Nitric Oxide–Donating Acetylsalicylic Acid Induces Apoptosis in Chronic Lymphocytic Leukemia Cells and Shows Strong Antitumor Efficacy In vivo

Dysregulation of apoptosis and subsequent accumulation of malignant, nonfunctional B cells are major characteristics of chronic lymphocytic leukemia (CLL; ref. 1). CLL patients show a very heterogeneous clinical course. Although some patients remain stable for years without the need of treatment, others advance within months requiring treatment immediately (2). Until recently, treatment was limited to conventional chemotherapy with chlorambucil, purine analogs, and other cytostatic drugs such as mitoxantrone and cyclophosphamide. Although combinations of these substances, for example fludarabine together with cyclophosphamide, have shown superior activity compared with single agent therapy, further treatment options more recently established include monoclonal antibodies such as rituximab (anti-CD20) and alemtuzumab (anti-CD52 ref. 3). In general, treatment focuses on controlling the disease and its symptoms rather than an outright cure. Thus, new targeted CLL cell-specific therapies to increase the number and duration of remissions are highly desirable.

Several aberrantly regulated signaling cascades have been implicated in the pathogenesis of CLL. One of them is the Wnt/β-catenin/Tcf (T-cell factor)/Lef-1 (lymphoid enhancer factor) pathway (4). While playing a crucial role in embryogenesis and several developmental processes (5), this pathway is mainly downregulated or even completely shut off in the adult organism. Aberrant activity of β-catenin/Tcf/Lef-1 has been associated with the development of a variety of cancers (6, 7). Aberrant Wnt/β-catenin/Tcf/Lef-1 signaling can be due to several defects within the signaling cascade all resulting in enhanced activity of the final effector of the cascade, the transcriptional β-catenin/Tcf/Lef-1 complex. Therefore, this pathway has been discussed to have a high potential for therapeutic interventions (8).

In CLL, Lef-1 was identified as one of the most overexpressed genes (9). A complex comprising a member of the TCF/Lef-1 transcription factor family and its coactivator β-catenin is responsible for expression of, for example, C-myc (10), CCND1, Cyclin D1 (11), or Lef-1 itself (12). These proteins are known for their roles in proliferation, cell growth, and also when expressed in an uncontrolled fashion, in tumorigenesis. Hence, the aberrant activity of Lef-1 in CLL might be involved in the pathogenesis of this disease and could be of potential interest for CLL therapy. Several compounds of the nonsteroidal anti-inflammatory drug (NSAID) family have been shown to induce cell death or inhibit proliferation in β-catenin/Tcf/Lef-1 positive cancers such as colorectal cancer (13), breast cancer (14), or CLL (15). Unfortunately, these compounds could not reach therapeutic plasma concentrations without producing significant toxicities. For example, R-etodolac was tested in CLL. IC₅₀ values in vitro were 800 μmol/L (15). When R-etodolac was tested in a phase II study, the maximum blood concentrations of 300 to 600 μmol/L achieved were not sufficient to have a potent effect on CLL cells. Higher concentrations of R-etodolac resulted in significant side effects in this study. Until now, the need for high concentrations of NSAIDs for a potent anticancer effect has limited their clinical use. To overcome this limitation, nitric oxide-donating acetylsalicylic acid (NO-ASA) has been developed (16). It consists of a traditional molecule of ASA to which the NO-donating moiety, −ONO₂, is covalently bound via a spacer. The general structure enables construction of several variants of NO-ASA depending on the position of the spacer-linked NO-donating group. Under physiologic conditions, locally generated NO is involved in gastric mucosa defense. The rational of the development of NO-ASA was, therefore, the creation of a gastrointestinal protective effect by NO-mediated increased mucosal defense, while maintaining the systemic effects of the drug (17). The meta-isomer of NO-ASA was shown to be well tolerated and to achieve high plasma levels in doses not leading to any side effects in humans (18, 19). Interestingly, the pharmacologic profile of NO-ASA seems to be dependent on the positional isomerism, with different isomers showing varying efficacies in different disease backgrounds (20). The para-isomer, for example, was shown to be a very potent inducer of apoptosis in colon cancer cell lines and in an intestinal cancer mouse model (21). The same group demonstrated a disruption of the β-catenin/Tcf/Lef-1 complex as a potential mechanism of NO-ASA action in colon cancer (13, 22), a tumor where 100% of cases show an aberrant activation of Wnt/β-catenin/Tcf/Lef-1 signaling due to an APC or a CTNNB1 (β-catenin) mutation. Also, in the human T-cell leukemia cell line Jurkat, the para-isomer of NO-ASA was shown to modulate the expression of β-catenin, going along with a strong induction of apoptosis (23). In addition, in a β-catenin/Tcf/Lef-1–positive breast cancer cell line, NO-ASA was shown to exhibit potent proapoptotic abilities associated with an interference with the β-catenin/Tcf/Lef-1 complex (14).

As most studies described a potent proapoptotic effect of the para-isomer of NO-ASA, especially in Wnt/β-catenin/Tcf/Lef-1–active cancers, we investigated the effect of this compound on primary CLL cells. We further studied the expression status of β-catenin/Lef-1 and known target genes of this complex upon treatment to get insight into the potential mode of action of NO-ASA in CLL. In addition, we investigated the efficacy of para-NO-ASA in a CLL xenograft nude mouse model. This study might identify NO-ASA as a potential therapeutic advancement in the treatment of CLL.

Peripheral blood was taken from CLL patients during routine diagnostic phlebotomy or from healthy volunteers after giving informed consent. All CLL patients had a confirmed diagnosis following standard criteria (24). Patient characteristics, such as age, gender, ZAP70 and CD38 status, Binet stage, and cytogenetic information, are listed in Supplementary Table 1. Peripheral blood monoculear cells (PBMC) were isolated by Ficoll-Hypaque density gradient centrifugation. CLL cells were enriched using RosetteSep (STEMCELL) according to the manufacturer's instructions. Patient samples were never older than 1 day. Primary cells were maintained in RPMI-1640 culture medium (Biochrom AG) which contained 1% penicillin/streptomycin (Biochrom AG) and 20% fetal bovine serum (FBS; Biochrom AG) and incubated at 37°C and 5% CO₂ in a humidified atmosphere.

The study was performed according to the World Medical Association's Declaration of Helsinki (6th version, Seoul/South Korea 2008) and authorized by the ethics committee at the University of Cologne (approval no. 04-231).

JVM-3 cells were purchased from the German Collection of Microorganisms and Cell Cultures (DSMZ) and maintained in RPMI-1640 that contained 20% FBS and 1% penicillin/streptomycin and cultured under abovementioned standard conditions.

Apoptotic cell death was assessed by flow cytometry (BD FACS Canto) using the Annexin V–FITC/PI Apoptosis Detection Kit (BD Pharmingen) following the manufacturer's instructions. FITC-labeled Annexin V binds phosphatidylserine on the surface of apoptotic cells, but not living cells, whereas PI cannot be excluded by dead cells. Hence, Annexin V–FITC and PI double-negative cells were considered viable cells, Annexin V–FITC-positive/PI-negative cells were considered apoptotic cells and Annexin V–FITC/PI double-positive cells were considered dead cells. All calculations were done relative to dimethylsulfoxide (DMSO) control. Lethal concentration 50 (LC₅₀) is the concentration at which 50% of cells were dead. Determination of LC₅₀ was done with Graph Pad Prism Software.

para-NO-ASA (2-(acetyloxy)-4-[(nitrooxy)methyl]phenyl ester, benzoic acid) was purchased from Cayman Chemical Company (Fig. 1). Aliquots of the stock solution with a final concentration of 10 mmol/L in DMSO were stored at −20°C. For in vivo experiments, the compound was suspended homogenously in 1% of carboxymethyl cellulose (CMC; Sigma-Aldrich). The suspension was freshly prepared before each application.

A caspase inhibitor (ApoBlock, BD Bioscience) was used to evaluate the role of caspases. CLL cells were incubated with 50 μmol/L of caspase inhibitor for 1 hour before the addition of para-NO-ASA or DMSO as a control for 24 hours. Cell survival was determined by Annexin V–FITC/PI staining and protein lysates were analyzed by immunoblotting.

Primary CLL cells were incubated either with 1, 10, and 20 μmol/L of para-NO-ASA, 0.5% of DMSO (vehicle-treated control), or left untreated for 24 hours. Equal amounts of total cell protein extracts were separated on NuPage Novex 4% to 12% Bis-Tris gels (Invitrogen GmbH) and transferred to nitrocellulose membranes (Invitrogen GmbH). Primary antibodies targeting Bcl-2, X-linked inhibitor of apoptosis (XIAP), Cyclin D1, β-catenin, and β-Actin were purchased from BD Bioscience. Primary antibodies against Lef-1, caspase 9, cleaved caspase 9, caspase 3, cleaved caspase 3, poly(ADP-ribose) polymerase (PARP), and cleaved PARP were obtained from Cell Signaling Technology. Anti-C-myc antibody was purchased from Santa Cruz Biotechnology. Secondary horseradish peroxidase (HRP)–labeled antibodies were obtained from DAKO Deutschland GmbH. Antibody binding was detected using Amersham ECL Western blotting detection reagents (GE Healthcare UK Limited). Band intensities were analyzed using ImageJ software. Average band intensities ± SEM are displayed in Supplementary Figures 1, 2, and 3.

Xenograft nude mouse models were established as described by Loisel and colleagues (25). Briefly, 1 × 10⁷ JVM-3 cells were injected subcutaneously into the right flank of 4-week-old CD-1 female nu/nu mice (Charles River Laboratories). When the tumor volume (TV) reached ∼ 60 mm³, mice were divided into 3 groups comprising 10 mice per group. Mice were treated daily for 22 days with 100 mg/kg of body weight of para-NO-ASA or vehicle (1% CMC) control p.o. (by mouth) via gavage. TV was evaluated every third day, using a standard vernier calliper, starting with the first day of treatment (day 0). Animals were visually examined for potential side effects during and after the treatment period. When the TV exceeded 1,000 mm³, mice were excluded from the experiment and sacrificed for ethical reasons.

The tumor volume was calculated according to the formula: TV = L × W × H/2, given that L is the tumor length (in mm), W the tumor width (in mm), and H the tumor height (in mm). The percentage of the inhibition rate (IR) of the TV was calculated according to the formula IR = (1 − RV_t/RV_c) × 100, as previously described (26).

Statistical analysis was performed using GraphPad Prism (GraphPad Software) for all experiments. Data are given as mean ± SEM. Unpaired Student's t test was used with 2-tailed comparison of the differences between test values and controls. P values < 0.05 were considered significant.

At first, we studied the effect of para-NO-ASA (Fig. 1) on its ability to reduce survival in CLL cells and healthy PBMCs. Primary CLL cells from 21 patients as well as healthy PBMCs from 4 voluntary donors were incubated with 1, 10, and 100 μmol/L of para-NO-ASA for 24 hours. The compound reduced the amount of annexin V/PI double-negative cells significantly with an LC₅₀ value of 8.72 ± 0.04 μmol/L. Healthy PBMCs were significantly less affected by para-NO-ASA. Because dose–response curves for healthy PBMCs did not reach sigmoidity, we could not calculate the LC₅₀ (Fig. 2A).

As determined, LC₅₀ values reached by para-NO-ASA treatment were very heterogeneous among studied CLL patients ranging from 2.16 to 114.50 μmol/L, we investigated correlations with patient characteristics (Supplementary Table 1). No correlation could be detected between either of the tested characteristics such as gender, age, Binet status, CD38 or ZAP70 status, percentage of CD5/CD19 double-positive cells, and cytogenetic abnormalities.

We further analyzed the time course of NO-ASA efficacy by assaying survival of primary CLL cells from 3 patients after treatment with para-NO-ASA (10 μmol/L) for different durations of time. An initial effect was seen after 6 hours when cell survival was already reduced down to almost 50% (53.4% ± 16.6%) of control levels (Fig. 2B). Survival reduction of 46.9% ± 8.5% after 9 hours of treatment and 43.6% ± 7.6% after 12 hours of treatment could not be increased by longer incubation times of up to 48 hours (data not shown).

In conclusion, para-NO-ASA reduced CLL cell survival in a time and concentration-dependent manner at low micromolar concentrations, whereas healthy PBMCs were affected to a significantly lesser extent.

Annexin V–FITC staining already demonstrated an apoptosis-mediated cell death induction. To further investigate the apoptotic mechanism induced by para-NO-ASA, we looked for major apoptotic prognostic markers via immunoblotting. Cells were treated with 1, 10, and 20 μmol/L of para-NO-ASA or vehicle control for 24 hours. In whole protein extracts, we could clearly detect concentration-dependent cleavage, hence activation of the initiator caspase 9, effector caspase 3, as well as the caspase 3 target PARP upon treatment (Fig. 3A). Caspase activation and cleavage of PARP occurred at concentrations between 1 and 10 μmol/L, therefore confirming the LC₅₀ of 8.72 ± 0.04 μmol/L.

As it is known that CLL is characterized by overexpression of antiapoptotic proteins, such as Bcl-2 and XIAP, we determined the expression levels of these proteins after treatment of CLL cells with 20 μmol/L of para-NO-ASA for 24 hours. Both Bcl-2 and XIAP were clearly downregulated upon treatment compared with vehicle control (Fig. 3B).

To gain further insight into the mechanistic background of NO-ASA action, we investigated the role of the β-catenin/Lef-1 transcriptional complex. Cells were treated with 1, 10, and 20 μmol/L para-NO-ASA or vehicle control for 24 hours. First, we determined protein levels of β-catenin and known β-catenin/Lef-1 targets such as Lef-1 itself, Cyclin D1, and C-myc. We detected a concentration-dependent reduction in full-length β-catenin protein levels, accompanied by the appearance of cleaved β-catenin protein upon para-NO-ASA treatment (Fig. 4A). Also, here the effect was visible at a concentration between 1 and 10 μmol/L. Furthermore, the β-catenin/Lef-1 targets Lef-1 itself, Cyclin D1 and C-myc were significantly reduced upon para-NO-ASA treatment with reduction increasing with concentration (Fig. 4B).

To determine whether β-catenin cleavage and subsequent reduction of its transactivational potential is a cause or consequence of apoptosis induction, we incubated CLL cells from 3 patients with a pan-caspase inhibitor prior to addition of para-NO-ASA (20 μmol/L). Pretreatment with the caspase inhibitor prevented the induction of apoptosis as indicated by a survival rate of 89.4% ± 4.1% compared with 32.7% ± 3.1% survival after para-NO-ASA treatment alone (Fig. 5A). This was supported by significantly reduced cleavage of caspase 3 and PARP upon para-NO-ASA treatment when cells were pretreated with the caspase inhibitor (Fig. 5B). Furthermore, levels of the antiapoptotic proteins Bcl-2 and XIAP remained unchanged when cells were incubated with both para-NO-ASA and the caspase inhibitor in contrast to para-NO-ASA treatment alone (Fig. 5C). Because we did not observed cleavage of β-catenin after para-NO-ASA treatment in the presence of the caspase inhibitor (Fig. 5D), it can be concluded that caspase activation precedes β-catenin cleavage and subsequent downregulation of β-catenin/Lef-1 target genes.

After obtaining promising in vitro results, we wanted to test the efficacy of NO-ASA in reducing and/or stabilizing tumor growth in a xenograft nude mouse model. Ten days postimplantation of JVM-3 cells into the right flank of irradiated CD1 female nu/nu mice, tumors had reached an average size of ∼ 60 mm³. Mice were divided into 3 groups with 10 mice each and treated daily with para-NO-ASA or vehicle control p.o. via gavage. TVs were measured every 3 days, including the first day of treatment, for 22 days. para-NO-ASA treatment led to a significantly reduced TV after 9 days of treatment compared with vehicle-treated control (126.4 ± 22.3 mm³ for para-NO-ASA vs. 290.0 ± 65.9 mm³ for vehicle control; P = 0.0303; Fig. 6). In the following days of treatment, tumor growth inhibition by para-NO-ASA became even more pronounced. Whereas the TV in vehicle treated controls reached 775.4 ± 219.6 mm³ after 21 days, the TV in the para-NO-ASA treated group remained stable at 128.7 ± 27.6 mm³ (P = 0.0091). The IR_max value of para-NO-ASA treatment versus vehicle control was 83.4%. During treatment, body weight remained stable and we did not observed visible side effects. Furthermore, histologic analysis of the liver, kidney, and spleen revealed no pathologic changes due to treatment (data not shown).

In conclusion, para-NO-ASA shows effective tumor growth inhibition in a xenograft nude mouse model. No obvious side effects occurred and the substance seemed to be well tolerated.

Here, we demonstrate that the para-isomer of NO-ASA induces apoptosis in CLL cells at low concentrations. Furthermore, this compound is highly selective toward malignant cells, as it affects healthy PBMCs significantly less. Despite highly variable LC₅₀ values ranging from 2.16 to 114.50 μmol/L upon para-NO-ASA treatment, no correlation could be detected between patient characteristics (Supplementary Table 1) and the sensitivity toward para-NO-ASA. Of the tested samples, one patient featured several adverse prognostic markers, such as high levels of ZAP70 and CD38 (68.0% and 46.6%, respectively) and, most important, a 17p deletion. 17p deletions are rather rare (about 7%) and are associated with short survival, quick disease progression (27) and an extensive drug resistance (28). With this background, it is especially interesting that this sample was highly sensitive toward para-NO-ASA with an LC₅₀ value of 6.31 μmol/L that is below the average LC₅₀ value of all tested samples (8.72 ± 0.04 μmol/L).

An interesting fact is that other cancers that respond to NO-ASA treatment are known to exhibit an aberrantly active β-catenin/Tcf/Lef-1 complex, such as colon cancer (22), pancreatic cancer (29), or leukemic Jurkat cells (23). Studies in these neoplasias frequently showed reduced transcriptional activity of the β-catenin/Tcf/Lef-1 complex after NO-ASA treatment. Also in CLL, β-catenin/Lef-1 signaling is known to be highly active and thought to play a significant role in the pathogenesis of the disease (4). We demonstrated cleavage of β-catenin and a clear decrease of specific β-catenin/Tcf/Lef-1 targets, such as Cyclin D1, C-myc, and Lef-1, after treatment with para-NO-ASA. As pretreatment with a pan-caspase inhibitor largely reduced the percentage of cell death, downregulation of antiapoptotic proteins and β-catenin cleavage, we conclude that β-catenin cleavage and subsequent inactivation is rather a consequence than a cause of apoptosis induction mediated by para-NO-ASA treatment. This finding is in line with other reports showing that β-catenin is cleaved by caspase 3, consequently resulting in a decrease in transactivation of Tcf/Lef-1–mediated transcription (30). At this point, we speculate that para-NO-ASA initiates the induction of apoptosis that is accompanied by downregulation of antiapoptotic proteins, such as Bcl-2 and XIAP, resulting in caspase 3 activation. Caspase 3 in turn cleaves β-catenin, subsequently reducing transactivation of the transcription factors Tcf/Lef-1 and diminishing their target gene expression. This might prevent the cells from antagonizing the progress of apoptotic events through deregulation of cell survival and proliferation via expression of survival-supporting target genes. β-catenin has been implicated in the regulation of apoptosis in several studies. For example, in rat hippocampal neurons, overexpression of a truncated β-catenin cleavage product accelerated apoptosis in these cells (31). Also, dominant negative Tcf was shown to induce apoptosis, indicating a substantial role of β-catenin/Tcf/Lef-1 signaling in induction, rather than simply being an effect of apoptosis.

Pharmacokinetic studies in rats showed that plasma peak concentrations (C_max) of NO-ASA are reached 6 hours after oral administration. In addition, after enzymatic metabolism, the salicylic acid and the NO species were described to show similar patterns and are therefore likely to exert their functions simultaneously (32). In humans, the time for NO-ASA to reach its C_max (T_max) was shown to be 5.4 ± 1.2 hours. In rats, plasma peak levels of 230 μmol/L (77.2 μg/mL) were achieved after oral administration of a single dose of 65 mg/kg of bodyweight without detectable side effects (33). Therefore, plasma concentrations of 5 μmol/L that resemble the LC₅₀ value calculated from our experiments should be easily achievable. It is also important that potential side effects, such as lowering of blood pressure due to the vasodilatory effect of NO, could not be detected even after administering high doses of NO-ASA (34). However, it has to be mentioned that many pharmacologic data concerning NO-ASA were obtained from studies with the meta-isomer and further data concerning the para-isomer still have to be determined.

To investigate the efficacy of NO-ASA in vivo, we utilized a CLL-like JVM3 xenograft mouse model. The JVM3 cell line, although originally derived from a B-prolymphocytic leukemia (PLL) patient, is a commonly utilized CLL cell line model. Furthermore, JVM3 cells exhibit increased protein levels of both Lef-1 and β-catenin compared with healthy B cells (data not shown). In addition, JVM3 cells were similarly sensitive to para-NO-ASA treatment compared with primary CLL cells (data not shown). During the long-term administration of 22 days p.o., para-NO-ASA inhibited tumor growth in the JVM3 xenograft model significantly as early as 9 days. This effect became more pronounced with ongoing treatment. We did not observe an actual tumor reduction in para-NO-ASA–treated animals possibly due to a protective effect of the microenvironment. The incomparableness of the microenvironment of the leukemic JVM3 cells in the xenograft model (solid, subcutaneous, immune-compromised) and the actual leukemia microenvironment (peripheral blood, bone marrow, lymphoid organs) is a limitation of this study. In general, NO-ASA was well tolerated and revealed no obvious toxicities.

In conclusion, para-NO-ASA is effective in reducing cell survival via induction of apoptosis and subsequent inhibition of β-catenin–mediated gene expression in CLL cells in vitro and could inhibit tumor growth in vivo. Further studies might clarify the exact mechanism of action. Given the high common acceptance of ASA and its derivates as well as the oral availability, para-NO-ASA is a promising compound for further evaluation in the treatment of CLL, possibly even for extensively therapy-resistant 17p-deleted cases.

No potential conflicts of interest were disclosed.

We thank Samir Tawadros for superior assistance with animal experiments.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.