Small molecule inhibitors of the CXXC domain, an epigenetic reader of DNA methylation

Medicinal-Pharmaceutical Chemistry

The CXXC domain functions as a reader of DNA methylation, binding to non-methylated CpGs. We will develop small molecule inhibitors and test effects on DNA methylation and growth of leukemia cells.

Research Interests
  • Drug development

We propose to establish a newresearch collaboration to assess therapeutic efficacy of small moleculeinhibitors targeting Acute Myeloid Leukemia (AML). Through combining chemical andstructural biology expertise (Bushweller), cell biology and epigeneticsexpertise (Garrett-Bakelman), and engineering approaches in biological systems(Caliari), we are poised to perform the proposed research goals.



MLL and leukemia. The Mixed Lineage Leukemia 1 gene (MLL1, also known as  KMT2A)at the 11q23 locus was first identified by its involvement in chromosometranslocations associated with acute leukemia. Patients with MLL1 rearrangements develop either AML,ALL or mixed lineage leukemia1-3.MLL leukemia accounts for around 10% of AML and ALL. Chromosomal translocationsinvolving the MLL1 gene producein-frame fusion proteins in which the N-terminal portion of the MLL1 protein isfused to the C-terminal portion of one of many protein partners. Regardless ofthe particular MLL1 fusion, MLL1 leukemia is associated with early relapse, andpatients are generally classified as poor-risk4,5.

DNA methylation in leukemia. Leukemia isclassified based on cellular phenotype, chromosomal translocations, and somaticmutations. More recently, DNA methylation has also been used to distinguishleukemia subtypes. Recent DNA methylation analyses have demonstrateddistinguishing DNA methylation changes which comprise genomic regions beyondpromoters, including other intronic and exonic CpG islands, CpG shores, andrepetitive sequences.  Using an Enhancedform of reduced representation bisulfite sequencing(ERRBS), which extends genomic coverage beyond CpG islands, the Figueroa labdetermined that MLL leukemias have extensive loss of DNA methylation whichaffects introns and distal intergenic CpG islands and shores 6.This distinguished MLL leukemia from other subtypes.

Readers of DNA methylation. Recognition of the methylation status of DNAis carried out by small protein domains capable of binding to CpG dinucleotidesand very specifically discriminating methylated versus unmethylated DNA.Methylated CpG dinucleotides are recognized by the well characterized methylbinding domains (MBD). The five members of the MBD containing family ofproteins are MBD1, MBD2, MBD3, MBD4 and MeCP2. To date, the only characterizeddomain capable of selective binding to unmethylated CpG dinucleotides is theCXXC domain. The CXXC domains from several proteins including MLL17,8, MBD1, and CGBP9 have been shown to bind DNA and recognizeunmethylated CpG dinucleotides. CXXC domains have a highly conservedspacing of eight cysteines that coordinate two zinc ions and fold into a saddle-likestructure that centers over an unmethylated CpG residue when bound to DNA10,11.CXXC domains are present in 12 different mammalian proteins including MLL1,MLL2, DNMT1, CGBP, TET1, TET3, KDM2A and KDM2B12.

Role of the MLL1 CXXC domain in MLL1 fusionprotein function. We have previously shown that MLL1 and MLL1 fusionproteins bind to specific clusters of CpG residues in the Hoxa9 locus and regulate expression of two transcripts within thislocus, Hoxa9 and miR-196b13,14.MLL1 functions to protect this specific region from DNA methylation13.We subsequently solved the solution structure of the MLL1 CXXC domain - DNAcomplex using NMR spectroscopy10. This provided structuralevidence showing how the CXXC domain distinguishes nonmethylated frommethylated CpG DNA. Based on this information point mutations were createdwhich disrupt the CXXC domain DNA binding ability. In the context of an MLL1fusion protein (MLL-AF9), these mutations resulted in increased DNA methylationof the specific Hoxa9 region,increased H3K9 methylation (a silencing histone mark), decreased expression ofthe Hoxa9-locus transcripts, andabrogated both immortalization potential and ability to cause leukemia in mice10. Thus, we established thatthe MLL1 CXXC domain interaction with DNA is a valid target for therapeuticintervention in MLL1 fusion leukemia.

Model systems to perform in vitro cellculture of AML. Hydrogels are water-swollen networks of polymers that haveemerged as promising cell culture platforms since they mimic salient elementsof native extracellular matrices (ECMs), have mechanics similar to many softtissues, and can support cell adhesion and protein sequestration 15. Hydrogels can revealfundamental phenomena regulating cell behavior in ways not possible withconventional culture substrates. In studies using hydrogels as models for drugscreening, cells grown on stiff, collagen-rich substrates show greaterresistance to chemotherapies compared to cells on softer substrates 16.In a more recent study, human leukemic cells displayed decreasedchemotherapeutic sensitivity in 3D culture compared to more conventional 2Dculture 17.These examples clearly motivate the need for 3D hydrogel cultures to bettermimic natural tissue microenvironments and more accurately assess cell responseto therapies.

The aims of thisproposal focus on development of small molecule inhibitors of the MLL1 CXXCdomain - DNA interaction and assessing their efficacy on MLL fusion patient-derivedcells encapsulated in engineered 3D hydrogels.


Preliminary Results

Assaysfor screening for inhibitors of the CXXC domain – DNA interaction. We have developed fluorescence polarization(FP) assays as a primary screening assay to identify compounds targeting theCXXC domain of MLL1 and inhibiting its interaction with DNA. Compoundsgenerated by the primary FP screen may interfere with the FP assay, i.e.produce “false-positives”. Therefore, we have two versions of the assay: DNAlabeled with fluorescein or with Texas red. Because these fluorophores havedifferent spectral properties, it is possible to identify false positivesarising from fluorescence interference effects. Becausethe fluorescence based assays cannot distinguish whether a compound binds tothe CXXC domain or DNA, it is essential to have a method to confirm to whichcomponent the inhibitors identified using the fluorescence assays actuallybind. To accomplish this, we are using NMR spectroscopy, namely 2D 15N-1HHSQC spectra of the CXXC domain in the presence and absence of compounds toidentify chemical shift perturbations.

Screening for inhibitors of the MLL CXXCdomain interaction. We have screened two fragment libraries of 1,500 and12,000 compounds using the FP assay described above, with IC50values ranging from 250 µM to >5 mM. Subsequent NMR analysis identified 30hits which show chemical shift perturbation in the HSQC spectra, confirmingthey bind to the CXXC domain. Because our previous work identified a Cysresidue that is located on the DNA binding interface on the CXXC domain 10, we have screened a Cysreactive library and identified 6 hits which inhibit DNA binding and are not activein screens of other unrelated targets.

Selectivityscreen. We have expressedand purified the CXXC domains from a total of 10 proteins to assess theselectivity of the fragments we have identified. We have shown that one of ourmost promising fragments has an IC50 for the MLL1 CXXC domain thatis at least 4-fold lower than seen for all the other CXXC domains save one, sothere is clear evidence we can achieve selectivity.

Hydrogelsas 3D cellular microenvironments. Hyaluronic acid (HA) is a naturally-derived non-sulfatedglycosaminoglycan that has been applied to a diverse set of cell culture and regenerativemedicine challenges due to its biocompatibility and its ease of chemicalmodification 18. HA hydrogels have previously been appliedsuccessfully as in vitro models ofhuman bone marrow 19. We have developed norbornene-modified HA(NorHA) hydrogels where critical cell-instructive properties such as mechanics,ligand presentation, and degradation can be independently tuned 20,21. Norbornenes react with thiols via alight-mediated thiol-ene addition reaction to form crosslinked polymericnetworks. Dithiol peptide crosslinkers can be synthesized where the amino acidsequence controls susceptibility to matrix metalloproteinase (MMP) cleavagewhile the amount of crosslinker controls mechanical properties (e.g., elasticmodulus).


Experimental Plan


Aim 1: Development of potentand specific inhibitors of the MLL CXXC domain (Bushweller lab).

            Wewill synthesize new inhibitors where we combine a fragment with a Cys reactiveinhibitor to achieve higher potency as well as prolonged duration of action. Wehave multiple fragments and Cys reactive inhibitors to choose from, so if onepair fails, we will move on to other pairings. The new compounds will besynthesized with varying length linkers between them to probe the effect oflinker length. Compounds will be assayed using the FP assays described above.


Aim 2: Demonstration of efficacy and on-target activity of CXXC domaininhibitors (Garrett-Bakelman and Caliari labs).

To assess for CXXC domain inhibitors on AMLcell phenotypes, we will perform in vitro cell culture using AML cell lineswith MLL translocations, specifically MV4-11 whch harbors a t(4;11)(q21;q23)translocation and MOLM13 which harbors a t(9;11)(p22;q23) translocation(MLL-AF4 and MLL-AF9 respectively) 22. Cells will first be exposed to optimizedcompounds from Aim 1 to determine halfmaximal inhibitor concentration (IC50) using growthinhibition as the phenotypic read out. This will be assessed by using an MTTassay and flow cytometry for cell viability and c-kit expression. Compoundswhich exhibit inhibitory effects on cell growth and/or survival will beselected for further study. Next, we will treat the cell lines with theoptimized compound concentrations and assess for functional effects on the MLLtranslocation through performing quantitiative DNA methylation profiling usingERRBS 23. We will determine if the compound demonstrateson-target effects through DNA methylation patterning analyses comparing theepigenetic patterning at MLL target genes 24 between cells exposed to thecompounds and those exposed to a vehicle control. Finally, to ensure the mostphysiological conditions are used to assess for compound functionality, we will establish a 3D in vitro culturesystem. We will encapsulate the cell lines in NorHA hydrogels engineered withstorage moduli on the order of 200-300 Pa mimicking bone marrow. We will testboth non-degradable and MMP-degradable hydrogel formulations. Since HA does notnatively support integrin-mediated cell adhesion we will also test formulationswith or without the cell-adhesive peptide RGD. The hydrogels which facilitatemaximal cell line survival will then be used to perform drug toxicitiyexperiments as described above. Cells will be harvested from the hydrogels andsubjected to growth inhibition assessment and DNA methylation profiling todetermine on target efficacy of the CXXC domain inhibitors.


We expect theproject to validate our hypothesis. The results obtained will serve as preliminarydata for a grant application (PI: Bushweller; Co-investogators: Caliari andGarrett-Bakelman). The proposed grant will evaluate the full range of biologicalmechanisms affected by the CXXC domain inhibitors, test for efficiacy of theinhibitor in primary AML patients samples using the 3D in vitro cultureapproach developed, and test for inhibitor efficiacy in mouse models of thedisease. Finally, we envision the grant including an Aim devoted to furtherchemical optimization of the CXXC domain inhibitors.







1          Stock,W. et al. Detection of MLL generearrangements in adult acute lymphoblastic leukemia. A Cancer and LeukemiaGroup B study. Leukemia 8, 1918-1922 (1994).

2          Rowley, J. D. Rearrangements involvingchromosome band 11Q23 in acute leukaemia. Seminarsin cancer biology 4, 377-385(1993).

3          Thirman, M. J. et al. Rearrangement of the MLL gene in acute lymphoblastic andacute myeloid leukemias with 11q23 chromosomal translocations. The New England journal of medicine 329, 909-914,doi:10.1056/NEJM199309233291302 (1993).

4          van der Linden, M. H., Creemers, S.& Pieters, R. Diagnosis and management of neonatal leukaemia. Seminars in fetal & neonatal medicine17, 192-195,doi:10.1016/j.siny.2012.03.003 (2012).

5          Khasawneh, M. K. & Abdel-Wahab, O.Recent discoveries in molecular characterization of acute myeloid leukemia. Current hematologic malignancy reports 9, 93-99, doi:10.1007/s11899-014-0200-y(2014).

6          Akalin, A. et al. Base-pair resolution DNA methylation sequencing revealsprofoundly divergent epigenetic landscapes in acute myeloid leukemia. PLoS Genet 8, e1002781, doi:10.1371/journal.pgen.1002781 (2012).

7          Wade, P. A. Methyl CpG-bindingproteins and transcriptional repression. Bioessays23, 1131-1137,doi:10.1002/bies.10008 (2001).

8          Birke, M. et al. The MT domain of the proto-oncoprotein MLL binds toCpG-containing DNA and discriminates against methylation. Nucleic acids research 30,958-965 (2002).

9          Lee, J. H., Voo, K. S. & Skalnik,D. G. Identification and characterization of the DNA binding domain ofCpG-binding protein. J Biol Chem 276, 44669-44676,doi:10.1074/jbc.M107179200 (2001).

10        Cierpicki, T. et al. Structure of the MLL CXXC domain-DNA complex and itsfunctional role in MLL-AF9 leukemia. NatStruct Mol Biol 17, 62-68(2010).

11        Allen, M. D. et al. Solution structure of the nonmethyl-CpG-binding CXXC domainof the leukaemia-associated MLL histone methyltransferase. Embo J (2006).

12        Long, H. K., Blackledge, N. P. &Klose, R. J. ZF-CxxC domain-containing proteins, CpG islands and the chromatinconnection. Biochemical Societytransactions 41, 727-740,doi:10.1042/BST20130028 (2013).

13        Erfurth, F. E. et al. MLL protects CpG clusters from methylation within the Hoxa9gene, maintaining transcript expression. ProcNatl Acad Sci U S A 105,7517-7522, doi:0800090105 [pii]


14        Popovic, R. et al. Regulation of mir-196b by MLL and its overexpression by MLLfusions contributes to immortalization. Blood113, 3314-3322,doi:10.1182/blood-2008-04-154310 (2009).

15        Caliari, S. R. & Burdick, J. A. Apractical guide to hydrogels for cell culture. Nature methods 13,405-414, doi:10.1038/nmeth.3839 (2016).

16        Nguyen, T. V., Sleiman, M., Moriarty,T., Herrick, W. G. & Peyton, S. R. Sorafenib resistance and JNK signalingin carcinoma during extracellular matrix stiffening. Biomaterials 35,5749-5759, doi:10.1016/j.biomaterials.2014.03.058 (2014).

17        Bruce, A. et al. Three-Dimensional Microfluidic Tri-Culture Model of theBone Marrow Microenvironment for Study of Acute Lymphoblastic Leukemia. PloS one 10, e0140506, doi:10.1371/journal.pone.0140506 (2015).

18        Highley, C. B., Prestwich, G. D. &Burdick, J. A. Recent advances in hyaluronic acid hydrogels for biomedicalapplications. Curr Opin Biotechnol 40, 35-40,doi:10.1016/j.copbio.2016.02.008 (2016).

19        Currao, M. et al. Hyaluronan based hydrogels provide an improved model tostudy megakaryocyte-matrix interactions. Experimentalcell research 346, 1-8,doi:10.1016/j.yexcr.2015.05.014 (2016).

20        Gramlich, W. M., Kim, I. L. &Burdick, J. A. Synthesis and orthogonal photopatterning of hyaluronic acidhydrogels with thiol-norbornene chemistry. Biomaterials34, 9803-9811, doi:10.1016/j.biomaterials.2013.08.089(2013).

21        Caliari, S. R., Vega, S. L., Kwon, M.,Soulas, E. M. & Burdick, J. A. Dimensionality and spreading influence MSCYAP/TAZ signaling in hydrogel environments. Biomaterials103, 314-323,doi:10.1016/j.biomaterials.2016.06.061 (2016).

22        Drexler, H. G., Quentmeier, H. &MacLeod, R. A. Malignant hematopoietic cell lines: in vitro models for thestudy of MLL gene alterations. Leukemia18, 227-232,doi:10.1038/sj.leu.2403236 (2004).

23        Garrett-Bakelman, F. E. et al. Enhanced reduced representationbisulfite sequencing for assessment of DNA methylation at base pair resolution.Journal of visualized experiments : JoVE,e52246, doi:10.3791/52246 (2015).

24        Bernt, K. M. et al. MLL-rearranged leukemia is dependent on aberrant H3K79methylation by DOT1L. Cancer cell 20, 66-78,doi:10.1016/j.ccr.2011.06.010 (2011).

Desired outcomes

The desired outcome of this research is the development of a potent selective inhibitor of the CXXC domain of MLL1 binding to DNA. This is highly likely to have therapeutic utility in the treatment of MLL fusion driven leukemia and perhaps other cancers as well. MLL fusion leukemia has a very poor prognosis, so the development of novel therapeutic approaches is clearly essential. The goal of this project is to advance the inhibitors to the stage where we have biological proof of principle of their activity in a cell culture setting and are poised to initiate studies in vivo in mice. This will make the project viable for application for NIH support to carry out the subsequent studies.