Optical implementation of the Hopfield model for two-dimensional associative memory

Optical implementation of Hopfield's neural network model [Proc. Natl. Acad. Sci. USA 79, 2554 (1982)] for two-dimensional associative memory is discussed. Two-state neuron elements are represented by a twisted nematic liquid-crystal optical switch array, and three-dimensional holographic interconnections are realized with these elements. Unipolar connections, created by adding a constant to bipolar interconnections and compensating them with an input-dependent thresholding operation, are realized. The 16- (4 x 4) neuron system model acts as a content-addressable associative memory with error-correction capability.

Modulation of cellular transglutaminase: protease-induced activation

Multiple molecular forms of transglutaminase are found in cells and each form is widely distributed. We find a 95 K dalton enzyme associated with membrane fractions. A 50 K dalton enzyme occurs primarily in epidermis and hair follicles. Cells after treatment with proteases show greater transglutaminase activity. The activated enzyme in rat chondrosarcoma cells is one of 95 K daltons, whereas mouse epidermal cells and rabbit endometrium cells after protease activation display enzymes of both 95 K daltons and 50 K daltons. The 95 K dalton enzyme, but not that of 80 K daltons, can be activated by proteases or sulfhydryl compounds after cell lysis. In cells that undergo terminal differentiation, e.g., reticulocytes, megakaryocytes, monocytes, chondrocytes, and epidermal cells, the forms of transglutaminase are modulated. Our findings suggest that these modulations in differentiating cells are the results of transglutaminase post-translational modifications that cause pronounced changes in catalytic activity.

Influence of pore and particle size on the frontal uptake of proteins. Implications for preparative anion-exchange chromatography

Several silica-based anion-exchange packings were synthesized with nominal pore sizes of 250, 500 and 1000 A in 10-, 20- and 50-micron particles. The static ("equilibrium") adsorption capacities for bovene serum albumin (mol. wt. 69,000), alpha-lactalbumin (17,500) and ferritin (440,000) were first measured using bulk material. The media were then packed into columns for frontal uptake experiments to measure adsorption from a flowing mobile phase. In general, frontal uptake was inversely related to both flow-rate and particle size. However, the magnitude of these relationships was strongly dependent on the pore to protein diameter ratio. More specifically, the uptake of bovine serum albumin was significantly more sensitive to linear velocity and particle size than alpha-lactalbumin. A mathematical model of the chromatographic process was used to calculate radial adsorption profiles across the chromatographic particle during frontal uptake. It was shown that restricted intraparticle diffusion due to insufficient pore size causes incomplete utilization of internal surface area. Under such conditions, protein is only bound within a finite shell on the outermost side of the particle; therefore, the effective loadability of the packing is greatly reduced. These data suggest that a pore size of at least 500 A will be required for the preparative chromatography of proteins with a molecular weight higher than ca. 100,000. This observation is especially evident when using particle sizes greater than 10 micron.

Efficient Tn10 transposition into a DNA insertion hot spot in vivo requires the 5-methyl groups of symmetrically disposed thymines within the hot-spot consensus sequence

Transposon Tn10 inserts preferentially at particular insertion "hot spots" that share a symmetrical 6-base-pair consensus sequence: 5' GCTNAGC 3'. The protein that recognizes this sequence is not known but is likely to be the Tn10-encoded transposase protein. We present evidence that the 5-methyl groups of the two thymines in this sequence are essential for efficient transposon insertion; in their absence the sequence is still recognized, but at lower efficiency. We have reached this conclusion by examination of a specific hot spot whose sequence is 5' GCCAGGC 3'. The innermost cytosines of this sequence happen to be substrates for methylation at their 5 positions by the bacterial dcm-encoded methylase. We find that Tn10 transposes into this site 15 times more frequently in a Dcm+ host than in a Dcm- host; in the Dcm- host, insertions still occur, but at a low frequency. Thus, at this site, the absence of pyrimidine 5-methyl groups at the third positions of the consensus sequence is sufficient to convert a strong insertion hot spot into a weaker but still recognizable hot spot. This observation supports the general proposition, suggested previously by comparisons among consensus sequences, that the presence or absence of these 5-methyl groups is one major feature that can make the difference between a strong and a weak Tn10 insertion hot spot.

Catalytic properties of inositol trisphosphate kinase: activation by Ca2+ and calmodulin

Inositol 1,4,5-trisphosphate (Ins-1,4,5-P3) is an important second-messenger molecule that mobilizes Ca2+ from intracellular stores in response to the occupancy of receptor by various Ca2+-mobilizing agonists. The fate of Ins-1,4,5-P3 is determined by two enzymes, a 3-kinase and a 5-phosphomonoesterase. The first enzyme converts Ins-1,4,5-P3 to Ins-1,3,4,5-P4, whereas the latter forms Ins-1,4-P2. Recent studies suggest that Ins-1,3,4,5-P4 might modulate the entry of Ca2+ from an extracellular source. In the current report, we describe the partial purification of the 3-kinase [approximately 400-fold purified, specific activity = 0.12 mumol/(min.mg)] from the cytosolic fraction of bovine brain and studies of its catalytic properties. We found that the 3-kinase activity is significantly activated by the Ca2+/calmodulin complex. Therefore, we propose that Ca2+ mobilized from endoplasmic reticulum by the action of Ins-1,4,5-P3 forms a complex with calmodulin, and that the Ca2+/calmodulin complex stimulates the conversion of Ins-1,4,5-P3, an intracellular Ca2+ mobilizer, to Ins-1,3,4,5-P4, an extracellular Ca2+ mobilizer. A rapid assay method for the 3-kinase was developed that is based on the separation of [3-32P]Ins-1,3,4,5-P4 and [gamma-32P]ATP by thin-layer chromatography. Using this new assay method, we evaluated kinetic parameters (Km for ATP = 40 microM, Km for Ins-1,4,5-P3 = 0.7 microM, Ki for ADP = 12 microM) and divalent cation specificity (Mg2+ much greater than Mn2+ greater than Ca2+) for the 3-kinase. Studies with various inositol polyphosphates indicate that the substrate-binding site is quite specific to Ins-1,4,5-P3. Nevertheless, Ins-2,4,5-P3 could be phosphorylated at a velocity approximately 1/20-1/30 that of Ins-1,4,5-P3.

Human T-cell leukemia virus type I infection in hemodialysis patients

Human T-cell leukemia virus Type I (HTLV-I) has been detected in a high proportion of individuals in Kumamoto prefecture (Kyushu) and Okinawa, where adult T-cell leukemia (ATL) is more likely to occur. The seroprevalence of antibody to HTLV-I was evaluated in hemodialysis patients and healthy individuals in the ATL endemic area. The prevalence of antibody to HTLV-I in 13,329 healthy controls was 3.6%, compared with 19.7% in 949 chronic hemodialysis patients. The prevalence of hemodialysis patients was significantly higher (P less than 0.01) than that in healthy individuals. Of 681 blood transfused patients who had undergone hemodialysis, 153 (22.5%) were seropositive, compared with 34 (12.7%) of the 268 who had not received the blood transfusion. The incidence was higher in dialysis patients who had been transfused than in those who had never received blood transfusions. Immune defects associated with uremia may have predisposed the patients to HTLV-I infection. Repeated antigenic stimulation from multiple infectious agents may have also played a role. The findings suggest an association between maintenance hemodialysis and HTLV-I infection, which can not be explained solely by blood or blood products.

Mathematics classrooms in Japan, Taiwan, and the United States

Observations were conducted in Chinese, Japanese, and American classrooms during mathematics classes. Activities in 20 representative classrooms were observed in each of 2 grades (1 and 5) and in each country. Some observations focused on individual children and others on the teachers. Large cross-cultural differences were found in many variables related to classroom structure and management. These differences paralleled differences in achievement in mathematics among the 3 countries. A number of these variables also were significantly related to average level of mathematics achievement within the American classrooms.

Purification and in vitro complementation of mutant histidinol dehydrogenases

The biochemistry of interallelic complementation within the Salmonella typhimurium hisD gene was investigated by in vitro protein complementation of mutant histidinol dehydrogenases (EC 1.1.1.23). Double-mutant strains were constructed containing the hisO1242 (constitutive overproducer) attenuator mutation and selected hisDa or hisDb mutations. Extracts from such hisDa986 and hisDb1799 mutant cells failed to show histidinol dehydrogenase activity but complemented to produce active enzyme. Inactive mutant histidinol dehydrogenases were purified from each of the two mutants by ion-exchange chromatography. Complementation by the purified mutant proteins required the presence of 2-mercaptoethanol and MnCl2, and protein-protein titrations indicated that heterodimers were strongly preferred in mixtures of the complementary mutant enzymes. Neither mutant protein showed negative complementation with wild-type enzyme. The Vmax for hybrid histidinol dehydrogenase was 11% of that for native enzyme, with only minor changes in Km values for substrate or coenzyme. Both purified mutant proteins failed to catalyze NAD-NADH exchange reactions reflective of the first catalytic step of the two-step reaction. The inactive enzymes bound 54Mn2+ weakly or not at all in the presence of 2-mercaptoethanol, in contrast to wild-type enzyme which bound 54Mn2+ to 0.6 sites per monomer under the same conditions. The mutant proteins, like wild-type histidinol dehydrogenase, behaved as dimers on analytical gel filtration chromatography, but dissociated to form monomers in the presence of 2-mercaptoethanol. This effect of 2-mercaptoethanol was prevented by low levels of MnCl2. It thus appears that mutant histidinol dehydrogenase molecules bind metal ion poorly. The complementation procedure may allow for formation of a functional Mn2+-binding site, perhaps at the subunit interface.

The effect of different brands of MacConkey and xylose lysine deoxycholate agar on the isolation of Aeromonas hydrophila

Aeromonas hydrophila has been reported to be involved in the opportunistic infection of human beings, therefore, its rapid diagnosis from the clinical specimens was considered to be extremely important for the prognosis. The effect of different brands of MacConkey (MAC) and xylose lysine deoxycholate (XLD) agars on the plating efficiency and colony size of A. hydrophila was evaluated by comparing with blood agar. Results indicate that (i) except Eiken's MAC have low plating efficiency, all the MAC from BBL, Difco, Mast, Merck, and Gibco have good plating efficiency; (ii) except Eiken's and Gibco XLD, all XLD from Difco, Mast, and Merck have larger bacterial colony size; and (iii) colonies of A. hydrophila appeared on MAC are larger than those on XLD of the same brands. We suggested that a clinical laboratory microbiologist should be alert in selecting the appropriate media to isolate the clinically important bacteria.

Parallel 2-d convolution on a mesh connected array processor

In this correspondence, a parallel 2-D convolution scheme is presented. The processing structure is a mesh connected array processor consisting of the same number of simple processing elements as the number of pixels in the image. For most windows considered, the number of computation steps required is the same as that of the coefficients of a convolution window. The proposed scheme can be easily extended to convolution windows of arbitrary size and shape. The basic idea of the proposed scheme is to apply the 1-D systolic concept to 2-D convolution on a mesh structure. The computation is carried out along a path called a convolution path in a systolic manner. The efficiency of the scheme is analyzed for windows of various shapes. The ideal convolution path is a Hamiltonian path ending at the center of the window, the length of which is equal to the number of window coefficients. The simple architecture and control strategy make the proposed scheme suitable for VLSI implementation.

Hypercalcemia and osteoclast proliferation in adult T-cell leukemia

Eighteen autopsy cases of adult T-cell leukemia (ATL) were investigated clinicopathologically. Thirteen of the patients had hypercalcemia during their clinical course. Nine of the thirteen had a high level of serum calcium at the terminal stage, even after extensive chemotherapy. Microscopic examination of the bone revealed proliferation of osteoclasts and bone resorption in eight patients. No osteoclast proliferation or bone resorption was found in the other nine normocalcemic patients. The infiltration of ATL cells was observed in only two patients--one was hypercalcemic and the other, normocalcemic. The factors affecting the serum calcium level were examined in two hypercalcemic patients. Hypercalcemia could not be accounted for by parathyroid hormone or prostaglandins E levels, which were in the normal range, or by 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D, which were low. Our findings are consistent with the mechanism proposed by several investigators, that the malignant T-lymphocytes produced an osteoclast-activating-factor-like substance that caused osteoclast proliferation and hypercalcemia.

Antimicrobial susceptibility of bacteria isolated in 1985--with special reference to prevalence of methicillin-resistant staphylococcus aureus and activities of cefazolin, cefotaxime and piperacillin

Antimicrobial susceptibility of nine species and one group of bacteria isolated from patients at the hospitals of Seoul National University, Severance, Hanyang University, and Kyungpuk University were tested by agar dilution method. S. aureus was most susceptible to cefazolin, methicillin and cotrimoxazole, and enterococci to ampicillin. Isolates of Enterobacteriaceae were most frequently susceptible to aminoglycosides and cefotaxime. Cefazolin susceptibility was markedly different from species to species. Aminoglycosides and piperacillin were more active than others against P. aeruginosa, and amikachin against A. anitratus. A large proportion of strains of several different species were conditionally susceptible to either tetracycline, ampicillin, cefazolin or cotrimoxazole suggesting the usefulness of these drugs for treatment of urinary tract infection. Activity of cefotaxime was highest against E. coli, and K. pneumoniae, while lowest against A. anitratus and P. aeruginosa. Decrease in the proportion of susceptible isolate was noted in E. coli and K. pneumoniae to cefazolin, K. pneumoniae, E. cloacae and S. marcescens to cotrimoxazole, and P. aeruginosa to tobramycin and gentamicin.

The effects of random directional distributed flow in nuclear magnetic resonance imaging

Capillary flow or microscopic random directional coherent flow as a model of perfusion is investigated both theoretically and experimentally. In the model, we assumed that molecular motion within a finite resolvable volume element (voxel) is a superposition of flow of randomly oriented small capillaries. In such a case, the observed signal from the capillary flow within a voxel will be attenuated in signal amplitude without any change in phase. Although this attenuation effect is similar to the diffusion phenomenon, it differs basically in the following aspects: since the motion in each capillary segment is coherent, phase cancellation occurs at even echoes due to spin rephasing, while the diffusion phenomenon is a purely random Brownian motion of the thermally agitated molecules, changing both in direction and speed during the measurement period. Because of the random character of diffusion, even-echo rephasing cannot be observed. Thus capillary flow or perfusionlike microscopic flow can be measured based on the above distinct flow characteristics, i.e., signal restoration at even echoes versus signal amplitude attenuation at odd echoes. By applying a suitable mathematical algorithm, information on the capillary flow alone can be extracted from the two separate distinct measurements, i.e., one with a single echo and the other with a double echo. Both a theoretical calculation of the capillary flow, as well as the experimental results with a human volunteer by a 0.6-T nuclear magnetic resonance imager, are presented.

Characterization of RNA polymerase associated with rice ragged stunt virus

The RNA polymerase associated with rice ragged stunt virus (RRSV) was characterized. Activity was optimum at 35-40 degrees in 0.1 MTris-HC1 (pH 8.5) and 6-8 mMMgCl2. S-Adenosyl-L-methionine stimulated the activity about 5- to 6-fold. It was also stimulated in the presence of chymotrypsin (200 micrograms/ml). The molecular weights of RNAs synthesized in vitro were calculated to be about half those of the respective genome segments. The synthesized RNAs hybridized to the genome RNAs, and the hybrids migrated identically to the genome RNAs in PAGE. These results indicate that RRSV particles transcribe full-length copies of the genome RNAs. The characteristics of the polymerase are discussed in relation to those of other members of the Reoviridae.

An improved nuclear magnetic resonance diffusion coefficient imaging method using an optimized pulse sequence

Two-dimensional diffusion coefficient maps (images) of a carefully controlled diffusion phantom have been measured by a new diffusion imaging sequence using a 0.6-T whole-body nuclear magnetic resonance (NMR) scanner having a gradient field strength of 2.5 mT/m. The free induction decay (FID) data for the diffusion coefficient images were collected by varying the duration of the readout gradient in the conventional two-dimensional Fourier imaging sequence. The experimental results obtained by the proposed NMR diffusion measurement technique indicate a close agreement with other previous measurements. The selection of optimum spin-echo time for maximum signal-to-noise ratio (SNR) in diffusion imaging is studied and also experimentally confirmed. Finally, a preclinical study with human volunteers has been performed and results are presented.