Cancer Gene Therapy (2009) 16, 731–740 r 2009 Nature Publishing Group All rights reserved 0929-1903/09 $32.00 www.nature.com/cgt

ORIGINAL ARTICLE Combined inhibition of PAK7, MAP3K7 and CK2a kinases inhibits the growth of MiaPaCa2 pancreatic cancer cell xenografts V Giroux, JL Iovanna, S Garcia and J-C Dagorn Centre de Recherche INSERM, Unite´ 624, Stress Cellulaire, Universite´ de la Me´diterrane´e, Case 915 Parc Scientifique de Luminy, Marseille, France

A panel of kinases whose inhibition increased apoptosis of pancreatic adenocarcinoma cells in vitro was recently established. The aim of this work was to observe in a mouse xenograft model whether inhibition of these kinases would alter pancreatic tumor growth. Rate of apoptosis, caspase-3 activity and cell viability were assessed in two pancreatic cancer cell lines, MiaPaCa2 and BxPC3, after inhibiting selected kinases by transfection of specific siRNAs. For in vivo experiments, MiaPaCa2 cells were injected into the pancreas of nude mice, where they formed tumors. Inhibition of kinases was obtained by repeated intraperitoneal (i.p.) injections of modified O-Methyl (OMe) siRNAs specific for the selected kinases. Tumor volumes were assessed after 21 days. Among selected kinases, PAK7, MAP3K7 and CK2a were those whose inhibition increased apoptosis the most in vitro. Simultaneous inhibition of two of them increased apoptosis up to five times. Moreover, inhibiting these kinases had little effect on 10 non-pancreatic cell lines, suggesting pancreatic specificity. In vivo, OMe-siRNAs induced significant but incomplete inhibition of kinase expression (45–75%). Nevertheless, such inhibition resulted in a twofold increase in caspase-3 activity in tumors and a strong reduction in tumor volume (about 75%). In vivo inhibition by OMe-siRNAs of three survival kinases apparently specific for pancreatic cancer cells, PAK7, MAP3K7 and CK2a, decreases significantly the growth of xenografted MiaPaCa2 cells. This strategy is therefore of potential clinical interest. Cancer Gene Therapy (2009) 16, 731–740; doi:10.1038/cgt.2009.22; published online 10 April 2009 Keywords: kinases; siRNAs; pancreatic cancer; in vivo experiments

Introduction farnesyltransferase, matrix metalloproteases, cyclooxy- genase-2 and lipoxygenase have entered clinical trials In spite of many efforts, the past decades have seen little but, unfortunately, the first evaluations are not encoura- 5 progress in the treatment of pancreatic adenocarcinoma.1 ging. The failure of the present anti-cancer strategies The 5-year survival rate remains under 5%, except for means that new directions should be explored. One patients whose cancer is diagnosed early enough to allow possibility would be to target the mechanisms by which surgical resection, for whom it can increase to 20%. cancer cells escape apoptosis. Kinases control the rever- However, they represent only 10–20% of the patients. sible process of phosphorylation and are deregulated Classical chemotherapeutic agents that are efficient in in many diseases, such as cancer. As protein and lipid other cancers have no effect in the pancreas. The cytosine phosphorylations control cell survival signaling, strategies analog gemcitabine is the only molecule that provides targeting kinases are very good candidates as new, some survival benefit to patients with advanced stage of more efficient therapies. The currently available kinase the disease,2 but improvement is limited, and attempts to inhibitors are effective treatments for some cancers. For increase gemcitabine efficacy by modulating pharmaco- example, drugs such as imatinib and gefitinib inhibit kinetic parameters or by using the drug in combination tumor growth by antagonizing specific survival kinases, with conventional cytotoxic drugs such as oxaliplatin thus reversing the malignant phenotype of chronic failed.3,4 More recently, new drugs such as inhibitors of myeloid leukemia and non-small-cell lung cancer. Un- fortunately, most tumors are heterogeneous and poorly characterized, making it difficult to treat them efficiently Correspondence: Dr J-C Dagorn, Centre de Recherche INSERM, Unite´624, Stress Cellulaire, Parc Scientifique et Technologique de with a single agent, and combination therapies targeting multiple signaling pathways are now privileged. The Luminy, Case 915, Marseille 13288 Cedex 9, France. 6 E-mail: [email protected] description of the complete human kinome allowed us Received 27 September 2008; revised 21 December 2008; accepted 23 to establish, in vitro, a catalog of pancreatic cancer cell January 2009; published online 10 April 2009 survival kinases.7 More recently, modified siRNAs Combining kinase inhibitors in pancreatic cancer V Giroux et al 732 allowing inhibition of specific genes in vivo have become histone-DNA fragments is expressed as fold increase in available. Using this tool, we could assess the influence of absorbance as compared with control (non-silencing) combined inhibition of survival kinases in vivo in a mouse siRNA. Experiments were carried out in triplicate and xenograft model of pancreatic adenocarcinoma. repeated three times.

Flow cytometry Materials and methods Cells were washed with phosphate-buffered saline (PBS) and fixed in cold ethanol 70% for 30 min at 4 1C. After a Cell lines and cell culture conditions wash with PBS, cells were treated with 50 ml RNAse A MiaPaCa-2, MCF7, SaOS-2, A375, SW872, MDA- (100 mg ml–1), labeled with 200 ml propidium iodide MB231, LS174T and U2OS cell lines were grown in (20 mg ml–1) and immediately analyzed by flow cytometry Dulbecco’s modified Eagle’s medium (DMEM) supple- (FACSCalibur, Becton Dickinson, Franklin-Lakes, NJ). mented with 10% fetal bovine serum, 2 mML-glutamine, –1 –1 Cell death analysis was carried out on 30 000 100 IU ml penicillin G and 100 mgml streptomycin. cells, evaluating the sub-G1 ratio (FlowJo, Ashland, BxPC3, HRT18, H1299 and H358 cells were cultivated in OR). The experiments were repeated at least three times RPMI 1640 medium in the presence of 2 mML-glutamine, –1 and each sample was assayed in triplicate. 4.5 g l glucose, 10 mM Hepes, 1 mM sodium pyruvate, 10% fetal bovine serum, and 100 IU ml–1 penicillin G and –1 Viability assay 100 mgml streptomycin. All cell lines were routinely Two days after siRNA transfection in 96-well plates, 10 ml cultivated in humidified 5% CO2 atmosphere. of the MTS reagent mixture (3-4,5-dimethylthiazol-2-yl)- 5-3-carboxymethoxyphenyl)-2-4-sulfophenyl)-2H-tetrazo- siRNA design lium) obtained from Promega, Madison, WI was added The siRNAs used in this study were designed by per well, the plates were incubated at 37 1C for 45 min, Qiagen, Courtaboeuf, France. The sequences are PAK7 and the absorbance at 490 nm was monitored. The 0 0 siRNA (sense 5 -GGUGUGCACGUUUCAUUAd(TT)-3 ), percentage of viable cells in a well was obtained from 0 MAP3K7 siRNA (sense 5 -AGAAUAUAUGAAAGU the ratio of absorbances of cells transfected with kinase 0 0 UCAAd(TT)-3 ), CK2a siRNA (sense 5 -CGUUGCUU siRNAs to cells transfected with control siRNA. GUGGAUUUAUAd(TT)-30) and non-silencing (control) 0 siRNA (sense 5 -UUCUCCGAACGUGUCACGUd Caspase-3 activity 0 (TT)-3 ). siRNAs with O-Methyl (OMe)-modified nucleo- Caspase-3 activity was measured using the Apo-ONE tides were used for in vivo experiments using this pattern Homogeneous Caspase-3 Assay Fluorometric Kit 0 (sense 5 -OMeX–OMeX–OMeX–OMeX–OMeX–XXXXX (Promega). Two days after siRNA transfection in 48-well 0 XXX–OMeX–OMeX–OMeX–OMeX–OMeX-d(TT)–3 ). plates, caspase-3 activity was measured by cleavage of the fluorogenic substrate Z-DEVD-R110 according to the Transfection manufacturer’s (Promega) instructions. Cells were seeded for 24 h in 96-well, 48-well and 6-well 2 plates 40 (40 000 cells per cm ). Unmodified siRNAs Semiquantitative RT-PCR analysis were transfected in cells with a mixture of 50 nM siRNA Total RNA was extracted using Trizol (Invitrogen, Carlsbad, and X-tremeGENE siRNA Transfection Reagent (Roche, CA, USA) following the manufacturer’s description. RNAs Neuilly, France). OMe-siRNAs were used at a concen- (1 mg) were analyzed by RT-PCR with the Promega One- tration of 50 nM using JetPEI Transfection Reagent Step RT-PCR system. RT-PCR was carried out using (Polyplus, Illkirch, France). Non-silencing unmodified various numbers of cycles to verify that the conditions or OMe-siRNA was used as negative control. chosen were within the linear range. The mRNA encoding PAK7 was specifically amplified with sense (50-AGAAGT Apoptosis assay TTACCGGCCT-30) and antisense (50-CTTTTCGGTCGT Two days after siRNA transfection in 96-well plates, GTAGT-30) primers. MAP3K7 was specifically amplified apoptosis was assessed by an ELISA that quantifies with sense (50-ACTCACTTGATGCGGT-30)andantisense cytoplasmic nucleosomes produced during apoptosis (Cell (50-CGGCGATCCTAGCTTC-30) primers. CK2a was spe- Death Detection ELISA plus, Roche). The 96-well plates cifically amplified with sense (50-ATCTTTCGGAAGGAG were centrifuged (200 g) for 10 min, the supernatant was CCATT-30)andantisense(50-TATCGCAGCAGTTTGTC discarded and lysis buffer was added. After lysis, the CAG-30) primers. As a control, the transcript coding for samples were centrifuged and 20 ml of the supernatant was GAPDH was specifically amplified with sense (50-GGGAA transferred to a streptavidin-coated microtitration plate. GCTCACTGGCATGGCCTTCC-30)andantisense(50-CA Biotin-labeled anti-histone antibodies and peroxidase- TGTGGGCCATGAGGTCCACCAC-30) primers. Reverse conjugated anti-DNA antibodies were added to each well, transcription was carried out for 45 min at 45 1C, followed by and the plate was incubated at room temperature for 2 h. 24–32 cycles of PCR, each cycle consisting in a denaturing After three washes with buffer, the peroxidase substrate step for 30 s at 95 1C, an annealing step for 1 min at 57 1C was added to each well to quantitate the captured and a polymerization step for 1 min at 72 1C. PCR products nucleosomes. After a 5-min incubation, the plates were were separated on a 2.0% agarose gel containing ethidium read at 405 nm in a microplate reader. The enrichment in bromide and photographed under ultraviolet light.

Cancer Gene Therapy Combining kinase inhibitors in pancreatic cancer V Giroux et al 733 Quantitative RT-PCR analysis Statistical analysis First-strand cDNA was synthesized in 20 ml reaction Results shown represent means±s.d. Statistical analysis with 1 mg total RNA using Expand Reverse Transcrip- was performed with Student’s t-test and ANOVA tests. tase (Roche). Initially, RNA was incubated with 1 ml random primers for 10 min at 65 1C and then quickly put on ice. On adding 4 ml reaction buffer, 2 ml DDT, 20 nmol for dNTP, 0.5 ml RNAse inhibitor and 1 ml reverse Results transcriptase, the reaction was incubated for 10 min at Combined inhibition of PAK7, MAP3K7 and CK2a 30 1C and then 45 min at 42 1C. Quantitative PCR was survival kinases increases apoptosis of pancreatic carried out with the Light Cycler system (Roche) and cancer cells in vitro Takara reagents. Five microliters of 10-fold diluted We have shown earlier that simultaneous inhibition of cDNA were mixed with 10 ml SYBR Premix Ex Taq two or three pancreatic cancer survival kinases could have (including Taq polymerase, reaction buffer, MgCl2, a stronger effect on apoptosis than a single inhibition. As SYBR green I dye and dNTP mix) and 4 nmol forward a follow-up to this work and to look for the most efficient and reverse primers (TATA-binding protein (TBP) combination, we chose in the recently published catalog primers are used as a control) in a volume of 20 ml. After of 56 pancreatic cancer survival kinases7 the 10 kinases an initial Taq activation for 10 s at 95 1C, Light Cycler whose inhibition showed the highest activity on apoptosis PCR was carried out using 45–55 cycles with the and monitored the effect of their combined inhibition. following cycling conditions: 95 1C for 5 s, 58 1C for 6 s These kinases are the following (name and access and 72 1C for 12 s. Fluorescence reading at the end of this number): PAK7 (p21(CDKN1A)-activated kinase 7, third segment, at high temperature to melt unspecific NM_177990); CK2a (casein kinase 2, a 1 polypeptide, PCR products, ensures quantification of the specific NM_177560); AK1 (adenylate kinase 1, NM_000476); product only. Directly after the PCR, the machine GRAF (GTPase regulator associated with focal adhesion performs a melting curve analysis by slowly increasing kinase pp125(FAK), NM_015071); MAP3K (mitogen- the temperature from 68 to 95 1C, with a continuous activated protein kinase kinase kinase 7, NM_003188); registration of changes in fluorescence emission intensity. DCK (deoxycytidine kinase, NM_000788); DGKH (diacyl- Each sample was analyzed in duplicate and the experi- glycerol kinase Z, NM_178009); TNNI3K (TNNI3 inter- ment was repeated three times. Results were analyzed acting kinase, NM_015978); PRPF4B (PRP4 pre-mRNA using RelQuant (Roche) and expressed as percent of processing factor 4 homolog B (yeast), NM_176800); TPK1 control values. (thiamin pyrophosphokinase 1, NM_022445). MiaPaCa2 cells were transfected with dual combina- In vivo experiments tions of survival kinase-specific siRNAs. Two days Six-week-old male Swiss Nude mice were anesthetized by after transfection, apoptosis was measured. Combi- i.p. injection of a ketamine–xylazine mix (100 mg kg–1 nations of MAP3K7/CK2a, PAK7/CK2a and PAK7/ 10 mgkg–1) and the abdomen was opened through a MAP3K7 showed the highest induction of apoptosis midline incision. MiaPaCa2 cells (107 cells in 200 ml PBS) (3.55, 3.48 and 3.42-fold, respectively) (Figure 1). These were injected directly into the pancreas. This procedure is three survival kinases were chosen for subsequent known to generate intrapancreatic xenografts within 21 experiments. days.8 The next day, 30 mg OMe-siRNAs þ and 25 ml To confirm earlier results showing that simultaneous JetPEI 4 were incubated in 160 ml JetPEI buffer (10 mM inhibition of two kinases might improve apoptosis HEPES–150 mM NaCl, pH 7.4) for 1 h and injected i.p. induction, MiaPaCa2 and BxPC3 cell lines were trans- The siRNA injection was repeated every third day (seven fected with siRNAs for the selected kinases, alone or in injections in total). Animals were weighed before each dual or triple combinations, and apoptosis was monitored injection. Twenty days after injection of the cells, the 2 days later (Figure 2). Single inhibitions resulted in signi- animals were killed and the presence of intrapancreatic ficant increases in apoptosis, as evidenced by cytoplasmic tumors was assessed. Dimensions of pancreatic tumors nucleosome assay, in both cell lines. For example, were measured with a caliper and tumor volume was inhibition of CK2a in MiaPaCa-2 increased the apoptosis calculated as (4p/3) (width/2)2 (length/2). 2.5-fold and inhibition of MAP3K7 in BxPC3 increased the apoptosis 4-fold. Simultaneous inhibition of two or Immunohistochemistry of active caspase-3 three kinases resulted in further increase, reaching fivefold Tissues were formalin-fixed and paraffin-embedded. induction of apoptosis in MiaPaCa2 cells and sixfold After deparaffinization and blocking of nonspecific in BxPC3 (Figure 2a). As a control, apoptosis was also binding, sections were incubated with an anti-human assessed by flow cytometry in MiaPaCa2 cells. In cells active-caspase-3 antibody (1:200; Promega). The immuno- transfected with control siRNA, the sub-G1 fraction peroxidase procedure was carried out using a Vectastain represented about 8% of the cells and increased upon goat anti-rabbit kit (Vector Laboratories, Burlingame, transfection with survival kinase siRNAs, reaching a CA). The active caspase-3-labeled surface was quantified maximum of 29% for the PAK7/MAP3K7 combination with an Olympus BX61 automated microscope ( 10 (Figure 2b). objective), using a Samba 2050 image analyzer (Samba The influence of kinase inhibition on cell viability was Technologies, Meylan, France). also assessed. Viability was significantly altered by kinase

Cancer Gene Therapy Combining kinase inhibitors in pancreatic cancer V Giroux et al 734 3.7

3.6

3.5

3.4

3.3

3.2

3.1

3.0 Induction of apoptosis

0

PAK7 X X X X X X X X X CK2 X X X X X X X X X AK1 X X X X X X X X X GRAF X X X X X X X X X MAP3K7 X X X X X X X X X DCK X X X X X X X X X DGKH X X X X X X X X X TNNI3K X X X X X X X X X PRPF4B X X X X X X X X X TPK1 X X X X X X X X X

Figure 1 Induction of apoptosis in MiaPaCa2 cells after inhibition of couples of selected kinases. The 10 kinases are those whose single inhibition is most efficient at promoting apoptosis. Inhibition was obtained by transfecting the cells with specific siRNAs as described in Materials and methods. Apoptosis is given as fold increase respective to control (transfection with scrambled siRNAs). Results represent mean±s.e. for three independent experiments. Black columns refer to couples whose inhibition generates the highest apoptosis.

inhibition. BxPC3 cells were apparently more sensitive O-Me-modified siRNAs (OMe-siRNAs) of survival than MiaPaCa2 cells as inhibition with any of the three kinases inhibit efficiently mRNA expression in vitro and single kinases resulted in diminution by half of the induce apoptosis viability, with no further effect of multiple inhibitions. The aim of this work was to observe whether inhibition of In MiaPaCa2 cells, however, single inhibitions had specific kinases would affect pancreatic cancer progres- more moderate effects. Dual inhibitions were apparently sion in a mouse xenograft model. As no chemical more efficient than single inhibitions, but this was not inhibitor was available for some of the kinases selected confirmed by statistical analysis. Finally, no improvement earlier in vitro, we decided to inhibit the kinases in vivo was observed when all three kinases were inhibited with the siRNA strategy, which proved successful in vitro. (Figure 2c). However, siRNA approaches in vivo require that (i) the To assess specificity of the selected kinases for siRNAs remain reasonably functional in an environment pancreatic cancer cells, caspase-3 activity and cell viability that contains nucleases and (ii) intact siRNAs be were assessed in MiaPaCa2 cells and in 10 non-pancreatic efficiently delivered to the target cells. Native siRNAs cancer cells after PAK7, MAP3K7 and CK2a siRNA are rapidly degraded, and mammalian cells do not readily transfections (Figure 3). In MiaPaCa2 cells, caspase-3 take up naked nucleic acids. We opted for chemically activity was strongly induced after siRNA transfection modified siRNAs in which the five first bases in 30 and 50 (3.4, 3.3 and 3.7 for PAK7, MAP3K7 and CK2a siRNAs, were 20-O-methylated. These modifications are supposed respectively), whereas cell viability was strongly decreased to increase siRNA resistance to nucleases while main- (by 36, 35 and 39% for PAK7, MAP3K7 and CK2a taining their activity.9 As a vector to target the siRNAs to siRNAs, respectively). None of the other cell lines showed pancreatic cancer cells, we used jet-PEI, a polyethyleni- significant increase of caspase-3 activity after inhibition of mine mixture of undisclosed composition, supposed to the three kinases. The only significant increases were allow very efficient transfection in vivo.10 This strategy observed in MDA-MB 231 cells after inhibition of had to be validated in vitro before working in vivo. MAP3K7 and CK2 and in HRT18 cells transfected with MiaPaCa2 cells were transfected with PAK7, MAP3K7 PAK7 or CK2 siRNA. After transfection with the three and CK2a OMe-siRNAs, each combined with JetPEI. siRNAs, cell viability was decreased by 30–40% in Two days after transfection, mRNA expression was MiaPaCa2 cells but remained unchanged in all other cell assayed by semiquantitative RT-PCR (Figure 4a) and lines, except for MCF7, in which a 20% decrease was quantitative RT-PCR (Figure 4b). The expressions of the observed. three kinases were indeed strongly decreased in cells

Cancer Gene Therapy Combining kinase inhibitors in pancreatic cancer V Giroux et al 735 6 * 48 h after transfection with PAK7, MAP3K7 and CK2a * specific OMe-siRNAs mixed with JetPEI and was 5 * * * * * found to be induced 2-, 2.2- and 1.8-fold, respectively * * (Figure 4c), about half the values observed with naked 4 * * siRNAs. Caspase-3 activity was increased about 2-fold 3 * * * (Figure 4d), again half the value recorded with the naked 2 siRNAs. However, similar to naked siRNAs, cell viability was found decreased by about 50%, as compared with 1 Induction of apoptosis control (Figure 4e). It was concluded that OMe-modifica- tions of siRNAs decreased their efficacy on kinase 0 expressions while keeping their effect on cell survival, CK2 PAK7 and could be used in vivo in our xenograft model. Control MAP3K7 + CK2 PAK7 + CK2 Inhibition of survival kinases in vivo with OMe-siRNAs/ MAP3K7 + CK2 PAK7 + MAP3K7 PAK7 + MAP3K7 JetPEI complexes reduces the size of tumors in nude 40 mice with pancreatic xenografts Nude mice were injected with 107 MiaPaCa2 cells in the 35 * pancreas in order to develop xenografts. Then, groups of * 30 * * 10 mice were randomly treated with combinations of one, two or three OMe-siRNAs specific for PAK7, MAP3K7, 25 * * * CK2a survival kinases, in a complex with JetPEI, 20 according to the schedule described in Materials and fraction(%) 1 15 methods. Their weight was monitored during the study and no difference was observed between groups (not 10 Sub-G shown), suggesting limited toxicity of siRNA administra- 5 tion in vivo, if any. Twenty days after cell injection, mice 0 were killed, tumors were removed and their volume was measured. The mean volume for control mice was CK2 3 PAK7 0.90±0.12 cm (Figure 5). Volumes after administration Control MAP3K7 + CK2 of single siRNAs were 0.51±0.14, 0.52±0.21 and PAK7 + CK2 ± 3 MAP3K7 + CK2 0.66 0.20 cm for PAK7, MAP3K7 and CK2a, respec- PAK7 + MAP3K7 PAK7 + MAP3K7 tively. Volumes decreased to 0.23±0.11, 0.22±0.19 and 3 1.2 0.51±0.14 cm after injections of PAK7/MAP3K7, PAK7/CK2a and MAP3K7/CK2a siRNAs, respectively. 1.0 Finally, treatment with the triple combination PAK7/ * 0.8 * MAP3K7/CK2a reduced the tumoral volume to * * 0.22±0.11 cm3. Survival kinase mRNA expression was * * * 0.6 * * checked by semiquantitative RT-PCR (Figure 6a) and * * 0.4 * * * quantitative RT-PCR in tumors removed from mice Cell viability (Figure 6b). We observed a strong decrease of mRNA 0.2 expression in tumors removed from mice injected 0 with their specific OMe-siRNAs (diminution of 45, 77 and 45% for PAK7, MAP3K7 and CK2a, respectively). CK2 PAK7 We also checked caspase-3 activity in the tumors by Control MAP3K7 + CK2 immunohistochemistry on tumor sections. Caspase-3 PAK7 + CK2 activity was strongly induced in tumors removed from PAK7+ MAP3K7MAP3K7 + CK2 PAK7 + MAP3K7 mice treated with single siRNAs (2.0-fold for MAP3K7), and dual combinations (2.3-fold for MAP3K7/CK2a), Figure 2 Effects of combined inhibition of PAK7, MAP3K7 and CK2 and was maximal in mice treated with all three siRNAs on apoptosis assessed by cytoplasmic nucleosome concentration (a) (3.2-fold) (Figure 6c). It is noteworthy that whatever the and by evaluating the sub-G1 ratio in flow cytometry (b), and on cell viability using the MTS test (c) in MiaPaCa2 (open columns) and number of kinases simultaneously inhibited (1, 2 or 3) the BxPC3 (hatched columns) pancreatic cancer cells. Inhibitions are total amount of OMe-siRNA remained the same (30 mg obtained by transfection with specific siRNAs. As control, cells are per injection). This means that mice received 10 mg of each transfected with scrambled siRNA. *Po0.05 vs control siRNA. OMe-siRNA when the three kinases were to be inhibited. transfected with their specific OMe-siRNAs (65–85% decrease of mRNA expression). We also tested whether Discussion the O-Me-siRNAs of PAK7, MAP3K7 and CK2a could induce apoptosis in MiaPaCa2 cells to the same extent Our main aim is to develop new therapeutic strategies for as their naked counterparts. Apoptosis was monitored pancreatic cancer because the only drugs presently

Cancer Gene Therapy Combining kinase inhibitors in pancreatic cancer V Giroux et al 736 4.0 * Control siRNA ** 3.5 PAK7 siRNA MAP3K7 siRNA 3.0 CK2 siRNA 2.5

2.0

1.5 Caspase-3 activity 1.0

0.5

0

A375 H358 MCF7 U2OS H1299 SaOS-2 SW872 LS174T HRT18 MiaPaCa2 MDA-MB 231

Control siRNA 1.8 PAK7 siRNA 1.6 MAP3K7 siRNA 1.4 CK2 siRNA

1.2

1.0

0.8 * Cell viability ** 0.6

0.4

0.2

0

A375 H358 MCF7 U2OS H1299 SaOS-2 SW872 LS174T HRT18 MiaPaCa2 MDA-MB 231 Figure 3 Influence of the inhibition of PAK7, MAP3K7 or CK2 on caspase-3 activity (a) and cell viability (b) in MiaPaCa2 cells and 10 non- pancreatic human tumor cell lines. Inhibitions are obtained by transfection with specific siRNAs. As control, cells are transfected with scrambled siRNA. Inhibitions are obtained by transfection with specific siRNAs. As control, cells are transfected with scrambled siRNA. *Po0.05 vs control.

available improve patients’ survival by a few weeks, at the inhibition induced significant apoptosis were identified. best. Pancreatic adenocarcinoma differs from most other Interestingly, simultaneous inhibition of two or more cancers in its absence of response to conventional anti- kinases potentiated apoptosis and similar experiments on cancer strategies5 because of its particular resistance to several pancreatic cell lines led to similar results. Such a the induction of apoptosis. Therefore, rather than trying strategy has the advantage of being systematic, without to fine-tune such therapeutic approaches, we chose to a priori opinion on the involvement of a given kinase. It explore possible strategies based on new concepts. Our has the drawback of showing unexpected kinases, some of hypothesis was that inhibition of specific kinases might them poorly described or only known by their gene, and impair pancreatic cancer progression. This idea is actually for which, obviously, no chemical inhibitor is available. not new as several kinase inhibitors are available to treat The demonstration that inhibition of kinases belonging some cancers11,12 but the strategy developed to select the to a restricted panel of 52 molecules altered pancreatic appropriate kinase(s) is, to our knowledge, original in cancer development in vitro was encouraging in view of a the context of pancreatic cancer: All kinases from the new therapeutic strategy. The next step towards a proof of human kinome were individually inhibited with siRNAs concept was to look whether increased apoptosis observed in pancreatic cancer cell cultures and those whose in vitro would translate in vivo into altered tumor

Cancer Gene Therapy Combining kinase inhibitors in pancreatic cancer V Giroux et al 737 3.0 Control siRNA PAK7 MAP3K7CK2 2.5 * * * 2.0

1.5

123 123 1.0 0.5

Induction of apoptosis 1.2 GAPDH 0 1.0

CK2 PAK7 0.8 Control MAP3K7 * * 0.6 * 1.2 2.5 0.4 Cell viability 1.0 * * * 2.0 0.2 0.8 1.5 0 0.6 CK2 1.0 PAK7 * Control 0.4 MAP3K7

* Caspase-3 activity 0.5 mRNA expression 0.2 * 0 0 CK2 PAK7 CK2 Control PAK7 MAP3K7 Control MAP3K7 Figure 4 Inhibition of PAK7, MAP3K7 and CK2 with OMe-siRNA/JetPEI complexes in MiaPaCa2 cells. Inhibition was controlled by assessing mRNA expression by (a) semi-quantitative RT-PCR (lane 1: PAK7; lane 2: MAP3K7; lane 3: CK2) and (b) quantitative RT-PCR. The consequences for MiaPaCa2 cells of kinase inhibitions were assessed by monitoring (c) apoptosis, (d) caspase-3 activity and (e) cell viability. *Po0.05 vs control siRNA.

development. The best mouse model of pancreatic cancer highest apoptosis (more than threefold increase), were presently available is the KrasG12D model.13 Unfortu- selected for further studies (Figure 1). nately, this model is not adapted to our study because As control, we compared the influences of their kinases to be targeted were selected in human cells and single, dual or triple inhibition on apoptosis and cell there is no guarantee that the mouse counterparts of these viability in two pancreatic cancer cell lines, MiaPaCa2 kinases play the same role in mouse pancreatic cancer. We and BxPC3 (Figure 2). Similar results were obtained in had no choice but to use a xenotransplant model already the two cell lines. Simultaneous inhibition of two kinases validated in our laboratory8 in spite of its limitations, the increased significantly the apoptosis, but concomitant most important of them being that it does not allow inhibition of the third one only provided a small assessment of the effects of the injected siRNAs on increment, as clearly evidenced by flow cytometry data normal pancreas and on other tissues because they are (Figure 2). Data on cell survival (Figure 2) led to similar directed at human kinases. When injected in the pancreas conclusions. of nude mice, MiaPaCa2 cells grow and form large We also wanted to assess the pancreatic specificity of tumors within 3 weeks. The question was whether the induction of apoptosis by the inhibition of the three inhibition of appropriate kinases, chosen in the panel kinases. As shown in Figure 3, none of the 10 non- established in vitro, would impair tumor growth in mice. pancreatic cell lines that we tested showed any increase in However, the important observation that simultaneous spontaneous apoptosis after inhibition of the selected inhibition of two kinases resulted in increased efficacy in kinases. Although these findings do not allow prejudicing vitro, and its potential therapeutic interest, led us to the situation in vivo, they suggest some tissue specificity of consider monitoring a similar setting in vivo in parallel the observed phenomenon. with single-kinase inhibition. To choose the best candi- In pancreatic cells, the three selected kinases are clearly dates, we started from the 10 kinases whose inhibition had involved in maintaining apoptosis. In fact, earlier reports resulted in vitro in the highest induction of apoptosis7 and show that they are implicated in cell survival. PAK7 can assessed on two pancreatic cancer cell lines which couples inhibit the induction of apoptosis by localizing to the of kinases would be most efficient. PAK7, MAP3K7 and mitochondria and by phosphorylating the BAD protein, CK2a, the three kinases whose dual inhibition induced thus blocking BAD translocation to the mitochondria.14

Cancer Gene Therapy Combining kinase inhibitors in pancreatic cancer V Giroux et al 738

1.0

0.8 )

3 * * * 0.6

* 0.4 * * Tumor volume (cm 0.2

0 CK2 PAK7 Control MAP3K7 PAK7 + CK2 MAP3K7 + CK2 PAK7 + MAP3K7 PAK7 + MAP3K7 CK2 Figure 5 Influence of PAK7, MAP3K7 and CK2 inhibitions on MiaPaCa2 pancreatic xenografts. MiaPaCa2 cells injected in the pancreas of nude mice developed xenografts. siRNAs specific for the three kinases were injected i.p., alone or in combinations, as OMe-siRNAs þ JetPEI complexes. Tumor volumes, measured after 21 days, are given (mean±s.e. for 10 mice) with representative examples of the tumors. *Po0.05 vs control siRNA.

PAK7 can phosphorylate BAD directly on serine-112 and (apoptotic repressor with caspase recruitment domain) indirectly on serine-136 through AKT activation. The that blocks the activity of caspase 8.26 Inhibitors of CK2 blockade of PAK7 relocalization to the mitochondria have been reported to trigger apoptosis and to increase inhibits its antiapoptotic properties.15 TGF-b is a the susceptibility of cancer cells to chemotherapeutic pluripotent regulatory polypeptide, which plays a role in agents or apoptotic stimuli.27–29 In fact, the survival differentiation, proliferation and tissue morphogenesis.16 function of CK2 in pancreatic cancer cells was already Protein kinase MAP3K7, also known as TAK1 (TGF-b- described.30 activated kinase 1), is one of the downstream mediators of Of the three selected kinases, CK2a is the only one for the receptors for TGF-b.17 which a chemical inhibitor is commercially available. MAP3K7 with its activator TAB1 (TAK1-binding However, because of similarities between kinase active protein 1) can activate NF-kB by an NIK (NF-kB- sites and lack of information on a vast majority of inducing kinase)-independent mechanism,18 but with a members of the human kinome, there is no guarantee that functional interaction with IKK (IkB kinase) a and such an inhibitor will be specific and that the observed IKKb.19 This activation of NF-kB has been implicated in result will actually be attributable to CK2a inhibition liver tumor formation.20 However, several members of only. By contrast, the use of appropriate antisense the IAP (inhibitors of apoptosis) family, such as XIAP, oligonucleotides or siRNAs allows specific inhibition. NAIP and ML-IAP, activate JNK1 survival pathway This is why we decided to use in vivo the same siRNA through interaction with MAP3K7 and its activator sequences as in vitro. However, to prevent rapid degrada- TAB1.21 Whether MAP3K7 can be pro-apoptotic or tion by nucleases, five bases were O-methylated at both promote survival in the same cell, depending on condi- ends, as recommended by others.9 We checked in vitro tions, or whether the two opposite functions are tissue that O-Me modifications had not altered their efficacy specific is unknown. (Figure 4) before proceeding to in vivo experiments. We CK2 is a ubiquitous and constitutively active serine/ monitored the efficacy of OMe protection in vivo with an threonine protein kinase that is implicated, among many OMe-siRNA to a non-relevant kinase (Mapk1). Inhibi- other functions, in apoptosis,22 cell survival23 and cell tion was maximum 3 days after i.p. injection, then transformation.24 The antiapoptotic action of CK2 occurs decreased, while remaining significant after 6 days (not through the phosphorylation of BID, a pro-apoptotic shown). To ensure maximal inhibition of the three member of the BCL-2 family, which becomes resistant to relevant kinases, OMe-siRNAs were injected every third caspase cleavage25 and the phosphorylation of ARC day during the 21 days of the experiment.

Cancer Gene Therapy Combining kinase inhibitors in pancreatic cancer V Giroux et al 739 1.2

1.0

Cont siRNA PAK7 siRNA 0.8 * 0.6 * PAK7 0.4 *

mRNA expression 0.2

0 GAPDH

Cont siRNA MAP3K7 siRNA CK2 siRNA PAK7 siRNA Control siRNA MAP3K7 siRNA

MAP3K7 4.0 * 3.5 * 3.0 * GAPDH 2.5 * * * 2.0 Cont siRNA CK2 siRNA 1.5

Caspase activity 1.0 CK2 0.5 0 CK2

GAPDH PAK7 Control MAP3K7 PAK7 + CK2 MAP3K7 + CK2 PAK7 + MAP3K7 PAK7 + MAP3K7 CK2 Figure 6 Inhibition of PAK7, MAP3K7 and CK2 expression in tumors removed from mice treated with the corresponding OMe-siRNAs. Expression of the three mRNAs was measured by (a) semi-quantitative RT-PCR and (b) quantitative RT-PCR. (c) Influence of kinase inhibition on caspase-3 activity was assessed in the tumors to observe whether the decreased tumor volumes reported on Figure 5 could be associated with increased apoptosis. *Po0.05 vs control siRNA.

When inhibited alone, the three selected kinases did not further reduction in cancer progression with strategies induce significant reduction of tumor volume. However, leading to more thorough kinase inhibition. simultaneous inhibition of MAP3K7 and CK2 reduced In conclusion, we have shown in a mouse xenograft model tumoral volume by 50%, maximum efficiency being that inhibition of couples of specific kinases could consider- obtained with PAK7 associated with MAP3K7 or CK2, ably slow down pancreatic adenocarcinoma progression. which reduced tumoral volume by 75% (Figure 5). The efficacy of modified siRNAs for in vivo inhibition is Simultaneous inhibition of the three kinases did not confirmed and could possibly be improved by refining the improve that reduction, although caspase-3 activity seemed injection schedule and siRNA amounts. These results to be higher in that group than with other combinations. suggest that inhibitors of two of the three selected kinases, Tumor reduction is probably because of increased PAK7 and MAP3K7, might soon enter into clinical trial. sensitivity to apoptosis. However, because siRNAs were administered only one day after cancer cells were injected into the pancreas, one cannot exclude the fact that tumor Acknowledgements volume reduction was because of the impedance of proper engraftment by kinase inhibition. It is interesting to note This work was supported by grants from the INSERM, (Figure 6) that inhibition of kinase expression in the tumor, the Canceropole PACA and the Association pour la as judged by RT-PCR, was far from complete but Recherche sur le Cancer (# 3932). JLI and JCD are remained in the 50% range, which might leave room for supported by a ‘Contrat Interface’ INSERM–AP–HM.

Cancer Gene Therapy Combining kinase inhibitors in pancreatic cancer V Giroux et al 740 References 16 Roberts AB, Sporn MB. Physiological actions and clinical applications of transforming growth factor-beta (TGF-beta). 1 Jemal A, Tiwari RC, Murray T, Ghafoor A, Samuels A, Growth factors 1993; 8: 1–9. Ward E et al. Cancer statistics, 2004 CA. Cancer J Clin 2004; 17 Yamaguchi K, Shirakabe K, Shibuya H, Irie K, Oishi I, 54: 8–29. Ueno N et al. Identification of a member of the MAPKKK 2 Burris III HA, Moore MJ, Andersen J, Green MR, family as a potential mediator of TGF-beta signal transduc- Rothenberg ML, Modiano MR et al. Improvements in tion. Science 1995; 270: 2008–2011. survival and clinical benefit with gemcitabine as first-line 18 Sakurai H, Shigemori N, Hasegawa K, Sugita T. TGF-beta- therapy for patients with advanced pancreas cancer: a activated kinase 1 stimulates NF-kappa B activation by randomized trial. J Clin Oncol 1997; 15: 2403–2413. an NF-kappa B-inducing kinase-independent mechanism. 3 Richards DA. Chemotherapeutic gemcitabine doublets in Biochem Biophys Res Commun 1998; 243: 545–549. pancreatic carcinoma. Semin Oncol 2005; 32: S9–S13. 19 Sakurai H, Miyoshi H, Toriumi W, Sugita T. Functional 4 Demols A, Peeters M, Polus M, Marechal R, Gay F, interactions of transforming growth factor beta-activated Monsaert E et al. Gemcitabine and oxaliplatin (GEMOX) in kinase 1 with IkappaB kinases to stimulate NF-kappaB gemcitabine refractory advanced pancreatic adenocarcino- activation. J Biol Chem 1999; 274: 10641–10648. ma: a phase II study. Br J Cancer 2006; 94: 481–485. 20 Arsura M, Panta GR, Bilyeu JD, Cavin LG, Sovak MA, 5 Friberg G, Kindler HL. Chemotherapy for advanced Oliver AA et al. Transient activation of NF-kappaB through pancreatic cancer: past, present, and future. Curr Oncol a TAK1/IKK kinase pathway by TGF-beta1 inhibits AP-1/ Rep 2005; 7: 186–195. SMAD signaling and apoptosis: implications in liver tumor 6 Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam formation. Oncogene 2003; 22: 412–425. S. The protein kinase complement of the human genome. 21 Sanna MG, da Silva Correia J, Ducrey O, Lee J, Nomoto K, Science 2002; 298: 1912–1934. Schrantz N et al. IAP suppression of apoptosis involves 7 Giroux V, Iovanna J, Dagorn JC. Probing the human distinct mechanisms: the TAK1/JNK1 signaling cascade and kinome for kinases involved in pancreatic cancer cell survival caspase inhibition. Mol Cell Biol 2002; 22: 1754–1766. and gemcitabine resistance. Faseb J 2006; 20: 1982–1991. 22 Guo C, Yu S, Davis AT, Wang H, Green JE, Ahmed K. A 8 Carracedo A, Gironella M, Lorente M, Garcia S, Guzma´n potential role of nuclear matrix-associated protein kinase M, Velasco G et al. Cannabinoids induce apoptosis of CK2 in protection against drug-induced apoptosis in cancer pancreatic tumor cells via endoplasmic reticulum stress- cells. J Biol Chem 2001; 276: 5992–5999. related genes. Cancer Res 2006; 66: 6748–6755. 23 Ahmed K, Gerber DA, Cochet C. Joining the cell survival 9 Judge AD, Bola G, Lee AC, MacLachlan I. Design of squad: an emerging role for protein kinase CK2. Trends Cell noninflammatory synthetic siRNA mediating potent gene Biol 2005; 12: 226–230. silencing in vivo. Mol Ther 2006; 13: 494–505. 24 Tawfic S, Yu S, Wang H, Faust R, Davis A, Ahmed K. 10 Grzelinski M, Urban-Klein B, Martens T, Lamszus K, Protein kinase CK2 signal in neoplasia. Histol Histopathol Bakowsky U, Ho¨bel S et al. RNA interference-mediated gene 2001; 16: 573–582. silencing of pleiotrophin through polyethylenimine-com- 25 Desagher S, Osen-Sand A, Montessuit S, Magnenat E, plexed small interfering RNAs in vivo exerts antitumoral Vilbois F, Hochmann A et al. Phosphorylation of bid by effects in glioblastoma xenografts. Hum Gene Ther 2006; 17: casein kinases I and II regulates its cleavage by caspase 8. 751–766. Mol Cell 2001; 8: 601–611. 11 Soverini S, Martinelli G, Iacobucci I, Baccarani M. Imatinib 26 Li PF, Li J, Muller EC, Otto A, Dietz R, von Harsdorf R. mesylate for the treatment of chronic myeloid leukemia. Phosphorylation by protein kinase CK2: a signaling switch Expert Rev Anticancer Ther 2008; 8: 853–864. for the caspase-inhibiting protein ARC. Mol Cell 2002; 10: 12 Cabebe E, Wakelee H. Role of anti-angiogenesis agents in 247–258. treating NSCLC: focus on bevacizumab and VEGFR 27 Faust RA, Tawfic S, Davis AT, Bubash LA, Ahmed K. tyrosine kinase inhibitors. Curr Treat Options Oncol 2007; Antisense oligonucleotides against protein kinase CK2-alpha 8: 15–27. inhibit growth of squamous cell carcinoma of the head and 13 Tuveson DA, Zhu L, Gopinathan A, Willis NA, Kachatrian neck in vitro. Head Neck 2000; 22: 341–346. L, Grochow R et al. Mist1-KrasG12D knock-in mice 28 Ravi R, Bedi A. Sensitization of tumor cells to Apo2 ligand/ develop mixed differentiation metastatic exocrine pancreatic TRAIL-induced apoptosis by inhibition of casein kinase II. carcinoma and hepatocellular carcinoma. Cancer Res 2006; Cancer Res 2002; 62: 4180–4185. 66: 242–247. 29 Ruzzene M, Penzo D, Pinna LA. Protein kinase CK2 14 Cotteret S, Jaffer ZM, Beeser A, Chernoff J. p21-Activated inhibitor 4,5,6,7-tetrabromobenzotriazole (TBB) induces kinase 5 (Pak5) localizes to mitochondria and inhibits apoptosis and caspase-dependent degradation of haemato- apoptosis by phosphorylating BAD. Mol Cell Biol 2003; poietic lineage cell-specific protein 1 (HS1) in Jurkat cells. 23: 5526–5539. Biochem J 2002; 364: 41–47. 15 Cotteret S, Chernoff J. Nucleocytoplasmic shuttling of Pak5 30 Hamacher R, Saur D, Fritsch R, Reichert M, Schmid RM, regulates its antiapoptotic properties. Mol Cell Biol 2006; 26: Schneider G. Casein kinase II inhibition induces apoptosis in 3215–3230. pancreatic cancer cells. Oncol Rep 2007; 18: 695–701.

Cancer Gene Therapy