Targeting the Oligomerization Domain of ETO Interferes with RUNX 1 / ETO Oncogenic Activity in t ( 8 ; 21 )-Positive Leukemic Cells

About 12% of all de novo acute myeloid leukemias are characterized by the translocation t(8;21), which generates the oncogenic fusion protein RUNX1/ETO. RUNX1/ETO has a modular structure and contains several docking sites for heterologous proteins, including transcriptional corepressors like N-CoR, SMART, and mSIN3A. RUNX1/ETO is found in high molecular weight complexes, which are crucial for the block in myeloid differentiation observed in RUNX1/ETO–transformed cells. Essential for high molecular weight complex formation is the nervy homology region 2 (NHR2) within ETO, which serves as interacting surface for oligomerization as well as association with members of the ETO protein family. Here, we show that the expression of a fusion peptide consisting of 128 amino acids (NC128), including the entire NHR2 domain of ETO, disrupts the stability of the RUNX1/ETO high molecular weight complexes, restores transcription of RUNX1/ETO target genes, and reverts the differentiation block induced by RUNX1/ETO in myeloid cells. In the presence of NC128, RUNX1/ETO–transformed cells lose their progenitor cell characteristics, are arrested in cell cycle progression, and undergo cell death. Our results indicate that selective interference with the oligomerization domain of ETO could provide a promising strategy to inhibit the oncogenic properties of the leukemia-associated fusion protein RUNX1/ ETO. [Cancer Res 2007;67(5):2280–9]


Introduction
Chromosomal translocations are frequent events during malignant cell transformation, particularly in leukemogenesis.The translocation t (8;21), one of the most frequent chromosomal anomaly in leukemia (1), involves the RUNX1 gene (also known as AML1, CBFa2, or PEBP2aB) on chromosome 21 and the ETO gene (also known as MTG8 or RUNX1T1) on chromosome 8.The ubiquitously expressed RUNX1 gene product acts as a transcription factor and belongs to the key regulators of hematopoietic cell differentiation toward the myeloid lineage (2).The nuclear protein ETO has been shown to make multiple contacts with molecules of the transcriptional repressor machinery, including N-CoR, SMRT, mSIN3A, and HDAC1-3 (3,4).Accordingly, the Drosophila homologue of ETO, nervy, was recently shown to act as a repressor of gene expression during Drosophila sensory organ development by interacting directly with the transcription factor daughterless (5).
Within the leukemic fusion protein, the first 177 amino acids of RUNX1, including the DNA binding domain, are fused to the almost entire ETO sequence (6).Thus, RUNX1/ETO retains the ability to bind to RUNX1 promoter sequences and represses transcription of RUNX1 target genes.Among others, RUNX1/ETO directly or indirectly down-regulates the expression of genes involved in myeloid differentiation, such as C/EBPa, PU.1, interleukin-3 (IL-3), granulocyte macrophage colony-stimulating factor (GM-CSF), and M-CSF receptor, and several granulocyte-specific proteins, such as myeloperoxidase, neutrophil elastase (NE), and granzyme B (7).In addition, RUNX1/ ETO inhibits the expression of the p14 ARF tumor-suppressor gene (8).In contrast, plakoglobin and cadherins, both genes of the Wnt signaling pathway, are found to be up-regulated by RUNX1/ETO and contribute to the immature phenotype found in mouse bone marrow cells transformed by RUNX1/ETO (9).These changes in gene expression may allow the cell to accumulate further genetic alterations, ultimately leading to transformation and the onset of leukemia.
ETO contains four regions of extensive homology with nervy, all of which are retained in RUNX1/ETO.In particular, the nervy homology region 2 (NHR2) mediates oligomerization between RUNX1/ETO proteins and is involved in the interaction of RUNX1/ ETO with the ETO family members MTGR1 and ETO-2.The resolution of the three-dimensional structure of the NHR2 domain revealed an a-helical tetramer critical for the activity of RUNX1/ ETO (10).The NHR2 domain plays a crucial role for RUNX1/ETO high molecular weight complex formation, repression of transcription, and inhibition of myeloid cell differentiation.Apparently, the proper assembling of the transcriptional repressive machinery is one essential feature of high molecular weight complexes as deletion of the oligomerization domain leads to almost complete loss of transformation potential (10)(11)(12)(13)(14).
In this study, we investigated the consequences of interfering with the NHR2-mediated oligomerization of RUNX1/ETO.We show that expression of a polypeptide (NC128) targeted to the oligomerization domain of RUNX1/ETO disrupts high molecular weight complex formation, restores expression of RUNX1/ETO target genes, and reverses the block in myeloid differentiation.Moreover, NC128 acts synergistically with known inhibitors of histone deacetylase (HDAC) activity to revert the RUNX1/ETOinduced differentiation block.RUNX1/ETO-transformed cells expressing NC128 lose progenitor cell characteristics, enter cell cycle arrest, and undergo cell death.Our data propose the NHR2 oligomerization domain of ETO as a promising target structure for a molecular intervention in RUNX1/ETO-positive leukemias.NHR2 domain (amino acids 490-578 in RUNX1/ETO).A Flag epitope and a nuclear localization signal linked via a glycine-serine spacer were included at the 5 ¶ end.NC128 and derivatives thereof (N89 and N82), as well as two control polypeptides [CP; derived from the first 54 amino acids of enhanced yellow fluorescent protein (eYFP) and BCR, comprising the a-helical dimerization motif of BCR, amino acids 1-72], were expressed from retroviral or lentiviral vectors coexpressing either eYFP or enhanced green fluorescent protein (eGFP).A further control plasmid was constructed by exchanging seven leucine residues within the NHR2 domain, as described in ref. 10, leading to N89m7.As the N89m7 polypeptide was rapidly degraded in cells, the fusion construct N89m7/eGFP was used in the experiments.The composition of all constructs was confirmed by DNA sequencing.Cells, cell culture, and retroviral and lentiviral transduction.293T and HeLa cells were maintained in DMEM (Life Technologies, Karlsruhe, Germany).The cell lines Kasumi-1, SKNO-1, HEL, KG-1, U937, U937-R/E (Zn 2+ -inducible RUNX1/ETO-expressing cells kindly provided by M. Ruthardt, University of Frankfurt, Medical Clinic II, Frankfurt, Germany), HL60, Jurkat, and HUT78 were cultured in RPMI 1640 containing 10% FCS (Life Technologies).The medium for Kasumi-1 cells was supplemented with 20% FCS.For SKNO-1 cells, 10 ng/mL GM-CSF was added to the medium.Human CD34 + stem cells were cultured in Iscove's modified Dulbecco medium (IMDM) with 20% FCS and 20 ng/mL Flt-3L, 20 ng/mL GM-CSF, 20 ng/mL SCF, 20 ng/mL TPO, 20 ng/mL IL-6, 10 ng/mL IL-3, 6 units/mL EPO, 100 units/mL penicillin/streptomycin, and 2 mmol/L L-glutamine.Retroviral transduction and long-term cultivation were done as previously described in detail (15).The retroviral expression plasmid MSCV-RUNX1/ ETO-IRES-eGFP was kindly provided by M. Scherr (Hannover Medical School, Hannover, Germany).Production and concentration of lentiviral supernatants were done by standard technologies (16).
Reverse transcription-PCR and PCR of genomic DNA.Total RNA was isolated using the RNeasy Kit (Qiagen, Hilden, Germany).Reverse transcription was done with random hexamers using MMLV-RT, RNase H-(Promega, Heidelberg, Germany), as suggested by the manufacturer.Used primer pairs for PCR are listed in Supplementary Fig. S1.Preparation of genomic DNA from Kasumi-1 cells was done using a DNeasy tissue kit (Qiagen).Amplification of h-actin, eYFP, and eGFP was done on 200 ng genomic DNA using AmpliTaq DNA polymerase (Roche, Mannheim, Germany).Both reverse transcription-PCR (RT-PCR) and PCR on genomic DNA were done on a Perkin-Elmer GeneAmp PCR system 2400 (Perkin-Elmer, Foster City, CA).
Preparation of whole-cell extracts and electrophoretic mobility shift assay.293T cells were transfected with RUNX1/ETO alone or together with NC128 or control plasmids.Preparation of whole-cell extracts, labeling of the probe, and separation on PAGE was done as described previously (18).The RUNX1 binding site within the RUNX3 promoter was used as a probe (5 ¶-GCCTGGTCCCTCAACCACAGAACCACAAGGCCAGGCCCT-3 ¶).Anti-ETO (Santa Cruz Biotechnology) was used for supershifting the RUNX1/ ETO complexes.
Proliferation, cell cycle distribution, and measurement of cell death.Cell viability was assessed in triplicates by trypan blue staining.For cell cycle analysis, cells were washed with PBS and incubated for 15 min with 2 Amol/L DRAQ5 (Alexis Biochemicals, San Diego, CA) at room temperature.Nuclear incorporation of DRAQ5 was measured by fluorescence-activated cell sorting [FACS; Becton Dickinson (Heidelberg, Germany), FACSCalibur].Multiplets of cells were excluded from analysis using the doublet discrimination module.Data analysis was done with Becton Dickinson Cell Quest Pro software.The percentage of cell death was determined by trypan blue staining, 7-amino-actinomycin D (7-AAD) staining (Beckman Coulter, Krefeld, Germany), or incubation of cells with phycoerythrin-or allophycocyanin-conjugated Annexin V (BD PharMingen) followed by FACS analysis according to standard protocols.

Results
Design of constructs targeting the oligomerization domain of RUNX1/ETO.Recently, we and others have shown that amino acids 334 to 430 of ETO (corresponding to amino acids 482-578 of RUNX1/ETO), including the entire NHR2 domain, are essential for repressor activity and the ability to form high molecular weight complexes (13,14).Based on these observations, we derived a series of polypeptides (NC128 and its derivatives N89 and N82) aiming to interfere with the capacity of RUNX1/ETO to form high molecular weight complexes.The structure and amino acid sequences of the derived constructs, including CPs, are shown in Fig. 1A and B. Proper expression of the different polypeptides and fusion proteins was detected in cell lysates obtained from lentivirally transduced Kasumi-1 or transfected 293T cells (Fig. 1C).Similar to RUNX1/ETO, NC128 was localized exclusively to the nucleus of transfected HeLa cells (Supplementary Fig. S2A).Coexpression of RUNX1/ETO and NC128 results in colocalization of both proteins in the nucleus of transfected cells (Supplementary Fig. S2B).
NC128 interacts with the NHR2 domain of RUNX1/ETO but fails to bind to nuclear repressor molecules.To investigate if NC128 interacts with RUNX1/ETO, both proteins were coexpressed in 293T cells.After immunoprecipitation with an anti-Flag antibody, samples were analyzed by Western blotting for the presence of RUNX1/ETO.A strong signal corresponding to RUNX1/ETO was observed in extracts obtained from cells cotransfected with both constructs (Fig. 2A).Similarly, immunoprecipitation of RUNX1/ETO followed by hybridization with the anti-Flag antibody showed the presence of the NC128 peptide in the complex.In contrast, neither RUNX1/ETO nor NC128 signals were detected in cell extracts immunoprecipitated with an anti-HA control antibody.Also, RUNX1/ETO was absent from anti-Flag immunoprecipitates derived from cells coexpressing RUNX1/ETO and CP.To further define the minimal sequences necessary for interaction with ETO, NC128 deletion constructs were analyzed equally.N89, which lacks the last 39 amino acids of NC128, still showed binding to ETO sequences to a similar extent as NC128.Further shortening of N89 by seven amino acids (N82) almost completely abolished binding capacity to ETO (Fig. 2A).The NHR2 domain of ETO was found to be the minimal region sufficient for interaction with NC128.GAL4-ETO deletion constructs lacking the NHR2 domain (GAL4-ETOD1-236DNHR2) were unable to bind to NC128 (Fig. 2B), whereas the NHR2containing proteins ETO, GAL4-ETO, and GAL4-ETOD1-236 clearly interacted with NC128.As expected, the dimerization motif of the control construct BCR as well as the oligomerization-defective N89m7/eGFP construct failed to bind to ETO.Furthermore, a GAL4-NHR2+C construct comprising the same ETO sequences as NC128 was also found to interact with NC128, implicating possible selfdimerization of NC128 polypeptides (Fig. 2B).As the NHR2 domain of ETO is known to recruit and to enhance binding affinity of corepressor molecules and ETO family members, immunoprecipitation experiments were also done from cell lines expressing corepressors known to associate with ETO.As shown in Fig. 2C, NC128 did not interact with the nuclear corepressor molecules N-CoR, SMRT, HDAC2, mSIN3A, and MTGR1, whereas binding of NC128 to endogenous ETO and RUNX1/ETO was confirmed in ETO-expressing HEL and RUNX1/ETO-expressing Kasumi-1 cells.
NC128 destabilizes high molecular weight complexes formed by RUNX1/ETO and overcomes transcriptional deregulation of RUNX1/ETO target genes.Because NC128 interacts with the oligomerization interface of RUNX1/ETO, we analyzed the influence of NC128 on high molecular weight complex formation by RUNX1/ETO.Cellular extracts from 293T cells cotransfected with RUNX1/ETO and NC128 expression plasmids were fractionated by size-exclusion chromatography and analyzed by Western blotting as described before (13).RUNX1/ETO forms complexes of f2 mDa in size (Fig. 3A, top).In mock-transfected or CPexpressing 293T cells, complex formation by RUNX1/ETO was not affected (middle).In contrast, cell extracts obtained from NC128expressing cells contained RUNX1/ETO complexes of reduced molecular weight ranging between 2 mDa and 440 kDa (bottom, fractions 8-14).Next, we addressed the question of whether NC128 was able to release the transcriptional block induced by RUNX1/ ETO.The transcription of the cell cycle regulatory genes p21 waf/cip and p27 kip , which are up-regulated in RUNX1/ETO-expressing cells (19,20), was down-regulated in cells expressing NC128 (Fig. 3B).This effect was specific for NC128 because expression of CP did not alter p21 waf/cip or p27 kip gene expression.In contrast, expression of PU.1, a master regulator of myeloid differentiation that is downregulated by RUNX1/ETO (21), was found to be up-regulated in Kasumi-1 cells expressing NC128.Similar results were found for M-CSF receptor and NE genes highly expressed in granulocytes and monocytes.Also, the tumor-suppressor p14 ARF gene, which is specifically suppressed by RUNX1/ETO in acute myeloid leukemia blasts (8), was up-regulated upon NC128 expression (Fig. 3B).
Next, we analyzed the DNA binding capacity of RUNX1/ETO in the presence of NC128.Lysates of 293T cells transfected with RUNX1/ETO alone or in combination with NC128 or CP were subjected to electrophoretic mobility shift assay.In the presence of NC128, the capacity of RUNX1/ETO to bind to the RUNX3 DNA response element was not affected (Fig. 3C).In fact, DNA/protein complexes of similar mobility were observed in extracts obtained from cells transfected with RUNX1/ETO alone or cotransfected with NC128.The presence of RUNX1/ETO in the complex was confirmed by a slower migrating band after incubation of the complex with an ETO antibody (supershift).Additional RUNX1/ ETO-DNA complexes of faster mobility were apparent when cell lysates obtained from NC128-transduced cells were used in Figure 2. NC128 interacts with the NHR2 sequences in ETO but does not contact nuclear corepressor molecules.A, RUNX1/ ETO (RE ) and Flag-tagged NC128, N89, N82, and CP were coexpressed in 293T cells.WB, Western blot.Immunoprecipitations (IP ) were done with an anti-Flag or anti-ETO antibody.An anti-HA antibody was used as control.Samples were subjected to SDS-PAGE and transferred onto a polyvinylidene difluoride membrane and hybridized with ETO or Flag antibodies.B, NC128 binds to NHR2 sequences in ETO.NC128 and derivatives thereof were coexpressed in 293T cells together with ETO or diverse GAL4-ETO constructs.Immunoprecipitations were done as described in (A).C, NC128 was either transfected into 293T cells or transduced into Kasumi-1 and HEL cells.NC128 complexes were immunoprecipitated with an anti-Flag antibody.Thereafter, immunoprecipitated complexes were analyzed for the presence of MTGR1, mSIN3A, HDAC2, N-CoR, and SMRT, all of which are endogenously expressed in 293T cells.Similarly, interaction with endogenous ETO and RUNX1/ETO was analyzed from cellular lysates of Kasumi-1 and HEL cells.
electrophoretic mobility shift assay (Fig. 3C, lanes 7 and 8).Most probably, this reflects binding of RUNX1/ETO complexes of lower molecular weight to the target oligonucleotide.From these observations, we conclude that NC128 interferes with the transcriptional regulatory properties of RUNX1/ETO by disrupting high molecular weight complex formation but does not affect the DNA binding properties of the oncoprotein.
NC128 enforces loss of progenitor cell characteristics and restores differentiation response in leukemic cells.Kasumi-1 cells share some common properties with hematopoietic progenitor cells, including high expression levels of CD34 and c-kit and the ability to form colonies in methylcellulose (22).CD34 and c-kit expression levels were analyzed in Kasumi-1 cells transduced with lentiviral vectors expressing eGFP and either NC128 or CP (Fig. 4A; Supplementary Fig. S3A).Although high levels of CD34 expression were observed in nontransduced or CP-transduced Kasumi-1 cells, cells expressing the NC128 peptide showed a 10-fold reduction in CD34 expression levels 10 days after transduction.This effect was not due to toxic effects arising from the transduction process because nontransduced cells continued to express the CD34 antigen and expanded in culture (Fig. 4A; NC128 day 10, top left quadrant).Similarly, c-kit expression levels decreased from 90% to f40% solely in NC128-transduced cells (Supplementary Fig. S3B).Furthermore, the clonogenic capability of Kasumi-1 cells was impaired in cells expressing NC128.Whereas 78% of the colonies obtained from mock-transduced cells did express eGFP, NC128transduced cells did not generate eGFP-positive colonies in semisolid medium (Supplementary Fig. S3C).Furthermore, Kasumi-1 cells expressing NC128 showed elevated expression levels of the early monocytic lineage marker CD116/GM-CSF receptor (Fig. 4A).Within the eGFP-positive population, an 18-fold increase in CD116 expression was observed 1 week after transduction, suggesting that NC128 allows partial differentiation of Kasumi-1 cells.In contrast, no alterations in GM-CSF receptor expression levels were observed in mock-or CP-transduced Kasumi-1 cells.
A central characteristic of RUNX1/ETO-positive leukemia cells is the disability to proceed to terminal differentiation, which results in the accumulation of blasts and absence of mature granulocytes.Therefore, we asked if expression of NC128 could influence differentiation in RUNX1/ETO-transformed cells.For this, hematopoietic cell lines were transduced with lentiviral vectors expressing either NC128 and eGFP or eGFP alone.Although the treatment of wild-type or mock-transduced Kasumi-1 cells with vitamin D3 and TGF-h resulted in an 8-to 10-fold increase in the numbers of mature cells, as estimated by the expression of the myeloid differentiation marker CD11b and CD14, a 20-fold increase in the number of CD11b-and CD14-positive cells was observed in NC128expressing cells (Fig. 4B).No increase in the number of mature cells was observed in the absence of cytokine stimulation, suggesting that expression of NC128 primes cells to cytokine-induced myeloid differentiation.
To verify that the effect of NC128 was specific for RUNX1/ETO, we tested the effect of NC128 in the cell line U937-R/E.In this cell line, expression of RUNX1/ETO is under the control of the Zn 2+inducible methallothionein promoter.In the absence of Zn 2+ , mock-transduced U937-R/E cells differentiate like wild-type cells with 90% and 80% of the cells expressing CD11b and CD14 after 2 days treatment with VitD3 and TGF-h, respectively (Fig. 4C).No effect of NC128 on the differentiation properties of U937-R/E cells was observed in the absence of Zn 2+ .The transcriptional activation of RUNX1/ETO by Zn 2+ results in a block in cytokine-induced myeloid differentiation, as estimated by the reduction in the numbers of CD11b-and CD14-positive cells.However, expression of NC128 in Zn 2+ -treated U937-R/E cells restored the response to cytokines and resulted in increased numbers of CD11b-and CD14positive cells.In contrast, expression of CP did not relieve the differentiation block induced by RUNX1/ETO (Fig. 4C).Hence, NC128 effectively reverted the RUNX1/ETO-mediated block in U937-R/E cell differentiation.
NC128 synergizes with the HDAC inhibitor VPA.Because HDAC inhibitors have been shown to be potent inducers of differentiation in leukemic cells (23), we asked if NC128 could synergistically act with HDAC inhibitors in inducing differentiation of RUNX1/ETO-transformed cells.Kasumi-1 cells expressing either eGFP (mock) or NC128 and eGFP were treated with different concentrations of VPA.After 4 days treatment with VPA, 15% to 30% of the Kasumi-1 cells expressed the myeloid differentiation marker CD13 (Fig. 4D).However, up to 70% of NC128-expressing cells expressed CD13 after VPA treatment (Fig. 4D).Altogether, these results show that expression of NC128 in combination with the HDAC inhibitor VPA is effective in reverting the differentiation block induced by RUNX1/ETO.
NC128 prevents proliferation and induces cell death in the myeloid cell lines Kasumi-1 and SKNO-1.During the course of these experiments, we noticed a reduction in the numbers of eGFPpositive Kasumi-1 cells expressing NC128.Thirty days after transduction, almost no eGFP-positive cells could be detected in this population, whereas mock-transduced, BCR, and CP-expressing cells continued to proliferate (Fig. 5A).This effect of NC128 was reproduced when the NC128 deletion mutant N89 was expressed in Kasumi-1 cells.In contrast, the construct N82, which lacks seven amino acids COOH-terminal to the recently modeled NHR2 domain (10), induced a significant slower growth retardation in Kasumi-1 cells compared with NC128 or N89 (Fig. 5A).N82expressing cells could be maintained thrice longer in culture.These results revealed the NHR2 domain as the minimal functional domain necessary for the growth-inhibitory effect of NC128 in RUNX1/ETO-dependent cell lines.Moreover, exchanging all seven leucine residues of the NHR2 domain within N89 (N89m7/eGFP) abrogated the growth-inhibitory effect, clearly demonstrating that the effect of N89 is dependent on binding to the NHR2 domain of RUNX1/ETO.Stable expression of NC128 in the RUNX1/ETOdependent cell line SKNO-1 resulted in a similar net decrease of eGFP-positive cells.This effect was specific for RUNX1/ETOexpressing cells, because no influence of NC128 expression on the growth of U937 cells was observed (Fig. 5A).Similar results were obtained for NC128-expressing HL-60, Jurkat, HEL, KG-1, and Hut78 cells (data not shown).To exclude any direct influence of NC128 on eGFP expression, genomic DNA obtained from transduced Kasumi-1 cells was analyzed for the presence of eGFP sequences by PCR.Although a strong PCR signal was observed in mock-transduced cells, only a weak PCR signal was detectable in NC128-expressing cells, indicating that these cells were selectively lost during cultivation (Supplementary Fig. S3).Further proliferation assays revealed that both RUNX1/ETO-dependent cell lines were growth arrested in the presence of NC128, whereas proliferation of CP-transduced cells was not affected (Fig. 5B).These observations were confirmed by cell cycle analysis.NC128expressing Kasumi-1 cells were arrested in the G 0 -G 1 phase of the cell cycle, whereas mock-transduced or CP-expressing cells behave like wild-type cells (Fig. 5C).Annexin V staining revealed a 5-fold increase in apoptotic Kasumi-1 cells expressing NC128 compared with wild-type or CP-transduced cells (data not shown).Moreover, long-term cultivation of NC128-transduced Kasumi-1 cells resulted in accumulation of dead cells.Two weeks after transduction, staining of Kasumi-1 cells with 7-AAD revealed increased cell death selectively in NC128-expressing cells, suggesting that NC128 acts exclusively on RUNX1/ETO-transformed cells (Fig. 5D).
NC128 blocks RUNX1/ETO-induced immortalization of human CD34-positive progenitor cells.The stable expression of RUNX1/ETO in primary CD34-positive progenitor cells triggers a strong self-renewal including sustained surface expression of CD34 Figure 5. NC128 prevents proliferation and induces cell death of RUNX1/ ETO-dependent cell lines.A, Kasumi-1 and SKNO-1 cells expressing NC128 or its derivatives are selectively lost during cultivation.Transduced cells were analyzed by FACS for eGFP-expression over a period of at least 4 wks.Initial transduction rates were set to 100%.As a control, NC128-expressing U937 cells were analyzed in parallel.B, NC128 prevents proliferation of Kasumi-1 and SKNO-1 cells.One week after transduction, 1 Â 10 5 wild-type, CP-expressing, and NC128-expressing cells (eGFP >95%) were plated into 24-well tissue culture plates.Total cell numbers were counted in triplicate daily.C, NC128 induces cell cycle arrest in Kasumi-1 cells.NC128-or CP-expressing cells were stained with DRAQ5 for cell cycle analysis.Nuclear incorporation of DRAQ5 was measured by FACS analysis 14 d after transduction.D, incorporation of 7-AAD in NC128-, CP-, and mock-transduced Kasumi-1 cells.7-AAD incorporation was measured by FACS analysis 2 wks after transduction.U937 cells were transduced in parallel and serve as control.
in ex vivo cultures (15).In agreement with this, CD34 + cells transduced with a MSCV-RUNX1/ETO construct coexpressing eGFP were enriched for RUNX1/ETO-expressing cells after 6 weeks in culture, whereas mock-transduced cells stopped proliferating after 40 days (Fig. 6A).Thereafter, RUNX1/ETO-immortalized progenitor cells were transduced with a vector expressing only eYFP or NC128 and eYFP.Expression of either RUNX1/ETO or NC128 was verified by Western blotting (Fig. 6B).In the NC128transduced population, a clear decrease in the percentage of CD34 + -expressing cells was observed ( from >30% to <11%), whereas no changes in the percentage of CD34 + -expressing cells was observed in the mock-transduced population (Fig. 6C).At the same time, we found a 2-fold increase in apoptotic cells in the NC128-transduced cell population, whereas no alteration was observed in untransduced or mock-transduced cells (Fig. 6C).Over time, the number of eYFP-expressing cells decreased to almost undetectable levels, whereas mock-transduced cells persisted in culture for up to 6 weeks.The decrease in eYFP-expressing cells was confirmed by PCR on genomic DNA at week 6 (Fig. 6D).Altogether, these data show that NC128 is able to revert RUNX1/ ETO-induced immortalization of primary human progenitor cells.

Discussion
Oligomerization of leukemia-associated fusion proteins, such as PML/RARa, BCR/ABL, and RUNX1/ETO, has been shown to be an essential feature for the oncogenic potential of these translocation products (11,12,24,25).As deletion of the oligomerization interface of RUNX1/ETO or mutations designed to disrupt oligomerization lead to loss of transforming properties (10)(11)(12)14), we reasoned that targeting the oligomerization domain of ETO by expressing polypeptides derived from this region in RUNX1/ETOtransformed cells could lead to abrogation of RUNX1/ETO function.A similar approach has been used to disrupt oligomerization of the chronic myeloid leukemia-associated fusion protein BCR/ABL.In analogy to RUNX1/ETO, BCR/ABL forms high molecular weight complexes important for ABL kinase activity (25).A peptide derived from the BCR oligomerization region was shown to sensitize BCR/ABL-positive cells to the kinase inhibitor STI571 and to block BCR/ABL-driven transformation of fibroblasts (26).In our studies, NC128 was found to bind specifically to the NHR2 domain of RUNX1/ETO.N89, which still contains the complete NHR2 domain, retained binding capacity and antiproliferative effects on RUNX1/ETO cell lines.However, substitution of seven leucines, critical for NHR2-mediated RUNX1/ETO oligomerization (10), or deletion of COOH-terminal amino acids within the NHR2 domain (N82) abolished binding to ETO, suggesting that a proper folded full-length NHR2 a-helix is essential and sufficient for the observed biochemical and cellular effects of NC128.Binding did not induce degradation of RUNX1/ETO protein nor interfered with nuclear localization or DNA binding properties of the oncoprotein.In addition to its role as oligomerization interface, the NHR2 domain contributes to the recruitment of nuclear corepressors like mSIN3A and several members of the nuclear corepressor protein family (13,14).Our observations do not indicate that NC128 binds to MTGR1, mSIN3A, HDAC2, N-CoR, and SMRT, and are in agreement with similar observations made with ETO mutants in which the oligomerization domain was disrupted (10).Binding to the NHR2 domain disrupts RUNX1/ETO oligomerization, impairs high molecular weight complex formation, and thereby most likely decreases the density of recruited corepressor proteins on promotor sequences, leading to transcriptional derepression of RUNX1/ETO target genes and reversion of the differentiation block.
Among others, we found increased levels of the myeloid transcription factor PU.1 in Kasumi-1 cell expressing NC128, which most likely contributed to the observed cellular effects.Indeed, previous work has shown that overexpression of PU.1 in Kasumi-1 cells leads to proliferation arrest and induces differentiation toward the monocytic lineage (21).The effects of NC128 on myeloid differentiation and clonogenic potential of RUNX1/ETOtransformed cells noticed in our studies are reminiscent of those observed with small interfering RNA targeted specifically against RUNX1/ETO mRNA (27,28).Also, peptides designed to block the interaction between RUNX1/ETO and N-CoR showed induction of repressed RUNX1 target genes and promotion of cytokinemediated differentiation (29).In all cases, disruption of RUNX1/ ETO function led to a reversion in the expression pattern of RUNX1/ETO target genes and to a VitD3/TGF-h-dependent myelomonocytic differentiation of transformed cells.Typical features of progenitor cells, such as high expression of the cell surface markers CD34 and c-kit, were rapidly lost in the presence of NC128, implying that a functional RUNX1/ETO high molecular weight complex is essential for the maintenance of the transformed phenotype.In addition, NC128 triggered cell death in RUNX1/ETOexpressing cell lines and in RUNX1/ETO-immortalized primary bone marrow cells but not in control cells, suggesting that the effects observed for NC128 are specific for RUNX1/ETO-expressing cells.In analogy to observations made for other oncogenes (30), our results and those of others suggest that expression of the fusion protein RUNX1/ETO is permanently required to maintain the transformed phenotype and implies that targeting RUNX1/ETO could be sufficient to eliminate the leukemic clone.These results are remarkable because a series of in vitro and in vivo experiments have shown that RUNX1/ETO is able to immortalize primary murine and human bone marrow cells (15,31,32), but requires a second hit for full transformation in mouse models (33,34).Deregulated gene expression and activated forms of tyrosine kinase receptors are frequent events observed in t(8;21)-transformed cells and are considered secondary events in leukemogenesis (35,36).The Kasumi-1 cell line used in our studies contains an active form of c-kit (N822K; ref. 37), whereas the SKNO-1 cell line contains an inactive form of p53 (38).Despite these additional mutations, impairment of RUNX1/ETO function seems to be sufficient to induce cytokine-dependent differentiation and proliferation arrest of transformed cells.Also, recent studies have shown that truncated forms of RUNX1/ETO, which lack the COOH-terminal N-CoR and SMRT binding domains, but still include the NHR2 sequences, have significant transforming potential (19,39).These truncated forms are present in Kasumi-1 and SKNO-1 cells and might contribute to their transformed status.Thus, the effect of NC128 and derivatives thereof on cell proliferation and differentiation does not depend on partners of RUNX1/ETO and strengthens the rationale for interference with the NHR2 oligomerization domain to prevent RUNX1/ETO function.The fact that suppression of RUNX1/ETO activity is sufficient to induce transcription of down-regulated genes and to restore cytokine-dependent myeloid differentiation suggests that RUNX1/ ETO target genes are not irreversibly epigenetically silenced.Indeed, studies on the RUNX1/ETO target gene c-FMS have shown that despite changes in the histone modification pattern and increased association of HDAC, the RUNX1/ETO complex does not alter binding of other transcription factors to RUNX1/ETO target genes (40).
Modifiers of chromatin, in particular HDAC inhibitors, have been used in vitro to treat leukemic cells, and several compounds are currently in phase I trials (41).For example, VPA, a potent HDAC inhibitor, was shown to partially relieve the differentiation block induced by RUNX1/ETO in hematopoietic progenitor cells (23).Besides its inhibitory effect on histone deacetylation, VPA also triggers RUNX1/ETO degradation via the ubiquitin pathway. 3We found that NC128 augments the effect of VPA in priming cells for differentiation, suggesting that the combined effect of both strategies may result in an effective reversion of the transformed phenotype.Further, a combination treatment might allow a reduction in the dosage of HDAC inhibitors and thus a reduction of toxic side effects.
We propose the oligomerization domain of RUNX1/ETO as a valid target for a molecular intervention in t(8;21) leukemias.Within the translocation protein, the oligomerization domain plays a central role for the oncogenic properties of the fusion protein.
As targeted disruption of RUNX1/ETO oligomerization leads to restoration of differentiation, growth inhibition, and increased cell death in t(8;21) leukemic cells, screening for small peptides or small molecular weight compounds, designed to block NHR2mediated oligomerization, could provide promising tools to improve the therapeutic outcome of acute myeloid leukemias carrying the t(8;21) translocation.

Figure 1 .
Figure 1.Design of the constructs used in this study.A, schematic diagram of NC128 and derivatives thereof.RHD, runt homology domain; NLS, nuclear localization signal; GS, glycine serine.B, amino acid sequences of the NHR2 constructs shown in (A).Numbers above the amino acid sequence, the corresponding amino acid position within the RUNX1/ETO fusion protein.ETO coding sequences (underlined ).Thick letters, amino acids within the core a-helical structure of NHR2.C, expression of NC128, N89, N82, BCR, CP, N89/eGFP, and N89m7/eGFP in transduced Kasumi-1 cells or transfected 293T cells was confirmed by Western blotting using an anti-Flag antibody.EGFP and PLCg were used as loading controls.

Figure 3 .
Figure 3. NC128 destabilizes high molecular weight complexes formed by RUNX1/ETO and reverts transcriptional deregulation induced by RUNX1/ETO.A, cellular extracts from 293T cells cotransfected with RUNX1/ETO and expression plasmids for NC128 or CP were fractionated by size-exclusion chromatography.Individual fractions were analyzed by Western blotting using anti-ETO antibody.The fraction number and the molecular mass of standard protein markers and their peak elution fraction are shown (top ).B, Kasumi-1 cells were transduced with lentiviral vectors expressing NC128 or CP.Total RNA was obtained 5 d after transduction, and the transcription pattern of the cell cycle regulatory genes p21 waf/cip and p27 kip , the tumor-suppressor gene p14 ARF , and genes important for myeloid differentiation (PU.1, MSCFR , and NE ) were analyzed by semiquantitative RT-PCR.h-Actin transcripts were used as an internal control.C, the DNA binding activity of RUNX1/ETO is not disturbed in the presence of NC128.Cellular extracts obtained from 293T cells transfected with RUNX1/ETO either alone or in combination with NC128 or BCR were incubated with a 32 P-labeled oligonucleotide containing the RUNX1/ETO binding sequence of the RUNX3 promoter.The amount of plasmid DNA used in the cotransfection experiments is indicated (table).*, faster migrating RUNX1/ETO complexes of lysates containing NC128 (lanes 7 and 8).Supershift (s.shift ) denotes complexes of higher molecular mass formed in the presence of the ETO antibody.

Figure 4 .
Figure 4. NC128 relieves differentiation block induced by RUNX1/ETO.A, Kasumi-1 cells expressing either NC128 or CP were analyzed by FACS for eGFP, CD34, and GM-CSF receptor (GM-CSF-R ) expression at different time points after transduction.Percentages of double-positive cells are shown.B, wild-type, mock-, and NC128-transduced Kasumi-1 cells were cultured in the presence of vitamin D3 (1 Amol/L) and TGF-h (5 nmol/L) to induce myeloid differentiation.Expression of the myeloid cell surface antigens CD11b and CD14 was measured by FACS 4 d after cytokine stimulation.Values of CD11b and CD14 represent percentage from eGFP-positive cells.C, NC128 effects are specific for RUNX1/ETO-expressing cells.Wild-type or transduced U937-R/E cells were analyzed for myeloid differentiation before and after treatment with Zn 2+ (120 nmol/L) to induce RUNX1/ETO expression.Coexpression of eGFP and CD11b was detected by FACS 48 h after induction of differentiation with 1 Amol/L vitamin D3 and 5 nmol/L TGF-h (D3 ).Percentage of wild-type (WT ), mock-, CP-, or NC128-transduced U937-R/E cells expressing the granulocytic differentiation markers CD14 or CD11b: Columns, mean of three independent experiments.D, NC128 synergizes with the HDAC inhibitor VPA in inducing myeloid differentiation of RUNX1/ETO-transformed cells.Mock-or NC128-transduced Kasumi-1 cells were treated with 0.5 or 1.0 mmol/L VPA.Four days after treatment, cells were analyzed for coexpression of eGFP and CD13 by FACS.

Figure 6 .
Figure 6.NC128 blocks RUNX1/ ETO-induced immortalization of human CD34-positive progenitor cells.A, CD34 + cells were transduced with a murine stem cell virus (MSCV)-based retroviral vector coexpressing RUNX1/ETO and eGFP.EGFP expression was measured by FACS at the indicated times.B, RUNX1/ ETO-immortalized cells were transduced with a retroviral vector expressing NC128 and eYFP.One week after transduction, protein levels of RUNX1/ETO and NC128 were verified by Western blotting with a-ETO and a-Flag antibodies.hBM, human bone marrow.C, percentage of CD34 and AnnexinV positive cells 6 d after transduction of RUNX1/ETO-immortalized primary human CD34 + cells with NC128.D, time course of the percentage of eYFP-positive cells from mock-and NC128-transduced RUNX1/ ETO-immortalized human progenitor cells.Initial transduction rates were set at 100%.PCR on genomic DNA for eYFP at week 6 verifies loss of NC128-transduced cells (bottom ).