An attention mechanism-based LSTM network for cancer kinase activity prediction.

Despite the endeavours and achievements made in treating cancers during the past decades, resistance to available kinase drugs continues to be a major problem in cancer therapies. Thus, it is highly desirable to develop computational models that can predict the bioactivity of a compound against cancer kinases. Here, we present a Long Short-Term Memory (LSTM) framework for predicting the activities of lead molecules against seven different kinases. A total of 14,907 compounds from the ChEMBL database were selected for model building. Two different molecular representations, namely, 2D descriptors and MACCS fingerprints were subjected to the LSTM method for the training process. We also successfully integrated an attention mechanism into our model, which helped us to interpret the contribution of chemical features on kinase activity. The attention mechanism extracted the significant chemical moieties more effectively by taking them into consideration during the activity prediction. The recorded accuracies in the test sets for both 2D descriptors and MACCS fingerprints-based models were 0.81 and 0.78, respectively. The receiver operating characteristic curve (ROC)-area under the curve (AUC) score for both models was in the range of 0.8-0.99. The proposed framework can be a good starting point for the development of new cancer kinase drugs.

Curing viruses in Pleurotus ostreatus by growth on a limited nutrient medium containing cAMP and rifamycin.

Oyster mushroom spherical virus (OMSV) and oyster mushroom isometric virus (OMIV) are the causative agents of a fruiting body deformation disease in the edible mushroom Pleurotus ostreatus. The curing of these mycoviruses was facilitated by a serial transfer of infected mycelia onto a limited nutrient medium containing 1mM of cAMP and 75 µg/ml of rifamycin (cAMP-rifamycin plate). The mycelia were grown on cAMP-rifamycin plates for 5 successive passages. ELISA and RT-PCR showed that the amount of mycoviruses inside the mycelia decreased significantly with increasing numbers of passages. The mycelia became free of viruses after 5 successive passages. Cultivation of the virus-cured mycelia on a mushroom compost medium produced a normal harvest, whereas the spawn infected with viruses failed to produce any fruiting bodies.

Isolation of a novel mycovirus OMIV in Pleurotus ostreatus and its detection using a triple antibody sandwich-ELISA.

A novel mycovirus was isolated from a diseased mushroom, Pleurotus ostreatus, using a purification procedure involving polyethylene glycol (PEG)-NaCl precipitation, differential centrifugation, and equilibrium centrifugation in a CsCl gradient. The virion was a 43 nm isometric virus encapsulating double-stranded RNA (dsRNA) genome of 2.1, 2.0, 1.9, and 1.7 kbp with a coat protein (CP) of 58 kDa. The new mycovirus was named Oyster Mushroom Isometric Virus (OMIV). A triple antibody sandwich-ELISA (TAS-ELISA) system was constructed to detect OMIV in the mushroom using an anti-OMIV mouse monoclonal antibody and an anti-OMIV rabbit polyclonal serum. The TAS-ELISA system was sensitive enough to allow detection of OMIV in the mushroom with the naked eye. It detected successfully virus particles from 0.6 mg of diseased tissue as well as 0.4 microg/ml purified virus preparation.

Gene structure and expression of the mouse adipocyte enhancer-binding protein.

The adipocyte enhancer-binding protein (AEBP1) is a transcriptional repressor with carboxypeptidase activity. AEBP1 expression is down-regulated during adipogenesis. Aortic carboxypeptidase-like protein (ACLP) is a non-nuclear isoform of AEBP1 that has an N-terminal extension of 380 amino acids. ACLP expression is up-regulated during vascular smooth muscle cell differentiation. To gain insight into the regulation of AEBP1 isoform expression, we have determined the structural organization of the mouse AEBP1 gene. This gene extends over 10 kb, has 21 exons, and gives rise to two mRNAs (AEBP1 and ACLP). The 9th intron is retained in the mature AEBP1 transcript. Thus, ACLP encodes an additional 380 amino acids N-terminal to the first ATG codon of AEBP1 which is located in exon 10. RT-PCR experiments showed that both transcripts are expressed ubiquitously in all mouse tissues examined, while Western blot analysis suggested that expression is translationally regulated. Our results provide evidence that two isoforms of AEBP1 with very different functions are produced by an alternative splicing mechanism. This represents a new example of regulation of subcellular localization by protein truncation.

Regulation of adipogenesis by a transcriptional repressor that modulates MAPK activation.

Mitogen-activated protein kinase (MAPK) is required for cell growth and cell differentiation. In adipogenesis, MAPK activation opposes the differentiation process. The regulatory mechanisms or the cellular factors that regulate the switch between growth and differentiation in the adipogenic lineage have been largely unelucidated. We show here that AEBP1, a transcriptional repressor that is down-regulated during adipogenesis, complexes and protects MAPK from its specific phosphatase in mammalian cells. We further show evidence that the modulation of MAPK activation by AEBP1 is a biologically relevant process in adipogenesis. Our results suggest that modulation of MAPK activation by the protective effect of AEBP1 may constitute a critical part in the determination between cell growth and differentiation in the adipogenic lineage. The proposed mode of action by which a transcription factor regulates MAPK activation is novel.

Structure and function of the tryptophan synthase alpha(2)beta(2) complex. Roles of beta subunit histidine 86.

To probe the structural and functional roles of active-site residues in the tryptophan synthase alpha(2)beta(2) complex from Salmonella typhimurium, we have determined the effects of mutation of His(86) in the beta subunit. His(86) is located adjacent to beta subunit Lys(87), which forms an internal aldimine with the pyridoxal phosphate and catalyzes the abstraction of the alpha-proton of L-serine. The replacement of His(86) by leucine (H86L) weakened pyridoxal phosphate binding approximately 20-fold and abolished the circular dichroism signals of the bound coenzyme and of a reaction intermediate. Correlation of these results with previous crystal structures indicates that beta-His(86) plays a structural role in binding pyridoxal phosphate and in stabilizing the correct orientation of pyridoxal phosphate in the active site of the beta subunit. The H86L mutation also altered the pH profiles of absorbance and fluorescence signals and shifted the pH optimum for the synthesis of L-tryptophan from pH 7.5 to 8.8. We propose that the interaction of His(86) with the phosphate of pyridoxal phosphate and with Lys(87) lowers the pK(a) of Lys(87) in the wild-type alpha(2)beta(2) complex and thereby facilitates catalysis by Lys(87) in the physiological pH range.

Catalytic mechanism of the tryptophan synthase alpha(2)beta(2) complex. Effects of pH, isotopic substitution, and allosteric ligands.

The mechanism of the tryptophan synthase alpha(2)beta(2) complex from Salmonella typhimurium is explored by determining the effects of pH, of temperature, and of isotopic substitution on the pyridoxal phosphate-dependent reaction of L-serine with indole to form L-tryptophan. The pH dependence of the kinetic parameters indicates that three ionizing groups are involved in substrate binding and catalysis with pK(a)1 = 6.5, pK(a)2 = 7.3, and pK(a)3 = 8.2-9. A significant primary isotope effect (approximately 3.5) on V and V/K is observed at low pH (pH 7), but not at high pH (pH 9), indicating that the base that accepts the alpha-proton (betaLys-87) is protonated at low pH, slowing the abstraction of the alpha-proton and making this step at least partially rate-limiting. pK(a)2 is assigned to betaLys-87 on the basis of the kinetic isotope effect results and of the observation that the competitive inhibitors glycine and oxindolyl-L-alanine display single pK(i) values of 7.3. The residue with this pK(a) (betaLys-87) must be unprotonated for binding glycine or oxindolyl-L-alanine, and, by inference, L-serine. Investigations of the temperature dependence of the pK(a) values support the assignment of pK(a)2 to betaLys-87 and suggest that the ionizing residue with pK(a)1 could be a carboxylate, possibly betaAsp-305, and that the residue associated with a conformational change at pK(a)3 may be betaLys-167. The occurrence of a closed to open conformational conversion at high pH is supported by investigations of the effects of pH on reaction specificity and on the equilibrium distribution of enzyme-substrate intermediates.

Enzymic characterization of a novel member of the regulatory B-like carboxypeptidase with transcriptional repression function: stimulation of enzymic activity by its target DNA.

The adipocyte-enhancer binding protein (AEBP) 1 is a novel transcriptional repressor with carboxypeptidase (CP) activity. AEBP1 binds to a regulatory sequence (termed adipocyte enhancer 1, AE-1) located in the proximal promoter region of the adipose P2 (aP2) gene, which encodes the adipocyte fatty-acid binding protein. Sequence comparisons and kinetic studies using known carboxypeptidase substrates, activators and inhibitors have characterized AEBP1 as a member of the regulatory B-like CP family. Significantly, the inherent CP activity of AEBP1 is stimulated by the AE-1 sequence. Our results indicate that AEBP1 is activated by a novel mechanism, wherby the direct binding of DNA enhances its protease activity. These results represent the first demonstration of DNA-mediated regulation of CP activity.

Transcriptional regulation by the gamma5 subunit of a heterotrimeric G protein during adipogenesis.

The adipocyte enhancer-binding protein (AEBP1) is a novel transcriptional repressor with carboxypeptidase activity. A two-hybrid screen was conducted to identify components of AEBP1 that might be important in regulating its activity. The gamma5 subunit of a heterotrimeric G protein was shown to bind specifically to AEBP1 and to attenuate its transcriptional repression activity. Adipogenic stimulation selectively decreased the Ggamma5 level and enhanced the transcriptional repression activity of AEBP1 during mitotic clonal expansion at the onset of adipogenesis. Thus, the actions of Ggamma5 and AEBP1 are directly linked, which could provide the basis for the regulation of transcription at the onset of differentiation. This report shows that a signal-transducing molecule is involved, by direct protein-protein interaction, in the regulation of transcription during adipogenesis.

Cloning and characterization of a novel zinc finger transcriptional repressor. A direct role of the zinc finger motif in repression.

We have identified a novel transcriptional repressor, AEBP2, that binds to a regulatory sequence (termed AE-1) located in the proximal promoter region of the aP2 gene that encodes the adipose fatty acid-binding protein. Sequence analysis of AEBP2 cDNA revealed that it encodes a protein containing three Gli-Krüppel (Cys2-His2)-type zinc fingers. Northern blot analysis revealed two transcripts (4.5 and 3.5 kilobases) which were ubiquitously expressed in every mouse tissue examined. In co-transfection assays, AEBP2 repressed transcription from the homologous aP2 promoter containing multiple copies of the AE-1 sequence. Moreover, a chimeric construct encoding a fusion AEBP2 protein with the Gal4 DNA-binding domain was able to repress the transcriptional activity of a heterologous promoter containing the Gal4-binding sequence. The transcriptional repression function of AEBP2 was completely abolished when one of the conserved histidine residues and a flanking serine residue in the middle zinc finger were replaced with an arginine residue. The defective transcriptional repression function of the mutant derivative was due neither to lack of expression nor to a failure to localize to the nucleus. Moreover, both the wild-type and mutant derivative of either the histidine-tagged recombinant AEBP2 proteins or the in vitro translated Gal4-AEBP2 fusion proteins were equally able to bind to the target DNA. These results suggest that a portion of the zinc finger structure may play a direct role in transcriptional repression function, but not in DNA binding.

Tryptophan synthase mutations that alter cofactor chemistry lead to mechanism-based inactivation.

Mutations in the pyridoxal phosphate binding site of the tryptophan synthase beta subunit (S377D and S377E) alter cofactor chemistry [Jhee, K.-H., et al. (1998) J. Biol. Chem. 273, 11417-11422]. We now report that the S377D, S377E, and S377A beta2 subunits form alpha2 beta2 complexes with the alpha subunit and activate the alpha subunit-catalyzed cleavage of indole 3-glycerol phosphate. The apparent Kd for dissociation of the alpha and beta subunits is unaffected by the S377A mutation but is increased up to 500-fold by the S377D and S377E mutations. Although the three mutant alpha2 beta2 complexes exhibit very low activities in beta elimination and beta replacement reactions catalyzed at the beta site in the presence of Na+, the activities and spectroscopic properties of the S377A alpha2 beta2 complex are partially repaired by addition of Cs+. The S377D and S377E alpha2 beta2 complexes, unlike the wild-type and S377A alpha2 beta2 complexes and the mutant beta2 subunits, undergo irreversible substrate-induced inactivation by L-serine or by beta-chloro-L-alanine. The rates of inactivation (kinact) are similar to the rates of catalysis (kcat). The partition ratios are very low (kcat/kinact = 0.25-3) and are affected by alpha subunit ligands and monovalent cations. The inactivation product released by alkali was shown by HPLC and by fluorescence, absorption, and mass spectroscopy to be identical to a compound previously synthesized from pyridoxal phosphate and pyruvate. We suggest that alterations in the cofactor chemistry that result from the engineered Asp377 in the active site of the beta subunit may promote the mechanism-based inactivation.

Characterization of three transcriptional repressor sites within the 3' untranslated region of the rat serine protease inhibitor 2.3 gene.

The activity of the rat serine protease inhibitor 2.3 gene (spi 2.3) is controlled by several positive promoter elements [Simar-Blanchet, A.-E., Paul, C., Mercier, L. & Le Cam, A. (1996) Eur. J. Biochem. 236, 638-648] and a negative element located in the 3' untranslated gene region (3' UTR) [Le Cam, A. & Legraverend, C. (1996) Eur. J. Biochem. 231, 620-627]. In the present studies, we dissected the 348-bp spi 2.3 3' UTR silencer to precisely define repressor sites and look for specifically interacting proteins. Three short elements referred to as A (nucleotides 1751-1776 in the cDNA), B (nucleotides 1812-1827) and C (nucleotides 1958-1974) sites repressed transcription from the homologous spi 2.3 promoter as well as from a heterologous minimal promoter containing the spi GAGA box enhancer. All three sites harbor a (TTTC) motif whose mutation affected silencer activity that was also dependent on flanking sequences. Those sites share the (TTTC) motif and a CCAAT/enhancer-binding-protein(C/EBP)-binding site with a fatty-acid-binding-protein gene promoter element shown to interact specifically with a transcriptional repressor [He, G. P., Muise, A., Wu Li, A. & Ro, H.-S. (1995) Nature 378, 92-96]. This repressor is however unlikely to mediate spi 2.3 3' UTR silencer action since it was not detected in rat hepatocytes. In vitro footprinting of the spi 2.3 3' UTR silencer region revealed a strong interaction with liver nuclear proteins. Among the six identified footprints, three of them (F-II, FIII and F-IV) bound C/EBPs and mapped in regions harboring the repressor function. Binding of C/EBPs to all three spi 2.3 3' UTR repressor sites, although rather weak, was confirmed by electrophoretic mobility shift assays that otherwise failed to reveal specific interactions with other liver nuclear proteins in vitro. However, none of the most largely liver expressed C/EBP species (i.e. alpha, beta and delta) activated the spi 2.3 3' UTR silencer function in NIH 3T3 cells, suggesting that binding of those transcription factors did not mediate the transcriptional repression.

Primary structure, sequence analysis, and expression of the thermostable D-hydantoinase from Bacillus stearothermophilus SD1.

The gene coding for the thermostable D-hydantoinase from the thermophilic bacterium Bacillus stearothermophilus SD1 was cloned and its nucleotide sequence was completely determined. The D-hydantoinase protein showed considerable amino acid sequence homology (20-28%) with other hydantoinases and functionally related allantoinases and dihydroorotases. Strikingly the sequence of the enzyme from B. stearothermophilus SD1 exhibited greater than 89% identity with hydantoinases from thermophilic bacteria. Despite the extremely high amino acid homology among the hydantoinases from thermophiles, the C-terminal regions of the enzymes were completely different in both sequence and predicted secondary structure, implying that the C-terminal region plays an important role in determining the biochemical properties of the enzymes. Alignment of the sequence of the D-hydantoinase from B. stearothermophilus SD1 with those of other functionally related enzymes revealed four conserved regions, and five histidines and an acidic residue were found to be conserved, suggesting a close evolutionary relationship between all these enzymes.

Site-directed mutagenesis of the amino acid residues in beta-strand III [Val30-Val36] of D-amino acid aminotransferase of Bacillus sp. YM-1.

The beta-strand III formed by amino acid residues Val30-Val36 is located across the active site of the thermostable D-amino acid aminotransferase (D-AAT) from thermophilic Bacillus sp. YM-1, and the odd-numbered amino acids (Tyr31, Val33, Lys35) in the strand are revealed to be directed toward the active site. Interestingly, Glu32 is also directed toward the active site. We first investigated the involvement of these amino acid residues in catalysis by alanine scanning mutagenesis. The Y31A and E32A mutant enzymes showed a marked decrease in k(cat) value, retaining less than 1% of the wild-type enzyme activity. The k(cat) values of V33A and K35A were changed slightly, but the Km of K35A for alpha-ketoglutarate was increased to 35.6 mM, compared to the Km value of 2.5 mM for the wild-type enzyme. These results suggested that the positive charge at Lys35 interacted electrostatically with the negative charge at the side chain of alpha-ketoglutarate. Site-directed mutagenesis of the Glu32 residue was conducted to demonstrate the role of this residue in detail. From the kinetic and spectral characteristics of the Glu32-substituted enzymes, the Glu32 residue seemed to interact with the positive charge at the Schiff base formed between the aldehyde group of pyridoxal 5'-phosphate (PLP) and the epsilon-amino group of the Lys145 residue.

A eukaryotic transcriptional repressor with carboxypeptidase activity.

Adipocyte differentiation involves the transcriptional activation of several genes in triglyceride metabolism, including the adipose P2 (aP2 or 422) gene that encodes the adipocyte lipid-binding protein ALBP. Within the mouse aP2 promoter region, the AE-1 sequence functions as either a positive or a negative element in the regulation of aP2 gene expression. The AE-1 sequence is the binding site for the positive murine (3T3) adipocyte factor C/EBP-alpha, several human preadipocyte factors, and a 3T3 preadipocyte factor(s) that has been implicated as a repressor of aP2 gene expression. Here we report the cloning of new complementary DNAs that encode the 3T3 preadipocyte factor (termed AEBP1) and demonstrate that AEBP1 expression is abolished during adipocyte differentiation. Furthermore, we show that an activity of a carboxypeptidase associated with AEBP1 is important in the transcriptional repression function of AEBP1. Thus AEBP1 might represent a new type of transcription factor that regulates transcription by cleavage of factors involved in transcription.

Molecular characterization of the mouse ribosomal protein S24 multigene family: a uniquely expressed intron-containing gene with cell-specific expression of three alternatively spliced mRNAs.

A family of 16 genes encoding the mouse ribosomal protein S24 was identified, and four members from this family were cloned. A single expressed intron-containing S24 gene (termed mrpS24) and one pseudogene (mrpS24p) were completely sequenced and characterized. The mrpS24 gene has seven exons and six introns spanning over 5.1 x 10(3) nucleotides (nt). The cap site of S24 was mapped to a G residue four nt upstream of a polypyrimidine tract and 15 nt downstream of a TATA-like (TATGA) element. The 5' region (-325 to +33) of the mrpS24 gene has a functional promoter that was able to express the fused chloramphenicol acetyltransferase (CAT) reporter gene. Two different forms of mouse S24 cDNA clones were previously isolated. Sequence analysis showed that one of these cDNA clones (termed S24a) lacks the entire exon V sequence (18 nt), and the deduced amino acid sequence is missing a C-terminal lysine residue encoded by the other cDNA (S24b). The pseudogene mrpS24p is flanked by an 11-bp direct repeat, and its sequence is almost identical to the S24 cDNA sequence, but it lacks two mini-exons, V and VI (20 nt), as in the cases of the human and rat S24 cDNAs. RT-PCR experiments demonstrated the existence of a third form (S24c) that similarly lacks both of the mini-exons, and suggested that different species of S24 mRNA might arise from alternative splicing of the mini-exons V and VI. Northern blot analysis showed that S24 expression is down- and up-regulated during adipocyte differentiation and in cellular transformation, respectively. RNase protection assays and RT-PCR experiments suggested that these cell-specific changes of S24 mRNA levels are mainly due to fluctuations in S24c mRNA level. Our results provide the first indication that a ribosomal protein gene is regulated by alternative usage of two mini-exons in a cell-specific manner.

A new biosensor for specific determination of sucrose using an oxidoreductase of Zymomonas mobilis and invertase.

A new biosensor for specific determination of sucrose was developed using an oxidoreductase of Zymomonas mobilis and invertase. Cells of Z. mobilis were permeabilized with toluene in order to utilize the enzymes of glucose-fructose oxidoreductase and gluconolactonase inside the intact cells. Permeabilized cells and invertase were coimmobilized in a gelatin membrane, and a whole cell enzyme electrode was constructed by fixing the membrane on a pH electrode. The production of hydrogen ion was detected using the biosensor-connected microcomputer, and the concentration of sucrose was determined by using both the initial rate and the steady-state methods. Optimum conditions for biosensor response were pH 6.2 and temperature 35 degrees C. The effect of interfering compounds on the electrode response was investigated, and the interference by various sugars was eliminated by determining sucrose concentration using the steady-state method. The biosensor developed is simple and reproducible, and the calibration curve for sucrose is linear up to 70 g/L.

The C/EBP-binding region and adjacent sites regulate expression of the adipose P2 gene in human preadipocytes.

Human preadipocytes contain nuclear factors that specifically bind to the AE-1 sequence, previously demonstrated as an enhancer element in the regulation of adipose P2 gene expression during 3T3 adipose differentiation. By transient transfection and in vivo competition experiments, the trans-acting factors were found to bind either to the C/EBP recognition site in the AE-1 sequence and act as a negative regulator or to the adjacent site (termed 3' AE-1) and act as a positive regulator of adipose P2 gene activity in human preadipocytes.

A direct role for C/EBP and the AP-I-binding site in gene expression linked to adipocyte differentiation.

Adipocyte differentiation is accompanied by the transcriptional activation of many new genes, including the gene encoding adipocyte P2 (aP2), an intracellular lipid-binding protein. Using specific deletions and point mutations, we have shown that at least two distinct sequence elements in the aP2 promoter contribute to the expression of the chloramphenicol acetyltransferase gene in chimeric constructions transfected into adipose cells. An AP-I site at -120, shown earlier to bind Jun- and Fos-like proteins, serves as a positive regulator of chloramphenicol acetyltransferase gene expression in adipocytes but is specifically silenced by adjacent upstream sequences in preadipocytes. Sequences upstream of the AP-I site at -140 (termed AE-1) can function as an enhancer in both cell types when linked to a viral promoter but can stimulate expression only in fat cells in the intact aP2 promoter. The AE-1 sequence binds an adipocyte protein identical or very closely related to an enhancer-binding protein (C/EBP) that has been previously implicated in the regulation of several liver-specific genes. A functional role for C/EBP in the regulation of the aP2 gene is indicated by the facts that C/EBP mRNA is induced during adipocyte differentiation and the aP2 promoter is transactivated by cotransfection of a C/EBP expression vector into preadipose cells. These results indicate that sequences that bind C/EBP and the Fos-Jun complex play major roles in the expression of the aP2 gene during adipocyte differentiation and demonstrate that C/EBP can directly regulate cellular gene expression.