Leukemia (2001) 15, 1852–1859  2001 Nature Publishing Group All rights reserved 0887-6924/01 $15.00 www.nature.com/leu Nitric oxide enhancement of fludarabine cytotoxicity for B-CLL lymphocytes DJ Adams1, MC Levesque2, JB Weinberg2, KL Smith1, JL Flowers1, J Moore1, OM Colvin1 and R Silber1,✠

1Department of Medicine, Duke University Medical Center, Durham, NC, USA; and 2Veterans Administration Medical Center, Durham, NC, USA

Fludarabine is active but not curative in the treatment of antibodies5 have been used to treat advanced or recurrent chronic lymphocytic (B-CLL). Nitric oxide (NO) sup- disease. plied from exogenous, NO-donating pro-drugs can also induce B-CLL cells generally manifest decreased apoptosis and apoptosis and death of acute leukemia cells. This study investi- 6–8 gated combinations of fludarabine with NO-donating pro-drugs resistance to apoptosis. This is likely mediated in part by for their cytotoxicity against freshly isolated B-CLL lympho- increased expression of bcl-26 and by chronic activation of cytes following a 72 h exposure in vitro. The median IC50 for endogenous inducible NO synthase (NOS2) and intracellular fludarabine was 2.2 ␮M (n = 85). The nitric oxide donors DETA- NO,7–9 a molecule that can inhibit apoptosis by nitrosylating NO, PAPA-NO, and MAHMA-NO were also cytotoxic, and their caspases.10–12 In contrast to low-level, endogenous NO, effects were inversely related to rates of NO release. Neither exogenous NO can induce apoptosis of normal and malignant DETA-NO depleted of NO nor DETA itself was effective, indicat- 13 ing that NO was required for cytotoxicity. Drug interactions cells, including leukemia cells. were evaluated by a modified combination index method. Syn- The purpose of this study was to determine the effect of ergy was observed in combinations of fludarabine or nelara- combining chemotherapeutic agents that target DNA metab- bine (506U78) with DETA-NO in 52% and 88% of samples, olism with NO-donating pro-drugs on B-CLL cells in vitro.We respectively. Interestingly, the combination of fludarabine and demonstrate that combinations of fludarabine with the NO DETA-NO was more cytotoxic in B-CLL cells less sensitive to fludarabine. DETA-NO did not enhance the activity of other donor, DETA-NO exhibit synergistic cytotoxicity for B-CLL DNA anti-metabolites, topoisomerase I and II inhibitors, or lymphocytes in a leukemia and drug-selective manner. More- alkylating agents. Finally, the anti-leukemic activity of fludara- over, B-CLL cells that exhibit emerging resistance to fludarab- bine alone or in combination with DETA-NO was found to corre- ine alone remain sensitive to the combination of fludarabine late with inhibition of cellular RNA synthesis. These results with DETA-NO. indicate that NO donors could enhance fludarabine therapy for B-CLL. Leukemia (2001) 15, 1852–1859. Keywords: chronic lymphocytic leukemia; nitric oxide; fludarabine; combination ; apoptosis Materials and methods Chemotherapeutic agents Introduction The chemotherapeutic agents used in this study were obtained B cell chronic lymphocytic leukemia (B-CLL) is the most com- from the following sources: fludarabine, Berlex Laboratories, mon adult leukemia in the western hemisphere, accounting Richmond, CA, USA; and 5-fluorouracil, Pharma- for a quarter of all (reviewed in Ref. 1). While cia, Kalamazoo, MI, USA; , Lilly, Indianapolis, IN, patients with early stage, indolent disease survive a median USA; SN-38, Research Triangle Institute, Research Triangle ␤ d 11 years, patients with advanced stage B-CLL have a median Park, NC, USA. (506U78; 2-amino-9- - -6- survival of 1.5–3 years with few therapeutic options that methoxy ara-guanine) was a gift from Dr Joanne Kurtzberg, impact overall survival. Furthermore, all B-CLL patients have Department of Pediatrics, Duke University Medical Center, impaired immune function; thus the natural course is charac- Durham, NC, USA. The diazenium diolate NO pro-drugs terized by increased infections, secondary malignancies, and were purchased from Cayman Chemical, Ann Arbor, MI, USA. autoimmune complications. Therefore, newtherapeutic Interleukin-4 (IL-4) and diethylene triamine (DETA) were strategies are needed for patients with B-CLL. purchased from Sigma Chemical, St Louis, MO, USA. Current treatment for B-CLL depends on disease stage. Indolent disease may not benefit from any treatment.1 Fludara- bine (9-␤-d-arabinofuranosyl-2-fluoroadenine-5Ј-monophos- Isolation of B-CLL lymphocytes phate; F-ara-AMP) is nowconsidered by some as an accept- able first-line therapy for B-CLL, since a direct comparison B-CLL patients from the Veteran’s Affairs (VA) Medical Center with demonstrated significant advantages in and Duke University Medical Center were enrolled in this both response rates and disease-free survival.2 However, study under a protocol approved by the Duke University and despite a response rate of 50–70% in previously untreated VA Institutional ReviewBoards. Blood wascollected into hep- patients, all patients eventually relapse with fludarabine- arinized Vacutainer CPT mononuclear cell preparation tubes refractory disease, and to date there is no standard salvage (Becton-Dickinson Vacutainer Systems, Franklin Lakes, NJ, therapy. Regimens combining alkylators with the topoisomer- USA), and centrifuged at 1500 g for 15 min at room tempera- ase II inhibitor, doxorubicin,3,4 and anti-CD52 monoclonal ture. The mononuclear cell layer was harvested, washed, and cultured overnight in RPMI 1640 growth medium (Gibco, Grand Island, NY, USA), supplemented with 10% heat-inacti- vated fetal bovine serum (HyClone, Logan, UT, USA), 1% Correspondence: DJ Adams, Duke Comprehensive Cancer Center, penicillin-streptomycin-amphotericin B (Sigma Chemical), DUMC 3843, 303 MSRB, Research Drive, Durham, NC, 27710, USA; Fax: 919-684-5653 and IL-4 (5 ng/ml) in culture flasks to remove ✠Deceased monocytes/macrophages. After overnight culture, T cells were Received 4 May 2001; accepted 20 July 2001 removed with magnetic immunobeads (Dynabeads, Dynal, Nitric oxide and fludarabine in B-CLL DJ Adams et al 1853 Princeton, NJ, USA) for samples with white blood cell counts Annexin V apoptosis assay less than 50 000/␮l. Residual red blood cells were then removed by centrifugation over Ficoll–Paque Plus (Pharmacia, The annexinV/propidium iodide assay (R&D Systems, Minnea- Uppsala, Sweden). polis, MN, USA) was used according to the manufacturer’s protocol to assess drug effects on B-CLL cell apoptosis. Briefly, B-CLL cells (5 × 106 in 4 ml) were cultured in six-well cluster Metabolic inhibition assay dishes in the presence or absence of study agents for various times. Cells were then washed and resuspended in Dulbecco’s The MTS cell proliferation assay (Celltiter 96 AQ, Promega, phosphate-buffered saline, incubated with the respective flu- Madison, WI, USA) was used to assess drug effects on cell orescent probes and analyzed by flowcytometry at 530 nm viability. Cells were plated at 2.5 × 105 cells/well in 96-well (annexin V FITC) and 585 nm (PI) on a FACScan flowcyto- microtiter plates (Corning Costar, Acton, VT, USA). Triplicate meter with Cell Quest software (Becton Dickinsen). Results wells were exposed to varying concentrations of agents for were expressed as the percentage of cells that were viable 72 h at 37°C. MTS reagent (20 ␮l) was then added to each (AV−/PI−), apoptotic (AV+/PI−), or dead (AV+/PI+). well for the final 3 h of culture, and the absorbance at 490 nm

(A490) was quantitated in an absorbance plate reader (Bio-Tek, Winooski, VT, USA). Response was defined as (mean drug- Inhibition of RNA synthesis − − × treated A490 blank)/(mean untreated control A490 blank) 100. The data were curve fit by weighted non-linear Drug effects on RNA synthesis were evaluated by measure- regression to determine the IC50, defined as the drug concen- ment of cellular incorporation of [3H]uridine.16 Briefly, B-CLL tration that inhibited MTS metabolism to 50% of untreated cells were plated in 96-well microplates at 2.5 × 105 cells/well controls. in a final volume of 200 ␮l. Twenty-four hours prior to the end of drug exposure, 0.2 ␮Ci of [3H]uridine (NewEngland Nuclear, Wilmington, DE, USA) were added. The assay was Assessment of drug interaction terminated by transferring the radiolabeled cells to glass fiber multiscreen suction plates (Millipore Corporation, Bedford, Drug interaction was assessed by a modification of the combi- MA, USA) that had been pre-wet with 0.1% Tween 20 in dis- nation index method of Chou and Talalay.14 Agents were tilled water. The samples were suctioned, lysed and washed combined in a ratio equal to the ratio of their individual IC50s with 0.1% Tween 20, washed with 95% ethanol, and suc- and evaluated in the MTS assay. The combination index, CI, tioned dry. After addition of 25 ␮l MicroScint 20 (Packard was assessed at the IC50 level of effect and defined as: Instruments, Downers Grove, IL, USA), retained radioactivity was quantified on a TopCount microplate scintillation counter IC of agent A in combination CI = 50 (Packard). Results were expressed as percent of untreated IC50 of A alone control c.p.m. incorporated. + IC50 of agent B in combination IC50 of B alone An additive interaction is then defined as Cl = 1, antagonism Results as CI Ͼ 1, and synergy as CI Ͻ 1. Since the mean 95% confi- dence interval for all interactions was 0.135, a CI Ͻ 0.6 was Cytotoxicity of fludarabine and nitric oxide against used to define a statistically significant level of synergy. cultured B-CLL lymphocytes An extension of the Chou and Talalay approach was also used to assess drug interaction. The three-dimentional model We sought to identify drug combinations containing fludara- of Kanzawa et al15 computes a combination effect (CE) surface bine that exhibit synergistic cytotoxicity toward B-CLL lym- that identifies the range of each drug concentration that yields phocytes. In particular, we wished to examine combinations either additive (CE = 0), greater than additive (CE Ͼ 0), or less with NO-donating agents and other agents known to affect than additive (CE Ͻ 0) interactions. This model provides a glo- nucleic acid metabolism. To this end, peripheral blood lym- bal viewof drug interaction that incorporates multiple drug phocytes were isolated from whole blood donated by 28 con- ratios and levels of effect in a single graphic, which in this senting B-CLL patients (median age 63, range 43–84 years). study is a color-coded contour map. The contour map rep- Twenty-one of 28 (75%) were male, and 9/28 (31%) were pre- resents the combination effect surface that is generated by viously untreated. Pilot studies aimed at optimizing culture subtracting a surface of theoretical additivity (TA), from an conditions revealed a significant amount of spontaneous observed response surface. Theoretical additivity is defined by apoptosis after 72 h incubation in RPMI growth medium. = + − the equation TA (fa)A (fa)B (fa)A(fa)B, where (fa)A and (fa)B Since this effect could cloud interpretation of chemosensitivity are the fractions of cells affected by drugs A and B, respect- assays, we examined various medium supplements to improve ively, and where the two drugs have mechanisms of action viability. We found that addition of interleukin 4 was sufficient that are not independent. In practice, data in an 8 × 8 well to maintain B-CLL viability over the required assay period and matrix format from six replicate plates were entered into a did not significantly affect drug interactions described below Microsoft Excel (Redmond, WA, USA) spreadsheet where the (data not shown). The cytotoxicity of fludarabine was then combination effect surface was computed and the result con- compared to that of other DNA anti-metabolites, the topoiso- verted to XYZ format. Contour plots were then produced using merase I inhibitor SN-38 (the active metabolite of the clinical SigmaPlot 2000 (SPSS, Chicago, IL, USA) scientific graphics agent, ), the topoisomerase II inhibitor doxorubicin, software. A contour plot of the respective fractions affected and the alkylating agent chlorambucil. Following drug for the combination was also constructed to correlate drug exposure for 3 days, the viability of B-CLL cells was determ- interaction with efficacy. ined with the MTS assay. The results are summarized in

Leukemia Nitric oxide and fludarabine in B-CLL DJ Adams et al 1854 Table 1. The IC50 of fludarabine was nearly a log greater than either SN-38 or doxorubicin, but was seven to 400 times less than other DNA anti-metabolities, and chlorambucil, which is the traditional first-line therapy in B-CLL. Based on observations that NO induces apoptosis and death of acute nonlymphocytic leukemia (ANLL) cells in vitro,13 we evaluated the ability of exogenous NO to kill B-CLL cells. Specifically, we examined 1-substituted diazen-1-ium-1,2- diolates as NO donors in which NO is released at known rates in vitro.17 The NO prodrugs were cytotoxic for B-CLL cells at relatively lowconcentrations. For example, the compound DETA-NO ((Z)-1[-2-aminoethyl]-N-(2-ammonioethyl)amino] diazen-1-ium-1,2-diolate), which releases half its NO in 20 h ° ␮ at 37 C, had an average IC50 of approximately 300 m in B- CLL specimens (Table 1). This represents an exposure rate of 0.25 ␮m NO/min, which is well within the steady state con- centration noted in vivo (0.01–1 ␮m).18–20 Moreover, DETA- NO achieved a higher level of cell kill than typically observed with chemotherapeutic agents in this assay. Cytotoxicity was due specifically to released NO, because DETA alone and DETA-NO that had been extensively pre-incubated prior to treatment were both inactive in six separate specimens (data not shown). To determine whether NO toxicity was a function of NO release rate, we evaluated the diazenium diolate analogues, MAHMA-NO and PAPA-NO, which have release half times of 1 and 15 min, respectively. The results shown in Figure 1a indicate that there is an inverse log relationship between NO release rate and in vitro cytotoxicity. Thus, exposure time is a critical factor in nitric oxide-mediated toxicity toward B-CLL Figure 1 Effect of nitric oxide release rate on cytotoxicity and on interaction with fludarabine in B-CLL cells. (a) Cytotoxicity of nitric cells. These results are consistent with those reported by oxide releasing agents was assessed by the MTS metabolic assay on 13 Shami et al, who found a similar correlation of release rate purified B-CLL lymphocytes from patient specimens as described in

to cytotoxicity for these NO donors when assayed against materials and methods. Results represent the interassay mean IC50 vs HL60 and U937 cell lines, representing acute myeloblastic release half time for MAHMA-NO (circle, n = 8), PAPA-NO (inverse and acute monoblastic lineages, respectively. triangle, n = 8), and DETA-NO (square, n = 32). Bars, standard error; line, linear regression fit of the data, r2 = 0.98. (b) Interaction of NO

releasing agents with fludarabine phosphate was assessed at the IC50 by the combination index method for MAHMA-NO (circles, n = 7), Interactions of fludarabine and nitric oxide PAPA-NO (inverse triangles, n = 7), and DETA-NO (squares, n = 29). A combination index Ͻ1 indicates synergy; CI = 1, additivity; CI Ͼ1, Next, we evaluated binary combinations of fludarabine using antagonism. Bars, standard error. the combination index method of Chou and Talalay.14 Agents were evaluated by simultaneous exposure and combined at

the ratio of their IC50s to minimize the number of leukemia below). An example of the interaction of fludarabine with cells required and to facilitate data analysis. Similarly, the DETA-NO is shown in Figure 2. The median effect plots for

combination index endpoint was also defined at the IC50 each agent alone and in combination are given in panel(a). which may not reflect optimal drug interaction conditions (see Because none of these lines is parallel, exclusivity of the inter- action cannot be determined.14 The resulting combination index plot in panel (b) describes an interaction that can be Table 1 Chemosensitivity of B-CLL cells in vitro antagonistic, additive or synergistic depending on the drug concentrations and related level of effect. At the IC (fraction Agent IC (␮M) Relative 50 50 = activitya affected 0.5), the interaction is moderately synergistic (ie = n Range Mean Median combination index 0.64), while at higher levels of effect, the interaction becomes increasingly more synergistic. SN-38 71 0.0–2.5 0.4 ± 0.1 0.3 1 The sequence of agent exposure can also be an important Doxorubicin 67 0.1–2.7 0.6 ± 0.1 0.4 1 parameter in drug interactions. In B-CLL cell specimens from Fludarabine 85 0.2–43.2 4.7 ± 0.7 2.2 7 three patients, sequential exposure to fludarabine and DETA- Gemcitabine 11 3–51 16 ± 41447 NO in either sequence after a 24 h interval produced a similar 506U78 11 7–158 33 ± 13 20 67 interaction when compared to simultaneous administration ± Chlorambucil 78 2–214 32 42273 (data not shown). This result suggests that the effect of NO DETA-NO 32 103–1219 274 ± 37 209 697 PAPA-NO 9 379–999 678 ± 78 646 2153 was not due to increased dephosphorylation of fludarabine 5- 17 116–1915 810 ± 112 848 2827 to its nucleoside form (F-ara-A) prior to drug uptake by cells. MAHMA-NO 9 603–2711 1802 ± 229 2022 6740 Accordingly, when the activity of fludarabine was compared directly to that of F-ara-A in 10 B-CLL patient samples, the a = The IC50 of an individual agent divided by the IC50 for SN-38, the results were equivalent (mean IC50 1.74 vs 1.68, most active agent tested. respectively). In addition, the interaction of F-ara-A with

Leukemia Nitric oxide and fludarabine in B-CLL DJ Adams et al 1855

Figure 3 Interaction of fludarabine and DETA-NO as a function of fludarabine sensitivity. Combination indices from 56 patient samples (circles) were plotted as a function of the sensitivity to fludarabine phosphate alone. The data was then fit to the hyperbolic function y = a/(x)n, where a = 0.88, n = 0.36. Dotted lines, 95% confidence intervals.

topoisomerase I and II inhibitors, and with chlorambucil, the traditional first-line agent in B-CLL. The results summarized in panel (a) of Figure 4 reveal a notable selectivity for the interac- tion of fludarabine and DETA-NO. While synergy was not spe- cific for fludarabine, the ara-G prodrug nelarabine was the only DNA to exhibit a comparable interaction. Figure 2 Interaction of fludarabine and DETA-NO in B-CLL lym- Almost 90% of the specimens treated with nelarabine + DETA- phocytes. (a) Median effect plots for the interaction of fludarabine and NO had a combination index Ͻ0.6, compared to 52% of DETA-NO were constructed with MTS cytotoxicity data from a B-CLL + patient specimen. The data and respective linear regression lines rep- samples exposed to fludarabine DETA-NO. In contrast, resent the effects of DETA-NO alone (squares, r2 = 0.99), fludarabine interactions with the other agents tested were additive and phosphate alone (circles, r2 = 0.94), and the combination (inverse tri- not synergistic. angles, r2 = 0.96). (b) The combination index (circles) was computed To judge the relative potency of the fludarabine–DETA-NO at increasing levels of effect for the specimen above where CI Ͻ1 combination, we examined other drug combinations in this = Ͼ indicates synergy; CI 1, additivity; CI 1, antagonism. series by the same method. Figure 4b demonstrates that the degree of interaction between fludarabine and DETA-NO is similar to that observed for the combination of fludarabine DETA-NO was similar (mean ratio of combination indices at with the topoisomerase II inhibitor doxorubicin (51% samples + + = ± the IC50 for (F DETA-NO)/(F-ara-A DETA-NO) 0.89 with CI Ͻ0.6). Likewise, combination of fludarabine with the 0.13). Of note, there was no significant effect of NO release topoisomerase I inhibitor SN-38 (the active metabolite of CPT- time on the interaction of fludarabine and NO on B-CLL 11) produced this level of synergy in 42% of specimens, while samples (Figure 1b). combinations of SN-38 with chlorambucil or doxorubicin A different perspective on the interaction of fludarabine and were less effective (22% and 12% specimens with CI Ͻ0.6, DETA-NO is shown in Figure 3, where the combination index respectively). (CI) is plotted as a function of sensitivity to fludarabine. The data exhibit a hyperbolic relationship that can be fit by the equation: Three-dimensional analysis of the fludarabine 0.88 interaction with DETA-NO CI at IC + = . 50 FLU DETA-NO (IC )0.36 50 FLU To obtain a more global perspective on the interaction of flu- This result suggests that NO may preferentially sensitize leuke- darabine with DETA-NO, we applied the three-dimensional mia cells to fludarabine that otherwise express emerging modeling approach of Kanzawa et al15 to four separate speci- resistance to fludarabine alone. mens from a representative patient. The resulting combination effect (CE) surface (shown as a contour plot in Figure 5a) revealed that greater than additive interaction (CE Ͼ 0) Specificity of the fludarabine–NO interaction occurred at concentration ranges of 30–150 ␮m for DETA-NO and 0.6–50 ␮m for fludarabine (dark blue area). This corre- To determine whether the synergistic interaction between sponds to 60–90% leukemia cell kill vs untreated controls fludarabine and NO was drug-specific, we evaluated combi- (Figure 5b). At concentrations above 200 ␮m DETA-NO in nations of DETA-NO with other DNA , with combination, the synergistic interaction of fludarabine and

Leukemia Nitric oxide and fludarabine in B-CLL DJ Adams et al 1856

Figure 5 Three-dimensional combination effect surface for the Figure 4 Selectivity of drug interactions with fludarabine. (a) In interaction of fludarabine and DETA-NO in B-CLL lymphocytes. The addition to fludarabine (FLU), the nitric oxide donor, DETA-NO was three-dimensional method of Kanzawa et al15,48 was used to obtain a combined with the DNA anti-metabolites nelarabine (506U78), gem- global viewof the interaction of fludarabine phosphate and DETA- citabine (GEM), and 5-flurouracil (5-Fura); the topoisomerase I inhibi- NO in a representative patient. (a) The mean combination effect sur- tor SN-38, the topoisomerase II inhibitor doxorubicin (DOX), and the face from four separate determinations reveals a continuum of alkylating agent chlorambucil (CHL). Bars represent the combination response from highly antagonistic (light) to highly synergistic (dark). Ͼ index at the IC50 level of effect; error bars are the interassay standard Superimposed white contour lines indicate regions of synergy ( 1) or error of the mean. The fraction within each bar indicates the number antagonism (Ͻ−1) that are significant at the 95% confidence level. of samples in which the combination index was less than or equal to The white star indicates the interaction at drug concentrations that

0.6 out of the total specimens tested. (b) Drug interactions were evalu- produce an IC50 level of effect for each agent alone (standard two- ated for combinations of the clinical agents with each other under the dimensional format). (b) Corresponding efficacy plot for (a) showing

same exposure conditions. the percent of B-CLL lymphocytes affected (Fa), ranging from minimal (dark) to maximal (light). Optimum synergy corresponds to a region of 80–90% cell kill. DETA-NO rapidly converted to additivity, and then to a region of antagonism (red area). While interpretation is clearly lim- ited to analyses of multiple samples from a single individual, fold increase in apoptosis with the combination of fludarabine this result indicates that evaluation of combinations at a single and DETA-NO at 24 h, and that this effect is eventually trans-

drug ratio (such as the IC50 of each agent) may not reveal lated into an increase in dead cells at 48 and 72 h compared optimal interaction, and that concentrations of fludarabine to either agent alone. In contrast, there was no significant loss and DETA-NO must be carefully controlled to maximize a of viability in untreated B-CLL cells over this time frame. synergistic anti-leukemia effect.

Mechanisms underlying the effects of fludarabine and Effect of fludarabine and DETA-NO on B-CLL DETA-NO in B-CLL apoptosis A recent report demonstrates that the mechanism of fludarab- Both fludarabine and NO are known to cause cell death via ine action in B-CLL involves inhibition of RNA synthesis as apoptosis, which is a primary pathway for cell death induced reflected by cellular [3H]uridine incorporation.23 Conse- by cytotoxic agents in hematological malignancies.21,22 quently, we determined whether there was a correlation Because the observation of a synergistic interaction between between cytotoxicity measured by inhibition of MTS metab- fludarabine and DETA-NO was made using a metabolic assay, olism and cytotoxicity measured by inhibition of [3H]uridine which can underestimate the number of apoptotic cells (data incorporation in B-CLL cells treated with fludarabine in the not shown), we performed B-CLL cell annexin-V apoptosis presence and absence of DETA-NO. In assays on three separ- assays. Figure 6 summarizes results from five patient speci- ate specimens, there was a high correlation between these two mens. The time course reveals that there is a significant two- endpoints of cytotoxicity (Table 2). When the combination

Leukemia Nitric oxide and fludarabine in B-CLL DJ Adams et al 1857 Discussion

Fludarabine is very active as salvage therapy for B-CLL, and some nowconsider it first-line therapy for B-CLL. The drug also has activity in other indolent lymphoid malignancies. Paradoxically, fludarabine (unlike the pyrimidine nucleoside analog gemcitabine) does not have comparable activity against solid tumors.24 Fludarabine is also an important part- ner for combination chemotherapy. Consistent with our results, Kano and colleagues25 found additive effects of fluda- rabine with doxorubicin in a B-CLL cell line. Additive effects were also seen with , 4-hydroperoxy-cyclophos- phamide and hydroxyurea, while synergy was observed with in vitro. Accordingly, fludarabine has demon- strated clinical activity in combination chemotherapy regi- mens with , , mitoxantrone + /prednisone, and .24,26 Inhibition of DNA replication via incorporation of F-ara- ATP and subsequent chain termination is thought to be the primary mechanism of fludarabine action in replicating cells.27 Several related molecular targets have been identified. These include , DNA polymerases ␣, ␤, ␥, and ⑀, DNA primase, and DNA ligase.27 Fludarabine inhibits DNA elongation, and once it is incorporated into nucleic acid, it is a poor substrate for DNA excision repair. In addition, fludarabine directly inhibits DNA repair enzymes.28 The effects of fludarabine on DNA replication are not likely the mechanism of cytotoxicity in these non-cycling cells. Four Figure 6 Induction of apoptosis by the combination of fludarabine other actions of fludarabine have been reported that could with DETA-NO in B-CLL lymphocytes. Drug interaction was evalu- account for anti-leukemia activity in B-CLL lymphocytes. (1) ated by annexin V/flowcytometric assay on five patient samples as F-ara-ATP is incorporated into DNA of quiescent cells by described in materials and methods. Viable cells (top panel) were dis- repair synthesis.29 The resulting chain termination and free criminated from apoptotic cells (middle panel) and dead cells (bottom DNA ends may trigger apoptosis. (2) Unlike ara-C and ara-A, panel) over the 72 h standard exposure time course. Results are fludarabine inhibits RNA synthesis due to preferential incor- presented for untreated controls (open bars), cells treated with fludar- abine phosphate alone (upward hatched bars), DETA-NO alone poration of F-ara-A into messenger RNA with resulting prema- 30 (downward hatched bars), and the simultaneous combination (filled ture termination of the transcript. Other DNA-directed drugs bars). Error bars are the interassay standard error of the mean. have also been found to target RNA in B-CLL. For example, Cohen et al16 reported a strong correlation between cytotoxic- ity and inhibition of RNA synthesis in B-CLL patient samples Table 2 Correlation of cytotoxicity to inhibition of RNA synthesis treated in vitro with the topoisomerase I inhibitor SN-38. in B-CLL lymphocytes Results reported here are consistent with this mechanism of cytotoxicity. While general inhibition of RNA synthesis is an Sample Cytotoxicitya Drug interactionb important mechanism, loss of specific transcripts might also be critical. Huang and colleagues23 observed loss of a 50 kDa n Pearson rc P value n Pearson rc P value protein in B-CLL cells after fludarabine treatment. (3) Another mechanism for fludarabine cytotoxicity that is independent of A 24 0.921 Ͻ0.0001 either DNA or RNA synthesis has been reported: direct acti- B 64 0.904 Ͻ0.0001 49 0.433 0.0019 vation of apoptosis. In a reconstituted, cell-free system, highly Ͻ Ͻ C 64 0.925 0.0001 49 0.766 0.0001 purified cytochrome-c was incubated with recombinant apoptosis protein activating factor 1 (Apaf-1) and recombinant a 3 Cytotoxicity was assessed by MTS assay or by [ H]uridine incor- procaspase-3 and -9.31 Addition of F-ara-ATP had a catalytic poration for n of number of drug combinations. The endpoints were percent of control absorbance at 490 nm (MTS assay) or percent efficiency almost 50 times that of the natural substrate, deoxy- of control CPM (uridine incorporation assay) ATP for causing cleavage of these pro-caspases to their active bDrug interaction was assessed by the combination index for n forms, which subsequently initiate apoptosis. (4) Finally, Frank number of drug combinations based on inhibition of metabolism vs et al32 have shown that fludarabine specifically depletes nor- inhibition of RNA synthesis. mal resting or activated human lymphocytes of the protein c Pearson pairwise correlation coefficient. and mRNA for STAT1, which is constitutively phosphorylated in B-CLL cells and may promote B-CLL cell viability.33 NO is also an important regulator of apoptosis in B-CLL index for drug interaction assessed by mitochondrial metab- lymphocytes. NO appears to be a double-edged sword olism (MTS assay) was compared to that evaluated by RNA depending on the redox state of the cell and whether cell synthesis (uridine incorporation), the correlation was weaker, exposure is endogenous or exogenous.8,9,34,35 For example, in yet still statistically significant. This suggests that inhibition of B-CLL lymphocytes, NOS2 can be increased 10- to 15-fold,34 RNA synthesis contributes to, but may not be the sole leading to elevated endogenous NO that exerts an anti-apop- mechanism of interaction between fludarabine and NO. totic effect.9 With brief exposures, exogenous NO can delay

Leukemia Nitric oxide and fludarabine in B-CLL DJ Adams et al 1858 apoptosis in cultured, mature murine B cells, an effect that is be achieved. Finally, combination of fludarabine and nitric correlated to stabilization of both the mRNA and protein lev- oxide could enhance drug activity in solid tumors, as well as els of the anti-apoptotic factor, bcl-2.8 Endogenous NO can in fludarabine-resistant leukemias. inhibit Fas-induced apoptosis in human B cell, T cell and monocytic cell lines36 probably via S-nitrosylation of the active site thiols of cysteine proteases, including the caspase Acknowledgements family. Finally, NOS2 inhibitors can induce apoptosis in 9,34 B-CLL cells. This work was supported in part by a grant from Berlex Lab- On the other hand, high level NO production is an oratories, the VA Research Service, the Leukemia and Lym- important component of host immune defense against neo- phoma Society of America, and NIH grants AR-39162 and Al- plasia, since NO is a prominent mediator of macrophage- 37 41764. This manuscript is dedicated to the memory of Dr mediated tumor cell cytotoxicity. Potential mechanisms of Robert Silber, who initiated the work and inspired the research toxicity include N-nitrosation of DNA and covalent modifi- 38 team. The authors are indebted to Dr Fumihiko Kanzawa for cation of tyrosine residues, inhibition of mitochondrial res- his guidance on application of the three-dimensional drug piration and inhibition of the glycolytic enzymes aconi- interaction model. tase39,40 and glyceraldehyde-3-phosphate dehydrogenase.41,42 Moreover, metal- and thiol-containing proteins are also major NO targets, and include DNA ligase43 and ribonucleotide reductase,38 which are also fludarabine targets and could References account for the observed synergy between NO donors and 1 Wierda WG, Kipps TJ. Chronic lymphocytic leukemia. Curr Opin fludarabine in B-CLL lymphocytes. Therefore, whether NO is Hematol 1999; 6: 253–261. an agent of life or death for cells appears to depend on the 2 Rai KR, Peterson B, Elias L, Shepherd L, Hines J, Nelson D, Cheson source, concentration and flux of NO, the type of cell and its B, Kolitz J, Schiffer CA. A randomized comparison of fludarabine redox state and the presence of competing NO scavengers and and chlorambucil for patients with previously untreated chronic NOS inhibitors. lymphocytic leukemia. A CALGB, SWOG, CTG/NCI-C and ECOG Despite the multiple cellular pathways regulated by nitric inter-group study. Blood 1996; 88: 552–552. 3 Jaksic B, Brugiatelli M, Krc I, Losonczi H, Holowiecki J, Planinc- oxide, surprisingly little is known about the interaction of this Peraica A, Kusec R, Morabito F, Iacopino P, Lutz D. High dose potent molecule with anti-cancer chemotherapeutic agents. chlorambucil vs Binet’s modified cyclophosphamide, doxorub- Wink and colleagues evaluated combinations of NO donors icin, , and prednisone regimen in the treatment of with the alkylating agents BCNU,28 ,44 and mel- patients with advanced B-cell chronic lymphocytic leukemia. phalan45 by clonogenic assay in replicating V79 Chinese ham- Results of an international multicenter randomized trial. Inter- ster lung fibroblasts and MCF-7 human breast cancer cells. national Society for Chemo-Immunotherapy, Vienna. Cancer 1997; 79: 2107–2114. In each case, NO released by the diazenium diolate donor, 4 Johnson S, Smith AG, Loffler H, Osby E, Juliusson G, Emmerich DEA/NO significantly enhanced cytotoxicity of the alkylating B, Wyld PJ, Hiddemann W. Multicentre prospective randomised agent, possibly by inhibition of DNA repair. However, trial of fludarabine versus vs cyclophosphamide, doxorubicin, and DEA/NO had cytotoxic activity alone, and an interaction prednisone (CAP) for advanced-stage chronic lymphocytic leu- endpoint analysis (eg the combination index we used here) kaemia. The French Group on CLL. Lancet 1996; 347: 1432–1438. was not determined. Hence, direct comparison of the results 5 Osterborg A, Dyer MJ, Bunjes D, Pangalis GA, Bastion Y, Catovsky D, Mellstedt H. Phase II multicenter study of human CD52 anti- of Wink et al with our results is not possible. In the case of body in previously treated chronic lymphocytic leukemia. Euro- quiescent B-CLL lymphocytes, we did not observe enhanced pean Study Group of CAMPATH-1H Treatment in Chronic Lym- activity of the alkylating agent chlorambucil with NO. Rather, phocytic Leukemia. J Clin Oncol 1997; 15: 1567–1574.

we found that the interaction of DETA-NO at its IC50 for B- 6 Jurlander J, Lai CF, Tan J, Chou CC, Geisler CH, Schriber J, Blu- CLL cells with the agents tested was highly selective for the menson LE, Narula SK, Baumann H, Caligiuri MA. Characteriz- DNA anti-metabolites, and that there were differences within ation of interleukin-10 receptor expression on B-cell chronic lym- phocytic leukemia cells. Blood 1997; 89: 4146–4152. this class of chemotherapeutic agent. The molecular basis for 7 Kim YM, Kim TH, Seol DW, Talanian RV, Billiar TR. Nitric oxide this selectivity is unknown, but it suggests the interaction of suppression of apoptosis occurs in association with an inhibition these agents could be clinically important. Indeed, in prelimi- of Bcl-2 cleavage and cytochrome c release. J Biol Chem 1998; nary studies, the combination of fludarabine with DETA-NO 273: 31437–31441. produced synergistic activity in cell culture models of human 8 Genaro AM, Hortelano S, Alvarez A, Martineza C, Bosca L. breast cancer, whereas fludarabine alone was virtually inac- Splenic B-lymphocyte programmed cell-death is prevented by nitric-oxide release through mechanisms involving sustained Bcl- tive at equivalent concentrations. Ultimately, the utility of 2 levels. J Clin Invest 1995; 95: 1884–1890. combinations of DNA anti-metabolites with NO donors will 9 Zhao H, Dugas N, Mathiot C, Delmer A, Dugas B, Sigaux F, Kolb depend on the specificity of the interaction for leukemia cells. JP. B-cell chronic lymphocytic leukemia cells express a functional Our preliminary data with normal peripheral blood mono- inducible nitric oxide synthase displaying anti-apoptotic activity. nuclear cells suggests that a therapeutic window of specificity Blood 1998; 92: 1031–1043. could exist. In four of five normal blood mononuclear cell 10 Mannick JB, Hausladen A, Liu LM, Hess DT, Zeng M, Miao QX, Kane LS, GowAJ, Stamler JS. Fas-induced caspase denitrosylation. samples, the interaction of fludarabine with DETA-NO was Science 1999; 284: 651–654. additive (CI = 1.0 ± 0.1), rather than synergistic. These 11 Li J, Billiar TR, Talanian RV, Kim YM. Nitric oxide reversibly samples contain only a small percentage of normal B cells, inhibits seven members of the caspase family via S-nitrosylation. and therefore do not represent a true normal control for B- Biochem Biophys Res Commun 1997; 240: 419–424. CLL. However, if NO-fludarabine combinations prove to be 12 Mohr S, Zech B, Lapetina EG, Brune B. Inhibition of caspase-3 by selectively cytotoxic for leukemia cells, such a combination S-nitrosation and oxidation caused by nitric oxide. Biochem Biophys Res Commun 1997; 238: 387–391. would be very useful clinically. With advances in develop- 13 Shami PJ, Sauls DL, Weinberg JB. Schedule and concentration- ment of NO pro-drugs such as those targeted to specific dependent induction of apoptosis in leukemia cells by nitric oxide. tissues and cell types,46,47 a very high therapeutic ratio might Leukemia 1998; 12: 1461–1466.

Leukemia Nitric oxide and fludarabine in B-CLL DJ Adams et al 1859 14 Chou T, Talalay P. Quantitative analysis of dose-effect relation- 30 Huang P, Plunkett W. Action of 9-beta-d-arabinofuranosyl-2-fluo- ships: the combined effects of multiple drugs or enzyme inhibitors. roadenine on RNA metabolism. Molec Pharmacol 1991; 39: Adv Enzyme Regul 1984; 22: 27–55. 449–455. 15 Kanzawa F, Nishio K, Fukuoka K, Fukuda M, Kunimoto T, Saijo 31 Genini D, Budihardjo I, Plunkett W, Wang X, Carrera CJ, Cottam N. Evaluation of synergism by a novel three-dimensional model HB, Carson DA, Leoni LM. Nucleotide requirements for the in for the combined action of cisplatin and etoposide on the growth vitro activation of the apoptosis protein-activating factor-1- of a human small-cell lung-cancer cell line, SBC-3. Int J Cancer mediated caspase pathway. J Biol Chem 2000; 275: 29–34. 1997; 71: 311–319. 32 Frank DA, Mahajan S, Ritz J. Fludarabine-induced immuno- 16 Cohen DP, Adams DJ, Flowers JL, Wall ME, Wani MC, Manikumar suppression is associated with inhibition of STAT1 signaling. Nat- G, Colvin OM, Silber R. Pre-clinical evaluation of SN-38 and ure Med 1999; 5: 444–447. novel analogs against human chronic B-cell lym- 33 Frank DA, Mahajan S, Ritz J. B lymphocytes from patients with phocytic leukemia lymphocytes. Leukemia Res 1999; 23: 1061– chronic lymphocytic leukemia contain signal transducer and acti- 1070. vator of transcription (STAT) 1 and STAT3 constitutively phos- 17 Keefer LK, Nims RW, Davies KM, Wink DA. ‘NONOates’ (1-sub- phorylated on serine residues. J Clin Invest 1997; 100: 3140– stituted diazen-1-ium-1,2-diolates) as nitric oxide donors: con- 3148. venient nitric oxide dosage forms: nitric oxide: Pt a – Sources and 34 Levesque MC, Adams DJ, Misukonis MA, Flowers J, Silber R, detection of No: No Synthase. Meth Enzymol 1996; 268: 281– Weinberg JB. Detection of inducible nitric oxide synthase (NOS2) 293. mRNA, antigen and enzyme activity in leukemia cells from 18 Jia L, Bonaventura C, Bonaventura J, Stamler JS. S-nitrosohaemo- patients with CLL. Blood 1998; 92: 1781. globin: a dynamic activity of blood involved in vascular control. 35 Kolb JP. Mechanisms involved in the pro- and anti-apoptotic role Nature 1996; 380: 221–226. of NO in human leukemia. Leukemia 2000; 14: 1685–1694. 19 Pinsky DJ, Patton S, Mesaros S, Brovkovych V, Kubaszewski E, 36 Mannick JB, Miao XQ, Stamler JS. Nitric oxide inhibits Fas- Grunfeld S, Malinski T. Mechanical transduction of nitric oxide induced apoptosis. J Biol Chem 1997; 272: 24125–24128. synthesis in the beating heart. Circul Res 1997; 81: 372–379. 37 Hibbs JJ. Synthesis of nitric oxide from l-arginine: a recently disco- 20 Vallance P, Patton S, Bhagat K, MacAllister R, Radomski M, Mon- vered pathway induced by cytokines with antitumour and anti- cada S, Malinski T. Direct measurement of nitric oxide in human microbial activity. Res Immunol 1991; 142: 565–569. beings. Lancet 1995; 346: 153–154. 38 Stamler JS. Redox signaling – nitrosylation and related target inter- 21 Daniel PT. Dissecting the pathways to death. Leukemia 2000; 14: actions of nitric-oxide. Cell 1994; 78: 931–936. 2035–2044. 39 Cooper CE. Nitric oxide and iron proteins. Biochim Biophys Acta 22 Solary E, Droin N, Bettaieb A, Corcos L, Dimanche-Boitrel MT, 1999; 1411: 290–309. Garrido O. Positive and negative regulation of apoptotic pathways 40 Brown GC. Nitric oxide and mitochondrial respiration. Biochim by cytotoxic agents in hematological malignancies. Leukemia Biophys Acta 1999; 1411: 351–369. 2000; 14: 1833–1849. 41 Beltran B, Orsi A, Clementi E, Moncada S. Oxidative stress and 23 Huang P, Sandoval A, Van Den Neste E, Keating MJ, Plunkett W. S-nitrosylation of proteins in cells. Br J Pharmacol 2000; 129: Inhibition of RNA transcription: a biochemical mechanism of 953–960. action against chronic lymphocytic leukemia cells by fludarabine. 42 Mohr S, Hallak H, de Boitte A, Lapetina EG, Brune B. Nitric oxide- Leukemia 2000; 14: 1405–1413. induced S-glutathionylation and inactivation of glyceraldehyde-3- 24 Hanauske A, Von Hoff D. Clinical development of fludarabine. phosphate dehydrogenase. J Biol Chem 1999; 274: 9427–9430. In: Cheson B, Keating M, Plunkett W (eds). Nucleoside Analogs 43 Graziewicz M, Wink DA, Laval F. Nitric oxide inhibits DNA ligase in Cancer Therapy, vol. 12, Basic and Clinical Oncology. Dekker: activity: potential mechanisms for NO-mediated DNA damage. NewYork, 1997, pp 113–158. 25 Kano Y, Akutsu M, Tsunoda S, Suzuki K, Ichikawa A, Furukawa Carcinogenesis 1996; 17: 2501–2505. Y, Bai L, Kon K. In vitro cytotoxic effects of fludarabine (2-F-ara- 44 Wink DA, Cook JA, Christodoulou D, Krishna MC, Pacelli R, Kim A) in combination with commonly used antileukemic agents by S, DeGraff W, Gamson J, Vodovotz Y, Russo A, Mitchell JB. Nitric isobologram analysis. Leukemia 2000; 14: 379–388. oxide and some nitric oxide donor compounds enhance the cyto- 26 Weiss MA, Glenn M, Maslak P, Rahman Z, Noy A, Zelenetz A, toxicity of cisplatin. Nitric Oxide-Biol Chem 1997; 1: 88–94. Scheinberg DA, Golde DW. Consolidation therapy with high-dose 45 Cook JA, Krishna MC, Pacelli R, DeGraff W, Liebmann J, Mitchell cyclophosphamide improves the quality of response in patients JB, Russo A, Wink DA. Nitric oxide enhancement of - with chronic lymphocytic leukemia treated with fludarabine as induced cytotoxicity. Br J Cancer 1997; 76: 325–334. induction therapy. Leukemia 2000; 14: 1577–1582. 46 Saavedra JE, Shami PJ, Wang LY, Davies KM, Booth MN, Citro 27 Plunkett W, Gandhi V. Nucleoside analogs: cellular pharma- ML, Keefer LK. Esterase-sensitive nitric oxide donors of the diazen- cology, mechanisms of action, and strategies for combination ther- iumdiolate family: in vitro antileukemic activity. J Med Chem apy. In: Cheson B, Keating M, Plunkett W (eds). Nucleoside Anal- 2000; 43: 261–269. ogs in Cancer Therapy, vol. 12, Basic and Clinical Oncology. 47 Saavedra JE, Billiar TR, Williams DL, Kim YM, Watkins SC, Keefer Dekker: NewYork, 1997, pp 1–35. LK. Targeting nitric oxide (NO) delivery in vivo. Design of a liver- 28 Laval F, Wink DA. Inhibition by nitric oxide of the repair protein, selective NO donor prodrug that blocks tumor necrosis factor- O6-methylguanine-DNA-methyltransferase. Carcinogenesis 1994; alpha-induced apoptosis and toxicity in liver. J Med Chem 1997; 15: 443–447. 40: 1947–1954. 29 Iwasaki H, Huang P, Keating MJ, Plunkett W. Differential incor- 48 Kanzawa F, Nishio K, Fukuoka K, Sunami T, Saijo N. In vitro inter- poration of ara-C, gemcitabine, and fludarabine replicating and actions of a newderivative of spicamycin, KRN5500, and other repairing DNA in proliferating human leukemia cells. Blood 1997; anticancer drugs using a three-dimensional model. Cancer Chem- 90: 270–278. other Pharmacol 1999; 43: 353–363.

Leukemia