Autocrine Signaling by Wnt-5a Deregulates Chemotaxis of Leukemic Cells and Predicts Clinical Outcome in Chronic Lymphocytic Leukemia

ROR1, a transmembrane receptor tyrosine-protein kinase, is upregulated in chronic lymphocytic leukemia (CLL). It is uniformly expressed on the surface of CLL cells (1–3), whereas it is almost absent in healthy B-cell subsets, with the notable exception of hematogones (4). Signaling through the ROR1 receptor is far from being completely understood. It is, however, widely accepted that ROR1 and closely related ROR2 act as receptors for Wnt ligands, mainly from the Wnt-5 family, and mediate the activation of the noncanonical Wnt pathway, referred to as the planar cell polarity (PCP) pathway. Wnt/PCP controls various aspects of cell polarity and migration with the most prominent role in embryonic development (5).

There is increasing evidence that ROR1 plays an important role in the pathogenesis of CLL. First, siRNA-mediated silencing of ROR1 (6) as well as treatment by specific anti-ROR1 antibodies can induce apoptosis of CLL cells (7). Second, we have recently shown that the PCP pathway, driven by ROR1, controls migratory properties of CLL cells in the chemokine gradient and that several PCP genes are expressed differently between IGHV-mutated (M-CLL) and unmutated (U-CLL) CLL subgroups (8). Third, it was shown that ROR1 overexpression can enhance leukemogenesis in the Eμ–TCL1 mouse model (9).

Strikingly, ROR1 expression in CLL patients is remarkably uniform. The amount of ROR1 does not correlate with markers of disease aggressiveness and does not predict clinical outcome (1, 2, 8). This is in distinct contrast with the fact that the ROR1-controlled noncanonical Wnt pathway has an impact on CLL progression (8, 9). These observations suggest that ROR1 activity and its downstream signaling in CLL are controlled by unknown factors.

We have shown earlier that acute stimulation by exogenous Wnt-5a, a putative ROR1 ligand, can modulate the chemotactic response of CLL cells (8). However, nothing is known if and how this is relevant for CLL biology. Here, we hypothesized that ROR1 ligands from the Wnt-5 family (Wnt-5a and Wnt-5b) control ROR1 activity in CLL cells and may help to explain the functional variability despite uniform ROR1 expression. We demonstrate that Wnt-5 ligands show dramatically varying expression among CLL patients and also among individual populations of healthy B cells, which implicate their role in the interaction of B cells with the secondary lymphoid tissue microenvironment. Furthermore, we provide evidence that autocrine Wnt-5a is a key factor, which contributes to the deregulated motility and chemotaxis of primary CLL cells. These observations identify Wnt-5a as the crucial regulator of ROR1 activity in CLL and further support the importance of the Wnt-5a-ROR1 axis for CLL pathogenesis.

CLL patients were monitored and treated at the Department of Internal Medicine – Hematology and Oncology, University Hospital Brno (Brno, Czech Republic) according to international criteria (10). All samples, including nonmalignant controls, were taken after written informed consent in accordance with the Declaration of Helsinki under protocols approved by the Ethical Committee of the University Hospital Brno. B lymphocytes from the peripheral blood of CLL patients and healthy volunteers and from nonmalignant tonsilar tissues were separated using non–B-cell depletion techniques (RosetteSep kits, StemCell, or magnetic separation–B-cell isolation kit II, Miltenyi Biotec; ref. 8). The separation efficiency was assessed by flow-cytometry and all tested samples contained ≥98% B cells. Individual tonsilar B-cell populations were separated by FACS sorting.

The CLL cohort selected for expression analyses represent patients monitored at the University Hospital where more aggressive cases with inferior prognosis are referred. The patients were examined for presence of mutations in TP53 and SF3B1 genes, for recurrent cytogenetic abnormalities, and for somatic hypermutations in IGHV gene. To validate correlation of WNT5 expression with less frequent abnormalities (trisomy 12, and SF3B1), basic cohort was enriched for patients carrying these aberrations. Serial samples of 36 patients were obtained to measure expression changes in time. For cohort characterization and further details, see Supplementary Information.

Relative gene expression of WNT5A, WNT5B, and ROR1 was assessed by quantitative real-time PCR. Briefly, cells were lysed in TriReagent (MRC) and total RNA was isolated by isopropanol precipitation. mRNA was reverse-transcribed using oligo(dT; Fermentas) and SuperScriptII (Invitrogen). ROR1 expression was measured using Power SYBR Green Master Mix (Applied Biosystems) and primers described previously (8). WNT5A/B genes were quantified using TaqMan Gene Expression Assays Hs00998537_m1 Wnt-5a and Hs01086864_m1 Wnt-5b (Life Technologies). The relative expression was calculated by the 2^−ddCt×100% method using HPRT1 and TBP genes as endogenous controls.

Immunoprecipitation and Western blotting were performed as previously described (11). For detection, Immobilon Western Chemiluminescent HRP Substrate (Millipore) was used. For more details, see Supplementary Information.

The fresh primary B cells were collected as described above and cultivated overnight in RPMI1640 (Life Technologies) supplemented with 1% FBS (Gibco) and antibiotics (penicillin/streptomycin, TPP) at 37 °C and 5% CO₂. The migration assay was performed in HTS Transwell 96-well plates (Corning Incorporated) with 5 μm pore size polycarbonate membranes following the manufacturer's instructions. After treatment (for details of individual treatments see Supplementary Information), 0.5 × 10⁶ cells were seeded in the Transwell upper insert and incubated for 6 hours at 37°C and 5% CO₂. Cell number in the bottom chamber was counted by Accuri C6 Flow Cytometer (BD Biosciences). The basal migration was defined as the percentage of transmigrated cells out of the total. Directed chemotaxis toward medium containing 200 ng/mL of CXCL12 or CCL19 (R&D Systems, 350-NS-010 and 361-MI-025) chemokines was defined as migration index (MI) calculated as ratio of migrated cells in the presence and absence of chemokine. The apoptosis was assessed in parallel by TMRE staining (2 μmol/L TMRE, 15 minutes at room temperature, T-66915; Invitrogen).

The distribution normality was tested by the Kolmogorov–Smirnov, Shapiro–Wilk, or D'Agostino and Pearson normality test. Parametric or nonparametric tests were used accordingly to assess the difference between two variables (unpaired t test and Mann–Whitney test), the difference in the paired samples (Wilcoxon test), and the correlation between two variables (Pearson and Spearman test). The Fisher exact test was used for categorical datasets. Differences in survival were analyzed using the log-rank test. The standard level of statistical significance was P ≤ 0.05. All statistical tests were performed as two-sided using GraphPad Prism 5 (GraphPad Software Inc.). Cutoffs stratifying patients into subgroups with short/long time to first treatment were determined using the CutOff Finder web application (12).

As the first step, we examined the ROR1 gene expression and observed that CLL B cells (n = 93), regardless of the IGHV mutational status, expressed uniformly high levels of ROR1 mRNA compared with B cells isolated from tonsils and peripheral blood (PB) from healthy volunteers (Fig. 1A). This correlated very well with the ROR1 protein expression on the cell surface, P = 0.0038 (Supplementary Fig. S1A, Pearson correlation), defined by flow cytometry. These findings are in agreement with the previous reports (1–3).

Although both Wnt-5a and Wnt-5b have been proposed to be ROR1 ligands, bona fide evidence of interaction was presented only for Wnt-5a (3). However, our coimmunoprecipitation analysis suggested that both Wnt-5a and Wnt-5b ligands can be found in complex with ROR1 (Fig. 1B). In order to test whether indeed these two ligands signal in primary CLL cells via ROR1, we stimulated primary CLL cells with recombinant Wnt-5a and Wnt-5b. As a control, we used Wnt-3a, a ligand known to signal exclusively via the Wnt/ß-catenin pathway (13), which is independent of ROR receptors (14). Ligands from both families could induce phosphorylation of Dishevelled (DVL) family of proteins, which can be subsequently used as a universal downstream readout (15). Out of the tested Wnt ligands, only Wnt-5a and Wnt-3a were able to trigger DVL3 phosphorylation in primary CLL cells (Fig. 1C, lanes 1–4). To prove that this effect is dependent on the ROR1 receptor, we pretreated cells with the blocking mouse monoclonal anti-ROR1 (2A2) antibody (16). Despite the fact that both Wnt-5a and Wnt-3a promote DVL3 phosphorylation in primary CLL cells, only the Wnt-5a effect is dependent on ROR1 receptor as it was blocked by prior treatment with the anti-ROR1 antibody (Fig. 1C, lanes 5–8). Effect of Wnt-3a is ROR1 independent and not affected by anti-ROR1 antibody. Recombinant Wnt-5b does not promote DVL3 phosphorylation in these cells; however, we cannot exclude that it acts via other pathways. Out of a panel of four anti-ROR1 antibodies targeting different epitopes in the ROR1 extracellular domain (Fig. 1D), only anti-ROR1 2A2 antibody was able to completely block Wnt-5a-induced DVL3 shift (Fig. 1D and Supplementary Fig. S1B/C) pinpointing the critical role of N-terminal immunoglobulin-like domain of ROR1 in the downstream signaling triggered by Wnt-5a. In summary, these data demonstrate that both Wnt-5a and Wnt-5b can interact with ROR1 and ROR1 is required for Wnt-5a–induced signaling toward DVL in CLL cells.

In order to understand the Wnt-5 role in CLL, mRNA expression of two genes encoding Wnt-5 ligands, WNT5A and WNT5B, was assessed in 137 previously untreated CLL patients (34% M-CLL, 66% U-CLL) and control nonmalignant samples from peripheral blood (N = 6) and tonsils (naïve, N = 8; centrocyte, N = 4; centroblast, N = 4; memory, N = 8; for details on separation/phenotyping see Supplementary Information). The levels of both WNT5s are homogeneously low in control peripheral blood B cells and naïve tonsilar B cells but more than ten times higher in tonsilar centroblasts and centrocytes (Fig. 2A and B). The WNT5 levels drop again in tonsilar memory B cells with the more profound decrease observed for WNT5A (Fig. 2A and B). CLL samples show highly variable expression of both WNT5 genes ranging from levels resembling healthy peripheral B cells or even lower to values exceeding median expression in germinal center B cells by more than 40 to 70 times in the case of WNT5A (Fig. 2A) and six to seven times in the case of WNT5B (Fig. 2B). Interestingly, WNT5A transcript was undetectable in a significant proportion of CLL patients (85/137 ≈62 %; referred as WNT5A-negative) whereas it was present, albeit at low levels, in most healthy (5/6) control peripheral B cells. The differences at the mRNA level translate into the increased protein amount (Fig. 2C), which suggest that cells high in WNT5A produce also functional Wnt-5a.

In U-CLL, the proportion of WNT5A-negative samples was significantly lower (50% vs. 85%, P < 0.0001, Fisher exact test) and the expression of WNT5B was remarkably higher (P < 0.0001, Mann–Whitney test; Fig. 2A and B). Although the association of the WNT5A and WNT5B expression with IGHV mutational status is comparably strong, we did not observe a direct correlation between the two genes (Fig. 2D). Furthermore, WNT5A and WNT5B expression does not correlate with the expression of ROR1 receptor (Supplementary Fig. S2A–S2D), which opens the possibility that ROR1 signaling might be regulated by the Wnt-5 ligands produced by CLL cells.

Higher WNT5A/B expression levels in a prognostically worse U-CLL subgroup raised the question whether the WNT5 genes could play a role in disease progression. To answer this issue, we measured their expression in paired samples from 36 patients. Median time between samplings was 36.2 months (range 9.3–92.3). Seventeen patients were examined in relapse after therapy administered between the samplings. Out of the remaining 19 patients, remarkable progression of the disease was observed in 8 patients, whereas 11 patients remained stable. We did not observe significant difference in WNT5 expression between the initial and subsequent samples (Wilcoxon paired t test, n.s.; Supplementary Fig. S2E and S2F). The fact that there are no remarkable changes in WNT5 expression comparable with the expression range within the studied cohort suggests that WNT5 expression is determined in the early disease state and is not, in most cases, affected by disease evolution.

Initial characterization of WNT5A and WNT5B expression (Fig. 2) uncovered significant differences between U-CLL and M-CLL, two major CLL subsets with a different clinical outcome. Subsequently, we retrospectively analyzed time to first treatment (TTFT) in patient cohorts divided according to the WNT5A (negative vs. positive cases, Fig. 2A) and WNT5B expression levels (high vs. low; Fig. 2B). As we show in Fig. 3A (WNT5A) and 3B (WNT5B), the subset of patients with lower WNT5A and WNT5B expression levels had a significantly longer TTFT (median 43.3 months for WNT5A-negative vs. 16.5 months for WNT5A-positive, P < 0.0001; median 56.8 months for WNT5B low vs. 22.2 months for WNT5B high, P < 0.0001).

Interestingly, increased WNT5A (Fig. 3C; positive vs. negative WNT5A expression: P = 0.0046, log-rank test) but not increased WNT5B (Supplementary Fig. S3A) expression also had prognostic value within M-CLL and identified a subgroup of M-CLL patients with shorter TTFT. WNT5A expression was dramatically higher in M-CLL patients with borderline mutated IGHV (97%–98% of germline identity; Fig. 3D, P < 0.0001, Spearman correlation). Similar but weaker association was also seen for WNT5B (Supplementary Fig. S3D; P = 0.0047, Spearman correlation). M-CLL patients with IGHV identity between 97% and 98% had shorter TTFT in our cohort (see Fig. 3C; TTFT in M-CLL below and above 97%; P = 0.0062, log-rank test), which is in agreement with the previously reported observation (17). Thus, M-CLL patients positive for WNT5A have more progressive disease and often carry less hypermutated IGHV gene in comparison with WNT5A-negative M-CLL patients. This correlation, however, did not hold true within U-CLL (Supplementary Fig. S3B and S3C).

In order to gain further insight into the role of WNT5 in CLL, we examined the correlation of WNT5A/B expression with other clinical parameters both in the whole cohort and in U-CLL and M-CLL separately (summed up in Supplementary Table S1). High expression of WNT5A and/or 5B expression in the whole cohort significantly positively correlated with numerous parameters associated with unmutated IGHV—leukocytosis, clinical stage at diagnosis, TP53 mutation/17p deletion, and 11q deletion. However, WNT5A and/or WNT5B gene expression levels were also significantly associated with several parameters even within U-CLL and/or M-CLL subgroups. Most strikingly, within U-CLL cohort, the WNT5A expression was increased in the most aggressive CLL defined by 17p deletion (P = 0.0071; Mann–Whitney test; Fig. 3E). Significantly higher WNT5A expression was also found in patients with aggressive CLL carrying a mutation in TP53 (P = 0.02; Mann–Whitney test; Fig. 3F) or in SF3B1 (Fig. 3G; SF3B1-mutated samples added to validate significance; P = 0.0218; Mann–Whitney test). Of note, these mutations are often mutually exclusive (18, 19), which suggests that high WNT5A is a hallmark of an aggressive disease. For WNT5B data, see Supplementary Fig. S4A/S4B. For complete data describing the WNT5 gene expression in patients stratified according to the FISH hierarchical risk model (20), see Supplementary Fig. S4C/S4D.

The association of WNT5A and WNT5B mRNA levels with markers of unfavorable prognosis observed here and the previously reported link between Wnt/PCP pathway signaling and CLL pathogenesis (8) prompted us to investigate the functional role of Wnt-5 ligands in the behavior of CLL cells. We analyzed 27 primary CLL samples (M-CLL, N = 12; U-CLL, N = 15) with variable WNT5A (i) and WNT5B expression (ii; Fig. 4A) for their ability to migrate in vitro in the absence or presence of chemokines using Transwell system (8). On the basis of the WNT5A and WNT5B expression, respectively, the samples were divided into two groups using the same criteria as in Fig. 2.

The basal migration capacity, defined as the proportion of CLL cells migrating in the absence of chemokine, varied strongly among individual samples, ranging from 0.02% to 7.74% (range, 10²–10⁴ cells) of migrated cells (median 0.3%, N = 27; Fig. 4B). Following stratification based on (i) IGHV mutational status, (ii) expression of WNT5A, and (iii) WNT5B, only high WNT5A expression could distinguish cells with a significantly higher migration capacity (median 1.08% vs. 0.23%, P = 0.0042, Mann–Whitney test; Fig. 4B). Importantly, high WNT5A expression also defined more motile cells within U-CLL cohort (Fig. 4B, iv; P = 0.0128, Mann–Whitney test, N = 27), which suggests that WNT5A, and not unmutated IGHV, is the primary parameter from those tested associated with the higher CLL cell migratory capacity.

Next, we studied the CLL cell ability to respond to chemokines CXCL12 and CCL19 known to stimulate via the receptors CXCR4 and CCR7, respectively (21). The chemotaxis was depicted as a migration index (MI) defined as the ratio of cells migrated in the presence versus absence of a chemokine. The MI varied ranging from 0.4 to 75.5 for CXCL12 (N = 25) and from 0.35 to 121 for CCL19 (N = 25; Fig. 4C and D) and cannot simply be explained by the expression levels of relevant receptors (ROR1 receptor and chemokine receptors CXCR4 and CCR7) since the surface levels of individual receptors, defined by flow cytometry, did not correlate significantly with migratory parameters (Supplementary Fig. S5A–S5C). In contrast, following patient stratification, we found significant differences between WNT5A-positive and WNT5A-negative groups (Fig. 4C and D) and in case of CCL19 also between WNT5B-high and WNT5B-low groups. Higher WNT5A expression defined the CLL cells generally less able to respond to the chemokine stimuli compared with WNT5A-negative group (MI CXCL12: P = 0.0143; MI CCL19: P = 0.0202; Mann–Whitney test). Similar to the basal migration, WNT5A-positive CLL samples showed a tendency toward a more impaired response to chemokines within the U-CLL patients, although in our sample, this trend was not significant. In summary, our functional data demonstrate that WNT5A-high CLL cells show deregulated migratory properties defined by higher basal motility and decreased response to chemokines.

The finding that WNT5A-positive CLL cells exhibited higher basal migration suggested a causative connection between Wnt-5 signaling and cell motility. To support this possibility, we first tested the effect of recombinant Wnt-5a on CLL cell migration. We found that recombinant Wnt-5a has a significant positive effect on the migration of cells lacking endogenous WNT5A expression in comparison with cells expressing WNT5A (P = 0.0223; N = 15; Fig. 5A). The differential response of CLL cells to WNT5A cannot be explained by the IGHV mutational status (Fig. 5A). In addition, overnight stimulation of WNT5A-negative CLL cells (N = 5) with Wnt-5a was able to decrease migratory response to CXCL12, which suggests that chronic Wnt-5a stimulation can desensitize CLL cells for chemotactic stimuli (Fig. 5B).

Next, we inhibited the Wnt/PCP pathway by (i) Rho/ROCK kinase inhibitor (100 μmol/L, Y-27632), (ii) porcupine inhibitor IWP2 (10 μmol/L), which specifically blocks secretion of Wnt ligands, (iii) soluble Frizzled-related protein (sFRP1, 0.3 μg/mL), a natural decoy receptor, which prevent Wnt ligands binding to their receptors; and (iv) Rac1 inhibitor (10 μmol/L, NSC23766). Out of these drugs, sFRP1 and IWP2 are specific for Wnt signal transduction, whereas inhibitors of Rho/Rac have a more general effect and may interfere with other signaling pathways regulating cytoskeleton. See Fig. 5C for a schematized mode of action for each drug. As expected, all these inhibitors with the exception of Rac1 inhibitor were able to decrease CLL cell migration in the 6-hour Transwell assay (Fig. 5D, N = 16 for Y-27632, N = 19 for IWP2, N = 21 for sFRP1, N = 10 for NSC23766). Cell viability was not significantly affected by any of these treatments (Fig. 5D).

Patient stratification (see Fig. 4) showed that only CLL cells positive for WNT5A but not unmutated IGHV (Fig. 5 E–G) or high WNT5B expression (Supplementary Fig. S6A–S6C) are significantly more prone to inhibition by the mechanistically unrelated Wnt/PCP pathway inhibitors Y-27632 (Fig. 5E, P = 0.0247, unpaired t test), IWP2 (Fig. 5F, P = 0.0429, unpaired t test), and sFRP1 (Fig. 5G, P = 0.0305, unpaired t test). We can exclude the possibility that the response associated primarily with U-CLL because WNT5A-positive cells responded significantly more to both Rho/ROCK inhibitor (Fig. 5E, iv, P = 0.0391) and porcupine inhibitor IWP2 (Fig. 5F, iv, P = 0.0214) also within U-CLL cohort. Interestingly, Wnt/PCP pathway inhibition can rescue also the chemotactic response and increase the MI index for both CXCL12 (Fig. 5H-i) and CCL19 (Fig. 5H-ii; Wilcoxon paired test of raw data shown in Supplementary Fig. S6D–S6F). The MI median was increased up to 5.83-fold in case of sFRP1 and CXCL12, and it was significant for all combinations with the exception of a 1.74-fold increase of MI CXCL12 by Rho/ROCK inhibitor Y-27632.

In summary, these data suggest that inhibiting the Wnt/PCP rescues migratory defects, namely increased basal migration and decreased chemotaxis, associated with WNT5A-positive cells and support causative connection between high Wnt-5 levels and deregulated migration.

In the present study, we have investigated the role of Wnt-5a and Wnt-5b, the two known ROR1 physiologic ligands, in CLL biology. Despite many similarities between Wnt-5a and Wnt-5b, several important distinctive features were observed. First, only Wnt-5a but not Wnt-5b can induce downstream DVL phosphorylation via ROR1, an event associated with the activation of the noncanonical WNT pathway (15). Second, only high WNT5A but not WNT5B expression correlated with the most aggressive form of CLL characterized by dysfunctional p53. Third, only patient stratification based on WNT5A but not WNT5B expression could explain the differences in the CLL migratory potential and their response to the inhibitors of the Wnt/PCP pathway. The typical features of WNT5A-positive cells are schematized in Fig. 6.

In agreement with previous reports, our study shows that the WNT/PCP pathway is deregulated in CLL at multiple levels. First, its activity can be controlled by the availability of the receptor, in CLL primarily by ROR1. It was shown that cell surface levels of ROR1 vary in CLL cells depending on its glycosylation (22, 23). Second, this work demonstrates that the WNT5 ligand expression differs dramatically among individual CLL patients, which suggests that the availability of ligands is another important control step in the pathway activation. Third, Wnt-5 ligand activity can be eliminated by several soluble pathway inhibitors, mainly by the soluble FZD decoy receptors from the sFRP family. Interestingly, sFRP1, the most efficient inhibitor of Wnt-5a induced signaling (24), and other proteins from the sFRP family were found to be epigenetically silenced in CLL but not in healthy CD19⁺ B cells (25).

We provide evidence that the high WNT5A expression and high Wnt-5a autocrine signaling define the subgroups of patients with poor prognosis whose CLL cells often have higher basal motility and attenuated response to chemokines. In general, it has been repeatedly shown that (i) circulating CLL lymphocytes display changed migratory and chemotactic capacity in comparison with normal B cells (26, 27) and (ii) the chemotactic potential differs dramatically among individual CLL samples (28–30). There is, however, no consensus how attenuated chemotaxis contributes to CLL pathogenesis and how it correlates with prognosis. Some studies argue that robust chemotaxis (high MI) associates with high CD38 and ZAP-70 (28, 29), which are generally markers of unfavorable prognosis (31, 32). Others, however, demonstrated that low CXCR4 defines patients with shorter overall survival (33). Moreover, it has been shown that B-cell receptor (BCR) stimulation attenuates (probably via decreased expression of CXCR4) chemotaxis toward CXCL12 and CLL with the strongest BCR-triggered downregulation of CXCR4 were associated with the worst progression-free survival (34). The reasons for these discrepancies are unclear but may reflect differences between in vivo and in vitro situation and/or between chronic and acute stimulation. For example, we have shown previously that acute treatment by Wnt-5a can promote chemotaxis of CLL cells (8), whereas CLL cells with high WNT5A expression respond to Wnt-5a worse despite high migration when unstimulated (as shown in this study). Herein, presented ability of WNT5A-low but not WNT5A-high cells to migrate better in response to Wnt-5a suggests the existence of negative feedback loops, which affect the final outcome.

The differences in chemotaxis cannot simply be explained by differences in chemokine receptor expression (27, 28). These studies implicate that other factors significantly modulate the response and we propose that autocrine Wnt-5a signaling via ROR1 is one of the key mechanisms in this process. Currently, it is not clear whether CLL cells can experience Wnt-5a produced by other cell types in the body in a paracrine fashion. WNT5A is, however, highly expressed by several immune cell types (http://www.immgen.org/) and germinal canter B cells functionally depend on Wnt-5a produced by dendritic cells (35). We can speculate that autocrine Wnt-5a protein production in WNT5A-high CLL cells can bypass the requirement for the ligand, which can be under physiologic conditions provided in a tightly regulated manner by supportive cells from the microenvironment.

In summary, our data demonstrate that varying levels of WNT5A (and to a lower extent also WNT5B) help to explain CLL behavior diversity and are useful markers of disease aggressiveness. Functionally, high autocrine Wnt-5a signaling leads to the impaired CLL cell migration. This opens the interesting possibility that permanent autocrine Wnt-5a signaling allows CLL cells to escape normal regulatory mechanisms controlling B-cell trafficking.

V. Bryja, M. Kaucka, S. Pospisilova, A. Kozubik, and K. Plevova are listed as co-inventors on a patent, which is owned by Masaryk University, Brno, that prevents the use of expression of Wnt-5a as a marker of CLL progression. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P. Janovska, L. Poppova, A. Kozubik, S. Pospisilova, S. Pavlova, V. Bryja

Development of methodology: P. Janovska, L. Poppova, M. Behal, M. Kaucka

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Janovska, L. Poppova, K. Plevova, H. Plesingerova, M. Behal, M. Kaucka, M. Hlozkova, O. Stehlikova, Y. Brychtova, M. Doubek, M. Machalova, S. Pavlova

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Janovska, L. Poppova, K. Plevova, P. Ovesna, M. Hlozkova, M. Borsky, O. Stehlikova, M. Doubek, S. Baskar, S. Pavlova, V. Bryja

Writing, review, and/or revision of the manuscript: P. Janovska, L. Poppova, K. Plevova, M. Kaucka, M. Doubek, S. Baskar, A. Kozubik, S. Pavlova, V. Bryja

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Janovska, L. Poppova, S. Baskar

Study supervision: S. Pospisilova, V. Bryja

The authors thank Lenka Bryjová and Hana Pecúchová (Masaryk University) for their technical assistance and the staff of the Department of Internal Medicine—Hematology and Oncology, Department of Transfusion Medicine and Department of Paediatric Otorhinolaryngology, University Hospital Brno for the help with collecting biological samples from CLL patients and non-CLL controls, and Mathiew Smith for language corrections.

This work was supported by grants from the Czech Science Foundation (301/11/0747), Masaryk University Student Grant for Specific Research MUNI/A/1180/2014 and MUNI/A/1398/2014, Ministry of Health of the Czech Republic (NT11217-5/2010, NT13493-4/2012, 15-29793A), Ministry of Education, Youth and Sports of the Czech Republic Developmental Fund project CZ.1.05/1.1.00/02.0068 (CEITEC), and Faculty Hospital Brno (FNBr65269705). This work was supported in part by the Intramural Research Program of the Center for Cancer Research, National Cancer Institute, NIH (SB).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.