Periodic Reporting for period 1 - DesProtDNA (Computational design of proteins binding nucleosomal DNA with specificity for therapeutic applications)
Berichtszeitraum: 2017-04-01 bis 2019-03-31
DNA encodes the biological instructions for the functioning of living organisms by folding into a complex three-dimensional structure. By wrapping around histone proteins, eukaryotic DNA forms nucleosome particles, which in turn self-associate to form a chromatin fiber with a structure that changes with the cell-cycle. DNA packaged into nucleosomes (“nucleosomal DNA”) is less accessible for most DNA-binding transcription factors and, accordingly, less active for gene expression. Considerable research efforts are being devoted to studying chromatin structure at distinct resolutions and to exploiting the therapeutic potential of nucleosomal DNA. Some proteins can bind to the exposed face of nucleosomal DNA and drive biological processes associated with cancer and viral infection. From a therapeutic point of view, drugs that bind to nucleosomal DNA with specificity hold great potential for blocking the action of proteins involved in a number of diseases, as they alter the accessibility of DNA to regulatory proteins.
The design of nucleosome-binding molecules specific for a DNA sequence remains an outstanding challenge. Nevertheless, nucleosomal DNA binding specificity is hard to achieve by repurposing currently existing chromatin-interacting proteins due to the lack of suitable natural templates. Recent advances in computational protein design provide a route to custom-tailor protein structures and interactions, and here we set out to design, for the first time, proteins customized to bind nucleosomal DNA with specificity. Such protein customization underlies the challenge of identifying and/or building a protein structure, its optimal placement around the DNA target site, and an amino-acid sequence that stabilizes the target protein-DNA complex.
DesProtDNA comprised three objectives. First, development of a general computational method to de novo design protein-DNA binding interfaces in the nucleosome with specificity. Second, experimentally validate the method by testing proteins designed to target a nucleosome with high-resolution structure available. Third, use the method to design proteins binding specific nucleosomal DNA sites involved in diseases. Additionally, the project aims to provide the researcher with new computational and experimental skills to implement new protein-DNA design schemes and validate them.
In conclusion, we have developed three different computational approaches that range from redesigning currently existing proteins to full de novo design to achieve optimal binding to nucleosomal DNA. The designed proteins have been experimentally characterized by expressing them recombinantly and assessing their binding properties. Novel nucleosome binding proteins have been obtained that where designed to target specific DNA sites and have binding interfaces similar as designed. Yet, further high-resolution structural information is necessary to confirm the actual binding mode of the designs. This is the first time proteins have been designed to bind the exposed face of nucleosomal DNA. Obtaining high-resolution structural conformation of our designs will lay the groundwork for developing new chromatin-based therapeutic strategies as well as innovative research tools for chromatin studies using our computational design method.
The design of nucleosome-binding molecules specific for a DNA sequence remains an outstanding challenge. Nevertheless, nucleosomal DNA binding specificity is hard to achieve by repurposing currently existing chromatin-interacting proteins due to the lack of suitable natural templates. Recent advances in computational protein design provide a route to custom-tailor protein structures and interactions, and here we set out to design, for the first time, proteins customized to bind nucleosomal DNA with specificity. Such protein customization underlies the challenge of identifying and/or building a protein structure, its optimal placement around the DNA target site, and an amino-acid sequence that stabilizes the target protein-DNA complex.
DesProtDNA comprised three objectives. First, development of a general computational method to de novo design protein-DNA binding interfaces in the nucleosome with specificity. Second, experimentally validate the method by testing proteins designed to target a nucleosome with high-resolution structure available. Third, use the method to design proteins binding specific nucleosomal DNA sites involved in diseases. Additionally, the project aims to provide the researcher with new computational and experimental skills to implement new protein-DNA design schemes and validate them.
In conclusion, we have developed three different computational approaches that range from redesigning currently existing proteins to full de novo design to achieve optimal binding to nucleosomal DNA. The designed proteins have been experimentally characterized by expressing them recombinantly and assessing their binding properties. Novel nucleosome binding proteins have been obtained that where designed to target specific DNA sites and have binding interfaces similar as designed. Yet, further high-resolution structural information is necessary to confirm the actual binding mode of the designs. This is the first time proteins have been designed to bind the exposed face of nucleosomal DNA. Obtaining high-resolution structural conformation of our designs will lay the groundwork for developing new chromatin-based therapeutic strategies as well as innovative research tools for chromatin studies using our computational design method.
We have developed three different computational design approaches combining Rosetta design and molecular dynamics simulations. In the first approach, we used helical bundles of extreme stability that were long and straight to be shape-complementary to the target DNA sites. These bundles were docked into target DNA grooves to maximize the number of good contacts between amino acids and DNA bases. Binding interfaces were verified in silico by protein-DNA docking and molecular dynamics simulations. In the second approach, we repurposed transcription activator-like (TAL) effectors to bind each groove of the target nucleosome supergrooves. Under this approach, small TAL effectors were docked separately in each of the two grooves, sequence-designed at the interface and linked with a short loop connection. In the third approach, we de novo-designed binding proteins (mini 3-helix bundles) containing two domains, where each domain was optimized for each groove of the supergrooves. We docked these mini proteins to each single major groove, similarly as before, and linked those well-oriented for a short loop connection.
The computational approaches were tested with the design of proteins targeting the nucleosome containing the 601-sequence, which is a DNA sequence strongly favoring nucleosomes and for which there are high-resolution crystal structures available. The designed proteins expressed solubly at high levels, except for the repurposed TAL effectors, which may not tolerate these modifications as these are marginally stable. For the helical bundles designed with the first approach we identified three proteins binding more strongly to nucleosomes than to naked DNA (in the low micromolar range). We carried out sedimentation velocity experiments and found that the design complexes were monodisperse, and one design bound with a stoichiometry of 2 protein:1 nucleosome. Currently, electron microscopy experiments are carried out to further elucidate the structure of this complex. Smaller versions of the designed proteins and that preserve the same binding interface were also found to bind nucleosomes with similar affinity, further supporting that the proteins bind through the intended interfaces. On the other hand, proteins de novo-designed in the third approach seem the most promising to achieve the highest specificity. Binding assays indicate novel binding proteins forming lower molecular weight species, which may support formation of protein-nucleosome complexes with 1:1 stoichiometry. Currently, additional experiments to get more structural characterization of these designs are in process.
We have presented our work in several international conferences, and we are currently preparing a manuscript for publication.
The computational approaches were tested with the design of proteins targeting the nucleosome containing the 601-sequence, which is a DNA sequence strongly favoring nucleosomes and for which there are high-resolution crystal structures available. The designed proteins expressed solubly at high levels, except for the repurposed TAL effectors, which may not tolerate these modifications as these are marginally stable. For the helical bundles designed with the first approach we identified three proteins binding more strongly to nucleosomes than to naked DNA (in the low micromolar range). We carried out sedimentation velocity experiments and found that the design complexes were monodisperse, and one design bound with a stoichiometry of 2 protein:1 nucleosome. Currently, electron microscopy experiments are carried out to further elucidate the structure of this complex. Smaller versions of the designed proteins and that preserve the same binding interface were also found to bind nucleosomes with similar affinity, further supporting that the proteins bind through the intended interfaces. On the other hand, proteins de novo-designed in the third approach seem the most promising to achieve the highest specificity. Binding assays indicate novel binding proteins forming lower molecular weight species, which may support formation of protein-nucleosome complexes with 1:1 stoichiometry. Currently, additional experiments to get more structural characterization of these designs are in process.
We have presented our work in several international conferences, and we are currently preparing a manuscript for publication.
The design of proteins binding nucleosomal DNA with specificity has not been achieved to date. Previous attempts to design protein-DNA interfaces were limited to repurposing already existing interfaces with naked DNA to change their binding specificities. The current proposal has extended the computational protein design methodology to build a class of protein-DNA interfaces that has not yet been observed in nature, advancing the state-of-the-art substantially. Here we have developed new methods for designing, de novo, protein-nuclesome interfaces (through nucleosome supergrooves) and experimentally tested their predicted properties. This a challenging task that has pushed the limits of current existing design protocols and opened the door for chromatin-based therapeutic approaches and innovative research tools for nucleosomes. For example, binding nucleosomal DNA to block the action of cancer-promoting transcription factors represents a novel anti-cancer strategy. In addition, the ability to design proteins to bind to specific nucleosomes could open new avenues to tag specific nucleosomes and may result in a new approach to study chromatin structure.