Mechanisms and clinical significance of BIM phosphorylation in chronic lymphocytic leukemia

Chronic lymphocytic leukemia (CLL) is a relatively common B-cell malignancy with a highly variable clinical course.^1,2  A valuable marker for predicting disease behavior is the mutational status of the B-cell receptor (BCR) immunoglobulin variable region genes. Those with unmutated IGHV genes (U-CLL) have a relatively poor prognosis compared with patients with mutated IGHV genes (M-CLL).^3,4  Expression of ZAP-70 and the cell surface activation marker CD38 is also associated with poorer clinical outcome.⁵ 

Antigen signaling plays a key role in the accumulation of CLL cells through increased cell survival and proliferation.^6-8  Engagement of antigen appears to be ongoing in CLL,^9-11  and differences in responses to surface IgM (sIgM) stimulation may be a key determinant of clinical behavior.^7,8  For example, U-CLL is generally more open to sIgM signaling responses in vitro, whereas more M-CLL cases have reduced sIgM signal capacity, displaying a stronger “anergic” phenotype.^10,12  Antigen-mediated activation of downstream signaling pathways, which lead to proliferation and survival, may contribute to accumulation of malignant cells in vivo and to disease progression. Hence, these pathways are an attractive target for therapeutic attack.⁷  Microenvironment signals derived from stromal cells and immunocytes also play a key role in CLL, promoting cell survival and supporting proliferation in vivo.¹³ 

One key target for survival signals in CLL cells is MCL1, a pro-survival BCL2 family protein. MCL1 is induced in cells stimulated with anti-IgM,^14-16  and its expression is associated with poor clinical outcome.^17,18  MCL1 also plays an important role in microenvironment-derived signaling^19,20  and is essential for survival of CLL cells cocultured with HK cells, a follicular dendritic cell line.²¹  A further potential player is the proapoptotic BCL2 family protein BIM, a preferred dimerization partner of MCL1. BIM is expressed as 3 isoforms (BIM_EL, BIM_L, and BIM_S; supplemental Figure 1, available on the Blood Web site; see the Supplemental Materials link at the top of the online article) and is a critical regulator of apoptosis in normal and malignant B cells.²²  BIM functions as a tumor suppressor in B-cell malignancies^23-25  and is a key determinant of BCR-induced apoptosis in normal B cells, where it is required for the deletion of autoreactive cells in vivo.^23,26 

BIM function is positively and negatively regulated by transcription, phosphorylation, and degradation.²⁷  The best understood pathway involves ERK1/2-mediated phosphorylation of BIM_EL at S⁶⁹, which has been shown to promote cell survival.^28,29  In murine B cells, this occurred after stimulation via IgM.³⁰  S⁶⁹ phosphorylation triggers release of survival molecules, such as MCL1, from sequestration by BIM_EL,³¹  and also enhances proteosomal degradation of BIM_EL,^29,31,32  although it is not clear whether degradation is an inevitable consequence of phosphorylation. In contrast to S⁶⁹, inducible phosphorylation at other sites enhances the proapoptotic function of BIM. For example, phosphorylation at T⁵⁶ in BIM_L (equivalent to T¹¹⁶ in BIM_EL) promotes apoptosis via reduced binding to cytoskeleton components and increased sequestration of BCL2.^33-35 

Given the key role of BCR signaling in CLL and the known involvement of BIM downstream of the BCR in other systems, we have investigated regulation of BIM in CLL. Previous studies have shown that CLL cells express variable levels of BIM isoforms.^16,36-38  Moreover, BIM has been shown to be functionally important in apoptosis control in CLL cells. In particular, studies using BH3-mimetic peptides have demonstrated that BIM triggers the release of cytochrome c from mitochondria unless it is effectively sequestered by high levels of BCL2, a feature of CLL cells.³⁶  Two previous studies have investigated BIM regulation in CLL cells; and although they have demonstrated modulation of the level and/or phosphorylation of BIM_EL by anti-IgM,^37,38  the clinical significance and mechanisms of BIM modulation remain unclear.

In this study, we demonstrate that BIM isoforms undergo phosphorylation in CLL cells stimulated with anti-IgM or cocultured with HK cells, a model for microenvironment support. BIM phosphorylation was dependent on MEK1/2 (the upstream activating kinase for ERK1/2) and was associated with release of MCL1. Importantly, the extent of anti-IgM–induced BIM phosphorylation was strongly associated with IGHV mutation status and with disease progression, even within M-CLL, suggesting that this pathway may be an important determinant of disease behavior.

A summary of the clinical background of the CLL samples used is provided in Table 1. The study was performed after approval from the Southampton and Southwest Hampshire Research Ethics Committee, and informed consent was provided in accordance with the Declaration of Helsinki. Blood was obtained from CLL patients who attended outpatient clinics at the University Hospitals of Leicester, Royal Wolverhampton Hospitals, Southampton University Hospitals Trust, Royal Berkshire NHS Foundation Trust, and the Queen Alexandra Hospital, Portsmouth (all United Kingdom). The majority of samples were obtained at diagnosis/before therapy. Where treatment had taken place, this was at least 3 months before sample collection.

Blood samples were processed as previously described.¹⁰  After cryopreservation, cells were allowed to recover by incubation for 1 hour at 37°C in RPMI 1640 (PAA Laboratories) supplemented with 10% (volume/volume) FBS (PAA Laboratories), 1mM pyruvate, 2mM glutamine, and nonessential amino acids (all from Lonza). Cell viability determined by trypan blue exclusion was more than or equal to 90%. The proportion of CD5⁺/CD19⁺ CLL cells was more than 90% in the majority of samples (range, 86%-99%). Normal B cells were isolated from peripheral blood of healthy donors using the B cell Isolation Kit II (Miltenyi Biotec) according to the manufacturer's protocol.

IGHV mutation status, expression of cell surface CD5, CD19, CD38, and sIgM, and intracellular ZAP-70 were determined as previously described.^10,12  IgM signaling capacity was determined by measuring the percentage of cells fluxing calcium after stimulation with goat F(ab′)₂ anti-IgM as described elsewhere (Table 1).¹⁰  Samples with more than 5% of responding cells were classed as positive intracellular Ca²⁺ signalers.¹⁰  Disease was considered “progressive” if the patient went on to receive treatment at any point after diagnosis.

Ramos cells were obtained from the European Collection of Cell Cultures. The HK cell line was a kind gift from Dr Yong Sung Choi (Laboratory of Cellular Immunology, New Orleans, LA). Both cell lines were cultured in supplemented RPMI 1640. For coculture experiments, cells were plated at a CLL:HK cell ratio of 20:1.

For sIgM stimulation, CLL cells were cultured at 1 × 10⁷/mL and incubated with soluble 20 μg/mL goat F(ab′)₂ anti–human IgM (Southern Biotechnology) as previously described.¹²  In experiments using chemical inhibitors, cells were pretreated with U0126 (10μM; Sigma-Aldrich), PD0325901 (100nM; Merck Chemicals), or MG-132 (1μM; Merck Chemicals) for 1 hour before stimulation with anti-IgM.

For immunoblotting experiments, cells were lysed directly in SDS sample loading buffer (Cell Signaling Technology) containing DTT, protease inhibitors, and phosphatase inhibitors (both from Sigma-Aldrich). After sonication, lysates were resolved on 10% to 20% Tris-HCl gradient gels (Bio-Rad), and immunoblotting was performed with the following antibodies: anti-BIM (2819), anti-S⁶⁹ phospho-BIM (4581), anti-ERK1/2 (9102), anti-T²⁰²/Y²⁰⁴ phospho-ERK1/2 (9101), anti-HA (C29F4 all from Cell Signaling Technology), anti-ubiquitin (P4D1), anti-MCL1 (S-19, both from Santa Cruz Biotechnology), anti-MCL1 (clone 22; BD Biosciences) and anti–β-actin (2066; Sigma-Aldrich). Secondary HRP-conjugated antibodies were from GE Healthcare. Densitometry analysis of images was performed using Quantity One Version 4.6.9 software (Bio-Rad). For quantitation of BIM phosphorylation, the faster migrating (nonphosphorylated) and slower migrating (phosphorylated) isoforms were quantified separately. For each BIM isoform, the proportion of phosphorylation was determined by calculating the amount of phosphorylated BIM as a percentage of total BIM. Maximal levels of BIM phosphorylation after stimulation with anti-IgM were used for further comparisons.

Induction of ERK1/2 phosphorylation was also determined by intracellular flow cytometry, using anti-T²⁰²/Y²⁰⁴ phospho-ERK1/2 conjugated to AlexaFluor-488 (BD Biosciences), as described elsewhere.³⁹ 

Cell pellets of 5 × 10⁶ Ramos cells were resuspended in 500 μL NP40 lysis buffer (1% [volume/volume] NP-40, 50mM Tris-HCl pH 8.0, 150mM NaCl, 2mM EDTA, protease inhibitors, and phosphatase inhibitors), lysed on ice for 15 minutes, and clarified by centrifugation. The lysate was rotated overnight at 4°C with 2 μg rat monoclonal anti-BIM (14A8; Millipore) or isotype control rat IgG2a (a kind gift of Professor M. J. Glennie, Southampton, United Kingdom). Alternatively, the lysate was incubated with 2 μg anti-MCL1 raised in rabbit (S-19; Santa Cruz Biotechnology) or with 2 μg anti-MCL1 raised in mouse (clone 22; BD Biosciences) and the appropriate isotype controls; polyclonal rabbit IgG (Abcam) or mouse IgG₁ (BD Biosciences). Protein G-Sepharose beads (GE Healthcare) were then added; and after 6 hours, the immune complexes were collected by centrifugation. The beads were washed 4 times in NP-40 lysis buffer, and bound proteins eluted in 50 μL of elution buffer (200mM glycine, pH 2.8). Densitometry analysis allowed quantitation of coprecipitated proteins. For comparison between experiments, levels were set to 1.0 in unstimulated cells, and any effects of IgM stimulation are shown as proportional to this amount.

pCAN-HA vectors containing HA-tagged wild-type or mutant rat BIM_EL have been described.²⁸  For transfection, 8 × 10⁶ Ramos cells were resuspended in 250 μL Optimem (Invitrogen) and electroporated at 250 V, 1050 μF, 900c in the presence of 15 μg of plasmid DNA. Equibio's Easyject Plus D2000 was used for electroporations. Cells were incubated at 1 × 10⁶/mL for 24 hours and then recounted before anti-IgM stimulation.

Cells were permeabilized in ice-cold 70% (volume/volume) ethanol and, after overnight storage at 4°C, washed and resuspended in phosphate-citrate buffer (0.2M Na₂HPO₄, 0.1M citric acid). After 10 minutes of incubation with 100 μg/mL RNase (QIAGEN), propidium iodide (PI; BD Biosciences) was added at 50 μg/mL. Cells were analyzed on a FACSCalibur flow cytometer (BD Biosciences). Alternatively, nonpermeabilized cells were resuspended in 300 μL annexin V binding buffer (BD Biosciences) containing 1.25 μL annexin V–FITC (Protein Core Facility, University of Southampton) and 0.05 μg PI. After 15 minutes of incubation in the dark, cells were analyzed on a FACSCalibur flow cytometer.

Statistical analyses were performed using GraphPad Prism Version 4.03 software. Mann-Whitney and paired t tests were 2-tailed with 95% confidence intervals.

We analyzed the expression of BIM in CLL using immunoblotting. Consistent with previous publications,^16,36-38  CLL cells expressed readily detectable levels of the 2 major BIM isoforms, BIM_L and BIM_EL (Figure 1A). Expression of the smallest isoform (BIM_S) was very low and generally only detected after long exposures of immunoblots. Using RT-PCR analysis, we confirmed that the isoforms expressed in CLL cells corresponded to the 3 major BIM splice variants (BIM_EL, BIM_L, and BIM_S), and there was no evidence for expression of substantial levels of other, noncanonical splice variants (data not shown).

We analyzed basal and sIgM-induced BIM phosphorylation in a cohort of 28 CLL samples comprising 11 U-CLL and 17 M-CLL, of which 14 were anti-IgM signaling responsive (intracellular Ca²⁺) and 14 were nonresponsive (Table 1). Consistent with previous signaling studies,¹⁰  the percentage of cells fluxing intracellular Ca²⁺ after stimulation with anti-IgM correlated positively with the presence of unmutated IGHV genes (P = .024), demonstrating that this cohort is representative of those previously studied (data not shown).

Immunoblot analysis revealed variable levels of basal and induced phosphorylation of BIM_EL and/or BIM_L, detected by accumulation of slower migrating isoforms (Figure 1A). BIM phosphorylation was quantified using digital image analysis. Repeat analyses of 6 CLL samples in up to 4 independent experiments demonstrated that quantitation of BIM phosphorylation was highly reproducible (SD values for repeat measurements ranged from 0.2% to 5.8%). We therefore selected a cut-off of more than 5% to categorize samples as positive for basal (ie, > 5% BIM_EL/BIM_L phosphorylation in unstimulated cells) or induced BIM_EL/L phosphorylation (ie, > 5% maximal increase in BIM phosphorylation in stimulated cells). We also compared anti-IgM–induced BIM phosphorylation in fresh and cryopreserved material from 3 samples, which showed a range of BIM_EL responses (7.1%-45% BIM_EL phosphorylation) to ensure that responses were not affected by cryopreservation. Overall, the results were similar for fresh and cryopreserved cells, and the variation in the percentage induced phosphorylation of BIM_EL and BIM_L was less than or equal to 4%.

Basal BIM_EL and BIM_L phosphorylation was detected in several samples but only exceeded the 5% cut-off in a small proportion (5 of 28; 18%; eg, CLL 291 in Figure 1A). In most cases, anti-IgM stimulation was associated with increased phosphorylation of BIM_EL and/or BIM_L (Figure 1A for representative examples; and Table 1). Increased phosphorylation was rapid (within 30 minutes) but transient and typically lost 2 to 4 hours after stimulation (Figure 1A). Using the more than 5% cut-off, increased BIM_EL/BIM_L phosphorylation was detected in 22 of 28 samples (79%) after stimulation with anti-IgM. In general, changes in BIM_EL and BIM_L were highly correlated (Figure 1B), but a small number of samples (3 of 28) showed selective phosphorylation of BIM_EL. In the samples with constitutive BIM_EL/BIM_L phosphorylation, anti-IgM stimulation caused a further increase in phosphorylation. Overall, there was a close correlation between intracellular Ca²⁺ and BIM phosphorylation responses (P = .014 and P = .009 for BIM_EL and BIM_L, respectively; Figure 1C), although there appeared to be a subset of intracellular Ca²⁺ nonresponsive samples that did show substantial BIM phosphorylation (especially for BIM_EL).

We also investigated BIM modulation in polyclonal B cells from blood of healthy donors. In contrast to CLL cells, anti-IgM induced selective phosphorylation of BIM_EL with no evidence of phosphorylation of BIM_L (n = 3; Figure 1D).

We next compared induction of BIM_EL/BIM_L phosphorylation with the clinical features of CLL (Figure 2A-B). The results showed a strong correlation between high levels of phosphorylation and unmutated IGHV genes (P = .003 and P < .0001 for BIM_EL and BIM_L, respectively). The levels of BIM_EL/BIM_L phosphorylation were higher in CD38⁺ or ZAP-70⁺ cases, but this did not reach statistical significance. There was a striking correlation between BIM phosphorylation and clinical behavior. Using requirement for treatment at any time after diagnosis as a surrogate marker of progressive disease, there was a very strong correlation between higher levels of induced BIM_EL phosphorylation and progressive disease (P = .0008). The same was true for BIM_L (P = .0005). The strong correlations remained if the stage A cases were considered alone (ie, one stage C and 2 stage B samples eliminated; P = .0009 and P = .0012 for BIM_EL and BIM_L, respectively; data not shown).

All cases of U-CLL showed increased anti-IgM–induced BIM phosphorylation (Figure 2A-B). In contrast, M-CLL fell into 2 relatively discrete subsets, particularly for BIM_EL, with 6 of 17 cases having a substantial response (> 50% phosphorylation) and 11 of 17 having much lower responses (< 25%). Interestingly, this was associated with progressive disease. Within the M-CLL subset, statistically significant correlations were found between BIM_EL/BIM_L phosphorylation and progressive disease (P = .043 and P = .013, for BIM_EL and BIM_L, respectively; Figure 2C).

Thus, higher levels of anti-IgM–induced BIM phosphorylation are a feature of U-CLL and are associated with progressive disease. In addition, the minor proportion of M-CLL cases with progressive disease appear to correlate with higher levels of sIgM-induced BIM phosphorylation.

ERK1/2 is activated downstream of sIgM in normal and CLL B cells and is a key upstream regulatory kinase for BIM_EL.^15,28,40  We therefore analyzed IgM-induced phosphorylation of ERK1/2 in CLL cases and compared these responses with BIM phosphorylation. Semiquantitative immunoblot analysis demonstrated an increase in phosphorylation of ERK1/2 at T²⁰²/Y²⁰⁴ within 2 to 5 minutes, which then declined at later time points (Figure 3A). Intracellular flow cytometry was used to quantify ERK1/2 as described previously (Table 1).³⁹  ERK1/2 phosphorylation increased rapidly in the majority of CLL samples after stimulation with anti-IgM. Consistent with previous results in CLL cells,¹⁵  we rarely detected anti-IgM induced activation of JNK1/2, a known BIM kinase³³ , and where present, it was extremely weak (immunoblot analysis; data not shown).

Using an arbitrary cut-off value of more than 1.2-fold compared to unstimulated cells, ERK1/2 responses were detected in 15 of 21 (71%) samples analyzed by intracellular flow cytometry. Phospho-ERK1/2 signaling capacity was positively associated with BIM_EL phosphorylation (P = .039; Figure 3B). There was a trend toward increased BIM_L phosphorylation and phospho-ERK1/2 responses, but this failed to reach significance (P = .080). The extent of ERK1/2 phosphorylation correlated with intracellular Ca²⁺ responses (P = .027; Figure 3C). Similar to BIM phosphorylation, phospho-ERK1/2 responses correlated positively with progressive disease (P = .0185; Figure 3D).

Given the positive correlations between anti-IgM–induced ERK1/2 phosphorylation, BIM phosphorylation, and progressive disease, we directly investigated the role of MEK1/2 in anti-IgM–induced BIM phosphorylation. MEK1/2 directly activates ERK1/2. CLL cells were pretreated with the MEK1/2 inhibitor U0126 before stimulation with anti-IgM. U0126 completely abrogated the increase in BIM_EL phosphorylation induced by anti-IgM (Figure 4A). Surprisingly, U0126 also prevented induction of BIM_L phosphorylation, even though this isoform lacks the ERK-docking domain found within BIM_EL (supplemental Figure 1). Immunoblot analysis of phospho-ERK1/2 confirmed that U0126 effectively inhibited MEK1/2 activity in CLL cells.

We selected Ramos cells as a transfectable cell line model to probe mechanisms and responses to BIM phosphorylation. Previous studies have used Ramos cells to study BIM phosphorylation.²⁹  To validate Ramos cells as an appropriate model, we first investigated their response to anti-IgM. Similar to CLL cells, engagement of IgM induced rapid phosphorylation of ERK1/2, BIM_EL, and BIM_L, and inhibition of MEK1/2 ablated phosphorylation of both BIM isoforms (Figure 4B).

An antibody specific for S⁶⁹-phosphorylated BIM was used to determine whether this site was undergoing phosphorylation in anti-IgM–treated cells. Stimulation of Ramos cells with anti-IgM clearly increased the levels of S⁶⁹-phosphorylated BIM_EL (Figure 5A). A similar response was observed in CLL cells from case CLL332.

To further investigate the role of S⁶⁹ in BIM_EL phosphorylation, Ramos cells were transfected with expression plasmids encoding wild-type and mutant (S⁶⁵ → A) rat HA-tagged BIM_EL proteins. Rat BIM expression plasmids were used because these have been used in previous BIM phosphorylation studies.²⁸  Rat S⁶⁵ is equivalent to human S⁶⁹. In cells transfected with wild-type HA-BIM_EL, anti-IgM induced high levels of phosphorylation with 78% ± 8% (mean ± SD of 3 experiments) of the detected HA-BIM_EL protein present as slower migrating forms (Figure 5B). In contrast, anti-IgM stimulation of cells transfected with the S⁶⁵ → A mutant expression construct resulted in significantly reduced HA-BIM_EL phosphorylation (26% ± 6%, mean ± SD of 3 experiments). Anti-IgM induced phosphorylation of either protein was completely blocked by the MEK1/2 inhibitors U0126 and PD035901 (Figure 5B). (PD035901 was used in combination with U0126 to ensure complete inhibition of MEK1/2 activity in Ramos cells.) The presence of phosphorylated HA-BIM_EL in the S⁶⁵ → A mutant demonstrates that BIM_EL is phosphorylated on additional sites to S⁶⁵, and the inhibitor data show that all of these phosphorylations are dependent on MEK1/2.

As mutation of S⁶⁵ failed to ablate BIM_EL phosphorylation, we extended the analysis to other potential phosphorylation sites. Combined mutation of S⁵⁵ → A, S⁶⁵ → A, S⁷³ → A, S¹⁰⁰ → A, T¹¹² → A, and S¹¹⁴ → G (rat amino acid numbering) completely ablated BIM_EL phosphorylation (Figure 5C). Mutation of S⁷³ → A or S¹⁰⁰ → A alone caused a modest (16% ± 5% and 9% ± 2%, respectively) decrease in BIM_EL phosphorylation relative to wild-type (mean ± SD, n = 3; Figure 5C). Mutation of S⁵⁵ → A had no effect on BIM_EL phosphorylation. Mutation of other individual phospho-acceptor sites gave variable results, which did not provide evidence for an important role.

In summary, BIM_EL S⁶⁹ appears to be a major target for MEK1/2-dependent phosphorylation downstream of sIgM, although other sites are probably involved.

Consequences of BIM_EL S⁶⁹ phosphorylation include reduced binding to pro-survival molecules and BIM_EL degradation.²⁷  However, we did not observe any substantial changes in BIM_EL expression in CLL cells after stimulation with anti-IgM for up to 8 hours, a time point by which BIM phosphorylation is typically reversed (Figure 1A). Longer time course experiments confirmed that, consistent with the transient nature of BIM_EL phosphorylation in CLL cells, BIM_EL expression did not decrease after sIgM stimulation, even after incubation times up to 40 hours (supplemental Figure 2). Moreover, cotreatment of cells with the proteasome inhibitor MG-132 also did not increase BIM_EL expression (supplemental Figure 2). This suggests BIM phosphorylation is not linked to substantial degradation in anti-IgM–stimulated CLL cells under these conditions.

MCL1 is a preferred binding partner for BIM. Similar to a previous study,¹⁵  analysis of MCL1 in primary CLL samples (Figure 6A) and Ramos cells (Figure 6B; and data not shown) demonstrated that sIgM stimulation with soluble anti-IgM did not trigger significant changes in MCL1 expression. Given that BIM phosphorylation is known to influence binding to prosurvival BCL2-family proteins, we used Ramos cells to analyze the effect of IgM-induced BIM phosphorylation on the association between BIM and MCL1. We focused on MCL1 because the molecular events controlling this interaction are relatively well characterized.³¹  MCL1 was detected in BIM immunoprecipitates, confirming that these 2 proteins interact in unstimulated cells (Figure 6B). After 30 minutes of stimulation with anti-IgM, BIM_EL phosphorylation substantially increased to 80% ± 5% (n = 6, mean ± SD; Figure 6C). Concurrent with this increase in BIM phosphorylation, the relative level of MCL-1 coprecipitated by BIM was reduced from 1.0 in unstimulated cells to 0.39 ± 0.10 (mean ± SD, n = 6; Figure 6C, representative example in Figure 6B). This demonstrates that anti-IgM–induced BIM phosphorylation is associated with reduced BIM/MCL1 binding.

Reverse immunoprecipitations using an MCL1 specific antibody confirmed that both BIM_EL and BIM_L isoforms were present in MCL1 complexes in untreated cells (Figure 6D). After stimulation with anti-IgM, the relative amount of BIM coimmunoprecipitated with MCL1 was reduced from 1.0 in unstimulated cells to 0.16% ± 0.07% for BIM_EL and 0.22% ± 0.03% for BIM_L (mean ± SD, n = 3; representative immunoblot experiment shown in Figure 6D). Residual BIM/MCL1 complexes contained phosphorylated BIM_EL/BIM_L, demonstrating that phosphorylated BIM isoforms retain some ability to bind to MCL1. To specifically investigate the binding between MCL1 and S⁶⁹-phosphorylated BIM_EL, MCL1 immunoprecipitates were immunoblotted using the S⁶⁹-phospho-specific antibody (Figure 6E). This analysis revealed an apparent low-level residual interaction between MCL-1 and S⁶⁹-phosphorylated BIM_EL. Therefore, phosphorylation of both BIM_EL and BIM_L is associated with a release of the MCL1 survival protein.

In addition to antigen, microenvironment-derived signals also play an important role in maintaining the survival of CLL cells.¹³  HK cells are a well-validated model to investigate pro-survival effects of the microenvironment on CLL cells.²¹  As previously shown, coculture with HK cells alone reduced spontaneous apoptosis of CLL cells (Figure 7A). In 6 of 10 samples, HK cell coculture was associated with detectable increases in phosphorylation of BIM_EL, but not BIM_L (Figure 7B). In general, levels of BIM_EL phosphorylation were lower than those induced by anti-IgM (median increase in responders was ∼ 50% for anti-IgM and ∼ 15% for HK cell coculture). Culture with HK cells increased ERK1/2 phosphorylation in the CLL cells, and U0126 completely prevented induction of BIM_EL phosphorylation (Figure 7C), demonstrating that the HK-induced BIM_EL phosphorylation detected is MEK1/2 pathway dependent.

Our results implicate modulation of BIM by ERK1/2 as an important survival response in CLL cells, and we therefore investigated whether MEK1/2 inhibition influenced CLL cell apoptosis. We first focused on the effects downstream of IgM engagement. In some samples, inhibiting MEK1/2 with U0126 enhanced apoptosis, but overall there was not a consistent response (data not shown). As previously described, anti-IgM stimulation increased, decreased, or had no effect on apoptosis, depending on the specific sample investigated.^14,15,40-42  The variable effect of U0126 presumably reflects the overall heterogeneity of responses to anti-IgM stimulation.

We therefore next investigated the effects of U0126 on HK cell-mediated CLL cell survival, where survival responses were relatively homogeneous (Figure 7A). In 2 samples (CLL189 and CLL383), U0126 treatment alone increased levels of apoptosis; and in 3 samples (CLL189, CLL383, and CLL431), U0126 reduced the pro-survival effects of HK cell coculture (Figure 7D). In the fourth sample (CLL410), HK cell coculture had a modest pro-survival effect, but apoptosis levels were unaffected by U0126. These results demonstrate the importance of the MEK1/2 pathway and its downstream effect on ERK1/2 and BIM phosphorylation for CLL cell survival.

Multiple lines of evidence support a role for antigen signaling and the microenvironment in development and progression of CLL.^7,8,13  There is a strong indication that CLL cells resist apoptosis, but the mechanisms potentially linking BCR signaling to this resistance are not completely understood. Previous studies of signaling have generally focused on upstream kinases. However, it is the impact of these pathways on downstream regulators of cell survival that ultimately influences disease behavior. In this study, we focus on the pivotal role of the proapoptotic molecule BIM. We find that the 2 major isoforms, BIM_EL and BIM_L, undergo phosphorylation in CLL cells stimulated with anti-IgM and that activation of this pathway in vitro is closely associated with disease behavior. Coculture with HK cells alone also promotes BIM_EL phosphorylation, suggesting that BIM may serve to coordinate microenvironment and antigen-mediated survival signals.

A key observation from this study is the close association between the ability of sIgM to induce BIM phosphorylation and progressive disease. This observation is to some extent expected, given the association between signaling and markers of poor prognosis disease, such as unmutated IGHV genes. However, the striking new observation is that IgM-induced BIM phosphorylation also correlated with progressive disease when analyzed specifically within the M-CLL subset. Previously, studies from our group and those of others, using a range of signaling measures including intracellular Ca²⁺ fluxes, have demonstrated that sIgM responses in vitro were more common in U-CLL.^10,12,42  In M-CLL, the overall frequency of such responses was reduced, but a significant proportion of individual M-CLL samples retained signaling ability.¹⁰  No clear association of the operation of this signaling pathway with prognosis in M-CLL emerged. In contrast, our present findings suggest that sIgM-mediated BIM phosphorylation may reveal a subset of M-CLL at increased risk of progression. Because BIM phosphorylation is more directly coupled to regulation of apoptosis than intracellular Ca²⁺ fluxes, it may provide a particularly strong marker of functionally relevant responses to antigen engagement, and hence of clinical behavior.

Multiple BIM phosphorylation sites have been identified which are linked to both suppression and induction of apoptosis. We confirmed previous findings,³⁷  demonstrating that BIM_EL S⁶⁹ was phosphorylated in anti-IgM–treated CLL cells, and this was also shown in Ramos cells. Using site-directed mutagenesis, we have now demonstrated that S⁶⁹ is probably the major site but that there are potential roles for S⁷⁷ and S¹⁰⁴, indicating that, as in other systems, multisite phosphorylation of BIM_EL occurs after sIgM stimulation. MEK1/2 inhibition blocked all BIM_EL phosphorylation, confirming ERK1/2 as the likely principal regulatory kinase. Consistent with this, anti-IgM–induced ERK1/2 phosphorylation correlated with BIM_EL phosphorylation, and also with disease progression.

We did not find strong evidence for anti-IgM–induced BIM_EL degradation in CLL cells, despite S⁶⁹ phosphorylation. This could be explained by our use of soluble rather than solid-phase anti-IgM. It is difficult to know the molecular form of candidate antigens in vivo, but they might include bacteria and/or apoptotic cells^43,44  and immobilized anti-IgM appears to have a stronger pro-survival effect in vitro.^15,41  It is possible that, in cells treated with soluble anti-IgM, ERK1/2-mediated phosphorylation fails to engage RSK1/2 phosphorylation activity at S⁹³, S⁹⁴, and S⁹⁸, which is required for binding of the E3 ubiquitin ligase βTrCP and subsequent BIM_EL degradation.⁴⁵  Regardless, BIM_EL phosphorylation appears to be a survival response because it was associated with release of MCL1. Interestingly, we detected a low level of interaction between MCL1 and S⁶⁹ phosphorylated BIM_EL, suggesting that, in contrast to BIM_EL-overexpressing HEK293 cells,³¹  S⁶⁹ phosphorylation is not sufficient to prevent MCL1 binding, but favors dissociation.

We also detected MEK1/2-dependent BIM_L phosphorylation in most responsive anti-IgM–treated CLL cells, and in Ramos cells. This is surprising because BIM_L lacks the serine residue equivalent to BIM_EL S⁶⁹ and the ERK docking domain required for BIM_EL phosphorylation.⁴⁶  Compared with BIM_EL, there was a weaker correlation between levels of BIM_L and ERK1/2 phosphorylation, which may reflect a requirement for additional intermediates to effectively engage ERK1/2 signaling in the absence of an ERK-docking domain. It is possible that ERK1/2 “gains access” to BIM_L by binding to partner proteins, such as MCL1, which also undergoes MEK1/2-dependent phosphorylation. Differential expression of “cofactors” specifically required for BIM_L phosphorylation may also explain why BIM_EL was selectively phosphorylated in sIgM-stimulated normal B cells. Alternatively, the selective phosphorylation of BIM_EL may be related to signal strength because BIM_L was not phosphorylated in CLL cells cocultured with HK cells, where overall levels of BIM phosphorylation were low compared with anti-IgM–treated cells. Further characterization of the mechanisms and functional consequences of BIM_L phosphorylation are required, but this response is also probably linked to survival because coimmunoprecipitation demonstrated that BIM_L also sequestered MCL1 and anti-IgM stimulation was associated with release of MCL1 from both BIM_EL and BIM_L in Ramos cells.

Engagement of sIgM in vivo by antigen, within the context of specific tissue microenvironments, is thought to play a key role in CLL, at least in part by activation of survival pathways leading to suppressed apoptosis and resistance to therapy. Previous studies have revealed a role for increased MCL1 expression in mediating both microenvironment and sIgM-mediated survival signals triggered by immobilized anti-IgM.^14-16,21  Our studies using soluble anti-IgM indicate that modulation of BIM phosphorylation is a parallel pathway activated in CLL cells and that coordinated regulation of the BIM:MCL1 complex is probably a regulatory “node,” critical for controlling CLL cell apoptosis in response to multiple signals (supplemental Figure 3). Thus, whereas sIgM and microenvironment signaling pathways can up-regulate MCL1 expression, parallel phosphorylation of BIM prevents inhibition of this pro-survival protein, by releasing it from sequestration.

Whereas previous studies have shown PI3K/AKT signaling to play a major role in the induction of MCL1 in response to sIgM stimulation,^14,15  our work suggests that MEK1/2/ERK1/2 signaling also functions as a survival pathway in CLL cells. Consistent with this, a recent study demonstrated that the Raf inhibitor sorafenib promotes apoptosis in CLL cells and overcomes survival effects of coculture with nurse-like cells.⁴⁷  However, it is unclear to what extent this is mediated via effects on the downstream ERK1/2 pathway, or to off-target effects of the inhibitor that can inhibit additional kinases. Moreover, inhibition of PI3Kδ using CAL-101 interferes with anti-IgM–induced ERK1/2 activation, demonstrating that ERK1/2 may lie downstream of PI3K after sIgM stimulation in CLL cells.⁴⁸  Thus, PI3K may coordinate survival signaling via both ERK1/2/BIM- and MCL1-dependent pathways. Our recent immunohistochemical studies have demonstrated ERK1/2 phosphorylation in proliferation centers of small lymphocytic lymphoma/CLL lymph nodes, confirming that this pathway is activated in vivo.⁴⁹  In this study, we have shown that MEK1/2 inhibition reverses the survival effect of coculture with HK cells, at least in cases where HK-mediated survival was evident. MEK1/2 inhibition results were more complex in cells treated with anti-IgM, perhaps expectedly as the effect of sIgM stimulation alone results in highly variable responses in vitro. The PI3Kδ inhibitor CAL-101 has shown impressive responses in early clinical trials, especially in CLL.⁵⁰  Dual inhibition of PI3K and MEK1/2 may be an effective therapeutic strategy to deprive CLL cells of sIgM and microenvironment-derived survival signaling in vivo.

The authors thank Dr Yong Sung Choi (New Orleans, LA) for the gift of HK cells; Professor Martin Glennie (Southampton, United Kingdom) for the gift of rat IgG2a antibody; Drs Vlad Malykh, Helen McCarthy, Abe Jacob, Ben Kennedy, and Henri Grech, and Mr Richard Palmer and colleagues for providing CLL samples and associated data; the patients who donated clinical samples; Mrs Isla Henderson for characterization of CLL samples; Professor Christian Ottensmeier for support; and Dr Andrew Steele for helpful comments.

A.P. was supported by a PhD studentship from the University of Southampton. This work was also supported by Kay Kendall Leukaemia Fund, Cancer Research UK, the Southampton Experimental Cancer Medicine Centre, and Tenovus Solentside.

Contribution: A.P., C.I.M., J.E.A., S.K., A.S.D., and K.N.P. performed the research and analyzed data; A.P., K.N.P., S.J.C., F.K.S., and G.P. designed the research and analyzed data; A.P. and G.P. wrote the initial draft of the manuscript; and all authors contributed to the modification of the draft and approved the final submission.

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

Correspondence: Graham Packham, Somers Cancer Sciences Building, Southampton General Hospital (MP824), Southampton, SO16 6YD, United Kingdom; e-mail: gpackham@soton.ac.uk.