Saccharomyces cerevisiae normally cannot assimilate mannitol, a promising brown macroalgal carbon source for bioethanol production. The molecular basis of this inability remains unknown. We found that cells capable of assimilating mannitol arose spontaneously from wild-type S. cerevisiae during prolonged culture in mannitol-containing medium. Based on microarray data, complementation analysis, and cell growth data, we demonstrated that acquisition of mannitol-assimilating ability was due to spontaneous mutations in the genes encoding Tup1 or Cyc8, which constitute a general corepressor complex that regulates many kinds of genes. We also showed that an S. cerevisiae strain carrying a mutant allele of CYC8 exhibited superior salt tolerance relative to other ethanologenic microorganisms; this characteristic would be highly beneficial for the production of bioethanol from marine biomass. Thus, we succeeded in conferring the ability to assimilate mannitol on S. cerevisiae through dysfunction of Tup1-Cyc8, facilitating production of ethanol from mannitol. Macroalgae, consisting of green, red, and brown algae, are promising sources of biofuels for several reasons: (i) macroalgae are more productive than land crops; (ii) arable land is not required for algal cultivation, obviating the necessity for irrigation, fertilizer, etc.; and (iii) macroalgae contain no lignin (1-4). Both red and brown algae contain high levels of carbohydrates, and a method for producing biofuel from these carbohydrates would be of tremendous economic and environmental benefit.Brown macroalgae contain up to 33% (wt/wt [dry weight]) mannitol, which is the sugar alcohol corresponding to mannose and a promising carbon source for bioethanol production (1, 5, 6). Although some bacteria, such as Escherichia coli and Zymobacter palmae, can assimilate mannitol, i.e., utilize mannitol and produce ethanol (6, 7), bacteria are generally sensitive to ethanol, as well as, several other growth-inhibitory compounds. Z. palmae and E. coli KO11 can produce ca. 1.3% (wt/vol) and 2.6% (wt/vol) ethanol from 3.8% (wt/vol) and 9.0% (wt/vol) mannitol, respectively; however, both strains are sensitive to 5% (wt/vol) ethanol (8,9). Yeast is currently considered to have several advantages over ethanologenic bacteria, including high tolerance to ethanol and inhibitory compounds (10). Several yeast strains, such as Pichia angophorae and Saccharomyces paradoxus NBRC0259-3, can produce ethanol from mannitol (8, 11). However, compared to the well-characterized model organism Saccharomyces cerevisiae, the host-vector systems of these yeasts are not well equipped, and their genetics and physiologies are poorly defined.Mannitol dehydrogenase is the key enzyme that catalyzes the pyridine nucleotide-dependent oxidation of D-mannitol to Dfructose (12). Despite the existence of genes encoding putative homologs of mannitol dehydrogenase (YEL070W and YNR073C), S. cerevisiae strains, including the S288C reference strain, are unable to assimilate mannitol for growth; a few exceptions exi...

Mannitol is a promising marine macroalgal carbon source. However, organisms that produce ethanol from mannitol are limited; to date, only the yeast Pichia angophorae and the bacterium Escherichia coli KO11 have been reported to possess this capacity. In this study, we searched a yeast strain with a high capacity to produce ethanol from mannitol and selected Saccharomyces paradoxus NBRC 0259 for its ability to produce ethanol from mannitol. This ability was enhanced after a 3-day cultivation of this strain in medium containing mannitol; the enhanced strain was renamed S. paradoxus NBRC 0259-3. We compared the ability of strain NBRC 0259-3 to produce ethanol from mannitol and glucose, under several conditions, with those of P. angophorae and E. coli KO11. As a result, we concluded that S. paradoxus NBRC 0259-3 strain is the most suitable yeast strain for the production of ethanol from mannitol.

An algorithm for start-point estimation of a signal from a frame is presented. In many applications of speech signal processing, the signal to be processed is often segmented into several frames, and then the frames are categorized into speech and non-speech frames. Instead, we focus on only the frame in which the speech starts. To simplify the problem, we assume that the speech is modeled by a number of complex sinusoidal signals. When a complex sinusoidal signal that starts in a frame is observed, it can be modeled as multiplication of a complex sinusoidal signal of which the length is infinite and a window function that has finite duration in the time domain. In the frequency domain, the spectrum of the signal of the frame is given by the shifted spectrum of the window function. Sharpness of the spectrum of the window function depends on the start point of the signal. Hence, the start point of the signal is estimated by the sharpness of the observed spectrum. This approach can be extended to the signal that consists of a number of complex sinusoidal signals. Simulation results using artificially generated signals show the validity of our method.