ORIGINAL ARTICLE Targeting Lipid Metabolism by The

Targeting lipid metabolism by the lipoprotein lipase inhibitor orlistat results in apoptosis of B-cell chronic lymphocytic leukemia cells

CP Pallasch1, J Schwamb1,SKo¨nigs1, A Schulz1, S Debey2, D Koﬂer1, JL Schultze2, M Hallek1, A Ultsch3 and C-M Wendtner1

1Laboratory of Cellular Immunotherapy, Clinic I of Internal Medicine, University of Cologne, Cologne, Germany; 2Molecular Tumor Biology and Tumor Immunology, Clinic I of Internal Medicine, University of Cologne, Cologne, Germany and 3Department of Mathematics and Computer Science, Databionics Research Group, Philipps-University Marburg, Marburg, Germany

Constitutively activated pathways contribute to apoptosis the analysis of BCR stimulation. Lipase activity was identified as resistance in chronic lymphocytic leukemia (CLL). Little is a prominent pathway in CLL cells and the recently identified known about the metabolism of lipids and function of lipases prognostic factor lipoprotein lipase (LPL) was seen to be in CLL cells. Performing gene expression profiling including 8,9 B-cell receptor (BCR) stimulation of CLL cells in comparison to specifically induced by stimulation of the BCR. LPL is a healthy donor CD5 þ B cells, we found significant overexpres- secreted enzyme, acts at the vascular side of endothelium and sion of lipases and phospholipases in CLL cells. In addition, we plays a central role for energy supply from serum triglycerides in observed that the recently defined prognostic factor lipoprotein peripheral tissues.10 Here, we found that the lipase inhibitor lipase (LPL) is induced by stimulation of BCR in CLL cells but orlistat, known as an effective inhibitor of LPL, acts as a not in CD5 þ normal B cells. CLL cellular lysates exhibited cytotoxic agent in CLL cells by its proapoptotic effects.11 significantly higher lipase activity compared to healthy donor controls. Incubation of primary CLL cells (n ¼ 26) with the lipase inhibitor orlistat resulted in induction of apoptosis, with a half- maximal dose (IC50) of 2.35 lM. In healthy B cells a significantly Materials and methods higher mean IC50 of 148.5 lM of orlistat was observed, while no apoptosis was induced in healthy peripheral blood mono- Patients and cells nuclear cells (PBMCs; Po0.001). Orlistat-mediated cytotoxicity was decreased by BCR stimulation. Finally, the cytotoxic After informed consent, blood was obtained from patients effects of orlistat on primary CLL cells were enhanced by the fulfilling diagnostic criteria for B-CLL. Only patients without simultaneous incubation with fludarabine (P ¼ 0.003). In sum- prior therapy or at least 6 months without prior chemotherapy mary, alterations of lipid metabolism are involved in CLL were included in this study. Fresh B-CLL samples were enriched pathogenesis and might represent a novel therapeutic target by applying B-RosetteSep (Stem Cell Technologies, Vancouver, in CLL. BC, Canada) and Ficoll-Hypaque (Seromed, Berlin, Germany) Leukemia (2008) 22, 585–592; doi:10.1038/sj.leu.2405058; published online 13 December 2007 density gradient purification resulting in purity of 498% of Keywords: lipoprotein lipase; chronic lymphocytic leukemia; CD19 þ /CD5 þ B-CLL cells. CLL cells were characterized orlistat; apoptosis for CD19, CD5, CD23, FMC7, CD38, ZAP70, sIgM, sIgG, CD79a and CD79b expression on a FACS Canto flow cytometer (BD PharMingen, Heidelberg, Germany). IgVH-hypermutational status of CLL patients was analyzed as previously published.12 Introduction Controls were isolated from healthy blood donors using anti-CD19- MACS-beads (Miltenyi, Bergisch-Gladbach, Germany) and B-cell chronic lymphocytic leukemia (B-CLL) is the most subsequent positive selection with CD5-PE (BD PharMingen) common leukemic disorder in the western hemisphere. It is and anti-PE-Beads (Miltenyi) resulting in at least 80% CD19 þ / characterized by an accumulation of mature B cells in the þ 1 CD5 B cells. This study was approved by the ethics committee blood, bone marrow and secondary lymphoid organs. The of the University of Cologne (Approval No. 01-163, 9 October accumulation of the malignant cells is caused not only by 2 2001), blood samples were given with informed consent dysregulation of apoptosis but also by a dynamic proliferation. according to the Helsinki protocol. A major factor in the pathogenesis of CLL is the microenvironment of the leukemic cell,3 especially the role of antigenic 4 stimulation via the B-cell receptor (BCR). Real-time PCR An unmutated status of the IgV gene was shown to exhibit H Quantitative PCR. Quantification of LPL mRNA was per- enhanced autoreactivity5 and to predict an unfavorable prog- formed by LightCycler Taqman Master (Roche, Penzberg, nosis in CLL.6 Furthermore, expression of ZAP-70 is associated Germany). Universal probe No. 13 (Roche) with forward primer with an enhanced signal transduction via the BCR complex.7 GCCCAGCTACAACCACCA and reverse primer TGTGAGCCA Based on the central relevance of BCR-signaling in CLL, we TAGTCCAGTGC was used for LPL amplification. b-2-Micro- applied gene expression profiling of primary CLL cells including globulin was applied as housekeeping gene standard by Universal probe No. 42 (Roche) with forward primer Correspondence: Professor C-M Wendtner, University of Cologne, TTCTGGCCTGGAGGCTATC and Reverse Primer TCAGGAAA Clinic I of Internal Medicine, Kerpener Strasse 62, D-50924 Cologne, TTTGACTTTCCATTC. All experiments were performed in Germany. E-mail: [email protected] replicates and crossing points were determined by second Received 16 August 2007; revised 6 November 2007; accepted 12 derivative maximum method. Relative quantification analysis November 2007; published online 13 December 2007 was performed by Exxor3 software package (Roche). Orlistat cytotoxicity in CLL CP Pallasch et al 586 Stimulation of cells each array. The parameters of the z-transformation were chosen Cells were cultivated in IMDM (Invitrogen, Carlsbad, CA, USA) in such a way that the distribution of transcripts that were not supplemented with 10% heat-inactivated fetal calf serum (FCS), expressed (unexpressed genes) is mapped to a standard normal À1 À1 2mML-glutamine, 100 U ml penicillin and 100 mgml distribution (unit transformation). The distribution of the streptomycin, and incubated at 37 1C in a 5% CO2 humidified expressed genes (z41) was analyzed to be log normal atmosphere. BCR stimulation was performed as previously distributed. This distribution could be normalized by a log described. Briefly, anti-IgM-polyacrylamid Immunobead reagent transformation. Ward’s hierarchical clustering algorithm was (Irvine Scientific, Santa Ana, CA, USA) was added to isolated used with correlation distance between CLL samples and B-CLL cells and incubated for 3 h.13 Anti-IgA-immunobeads controls (Figure 1a). The same distance measure was used in served as additional negative control. multidimensional scaling (MDS). MDS was applied to project the distances of the 78 genes into a two dimensional space (Figure 1b). Robust mean and variance are calculated for a Gene expression profiling two-group t-statistic. This statistic was used to identify differentially High-quality total RNA was extracted by Trizol (Invitrogen) lysis expressed genes (Figure 1c). and subsequent RNA-clean-up (Qiagen, Hilden, Germany). Total RNA (100 ng) was used for reverse transcription, second strand synthesis and in vitro transcription for generation of biotinylated cRNA according to the manufacturers’ protocols. Results For gene expression analysis Illumina WG6 Bead Chip Arrays (Illumina, San Diego, CA, USA) containing 447 000 transcripts Lipases and lipase activity associated genes are were hybridized, scanned and decoded as previously described.14 overexpressed in B-CLL patients To identify new pathogenic pathways in CLL, we assessed genome-wide transcription of CLL cells (n ¼ 8) compared to Treatment of cells CD5 þ B cells derived from healthy donors (n ¼ 5). Addition- Orlistat (Sigma, Steinheim, Germany) was dissolved in dimethyl ally, all samples were BCR stimulated to reveal functional sulfoxide (DMSO) and applied for CLL inhibition using aspects of BCR signaling as a central functional component in 1–100 mM concentrations. Fludarabine (Schering, Berlin, Germany) was dissolved in water and applied in 10 mM concentration. DMSO-treated cells were taken as baseline control; effective DMSO concentration in all treatments was 0.1%.

Lipase activity assay Lipase activity of cellular LPL was assessed by using Confluolip fluorimetric lipase activity assay (Progen, Heidelberg, Germany) according to the manufacturers’ protocol. Briefly, CLL cells and peripheral blood mononuclear cells (PBMC) were cultured for 24 h, washed and lysed using Triton X-100. For analysis of lipase activity, 100 mg of cellular lysate was added to 1-trinitrophenyl- aminododecanoyl-2-pyrendecanoyl-3-0-hexadecyl-sn-glycerol as substrate solution kept at 37 1C. Fluorescence of quenched pyrene, after adding lipase containing samples, was measured in a kinetic assay using 342 nm excitation and 400 nm emission in a thermostated cuvette at 37 1C. Purified LPL (Sigma) served as positive control. Serial dilutions of fluorescence standard were used for calibration of measurement of cellular lysates and positive control.

Apoptosis and cell viability Apoptosis was determined by flow cytometry using Annexin V-FITC/7AAD staining (BD PharMingen). Cell viability was additionally analyzed using XTT assays with cell proliferation Kit II (Roche). Analysis for poly(ADP-ribose) polymerase (PARP) cleavage was done by applying equal amounts of cells for lysis and western blotting; anti-PARP (NEB, Ipswich, MA, USA) was used for detection of PARP cleavage. Curve fitting was performed using SIGMAPLOT (SPSS Inc., Figure 1 Differential expression of lipases in chronic lymphocytic Chicago, IL, USA). IC50 values were determined by fitting data to leukemia (CLL). (a) Hierarchical clustering with genes included by b b b the Gene Ontology term GO:0016298 (lipase activity) comparing the Hill’s equation, that is, y ¼ y0 þðax Þ=ðc þ x Þ. gene expression profiles of native (n ¼ 8) and B-cell receptor (BCR)-stimulated CLL cells (n ¼ 8), healthy unstimulated (n ¼ 5) and BCR-stimulated (n ¼ 5) CD5 þ B cells. (b) Multidimensional scaling Statistical analysis of gene expression data analysis of lipase activity-associated gene expression data sets The log2 expression values of the Illumina Whole Genome Bead revealing distinct subgroups defined by lipases. (c) Lipase activity Chip Array data were used in this analysis. In order to compare associated genes with significant differential expression in CLL the data from different arrays, a z-transformation was applied to samples (red bars) and healthy CD5 þ B cells (blue bars).

Leukemia Orlistat cytotoxicity in CLL CP Pallasch et al 587 B cells. After purification, cells were incubated for 3 h with anti-human IgM-immunobead reagent for BCR stimulation, with anti-human IgA-immunobeads as negative control, or remained untreated in culture medium. Total RNA was extracted from both BCR-stimulated and BCR-unstimulated samples and applied to the Illumina Whole Genome Bead Chip Array containing 447 000 transcripts. The anti-IgA control revealed similar gene expression profiles as untreated negative controls. Subsequent analysis of gene expression profiles revealed significant changes in lipid metabolism associated genes when comparing CLL cases with healthy controls. Further analysis based on the Gene Ontology (GO) database AmigO revealed 78 genes defined by the GO-term ‘lipase activity’ as discriminator of healthy donor samples and diseased CLL samples. Hierarchical clustering and MDS analysis showed Figure 2 Lipase activity assay of chronic lymphocytic leukemia (CLL) that 78 genes exhibiting lipase activity or bearing a close and healthy donor cellular lysates. Fluorimetric assessment of lipase relation to lipase activity clearly define the healthy and the activity in CLL cell lysates (n ¼ 6) showed significantly higher levels diseased subset (Figures 1a and b). A total of 21 of 78 genes (1248.3 U mgÀ1) compared to lipase activity in healthy donor B cells according to the lipase activity GO-term were identified to be (n ¼ 8; 273.1 U mgÀ1; P ¼ 0.012) and healthy donor PBMCs (n ¼ 4; À1 the principle components of differential expression. Of these 21 72 U mg ; P ¼ 0.0021; left box). Orlistat treatment significantly ¼ differentially expressed genes, 20 were overexpressed in CLL reduces lipase activity in CLL cell lysates (P 0.0062; right box) while no effect could be seen for control cells (data not shown). (Figure 1c). The majority of genes in this cluster exhibited phospholipase activity. Most prominently, seven of these genes were phospho- Figure 2 (P ¼ 0.0021) and healthy donor B cells (237.5 U mgÀ1; lipase A2 (PLA2) enzymes or showed PLA2 domain homologies. P ¼ 0.012). PLA2 g (PLA2G4C), group IIF phospholipase (PLA2G2F) and group IIE phospholipase (PLA2G2E) induce the release of arachidonate from phospholipids.15–17 LEP14 and LEP1318 were BCR stimulation significantly increases LPL expression predicted in silico to have PLA2 activity, and PATE bears a in B-CLL cells putative PLA2 motif.19 Adiponutrin is similar to patatin-like Gene expression profiling further indicated that LPL expression phospholipase domain-containing protein 3 and also shows was specifically induced by BCR stimulation in CLL cells. No PLA2 activity.20 expression was seen in healthy donor control CD5 þ B cells Phospholipase C activity or relation was found in five genes; after BCR stimulation. To verify the expression data, we phospholipase C-b (PLCb1) catalyzes the generation of inositol performed real-time PCR for quantitative analysis of LPL. In 1,4,5-trisphosphate and diacylglycerol from phosphatidylinosi- CD5 þ healthy donor B cells (n ¼ 9), no significant expression of tol 4,5-bisphosphate, a key step in the intracellular transduction LPL could be detected either in unstimulated (median expres- of many extracellular signals,21 whereas PLC-b2 prefers sion ratio (ER) 0.3) or in BCR-stimulated cells (mean ER 0.58; phosphatidyl-inositol 4,5-bisphosphate to phosphatidylinositol P40.05). In CLL cells (n ¼ 28) a significant increase of LPL as substrate.22 The bradykinin receptor B2 is directly interacting expression was observed after BCR stimulation (median ER 0.21 23,24 with phospholipase C and provides mitogenic stimuli. Two versus 14.59; Po0.01). When separating IgVH-mutated and - genes associated with phospholipase D activity were also unmutated CLL samples, induction of LPL was seen in both identified: first phospholipase D2 (PLD2) catalyzes the hydro- subsets. In IgVH-unmutated samples (n ¼ 11) we found a 4.75- lysis of PC to produce phosphatidic acid and choline, PLD2 was fold increase after BCR stimulation (median ER 3.2 versus 15.23, 25,26 shown to prevent apoptosis in cancer cell lines and Po0.05). In IgVH-mutated samples (n ¼ 17) unstimulated cells increases Syk-mediated signaling in mast cells.27 We also exhibited low LPL expression (median ER ¼ 0.08). Nevertheless, identified LPL as CLL-specific expressed gene in agreement after BCR stimulation LPL expression was induced (median with previously published data.8,28 As expected, LPL expression ER ¼ 0.08 versus 7.87, Po0.01) (Figure 3). was found in samples from unstimulated IgVH-unmutated CLL patients, but not from healthy donor cells or BCR-unstimulated IgVH-mutated patients. Effects of orlistat treatment in primary CLL samples Among the genes associated with lipase activity, LPL was a particularly interesting candidate gene. First, it has been shown Cellular lipase activity in CLL and healthy donor cells recently that this gene might serve as a surrogate marker for For detection of intracellular and cellular bound lipases, a total unfavorable prognosis in CLL. Second, the overexpression of LPL lipase activity assay was performed using 1-trinitrophenyl- and its regulation by BCR stimulation in CLL allowed to aminododecanoyl-2-pyrendecanoyl-3-0-hexadecyl-sn-glycerol hypothesize that this protein might not only serve as a surrogate as substrate for triglyceride specific hydrolysis by cellular lysates marker but might also be involved in the pathophysiology of of CLL (n ¼ 6) and healthy donor control B-cells and PBMC CLL. Third, as shown above, CLL cells exhibit significantly (n ¼ 4). Fluorimetric kinetic measurement of cellular lysates higher lipase activity compared to healthy donor controls including purified LPL þ control revealed activity of both (Figure 3). Moreover, LPL can be targeted with the specific hepatic and LPL, a pH of 8.2 allowed analysis of LPL activity lipase inhibitor orlistat (tetrahydrolipstatin). as main part of total lipase activity (Figure 2). To address the potential role of LPL, we tested the effects of a Comparison of lipase activity in CLL-cell lysates revealed an lipase inhibitor, orlistat (tetrahydrolipstatin), which has a rather activity of 1248.3 U mgÀ1 while healthy donor PBMC showed broad specificity for lipases and is approved for the treatment of significantly lower lipase activity of 72 U mgÀ1 as shown in obesity. When assessing cell survival of B-CLL cells exposed to

Leukemia Orlistat cytotoxicity in CLL CP Pallasch et al 588 increasing concentrations of orlistat, a specific and concentra- apoptosis was analyzed by Annexin V/7-AAD flow cytometry. tion-dependent induction of apoptosis and cell death in primary Annexin V/7-AADÀ cells were defined as viable cells and CLL cells was observed. background apoptosis rates were subtracted for determination of Freshly isolated CLL cells (n ¼ 12) were incubated with IC50. With this method, the half-maximal inhibitory concentra- increasing concentrations of orlistat (1–100 mM). The influence tion (IC50) of orlistat was determined at 2.35 mM after 24 h on cell viability was measured by XTT assay. The induction of incubation (after 48 h at 1.58 mM). A similar IC50 value of 5.48 mM of orlistat was found by the XTT cytotoxicity assay. In healthy donor B cells (n ¼ 6), orlistat showed a significantly higher IC50 of 148.5 mM. On healthy donor PBMCs without prior cell sorting, 100 mM orlistat induced only marginal apoptotic effects with 13.1% Annexin V þ cells (Figure 4a). We further investigated the impact of LPL expression, clinical stage, IgVH-mutational status and ZAP70 expression on sensitivity of CLL cells toward orlistat. Suboptimal doses of 20 mM orlistat incubated with CLL cells for 24 and 48 h were used. After 24 h, LPLÀ patients (n ¼ 8) showed 40.8% viable cells while LPL þ cells (n ¼ 6) showed no significant difference with 41.5% viability (P ¼ 0.9). After 48 h, LPLÀ patients showed 22.1% viability, in LPL þ a reduced viability of 14.2% was seen. Nevertheless, no significant difference could be demonstrated (P ¼ 0.12) indicating no significant susceptibility toward orlistat in the LPL-defined patient subgroups (Figure 4b). Samples Figure 3 Regulation of lipoprotein lipase (LPL) expression. Analysis derived from patients in clinical stage Binet A (n ¼ 6) showed of LPL-expression in IgVH-mutated (n ¼ 17) chronic lymphocytic 68.5% apoptosis, those from Binet B stage patients (n ¼ 7) leukemia (CLL) cells, unmutated (n ¼ 11) CLL cells and in healthy 66.0% and Binet C stage patients (n ¼ 8) 55.0% apoptosis, donor CD5 þ B cells (n ¼ 9). Expression of LPL mRNA expression was indicating that there was no significant difference shown in analyzed by real-time PCR. In native, unstimulated cells LPL orlistat response with regard to Binet stage. Analyzing the expression is only seen in IgV -unmutated CLL cells. B-cell receptor H correlation of the response to orlistat with the ZAP70 expression, stimulation of 3 h significantly induces LPL expression in IgVH-mutated CLL (Pp0.01) and increases LPL expression in IgVH-unmutated CLL 20 mM orlistat (for 24 h) induced a 63.7% apoptosis induction in (Pp0.05), no induction is seen in healthy donor CD5 þ B cells. ZAP70À patients (n ¼ 16) and 55.7% apoptosis induction in

Figure 4 (a) Determination of orlistat IC50 in chronic lymphocytic leukemia (CLL), PBMCs and healthy B cells. Orlistat-induced apoptosis was assessed by Annexin V-FITC/7-AAD staining. Half-maximal dosage (IC50) of orlistat for apoptosis induction in CLL cells was 2.35 mM orlistat; IC50 in healthy B cells was significantly increased to 148.5 mM; only marginal cytotoxic orlistat effects were seen in healthy PBMCs. (b) Orlistat susceptibility of CLL samples based on baseline LPL expression. Viability of CLL cells upon treatment with 20 mM orlistat after 24 and 48 h in LPLÀ (n ¼ 8) and LPL þ patients (n ¼ 6). (c) Poly (ADP-ribose) polymerase (PARP) cleavage by orlistat in CLL samples. Equal amounts of CLL cells were treated for 24 h with DMSO as negative control or rising concentrations of orlistat (1, 20, 100 mM), lysed and PARP cleavage was detected by immunoblotting. Cleaved and activated PARP is detected at 89 kDa. Orlistat treatment induces complete PARP cleavage in CLL cells while no orlistat specific cleavage is seen in healthy PBMCs. (d) Caspase-dependent induction of apoptosis by orlistat. Significantly increased viability by inhibition with pan-caspase inhibitor z-VAD in 20 mM orlistat-treated CLL cells compared to CLL cells solely treated with 20 mM orlistat without z-VAD (42 versus 9.9% viable cells; P ¼ 0.016).

Leukemia Orlistat cytotoxicity in CLL CP Pallasch et al 589 The induction of apoptosis was additionally examined by detection of cleaved PARP in orlistat-treated cells. Orlistat treatment caused a dose-dependent effect on PARP with a total cleavage of the 116 kDa isoform to the 89 kDa cleaved PARP at 100 mM orlistat in all three CLL samples analyzed. No specific cleavage of PARP was detected using PBMCs derived from healthy donors (Figure 4c). The proapoptotic effects of orlistat treatment in CLL cells was further analyzed by caspase inhibition with the pan-caspase inhibitor z-VAD. CLL cells (n ¼ 5) were incubated with 20 mM orlistat and 20 mM z-VAD pan-caspase inhibitor for 48 h. Caspase inhibition by z-VAD in 20 mM orlistat-treated CLL cells revealed significantly enhanced viability of 42 versus 9.9% in solely 20 mM orlistat-treated CLL cells (P ¼ 0.016) (Figure 4d). Again, for detection of intracellular and cellular bound lipases, a total lipase activity assay applied to cellular lysates of CLL cells (n ¼ 6) was treated with orlistat (Figure 2). Incubation of CLL cells with orlistat revealed dose-dependent À1 decrease of lipase activity to 721.9 U mg by 5 mM orlistat treatment (P ¼ n.s.) and significantly decreased to 617 U mgÀ1 by 20 mM orlistat treatment (P ¼ 0.0062) (Figure 2). To assess the relevance of triglyceride catalysis and accessi- bility of free fatty acids for survival of CLL cells we substituted 15 mM palmitic acid solved in DMSO to orlistat-treated CLL cells (n ¼ 6) and healthy donor controls (n ¼ 4) (Figure 5a). Addition of palmitic acid revealed a partly reduced cytotoxicity of low- dose orlistat (1 mM) in CLL cells (P ¼ 0.02), whereas no significant effect could be observed in healthy donor B cells. Nevertheless, a complete depletion of orlistat toxicity could not be achieved by palmitic acid substitution. As shown above, various target genes can be postulated as mediators of specific orlistat toxicity in CLL cells. Fatty acid synthase was previously published as orlistat target. Fatty acid synthase inhibitor C75 (Alexis) was used for further elucidation of underlying pathways of CLL orlistat susceptibility (Figure 5b). We assessed C75 as a selective fatty acid synthase inhibitor for toxicity assays in CLL patients (n ¼ 4) and healthy donor B cells (n ¼ 4). In CLL patient cells, IC50 of C75 treatment was determined for 20.03 mM, in healthy donors a slightly increased IC50 of 44.82 mM was observed (P ¼ 0.0059). Since LPL is induced by BCR stimulation, we analyzed the influence of BCR stimulation on orlistat sensitivity. BCR- unstimulated and BCR-stimulated CLL cells were treated with 20 and 100 mM orlistat, cell viability was assessed by Annexin V/ 7-AAD flow cytometry and was compared to DMSO control. After BCR stimulation, the induction of apoptosis by orlistat was partially inhibited; at 20 mM orlistat the rate of viable cells was Figure 5 (a) Substitution of palmitic acid in orlistat treatment. 24.2% in unstimulated cells and 44.2% in BCR-stimulated CLL Reduced cytotoxicity of submaximal dosed orlistat in CLL cells after substitution of free fatty acid by 15 mM palmitic acid (P ¼ 0.033). (b) cells (n ¼ 18; Po0.001). Using 100 mM orlistat, the rate of viable Cytotoxicity and determination of IC50 of fatty acid synthase inhibitor cells was 13.9% in unstimulated cells versus 37.2% in BCR- C75. Using C75, CLL cells (n ¼ 4) showed a decreased IC50 (20.03 mM) stimulated CLL cells (n ¼ 18; P ¼ 0.001) (Figure 5c). compared to healthy donor B cells (n ¼ 4; IC50 44.82 mM; P ¼ 0.0059). When combined with fludarabine, orlistat showed additive (c) Cytotoxic effects of orlistat in B-CLL cells after BCR stimulation. cytotoxic effects on CLL cells (n ¼ 6). Submaximal doses of Effects of orlistat treatment (20, 100 mM) on viability in CLL cells were 10 mM fludarabine and of 20 mM orlistat yielded an induction of assessed in unstimulated and BCR-stimulated cells. The percentage of viable cells was detected by Annexin V/7-AAD flow cytometry and apoptosis in 32.0 and 66% of CLL cells, respectively. The normalized by DMSO control. BCR-stimulated cells show enhanced combination of 10 mM fludarabine and 20 mM orlistat induced the viability versus unstimulated cells applying 20 mM orlistat (Po0.001) apoptosis in 80.5% of CLL cells, which was superior compared and 100 mM orlistat (P ¼ 0.001). to orlistat alone (P ¼ 0.03) (Figures 6a and b).

ZAP70 þ patients (n ¼ 7) (P ¼ n.s.). The comparison for IgVH- mutational status did also not reveal any significant difference. Discussion Sensitivity toward orlistat treatment of three patients harboring 17 deletion were analyzed revealing individual IC50 values of 5, The application of gene expression analysis on CLL cells 10 and 8.5 mM, indicating susceptibility to orlistat treatment in compared to CD5 þ B cells as their physiological counterpart the p53 deleted subgroup. led to the identification of a gene cluster defined by the

Leukemia Orlistat cytotoxicity in CLL CP Pallasch et al 590 might be critically involved in CLL pathogenesis. We could show that LPL is regulated by BCR stimulation in vitro in both IgVH-mutated and IgVH-unmutated CLL patient cells. Recently, IgVH-unmutated CLL cells were demonstrated to exhibit autoreactive BCR reactivity.5 Two lines of evidence suggest that LPL might indeed be regulated by BCR activation in vivo: first, LPL expression in vivo is almost exclusively seen in IgVH-unmutated CLL, which are associated with autoreactive BCR activity;5 second, we demonstrate here that LPL is induced or significantly increased by BCR stimulation ex vivo. Therefore, LPL expression has to be determined as a prerequisite feature of CLL cells. The functional role of the cytoplasmatic expression of LPL in CLL cells remains unclear, since LPL is a secreted enzyme, acts at the vascular side of endothelium and plays a central role for energy supply from serum triglycerides in peripheral tissues.10 The relation of LPL to BCR stimulation suggests a central role in CLL pathogenesis and supports the hypothesis of a persistent autoreactive BCR stimulation of IgVH-unmutated CLL cells in the microenvironment of the malignant cell. This CLL-specific expression pattern elucidates LPL as a potential therapeutic Figure 6 Additive effects of orlistat and fludarabine treatment. (a) target. Apoptosis was detected by Annexin V-FITC/7-AAD staining of freshly The identification of numerous lipase activity exhibiting genes isolated chronic lymphocytic leukemia (CLL) cells and incubation for in CLL led us to screen for putative drugs that act by inhibition of 48 h with DMSO negative control, 20 mM orlistat, 10 mM fludarabine m m lipases. Orlistat is a broadly applied drug in obesity treatment and combination of 20 M orlistat and 10 M fludarabine. A flow 35 cytometric analysis is shown using a representative primary CLL sample. inhibiting gastro-intestinal lipases. It shows a broad activity (b) Additive cytotoxic effects were seen in combination of fludarabine and against, for example, gastric and pancreatic lipases. Only orlistat (n ¼ 7; P ¼ 0.03) compared to orlistat treatment alone. recently, orlistat was identified as an effective inhibitor of LPL,11 phospholipases36 and PLA2.37 Applying the clinically used lipase inhibitor to study the GO-term ‘lipase activity’,29 reliably distinguishing between both potential role of LPL in B-CLL we obtained a rather surprising subsets. Gene expression of 78 lipases and lipase activity result. Orlistat effectively induced apoptosis in CLL cells with an associated genes was demonstrated as significant difference IC50 of 2.35 mM, while PBMC were basically not affected and the between healthy and malignant B cells. To pursue the gene IC50 for purified B cells from healthy donors was almost two expression-based findings of lipase activity as CLL-specific magnitudes higher. Orlistat has been described to inhibit feature we performed functional assessment of lipase activity prostate cancer cell lines by inhibiting fatty acid synthase,38 exhibited by CLL cells in comparison to healthy donor controls. similar effects have been observed in breast cancer cell lines.39 Triglyceride-specific lipase activity could be significantly These effects have been mainly attributed to the overexpression increased in CLL cells compared to healthy donor controls. of fatty acid synthase in these cancer cell lines.40 Based on our Here we could link descriptive microarray data to enzymatic expression data, no significant expression of fatty acid synthase function to propose lipid metabolism as functional relevant in CLL was seen. In addition, we targeted fatty acid synthase by feature in CLL. This suggests that the role of lipid metabolism the FAS-specific inhibitor C75. High dosage of half-maximal might have been underestimated in B-CLL in the past. Principle 20 mM is necessary to induce cytotoxicity. In healthy donor component analysis of our microarray data set revealed 21 controls a significantly increased IC50 of 44.82 mM was observed. genes to be the major constituents in definition of functional Nevertheless, high dosage of C75 is necessary and comparable difference between healthy B cells and CLL. Seven of these amounts induce apoptosis in healthy donor cells as well. Fatty genes were identified as members of the PLA2 family. PLA2 acid synthase might by free fatty acid supply in part contribute to cytoplasmatic level correlates with apoptosis30 and PLA2 is CLL cell survival. important for maintaining the phospholipids composition of the To reveal further insight in orlistat cytotoxicity we performed membrane especially during oxidative stress.31 substitution of palmitic acid as free fatty acid during orlistat. This Another interesting link between apoptosis and lipid meta- resulted in a significant reduction of orlistat cytotoxicity. A bolisms originates from studies on PLD2, which we also complete termination of orlistat effects could not be achieved, identified as upregulated in B-CLL cells. In fact, PLD2 was indicating that energy supply through fatty acid may in part play shown to prevent apoptosis in cancer cell lines25,26 and a role in exhibiting orlistat cytotoxicity. In addition, orlistat increases Syk-mediated signaling in mast cells.27 Similarly, seems to exert apoptosis induction in a caspase-dependent PLC is an important enzyme in intracellular signaling and was manner, as shown by countering orlistat toxicity with z-VAD recently identified to be involved in CLL BCR signaling.32 Here pan-caspase inhibitor. we identified five PLC genes overexpressed in CLL indicating the Several mechanisms of orlistat cytotoxicity in CLL cells can be activation of this pathway in CLL. hypothesized from our data. First, orlistat can decrease energy It was recently shown that LPL expression in CLL is associated sources of the CLL cell by inhibition of LPL. Second, the 8,28,33,34 with unmutated IgVH-hypermutational status, providing disruption of intracellular phospholipase-dependent mediator better predictive value compared to ZAP70 expression.2 Our pathways can influence survival signals of the CLL cell. Third, study adds to this knowledge by showing that LPL as a recently membrane stability can be decreased by blocking PLA2 activity. identified prognostic marker in CLL is induced by BCR Here we could identify the availability of free fatty acid as at stimulation. This indicates that pathways of the lipid metabolism least one component of orlistat toxicity. This finding is supported

Leukemia Orlistat cytotoxicity in CLL CP Pallasch et al 591 by functional lipase activity analysis in orlistat-treated CLL cells, 12 Matthews C, Catherwood M, Morris TC, Alexander HD. Routine harboring a decreased triglyceride specific lipase activity by analysis of IgVH mutational status in CLL patients using BIOMED-2 orlistat treatment. A miscellaneous mechanism might be the standardized primers and protocols. Leuk Lymphoma 2004; 45: basis for induction of apoptosis by the broad-spectrum lipase 1899–1904. 13 Kofler DM, Buning H, Mayr C, Bund D, Baumert J, Hallek M et al. inhibitor orlistat specifically in CLL and future studies have to Engagement of the B-cell antigen receptor (BCR) allows efficient further elucidate the metabolic changes induced by orlistat. transduction of ZAP-70-positive primary B-CLL cells by recombi- We compared orlistat to fludarabine as a well-established and nant adeno-associated virus (rAAV) vectors. Gene Therapy 2004; potent drug in CLL treatment41 and found additive cytotoxic 11: 1416–1424. effects in combining orlistat with fludarabine. Fludarabine acts 14 Debey S, Zander T, Brors B, Popov A, Eils R, Schultze JL. A highly as an antimetabolite in nucleoside metabolism,42 and so far no standardized, robust, and cost-effective method for genome-wide transcriptome analysis of peripheral blood applicable to large- interference with lipid metabolism was reported. In oral scale clinical trials. Genomics 2006; 87: 653–664. application as anti-obesity treatment, orlistat shows little side 15 Underwood KW, Song C, Kriz RW, Chang XJ, Knopf JL, Lin LL. effects. Overdosage in humans with increased systemic uptake A novel calcium-independent phospholipase A2, cPLA2-gamma, did not reveal severe side effects43 and animal models with that is prenylated and contains homology to cPLA2. J Biol Chem extreme overdosage did not reveal cytotoxic effects of orlistat 1998; 273: 21926–21932. either (FDA hearing protocol). 16 Murakami M, Yoshihara K, Shimbara S, Lambeau G, Singer A, These novel findings allow to speculate that drugs targeting Gelb MH et al. Arachidonate release and eicosanoid generation by group IIE phospholipase A(2). Biochem Biophys Res Commun the lipid metabolism might be a new class of agents for the 2002; 292: 689–696. treatment of CLL. Future investigations should aim at investigat- 17 Murakami M, Yoshihara K, Shimbara S, Lambeau G, Gelb MH, ing these agents in animal models. Singer AG et al. Cellular arachidonate-releasing function and inflammation-associated expression of group IIF secretory phospholipase A2. J Biol Chem 2002; 277: 19145–19155. Acknowledgements 18 Marshall D, Hardman MJ, Nield KM, Byrne C. Differentially expressed late constituents of the epidermal cornified envelope. We thank Reinhild Brinker for excellent technical support. CPP, Proc Natl Acad Sci USA 2001; 98: 13031–13036. 19 Bera TK, Maitra R, Iavarone C, Salvatore G, Kumar V, Vincent JJ DMK and CMW were supported by a grant from Else-Kro¨ner- et al. PATE, a gene expressed in prostate cancer, normal prostate, Fresenius-Stiftung (P17/06//A62/05), Ko¨ln Fortune (050808) and and testis, identified by a functional genomic approach. Proc Natl the ‘Deutsche Jose´ Carreras Leuka¨mie Stiftung’ (DJCLS R06/16). Acad Sci USA 2002; 99: 3058–3063. 20 Jenkins CM, Mancuso DJ, Yan W, Sims HF, Gibson B, Gross RW. Identification, cloning, expression, and purification of three novel References human calcium-independent phospholipase A2 family members possessing triacylglycerol lipase and acylglycerol transacylase 1 Caligaris-Cappio F, Hamblin TJ. B-cell chronic lymphocytic activities. J Biol Chem 2004; 279: 48968–48975. leukemia: a bird of a different feather. J Clin Oncol 1999; 17: 21 Caricasole A, Sala C, Roncarati R, Formenti E, Terstappen GC. 399–408. Cloning and characterization of the human phosphoinositide- 2 Messmer BT, Messmer D, Allen SL, Kolitz JE, Kudalkar P, Cesar D specific phospholipase C-beta 1 (PLC beta 1). Biochim Biophys et al. In vivo measurements document the dynamic cellular Acta 2000; 1517: 63–72. kinetics of chronic lymphocytic leukemia B cells. J Clin Invest 22 Park D, Jhon DY, Kriz R, Knopf J, Rhee SG. Cloning, sequencing, 2005; 115: 755–764. expression, and Gq-independent activation of phospholipase 3 Ghia P, Granziero L, Chilosi M, Caligaris-Cappio F. Chronic B cell C-beta 2. J Biol Chem 1992; 267: 16048–16055. malignancies and bone marrow microenvironment. Semin Cancer 23 Greco S, Muscella A, Elia MG, Romano S, Storelli C, Marsigliante Biol 2002; 12: 149–155. S. Mitogenic signalling by B2 bradykinin receptor in epithelial 4 Stevenson FK, Caligaris-Cappio F. Chronic lymphocytic leukemia: breast cells. J Cell Physiol 2004; 201: 84–96. revelations from the B-cell receptor. Blood 2004; 103: 4389–4395. 24 Hess JF, Borkowski JA, Macneil T, Stonesifer GY, Fraher J, 5 Herve M, Xu K, Ng YS, Wardemann H, Albesiano E, Messmer BT Strader CD et al. Differential pharmacology of cloned human et al. Unmutated and mutated chronic lymphocytic leukemias and mouse B2 bradykinin receptors. Mol Pharmacol 1994; 45: derive from self-reactive B cell precursors despite expressing 1–8. different antibody reactivity. J Clin Invest 2005; 115: 1636–1643. 25 Yamada M, Banno Y, Takuwa Y, Koda M, Hara A, Nozawa Y. 6 Damle RN, Wasil T, Fais F, Ghiotto F, Valetto A, Allen SL et al. Ig V Overexpression of phospholipase D prevents actinomycin gene mutation status and CD38 expression as novel prognostic D-induced apoptosis through potentiation of phosphoinositide indicators in chronic lymphocytic leukemia. Blood 1999; 94: 3-kinase signalling pathways in Chinese-hamster ovary cells. 1840–1847. Biochem J 2004; 378: 649–656. 7 Chen L, Widhopf G, Huynh L, Rassenti L, Rai KR, Weiss A et al. 26 Zhong M, Shen Y, Zheng Y, Joseph T, Jackson D, Foster DA. Expression of ZAP-70 is associated with increased B-cell receptor Phospholipase D prevents apoptosis in v-Src-transformed rat signaling in chronic lymphocytic leukemia. Blood 2002; 100: fibroblasts and MDA-MB-231 breast cancer cells. Biochem 4609–4614. Biophys Res Commun 2003; 302: 615–619. 8 Oppezzo P, Vasconcelos Y, Settegrana C, Jeannel D, Vuillier F, 27 Lee JH, Kim YM, Kim NW, Kim JW, Her E, Kim BK et al. Legarff-Tavernier M et al. The LPL/ADAM29 expression ratio is a Phospholipase D2 acts as an essential adaptor protein in the novel prognosis indicator in chronic lymphocytic leukemia. Blood activation of Syk in antigen-stimulated mast cells. Blood 2006; 2005; 106: 650–657. 108: 956–964. 9 Bilban M, Heintel D, Scharl T, Woelfel T, Auer MM, Porpaczy E 28 Heintel D, Kienle D, Shehata M, Kro¨ber A, Kroemer E, et al. Deregulated expression of fat and muscle genes in B-cell Schwarzinger I et al. High expression of lipoprotein lipase in poor chronic lymphocytic leukemia with high lipoprotein lipase risk B-cell chronic lymphocytic leukemia. Leukemia 2005; 19: expression. Leukemia 2006; 20: 1080–1088. 1216–1223. 10 Preiss-Landl K, Zimmermann R, Hammerle G, Zechner R. 29 Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM Lipoprotein lipase: the regulation of tissue specific expression et al. Gene ontology: tool for the unification of biology. The Gene and its role in lipid and energy metabolism. Curr Opin Lipidol Ontology Consortium. Nat Genet 2000; 25: 25–29. 2002; 13: 471–481. 30 Dong M, Johnson M, Rezaie A, Ilsley JN, Nakanishi M, Sanders 11 Lookene A, Skottova N, Olivecrona G. Interactions of lipoprotein MM et al. Cytoplasmic phospholipase A2 levels correlate with lipase with the active-site inhibitor tetrahydrolipstatin (Orlistat). apoptosis in human colon tumorigenesis. Clin Cancer Res 2005; Eur J Biochem 1994; 222: 395–403. 11: 2265–2271.

Leukemia Orlistat cytotoxicity in CLL CP Pallasch et al 592 31 Asai K, Hirabayashi T, Houjou T, Uozumi N, Taguchi R, Shimizu phospholipase A(2) in obese patients with metabolic syndrome. T. Human group IVC phospholipase A2 (cPLA2gamma). Roles in Atherosclerosis 2006; 193: 428–437. the membrane remodeling and activation induced by oxidative 38 Kridel SJ, Axelrod F, Rozenkrantz N, Smith JW. Orlistat is a novel stress. J Biol Chem 2003; 278: 8809–8814. inhibitor of fatty acid synthase with antitumor activity. Cancer Res 32 Petlickovski A, Laurenti L, Li X, Marietti S, Chiusolo P, Sica S et al. 2004; 64: 2070–2075. Sustained signaling through the B-cell receptor induces Mcl-1 and 39 Menendez JA, Vellon L, Lupu R. Antitumoral actions of the anti- promotes survival of chronic lymphocytic leukemia B cells. Blood obesity drug orlistat (XenicalTM) in breast cancer cells: blockade 2005; 105: 4820–4827. of cell cycle progression, promotion of apoptotic cell death and 33 Nuckel H, Huttmann A, Klein-Hitpass L, Schroers R, Fu¨hrer A, PEA3-mediated transcriptional repression of Her2/neu (erbB-2) Sellmann L et al. Lipoprotein lipase expression is a novel oncogene. Ann Oncol 2005; 16: 1253–1267. prognostic factor in B-cell chronic lymphocytic leukemia. Leuk 40 Menendez JA, Vellon L, Mehmi I, Oza BP, Ropero S, Colomer R Lymphoma 2006; 47: 1053–1061. et al. Inhibition of fatty acid synthase (FAS) suppresses HER2/neu 34 van’t Veer MB, Brooijmans AM, Langerak AW, Verhaaf B, (erbB-2) oncogene overexpression in cancer cells. Proc Natl Acad Goudswaard CS, Graveland WJ et al. The predictive value of Sci USA 2004; 101: 10715–10720. lipoprotein lipase for survival in chronic lymphocytic leukemia. 41 Elter T, Borchmann P, Schulz H, Reiser M, Trelle S, Schnell R et al. Haematologica 2006; 91: 56–63. Fludarabine in combination with alemtuzumab is effective and feasible 35 Hvizdos KM, Markham A. Orlistat: a review of its use in the in patients with relapsed or refractory B-cell chronic lymphocytic management of obesity. Drugs 1999; 58: 743–760. leukemia: results of a phase II trial. J Clin Oncol 2005; 23: 7024–7031. 36 Migas I, Severson DL. Diacylglycerols derived from membrane 42 Wendtner CM, Eichhorst BF, Hallek MJ. Advances in chemother- phospholipids are metabolized by lipases in A10 smooth muscle apy for chronic lymphocytic leukemia. Semin Hematol 2004; 41: cells. Am J Physiol 1996; 271: C1194–C1202. 224–233. 37 Filippatos TD, Gazi IF, Liberopoulos EN, Athyros VG, Elisaf MS, 43 Zhi J, Melia AT, Eggers H, Joly R, Patel IH. Review of limited Tselepis AD et al. The effect of orlistat and fenoﬁbrate, alone or systemic absorption of orlistat, a lipase inhibitor, in healthy human in combination, on small dense LDL and lipoprotein-associated volunteers. J Clin Pharmacol 1995; 35: 1103–1108.

Leukemia