Structure-activity relationships derived by machine learning: the use of atoms and their bond connectivities to predict mutagenicity by inductive logic programming.

We present a general approach to forming structure-activity relationships (SARs). This approach is based on representing chemical structure by atoms and their bond connectivities in combination with the inductive logic programming (ILP) algorithm PROGOL. Existing SAR methods describe chemical structure by using attributes which are general properties of an object. It is not possible to map chemical structure directly to attribute-based descriptions, as such descriptions have no internal organization. A more natural and general way to describe chemical structure is to use a relational description, where the internal construction of the description maps that of the object described. Our atom and bond connectivities representation is a relational description. ILP algorithms can form SARs with relational descriptions. We have tested the relational approach by investigating the SARs of 230 aromatic and heteroaromatic nitro compounds. These compounds had been split previously into two subsets, 188 compounds that were amenable to regression and 42 that were not. For the 188 compounds, a SAR was found that was as accurate as the best statistical or neural network-generated SARs. The PROGOL SAR has the advantages that it did not need the use of any indicator variables handcrafted by an expert, and the generated rules were easily comprehensible. For the 42 compounds, PROGOL formed a SAR that was significantly (P < 0.025) more accurate than linear regression, quadratic regression, and back-propagation. This SAR is based on an automatically generated structural alert for mutagenicity.

On the reaction mechanism of l-lactate oxidase: Quantitative structure-activity analysis of the reaction with para-substituted l-mandelates

The rate constants for reduction of the flavoenzyme, l-lactate oxidase, and a mutant (in which alanine 95 is replaced by glycine), by a series of para-substituted mandelates, in both the 2-1H- and 2-2H- forms, have been measured by rapid reaction spectrophotometry. In all cases, significant isotope effects (1H/2H = 3–7) on the rate constants of flavin reduction were found, indicating that flavin reduction is a direct measure of α-C-H bond breakage. The rate constants show only a small influence of the electronic characteristics of the substituents, but show a good correlation when combined with some substituent volume parameters. A surprisingly good correlation is found with the molecular mass of the substrate. The results are compatible with any mechanism in which there is little development of charge in the transition state. This could be a transfer of hydride to the flavin N(5) position or a synchronous mechanism in which the α-C-H is formally abstracted as a H+ while the resulting charge is simultaneously neutralized by another event.

High yield conversion of doxorubicin to 2-pyrrolinodoxorubicin, an analog 500-1000 times more potent: structure-activity relationship of daunosamine-modified derivatives of doxorubicin.

A convenient, high yield conversion of doxorubicin to 3'-deamino-3'-(2''-pyrroline-1''-yl)doxorubicin is described. This daunosamine-modified analog of doxorubicin is 500-1000 times more active in vitro than doxorubicin. The conversion is effected by using a 30-fold excess of 4-iodobutyraldehyde in anhydrous dimethylformamide. The yield is higher than 85%. A homolog of this compound, 3'-deamino-3'-(1'',3''-tetrahydropyridine-1''-yl)doxorubicin, was also synthesized by using 5-iodovaleraldehyde. In this homolog, the daunosamine nitrogen is incorporated into a six- instead of a five-membered ring. This analog was 30-50 times less active than its counterpart with a five-membered ring. A similar structure-activity relationship was found when 3'-deamino-3'-(3''-pyrrolidone-1''-yl)doxorubicin (containing a five-membered ring) and 3'-deamino-3'-(3''-piperidone-1''-yl)doxorubicin (with a six-membered ring) were tested in vitro, the former being 5 times more potent than the latter. To further elucidate structure-activity relationships, 3'-deamino-3'-(pyrrolidine-1''-yl)doxorubicin, 3'-deamino-3'-(isoindoline-2''-yl)doxorubicin, 3'-deamino-3'-(2''-methyl-2''-pyrroline-1''-yl)doxorubicin, and 3'-deamino-3'-(3''-pyrroline-1''-yl)doxorubicin were also synthesized and tested. All the analogs were prepared by using reactive halogen compounds for incorporating the daunosamine nitrogen of doxorubicin into a five- or six-membered ring. These highly active antineoplastic agents can be used for incorporation into targeted cytotoxic analogs of luteinizing hormone-releasing hormone intended for cancer therapy.

Vascular imprints of neuronal activity: Relationships between the dynamics of cortical blood flow, oxygenation, and volume changes following sensory stimulation

Modern functional neuroimaging methods, such as positron-emission tomography (PET), optical imaging of intrinsic signals, and functional MRI (fMRI) utilize activity-dependent hemodynamic changes to obtain indirect maps of the evoked electrical activity in the brain. Whereas PET and flow-sensitive MRI map cerebral blood flow (CBF) changes, optical imaging and blood oxygenation level-dependent MRI map areas with changes in the concentration of deoxygenated hemoglobin (HbR). However, the relationship between CBF and HbR during functional activation has never been tested experimentally. Therefore, we investigated this relationship by using imaging spectroscopy and laser-Doppler flowmetry techniques, simultaneously, in the visual cortex of anesthetized cats during sensory stimulation. We found that the earliest microcirculatory change was indeed an increase in HbR, whereas the CBF increase lagged by more than a second after the increase in HbR. The increased HbR was accompanied by a simultaneous increase in total hemoglobin concentration (Hbt), presumably reflecting an early blood volume increase. We found that the CBF changes lagged after Hbt changes by 1 to 2 sec throughout the response. These results support the notion of active neurovascular regulation of blood volume in the capillary bed and the existence of a delayed, passive process of capillary filling.

Solution structure of DFF40 and DFF45 N-terminal domain complex and mutual chaperone activity of DFF40 and DFF45

Apoptotic DNA fragmentation is mediated by a caspase-activated DNA fragmentation factor (DFF)40. Expression and folding of DFF40 require the presence of DFF45, which also acts as a nuclease inhibitor before DFF40 activation by execution caspases. The N-terminal domains (NTDs) of both proteins are homologous, and their interaction plays a key role in the proper functioning of this two-component system. Here we report that the NTD of DFF45 alone is unstructured in solution, and its folding is induced upon binding to DFF40 NTD. Therefore, folding of both proteins regulates the formation of the DFF40/DFF45 complex. The solution structure of the heterodimeric complex between NTDs of DFF40 and DFF45 reported here shows that the mutual chaperoning includes the formation of an extensive network of intermolecular interactions that bury a hydrophobic cluster inside the interface, surrounded by intermolecular salt bridges.

X-ray structure of clotting factor IXa: active site and module structure related to Xase activity and hemophilia B.

Hereditary deficiency of factor IXa (fIXa), a key enzyme in blood coagulation, causes hemophilia B, a severe X chromosome-linked bleeding disorder afflicting 1 in 30,000 males; clinical studies have identified nearly 500 deleterious variants. The x-ray structure of porcine fIXa described here shows the atomic origins of the disease, while the spatial distribution of mutation sites suggests a structural model for factor X activation by phospholipid-bound fIXa and cofactor VIIIa. The 3.0-A-resolution diffraction data clearly show the structures of the serine proteinase module and the two preceding epidermal growth factor (EGF)-like modules; the N-terminal Gla module is partially disordered. The catalytic module, with covalent inhibitor D-Phe-1I-Pro-2I-Arg-3I chloromethyl ketone, most closely resembles fXa but differs significantly at several positions. Particularly noteworthy is the strained conformation of Glu-388, a residue strictly conserved in known fIXa sequences but conserved as Gly among other trypsin-like serine proteinases. Flexibility apparent in electron density together with modeling studies suggests that this may cause incomplete active site formation, even after zymogen, and hence the low catalytic activity of fIXa. The principal axes of the oblong EGF-like domains define an angle of 110 degrees, stabilized by a strictly conserved and fIX-specific interdomain salt bridge. The disorder of the Gla module, whose hydrophobic helix is apparent in electron density, can be attributed to the absence of calcium in the crystals; we have modeled the Gla module in its calcium form by using prothrombin fragment 1. The arched module arrangement agrees with fluorescence energy transfer experiments. Most hemophilic mutation sites of surface fIX residues occur on the concave surface of the bent molecule and suggest a plausible model for the membrane-bound ternary fIXa-FVIIIa-fX complex structure: fIXa and an equivalently arranged fX arch across an underlying fVIIIa subdomain from opposite sides; the stabilizing fVIIIa interactions force the catalytic modules together, completing fIXa active site formation and catalytic enhancement.

Position- and orientation-independent activity of the Schizosaccharomyces pombe meiotic recombination hot spot M26

The activity of the M26 meiotic recombination hot spot of Schizosaccharomyces pombe depends on the presence of the heptamer 5′-ATGACGT-3′. Transplacement of DNA fragments containing the ade6-M26 gene to other chromosomal loci has previously demonstrated that the heptamer functions in some, but not all, transplacements, suggesting that hot spot activity depends on chromosomal context. In this study, hot spot activity was tested in the absence of gross DNA changes by using site-directed mutagenesis to create the heptamer sequence at novel locations in the genome. When created by mutagenesis of 1–4 bp in the ade6 and ura4 genes, the heptamer was active as a recombination hot spot, in an orientation-independent manner, at all locations tested. Thus, the heptamer sequence can create an active hot spot in other chromosomal contexts, provided that the gross chromosomal structure is not altered; this result is consistent with the hypothesis that a specific higher-order chromatin structure is required for M26 hot spot activity.

ACTIVITY: a database on DNA/RNA sites activity adapted to apply sequence-activity relationships from one system to another

ACTIVITY is a database on DNA/RNA site sequences with known activity magnitudes, measurement systems, sequence-activity relationships under fixed experimental conditions and procedures to adapt these relationships from one measurement system to another. This database deposits information on DNA/RNA affinities to proteins and cell nuclear extracts, cutting efficiencies, gene transcription activity, mRNA translation efficiencies, mutability and other biological activities of natural sites occurring within promoters, mRNA leaders, and other regulatory regions in pro- and eukaryotic genomes, their mutant forms and synthetic analogues. Since activity magnitudes are heavily system-dependent, the current version of ACTIVITY is supplemented by three novel sub-databases: (i) SYSTEM, measurement systems; (ii) KNOWLEDGE, sequence-activity relationships under fixed experimental conditions; and (iii) CROSS_TEST, procedures adapting a relationship from one measurement system to another. These databases are useful in molecular biology, pharmacogenetics, metabolic engineering, drug design and biotechnology. The databases can be queried using SRS and are available through the Web, http://wwwmgs.bionet.nsc.ru/systems/Activity/.

Structural studies of p21Waf1/Cip1/Sdi1 in the free and Cdk2-bound state: conformational disorder mediates binding diversity.

The cyclin-dependent kinase (Cdk) inhibitor p21Waf1/Cip1/Sdi1, important for p53-dependent cell cycle control, mediates G1/S arrest through inhibition of Cdks and possibly through inhibition of DNA replication. Cdk inhibition requires a sequence of approximately 60 amino acids within the p21 NH2 terminus. We show, using proteolytic mapping, circular dichroism spectropolarimetry, and nuclear magnetic resonance spectroscopy, that p21 and NH2-terminal fragments that are active as Cdk inhibitors lack stable secondary or tertiary structure in the free solution state. In sharp contrast to the disordered free state, however, the p21 NH2 terminus adopts an ordered stable conformation when bound to Cdk2, as shown directly by NMR spectroscopy. We have, thus, identified a striking disorder-order transition for p21 upon binding to one of its biological targets, Cdk2. This structural transition has profound implications in light of the ability of p21 to bind and inhibit a diverse family of cyclin-Cdk complexes, including cyclin A-Cdk2, cyclin E-Cdk2, and cyclin D-Cdk4. Our findings suggest that the flexibility, or disorder, of free p21 is associated with binding diversity and offer insights into the role for structural disorder in mediating binding specificity in biological systems. Further, these observations challenge the generally accepted view of proteins that stable secondary and tertiary structure are prerequisites for biological activity and suggest that a broader view of protein structure should be considered in the context of structure-activity relationships.

The mode of action and the structure of a herbicide in complex with its target: binding of activated hydantocidin to the feedback regulation site of adenylosuccinate synthetase.

(+)-Hydantocidin, a recently discovered natural spironucleoside with potent herbicidal activity, is shown to be a proherbicide that, after phosphorylation at the 5' position, inhibits adenylosuccinate synthetase, an enzyme involved in de novo purine synthesis. The mode of binding of hydantocidin 5'-monophosphate to the target enzyme was analyzed by determining the crystal structure of the enzyme-inhibitor complex at 2.6-A resolution. It was found that adenylosuccinate synthetase binds the phosphorylated compound in the same fashion as it does adenosine 5'-monophosphate, the natural feedback regulator of this enzyme. This work provides the first crystal structure of a herbicide-target complex reported to date.

Structure-activity analysis of thanatin, a 21-residue inducible insect defense peptide with sequence homology to frog skin antimicrobial peptides.

Immune challenge to the insect Podisus maculiventris induces synthesis of a 21-residue peptide with sequence homology to frog skin antimicrobial peptides of the brevinin family. The insect and frog peptides have in common a C-terminally located disulfide bridge delineating a cationic loop. The peptide is bactericidal and fungicidal, exhibiting the largest antimicrobial spectrum observed so far for an insect defense peptide. An all-D-enantiomer is nearly inactive against Gram-negative bacteria and some Gram-positive strains but is fully active against fungi and other Gram-positive bacteria, suggesting that more than one mechanism accounts for the antimicrobial activity of this peptide. Studies with truncated synthetic isoforms underline the role of the C-terminal loop and flanking residues for the activity of this molecule for which we propose the name thanatin.

Crystal and molecular structure of paclitaxel (taxol).

Paclitaxel (formerly called taxol), an important anticancer drug, inhibits cell replication by binding to and stabilizing microtubule polymers. As drug-receptor interactions are governed by the three-dimensional stereochemistries of both participants, we have determined the crystal structure of paclitaxel to identify its conformational preferences that may be related to biological activity. The monoclinic crystals contain two independent paclitaxel molecules in the asymmetric unit plus several water and dioxane solvent molecules. Taxane ring conformation is very similar in both paclitaxel molecules and is similar to the taxane ring conformation found in the crystal structure of the paclitaxel analogue docetaxel (formerly called taxotere). The two paclitaxel molecules have carbon-13 side-chain conformations that differ from each other and from that of the corresponding side chain in the docetaxel crystal structure. The carbon-13 side-chain conformation of one paclitaxel molecule is similar to what was proposed from NMR studies done in polar solvents, while that of the other paclitaxel molecule is different and hitherto unobserved. The paclitaxel molecules interact with each other and with solvent atoms through an extensive network of hydrogen bonds. Analysis of the hydrogen-bonding network together with structure-activity studies may suggest which atoms of paclitaxel are important for binding to microtubule receptors.