A prototype nonpeptidyl, hydrazone class, thrombopoietin receptor agonist, SB-559457, is toxic to primary human myeloid leukemia cells

Thrombopoietin (Tpo) is the primary regulator of megakaryopoiesis and platelet production.^1,2  On binding to its cognate receptor, Mpl, it initiates a signaling cascade that results in the proliferation of human megakaryocyte progenitor cells, and the subsequent production of platelets.^1,3,4  Neutralization, or knockout, of either Tpo or Mpl leads to profound thrombocytopenia and hemorrhage.^1,2 

The cloning and expression of recombinant human Tpo (rhTpo)^5,6  was heralded as a major treatment advance for thrombocytopenic patients.²  Unfortunately, some people receiving a pegylated form of Tpo developed cross-reacting, antiendogenous Tpo antibodies and profound thrombocytopenia.⁷  To avoid this calamity, alternative Mpl-stimulating agents were developed.^2,8,9  One of these, a nonpeptidyl, hydrazone class small molecule, named Eltrombopag (E; SB-497115), was recently approved by the Food and Drug Administration for use in adult chronic idiopathic thrombocytopenic purpura patients.^10,11 

Because Mpl and/or Tpo is expressed by many malignant hematopoietic cell types,^12-14  concern that Mpl agonists might exacerbate a leukemia patient's disease is appropriate. We investigated this issue using primary human leukemia cells and a prototype nonpeptidyl, hydrazone class, Mpl agonist, SB-559457 (SB; Figure 1A) whose in vitro megakaryocytopoietic properties are virtually indistinguishable from rhTpo (5μM SB = 100 ng/mL; 2.8μM rhTpo).¹⁵  Counterintuitively, SB proved specifically toxic to myeloid leukemia cells.¹⁶ 

SB-559457 and Eltrombopag were supplied by Glaxo SmithKline Pharmaceuticals. The compounds were dissolved in 100% dimethyl sulfoxide (DMSO) to prepare 10mM stock solution and then diluted as needed. rhTpo (R&D Systems) was dissolved in Iscove modified Dulbecco medium (Invitrogen) to a final concentration of 5 ng/mL.

Primary human leukemia cells were obtained from the Stem Cell and Xenograft Core of the Abramson Cancer Center, University of Pennsylvania. This Core is Institutional Review Board–approved by the University of Pennsylvania to provide material from consenting patients. MO, M1, M2, M4, and M5 subtypes were represented in acute myeloid leukemia (AML) samples; precursor B-cell, precursor T-cell, and L3/Burkitt in acute lymphoblastic leukemia (ALL) samples; and both chronic and blast phase in chronic myeloid leukemia (CML) samples. Cells were grown in suspension culture (37°C, 100% in humidity, 5% CO₂) using Endothelial Cell Medium-2 plus 2% FBS (Cambrex).

The reverse-transcribed polymerase chain reaction (RT-PCR) analysis was performed as previously published¹⁷  using the following primers: REDD1: forward-dTGTTTAGCTCCGCCAACTCT; reverse-dCACCCCAAAAGTTCAGTCGT; GAPDH: forward-dGACAGTCAGCCGCATCTTCTT; reverse dCCAATACGACCAAATCCGTTGAC; 18S ribosomal RNA: forward-dGGACATCTAAGGGCATCACAGACC; reverse-dTGACTCA-ACACGGGAAACCTCAC. Quantitative analysis of PCR data used the 2^−ΔΔCT method.¹⁸ 

Western blots were performed as previously published.¹⁷  Equivalent protein amounts were resolved on a 10% polyacrylamide gel, and then blotted with antibodies for phosphorylated and total protein p70S6K, and S6K (Cell Signaling Technology), or for Mpl (Millipore).

Cells were stained with annexin V and propidium iodide (PI; R&D Systems). A FACSScan (BD Biosciences) was used for data acquisition. Analysis was performed using FlowJo software (TreeStar).

Total RNA was isolated from 5 × 10⁶ primary AML cells using QIAGEN RNeasy kits after 6 hours of exposure to 2.86μM TPO or 5μM SB. RNA was submitted to the Cancer Center Microarray Facility for quality check and subsequent analysis using Affymetrix GeneChip U133A, Version 2. Data analysis was performed using the GCRMA algorithm and Array Assist lite 3.4 program. The microarray data can be viewed at the Gene Expression Omnibus under accession number GSE18673.

SB and rhTpo effects on primary human leukemia cell growth were evaluated in suspension culture at doses previously shown to be optimal for megakaryocyte growth stimulation (5μM and 2.8μM, respectively).¹⁵  SB not only failed to stimulate leukemia cell proliferation, in myeloid cell samples it proved toxic (Figure 1B-F). In 24 of 26 AML, 0 of 6 ALL, and 3 of 6 CML samples, SB (5μM) exposed cells died between 3 and 6 days of culture. In this small study, no correlation between cell death and myeloid disease subtype was observed. Of note, neither rhTpo (2.8μM), nor a DMSO alone control solution had deleterious effects on cell proliferation compared with untreated controls.

A cytotoxic mechanism was sought. Induction of apoptosis was assessed by flow cytometry using annexin V and PI staining. The histogram of AML 774 is typical of a responsive patient (Figure 1H). Compared with control, annexin V staining was not apparent before 24 hours but demonstrated an approximate 22% increase at 48 hours (Figure 1H), and an approximate 50% increase at 72 hours. At all time points, PI⁺ cells increased concomitantly with the detection of annexin V, suggesting a nonapoptotic, necrotic cell death mechanism. Notably, when E's effect on 3 additional AML samples was evaluated, similar flow cytometric, but even greater antiproliferative effects, were observed suggesting a class effect. The dot plot of AML patient 710, whose cells were unaffected by exposure to SB, failed to show an increase in either PI or annexin V staining.

A relationship between cytotoxic effect and Mpl receptor expression was then investigated but not identified. In agreement with previous reports,^12,13,19  Mpl protein was detectable in various amounts on virtually all cells examined, including ALL cells (Figure 2A). However, no quantitative relationship between Mpl expressed (as measured by Western blot) and response was discerned. Further, competition experiments revealed that rhTpo could neither compete for, nor block, the cytotoxic effect of SB. These results suggest the possibility that hydrazone class Mpl agonists might kill through a non–Mpl-dependent mechanism. However, because (1) SB and E are nontoxic to normal marrow cells at antileukemic concentrations, (2) SB and rhTpo engage the Mpl receptor in physically different ways²⁰  (explaining the competition assay results), and (3) Mpl expressed on ALL cells is reportedly inactive,¹⁴  we hypothesize that SB's interaction with Mpl is integral to the cytopathic effect.

Differential engagement of Mpl by SB and rhTpo was pursued as a possible cytopathic mechanism by examining the signaling pathways activated by each in 3 SB responsive primary AML samples. Cells were exposed to SB (5μM) or rhTpo (2.86μM) in suspension culture for 1, 3, or 5 hours. By 3 hours, SB-stimulated cells showed considerably increased phosphorylation of p70S6K at Thr421/Ser424, and ribosomal kinase S6, compared with unstimulated controls, or cells stimulated with rhTpo (Figure 2B). No significant differences in p70S6K phosphorylation compared with controls were observed in nonresponding sample AML710 (supplemental Figure 1, available on the Blood website; see the Supplemental Materials link at the top of the online article). Interestingly, similar p70S6 and S6 kinase phosphorylation patterns have been demonstrated in CML cells exposed to arsenic²¹  and hepatocytes exposed to phosphatase-inhibitory toxins.²²  To further implicate the p70S6K pathway in SB's cytopathic effect, we first inhibited p70SK phosphorylation with rapamycin in AML 774 (primary cells) and MOLM 14 cells, and then exposed those cells to SB. In each case, the previously observed antiproliferative effect was still observed but was even more pronounced in primary cells. As before, rhTpo had no effect on cell growth (supplemental Figure 2). These findings suggest that SB might trigger an autophagic cell death because the mTOR pathway has been reported to negatively regulate autophagy.²³  Still, the interplay between apoptosis and autophagy is complex, and additional experimentation will be required to resolve this issue.

To help elucidate the transcriptional consequences of differential signaling initiated by SB (5μM) and rhTpo (2.86μM), expression profiling was conducted on 5 of the 24 sensitive primary AML samples after a 6-hour exposure to each molecule. This time point was chosen because by this time clear differences in proliferation and p70S6/S6 kinase phosphorylation began to emerge. We found statistically significant, more than 2-fold changes in glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and regulated in development and DNA-damage responses 1 (REDD1), confirmed by quantitative RT-PCR (Figure 2C). Because expression of both genes has been linked to cell death signaling pathways^24-27  and was not observed in SB-insensitive cells (supplemental Figure 3), we further pursued their role in SB-induced cytotoxicity using an siRNA knockdown strategy. In SB-treated MOLM 14 cells, a GAPDH knockdown had no significant effect on SB-induced cell death. A REDD1 knockdown, in contrast, enhanced SB toxicity (supplemental Figure 3). The mechanistic explanation of these observations will also require further investigation.

In conclusion, the challenge of defining the precise cell death pathway of nonpeptidyl hydrazone class Mpl agonists remains but in our view should not deter the initiation of studies using these agents in patients with AML, and perhaps kinase-resistant CML patients as well.¹⁶  Elderly AML patients who are not otherwise candidates for more conventional therapies would seem to be particularly appropriate candidates because our data suggest the distinct possibility for clinical benefit, and little potential for harm with these otherwise well-tolerated drugs.

The authors thank Drs Elizabeth Hexner and Joanna Opalinska for helpful discussions and comments during the preparation of this manuscript.

This work was supported in part with research funds provided by Glaxo SmithKline, Collegeville, PA, and the Abramson Cancer Center at the University of Pennsylvania (Tobacco Settlement Grant; A.M.G.).

Contribution: A.K. designed and performed experiments, analyzed data, and prepared a first draft of the manuscript; and A.M.G. designed and supervised the research study and prepared the final draft of the manuscript.

Conflict-of-interest disclosure: A.M.G. received research funding from GlaxoSmithKline. A.K. declares no competing financial interests.

Correspondence: Alan M. Gewirtz, Division of Hematology/Oncology, Department of Medicine, Rm 713, BRB II/III, University of Pennsylvania School of Medicine, 421 Curie Blvd, Philadelphia, PA 19104; e-mail: gewirtz@mail.med.upenn.edu.