Targeting a Cryptic Pocket in a Protein-Protein Contact by Disulfide-Induced Rupture of a Homodimeric Interface.

Interference with protein-protein interfaces represents an attractive as well as challenging option for therapeutic intervention and drug design. The enzyme tRNA-guanine transglycosylase, a target to fight Shigellosis, is only functional as a homodimer. Although we previously produced monomeric variants by site-directed mutagenesis, we only crystallized the functional dimer, simply because upon crystallization the local protein concentration increases and favors formation of the dimer interface, which represents an optimal and highly stable packing of the protein in the solid state. Unfortunately, this prevents access to structural information about the interface geometry in its monomeric state and complicates the development of modulators that can interfere with and prevent dimer formation. Here, we report on a cysteine-containing protein variant in which, under oxidizing conditions, a disulfide linkage is formed. This reinforces a novel packing geometry of the enzyme. In this captured quasi-monomeric state, the monomer units arrange in a completely different way and, thus, expose a loop-helix motif, originally embedded into the old interface, now to the surface. The motif adopts a geometry incompatible with the original dimer formation. Via the soaking of fragments into the crystals, we identified several hits accommodating a cryptic binding site next to the loop-helix motif and modulated its structural features. Our study demonstrates the druggability of the interface by breaking up the homodimeric protein using an introduced disulfide cross-link. By rational concepts, we increased the potency of these fragments to a level where we confirmed their binding by NMR to a nondisulfide-linked TGT variant. The idea of intermediately introducing a disulfide linkage may serve as a general concept of how to transform a homodimer interface into a quasi-monomeric state and give access to essential structural and design information.

Identification and in silico structural analysis of Gallus gallus protein arginine methyltransferase 4 (PRMT4).

Protein arginine methyltransferase 4 (PRMT4) is an essential epigenetic regulator of fundamental and conserved processes during vertebrate development, such as pluripotency and differentiation. Surprisingly, PRMT4 homologs have been identified in nearly all vertebrate classes except the avian genome. This raises the possibility that in birds PRMT4 functions are taken over by other PRMT family members. Here, we reveal the existence of a bona fide PRMT4 homolog in the chicken, Gallus gallus. Using a biochemical approach, we initially purified a putative chicken PRMT4 protein and thus provided the first evidence for the presence of an endogenous PRMT4-specific enzymatic activity toward histone H3 arginine 17 (H3R17) in avian cells. We then isolated a G. gallus PRMT4 (ggPRMT4) transcript encompassing the complete open reading frame. Recombinant ggPRMT4 possesses intrinsic methyltransferase activity toward H3R17. CRISPR/Cas9-mediated deletion of ggPRMT4 demonstrated that the transcript identified here encodes avian PRMT4. Combining protein-protein docking and homology modeling based on published crystal structures of murine PRMT4, we found a strong structural similarity of the catalytic core domain between chicken and mammalian PRMT4. Strikingly, in silico structural comparison of the N-terminal Pleckstrin homology (PH) domain of avian and murine PRMT4 identified strictly conserved amino acids that are involved in an interaction interface toward the catalytic core domain, facilitating for the first time a prediction of the relative spatial arrangement of these two domains. Our novel findings are particularly exciting in light of the essential function of the PH domain in substrate recognition and methylation by PRMT4.

Price for Opening the Transient Specificity Pocket in Human Aldose Reductase upon Ligand Binding: Structural, Thermodynamic, Kinetic, and Computational Analysis.

Insights into the thermodynamic and kinetic signature of the transient opening of a protein-binding pocket resulting from accommodation of suitable substituents attached to a given parent ligand scaffold are presented. As a target, we selected human aldose reductase, an enzyme involved in the development of late-stage diabetic complications. To recognize a large scope of substrate molecules, this reductase opens a transient specificity pocket. The pocket-opening step was studied by X-ray crystallography, microcalorimetry, and surface plasmon resonance using a narrow series of 2-carbamoyl-phenoxy-acetic acid derivatives. Molecular dynamics simulations suggest that pocket opening occurs only once an appropriate substituent is attached to the parent scaffold. Transient pocket opening of the uncomplexed protein is hardly recorded. Hydration-site analysis suggests that up to five water molecules entering the opened pocket cannot stabilize this state. Sole substitution with a benzyl group stabilizes the opened state, and the energetic barrier for opening is estimated to be ∼5 kJ/mol. Additional decoration of the pocket-opening benzyl substituent with a nitro group results in a huge enthalpy-driven potency increase; on the other hand, an isosteric carboxylic acid group reduces the potency 1000-fold, and binding occurs without pocket opening. We suggest a ligand induced-fit mechanism for the pocket-opening step, which, however, does not represent the rate-determining step in binding kinetics.

High resolution crystal structure of Clostridium propionicum ß-alanyl-CoA:ammonia lyase, a new member of the "hot dog fold" protein superfamily.

Clostridium propionicum is the only organism known to ferment ß-alanine, a constituent of coenzyme A (CoA) and the phosphopantetheinyl prosthetic group of holo-acyl carrier protein. The first step in the fermentation is a CoA-transfer to ß-alanine. Subsequently, the resulting ß-alanyl-CoA is deaminated by the enzyme ß-alanyl-CoA:ammonia lyase (Acl) to reversibly form ammonia and acrylyl-CoA. We have determined the crystal structure of Acl in its apo-form at a resolution of 0.97 Å as well as in complex with CoA at a resolution of 1.59 Å. The structures reveal that the enyzme belongs to a superfamily of proteins exhibiting a so called "hot dog fold" which is characterized by a five-stranded antiparallel ß-sheet with a long α-helix packed against it. The functional unit of all "hot dog fold" proteins is a homodimer containing two equivalent substrate binding sites which are established by the dimer interface. In the case of Acl, three functional dimers combine to a homohexamer strongly resembling the homohexamer formed by YciA-like acyl-CoA thioesterases. Here, we propose an enzymatic mechanism based on the crystal structure of the Acl·CoA complex and molecular docking.

Identification of a novel polyfluorinated compound as a lead to inhibit the human enzymes aldose reductase and AKR1B10: structure determination of both ternary complexes and implications for drug design.

Aldo-keto reductases (AKRs) are mostly monomeric enzymes which fold into a highly conserved (α/ß)8 barrel, while their substrate specificity and inhibitor selectivity are determined by interaction with residues located in three highly variable external loops. The closely related human enzymes aldose reductase (AR or AKR1B1) and AKR1B10 are of biomedical interest because of their involvement in secondary diabetic complications (AR) and in cancer, e.g. hepatocellular carcinoma and smoking-related lung cancer (AKR1B10). After characterization of the IC50 values of both AKRs with a series of polyhalogenated compounds, 2,2',3,3',5,5',6,6'-octafluoro-4,4'-biphenyldiol (JF0064) was identified as a lead inhibitor of both enzymes with a new scaffold (a 1,1'-biphenyl-4,4'-diol). An ultrahigh-resolution X-ray structure of the AR-NADP(+)-JF0064 complex has been determined at 0.85 Šresolution, allowing it to be observed that JF0064 interacts with the catalytic residue Tyr48 through a negatively charged hydroxyl group (i.e. the acidic phenol). The non-competitive inhibition pattern observed for JF0064 with both enzymes suggests that this acidic hydroxyl group is also present in the case of AKR1B10. Moreover, the combination of surface lysine methylation and the introduction of K125R and V301L mutations enabled the determination of the X-ray crystallographic structure of the corresponding AKR1B10-NADP(+)-JF0064 complex. Comparison of the two structures has unveiled some important hints for subsequent structure-based drug-design efforts.

Ligand binding stepwise disrupts water network in thrombin: enthalpic and entropic changes reveal classical hydrophobic effect.

Well-ordered water molecules are displaced from thrombin's hydrophobic S3/4-pocket by P3-varied ligands (Gly, d-Ala, d-Val, d-Leu to d-Cha with increased hydrophobicity and steric requirement). Two series with 2-(aminomethyl)-5-chlorobenzylamide and 4-amidinobenzylamide at P1 were examined by ITC and crystallography. Although experiencing different interactions in S1, they display almost equal potency. For both scaffolds the terminal benzylsulfonyl substituent differs in binding, whereas the increasingly bulky P3-groups address S3/4 pocket similarly. Small substituents leave the solvation pattern unperturbed as found in the uncomplexed enzyme while increasingly larger ones stepwise displace the waters. Medium-sized groups show patterns with partially occupied waters. The overall 40-fold affinity enhancement correlates with water displacement and growing number of van der Waals contacts and is mainly attributed to favorable entropy. Both Gly derivatives deviate from the series and adopt different binding modes. Nonetheless, their thermodynamic signatures are virtually identical with the homologous d-Ala derivatives. Accordingly, unchanged thermodynamic profiles are no reliable indicator for conserved binding modes.