My laboratory uses single-molecule optical microscopy to study fundamental interactions between proteins and nucleic acids—we literally watch individual proteins as they interact with their DNA substrates. Our overall goal is to reveal the molecular mechanisms cells use to repair, maintain, and decode their genetic information. This research combines aspects of biochemistry, physics, and nanoscale technology to answer questions about complex biological problems that are difficult to address through traditional biochemistry. As part of our work, we are developing new experimental platforms that enable "high-throughput" single-molecule imaging.
The primary advantages of our approaches are that we can actually see thousands of individual molecules. We can see what proteins are bound to DNA, where they are bound, how they move, and how they influence other components of the system—all in real time, at the level of a single reaction. We are beginning to use these new research tools to directly visualize the molecular basis of DNA repair and chromatin structure, both of which are relevant to diseases such as cancer.
Technology Development and Curtains of DNA
With a background in biological research, I became intrigued by of the power of single-molecule imaging, but I was also aware of its significant limitations. There are two particular problems that plague single-molecule research and prevent it from being applied to many complex biological problems. The first problem is nonspecific surface adsorption of biological molecules to the inherently artificial environment of a microscope slide surface. The second problem is statistics. Single-molecule techniques tend to be technically demanding, and it is difficult to obtain enough data to reliably interpret any experimental observations.
I knew that lipid bilayers offered the potential for making an inert surface comparable to the microenvironment that biological molecules such as proteins and DNA would normally experience inside of a living cell. At the same time it was clear that the lipid bilayer would cause a new problem: the bilayers are fluid, and DNA anchored to the bilayer would be free to move in two dimensions. As a consequence, we would only get a fleeting glimpse of the DNA as it moved through the field of view. I also realized, however, that we could potentially align all of the anchored DNA molecules along the edges of mechanical barriers that would disrupt the continuity of the bilayer.
We began experimenting with microscale barriers to lipid diffusion that we made by manually etching a fused silica slide glass. From the outset we were able to align the DNA molecules into molecular curtains, which allowed us to image on the order of 100 molecules in a single field of view. We have since adopted nanolithography to make patterns of barriers that can precisely align thousands of DNA molecules. For example, we can control the distance between adjacent DNA molecules, anchor both ends of the DNA curtains to the surface, and organize the DNA molecules into crisscrossing patterns; these new tools will allow us to probe more and more complex biological questions.
Movement of Postreplicative Mismatch Repair Proteins along DNA
The machinery responsible for replicating our genomes is not perfect; it occasionally adds an incorrect nucleotide. If these mismatches are not repaired, they can lead to permanent mutations. The importance of the proteins involved in this process is evidenced by their conservation throughout evolution; organisms ranging from bacteria to humans use the same basic set of proteins and reactions to complete the repair reaction. Moreover, the most common causes of hereditary colon cancer are linked to defects in the mismatch repair (MMR) pathway. In bacteria, the mismatch repair protein MutS is responsible for identifying mispaired bases and initiating the repair pathway. Once a mismatch is located, MutS recruits MutL, which in turn interacts with the nicking endonuclease MutH. Together these proteins locate the nearest hemimethylated dGATC site, which can be up to 2 kb away from the mismatch, and they use this site as a marker to distinguish between the parental and daughter DNA strands. This ensures that the newly synthesized daughter DNA strand is targeted for removal, thereby preventing any mutations from occurring in the original parental DNA sequence.
We are trying to dissect mismatch repair by directly visualizing the MMR proteins using our DNA curtain technology. We have shown that Msh2-Msh6 (a eukaryotic homolog of MutS) slides along DNA, exhibiting a one-dimensional (1D) diffusion coefficient consistent with rotation of the protein around the DNA while tracking the phosphate backbone. Msh2-Msh6 can also enter a nondiffusive state, but ATP binding induces a conformational change triggering reentry into the diffusive state. More recently, we found that the Mlh1-Pms1 protein complex (a eukaryotic homolog of MutL) also slides on DNA, but it moves more than 10 times faster than Msh2-Msh6. These results highlight that 1D diffusion plays a key role in how these proteins move along DNA. We are currently trying to reconstitute a complete reaction that includes both Msh2-Msh6 and Mlh1-Pms1, as well as engineered DNA substrates containing a single mispaired base at defined locations. By using multicolor imaging of DNA curtains, we hope to determine how these proteins interact with one another as they move along DNA, and to visualize exactly what happens throughout the repair reaction.
Imaging Individual Nucleosomes and Patterns of Nucleosome Deposition
Nucleosomes and chromatin affect virtually all aspects of eukaryotic biology. In an effort to mimic these natural in vivo substrates, we are using our DNA curtain technology to study fluorescently tagged nucleosomes. In our initial studies we have measured the intrinsic energy landscapes for nucleosome deposition on two model DNA substrates: the λ-phage genome and a fragment of the human β-globin locus. Our findings have demonstrated that intrinsically preferred nucleosome positions are well described by the most recent generation of theoretical algorithms that use DNA sequence composition to predict nucleosome positions, and we have confirmed that poly(dA-dT) exerts a strong exclusionary effect on nucleosome binding. We have also demonstrated that intrinsically preferred nucleosome-binding sites coincide with promoters and regulatory regions within the human β-globin DNA. This is the first measurement of an intrinsic energy landscape for a human DNA molecule; it supports the notion that a limited number of strongly positioned nucleosomes at critical locations may drive "statistical packing" of nucleosomes at flanking sites, suggesting that this mechanism may play a dominant role in organizing eukaryotic chromatin.
We have also imaged the deposition patterns of nucleosomes harboring histone variants and centromere-specific nonhistone proteins. These studies revealed that the variants H2AZ and CenH3 had little impact on the overall nucleosome deposition patterns on the DNA curtains. However, the nonhistone protein Scm3 altered the patterns of nucleosome deposition and allowed the nucleosomes to overcome the otherwise exclusionary effects of poly(dA-dT). This altered sequence preference may help Scm3 target centromeric nucleosomes to the poly(dA-dT)-rich sequences that are found at yeast centromeres. Looking forward, we anticipate being able to visualize how nucleosomes are affected by chromatin-remodeling proteins and various DNA-translocating proteins, and we are also beginning to incorporate these substrates into experiments designed to test the impact of nucleosomes on DNA repair mechanisms.
Grants from Columbia University, the National Institutes of Health, the National Science Foundation, and the Irma T. Hirschl and Monique Weill-Caulier Trusts provide partial support for these projects.
As of May 30, 2012