Oncogene (2003) 22, 4459–4468 & 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00 www.nature.com/onc ORIGINAL PAPERS BCL6 overexpression prevents increase in reactive oxygen species and inhibits apoptosis induced by chemotherapeutic reagents in B-cell cells

Tetsuya Kurosu1, Tetsuya Fukuda1, Tohru Miki1 and Osamu Miura*,1

1Department of Hematology and Oncology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyoku, Tokyo 113-8519, Japan

Chromosomal translocations and somatic mutations diffuse large B-cell lymphoma (DLBCL) (Kerckaert occurring in the 50 noncoding region of the BCL6 , et al., 1993; Ye et al., 1993; Miki et al., 1994). The BCL6 encoding a transcriptional repressor, are most frequent gene encodes a transcriptional repressor containing a genetic abnormalities associated with non-Hodgkin B-cell carboxy-terminal DNA-binding zinc-finger domain and lymphoma and result in deregulated expression of BCL6. an amino-terminal BTB/POZ domain, which can bind However, the significance of deregulated expression of to transcriptional repressor cofactors, including N-CoR BCL6 in lymphomagenesis and its effect on clinical and SMRT, and, thereby, recruit the outcomes of lymphoma patients have remained elusive. In complex (reviewed in Dent et al., 2002; Niu, the present study, we established Daudi and Raji B-cell 2002). BCL6is preferentially expressed in germinal lymphoma cell lines that overexpress BCL6 or its mutant, center B cells and has been demonstrated to be required BCL6-Ala333/343, in which serine residues required for for the formation of germinal centers (Cattoretti et al., degradation through the proteasome pathway in B-cell 1995; Onizuka et al., 1995; Dent et al., 1997; Fukuda -stimulated cells are mutated. BCL6 overexpres- et al., 1997; Ye et al., 1997). A recent study with DNA sion did not have any significant effect on cell prolifera- microarray analysis revealed that BCL6represses a tion, but significantly inhibited apoptosis caused by discrete set of involved in B-cell activation, etoposide, which induced a proteasome-dependent degra- terminal differentiation, and inflammation (Shaffer dation of BCL6. BCL6-Ala333/343 was not degraded et al., 2000). It has thus been hypothesized that after etoposide treatment and strongly inhibited apoptosis. downregulation of BCL6is required for germinal center In these lymphoma cell lines, etoposide increased the B cells to differentiate into plasma cells. Consistent with generation of reactive oxygen species (ROS) and reduced this hypothesis, BCL6expression is regulated by signals mitochondria membrane potential, both of which were important for the differentiation of germinal center B inhibited by the antioxidant N-acetyl-l-cysteine (NAC). cells. For instance, the expression of BCL6is down- NAC also inhibited apoptosis. Furthermore, BCL6 over- regulated at the transcriptional level by the CD40 expression was found to inhibit the increase in ROS levels receptor signaling (Allman et al., 1996) and at the and apoptosis in response to etoposide and other protein level by the B-cell receptor signaling, which chemotherapeutic reagents. These results raise the possi- induces phosphorylation of Ser333 and Ser343 of BCL6 bility that deregulated expression of BCL6 may endow through activation of Erks and, thereby, targets BCL6 lymphoma cells with resistance to chemotherapeutic for degradation by the ubiquitin proteasome pathway reagents, most likely by enhancing the antioxidant defense (Niu et al., 1998). systems. Chromosomal translocations involving the BCL6 Oncogene (2003) 22, 4459–4468. doi:10.1038/sj.onc.1206755 gene represent the most common genetic abnormality in non-Hodgkin B-cell and have been Keywords: BCL6; apoptosis; chemotherapeutic re- demonstrated in 27–45% of DLBCL and 5–14% of agents; reactive oxygen species; etoposide follicular lymphomas (Bastard et al., 1994; Lo Coco et al., 1994; Otsuki et al., 1995; Ohno and Fukuhara, 1997). These translocations typically juxtapose the intact BCL6-coding sequences with heterologous pro- moters from genes expressed in B cells, including the Introduction immunoglobulin genes, thus causing deregulated expres- sion of BCL6(Ye et al., 1995; Chen et al., 1998). Besides The BCL6gene was originally identified on the break this promoter substitution mechanism, occasional mi- point of a chromosomal translocation involving 3q27 in crodeletions and frequent multiple somatic mutations in the 50 noncoding negative autoregulatory region of the *Correspondence: O Miura; E-mail: [email protected] BCL6gene have been shown to deregulate the expres- Received 11 February 2003; revised 1 May 2003; accepted 1 May 2003 sion of BCL6in lymphoma cells (Kikuchi et al., 2000; Antiapoptotic effect of BCL6 in lymphoma cells T Kurosu et al 4460 Nakamura, 2000; Wang et al., 2002). Since the BCL6 a gene is most frequently affected in non-Hodgkin B-cell Daudi Daudi/Ton-BCL6 lymphomas, it has been suggested that deregulated DOX - - + expression of BCL6should play an important role in Blot: lymphomagenesis. However, the oncogenic ability of anti-BCL6 BCL6in B cells has remained elusive. It has also remained very controversial as to whether the BCL6 anti-α -tubulin gene alterations affect the clinical outcome of patients with Non-Hodgkin lymphomas (reviewed in Lossos et al., 2001; Barrans et al., 2002; Dent et al., 2002; Niu, b 2002). 120% To gain insight into the significance of deregulated BCL6expression in the pathogenesis and prognosis of 100% non-Hodgkin lymphomas, we overexpressed BCL6in germinal center B-cell-derived lymphoma cell lines, 80% Daudi and Raji, and examined the effects on cell proliferation and apoptosis. BCL6overexpression sig- DOX(-) 60% nificantly inhibited apoptosis of lymphoma cells in DOX(+)

response to etoposide and other chemotherapeutic Viable cells reagents. BCL6overexpression also inhibited the 40% increase in reactive oxygen species (ROS) levels and the reduction of mitochondria transmembrane potential 20% (Dcm) in response to etoposide. The ROS scavenger N- l acetyl- -cysteine (NAC) also inhibited the reduction of 0% Dcm and apoptosis as well as the increase in ROS levels 0 0.1 0.25 0.5 1 2 5 10 20 in etoposide-treated lymphoma cells. These observations Etoposide (µM) indicate that BCL6overexpression inhibits apoptosis of lymphoma cells treated with chemotherapeutic reagents, Figure 1 Inducible overexpression of BCL6enhances the survival most likely through enhancement of the antioxidant of Daudi cells treated with etoposide. (a) Parental Daudi cells and Daudi/Ton-BCL6cells were cultured overnight in the absence ( À) defense systems. or presence ( þ ) of DOX (1 mg/ml), as indicated, and harvested. Total cell lysates were subjected to Western blot analysis with anti- BCL6antibody followed by reprobing with anti- a-tubulin. (b) Daudi/Ton-BCL6cells, precultured overnight in the absence ( À)or Results presence ( þ ) of DOX as indicated, were treated with indicated concentrations of etoposide for 24 h in the absence (À) or presence BCL6 overexpression inhibits apoptosis of B-cell ( þ ) of DOX. Numbers of viable cells were measured by the XTT assay. Each data point represents the mean of triplicate determina- lymphoma cells treated with etoposide tions, with error bars indicating standard errors, and is expressed as To explore the possibility that overexpression of BCL6 a percentage of cell numbers without etoposide may have an effect on the proliferation of lymphoma cells or on their responses to chemotherapeutic reagents, we established a clone of the Burkitt’s lymphoma Daudi cell line, Daudi/Ton-BCL6, as described under Materi- absence of DOX, and the numbers of viable cells were als and methods. As shown in Figure 1a, Daudi/Ton- measured by the XTT assay. As shown in Figure 1b, BCL6cells express a low level of BCL6,comparable etoposide decreased the number of viable cells in a dose- with that in parental Daudi cells, in the absence of dependent manner in the presence or absence of DOX. doxycycline (DOX) and inducibly overexpress BCL6 However, a significantly higher number of cells survived when cultured in the presence of DOX. We first the treatment of etoposide at concentrations as high as examined the effect of BCL6overexpression on cell 2 mm when cultured with DOX, which indicates that growth by culturing Daudi/Ton-BCL6cells in the Daudi cells overexpressing BCL6are more resistant to presence or absence of DOX and by measuring the etoposide treatment than cells expressing the endogen- number of viable cells for up to 4 days by the XTT ous level of BCL6. colorimetric assay, as described under Materials and Chemotherapeutic reagents, including etoposide, have methods. However, we did not observe any significant been known to induce apoptosis of cancer cells. There- effect of BCL6overexpression on cell proliferation in fore, we next examined the effect of BCL6overexpres- repeated experiments (negative data not shown). sion on etoposide-induced apoptosis by the annexin V We next examined the effect of BCL6overexpression staining method, as described under Materials and on the response of Daudi cells to etoposide, a methods. Etoposide induced apoptosis of Daudi/Ton- topoisomerase II inhibitor commonly used as a che- BCL6cells in time- and dose-dependent manners motherapeutic reagent in the treatment of lymphoma. (Figure 2 and data not shown). Notably, the fraction Daudi/Ton-BCL6cells were cultured for 24 h with of annexin V-positive, apoptotic cells was consistently various concentrations of etoposide in the presence or smaller in the presence as compared with the absence of

Oncogene Antiapoptotic effect of BCL6 in lymphoma cells T Kurosu et al 4461 a DOX (-) DOX (+) DOX (-) DOX (+) Etoposide (h) 0 2 4 6 0 2 4 6 Blot: 6.4 % 5.5 % Control anti-PARP * 0.7 % 1.3 %

anti-α-tubulin

Etoposide 20.7 % 17.3 % anti-BCL6 16.3 % 12.7 % PI Figure 3 Reduction in etoposide-induced cleavage of PARP by BCL6overexpression and etoposide-induced downregulation of Annexin V BCL6. Daudi/Ton-BCL6 cells, precultured in the absence (À)or presence ( þ ) of DOX as indicated, were further cultured with 50 mm etoposide for indicated times. Cells were lysed and subjected b to Western blot analysis with anti-PARP antibody. The membrane 80% was sequentially reprobed with anti-a-tubulin and anti-BCL6, as indicated. The position of cleaved PARP is indicated by an asterisk

60%

DOX(-) hand, DOX did not show any significant effect on 40% DOX(+) etoposide-induced apoptosis, as estimated by annexin V binding as well as by cleavage of PARP, in Daudi/Ton cells. These data indicate that overexpression of BCL6 20% significantly inhibits etoposide-induced apoptosis invol-

Annexin V positive cells Annexin V positive ving caspase activation in Daudi cells. To rule out the possibility that the anti-apoptotic 0% 0 8 16 24 effect of BCL6overexpression observed in Daudi cells is Etoposide (h) unique to this cell line, we also examined Raji cells, another Burkitt’s lymphoma cell line. As described Figure 2 Overexpression of BCL6mitigates etoposide-induced under Materials and methods, we established a Raji apoptosis in Daudi cells. Daudi/Ton-BCL6cells, precultured overnight in the absence (À) or presence ( þ ) of DOX as indicated, clone, Raji-BCL6, overexpressing wild-type BCL6, and were further cultured with 2 mm etoposide or without etoposide a clone, Raji-BCL6-Ala333/343, overexpressing the (control), as indicated, for 16h ( a) or with 2 mm etoposide for BCL6-Ala333/343 mutant, in which main phosphoryla- indicated times (b) in the absence (À) or presence ( þ ) of DOX. tion targets for Erks, Ser333 and Ser343 (Niu et al., Cells were then stained with annexin V-FITC as well as PI and analysed by flow cytometry. Percentages of cells staining positive 1998) are mutated. As shown in Figure 4, etoposide- for both annexin V and PI or annexin V alone are indicated induced apoptosis, as estimated by annexin V binding as well as by cleavage of PARP, was moderately or more remarkably inhibited in Raji-BCL6or Raji-BCL6- Ala333/343 cells, respectively, as compared with that DOX at various incubation periods as well as at various in parental Raji cells (Figure 4). Similar to the concentrations of etoposide in repeated experiments observation in Daudi cells, the expression level of (Figure 2 and data not shown). We also examined BCL6in parental Raji cells significantly declined after poly(ADP-ribose) polymerase (PARP), which is cleaved etoposide treatment, while that in Raji-BCL6cells by caspase-3 or caspase-7 in cells undergoing apoptosis, showed a more moderate decline. On the other hand, in Daudi/Ton-BCL6cells treated with etoposide by the expression level of BCL6-Ala333/343 in Raji-BCL6- Western blot analysis. As shown in Figure 3, PARP was Ala333/343 did not show any decline, but transiently cleaved as early as 2 h after treatment with 50 mm increased at 8 h after etoposid treatment in repeated etoposide in the presence or absence of DOX. However, experiments (Figure 4b and data not shown). Thus, a significantly smaller fraction of PARP was cleaved at these results have confirmed the antiapoptotic effect of 2 h in cells cultured with DOX as compared with that in BCL6overexpression in B-cell lymphoma cells treated cells cultured without DOX. Unexpectedly, reprobing with etoposide and revealed that etoposide induces the with anti-BCL6showed that the expression level of decline in BCL6level, which correlated with apoptosis. BCL6, particularly when expressed at a low, endogen- Furthermore, it was demonstrated that Ser333 and ous level, decreased rapidly with etoposide treatment, Ser343 are not required for the antiapoptotic effect, but while equal loading of samples was confirmed by may be involved in etoposide-induced decline in BCL6 reprobing with anti-a-tubulin (Figure 3). On the other level.

Oncogene Antiapoptotic effect of BCL6 in lymphoma cells T Kurosu et al 4462 a a Raji-BCL6 DMSO/ Z-VAD-fmk/ Raji Raji-BCL6 -Ala333/343 Etoposide Etoposide 0 2 4 6 2 4 6 (h) Blot : Control 1.4% 1.7% 1.9% anti-BCL6

0.5% 0.7% 0.9% anti-PARP * b MG132/ Etoposide 16.3% 10.6% 5.3% MG132 Etoposide Etoposide 0 2 4 6 2 4 6 2 4 6 (h) 16.4% 9.2% 4.4% Blot : PI anti-BCL6 Annexin V anti-PARP b * Raji-BCL6 Raji Raji-BCL6 -Ala333/343 c MG132/ Etoposide (h) 0 8 24 48 0 8 24 48 0 8 24 48 MG132 Etoposide Etoposide Blot : 0 3 6 3 6 3 6 (h) anti-BCL6 Blot : anti-PARP anti-BCL6 * anti-PARP Figure 4 Overexpression of wild-type BCL6or the BCL6-Ala333/ * 343 mutant mitigates etoposide-induced apoptosis in Raji cells. Raji-BCL6or Raji-BCL6-Ala333/343 cells, which overexpress Figure 5 Effects of Z-VAD-fmk and MG132 on etoposide- wild-type BCL6or the BCL6-Ala333/343 mutant, respectively, as induced degradation of BCL6and cleavage of PARP in lymphoma m cells. (a) Daudi cells were pretreated with 0.1% DMSO (solvent well as parental Raji cells were cultured with 1 m etoposide for m 48 h and subjected to the apoptosis assay with annexin V staining control) or the pan-caspase inhibitor Z-VAD-fmk (50 m ), as m indicated, for 30 min and further cultured for indicated times with (a) or cultured with 1 m etoposide for indicated times and m subjected to Western blot analysis with anti-PARP followed by 50 m etoposide. Daudi (b) or Raji (c) cells were pretreated with 0.1% DMSO or the proteasome inhibitor MG132 (50 mm) for reprobing with anti-BCL6( b). The position of cleaved PARP is m indicated by an asterisk 30 min and further cultured with or without 50 m etoposide, as indicated, for indicated times. Cells were then harvested for Western blot analysis with anti-BCL6followed by reprobing with Etoposide induces a proteasome-dependent degradation of anti-PARP. The positions of cleaved PARP are indicated by BCL6 asterisks As the expression level of BCL6declined rapidly in cells undergoing etoposide-induced apoptosis, we next in- compared with MG132, and confirmed its inhibitory vestigated the mechanism for this decline. First, we effect on etoposide-induced BCL6degradation in both examined the effect of the broad spectrum caspase Daudi and Raji cells (data not shown). These results inhibitor benzyloxycarbonyl-Val-Ala-Asp fluoromethyl indicate that BCL6is rapidly degraded in etoposide- ketone (Z-VAD-fmk). As expected, pretreatment with treated cells, not by caspases, but through the protea- 50 mm Z-VAD-fmk inhibited the cleavage of PARP in some pathway. Daudi cells treated with etoposide (Figure 5a). However, Z-VAD-fmk did not show any inhibitory effect on the etoposide-induced decline in the BCL6expression level. BCL6 overexpression represses the etoposide-induced We next examined the effect of N-CBZ-Leu-Leu-Leu- increase in ROS levels and reduction of Dcm AL (MG132), a proteasome inhibitor previously shown In an attempt to gain insight into the mechanisms by to inhibit the degradation of BCL6in germinal center B which overexpression of BCL6confers resistance to cells treated with anti-IgM antibody (Niu et al., 1998). etoposide, we next examined whether etoposide en- As shown in Figure 5b and c, pretreatment with MG132 hances the generation of ROS, which have been shown significantly inhibited the decline in the BCL6expres- to play important roles in the initiation and execution of sion level in both Daudi and Raji cells treated with apoptosis (reviewed in Fleury et al. (2002), Jabs (1999)). etoposide, while MG132 increased or decreased the As shown in Figure 6a, treatment of Raji cells with 1 mm PARP cleavage induced by etoposide in Daudi or Raji etoposide for 12–48 h induced a time-dependent increase cells, respectively. We also examined the effect of in ROS levels, as estimated fluorometrically using a lactacystin, a more specific proteasome inhibitor as fluorescent probe, dihydroethidium (HE). Furthermore,

Oncogene Antiapoptotic effect of BCL6 in lymphoma cells T Kurosu et al 4463 a population of cells showing moderately increased ROS Etoposide (h) levels even at 48 h of etoposide treatment (Figure 6a). 12 24 48 The etoposide-induced increases in ROS generation in these clones, thus, correlated with their sensitivities to etoposide-induced apoptosis, which raises a possibility Raji that BCL6overexpression may mitigate etoposide- induced apoptosis by reducing ROS levels. Since mitochondrial membrane permeabilization also plays an important role in the induction of apoptosis (reviewed in Fleury et al. (2002), Jabs (1999), Zamzami and Kroemer (2001)), we next examined Dcm by using a fluorescent probe, as described under Materials and Raji-BCL6 methods. As demonstrated in Figure 6b, Raji cells showing reduced Dcm gradually increased in number from 12 to 24 h of etoposide treatment, and only a minor fraction of cells retained Dcm intact at 48 h. On the other hand, Raji-BCL6cells showed a less significant reduction in Dcm during etoposide treatment as Raji-BCL6 compared with parental Raji cells. Furthermore, Raji- -Ala333/343 BCL6-Ala333/343 did not show any significant increase in the fraction of cells showing reduced Dcm before 24 h, and a significant portion of the cells retained Dcm intact at 48 h (Figure 6b). These results suggest that over- expression of BCL6may also play an inhibitory role in b Etoposide (h) etoposide-induced reduction of Dcm. Control 12 24 48 To examine the effects of increased levels of ROS on reduction of Dc and apoptosis in cells treated with M1 M2 M1 M2 M1 M2 M1 M2 m etoposide, we pretreated Raji cells with the ROS Raji 10.8% 24.2% 80.9% scavenger NAC and further treated cells with etoposide as well as NAC for 24–48 h. As shown in Figure 7a, NAC partially inhibited the etoposide-induced increase M1 M2 M1 M2 M1 M2 M1 M2 in ROS levels. In addition, NAC also partially reduced Raji-BCL6 4.0% 5.9% 12.6% 67.2% the loss of Dcm and apoptosis in Raji cells treated with etoposide (Figure 7b, c). These results suggest that the etoposide-induced increase in ROS levels may play a role in the induction of the loss of Dc , thus leading to M1 M2 M1 M2 M1 M2 M1 M2 m Raji-BCL6 apoptosis, and support the idea that BCL6overexpres- 3.5% 4.0% 5.5% 33.9% -Ala333/343 sion inhibits apoptosis by reducing ROS levels. We also examined the correlation between BCL6 overexpression and the etoposide-induced ROS genera- Figure 6 Overexpression of wild-type BCL6or the BCL6-Ala333/ tion in Daudi/Ton-BCL6cells. Unlike Raji cells, the 343 mutant reduces etoposide-induced increase in ROS levels and increase in ROS levels in these cells after etoposide reduction of Dcm in Raji cells. (a) Raji, Raji-BCL6, or Raji-BCL6- Ala333/343 cells were incubated with (filled area) or without (open treatment was not observed in whole population of cells, area) 1 mm etoposide for indicated times and subjected to the flow but in some fraction of cells. However, the fraction of cytometric analysis for ROS generation by using the fluorescent cells showing high levels of ROS after treatment with probe HE, as described under Materials and methods. (b) Raji, Raji-BCL6, or Raji-BCL6-Ala333/343 cells were incubated with various concentrations with etoposide was significantly 1 mm etoposide for indicated times or without etoposide for 48 h as reduced when cells were cultured with DOX to over- control (control) and subjected to the flow cytometric analysis for express BCL6(Figure 8). Therefore, etoposide-induced Dcm as described under Materials and methods. Percentages of increase in ROS levels correlated with sensitivity to cells showing reduced Dcm are indicated etoposide-induced apoptosis, which is in accordance with the idea that BCL6overexpression inhibits at 48 h after etoposide treatment, a population of cells apoptosis by reducing ROS levels. that exhibited drastically increased levels of ROS was additionally observed. In Raji-BCL6cells, both etopo- BCL6 overexpression prevents the increase in ROS levels side-induced time-dependent increase in ROS levels at and apoptosis in cells treated with various 12–24 h and the fraction of cells that exhibited chemotherapeutic reagents drastically increased levels of ROS at 48 h were less significant as compared with those observed in Raji To examine whether the antiapoptotic effect of BCL6 cells. Furthermore, the etoposide-induced increase in overexpression is specific for the topoisomerase II ROS levels was most significantly inhibited in Raji- inhibitor etoposide, we examined three other che- BCL6-Ala333/334 cells, which exhibited only a single motherapeutic reagents that are in common use for

Oncogene Antiapoptotic effect of BCL6 in lymphoma cells T Kurosu et al 4464 a a Etoposide (h) DOX(-) DOX(+)

24 48 M1 M2 M1 M2 Control 7.7 % 5.5 % Control

M1 M2 M1 M2 63.5 % Etoposide 42.7 %

NAC

b 80 % b c Etoposide (h) Etoposide (h) 60 % 24 48 48 DOX(-) M1 M2 M1 M2 40 % 9.6% Control DOX(+) 45.4% 66.2 % 20 % 21.6% %cells staining with HE 0% M1 M2 M1 M2 0 1 0.1 NAC 32.6% 40.0 % 0.01 0.05 0.5 11.6% Etoposide (µΜ)

4.0% Figure 8 BCL6overexpression reduces etoposide-induced in- crease in ROS levels in Daudi cells. Daudi/Ton-BCL6cells, PI precultured overnight in the absence (À) or presence ( þ )of DOX as indicated, were further cultured for 24 h with 0.5 mm Annexin V etoposide or without etoposide (control), as indicated (a) or with Figure 7 The antioxidant NAC reduces etoposide-induced in- indicated concentrations of etoposide (b). Cells were then subjected to the flow cytometric analysis for ROS generation by using HE. crease in ROS levels, reduction of Dcm, and apoptosis in Raji cells. (a) Raji cells were pretreated with or without 25 mm NAC, as Percentage of cells staining positive with HE staining are indicated indicated. Subsequently, cells were further cultured with (filled in (a) and plotted in (b) areas) or without (open areas) 1 mm etoposide for indicated times and subjected to the flow cytometric analysis for ROS (a), cultured with 1 mm for indicated times and subjected to the flow cytometric Furthermore, treatment of Daudi/Ton-BCL6cells with analysis for Dcm (b), or cultured with 1 mm for 48 h and subjected to daunorubicin, Ara-C, or vincristine increased the ROS the annexin V binding assay for apoptosis (c). Percentages of cells levels, which was reduced by overexpression of BCL6 showing reduced Dcm and those of cells staining positive for both annexin V and PI or annexin V alone are indicated induced by DOX (Figure 9b). On the other hand, DOX did not have any effect on the increase in ROS levels and apoptosis in Daudi/Ton cells treated with these che- motherapeutic reagents (data not shown). These results treatment of hematopoietic and have differ- indicate that overexpression of BCL6endows resistance ent mechanisms of action (Chabner and Longo, 2001). to various chemotherapeutic reagents most likely by Daunorubicin also inhibits topoisomerase II to induce reducing the ROS generation. DNA strand breakage as well as intercalates between DNA base pairs. Ara-C is phosphorylated to Ara-CTP, competes with dCTP for incorporation into newly Discussion synthesized DNA, and terminates DNA chain elonga- tion. Vincristine inhibits tubulin polymerization and In the present study, we inducibly or constitutively prevents the formation of mitotic spindle (Chabner & overexpressed BCL6in germinal center B-cell-derived Longo, 2001). In spite of different mechanisms of action lymphoma cells, Daudi or Raji cell lines, respectively, for these reagents, induction of BCL6overexpression by and examined the effects of BCL6overexpression on cell DOX treatment in Daudi/Ton-BCL6cells reduced proliferation and apoptosis. In both cell lines, BCL6 apoptosis induced by each of these three reagents, as overexpression neither affected cell proliferation nor estimated by annexin V binding as well as by cleavage of induced apoptosis. However, apoptosis induced by PARP (Figure 9a and data not shown). Similarly, Raji various chemotherapeutic reagents were partially but clones overexpressing BCL6or the BCL6-Ala333/343 significantly prevented by BCL6overexpression in both mutant exhibited resistance to each of these reagents as cell lines, as examined by the annexin V binding assay. compared with parental cells (data not shown). Further studies in detail on etoposide-treated cells

Oncogene Antiapoptotic effect of BCL6 in lymphoma cells T Kurosu et al 4465 a impaired S phase progression and apoptosis. It should DOX (-) DOX (+) be noted, however, that the cell lines utilized in these studies are nonlymphoid cells, which do not normally 0 2 4 6 12 0 2 4 6 12 (h) express BCL6. Thus, it is speculated that the effects of BCL6overexpression in nonlymphoid cells that do not normally express BCL6are quite different from those in DNR * lymphoma cells of germinal center B-cell origin. Yamochi et al. (1999) also reported that apoptosis Ara-C induced by BCL6overexpression in CV-1 and HeLa * cells was preceded by downregulation of apoptosis repressors BCL2 and BCL-XL. In accordance with this, VCR Tang et al. (2002) recently reported that the forced expression of a constitutively active mutant of AFX, a * member of the forkhead family of factors, in HeLa cells induced apoptosis as well as overexpres- b sion of BCL6and downregulation of BCL-X L. BCL6 80% was further shown to bind multiple sites in the BCL-XL

E promoter in vitro and repress the BCL-XL promoter

H

h 60% activity in 293E cells. However, we did not observe any

t

i significant effect of BCL6overexpression on BCL-X L

w

g protein levels in Daudi and Raji cells (unpublished

n DOX(-) i 40%

n observation). Furthermore, we did not observe any

i DOX(+)

a

t endogenous BCL2 expression in these cells, which is in

s

s accordance with the fact that germinal center B cells do

l

l 20%

e not generally express BCL2 (Shaffer et al., 2001). Thus,

c

% it is speculated that BCL6overexpression in hetero- 0% logous cells may have different cellular effects by repressing physiologically irrelevant genes. ol e R C R tr id N a- In accordance with the present study showing the n os r VC o p D A C to antiapoptotic effect of BCL6, we previously demon- E strated that BCL6expression is upregulated during differentiation of myocytes and that adenovirus- Figure 9 BCL6overexpression reduces apoptosis and the increase in ROS levels induced by various chemotherapeutic reagents in mediated overexpression of BCL6prevented apoptosis Daudi cells. (a) Daudi/Ton-BCL6cells, precultured overnight in during myocyte differentiation (Kumagai et al., 1999). the absence (À) or presence ( þ ) of DOX as indicated, were further In addition, a recent study in Bcl6-deficient mice cultured with 10 mm of daunorubicin (DNR), Ara-C, or vincristine suggested a role for BCL6in protecting apoptosis of (VCR), as indicated, for indicated times. Cells were then lysed and subjected to Western blot analysis with anti-PARP. The positions spermatocytes, which normally express BCL6, in re- of cleaved PARP are indicated by asterisks. (b) Daudi/Ton-BCL6 sponse to stresses (Kojima et al., 2001). Furthermore, cells were similarly treated and cultured without any chemother- Shaffer et al. (2000) reported that inhibition of BCL6 apeutic reagent as control (control) or with 0.5 mm etoposide, function in Raji cells by a dominant-negative form of 0.01 mm DNR, 2 mm Ara-C, and 0.1 mm VCR, as indicated, for 24 h. BCL6, consisting solely of the DNA-binding zinc-finger Cells were then subjected to the flow cytometric analysis for ROS generation using HE domain, inhibited the cell cycle progression and induced apoptosis. Together, these observations strongly suggest that BCL6plays an antiapoptotic role in various types revealed that BCL6overexpression has inhibitory effects of cells endogenously expressing BCL6, including on decrease in Dcm generation of ROS, and caspase- germinal center B cells and lymphoma cells derived mediated cleavage of PARP in these cells. Intriguingly, from these cells. BCL6was found to undergo a proteasome-dependent In the present study, an increase in the production of degradation in etoposide-treated cells. These findings ROS and a decline in Dcm were observed in Daudi and strongly suggest that the deregulated BCL6expression Raji cells treated with etoposide. Furthermore, inhibi- in B-cell lymphomas plays a role in the modulation of tion of increase in ROS by the antioxidant NAC apoptotic responses of these cells to chemotherapeutic significantly reduced apoptosis as well as reduction in reagents and, thus, may have a significant effect on Dcm, thus suggesting that an increase in the production clinical outcomes of lymphoma patients treated with of ROS may play an important role in etoposide- chemotherapy. induced apoptosis in lymphoma cells. In accordance In contrast to these results, Yamochi et al. (1999) with this, previous studies have demonstrated that ROS previously reported that adenovirus-mediated overex- act upstream of mitochondrial membrane depolariza- pression of BCL6in CV-1 and HeLa cells induced cell tion and execution of apoptosis by activation of cycle arrest at the G2/M phase and apoptosis. Similarly, caspases (reviewed in Fleury et al. (2002), Jabs (1999)). Albagli et al. (1999) reported that tetracycline-regulated An early increase in ROS levels preceding mitochondrial overexpression of BCL6in U2OS osteosarcoma cells membrane permeabilization has also been reported in

Oncogene Antiapoptotic effect of BCL6 in lymphoma cells T Kurosu et al 4466 various models, including apoptosis induced by etopo- BCL6in cells undergoing apoptosis. Intriguingly, our side (Fleury et al., 2002). However, previous studies also preliminary results have shown that Erks are activated suggest that reduction of Dcm, concomitant with in Raji cells in response to etoposide, although uncoupling of mitochondrial electron transfer and PD98059, an MEK inhibitor, did not show any ATP synthesis, causes increased generation of ROS. significant effect on etoposide-induced BCL6degrada- Thus, a dual role has been implicated for ROS in the tion (unpublished data). Further studies are also under- apoptotic process: first, as a facultative signal during the way to elucidate the mechanisms for BCL6degradation induction phase, and, second, as a common consequence and its significance in the regulation of apoptosis of mitochondrial permeability transition leading to the induced by chemotherapeutic reagents. final destruction of the cell (Jabs, 1999; Fleury et al., The results of the present study predict that patients 2002). We speculate that a fraction of Raji cells that with non- with deregulated expres- showed dramatic increases in ROS levels at 48 h after sion of BCL6may have poor clinical outcomes, because etoposide treatment may be due to excessive generation BCL6overexpression significantly affected apoptosis of of ROS caused by loss of Dcm. In accordance with this lymphoma cells treated with chemotherapeutic reagents. speculation, this fraction was significantly smaller or However, the prognostic significance of BCL6gene was barely detectable in Raji-BCL6or Raji-BCL6- rearrangements or BCL6expression in DLBCL has Ala333/343 cells, respectively, thus showing a correla- remained very controversial, and many contradictory tion with the fraction of apoptotic cells. More impor- results have so far been reported (Lossos et al., 2001; tantly, the early increase in ROS levels after etoposide Barrans et al., 2002; Dent et al., 2002; Niu, 2002). This treatment was also reduced in Raji cells overexpressing may be at least partly caused by the heterogeneity of BCL6or its mutant and correlated with their suscept- DLBCL. In particular, a recent profiling ibility to etoposide-induced apoptosis. Inducible over- study with DNA microarray techniques has classified expression of BCL6in Daudi cells also reduced the DLBCL into two molecularly distinct forms: germinal increase in ROS levels in response to etoposide as well as center B-like and activated B-like forms (Alizadeh et al., other chemotherapeutic reagents and mitigated apopto- 2000). Germinal center B-like DLBCL is characterized sis. Therefore, it is speculated that BCL6overexpression by the expression of genes normally expressed in prevents apoptosis of lymphoma cells in response to germinal center B cells, including the BCL6gene, and chemotherapeutic reagents by inhibiting the increase in is associated with better overall survival. Accordingly, a ROS levels. Intriguingly, our preliminary data show that recent study has defined the BCL6gene expression as a BCL6overexpression also inhibits apoptosis of Daudi favorable prognostic factor in DLBCL, although the and Raji cells cultured under fetal calf serum (FCS)- expression levels of BCL6protein were not examined deprived conditions (unpublished observations), which (Lossos et al., 2001). However, in a more recent study, raises the possibility that the deregulated expression of the BCL6gene rearrangements were more frequently BCL6in B cells may play a role in lymphomagenesis observed in cases not expressing the germinal center B- through its antiapoptotic effects on various stresses. cell phenotype and were associated with a poor Currently, studies are also underway in our laboratory prognosis as revealed by multivariate analysis of a pure to determine the effects of BCL6on the antioxidant cohort of nodal DLBCL (Barrans et al., 2002). There- defense systems as well as on the defense systems for fore, it is speculated that the deregulated expression of mitochondrial integrity. BCL6, but not its expression per se, predicts a poor The present study has also revealed that the level of clinical outcome in DLBCL. Furthermore, a very recent BCL6declines rapidly in lymphoma cells treated with study has demonstrated that BCL6is inactivated by etoposide or other chemotherapeutic reagents (Figures acetylation and showed that BCL6is acetylated under 2–4 and data not shown). This is most likely caused by physiologic conditions in normal germinal center B cells the proteasome-mediated degradation, because it was and in germinal center-derived B-cell lymphoma cells prevented by the proteasome inhibitors, MG132 and (Bereshchenko et al., 2002). Therefore, future studies lactacystin, but not by the broad-spectrum caspase examining the expression level and acetylation status of inhibitor Z-VAD-fmk (Figure 5). However, it is also BCL6protein in lymphoma cells may shed more light on possible that the decline in BCL6level is partly caused at the prognostic significance of BCL6expression and its the transcriptional level, because the level of over- effect on sensitivities of lymphoma cells to chemother- expressed BCL6declined more gradually than that of apeutic reagents. Future studies aiming to exploit BCL6 endogenous BCL6(Figures 3 and 4). Notably, the level in lymphoma cells as a molecular target for therapy, for of BCL6-Ala333/343 did not decline, but transiently instance by using inhibitors for deacetylases, are also increased in cells treated with etoposide in repeated needed (Bereshchenko et al., 2002). experiments, which may explain the observation that Raji-BCL6-Ala333/343 cells were more resistant to etoposide than Raji-BCL6cells (Figure 3 and data not Materials and methods shown). It is also speculated that Ser333 and Ser343 of BCL6, sites involved in phosphorylation by Erks and Cells and reagents degradation mediated by the proteasome in B-cell Burkitt’s lymphoma cells lines, Daudi and Raji, were obtained receptor signaling (Niu et al., 1998), may also be from the Japanese Cancer Resources Bank (JCRB, Tokyo, involved in proteasome-dependent degradation of Japan). Daudi and Raji cells were cultured in RPMI1640

Oncogene Antiapoptotic effect of BCL6 in lymphoma cells T Kurosu et al 4467 medium (Nissui, Tokyo, Japan) supplemented with 20% or analysis of total cell lysates, samples were prepared by mixing 10% FCS, respectively. 293 T cells were kindly provided by Dr an aliquot of cell lysates with an equal volume of 2 Â S Yamaoka and cultured in DMEM with 10% FCS. Laemmli’s sample buffer and heating at 1001C. Samples were DOX, vincristine, daunorubicin, cytosine arabinoside (Ara- separated by SDS–polyacrylamide gel electrophoresis (PAGE) C), MG132 and NAC were purchased from Sigma (St Louis, and electrotransferred to Immobilon P membrane (Millipore, MO, USA). Etoposide, G418, and hygromycin B were Bedford, MA, USA). The membranes were probed with a purchased from Wako (Osaka, Japan). Z-VAD-fmk and relevant antibody followed by detection using enhanced lactacystin were obtained from Calbiochem (La Jolla, CA, chemiluminescence Western blotting detection system (Amer- USA). A rabbit polyclonal antibody against BCL6was sham Pharmacia Biotech). For reprobing of the membranes, described previously (Nakamura, 2000). A mouse monoclonal the membranes were treated with stripping buffer composed of antibody against a-tubulin (B-7) was purchased from Sigma, 100 mm 2-mercaptoethanol, 2% SDS, and subsequently and a rabbit polyclonal antibody against PARP was obtained probed with a different antibody. from New England Biolabs (Beverly, MA, USA). Cell viability assay Construction of expression plasmids Cell viability was measured by the sodium 30-[1-(phenylamino- Point mutations causing substitution of both Ser333 and carbonyl)-3,4-tetrazolium]-bis(4-methoxy-6-nitro)benzene sulfo- Ser343 with alanine residues were introduced into a full-length nic acid hydrate (XTT) colorimetric assay using the Cell human BCL6cDNA (Miki et al., 1994) by PCR-based Proliferation Kit II (Roche Molecular Biochemicals, Man- methods, as described previously (Niu et al., 1998), and nheim, Germany) according to the manufacturer’s instructions. confirmed by sequencing. The sequences encoding for the Flag In brief, cells were cultured in a 96-well dish at a density of tag were added upstream and in frame with sequences coding 5 Â 104 cells per each well in a final volume of 100 mlwith for wild-type or mutant BCL6by PCR-based methods and indicated concentrations of a chemotherapeutic reagent and confirmed by sequencing. Subsequently, the wild-type and cultured for 24 h. Then, 50 mlofXTTlabelingmixturewas mutant BCL6cDNA clones were subcloned into the pCAGGS added, and cells were further incubated for 4 h before the optical expression plasmid, kindly provided by Dr J Miyazaki, to give density at 490 nm was measured with a microplate reader. pCAGGS-BCL6and pCAGGS-BCL6-Ala333/343,respec- tively. The wild-type BCL6cDNA clone was also subcloned Flow cytometric analysis of apoptosis, ROS production and Dc into pRevTRE (Clontech) to give pRevTRE-BCL6. m For flow cytometric analysis of apoptosis, phosphatidylserine Cell transfection and infection exposure on the outer leaflet of the plasma membrane was detected using the TACS Annexin V Kit (Trevigen, Gaithers- Daudi cells were first transfected with pTet-On (Clontech) by burg, MD, USA) according to the manufacturer’s instructions. electroporation using Gene Pulser (Bio-Rad, Hercules, CA, In brief, 1 Â 106 cells were resuspended in 100 ml of binding USA) at 330V and 960 mF. Transfected cells were selected in buffer containing annexin V-FITC and propidium iodide (PI) medium containing G418 and clones were selected by limiting and incubated for 15 min. After 400 ml of binding buffer was dilution. One selected clone, Daudi/Ton, was used for control added, the cells were analysed using a Becton Dickinson experiments and for infection with a retrovirus carrying the wild- FACSscan flow cytometer (Mountain View, CA, USA). type BCL6cDNA clone, which was prepared by transfecting The production of ROS was estimated fluorometrically À À 293 T cells with pRevTRE-BCL6, SV-c -e , and pHCMV-G using a fluorescent probe, dihydroethidium (HE; Molecular (Yee et al., 1994), kindly provided by Dr J Burns and Dr T probes, Eugene, OR, USA), which is oxidized to the Friedmann, by using FuGene6(Roche) according to the fluorescent intercalator, ethidium, by cellular oxidants, parti- manufacturer’s instructions. After 48 h of infection, cells were cularly superoxide radicals (Rothe and Valet, 1990), as selected in a medium containing G418 and hygromycin B. Several described previously (Dvorakova et al., 2001). In brief, clones were isolated by limiting dilution and examined for the 1 Â 106 cells were incubated with 2 mm HE for 30 min at induction of BCL6expression by the addition of DOX, and one 371C, resuspended in PBS, and analysed by flow cytometry clone, Daudi/Ton-BCL6, was used for subsequent experiments. (excitation: 488 nm, emission: 620 nm). Raji cells were transfected with pCAGGS-BCL6or The Dcm was estimated by using the DePsipher kit pCAGGS-BCL6-Ala333/343 and pSV2neo (Clontech) by (Trevigen), according to the manufacturer’s instructions. In electroporation, as described above. Selected clones were brief, 1 Â 106 cells were incubated with a lipophilic cation (5, 50, examined for BCL6expression, and the clones expressing the 6,60, tetrachloro-1, 10,3,30-tetraethylbenzimidazolyl carbo- highest level of wild-type or mutant BCL6, Raji-BCL6 and cyanin iodide) for 20 min at 371C, washed in PBS and analysed Raji-BCL6-Ala333/343, respectively, were used for subsequent by flow cytometry. experiments. Acknowledgements Immunoblotting We are grateful to Drs Shoji Yamaoka, Jun-ichi Miyazaki, Cells were lysed in a lysis buffer containing 1% Triton X-100, Jane C Burns, Theodore Friedmann for the generous gifts of 20 mm Tris-HCl (pH 7.5), 150 mm NaCl, 1 mm EDTA, 1 mm experimental materials. This work was supported in part by sodium orthovanadate, 1 mm phenylmethylsulfonyl fluoride, grants from the Ministry of Education, Science, Sports and and 10 mg/ml each of aprotinin and leupeptin. For immunoblot Culture of Japan.

References

Albagli O, Lantoine D, Quief S, Quignon F, Englert C, Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Kerckaert JP, Montarras D, Pinset C and Lindon C. (1999). Rosenwald A, Boldrick JC, Sabet H, Tran T, Yu X, Powell Oncogene, 18, 5063–5075. JI, Yang L, Marti GE, Moore T, Hudson Jr J, Lu L, Lewis

Oncogene Antiapoptotic effect of BCL6 in lymphoma cells T Kurosu et al 4468 DB, Tibshirani R, Sherlock G, Chan WC, Greiner TC, Lo Coco F, Ye BH, Lista F, Corradini P, Offit K, Knowles Weisenburger DD, Armitage JO, Warnke R, Levy R, DM, Chaganti RS and Dalla-Favera R. (1994). Blood, 83, Wilson W, Grever MR, Byd JC, Botstein D, Brown PO 1757–1759. and Staudt LM. (2000). Nature, 403, 503–511. Lossos IS, Jones CD, Warnke R, Natkunam Y, Kaizer H, Allman D, Jain A, Dent A, Maile RR, Selvaggi T, Kehry MR Zehnder JL, Tibshirani R and Levy R. (2001). Blood, 98, and Staudt LM. (1996). Blood, 87, 5257–5268. 945–951. Barrans SL, O’Connor SJ, Evans PA, Davies FE, Owen RG, Miki T, Kawamata N, Hirosawa S and Aoki N. (1994). Blood, Haynes AP, Morgan GJ and Jack AS. (2002). Br. J. 83, 26–32. Haematol., 117, 322–332. Nakamura Y. (2000). Leuk. Lymphoma, 38, 505–512. Bastard C, Deweindt C, Kerckaert JP, Lenormand B, Rossi A, Niu H. (2002). Hematol. Oncol., 20, 155–166. Pezzella F, Fruchart C, Duval C, Monconduit M and Tilly Niu H, Ye BH and Dalla-Favera R. (1998). Genes Dev., 12, H. (1994). Blood, 83, 2423–2427. 1953–1961. Bereshchenko OR, Gu W and Dalla-Favera R. (2002). Nat. Ohno H and Fukuhara S. (1997). Leuk. Lymphoma, 27, 53–63. Genet., 32, 606–613. Onizuka T, Moriyama M, Yamochi T, Kuroda T, Kazama A, Cattoretti G, Chang CC, Cechova K, Zhang J, Ye BH, Falini Kanazawa N, Sato K, Kato T, Ota H and Mori S. (1995). B, Louie DC, Offit K, Chaganti RS and Dalla-Favera R. Blood, 86, 28–37. (1995). Blood, 86, 45–53. Otsuki T, Yano T, Clark HM, Bastard C, Kerckaert JP, Jaffe Chabner BA and Longo DL (eds) (2001). Cancer Chemother- ES and Raffeld M. (1995). Blood, 85, 2877–2884. apy and Biotherapy: Principles and Practice. Lippincott Rothe G and Valet G. (1990). J. Leukoc. Biol., 47, 440–448. Williams and Wilkins: Philadelphia. Shaffer AL, Rosenwald A, Hurt EM, Giltnane JM, Lam LT, Chen W, Iida S, Louie DC, Dalla-Favera R and Chaganti RS. Pickeral OK and Staudt LM. (2001). Immunity, 15, 375–385. (1998). Blood, 91, 603–607. Shaffer AL, Yu X, He Y, Boldrick J, Chan EP and Staudt LM. Dent AL, Shaffer AL, Yu X, Allman D and Staudt LM. (2000). Immunity, 13, 199–212. (1997). Science, 276, 589–592. Tang TT, Dowbenko D, Jackson A, Toney L, Lewin DA, Dent AL, Vasanwala FH and Toney LM. (2002). Crit. Rev. Dent AL and Lasky LA. (2002). J. Biol. Chem., 277, 14255– Oncol. Hematol., 41, 1–9. 14265. Dvorakova K, Waltmire CN, Payne CM, Tome ME, Briehl Wang X, Li Z, Naganuma A and Ye BH. (2002). Proc. Natl. MM and Dorr RT. (2001). Blood, 97, 3544–3551. Acad. Sci. USA, 99, 15018–15023. Fleury C, Mignotte B and Vayssiere JL. (2002). Biochimie, 84, Yamochi T, Kaneita Y, Akiyama T, Mori S and Moriyama M. 131–141. (1999). Oncogene, 18, 487–494. Fukuda T, Yoshida T, Okada S, Hatano M, Miki T, Ishibashi Ye BH, Cattoretti G, Shen Q, Zhang J, Hawe N, de Waard R, K, Okabe S, Koseki H, Hirosawa S, Taniguchi M, Miyasaka Leung C, Nouri-Shirazi M, Orazi A, Chaganti RS, Rothman N and Tokuhisa T. (1997). J. Exp. Med., 186, 439–448. P, Stall AM, Pandolfi PP and Dalla-Favera R. (1997). Nat. Jabs T. (1999). Biochem Pharmacol, 57, 231–245. Genet., 16, 161–170. Kerckaert JP, Deweindt C, Tilly H, Quief S, Lecocq G and Ye BH, Chaganti S, Chang CC, Niu H, Corradini P, Chaganti Bastard C. (1993). Nat. Genet., 5, 66–70. RS and Dalla-Favera R. (1995). EMBO J., 14, 6209–6217. Kikuchi M, Miki T, Kumagai T, Fukuda T, Kamiyama R, Ye BH, Lista F, Lo Coco F, Knowles DM, Offit K, Chaganti Miyasaka N and Hirosawa S. (2000). Oncogene, 19, 4941– RS and Dalla-Favera R. (1993). Science, 262, 747–750. 4945. Yee JK, Miyanohara A, LaPorte P, Bouic K, Burns JC and Kojima S, Hatano M, Okada S, Fukuda T, Toyama Y, Yuasa Friedmann T. (1994). Proc. Natl. Acad. Sci. USA, 91, 9564– S, Ito H and Tokuhisa T. (2001). Development, 128, 57–65. 9568. Kumagai T, Miki T, Kikuchi M, Fukuda T, Miyasaka N, Zamzami N and Kroemer G. (2001). Nat. Rev. Mol. Cell Biol., Kamiyama R and Hirosawa S. (1999). Oncogene, 18, 2, 67–71. 467–475.

Oncogene