Annotating gene function by combining expression data with a modular gene network

MOTIVATION: A promising and reliable approach to annotate gene function is clustering genes not only by using gene expression data but also literature information, especially gene networks. RESULTS: We present a systematic method for gene clustering by combining these totally different two types of data, particularly focusing on network modularity, a global feature of gene networks. Our method is based on learning a probabilistic model, which we call a hidden modular random field in which the relation between hidden variables directly represents a given gene network. Our learning algorithm which minimizes an energy function considering the network modularity is practically time-efficient, regardless of using the global network property. We evaluated our method by using a metabolic network and microarray expression data, changing with microarray datasets, parameters of our model and gold standard clusters. Experimental results showed that our method outperformed other four competing methods, including k-means and existing graph partitioning methods, being statistically significant in all cases. Further detailed analysis showed that our method could group a set of genes into a cluster which corresponds to the folate metabolic pathway while other methods could not. From these results, we can say that our method is highly effective for gene clustering and annotating gene function.     

Effect of photoperiod on clock gene expression and subcellular distribution of PERIOD in the circadian clock neurons of the blow fly Protophormia terraenovae

We examined the effect of photoperiod on the expression of circadian clock genes period (per) and timeless (tim), using quantitative real-time polymerase chain reaction (PCR), and the effect of photoperiod on subcellular distribution of PERIOD (PER), using immunocytochemistry, in the blow fly, Protophormia terraenovae. Under both short-day and long-day conditions, the mRNA levels of per and tim in the brain oscillated, and their peaks and troughs occurred around lights-off and lights-on, respectively. The oscillations persisted even under constant darkness. In the large ventral lateral neurons (l-LN_vs), small ventral lateral neurons (s-LN_vs), dorsal lateral neurons (LN_ds), and medial dorsal neurons (DN_ms), the subcellular distribution of PER-immunoreactivity changed with time. The number of cells with PER-immunoreactivity in the nucleus was highest 12 h after lights-off and lowest 12 h after lights-on, regardless of photoperiod, suggesting that PER nuclear translocation entrains to photoperiod. When temporal changes in the nuclear localization of PER were compared, the neurons could be classified into 2 groups: the l-LN_vs were similar to the s-LN_vs, and the LN_ds were similar to DN_ms. In LN_ds and DN_ms, decreasing rates of the number of cells with PER immunoreactivity in the nucleus per brain from the maximum were large as compared with those in l-LN_vs and s-LN_vs under short-day conditions. These results suggest that photoperiodic information is reflected in the expression patterns of circadian clock genes per and tim and in the subcellular distribution of PER. This observation suggests that the 2 different groups of clock neurons respond to photoperiod in slightly different manners.     

Structure of the transition state analog of medium-chain acyl-CoA dehydrogenase. Crystallographic and molecular orbital studies on the charge-transfer complex of medium-chain acyl-CoA dehydrogenase with 3-thiaoctanoyl-CoA

The flavoenzyme medium-chain acyl-CoA dehydrogenase (MCAD) eliminates the alpha-proton of the substrate analog, 3-thiaoctanoyl-CoA (3S-C8-CoA), to form a charge-transfer complex with deprotonated 3S-C8-CoA. This complex can simulate the metastable reaction intermediate immediately after the alpha-proton elimination of a substrate and before the beta-hydrogen transfer as a hydride, and is therefore regarded as a transition-state analog. The crystalline complex was obtained by co-crystallizing MCAD in the oxidized form with 3S-C8-CoA. The three-dimensional structure of the complex was solved by X-ray crystallography. The deprotonated 3S-C8-CoA was clearly located within the active-site cleft of the enzyme. The arrangement between the flavin ring and deprotonated 3S-C8-CoA is consistent with a charge transfer interaction with the negatively charged acyl-chain of 3S-C8-CoA as an electron donor stacking on the pyrimidine moiety of the flavin ring as an electron acceptor. The structure of the model complex between lumiflavin and the deprotonated ethylthioester of 3-thiabutanoic acid was optimized by molecular orbital calculations. The obtained theoretical structure was essentially the same as that of the corresponding region of the X-ray structure. A considerable amount of negative charge is transferred to the flavin ring system to stabilize the complex by 9.2 kcal/mol. The large stabilization energy by charge transfer probably plays an important role in determining the alignment of the flavin ring with 3S-C8-CoA. The structure of the highest occupied molecular orbital of the complex revealed the electron flow pathway from a substrate to the flavin ring.     

APAF-1-ALT,a novel alternative splicing form of APAF-1, potentially causes impeded ability of undergoing DNA damage-induced apoptosis in the LNCaP human prostate cancer cell line

We have found a novel alternative splicing product of the apoptotic protease activating factor 1 (APAF-1), termed APAF-1-ALT, in the LNCaP human prostate cancer cell line. APAF-1-ALT harbors the caspase recruitment domain and an incomplete CED-4 like/ATPase domain, but lacks the WD-40 repeat units. The LNCaP cell expressed the full-length APAF-1 weakly and APAF-1-ALT rather abundantly, especially after mycoplasma infection. LNCaP underwent a retarded DNA damage-induced apoptosis, which was independent of caspase 9 activity. APAF-1-ALT functioned less effectively in inducing apoptosis than did APAF-1-XL, the full-length APAF-1, in transient transfection. Moreover, APAF-1-ALT interfered with APAF-1-XL's ability to induce apoptosis in transient double transfection experiment. These results indicate that APAF-1-ALT is a molecule that is deficient and impeded for mediating apoptosis and that it may contribute to the resistance to DNA damage-induced treatment observed in LNCaP.     

Resonance Raman study on the flavin in the purple intermediates of D-amino acid oxidase

Resonance Raman (RR) spectra were measured for the purple intermediates of D-amino acid oxidase reconstituted with isotopically labelled FAD's, i.e., [4a-13C]-, [4,10a-13C2]-, [2-13C]-, [5-15N]-, and [1,3-15N2]flavin adenine dinucleotides, and compared with those with the native enzyme. The RR lines around 1605 cm-1 with D-alanine or D-proline as a substrate and at 1548 cm-1 with D-alanine undergo isotopic shifts upon [4a-13C]- and [4,10a-13C2]-labelling. These lines are assigned to the vibrational modes associated with C(10a) = C(4a) - C(4) = O moiety of reduced flavin, providing the first assignment of RR lines of reduced flavin and conclusive evidence that reduced flavin is involved in this intermediate.     

On the structures of flavoprotein D-amino acid oxidase purple intermediates. A resonance Raman study

Resonance Raman (RR) spectra were obtained in H2O or D2O solution for the purple intermediates of D-amino acid oxidase (DAO) with isotopically labeled substrates, i.e., [1-13C]-, [2-13C]-, [3-13C]-, [15N]-, and [3,3,3-D3]alanine; [carboxyl-13C]- and [15N]proline. RR spectra were also measured for the intermediates of DAO reconstituted with isotopically labeled FAD's, i.e., [4a-13C]-, [4,10a-13C2]-, [2-13C]-, [5-15N]-, and [1,3-15N2]FAD in D2O. The isotopic shift of the 1692 cm-1 band upon [15N]- or [2-13C]-substitution of alanine indicates that the band is due to the C = N stretching mode of an imino acid derived from D-alanine, i.e., alpha-iminopropionate. The 1658 cm-1 band with D-proline was also assigned to the C = N stretching mode of an imino acid derived from D-proline, i.e., delta 1-pyrrolidine-2-carboxylate, since the band shifts to 1633 cm-1 upon [15N]-substitution and its stretching frequency is generally found in this frequency region. Since the band shifts to low frequency in D2O, the imino acid should have a protonated imino group such as the C = N+1H form. The intense band at 1363 cm-1 with D-alanine was assigned to a mixing of the CO2- symmetric stretching and CH3 symmetric deformation modes in alpha-iminopropionate, based on the isotope effects. The 1359 cm-1 band with D-proline has probably contributions of CO2- symmetric stretching and CH2 wagging, considering the isotope effects with [carboxyl-13C]proline. The 1359 cm-1 band with D-proline was split into 1371 cm-1 and 1334 cm-1 bands in D2O. As this splitting of the 1359 cm-1 band with D-proline in D2O can not be interpreted only by the replacement of the C = N+1-H proton by deuterium, the carboxylate of the imino acid probably interacts with the enzyme through some proton(s) exchangeable by deuterium(s) in D2O. The bands around 1605 cm-1 which shift upon [4a-13C]- and [4,10a-13C2]-labeling of FAD are derived from a fully reduced flavin, because the isotopic shifts of the band are very different from those of the bands of oxidized or semiquinoid flavin observed near 1605 cm-1.(ABSTRACT TRUNCATED AT 400 WORDS)     

Association-dissociation of the flavoprotein hog kidney D-amino acid oxidase. Determination of the monomer-dimer equilibrium constant and the energetics of subunit association.

The enzyme concentration dependence of spectrophotometric titrations of hog kidney D-amino acid oxidase [EC 1.4.3.3] with p-aminobenzoate was studied. The monomer-dimer equilibrium constant of the oxidized holoenzyme at 25 degrees C was estimated to be 7 X 10(5)M-1 at pH 7.5 and 4X 10(6)M-1 at pH 8.3. The energetics of subunit association are discussed.