Chromatin Regulation and Human Cancer
Recently the ability to rapidly sequence DNA has permitted the remarkably detailed analysis of mutations in human cancer, revealing new cancer-causing genes and mechanisms. One observation was that the genes encoding chromatin regulators are frequently mutated in cancer. Among the most frequently mutated are the subunits of mSWI/SNF (BAF) complexes. These complexes derive energy from the alternative ATPases Brg or Brm, which are paired with a second ATPase, β-actin. These complexes resemble yeast SWI/SNF complexes, but have lost, gained, and shuffled subunits, likely as a result of evolutionary pressures, and seem to have adopted new mechanisms in vertebrates. We isolated and cloned the subunits of the BAF complexes several years ago. Our early studies had suggested they were tumor suppressors, but we did not know how important their role would be. To determine the precise frequency of mutation in cancer, Cigall Kadoch, then a graduate student in our laboratory (now at Harvard Medical School), conducted proteomic and biochemical analyses of BAF complexes in different cell types and defined for the first time what appears to be the full 15-subunit composition of the complex. She coupled her efforts with the bioinformatic work of Courtney Hodges and Diana Hargreaves in our lab to reveal that more than 20 percent of all human cancers bear mutations in the subunits of these complexes, thus making BAF complexes the most frequently and broadly mutated chromatin regulators in human cancer. But what was the underlying mechanism?
A clue to the mechanism of action of BAF complexes in tumor suppression came from our proteomic analysis in different cell types, which revealed that topoisomerase II (TopoII) copurified with BAF. This was informative because Hargreaves (now at the Salk Institute) and Emily Dykhuizen (now at Purdue University) found that rapid conditional deletion of the oncogenic BAF subunits led to cell cycle arrest with anaphase bridges, reflecting the inability of cells to untangle DNA at anaphase, normally the job of TopoII. Genomic studies indicated that BAF is necessary for TopoII binding over the genome and that when the gene for the Brg ATPase was deleted, 70 percent of TopoII sites were lost. Thus, our present understanding of the mechanism of tumor suppression is that BAF helps TopoII resolve tangled DNA, allowing it to segregate normally to daughter cells. When an oncogenic subunit of the BAF complex is mutated, DNA is not untangled at anaphase, leading to breaks with defective repair. Eventually these breaks lead to mutations that can drive tumor progression. Presently, we are studying the molecular mechanism underlying the ability of BAF to allow TopoIIα binding and function.
BAF Complexes Can Be Oncogenes as well as Tumor Suppressors
The genetics of the vast majority of BAF subunit–related cancers indicate that they are genetically dominant tumor suppressors. However, one rare cancer caught our attention, which appeared to be decidedly different. Synovial sarcoma is a nearly untreatable cancer of young people that in essentially all cases have the t(X;18) chromosomal translocation, which results in the fusion of 78 amino acids of the SSX protein onto the SS18 BAF subunit (Figure 1). SS18 was discovered by Cigall Kadoch (in our lab) to be a dedicated subunit of BAF complexes. Kadoch found that the oncogenic SS18-SSX fusion protein inserted itself into the BAF complex, thereby displacing the product of the normal allele. Remarkably, this mechanism of oncogenesis was reversible: when the product of the normal SS18 allele was overexpressed, it displaced the oncogenic fusion protein and led to the arrest and death of the synovial sarcoma cells. We found that Sox2 drove the proliferation of the synovial sarcoma cells and that the SS18-SSX fusion removed repressive polycomb complexes from the Sox2 gene, resulting in its activation. Kadoch is continuing this work in her laboratory at Harvard.
Development of the Chromatin Indicator and Assay (CiA) System to Study Chromatin and Gene Regulatory Events
For many years we and our colleagues used various in vitro methods that rely on nucleosomal templates to study the mechanisms involved in chromatin regulation. Using these methods, we found that chromatin regulatory complexes position nucleosomes, evict them, exchange them, or slide them to specific positions. However, the inability to assemble chromatin that faithfully resembles that in a real cell severely limited this technique. For example, nucleosomes assembled in vitro lack tissue-specific chromatin states, posttranslational modifications, methylated DNA, topological features, and most other important aspects of chromatin structure. To get around these problems and to define mechanisms involved in chromatin dynamics, I came up with the idea of letting the cell assemble the chromatin and then "pipetting" a specific chromatin regulatory activity to one allele of a single gene. The "pipette" is a small-molecule chemical inducer of proximity (CIP) with surfaces that bind one peptide tag on one side and another tag on the other side (Figure 2), such as rapamycin.
To make a mouse that can be a source of tissue-specific, topologically faithful chromatin, we inserted, by homologous recombination, arrays of DNA-binding sites upstream of the Oct4 gene. We also inserted a GFP sequence in reading frame into the Oct4 gene. Thus we had a test allele and a wild-type allele (Figure 2). The Oct4 gene is a classic example of an epigenetically suppressed locus, in that adding pluripotency transcription factors is not sufficient for immediate activation of the gene, but that gene activation requires reversal of chromatin barriers over many days. We call this mouse the CiAO mouse, since the insertion is at the Oct4 gene. By adding the CIP to cells from the mouse, one can probe chromatin in the active form in pluripotent cells as well as the repressed form in somatic cells. Adding the CIP produces a cloud of the tagged recruited protein because the on-rate for these molecules is limited by diffusion and the off-rate is seconds. Thus all possible topologies are explored when the chromatin regulatory activity is recruited and the regulator is allowed to bind to this locus in a normal manner, quite different from using a rigid fusion protein.
Using this system in embryonic stem (ES) cells, in which the Oct4 gene is in an open and transcriptionally active configuration, Nathaniel Hathaway (now at the University of North Carolina) and Oliver Bell (now at the Institute of Molecular Biotechnology, Vienna) recruited HP1, the protein critical for heterochromatin formation. They measured, for the first time, precise rates of propagation along the chromosome to about 20,000 bp, as well as rates of dissolution (Figure 3). The measurement of these two rates allowed Courtney Hodges to build a model that explains the topologic features of more than 99 percent of all heterochromatic domains (except those near the centrosome). We call this the balanced intrinsic reaction rate model (Figure 4) of chromatin boundary formation, and we are applying this to other chromatin marks. Hathaway and Bell were also able to test epigenetic memory by simply removing the CIP and following the kinetics of resumption of the active state at the Oct4 gene, which we could monitor at the single-cell level. Surprisingly fast dissolution of epigenetic repression was observed (about three cell divisions), indicating that the H3K9Me3 mark might not be as epigenetically stable as commonly presumed.
Chromatin Regulation in the Development of the Nervous System
One of the most distinctive steps in the development of the nervous system occurs at mitotic exit when neural progenitor cells lose the potential to become many types of neurons and adapt fates, morphologies, and functions that will persist for centuries (at least in the case of certain turtles). In this sense, a postmitotic neuron has perhaps the most stable epigenetic state of any cell type. Our early genetic studies indicated that this transition requires the subunits of SWI/SNF-like BAF complexes. Purification of the complexes from various stages of developing neurons led us to discover that neural stem cells and postmitotic neurons have specific BAF complexes distinguished by their composition. Further genetic studies revealed that the progenitor complex is essential for self-renewal of neural stem cells and the postmitotic complex is necessary for dendritic development and synapse formation. Andrew Yoo (now at Washington University in St. Louis) and graduate student Brett Staahl in our laboratory found that switching of subunits is controlled by a triple-negative genetic circuit in which REST represses miR9 and miR124, which in turn repress BAF53a, leading to the switching of the other subunits (Figure 5).
Remarkably, Yoo and graduate student Alfred Sun in the lab found that forcing this epigenetic switch can convert human fibroblasts to neurons and even make types of neurons that were never possible before. This instructive role of BAF complexes was surprising, but soon we gained additional insight into it from a wave of genome-sequencing studies in human neurologic diseases. These studies found that a remarkable number of different neurologic diseases emerged from mutations in the subunits of the complexes. Both npBAF and nBAF subunits were found to be mutated, pointing to critical roles for BAF complexes in both stem cells and postmitotic neurons (Figure 5). We are presently exploring the molecular mechanisms involved.
These studies were also supported by grants from the National Institutes of Health, the California Institute for Regenerative Medicine (CIRM), and the Simons Foundation Autism Research Initiative (SFARI).
As of June 16, 2015