Low catalase expression confers redox hypersensitivity and identifies an indolent clinical behavior in CLL

Chronic lymphocytic leukemia (CLL), the most prevalent leukemia in Europe and North America, is characterized by the expansion of a population of mature B cells that accumulate in the bone marrow, lymphoid tissues, and blood.^1,2  Clinical presentation, natural course of the disease, and response to treatment are all extremely variable, with some patients having indolent disease and others experiencing a more accelerated course, treatment resistance, and a dismal outcome.^3,4  Several biological parameters have been shown to be associated with clinical outcomes in patients with CLL, including the presence or absence of somatic mutations within the immunoglobulin variable heavy chain genes (IGHV), specific chromosomal abnormalities and gene mutations, the expression of the ZAP70 tyrosine kinase, and CD38 antigen.⁵  Although the reason for this clinical disparity is not fully understood, B-cell receptor (BCR) signaling, which is central to B-cell development, is considered a key determinant of variable clinical behavior and is a target for therapeutic interventions in CLL.^6-10 

BCR stimulation by antigen induces increase of intracellular calcium; activation of receptor proximal protein tyrosine kinases, such as spleen tyrosine kinase (SYK); and the tyrosine phosphorylation of several proteins of the BCR signaling network.¹¹  However, BCR stimulation also recruits negative regulators such as protein tyrosine phosphatases (PTPs) that counterbalance protein kinase activity.^12,13  Initiation, transmission, and strength of the BCR signaling are indeed regulated by the interplay between kinases and PTPs, and the full activation of the signaling cascade requires not only activation of kinases but also inhibition of PTPs.^12,13  PTPs may be reversibly inhibited through oxidation of the catalytic cysteine by hydrogen peroxide (H₂O₂), which is the primary reactive oxygen species (ROS) produced by B cells. Hydrogen peroxide is, in turn, generated by calcium-dependent NADPH oxidases (NOX), which are recruited and activated in B cells after BCR stimulation.^12-15  Thus, BCR-induced hydrogen peroxide acts as a positive feedback that plays an important role in the amplification of the BCR signal.

Increased ROS levels have been detected in various cancers, where they activate protumorigenic signals; enhance cell survival, proliferation, and chemoresistance; and cause DNA damage and genetic instability.^16-18  However, escalated levels of ROS can also promote antitumorigenic signals, resulting in an increase of oxidative stress and induction of cancer cell death.^16,19-21  For this reason, therapies aimed at either reverting the increased ROS levels or further escalating ROS may be potentially effective against cancer.²² 

CLL cells accumulate higher levels of ROS than normal B cells.²³  ROS levels are extremely variable across patients’ samples, and higher ROS levels are associated with favorable prognostic features and a slower disease progression.²⁴  Moreover, augmented levels of ROS confer increased sensitivity to anticancer agents that induce apoptosis in leukemia cells.²⁵  Altogether, escalated levels of ROS seem to account for a lesser aggressive behavior of CLL cells. However, it is still unclear how CLL B cells sense and respond to redox cues and what effect they have on disease.

In this study, we used phosphospecific flow cytometry to characterize redox signaling sensitivity in prognostic subsets of leukemic cells and to investigate possible mechanisms regulating response of leukemic cells to redox cues in CLL.

Peripheral blood mononuclear cell (PBMC) samples from 42 untreated patients with CLL and from 9 age-matched healthy donors (HDs) were collected and cryopreserved at the Hematology Unit, Azienda Ospedaliera Universitaria Integrata, Verona, Italy, on approval from the local Ethics Committee (Comitato Etico per la Sperimentazione, Azienda Ospedaliera Universitaria Integrata). In accordance with the Declaration of Helsinki, all patients and donors provided written informed consent. Details on inclusion criteria and characteristics of patients at diagnosis are summarized in supplemental Methods and supplemental Table 1, available on the Blood Web site.

PBMCs were isolated by Ficoll-hypaque centrifugation (Lymphoprep; Nicomed, Oslo, Norway) and stored in liquid nitrogen. Cell viability assessment and B-cell purification are detailed in supplemental Methods.

After 24-hour rest at 37°C, PBMCs were treated in bulk for 10 minutes with 3.3 mM H₂O₂ (Sigma-Aldrich, Milan, Italy) and/or 20 μg/mL goat F(ab′)₂ anti-human immunoglobulin M (IgM; SouthernBiotech, Birmingham, AL), as detailed in supplemental Methods. H₂O₂ efficacy in inhibiting PTPs was assessed using a tyrosine phosphatase assay (supplemental Information). When indicated, before modulation with 3.3 mM H₂O₂, PBMCs were incubated with the enzyme inhibitors 3-amino-1,2,4-triazole (ATZ; 1 mM), mercaptosuccinic acid (MCA; 100 μM), and sodium diethyldithiocarbamate (DDTC; 70 μM; all from Sigma-Aldrich) for 24 hours at 37°C; the antioxidant N-acetylcysteine (NAC; 5 mM), exogenous catalase (1000 U/mL; both from Sigma-Aldrich), or the SYK inhibitor PRT-060318 (2 μM; Selleck Chemicals, Munich, Germany) for 30 minutes.

Phosphospecific flow cytometry was performed as previously described.^7,8  Briefly, cells were fixed and permeabilized with PerFix Expose kit (Beckman Coulter, Miami, FL), according to the manufacturer’s protocol. Permeabilized cells were stained with fluorochrome-conjugated antibodies. For staining phosphoproteins, the following antibodies were used: anti-pSYK-PECy7 (pY352)/ZAP70 (pY319) (clone 17A/P-ZAP70), anti-pp38-PE (pT180/pY182, clone 36/p38), anti-pNF-κB-PECy7 p65 (pS529, clone K 10-895.12.50), anti-pJNK-AlexaFluor647 (pT183/pY185, clone N9-66; all from BD Biosciences, San Jose, CA), and anti-pERK1/2-AlexaFluor488 (pT202/pY204, clone D13.14.4E; Cell Signaling Technology, Danvers, MA). The complete list of antibodies used is summarized in supplemental Table 2. Although antibody against pSYK recognizes the activating tyrosine residue in SYK (pY352) and ZAP70 (pY319), in CLL, phosphorylation of the activating tyrosine in ZAP70 is highly inefficient and negligible when compared with phosphorylation of SYK.²⁶  Therefore, we can assume that phosphorylation events recognized in our study mostly refer to SYK phosphorylation at Y352 site.

Approximately 30 000 gated events were acquired for each sample on a FACSCanto II cytometer (Becton Dickinson, Franklin Lakes, NJ). Samples were run in duplicate. Flow cytometry data processing and analysis are described in supplemental Methods.

Real-time quantification for CAT (catalase), GPX1 (glutathione peroxidase1), SOD1 (Cu,Zn-superoxide dismutase), SOD2 (Mn-superoxide dismutase), and gp91phox (Nox subunit) was performed as detailed in supplemental Methods.

Protein expression of catalase, GPX, CuZnSOD, and MnSOD was determined using monoclonal antibodies and flow cytometry, as described in supplemental Methods.

Levels of reduced thiols were detected by ThiolTracker Violet (Thermo Fisher Scientific), according to the manufacturer’s instructions, as described in supplemental Methods.

Cell antioxidant capacity was measured in cell lysates (5 × 10⁶ cells), using a colorimetric Antioxidant Assay Kit (Cayman Chemical, Ann Arbor, MI), following the manufacturer’s instructions.

Total and mitochondrial ROS were, respectively, measured using 2′,7′-dichlorodihydrofluorescein diacetate (Thermo Fisher Scientific) or MitoSOX Red mitochondrial superoxide indicator (Thermo Fisher Scientific), according to the manufacturer’s instruction, as described in supplemental Methods.

Mitochondria were labeled with MitoTracker Green (Thermo Fisher Scientific), according to the manufacturer’s instruction, as described in supplemental Methods.

Student t test, 2-sample Wilcoxon signed-rank sum test, Mann-Whitney U test, or Kruskal-Wallis test were used as appropriate. Time-to-first-treatment (TTFT) was calculated from the date of diagnosis to the date of initial therapy.^4,27  Patients who did not receive any treatment during follow-up were censored at their last follow-up date. TTFT curves estimated using the Kaplan-Meier method were compared using the log-rank (Mantel-Cox) test. Time-to-events models for TTFT were generated using Cox proportional hazards regression, and the proportional hazard assumption was tested using the Schoenfeld’s test. Cutoffs for grouping patients on the basis of signaling sensitivity to H₂O₂ and enzyme expression were calculated as described in supplemental Information.

Differences were considered statistically significant for P values ≤ .05. Graphing and statistical analyses were performed using GraphPad Prism software (v.7.03; GraphPad Software Inc., La Jolla, CA). Gaussian kernel density estimates and univariate/multivariate time-to-event analyses were performed using Mathematica software (v.11.1.1.0; Wolfram Research Inc., Champaign, IL).

The phosphorylation levels of 5 proteins downstream of the BCR signaling (namely, SYK, ERK1/2, p38, NF-κB p65, and JNK) were analyzed at the single-cell level in 42 CLL cell samples, using phosphospecific flow cytometry. Phosphorylation of the BCR signaling proteins was measured in the basal condition (ie, unmodulated) and under stimulation with H₂O₂ (ie, H₂O₂ modulated). Circulating B cells from HDs were analyzed as controls (Figure 1A). Although the basal phosphorylation statuses of signaling proteins were highly heterogeneous among CLL samples (σ² ≥ 1.25 for pSYK, pp38, and pNF-κB p65; Figure 1B), CLL B cells showed more elevated average basal levels of pSYK, pp38, and pNF-κB p65 than HD B cells (Figure 1B). H₂O₂ induced a significant increase in the average phosphorylation of ERK1/2 and p38 in both HD and CLL B cells, whereas a significant increase of SYK phosphorylation was observed only in B cells from patients with CLL (Figure 1A,C). Remarkably, redox sensitivity, measured as the log2-fold change in phosphorylation between H₂O₂-modulated and unmodulated conditions, was significantly higher in CLL than HD B cells for SYK, ERK1/2, p38, and NF-κB p65 (Figure 1D). Redox signaling responses were unrelated to basal phosphorylation levels for each signaling protein except JNK (supplemental Figure 5). Pretreating CLL cells with the antioxidant NAC before stimulation with H₂O₂ resulted in a significant reduction of redox signaling sensitivity (Figure 1E; supplemental Figure 6).

To assess whether phosphorylation events were occurring simultaneously in each cell, we evaluated in individual B cells the relationships between phosphorylation responses that significantly increased on H₂O₂ modulation (ie, pSYK, pERK1/2, pp38). H₂O₂ induced the coordinated phosphorylation of ERK1/2 and p38 only in a fraction of the HD cell population, whereas responses of pERK1/2 and pp38 to H₂O₂ occur simultaneously with the upstream phosphorylation of SYK in most individual CLL cells (Figure 1F). SYK was phosphorylated in concert with ERK1/2 and p38 across individual CLL samples (supplemental Figure 7), thus suggesting that H₂O₂ triggers an interconnected network of signaling pathways that is common across patients with CLL. Therefore, CLL cells exhibit a specific redox sensitivity pattern with an overall higher and more heterogeneous redox response than HD B cells.

We examined the relationship between redox sensitivity and BCR signaling, measured as ERK1/2 phosphorylation in response to anti-IgM modulation, in the CLL subsets characterized by different redox sensitivity. ERK1/2 response was expressed as the log2-fold change in phosphorylation between anti-IgM-modulated and unmodulated conditions.⁷  An overall H₂O₂ sensitivity index for each patient with CLL was obtained by summing the log2-fold change response to H₂O₂ for pSYK, pERK1/2, and pp38, which significantly increased in CLL cells on H₂O₂ modulation (Figure 1A,C). The median of the new distribution was used to discriminate between high and low H₂O₂-sensitive CLL. As shown in supplemental Figure 8, no differences were detected between anti-IgM ERK1/2 response and H₂O₂ sensitivity.

To assess whether H₂O₂ could perturb the signaling response induced by BCR engagement, we measured ERK1/2 response to anti-IgM in the presence or absence of H₂O₂. BCR-responder and BCR-nonresponder CLL was referred to the median value of the ERK1/2 response distribution. The presence of H₂O₂ significantly potentiated signaling response of CLL cells to anti-IgM modulation in BCR-nonresponder, as well as the BCR-responder CLL (Figure 2), thus extending data from previous studies.^28,29 

To investigate whether CLL cells from different prognostic classes could differentially respond to external redox cues, redox sensitivity was considered in relationship to standard prognostic parameters. Phosphorylation response of SYK to H₂O₂ was significantly higher in the patient subsets defined by M-IGHV, CD38, and ZAP70-negative expression (Figure 3A). Moreover, redox pSYK and pp38 hypersensitivity was associated with early-stage (Binet stage A) disease, whereas redox pSYK and pERK1/2 hypersensitivity was more frequently detected in patients who did not require treatment during the follow-up (Figure 3A).

To assess the potential link between redox sensitivity and disease behavior, measured as TTFT, we examined time-to-event modeling using signaling data in response to H₂O₂, IGHV status, ZAP70, and CD38 expression; cytogenetic alterations; and Binet clinical stage in 41 patient samples. Univariate time-to-event analysis showed that low pSYK (P = .0025) and pERK1/2 (P = .0436) responses to H₂O₂ (Figure 3B), UM-IGHV (P< .0001), ZAP70-positive (P = .0050), and Binet stage B-C (P< .0001) were significantly associated with shorter TTFT. Consistent with Myklebust et al,³⁰  grouping patients into low and high responses to H₂O₂ on the basis of the sum of pSYK, pERK1/2, and pp38 responses allowed a better stratification of different TTFT than using individual phosphoproteins (Figure 3B). Comprehensive multivariate hazards models using significant parameters, as resulted from the univariate analysis, showed that pSYK or pERK1/2 responses to H₂O₂ do not influence clinical outcome independent of standard prognostic parameters (data not shown). Then, we investigated a possible role for H₂O₂-induced pSYK in promoting antitumor signals, such as cell death. Analysis of cell viability after treatment with H₂O₂ in the presence of the SYK inhibitor PRT-060318 showed that SYK inhibition does not influence H₂O₂-induced cell death (supplemental Figure 9)

The next step was to investigate potential mechanisms underlying different redox sensitivity in CLL cells.

We hypothesized that redox hypersensitivity could be the result of an increased accumulation of exogenous H₂O₂ within the cells, which would induce a higher inhibition of phosphatases with the consequent shift of the enzymatic balance toward phosphorylation. One possibility is that higher accumulation of exogenous H₂O₂ is a consequence of reduced antioxidant capacity of the cells. Therefore, we analyzed major cellular antioxidant systems, enzymatic and nonenzymatic, and compared them between the 2 CLL subsets characterized by different redox sensitivity, measured by the sum of pSYK, pERK1/2, and pp38 H₂O₂ responses. Samples exhibiting high sensitivity to H₂O₂ expressed lower levels of catalase and glutathione peroxidase (GPX), which convert hydrogen peroxide to molecular oxygen (Figure 4A). In contrast, expression levels of superoxide dismutase, both copper-zinc (CuZnSOD) and manganese (MnSOD) types, which catalyze the dismutation of the superoxide anions to hydrogen peroxide, as well as thiols, which represent the major nonenzymatic antioxidant scavenging ROS, showed no significant differences between CLL cells with different redox sensitivity (Figure 4A). Overall, antioxidant capacity of high H₂O₂-sensitive CLL appeared decreased by trend (Figure 4A). The expression of catalase, GPX, CuZnSOD, and MnSOD was confirmed at protein level (Figure 4B).

We compared catalase or GPX with IGHV status and detected that levels of catalase and GPX were significantly lower within CLL with M-IGHV (Figure 5A). Moreover, both catalase and GPX levels were positively correlated with BCR signaling, measured as pERK1/2 response to BCR modulation (Figure 5B). Then, we investigated potential correlations between catalase or GPX expression levels and disease behavior. As shown in Figure 5C, Kaplan-Meier curves showed that low levels of catalase and GPX were significantly associated with a longer TTFT.

These data suggest that differences in the antioxidant machinery may underlie divergent redox signaling sensitivity of BCR proteins, as well as clinical behavior in patients with CLL.

To assess a role for catalase and GPX in the redox sensitivity of CLL cells, first we blocked activity of catalase, GPX, or CuZnSOD in CLL cells before stimulation with H₂O₂. Cell treatment with ATZ, which specifically inhibits catalase activity, resulted in a further increase of protein phosphorylation without affecting signal transduction induced by BCR engagement (supplemental Figure 10). In contrast, blocking GPX with MCA induced a slight phosphorylation reduction. Inhibition of CuZnSOD with DDTC before H₂O₂ treatment had no effects on H₂O₂-induced phosphorylation of the BCR signaling proteins (Figure 6A-B). Remarkably, preincubation with exogenous catalase before the addition of H₂O₂ completely reversed protein phosphorylation induced by H₂O₂ (Figure 6C-D).

These findings establish a functional link between catalase and redox sensitivity and support the hypothesis that a lower catalase activity, a condition that would determine the accumulation of exogenous hydrogen peroxide within leukemic cells, can determine redox hypersensitivity in CLL.

In addition to determining redox hypersensitivity, a reduced antioxidant activity might also induce a higher accumulation of endogenous ROS in leukemic cells. To assess a possible association between redox sensitivity and ROS levels, we compared cellular and mitochondrial levels of ROS in the patient groups defined by high and low sensitivity to H₂O₂. Total cellular levels of ROS, but not specific mitochondrial superoxide, were significantly increased in high H₂O₂-sensitive CLL cells (Figure 7A-B). Mitochondria and the NOX family of enzymes are the best-characterized intracellular sources of ROS¹⁵ ; therefore, to identify the possible source of ROS, we compared mitochondrial amounts and NOX catalytic subunit (gp91) expression in the patient groups defined by high and low sensitivity to H₂O₂. We detected higher amounts of mitochondria in high H₂O₂-sensitive CLL cells (Figure 7C) while observing a trend toward association between high levels of gp91expression and high-H₂O₂-sensitivity CLL (Figure 7D).

This study shows that sensitivity of BCR signaling proteins to redox cues is of clinical relevance in CLL, as redox hypersensitivity is associated with a lesser aggressive behavior of the disease. We identified low catalase expression as a possible mechanism accounting for redox hypersensitivity. The key advances of this study are defining redox hypersensitivity linked to lower antioxidant capability as intrinsic characteristic of CLL with an indolent clinical course, and gaining deeper insights into mechanisms controlling redox and signaling heterogeneity of CLL.

According to previous reports,^7,29  our data document a tonic, antigen-independent signaling in CLL.^30-32  Inactivation of phosphatases by H₂O₂, the redox species that can be endogenously produced in various cells and tissues in response to diverse signals,³³  increases the levels of tonic BCR signaling in B cells from patients with CLL, indicating that negative regulatory phosphatase activity is generally required to control tonic BCR signaling in leukemic cells. Redox sensitivities are reverted by the antioxidant NAC, thus supporting a mechanistic link between ROS and phosphorylation of BCR proteins. Phosphorylation responses of ERK1/2 and p38 to H₂O₂ occur simultaneously with the upstream phosphorylation of SYK in each CLL cell, as well as across individual CLL. Although a kinetic analysis of signaling would be necessary to reveal the temporal sequence of signaling protein activation, these data suggest that downstream signals triggered by H₂O₂ may depend on the level of SYK phosphorylation. As in previous studies on primary healthy B cells,^29,34  the same level of H₂O₂ is ineffective in phosphorylating SYK in normal B cells, whereas it induces minimal activation of ERK1/2 and p38 signaling proteins. Remarkably, in CLL cells, H₂O₂ potentiates signal transduction triggered by the BCR engagement. Taken together, these results suggest that exogenous H₂O₂ interrupts phosphatase activity and perturbs tonic as well as induced BCR signaling in CLL.

CLL is characterized by a highly variable natural clinical course.²  Multiple defects in the dysregulated response to signals from the microenvironment can confer a selective survival advantage to CLL cells with a worse prognosis.³⁵  The elucidation of mechanisms regulating these responses is of preeminent importance to understand CLL pathogenesis and to develop strategies for improving clinical management of the disease. In this study, we show that signaling sensitivity to H₂O₂ is highly variable among patients with CLL, and that redox hypersensitivity identifies a subgroup of treatment-naive patients with mutated IGHV genes and a slower clinical progression. Intriguingly, CLL with mutated IGHV derives from a postgerminal center B-cell subset,³⁶  which has been shown to exhibit redox signaling hypersensitivity.³⁷  These findings suggest that mechanisms underlying redox sensitivity could be still functional in M-CLL despite oncogenic transformation. Moreover, our findings are in agreement with previous data demonstrating that in CLL, in vitro apoptotic proficiency, in response to chemotherapeutic agents, is associated with a higher signaling response to H₂O₂.²⁹  Taken together, these data point to redox hypersensitivity as a feature of CLL with a lower selective growth advantage.

SYK hyperactivity has been described as inducing cell death in acute lymphoblastic leukemia cells³⁸  and acting synergistically with therapy-induced ROS in mediating apoptosis in CLL.³⁹  In contrast, we document that SYK activation does not affect CLL cell death induced by H₂O₂. High H₂O₂ concentration in our experimental setting may overcome a possible contribution of SYK. However, further investigations are needed to address the mechanistic role of SYK hyperactivation in CLL cell death and clinical behavior.

Differential responsiveness of BCR proteins evoked by H₂O₂ in the absence of BCR engagement is associated with divergent expression levels of catalase and GPX across patients with CLL. Lower levels of catalase and GPX expression are more frequently detected in patients with favorable prognostic parameters (M-CLL) and a slower progression of the disease. In contrast, high levels of catalase are detected in CLL with a more aggressive disease. Consistently, increased catalase expression has been observed in some cancers^40-42  and in HL-60 cancer cells, where it accounts for low sensitivity to H₂O₂ exposure.^43,44  However, catalase downregulation also has been detected in cancer and has been associated with proliferation, migration, and invasion in cancer cells.^45,46  Interestingly, a dual role in carcinogenesis has been also reported for GPX.⁴⁷  However, only catalase inhibition increases the phosphorylation response of leukemic cells to H₂O₂. Moreover, exogenous catalase completely reverses protein phosphorylation induced by H₂O₂. Taken together, these results point to catalase as a major element of the antioxidant machinery responsible for regulating redox signaling sensitivity in CLL cells. Theoretically, lower catalase activity may account for an escalated accumulation of H₂O₂ within leukemic cells, with the consequent greater inhibition of phosphatases and signaling increase. In contrast, higher catalase activity can be responsible for a greater H₂O₂ conversion and a lower, if any, effect on signaling. Differential catalase expression in CLL supports the existence of 2 main disease subtypes characterized by a disparity in clinical outcome, probably as a consequence of differences not only in underlying genetic lesions, epigenetic changes, activated signaling pathways, and interactions with the microenvironment, but also in the redox machinery.

Although some studies are shedding light on the regulation of catalase at both transcriptional^45,48  and activity levels, such as the p53–catalase direct interaction,⁴⁹  the precise molecular mechanisms controlling its expression and activity are still poorly understood. Future studies aimed at deciphering the signaling pathways and the molecular mechanisms that regulate catalase expression and activity in CLL could be of crucial relevance for the development of therapies targeting redox pathways.

In the subset of patients with CLL characterized by redox hypersensitivity and a lesser aggressive disease, in addition to detecting low catalase and GPX levels, we also document an elevated accumulation of cellular ROS. Consistently, downregulation of antioxidant enzymes in cancer is mainly associated with a high production of H₂O₂.^50-52  Moreover, our data are in accordance with a recent study showing that higher levels of ROS in CLL cells are associated with a slower disease progression.²⁴  One possibility is that catalase downregulation induces escalated levels of ROS that promote antitumor signals such as cell death or susceptibility to apoptosis in CLL, thus accounting for less aggressive behavior of cancer cells. Accordingly, Linley et al identified a significant inverse correlation between ROS levels and the proportion of viable cells.²⁴ 

In addition to decreased expression of antioxidant enzymes, other mechanisms may be responsible for the generation of an increased amount of ROS in malignant cells, including an enhanced mitochondrial respiration and NOX enzymes.^15,51  A recent study showed that CLL cells accumulate higher levels of ROS than normal B cells through an elevated respiratory capacity linked to an increased mitochondrial mass.²⁸  Thus, differences in mitochondrial amount with a consequent increase in oxidative phosphorylation may also be responsible for variation in ROS levels among individual CLL samples. In agreement, we show higher amounts of mitochondria in redox hypersensitive CLL cells as compared with redox low-sensitivity cells. However, this increase is not associated with augmented mitochondrial superoxide ions. One possible explanation is that in redox hypersensitive CLL cells, superoxide ions may be rapidly converted to cytosolic H₂O₂, whose levels may accumulate in CLL with low catalase. However, the role of mitochondria in generating different ROS levels in CLL subsets is not clear and deserves to be further investigated.

In conclusion, this study shows that differential redox profiles in CLL are associated with divergent clinical behaviors and advances our understanding of the redox and signaling heterogeneity of CLL. Future challenges are to design therapeutic strategies targeting redox pathways that could implement the effectiveness of current therapies and overcome drug resistance in CLL.

The authors thank all patients who have donated samples for this study.

This work was supported by grants from Fondazione Cassa di Risparmio di Verona, Vicenza, Belluno e Ancona and Associazione Italiana Ricerca sul Cancro (grant 6599) (M.T.S.); Associazione Italiana Ricerca sul Cancro (grant IG16797) (C.L.); Fondazione Cassa di Risparmio di Verona, Vicenza, Belluno e Ancona (grant 2015.0872) (M.T.S.); and Basic Research Program 2015, University of Verona (RATs 2015) (R.C.). I.D. is a fellow of Fondazione Umberto Veronesi.

Contribution: C.C. designed and performed experiments, analyzed data, and contributed to writing the manuscript; R.C. performed modeling analyses and interpreted data; I.D. performed and analyzed reverse transcription polymerase chain reaction experiments; O.P. and O.L. performed flow cytometry experiments; E.M. managed clinical data; C.L., G.P., and M.D. contributed to the study design; M.T.S. designed and coordinated the study, interpreted data, and wrote the manuscript; and all authors reviewed the manuscript.

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

Correspondence: Maria Teresa Scupoli, Laboratorio Universitario di Ricerca Medica, Policlinico G.B. Rossi, P.le L. A. Scuro, 10, 37134 Verona, Italy; e-mail: mariateresa.scupoli@univr.it.