Blocking of Evi 1 Binding
Testing a new synthetic polyamide for the ability to inhibit the function of a leukemogenic DNA binding protein.
We are interested in developing novel cures for acute myeloid leukemia (AML), a lethal neoplasm that affects thousands per year in the US. Current treatments for the disease are inadequate and despite therapy, the median five-year survival is less than 25% (http://www.leukemia-lymphoma.org/hm_lls). Advances in treatment require a better understanding of the molecular mechanisms involved in the genesis of AML and its resistance to therapy. Our goal is to define molecular targets for drug development in MDS and develop novel therapeutics to inhibit their action. The zinc finger protein EVI1 plays a critical role in AML, both through dysregulated expression associated with chromosomal translocation, and through overexpression in the absence of genomic rearrangement. The EVI1 protein binds to DNA in a sequence-specific manner, and is involved in the regulation of gene expression. Our lab has identified a number of EVI1 target genes, and their role in leukemogenesis is under study. Of critical importance are our findings 1) that suppression of EVI1 expression in leukemic cells results in apoptosis via the intrinsic (mitochondrial) pathway; and 2) that malignant transformation by EVI1 is completely dependent on the protein’s ability to bind DNA in a sequence-specific manner. Our data clearly demonstrates the critical role for EVI1 in blocking apoptosis in hematopoietic cells, and strongly suggests that by blocking the ability of EVI1 to bind DNA with a therapeutic agent, we can specifically kill AML cells.
We hypothesize that EVI1 is essential for dysplastic growth of bone marrow cells, and provides a key target for therapy in the treatment of MDS. Furthermore, we hypothesize that through the use of rationally designed polyamides, we can develop agents that specifically block EVI1 binding to DNA, and that these may be specific and effective in the treatment of AML.
this hypothesis, we wish to test the ability of DAS-1-39, an
EVI1-specific polyamide we have synthesized, to
bind to the EVI1 binding site and to block EVI1 binding both
in vitro and in vivo.
To suppress the growth of cancer cells, most chemotherapeutic drugs impair cell division. As a result, chemotherapy inevitably damages the normal cells that are also fast-dividing, leading to serious side effects on the immune system, red cell and platelet production, the gastrointestinal tract, and the skin. Leukemic therapy is further hampered by the frequent acquisition of resistance to chemotherapy by the leukemic cells, due in part to P1 glycoprotein-type multidrug resistance mechanisms, and more commonly resistance to apoptotic cell death (e.g., (20)). To overcome this problem, it is necessary to disrupt the underlying mechanism of anti-apoptosis that confers chemoresistance. Data from our lab (shown below) and others (21) indicates that one important effect of EVI1 in malignant transformation is to block apoptosis, particularly that induced by chemotherapy. Thus, rationally designed cancer therapy that targets EVI1 may overcome this chemoresistance.
One perceived problem is that EVI1 is a transcription factor, and these have been viewed as poor drug targets. However, recent experiments from our collaborator, Peter Dervan at Cal Tech, indicate that one can inhibit transcription factor binding to DNA via the use of polyamides that bind to DNA in a sequence-specific manner. Polyamides are synthetic oligomers programmed to recognize specific DNA sequences. By binding to the minor groove of DNA, these small molecules can function as allosteric inhibitors of transcription factors. Polyamides containing the aromatic amino acids N-methypyrrole (Py), N-methylimidazole (Im), and N-methyl-3-hydroxypyrrole (Hp) form the basis of Dervan’s molecular code. Pairs of these heterocyclic rings bind Watson-Crick base pairs in the minor groove in a highly specific way: Im/Py and Py/Im distinguishes G·C from C·G, Hp/Py and Py/Hp distinguishes T·A from A·T (22). Based on these pairing rules, polyamides can be designed to target desired DNA sequences. The validity of these rules have been confirmed through characterization of synthesized polyamides via DNase I footprinting, affinity cleavage, two dimensional nuclear magnetic resonance (NMR) (23), and X-ray crystallography (24). Besides its specificity, the affinity of polyamide to DNA is also remarkable. The Km of polyamide binding to specific DNA sequence is typically in the nanomolar range, comparable to that of naturally occurring DNA-binding proteins (25). As TFs typically bind in major groove of DNA and polyamides bind in minor groove of DNA, polyamides can serve as an allosteric inhibitor of TFs (22). The heterocyclic rings of polyamide can be constructed into a variety of motifs, including hairpin, cycle, -substituted, H-pin, U-pin, turn-to-turn tandem, and candy cane overlay (26). Each motif confers a unique structural advantage for the polyamide. Furthermore, the repertoire of monomers, consisting of Py and Im initially, has also been expanded, providing more options to optimize polyamide-DNA interaction. Pyrazole (Pz) and 1H-pyrrole (Nh), functionally akin to Py, project a hydrogen with positive potential toward the DNA; 5-methylthiazole (Nt), furan (Fr), akin to Im, project an sp2 lone pair from nitrogen or oxygen; 3-hydroxythiophene (Ht), akin to Hp, projects a hydroxyl group; 4-methylthiazole (Th) and 4-methylthiophene (Tn) project a large, polarizable sulfur atom (22, 27). As the technology of polyamide design matures, polyamide will become a powerful tool for both molecular biology and medicine. In addition, polyamides can be modified for optimal cellular and nuclear uptake. Thus, polyamides are ideal candidates for chemotherapeutic drugs. In collaboration with Peter Dervan’s lab we have recently synthesized a polyamide (DAS-1-39) designed to block DNA binding of EVI-1 .
Background Data to Project
Generation of missense mutants of
EVI1 that fail to bind DNA in a sequence-specific manner. Studies
by Delwel et al indicated
that zinc fingers 4-7 within zinc finger Domain 1 of EVI1 are
necessary for that domain to bind to DNA with sequence specificity
(28). From structural studies of zinc finger motifs, it is
known that the C-terminal side is comprised of an a-helix that
is held in place by an internal hydrophobic core mediated by
the interaction of the side chains of the conserved phenylalanine
residue and the leucine residue (also invariant) located in
the central portion of the a-helix . The amino
acid side chains that interact with DNA are situated on the
outer face of the a-helical region, specifically those located
at positions 2, 3, and 6 of the a-helix, but also include the
amino acid at position –1 relative to the a-helix (29).
Through in vitro and in vivo structure-function studies, we
determined that amino acids Q199 and R205 residing in zinc
finger six are critical for sequence-specific DNA binding.
This analysis allowed us to generate single amino acid missense
mutations (R205N and Q199D) that are essentially devoid of
DNA binding ability (30). Zinc finger Domain 2 (fingers 8-10),
binds with high affinity to GAAGATGAG motif (31). To create
a mutant of EVI1 that fail to bind via the second set of zinc
fingers, we chose to alter R769, since it is situated within
the a-helical segment of zinc finger 9 at a position that is
expected to make contact with DNA , and since
arginine is known to have the capacity to hydrogen bond guanine
residues (29). Wildtype and R769C mutant second zinc finger
domain protein purified from E. coli showed negligible
capacity to bind the GAAGATGAG motif by electromobility shift
assay (EMSA) .
Suppression of EVI1 in DA-1 cells using shRNA results in apoptosis via the intrinsic (mitochondrial) pathway. For EVI1 to constitute a viable target for anti-leukemia therapy, leukemic cells must be completely dependent on its continued expression for survival. To determine if this is the case, we suppressed EVI1 in DA-1 leukemic cells, which bear a proviral insertion that activates Evi1, and followed the effects of the suppression on cell growth. Two different short hairpin RNAs (sh11 and sh54) were identified that suppressed EVI1 expression in leukemic DA-1 cells. We used high titer retroviral infection followed by drug selection (puromycin) to introduce these into DA-1 cells. As assessed by quantitative PCR, western blot, and immunofluorescent analyses of DA-1 cells at 4 days post-infection, both sh11 and sh54 were able to suppress EVI1 expression, as compared to the lucferase-specific (shLuc) and scrambled sequence (shScr) shRNA controls (data not shown). Cytospin analysis of sh11- and sh54-transduced cells revealed nuclear fragmentation, suggestive of apoptosis; control shRNA-transduced cultures lacked these features (data not shown). To determine if this morphology was due to apoptosis, we performed TUNEL analysis on both sh11- and sh54-infected cells as well as control cells (infected with shScr or shLuc viruses). Cytospin smears of suspension cells were stained for free ends of DNA and enumerated, which revealed a markedly higher number of apoptotic cells in the sh11- and sh54-treated populations . To confirm these findings we also performed measurements of histone release, which provides a quantitative measurement of degree of apoptosis in the population. This too showed increased apoptosis in cells with suppressed EVI1 expression . Apoptosis can occur via either an intrinsic or an extrinsic pathway. In order to further define the exact effect of EVI1 on the leukemic cells, we sought to determine upon which of these pathways EVI1 acts. The intrinsic or mitochondrial pathway is characterized by loss of mitochondrial membrane potential, release of cytochrome C, and activation of caspases 9 and 3, while activation of the extrinsic pathway is characterized by death receptor ligand engaging and activating receptor, leading to caspases 8 and 3 activation. We examined loss of mitochondrial potential using MitoTracker dye, which is taken up by and retained by intact mitochondria, but is lost by the apoptotic nuclei. Flow cytometric analysis revealed that while control cells showed a single peak of strongly staining cells, suppression of EVI1 resulted in significant loss of mitochondrial membrane potential , consistent with activation of the intrinsic apoptotic pathway. We then performed Western blot analyses of caspases 3, 8, and 9, to determine the amount of cleaved product. This revealed strong activation of caspases 3 and 9 upon suppression of EVI1 expression . In the absence of any stimulation, a small amount of cleaved caspases 8 is present in the cell; suppression of EVI1 results in a slight increase in activated caspase 8. Treatment of control DA-1 cells with TNFa resulted in no detectable increase in cleaved caspases 8. Following shRNA-mediated suppression of EVI1, there was a demonstrable increase in sensitivity. However, this increase was not as dramatic as that seen with caspases 3 and 9 . These studies indicate that Evi1 expression acts to prevent apoptosis primarily via the intrinsic pathway.
Suppression of EVI1 slows cell growth but does not prolong the cell cycle transit time. Previous studies have shown that EVI1 plays a role in accelerating the cell cycle (18). However, these studies were performed in fibroblasts rather than in leukemic cells. To determine if suppression of EVI1 in DA-1 cells slowed cell growth, a growth curve was performed, which revealed that the doubling time was prolonged from about 17 hrs in control cells to 18-22 hrs in EVI1-suppressed cells . To assess whether one particular phase of the cell cycle was prolonged, we performed flow cytometric analysis of propidium iodide-stained cells. This revealed that both sh11- and sh54-transduced cells had a slightly higher percentage of cells in G2/M relative to control cells . Using these data we calculated the duration of each phase of the cell cycle: both sh11 and sh54 caused a prolongation of the G2/M phase (33). To confirm and extend these studies, we performed cell cycle analysis of shRNA-transduced DA-1 cells using pulse BrdU labeling and flow cytometric analysis, which yields a more accurate picture of cell cycle kinetics (34). These data revealed that the length of the cell cycle in shLuc DA-1 cells is 10.1 hours, shorter than that calculated by the hemacytometer method, indicating that the growth fraction is not 100%. They also revealed that there was no significant lengthening of the cell cycle in EVI1-suppressed cells: sh11-treated cells had a dt of 12.6 hrs and sh54-treated cells, 10.2 hrs . Thus, a change in cycling time could not fully explain the slower growth observed by cell counting, suggesting an increase in cells undergoing apoptosis.
We hypothesize that EVI1 induces leukemia through the activation of cell survival pathways, and that continued EVI1 expression is necessary for continued leukemic growth. If this hypothesis is correct, then it follows that inhibition of EVI1 binding to DNA using specific polyamides should block the growth of EVI1-driven leukemias.
Rotation Project. Can polyamide DAS-1-39 bind to the GACAAGATA motif, and can it inhibit EVI1 binding to DNA?
i. Quantitative DNase I Footprint Titration Analysis. To determine how well DAS-1-39 binds to the DNA sequence GACAAGATA. we can determine the polyamide:DNA equilibrium dissociation constant through quantitative DNase I footprint titration analysis. DNAs containing the EVI-1/DAS-1-39 recognition site and end-labeled on one strand with 32P will be used for DNAseI footprint analysis with increasing amounts of DAS-1-39; this should allow us to determine the location of polyamide binding and to calculate an equilibrium binding constant using a modified Hill equation as described by Trauger, et al (27).
ii. Gel Mobility Shift Assay. To determine whether DAS-1-39 inhibits the binding of EVI-1 to its recognition site we can perform electromobility shift assays using a radiolabeled double-stranded DNA containing the EVI1 binding site and purified EVI-1 derived from E. coli (30), which routinely binds tightly to the EVI1 binding site. We can then titrate in increasing amounts of DAS-1-39, and from this determine whether the polyamide can displace the protein. Conversely, we can add protein after the binding of polyamide, to determine if the protein can displace the polyamide. From these experiments, one can determine the relative affinities of the protein and the polyamide for the target DNA sequence.
iii. Nuclear Localization of Polyamide-Fluorescein Conjugates in Cell Culture. To determine if cells effectively take up DAS-1-39 and whether it localizes to the nucleus, one can follow the fate of DAS-1-39-flourescein conjugates that are added to cell culture. After incubation for a period of time, confocal microscopy will be used to visualize cellular localization of DAS-1-39. In a recent study, Edelson, et al assessed the nuclear localization properties of 100 polyamide-flourophore conjugates, from which guidelines were established for designing cell-permeable polyamides. DAS-1-39 was designed in accordance with these guidelines, and so we expect that it will localize to the nucleus.
iv. Inhibition of EVI1 binding in vivo. If the polyamide is able to enter the nucleus, as we suspect it will be, we will then determine if it can inhibit EVI1 action within living cells in culture. We have developed a cell line, termed 6D-AP17, that encodes a EVI1-VP16 fusion protein under control of Tet operons so that it can be activated by tetracycline removal; 6D-AP17 also contains a CAT reporter that harbors 5 copies of the EVI1 binding motif (GACAAGATA) within its promoter. The EVI1-VP16 has the binding specificity of EVI1 and the potent transcriptional activation properties of the Herpes Simplex Virus VP16 protein. We have shown that in the presence of tetracycline, there is background levels of CAT; upon removal of tetracycline, there is marked increase in CAT activity. We have shown that this is dependent on high affinity binding of EVI1 to its binding motif in the CAT promoter, since control reporters lacking this sequence are not induced by EVI1-VP16, and a mutant EVI1-VP16, with a single missense mutation in zinc finger six (R205N) is unable to activate the reporter (30). Thus, we can use this cell line to determine if DAS-1-39 is able to inhibit the binding of EVI1 to DNA. 6D-AP17 cells will be cultured in the absence or in the presence of increasing amounts of DAS-1-39. Tetracycline will then be removed, resulting in the production of EVI1-VP16. At 16 hrs post induction, we will then lyse the cells and measure CAT activity. As a control, increasing amounts of DAS-1-39 will be incubated with parental S2-6 cells transfected with a constitutively active CAT reporter; this will control for non-specific effects of DAS-1-39 on transcription. A second control will be to incubate 6D-AP17 cells with increasing amounts of the “mismatch” polyamide with similar structure to DAS-1-39, but which should not bind to the GACAAGATA motif. This should not inhibit CAT activity. We will also conduct a time course of CAT activation, with DAS-1-39 or the mismatch, to assess whether we have chosen the correct time point to assay the cells following removal of tetracycline. In all of these experiments, we will assure that cell viability is not altered by the addition of the polyamides. The CAT assay will be normalized based on protein concentration in the lysates.
Conclusion and Future Directions.
The proposed project will lay the groundwork for the development of a new drug to cure acute myeloid leukemia. Considering the limitations of today’s chemotherapeutic drugs, it is highly desirable to develop drugs with specific molecular targets so that normal cells can be spared. The molecular code developed by Dervan and his colleagues may allow us to realize this goal in the context of disease-causing DNA binding proteins. If the experiments proposed in SA2 show promise, we will modify the polyamide to obtain higher affinity derivatives. We will also test whether DAS-1-39 or better derivatives have the ability to block EVI1 transcriptional activation of endogenous target genes that we have identified (30). We will determine if they are able to induce apoptosis in leukemic cell lines that express EVI1. If these studies are promising, we can proceed with in vivo experiments in mice. With potential clinical impacts, this promising project may bring great benefits to patients with MDS.
1. Buonamici S, D L, Chi Y, Zhao R, Wang X, Brace L, et al.
EVI1 induces myelodysplastic syndrome in mice. J Clin Invest.