Tiam1/Rac1 signals contribute to the proliferation and chemoresistance, but not motility, of chronic lymphocytic leukemia cells

Chronic lymphocytic leukemia (CLL) is the most prevalent leukemia in the Western world and is still considered incurable. It is increasingly recognized that microenvironmental interactions of CLL cells within bone marrow and lymph nodes essentially affect CLL pathophysiology, and that disrupting these interactions bears a high therapeutic potential.¹  Although almost all CLL cells in the peripheral blood are cell cycle-arrested, aggregates of proliferating CLL cells are found in the lymph nodes and, to a lesser extent, in the bone marrow.²  Here, CLL cell proliferation appears to be primarily induced by their crosstalk with CD4⁺ T cells.^3,4  Other accessory cells, and in particular stromal cells, further support the activation and survival of malignant cells.^5-8 

The correct localization of CLL cells adjacent to favorable microenvironmental signals provided by bystander cells requires the integration of adhesive and migratory signals into cytoskeletal responses. In addition, coordinated and dynamically regulated changes in the cytoskeleton are mandatory for tumor cell proliferation. Rho GTPases play a major role in the organization of the actin cytoskeleton, as well as in the control of gene expression programs involved in proliferation and survival.^9,10  In hematopoietic stem and progenitor cells, Rho GTPases have been described as key mediators of microenvironmental signals downstream of the chemokine receptor CXCR4 and β1 integrins, which are also known to orchestrate CLL cell migration and survival.^11-13  Rho proteins mainly function as guanine nucleotide-dependent molecular binary switches alternating between an inactive GDP-bound and an active GTP-bound conformational state, which is induced by guanine nucleotide exchange factors (GEFs). The best-studied members of the Rho GTPase family in hematopoietic cells are RhoA and Rac1.^11,14 

In CLL, RhoA has been suggested as a contributor to the induction of chemokine-induced integrin activation and migration.^15,16  The role of Rac activity and usage in CLL is still elusive. We investigated RhoA and Rac1 function in CLL cell motility and proliferation. We found that the motility of resting CLL cells was dependent on the RhoA effector molecule Rho-associated coiled-coil forming protein kinase (ROCK) but was independent of active Rac1. This independency was accompanied by a severe downregulation of the major Rac GEF T lymphoma invasion and metastasis 1 (Tiam1) in resting CLL cells. However, once activated, CLL cells restored Tiam1 expression, and CLL cell proliferation involved the Rac1/Tiam1 axis, suggesting a dichotomy of Tiam1/Rac usage in resting vs proliferating CLL cells.

Fluorochrome-labeled monoclonal antibodies were purchased from Becton-Dickinson, Dako-Cytomation, Beckman Coulter, and eBioscience. Chemokines were supplied by R&D Systems. All other reagents and chemicals were purchased from Sigma-Aldrich unless otherwise indicated.

Blood samples from patients with CLL at the Third Medical Department Salzburg were obtained on written informed consent and under the approval of the Salzburg Ethics Committee, Salzburg, Austria (415-E/1287/4-2011, 415-E/1287/8-2011). This study was conducted in accordance with the Declaration of Helsinki. Analyzed patients were chemonaive or had received no therapy for the last 6 months. For patient characteristics, see supplemental Table 1, available on the Blood Web site. Peripheral blood mononuclear cells from patients with CLL or healthy donors were isolated by density gradient centrifugation and either were used freshly or were viably frozen. Thawed cells were cultured in RPMI-1640 containing 10% fetal calf serum, glutamine, and antibiotics. Negative or positive selection of healthy B lymphocytes or CLL cells was performed using the EasySep system from Stem Cell Technologies or the anti-CD19 antibody-based MACS separation system (Miltenyi Biotec). CLL cell purity was higher than 95%.

Total RNA from CD19-sorted B lymphocytes or CLL cells was isolated using the QIAamp RNA Blood Mini Kit (Qiagen). Reverse transcription of RNA was performed using the i-script cDNA Synthesis Kit (Bio-Rad). Quantitative real-time polymerase chain reaction (PCR) using Taqman Gene Expression Assays was performed using the 7500 Real-Time PCR System (Applied Biosystems). Results were quantified using the ΔΔC_T method, with 18S rRNA as a control (for ΔC_T and C_T values, see supplemental Table 2).

µ-slides (ibidi) were coated with 100 µg/mL collagen IV (Stem Cell Technologies) overnight at 4°C, followed, where indicated, by 1 hour coating with 100 ng/mL CXCL12, washing, and blocking with 2% human serum albumin. For inhibition experiments, CLL cells were pretreated with 50 µM NSC-23766, 10 µM Y-27632, 100 ng/mL Pertussis toxin, 10 µM src inhibitor PP2, or solvent for 30 minutes. The chambers were placed on a heated (37°C) stage of an inverted phase-contrast microscope (Olympus IX81), and at least 60 cells per field were recorded by time-lapse videomicroscopy, using a 20-fold magnification. Motile cells were determined by manual tracking of single cells.^13,17  Cells moving more than 5 cell diameters during the observation time of 1 hour were considered motile. Representative differential interference contrast (DIC) images were taken with ×96 magnification.

CLL cell proliferation was induced as described.¹⁸  PBMCs from patients with CLL were cultured with fibroblasts (NIH/3T3 cell line) with or without 10 µL/mL of activating anti-CD3/CD28 microbeads (Life Technologies) for 5 days, and cell proliferation of CD5⁺/CD19⁺ cells was assessed by an intracellular staining, using an antibody against Ki-67 or isotype control, and analyzed by flow cytometry. Alternatively, CLL cells were activated via their coculture with CD40L-transfected fibroblasts, as described.¹⁸ 

PBMCs from patients with CLL were resuspended in Opti-MEM reduced serum cell culture medium (Life Technologies) and transfected with 100 pmol small interfering RNA (siRNA), using the Amaxa nucleofector apparatus and the C-05 program according to the manufacturer's instructions (Lonza AG). After transfection with nontargeting-siRNA, Tiam1-siRNA, or Rac1-siRNA (Ambion), cells were activated using 10 µL/mL anti-CD3/C28 beads for 24 hours. RNA was isolated from anti-CD19 MACS-sorted CLL cells and further used for real-time PCR.

CD5 and CD19 expression was assessed using fluorochrome-labeled antibodies, and cell viability was cytometrically analyzed using annexin V-fluorescein isothiocyanate and 7-aminoactinomycin D and a FC-500 flow cytometer (Beckman Coulter). Viable cells were defined as annexin V/7-AAD double-negative.

The activity of RhoA and Rac1 was measured using the RhoA and Rac1 G-LISA assays (Cytoskeleton). CLL cells were harvested in ice-cold phosphate-buffered saline after stimulation and lysed immediately. Equal protein loading was confirmed using a Bradford assay. Activated Rho or Rac was captured in 96-well plates using its substrates Rhotekin or Pak for RhoA and Rac1, respectively. After washing away unbound nonactivated GTPase, activated RhoA or Rac1 was detected using specific horseradish peroxidase-conjugated antibodies. Chemiluminescence was detected by a DTX880 multiplate reader (Beckman Coulter).

Cell lysates of CD19⁺-selected CLL cells were separated using 10% sodium dodecyl sulfate/polyacrylamide gel electrophoresis. After transferring the proteins onto a polyvinylidene fluoride membrane, immune detection of specific proteins was performed using appropriate antibodies (1:1000 to 1:2000) and enhanced chemiluminescence detection. Images were analyzed using the ImageJ software, V1.45s.

Statistical analyses were performed using the GraphPad Prism software. All data were tested for normal distribution. Comparisons of normally distributed data from 2 groups were done with t tests (paired or unpaired Student t tests, respectively), and nonparametric data were compared using Wilcoxon signed rank test (paired analysis) or Mann-Whitney test (unpaired analysis). For comparing 3 or more groups, 1-way analysis of variance and respective nonparametric tests and the corresponding post hoc tests were used. Differences were considered statistically significant when P < .05, with values at P < .05 marked as *, <.01 as **, and <.001 as ***. Nonsignificant differences are marked as n.s.

To get a first insight into the expression of different members of the Rho GTPase family in CLL, we profiled the transcript levels of common lymphocyte-relevant Rho GTPases and GEFs by real-time PCR. Comparing CLL cells with healthy B cells, we found significantly higher messenger RNA (mRNA) levels of RhoA, RhoC, and RhoH, but lower expression of the GEF Vav1 (Figure 1A-E). Notably, the GEF Tiam1 was downregulated to almost a complete loss in CLL cells compared with normal B cells (Figure 1E), irrespective of the clinical risk profile of the CLL samples (supplemental Figure 1).

Directed lymphocyte migration is controlled by the dynamic activation of RhoA and Rac1.¹⁹  Tiam1 has been identified as a Rac-specific GEF^20,21  that organizes T-cell polarization, migration, and homing.²²  Moreover, RhoH is a hematopoiesis-specific Rho GTPase²³  with constitutive activity and potential Rac-inhibitory function,²⁴  which has recently been identified as being involved in tumor development and CLL cell–microenvironment interactions.^25,26  Hypothesizing that the lack of Tiam1 combined with the overexpression of the potential Rac antagonist RhoH may affect Rac activation and, consequently, cell motility in CLL, we investigated chemokine-induced Rac1 activation compared with RhoA activation. Indeed, using quantitative G-LISA assays,^27,28  we observed CXCL12-induced RhoA in CLL cells that were interacting with a CXCL12-collagen substrate, inducing their motility (Figure 2A, left). We also detected CXCL12-induced RhoA activation in healthy B cells (supplemental Figure 2A). However, we did not detect Rac1 activation in CLL cells in contact with the CXCL12-collagen substrate (Figure 2A, right). Stimulation of CLL cells with soluble CXCL12 did not induce Rac1 activation either (data not shown). As an additional technical control, we used a classical Rac1 activation pull-down assay. CXCL12-induced Rac1 activation of peripheral blood mononuclear cells from healthy donors was clearly detectable in this system, whereas a CLL-derived Rac activation signal could not be observed (supplemental Figure 2B). A second chemokine, the CCR7 ligand CCL19, also failed to provoke Rac1 activation in CLL cells (Figure 2B).

To explore the consequences of this Rac activation defect on CLL cell polarization and motility, we next tested the phosphorylation state of key components downstream of CXCL12. Stimulation of CLL cells with a CXCL12 substrate induced the dephosphorylation of proteins of the ezrin/radixin/moesin family, which is required for correct microvillar collapse during lymphocyte polarization.²⁹  Furthermore, CXCL12 stimulation induced the phosphorylation of the master GEF Vav1 (Figure 2C), indicating the integrity of the migratory machinery beyond Rac/Tiam signals. As conventional readouts of chemokine receptor downstream signaling, we also detected robust CXCL12-induced phosphorylation of Akt and ERK1/2 (supplemental Figure 2C).

We next sought to assess the functional contribution of the RhoA and Rac axes to CLL cell motility. Using time-lapse videomicroscopy, we addressed the ability of the Tiam1/Rac inhibitor NSC-23766 and the RhoA/ROCK antagonist Y-27632 to specifically interfere with CXCL12-induced polarization and migration of CLL cells on a collagen substrate (Figure 2D-F). We observed that the presence of collagen on its own only minimally supported random migration of CLL cells. In contrast, coimmobilization of CXCL12 robustly induced CLL cell motility on this substrate (supplemental Video 1), similar to our recent observations on hyaluronan/CCL21 substrates.¹⁷  Notably, pseudo-DIC images demonstrated the extraordinary phenotype of CLL cell polarization and motility, with its low degree of spreading (Figure 2D).

Notably, Rac activation was not involved in this migratory mode, as pretreatment with the Rac/Tiam1 inhibitor NSC-23766 did not affect polarization or motility (Figure 2E). The efficacy of this inhibitor was ensured by control experiments using NIH/3T3 cells (supplemental Figure 3A). In contrast, chemokine-induced CLL cell motility was strongly reduced in the presence of Y-27632, with complete loss of the polarized phenotype (Figure 2F), indicating the involvement of the RhoA/ROCK axis in this process. Y-27632 did not alter viability or cell size, nor did it activate CLL cells (supplemental Figure 3B). Classically, chemokine-mediated CLL cell motility was dependent on Gα_i proteins and could be abrogated by pertussis toxin (supplemental Figure 3C). CLL cell motility was also diminished in the presence of PP2, a src kinase inhibitor³⁰  (supplemental Figure 3D). It is worth mentioning that neither NSC-23766 nor Y-27632 blocked CXCL12-triggered very late antigen 4-mediated arrests under shear flow (supplemental Figure 4). These data suggest that the very first step of CLL cell extravasation, the arrest on the endothelium, likely involves GTPases distinct from both RhoA and Rac, in contrast to the RhoA usage during motility.

Collectively, our results suggest that resting CLL cells demonstrate a specific Gα_i-coupled, RhoA-dependent, but Rac/Tiam1-independent mode of lateral chemokine-induced motility.

Rac is not only a master component of cellular motility but also an established element of tumor cell proliferation (for review, see Parri and Chiarugi³¹ ). Therefore, we next examined RhoA and Rac usage in CLL cells under conditions that mimic the situation in lymphoid CLL proliferation centers, employing our recently established in vitro system.¹⁸  There, anti-CD3/CD28 microbeads serve to activate the autologous T cells within PBMCs from patients with CLL. In turn, CLL cells receive activation signals from these activated T cells, as well as survival signals from an additional layer of fibroblasts. The activation of CLL cells, starting on 8- to 24-hour coculture, is followed by their proliferation after 3 to 5 days.¹⁸  First, to detect early activation-based alterations, we assessed the expression of the GTPases RhoA, RhoH, and Rac1 and GEFs Vav1, Trio, and Tiam1 in resting CLL cells compared with CLL cells that were activated by T cells after 24 hours. We observed slightly increased transcript levels of all investigated targets in activated CLL cells. However, only Rac and Tiam1 transcription was robustly induced on activation, with a mean upregulation of 2-fold for Rac1 and more than 15-fold for Tiam1 in activated CLL cells compared with nonactivated cells (Figure 3A), irrespective of their risk profile (supplemental Figure 5A-B). In contrast, Tiam1 transcription in normal B cells was not further increased above the nonreduced basal levels by activation (supplemental Figure 5C; supplemental Table 2; compare Figure 1E). The GEF Trio was slightly but not significantly downregulated in activated CLL cells compared with in nonactivated cells (Figure 3A). Notably, the activation-induced upregulation of Rac1 and Tiam1 in CLL cells required their direct cell–cell contact with T cells and could not be observed when activated T cells were separated from the CLL cells by a 0.4-µm pore membrane filter (Figure 3B).

To test whether CD40L-CD40 signals are sufficient to induce Tiam1 transcription, we next cocultured CLL cells with mock- or CD40L-transfected NIH/3T3 cells. We found upregulated Tiam1 levels in CD40-activated CLL cells comparable to the observed Tiam1 upregulation in CLL, using the T-cell activation setting (Figure 3C; supplemental Figure 5D). Next, we wished to explore the consequences of the restored Tiam1 expression on Rac1 activation. We stimulated resting and activated CLL cells with the chemokines CXCL12 or CCL19 and performed Rac1 activation assays. Activated CLL cells displayed reduced CXCR4 expression but increased CCR7 expression, rendering the CCR7 binding ligand CCL19 the relevant control chemokine for testing Rac1 activation of these cells. We found slightly elevated levels of GTP-bound Rac1 in the absence of chemokine stimulation in activated CLL cells compared with resting CLL cells (Figure 3D). However, we could not detect any further increase in GTP-bound Rac1 in activated CLL cells on their stimulation with CCL19 or CXCL12 (data not shown), suggesting a higher intrinsic rather than chemokine-inducible Rac1 activation in activated CLL cells.

To further assess the functional role of the restored Tiam1 expression and increased intrinsic Rac activation, we tested the effects of Tiam1/Rac inhibition vs RhoA/ROCK inhibition on CLL cell proliferation. We measured the number of Ki-67⁺ CLL cells in the presence or absence of NSC-23766 or Y-27632 and activated T cells/fibroblasts after 5 days. We found a strong, significant decrease in the number of Ki-67⁺ CLL cells in the presence of NSC-23766 (Figure 4A), whereas autologous T-cell proliferation was not significantly impaired (supplemental Figure 6A). ROCK inhibition by Y-27632 did not affect the number of Ki-67⁺ CLL cells in this setting (Figure 4B).

To get insight into the responsible mechanisms, we assessed the expression of c-Myc, a master regulator of proliferation. In a time course analysis, we observed that the upregulation of Tiam1 mRNA is a rapid event in T-cell-mediated CLL activation, with a maximum increase after 8 hours and a progressive decline thereafter (Figure 4C, left). As expected, inhibition of Tiam/GEF binding to Rac by NSC-23766 did not change the expression level of Tiam1 after 24 hours (Figure 4C, right). However, at the same time, Rac inhibition lead to the blockade of c-Myc induction with a maximum after 8 to 24 hours (Figure 4D, left) and significantly reduced c-Myc levels at 24 hours (Figure 4D, right). By immunoblotting, we confirmed that these transcriptional alterations were translated to the protein level (Figure 4E). CLL cell activation (CD69 expression) was not altered by the inhibition (supplemental Figure 6B), suggesting a c-Myc specific mechanism rather than a general decrease in activation responsible for the observed antiproliferative effects. As an additional control for the specificity of the used Tiam/Rac inhibitor, we used Tiam1 or Rac1 siRNA, which confirmed the necessity for Tiam1 and Rac1 in the T-cell-induced induction of c-Myc transcription that preceded CLL cell cycle induction (Figure 4F). In an additional time course experiment, the activation-mediated induction of Ki-67 mRNA was abrogated from 100-fold, compared with untreated control, to 40-fold after NSC-23766 treatment (supplemental Figure 6C).

The DNA intercalating purine analog fludarabine is still the backbone of current chemotherapeutic interventions in CLL. Strikingly, fludarabine improves progression-free survival, but not overall survival,³²  probably as a result of drug resistance development. For one, signals from activated T cells can contribute to fludarabine resistance of CLL cells.³³  Therefore, we next investigated the contribution of Tiam1/Rac signaling to T-cell-mediated fludarabine resistance of CLL cells. After 24 hours of anti-CD3/CD28 microbead stimulation, we treated the cells with 10 µM fludarabine and/or NSC-23766 and measured cell viability after 48 hours. As expected, the cell viability of nonactivated CLL cells dropped from 63.4% to 33.7% after treatment, but activated CLL cells developed a fludarabine resistance (84.5% vs 79.0% for untreated and treated activated CLL cells, respectively). Importantly, the additional application of the Tiam1/Rac inhibitor NSC-23766 lead to the resensitization of CLL cells toward fludarabine and a cell viability rate of 49.2%. NSC-23766 alone did not induce significant cell death in activated CLL cells (Figure 5) and did not alter the viability or the extent of fludarabine-induced apoptosis of unstimulated CLL cells (supplemental Figure 7).

In summary, we suggest that CLL cells use a Rho/ROCK-dependent mode of cell motility but a Tiam1/Rac-dependent mode of cell proliferation and drug resistance in CLL (Figure 6).

CLL cells cycle between the peripheral blood and lymphoid compartments. Activating signals from the lymphoid microenvironment promote the proliferation, survival, and chemoresistance of the malignant cells.^34,35  In this study, we found an opposing contribution of functional RhoA and Tiam1/Rac1 to motility of resting CLL cells and microenvironment-induced survival/proliferation, respectively (Figure 5B).

To get a first insight into the role of Rho GTPases in CLL, we profiled the transcript levels of common Rho GTPases and GEFs in CLL cells compared with normal B lymphocytes. We found a severe downregulation of Tiam1 transcript levels in resting CLL cells. However, Tiam1 expression was restored once CLL cells received activating microenvironmental signals; for example, by activated autologous T lymphocytes. Tiam1 is considered an oncogene in a variety of tumors, acting as a modifier of Ras-mediated tumor transformation and progression and as a mediator of tumor cell-microenvironment interactions.^36-39  A recent report suggests that a balance of Tiam1-Rac1- and RhoA-mediated signals is important for pancreatic cancer cell proliferation and invasion, respectively.⁴⁰  Knocking down Tiam1 in the tumor cells increased their migratory capacity, accompanied by increased RhoA activity.⁴⁰  Because Tiam1/Rac signals have previously been considered imperative for lamellipodium formation, chemokine-induced motility, and homing of multiple other nontransformed cell types,⁴¹  this observation points to a differential usage of Tiam1/Rac in cancer cells.

Correspondingly, we observed that chemokine-induced CLL cell motility was independent from Tiam1 expression and chemokine-induced Rac activation, whereas a functional RhoA/ROCK axis was mandatory. It is worth mentioning that the migratory phenotype of CLL cells on chemokine-bearing substrates was extraordinary, with a low degree of spreading compared with T lymphocytes, as previously depicted.¹⁷  This mode of migration might be the consequence of an altered cytoskeletal organization as an adaption to impaired Tiam1-induced Rac activation. This is consistent with the observation of RhoH overexpression in CLL, confirming a recent report by Troeger et al.²⁶  Deletion of RhoH, a Rac antagonist in hematopoietic progenitor cells, lead to increased CXCL12-induced chemotaxis.⁴²  In contrast, RhoH knockout CLL cells showed impaired CXCL12-directed migration²⁶  that might reflect their aberrant RhoH/Rac balance. Whether other Rho GTPases, such as Rab5⁴³  or RhoG,⁴⁴  compensate for impaired Tiam1/Rac1 signals and orchestrate the CLL-specific mode of migration remains to be elucidated.

An intriguing finding of our study was the contribution of restored Tiam1/Rac expression to CLL cell proliferation and chemoresistance. Mimicking the proliferation-inducing lymphoid microenvironment in 2 reductive optimized coculture setups,¹⁸  we observed the restored transcription and translation of Tiam1 protein in activated CLL cells. Little is known of the transcriptional regulation of Tiam1. Increased expression of the transcription factor AP-1 on CD40 stimulation of B cells⁴⁵  might contribute to its restoration, as the Tiam1 promoter contains multiple putative AP-1 binding sites. In colon cancer subsets, Tiam1 expression can be suppressed by promoter methylation.⁴⁶  At any rate, the upregulation of Tiam1, accompanied by increased Rac1 levels, preceded the induction of c-Myc and Ki-67 and the subsequent proliferation of CLL cells. c-Myc is a master regulator of cellular proliferation.⁴⁷  In CLL, c-Myc expression correlates with cell cycle entry and disease progression.^48-51  By using both the Tiam1/Rac inhibitor NSC-23766 as well as Tiam1 and Rac1 siRNA approaches, we demonstrated the direct dependency of c-Myc transcription on functional Tiam1 and Rac1 expression. Our suggestion of Tiam1 as a critical component of tumorigenesis is consistent with recent reports that attribute the connection of Tiam1/Rac and c-Myc to the Wnt pathway.^52-54  For example, in colorectal cancer cells, a complex of Tiam1, Rac1, and the key Wnt-mediator β-catenin binds to the Wnt-responsive promoters of c-Myc and cyclin D1, thereby enhancing their transcription.⁵³  Furthermore, Rac1 is involved in regulating the phosphorylation of the signal transducer and activator of transcription 3, which results in the transcriptional induction of c-Myc.⁵⁵  It is conceivable that a similar mechanism accounts for the parallel induction of Tiam1 and c-Myc in CLL cells.

Our finding of reduced chemoresistance of CLL cells on Tiam1/Rac inhibition implies therapeutic consequences. We observed that CLL cells acquired absolute resistance toward fludarabine when cocultured with activated T cells and fibroblasts. However, they could be resensitized by Tiam1/Rac inhibition. There are several possible mechanisms by which Rac1 could contribute to an antiapoptotic outcome of activated CLL cells. First, Rac1 was shown to interact with Bcl-2 and to stabilize its antiapoptotic activity in lymphoma.⁵⁶  Concurrent exposure to NSC-23766 increased the apoptotic effect of the Bcl-2 inhibitor and BH3 mimetic ABT-737 in leukemia cells.⁵⁷  Bcl-2 overexpression is a hallmark of CLL⁵⁸ . Coherently, BH3 mimetics have proven to be efficient in patients with CLL, but with heterogeneity,⁵⁹  and in microenvironment-mediated drug resistance.⁶⁰  It is possible that additional targeting of Rac could overcome some of these limitations. Second, the antiproliferative effects of Rac1 targeting have been shown in p53-deficienct lymphoma cells, which involved Akt signals.⁶¹  We recently observed that Akt targeting can overcome p53-deficiency-induced fludarabine resistance of CLL cells,⁶²  suggesting further investigation of Rac1 and Akt signaling cascade in p53-deficient and fludarabine-resistant CLL cells.

Remarkably, described mechanisms appear to act independent of the expression of prognostic markers, such as CD38 or CD49d expression and active Tiam1/Rac1, and to be intrinsic to the general lymphoid activation of CLL cells. As a consequence, the use of Rac targeting approaches could be a risk group-independent therapeutic strategy to selectively target the proliferative and resistant pool of CLL cells, which is considered responsible for recurrent relapses of the patients.

We thank all patients for participation, Karin Stefanon for sample preparation, and Dr Andrea Zurl for providing access to the multiplate reader. This work has been supported by the Austrian Science Fund FWF (SFB-P021 to R.G. and P25015-B13 to T.N.H.), the Paracelsus Medical University Salzburg (PMU-FFF E-12/15/074-HAH and E-10/11/058-HAR to T.N.H.), the Salzburg Cancer Research Institute - Laboratory for Immunological and Molecular Cancer Research GmbH, and the province of Salzburg.

Contribution: T.N.H., S.W.H., and R.G. designed research; S.W.H., P.W.K., S.G., K.O., U.P., and M.W., performed research; S.W.H., T.N.H., P.W.K., D.A., H.K., R.H., and R.G. analyzed and interpreted data; and S.W.H. and T.N.H. wrote the manuscript.

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

Correspondence: Tanja N. Hartmann, Laboratory for Immunological and Molecular Cancer Research, Third Medical Department with Hematology, Oncology, Hemostaseology, Infectiology, and Rheumatology, Paracelsus Medical University, Salzburg, Austria; e-mail: t.hartmann@salk.at.