Using disease-associated mutations to understand the biochemical regulation of a multi-subunit histone methyltransferase complex
Eukaryotic DNA is compacted into chromatin, which must be continually remodeled to allow for DNA dependent processes such as transcription. The basic repeating unit of chromatin, the nucleosome, is composed of an octomer of histone proteins around which 147 base pairs of DNA is wrapped. One way chromatin remodeling is achieved is by posttransitional modification of histones.
In the Cosgrove lab, we are interested in the molecular mechanisms that regulate histone H3 lysine 4 (H3K4) methylation, which is a histone modification required for the inheritance of transcriptionally active states of chromatin. Lysine residues can be mono-, di-, or tri-methylated with each modification leading to distinct functional outcomes. However, the mechanisms that regulate the degree of H3K4 methylation are poorly understood. In humans, the SET1 family of histone methyltransferases are the major implementers of H3K4 methylation. The SET1 family is composed of the mixed-lineage leukemia proteins (MLLs) 1-4 and Setd1a/b that all share the conserved Su(var)E(z)Trx (SET) domain. Mutations in the genes encoding the SET1 family are associated with leukemias, solid tumors, and developmental disorders. All SET1 family members interact with a conserved sub-complex of proteins that include WDR5, RbBP5, ASH2L, and DPY-30 (WRAD). The interaction of a SET1 member with WRAD is known as the core complex and it is required for multiple methylation of H3K4.
We focus on the most well studied human SET1 family member, MLL1, which has served as a paradigm for understanding how the activity of the SET1 family is regulated. We previously found that MLL1 catalyzes H3K4 mono-methylation in the absence of WRAD. However, the conversion to di-methylation is catalyzed by the MLL1 core complex and requires an unknown surface from MLL1 that is distinct from the canonical MLL1 active site. Due to the lack of a high-resolution structure of the MLL1 core complex, the surfaces involved in the formation of the MLL1 core complex and the di-methyltransferase active site remain unknown. Recent genome sequencing studies have identified a number of disease-associated missense mutations that localize to the SET domains of several SET1 family members. Stephen Shinsky, a graduate student in the Cosgrove lab, modeled five MLL-2 missense mutations associated with human Kabuki Syndrome (KS) in the MLL1 core complex. He found that a subset of KS mutations map to a common solvent exposed surface in the MLL1 SET domain (Figure 1). Although all the residues mutated in KS are oriented away from the canonical MLL1 active site, Stephen found that all KS variants displayed loss of H3K4 di-methylation when assembled into the core complex. Stephen discovered that the loss of di-methylation activity is associated with a weakened ability of KS variants to interact with WRAD or the RbBP5-ASH2L heterodimer, which is required for formation of the di-methyltransferase active site (Figure 1). These results suggest that amino acids from this surface of MLL1, which we term the Kabuki interaction surface (KIS), are required for formation of the core complex and di-methyltransferase active site within SET1 family core complexes (Figure 1). Stephen's work was recently published in the Journal of Molecular Biology.
Figure 1. KS missense mutations define a Kabuki interaction surface (KIS) that is crucial for the interaction with the RbBP5/ASH2L heterodimer and for H3K4 di-methylation. (Upper Panel) The two-active site model for multiple lysine methylation by Set1 family complexes. The MLL1 core complex mono-methylates H3K4 at the canonical SET domain active site (indicated by the number 1), which is followed by H3K4 dimethylation at a second active site formed at the interface between MLL and WRAD (indicated by number 2). KIS localized KS mutations in the MLL SET domain weaken the interaction with RbBP5 and ASH2L resulting in the loss of H3K4 di-methylation activity (lower left panel). Alternatively, substitution of R3765 with leucine in the WDR5 interaction (Win) motif destabilizes the interaction between MLL and WRAD thus also preventing formation of the di-methyltransferase active site (lower right panel).
References: Shinsky SA, Hu M, Vought VE, Ng SB, Bamshad MJ, Shendure J and Cosgrove MS. A non-active site SET domain surface crucial for the interaction of MLL1 and the RbBP5-ASH2L heterodimer within MLL family core complexes. Journal of Molecular Biology (2014) 426: 2283-2299.
Graduate student Dimitra Aggeli, working in the Amberg Lab, has discovered how to turn the small actin binding protein cofilin into an actin filament destroying machine. Building off the previous work by graduate student Mike Clark, Dimitra has been trying to understand why some of Mike's mutants are hyperactive for actin filament disassembly. Turns out these mutants disrupt one of the two actin binding sites (the secondary filament specific binding site) on cofilin (see Figure 1) thus allowing Dimitra to isolate and study the effects of the primary binding site on actin filament stability.
Dimitra has found that normal cofilin actually promotes actin polymerization and this activity is completely lost by the mutants defective in the secondary binding site. In fact, the mutants rapidly disassemble pre-formed actin filaments. To understand the mechanism of filament destabilization, Dimitra used fluorescently labeled actin filaments to watch the process of filament destruction by the mutants under a microscope. What she found is that the mutants sever (break in the middle) the actin filaments (see the movie below). This is an activity of cofilin that is activated by the protein Aip1p that was discovered by Dr Amberg during his post-doc years. Dimitra's observations not only explain how cofilin severs, an answer long sought by many labs, but they also explain how cofilin severing is regulated and that is by disrupting the secondary binding site. Dimitra is currently writing a paper on her observations. This paper will include data from a previous graduate student of Stephan Wilkens', Erik Kish-Trier, who during a brief stint in the Amberg lab, solved the crystal structure of the three cofilin mutants; one at 1.90Å, one at 1.45Å, and the third at a stunning 1.10Å. Not bad for Erik's first foray into crystallography!
Understanding ε-mediated inhibition of bacterial F-type ATP synthase to develop drugs against Myobacterium tuberculosis in the Duncan Lab
Mycobacterium tuberculosis (MTB) is an infectious pathogen that causes Pulmonary Tuberculosis and kills over one million people every year. It is also a major cause of death in HIV patients. Evolution of extreme drug-resistent strains in MTB poses a serious problem towards its treatment. As a result, there is an ongoing need to develop novel drugs that can effectively fight the resistant strains. The F-type ATP synthase is responsible for energy production in living organisms and was recently identified as a drug target for treatment of MTB. It is a unique rotary motor enzyme that can synthesize as well as hydrolyze ATP depending on its direction of rotation.
The ATP synthase is often called F0F1. It has a membrane embedded F0 complex that transports protons by rotation of its c subunits and an external catalytic F1 complex that is composed of 5 subunits with the stoichiometry of α3β3γδε. The ε subunit is involved in inhibition of many bacterial ATP synthases wherein its C-terminal domain inhibits both ATP synthesis and hydrolysis whereas the N-terminal domain is necessary for functional assembly of the enzyme. Work in the Duncan lab is focused on understanding of ε mediated inhibition of bacterial ATP synthase. Recently, the Duncan lab determined the first high resolution crystal structure of F1 in Escherichia coli (EF1) in an autoinhibited conformation. The structure showed ε's CTD in a highly extended conformation (εx) that engaged in interaction with rotor and stator subunits inside the central cavity of EF1. These interactions of ε with other subunits are responsible for inhibition of the eyzyme.
Naman Shah, a graduate student in the Duncan lab, is studying these interactions of ε within F1 with the help of an optical assay that measures binding and dissociation kinetics of F1/ ε. The binding and dissociation of wild type and different mutants of ε are correlated with their inhibitory effects on the enzyme. A paper describing his results was recently published in the Journal of Biological Chemistry. This work has provided new insights on ε's confirmation when treated with different catalytic-site ligands as well as at different rotation angles of the enzyme. Understanding ε's interaction within F0F1 could lead to development of drugs that specifically target F0F1 in MTB as well as many other pathogenic bacteria. Drugs that can enhance the ε mediated inhibition of F0F1 are expected to kill the pathogen by halting energy metabolism. These drugs should be highly specific towards the pathogenic bacteria as ε's homolog does not have this regulatory role in mitochondrial F0F1.
References: Shah NB, Hutcheon ML, Haarer BK, Duncan TM. (2013) F1-ATPase of Escherichia coli: The ε -inhibited State Forms after ATP Hydrolysis, is Distinct from the ADP-Inhibited State, and Responds Dynamically to Catalytic-Site Ligands. Journal of Biological Chemistry (epub version)
Cingolani G and Duncan TM (2011) Structure of the ATP synthase catalytic complex (F1) from Escherichia coli in an autoinhibited conformation. Nature Structure Molecular Biology 18, 701-707.
The vacuolar ATPase (V-ATPase) is a rotary molecular motor enzyme that functions to acidify the lumen of subcellular organelles in all eukaryotic cells. V-ATPase function is involved in a number of fundamental cellular processes including pH/ion homeostasis, endocytosis, vesicular traffic and antigen processing. In the Wilkens Lab, we are studying the structure of the enzyme to gain a more detailed understanding of its mechanism and its unique mode of regulation.
V-ATPase regulation involves a major structural rearrangement wherein the soluble and membrane sectors dissociate from one another (Fig 1). While dissociated, the activity of both V1 (ATPase) and Vo (proton transport) are silenced and a single enzyme subunit (C) is released into the cytosol. The structural basis for this unique mode of regulation is being studied by graduate student, Rebecca Oot, whose work is focused upon subunits that play a role in linking the V1 and Vo sectors during normal enzyme function. Interactions between these subunits, C, H, E, G and aNT, (in color, Fig 1) function to resist the torque generated during rotary catalysis. Interestingly, these interactions are required to be strong enough to maintain the structural integrity of the enzyme during rotary catalysis but must be vulnerable to breakage for regulated enzyme disassembly to occur.
Previously, Rebecca found that subunit EG heterodimer binds subunit C (via Chead) with high affinity (delta G ~-40kJ/mol), an interaction that greatly stabilizes EG in vitro. Recently, Rebecca used purified recombinant subunits and subunit domains for quantitative, biophysical characterization of the interactions between subunit C, EG heterodimer and a soluble subdomain of a membrane embedded subunit, a (aNT). Here, Rebecca found that subunit C (Cfoot domain) forms a ternary binding interface formed from low affinity interactions with another copy of the EG heterodimer and aNT.
We speculate that this interface, composed of Cfoot-EG-aNT, results in a high avidity interaction that may be tunable by environmental signals, such as pH. In the assembled enzyme, this interaction would be equivalent to the product of the individual affinities (low nanomolar Kd). However, breaking of even one of the interactions (EG-aNT or EG-Cfoot) would compromise the integrity of the linkage to the membrane and may facilitate enzyme dissociation. It has been shown that enzyme function is required for regulation to occur and we hypothesize that upon the weakening of this interaction, continued enzyme rotation would serve to physically rip apart the high affinity EG-Chead interaction, leading to loss of subunit C from the enzyme. Importantly, the delta G of association of the high affinity interaction is similar to that of ATP hydrolysis, suggesting that the energetic cost of breaking the interaction could be balanced by hydrolysis of one ATP.
References: Oot RA and Wilkens S. (2012) Subunit Interactions at the V1-Vo Interface in Yeast Vacuolar ATPase. Journal of BIological Chemistry 287: 13396-13406.
Oot RA and Wilkens S. (2010) Domain Characterization and Interaction of the Yeast Vacuolar ATPase Subunit C with the Peripheral Stator Stalk Subunits E and G. Journal of Biological Chemistry 285: 24654-24664.
Cell identity in multi-cellular organisms is determined in part by factors that regulate gene accessibility within the context of eukaryotic chromatin. Post-translational modifications of histone proteins are central in the establishment of heritable gene expression programs through the regulation of chromatin structure. While a large number of distinct post-translational modifications have been identified in recent years, the molecular mechanisms by which these modifications regulate gene accessibility are not well understood.
We are focusing on identifying the molecular mechanisms for the regulation of histone H3 lysine 4 (H3K4) methylation - an epigenetic mark correlated with transcriptional activation. H3K4 methylation is critical for a number of important biological processes including stem cell differentiation, metazoan development, transcription, and the pathogenesis of cancer. H3K4 can be mono-, di-, or trimethylated, often with distinct functional outcomes. The molecular mechanisms that control the degree of H3K4 methylation are not well understood.
In our laboratory we use molecular, structural and biophysical approaches to investigate the molecular mechanisms of H3K4 methylation by the human Mixed Lineage Leukemia protein-1 (MLL1) core complex. The human MLL1 gene resides at chromosome 11 band q23 and is frequently translocated in aggressive infant and adult acute leukemias. MLL1's H3K4 methyltransferase activity regulates the expression of HOX genes during hematopoiesis and development. MLL1 contains an evolutionarily conserved SET domain that is required for its catalytic activity. MLL1's catalytic activity is regulated through interaction with a conserved core complex of proteins that include WDR5 (WD-repeat protein-5), RbBP5 (retinoblastoma binding protein-5), Ash2L (Absent, small, homeotic dics-2-like) and DPY-30 (Dumpy-30). Together, these proteins form the MLL1 core complex, which is required for the regulation of the degree of H3K4 methylation in eukaryotes. We have recently discovered that the WDR5-RbBP5-Ash2L sub-complex possesses a SET domain-independent methyltransferase activity that catalyzes H3K4 dimethylation within MLL1 core complex. Our results suggest that eukaryotes have evolved a highly intricate system for regulating the degree of H3K4 methylation.
Our long-term goal is to understand the protein-structural features responsible to the enzymatic activity and regulation of H3K4 methylation by the human MLL1 core complex. This information is essential for a broader understanding of the role of H3K4 methylation in the pathways that regulate gene expression in the context of eukaryotic chromatin. In addition, the knowledge gained from work may lead to better diagnostics and the rational design of anti-cancer drugs for the treatment of certain forms of leukemia and other malignancies.
Projects in the lab are available to investigate the structure, enzymology and cell biology of MLL family complexes. Interested Ph.D. applicants should apply at the departmental website. Postdoctoral applicants should contact me directly at: firstname.lastname@example.org
References: Patel, A., Vough, V., Dharmarajan, V., and Cosgrove, M.S. (2011) A novel non-SET domain multi-subunit methyltransferase required for sequential nucleosomal histone H3 methylation by the MLL1 core complex. Journal of Biological Chemistry 286, 3359-3369.
Patel, A., Dharmarajan, V., Vough, V.E., and Cosgrove, M.S. (2009) On the mechanism of multiple lysine methylation by the human Mixed Lineage Leukemia protein-1 (MLL1) core complex. Journal of Biological Chemistry 284, 24242-24256. (Selected as JBC's Paper of the Week).
Research Highlight: An Evolutionarily conserved DNA recombination/repair protein identified in mitochondria
Mitochondrial DNA (mtDNA), encoding integral components of the energy-producing oxidative phosphorylation pathway on the mitochondrial inner membrane, is extremely vulnerable to damages by oxidative, chemical, irradiational and metabolic stresses. However, although mtDNA has been discovered almost 50 years ago, how damaged mtDNA is repaired is poorly understood. Homologous recombination (HR) is a universal molecular process conserved from batcheriophage to human. It has primarily evolved as a mechanism for the repair of double stranded DNA breaks and for the re-initiation of DNA synthesis from collapsed replication forks.
Can the universality of HR be extended to mitochondria? A crucial step to address this question is to demonstrate the presence of DNA recombination machineries in mitochondria. We discovered the yeast MGM101 gene that is essential for mtDNA maintenance in 1993. Other investigators subsequently demonstrated that Mgm101 is a mt-nucleoid protein required for mtDNA repair. Although Mgm101 has also been shown to share some sequence similarities with the recombination protein, Rad52, the biochemical and functional study has been held back in the last decade by the difficulty to produce the recombinant protein. Three years ago, MacMillan Mbantenkhu, a graduate student joined our lab and attempted to express Mgm101 in E. coli. By using the MBP-fusion strategy, Mac was able to produce Mgm101 in large quantities in a soluble form. This important accomplishment permitted him to show that Mgm101 tends to oligomerize in vitro. Xiaowen Wang, a research scientist in the group, then found that like Rad52, Mgm101 preferentially binds to ssDNA. More importantly, she also found that Mgm101 catalyzes single strand DNA annealing even in the presense of the mitochondrial single strand binding protein, Rim1. In collaboration with Stephen Wilkens, we showed that Mgm101 forms distinct oligomeric rings (see Figure, panels A & B) with a diameter of ~200 Å, and highly compressed helical filaments with a pitch of ~50 Å (panels C & D). In the presense of short (panel E) and long (panel F) ssDNA substrates, Mgm101 forms thin and condensed nucleoprotein filaments respectively. When analyzed by sedimentation velocity analytical ultracentrifugation, our collaboratrs, Michael Cosgrove and Anamika Patel, established that each Mgm101 ring contains ~14 subunits. These biochemical and structural properties remarkably resemble those of the Rad52-type recombination proteins such as Redβ, Erf, RecT and Sak from bacteriophages. Together with Jonathan Nardozzi, a post-doctoral fellow in our lab, we have been able to show that several mutations compromising Mgm101 function in vivo also affected Mgm101 oligomerization in vitro. This led to our hypothesis that the Mgm101 rings may act as stores in the mitochondrial nucleoids, which stabilizes the protein by preventing the aggregation of the otherwise highly unstable Mgm101 protomers. Other contributors to this work include Elizabeth Hoffman, a SURF student who helped to show that defect in Mgm101 function affects mtDNA recombination in vivo.
In summary, our finding strongly supports the existence of a recombination system of bacteriophage-origin in mitochondria. This finding could have important implications for better understanding the mechanisms of repair and probably also replication of mtDNA. This piece of work was recently published online in the Journal of Biological Chemistry.
Reference: Mbantenkhu M*, Wang X*, Nardozzi JD, Wilkens S, Hoffman E, Patel A, Costrove MS and Chen XJ. Mgm101 is a Rad52-related protein required for mitochondrial DNA recombination. J Biol Chem 2011; Oct 25 [Epub ahead of print]
The Loh lab designs protein-based switches as platforms for biosensing. Proteins recognize many interesting ligands, but most proteins don't change their structure upon binding. This lack of conformational change makes it difficult to detect the binding event. Therefore, a major challenge is to develop general mechanisms by which ordinary binding proteins can be converted to molecular switches. Meg Stratton, a PhD student in the Loh lab, created a flourescent calcium sensor (calbindin-AFF) based on the calcium-binding protein calbindin D9k. The method she helped develop, called 'alternate frame folding', employs partial amino acid sequence duplication to create two native states of calbindin-AFF (N and N'; top figure). One of the states cannot bind calcium and the other can. Calcium binding drives the conformational change from the former state to the latter (bottom figure). In collaboration with David Eliezer's group at Cornell Weill Medical College, Meg used nuclear magnetic resonance spectroscopy to characterize the solution structures of the calbindin-AFF switch in each conformation. She showed that the duplicated sequences undergo matched order-disorder transitions, so that one segment folds as its twin unfolds (and vice versa), as calbindin-AFF switches from one state to the other. This finding is signficant because the coupled folding-unfolding event is a large, physical change that forms the basis for flourscent and other optical detection in calbindin-AFF and other similarly-engineered protein sensors.
References: Stratton, M.M., McClendon, S., Eliezer, D. and Loh, S.N. (2011). Structural characterization of two alternate conformations in a calbindin D9k-based molecular switch. Biochemistry 50, 5583-5589
Stratton, M.M. & Loh, S.N. (2011). Converting a protein into a switch for biosensing and functional regulation. Protein Sci. 20, 19-29
Stratton, M.M. & Loh, S.N. (2010. On the mechanism of protein fold-switching by a molecular sensor. Proteins: Struct. Funct. Bioinf. 78, 3260-3269
Several years ago, a graduate student in the Hanes laboratory, Wencheng Zhu, was studying embryonic development in Drosophila melanogaster (fruit fly) and discovered a novel methyltransferase called Bin3 (Bicoid-interacting protein 3; Zhu & Hanes, Gene, 2000). The function of this protein remained a mystery for a long time. Then, a group in Montreal found a human homolog of Bin3 and showed it methylates a small non-coding RNA called 7SK (Jeronimo et al., Mol. Cell, 2007). Now, a postdoc in the Hanes laboratory, Vimi Singh, discovered the biological role of Bin3 (Singh et al., Dev. Biol., 2011). It turns out to play a critical role in translation regulation during embryo development. Without its function, normal head development, at least in the fly, cannot occur.
Specifically, Bin3 appears to methylate 7SK RNA during the process of oogenesis and early blastoderm development. Methylation helps stabilize the 7SK RNA, which in turn acts like a scaffold to allow formation of a protein complex that shuts off translation of caudal mRNA. Bin3 and Bicoid are key members of this complex and work together to ensure that Caudal protein is not expressed in the anterior of the embryo. This allows head and thoracic development to occur (see Figure). In bin3 mutants, Caudal protein is expressed in the head and development is defective leading to embryonic death. Whether human Bin3 plays an analogous role in human embryo development is not yet known, but it would not be surprising if it did.
Reference: Singh, N., Morlock, H. and Hanes, S. D. (2011). The Bin3 RNA methyltransferase interacts with Bicoid and is required for caudal repression in the Drosophila embryo. Dev. Biol. 352:104-115