle, induces cell death mediated by proteasome inhibition (68). we cannot rule out that the inhibition of the generation of the

1830 Monticone et al.

NICD observed contributes to some extent to the killing of solved at the 7.5 mmol/L concentration in DMSO, aliquoted, and these cells. However, our gene expression results along with stored in the dark at −20°C. Cells were treated with 7.5 μmol/L those obtained with lactacystin lead us to favor the hypothesis LLNle or DAPT or vehicle alone (DMSO 0.1%), unless other- that proteasome inhibition is the key process unleashed by wise indicated, and kept in a humidified 5% CO2 atmosphere LLNle, which drives GBM TICs to programmed cell death. at 37°C for the indicated time period. This concentration is The observation that the inhibitory effects in GBM TICs on in the low range reported by several previous in vitro studies both γ-secretase and proteasome activity by LLNle are dose (38, 74-77). The γ-secretase inhibitor L-685,458 and the pro- dependent and occur concomitantly at the same concentration teasome inhibitor lactacystin were obtained from Sigma- range raises a caveat on the use of this molecule as a tout court Aldrich, dissolved at 10 mmol/L in DMSO and 30 mmol/L γ-secretase inhibitor. However, we believe that this dual spec- in H2O, respectively, aliquoted, and stored in the dark at −20°C. ificity may be exploited to target at the same time two relevant pathways for cancer onset and progression (69, 70). Interesting- Western Blot Analysis ly, the association of γ-secretase inhibitors with chemotherapy Cells were detached from Matrigel-coated flasks by incu- has been proposed very recently as a novel approach for the bation in PBS containing 1 mmol/L EDTA and harvested by treatment of metastatic colon cancer (71), whereas the use of centrifugation at 450 g for 5 min. Protein cell extracts were proteasome inhibitors into the clinic dates back of a decade prepared by lysis in 5 mmol/L EDTA, 1% Triton ×100, and and now is part of the standard treatment of some hematologic 0.1% SDS in PBS buffer on ice for 30 min. Cell lysates were malignancies (72). Although a molecule having dual specificity then centrifuged through Qiashredder (Qiagen) to homogenate may be associated to a higher toxicity with respect to more spe- the sample and fragment the DNA. Protein concentrations cific inhibitors, it may have the advantage of a later onset of were quantified (BCA protein assay kit, Thermo Scientific) drug resistance due to the possibility of targeting efficiently and equal amounts of protein were run on NuPage 4-12% two targets (72, 73). Furthermore, our results showing that on Bis-Tris Glycine polyacrylamide gels (Invitrogen) under re- a molar basis LLNle is more effective than lactacystin in GBM ducing conditions and were blotted on Immobilon membrane TIC in vitro killing suggest the possibility that similar results (Millipore) according to the manufacturer's instructions. After could be achieved also in vivo at lower doses with respect to blotting, the membrane was briefly stained with Ponceau S other proteasome specific inhibitors. solution (Sigma-Aldrich); washed with deionized water; This study shows effective in vitro killing activity by LLNle on GBM TICs via apoptotic cell death following γ-secretase and proteasome inhibition. On these bases, future studies in animal models bearing human GBM orthotopic transplant may be foreseen with the goal to assess the efficacy of drugs targeting both the proteasome and γ-secretase, either alone or in combi- nation with other chemotherapic or biological response modi- fiers, in the therapy of this fatal disease.

Materials and Methods Cell Culture GBM TICs were obtained from tumor surgical samples provided by the Neurosurgery Department of the San Martino Hospital in Genoa. An informed consent was obtained from all patients before surgery as required by the Ethic Board. Cell isolation was described in detail elsewhere (30). Prolif- eration medium was DMEM-F12/Neurobasal additioned with 1% v/v B27 supplement, (Life Technologies Ltd), 2 mmol/L L-glutamine (Life Technologies Ltd), recombinant human fibro- blast growth factor (FGF-2; 10 ng/mL; Peprotech), and recombinant human epidermal growth factor (20 ng/mL; Peprotech). The medium was changed twice a week. Under these conditions, TICs grown as a monolayer in flasks coated with Matri- gel (BD Biosciences) expressed stem cell markers, maintained intact self-renewal capacity, maintained partial multilineage dif- FIGURE 7. LLNle induces Hsp70 expression, induces the accumulation of poly-ubiquitinylated proteins, and reduces the expression of NICD ferentiation ability, and gave rise to tumor when injected ortho- in a dose-dependent fashion. Western blot analysis of GBM TICs from PT1 topically in nude mice (30). Cells were seeded the day before treated for 48 hwithLLNle at theindicated concentration and withDMSO challenging with γ-secretase inhibitors and monitored by bright (vehicle) 3.75% (v/v), corresponding to its highest amount present in the highest concentration point of LLNle used. Equal amount of total proteins field microscopy at each experimental time point. were loaded in eachlane, and after blotting, themembrane was stained The γ-secretase inhibitors LLNle and DAPT were obtained, using Ponceau Red to show integrity of the total protein pattern and load- ≥ ing before challenging it with specific antibodies. Protein standard markers lyophilizedwithpurityof 95%byhighperformanceliquid (STD) are indicated by arrowheads on the right along with their mass chromatography (Calbiochem, Merck Chemicals Ltd) and dis- values expressed in kDa.

LLNle Kills Human Glioblastoma Cells 1831

FIGURE 8. LLNle induces cell cycle arrest in G2-M phase. High-resolution flow cytometric analysis of DAPI-stained nuclei of PT1 GBM TICs treated with7.5 μmol/L LLNle (right panels) reveals a substantial increase of nuclei in the S, G2, and M phase of the cell cycle at 24 h followed by an increase of nuclei/fragment having a sub-G1 DNA content (arrowheads) at 48 hwithrespect to cells treated with0.1% DMSO (v/v); left panels. saturated with 5% milk in 20 mmol/L Tris-HCl (pH 7.5), 500 (HBO-100 long life, 100W) and the emitted fluorescence mmol/L NaCl, and 0.05% Tween 20; and then challenged was collected by the Gratz setting (488 blue solid laser shut- with a primary antibody. Horse radish peroxidase–conjugated down; 435 nm long pass filter). goat anti-rabbit or goat anti-mouse IgGs and the SuperSignal To determine the extent of apoptosis, the Vybrant apoptosis west Pico Chemiluminescent substrate (Thermo Scientific) assay kit (Invitrogen) was used. Cells were labeled according to were used for immunodetection. Chemiluminescent signals the manufacturer's instruction. Samples were subjected to flow were acquired by a ChemiDoc XRS apparatus (Bio-Rad Lab- cytometric analysis using a CyAn ADP analyzer equipped with oratories). The rabbit polyclonal antiserum ab8925 specific for NICD and the mouse monoclonal ab2787 specific for Hsp70 were from Abcam. The mouse monoclonal antibody anti– poly-ubiquitinylated-protein (04-262) was purchased from Millipore.

Flow Cytometric Analysis Cells harvested with their media to ensure collection of floating cells along with adherent cells were centrifuged at 980 g for 5 min. To measure the DNA content in cell nuclei, each sample was stained with DAPI as described by Otto (78). Briefly, 1 million cells were incubated in 2 mL of detergent solution (0.1 mol/L Citric acid, 0.5% Tween 20 in water) for 20 min at room temperature. Nuclei were passed over a nylon sieve with a pore size of 50 μm (CellTrics yellow filters, Partec GmbH), pelletted, resuspended in 200 μL of detergent solution, and incubated for 10 min at room temperature with FIGURE 9. The proteasome inhibitor lactacystin kills GBM stem cells in gentle shaking. Finally, sixvolumes of staining solution culture. Survival curves of GBM TIC cells derived from PT1 are based on the results of MTT assays after continuous exposure to drugs for 48 h. (0.4 mol/L Na2HPO4, 5 μmol/L DAPI in water) were added. Average values from at least three independent experiments are shown; Each sample was then analyzed, after 15 min of incubation, bars, SD. Concentration (v/v) for H2O (vehicle; solid line)was(1) 0.027%, (2) 0.081%, (3) 0.0167%, (4) 0.243%, (5) 0.729%, and (6) 2.187%. Con- using a CyflowML multiparameter flow cytometer (Partek). centration for Lactacystin (dashed line)was(1) 2.7 μmol/L, (2) 8.1 μmol/L, Excitation of DAPI was provided by an UV mercury lamp (3) 24.3 μmol/L, (4) 72.9 μmol/L, (5) 218.7 μmol/L, and (6) 656.1 μmol/L.

1832 Monticone et al.

488 and 635 nm lasers and 530/40, 575/25, and 665/20 nm band Individual genes with different expression levels, among the pass filters to detect the three fluorochromes (Dako). three groups, were identified using SAM (81). The false dis- covery rate expressed as q-value was used to evaluate statistical RNA Extraction and Quality Analysis significance, and its threshold was set at 0.02 (2%). For com- Total RNA was isolated using miRNeasy mini kit (Qia- parison purposes, an arbitrary filter was applied excluding all gen) with DNase treatment. RNA concentration and purity genes that did not exhibit a difference in expression of at least were determined from measuring absorbance at 260 and 2-fold. Genes differentially expressed were investigated using μ 280 nm; 2 g total RNA were run on a 1% denaturing gel (a) a multiclass analysis to test differences among the three and 100 ng were loaded on the 2100 Bioanalyzer (Agilent) groups of cells and (b) a two-class analysis within each pair to verify RNA integrity. groups to specify expression changes. Amplification of RNA and Array Hybridization We used PCA to reduce the complexity of high-dimensional According to the recommendations of the manufacturer, data and to simplify the task of identifying patterns and sources 100 ng of total RNA were used in the first-round synthesis of variability in these large data sets. of double-stranded cDNA. The RNA was reverse transcribed The results from SAM were visualized using HCL (82). using a Whole Transcript cDNA synthesis and amplification All the microarray information has been submitted to the National Center for Biotechnology Information Gene Expres- kit (AffymetrixUK Ltd.). The resulting biotin-labeled cRNA 8 was purified using an IVT clean-up kit (Affymetrix) and sion Omnibus Web site. quantified using a UV spectrophotometer (A260/280; Beck- man). An aliquot (15 μg) of cRNA was fragmented by heat Pathways Identification by Expression Analysis Systemic and ion-mediated hydrolysis at 94°C for 35 min. Fragmented Explore cRNA, run on the Bioanalyzer (Agilent Technologies,) to ver- Gene lists from Affymetrixresults were examinedusing the Expression Analysis Systemic Explore program, accessible via ify the correct electropherogram, was hybridized in a hybrid- 9 ization oven (16 h, 45°C) to a Human Gene 1.0 ST array the DAVID Bioinformatics Resources Web site. Expression (Affymetrix) representing whole-transcript coverage. Each Analysis Systemic Explore is a customized stand-alone soft- one of the 28869 genes is represented on the array by ∼26 ware application with statistical functions for discovering biological themes within gene lists. This software assigns genes of probes spread across the full length of the gene, providing a 10 more complete and more accurate picture of gene expression interest into functional categories based on the GO database than the 3′-based expression array design. The washing and and uses the Fisher's exact test statistics to determine the prob- staining procedures of the arrays with phycoerythrin-conjugat- ability of observing the number of genes within a list of interest ed streptavidin (Invitrogen) was completed in the Fluidics Sta- versus the number of genes in each category on the array. A tion 450 (Affymetrix). The arrays were subsequently scanned more detailed analysis of the genes' association with physiolog- ic pathways was done using the Kyoto Encyclopedia of Genes using a confocal laser GeneChip Scanner 3000 7G and the 11 GeneChip Command Console (Affymetrix). and Genomes (KEGG). Each identified process was confirmed through PubMed/Medline.12 GeneChipMicroarray Analysis and Data Normalization Affymetrixraw data files (cell intensity files) were used Reverse Transcription-PCR Analysis as input files in expression console environment (Affyme- Starting from ∼1 μg of total RNA, cDNA was synthesized trix). Briefly, cell intensity files were processed using the by using Oligo(dT)20, random hexamers mix, and a Super- Robust Multi-Array Analysis procedure (79), an algorithm script III first-strand synthesis system supermixfor reverse that is publicly available at the Bioconductor.org website.5 transcription-PCR (RT-PCR; Invitrogen). cDNAs were diluted The Robust Multi-Array Analysis method was used to con- 5 to 20 times, then subjected to PCR analysis. vert the intensities from the multiple probes of a probe set Relative quantification was done by real-time quantitative into a single expression value with greater precision and re- RT-PCR. Briefly, quantitative RT-PCR was done and analyzed duced background noise (relying on the perfect match probes using real-time PCR (ABI Prism 7700 Sequence Detector, only and thus ignoring the mismatch probes) and then to Applied Biosystems). Primers were designed across a common normalize by sketch quantile normalization. Quality assess- exon-exon splice junction by the Primer Express 2.0 software ments were also done in the expression console environment. (Applied Biosystems) to avoid possible signal production from This procedure, based on various metrics, allowed us to potential contaminating genomic DNA. Reactions were carried identify a chip as an outlier (see for details Quality assess- out in triplicates and amplicons were measured by SYBR ment of exon and gene arrays).6 Significance Analysis of Green fluorescence (Applied Biosystems) as per manufac- Microarrays (SAM), PCA of variance and HCL, after mean turer's recommendations. Dissociation curve analysis defined scaling and log2 transformation were performed with the the specificity of the products by the presence of a single dis- software tool from The Institute for Genomic Research mul- sociation peak on the thermal melting curve. tiple experimental viewer7 (80).

8 http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE14581 5 http://www.bioconductor.org 9 http://david.abcc.ncifcrf.gov/ 6 http://www.affymetrix.com/support/technical/whitepapers/exon_gene_arrays_ 10 http://www.geneontology.org/index.shtml qa_whitepaper.pdf 11 http://www.genome.jp/kegg/pathway.html 7 http://www.tigr.org/software/tm4/mev.html 12 http://www.ncbi.nlm.nih.gov/sites/entrez?db=pubmed

LLNle Kills Human Glioblastoma Cells 1833

The gene coding for the low-density lipoprotein receptor– 21. Shih AH, Holland EC. Notch signaling enhances nestin expression in glio- – related protein-associated protein 1 (LRPAP1) was used as the mas. Neoplasia 2006;8:1072 82. 22. Zhang XP, Zheng G, Zou L, et al. Notch activation promotes cell prolifera- endogenous control for normalization because, in the microar- tion and the formation of neural stem cell-like colonies in human glioma cells. ray data, it showed in all conditions the steadiest expression in Mol Cell Biochem 2008;307:101–8. our experimental setting when compared with other housekeep- 23. Fan X, Matsui W, Khaki L, et al. Notch pathway inhibition depletes stem-like ing genes. cells and blocks engraftment in embryonal brain tumors. Cancer Res 2006;66: 7445–52. To avoid possible signal arising from potential contaminat- 24. Eyler CE, Rich JN. Survival of the fittest: cancer stem cells in therapeutic inggenomicDNA,specificprimersforeachgenewerede- resistance and angiogenesis. J Clin Oncol 2008;26:2839–45. signed across a common exon-exon splice junction by the 25. Kornilova AY, Das C, Wolfe MS. Differential effects of inhibitors on the γ- Primer Express software (Applied Biosystems; see Supplemen- secretase complex. Mechanistic implications. J Biol Chem 2003;278:16470–3. tary Data S6). 26. Imbimbo BP. Therapeutic potential of γ-secretase inhibitors and modulators. Curr Top Med Chem 2008;8:54–61. 27. Siemers ER, Quinn JF, Kaye J, et al. Effects of a γ-secretase inhibitor in a – Disclosure of Potential Conflicts of Interest randomized study of patients with Alzheimer disease. Neurology 2006;66:602 4. The authors declare that they have no competing interests in relation to this 28. Siemers ER, Dean RA, Friedrich S, et al. Safety, tolerability, and effects on β γ article. plasma and cerebrospinal fluid amyloid- after inhibition of -secretase. Clin Neuropharmacol 2007;30:317–25. 29. Fleisher AS, Raman R, Siemers ER, et al. Phase 2 safety trial targeting amyloid β production with a γ-secretase inhibitor in Alzheimer disease. Arch Neurol References 2008;65:1031–8. 1. Singh SK, Clarke ID, Terasaki M, et al. Identification of a cancer stem cell in 30. Gangemi RM, Griffero F, Marubbi D, et al. SOX2 silencing in glioblastoma human brain tumors. Cancer Res 2003;63:5821–8. tumor initiating cells causes stop of proliferation and loss of tumorigenicity. Stem 2. Hemmati HD, Nakano I, Lazareff JA, et al. Cancerous stem cells can arise Cells 2008. from pediatric brain tumors. Proc Natl Acad Sci U S A 2003;100:15178–83. 31. Hosack DA, Dennis G, Jr., Sherman BT, Lane HC, Lempicki RA. Identifying 3. Bao S, Wu Q, Sathornsumetee S, et al. Stem cell-like glioma cells promote biological themes within lists of genes with EASE. Genome Biol 2003;4:R70. tumor angiogenesis through vascular endothelial growth factor. Cancer Res 2006; 32. Sansone P, Storci G, Tavolari S, et al. IL-6 triggers malignant features in 66:7843–8. mammospheres from human ductal breast carcinoma and normal mammary 4. Vermeulen L, Sprick MR, Kemper K, Stassi G, Medema JP. Cancer stem cells- gland. J Clin Invest 2007;117:3988–4002. old concepts, new insights. Cell Death Differ 2008;15:947–58. 33. Egan SE, St-Pierre B, Leow CC. Notch receptors, partners and regulators: 5. Salmaggi A, Boiardi A, Gelati M, et al. Glioblastoma-derived tumorospheres from conserved domains to powerful functions. Curr Top Microbiol Immunol identify a population of tumor stem-like cells with angiogenic potential and en- 1998;228:273–324. hanced multidrug resistance phenotype. Glia 2006;54:850–60. 34. Maier MM, Gessler M. Comparative analysis of the human and mouse Hey1 6. Liu G, Yuan X, Zeng Z, et al. Analysis of gene expression and chemoresis- promoter: Hey genes are new Notch target genes. Biochem Biophys Res Com- tance of CD133+ cancer stem cells in glioblastoma. Mol Cancer 2006;5:67. mun 2000;275:652–60. 7. Hambardzumyan D, Becher OJ, Rosenblum MK, Pandolfi PP, Manova-Todor- 35. Bush KT, Goldberg AL, Nigam SK. Proteasome inhibition leads to a heat- ova K, Holland EC. PI3K pathway regulates survival of cancer stem cells residing shock response, induction of endoplasmic reticulum chaperones, and thermotoler- in the perivascular niche following radiation in medulloblastoma in vivo. Genes ance. J Biol Chem 1997;272:9086–92. Dev 2008;22:436–48. 36. Lee DH, Goldberg AL. Proteasome inhibitors cause induction of heat shock 8. Oka N, Soeda A, Inagaki A, et al. VEGF promotes tumorigenesis and angio- proteins and trehalose, which together confer thermotolerance in Saccharomyces genesis of human glioblastoma stem cells. Biochem Biophys Res Commun 2007; cerevisiae. Mol Cell Biol 1998;18:30–8. 360:553–9. 37. Weihofen A, Lemberg MK, Friedmann E, et al. Targeting presenilin-type 9. Bao S, Wu Q, McLendon RE, et al. Glioma stem cells promote radioresis- aspartic protease signal peptide peptidase with γ-secretase inhibitors. J Biol Chem tance by preferential activation of the DNA damage response. Nature 2006;444: 2003;278:16528–33. 756–60. 38. Curry CL, Reed LL, Broude E, Golde TE, Miele L, Foreman KE. Notch 10. Dalerba P, Cho RW, Clarke MF. Cancer stem cells: models and concepts. inhibition in Kaposi's sarcoma tumor cells leads to mitotic catastrophe through Annu Rev Med 2007;58:267–84. nuclear factor-κB signaling. Mol Cancer Ther 2007;6:1983–92. 11. Jarriault S, Brou C, Logeat F, Schroeter EH, Kopan R, Israel A. Signalling 39. Akiyoshi T, Nakamura M, Yanai K, et al. γ-secretase inhibitors enhance tax- downstream of activated mammalian Notch. Nature 1995;377:355–8. ane-induced mitotic arrest and apoptosis in colon cancer cells. Gastroenterology 12. Lowell S, Benchoua A, Heavey B, Smith AG. Notch promotes neural lineage 2008;134:131–44. entry by pluripotent embryonic stem cells. PLoS Biol 2006;4:e121. 40. Nefedova Y, Sullivan DM, Bolick SC, Dalton WS, Gabrilovich DI. Inhibi- 13. Tanigaki K, Nogaki F, Takahashi J, Tashiro K, Kurooka H, Honjo T. Notch1 tion of Notch signaling induces apoptosis of myeloma cells and enhances sensi- and Notch3 instructively restrict bFGF-responsive multipotent neural progenitor tivity to chemotherapy. Blood 2008;111:2220–9. cells to an astroglial fate. Neuron 2001;29:45–55. 41. Konishi J, Kawaguchi KS, Vo H, et al. γ-secretase inhibitor prevents Notch3 14. Chung HM, Struhl G. Nicastrin is required for Presenilin-mediated trans- activation and reduces proliferation in human lung cancers. Cancer Res 2007;67: membrane cleavage in Drosophila. Nat Cell Biol 2001;3:1129–32. 8051–7. 15. Lee SF, Shah S, Li H, Yu C, Han W, Yu G. Mammalian APH-1 interacts with 42. Yao J, Duan L, Fan M, Wu X. γ-secretase inhibitors exerts antitumor activity presenilin and nicastrin and is required for intramembrane proteolysis of amy- via down-regulation of Notch and Nuclear factor κ B in human tongue carcinoma β – loid- precursor protein and Notch. J Biol Chem 2002;277:45013 9. cells. Oral Dis 2007;13:555–63. 16. Yu G, Nishimura M, Arawaka S, et al. Nicastrin modulates presenilin-mediat- 43. Williams CK, Segarra M, Sierra Mde L, Sainson RC, Tosato G, Harris AL. ed notch/glp-1 signal transduction and βAPP processing. Nature 2000;407:48–54. Regulation of CXCR4 by the Notch ligand δ-like 4 in endothelial cells. Cancer 17. Selkoe DJ, Wolfe MS. Presenilin: running with scissors in the membrane. Res 2008;68:1889–95. – Cell 2007;131:215 21. 44. Heo JS, Lee MY, Han HJ. Sonic hedgehog stimulates mouse embryonic stem 18. Shih Ie M, Wang TL. Notch signaling, γ-secretase inhibitors, and cancer cell proliferation by cooperation of Ca2+/protein kinase C and epidermal growth therapy. Cancer Res 2007;67:1879–82. factor receptor as well as Gli1 activation. Stem Cells 2007;25:3069–80. 19. Hulleman E, Quarto M, Vernell R, et al. A role for the transcription factor 45. Liu CX, Ranganathan S, Robinson S, Strickland DK. γ-Secretase-mediated HEY1 in glioblastoma. J Cell Mol Med 2008. release of the low density lipoprotein receptor-related protein 1B intracellular domain suppresses anchorage-independent growth of neuroglioma cells. J Biol 20. Purow BW, Haque RM, Noel MW, et al. Expression of Notch-1 and its li- – gands, Δ-like-1 and Jagged-1, is critical for glioma cell survival and proliferation. Chem 2007;282:7504 11. Cancer Res 2005;65:2353–63. 46. Tohda S, Sato T, Kogoshi H, Fu L, Sakano S, Nara N. Establishment of a

1834 Monticone et al.

novel B-cell lymphoma cell line with suppressed growth by γ-secretase inhibi- 64. Song LL, Peng Y, Yun J, et al. Notch-1 associates with IKKα and regulates tors. Leuk Res 2006;30:1385–90. IKK activity in cervical cancer cells. Oncogene 2008;27:5833–44. 47. Gudmundsson KO, Thorsteinsson L, Sigurjonsson OE, et al. Gene expres- 65. Bansal K, Narayana Y, Patil SA, Balaji KN. M. bovis BCG induced expression analysis of hematopoietic progenitor cells identifies Dlg7 as a potential stem sion of COX-2 involves nitric oxide-dependent and -independent signaling path- cell gene. Stem Cells 2007;25:1498–506. ways. J Leukoc Biol 2009. 48. Xie Y, Varshavsky A. RPN4 is a ligand, substrate, and transcriptional 66. Rosati E, Sabatini R, Rampino G, et al. Constitutively activated Notch sig- regulator of the 26S proteasome: a negative feedback circuit. Proc Natl Acad naling is involved in survival and apoptosis resistance of B-CLL cells. Blood Sci U S A 2001;98:3056–61. 2009;113:856–65. 49. Wojcik C, DeMartino GN. Analysis of Drosophila 26 S proteasome using 67. Rasul S, Balasubramanian R, Filipovic A, Slade MJ, Yague E, Coombes RC. – RNA interference. J Biol Chem 2002;277:6188 97. Inhibition of γ-secretase induces G2-M arrest and triggers apoptosis in breast can- – 50. Meiners S, Heyken D, Weller A, et al. Inhibition of proteasome activity in- cer cells. Br J Cancer 2009;100:1879 88. duces concerted expression of proteasome genes and de novo formation of Mam- 68. Han J, Ma I, Hendzel MJ, Allalunis-Turner J. The cytotoxicity of γ-secretase malian proteasomes. J Biol Chem 2003;278:21517–25. inhibitor I to breast cancer cells is mediated by proteasome inhibition, not by γ 51. Lundgren J, Masson P, Realini CA, Young P. Use of RNA interference and -secretase inhibition. Breast Cancer Res 2009;11:R57. complementation to study the function of the Drosophila and human 26S protea- 69. Miele L, Miao H, Nickoloff BJ. NOTCH signaling as a novel cancer thera- some subunit S13. Mol Cell Biol 2003;23:5320–30. peutic target. Curr Cancer Drug Targets 2006;6:313–23. 52. Fels DR, Ye J, Segan AT, et al. Preferential cytotoxicity of bortezomib to- 70. Yang Y, Kitagaki J, Wang H, Hou DX, Perantoni AO. Targeting the ubiqui- ward hypoxic tumor cells via overactivation of endoplasmic reticulum stress path- tin-proteasome system for cancer therapy. Cancer Sci 2009;100:24–8. ways. Cancer Res 2008;68:9323–30. 71. Meng RD, Shelton CC, Li YM, et al. γ-Secretase inhibitors abrogate oxali- 53. Kawazoe Y, Nakai A, Tanabe M, Nagata K. Proteasome inhibition leads to platin-induced activation of the Notch-1 signaling pathway in colon cancer cells the activation of all members of the heat-shock-factor family. Eur J Biochem resulting in enhanced chemosensitivity. Cancer Res 2009;69:573–82. 1998;255:356–62. 72. Orlowski RZ, Kuhn DJ. Proteasome inhibitors in cancer therapy: lessons 54. Kim D, Kim SH, Li GC. Proteasome inhibitors MG132 and lactacystin from the first decade. Clin Cancer Res 2008;14:1649–57. hyperphosphorylate HSF1 and induce hsp70 and hsp27 expression. Biochem 73. Ruckrich T, Kraus M, Gogel J, et al. Characterization of the ubiquitin-pro- Biophys Res Commun 1999;254:264–8. teasome system in bortezomib-adapted cells. Leukemia 2009. 55. Rock KL, Gramm C, Rothstein L, et al. Inhibitors of the proteasome block 74. Hartmann T, Bieger SC, Bruhl B, et al. Distinct sites of intracellular pro- the degradation of most cell proteins and the generation of peptides presented on duction for Alzheimer's disease A β40/42 amyloid peptides. Nat Med 1997;3: MHC class I molecules. Cell 1994;78:761–71. 1016–20. 56. Sasaki T, Kishi M, Saito M, et al. Inhibitory effect of di- and tripeptidyl al- – 75. Gal H, Makovitzki A, Amariglio N, Rechavi G, Ram Z, Givol D. A rapid dehydes on calpains and cathepsins. J Enzyme Inhib 1990;3:195 201. assay for drug sensitivity of glioblastoma stem cells. Biochem Biophys Res Com- 57. Qin JZ, Stennett L, Bacon P, et al. p53-independent NOXA induction over- mun 2007;358:908–13. comes apoptotic resistance of malignant melanomas. Mol Cancer Ther 2004;3: 76. Kogoshi H, Sato T, Koyama T, Nara N, Tohda S. γ-Secretase inhibitors – 895 902. suppress the growth of leukemia and lymphoma cells. Oncol Rep 2007;18: 58. Cuevas IC, Slocum AL, Jun P, et al. Meningioma transcript profiles reveal 77–80. deregulated Notch signaling pathway. Cancer Res 2005;65:5070–5. 77. Aleksic T, Feller SM. γ-Secretase inhibition combined with platinum com- 59. Bellavia D, Mecarozzi M, Campese AF, et al. Notch3 and the Notch3-upre- pounds enhances cell death in a large subset of colorectal cancer cells. Cell Com- gulated RNA-binding protein HuD regulate Ikaros alternative splicing. EMBO J mun Signal 2008;6:8. 2007;26:1670–80. 78. Otto FJ. High-resolution analysis of nuclear DNA employing the fluoro- – 60. Ban J, Bennani-Baiti IM, Kauer M, et al. EWS-FLI1 suppresses NOTCH- chrome DAPI. Methods Cell Biol 1994;41:211 7. activated p53 in Ewing's sarcoma. Cancer Res 2008;68:7100–9. 79. Irizarry RA, Hobbs B, Collin F, et al. Exploration, normalization, and sum- maries of high density oligonucleotide array probe level data. Biostatistics 2003; 61. Ferretti E, Tosi E, Po A, et al. Notch signaling is involved in expression of 4:249–64. thyrocyte differentiation markers and is down-regulated in thyroid tumors. J Clin Endocrinol Metab 2008;93:4080–7. 80. Saeed AI, Sharov V, White J, et al. TM4: a free, open-source system for – 62. Lee CW, Raskett CM, Prudovsky I, Altieri DC. Molecular dependence of microarray data management and analysis. Biotechniques 2003;34:374 8. estrogen receptor-negative breast cancer on a notch-survivin signaling axis. Can- 81. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied cer Res 2008;68:5273–81. to the ionizing radiation response. Proc Natl Acad Sci U S A 2001;98:5116–21. 63. Narayana Y, Balaji KN. NOTCH1 up-regulation and signaling involved in 82. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and dis- Mycobacterium bovis BCG-induced SOCS3 expression in macrophages. J Biol play of genome-wide expression patterns. Proc Natl Acad Sci U S A 1998;95: Chem 2008;283:12501–11. 14863–8.

z-Leucinyl-Leucinyl-Norleucinal Induces Apoptosis of Human Glioblastoma Tumor−Initiating Cells by Proteasome Inhibition and Mitotic Arrest Response

Massimiliano Monticone, Emanuela Biollo, Andrea Fabiano, et al.

Mol Cancer Res 2009;7:1822-1834. Published OnlineFirst October 27, 2009.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-09-0225

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2010/10/26/1541-7786.MCR-09-0225.DC1

Cited articles This article cites 78 articles, 33 of which you can access for free at: http://mcr.aacrjournals.org/content/7/11/1822.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/7/11/1822.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/7/11/1822. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.