Leukemia (2011) 25, 89–100 & 2011 Macmillan Publishers Limited All rights reserved 0887-6924/11 www.nature.com/leu
ORIGINAL ARTICLE Chemoproteomics-based kinome profiling and target deconvolution of clinical multi-kinase inhibitors in primary chronic lymphocytic leukemia cells U Kruse1,3, CP Pallasch2,3, M Bantscheff1,3, D Eberhard1, L Frenzel2, S Ghidelli1, SK Maier1, T Werner1, CM Wendtner2 and G Drewes1 1
Cellzome AG, Heidelberg, Germany and 2Laboratory of Molecular Biology and Immunology of CLL, Department I of Internal Medicine, Center of Integrated Oncology and Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, Cologne, Germany
The pharmacological induction of apoptosis in neoplastic B cells presents a promising therapeutic avenue for the treatment of chronic lymphocytic leukemia (CLL). We profiled a panel of clinical multi-kinase inhibitors for their ability to induce apoptosis in primary CLL cells. Whereas inhibitors targeting a large number of receptor and intracellular tyrosine kinases including c-KIT, FLT3, BTK and SYK were comparatively inactive, the CDK inhibitors BMS-387032 and flavopiridol showed marked efficacy similar to staurosporine. Using the kinobeads proteomics method, kinase expression profiles and binding profiles of the inhibitors to target protein complexes were quantitatively monitored in CLL cells. The targets most potently affected were CDK9, cyclin T1, AFF3/4 and MLLT1, which may represent four subunits of a deregulated positive transcriptional elongation factor (p-TEFb) complex. Albeit with lower potency, both drugs also bound the basal transcription factor BTF2/TFIIH containing CDK7. Staurosporine and geldanamycin do not affect these targets and thus seem to exhibit a different mechanism of action. The data support a critical role of p-TEFb inhibitors in CLL that supports their future clinical development. Leukemia (2011) 25, 89–100; doi:10.1038/leu.2010.233; published online 14 October 2010 Keywords: BMS-387032; chemical proteomics; chronic lymphocytic leukemia; flavopiridol; kinase inhibitors
Introduction In the past decade, kinase inhibitors were introduced for the therapy of malignant diseases, inaugurating a revolution in the treatment of chronic myelogenous leukemia (CML).1 On the basis of key etiological events such as the BCR–ABL fusion protein in CML, KIT and FLT3 mutations in acute myeloid leukemia, JAK2 mutations in myeloproliferative disorders or BRAF mutations in malignant myeloma, kinase inhibitors like imatinib were developed to target aberrant kinase activities.2,3 In some cases, dedicated inhibitors developed for a given indication show promise for the treatment of other diseases. For example, the ABL inhibitor imatinib was initially developed for the treatment of CML, but was subsequently discovered as a Correspondence: Professor Dr CM Wendtner, Laboratory of Molecular Biology and Immunology of CLL, Department I of Internal Medicine, Klinik I fu¨r Innere Medizin, Klinikum der Universita¨t zu Ko¨ln, Kerpener Str. 62, Ko¨ln D-50937, Germany or Dr G Drewes, Cellzome AG, Meyerhofstrasse 1, D-69117 Heidelberg, Germany. E-mails:
[email protected] or
[email protected] 3 These authors contributed equally to this study. Received 21 December 2009; revised 30 July 2010; accepted 23 August 2010; published online 14 October 2010
potent inhibitor of the KIT tyrosine kinase mutated in gastrointestinal stromal tumors. Recently, dasatinib, a drug that was first approved for imatinib-refractory CML, showed promising in vitro activity in chronic lymphocytic leukemia (CLL) cells.4 However, to date no kinase inhibitor has been approved for the treatment of CLL. This disease is generally characterized by the accumulation of CD5 þ /CD19 þ B-cells, which seem to escape apoptosis, in the peripheral blood and lymphoid organs. Microenvironmental influences and stimulus by an autoreactive BCR have a key role in CLL pathogenesis. The standard therapeutic regimen consists of combined immuno- and chemotherapies but hitherto CLL represents an incurable disease.5 Protein kinase inhibitors like dasatinib or flavopiridol are currently investigated in preclinical development and in clinical trials.4,6 In the variable clinical course of the disease, the expression of the tyrosine kinase ZAP-70 represents a prognostic factor of poor clinical course.7 However, the relevant signaling pathways and targets as well as the clinical impact of kinase inhibitors in CLL remain to be elucidated.8 Recent advances in quantitative proteomics have enabled direct and unbiased approaches to elucidate a drug’s mechanism of action directly in the complex native proteome of the target cell or the tissue of interest.9–11 These approaches represent an important advantage in the study of small-molecule kinase therapeutics as efficacious drugs often exhibit polypharmacology; they affect multiple targets and off-targets because protein kinases display a high degree of structural conservation around the ‘druggable’ ATP binding site.12,13 In therapeutic applications in oncology, the increased efficacy gained by the inhibition of a spectrum of targets rather than the selective inhibition of a single target may help overcome problems with efficacy and drug resistance, but may come at a price because of the increased potential for toxic liabilities. As the rational design of drug candidates with a defined spectrum of targets remains a challenge,14 the discovery of compounds with polypharmacology is frequently achieved by cell-based screening in tumor cell lines. However, if the molecular targets of such compounds are incompletely or not at all understood, the subsequent optimization and development efforts can be problematic. Recently, a chemical proteomics strategy based on a mixed inhibitor affinity matrix (kinobeads) and quantitative mass spectrometry has enabled the straightforward determination of the target profiles of small-molecule kinase drugs in cells and tissue lysates. This method allows the generation of quantitative target binding data under close-to-physiological conditions and has recently been applied to identify novel targets of ABL inhibitors including imatinib in an immortalized myelogenous leukemia cell line.9
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
90 In the present study, we apply the kinobeads chemical proteomics approach towards the profiling of clinical-stage kinase inhibitors in primary CLL cells, by correlating their ability to induce apoptosis ex vivo with their kinase target profiles determined directly in patient samples. Under the experimental conditions, many of the target proteins in the sample are preserved as native protein complexes. The results shed new light on the central role of a positive transcriptional elongation factor (p-TEFb) complex consisting of at least four subunits in B-cell malignancies, and demonstrate that the kinobeads methodology offers a powerful approach to translational oncology research. This approach is generally applicable to patient-derived cell samples and informative target profiles of compounds are obtained with limited sample material.
Materials and methods
Reagents and drugs All reagents were purchased from Sigma (St Louis, MO, USA) unless otherwise noted (see Supplementary Information online).
Patient samples and cell culture This study was approved by the ethics committee of the University of Cologne (approval 01-163). Blood samples were obtained from patients fulfilling the diagnostic criteria for CLL with informed consent according to the Helsinki protocol. To ensure optimal conditions, patient samples were processed immediately and only freshly withdrawn blood samples were included. A significant peripheral lymphocytosis of more than 30 000 cells/ml was required for isolation of sufficient cell numbers for proteomics analysis. Only treatment-naı¨ve patients or patients with a minimal treatment-free interval of 12 months were included. Stratification due to FISH, IgVH or CD38 expression was not applied. Clinical features of the patients included in this study are provided in Supplementary Table S1. Fresh blood samples were enriched by applying B-RosetteSep (Stem Cell Technologies, Vancouver, BC, Canada) and by density gradient purification (see Supplementary Information online). Leukapheresis was performed on a COBE spectra apheresis system (Caridian) and CLL cells were isolated with 98% purity. Freshly purified CLL cells were cultured in RPMI media (see Supplementary Information online). Compounds were dissolved in dimethyl sulfoxide (DMSO) at a final concentration of 0.1% and DMSO-treated cells were taken as baseline control. The procedures for HS-5 bone marrow stroma coculture, B-cell receptor and CD40L stimulation were performed as described.15
Apoptosis and cell viability assays Apoptosis was determined by flow cytometry using Annexin V-FITC/7AAD staining (BD Pharmingen, San Diego, CA, USA) after 24 and 48 h. The cellular potency of compounds as defined by half-maximal induction of apoptosis in primary CLL cells was determined using concentrations up to 100 mM. The fraction of viable cells was determined by counting annexin-V/7-AAD double-negative cells for each individual dosage. Median values were subsequently applied for regression analysis and calculation of the half-maximal dosage effect (EC50). Curve fitting was performed using Sigmaplot. EC50 values were determined by fitting data to the Hill Equation. y ¼ y0 þ (axb)/(cb þ xb).
Quantitative immunopurification A volume of 150 mg of anti-CDK9 antibody (ab6544, Abcam, Cambridge, MA, USA) was immobilized on 100 ml AminoLink Leukemia
resin (Thermo Fisher Scientific, Rockford, IL, USA). Whole-cell extracts from Ramos cells and from CLL cells obtained by leukapheresis (10-mg total protein content) were incubated with pre-washed immuno resin on a shaker for 2 h at 4 1C. Beads were washed in lysis buffer containing 0.4% Igepal-CA630 (Sigma) and eluted in 100 ml 2 sodium dodecyl sulfate (SDS) sample buffer. Protein samples were reduced, alkylated and separated by SDS-PAGE. As an isotype-matched specificity control, IgG from the same species was used in an analogous ‘mock immunoaffinity purification (IP)’ carried out in parallel from an aliquot of the Ramos cell sample. After iTRAQ labeling, samples were combined for LC-MS/MS analysis. The enrichment of proteins in CDK9 IPs versus an isotype control antibody IP was calculated from summed-up reporter ion responses A as follows: Enrichment ¼ (A(CDK9-IP)A(mock IP))/(A(CDK9-IP) þ A(mock IP)), such that it scales from 1 to 1.
Kinobeads profiling and mass spectrometry Kinobeads were prepared by immobilization of ATP-mimetics on sepharose beads as described9 with minor modifications (see Supplementary Information online). In addition to the previously reported set of seven ligands, BMS-387032 was included on the beads by immobilization via the piperidine moiety. Mass spectrometry procedures were essentially as described.9,16 In brief, gel lanes were cut into slices across the separation range and subjected to in-gel tryptic digestion followed by labeling with iTRAQ reagents (Applied Biosystems, Foster City, CA, USA) as described.17 Labeled peptide samples were combined as indicated in Supplementary Table S2 to generate eight labeled samples for each dose–response curve, and sequencing was performed by LC-MS/MS on an Eksigent 1D þ high-pressure liquid chromatography (HPLC) system coupled to a LTQOrbitrap mass spectrometer (Thermo Scientific). Peptide extracts of vehicle controls were labeled with iTRAQ reagent 117 and combined with extracts from compound-treated samples labeled with iTRAQ reagents 114–116. Tandem mass spectra were generated using pulsed-Q dissociation, enabling detection of iTRAQ reporter ions.9,16 Peptide mass and fragmentation data were used to query the IPI database using Mascot (Matrix Sciences, Boston, MA, USA). Protein identifications were validated using a decoy database. Quantification based on reporter ion iTRAQ was performed with in-house developed software. All quantification data are listed in the Supplementary Tables. Detailed information on methods is provided in Supplementary Information online.
Results To assess the therapeutic potential of kinase inhibitor drugs in CLL, we first tested a panel of multi-targeted kinase inhibitors in an ex vivo apoptosis assay in patient-derived CLL cells. In the second part of the study, we delineated ‘druggable’ kinase targets by proteomic target deconvolution of the most potent apoptosis-inducing compounds (Figure 1), and conducted a quantitative analysis of protein kinase expression in primary CLL cells by means of the kinobeads approach.9,16
Profiling of multi-kinase inhibitors in CLL cells Multi-kinase inhibitors provide a rapid means of probing the relevance of a sizable pool of putative therapeutic targets in a small set of primary cell-based assays. We selected a set of 13 compounds consisting of pan-kinase tool inhibitors and
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
91
Figure 1 Proteomic drug target deconvolution by a kinobeads competition assay. (a) In the first approach, CLL cells are isolated from patient blood and cell lysates are prepared. The compound of interest is added to lysate aliquots at different concentrations followed by addition of the kinobeads affinity reagent. The kinobeads reagent consists of ATP mimetics immobilized to beads via a linker, and binds to many kinases and other purine binding proteins.9 The drug added to the lysate competes with the kinobeads for binding to the target proteins. As a result, the targets of the drugs present in the lysate are depleted on the beads in a dose-dependent manner. For example, if a target in the lysate is fully complexed with the free drug, it will not be captured on the kinobeads. This depletion is measured relative to a vehicle control by a process consisting of the following steps: (i) elution of the bead-bound proteins; (ii) digestion with trypsin; (iii) tagging of all resulting peptides with four different isobaric tags (iTRAQ reagents); (iv) quantitative mass spectrometry of the combined tagged samples; and (v) identification of target proteins and computation of dose-response curves from the peptide spectra. (b) In the second approach CLL cells isolated from patient blood samples are kept in cell culture and treated with various concentrations of the drug. Subsequently, cell lysates are prepared, incubated with kinobeads and captured proteins are analyzed as described above.
clinical multi-kinase-targeted drugs. The tool compound set consisted of the pan-kinase inhibitor staurosporine,18 the PKC/GSK3 inhibitor Ro31-7549 and PD173955, a nonselective SRC family/ABL kinase inhibitor. The set of approved drugs comprised the multi-targeted tyrosine kinase inhibitors dasatinib and sunitinib, and the VEGFR/RAF inhibitor sorafenib. The set of clinical drug candidates consisted of the multi-tyrosine kinase inhibitors vandetanib (also known as ZD6474), bosutinib (SKI-606), TKI258 (CHIR-258), pazopanib (GW786034), axitinib (AG013736), the SYK/FLT3 inhibitor R406 and the CDK inhibitor BMS-387032 (SNS-032). Sunitinib and sorafenib (phase II) and BMS-387032 (phase I) are currently evaluated in clinical trials for relapsed/refractory CLL (http://www. clinicaltrials.gov). All compounds were tested for their ability to induce apoptosis of the purified primary malignant CLL cells ex vivo.19 The pan-kinase inhibitor staurosporine is known to induce apoptosis in CLL cells and therefore represents a positive control.20 The initial testing was performed at a drug concentration of 1 mM, representing a realistic exposure level typically achieved with kinase inhibitors in patient serum.21 We estimated that our set of compounds (excluding staurosporine) may, at the selected concentration of 1 mM, affect a pool consisting of around 120 distinct kinase targets, or roughly one quarter of the human kinome (for a list of compounds and their targets, see Supplementary Table S3 and references therein). This estimate was derived from the published target potency data of these compounds,9,18,22,23 but does not take into account whether or not the target is expressed in CLL cells. Taking into consideration the typically observed 10-fold potency drop when potencies determined in cell-based assays
are compared with biochemical assays,24 we only listed targets with a reported target potency of at least 100 nM in biochemical enzyme inhibition or binding assays (reported as Kd or Ki/IC50). Each inhibitor was applied to the CLL cells (n ¼ 4 patients) for 24 h, and cell viability was monitored by flow cytometry and staining of apoptotic cells using Annexin V-FITC with co-staining for membrane integrity by 7-AAD. Background apoptosis was low for untreated or vehicle-treated cells with 79.9 and 81.5% viable cells remaining, respectively (Figure 2). As expected, staurosporine exhibited pronounced cytotoxic activity with only 4% viable cells remaining. No substantial impact on cell viability was observed for R-406, Ro31-7549, sorafenib, vandetanib, TKI258, pazopanib and axitinib, and a slight decrease in viability was observed with bosutinib (71% remaining viable cells), sunitinib (68%), dasatinib (67%) and PD173955 (65%). Only BMS-387032 caused a dramatic reduction of viable cells similar to staurosporine (3.3%). By repeating the experiment over a range of drug concentrations, the EC50 for BMS-387032-induced apoptosis in CLL cells was determined as 122 nM (n ¼ 8 patients) (Figure 3a). To assess the specificity of BMS-387032 for CLL cells, the drug was also tested with healthy donor-derived sorted CD19-positive peripheral B cells. We observed a sixfold lower efficacy in the healthy donor-derived cells (EC50 ¼ 870 nM; n ¼ 6 healthy donors) when compared with the CLL cells (Figure 3b). Conditions mimicking microenvironmental stimuli relevant in CLL pathogenesis, including coculture with bone marrow stroma cells, and stimulation of B-cell receptor or CD40L, had no substantial impact on the potency of BMS-387032 (Supplementary Figure S1). Leukemia
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
92
Figure 2 Effect of kinase inhibitors on the survival of CLL cells as measured with an ex-vivo apoptosis assay. Cells were freshly isolated from CLL patient blood samples (n ¼ 4) and were treated with compounds at 1 mM final concentration (in 0.1% DMSO) for 24 h. As a vehicle control 0.1% DMSO was used. Cells were co-stained with Annexin V/7-AAD as marker for apoptosis and subjected to flow cytometry analysis. The percentage of Annexin V/7-AAD double-negative viable cells is shown. (a) Example of fluorescence-activated cell sorting (FACS) analysis for induction of apoptosis by kinase inhibitors. (b) Efficacy of kinase inhibitors for the induction of apoptosis. Untreated cells (native) and DMSO-treated cells served as negative controls. Compounds were used at 1 mM concentration for 24 h incubation. Staurosporine served as positive control and significant induction of apoptosis was observed for geldanamycin, flavopiridol and BMS-387032.
Because BMS-387032 was developed as a CDK2 inhibitor,25 we decided to test two additional clinical CDK inhibitors, flavopiridol (alvocidib), which is currently evaluated in clinical trials for relapsed CLL,26 and R-roscovitine (seliciclib). Leukemia
Flavopiridol exhibited an EC50 of 42 nM in CLL cells versus 23 nM in B cells obtained from healthy donors (Figure 3c), whereas roscovitine showed a markedly lower efficacy in both CLL cells (EC50 ¼ 7.1 mM) and in healthy donor B cells
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
93
Figure 3 Effect of CDK inhibitors on the survival of CLL and normal B cells isolated from healthy donors. Induction of apoptosis (EC50 values) in CLL patient cells and healthy donor B cells was assessed by Annexin V-FITC/7-AAD flow cytometry for the quantification of double-negative viable cells. (a) Induction of apoptosis in CLL cells in vitro by CDK inhibitors. R-roscovitine shows moderate effects with EC50 ¼ 7.1 mM (n ¼ 6 patients). More potent cytotoxic effects were seen for flavopiridol (EC50 42 nM, n ¼ 6 patients) and BMS-387032 (EC50 122 nM, n ¼ 8 patients). (b–d) Cytotoxicity of CDK inhibitors in CLL patient cells (solid lines) versus healthy donor B cells (dashed lines, N ¼ 6 patients). (b) BMS-387032 exhibited more pronounced toxicity of 122 nM in CLL cells versus 870 nM in healthy donor B cells. No substantial difference of EC50 was observed in (c) flavopiridol-treated cells (42 versus 23 nM) and (d) R-roscovitine-treated cells (7.1 versus 7.2 mM).
(EC50 ¼ 7.2 mM) (Figure 3d). In summary, BMS-387032 and flavopiridol seemed equally potent for the induction of apoptosis in CLL cells, whereas R-roscovitine in comparison showed only moderate activity.
Differential proteomic profiling of kinases in primary CLL cells and Ramos cells Research on B cells frequently uses cell lines derived from rapidly growing B-cell lymphomas, like Ramos or Raji cells. As difficulties in establishing permanent cell lines have hampered studies on CLL,27 we obtained CLL cells with moderate Zap-70 expression from a patient by leukapheresis and purified the cells by untouched depletion.28 Both CLL and B-cell lymphomas are known to be sensitive to inhibitors of B-cell receptor signaling like Rituximab.29 Hence, we quantitatively compared the complement of expressed kinase targets in CLL cells with Ramos lymphoma cells, by performing differential mapping of kinobeads-captured proteins. To test whether the kinobeads matrix might be further optimized toward identification of relevant targets by inclusion of the most active compound in the CLL apoptosis assay, BMS-387032, we performed a parallel experiment, replacing kinobeads with beads containing immobilized BMS-387032. Lysates of CLL and Ramos cells were subjected to precipitation with either kinobeads or BMS-387032-beads, the captured proteins were digested with trypsin, labeled with isobaric tags, mixed, purified by highpressure liquid chromatography and analyzed quantitatively with tandem mass spectrometry (LC-MS/MS). Approximately 1200 proteins were differentially quantified, including 190 protein kinases (Supplementary Figure S2, and Supplementary Tables S4 and S5). The profiles of the Ramos lymphoma cells and the CLL cells reveal considerable differences in kinase
expression (Figure 4a). Of note, we also identified many proteins known to be associated with kinases (using information from the Human Protein Reference Database, hprd.org). Many protein complexes are identified, including CDK–cyclin complexes, complexes of Syk with BCR subunits and proximal SRC family tyrosine kinase complexes, indicating that complexes are largely preserved under the experimental conditions (Figure 4b). The BMS-387032 matrix captured a few kinases not identified with the kinobeads matrix, like the PCTAIRE and PFTAIRE kinases of the CDK family, which have been linked to genotoxicity.30 We therefore decided to include BMS-387032 as a capturing ligand in the kinobeads for the subsequent inhibitor profiling experiments.
Proteomic profiling of kinase inhibitors in primary CLL cells Target deconvolution of the active compounds BMS-387032, flavopiridol and R-roscovitine was performed using primary CLL cells. To assess the quantitative target binding in the CLL cells, we employed the kinobeads competition-binding assay in combination with isobaric peptide tagging and LC-MS/MS.9,16 In brief, aliquots of the lysate were spiked with vehicle or with drug in concentrations ranging from 40 nM to 10 mM. Subsequently, kinases and associated or related proteins in each sample were captured on kinobeads. The principle of the kinobeads assay is that the free drug and the immobilized ATPsite mimetics on the beads compete for binding to the target proteins in the lysate. As a consequence, the targets of the drugFand any proteins physically associated with the target in a complexFare prevented from binding to the beads in a dosedependent manner, while the binding of other proteins to the beads is not affected (Figure 1). Protein binding to the beads is Leukemia
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
94
Figure 4 Differential kinase-expression heat maps of primary CLL cells and the Ramos B-cell lymphoma cell line. Proteins were captured from lysates of CLL or Ramos cells either with kinobeads or with immobilized BMS-387032 as indicated, and relative protein quantification was performed by isobaric tagging and mass spectrometry. A total of 1200 proteins were differentially quantified from both the cell samples, including 190 protein kinases (for complete data see Supplementary Figure S1 and Supplementary Table S3). (a) Heat map representing relative levels of captured protein kinases for the 20 kinases, which differ most prominently between CLL and Ramos cells. The CLL cells express cytoplasmic tyrosine kinases of the SRC family at more than 10-fold higher levels, whereas Ramos cells express some mitotic kinases like Aurora A and B, MYT1, and PLK4 at 4–5-fold higher levels. (b) Heat map representing relative levels of captured protein kinases, and co-precipitated interacting proteins. Many proteins previously described as kinase interactors are captured on the beads along with their kinase partners. Examples include CDK–cyclin complexes, complexes of Syk with BCR subunits and proximal SRC-family tyrosine kinase complexes. Protein interaction data are from http://www.hprd.org.
monitored by quantitative LC-MS/MS of the combined peptide pools from compound-treated and untreated lysate samples after isobaric tagging. For each peptide detected by MS/MS, the decrease of signal intensity relative to the vehicle control reflects competition by the ‘free’ drug for its target (Supplementary Figure S3). From this dataset, dose–response binding profiles were calculated for all proteins detected, including between 130 and 140 protein kinases (Figure 5, Supplementary Table S5). As expected, staurosporine was found to bind to many kinases in the CLL cell lysate exemplifying its pleiotropic effects; 39 out of 133 kinases were found to exhibit more than 50% binding reduction on kinobeads when 1 mM drug was present in the lysate. Among the potently targeted kinases are CaMKK2, AMPK, MST1, CaMK2, PDK1, SYK, PKCb and FER (IC50 values in the lysate all below 0.2 mM (Figure 5)). The two potent CDK inhibitors showed distinct but more selective, partially overlapping target profiles (Figure 5). For both flavopiridol and BMS-387032 the most potently inhibited kinase is CDK9, with IC50 values of 30 nM for flavopiridol and 50 nM for BMS-387032 consistent with the slightly more pronounced effect of flavopiridol on cell viability. CDK9 is known to be part of the p-TEFb complex, which has a key role in RNA polymerase II transcriptional elongation, and in human cells consists of the Leukemia
CDK9 catalytic subunit and one of three cyclins, T1, T2 or K.31 The proteomics approach used in this study can identify intact protein complexes because the conditions used for lysate preparation and drug binding largely preserve intact protein complexes.9 Consequently, the binding of the drugs to the p-TEFb complex was detected by matching dose–response binding curves for the catalytic subunit CDK9 and the regulatory subunit cyclin T1 (CCNT1). Of note, we detected two additional proteins with inhibition profiles closely matching those of p-TEFb: the putative transcriptional activator AFF4 (MCEF), which was previously described as a positive regulator in transcriptional elongation,32 and MLLT1 (ENL), a protein encoded by a gene involved in chromosomal translocations in acute leukemias33 (Figure 6a). Cyclin K (CCNK) was affected by both drugs, but with lower potencies compared with the other p-TEFb subunits, indicating that cyclin K may be a part of a distinct complex in CLL cells (Supplementary Table S6). The third CDK inhibitor studied, R-roscovitine, exerted a 100-fold lower efficacy on CLL cell viability and consistently it was found to inhibit p-TEFb only at micromolar concentrations. R-roscovitine was about equipotent for the CDK-activating kinase (CAK) complex with the core components CDK7, cyclin H (CCNH) and menage a trois (MNAT1). This complex was also
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
95 found to be targeted by BMS-387032 and flavopiridol with IC50 values of 1–3 mM, and therefore it is unlikely to represent a primary target for the proapoptotic effects on CLL cells (Figure 6b). An additional protein, ERCC2, was found to be
inhibited with dose–response profiles matching CDK7, CCNH and MNAT1 and hence is likely to represent a further subunit of the BTF2/TFIIH complex or sub-complex targeted by the inhibitors. The target profiles of flavopiridol and BMS-387032 contain some notable differences, with the latter drug generally appearing to be more selective. Both drugs inhibit CDK10, CRK7 and PFTAIRE1 with similar potencies. BMS-387032 inhibits PCTAIRE2 (30-fold) and GSK3a (ca. 10-fold) more strongly, whereas flavopiridol is more potent for CDK5 and inhibited several additional kinases not affected by BMS-387032. A protein of unknown function encoded by C14ORF129 has been described as a GSK3 interacting protein,34 but exhibits a very similar inhibition profile to PCTAIRE and may represent an interactor of this kinase in CLL cells.
Target profiling in live CLL cells To address the interaction of the drugs with their targets more directly, we examined the targets of geldanamycin and BMS387032 in living primary cells as outlined in Figure 1. CLL cells were cultured in the presence of the drugs for 6 h and lysates were subjected to kinobeads profiling and LC-MS/MS analysis as described above. Derivatives of the heat shock protein 90 (HSP90) inhibitor geldanamycin are currently in clinical trials for CLL and have been reported to induce apoptosis in CLL cells by triggering the degradation of HSP90 client proteins, among which are many kinases. We found geldanamycin to be a moderately effective inducer of apoptosis in CLL cells (Figure 2). When cells were treated with geldanamycin, we observed partial downregulation of several out of 65 kinases detected, with the most pronounced effect on BLK. Several CDKs were detected but were not substantially affected, indicating that downregulation of CDKs does not probably account for the efficacy of geldanamycin (Supplementary Figure S4). Because cell numbers were limited, the profile of BMS-387032 obtained in live CLL cells contained data for only 92 kinases compared with the 133 kinases detected in the above lysate profiling experiment. Consistent with the lysate profile, we again detected potent binding to CDK9 and PCTK2 (EC50 ¼ 0.1 mM), and less potent binding to GSK3A, CDK7 and CDK5 (Supplementary Figure S4).
Characterization of the CDK9 complex as a drug target in CLL cells To further analyze the composition of the CDK9/p-TEFb complex in CLL cells, we performed a quantitative IP from the same CLL cells that were used in the inhibitor profiling
Figure 5 Kinase-binding profiles in primary CLL cells for staurosporine and the CDK inhibitors BMS-387032 and flavopiridol. The bar graphs represent kinobeads competition binding profiles across a set of protein kinases, simultaneously identified from primary CLL cells. pIC50 values for individual kinases, defined as the log concentration of drug at which half-maximal competition of kinobeads binding is observed, are indicated for staurosporine (green), flavopiridol (orange) and BMS-387032 (blue). Numbers printed above the bars represent IC50 values in mM. CLL cells were purified from a leukapheresis sample and a cell lysate was prepared. Compounds were added to lysate aliquots at various concentrations (10, 3, 1, 0.3, 0.1, 0.03 mM, 0.5% DMSO as vehicle) followed by addition of kinobeads. Relative amounts of captured proteins from each aliquot were monitored by quantitative mass spectrometry. Proteins in the lysate that bind the drugs show reduced binding to kinobeads. Leukemia
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
96
Figure 6 Proteomic profiling indicates a p-TEFb complex consisting of four proteins as main target of BMS-387032 and BTF2/TFIIH complexes as additional targets in CLL cells. (a) Binding of BMS-387032 to the p-TEFb complex is detected by matching dose–response binding curves for the catalytic kinase subunit CDK9 and the regulatory subunit cyclin T1 (CCNT1). Two additional proteins were detected with matching inhibition profiles, AFF4 (MCEF) and MLLT1, a member of the AF4 family of transcription factors, likely representing additional robust associations in the p-TEFb complex. (b) Binding of BMS-387032 to the BTF2/TFIIH complex is detected by CDK-activating kinase (CAK) complex with the core components CDK7, cyclin H (CCNH) and menage a trois (MNAT1). A fourth protein, ERCC2 was found to be inhibited with dose–response profiles matching CDK7, CCNH and MNAT1 and hence is likely to represent an additional subunit of the BTF2/TFIIH complexes.
Figure 7 Mapping of CDK9-associated proteins by immunoaffinity purification (IP) and quantitative mass spectrometry confirms the complex with AFF3/4 and MLLT1. CDK9 was immunoprecipitated from CLL cells or Ramos cells, and co-enriched proteins were quantified relative to an isotypematched specificity control antibody by isobaric tagging and LC-MS/MS. The enrichment of proteins in CDK9 IPs versus an isotype control antibody IP was calculated such that it scales from 1 to 1 (see Materials and methods). The size of the squares indicates the strength of identification (spectral counts). (a) Proteins enriched versus control from CLL versus Ramos cells. The box highlights proteins that show more than 10-fold enrichment in both cell types. (b) The same plot as in a, but only showing proteins that were also identified in the kinobeads competition binding experiments with CDK9 inhibitors BMS-387032 and flavopiridol. (c) Heat map representation of IC50 values and relative abundances of strongly enriched proteins (blue box in b) in IPs from CLL and Ramos cells using a heat map. Dark green indicates high relative abundance and light green low relative abundance.
experiments and from Ramos cells. Protein enrichment in the IP samples was quantified by LC-MS/MS relative to a mock-IP with control IgG (Figure 7a, Supplementary Table S7). Leukemia
We determined a substantial (410-fold) enrichment for 51 proteins in the CDK9 IP sample (Supplementary Figure S5). As false positives, because of limited specificity of the antibody
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
97
Figure 8 Kinobeads expression profiling of CDK9-associated proteins by quantitative mass spectrometry indicates overexpression of AFF3/4 and MLLT1 in CLL cells from five CLL patients (orange bars: Zap-70positive patients, red bars: Zap-70-negative patients, see Supplementary Table S1) and normal B cells. Kinases and associated proteins were isolated from CLL cell samples and from a pool of B cells from six healthy donors, and were subjected to quantitative LC-MS/MS profiling.
or of other factors, are a common problem in IP-mass spectrometry experiments, and previous data have shown that the active CDK9 target complex was captured on kinobeads, we focused on the proteins that were detected both in the IP and in the kinobeads inhibitor profiling samples. This approach clearly delineated a defined protein complex formed by CDK9, cyclin T1, AFF3/4 and MLLT1 (Figure 7b). Finally, we asked whether this complex or any of its components were differentially expressed in CLL cells compared with normal B cells. We collected CLL cells from five patients and used normal B cells from six healthy subjects as control. The control samples were pooled, as they were available in limited amounts only. We performed differential mapping of the kinobeads-captured CDK9 complex proteins in the patient cells and the pooled normal B cells by LC-MS/MS (Figure 7c, Supplementary Table S8). Whereas the core complex components CDK9 and cyclin T1 are only present at marginally higher levels in the cells from the five CLL patients compared with the normal B-cell pool, several components of the CDK9 target complex, AFF3/4 and MLLT1, are overexpressed in the CLL cells from all five patients (Figure 8).
Discussion The profiling of a panel of multi-targeted kinase inhibitors led to the identification of BMS-387032 as a potent inducer of apoptosis in CLL cells. By contrast, the majority of the tested tyrosine and serine/threonine kinase inhibitors showed little or no effect. For many of the inhibitors tested, the spectrum of
kinase targets is defined through extensive published in vitro selectivity data.9,18 In total, the combined target list of the inhibitors that tested ineffective in the current study is projected to add up to around 120 distinct kinases, which thus appear not to be critical for proapoptotic mechanisms in peripheral CLL cells (Supplementary Table S1). The lack of efficacy of these compounds effectively restricts the list of useful kinase targets for the induction of apoptosis in CLL cells in the periphery. It should be noted that our results may not exclude the option to consider any of these drugs for the treatment of CLL, because the targets of these inhibitors may still have a role in dividing CLL cells in the proliferation centers in the spleen and bone marrow.35 The quantitative expression data on the kinobeadscaptured sub-proteome generated in the present study show the prominent expression of several SRC family kinases in CLL compared with Ramos cells, confirming published data on Lyn.36 For Lyn in particular, we observed a different set of interacting proteins expressed in CLL versus Ramos cells (Figure 4b), supporting the notion of aberrant regulation of this kinase in CLL. However, pan-SRC kinase inhibitors including dasatinib, bosutinib, TKI258 and PD173955 were inactive or marginally active in our assays. Therefore, these kinases seem not to be directly involved in cell survival mechanisms. Our results for dasatinib are in agreement with published data4 and suggest that this drug, despite targeting a multitude of kinase pathways including BCR signaling through BTK,37 does not directly target any pathways that peripheral CLL cells may be ‘addicted’ to. In line with the notion that the viability of peripheral CLL cells may not critically depend on BCR signaling, we did not likewise observe any substantial effect of the SYK inhibitor R406.38 However, BCR-dependent responses of CLL cells may still be of importance in cell migration and survival in tissue microenvironments, and were recently reported to be susceptible to micromolar R406 concentrations.39 Prompted by the identification of the CDK inhibitor BMS387032 as an effective inducer of CLL cell death in line with recent observations,40 we also tested two structurally unrelated CDK inhibitors, flavopiridol and R-roscovitine . Flavopiridol was slightly more effective than BMS-387032, whereas roscovitine is approximately 100-fold less potent. Our results are broadly in agreement with a recent report showing that BMS-387032 effectively killed CLL cells regardless of prognostic factors or treatment history. In this study, BMS-387032 was 10–20-fold more potent than flavopiridol, whereas roscovitine was two orders of magnitude less effective in the induction of cell death.40,41 With BMS-387032 described recently as a potent and selective inhibitor of several CDK enzymes based on recombinant protein,40 we decided to assess its mechanism of action as compared with flavopiridol and roscovitine by the kinobeads method, which represents an unbiased proteomics approach to profile the targets of a drug in a cell lysate.9 The target binding profiles of the drugs were determined from the endogenous proteins monitored directly in patient-derived CLL cells by quantitative LC-MS/MS, and protein complexes were detected by matching concentration–response binding data for proteins in a stoichiometric complex. The data clearly indicate that the p-TEFb complex is the main target for mediating the proapoptotic effects of BMS-387032 and flavopiridol. Different from other CDK/cyclin complexes, the p-TEFb containing CDK9 is not involved in regulation of the cell cycle but has a key role in RNA polymerase II elongation. The p-TEFb complex is thought to be comprised of CDK9 as the catalytic subunit and one of three cyclins T1, T2 or K, as the regulatory subunit.31,42 Of note, we detected two additional proteins with inhibition profiles closely matching those of CDK9 and cyclin T1, the Leukemia
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
98 putative transcriptional activator AFF4 (MCEF), an AF-4 domain protein previously described as a positive regulator of transcriptional elongation,32 and MLLT1 (ENL), a YEATS domain protein33 (Figure 6). AFF4 was found previously to associate with p-TEFb by immunoprecipitation and MLLT1 has been reported to bind to AFF4, and both proteins were found to be substrates for CDK9.32,43,44 As AFF4 and MLLT1 do not seem to be ‘druggable’ proteins, the parsimonious explanation for the matching dose–response profiles obtained for these proteins is that they represent additional regulatory or targeting subunits of the p-TEFb complex. The composition of a defined protein complex formed by CDK9, cyclin T1, AFF3/4 and MLLT1 in CLL cells and in Ramos cells was confirmed by quantitative IP. MS/MS-based quantification of the complex subunits in five CLL patients and in normal B cells indicated overexpression of the AFF3/4 and MLLT1 subunits but not the core components CDK9 and cyclin T1 in CLL cells from all five patients. Interestingly, both AFF4 and MLLT1 constitute fusion partners of the mixed-lineage leukemia (MLL) gene, a frequent target for recurrent translocations in acute leukemias. MLL encodes a DNA-binding protein with methyltransferase activity toward histone H3 and stimulates target gene expression including HOX genes.45 It is tempting to speculate that fusions between AFF4 or MLLT1 and MLL may recruit the p-TEFb complex to MLL, converting it into a leukemogenic oncoprotein. While the current study was under review, additional evidence for this mechanism in MLL was published.46,47 Hence, leukemias caused by these translocations should be sensitive to p-TEFb inhibitors like BMS-387032 and flavopiridol. Regarding the role of p-TEFb in CLL, it has recently been suggested that the inhibition of CDK9 by BMS-387032 would lead to a decrease in transcription and the level of short-lived antiapoptotic proteins such as Mcl-1 and XIAP resulting in cell death.40,41 However, because of its role in transcriptional regulation, the inhibition of CDK9 may lead to mechanism-based toxicity. A recent phase I study showed that BMS-387032 caused tumor lysis syndrome and myelosuppression.48 In addition to the effect on p-TEFb, our data show that the CDK-related kinase PCTAIRE is a novel potently inhibited target of BMS-387032, but not flavopiridol, whereas the related kinase PFTAIRE is moderately inhibited by both drugs. A protein of unknown function encoded by C14ORF129 exhibits an inhibition dose response very similar to PCTAIRE and may represent a regulatory subunit of this kinase. The inhibition of PCTAIRE and PFTAIRE kinases by small-molecule inhibitors has recently been linked to toxicity caused by chromosomal damage.30 The kinobeads approach is not limited to the profiling of ATPcompetitive kinase inhibitors, but is also suitable to characterize agents that affect kinase abundance or activity in cells by other mechanisms, for example by inhibiting HSPs. The function of HSP90 is to maintain client proteins in a correctly folded conformation. Inhibition of HSP90 leads to proteasomal degradation of client proteins including several cancer targets and HSP90 inhibitors are in clinical development for the treatment of cancer. Geldanamycin was reported to induce apoptosis in CLL cells after 4 h of treatment with 30–100 nM drug.49 The primary cells used in our experiments were ZAP-70 negative and consistently we did not detect this kinase. Geldanamycin exerted moderate effects on cell survival, and subsequent kinobeads and LC-MS/MS analysis indicated the downregulation of BLK, a kinase involved in the control of proliferation during B-cell development and possibly in the formation B-lymphoid tumors.50 Together with the fact that BLK was also among the targets potently affected by staurosporine and therefore may contribute to its proapoptotic activity, the Leukemia
data represent circumstantial evidence that BLK may represent a suitable therapeutic target in CLL. In conclusion, the combination of ex vivo cell assays with the kinobeads-based quantitative proteomics approach enables, for the first time, the determination of the affinity of small-molecule inhibitors for their target protein complexes directly in patientderived samples. The observed lack of proapoptotic efficacy of a panel of clinical multi-kinase inhibitors suggests that the inhibition of major tyrosine kinase pathways, including those downstream of the BCR, does not efficiently target CLL cells in the periphery. By contrast, the observed strong effects of CDK9/p-TEFb inhibitors reveal a promising future for selective CDK9 drugs that do not affect cell cycle CDKs, inhibition of which might cause serious side effectsFand might be incorporated in comprehensive combinatorial treatment regimens.
Conflict of interest UK, MB, DE, SG, TW and GD are employees of Cellzome AG. CPP and CMW received research funding from Cellzome AG. All the authors declare no other conflict of interest.
Acknowledgements We are thankful to Reinhild Brinker, Anja Podszuweit and Jessica Perrin for expert technical assistance, to Vale´rie Reader for compound synthesis, to Judith Schlegl, Christine Gmu¨nd and Vincent Wolowski for software and database development, and to Yann Abraham for help with data analysis. We also thank Frank Weisbrodt for help with the figures. We would like to thank Tim Edwards, Jason Fisherman, Carsten Hopf, Gitte Neubauer and David Simmons for suggestions and support. This work was supported by a grant from the German Bundesministerium fu¨r Bildung und Forschung (Spitzencluster BioRN, Verbundprojekt Inkubator/Teilprojekt BioRN-IND-TP02) to Cellzome AG. CPP and CMW were supported by the Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), Ko¨ln Fortune, the CLL Global Research Foundation, and the Marga and Walter Boll-Stiftung.
Author contributions UK, CPP and MB designed and performed research, analyzed data and wrote the paper; DE, LF, SKM, SG and TW performed research; CMW provided CLL patient samples and wrote the paper; GD conceptualized the project, analyzed data and wrote the paper.
References 1 Druker BJ. Translation of the Philadelphia chromosome into therapy for CML. Blood 2008; 112: 4808–4817. 2 Kumar C, Purandare AV, Lee FY, Lorenzi MV. Kinase drug discovery approaches in chronic myeloproliferative disorders. Oncogene 2009; 28: 2305–2313. 3 Zhang J, Yang PL, Gray NS. Targeting cancer with small molecule kinase inhibitors. Nat Rev Cancer 2009; 9: 28–39. 4 Veldurthy A, Patz M, Hagist S, Pallasch CP, Wendtner CM, Hallek M et al. The kinase inhibitor dasatinib induces apoptosis in chronic lymphocytic leukemia cells in vitro with preference for a subgroup of patients with unmutated IgVH genes. Blood 2008; 112: 1443–1452.
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
99 5 Wendtner CM, Eichhorst BF, Hallek MJ. Advances in chemotherapy for chronic lymphocytic leukemia. Semin Hematol 2004; 41: 224–233. 6 Phelps MA, Lin TS, Johnson AJ, Hurh E, Rozewski DM, Farley KL et al. Clinical response and pharmacokinetics from a phase 1 study of an active dosing schedule of flavopiridol in relapsed chronic lymphocytic leukemia. Blood 2009; 113: 2637–2645. 7 Wiestner A, Rosenwald A, Barry TS, Wright G, Davis RE, Henrickson SE et al. ZAP-70 expression identifies a chronic lymphocytic leukemia subtype with unmutated immunoglobulin genes, inferior clinical outcome, and distinct gene expression profile. Blood 2003; 101: 4944–4951. 8 Gaiger A, Heintel D, Jager U. Novel molecular diagnostic and therapeutic targets in chronic lymphocytic leukaemia. Eur J Clin Invest 2004; 34 (Suppl 2): 25–30. 9 Bantscheff M, Eberhard D, Abraham Y, Bastuck S, Boesche M, Hobson S et al. Quantitative chemical proteomics reveals mechanisms of action of clinical ABL kinase inhibitors. Nat Biotechnol 2007; 25: 1035–1044. 10 Nita-Lazar A, Saito-Benz H, White FM. Quantitative phosphoproteomics by mass spectrometry: past, present, and future. Proteomics 2008; 8: 4433–4443. 11 Sharma K, Weber C, Bairlein M, Greff Z, Keri G, Cox J et al. Proteomics strategy for quantitative protein interaction profiling in cell extracts. Nat Methods 2009; 6: 741–744. 12 Apsel B, Blair JA, Gonzalez B, Nazif TM, Feldman ME, Aizenstein B et al. Targeted polypharmacology: discovery of dual inhibitors of tyrosine and phosphoinositide kinases. Nat Chem Biol 2008; 4: 691–699. 13 Peters EC, Gray NS. Chemical proteomics identifies unanticipated targets of clinical kinase inhibitors. ACS Chem Biol 2007; 2: 661–664. 14 Hopkins AL, Mason JS, Overington JP. Can we rationally design promiscuous drugs? Curr Opin Struct Biol 2006; 16: 127–136. 15 Pallasch CP, Schulz A, Kutsch N, Schwamb J, Hagist S, Kashkar H et al. Overexpression of TOSO in CLL is triggered by B-cell receptor signaling and associated with progressive disease. Blood 2008; 112: 4213–4219. 16 Bantscheff M, Boesche M, Eberhard D, Matthieson T, Sweetman G, Kuster B. Robust and sensitive iTRAQ quantification on an LTQ Orbitrap mass spectrometer. Mol Cell Proteomics 2008; 7: 1702–1713. 17 Ross PL, Huang YN, Marchese JN, Williamson B, Parker K, Hattan S et al. Multiplexed protein quantitation in Saccharomyces cerevisiae using amine-reactive isobaric tagging reagents. Mol Cell Proteomics 2004; 3: 1154–1169. 18 Karaman MW, Herrgard S, Treiber DK, Gallant P, Atteridge CE, Campbell BT et al. A quantitative analysis of kinase inhibitor selectivity. Nat Biotechnol 2008; 26: 127–132. 19 Rawstron AC, Bennett F, Hillmen P. The biological and clinical relationship between CD5+23+ monoclonal B-cell lymphocytosis and chronic lymphocytic leukaemia. Br J Haematol 2007; 139: 724–729. 20 Kitada S, Zapata JM, Andreeff M, Reed JC. Protein kinase inhibitors flavopiridol and 7-hydroxy-staurosporine down-regulate antiapoptosis proteins in B-cell chronic lymphocytic leukemia. Blood 2000; 96: 393–397. 21 Gill AL, Verdonk M, Boyle RG, Taylor R. A comparison of physicochemical property profiles of marketed oral drugs and orally bioavailable anti-cancer protein kinase inhibitors in clinical development. Curr Top Med Chem 2007; 7: 1408–1422. 22 Brehmer D, Godl K, Zech B, Wissing J, Daub H. Proteome-wide identification of cellular targets affected by bisindolylmaleimidetype protein kinase C inhibitors. Mol Cell Proteomics 2004; 3: 490–500. 23 Wisniewski D, Lambek CL, Liu C, Strife A, Veach DR, Nagar B et al. Characterization of potent inhibitors of the Bcr-Abl and the c-kit receptor tyrosine kinases. Cancer Res 2002; 62: 4244–4255. 24 Knight ZA, Shokat KM. Features of selective kinase inhibitors. Chem Biol 2005; 12: 621–637. 25 Misra RN, Xiao HY, Kim KS, Lu S, Han WC, Barbosa SA et al. N-(cycloalkylamino)acyl-2-aminothiazole inhibitors of cyclindependent kinase 2. N-[5-[[[5-(1,1-dimethylethyl)-2-oxazolyl]methyl]thio]-2-thiazolyl]-4- piperidinecarboxamide (BMS-387032),
26 27 28
29 30
31 32
33 34 35
36
37
38
39
40
41
42 43
44
45
a highly efficacious and selective antitumor agent. J Med Chem 2004; 47: 1719–1728. Chen R, Keating MJ, Gandhi V, Plunkett W. Transcription inhibition by flavopiridol: mechanism of chronic lymphocytic leukemia cell death. Blood 2005; 106: 2513–2519. Loisel S, Ster KL, Quintin-Roue I, Pers JO, Bordron A, Youinou P et al. Establishment of a novel human B-CLL-like xenograft model in nude mouse. Leuk Res 2005; 29: 1347–1352. Essakali S, Carney D, Westerman D, Gambell P, Seymour JF, Dobrovic A. Negative selection of chronic lymphocytic leukaemia cells using a bifunctional rosette-based antibody cocktail. BMC Biotechnol 2008; 8: 6. Hillmen P. Advancing therapy for chronic lymphocytic leukemia– the role of rituximab. Semin Oncol 2004; 31 (Suppl 2): 22–26. Olaharski AJ, Gonzaludo N, Bitter H, Goldstein D, Kirchner S, Uppal H et al. Identification of a kinase profile that predicts chromosome damage induced by small molecule kinase inhibitors. PLoS Comput Biol 2009; 5: e1000446. Bres V, Yoh SM, Jones KA. The multi-tasking P-TEFb complex. Curr Opin Cell Biol 2008; 20: 334–340. Bitoun E, Oliver PL, Davies KE. The mixed-lineage leukemia fusion partner AF4 stimulates RNA polymerase II transcriptional elongation and mediates coordinated chromatin remodeling. Hum Mol Genet 2007; 16: 92–106. Tkachuk DC, Kohler S, Cleary ML. Involvement of a homolog of Drosophila trithorax by 11q23 chromosomal translocations in acute leukemias. Cell 1992; 71: 691–700. Ewing RM, Chu P, Elisma F, Li H, Taylor P, Climie S et al. Large-scale mapping of human protein-protein interactions by mass spectrometry. Mol Syst Biol 2007; 3: 89. Smit LA, Hallaert DY, Spijker R, de GB, Jaspers A, Kater AP et al. Differential Noxa/Mcl-1 balance in peripheral versus lymph node chronic lymphocytic leukemia cells correlates with survival capacity. Blood 2007; 109: 1660–1668. Contri A, Brunati AM, Trentin L, Cabrelle A, Miorin M, Cesaro L et al. Chronic lymphocytic leukemia B cells contain anomalous Lyn tyrosine kinase, a putative contribution to defective apoptosis. J Clin Invest 2005; 115: 369–378. Hantschel O, Rix U, Schmidt U, Burckstummer T, Kneidinger M, Schutze G et al. The Btk tyrosine kinase is a major target of the Bcr-Abl inhibitor dasatinib. Proc Natl Acad Sci USA 2007; 104: 13283–13288. Braselmann S, Taylor V, Zhao H, Wang S, Sylvain C, Baluom M et al. R406, an orally available spleen tyrosine kinase inhibitor blocks fc receptor signaling and reduces immune complexmediated inflammation. J Pharmacol Exp Ther 2006; 319: 998–1008. Quiroga MP, Balakrishnan K, Kurtova AV, Sivina M, Keating MJ, Wierda WG et al. B cell antigen receptor signaling enhances chronic lymphocytic leukemia cell migration and survival: specific targeting with a novel Syk inhibitor, R406. Blood 2009; 114: 1029–1037. Conroy A, Stockett DE, Walker D, Arkin MR, Hoch U, Fox JA et al. SNS-032 is a potent and selective CDK 2, 7 and 9 inhibitor that drives target modulation in patient samples. Cancer Chemother Pharmacol 2009; 64: 723–732. Chen R, Wierda WG, Chubb S, Hawtin RE, Fox JA, Keating MJ et al. Mechanism of action of SNS-032, a novel cyclin-dependent kinase inhibitor, in chronic lymphocytic leukemia. Blood 2009; 113: 4637–4645. Wang S, Fischer PM. Cyclin-dependent kinase 9: a key transcriptional regulator and potential drug target in oncology, virology and cardiology. Trends Pharmacol Sci 2008; 29: 302–313. Estable MC, Naghavi MH, Kato H, Xiao H, Qin J, Vahlne A et al. MCEF, the newest member of the AF4 family of transcription factors involved in leukemia, is a positive transcription elongation factor-b-associated protein. J Biomed Sci 2002; 9: 234–245. Mueller D, Bach C, Zeisig D, Garcia-Cuellar MP, Monroe S, Sreekumar A et al. A role for the MLL fusion partner ENL in transcriptional elongation and chromatin modification. Blood 2007; 110: 4445–4454. Krivtsov AV, Armstrong SA. MLL translocations, histone modifications and leukaemia stem-cell development. Nat Rev Cancer 2007; 7: 823–833. Leukemia
Proteomic profiling of multi-kinase inhibitors in cll U Kruse et al
100 46 Lin C, Smith ER, Takahashi H, Lai KC, Martin-Brown S, Florens L et al. AFF4, a component of the ELL/P-TEFb elongation complex and a shared subunit of MLL chimeras, can link transcription elongation to leukemia. Mol Cell 2010; 37: 429–437. 47 Yokoyama A, Lin M, Naresh A, Kitabayashi I, Cleary ML. A higherorder complex containing AF4 and ENL family proteins with P-TEFb facilitates oncogenic and physiologic MLL-dependent transcription. Cancer Cell 2010; 17: 198–212. 48 Tong WG, Chen R, Plunkett W, Siegel D, Sinha R, Harvey RD et al. Phase I and pharmacologic study of SNS-032, a potent and selective Cdk2, 7, and 9 inhibitor, in patients with advanced
chronic lymphocytic leukemia and multiple myeloma. J Clin Oncol 2010; 28: 3015–3022. 49 Jones DT, Addison E, North JM, Lowdell MW, Hoffbrand AV, Mehta AB et al. Geldanamycin and herbimycin A induce apoptotic killing of B chronic lymphocytic leukemia cells and augment the cells’ sensitivity to cytotoxic drugs. Blood 2004; 103: 1855–1861. 50 Malek SN, Dordai DI, Reim J, Dintzis H, Desiderio S. Malignant transformation of early lymphoid progenitors in mice expressing an activated Blk tyrosine kinase. Proc Natl Acad Sci USA 1998; 95: 7351–7356.
Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)
Leukemia
