Mitochondrial metabolism contributes to oxidative stress and reveals therapeutic targets in chronic lymphocytic leukemia

Metabolic reprogramming is a central feature of cancer cells and has emerged as an attractive target in novel therapeutic strategies.¹  Chronic lymphocytic leukemia (CLL) is the leukemia with the highest incidence among adults in the Western countries. However, CLL cells’ metabolism remains a relatively unexplored area.

Results from few studies indicate that CLL cells might contribute to the so-called cancer-associated oxidative stress.^2,3  This metabolic condition is regularly seen in cancer patients and results from accumulating reactive oxygen species (ROS).⁴  The underlying mechanisms of oxidative stress in cancer patients often remain elusive. Tumor cells⁵  as well as tumor-associated cells (eg, activated granulocytes⁶ ) can generate abundant ROS. Under physiological conditions, the mitochondrial respiratory chain is the main site of ROS production.⁷  Cancer-related mitochondrial alterations such as defective oxidative phosphorylation⁸  or an enhanced mitochondrial biogenesis^9,10  promote (mitochondrial) ROS output.¹¹  However, nonmitochondrial ROS production by the membrane-bound NADPH-oxidase (NOX) complex can also play a role in cancer. An enhanced NOX activity has actually been identified in several entities.^12,13 

Escalated ROS levels promote genetic instability and development of drug resistance, harness cell-signaling,^4,14  and altogether account for cancer cells’ aggressive behavior.⁵  Oxidative stress additionally attenuates immune responses by leading to dysfunctions and even apoptosis of NK- and T-cells,^15-17  suggesting a role in tumor immune-escape.¹⁸ 

Malignant cells on their part appear adapted to permanent (intrinsic) oxidative stress. They upregulate protective pathways,¹⁹  which in addition favor resistance toward anticancer agents.^20,21 

Increasing evidence suggests a connection between tumor-specific metabolism and a surplus of ROS production. Metabolic alterations can render malignant cells more susceptible to perturbations within the metabolic framework and therefore pose an opportunity for novel therapeutic interventions.¹ 

We first sought to investigate the prevalence of oxidative stress in CLL patients and its systemic impact, namely the link to immune deregulations already found at early disease stages.²²  Next, we focused on deciphering the metabolic processes responsible for the observed ROS overproduction. Using a comprehensive and multimodal approach, we identified an interrelationship between the altered mitochondrial metabolism in CLL cells, ROS generation, their adaptation to intrinsic oxidative stress, and mitochondrial biogenesis. Taking advantage of their peculiar bioenergetic profile, we were able to selectively target CLL cells by pharmacologically interfering with their mitochondrial metabolism.

A better understanding of the CLL-specific metabolic pathways is inevitable for developing novel targeted approaches. Agents directed against the unique metabolic features of CLL cells might represent an opportunity to improve current treatments and even to overcome drug resistance within cancer cells.

Samples were collected (during October 2011 through April 2013) on approval from the Ethics Committee of the University of Erlangen-Nuremberg (protocol number: 200_12) and after receiving patients’ written informed consent. This study was conducted in accordance with the Declaration of Helsinki. Peripheral blood mononuclear cells (PBMCs) and serum from a total of 75 untreated CLL patients and from 16 healthy donors (HDs) were collected prospectively (patients’ characteristics are detailed in supplemental Table 1).

PBMCs were obtained using Ficoll-Paque (GE Healthcare, Piscataway Township, NJ). CLL cells and B-cells were purified by magnetic bead-based negative selection (B-cell Isolation Kit II, Miltenyi Biotec, Bergisch Gladbach, Germany).

Products of DNA (8-hydroxydeoxyguanosine/8-OdHG, Cell Biolabs, San Diego, CA) and lipid (malondialdehyde/MDA, TBARS Assay Kit, Cayman Chemical, Ann Arbor, MI) oxidation were quantified in sera using a VersaMax ELISA (enzyme-linked immunosorbent assay) microplate reader (Molecular Devices, Sunnyvale, CA).

ATP content was assessed in B-cells and CLL cells using a colorimetric ATP Assay Kit (Abcam, Cambridge, United Kingdom).

NADPH levels were determined in 5 × 10⁶ purified HD B-cells and CLL cells using a colorimetric NADP/NADPH assay kit (BioAssay Systems, Hayward, CA).

Antioxidant capacity was measured in cell lysates using a colorimetric Antioxidant Assay Kit (Cayman Chemical) following the manufacturer’s instructions.

The IFN-γ concentration was quantified using an IFN-γ Quantikine ELISA Kit (R&D, Minneapolis, MN).

Superoxide generation by membrane-bound NOX was measured in cytochrome c reduction assays as described previously²³  using a Multiscan device (ThermoScientific, Waltham, MA).

Cells were stained according to the manufacturer's recommendations using fluorochrome-coupled antibodies (supplemental Table 2). Staining was performed in presence of Fc-receptor blocking solution (Biolegend, San Diego, CA). The LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Life Technologies, Carlsbad, CA) was used for dead-cell exclusion. Cell counts were determined using BD Trucount tubes (BD Biosciences, San Jose, CA). For intracellular staining, cells were treated with BD Cytofix/Cytoperm Fixation/Permeabilization Kit (BD Biosciences). Cells were analyzed using an FACS Canto II cytometer (BD Biosciences) and the FlowJo Version 9.5 software (TreeStar, Ashland, OR).

Viable cells are defined as the 7-AAD^neg/Annexin-V^neg cells in FACS. Percentage of specific cell death was calculated according to the following formula: specific cell death (%) = 100 × (% dead cells − % baseline dead cells) / (100% − % baseline dead cells).

Total and mitochondrial ROS were quantified by FACS using CellROX and MitoSOX dyes, respectively (Life Technologies). For evaluating the contribution of NOX and mitochondria to ROS production, cells were pretreated for 1 hour with 0.5 μM of the NOX inhibitor diphenyleneiodonium chloride (DPI, Sigma Aldrich) or 500 nM of the mitochondria-targeting antioxidant MitoQ (kindly provided by M. Murphy).

Glutathione was stained with Thioltracker Violet (Life Technologies) and was semiquantified using FACS.

Cells were stained with the mitochondrial specific dye MitoTracker Green (Life Technologies) according to the manufacturer’s instructions and were analyzed by FACS.

Mitochondrial electrochemical membrane potential (ΔψM) was assessed using the potentiometric dye JC-1 (JC-1 Mitochondrial Assay Kit, Cayman Chemical). Cells were stained with JC-1 according to the manufacturer’s instructions for 15 minutes at 37°C followed by a FACS analysis.

RNA and DNA were extracted from cell lysates (AllPrep DNA/RNA/Protein Mini Kit, Qiagen, Hilden, Germany) and DNA was prepared (Superscript First Strand Synthesis System, Life Technologies) using a Mastercycler nexus (Eppendorf, Hamburg, Germany). The mRNA levels were quantified by qPCR (Quantitect SYBR Green PCR Kit; Qiagen) on a Rotor Gene Q (Qiagen). Relative gene expression was determined by normalizing the expression of each target gene to β2-microglobulin and/or β-actin using gene-specific primers (supplemental Table 3).

Bioenergetics of CLL cells and B-cells were determined using the XF96e Extracellular Flux Analyzer (Seahorse Bioscience, North Billerica, MA). Cells were seeded in specialized tissue culture plates in an optimized concentration of 240,000 cells per well and were subsequently immobilized using CELL-TAK (BD Biosciences). One hour before measurement, cells were incubated at 37°C in a CO₂-free atmosphere. First, basal oxygen consumption rate (OCR) (an indicator for mitochondrial respiration) and extracellular acidification rate (ECAR) (an indicator for lactic acid production or glycolysis) were detected. Next, OCR and ECAR responses toward the application of oligomycin (1 µM), FCCP (2.5 µM), and the combination of antimycin (3 µM) and rotenone (3 µM) (XF Cell Mito Stress Test Kit, Seahorse Bioscience) as well as glucose and PK11195 (Sigma Aldrich) were evaluated. All experiments were performed in at least hexaplicates.

Cells were stained with anti-CD19-APC (BD Biosciences). Mitochondria were visualized using MitoTracker Green FM (Life Technologies), and Hoechst 33258 (Sigma Aldrich) was added for nuclear staining. Cells were immediately analyzed using a fluorescence microscope (Axio-Imager M2, Zeiss, Jena, Germany) at a 630-fold magnification.

CLL cells and B-cells were pelleted and prepared for TEM as described previously²⁴  with 1 modification: after fixation with osmium tetroxide, cells were washed and the pellets were embedded in agar (2% in H₂O). Following hardening on ice, the agar block was cut in pieces and washed before performing dehydration. Samples were analyzed and photographed with a Zeiss EM10 electron microscope (Zeiss) and a Gatan SC1000 OriusTM CCD camera in combination with the Digital Micrograph TM software (Gatan, Pleasanton, CA).

Dimethyl sulfoxide, phorbol-12-myristate-13-acetate (PMA), N-acetyl-cysteine (NAC), cytochrome c, and PK11195 were obtained from Sigma Aldrich, Ca²⁺ and Mg²⁺ from PAA (Piscataway, NJ), NADPH from Roth (Karlsruhe, Germany), and tin-protoporphyrin (SnPP) from Frontier Scientific (Logan, UT).

Differences in means were evaluated with parametric (2-tailed Student or paired t test) or nonparametric (Mann-Whitney U or Wilcoxon) tests based on the distribution levels. All statistical analyses were performed using GraphPad Prism Version 5 (GraphPad Prism Software Inc.) at a significance level of P < .05.

First, we evaluated whether CLL patients display signs of oxidative stress. Because chemotherapy induces ROS,²⁵  only untreated patients were included in all our investigations. Established surrogate markers for ROS, namely products of DNA- (8-OHdG) and lipid-oxidation (MDA), were measured in patients’ and HDs’ sera. As anticipated, increased levels of oxidized biomolecules were prevalent in the CLL patients when compared with controls (Figure 1A). CLL samples were scored according to the 8-OdHG and MDA concentrations and were grouped into a cohort of intermediate/low (ROS^int/low) and high ROS (ROS^high) levels (Figure 1A).

Production of ROS is a hallmark of myeloid cells, and their accumulation can lead to oxidative stress.⁶  Numbers of monocytes and neutrophils were comparable in ROS^high and ROS^int/low patients (Figure 1B-C). Actually, higher ROS levels correlated with the number of circulating CLL cells (Figure 1D). This finding could reflect the higher ROS production in CLL cells compared with B-cells (Figure 1E). In addition, a larger fraction of ROS^high patients had experienced lymphocyte doubling (Figure 1F, supplemental Figure 1).

Next, we analyzed the lymphocyte compartment in conjunction with ROS levels. ROS^high patients had CD4⁺ T-cells of a less activated phenotype accompanied by a reduced CD3ζ-chain expression, and less circulating terminally differentiated effector memory cells (Figure 2A). Furthermore, CD8⁺ T-cells of ROS^high patients displayed significantly lower CD3ζ-chain levels (Figure 2B, supplemental Figures 3 and 4). Number of T-cells, CD4⁺/CD8⁺ T-cell ratio, and proliferative status were not different (supplemental Figures 4-6). These alterations are in line with reports on ROS effects on T-cells²⁶  and observations that CD4⁺ T-cells are particularly deregulated in CLL.²²  The addition of the antioxidant NAC when stimulating T-cells in the presence of CLL cells significantly bolstered T-cell activation as exemplified by an increased expression of HLA-DR, CD137 (both after 96 hours), CD69 (after 24 hours as an early activation marker), and IFN-γ production (Figure 2C). It is well established that NK-cells are sensitive toward ROS.¹⁶  We observed that increased ROS levels were associated with higher proportions of CD56^brightCD16^dim NK-cells (Figure 2D). Circulating NK-cells (supplemental Figure 7), expression of activating molecules such as DNAM1 and NKG2D, and CD3ζ-chain (Figure 2E) were similar in ROS^high and ROS^int/low patients.

Malignant cells (eg, melanoma¹³  or chronic myelogenous leukemia²⁷ ) can generate large amounts of superoxide through an increased activity of NOX. We evaluated NOX components in CLL cells and B-cells and found that the gp91 subunit is significantly lower expressed in CLL cells (Figure 3A). We measured superoxide production by NOX based on extracellular cytochrome c reduction.²³  Superoxide levels were similar in B-cells and CLL cells before (baseline) and after the addition of a NOX substrate (ie, NADPH) as well as on respiratory burst induction by PMA (Figure 3B). The respiratory burst elicited only in B-cells a significant increase of cellular ROS (Figure 3C), which further corroborates the notion that NOX is not responsible for the ROS overproduction seen in CLL cells.

Suppression of antioxidant molecules²⁸  can also lead to increased ROS levels in cancer cells. We assessed major antioxidants and found in CLL cells a significantly increased expression of catalase and heme oxygenase-1 (HO-1), whereas manganese superoxide dismutase (MnSOD) located in the mitochondria was found reduced (Figure 3D). Thiols represent the major nonenzymatic antioxidant scavenging ROS. CLL cells possessed significantly more thiols (Figure 3E) and expressed higher levels of the catalytic subunit of glutamate-cysteine ligase (GCLC) and glucose-6-phosphate dehydrogenase (G6PDH) (Figure 3F-G). GCLC is the rate-limiting enzyme for glutathione synthesis, the most abundant thiol. G6PDH is the key enzyme of the pentose phosphate pathway, which provides the principal cellular reductant NADPH (Figure 3H), which is necessary for glutathione recycling. Overall, redox-buffering capacity of CLL cells appears increased by trend (Figure 3I). In fact, these seemingly contradictory findings could be explained by a compensatory adaptation of CLL cells to persistent oxidative stress.⁴ 

Complexes I and III of the respiratory chain are considered the main generators of mitochondrial ROS.¹¹  Using a probe that selectively detects mitochondrial superoxide, we found significantly increased levels in CLL cells (Figure 4A). MnSOD plays a decisive role in metabolizing superoxide. Therefore, reduced MnSOD expression (Figure 3D) could contribute to increased mitochondrial ROS in CLL cells. To dissect better the respective role of mitochondria and NOX as ROS sources, we treated CLL cells with MitoQ and DPI, respectively. MitoQ is a mitochondria-targeted derivate of the antioxidant ubiquinone.²⁹  DPI at the used dosage blocks NOX enzymes without affecting mitochondrial ROS production.³⁰  Only MitoQ led to a significant reduction (of mitochondrial and at the same time) of total cellular ROS (Figure 4B). These data strongly point to mitochondria as the source of ROS overproduction in CLL cells (Figure 1E), supporting our previous results (Figure 3).

Next, we were interested in whether mitochondrial mass was altered in CLL cells. Using various modalities we found a substantial increase of mitochondrial mass, of relative mitochondrial DNA copies, and of mitochondrial number in CLL cells (Figure 4C-F), which is in line with previous reports.¹⁰  An increased mitochondrial biogenesis could by itself at least contribute to cellular oxidative stress (Figure 4G), because mitochondria represent the main source for (basal) ROS.

Defective mitochondria promote a compensatory increase of mitochondrial biogenesis.⁷  For assessing mitochondrial fitness we measured the mitochondrial membrane potential (∆ΨM) and cellular ATP production. Energy produced during mitochondrial respiration is maintained as a high ∆ΨM, whereas declining mitochondrial function is associated with ∆ΨM loss. We found a significantly higher ∆ΨM in CLL cells accompanied by a significantly elevated energy production (Figure 5A-B). Both are in opposition to a disrupted mitochondrial function. The functionality of the CLL cells’ mitochondria is further underlined by the increased respiratory rate (21.5^+/−5 pmol/min OCR in CLL cells vs 3.5^+/−1 pmol/L OCR in B-cells, P = .01) (Figure 5C). Aerobic glycolysis characterized by lactate production as measured by ECAR, unlike in many other cancer entities,³¹  does not appear overly activated in CLL cells (Figure 5C-D). Molecules of the glucose metabolism including glut1, glut3, hexokinase 2/hk2, lactate dehydrogenase a/ldha, and pyruvate dehydrogenase kinase 1/pdk1 were not increased in CLL cells (Figure 5E).

Treatment with the complex V inhibitor oligomycin reduced average OCR to 47^+/−7% (of the baseline level) in CLL cells and to 66^+/−12% in B-cells, respectively (Figure 5F). Because mitochondrial complex V (F₀F₁-ATPase) couples the electron chain to ATP synthesis, these data indicate that in CLL cells a significantly higher fraction of oxygen consumption was coupled to ATP-synthesis than in B-cells (53% vs 33%, P = .0002). We found no differences regarding the coupling efficacy and the expression of key uncoupling proteins (supplemental Figure 8). Next, we examined the effect of uncoupling respiration from energy generating oxidative phosphorylation on OCR so as to measure the cells’ maximal respiratory capacity. Addition of the uncoupling agent FCCP resulted in a significantly stronger OCR increase in CLL cells compared with B-cells (218^+/−82% vs 121^+/−14%, P = .002) (Figure 5F). This elevated spared or reserve respiratory capacity can be well explained by the observed increased mitochondrial biogenesis (Figure 4).³²  Furthermore, CLL cells’ stronger dependency on maintaining the mitochondrial energetic flux was further highlighted by the effects elicited after the cells’ sequential treatment with oligomycin and FCCP, which block mitochondrial function (Figure 5G). CLL cells displayed a significantly higher compensatory increase of the aerobic glycolysis for meeting their energetic demands (230^+/−62% in CLL cells vs 125^+/−21% in B-cells after oligomycin and 354^+/−57% in CLL cells vs 151^+/−28% in B-cells after FCCP).

We analyzed the expression of molecules involved in mitochondrial biogenesis and/or mitochondrial translation including tfam, nuclear respiratory factor 1/nrf1, nuclear respiratory factor 2/nrf2, estrogen-related receptor α/erra, and the peroxisome proliferator-activated factor α/ppara.¹⁰  We found that tfam was significantly higher as expressed in CLL cells (Figure 6A). Previous reports have shown that ROS can act as an activator of tfam expression.^33,34  Treating CLL cells with the antioxidants NAC and MitoQ resulted in a significant downregulation of tfam (Figure 6B). Moreover, redox modulation also led to a diminution of the highly expressed antioxidant HO-1 (Figures 3D and 6C).²¹  HO-1 is a stress-responsive molecule with strong antioxidant properties. Recent data suggest that its product CO is involved in mitochondrial biogenesis.^35,36  In line with these observations, blocking HO-1-activity by SnPP resulted in a significant reduction of tfam expression (Figure 6D), further corroborating the (indirect) interconnection between ROS, cellular adaptation to ROS, and mitochondrial biogenesis.

The idea of killing cancer cells by ROS-mediated mechanisms that take advantage of their intrinsic oxidative stress was proposed a decade ago.³⁷  We hypothesized that the mitochondrial alterations (Figure 4) together with the CLL cells’ energetic reliance on oxidative phosphorylation (Figure 5F-G) would render them preferentially vulnerable to agents that induce further mitochondrial ROS by affecting the mitochondrial respiratory chain. The benzodiazepine derivate PK11195 generates mitochondrial superoxide in a concentration-dependent manner by inhibiting the mitochondrial F₁F₀-ATPase³⁸  (Figure 6E-F). Treating CLL cells with PK11195 appears to shift the redox balance above their increased tolerability threshold and thereby induces cell death (Figure 6G). Furthermore, PK11195 did not elicit increased cell death in patients’ T-cells and healthy control-derived B- and T-cells (Figure 6G). CLL cells could not be rescued by human bone marrow stromal cells that modulate their redox status and harness their resilience toward ROS-inducing agents³⁹  (Figure 6H).

Malignant transformation is inevitably linked to metabolic reprogramming in cancer cells.¹  As yet, and despite the high incidence of CLL, little is known about CLL cells’ metabolism. Oxidative stress represents a metabolic imbalance, which is regularly observed in cancer patients.²⁵  In line with findings from a recent study evaluating ROS-related damages in lymphocytes from patients with monoclonal B-lymphocytosis and CLL,³  we found increased amounts of oxidatively modified DNA and lipids in the sera of untreated CLL patients. In other entities (eg, pancreatic cancer) it has been shown that the oxidative phenotype of cancer cells can correlate with poor prognosis.^5,40  In a similar fashion, a significantly larger proportion of patients from the ROS^high group have experienced lymphocyte count doubling, which is an early marker for disease activity that has prognostic relevance.

Increased ROS levels can be detrimental for immune cells resulting in loss of function or even apoptosis of, for example, effector T-cells^6,15  and NK-cells.^16,27  On the basis of these observations, tumor-associated oxidative stress appears as a contributor to tumor immune escape.¹⁸  In fact, increased ROS levels in CLL patients were associated with a less activated phenotype of CD4⁺ T-cells and downregulation of the signal-transducing CD3ζ-chain in both the CD4⁺ and CD8⁺ T-cells. This is in line with data from other types of malignancies such as breast, colon, and pancreatic cancer.⁶  Furthermore, this link could at least partly explain the observed deregulations of CD4⁺ T-cells in patients with very early CLL.²²  Interestingly, we found a higher proportion of circulating CD56^bright NK-cells in patients with higher ROS levels. This circumstance additionally highlights how ROS fine-tune immune balances: this NK-cell subset has been shown to be more resilient toward ROS-induced cell death,¹⁶  which could be decisive for its preferential enrichment in CLL and other cancers.⁴¹  We could furthermore show that stimulating patients’ T-cells in the presence of the antioxidant NAC reverted some of the ROS-mediated effects. This provides on the one hand a strong indication that oxidative stress has a relevant impact on the CLL patients’ immune system and suggests on the other hand pharmacological redox-remodeling as a potential mean for harnessing immune responses in CLL. Such strategies are already exploited in various entities with the most promising results being achieved in AML by combining antioxidants with mild immune stimulation.⁴² 

Pathomechanisms responsible for ROS overproduction often remain vague. Our data suggest that CLL cells are strong contributors to patients’ oxidative stress. This is in line with studies showing that B-cell stimulation promotes ROS production²³  and CLL cells partially resemble chronically activated B-cells.⁴³  Several types of solid (eg, melanoma¹³ ) and hematological malignancies (eg, hairy cell leukemia¹² ) display increased superoxide production by the membrane-bound NOX. Expression of NOX subunits and NOX activity were not increased in CLL cells. In fact, we found lower levels of the gp91 subunit. Interestingly, it has been recently shown that gp91-deficiency in B-cells is associated with a stronger proliferative capacity,⁴⁴  which is a prerequisite for CLL development.

Although lower antioxidant levels have been described, for example, in lung cancer cells,²⁸  in most malignancies the opposite is the case. Because cancer cells are continuously exposed to high amounts of endogenous ROS, cells capable of an according stress adaptation are preferentially selected. Consequently, an enhanced antioxidant capacity is regularly seen in tumors,¹⁹  which also has implications for the development of drug resistance.²⁰  Several antioxidants were increased in CLL cells, which is partly (eg, for HO-1) in line with previous reports.²¹  Expression of G6PDH catalyzing the entry of glucose-6-phosphate into the pentose phosphate pathway to generate nucleotides, NAPDH, and thiols was also increased. Thiols themselves hold crucial antioxidant functions⁴⁵  and were elevated in CLL cells, potentially contributing to the (by trend) enhanced antioxidant capacity.

Our data altogether point to mitochondria as the main cause of oxidative stress in CLL cells. This is in accord with recent studies showing that B-cell stimulation leads to an increase in mitochondrial ROS, which are essential for an adequate B-cell response.²³  We could show that CLL cells similar to AML-blasts⁹  exhibit increased mitochondrial biogenesis that translates into increased cellular respiratory rate and respiratory reserve capacity. In contrast to many other types of cancer, CLL cells do not appear to be overly utilizing aerobic glycolysis, also known as the Warburg effect.³¹  Our results suggest that CLL cells use primarily oxidative phosphorylation for generating energy. The increased mitochondrial biogenesis could theoretically reflect feedback mechanisms compensating for mitochondrial defects.³³  However, CLL cells’ mitochondria appeared functional. In fact, increased respiratory rates (eg, by leading to more leaking protons in the electron transport chain) and high ∆ΨM (eg, by increasing stability of ROS) can positively correlate with mitochondrial ROS levels as established in previous studies¹¹  and could thereby explain the observed oxidative stress.

Furthermore, increasing evidence suggests a close interrelationship between oxidative stress and mitochondrial biogenesis. Cellular adaptation mechanisms—in our case the upregulation of the stress-responsive HO-1³⁵ —toward abundant ROS can act as positive regulators of mitochondrial transcription factors such as TFAM,³⁴  which was upregulated in CLL cells (Figure 7). This process is thought to represent a protective measure: increased mitochondrial ROS lead to mitochondrial damages and subsequent decrease of energy production. The according activation of the redox adaptation cascade would then additionally promote biogenesis of novel and functional mitochondria.⁴⁶  Because more mitochondria can result in more mitochondrial ROS, this process turns into a vicious cycle in CLL cells (Figure 7). As a proof of principle, we were able to target this self-amplifying loop in vitro by treating CLL cells with antioxidants that led to downregulation of both HO-1 and TFAM. This interventional approach could represent an interesting strategy aiming (a) directly to antagonize ROS-mediated effects and (b) indirectly to prevent mitochondrial biogenesis.

Furthermore, this oxidative phenotype of CLL cells provides the mechanistic basis for the efficacy and specificity of PK11195. This benzodiazepine derivate inhibits the F1F0-ATPase,³⁸  thereby blocking oxidative phosphorylation. The resulting surplus production of mitochondrial superoxide leads to a rather selective apoptosis in CLL cells. The targeted action of PK11195 is most likely attributed to the metabolic differences between malignant and nonmalignant cells. Our data are consistent with recent studies evaluating F1F0-ATPase inhibitors for eliminating alloreactive T-cells in graft-versus-host disease, which share metabolic similarities with CLL cells.⁴⁷ 

In conclusion, this study shows that the CLL cells’ mitochondria are a key source for the patients’ oxidative stress linked to immune alterations. Our observations provide a rationale for 2 opposite therapeutic redox-modulating strategies in CLL. First, cellular adaptation to oxidative stress appears as an indirect promoter of ROS production; a vicious cycle that could be breached by redox-manipulation using antioxidants that could also prove beneficial for enhancing intrinsic immune responses. Second, using agents that further aggravate mitochondrial ROS production and thereby overwhelm the cancer cells’ protective systems could selectively impact CLL cells. Targeting metabolic pathways could help to increase the efficacy of already-existing therapeutic approaches and to overcome drug resistances in CLL.

The authors are very grateful to M. Murphy from Cambridge University for providing MitoQ and to N. Schröder-Kreß for assisting with electron microscopy.

This work was supported by the IZKF Erlangen, the Max-Eder program of the Deutsche Krebshilfe, the European Hematology Association, and the Ria-Freifrau von Fritsch Foundation.

Contribution: R.J. and A.D.H. performed research, analyzed data, and helped to write the manuscript; H.B., A.G., J. Bricks, J. Berger, and D.S. performed research; M.J.E. collected patient material and provided patient data; A.M. helped to design the study and to write the manuscript; and D.M. designed the study, analyzed data, and wrote the manuscript.

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

Correspondence: Dimitrios Mougiakakos, Department of Internal Medicine 5, Hematology and Oncology, University of Erlangen-Nuremberg, Ulmenweg 18, 90154 Erlangen, Germany; e-mail: dimitrios.mougiakakos@uk-erlangen.de.