Deep learning guided design of dynamic proteins.

Deep learning has greatly advanced design of highly stable static protein structures, but the controlled conformational dynamics that are hallmarks of natural switch-like signaling proteins have remained inaccessible to de novo design. Here, we describe a general deep-learning-guided approach for de novo design of dynamic changes between intra-domain geometries of proteins, similar to switch mechanisms prevalent in nature, with atom-level precision. We solve 4 structures validating the designed conformations, show microsecond transitions between them, and demonstrate that the conformational landscape can be modulated by orthosteric ligands and allosteric mutations. Physics-based simulations are in remarkable agreement with deep-learning predictions and experimental data, reveal distinct state-dependent residue interaction networks, and predict mutations that tune the designed conformational landscape. Our approach demonstrates that new modes of motion can now be realized through de novo design and provides a framework for constructing biology-inspired, tunable and controllable protein signaling behavior de novo.

Identifying ligands for the PHD1 finger of KDM5A through high-throughput screening.

PHD fingers are a type of chromatin reader that primarily recognize chromatin as a function of lysine methylation state. Dysregulated PHD fingers are implicated in various human diseases, including acute myeloid leukemia. Targeting PHD fingers with small molecules is considered challenging as their histone tail binding pockets are often shallow and surface-exposed. The KDM5A PHD1 finger regulates the catalytic activity of KDM5A, an epigenetic enzyme often misregulated in cancers. To identify ligands that disrupt the PHD1-histone peptide interaction, we conducted a high-throughput screen and validated hits by orthogonal methods. We further elucidated structure-activity relationships in two classes of compounds to identify features important for binding. Our investigation offers a starting point for further optimization of small molecule PHD1 ligands.

Chromatin Sensing by the Auxiliary Domains of KDM5C Regulates Its Demethylase Activity and Is Disrupted by X-linked Intellectual Disability Mutations.

The H3K4me3 chromatin modification, a hallmark of promoters of actively transcribed genes, is dynamically removed by the KDM5 family of histone demethylases. The KDM5 demethylases have a number of accessory domains, two of which, ARID and PHD1, lie between the segments of the catalytic domain. KDM5C, which has a unique role in neural development, harbors a number of mutations adjacent to its accessory domains that cause X-linked intellectual disability (XLID). The roles of these accessory domains remain unknown, limiting an understanding of how XLID mutations affect KDM5C activity. Through in vitro binding and kinetic studies using nucleosomes, we find that while the ARID domain is required for efficient nucleosome demethylation, the PHD1 domain alone has an inhibitory role in KDM5C catalysis. In addition, the unstructured linker region between the ARID and PHD1 domains interacts with PHD1 and is necessary for nucleosome binding. Our data suggests a model in which the PHD1 domain inhibits DNA recognition by KDM5C. This inhibitory effect is relieved by the H3 tail, enabling recognition of flanking DNA on the nucleosome. Importantly, we find that XLID mutations adjacent to the ARID and PHD1 domains break this regulation by enhancing DNA binding, resulting in the loss of specificity of substrate chromatin recognition and rendering demethylase activity lower in the presence of flanking DNA. Our findings suggest a model by which specific XLID mutations could alter chromatin recognition and enable euchromatin-specific dysregulation of demethylation by KDM5C.

Recognition of Histone H3 Methylation States by the PHD1 Domain of Histone Demethylase KDM5A.

PHD reader domains are chromatin binding modules often responsible for the recruitment of large protein complexes that contain histone modifying enzymes, chromatin remodelers, and DNA repair machinery. A majority of PHD domains recognize N-terminal residues of histone H3 and are sensitive to the methylation state of Lys4 in histone H3 (H3K4). Histone demethylase KDM5A, an epigenetic eraser enzyme that contains three PHD domains, is often overexpressed in various cancers, and its demethylation activity is allosterically enhanced when its PHD1 domain is bound to the H3 tail. The allosteric regulatory function of PHD1 expands roles of reader domains, suggesting unique features of this chromatin interacting module. Our previous studies determined the H3 binding site of PHD1, although it remains unclear how the H3 tail interacts with the N-terminal residues of PHD1 and how PHD1 discriminates against H3 tails with varying degrees of H3K4 methylation. Here, we have determined the solution structure of apo and H3 bound PHD1. We observe conformational changes occurring in PHD1 in order to accommodate H3, which interestingly binds in a helical conformation. We also observe differential interactions of binding residues with differently methylated H3K4 peptides (me0, me1, me2, or me3), providing a rationale for PHD1's preference for lower methylation states of H3K4. We further assessed the contributions of various H3 interacting residues in the PHD1 domain to the binding of H3 peptides. The structural details of the H3 binding site could provide useful information to aid the development of allosteric small molecule modulators of KDM5A.

Probing the KRas Switch II Groove by Fluorine NMR Spectroscopy.

While there has been recent success in the development of KRasG12C inhibitors, unmet needs for selective inhibitors of KRasG12D and the remaining oncogenic KRas proteins remain. Here, we applied trifluoromethyl-containing ligands of KRas proteins as competitive probe ligands to assay the occupancy of the switch II pocket by 19F NMR spectroscopy. Structure-activity-relationship studies of probe ligands increased the sensitivity of the assay and identified structures that differentially detected each nucleotide state of KRasG12D. These differences in selectivity, combined with the high resolution of 19F NMR spectroscopy, enabled this method to be expanded to assay both nucleotide states of the protein simultaneously.

Systems-level effects of allosteric perturbations to a model molecular switch.

Molecular switch proteins whose cycling between states is controlled by opposing regulators1,2 are central to biological signal transduction. As switch proteins function within highly connected interaction networks3, the fundamental question arises of how functional specificity is achieved when different processes share common regulators. Here we show that functional specificity of the small GTPase switch protein Gsp1 in Saccharomyces cerevisiae (the homologue of the human protein RAN)4 is linked to differential sensitivity of biological processes to different kinetics of the Gsp1 (RAN) switch cycle. We make 55 targeted point mutations to individual protein interaction interfaces of Gsp1 (RAN) and show through quantitative genetic5 and physical interaction mapping that Gsp1 (RAN) interface perturbations have widespread cellular consequences. Contrary to expectation, the cellular effects of the interface mutations group by their biophysical effects on kinetic parameters of the GTPase switch cycle and not by the targeted interfaces. Instead, we show that interface mutations allosterically tune the GTPase cycle kinetics. These results suggest a model in which protein partner binding, or post-translational modifications at distal sites, could act as allosteric regulators of GTPase switching. Similar mechanisms may underlie regulation by other GTPases, and other biological switches. Furthermore, our integrative platform to determine the quantitative consequences of molecular perturbations may help to explain the effects of disease mutations that target central molecular switches.

Expanding the space of protein geometries by computational design of de novo fold families.

Naturally occurring proteins vary the precise geometries of structural elements to create distinct shapes optimal for function. We present a computational design method, loop-helix-loop unit combinatorial sampling (LUCS), that mimics nature's ability to create families of proteins with the same overall fold but precisely tunable geometries. Through near-exhaustive sampling of loop-helix-loop elements, LUCS generates highly diverse geometries encompassing those found in nature but also surpassing known structure space. Biophysical characterization showed that 17 (38%) of 45 tested LUCS designs encompassing two different structural topologies were well folded, including 16 with designed non-native geometries. Four experimentally solved structures closely matched the designs. LUCS greatly expands the designable structure space and offers a new paradigm for designing proteins with tunable geometries that may be customizable for novel functions.

Computational design of a modular protein sense-response system.

Sensing and responding to signals is a fundamental ability of living systems, but despite substantial progress in the computational design of new protein structures, there is no general approach for engineering arbitrary new protein sensors. Here, we describe a generalizable computational strategy for designing sensor-actuator proteins by building binding sites de novo into heterodimeric protein-protein interfaces and coupling ligand sensing to modular actuation through split reporters. Using this approach, we designed protein sensors that respond to farnesyl pyrophosphate, a metabolic intermediate in the production of valuable compounds. The sensors are functional in vitro and in cells, and the crystal structure of the engineered binding site closely matches the design model. Our computational design strategy opens broad avenues to link biological outputs to new signals.

A Cell-Penetrating Scorpion Toxin Enables Mode-Specific Modulation of TRPA1 and Pain.

TRPA1 is a chemosensory ion channel that functions as a sentinel for structurally diverse electrophilic irritants. Channel activation occurs through an unusual mechanism involving covalent modification of cysteine residues clustered within an amino-terminal cytoplasmic domain. Here, we describe a peptidergic scorpion toxin (WaTx) that activates TRPA1 by penetrating the plasma membrane to access the same intracellular site modified by reactive electrophiles. WaTx stabilizes TRPA1 in a biophysically distinct active state characterized by prolonged channel openings and low Ca2+ permeability. Consequently, WaTx elicits acute pain and pain hypersensitivity but fails to trigger efferent release of neuropeptides and neurogenic inflammation typically produced by noxious electrophiles. These findings provide a striking example of convergent evolution whereby chemically disparate animal- and plant-derived irritants target the same key allosteric regulatory site to differentially modulate channel activity. WaTx is a unique pharmacological probe for dissecting TRPA1 function and its contribution to acute and persistent pain.

The centrosomin CM2 domain is a multi-functional binding domain with distinct cell cycle roles.

The centrosome serves as the main microtubule-organizing center in metazoan cells, yet despite its functional importance, little is known mechanistically about the structure and organizational principles that dictate protein organization in the centrosome. In particular, the protein-protein interactions that allow for the massive structural transition between the tightly organized interphase centrosome and the highly expanded matrix-like arrangement of the mitotic centrosome have been largely uncharacterized. Among the proteins that undergo a major transition is the Drosophila melanogaster protein centrosomin that contains a conserved carboxyl terminus motif, CM2. Recent crystal structures have shown this motif to be dimeric and capable of forming an intramolecular interaction with a central region of centrosomin. Here we use a combination of in-cell microscopy and in vitro oligomer assessment to show that dimerization is not necessary for CM2 recruitment to the centrosome and that CM2 alone undergoes significant cell cycle dependent rearrangement. We use NMR binding assays to confirm this intramolecular interaction and show that residues involved in solution are consistent with the published crystal structure and identify L1137 as critical for binding. Additionally, we show for the first time an in vitro interaction of CM2 with the Drosophila pericentrin-like-protein that exploits the same set of residues as the intramolecular interaction. Furthermore, NMR experiments reveal a calcium sensitive interaction between CM2 and calmodulin. Although unexpected because of sequence divergence, this suggests that centrosomin-mediated assemblies, like the mammalian pericentrin, may be calcium regulated. From these results, we suggest an expanded model where during interphase CM2 interacts with pericentrin-like-protein to form a layer of centrosomin around the centriole wall and that at the onset of mitosis this population acts as a nucleation site of intramolecular centrosomin interactions that support the expansion into the metaphase matrix.

A 31-residue peptide induces aggregation of tau's microtubule-binding region in cells.

The self-propagation of misfolded conformations of tau underlies neurodegenerative diseases, including Alzheimer's. There is considerable interest in discovering the minimal sequence and active conformational nucleus that defines this self-propagating event. The microtubule-binding region, spanning residues 244-372, reproduces much of the aggregation behaviour of tau in cells and animal models. Further dissection of the amyloid-forming region to a hexapeptide from the third microtubule-binding repeat resulted in a peptide that rapidly forms fibrils in vitro. We show that this peptide lacks the ability to seed aggregation of tau244-372 in cells. However, as the hexapeptide is gradually extended to 31 residues, the peptides aggregate more slowly and gain potent activity to induce aggregation of tau244-372 in cells. X-ray fibre diffraction, hydrogen-deuterium exchange and solid-state NMR studies map the beta-forming region to a 25-residue sequence. Thus, the nucleus for self-propagating aggregation of tau244-372 in cells is packaged in a remarkably small peptide.

Allosteric N-WASP activation by an inter-SH3 domain linker in Nck.

Actin filament networks assemble on cellular membranes in response to signals that locally activate neural Wiskott-Aldrich-syndrome protein (N-WASP) and the Arp2/3 complex. An inactive conformation of N-WASP is stabilized by intramolecular contacts between the GTPase binding domain (GBD) and the C helix of the verprolin-homology, connector-helix, acidic motif (VCA) segment. Multiple SH3 domain-containing adapter proteins can bind and possibly activate N-WASP, but it remains unclear how such binding events relieve autoinhibition to unmask the VCA segment and activate the Arp2/3 complex. Here, we have used purified components to reconstitute a signaling cascade driven by membrane-localized Src homology 3 (SH3) adapters and N-WASP, resulting in the assembly of dynamic actin networks. Among six SH3 adapters tested, Nck was the most potent activator of N-WASP-driven actin assembly. We identify within Nck a previously unrecognized activation motif in a linker between the first two SH3 domains. This linker sequence, reminiscent of bacterial virulence factors, directly engages the N-WASP GBD and competes with VCA binding. Our results suggest that animals, like pathogenic bacteria, have evolved peptide motifs that allosterically activate N-WASP, leading to localized actin nucleation on cellular membranes.

Biophysical methods for identifying fragment-based inhibitors of protein-protein interactions.

Fragment-based lead discovery complements high-throughput screening and computer-aided drug design for the discovery of small-molecule inhibitors of protein-protein interactions. Fragments are molecules with molecular masses ca 280 Da or smaller, and are generally screened using structural or biophysical approaches. Several methods of fragment-based screening are feasible for any soluble protein that can be expressed and purified; specific techniques also have size limitations and/or require multiple milligrams of protein. This chapter describes some of the most common fragment-discovery methods, including surface plasmon resonance, nuclear magnetic resonance, differential scanning fluorimetry, and X-ray crystallography.

A Fragment-Based Ligand Screen Against Part of a Large Protein Machine: The ND1 Domains of the AAA+ ATPase p97/VCP.

The ubiquitous AAA+ ATPase p97 functions as a dynamic molecular machine driving several cellular processes. It is essential in regulating protein homeostasis, and it represents a potential drug target for cancer, particularly when there is a greater reliance on the endoplasmic reticulum-associated protein degradation pathway and ubiquitin-proteasome pathway to degrade an overabundance of secreted proteins. Here, we report a case study for using fragment-based ligand design approaches against this large and dynamic hexamer, which has multiple potential binding sites for small molecules. A screen of a fragment library was conducted by surface plasmon resonance (SPR) and followed up by nuclear magnetic resonance (NMR), two complementary biophysical techniques. Virtual screening was also carried out to examine possible binding sites for the experimental hits and evaluate the potential utility of fragment docking for this target. Out of this effort, 13 fragments were discovered that showed reversible binding with affinities between 140 µM and 1 mM, binding stoichiometries of 1:1 or 2:1, and good ligand efficiencies. Structural data for fragment-protein interactions were obtained with residue-specific [U-(2)H] (13)CH3-methyl-labeling NMR strategies, and these data were compared to poses from docking. The combination of virtual screening, SPR, and NMR enabled us to find and validate a number of interesting fragment hits and allowed us to gain an understanding of the structural nature of fragment binding.

Histone demethylase KDM5A is regulated by its reader domain through a positive-feedback mechanism.

The retinoblastoma binding protein KDM5A removes methyl marks from lysine 4 of histone H3 (H3K4). Misregulation of KDM5A contributes to the pathogenesis of lung and gastric cancers. In addition to its catalytic jumonji C domain, KDM5A contains three PHD reader domains, commonly recognized as chromatin recruitment modules. It is unknown whether any of these domains in KDM5A have functions beyond recruitment and whether they regulate the catalytic activity of the demethylase. Here using biochemical and nuclear magnetic resonance (NMR)-based structural studies, we show that the PHD1 preferentially recognizes unmethylated H3K4 histone tail, product of KDM5A-mediated demethylation of tri-methylated H3K4 (H3K4me3). Binding of unmodified H3 peptide to the PHD1 stimulates catalytic domain-mediated removal of methyl marks from H3K4me3 peptide and nucleosome substrates. This positive-feedback mechanism--enabled by the functional coupling between a reader and a catalytic domain in KDM5A--suggests a model for the spread of demethylation on chromatin.

Elucidating the mechanism of substrate recognition by the bacterial Hsp90 molecular chaperone.

Hsp90 is a conformationally dynamic molecular chaperone known to promote the folding and activation of a broad array of protein substrates ("clients"). Hsp90 is believed to preferentially interact with partially folded substrates, and it has been hypothesized that the chaperone can significantly alter substrate structure as a mechanism to alter the substrate functional state. However, critically testing the mechanism of substrate recognition and remodeling by Hsp90 has been challenging. Using a partially folded protein as a model system, we find that the bacterial Hsp90 adapts its conformation to the substrate, forming a binding site that spans the middle and C-terminal domains of the chaperone. Cross-linking and NMR measurements indicate that Hsp90 binds to a large partially folded region of the substrate and significantly alters both its local and long-range structure. These findings implicate Hsp90's conformational dynamics in its ability to bind and remodel partially folded proteins. Moreover, native-state hydrogen exchange indicates that Hsp90 can also interact with partially folded states only transiently populated from within a thermodynamically stable, native-state ensemble. These results suggest a general mechanism by which Hsp90 can recognize and remodel native proteins by binding and remodeling partially folded states that are transiently sampled from within the native ensemble.

13C NMR metabolomic evaluation of immediate and delayed mild hypothermia in cerebrocortical slices after oxygen-glucose deprivation.

BACKGROUND: Mild brain hypothermia (32°-34°C) after human neonatal asphyxia improves neurodevelopmental outcomes. Astrocytes but not neurons have pyruvate carboxylase and an acetate uptake transporter. C nuclear magnetic resonance spectroscopy of rodent brain extracts after administering [1-C]glucose and [1,2-C]acetate can distinguish metabolic differences between glia and neurons, and tricarboxylic acid cycle entry via pyruvate dehydrogenase and pyruvate carboxylase. METHODS: Neonatal rat cerebrocortical slices receiving a C-acetate/glucose mixture underwent a 45-min asphyxia simulation via oxygen-glucose-deprivation followed by 6 h of recovery. Protocols in three groups of N=3 experiments were identical except for temperature management. The three temperature groups were: normothermia (37°C), hypothermia (32°C for 3.75 h beginning at oxygen--glucose deprivation start), and delayed hypothermia (32°C for 3.75 h, beginning 15 min after oxygen-glucose deprivation start). Multivariate analysis of nuclear magnetic resonance metabolite quantifications included principal component analyses and the L1-penalized regularized regression algorithm known as the least absolute shrinkage and selection operator. RESULTS: The most significant metabolite difference (P<0.0056) was [2-C]glutamine's higher final/control ratio for the hypothermia group (1.75±0.12) compared with ratios for the delayed (1.12±0.12) and normothermia group (0.94±0.06), implying a higher pyruvate carboxylase/pyruvate dehydrogenase ratio for glutamine formation. Least Absolute Shrinkage and Selection Operator found the most important metabolites associated with adenosine triphosphate preservation: [3,4-C]glutamate-produced via pyruvate dehydrogenase entry, [2-C]taurine-an important osmolyte and antioxidant, and phosphocreatine. Final principal component analyses scores plots suggested separate cluster formation for the hypothermia group, but with insufficient data for statistical significance. CONCLUSIONS: Starting mild hypothermia simultaneously with oxygen-glucose deprivation, compared with delayed starting or no hypothermia, has higher pyruvate carboxylase throughput, suggesting that better glial integrity is one important neuroprotection mechanism of earlier hypothermia.

1H nuclear magnetic resonance brain metabolomics in neonatal mice after hypoxia-ischemia distinguished normothermic recovery from mild hypothermia recoveries.

BACKGROUND: Mild brain hypothermia (31-34 °C) after neonatal hypoxia-ischemia (HI) improves neurodevelopmental outcomes in human and animal neonates. Using an asphyxia model with neonatal mice treated with mild hypothermia after HI, we investigated whether (1)H nuclear magnetic resonance (NMR) metabolomics of brain extracts could suggest biomarkers and distinguish different treatments and outcome groups. METHODS: At postnatal day 7 (P7), CD1 mice underwent right carotid artery occlusion, 30 min of HI (8% oxygen), and 3.5 h of either hypothermia (31 °C) or normothermia (37 °C). Whole brains were frozen immediately after HI, immediately after 3.5 h of hypothermia or normothermia treatments, and 24 h later. Perchloric acid extractions of 36 metabolites were quantified by 900 MHz (1)H NMR spectroscopy. Multivariate analyses included principal component analyses (PCA) and a novel regression algorithm. Histological injury was quantified after HI at 5 d. RESULTS: PCA scores plots separated normothermia/HI animals from hypothermia/HI and control animals, but more data are required for multivariate models to be predictive. Loadings plots identified 11 significant metabolites, whereas the regression algorithm identified 6. Histological injury scores were significantly reduced by hypothermia. CONCLUSION: Different treatment and outcome groups are identifiable by (1)H NMR metabolomics in a neonatal mouse model of mild hypothermia treatment of HI.

Design of a phosphorylatable PDZ domain with peptide-specific affinity changes.

Phosphorylation is one of the most common posttranslational modifications controlling cellular protein activity. Here, we describe a combined computational and experimental strategy to design new phosphorylation sites into globular proteins to regulate their functions. We target a peptide recognition protein, the Erbin PDZ domain, to be phosphorylated by cAMP-dependent protein kinase. Comparing the five successful designs to the unsuccessful cases, we find a trade-off between protein stability and the ability to be modified by phosphorylation. In two designs, Erbin's peptide binding function is modified by phosphorylation, where the presence of the phosphate group destabilizes peptide binding. One of these showed an additional switch in specificity by introducing favorable interactions between a designed arginine in the peptide and phosphoserine on the PDZ domain. Because of the diversity of PDZ domains, this opens avenues for the design of related phosphoswitchable domains to create a repertoire of regulatable interaction parts for synthetic biology.

Metabolomics of oxidative stress in recent studies of endogenous and exogenously administered intermediate metabolites.

Aerobic metabolism occurs in a background of oxygen radicals and reactive oxygen species (ROS) that originate from the incomplete reduction of molecular oxygen in electron transfer reactions. The essential role of aerobic metabolism, the generation and consumption of ATP and other high energy phosphates, sustains a balance of approximately 3000 essential human metabolites that serve not only as nutrients, but also as antioxidants, neurotransmitters, osmolytes, and participants in ligand-based and other cellular signaling. In hypoxia, ischemia, and oxidative stress, where pathological circumstances cause oxygen radicals to form at a rate greater than is possible for their consumption, changes in the composition of metabolite ensembles, or metabolomes, can be associated with physiological changes. Metabolomics and metabonomics are a scientific disciplines that focuse on quantifying dynamic metabolome responses, using multivariate analytical approaches derived from methods within genomics, a discipline that consolidated innovative analysis techniques for situations where the number of biomarkers (metabolites in our case) greatly exceeds the number of subjects. This review focuses on the behavior of cytosolic, mitochondrial, and redox metabolites in ameliorating or exacerbating oxidative stress. After reviewing work regarding a small number of metabolites-pyruvate, ethyl pyruvate, and fructose-1,6-bisphosphate-whose exogenous administration was found to ameliorate oxidative stress, a subsequent section reviews basic multivariate statistical methods common in metabolomics research, and their application in human and preclinical studies emphasizing oxidative stress. Particular attention is paid to new NMR spectroscopy methods in metabolomics and metabonomics. Because complex relationships connect oxidative stress to so many physiological processes, studies from different disciplines were reviewed. All, however, shared the common goal of ultimately developing "omics"-based, diagnostic tests to help influence therapies.