Simvastatin and downstream inhibitors circumvent constitutive and stromal cell-induced resistance to doxorubicin in IGHV unmutated CLL cells

The immunoglobulin heavy-chain variable region (IGHV) mutational status is a strong determinant of remission duration in chronic lymphocytic leukemia (CLL). The aim of this work was to compare the multidrug resistance (MDR) signature of IGHV mutated and unmutated CLL cells, identifying biochemical and molecular targets potentially amenable to therapeutic intervention. We found that the mevalonate pathway-dependent Ras/ERK1–2 and RhoA/RhoA kinase signaling cascades, and the downstream HIF-1α/P-glycoprotein axis were more active in IGHV unmutated than in mutated cells, leading to a constitutive protection from doxorubicin-induced cytotoxicity. The constitutive MDR phenotype of IGHV unmutated cells was partially dependent on B cell receptor signaling, as shown by the inhibitory effect exerted by ibrutinib. Stromal cells further protected IGHV unmutated cells from doxorubicin by upregulating Ras/ERK1–2, RhoA/RhoA kinase, Akt, HIF-1α and P-glycoprotein activities. Mevalonate pathway inhibition with simvastatin abrogated these signaling pathways and reversed the resistance of IGHV unmutated cells to doxorubicin, also counteracting the protective effect exerted by stromal cells. Similar results were obtained via the targeted inhibition of the downstream molecules ERK1–2, RhoA kinase and HIF-1α. Therefore, targeting the mevalonate pathway and its downstream signaling cascades is a promising strategy to circumvent the MDR signature of IGHV unmutated CLL cells.


INTRODUCTION
Chronic lymphocytic leukemia (CLL) is chara cterized by a highly heterogeneous clinical course and a poor curability with conventional chemotherapy treatment approaches. The immunoglobulin heavychain variable region (IGHV) mutational status has emerged as a powerful prognosticator: patients with IGHV unmutated (UM) CLL cells display inferior survival rates [1,2] and shorter complete remission duration than IGHV mutated (M) patients [3].
Stromal cells (SCs) are known to protect CLL cells from spontaneous apoptosis and druginduced cytotoxicity. We previously reported that IGHV UM cells are more dependent than M cells on microenvironmentmediated

RESULTS
The Ras/ERK1-2 and RhoA/RhoA kinase signaling pathways and the HIF-1α/Pgp axis are more active in IGHV UM than M CLL cells The activity of Ras and RhoAdependent signaling pathways was analyzed in IGHV M and UM CLL cells (>90% pure as described below) after ex vivo culture for 24 hours. Both type of cells exhibited detectable amounts of nonisoprenylated cytosolic Ras and unphosphorylated ERK1-2, but only IGHV UM cells showed high intracellular levels of the Ras GTPbound active form and the Rasdownstream effector kinase phospho-ERK1-2 ( Figure 1A, left), in keeping with their accelerated Mev pathway activity [15]. Similarly, the amount of active GTPbound RhoA and the activity of the downstream RhoA kinase were significantly higher in IGHV UM than M cells (p always = 0.001) ( Figure 1A, right).
Both ERK1-2 and RhoA kinase phosphorylate and activate the transcription factor HIF-1α [16,17]. Accordingly, the activation of Ras/ERK1-2 and RhoA/ RhoA kinase signalling pathways in IGHV UM cells led to the phosphorylation of HIF-1α (Supplementary Figure  S1) and to the increase of its transcriptional activity, as shown by the significantly higher amounts of nuclear HIF-1α bound to its specific DNA target sequence (p = 0.002) ( Figure 1B, left). As a consequence, IGHV UM cells showed higher MDR1 mRNA expression ( Figure 1B, right) and lower Doxo accumulation than IGHV M cells (p always = 0.001) ( Figure 1C).
In agreement with our previous data [4], both untreated IGHV M and UM cells showed high levels of viability after ex vivo culture for 48 hours, whereas Doxo treatment induced a significant decrease of viability in IGHV M compared to UM CLL cells (p = 0.001) ( Figure  1D and Supplementary Figure S2).

The constitutive MDR phenotype of IGHV UM cells depends on BCR signaling and is further upregulated by SC-mediated extrinsic signals
IGHV UM cells display an increased signaling through the B cell receptor (BCR) compared to IGHV M cells [18]. To study the effect of BCR signaling on Mev regulated pathways and MDR phenotype we incubated IGHV M and UM cells with antiIgM antibody, in the presence or absence of the Bruton tyrosine kinase (BTK) inhibitor ibrutinib. AntiIgMmediated BCR stimulation did not increase the baseline production of cholesterol and FPP (Figure 2A), the amount of GTPbound RhoA and the activity of RhoA kinase ( Figure 2B), the transcriptional activity of HIF-1α and the expression of its target gene MDR1 ( Figure 2C) in both IGHV M and UM cells. Of note, ibrutinibmediated BCR inhibition significantly decreased the constitutively higher levels of cholesterol and FPP production, RhoA and RhoA kinase activity, HIF-1α transcriptional activity and MDR1 expression in IGHV UM cells (p always ≤ 0.05), whereas it did not affect the Mev pathway and Mevregulated signaling in IGHV M cells.
SCs are known to confer a chemoresistant phenotype to CLL cells [19][20][21]. Therefore, we also sought whether IGHV M and UM cells cocultured with the murine stromal cell line M2-10B4 upregulated the Mev pathway and the RhoA/RhoA kinase signaling pathways. IGHV M and UM cells showed high and comparable levels of cell viability after 24hour ex vivo culture, both in the presence and in the absence of SCs (data not shown). After exposure to M2-10B4 SCs, the production of cholesterol and FPP ( Figure 3A), the amount of GTPbound RhoA and the activity of RhoA kinase ( Figure 3B), the transcriptional activity of HIF-1α, and the expression of its target gene MDR1 ( Figure 3C) were significantly increased in IGHV UM but not in M CLL cells (p always ≤ 0.03).

SIM effectively reverses the MDR phenotype of IGHV UM cells and restores Doxo-induced cytotoxicity
We next examined whether SIM, which switches off the Mev pathway downstream to the ratecontrolling enzyme 3hydroxy3methylglutarylCoA reductase, was effective in reversing the MDR phenotype of IGHV UM cells.
This SIMmediated inhibition of the Mev pathway and downstream signaling pathways was paralleled by a lower HIF-1α phosphorylation (Supplementary Figure S3) and a statistically significant decrease in HIF-1α activity and MDR1 expression in IGHV UM cells, both in the presence or absence of SCs (p always < 0.003) ( Figure 5A, 5B). Thus, SIM significantly increased the accumulation of intracellular Doxo, even in the presence of SCs (p always ≤ 0.002) ( Figure  5C). The SIMinduced increase in Doxo accumulation determined an enhanced cytotoxic death of IGHV UM cells after exposure to the combination SIM+Doxo for 48 hours (p ≤ 0.0001). Interestingly, SIM was also effective in reversing the protective effect exerted by M2-10B4 SCs (p ≤ 0.0001) ( Figure 5D and Supplementary Figure S4).

SIM effectively abrogates SC-induced Akt upregulation in IGHV UM cells
We next examined the effect of another Mev pathway inhibitor, ZA, which switches off the pathway downstream to FPP-synthase. ZA significantly abrogated cholesterol and  To further elucidate the difference between ZA and SIM, we also evaluated the effects on Akt and NF-kB, which are wellknown prosurvival factors for CLL cells [22]. The baseline activity of Akt, but not that of NF-kB, was significantly higher in IGHV UM than M cells (p ≤0.001) ( Figure 6A, 6B). The same results were confirmed by evaluating the expression of the active phosphorylated form of Akt and the nuclear translocation of the NF-kB components p50 and p65 by western blot analyses (Supplementary Figure S8A, S8B).
Akt activity was further upregulated by the presence of M2-10B4 SCs ( Figure 5C and Supplementary Figure  S9A), whereas NF-kB activity was not, as previously reported [4]. ZA significantly increased Akt activity in IGHV UM cells (p ≤ 0.009), and this effect was particularly evident when ZA and SCs were used in combination (p ≤ 0.03) (Supplementary Figure S9A). Of note, ZA also increased NF-kB activity in IGHV UM cells, both in the absence and in the presence of SCs (p ≤ 0.05 and p ≤ 0.04, respectively) (Supplementary Figure S9B). Unlike ZA, SIM reduced the baseline Akt and NF-kB activities (p ≤ 0.02), and significantly abrogated the SC-induced up-regulation of Akt and NF-kB activity (p ≤ 0.03; Figure 6C, 6D), Akt phosphorylation (Supplementary Figure S8C) and NF-kB nuclear translocation (Supplementary Figure S8D).

Specific inhibitors of ERK1-2, RhoA kinase and HIF-1α reverse the MDR phenotype of IGHV UM cells
We also tested the activity of specific inhibitors targeting ERK1-2 kinases (PD98059, PD), RhoA kinase (Y27632, Y276) and HIF-1α (YC-1) in IGHV UM cells cultured alone or in the presence of SCs. Of note, the three inhibitors, as well as the upstream Mev pathway inhibitor SIM, had no impact on SCs viability after 48 hours of culture and did not increase the cytotoxic activity of Doxo toward SCs (Supplementary Figure S10).

DISCUSSION
This study shows that IGHV UM cells are endowed with a MDR phenotype which grants them a greater resistance to Doxoinduced cytotoxicity compared with IGHV M cells. Mechanistic insights have shown that IGHV UM cells, according to their accelerated Mev pathway activity [15], had significantly higher activity of the Mev-regulated Ras/ERK1-2 and RhoA/RhoA kinase signaling cascades, increased HIF-1α phosphorylation and activity, higher MDR1 gene expression, very effective Doxo extrusion and enhanced survival compared with IGHV M cells. The MDR signature of IGHV UM cells is at least partially dependent on BCR signaling, as shown by the inhibitory effect exerted by ibrutinib on the Mev pathway and Mevregulated signaling.
The constitutive MDR features of IGHV UM cells are further exacerbated by the incubation with SCs, which stimulates the Mev pathway and its downstream signaling cascades. Mev pathway manipulation with SIM, and specific ERK1-2, RhoA kinase or HIF-1α inhibitors reset the chemosensitivity of IGHV UM cells to the same levels of IGHV M cells, and neutralize the MDRenhancing effect operated by SCs.
A role for MDR1 in CLL chemoresistance has previously been suggested by the association of Pgp expression with IGHV UM genes and poor prognosis cytogenetics [23]. Our data confirm that the MDR1 gene, which is a primary transcriptional target of HIF-1α [24,25], is constitutively more expressed in IGHV UM than in M cells. HIF1 is a heterodimeric transcription factor composed of an oxygen-induced HIF-1α and a constitutively expressed HIF-1β subunit. During hypoxia, HIF-1α is positively regulated by the activation of the RhoAdependent signaling cascade [16]. Under non hypoxic conditions, the expression and transcriptional activity of HIF-1α are regulated by growth factors and cytokines through the activation of kinase pathways, such as the Ras/ERK1-2 and the PI3k/Akt pathways [17]. Unlike their normal counterparts, CLL cells express HIF-1α even under normoxia [26], and its transcriptional activity regulates CLL cells survival by inducing the expression of growth factors, such as VEGF, and the cytokines macrophage migrationinhibition factor (MIF) and midkine [26][27][28]. This is the first report showing that HIF-1α activity is significantly higher in IGHV UM than in M cells, due to the higher activity of the upstream Ras/ERK1-2 and RhoA/RhoA kinase signaling cascades in these cells. The MDR signature of IGHV UM cells is partially dependent on their tonic BCR signaling, as shown by the inhibitory effect exerted by ibrutinib but the lack of stimulatory effect exerted by antiIgM.
Differently from antiIgMmediated BCR stimulation, SCs exposure further increase the activity of Ras/ERK1-2 and RhoA/RhoA kinase signaling pathways and HIF-1α in IGHV UM cells. As a result, the target gene MDR1 is over-expressed, Doxo is more efficiently extruded and IGHV UM cells are further protected from Doxoinduced cytotoxicity. SCs are known to confer chemoresistance to CLL cells. One of the main players of the SCinduced MDR is the CXCL12/CXCR4 axis, which is known to activate the signal transducers ERK1-2 and Akt, and trigger LFA1 affinity to the integrin ICAM-1, through a RhoA-mediated mechanism [29]. It is already known that the BCR signaling modulates the activity of the CXCL12/CXCR4 axis, in fact the blockade of the BCR signaling with a BTK inhibitor decreases the migration of CXCR4expressing CLL cells toward CXCL12 produced by SCs [30]. IGHV UM cells were exposed to 1 μM Doxo, 1 μM SIM, and the combination SIM+Doxo for 48 hours, both in the absence and in the presence of M2-10B4 SCs. Cell viability was determined by Ann-V staining and cytofluorimetric analysis on CD19+/CD5+ cells. No decrease in viability was observed when IGHV UM cells were exposed to SIM and Doxo used alone (91% ± 2% and 90% ± 1% of CD19+/CD5+ viable cells, respectively). By contrast, a significant reduction in cell viability was induced by the combination SIM+Doxo, both in the absence (19, 5% ± 7% CD19+/CD5+ viable cells, p < 0.0001) and in the presence (10% ± 3% CD19+/CD5+ viable cells, p < 0.0001) of SCs. In panels A-D results are from 8 side-by-side experiments. Box and whiskers plot represent median values, first and third quartiles, and minimum and maximum values for each dataset. www.impactjournals.com/oncotarget The Mev pathway and the CXCL12/CXCR4 axis may reciprocally affect their activity, as shown by at least two observations: 1) for optimal signaling CXCR4 must be incorporated into membrane lipid rafts, whose formation require membrane cholesterol, and which orchestrate the interaction of the small GTPases Rac and Rho with their downstream transducers [31]; 2) hypercholesterolemia induces the secretion of CXCL12 and drives the migration of CD19+/CXCR4+ B lymphocytes from the bone marrow to the peripheral blood by interfering with the CXCL12/ CXCR4 axis [32]. Further studies aiming at elucidating the connections between the Mev pathway and the CXCL12/ CXCR4 axis in CLL are currently ongoing in our laboratory.
Statins and aminobisphosphonates, by targeting key enzymes in the Mev pathway, cause the intracellular deprivation of isoprenoid moieties such as FPP and GGPP, thus preventing Ras and RhoA prenylation and the activation of their downstream signaling pathways [33]. We have recently reported that ZA effectively interrupts Ras and RhoAdependent downstream signalling pathways, abrogates Pgp expression, and restores Doxo induced cytotoxicity in MDR+ human cancer cell lines derived from solid tumors [14]. In CLL, ZA was not an ideal chemosensitizing agent since it inhibited Ras/ERK1-2 and RhoA/RhoA kinase signaling pathways, HIF-1α activity, and MDR1 expression, but did not increase Doxo induced cytotoxicity. By contrast, SIM was effective in reversing the resistance of IGHV UM cells to Doxo, also in the presence of the protective effect exerted by SCs. This different efficacy was due to a stronger ability of SIM to increase the intracellular amount of Doxo, especially in the presence of SCs, but also to a differential regulatory effect exerted by the two compounds on the prosurvival factors Akt and NF-kB. We have already reported that ZA induces the activation of NF-kB in tumorassociated macrophages, reverting their phenotype from tumorpermissive into tumoricidal cells [34]. A similar upregulation of the Akt/NF-kB pathway was observed in IGHV UM cells after ZA, but not after SIM exposure. This at least partially accounted for a ZAmediated prosurvival activity which most likely prevailed on the death signals triggered by Doxo.
Among the chemosensitizing agents that we have tested, statins are the best candidate for clinical translation, since they are commonly used in the clinical practice as cholesterollowering agents. Previous data have shown that SIM, although potently effective in blocking the proliferation and inducing apoptosis of CLL cells in vitro [35][36][37], is not capable of significantly reducing the tumor burden when administered as a single agent in previously untreated

Figure 6: SIM effectively counteracts SC-induced Akt upregulation in IGHV UM cells. IGHV UM cells had constitutive
higher levels of Akt activity (p < 0.001) compared to IGHV M cells A. By contrast, there was no difference in baseline NF-kB activity between IGHV M and UM cells B. The Akt activity was further upregulated by the co-culture of IGHV UM cells with M2-10B4 SCs (p < 0.0001). SIM reduced baseline levels of Akt activity (p = 0.0082), and significantly counteracted SC-induced Akt upregulation (p = 0.0004) C. NF-kB activity was not upregulated by SCs, but it was significantly reduced by SIM exposure, both in the absence (p = 0.02) or presence of SCs (p = 0.034) D. In panels A and B bars represent mean values ± SEM. In panels C and D results are from 8 side-by-side experiments, and box and whiskers plot represent median values, first and third quartiles, and minimum and maximum values for each dataset. www.impactjournals.com/oncotarget CLL patients [37]. By contrast, statins effectively increase the susceptibility of both drugsensitive and drugresistant CLL and lymphoma cells to dexamethasone and cytotoxic drugs [38], even synergizing with purine analogs to induce apoptosis of CLL cells [35]. Schmidmaier et al. have already reported that the Mev pathway and the downstream RhoA/ RhoA kinase signaling pathway mediate SCinduced MDR, and that targeting of this pathway by SIM may improve the efficacy of anti-myeloma therapies by reduction of SCinduced MDR [39]. The same authors showed in a phase II clinical trial that SIM could overcome MDR in refractory myeloma receiving chemotherapy with bortezomib or bendamustine [40]. In the CLL setting, Chae et al. recently reported data on a retrospective analysis showing that the use of statin and aspirin is associated with improved outcome in CLL patients receiving salvage fludarabine, cyclophosphamide and rituximab (FCR) chemotherapy [41], providing the rationale for a prospective study aimed at evaluating the effects of statins in CLL patients receiving chemoimmunotherapy. Since none of the drugs included in the FCR regimen are substrates of the Pgp, further studies aimed at determining the role and mechanism of action of statins in chemosensitizing CLL cells are warranted.
Doxo accumulation as a readout assay to test Pgp activity could be considered a limitation of our study, due to previous data showing that the addition of anthracyclines to alkylating agents did not improve the clinical outcome of CLL patients [8]. More recently, however, the addition of mitoxantrone or epirubicine to fludarabine-based regimens has been shown to improve response rates [42][43][44]. Skribek et al. have reported that anthracyclines can have direct cytotoxic effects on CLL cells in vitro irrespectively to age, clinical stage or cytogenetic markers like del(17p) [45]. These observations and our data are strong incentives B. MDR1 expression. Twenty four-hour co-culture with M2-10B4 SCs significantly increased the expression of MDR1 in IGHV UM cells (p < 0, 0001). After PD, Y276 and YC-1 treatment a significant decrease in MDR1 expression was observed in IGHV UM cells cultured alone (p always ≤ 0.001) and in the presence of SCs (p always ≤ 0.0001). C. Doxo intracellular accumulation. Intracellular Doxo was significantly lower in IGHV UM cells cultured with SCs than in IGHV UM cells cultured alone (p = 0.02). After 48hour exposure to PD, Y276 and YC-1, Doxo accumulation was significantly increased, both in the absence (p always ≤ 0.003) and in the presence of M2-10B4 SCs (p always ≤ 0.0012). D. Percentage of CD19+/CD5+ viable cells. IGHV UM cells were exposed to 1 μM Doxo, used alone and in combination with each inhibitor (i.e. PD+Doxo, Y276+Doxo, YC-1+Doxo), both in the absence and in the presence of the M2-10B4 SCs. Cell viability was determined by Ann-V staining and cytofluorimetric analysis on CD19+/CD5+ cells after 48-hours of culture. Doxo alone, as well as the three inhibitors used as single agents, did not induce a decrease in cell viability. By contrast, the combinations PD+Doxo, Y276+Doxo and YC-1+Doxo significantly reduced the viability of IGHV UM cells both in the absence and in the presence (p always < 0.0001) of SCs. In panels A-D results are from 7 side-by-side experiments. Box and whiskers plot represent median values, first and third quartiles, and minimum and maximum values for each dataset.
to reconsider the therapeutic potential of anthracyclines in CLL, especially when used in combination with chemo sensitizing agents targeting the Mev pathway such as statins, to treat specific subsets of high-risk CLL patients (e.g. IGHV UM patients). Moreover, the chemosensitizing effects that we observed by targeting ERK1-2 and RhoA kinase, as well as the transcription factor HIF-1α, in addition to providing a confirmatory evidence about the mechanistic role played by these pathways in the MDR+ phenotype of IGHV UM cells, opens new perspectives on innovative targets amenable to therapeutic interventions.
Interestingly, recent data have shown that Pgp may be involved in the decreased anti-cancer efficiency and modified pharmacological properties of novel targeted agents, such as imatinib [46] and secondgeneration tyrosine kinase inhibitors [47]. Therefore, the study of Pgp activity and its modulation might have a broad impact in the current era of targeted therapies, providing important information on the pharmacokinetic and the anticancer effects of these novel agents.

Patients
Peripheral blood (PB) samples were collected from 63 untreated CLL patients between March 2008 and November 2012 after informed consent. Patient demographics and clinical characteristics are reported in Supplementary Table S1. CLL was defined by clinical examination, PB cell count and immunophenotypic criteria. Tumor IGHV rearrangements were amplified starting from genomic DNA and sequences with deviations of < 2% or ≥ 2% from the germline IGHV sequence were considered IGHV UM or M as previously reported [48].

Cell purification and cell lines
Peripheral blood mononuclear cells (PBMC) from IGHV M and UM patients were isolated from fresh samples by density gradient centrifugation (Ficoll-Histopaque, PAA Laboratories GmbH, Linz, Austria). PBMC were used without further manipulation when they contained more than 90% of CLL cells (CD19+/CD5+). When CLL cells were ≤ 90% they were purified by negative selection using a Magnetic BeadActivated Cell Sorting (MACS ® ) with a B Cell Isolation Kit II (Miltenyi Biotec, Bologna, Italy).

Chemicals
Electrophoresis reagents were obtained from Bio-Rad Laboratories (Hercules, CA, USA). The protein content of cell lysates was assessed with the bicinchoninic acid kit from Sigma Chemical Co (St. Louis, MO, USA). ZA was a kind gift from Novartis (Basel, Switzerland). SIM and Y276 were purchased from Calbiochem (San Diego, CA, USA). When not otherwise specified, all the other reagents were purchased from Sigma Chemical Co.
For co-culture experiments, the M2-10B4 murine SCs were harvested through Trypsin-EDTA (Sigma-Aldrich, Milano, Italy) digestion and plated (150 × 10 3 cells/well) in complete culture medium for 24 hours. CLL cells (1 × 10 6 cells/well) were added to the culture the following day.

Mev pathway activity
After 24 hours of culture, 1 × 10 6 CLL cells were incubated for another 24 hours with 1 μCi of [ 3 H]acetate (3600 mCi/mmol; Amersham International, Piscataway, NJ, USA) to measure the intracellular synthesis of cholesterol and FPP. Lipids were extracted in methanol/ hexane, resolved by thin layer chromatography and quantified by liquid scintillation counting as reported in [15]. According to the titration curve, the results are expressed as fmol/1 × 10 6 cells.

Ras and RhoA activity
To evaluate Ras and RhoA activity, their GTPbound fraction, taken as an index of the Gprotein activation [49], was measured after 24 hours of culture. RasGTP was detected in a pulldown assay as previously reported [50]; RhoAGTP binding was measured with the GLISA RhoA Activation Assay Biochem Kit (Cytoskeleton Inc, Denver, CO, USA), according to the manufacturer's instructions. For each set of experiments a titration curve was prepared using serial dilution of the RhoAGTP positive control of the kit. Data are expressed as U absorbance/mg cell proteins (U mg/prot).

RhoA kinase activity
RhoA kinase activity was measured after 24 hours of culture on 2 × 10 6 cells with the CycLex Rho Kinase Assay www.impactjournals.com/oncotarget Kit (CycLex Co., Nagano, Japan), as already described [50]. For each set of experiments a titration curve was set using serial dilution of recombinant RhoA kinase (Rock2, MBL Inc., Woburn, MA, USA). Data are expressed as U absorbance/mg cell proteins (U mg/prot).

HIF-1α activity
Nuclear proteins from 2 × 10 6 cells were extracted after 24-hour culture using the Nuclear Extract Kit (Active Motif, Rixensart, Belgium), and quantified. The activity of HIF-1α was assessed on 10 μg nuclear extracts using the TransAM™ HIF-1α Transcription Factor Assay Kit (Active Motif), according to manufacturer's instructions.
To assess the procedure specificity, a competition assay was performed by adding an excess (20 pmol) of HIF-1α consensus oligonucleotide to nuclear extracts derived from CLL cells. Data are expressed as U absorbance/mg cell proteins (U mg/prot).

NF-kB and Akt activity
NF-kB activity was measured on 10 μg nuclear extracts using the TransAM Flexi NF-kB Family assay kit (Active Motif), according to the manufacturer's instructions. To assess the procedure specificity, a competition assay was performed by adding an excess (20 pmol) of the probe containing the NF-kB consensus sequence to the positive control nuclear extracts provided by the kit.
Akt activity was measured on 2 × 10 6 cells with the CycLex Akt/PKB Kinase Assay/Inhibitor screening kit (CycLex Co.) in 96well plates precoated with the Akt substrate AkTide2T, according to the manufacturer's instructions. The titration curve was prepared using serial dilutions of recombinant Akt (CycLex Co.).
NF-kB and Akt activity are expressed as U absorbance/mg cell proteins (U mg/prot).

Intracellular doxorubicin accumulation
To evaluate the activity of Pgp the intracellular accumulation of the Pgp substrate Doxo was measured. To this aim, 1 × 10 6 cells were incubated with 1 μM Doxo cultured alone or in presence of inhibitors for 48 hours; intracellular Doxo was measured spectrofluorimetrically, at excitation and emission wavelengths of 475 and 553 nm, as reported earlier [50]. A blank was prepared in the absence of cells in each set of experiments, and its fluorescence was subtracted from that measured in each sample. Fluorescence was converted in nmol Doxo/mg cell proteins (nmol/mg prot) using a previously prepared calibration curve.

Quantification of apoptotic and viable cells
After 48 hours of culture, cells were harvested, washed in PBS and stained with antiCD19 PerCP (Beckman Coulter, Milano, Italy) and antiCD5 APC (Dako SpA, Milano, Italy) monoclonal antibodies. The percentage of apoptotic cells was determined by Annexin-V (Ann-V) staining with the MEBCYTO-Apoptosis Kit (MBL Medical and Biological Laboratories, Naka-ku Nagoya, Japan). Flow cytometry was performed with a FACSCalibur and CELLQuest software (Becton Dickinson, Mountain View, CA, USA).

Statistical analysis
Statistical analysis was performed with the SigmaStat 3.5 software (Systat Software Inc., Chicago, IL, USA) and with GraphPad Prism version 5.01 for Windows (GraphPad Software, San Diego, CA, USA). All datasets were evaluated with a normality test: data with Gaussian or approximately Gaussian distribution were compared by ttest, in all other cases a MannWhitney rank sum test was used. When experiments involved different culture conditions of the same sample a repeatedmeasures analysis was done.
Results are expressed as mean ± SEM, unless otherwise specified.
Statistical significance was defined as p value < 0.05. Italy; and local funds of the University of Turin (ex 60%). Valentina Griggio was the recipient of a "Giorgio Bissolotti e Teresina Bosio" fellowship from Fondazione "Angela Bossolasco", Torino, Italy and currently is the recipient of a "Anna Nappa" fellowship from the Italian Association for Cancer Research (AIRC, Ref 16343); Micol Rigoni was the recipient of a fellowship from Fondazione Cassa di Risparmio di Torino (CRT) and currently is the recipient of a fellowship from Fondazione "Angela Bossolasco", Torino, Italy.

ACKNOWLEDGMENTS AND FUNDING
We are grateful to Mr. Andrew Martin Garvey for editorial assistance.