Discovery of an Inhibitor of Replication Protein A’s Protein-Protein Interactions
By: Carol A. Rouzer, VICB Communications
Published: October 30, 2013
Fragment-based drug discovery combined with structure-based design yields a molecule that inhibits RPA’s interactions with DNA damage response proteins.
DNA damage is a hallmark of cancer, and many cancer treatments, including radiation therapy and most chemotherapeutic agents, inflict additional DNA damage. Consequently, DNA damage response pathways are often activated in cancer cells and may contribute to the development of resistance to treatment. This has led to the hypothesis that inhibition of the DNA damage response and subsequent DNA repair may be an effective therapeutic target for some cancers. Now, VICB members Stephen Fesik and Walter Chazin report the discovery of a potent small molecule inhibitor of replication protein A, a key player in the DNA damage response [A. O. Frank, et al. (2013) J. Med. Chem., published online October 22, doi:10.1021/jm401333u].
RPA is a single-stranded DNA (ssDNA) binding protein comprising 70 kDa, 32 kDa, and 14 kDa subunits (Figure 1). The protein contains four oligonucleotide binding (OB) folds, designated A, B, C, and D, that bind ssDNA in decreasing order of affinity. RPA binds to ssDNA that is exposed during replication, DNA damage, and DNA repair. It protects the DNA from digestion by nucleases and prevents hairpin formation. The N-terminal domain of RPA’s 70 kDa subunit (RPA70N) also contains an OB fold that weakly binds ssDNA. Instead, it serves primarily as a docking point for multiple proteins. Thus, RPA also serves as the focal point for the assembly of DNA replication and repair complexes. Binding of proteins, such as ATRIP, RAD9, MRE11, and p53, to RPA70N is critical to the DNA damage response and DNA repair. Thus, blocking these interactions may have a therapeutic benefit in cancer. However, it is likely that complete inhibition of all of RPA’s functions would be highly toxic. This led the Fesik and Chazin labs, in collaboration with David Cortez and his laboratory, to search for a molecule that blocks only RPA70N protein-protein interactions.
Figure 1. Ribbon model of the structure of RPA domains showing the portions of the molecule for which crystal structure data are available. The dotted lines connecting these domains indicate flexible linkers that are the source of the dynamic architecture of the complete protein. Image provided by Walter Chazin.
Discovery of inhibitors of protein-protein interactions is often challenging because the binding sites are large and lack well-defined pockets into which a small molecule will fit neatly. This is true for the basic cleft that serves as the protein binding site on RPA70N. To address this problem, the Fesik lab took a fragment-based approach, which began by screening a library of molecular “fragments” (M.W. < 350), in search of compounds that exhibit binding to RPA70N, even at low affinity. Their goal was to identify fragments that interact with the protein at multiple sites, and then link them together to obtain a single, high affinity molecule.
The Fesik lab used 1H, 15N HMQC NMR (heteronuclear quantum coherence nuclear magnetic resonance) to screen their fragment library. This technique provided a signal for each H-N bond in RPA70N displayed as a series of spots on a two-dimensional graph (Figure 2). Data were acquired in the absence and presence of each test compound. Movement of the position of a spot indicated binding of the fragment and provided information regarding the exact location on the protein where the interaction occurred. The results revealed two binding sites on RPA70N. Site-1, located at serine-55 and Site-2, located at threonine-60, were positioned at opposite ends of the protein binding cleft.
Figure 2. 1H, 15N HMQC NMR data from the screening of a fragment library for binding to RPA70N. Each spot on the spectrum represents an N-H bond in the protein. Black spots are data obtained in the absence of a fragment, and red spots are obtained in the presence of a fragment hit. (A) Spectrum from the analysis of a fragment hit that binds to Site-1. Note the shift in the position of the spot corresponding to serine-55 (inset). (B) Spectrum from the analysis of a fragment hit that binds to Site-2. Note the shift in the position of the spot corresponding to threonine-60 (inset). Image reproduced with permission from A. O. Frank, et al. (2013) J. Med. Chem., published online October 22, doi:10.1021/jm401333u], copyright 2013, American Chemical Society.
The investigators screened 14,976 fragments, obtaining 149 confirmed hits of which 81 bound to both sites, 52 bound to Site-1 only, and 16 bound to Site-2 only. The binding affinities for these interactions fell in the range of 500 to 5000 μM. Based on the structures of the hit fragments, the researchers purchased and screened additional compounds in search of more detailed structure-activity relationships. In addition, they obtained co-crystal structures for 12 of the hits bound to RPA70N to assess how the molecules interact with the protein. The results led to a group of previously published inhibitors of RPA70N, with affinities in the 10 μM range [J. D. K. Patrone, et al. (2013) ACS Med. Chem. Lett., 4, 601]. In search of more potent inhibitors, however, the investigators further explored compounds 2 and 4 (Figure 3). Co-crystal structures of these molecules with RPA70N indicated that each binds to the protein at both Site-1 and Site-2, albeit with different affinities. Although both molecules exhibited the highest affinity for Site-1, they were able to bind simultaneously to the protein, as indicated by X-ray crystallography data. Under these conditions, compound 2’s higher affinity for Site-2 allowed it to occupy that position, while compound 4 occupied Site-1 (Figure 4).
Figure 3. Structures of compounds 2 and 4. The circle shows the points on each molecule that were in close proximity when bound to RPA70N. These data led to the synthesis of compounds 7 and 8, designed on the principle of linking compounds 2 and 4.
Figure 4. Co-crystal structures of RPA70N in complex with fragment hits. (A) Surface depiction of the basic cleft showing the two binding sites. Structures of compound 2 (green) (B), compound 4 (yellow) (C), and compounds 2 and 4 (D) bound to the protein. Water molecules are shown as red spheres. H-bonds are dashed lines. Residues within 3Å of a binding site are blue, and key binding site residues (serine-55 for Site-1 and threonine-60 for Site-2) are in magenta. Image reproduced with permission from A. O. Frank, et al. (2013) J. Med. Chem., published online October 22, doi:10.1021/jm401333u], copyright 2013, American Chemical Society.
The co-crystal structure, including both compounds 2 and 4 bound to RPA70N, revealed that the closest heavy atoms - the 4-position of the phenyl ring of compound 2 and the 4-position of the pyrazole phenyl ring of compound 4 (Figure 3) - were only 4.9 Å apart. This suggested that the two compounds could easily be linked together to form a single larger molecule with higher affinity. The investigators did just that, with their synthesis of compound 7 (Figure 3), comprising all of compounds 4 and 2 (with the exception of the chlorine substituent on compound 2) connected with a carbon-oxygen linker. As expected, compound 7 displayed a much higher affinity for RPA70N (20 μM) than either compound 2 or 4. In addition, compound 7 could displace ATRIP from the RPA70N binding site in a fluorescence polarization anisotropy (FPA) assay, indicating that it possessed the desired biological activity.
Despite the promising results with compound 7, its affinity was not as high as predicted based on the starting affinities of compounds 2 and 4. X-ray co-crystal structures indicated that the 5-position phenyl ring of compound 7 was shifted relative to its position in the co-crystal structure of compound 4 bound to RPA70N (Figure 5). Careful analysis of the structural data suggested modifications to the molecule that would improve its interaction with the RPA70N binding cleft. The resulting compound 8, indeed, showed much higher affinity (0.19 μM) for the protein.
Figure 5. Co-crystal structures of RPA70N in complex with compound 7 (light blue) (A and C) and compound 8 (tan) (B and D). Compounds 2 and 4 are also shown in (B) and (D) for comparison. Water molecules are shown as red spheres. H-bonds are dashed lines. Residues within 3Å of a binding site are blue, and key binding site residues (serine-55 for Site-1 and threonine-60 for Site-2) are in magenta. Image reproduced with permission from A. O. Frank, et al. (2013) J. Med. Chem., published online October 22, doi:10.1021/jm401333u], copyright 2013, American Chemical Society.
Compound 8 readily displaced ATRIP from RPA70N in the FPA assay. However, it was equally important that compound 8 should not interfere with RPA’s other functions. Consequently, the investigators evaluated the compound’s ability to interfere with the ability of the A and B domains of RPA70 to bind ssDNA. Addition of these domains to form RPA70NAB had little effect on compound 8’s ability to displace ATRIP from the protein binding site. Although compound 8 was also able to displace ssDNA from RPANAB, the required concentrations were over 10-fold higher than those needed to displace ATRIP, confirming the desired selectivity.
Compound 8 represents the first molecule that can selectively interfere with RPA70N’s interactions with DNA damage response proteins at submicromolar concentrations. With this tool, scientists can now begin to probe the role of RPA-dependent DNA damage responses in cancer cell survival and resistance to therapy. Compound 8 also serves as an important lead molecule for the potential development of cancer therapeutics targeting RPA.