CBP Modulates Sensitivity to Dasatinib in Pre-BCR⁺ Acute Lymphoblastic Leukemia

Although the treatment of acute lymphoblastic leukemia (ALL) has improved during the last decades (1), 25% of children and 65% of adults die within 5 years after diagnosis with relapse and chemotherapy-related complications being among the main causes of death (2, 3). Several ALL subtypes have been described on the basis of their karyotype, cell type, immunophenotype, and gene expression profile. The pre-BCR⁺ ALL subtype is characterized by expression of the pre-B-cell receptor (pre-BCR; ref. 4) and in 50% of cases is associated with the chromosomal translocation t(1;19), coding for the chimeric fusion protein E2A-PBX1 (5), which is present in about 5% of pediatric and adult ALL (6). Using preclinical models, we and others have suggested that dasatinib, a small-molecule multi-tyrosine kinase inhibitor, is a candidate therapy in pre-BCR⁺ ALL (4, 7–10), and early clinical evidence supports our observations (11). However, novel therapies in ALL including small-molecule inhibitors such as dasatinib are limited by the development of resistance and short-lived responses (12). An understanding of drug sensitivity and resistance mechanisms to modern targeted small molecules will support the development of more rational combination therapies to improve efficacy and survival of patients and reduce adverse effects.

Using unbiased functional genomic screens and global transcriptomic analytic approaches, we studied intrinsic primary and acquired mechanisms of dasatinib sensitivity and resistance in pre-BCR⁺ ALL. Our studies identified CREB-binding protein (CBP) as a major modulator of dasatinib sensitivity and demonstrate that its targeting by small-molecule inhibitors shows promising preclinical efficacy increasing sensitivity to dasatinib in ALL.

Human leukemia cell lines were cultured in RPMI1640 medium supplemented with 10% FBS, 100 U/mL penicillin/streptomycin, and 0.29 mg/mL l-glutamine. E2A-PBX1–positive cell lines (697, RCH-ACV, and Kasumi-2) were authenticated using Western blot analysis for E2A-PBX1 fusion protein expression. E2A-PBX1–negative cell lines (SEM, REH, and HAL-01) were used and not further authenticated. All cell lines were obtained from DSMZ in 2013. Cell lines were negatively tested for Mycoplasma contamination by PCR. The REH cell line was found listed in the database of commonly misidentified cell lines maintained by International Cell Line Authentication Committee, and was used here as control. Pre-BCR status was assessed by flow cytometry for surface VPREB1. Fresh human umbilical cord blood was collected from newborn placenta (with informed consent and Stanford University Institutional Review Board approval), CD34⁺ cells were enriched and cultured as described previously (13). Primary ALL cells were obtained from the Tissue Bank of the Department of Hematology/Oncology, University Medical Center of Freiburg (Freiburg, Germany) and of the German cooperative ALL-BFM study group, University Medical Center Schleswig-Holstein, Campus Kiel (Kiel, Germany). Experiments with human tissue samples were performed in accordance with the tenets of the declaration of Helsinki and were approved by the Ethics Commission from the University of Freiburg (ethical vote, approval no. 279/17). All patients were informed by a physician and signed an informed consent form. Human cell lines, primary leukemia cells, and cord blood CD34⁺ cells were treated with dasatinib (LC Laboratories), saracatinib (Selleckchem), bosutinib (Selleckchem), dexamethasone (Sigma-Aldrich), XX-650-23 (14), and ICG-001 (Selleckchem) at the indicated concentrations. Titration curves were performed using increasing drug concentrations and cells were counted by Trypan blue assay after 4 days.

The shRNA screen was performed as described elsewhere (15, 16). Three shRNA lentivirus sublibraries containing a total of about 106,249 shRNA sequences and targeting 3,158 human genes (about 25 shRNAs for most genes and 50 shRNAs for kinases/phosphatases) and containing 13,774 negative control shRNAs were used to transduce RCH-ACV cells in four pools. After puromycin selection and expansion, RCH-ACV cells stably expressing pooled shRNAs were split and subjected to four rounds of treatment with dasatinib (20 nmol/L) or vehicle, in two replicates. For each round, cells were treated with dasatinib for 72 hours. The dasatinib resistance/sensitivity conferred by an individual shRNA was calculated using Mann–Whitney U test and casTLE, a novel statistical framework for cas9 and shRNA screening technologies (v1.0 available from: https://bitbucket.org/dmorgens/castle; ref. 17). The enrichment score for each shRNA was normalized by the distribution of enrichment scores of the negative control shRNAs. With this normalization, drug resistance phenotype is independent from growth phenotype (15). Candidate shRNA shown to confer sensitivity or resistance are shown in Supplementary Table S1.

Individual shRNA sequences (Supplementary Table S2) were cloned into the BstXI site of the p309 lentiviral vector (15), for stable transduction in human ALL cell lines. For mutagenesis studies, RCH-ACV cells were transduced with a modified vector expressing CRISPR/Cas9 (LentiCas9-Blast, Addgene 52962), and selected with blasticidin. Stably transduced cells were infected with a pU6-sgRNA EF1Alpha-puro-T2A-GFP lentiviral vector containing sgRNAs cloned into the BstXI/BlpI site (18). sgRNA sequences were selected for CBP from a human genome-wide library and from Gecko v2 database and are listed in Supplementary Table S3.

Lentivirus generation and transduction of human cells are described elsewhere (19). After puromycin selection, transduced cells were mixed 50/50 for competition assays using cells with control vectors and cells with shRNA knockdown or sgRNA knockout for the gene of interest. Mixed cells were split for vehicle and drug treatment. Indel frequency was analyzed by Sanger sequencing and TIDE assay and performed as described previously (20).

Apoptosis and cell-cycle analysis were described elsewhere (21). Briefly, apoptotic cells were quantified on the basis of Annexin V staining (eBioscience) by flow cytometry. BrdU/7-AAD incorporation assays were performed according to the manufacturer's protocol (catalog no. 559619, BD Pharmingen).

For xenograft bone marrow transplantation, sublethally (2 Gy) irradiated female NSG mice of 8–9 weeks of age were injected via tail vein with 1000 RCH-ACV leukemia cells. After randomization, mice were treated daily, intraperitoneally, with dasatinib starting 7 days after transplantation with vehicle (30% PEG1500, 1% Tween 80, 2.5% DMSO dissolved in PBS) or 2.5 mg/kg dasatinib for 20 days. For combination treatment with dasatinib and XX-650-23, mice were treated with dasatinib as described previously (9). Cohorts of mice treated with XX-650-23 received the drug dissolved in PBS daily, intravenously, starting 7 days after transplantation with 1.25 mg/kg (14). The dose was increased every 4 days, in 1.25 mg/kg steps, to reach a maximal drug dose of 5 mg/kg daily. Mice were treated for 25 days. Analysis was performed in a nonblinded fashion, and statistical methods were not used to estimate sample size. All data collected from mice were included for analysis. Disease-free survival was defined by signs of illness including general lymphadenopathy, lethargy, weight loss, and shivering. Moribund mice were euthanized.

RNA was isolated using the RNeasy Mini Kit (Qiagen) and cDNA was synthesized using Superscript Reverse Transcriptase IV Kit (Life Technologies) following the manufacturer's recommendations. Relative expression was quantified using an ABI 7900HT Thermocycler. TaqMan Master mix (Applied Biosystems) following Taqman gene expression assays from Life Technologies (Supplementary Table S4). All signals were quantified using the ΔΔC_t method and normalized to ΔΔC_t values of the Actb gene expression levels.

Western blot analysis for protein quantification was performed using a modified RIPA lysis buffer as described previously (9). The following antibodies were used for immunodetection: rabbit anti-CBP (clone C-20; Santa Cruz Biotechnology), mouse anti-mouse AcH3K27 (catalog no. 17-683, EMD Millipore), rabbit anti-histone H3 (#Ab1791, Abcam), rabbit anti-Gapdh (#G9545, Sigma-Aldrich), rabbit anti-HA (#Ab9110, Abcam), and mouse anti-RAC1 (clone 23A8, #05-389, EMD Millipore). Secondary antibodies were rabbit anti-mouse (#816729, Life Technologies) and goat anti-rabbit (#G21234, Life Technologies) coupled to horseradish peroxidase. Quantification by densitometry was performed using ImageJ software (NIH, Bethesda, MD).

RNA from dasatinib-sensitive and dasatinib-resistant RCH-ACV cells from two independent sublines were used for RNAseq analysis following isolation using an RNeasy Mini Kit (Qiagen) and library preparation using the TruSeq Stranded mRNA Kit (#RS-122-2101, Illumina). Sequencing was performed in the Stanford Genome Center and Personalized Medicine using an Illumina HiSeq2500 (paired end 101 bp). RNA sequences were aligned and analyzed in the web-based platform DNAnexus using following software tools (version): bedtools (2.19.0), bwa-aln (0.6.2-r126), bwa-mem (0.7.7-r441), fastqc (0.10.1), Picard Tools [1.107(1667)], and samtools (0.1.19-44428cd).

Differential gene expression analysis was performed by DESeq2 in R platform using a P < 0.05 as cutoff. Gene ontology enrichment of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways was analyzed using DAVID bioinformatics resources (22) and row Z scores of were calculated and visualized in heatmaps using CIMminer. For gene set enrichment analysis (23), RPKM values from expressed genes were plotted using GraphPad Prism software (GraphPad Inc.).

For SNV/indel calling, annotation and effect prediction, Varscan 2 (24) and SNPeff (25) software tools were employed using the following parameters: 50 transcripts minimum, P <0.05, heterozygosity defined as >0.25 of transcripts and are shown in Supplementary Table S5. Germline, noncoding, and synonymous SNV/indels were filtered out. SNV/indels were confirmed and visualized by integrative genomics viewer (26). The CBP point mutation R1748H was validated after PCR amplification and Sanger sequencing using specific primers (Supplementary Table S6).

cDNA of CTNNB1 with 4 phosphorylation site mutations in the N-terminal linked to hemagglutinin gene from mouse was amplified using following primers (5′ primer ATGGCTACTCAAGCTGACCTG and 3′ primer CAGGTCAGTATCAAACCAGGC). PCR products were isolated and cloned into the pMYS-IRES-GFP retroviral vector using the In-Fusion HD Cloning Kit (Clontech Laboratories Inc) as described previously (9).

Statistical differences between two groups were analyzed by two-sided Mann–Whitney U test or by Student t test. Dose–response curves were compared using the extra sum of squares F test. The Bliss independence model (27) was used to evaluate synergy between drug combinations. A P < 0.05 was considered statistically significant. Statistical differences from Kaplan–Meier curves were analyzed by log-rank (Mantel–Cox) test. Statistical analysis and graphs were performed using GraphPad Prism software (GraphPad Inc.).

All experiments on mice were performed with the approval of and in accordance with Stanford's Administrative Panel on Laboratory Animal Care (APLAC, Protocol 9839).

A functional genomic screen was performed to prospectively identify novel genes and pathways involved in dasatinib sensitivity and resistance (15, 16). The pre-BCR⁺ ALL cell line RCH-ACV was transduced using lentiviral vectors harboring pooled shRNA sublibraries targeting in total 3,158 human genes (106,249 shRNAs) including negative controls (Fig. 1A). The sublibraries employed were selectively enriched for shRNAs that targeted either kinases, phosphatases, epigenetic factors, or cancer-related pathways. Stably transduced cells were treated with dasatinib for 12 days (4 pulses) of in vitro growth (Fig. 1B). Dasatinib- and vehicle-treated cells were then analyzed by deep-sequencing to quantify shRNA relative abundance. Several candidate shRNAs were observed to confer dasatinib resistance or increase dasatinib sensitivity (Fig. 1C; Supplementary Table S1). Gene ontology analysis of shRNAs that increased dasatinib sensitivity identified the B-cell receptor signaling pathway as significantly enriched (P = 5.7 × 10⁻⁹; FDR = 6.9 × 10⁻⁶), including kinases ZAP70 and SYK, consistent with previous observations of their roles in pre-BCR⁺ ALL (10) and validating the shRNA screening approach. Notably, shRNAs targeting CREB-binding protein (CBP) were markedly depleted in the dasatinib-treated population compared with control shRNAs (Supplementary Table S1), suggesting that depletion of the coactivator CBP increases dasatinib sensitivity in human pre-BCR⁺ ALL cells.

Competition growth assays were performed to validate single shRNAs identified in the functional genomics screen (10). shRNA-mediated knockdown of CBP in RCH-ACV ALL cells (Fig. 2A) had significant stimulatory effects on cell growth in the absence of dasatinib (Fig. 2B), consistent with the previously described tumor suppressor function of CBP in ALL (28) and lymphoid malignancies (29). In contrast, CBP knockdown substantially inhibited proliferation of dasatinib-treated RCH-ACV cells compared with control transduced cells (Fig. 2B). The interactive effects of CBP knockdown and dasatinib treatment were observed in a second E2A-PBX1⁺/pre-BCR⁺ cell line, 697, but were not observed in representative pre-BCR⁻ or non-E2A-PBX1 cells (Supplementary Fig. S1A and S1B). Mechanistically, shRNA-mediated repression of CBP results in decreased expression levels of pre-BCR–associated genes, previously associated with dasatinib sensitivity and resistance including ZAP70, SYK, and LCK (10) and novel identified genes in our shRNA screen as BTK and LYN (Fig. 2C and D). These results demonstrated a synthetic lethal interaction between CBP and dasatinib in two independent pre-BCR⁺ ALL cell lines consistent with the shRNA screen results, and reinforced the role of CBP as a master transcriptional regulator of pre-BCR/BCR–associated genes (28–29).

To further confirm a role for CBP in dasatinib sensitivity, mutagenesis studies were performed using a CRISPR/Cas9 approach in the context of competition growth assays. RCH-ACV cells expressing sgRNAs that efficiently targeted CBP (Supplementary Fig. S2A and S2B) showed a proliferative advantage in the presence of vehicle but considerably decreased competitive fitness in the presence of dasatinib consistent with the knockdown studies (Supplementary Fig. S2C). Thus, two different genetic approaches confirmed that CBP modulates dasatinib sensitivity in pre-BCR⁺ ALL cells.

Given the foregoing genetic results, a pharmacologic approach was employed to test the effects of CBP inhibitors in combination with dasatinib. The inhibitory compound XX-650-23, which antagonizes the interaction between the KIX domain of CBP and the KID domain of the transcription factor CREB (14), showed a synergistic effect in combination with dasatinib on cell growth inhibition of pre-BCR⁺ cell lines RCH-ACV (Fig. 3A and B) and 697 (Supplementary Fig. S3). Pharmacologic inhibition of CBP also increased dasatinib sensitivity in Kasumi-2 cells (which are E2A-PBX1⁺ pre-BCR⁻) and REH (TEL-AML1⁺ ALL), but not SEM, HAL-01, and SUP-B15, which lack E2A-PBX1 (Supplementary Fig. S3), suggesting some selectivity. Dasatinib in combination with XX-650-23 did not inhibit growth of human cord blood cells (IC₅₀ >10 μmol/L) cultured in vitro (Supplementary Fig. S4). These data suggest that the combination targeted therapy might be safe and effective in a subset of leukemias, particularly pre-BCR⁺ ALLs.

The preclinical efficacy of dasatinib in combination with XX-650-23 was tested in vivo after xenotransplantation of RCH-ACV cells in NSG mice. As expected, in vivo treatment with dasatinib alone prolonged disease-free survival, consistent with previous observations using a conditional E2A-PBX1 transgenic mouse model (9). Notably, cotreatment with dasatinib and XX-650-23 modestly, but significantly, increased the disease-free survival of transplanted NSG mice compared with either drug alone (Fig. 3C, median disease-free survival for XX-650-23, 30 days; for dasatinib, 39 days; and for combination, 43 days). Consistent with the shRNA results, pharmacologic inhibition of CBP with XX-650-23 resulted in decreased expression levels of pre-BCR–associated genes implicated in dasatinib sensitivity in RCH-ACV cells (Fig. 3D and E), and in a subset of primary ALL cells with selected karyotypes including E2A-PBX1⁺ ALL (Supplementary Fig. S5). These data further establish a novel combination therapy that inhibits a subset of human leukemias compared with normal hematopoietic progenitors and shows promising preclinical efficacy in vivo.

To prospectively define mechanisms of acquired dasatinib resistance in pre-BCR⁺ human leukemias, two independent dasatinib-resistant RCH-ACV sublines were generated by culture through multiple passages in the presence of increasing concentrations of dasatinib for several months. Acquired resistance to dasatinib was confirmed in vitro (Fig. 4A; Supplementary Fig. S6A) and in vivo in xenotransplantation assays (Fig. 4B). Functionally, dasatinib-resistant cells showed reduced apoptosis/cell death (Supplementary Fig. S6B) and increased cycling cell frequencies compared with dasatinib-sensitive cells in the presence of dasatinib (Supplementary Fig. S6C). Dasatinib-resistant cells also displayed cross-resistance to the chemically distinct SRC family kinase inhibitors bosutinib and saracatinib, but not to the corticosteroid dexamethasone (Supplementary Fig. S7A–S7C). These data suggest that dasatinib-resistant cells developed a specific mechanism of drug resistance to SRC family kinase inhibitors but not in general to other chemotherapeutics.

The most frequent mechanism of resistance to small-molecule inhibitors including dasatinib is point mutations in the targeted kinase domain (30, 31). Therefore, RNA-seq was performed to detect SNVs and indels in transcribed genes in dasatinib-resistant compared with dasatinib-sensitive RCH-ACV cells. RNA-seq data showed that various SNVs and indels were acquired in the dasatinib-resistant sublines (Fig. 4C; Supplementary Table S5); however, none affected SRC family kinases or other dasatinib targets (HMS LINCS database; ref. 32), suggesting a possible bystander effect in the acquisition of dasatinib resistance.

Conversely, several SNVs originally present in the dasatinib-sensitive cells were unexpectedly absent in the dasatinib-resistant sublines (11 SNVs in each). One of these was a R1748H substitution in the CBP gene present in dasatinib-sensitive RCH-ACV cells (allele frequency of 12% in dasatinib-sensitive subline #1 and 58% in dasatinib-sensitive subline #2 in RNA-seq analysis), but not in dasatinib-resistant sublines as confirmed by Sanger sequencing (Fig. 4D; Supplementary Table S5). This SNV, which has not been previously described, is located in proximity to the histone acetyl-transferase domain of CBP. Mutations in this region can impair histone acetylation and transcriptional regulation of CBP target genes, including genes involved in signal transduction by the B-cell receptor and B-cell–specific transcription factors (28, 29). These data suggest that RCH-ACV cells heterozygous for R1748H may be partially insufficient for CBP tumor suppressor function, but this SNV was lost independently in both sublines during acquisition of dasatinib resistance likely due to selective pressure for restoration of wild-type CBP function.

The observed selection on the CBP gene suggested that RCH-ACV cells might acquire dasatinib resistance through transcriptional plasticity by an epigenetic mechanism, as described for several enzymatic and epigenetic inhibitors (33–35). Because epigenetic states are dynamic and undergo reversal in the absence of selective conditions, we assessed for potential loss of resistance over time. After 2 months culture off-drug, dasatinib-resistant cells partially reversed their resistance phenotypes (Fig. 5A), suggesting transcriptional plasticity as a possible contributory cause of dasatinib-acquired resistance.

Whole transcriptome analysis of RNA-seq data from dasatinib-sensitive and resistant ALL cells from two independent sublines (Fig. 5B) identified 1,810 downregulated genes and 1,669 upregulated genes in common (P ≤ 0.05). Gene ontology analysis of common regulated genes identified several pathways involved in acquired dasatinib resistance (Fig. 5C and D; Supplementary Fig. S8A and S8B). Pathways depleted in dasatinib-resistant sublines included DNA replication/cell cycle, p53 signaling, and DNA repair pathways. Consistent with the transcriptional profile, dasatinib-resistant sublines showed a proliferative disadvantage in vitro in competition assays (Supplementary Fig. S9A and S9B) as well as in vivo after xenograft transplantation (Supplementary Fig. S9C). Depletion of the p53 signaling pathway was consistent with resistance to apoptosis/cell death (Supplementary Fig. S6B). These data suggest a mechanistic model in which dasatinib-resistant cells may bypass specific pathways inhibited by dasatinib through compensatory activation of alternative pathways.

Gene ontology analysis identified the WNT, mTOR, and MAPK signaling pathways enriched in dasatinib-resistant RCH-ACV cells (Fig. 5D; Supplementary Fig. S10A). Furthermore, RNAseq data showed that expression of several members of the WNT signaling pathway was increased in dasatinib-resistant RCH-ACV cells (Supplementary Fig. S10A and S10B), which was confirmed by qRT-PCR (Supplementary Fig. S10C). To validate the WNT pathway as a contributor to the dasatinib-resistant phenotype, a stabilized β-catenin protein was expressed in RCH-ACV–sensitive cells (Fig. 6A), which were subsequently treated with increasing dasatinib concentrations. Cells expressing exogenous β-catenin were significantly more resistant to dasatinib than cells transduced with empty vector (Fig. 6B). To further confirm the role of the WNT pathway in primary and acquired dasatinib resistance, we knocked down the GTPase RAC1, a positive regulator of the WNT signaling pathway (36) that was also identified as a sensitizer in our shRNA screen. Depletion of RAC1 in RCH-ACV–sensitive and -resistant cells (Supplementary Fig. S11A) caused decreased competitive fitness in the presence of dasatinib compared with control transduced cells (Supplementary Fig. S11B). These data suggest that the WNT pathway is involved in primary and acquired dasatinib resistance in pre-BCR⁺ ALLs.

Given the central role of CBP in dasatinib sensitivity and its interaction with β-catenin in regulation of gene expression (37), we tested the effect of ICG-001, an inhibitor of CBP interaction with β-catenin (38), in acquired dasatinib resistance. Combination treatment with ICG-001 and dasatinib displayed a synergistic interaction in two independent dasatinib-resistant RCH-ACV sublines (Fig. 6C and D; Supplementary Fig. S12A). The KIX domain inhibitor XX-650-23 also increased dasatinib sensitivity in the dasatinib-resistant sublines (Fig. 6E; Supplementary Fig. S12B), indicating that CBP can modulate intrinsic dasatinib sensitivity through multiple domains.

Our studies using unbiased genetic approaches identified CBP as a major regulator of dasatinib sensitivity in a subset of acute leukemia. Pharmacologic inhibition of CBP enhances de novo sensitivity to dasatinib, overcomes acquired dasatinib resistance, and shows encouraging preclinical efficacy in vitro and in vivo.

Dasatinib is an FDA-approved drug for the treatment of BCR-ABL⁺ CML and ALL (39). Known mechanisms of dasatinib resistance in these clinical settings include somatic mutations in the ABL kinase domain and BCR-ABL amplifications (30, 31). Novel mechanisms of dasatinib resistance in experimental models of BCR-ABL⁺ ALL have been proposed to involve upregulation of the antiapoptotic oncogene BCL6 (40) as well as priming of mesenchymal stem cells (41). Preclinical studies demonstrate that dasatinib may also be an efficacious therapy for the subset of ALL that is pre-BCR⁺ through inhibition of SRC-family kinases including LCK (42). Functional studies suggested that depletion or inhibition of the kinases LCK, ZAP70, and SYK enhances dasatinib sensitivity in pre-BCR⁺ ALL. Interestingly, LCK, ZAP70, and SYK expression is directly regulated by E2A-PBX1, a common driver oncogene in pre-BCR⁺ leukemias (9, 10). Notably, the dasatinib IC₅₀ concentration for SYK was described at 2.9 μmol/L (32), which is likely very similar to the IC₅₀ concentration for its closely related tyrosine kinase family member ZAP70. This is not a physiologically achievable concentration, suggesting that the observed effect of dasatinib on pre-BCR⁺ ALL is likely mediated through the inhibition of SRC family kinases including BTK, LCK, and LYN. Altogether, we hypothesized that the spectrum of drug resistance mechanisms would be different compared with BCR-ABL–driven leukemias involving gene expression and transcriptional plasticity.

This was addressed through shRNA screens, which can identify synthetic lethal genetic interactions (43) and drug resistance mechanisms to targeted therapies (44), thus providing a powerful platform to identify effective drug combinations to overcome resistance. Our screen identified several genes, most notably CBP, involved in dasatinib sensitivity in human pre-BCR⁺ ALL cells. The role of CBP as a regulator of dasatinib sensitivity was validated using genetic and pharmacologic techniques. The observed increased sensitivity to dasatinib through pharmacologic inhibition and genetic inactivation of CBP in RCH-ACV compared with 697 cells might be due to higher CBP expression in the former (Supplementary Fig. S1; Fig. 2). Our results support a crucial role in drug sensitivity mechanisms and intensify the focus on CBP as a potential drug target in ALL.

CBP is a large protein (265 kDa), with multiple domains amenable to potential drug targeting. It functions as a coactivator of gene transcription with histone acetyltransferase activity involved in many different cellular processes (45). CBP loss of function has been previously implicated in ALL relapsing disease and pathogenesis (28, 46) consistent with its role as a haploinsufficient tumor suppressor in lymphoma (29). Conversely, CBP also serves a pro-oncogenic function in leukemia, and is a promising drug target in AML where the KIX domain inhibitor XX-650-23 shows antileukemic activity (14). Moreover, an inhibitor of CBP-β-catenin interaction showed efficacy as a single agent and in combination with chemotherapy in ALL (47). Mechanistically, CBP regulates pre-BCR/BCR-associated genes (28, 29), involved in dasatinib sensitivity and resistance in our previous work (10) or identified in our shRNA screen. Genetic and pharmacologic inhibition of CBP results in downregulation of pre-BCR pathway signaling members increasing synergistically sensitivity to dasatinib in pre-BCR⁺ ALLs. Dasatinib is known to inhibit BTK, LCK, and LYN at the nanomolar range (32). Thus, there is some overlap between kinases inhibited by dasatinib and regulated by CBP, suggesting that shRNA-mediated knockdown of CBP is pulling out several signals with synthetic lethality. Our studies credential CBP as a target for combination therapy with dasatinib and establish its role to enhance drug sensitivity in a subset of ALL.

Global transcriptomic studies identified several pathways depleted in dasatinib-resistant ALL cells including DNA replication/cell cycle, p53 signaling, and DNA repair machinery, which correlate with the resistance cellular phenotype. Conversely, several pathways are enriched in dasatinib-resistant cells, including WNT, MAPK, and mTOR signaling, which are amenable to pharmacologic inhibition and have been associated with drug resistance mechanisms. Because of the critical role of the CBP–β-catenin interaction in gene regulation, we focused on the WNT signaling pathway. Genetic perturbation of the pathway by overexpression of β-catenin or RAC1 knockdown confirmed an alternative pathway to overcome dasatinib resistance. Indeed, the CBP–β-catenin inhibitor ICG-001 showed a synergistic effect in combination with dasatinib further highlighting the role of CBP in modulating dasatinib sensitivity through bypass of targeted signaling pathways via WNT. Our observations extend recent studies of the role of the WNT signaling pathway in primary resistance to bromodomain inhibitors in AML (34, 35) as well as acquired resistance to doxorubicin in neuroblastoma (48), suggesting a convergence to a common WNT pathway of resistance to multiple anticancer therapies. In these experiments, we used an in vitro system with human cell lines and therefore, the data should be interpreted with caution. Future studies are required to elucidate whether same or similar mechanisms occur in patients with pre-BCR⁺ ALL treated with dasatinib.

Mutations of drug targets are a common mechanism of acquired resistance to kinase targeted therapies. Although mutational analysis showed acquisition of SNVs/indels in dasatinib-resistant cells, they did not affect known kinases targeted by dasatinib (HMS LINCS database; ref. 32), suggesting that they may be passenger mutations or have an indirect effect on dasatinib resistance through activation of alternative pathways. However, we observed strong selection for loss of the R1748H SNV in the CBP gene in dasatinib-resistant cells. Several loss-of-function variants affecting the histone acetyl-transferase domain of CBP or regions nearby have been described and well characterized in ALL (28). The pre-BCR⁺ ALL cell line RCH-ACV was established from bone marrow cells at relapse after chemotherapy, suggesting similar origin and probable loss-of-function as described mutations in relapsed ALL (49). We hypothesize, that pre-BCR⁺ ALL cells with impaired CBP function are more sensitive to dasatinib as further confirmed by pharmacologic studies. Collateral sensitivity of primary or relapsed ALL (28) and B-cell lymphomas with impaired dysfunctional CBP (50) might extend the indication for dasatinib treatment. Of note, the combination of XX-650-23 and dasatinib has a synergistic effect on cell growth inhibition not only of E2A-PBX1⁺/pre-BCR⁺ cells (RCH-ACV and 697) but also of E2A-PBX1⁺/pre-BCR- (Kasumi-2) and non-E2A-PBX1 cells (REH, THP-1). Hence, we observed a modest but statistically significant prolonged disease-free survival of dasatinib and XX-650-23 in xenograft transplantation experiments. The synergistic effect was observed at 2–4 μmol/L concentration of XX-650-23 in vitro and these concentrations were likely not reached in vivo. Our data suggest that the development of small-molecule inhibitors targeting CBP with improved pharmacodynamics and pharmacokinetics profile might provide a therapeutic option for a subset of ALLs particularly those that are pre-BCR⁺. Prospective studies and further experimentation beyond pre-BCR⁺ ALL are needed to establish how broad the indication is for combination therapy using dasatinib and CBP inhibitors.

In conclusion, using two different unbiased approaches, we have elucidated mechanisms of dasatinib sensitivity in human pre-BCR⁺ ALL and identified CBP as a master regulator to facilitate activation of alternative signaling pathways such as pre-BCR and WNT. Small molecules targeting CBP show promising preclinical efficacy modulating dasatinib sensitivity and should be considered in future clinical trials to prevent the emergence of resistant/relapsing ALL clones.

The RNA sequencing data that support the findings of this study have been deposited in the Gene Expression Omnibus (http://ncbi.nlm.nih.gov/geo) with the accession number GSE97352.

No potential conflicts of interest were disclosed.

Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the NSF.

Conception and design: J. Duque-Afonso, C.-H. Lin, H.-D. Chae, K.M. Sakamoto, M.C. Bassik, M.L. Cleary

Development of methodology: J. Duque-Afonso, C.-H. Lin, L. Zhu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Duque-Afonso, C.-H. Lin, K. Han, J. Jeong, L. Zhu, M. Schrappe, G. Cario, M.L. Cleary

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Duque-Afonso, C.-H. Lin, K. Han, D.W. Morgens, E.E. Jeng, S.H.K. Wong, J. Duyster, M.L. Cleary

Writing, review, and/or revision of the manuscript: J. Duque-Afonso, C.-H. Lin, J. Jeong, L. Zhu, M.C. Wei, H.-D. Chae, M. Schrappe, G. Cario, K.M. Sakamoto, M.L. Cleary

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.-H. Lin, J. Jeong, X. Xiao

Study supervision: J. Duque-Afonso, M.L. Cleary

Other (library construction and sequencing): Z. Weng

We thank Maria Ambrus and Cita Nicolas for technical assistance, members of the Cleary laboratory for helpful discussions, and Carlos Duque-Afonso for graphic design. This work was supported in part by grants from the NIH (CA214888), the William Lawrence and Blanche Hughes Foundation (to M.L. Cleary), the German Research Foundation (Deutsche Forschungsgemeinschaft, ref. DU 1287/2-1 and 1287/3-1) and Research Commission of the University of Freiburg Medical School (Forschungskommission, ref. DUQ1106/16 to J. Duque-Afonso), the Lucile Packard Foundation for Children's Health, the Child Health Research Institute and the Stanford NIH-NCATS-CTSA grant #UL1 TR001085 (to M.L. Cleary, J. Duque-Afonso, and C.-H. Lin), Alex's Lemonade Stand Foundation for Childhood Cancer (to S.H.K. Wong). K. Han was supported by The Walter V. and Idun Berry Postdoctoral Fellowship Program and M.C. Bassik was supported by a grant from Stanford ChEM-H and an NIH Directors New Innovator Award (1DP2HD08406901). K.M. Sakamoto has been supported by the Maxfield Foundation, SPARK program and Child Health Research Institute Lucile Packard Foundation for Children's Health, Leukemia and Lymphoma Society of America (SLP-8009-15), Pediatric Cancer Research Foundation, Hyundai Hope on Wheels (#02500CA), USC Parker Hughes Institute for Childhood Cancer Research/William Lawrence & Blanche Hughes Foundation, St. Baldrick's Foundation, and Bear Necessities/Rally Foundation. E.E. Jeng was supported by a Hubert Shaw and Sandra Lui Stanford Graduate Fellowship. This material is based on work supported by the National Science Foundation (NSF) Graduate Research Fellowship (grant DGE-114747, to D.W.Morgens).

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