Research ArticleCancer

PP2A inhibition sensitizes cancer stem cells to ABL tyrosine kinase inhibitors in BCR-ABL+ human leukemia

See allHide authors and affiliations

Science Translational Medicine  07 Feb 2018:
Vol. 10, Issue 427, eaan8735
DOI: 10.1126/scitranslmed.aan8735

Drug pair enABLes killing of leukemia

Imatinib, the classic targeted drug for the treatment of cancer, was designed to target the BCR-ABL fusion protein in chronic myeloid leukemia and has saved many patients’ lives. Unfortunately, some leukemias are resistant to imatinib despite having the BCR-ABL translocation, and others can develop resistance during treatment. Moreover, imatinib generally does not eradicate the leukemic stem cells and therefore requires continued treatment to maintain efficacy, so combination approaches are still needed. Lai et al. discovered that protein phosphatase 2A is a therapeutic target in imatinib-insensitive leukemia cells, including stem cells, and that the combination of imatinib and related drugs with PP2A inhibition effectively kills these cancer cells.


Overcoming drug resistance and targeting leukemic stem cells (LSCs) remain major challenges in curing BCR-ABL+ human leukemia. Using an advanced drug/proliferation screen, we have uncovered a prosurvival role for protein phosphatase 2A (PP2A) in tyrosine kinase inhibitor (TKI)–insensitive leukemic cells, regulated by an Abelson helper integration site–1–mediated PP2A–β-catenin–BCR-ABL–JAK2 protein complex. Genetic and pharmacological inhibition of PP2A impairs survival of TKI nonresponder cells and sensitizes them to TKIs in vitro, inducing a dramatic loss of several key proteins, including β-catenin. We also demonstrate that the clinically validated PP2A inhibitors LB100 and LB102, in combination with TKIs, selectively eliminate treatment-naïve TKI-insensitive stem and progenitor cells, while sparing healthy counterparts. In addition, PP2A inhibitors and TKIs act synergistically to inhibit the growth of TKI-insensitive cells, as assessed by combination index analysis. The combination eliminates infiltrated BCR-ABL+ blast cells and drug-insensitive LSCs and confers a survival advantage in preclinical xenotransplant models. Thus, dual PP2A and BCR-ABL inhibition may be a valuable therapeutic strategy to synergistically target drug-insensitive LSCs that maintain minimal residual disease in patients.


Protein tyrosine kinases are major therapeutic targets in various cancers, and several tyrosine kinase inhibitors (TKIs) have been successfully applied as molecularly targeted cancer therapies. In particular, the TKIs imatinib mesylate (IM), dasatinib (DA), and nilotinib (NL), which specifically target the kinase activity of BCR-ABL in chronic myeloid leukemia (CML), have transformed CML from a fatal disease to a manageable disease (14). Although these TKIs have demonstrated remarkable clinical efficacy in chronic phase (CP) CML, TKI monotherapies are not curative, and only about 10% of CP CML patients can discontinue TKI treatment and maintain therapy-free remission (57). Reduced efficacy of TKIs in treating accelerated phase and blast crisis (BC) CML and the development of primary and acquired resistance to these compounds remain major challenges (812). BCR-ABL+ acute lymphoblastic leukemia (ALL) closely resembles the aggressive lymphoid BC of CML and is also prone to relapse with current TKI monotherapies (13, 14). Moreover, TKIs are much less effective in eradicating leukemic stem cells (LSCs), the key cell population that generates minimal residual disease (MRD) and drives disease relapse (7, 8, 1517). Hence, therapeutic agents or treatment strategies that simultaneously overcome resistance to TKIs and target LSCs are urgently needed.

One candidate is Abelson helper integration site–1 (AHI-1, also known as Jouberin), a scaffold protein with multiple protein interaction domains (an N-terminal coiled coil domain, SH3 domain, two PEST (proline-glutamic acid-serine-threonine) motifs, and seven WD40 repeats), which is up-regulated in CML LSCs, together with BCR-ABL (18, 19). Suppression of AHI-1 increases the TKI sensitivity of IM nonresponder stem/progenitor cells or blast cells from patients in BC (20). AHI-1 directly interacts with BCR-ABL and JAK2 and mediates their signaling activity via its WD40 domain and N-terminal region, respectively (21). A combination of TKI and JAK2 inhibitors inhibits the growth of TKI-insensitive CML stem/progenitor cells compared to single agents, in vitro, and in animal models of leukemia (21, 22). These observations suggest an important role for AHI-1 in stabilizing the AHI-1–BCR-ABL–JAK2 complex, validating it as a potential therapeutic target. However, no specific AHI-1 inhibitors are currently known. Moreover, the abnormal expression of AHI-1 and its mutations has been reported in many other diseases (2326). In particular, AHI-1 interacts with β-catenin and facilitates its nuclear translocation in kidney cells in cystic kidney ciliopathy (27), and it remains to be determined whether it can also facilitate β-catenin signaling in BCR-ABL+ CML/ALL.

Protein phosphatase 2A (PP2A) is a serine/threonine phosphatase composed of scaffold (A), regulatory (B), and catalytic (C) subunits, with the specificity and activity of the PP2A holoenzyme being regulated by the B regulatory subunit (28, 29). The different subunits of PP2A can theoretically give rise to four different combinations of PP2A A to C heterodimers (30) and 92 heterotrimers, which provide the means for the vast heterogeneity and specificity to regulate numerous signaling cascades involved in proliferation, cell cycle control, adhesion, migration, and metabolism (29). Targeting PP2A as a therapeutic strategy has recently gathered a lot of momentum, yet the complexity and heterogeneity of PP2A holoenzymes have meant that the antitumor effects of activating or inhibiting PP2A activity are still largely not understood. In hematological cancers, pharmacological activation of PP2A, either by increasing cyclic adenosine monophosphate or by inhibiting SET (an endogenous inhibitor of PP2A), can suppress cell growth and enhance apoptosis of CML blast cells (31, 32). Several studies demonstrated the efficacy of PP2A inhibitors, especially in combination with chemo- or radiotherapies in various cancers [reviewed by (33)], and it remains to be determined whether this extends to hematological cancers and, more specifically, to TKI-insensitive stem cells.

Here, we used the Prestwick compound library/proliferation screen and identified a role for PP2A in regulation of the properties of LSCs. Its inhibition, in combination with a TKI, synergistically targets BCR-ABL+ blast cells and drug-insensitive LSCs in vitro and in vivo. Mechanistically, dual inhibition of PP2A and BCR-ABL disrupts several AHI-1–mediated signaling molecules, particularly PP2A-mediated protein degradation of β-catenin and inhibition of its downstream target genes.


A specific PP2A inhibitor is identified in AHI-1–transduced CML cells using an advanced drug/proliferation screen

To identify potential inhibitors of AHI-1, we screened the Prestwick Chemical Library, containing 1200 U.S. Food and Drug Administration–approved drugs with known bioavailability and safety in humans, using AHI-1–transduced K562 cells expressing a yellow fluorescent protein (YFP) marker, alone or in combination with 0.1 μM IM. We used a low dose of IM because it reduced BCR-ABL phosphorylation without affecting the viability of the cells (fig. S1, A to C). After treatment, the cells were analyzed using a high-content screening system, which allowed for comparison of the rate of growth and assessment of AHI-1 expression by measuring YFP intensity. In this screen, 28 compounds demonstrated >80% growth inhibition, with three inhibiting AHI-1 expression by >80%. Cantharidin (CAN), an inhibitor of PP2A, was further selected from this list of compounds, because it inhibited AHI-1 expression and the combination of CAN plus IM resulted in 93% growth inhibition in AHI-1–transduced cells, as opposed to about 30% with CAN and 15% with IM (0.1 μM) alone (fig. S1D). To confirm this observation, AHI-1 was suppressed by short hairpin RNA (shRNA) (65% suppression, fig. S1E) in K562 cells, and these cells were then treated with less toxic analogs of CAN, such as the water-soluble LB100 (in clinical trial, (3335) and lipid-soluble LB102, which resulted in a significant decrease in viability compared to parental K562 cells (P < 0.05; fig. S1F). Treatment of AHI-1–suppressed cells with IM also decreased viability compared to parental K562 cells, consistent with the role of AHI-1 as an important mediator of BCR-ABL activity (20).

Dual inhibition of PP2A and BCR-ABL results in synergistic cytotoxicity in IM-resistant cells

To determine the specificity and kinetics of the PP2A inhibitors (LB100 and LB102) in inhibition of Ser/Thr phosphatase activity of PP2A, we performed a phosphatase assay using the catalytic subunit of recombinant human PP2A and phosphopeptide (K-R-pT-I-R-R) as a substrate. The inhibitory effects of LB100/LB102 on phosphatase activity of PP2A were demonstrated in a dose-dependent manner after 15 min of treatment with various concentrations of either LB100 or LB102 (Fig. 1A). This result was further supported by additional phosphatase assays using p-nitrophenyl phosphate as a substrate (fig. S2A). We then tested LB100 and LB102 in K562, K562 IM-resistant (K562-IMR, a spontaneously derived cell line without BCR-ABL mutations), and BV173 (BCR-ABL+ blast cells derived from late-stage BCR-ABL+ALL), and found that the IC50s (the half maximal inhibitory concentrations) for these cell lines after 48 hours were in the low micromolar range (K562, 3.5 μM for LB100 and 3 μM for LB102; K562-IMR, 5 μM for LB100 and 4 μM for LB102; BV173, 1 μM for LB100/LB102; fig. S2B). Immunoprecipitation (IP) phosphatase assays demonstrated the specificity of these compounds for PP2A, with Ser/Thr phosphatase activity decreasing to 25% after 5 hours of LB100 or LB102 treatment (Fig. 1B). Combination treatments with IM and LB100 or LB102 further reduced the viability of K562 (>80%; P < 0.005) and IM-resistant cells (>60%; P < 0.005) compared to single treatments (Fig. 1C), which was further supported by a proliferation assay (>80% inhibition; P < 0.005, Fig. 1D). The combination also significantly increased annexin V+ apoptotic cells compared to either treatment alone (P < 0.01; Fig. 1E), which was accompanied by increased cleavage of caspase 8 and caspase 3 (Fig. 1F). To determine whether the combination of PP2A inhibitors and IM had synergistic or additive effects (36), we performed viability assays with graded doses of LB100, LB102, and IM, alone or in combination. The average combination index (CI) for effective dosages (EDs; ED50, ED75, and ED90) was calculated to be 0.59 for the combination of IM and LB100 and 0.64 for IM and LB102 in K562-IMR cells, and 0.21 for the combination of IM and LB100 and 0.43 for IM and LB102 in BV173 cells (fig. S2C). These results indicate that the combinations are highly synergistic.

Fig. 1 LB100 and LB102 specifically inhibit PP2A phosphatase activity and the growth of BCR-ABL+ cells.

(A) The inhibitory effects of LB100 and LB102 on phosphatase activity of protein phosphatase 2A (PP2A) were measured in the presence of various concentrations of either LB100 or LB102 for 15 min in a phosphatase assay. (B) K562 cells were treated with 5 μM LB100/LB102, and PP2A phosphatase activity was measured at the time points indicated using a PP2A immunoprecipitation (IP) phosphatase assay kit (left). K562 and K562 imatinib mesylate–resistant (K562-IMR) cells were treated with 5 μM LB100/LB102, and BV173 cells were treated with 2.5 μM LB100/LB102 and then assayed for phosphatase activity after 12 hours (right). (C) K562 and K562-IMR cells were treated with IM (0.5 μM) with or without LB100/LB102 (2.5 or 5 μM) for 48 hours and assayed for cell viability. The cell lines were treated with 0.5 μM IM (K562) or 5 μM IM (K562-IMR) with or without 5 μM LB100/LB102 and assayed for cell proliferation (D) and apoptosis (E). The fraction of apoptotic cells was determined by annexin V+ staining. (F) Representative Western blot analysis of total and cleaved caspase 8 and caspase 3 in K562 cells after 24 and 48 hours of treatment. Data are mean and SEM from three independent experiments. P values were calculated using paired two-tailed Student’s t test. Ctrl, untreated cells.

BCR-ABL inhibition does not ameliorate PP2A inhibition–dependent mitotic arrest in CML cells

PP2A has a major role in the regulation of cell division, particularly dephosphorylating CDK1-phosphorylated substrates, which prevent cell entry into mitosis (37). A PP2A complex involving the PR55α regulatory subunit prevents transition from G2 to M and is also thought to control mitotic spindle breakdown, chromatin decondensation, and postmitotic reassembly of the nuclear envelope and Golgi apparatus (38). To determine whether increased apoptotic cell death and decreased viability after combination treatment were linked to changes in cell cycle regulation, we examined cell cycle distributions of K562 and K562-IMR cells after combination treatment with IM and LB100 or LB102. LB100/LB102 treatment resulted in a significant, dose-dependent increase in the proportion of G2-M cells compared to untreated cells (K562 = 32 and 28% versus 10% and K562-IMR = 30 and 36% versus 6%; P < 0.05, Fig. 2A and fig. S3A). Although PP2A inhibition significantly increased the proportion of K562 cells with disrupted mitotic spindles (a key feature of mitotic catastrophe) by 40% (P < 0.001; Fig. 2, B and C, and fig. S3B), a similar G2-M shift was observed after combination treatment with PP2A inhibitors and IM (Fig. 2A), demonstrating that BCR-ABL inhibition does not ameliorate cell cycle dysfunction and that the increase in the proportion of G2-M cells was due to PP2A inhibition alone.

Fig. 2 Inhibition of PP2A disrupts cell cycle control and induces mitotic arrest in CML cells.

(A) Representative fluorescence-activated cell sorting (FACS) plots showing cell cycle distribution in K562 and K562-IMR cells after treatment with 5 μM LB100 alone or in combination with IM (0.5 and 5 μM, respectively). Arrows indicate enhancement in the G2-M population after LB100 treatment. (B) Representative confocal microscopy images of K562 cells stained with an anti–α-tubulin antibody (green) and DNA binding dye 4′,6-diamidino-2-phenylindole (DAPI) (blue) in the presence or absence of 5 μM LB100. Scale bars, 10 μm. (C) Percentage of noncatastrophic cells after treatment with 5 μM LB100 or 5 μM LB102. Values were determined by counting the number of catastrophic cells, which displayed disrupted spindles with multipolar cell division, in PP2A inhibitor–treated cells compared to control cells. At least 100 cells per treatment group were counted in the field of view, and the percentages of noncatastrophic cells were calculated and expressed as percentages of untreated control cells. Data are mean and SEM from three independent experiments. P values were calculated using paired two-tailed Student’s t test.

Combination treatment reduces protein phosphorylation and expression of several key signaling proteins and increases β-catenin protein degradation

We then evaluated the effects of PP2A inhibitors, alone or in combination with IM, on key proteins in BCR-ABL signaling. IM treatment reduced phosphorylation of BCR-ABL, JAK2, and STAT5 in K562 and K562-IMR cells, without changes in total protein (Fig. 3A and fig. S4A), consistent with previous observations (21, 22). IM also slightly reduced expression of AHI-1 and β-catenin in K562-IMR, but not K562 cells. LB100 and LB102 did not affect the expression of major signaling proteins. However, the combination of IM and LB100 or LB102 in K562-IMR and K562 cells resulted in a marked reduction in the protein expression of BCR-ABL, JAK2, and STAT5, with a corresponding loss of P-BCR-ABL, P-JAK2, and P-STAT5. IM treatment alone did not reduce phosphorylation of BCR-ABL at Y177, the critical residue where BCR-ABL interacts with GRB2 and activates the phosphoinositide 3-kinase(PI3K)/protein kinase B (AKT) and RAS/mitogen-activated protein kinase (MAPK) pathways, but the combination of IM and LB100/LB102 did (Fig. 3A), which reduced extracellular signal-regulated kinase (P-ERK) (Y204), a downstream substrate of the RAS/MAPK pathway (39). Examination of signaling proteins in the PI3K/AKT and RAS/MAPK pathways also demonstrated reduced P-AKT (S473) and P-P38 (T180/182), with corresponding decreases in AKT and P38 protein expression after combination treatment. In addition, protein expression of SET (also known as I2PP2A) was not changed by these treatments (Fig. 3A). Notably, AHI-1 and BCR-ABL protein expression was inhibited to a greater extent by the combination than by IM or PP2A inhibitors alone (Fig. 3A). We then examined whether the reduction in AHI-1 and BCR-ABL protein expression due to the combination treatment was mediated by induction of protein degradation or an inhibitory effect on protein synthesis; we used MG132, to block proteasome-mediated protein degradation, or cycloheximide (CHX), to block protein synthesis. We demonstrated increased AHI-1 protein expression in BCR-ABL and AHI-1–cotransduced cells treated with IM and MG132, but this effect was not observed in the same cells treated with LB100 and MG132 (fig. S4B, left). Moreover, IM or LB100 alone increased BCR-ABL expression in the presence of MG132, and the combination of IM and LB100 did not result in a further increase, possibly due to saturation of highly increased BCR-ABL expression induced by such an inhibitor alone. These results suggest that protein degradation of AHI-1 is mostly caused by IM, whereas both IM and LB100 contribute to the protein degradation of BCR-ABL (fig. S4B, left). In contrast, we did not observe obvious changes when CHX was used under the same treatment conditions with IM and/or LB100 (fig. S4B, right).

Fig. 3 Combination treatment with PP2A inhibitors and IM disrupts phosphorylation and expression of key signaling proteins.

K562-IMR cells were treated with 5 μM IM and 5 μM LB100/LB102, either alone or in combination for 24 or 48 hours, and cell lysates were harvested for Western blotting of (A) key signaling proteins after 48 hours and (B) phosphorylation of specific β-catenin residues after 24 and 48 hours. Representative Western blots from three repeats are shown. (C) RNA from the treated cells was used for quantitative reverse transcription polymerase chain reaction analysis of β-catenin downstream target genes cyclin D1 (CCND1), MYC, TCF1, and LEF1 in K562-IMR cells after 48 hours of treatment. (D) Schematic diagram of how BCR-ABL–mediated tyrosine phosphorylation (Y86) and PP2A-mediated threonine and serine dephosphorylation (T41, S45) change in response to tyrosine kinase inhibitors (TKIs) and PP2A inhibitors in chronic myeloid leukemia (CML) cells. Data are mean and SEM from three independent experiments. P values were calculated using paired two-tailed Student’s t test. ERK, extracellular signal–regulated kinase; JNK, c-Jun N-terminal kinase; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Inhibition of BCR-ABL in combination with PP2A inhibitors resulted in the loss of β-catenin protein expression in K562 and K562-IMR cells (Fig. 3A and fig. S4A). We further examined β-catenin stability and phosphorylation after treatment with IM and LB100 or LB102. PP2A inhibitors increased β-catenin phosphorylation at T41 and S45 (Fig. 3B), consistent with the role of PP2A in dephosphorylating β-catenin at those sites (40, 41). Also consistent with previous observations, IM reduced Y86 phosphorylation of β-catenin, a signaling site thought to be critical in promoting β-catenin stabilization (42). IM treatment led to a decrease in total β-catenin after 48 hours compared to untreated cells, which was compounded to near-total loss with IM plus LB100/LB102. The inhibition of β-catenin protein in K562-IMR cells after 48 hours of combination treatment was also confirmed by a significant reduction in transcripts of several β-catenin downstream target genes, including CCND1, MYC, LEF1, and TCF1, compared to IM treatment alone (P < 0.05; Fig. 3C). These results demonstrate that concurrent BCR-ABL and PP2A inhibition greatly reduced the expression of key signaling proteins, as well as increased the phosphorylation of T41 and S45 and decreased the phosphorylation of Y86 residues of β-catenin (Fig. 3, A to C), which may contribute to blocked β-catenin signal transduction and induction of protein degradation (Fig. 3D).

Knockdown of the PP2A catalytic C subunit increases sensitivity of CML cells to IM and replicates the combination effects of IM and PP2A inhibitors

To demonstrate that the combination effects observed were due to specific perturbations in PP2A, we knocked down the PP2A catalytic C subunit (PP2A-C) (43), reducing its expression by ~66% in K562-IMR cells (shPP2A-C, Fig. 4A). The expression of BCR-ABL, β-catenin, and AHI-1 in shPP2A-C cells was greatly reduced by IM treatment in the absence of any PP2A inhibitors compared to control K562-IMR cells (Scr, Fig. 4A, lane 2 compared to lane 6). Moreover, phosphorylated BCR-ABL, JAK2, AKT, and STAT5 were all reduced in the shPP2A-C cells, whereas total JAK2, AKT, and STAT5 remained unchanged. Thus, these molecular changes in IM-treated shPP2A-C cells resembled those in K562-IMR cells after the combination treatment with PP2A inhibitors and IM (Figs. 3A and 4A). The viability of shPP2A-C cells significantly decreased in response to IM treatment, compared to control cells, after 48 hours (56% decrease, P < 0.05; Fig. 4B), similar to LB100 and IM treatment (Fig. 1C). Similarly, we observed a significant increase in apoptosis in IM-treated shPP2A-C cells compared to control cells (P < 0.001; Fig. 4B). These results demonstrate that the cellular and molecular effects of the combination treatment depend on dual inhibition of BCR-ABL and PP2A-C.

Fig. 4 Knockdown of PP2A-C sensitizes K562-IMR cells to IM treatment.

K562-IMR cells were transduced with short hairpin RNA (shRNA) targeting the PP2A catalytic C subunit (shPP2A-C) or scrambled shRNA (Scr) and treated with 5 μM IM and 5 μM LB100, either alone or in combination, for 48 hours. After 48 hours, some of the cells were used to make cell lysates for Western blotting [representative blot is shown in (A), whereas the rest of the cells were used for cell viability and apoptosis assays (B)]. Data are mean and SEM from three independent experiments. P values were calculated using paired two-tailed Student’s t test. (C) Distinct Abelson helper integration site–1 (AHI-1) prey proteins, including PP2A subunit B (PR55α, P63151), identified by IP coupled to mass spectrometry (IP-MS) in 293T cells transfected with two different human AHI-1 constructs: N-terminal FLAG-tagged AHI-1 (N-AHI-1) and C-terminal FLAG-tagged AHI-1 (C-AHI-1). The percent sequence coverage, statistical confidence [log(e)], and the number of unique peptides observed are presented; the results are from three replicate IPs.

PP2A, through its B subunit PR55α, and β-catenin both physically interact with AHI-1

Because BCR-ABL interacts with AHI-1 (21) and PP2A inhibitors were identified in a screen for AHI-1 inhibitors, we postulated that AHI-1 may mediate the effects of combination treatment through direct interaction with PP2A. N-terminal– or C-terminal–tagged human AHI-1 constructs were transduced into 293T cells to identify AHI-1 interacting proteins by IP coupled to mass spectrometry (IP-MS, fig. S4C) (44). The PP2A B subunit, PR55α (accession number, P63151), was the most abundant protein that interacted with AHI-1 (Fig. 4C). IP–Western blot analysis further confirmed the interaction in AHI-1–transduced 293T cells and BaF3 cells, but not in control cells (Fig. 5A). These results demonstrate that AHI-1 interacts with PP2A through PR55α, potentially mediating the effects of BCR-ABL and PP2A inhibition.

Fig. 5 PR55α, a B subunit of PP2A, and β-catenin interact with AHI-1.

(A) PR55α was immunoprecipitated from cell lysates of 293T cells transfected with a human influenza hemagglutinin (HA)–tagged full-length AHI-1 construct (left) and BaF3 cells stably transduced with full-length AHI-1 (right). The immunoprecipitates were probed with anti–AHI-1 antibodies and an anti-PR55α antibody. Mouse serum was used as an immunoglobulin G (IgG) control to replace a specific IP antibody as a negative control. Western blots (WB) shown are representative of three independent experiments. (B) HA was immunoprecipitated from 293T cells overexpressing HA-tagged full-length AHI-1. The immunoprecipitates were probed with either anti–β-catenin or anti–AHI-1 antibody. (C) AHI-1 was immunoprecipitated from cell lysates of BCR-ABL–transduced BaF3 cells or BCR-ABL–transduced BaF3 cells overexpressing full-length AHI-1. The immunoprecipitates were then probed with either anti–β-catenin or anti–AHI-1 antibody. (D) Schematic diagram of how targeting the AHI-1–mediated complex by dual inhibition of PP2A and BCR-ABL disrupts several key signaling molecules and activities of PP2A/β-catenin and downstream target genes, leading to reduced survival and increased TKI response of drug nonresponder cells.

One downstream effect of the combination treatment was increased degradation of β-catenin. Previous observations suggested a role for AH1-1 in the nuclear transport of β-catenin in kidney cells (27). Hence, we examined whether AHI-1 also directly interacts with β-catenin in hematopoietic cells. Human influenza hemagglutinin (HA)–tagged full-length AHI-1 constructs were transduced into human 293T and murine BaF3 cells. Co-IP with HA produced a strong band for β-catenin, demonstrating that AHI-1 can interact directly with β-catenin (Fig. 5B). This interaction was further confirmed in BaF3 cells cotransduced with BCR-ABL and AHI-1 (Fig. 5C). The AHI-1–β-catenin interaction was also detectable at endogenous levels in BCR-ABL–transduced BaF3 cells (Fig. 5C). These results suggest that AHI-1 facilitates the PP2A-dependent dephosphorylation of β-catenin through the PR55α subunit and that BCR-ABL and PP2A inhibition results in numerous cellular effects by disrupting BCR-ABL/JAK2/STAT5 and β-catenin signaling (Fig. 5D).

Dual BCR-ABL and PP2A inhibition more effectively targets CML stem/progenitor cells than either treatment alone

To investigate whether dual inhibition of BCR-ABL and PP2A may be therapeutically more effective for CML patients who do not respond adequately to single TKI treatment, we performed viability and apoptosis assays on CD34+ stem/progenitor cells obtained at diagnosis from patients classified retrospectively as IM nonresponders (n = 5) (45, 46). Because DA is a second-generation TKI, more potent than IM, it was included in the remaining studies. As we have previously reported (47), only 50% of pretreated CD34+ CML stem/progenitor cells from IM nonresponders responded to IM or DA treatment. Combination treatments of IM or DA with LB100/LB102 significantly decreased the viability of these cells (>90% inhibition; P < 0.02), whereas the viability of CD34+ healthy bone marrow (BM) cells remained relatively unchanged or only slightly reduced (Fig. 6A). The increased efficacy of the combination treatment on IM nonresponder cells was accompanied by a significant increase in apoptotic cells compared to cells treated with DA alone (~2-fold, P < 0.04; Fig. 6B).

Fig. 6 Combination treatment with PP2A inhibitors and TKIs effectively targets CD34+ CML cells.

CD34+ normal bone marrow (NBM, n = 3) and CML cells from IM nonresponders (n = 5) were cultured for 72 hours with single and combination treatments of 5 μM IM, 150 nM dasatinib (DA), and 2.5 μM LB100/LB102. After 72 hours, (A) viability and (B) apoptotic cells were measured. (C) CD34+ NBM and CML cells from IM nonresponders were plated in colony-forming cell (CFC) assays with single and combination treatments of TKIs and PP2A inhibitors. CFCs were expressed as a percentage of colonies from untreated control cells. (D) Long-term culture-initiating cell (LTC-IC) assays were performed from CD34+ CML cells obtained from the same CML patient samples that were used to perform CFC assays with or without indicated inhibitors, alone or in combination. DMSO, dimethyl sulfoxide.

To further determine whether the combination of TKI and LB100/LB102 can eliminate LSCs and their progenitor cells from IM nonresponders, we performed colony-forming cell (CFC) assays (a progenitor assay) and long-term culture-initiating cell (LTC-IC) assays (a stem cell assay). The combination decreased CML CD34+ progenitors, as demonstrated by significant inhibition of colony growth after 2 weeks of treatment relative to TKI or LB100 or LB102 alone (~70 inhibition; P < 0.05, Fig. 6C). Both LB100 and LB102 had no toxicity on CD34+ healthy BM cells (n = 3) in CFC assays, which is a vast improvement over the first-generation PP2A inhibitor CAN (P < 0.01, Fig. 6C and fig. S5). Moreover, LTC-IC assays showed that primitive leukemic cells were more significantly eliminated by combination treatment, particularly by NL with LB100 or LB102 (P < 0.05, Fig. 6D). These results indicate that dual inhibition of BCR-ABL and PP2A is more effective at eliminating IM nonresponder stem cells and their progenitors than TKI or PP2A inhibitors alone in vitro.

The combination of DA and LB100 decreases engraftment of leukemic blast cells and prolongs the survival of leukemic mice

The combination of PP2A inhibitors with TKIs was further investigated in an aggressive BCR-ABL+ blast cell (BV173) model from late-stage disease where TKI monotherapy is ineffective (13, 14). BV173-YFP+ cells carrying a luciferase reporter (2.5 × 106 per mouse) were intravenously injected into sublethally irradiated nonobese diabetic (NOD)/severe combined immunodeficient (SCID) interleukin-2 receptor γ-chain–deficient (NSG) mice (21). Two weeks after transplantation, mice were treated with vehicle control (propylene glycol), DA (15 mg/kg), LB100 (1.5 mg/kg), or DA plus LB100 for 2 weeks by oral gavage and/or intraperitoneal injection. The biological effects were closely monitored after discontinuation of treatment. Bioluminescence imaging demonstrated that mice treated with either DA alone or DA plus LB100 had dramatically lower bioluminescence signals, or no detectable signals, compared to those treated with vehicle or LB100 alone, 5.5 weeks after transplantation (Fig. 7A). At 7 weeks after transplantation (3 weeks after discontinuation of treatment), when vehicle- and LB100-treated mice were moribund, these mice presented with enlarged hematopoietic organs, including spleen and liver, which did not occur with DA and LB100 combination treatment (Fig. 7B). DA-treated mice displayed slight splenomegaly (0.05 g compared to 0.01 g) and hepatomegaly (2.12 g compared to 1.42 g) compared to combination-treated mice and control mice without injection of leukemic cells (Fig. 7B). Histological analysis [hematoxylin and eosin (H&E) staining] of these organs showed that vehicle- and LB100-treated groups had the greatest infiltration of leukemic cells, followed by DA (low infiltration, indicated by arrows) and combination treatment (no detectable infiltration, Fig. 7C). Decreased engraftment of human leukemic cells after combination treatment was confirmed by a significant reduction of BCR-ABL transcripts (P < 0.01, Fig. 7D) and much fewer YFP+ engrafted cells (fig. S6A) in the BM, spleen, and liver of these mice compared to mice treated with vehicle, LB100, or DA alone. There was also a high amount of phosphorylated STAT5 and CRK-like protein (CRKL), and increased total BCR-ABL protein in BM from vehicle- and LB100-treated mice, but these were very low/undetectable in the DA- and combination-treated mice (fig. S6B).

Fig. 7 Treatment with LB100 and DA reduces engraftment of leukemic blast cells and prolongs survival of leukemic mice.

(A) BV173-YFP (yellow fluorescent protein) cells were intravenously injected into sublethally irradiated interleukin-2 receptor γ-chain–deficient (NSG) mice. Treatment with DA or LB100, alone or in combination, was initiated 2 weeks after transplantation for 2 weeks. A noninvasive bioluminescent imaging assay was performed 1 week after the end of treatment, and representative images from each treatment group (out of 6 to 8 mice) are presented. (B) One mouse from each group was sacrificed, and hematopoietic tissues were collected for analysis 7 weeks after transplantation. Images and weights of spleen and liver from each mouse are shown. (C) Hematoxylin and eosin staining of spleen and liver of mice with or without treatments. Arrows indicate areas where infiltrated leukemic cells are found. Scale bar, 0.1 mm. (D) Fold difference in BCR-ABL transcripts in BM, spleen, and liver of treated groups compared to vehicle-treated control mice. (E) Representative bioluminescent images from a negative control mouse (ctrl), a DA-treated mouse, and a mouse that received the combination (out of 4 to 8 mice per group) 10 weeks after transplantation. Mice from vehicle and LB100 treatment groups had already died. (F) FACS plots of engraftment of human leukemic cells (YFP+CD19+) in BM and spleen from mice in each treatment group and percentages of these double-positive cells. (G) Images and weights of spleen and liver from each mouse. (H) Fold difference in BCR-ABL transcripts from BM and spleen of treated mice compared to vehicle-treated controls. (I) Western blots of whole protein extracts from the BM of mice treated with vehicle, DA, or DA plus LB100. (J) Overall survival of leukemic mice treated with vehicle, LB100 (1.5 mg/kg), and/or DA (15 mg/kg), alone or in combination once a day for 2 weeks. Median survival with the combination of LB100 and DA versus LB100 alone was 83 days versus 52 days, ratio = 0.65, 95% confidence interval (CI) = 0.226 to 1.879, P = 0.0011; versus DA alone, 71 days, ratio = 0.924, 95% CI = 0.298 to 2.865, P = 0.0184. Ctrl, mice without injection of BV173-YFP cells, as a negative control. Vehicle, mice injected with BV173-YFP cells treated with propylene glycol, as a positive control. ND, not detectable.

The difference in infiltration of leukemic cells in hematopoietic organs between mice treated with DA or DA plus LB100 was more pronounced at 10 weeks after transplantation (6 weeks after discontinuation of treatment). Lower bioluminescence intensity was detected in the combination group compared to DA treatment alone, which corresponded with dramatically reduced human leukemic cell engraftment in these mice (Fig. 7, E and F). Enlarged spleens and livers were observed in mice treated with DA alone, and histological analysis revealed that these organs had massive infiltration of human leukemic cells compared to mice treated with DA plus LB100 (Fig. 7G and fig. S7). BCR-ABL transcripts were very low in the BM and spleen of DA- and LB100-treated mice, unlike the mice treated with DA alone (P < 0.01, Fig. 7H). In addition, we observed reduced phosphorylation of CRKL and STAT5 and total BCR-ABL protein, confirming reduced engraftment of human leukemic cells in mice treated with DA plus LB100 compared to DA treatment alone (Fig. 7I). This translated to significantly longer survival in these mice compared to those treated with DA or LB100 alone, even 8 weeks after discontinuation of treatment (combination versus LB100, 83 days versus 52 days, P = 0.0011; combination versus DA, 83 days versus 71 days, P = 0.0184, Fig. 7J). Overall, these results demonstrate that LB100 in combination with DA is more effective than DA or LB100 treatment alone in eliminating leukemic blast cells that can generate aggressive leukemia in mice, conferring a survival advantage in a preclinical xenotransplantation model.

PP2A inhibition sensitizes BCR-ABL+ LSCs to TKIs and prolongs the survival of leukemic mice in a transgenic mouse model of CML

To determine whether the drug combination more effectively targets BCR-ABL+ stem cells, we further examined the effects in an inducible BCR-ABL transgenic mouse model of CML (48, 49). In these mice, withdrawal of tetracycline results in reversible induction of BCR-ABL expression and induction of a CML-like myeloproliferative disease characterized by neutrophilic leukocytosis and splenomegaly, which provides a robust model that has been successfully used for preclinical studies of therapies against CML stem cells in vivo (50, 51). After induction of leukemia by tetracycline withdrawal, BM cells from SCLtTA/TRE–BCR-ABL mice were transplanted into lethally irradiated Pep3B mice and were subsequently dosed with DA, LB100, or a combination of both for 2 weeks (Fig. 8A). Two and a half weeks after discontinuation of treatment, combination-treated mice displayed a marked decrease in splenic (0.06 g) and hepatic (0.85 g) size compared to vehicle-treated (spleen = 0.24 g, liver = 1.35 g) and DA-treated mice (spleen = 0.14 g, liver = 1.35 g; Fig. 8B). A decrease in donor-derived BCR-ABL+ cells (CD45.2+) and a ~2-fold decrease in long-term hematopoietic stem cells (HSCs) (linSca1+cKit+Flt3CD150+CD48) were observed in the BM of LB100-treated mice (38 of 1.01 × 106 linCD45.2+ cells) and combination-treated mice (46 of 1.08 × 106 linCD45.2+ cells), compared to vehicle (93 of 1.18 × 106 linCD45.2+ cells) and DA treatment alone (85 of 1.08 × 106 linCD45.2+ cells, Fig. 8C). There was also a ~2-fold decrease in granulocyte macrophage progenitor cells (GMPs; lincKit+FcγR+CD34+) in the combination- versus vehicle-treated mice (1935 of 645,000 linCD45.2+ cells) or 0.3% compared to 0.6% for vehicle (2161 of 366,200, fig. S8). DA (0.36%) and LB treatment alone (0.38%) also showed a ~2-fold decrease in GMPs compared to vehicle. Furthermore, combination treatment dramatically reduced BCR-ABL+ donor-derived cells in peripheral blood (PB) compared to vehicle-, DA-, and LB100-treated mice (6.8% compared to 59, 62, and 47% respectively; Fig. 8D). The median survival of combination-treated mice compared to those treated with vehicle or single agents increased (Fig. 8E), consistent with in vivo data obtained from an aggressive human leukemia model (Fig. 7). These results demonstrate that the combination of LB100 and DA markedly reduces the number of LSCs capable of multilineage engraftment and prolongs survival compared to individual treatments alone.

Fig. 8 LB100 enhances the efficacy of DA against engrafted BCR-ABL+ stem cells in a BCR-ABL transgenic mouse model.

(A) Experimental design using the SCLtTA/TRE-BCR-ABL transgenic mouse model. (B) A group of mice was sacrificed and analyzed 6.5 weeks after transplantation. Images and weights of spleen and liver from each mouse are shown. (C) FACS plots from harvested BM showing gating strategies for detection of stem cells [long-term (LT)–HSCs; CD45.2+linSca1+cKit+Flt3CD48CD150+]. The absolute cell numbers calculated for LT-HSCs (red) and short-term (ST)–HSCs (black) per sample are indicated. (D) FACS plots from peripheral blood collected at the end point, when mice displayed >20% weight loss, increased white blood cell counts (3- to 5-fold), breathing difficulties, reduced alertness/responsiveness, lethargy, etc., showing BCR-ABL+ granulocytes (CD45.2+Gr1+Mac1+) with percentage of total CD45.2+ cells (black) and percentage of Gr1+Mac1+ cells (red) outlined. (E) Overall survival of mice treated with vehicle, LB100 (1.5 mg/kg), and/or DA (15 mg/kg), alone or in combination. Median survival with the combination versus LB100 alone was >82 days versus 60 days, and that with the combination versus DA alone was >82 days versus 75 days, as indicated in the figure.


Overcoming drug resistance and eradicating cancer stem cells to overcome MRD remain major challenges in the treatment of BCR-ABL+ human leukemia and other cancers. Here, we provide preclinical evidence that a combination of TKI and PP2A inhibitors is an improved strategy to target drug-insensitive stem/progenitor cells. Although PP2A functions as a tumor suppressor in several types of cancer (52, 53), we demonstrated a prosurvival role for PP2A in TKI-insensitive leukemic stem/progenitor cells. PP2A is an important regulator of G2-M checkpoint entry, dephosphorylating CDK1 substrates that facilitate mitotic exit (37, 54); thus, PP2A inhibitors force premature cell cycle entry. This works synergistically with DNA-damaging agents, such as DNA-alkylating agents or ionizing radiation, to increase tumor killing in several cancer models (33, 34, 55). When we tested the effects of the PP2A inhibitors LB100/LB102 on CML cells, we observed an expected increase in G2-M fraction and mitotic catastrophe, but this effect was not enhanced by TKIs. In contrast, we found that the efficacy of PP2A inhibition was enhanced by the addition of TKIs with respect to their ability to disrupt key signaling pathways in BCR-ABL+ cells. We demonstrated that PP2A inhibition impairs survival and sensitizes these cells, including drug-insensitive LSCs, to TKI treatments in vitro, while reducing engraftment of blast cells and LSCs in mice. This resulted in a survival advantage after discontinuation of treatment compared to TKI treatment alone. This combination effect appears to be due to the disruption of key protein interactions and their signaling networks, rather than cell cycle dysfunction. Moreover, by demonstrating that the knockdown of PP2A-C by shRNA in drug-resistant cells combined with TKI treatment mirrors the dramatic loss of signaling proteins and cell viability seen after combination treatment in IM-resistant cells, we confirmed that this combination treatment requires concurrent inhibition of both PP2A-C and BCR-ABL.

Mechanistically, we demonstrated that the inhibition of BCR-ABL and PP2A disrupts several molecules involved in AHI-1–mediated signaling, particularly β-catenin, which is critical in the maintenance of LSCs in CML (5658). β-Catenin is tightly regulated by the β-catenin degradation complex, and phosphorylation of β-catenin, including N-terminal residues T41 and S45, primes β-catenin for proteasome-mediated degradation (59). These residues can be directly dephosphorylated by PP2A via its PR55α subunit (40), and as expected, we observed a dramatic increase in phosphorylation of these residues 24 hours after PP2A inhibition. This was followed by degradation of β-catenin 24 hours later. The regulation of β-catenin signaling occurs on two separate levels: through cytoplasmic stabilization and/or nuclear translocation processes (60, 61). In neuronal cells, knockdown of AHI-1 by small interfering RNA abolishes nuclear transport of β-catenin, resulting in decreased β-catenin transcriptional activity (27). In addition, AHI-1 binds to β-catenin in the cytoplasm and facilitates its cytoplasm-nucleus translocation. We now present evidence from IP-MS and IP–Western blot analyses that both β-catenin and PP2A PR55α can directly interact with AHI-1 in CML cells and in AHI-1– and BCR-ABL–transduced cells. This provides mechanistic insights into AHI-1–mediated dephosphorylation of β-catenin by PP2A. AHI-1 has also been previously found to interact with BCR-ABL and JAK2 (21). Moreover, BCR-ABL promotes β-catenin stability and increases nuclear translocation in CML cells by phosphorylating Y86 of β-catenin (42). Hence, AHI-1 appears to play a central role as a mediator of β-catenin stability (via regulation of T41 and S45 dephosphorylation by PP2A), promoting BCR-ABL–dependent phosphorylation of β-catenin at Y86 and facilitating its translocation to the nucleus. However, additional studies are needed to confirm these critical findings, such as by targeted mutagenesis, which introduces phosphomimic mutations to these sites or replaces them with a nonphosphorylatable amino acid. Nevertheless, our studies demonstrate that disruption of AHI-1 expression and its complex after combination treatment could potentiate the elimination of leukemia stem/progenitor cells, in part via the loss of β-catenin signaling, resulting in decreased engraftment of leukemic cells as seen in our two in vivo models of leukemia. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis of CML cells after combination treatment revealed suppression of β-catenin downstream target genes, confirming loss of β-catenin in these cells.

The disruption of a multitude of key signaling molecules, such as BCR-ABL, STAT5, c-Jun N-terminal kinase (JNK), and β-catenin, after combination treatment points to a major collapse of the signaling network in these cells. Because the expression of certain signaling proteins, such as ERK (and to a certain extent AKT and glycogen synthase kinase-3β), was not severely affected by the combination treatment, we inferred that the loss of these proteins was due to a specific (rather than a global) change in protein expression/degradation. As a scaffold adaptor protein, AHI-1 can have an important role in amplification of oncogenic signaling pathways by acting as a molecular bridge to facilitate protein complex formation. We have previously demonstrated an increase in BCR-ABL, JAK2, and STAT5 expression when AHI-1 is overexpressed, as well as a decrease when AHI-1 is suppressed (20). Moreover, the role of AHI-1 phosphorylation by these signaling proteins is still largely unknown, because there are at least 87 predicted phosphorylation sites on AHI-1 (NetPhos 2.0) (62), indicating its potential to interact with multiple proteins. In addition, a putative c-FOS binding site has been identified on AHI-1 (63), which suggests that the loss of AHI-1 expression could be positively reinforced by the loss of P-ERK and JNK expression as a result of combination treatment. Thus, it is possible that the disruption of key signaling proteins with the combination treatment destabilizes AHI-1–mediated protein-protein complexes and potentiates further shutdown of CML/ALL signaling networks. It has been reported that IM plus arsenic sulfide (AS) exerts more profound therapeutic effects than IM or AS treatment alone in a mouse model of CML, and AS seems to target BCR-ABL through the ubiquitination of key lysine residues, causing its proteasomal degradation (64). Similarly, we have demonstrated that the inhibitory effects of PP2A inhibitors in combination with TKIs contribute to the induction of protein degradation of AHI-1 and BCR-ABL, which, at least in part, contribute to the destabilization of AHI-1–mediated protein-protein complexes. It would also be interesting to pinpoint other key mechanisms of action that contribute to this synergistic effect, possibly through the regulation of the switch from inactive TK to active TK that can subsequently serve as a target for inhibition by a TKI.

We did not observe a change in SET protein expression even after combination treatment or suppression of PP2A-C. This contrasts with a previous study demonstrating that expression of SET, an endogenous inhibitor of PP2A, was increased in BC CML cells (31) and that PP2A-activating drugs were able to inhibit these cells (32). Here, we did not observe in vivo anti-leukemic activity with PP2A inhibitors alone, but PP2A inhibition greatly improved TKI efficacy against drug-insensitive leukemic cells regulated by the AHI-1–mediated protein complex, including the PP2A B subunit PR55α and β-catenin. In addition, it has recently been suggested that PP2A interrupts glucose metabolism in BCR-ABL+ ALL and that inhibition of PP2A induces cell death (65). Given the extremely complex role that PP2A plays in the regulation of multiple signaling cascades involved in cell cycle control, adhesion, migration, and metabolism in various cancers (29), it is likely that the activity of PP2A and its specific interacting proteins and pathways are differentially regulated during CML/ALL development or in specific subpopulations. We noticed that dual inhibition of PP2A and BCR-ABL seems to be more effective against BCR-ABL+ ALL blast cells in a preclinical xenotransplant model as compared to a BCR-ABL–transgene/transplantation model that aims to target BCR-ABL+ stem cells and myeloid progenitor cells. Relatively high expression of BCR-ABL in the BCR-ABL–transgene/transplantation model and differing cell types targeted in these two models may contribute to differential drug responses observed (21, 22, 4851). Notably, we observed reduced HSC numbers in BCR-ABL–transgenic mice treated with LB100, but these LB100-treated mice did not demonstrate survival benefits compared to mice receiving the combination treatment. When we analyzed these LB100-treated mice, they were already in late-stage disease compared to the other mice. Hence, the inhibitory effect on HSCs can be interpreted, in part, as BM failure and exhaustion of the HSCs commonly observed in late-stage diseased mice. Additional studies would be needed to determine whether this observation can be confirmed in additional LB100-treated mice and to analyze such mice at an earlier point. Perhaps a prolonged treatment period would achieve a better outcome in these aggressive leukemia models, as previously reported (66, 67). Nevertheless, our study extensively demonstrates that PP2A inhibition with TKIs can synergistically target drug-insensitive CML stem/progenitor cells and BCR-ABL+ ALL cells in preclinical in vitro and in vivo models. A phase 1 clinical trial of the PP2A inhibitor LB100 for the treatment of relapsed solid tumors has been successfully completed without safety issues ( (35), which opens a promising avenue for combination cancer therapies. Here, we have demonstrated that both LB100 and LB102 are not toxic to healthy normal stem/progenitor cells, which is a vast improvement over the first-generation PP2A inhibitor CAN. Thus, our study provides proof of principle for the development of therapies to overcome drug resistance in BCR-ABL+ leukemia and possibly other cancers with activation of PP2A.


Study design

This study was designed to identify specific growth inhibitory compounds that disrupt an AHI-1–mediated protein complex and develop a more effective, mechanism-based treatment strategy that targets drug-insensitive LSCs and progenitors both in vitro and in vivo. Using an advanced drug/proliferation screen of the Prestwick Chemical Library, we identified a specific PP2A inhibitor in AHI-1–transduced cells. The specificity of PP2A inhibitors (LB100 and LB102) on inhibition of PP2A phosphatase activity was investigated by phosphatase assays, and their inhibitory effects, including potential synergistic or additive effects with TKIs, were tested in BCR-ABL+ cells and IM-resistant cell lines. To investigate molecular mechanisms of drug actions that disrupt the AHI-1–mediated protein complex, we performed various molecular assays, including Western blot analysis, co-IP, and MS in IM-resistant cells and shRNA-mediated PP2A-C knockdown cells. The efficacy of PP2A inhibitors, alone or in combination with TKIs, was then investigated in CML stem/progenitor cells from TKI nonresponders in vitro. Two in vivo models were specifically designed to investigate the efficacy of dual inhibition of PP2A and BCR-ABL: the first against BCR-ABL+ ALL blast cells in a preclinical xenotransplant model and the second against BCR-ABL+ stem cells and myeloid progenitor cells in a BCR-ABL–transgene/transplantation model. In each experiment, mice were randomly assigned to treatment groups, and n > 6 mice per group were used to achieve statistical significance. DA (15 mg/kg), LB100 (1.5 mg/kg), or DA plus LB100 was administered daily by oral gavage and/or intraperitoneal injection for 2 weeks after the induction of leukemia in mice. Survival of leukemic mice and engraftment of leukemic blast cells, as well as stem cell populations and their progenitor cells, were monitored at a number of time points and extensively analyzed by noninvasive bioluminescence imaging, flow cytometry, and histological analyses. All sample measurements were blinded, and no animals were excluded from analysis. Each in vitro and in vivo experiment was performed in triplicate and repeated three times.

Human cells

Primary CML cells were obtained from CP CML patients at diagnosis, before the initiation of TKI therapy, and were clinically classified as IM responders and IM nonresponders based on the European LeukemiaNet treatment guidelines (46). Fresh normal BM (NBM) cells were also obtained from healthy adult donors (ALLCELLS). Informed consent was obtained, and the procedures used were approved by the Research Ethics Board of the University of British Columbia. CD34+ cells were enriched using EasySep CD34 selection kits (STEMCELL Technologies), and purity was verified by fluorescence-activated cell sorting (FACS) (21). BCR-ABL+ human cell lines, including K562 [American Type Culture Collection (ATCC) CCL-243], BV173 cells (CLS 330-133), K562-IMR (from A. Turhan’s laboratory, Paris-Sud University Hospital), parental BaF3 cells (ATCC HB-283), human 293T cells (ATCC CRL-3216), and primary CD34+ cells, were cultured as previously described (21) and as also detailed in Supplementary Materials and Methods. Cell viability was assessed using trypan blue dye exclusion.

Prestwick Chemical Library/proliferation screen

A Prestwick Chemical Library (Prestwick Chemical) was screened against AHI-1–transduced K562 cells containing a YFP marker. K562 control cells and AHI-1–transduced cells were seeded into 96-well cell culture plates (Falcon, BD Biosciences) at 3000 cells per well in 100 μl of medium with either dimethyl sulfoxide or 0.1 μM IM for 24 hours. Prestwick chemical compounds were then added to the cell plates at a final concentration of 10 μM. Plates were incubated for 48 hours and fixed with 3.7% paraformaldehyde containing Hoechst 33342 (Sigma-Aldrich). Plates were scanned and analyzed using a high-content screening system (ArrayScan, Cellomics).


IM, DA, and NL were obtained from Selleckchem. PP2A inhibitors LB100 and LB102 were provided by Lixte Biotechnology. CAN was purchased from Sigma-Aldrich.

Proliferation, apoptosis, and cell cycle assays

Cell proliferation was measured using [3H]thymidine incorporation. Briefly, 1 μCi of [3H]thymidine was added after 48 hours of drug exposure, and thymidine incorporation was measured after 4 hours with an LKB Betaplate scintillation counter (LKB Wallace-PerkinElmer). For apoptosis assays, cells were stained with annexin V and propidium iodide (PI) and analyzed on a FACSCalibur flow cytometer (BD Biosciences). Total apoptotic cells were determined as the sum of “early” (annexin V+) and “late” apoptotic cells (annexin V+/PI+). For cell cycle analysis, cells were fixed, permeabilized with ice-cold 100% ethanol, stained with PI (50 μg/ml), and analyzed by FACS.

IP and Western blotting

IP assay and Western blotting analysis were performed as previously described (15, 20). Antibodies used for IP, immunofluorescence, and Western blotting are detailed in Supplementary Materials and Methods.

IP and MS analysis

N- and C-tagged full-length AHI-1 complementary DNAs were constructed in V180 and V181 vectors and verified by sequencing. Human 293T cells were transiently transfected with constructs expressing either empty vector controls or human N- or C-terminal 3× FLAG human AHI-1. Cells were harvested and lysed, and protein complexes were immunoprecipitated with anti-FLAG monoclonal antibody–agarose. Protein complexes were separated on a 4 to 12% SDS–polyacrylamide gel electrophoresis gel, and experimental bands of interest and their cognate control lane regions were excised for MS analysis. The remaining lanes were divided into 1.5-mm slices and processed individually. Tryptic peptides were analyzed by reverse-phase liquid chromatography and nano-electrospray tandem MS on an Agilent 1100 high-performance liquid chromatography system coupled to an AB Sciex 4000 QTrap mass spectrometer. MS spectra were assigned to proteins using Mascot and X!Tandem software. Interacting proteins were those that were identified with at least two unique peptides in two or more experimental samples, not found in the control samples, and had a Mascot score of greater than 50 and an X!Tandem log(e) score of less than −3. Details of the MS method can be found in (44).

CFC and LTC-IC assays

CFC and LTC-IC assays were performed as previously described (21). For CFC assays, 1000 primary CD34+ cells were mixed with MethoCult H4230 (STEMCELL Technologies) and growth factor cocktail, plus or minus inhibitors. Colonies were counted after 14 days. For LTC-IC assay, CD34+ cells were plated on humanized feeders in MyeloCult H5100 medium (STEMCELL Technologies) and maintained for 6 weeks with weekly half-medium changes. Inhibitors were added on the first day. After 6 weeks, 104 viable cells were plated for CFC assays.

PP2A-C shRNA knockdown

Lentiviral construct (pGFP-C-shLenti with loop: 5′-TCAAGAG-3′, OriGene) targeting PP2A-C (5′-TGGAACTTGACGATACTCTAA-3′) (43) was purchased from OriGene, packaged into lentiviral particles, and transduced into BCR-ABL+ cells. The transduced cells were enriched by FACS of GFP+ cells and knockdown confirmed by Western blotting.

qRT-PCR assays

Total RNA was extracted using the TRIzol RNA extraction method (Life Technologies). Glycogen (10 μg/ml; Life Technologies) was used to visualize the RNA pellet. qRT-PCR was performed as previously described (15). Primers are detailed in Supplementary Materials and Methods.

Engraftment of BCR-ABL+ cells in immunodeficient mice

A total of 2.5 × 106 BV173 YFP+ cells per mouse were injected intravenously into 8- to 10-week-old, sublethally irradiated NSG mice. Two weeks after transplantation, mice from each treatment group were given d-luciferin (50 mg/kg) intraperitoneally and imaged by noninvasive bioluminescence imaging using the IVIS Lumina II CCD camera system to confirm engraftment. Mice were then treated once a day for 2 weeks with vehicle (propylene glycol) or DA (15 mg/kg) by oral gavage and/or LB100 (1.5 mg/kg) intraperitoneally. The extent of engraftment for each mouse was analyzed at different time points after discontinuation of treatment. Engrafted cells in BM, spleen, and liver were analyzed for detection of YFP+ (CD19+) cells by FACS. Liver and spleen sections from the animals were fixed with 10% (v/v) formalin, paraffin-embedded, and stained with H&E for histological analyses.

For the SCLtTA/TRE-BCR-ABL mouse model, leukemia was induced by tetracycline withdrawal, and after 2.5 weeks, 106 unfractionated BM cells were injected intravenously into lethally irradiated Pep3B mice. Healthy Pep3B BM cells (2 × 105) were injected as supporting cells. Three weeks after transplantation, PB from each mouse was analyzed by FACS for detection of CD45.2+ donor cells, to confirm engraftment. The mice were treated as described above. One mouse from each group was sacrificed, and BM, spleen, liver, and PB were analyzed by FACS. All animal experiments were performed in the Animal Resource Center at the British Columbia Cancer Research Centre, and procedures used were approved by the Animal Care Committee of the University of British Columbia. The mice were monitored daily for body weight loss and lethargy.

Statistical analysis

All data are presented as the mean and SEM from at least three independent experiments. Differences between groups were evaluated using paired two-tailed Student’s t test, with P < 0.05 considered statistically significant. For mouse survival curve analysis, median survival of mice from different groups was presented, and log-rank tests were used to evaluate differences between the groups.


Materials and Methods

Fig. S1. Identification of a growth inhibitory compound CAN in AHI-1–transduced CML cells.

Fig. S2. Synergistic cytotoxicity in CML cells by combination of PP2A inhibitors with IM.

Fig. S3. Disruption of cell cycle control and abnormal formation of mitotic spindles in CML cells by PP2A inhibitors.

Fig. S4. Reduced phosphorylation and expression of signaling proteins in K562 cells after treatment with PP2A inhibitors and IM.

Fig. S5. Effects of PP2A inhibitors and IM on CD34+ NBM cells.

Fig. S6. Analysis of mice treated with LB100, DA, or a combination, 7 weeks after transplantation.

Fig. S7. Analysis of mice treated with LB100, DA, or a combination, 10 weeks after transplantation.

Fig. S8. Analysis of progenitor populations in the BM of BCR-ABL transgenic mice.


Acknowledgments: We thank the Stem Cell Assay Laboratory for processing patient samples, members of the Leukemia/BMT Program of British Columbia and the Hematology Cell Bank of British Columbia for providing patient samples, the Terry Fox Laboratory FACS Facility for cell sorting, STEMCELL Technologies for culture reagents, and J. Kovach from Lixte Biotechnology Holdings Inc. for LB100 and LB102 inhibitors and helpful discussions. Funding: This work was supported by the Canadian Institutes of Health Research and in part by the Canadian Cancer Society, the Leukemia & Lymphoma Society of Canada, and the Cancer Research Society (X.J.), and the Leukemia & Lymphoma Society of Canada (C.J.E.). J.S. and K.R. are the recipients of MITACS postdoctoral fellowships. Author contributions: D.L., M.C., J.S., X.L., and K.R. designed and performed experiments and analyzed the data. K.H. performed drug screen experiments. G.B.M. helped design and interpret the IP-MS experiments. D.L.F. provided clinical data. C.J.E. provided expertise in BCR-ABL–transgenic mouse model and in analyzing murine stem cell populations. X.J. developed the concept, designed the experiments, and supervised the study. D.L., J.S., and X.J. wrote the manuscript, and all authors commented on the manuscript. Competing interests: The authors declare that they have no competing interests. Materials and data availability: LB100 and LB102 are available from X.J. under a material transfer agreement with Lixte Biotechnology Holdings Inc. and the British Columbia Cancer Agency.

Stay Connected to Science Translational Medicine

Navigate This Article