10,000 publications from this institution
α/β hydrolases make up a large and diverse protein superfamily. In natural product biosynthesis, <i>cis</i>-acting thioesterase α/β hydrolases can terminate biosynthetic assembly lines and release products by hydrolyzing or cyclizing the biosynthetic intermediate. Thioesterases can also act in <i>trans</i>, removing aberrant intermediates and restarting stalled biosynthesis. Knockout of this "editing" function leads to reduced product titers. The borrelidin biosynthetic gene cluster from <i>Streptomyces parvulus</i> Tü4055 contains a hitherto uncharacterized stand-alone thioesterase, <i>borB</i>. In this work, we demonstrate that purified BorB cleaves acyl substrates with a preference for propionate, which supports the hypothesis that it is also an editing thioesterase. The crystal structure of BorB shows a wedgelike hydrophobic substrate binding crevice that limits substrate length. To investigate the structure-function relationship, we made chimeric BorB variants using loop regions from characterized homologues with different specificities. BorB chimeras slightly reduced activity, arguing that the modified region is a not major determinant of substrate preference. The structure-function relationships described here contribute to the process of elimination for understanding thioesterase specificity and, ultimately, engineering and applying <i>trans</i>-acting thioesterases in biosynthetic assembly lines.
We have found that a thin silicon dioxide skin stabilizes the float zone in ribbon-to-ribbon single crystal growth. 40-μm-thick single crystal ribbons, oriented 〈100〉 and scanned in the 〈011〉 direction, were grown completely free of subgrain as well as grain boundaries. This surprising suppression of low angle grain boundaries may be related to a similar effect which has been seen in supported thick silicon on SiO2 films.
subunits. The an3 T-cell receptors rec-ognize short peptides derived from protein antigens,embedded in the groove of antigen-presenting mol-ecules encoded by genes of the major histocompati-bility complex (MHC). In their roles as helper (CD4+)and cytotoxic (CD8+) T cells, ac T cells constitute themajor component of T-cell immunity to infectiousagents and other antigens. Several years ago, a distinctsubset of T cells was discovered that express an antigenreceptor composed of variable y and 8 subunits.Although a great deal has been learnt of the diversity,development and subtype heterogeneity of y8 T cells[1], an understanding of their physiological roles in theimmune response and their antigenic targets has beenelusive. Recent progress in both of these areas holdsout the promise that significant advances may soon beforthcoming.The recent production of mice deficient in expressionof cat and/or y5 T-cell receptors, by means of targetedmutations that inactivate their genes for and/or 8T-cell receptor chains, allows for the first time directanalysis of the specific roles played by the differenttypes of T cell in immunity to infectious diseases.Obviously, if '6 T cells have a unique role in immunityto specific infectious diseases, T-cell receptor 8 chainmutant mice, which lack 68 T cells but have apparentlynormal c T cells, should be highly sensitive to suchdiseases. Examples of this sort may emerge with time,but none has yet been reported. Some infectiousagents, on the other hand, may be attacked by botha3 and y6 T cells. In this case, mice mutant for thegenes encoding both
In a programmable (multistage) cellular neural network (CNN) structure, the CPU is a CNN universal chip which supports massively parallel computations on patterns and images, including videos. In this paper, we decompose the structure of a class of simultaneous recurrent networks (SRN) into a CNN program and run it on a von Neumann-like stored program CNN structure. To train the SRN, we map the back-propagation-through-time (BTT) learning algorithm into a sequence of CNN subroutines to achieve real-time performance via a CNN universal chip. By computing in parallel, the CNN universal chip can be programmed to implement in real time the BTT learning algorithm, which has a very high time complexity. An estimate of the time complexity of the BTT learning algorithm based on the CNN universal chip is presented. For small-scale problems, our simulation results show that a CNN implementation of the BTT learning algorithm for a two-dimensional SRN is at least 10,000 times faster than that based on state-of-the-art sequential workstations. For the few large-scale problems which we have so far simulated, the CNN implemented BTT learning algorithm maintained virtually the same time complexity with a learning time of a few seconds, while those implemented on state-of-the-art sequential workstations dramatically increased their time complexity, often requiring several days of running time. Several examples are presented to demonstrate how efficiently a CNN universal chip can speed up the learning algorithm for both off-line and on-line applications.
