Pan-HSP90 ligand binding reveals isoform-specific differences in plasticity and water networks.

Isoforms of heat shock protein 90 (HSP90) fold oncoproteins that facilitate all 10 hallmarks of cancer. However, its promise as a therapeutic target remains unfulfilled as there is still no FDA-approved drug targeting HSP90 in disease. Among the reasons hindering progress are side effects caused by pan-HSP90 inhibition. Selective targeting of the four isoforms is challenging due to high sequence and structural similarity. Surprisingly, while decades of drug discovery efforts have produced almost 400 human HSP90 structures, no single ligand has been structurally characterized across all four human isoforms to date, which could reveal structural differences to achieve selectivity. To better understand the HSP90 landscape relevant for ligand binding and design we take a three-pronged approach. First, we solved the first complete set of structures of a single ligand bound to all four human isoforms. This enabled a systematic comparison of how side-chains and water networks respond to ligand binding across isoforms. Second, we expanded our analysis to publicly available, incomplete isoform-ligand series with distinct ligand chemistry. This highlighted general trends of protein and water mobility that differ among isoforms and impact ligand binding. Third, we further probed the Hsp90α conformational landscape for accommodating a congeneric series containing the purine scaffold common to HSP90 inhibitors. This revealed how minor ligand modifications flip ligand poses and perturb water and protein conformations. Taken together, this work illustrates how a systematic approach can shed new light on an "old" target and reveal hidden isoform-specific accommodations of congeneric ligands that may be exploited in ligand discovery and design.

Water Networks Repopulate Protein-Ligand Interfaces with Temperature.

High-resolution crystal structures highlight the importance of water networks in protein-ligand interactions. However, as these are typically determined at cryogenic temperature, resulting insights may be structurally precise but not biologically accurate. By collecting 10 matched room-temperature and cryogenic datasets of the biomedical target Hsp90α, we identified changes in water networks that impact protein conformations at the ligand binding interface. Water repositioning with temperature repopulates protein ensembles and ligand interactions. We introduce Flipper conformational barcodes to identify temperature-sensitive regions in electron density maps. This revealed that temperature-responsive states coincide with ligand-responsive regions and capture unique binding signatures that disappear upon cryo-cooling. Our results have implications for discovering Hsp90 selective ligands, and, more generally, for the utility of hidden protein and water conformations in drug discovery.

Structures of REV1 UBM2 Domain Complex with Ubiquitin and with a Small-Molecule that Inhibits the REV1 UBM2-Ubiquitin Interaction.

REV1 is a DNA damage tolerance protein and encodes two ubiquitin-binding motifs (UBM1 and UBM2) that are essential for REV1 functions in cell survival under DNA-damaging stress. Here we report the first solution and X-ray crystal structures of REV1 UBM2 and its complex with ubiquitin, respectively. Furthermore, we have identified the first small-molecule compound, MLAF50, that directly binds to REV1 UBM2. In the heteronuclear single quantum coherence NMR experiments, peaks of UBM2 but not of UBM1 are significantly shifted by the addition of ubiquitin, which agrees to the observation that REV1 UBM2 but not UBM1 is required for DNA damage tolerance. REV1 UBM2 interacts with hydrophobic residues of ubiquitin such as L8 and L73. NMR data suggest that MLAF50 binds to the same residues of REV1 UBM2 that interact with ubiquitin, indicating that MLAF50 can compete with the REV1 UBM2-ubiquitin interaction orthosterically. Indeed, MLAF50 inhibited the interaction of REV1 UBM2 with ubiquitin and prevented chromatin localization of REV1 induced by cisplatin in U2OS cells. Our results structurally validate REV1 UBM2 as a target of a small-molecule inhibitor and demonstrate a new avenue to targeting ubiquitination-mediated protein interactions with a chemical tool.

Small-molecules that bind to the ubiquitin-binding motif of REV1 inhibit REV1 interaction with K164-monoubiquitinated PCNA and suppress DNA damage tolerance.

REV1 protein is a mutagenic DNA damage tolerance (DDT) mediator and encodes two ubiquitin-binding motifs (i.e., UBM1 and UBM2) that are essential for the DDT function. REV1 interacts with K164-monoubiquitinated PCNA (UbPCNA) in cells upon DNA-damaging stress. By using AlphaScreen assays to detect inhibition of REV1 and UbPCNA protein interactions along with an NMR-based strategy, we identified small-molecule compounds that inhibit the REV1/UbPCNA interaction and that directly bind to REV1 UBM2. In cells, one of the compound prevented recruitment of REV1 to PCNA foci on chromatin upon cisplatin treatment, delayed removal of UV-induced cyclopyrimidine dimers from nuclei, prevented UV-induced mutation of HPRT gene, and diminished clonogenic survival of cells that were challenged by cyclophosphamide or cisplatin. This study demonstrates the potential utility of a small-molecule REV1 UBM2 inhibitor for preventing DDT.

Identification of the first small-molecule inhibitor of the REV7 DNA repair protein interaction.

DNA interstrand crosslink (ICL) repair (ICLR) has been implicated in the resistance of cancer cells to ICL-inducing chemotherapeutic agents. Despite the clinical significance of ICL-inducing chemotherapy, few studies have focused on developing small-molecule inhibitors for ICLR. The mammalian DNA polymerase ζ, which comprises the catalytic subunit REV3L and the non-catalytic subunit REV7, is essential for ICLR. To identify small-molecule compounds that are mechanistically capable of inhibiting ICLR by targeting REV7, high-throughput screening and structure-activity relationship (SAR) analysis were performed. Compound 1 was identified as an inhibitor of the interaction of REV7 with the REV7-binding sequence of REV3L. Compound 7 (an optimized analog of compound 1) bound directly to REV7 in nuclear magnetic resonance analyses, and inhibited the reactivation of a reporter plasmid containing an ICL in between the promoter and reporter regions. The normalized clonogenic survival of HeLa cells treated with cisplatin and compound 7 was lower than that for cells treated with cisplatin only. These findings indicate that a small-molecule inhibitor of the REV7/REV3L interaction can chemosensitize cells by inhibiting ICLR.

Structural Basis for the Interaction between Pyk2-FAT Domain and Leupaxin LD Repeats.

Proline-rich tyrosine kinase 2 (Pyk2) is a nonreceptor tyrosine kinase and belongs to the focal adhesion kinase (FAK) family. Like FAK, the C-terminal focal adhesion-targeting (FAT) domain of Pyk2 binds to paxillin, a scaffold protein in focal adhesions; however, the interaction between the FAT domain of Pyk2 and paxillin is dynamic and unstable. Leupaxin is another member in the paxillin family and was suggested to be the native binding partner of Pyk2; Pyk2 gene expression is strongly correlated with that of leupaxin in many tissues including primary breast cancer. Here, we report that leupaxin interacts with Pyk2-FAT. Leupaxin has four leucine-aspartate (LD) motifs. The first and third LD motifs of leupaxin preferably target the two LD-binding sites on the Pyk2-FAT domain, respectively. Moreover, the full-length leupaxin binds to Pyk2-FAT as a stable one-to-one complex. Together, we propose that there is an underlying selectivity between leupaxin and paxillin for Pyk2, which may influence the differing behavior of the two proteins at focal adhesion sites.

Structural and mechanistic insights into the interaction between Pyk2 and paxillin LD motifs.

Proline-rich tyrosine kinase 2 (Pyk2) is a member of the focal adhesion kinase (FAK) subfamily of cytoplasmic tyrosine kinases. The C-terminal Pyk2-focal adhesion targeting (FAT) domain binds to paxillin, an adhesion molecule. Paxillin has five leucine-aspartate (LD) motifs (LD1-LD5). Here, we show that the second LD motif of paxillin, LD2, interacts with Pyk2-FAT, similar to the known Pyk2-FAT/LD4 interaction. Both LD motifs can target two ligand binding sites on Pyk2-FAT. Interestingly, they also share similar binding affinity for Pyk2-FAT with preferential association to one site relative to the other. Nevertheless, the LD2-LD4 region of paxillin (paxillin(133-290)) binds to Pyk2-FAT as a 1:1 complex. However, our data suggest that the Pyk2-FAT and paxillin complex is dynamic and it appears to be a mixture of two distinct conformations of paxillin that almost equally compete for Pyk2-FAT binding. These studies provide insight into the underlying selectivity of paxillin for Pyk2 and FAK that may influence the differing behavior of these two closely related kinases in focal adhesion sites.

NMR structural analysis of the soluble domain of ZiaA-ATPase and the basis of selective interactions with copper metallochaperone Atx1.

A Cu(I) metallochaperone, Atx1, interacts with the amino-terminal domain of a Cu(I)-transporting ATPase, PacS(N), but not with a domain of related Zn-transporting ATPase, ZiaA(N) in Synechocystis PCC 6803. This is thought to prevent ZiaA(N) from acquiring Cu(I), which it binds more tightly than Zn. Solution structures of Atx1, PacS(N), and the heterodimer were previously described. Here we report solution structural studies of the ZiaA(N) soluble domain. Apo-ZiaA(N) has a typical ferredoxin-like fold followed by an atypical 34 residues of unstructured polypeptide containing a His(7) motif. ZiaA(N) competes with the metallochromic indicator 4-(2-pyridylazo)resorcinol for 1 equiv of Zn, which can be displaced by thiol-modifying p-mercuriphenylsulfonic acid, establishing that a high-affinity site involves thiols of the CXXC motif within the ferredoxin-like fold. A single equivalent of Zn affects nuclear magnetic resonance signals arising from the CXXC motif as well as all seven His residues. The presence of NMR-line broadening in both sites implies that Zn(1)-ZiaA(N) undergoes exchange phenomena, consistent with CXXC-bound Zn coincidentally sampling various His ligands. These Zn-dependent dynamic changes could either aid metal transfer or alter intramolecular interactions. No formation of Atx1-Cu(I)-ZiaA(N) heterodimers was observed, and in the presence of equimolar ZiaA(N) and PacS(N), only Atx1-Cu(I)-PacS(N) complexes were detected. Residues flanking the CXXC motif of PacS(N) (R(13)-ASS(20)) differ in charge and bulk from those of ZiaA(N) (D(18)-KLK(25)) and make contacts in the Atx1-Cu(I)-PacS(N) complex. Crucially, swapping these residues flanking the CXXC motifs of ZiaA(N) and PacS(N) reciprocally swaps partner choice by Atx1. These few residues of the two ATPases have diverged during evolution to bias Atx1 interactions in favor of PacS(N) rather than ZiaA(N.).

Structural insights into membrane targeting by the flagellar calcium-binding protein (FCaBP), a myristoylated and palmitoylated calcium sensor in Trypanosoma cruzi.

The flagellar calcium-binding protein (FCaBP) of the protozoan Trypanosoma cruzi is targeted to the flagellar membrane where it regulates flagellar function and assembly. As a first step toward understanding the Ca(2+)-induced conformational changes important for membrane-targeting, we report here the x-ray crystal structure of FCaBP in the Ca(2+)-free state determined at 2.2A resolution. The first 17 residues from the N terminus appear unstructured and solvent-exposed. Residues implicated in membrane targeting (Lys-19, Lys-22, and Lys-25) are flanked by an exposed N-terminal helix (residues 26-37), forming a patch of positive charge on the protein surface that may interact electrostatically with flagellar membrane targets. The four EF-hands in FCaBP each adopt a "closed conformation" similar to that seen in Ca(2+)-free calmodulin. The overall fold of FCaBP is closest to that of grancalcin and other members of the penta EF-hand superfamily. Unlike the dimeric penta EF-hand proteins, FCaBP lacks a fifth EF-hand and is monomeric. The unstructured N-terminal region of FCaBP suggests that its covalently attached myristoyl group at the N terminus may be solvent-exposed, in contrast to the highly sequestered myristoyl group seen in recoverin and GCAP1. NMR analysis demonstrates that the myristoyl group attached to FCaBP is indeed solvent-exposed in both the Ca(2+)-free and Ca(2+)-bound states, and myristoylation has no effect on protein structure and folding stability. We propose that exposed acyl groups at the N terminus may anchor FCaBP to the flagellar membrane and that Ca(2+)-induced conformational changes may control its binding to membrane-bound protein targets.

NMR structure of DREAM: Implications for Ca(2+)-dependent DNA binding and protein dimerization.

DREAM (calsenilin/KChIP3) is an EF-hand calcium-binding protein that binds to specific DNA sequences and regulates Ca2+-induced transcription of prodynorphin and c-fos genes. Here, we present the atomic-resolution structure of Ca2+-bound DREAM in solution determined by nuclear magnetic resonance (NMR) spectroscopy. Pulsed-field gradient NMR diffusion experiments and 15N NMR relaxation analysis indicate that Ca2+-bound DREAM forms a stable dimer in solution. The structure of the first 77 residues from the N-terminus could not be determined by our NMR analysis. The C-terminal DREAM structure (residues 78-256) contains four EF-hand motifs arranged in a tandem linear array, similar to that seen in KChIP1, recoverin, and other structures of the neuronal calcium sensor (NCS) branch of the calmodulin superfamily. Mg2+ is bound at the second EF-hand, whereas Ca2+ is bound functionally at the third and fourth sites. The first and second EF-hands form an exposed hydrophobic groove on the protein surface lined by side-chain atoms of L96, F100, F114, I117, Y118, F121, F122, Y151, L155, L158, and L159 that are highly conserved in all NCS proteins. An exposed leucine near the C-terminus (L251) is suggested to form intermolecular contacts with leucine residues in the hydrophobic groove (L155, L158, and L159). Positively charged side chains of Arg and Lys (Lys87, Lys90, Lys91, Arg98, Lys101, Arg160, and Lys166) are clustered on one side of the protein surface and may mediate electrostatic contacts with DNA targets. We propose that Ca2+-induced dimerization of DREAM may partially block the putative DNA-binding site, which may suggest as to how Ca2+ abolishes DREAM binding to DNA to activate the transcription of prodynorphin and other downstream genes in pain control.

The delivery of copper for thylakoid import observed by NMR.

The thylakoid compartments of plant chloroplasts are a vital destination for copper. Copper is needed to form holo-plastocyanin, which must shuttle electrons between photosystems to convert light into biologically useful chemical energy. Copper can bind tightly to proteins, so it has been hypothesized that copper partitions onto ligand-exchange pathways to reach intracellular locations without inflicting damage en route. The copper metallochaperone Atx1 of chloroplast-related cyanobacteria (ScAtx1) engages in bacterial two-hybrid interactions with N-terminal domains of copper-transporting ATPases CtaA (cell import) and PacS (thylakoid import). Here we visualize copper delivery. The N-terminal domain PacS(N) has a ferredoxin-like fold that forms copper-dependent heterodimers with ScAtx1. Removal of copper, by the addition of the cuprous-ion chelator bathocuproine disulfonate, disrupts this heterodimer, as shown from a reduction of the overall tumbling rate of the protein mixture. The NMR spectral changes of the heterodimer versus the separate proteins reveal that loops 1, 3, and 5 (the carboxyl tail) of the ScAtx1 Cu(I) site switch to an apo-like configuration in the heterodimer. NMR data ((2)J(NH) couplings in the imidazole ring of (15)N ScAtx1 His-61) also show that His-61, bound to copper(I) in [Cu(I)ScAtx1](2), is not coordinated to copper in the heterodimer. A model for the PacS(N)/Cu(I)/ScAtx1 complex is presented. Contact with PacS(N) induces change to the ScAtx1 copper-coordination sphere that drives copper release for thylakoid import. These data also elaborate on the mechanism to keep copper(I) out of the ZiaA(N) ATPase zinc sites.

Protein stability and mutations in the axial methionine loop of a minimal cytochrome c.

The minimal mono-heme ferricytochrome c from Bacillus pasteurii, containing 71 amino acids, has been further investigated through mutagenesis of different positions in the loop containing the iron ligand Met71. These mutations have been designed to sample different aspects of the loop structure, in order to obtain insights into the determinants of the stability of the iron(III) environment. In particular, positions 68, 72 and 75 have been essayed. Gln68 has been mutated to Lys to provide a suitable alternate ligand that can displace Met71 under denaturing conditions. Pro72 has been mutated to Gly and Ala to modify the range of allowed backbone conformations. Ile75, which is in van der Waals contact with Met71 and partly shields a long-lived water molecule in a protein cavity, has been substituted by Val and Ala to affect the network of inter-residue interactions around the metal site. The different contributions of the above amino acids to protein parameters such as structure, redox potential and the overall stability against unfolding with guanidinium hydrochloride are analyzed. While the structure remains essentially the same, the stability decreases with mutations. The comparison with mitochondrial c-type cytochromes is instructive.