Ser70 phosphorylation of Bcl-2 by selective tyrosine nitration of PP2A-B56δ stabilizes its antiapoptotic activity

PP2A is an essential phosphatase involved in numerous cellular processes, such as the regulation of cell-cycle progression, cytoskeleton dynamics, and cellular motility, and suppression of tumorigenesis via regulating cell proliferation and death pathways.^1,2  To date, several important signaling pathways have been implicated in the anti-tumorigenic activity of PP2A. For instance, PP2A holoenzyme containing B56α was reported to be a direct regulator of c-Myc and Bcl-2, with the dephosphorylation of the former oncoprotein at Ser62 required for its degradation,³  whereas dephosphorylation of the latter at Ser70 leads to the inactivation of its antiapoptotic activity.⁴  Of note, inhibition of PP2A by okadaic acid or by polyamine depletion results in the specific accumulation of S70 phosphorylated Bcl-2 (S70pBcl-2), which in turn inhibited drug-induced apoptosis.^5,6 

Bcl-2 is a key regulator of apoptosis in various lymphoid and other malignancies. Bcl-2 inhibits cell death by sequestering pro-apoptotic proteins Bax and Bak at the outer mitochondrial membrane (OMM) and preventing the release of pro-apoptotic factors such as cytochrome c, AIF, and SMAC/DIABLO.⁷  The antiapoptotic activity of Bcl-2 is regulated by posttranslational modifications. Three residues (T69, S70, and S87) within the flexible loop domain of Bcl-2^8-10  are phosphorylated mainly by the mitogen-activated protein (MAP) kinases such as c-Jun N-terminal kinase, extracellular signal-regulated kinase (ERK), and p38.¹¹  Mono-site (S70) phosphorylation increases the antiapoptotic activity of Bcl-2 by enhancing dimerization with Bax.^5,6  The 2 adjacent phosphorylation sites provide additional functional modulation of Bcl-2 in a context-dependent manner.¹² 

In addition to its canonical antiapoptotic activity, we have recently shown that Bcl-2 overexpression in hematopoietic tumors is associated with a pro-oxidant intracellular milieu, in particular an increased in intracellular superoxide (O₂⁻) level. More importantly, inhibiting O₂⁻ production abrogated the antiapoptotic activity of Bcl-2.¹³  Although O₂⁻ is implicated in Bcl-2-mediated chemoresistance, the underlying molecular mechanism(s) is/are poorly understood.

Reactive oxygen species (ROS) such as O₂⁻ and H₂O₂ are important signaling molecules that have been implicated in various cellular processes. Redox homeostasis is maintained by antioxidant machineries such as superoxide dismutase (SOD), peroxiredoxins, glutathiones, and catalases. Defects in antioxidant defenses are invariably associated with pathological disease states; for example, SOD1 deficiency is linked with cancer and Alzheimer disease.^14,15 

Interestingly, ROS can regulate the activity of diverse kinases and phosphatases via diverse mechanisms.^16-18  Here we show that physiologic increases in intracellular O₂⁻ induced by pharmacological inhibition and knockdown of SOD1 stabilize the antiapoptotic activity of Bcl-2 by inhibiting its dephosphorylation at S70. We also report that this is a function of selective nitration of a specific tyrosine residue (Y289) on the Bcl-2-bound B56δ subunit of PP2A; site-specific nitration of B56δ releases the PP2A catalytic subunit from Bcl-2, leading to increased S70 phosphorylation. Importantly, the nitrosative regulation of PP2A-mediated dephosphorylation of Bcl-2 is also observed in primary cells derived from clinical human lymphomas. These data provide mechanistic insights into the role of a pro-oxidant state in promoting chemoresistance via inactivating PP2A, thereby undermining its tumor suppressor activity.

Jurkat, Hela, MDA-MB-231, and HK-1 tumor cell-lines were purchased from ATCC (Rockville, MD). All cell lines were cultured and maintained according to supplier’s recommendations.

Cells were lysed with coimmunoprecipitation (co-IP) lysis buffer and precleared with protein A agarose beads (Santa Cruz) under constant mixing at 4°C for an hour. Samples were then normalized, added with 3 μg of primary antibody, rotated overnight at 4°C, and then immunoprecipitated with protein A agarose beads. Subsequently, beads were washed three times and then analyzed via western blot.

Cells were fixed on slides and incubated with the primary antibodies of interests at the desired dilutions, followed by incubation with 1:200 dilution of Alexafluor-488 anti-mouse secondary antibody and Alexafluor-568 anti-rabbit secondary antibody (Invitrogen). Images were captured with Olympus Fluoview-FV1000 confocal microscope.

Cells were mixed with 4 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-dimethyltetrazolium bromide (MTT; Sigma Aldrich) in a 96-well format and incubated for 2 hours at 37°C. The formazan crystals formed were then spun down, dissolved in dimethylsulfoxide, and quantitated spectrophotometrically at an absorbance wavelength of 570 nm.

Adherent cells were incubated with crystal violet solution for 10 minutes at room temperature. Nonviable cells were then washed off using distilled water and images were captured thereafter.

Site-directed mutagenesis of tyrosine-289 in B56δ to a phenylalanine residue was performed using the Quikchange II XL Site-directed Mutagenesis Kit provided by Stratagene.

All experiments were performed for at least 3 times, except for experiments involving clinical lymphoma biopsies. Numerical data were presented as mean ± standard deviation. The paired Student t test (2-tail, unequal variance) was used when comparisons were made between 2 groups. Statistical significance was set at P < .05. Refer to the detailed experimental protocols in the supplemental Experimental Procedures on the Blood Web site.

To evaluate the effect of O₂⁻ on the phosphorylation of Bcl-2, intracellular O₂⁻ was induced in Jurkat cells by inhibition of SOD1 with diethyldithiocarbamate (DDC; 50-600 μM). Increase in O₂⁻ (supplemental Figure 1A) augmented S70pBcl-2 in a dose-dependent manner (Figure 1A), whereas the other 2 phosphorylation sites (T69 and S87) remained unchanged. A similar increase in S70pBcl-2 was also observed upon small interfering RNA (siRNA)-mediated silencing of SOD1 using 4 independent siRNA sequences (Figure 1B). Conversely, transient overexpression of SOD1 (Figure 1C) as well as the exogenous addition of bovine SOD1 (Figure 1D), significantly reduced the level of S70pBcl-2. Similarly, scavenging intracellular O₂⁻ (tiron; 1-5 mM) abrogated DDC-induced phosphorylation of S70 (Figures 1E; supplemental Figure 1B). Of note, O₂⁻-induced phosphorylation of Bcl-2 at S70 is not exclusive to Jurkat cells, as evidenced in Hela cervical, MDA231 breast, and HK-1 nasopharyngeal cancer cell lines (supplemental Figure 1C-E).

Notably, as previously reported, an increase in intracellular O₂⁻ (DDC or siSOD1) significantly inhibited the antitumor activity of 2 commonly used chemotherapeutic agents, etoposide and doxorubicin (Figure 1F; supplemental Figure 1F-G). Conversely, overexpression of SOD1 or inhibition of O₂⁻ production via diphenyliodonium (DPI) pretreatment sensitized Jurkat cells to 5-fluorouracil (5FU) and cisplatin (Figure 1H); Jurkat cells are relatively refractory to 5FU and cisplatin (Figure 1F). So far, these data implicate S70 Bcl-2 phosphorylation in O₂⁻-induced drug resistance.^19-21 

To directly test if S70 phosphorylation of Bcl-2 is required for O₂⁻-mediated chemoresistance, we assessed the effect of transient expression of a phosphomimetic (S70E) or nonphosphorylatable (S70A) mutant of Bcl-2 on Jurkat cells sensitivity to etoposide. Whereas cells expressing phosphomimetic S70E Bcl-2 were significantly resistant to etoposide-induced cell death even in the absence of enhanced O₂⁻ production (Figure 1I), the expression of S70A Bcl-2 had the exact opposite effect and were no longer protected by enhanced O₂⁻. These data indicate that S70pBcl-2 is a crucial mediator of O₂⁻-induced chemoresistance in tumor cells.

Regulation of protein phosphorylation is achieved by a dynamic balance between the activities of specific kinases and phosphatases. Although the stress-activated MAP kinases are frequently implicated in the regulation of Bcl-2 phosphorylation,¹¹  neither c-Jun N-terminal kinase nor p38 were activated by increased O₂⁻ (supplemental Figure 2A). Although ERK1/2 was activated upon O₂⁻ induction in Jurkat cells, inhibition of ERK1/2 with PD98059 had no effect on S70 phosphorylation. These data suggest against the involvement of MAP kinases in O₂⁻-induced S70 Bcl2 phosphorylation (supplemental Figure 2B).

To evaluate if O₂⁻-induced upregulation of S70pBcl-2 could be due to an inactivation of Bcl-2 phosphatases, we tested if DDC treatment could affect the interaction of Bcl-2 with the catalytic subunits of 2 major Bcl-2 phosphatases, PP2A (PP2A-C) and PP1 (PP1-C). Although Bcl-2 co-immunoprecipitated with both PP2A-C and PP1-C (Figure 2A), only the interaction of Bcl-2 with PP2A-C was disrupted upon O₂⁻ induction. This was accompanied by a concomitant increase in S70pBcl-2 (Figure 2A). Similar findings were observed in Hela and MDA231 cell lines (Figure 2B) or when endogenous PP2A-C was immunoprecipitated (Figure 2C). Further supporting the hypothesis that Bcl-2 is downstream of PP2A, we found that pharmacological inhibition (okadaic acid) or activation (FTY720) of PP2A respectively enhanced or decreased the phosphorylation of Bcl-2 at S70 (supplemental Figure 2C). These data suggest that O₂⁻-induced upregulation of S70pBcl-2 could be due to the inhibition of PP2A-C mediated dephosphorylation of Bcl-2.

PP2A is reported to bind to Bcl-2 at the OMM.⁶  We confirmed that the A and C core subunits of PP2A localize to the OMM of purified mitochondria together with Bcl-2 (Figure 2D). Confocal microscopy using Mitotracker also demonstrated that majority of Bcl-2 is localized at the mitochondria (supplemental Figure 2D). We next tested if the mitochondrial localization of the AC core enzyme was affected by increased intracellular O₂⁻. Interestingly, in both Jurkat and MDA231 cells, mitochondrial PP2A-A and PP2A-C expression decreased following an increase in intracellular O₂⁻, concomitant with phosphorylation of Bcl-2 S70, and both effects were reversed by the O₂⁻ scavenger tiron (Figure 2E). These data were further corroborated by confocal microscopy, demonstrating that DDC treatment or SOD1 knockdown decreased the colocalization of PP2A-C with Bcl-2 (Figure 2F-G).

PP2A is a heterotrimer, with a core A and C heterodimer whose substrate specificity is determined by its B regulatory subunit.²²  B56α has been previously identified as a regulatory subunit responsible for PP2A-mediated dephosphorylation of Bcl-2.⁴  Because the B56 family of targeting subunits can be redundant and tissue-specific,^23-25  we tested the interaction of all PP2A-B56 subunits with Bcl-2 in our system. Co-IP assays revealed HA-B56δ as a major interacting partner of Bcl-2 in both Jurkat (Figure 3A) and MDA231 cells (supplemental Figure 3A). This interaction appears physiologic, because immunoprecipitation of endogenous Bcl-2 also pulled down endogenous B56δ (Figure 3B; supplemental Figure 3B). Consistent with the model that B56δ recruits PP2A to dephosphorylate Bcl-2, expression of HA-B56δ resulted in a decline in S70pBcl-2 level (Figure 3A), whereas silencing of B56δ resulted in increased S70pBcl-2 (Figure 3C). These data indicate that a B56δ-containing PP2A heterotrimer (PP2A_B56δ) is involved in the dephosphorylation of Bcl-2 at S70 in our model of study.

We next tested if increase in O₂⁻ affected the binding of B56δ to Bcl-2. Increasing intracellular O₂⁻ decreased the interaction of Bcl-2 with PP2A-A and PP2A-C subunits (Figure 3B,D), but not B56δ (Figure 3B,E). These data indicate that O₂⁻ inhibits the association of Bcl-2-bound B56δ from PP2A-AC heterodimer. Consistent with this, PP2A-C binding to both Bcl-2 and B56δ was reduced under similar conditions (Figure 3D,F), whereas the PP2A-AC heterodimer remained intact despite not being in association with B56δ (Figure 3D). In purified mitochondria, B56δ abundance was unaffected by siSOD1 despite a marked decrease in the amount of mitochondrial PP2A-A and -C (Figure 3G). Likewise, DDC treatment reduced the colocalization of Bcl-2 with PP2A-C but not with B56δ (Figure 3H). Taken together, the data suggest that increased intracellular O₂⁻ inhibits the recruitment of PP2A-AC heterodimer to Bcl-2-bound B56δ.

We sought to understand how increased O₂⁻ production might inhibit the assembly of the PP2A heterotrimer. O₂⁻ does not readily react with most biological molecules,²⁶  thus strongly suggesting the involvement of downstream products of O₂⁻ such as peroxynitrite (ONOO⁻) and H₂O₂. Because O₂⁻ production was induced by preventing the dismutation of O₂⁻ to H₂O₂, the involvement of H₂O₂ is unlikely. The alternative reaction of O₂⁻ with nitric oxide (NO) to form ONOO⁻ may therefore be crucial for O₂⁻-mediated signaling on PP2A.

The production of ONOO⁻ resulting from the reaction of O₂⁻ with NO consumes intracellular NO. Because there is no reliable commercial ONOO⁻ probe, we sought to measure the decrease of intracellular NO as a surrogate indicator of ONOO⁻ production upon O₂⁻ induction. Indeed, a decline in the fluorescence of the NO-sensitive probe, DAF-FM, was clearly detected upon an increase in O₂⁻ in tumor cells (Figure 4A; supplemental Figure 4A). Pretreatment with the O₂⁻ scavenger tiron abrogated the drop in NO induced by DDC in both cell lines, consistent with the generation of ONOO⁻.

We next investigated if ONOO⁻ was responsible for O₂⁻-mediated induction of S70pBcl-2 in tumor cells. O₂⁻-mediated phosphorylation of Bcl-2 at S70 was blocked by FeTPPS (a porphyrin-based catalyst of ONOO⁻ degradation) pretreatment in a dose-dependent manner (Figure 4B). Removal of ONOO⁻ by FeTPPS also sensitized Jurkat cells to 5FU and cisplatin (supplemental Figure 4E). Conversely, addition of exogenous ONOO⁻ caused a dose-dependent increase in S70pBcl-2 that peaked at 10 μM (Figure 4C). This was accompanied by a decline in the interaction between Bcl-2 and the PP2A-AC heterodimer (Figure 4D; supplemental Figure 4B), a drop in mitochondria localization of PP2A-A and -C (Figure 4E), and an augmented resistance against drug-induced apoptosis (supplemental Figure 4C-D). These data directly implicate ONOO⁻ as an essential intermediate in the O₂⁻-mediated induction of S70pBcl-2 in tumor cells.

We next sought to identify the precise biochemical modification by ONOO⁻ that resulted in the decreased interaction of the PP2A-AC heterodimer with Bcl-2-bound B56δ. We considered it unlikely that the core AC heterodimer enzyme would be a target of O₂⁻/ONOO⁻ because inhibition of global PP2A enzymatic activity would be highly toxic. We therefore examined if B56δ could be a target for nitrosative modification. The 2 most common protein modifications by ONOO⁻ are the S-nitrosylation of cysteine residues and the nitration of tyrosine residues.²⁶  We examined a multiple sequence alignment of various B56 isoforms (both mammalian and nonmammalian origins) for highly conserved cysteine or tyrosine residues that could serve as candidate nitrosative modification sites. Interestingly, we identified a highly conserved tyrosine residue (corresponding to Y289 of human B56δ; Figure 4F) that not only falls within the A subunit binding domain of the B56 subunits,²⁷  but also serves as 1 of the interacting residues between the B56 isoforms and the A-scaffolding subunit.^28,29  This tyrosine residue was an attractive target for additional study because nitrative modification of this residue would be predicted to block the interaction between PP2A-AC heterodimers and B56 subunits bound to specific targets such as Bcl-2.

Serendipitously, analysis of sequence adjacent to Y289 revealed that this tyrosine residue is also particularly prone to nitrative processes. Protein tyrosine nitration is governed by the local context of the tyrosine, and specific criteria that identify favorable sites have been defined (supplemental Figure 5C).^30,31  Notably, all of these criteria favor modification of Y289 in B56δ (Figure 4F; supplemental Figure 5A-B).

Based on these in silico findings, we tested if B56δ was tyrosine-nitrated in an ONOO⁻-dependent manner. Indeed, 3-nitrotyrosine (3-NT) was detected in B56δ immunoprecipitated from Jurkat and Hela lysates and was markedly augmented upon an increase in O₂⁻ or ONOO⁻ treatment (Figure 4G; supplemental Figure 5D). Importantly, O₂⁻/ONOO⁻-induced B56δ tyrosine nitration was always paralleled by a decline in B56δ-PP2A-C interaction and an increase in S70pBcl-2, whereas the interaction between B56δ and Bcl-2 remained constant despite the redox modification of B56δ (Figure 4G; supplemental Figure 5D). In line with this, FeTPPS pretreatment blocked the increase in DDC-induced B56δ tyrosine nitration, restored the binding of B56δ to PP2A-C, and allowed for the dephosphorylation of S70 of Bcl-2 (Figure 4H). Similar results were obtained when DPI was used to block the increase in intracellular O₂⁻ levels induced by SOD1 knockdown (Figure 4I; supplemental Figure 5E). Conversely, overexpression of SOD1 resulted in a decrease in B56δ tyrosine nitration accompanied by an increase in the formation of PP2A-AC-B56δ heterotrimers and dephosphorylation of S70pBcl-2 (Figure 4J). The effect of O₂⁻ on PP2A_B56δ holoenzyme assembly is also reflected by the changes in its phosphatase activity. DDC treatment resulted in a decline in PP2A_B56δ activity, whereas pretreatment with DPI or FeTPPS ameliorated the effect of DDC (supplemental Figure 5F). Collectively, these data indicate that O₂⁻-induced S70 phosphorylation of Bcl-2 is driven by ONOO⁻-mediated tyrosine nitration of B56δ and the consequential impairment of PP2A holoenzyme assembly.

To further investigate if Y289 of B56δ (B56δ^Y289) was indeed nitrated as predicted by our in silico analysis, we generated a custom antibody specifically raised against a synthetic peptide comprising nitrated B56δ^Y289 and flanking sequence (supplemental Figure 6A). Affinity-purified antibody detected a single prominent band corresponding to the molecular mass of B56δ (69.9 kDa) (Figure 5A), indicating that B56δ^Y289 was nitrated in Jurkat cells. More importantly, nitrated B56δ^Y289 was markedly enhanced by increasing intracellular O₂⁻ or ONOO⁻ and paralleled the abundance of S70pBcl-2 (Figure 5A). Moreover, degradation of ONOO⁻ by FeTPPS abrogated O₂⁻-induced B56δ^Y289 nitration. However, inhibition of O₂⁻ production did not affect the ONOO⁻-induced nitration of B56δ^Y289 nor the upregulation of S70pBcl-2, further supporting the model that ONOO^- is downstream of O₂⁻ in the modification of B56δ and regulation of S70 Bcl-2 phosphorylation.

We next asked if nitration of Y289 is required for O₂⁻-induced upregulation of S70pBcl-2. Wild-type (WT-B56δ) or Y289F mutant (Y289F-B56δ) HA-B56δ (supplemental Figure 6B) was expressed in Jurkat cells and the response to O₂⁻ assessed. Overexpression of WT-B56δ resulted in a modest delay in DDC-induced S70 Bcl-2 phosphorylation compared with controls, whereas expression of Y289F-B56δ completely abolished O₂⁻-induced S70 phosphorylation (Figure 5B). In addition, we tested if Y289F mutation of B56δ could reverse the effects of O₂⁻ on PP2A. Expression of Y289F-B56δ blocked O₂⁻ or ONOO⁻-induced nitration of B56δ, and at the same time blocked phosphorylation of Bcl-2 at S70 (Figure 5C,E; supplemental Figure 6C). O₂⁻ reduced the interaction of B56δ with PP2A-C, and this was prevented by expression of Y289F-B56δ (Figure 5C, compare lane 6 with lane 8). Furthermore, the protective effect of O₂⁻ on drug-induced cell death is alleviated in cells expressing Y289F-B56δ (Figure 5D,F). These data confirm B56δ^Y289 nitration as a critical step in O₂⁻-mediated inhibition of both PP2A_B56δ holoenzyme assembly and the Bcl-2 phosphatase activity that accompanies it.

To assess the in vivo relevance of these findings, we analyzed the expression and phosphorylation status of relevant proteins in primary patient-derived lymphoma samples. Consistent with the in vitro findings, S70pBcl-2 levels were inversely related to SOD1 abundance (Figure 6A-B; supplemental Figure 7). There were, however, 2 outliers, P1 and P3, in which S70pBcl-2 was observed despite the high expression of SOD1. Notably, P1 and P3 lack B56δ expression, which may explain why S70 is not dephosphorylated.

Of the 14 primary lymphoma samples tested, 5 (denoted with *) were derived from tumors that were sufficiently large for further experimental analysis via co-IP assay. Using these lymphoma biopsies, we asked if B56δ was tyrosine nitrated in primary lymphoma tumors and if this correlated with S70pBcl-2. As predicted, a robust interaction between B56δ and Bcl-2 was detected in all 5 samples (Figure 6C). B56δ tyrosine nitration was also detected in tumor biopsies derived from P9 and P10, and to a lesser extent in tumors P6 and P7; this correlated with the level of SOD1. Importantly, the highly nitrated B56δ in P9 and P10 interacted poorly with PP2A-C (Figure 6C), which correlated with increased S70pBcl-2 in the immunoblots in Figure 6A. In 2 Hodgkin lymphoma samples (P6 and P7) with abundant SOD1, B56δ nitration was low, and B56δ interacted with the PP2A-C subunit. P4 was an exception. B56δ was unable to bind PP2A-C despite being poorly nitrated, suggesting a potential heterogeneity in the regulation of PP2A subunits in primary tumors. In P4, it is plausible that B56δ may have lost its initial ability to bind PP2A-C, rendering B56δ tyrosine nitration less critical in the regulation of S70pBcl-2. Whether this phenomenon results from an inherent mutation on B56δ or PP2A-C warrants further investigation. We also tested if the suppression of SOD1 activity elicited a similar signaling response in primary cells from a mantle lymphoma. As predicted, PP2A-C coprecipitated with Bcl-2 and treatment with DDC blocked the interaction and augmented the S70pBcl-2 level (Figure 6D).

To check if B56δ^Y289 nitration is also synchronized with the inverse status of S70pBcl-2/SOD1, we analyzed the levels of nitrated B56δ^Y289, S70pBcl-2, and SOD1 in 6 additional (P15-P20) primary lymphoma samples. Consistent with our in vitro findings, high levels of nitrated B56δ^Y289 correlated well with elevated S70pBcl-2 and low SOD1 (P16, P18, and P20), whereas the converse is true for samples with low nitrated B56δ^Y289 level (P15, P17, and P19) (Figure 6E). Of note, nitrated B56δ^Y289 could not be detected in 3 other nonmalignant cell lines, whereas little or no S70pBcl-2 was present in these cell lines as well (Figure 6E).

Last, we calculated the ratio of S70 pBcl-2 to SOD1 for P1-P20 and correlated it with the respective prognosis of each lymphoma patient (supplemental Figure 7). Interestingly, we found that in all 5 primary lymphomas with S70pBcl-2:SOD1 ratio greater than 1, the patient had a poor prognosis and suffered from a relapse, whereas this was true in only 3 (of 11) patients with a ratio lower than 1 (Figure 6B,F). These data highlight the potential for using both S70pBcl-2 and SOD1 as prognostic markers in clinical lymphomas. In summary, our study supports the hypothesis that O₂⁻/ONOO⁻ production stimulates B56δ^Y289 nitration leading to an increase in S70 Bcl-2 phosphorylation that in turn promotes tumor chemoresistance.

ROS-governed cellular processes are highly intricate in nature, often involving a well-orchestrated series of events that necessitate the tight regulation of numerous parameters. Likewise, ROS-dependent cell fate processes require a delicate balance of different intracellular ROS species, deregulation of which often leads to uncontrolled cell proliferation and cancer. Here, we describe a novel mechanism whereby an increase in intracellular O₂⁻ induced by either pharmacological inhibition or genetic knockdown of SOD1 promotes chemoresistance by augmenting the phosphorylation (and activation) of Bcl-2 at S70. By preventing SOD1-mediated conversion of O₂⁻ to H₂O₂, O₂⁻ could react more readily with NO to form ONOO⁻. This led to the inhibitory nitration of the Bcl-2 phosphatase, PP2A_B56δ, and the consequential increase in S70pBcl-2 level. Moreover, we identified a nitration-prone tyrosine residue on the B56δ regulatory subunit of PP2A (B56δ^Y289) that is also a known interacting site between B56δ and the A scaffolding subunit of PP2A. The nitration of the tyrosine residue impeded the association between the catalytic core and the B56δ-Bcl-2 complex and inhibited the recruitment of the AC core to the mitochondria. As a result, the dephosphorylation of Bcl-2 at S70 was inhibited, and the accumulation of phosphorylation-activated Bcl-2 ensued. These findings reveal a novel mechanism that enables cancer cells to harness the pro-survival properties of O₂⁻ to acquire a resistant phenotype (Figure 7).

The regulatory role of ROS in cell death and proliferation processes is akin a double-edged sword. Although a mild pro-oxidant state dominated by O₂⁻ anions creates an intracellular milieu that favors cell survival,^19,20,32  excessive production of intracellular ROS triggers cell death via the oxidative activation of various death effectors.³³  There is a dynamic interplay among the various ROS in the regulation of cell death processes. For example, we found that a tilt in cellular redox equilibrium in favor of O₂⁻ anions facilitated the formation of ONOO⁻ via the radical fusion of O₂⁻ and NO. The resultant modest increase in ONOO⁻ drives nitrosative modification specifically of nitration-labile tyrosine residues. The consequences of increased ONOO⁻ are dose-dependent; low or moderate levels of ONOO⁻ will only affect nitration-sensitive targets, whereas higher levels of ONOO⁻ cause widespread nitration of less reactive targets as well. In fact, more is not always better: others have found that high levels of exogenous ONOO⁻ (>75 μM) can nitrate and activate the catalytic subunit of PP2A.^34,35  This could account for the biphasic effect of ONOO⁻ observed in Figure 4C and supplemental Figure 4D, where initial increase in S70pBcl-2 and cell viability at low doses of ONOO⁻ was followed by falling Bcl-2 phosphorylation and cell death at high-dose ONOO⁻.

O₂⁻-induced chemoresistance may provide the rationale for a chemotherapeutic strategy. S70 Bcl-2 phosphorylation predicts chemoresistance.^9,36  In these cases where Bcl-2 is activated by O₂⁻, suppression of O₂⁻ might enhance the efficacy of conventional cytotoxic agents. Indeed, the idea of redox modulation as a chemotherapeutic strategy has been suggested by studies demonstrating that O₂⁻ suppressing agents such as DPI and Tempol (SOD1 mimetics) can bypass the antiapoptotic activity of Bcl-2 and sensitize tumors to drug-induced apoptosis.^13,37  Pharmacological strategies that involve the suppression of intracellular O₂⁻ level may be included in the future in the design and development of effective combination therapies.

The identification of O₂⁻-driven nitration of PP2A_B56δ and changes in S70pBcl-2 may also provide valuable clinical biomarkers of redox-driven neoplastic processes. The marked inverse correlation between the abundance of SOD1 and the amount of both S70pBcl-2 and B56δ nitration in clinical lymphoma biopsies is a potential biological signature that could be exploited for prognostic and therapeutic purposes. For instance, the detection of lymphomas with low SOD1 but high S70pBcl-2 and nitrated B56δ might indicate greater drug resistance, and suggest the inclusion of redox modulators in the chemotherapeutic regime. Therefore, our findings could provide the basis of a novel approach for stratifying clinical lymphomas based on tumor aggressiveness and resistance to chemotherapy.

The authors thank Dr Jayshree Hirpara, Dr Goh Boon Cher, and Ms Goh Han Lee for technical assistance on the primary lymphoma samples and Mr Stephen Chong for the synthesis of the Bcl-2 mutant plasmids.

This work is supported by grants from the National Medical Research Council of Singapore (grant numbers NPRC10/NM06M and NMRC/1241).

Contribution: I.C.C.L. designed and performed the experiments, conceived the study, analyzed the data, and wrote the manuscript; T.L. provided the clinical materials and surgical specimen; Y.H. assisted in the prognostic scoring of lymphoma biopsies; D.M.V. provided anti-B56δ and anti-PP2A-C antibodies as well as the HA-tagged B56 subunit plasmid constructs, provided critical comments, and wrote the manuscript; and S.P. is the lead principal investigator, provided funding for the study, conceived the study, analyzed the data, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Shazib Pervaiz, Department of Physiology, Yong Loo Lin School of Medicine, National University of Singapore, 2 Medical Dr, MD9 #01-05, Singapore 117597; e-mail: phssp@nus.edu.sg.