Multitasking via baseline control in recurrent neural networks.

Changes in behavioral state, such as arousal and movements, strongly affect neural activity in sensory areas, and can be modeled as long-range projections regulating the mean and variance of baseline input currents. What are the computational benefits of these baseline modulations? We investigate this question within a brain-inspired framework for reservoir computing, where we vary the quenched baseline inputs to a recurrent neural network with random couplings. We found that baseline modulations control the dynamical phase of the reservoir network, unlocking a vast repertoire of network phases. We uncovered a number of bistable phases exhibiting the simultaneous coexistence of fixed points and chaos, of two fixed points, and of weak and strong chaos. We identified several phenomena, including noise-driven enhancement of chaos and ergodicity breaking; neural hysteresis, whereby transitions across a phase boundary retain the memory of the preceding phase. In each bistable phase, the reservoir performs a different binary decision-making task. Fast switching between different tasks can be controlled by adjusting the baseline input mean and variance. Moreover, we found that the reservoir network achieves optimal memory performance at any first-order phase boundary. In summary, baseline control enables multitasking without any optimization of the network couplings, opening directions for brain-inspired artificial intelligence and providing an interpretation for the ubiquitously observed behavioral modulations of cortical activity.

Anterior Chamber Shallowing from the Early Stage of Surgery and Suprachoroidal Effusion during Clear Corneal Small-Incision Cataract Surgery: A Case Report.

Suprachoroidal effusion (SCE) is a rarely observed complication due to the recent generalization of clear corneal small-incision cataract surgery. We report a case of anterior chamber shallowing (ACS) from the early stage of surgery and SCE during clear corneal small-incision cataract surgery. A 69-year-old man was referred to our department for primary open-angle glaucoma and grade 2 nuclear cataract. The intraocular pressure (IOP) was 18 and 12 mm Hg in the right and left eyes with the instillation of three antiglaucoma eye drops in both eyes, respectively, and deep anterior chamber and normal axial length were observed. At the age of 70 years, which was 4 months after the initial visit to our department, the IOP of the right eye increased to 30 mm Hg. Hence, cataract surgery and microhook ab interno trabeculotomy (µLOT) of the right eye were scheduled. Mild ACS was observed during continuous curvilinear capsulorhexis (CCC), and ACS worsened as the surgery progressed, making the surgery progressively challenging. SCE was observed by fundus examination after phacoemulsification and cortex removal, and the wound was immediately closed with a suture. The IOP was 28 mm Hg on postoperative day (POD) 1 and decreased to 14 mm Hg on POD 5. SCE disappeared on POD 12. On POD 18, intraocular lens implantation into the bag and µLOT were performed under general anesthesia. Subsequently, the IOP decreased to 15 mm Hg 3 months after the surgery. Mild ACS was already present at the time of CCC, so it is possible that SCE occurred in the early stage of surgery. If ACS is observed intraoperatively, especially if there are SCE risk factors, such as hypertension, glaucoma, and lung cancer, as in this case, and even if the eye has deep anterior chamber and normal axial length preoperatively, fundoscopic examination should be performed even at an early stage of clear corneal small-incision cataract surgery to rule out SCE.

Edge of Chaos and Avalanches in Neural Networks with Heavy-Tailed Synaptic Weight Distribution.

We propose an analytically tractable neural connectivity model with power-law distributed synaptic strengths. When threshold neurons with biologically plausible number of incoming connections are considered, our model features a continuous transition to chaos and can reproduce biologically relevant low activity levels and scale-free avalanches, i.e., bursts of activity with power-law distributions of sizes and lifetimes. In contrast, the Gaussian counterpart exhibits a discontinuous transition to chaos and thus cannot be poised near the edge of chaos. We validate our predictions in simulations of networks of binary as well as leaky integrate-and-fire neurons. Our results suggest that heavy-tailed synaptic distribution may form a weakly informative sparse-connectivity prior that can be useful in biological and artificial adaptive systems.

Simple method for analyzing the purity of protease-containing samples by acid-treatment SDS-PAGE.

Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) is a widely used technique to analyze the purity of a protein. However, it is necessary to denature (via boiling) the samples before subjecting them to electrophoresis. In the case of protease-containing samples, autolysis of the protease can occur, affecting the accuracy of results. In this study, we investigated the methods for analyzing the purity of Dispase I, a thermolysin-like neutral protease. When we analyzed D protease, a neutral metalloprotease component of Dispase I and highly purified Dispase I using the conventional SDS-PAGE method, a large number of bands were detected in both cases. These bands (putative D protease fragments) were assumed to result from autolysis. To inactivate D protease (optimal pH 7-8), 0.05 M sulfuric acid was utilized (pH 0.7-2.5). Using a conventional sample preparation solution, acid-treated Dispase I samples (without boiling) were made, and SDS-PAGE (15% w/v gel) was carried out. Our findings show that autolysis was inhibited under strong acidic conditions, and protein denaturation was achieved by treatment with sulfuric acid and SDS without boiling. Using this modified SDS-PAGE method, the purities of Dispase I and the purified enzyme were determined to be approximately 80% and 98%, respectively. Furthermore, we demonstrated that this method can be applied for the analysis of other samples including non-acidic proteases (e.g., thermolysin, subtilisin, and trypsin) and protease-contaminated samples (a mixed solution of albumin and D protease).

A Novel Method for the Objective Identification of Hyperautofluorescent Ring in Retinitis Pigmentosa Using Binarization Processing.

PURPOSE: To assess precision and accuracy of a new objective algorithm using binarization in a software for identifying the hyperautofluorescent ring (AF ring) in retinitis pigmentosa (RP) compared with subjective visual inspection. METHODS: Ultra-widefield AF images were obtained from 23 eyes of 13 patients with retinitis pigmentosa (RP). We defined the borders of the AF rings using semiautomatic binarization algorithm in Fiji software. We compared the degree of precision (intra- and interrater agreements) of this algorithm and that of subjective visual inspection (freehand method) using Jaccard indices (JIs). To compare the classification performance (whether 68 points of Humphrey Field Analyzer is classified as inside, on, or outside AF rings), we calculated percent agreement and weighted kappa statistic between the two methods. The relationship between the distance from the AF ring and retinal sensitivities was also investigated. RESULTS: The binarization method showed significantly higher JIs than the freehand method (for interrater: 0.94-0.95 vs. 0.73-0.78, respectively, P = 0.002; for intrarater: 0.95 vs. 0.68-0.71, respectively, P = 0.005). Percent agreement for classification between the two methods were 0.94 and weighted kappa statistic was 0.94 (P < 0.001). The retinal sensitivities decreased significantly and eccentrically from 2° inside to 3° outside the AF ring. CONCLUSIONS: Defining the AF ring in RP using the binarization algorithm showed significantly higher precision and the same degree of accuracy compared with visual inspection. TRANSLATION RELEVANCE: This novel method may enable quantitative analysis of the AF ring, an indicator of retinal function in RP.

Characterization of Three Fungal Isomaltases Belonging to Glycoside Hydrolase Family 13 That Do not Show Transglycosylation Activity.

α-1,6-Glucosidase (isomaltase) belongs to glycoside hydrolase (GH) families 13 and 31. Genes encoding 3 isomaltases belonging to GH family 13 were cloned from filamentous fungi, Aspergillus oryzae (agl1), A. niger (agdC),and Fusarium oxysporum (foagl1), and expressed in Escherichia coli. The enzymes hydrolyzed isomaltose and α-glucosides preferentially at a neutral pH, but did not recognize maltose, trehalose, and dextran. The activity of AgdC and Agl1 was inhibited in the presence of 1 % glucose, while Foagl1 was more tolerant to glucose than the other two enzymes were. The three fungal isomaltases did not show transglycosylation when isomaltose was used as the substrate and a similar result was observed for AgdC and Agl1 when p-nitrophenyl-α-glucoside was used as the substrate.

Linear and nonlinear response of the Vlasov system with nonintegrable Hamiltonian.

Linear and nonlinear response formulas taking into account all Casimir invariants are derived without use of angle-action variables of a single-particle (mean-field) Hamiltonian. This article deals mainly with the Vlasov system in a spatially inhomogeneous quasistationary state whose associating single-particle Hamiltonian is not integrable and has only one integral of the motion, the Hamiltonian itself. The basic strategy is to restrict the form of perturbation so that it keeps Casimir invariants within a linear order, and the single particle's probabilistic density function is smooth with respect to the single particle's Hamiltonian. The theory is applied for a spatially two-dimensional system and is confirmed by numerical simulations. A nonlinear response formula is also derived in a similar manner.

Conditions for predicting quasistationary states by rearrangement formula.

Predicting the long-lasting quasistationary state for a given initial state is one of central issues in Hamiltonian systems having long-range interaction. A recently proposed method is based on the Vlasov description and uniformly redistributes the initial distribution along contours of the asymptotic effective Hamiltonian, which is defined by the obtained quasistationary state and is determined self-consistently. The method, to which we refer as the rearrangement formula, was suggested to give precise prediction under limited situations. Restricting initial states consisting of a spatially homogeneous part and small perturbation, we numerically reveal two conditions that the rearrangement formula prefers: One is a no Landau damping condition for the unperturbed homogeneous part, and the other comes from the Casimir invariants. Mechanisms of these conditions are discussed. Clarifying these conditions, we validate to use the rearrangement formula as the response theory for an external field, and we shed light on improving the theory as a nonequilibrium statistical mechanics.

Landau-like theory for universality of critical exponents in quasistationary states of isolated mean-field systems.

An external force dynamically drives an isolated mean-field Hamiltonian system to a long-lasting quasistationary state, whose lifetime increases with population of the system. For second order phase transitions in quasistationary states, two nonclassical critical exponents have been reported individually by using a linear and a nonlinear response theories in a toy model. We provide a simple way to compute the critical exponents all at once, which is an analog of the Landau theory. The present theory extends the universality class of the nonclassical exponents to spatially periodic one-dimensional systems and shows that the exponents satisfy a classical scaling relation inevitably by using a key scaling of momentum.

Nonlinear response for external field and perturbation in the Vlasov system.

A nonlinear response theory is provided by use of the transient linearization method in the spatially one-dimensional Vlasov systems. The theory inclusively gives responses to external fields and to perturbations for initial stationary states, and is applicable even to the critical point of a second-order phase transition. We apply the theory to the Hamiltonian mean-field model, a toy model of a ferromagnetic body, and investigate the critical exponent associated with the response to the external field at the critical point in particular. The obtained critical exponent is the nonclassical value 3/2, while the classical value is 3. However, interestingly, one scaling relation holds with another nonclassical critical exponent of susceptibility in the isolated Vlasov systems. Validity of the theory is numerically confirmed by directly simulating temporal evolutions of the Vlasov equation.

Dynamical pattern formation in two-dimensional fluids and Landau pole bifurcation.

A phenomenological theory is proposed to analyze the asymptotic dynamics of perturbed inviscid Kolmogorov shear flows in two dimensions. The phase diagram provided by the theory is in qualitative agreement with numerical observations, which include three phases depending on the aspect ratio of the domain and the size of the perturbation: a steady shear flow, a stationary dipole, and four traveling vortices. The theory is based on a precise study of the inviscid damping of the linearized equation and on an analysis of nonlinear effects. In particular, we show that the dominant Landau pole controlling the inviscid damping undergoes a bifurcation, which has important consequences on the asymptotic fate of the perturbation.

Non-mean-field critical exponent in a mean-field model: dynamics versus statistical mechanics.

Mean-field theory tells us that the classical critical exponent of susceptibility is twice that of magnetization. However, linear response theory based on the Vlasov equation, which is naturally introduced by the mean-field nature, makes the former exponent half of the latter for families of quasistationary states having second order phase transitions in the Hamiltonian mean-field model and its variances, in the low-energy phase. We clarify that this strange exponent is due to the existence of Casimir invariants which trap the system in a quasistationary state for a time scale diverging with the system size. The theoretical prediction is numerically confirmed by N-body simulations for the equilibrium states and a family of quasistationary states.

Spectral and formal stability criteria of spatially inhomogeneous stationary solutions to the Vlasov equation for the Hamiltonian mean-field model.

Stability of spatially inhomogeneous solutions to the Vlasov equation is investigated for the Hamiltonian mean-field model to provide the spectral and formal stability criteria in the form of necessary and sufficient conditions. These criteria determine stability of spatially inhomogeneous solutions whose stability has not been decided correctly by using a less refined formal stability criterion. It is shown that some of such solutions can be found in a family of stationary solutions to the Vlasov equation, which is parametrized with macroscopic quantities and has a two-phase coexistence region in the parameter space.

Linear response theory in the Vlasov equation for homogeneous and for inhomogeneous quasistationary states.

Response to small external forces is investigated in quasistationary states of Hamiltonian systems having long-range interactions. Quasistationary states are recognized as stable stationary solutions to the Vlasov equation, and, hence, the linear response theory to the Vlasov equation is proposed for spatially one-dimensional systems with periodic boundary condition. The proposed theory is applicable both to homogeneous and to inhomogeneous quasistationary states and is demonstrated in the Hamiltonian mean-field model. In the homogeneous case magnetic susceptibility is explicitly obtained, and the Curie-Weiss like law is suggested in a high-energy region. The linear response is also computed in the inhomogeneous case, and resonance absorption is investigated to extract nonequilibrium dynamics in the unforced system. Theoretical predictions are examined by direct numerical simulations of the Vlasov equation.

Active suppression of EndoPG IV by ligation of the pro-sequence from Stereum purpureum Pro-EndoPG I.

We attempted to inactivate endopolygaolacturonase from Stereum purpureum (EndoPG) IV of identical origin by linking the pro-sequence of S. purpureum Pro-EndoPG I to the C-terminus. The recombinant Pro-EndoPG IV, expressed in Escherichia coli, had no polygalacturonase (PG) activity, but activity was acquired after partial degradation of the pro-sequence with V8 protease, as was the case for Pro-EndoPG I. These results indicate that the pro-sequence of Pro-EndoPG I can suppress the PG activity of EndoPG IV.

Effect of individual suppression of RBCS multigene family on Rubisco contents in rice leaves.

In higher plants, a small subunit of Rubisco is encoded for by an RBCS multigene family in the nuclear genome. However, it is unknown how each multigene member contributes to the accumulation of Rubisco holoenzyme. Here, four RBCS genes that are highly expressed in leaf blaedes of rice (Oryza sativa L.) were individually suppressed by RNAi, and the effects on leaf Rubisco content were examined at seedling, vegetative and reproductive stages. Rubisco contents in each transgenic line declined irrespective of growth stage, and the ratios of Rubisco-N to total N were 66-96% of wild-type levels. The mRNA levels of the suppressed RBCS genes declined significantly, whereas those of the unsuppressed ones did not change drastically. These results indicate that four RBCS genes all contribute to accumulation of Rubisco holoenzyme irrespective of growth stage and that suppression of one RBCS gene is not fully compensated by other RBCS genes. Additionally, the mRNA levels of the large subunit of Rubisco showed a change similar to that of total RBCS mRNA level irrespective of genotype and growth stage. These results suggest that gene expression of RBCS and RBCL is regulated in a coordinated manner at the transcript level in rice.

Precise determination of the nonequilibrium tricritical point based on Lynden-Bell theory in the Hamiltonian mean-field model.

Existence of a nonequilibrium tricritical point has been revealed in the Hamiltonian mean-field model by a nonequilibrium statistical mechanics. This statistical mechanics gives a distribution function containing unknown parameters, and the parameters are determined by solving simultaneous equations depending on a given initial state. Due to difficulty in solving these equations, pointwise numerical detection of the tricritical point has been unavoidable on a plane characterizing a family of initial states. In order to look into the tricritical point, we expand the simultaneous equations with respect to the order parameter and reduce them to one algebraic equation. The tricritical point is precisely identified by analyzing coefficients of the reduced equation. Reentrance to an ordered phase in a high-energy region is revisited around the obtained tricritical point.

The pro-form of Stereum purpureum endopolygalacturonase I is inactivated by a pro-sequence in the C-terminal region.

The Pro-form (Pro-EndoPG I) of Stereum purpureum endopolygalacturonase I has a unique C-terminal region (pro-sequence) that is lacking in PGs of other origins. Mature EndoPG I purified from the culture filtrate of this fungus does not have the 44-amino-acid pro-sequence present in Pro-EndoPG I. We expressed Pro-EndoPG I in Escherichia coli and examined its activity. It was found that Pro-EndoPG I had no PG activity, but that PG activity was acquired after the degradation of part of the pro-sequence with V8 protease. These results suggest that the pro-sequence inactivates auto-PG activity. No similar characteristic has been reported for any glycoside hydrolase. We then constructed EndoPG I mutants and identified two Glu residues, E364 and E366, that were related to auto-inactivation. A test involving injection of the enzyme into apple trees showed that Pro-EndoPG I induced the same silver-leaf symptoms as mature EndoPG I.

Expression, purification, and analyses of glycosylation and disulfide bonds of Stereum purpureum endopolygalacturonase I in Pichia pastoris.

We have succeeded in the expression of Stereum purpureum endopolygalacturonase I (EndoPG I) using the Pichia expression system and in purification of the three kinds of recombinant EndoPG I, which have one to three sugar chains by using CM52 column chromatography. The sugar chains which were added to EndoPG I were the M8, M9, and/or M10 high-mannose type. The results of LC-MS analysis showed that recombinant EndoPG Is were randomly glycosylated at four N-glycosylation sites. From the thermal denaturation curves of the recombinant enzymes, it was suggested that EndoPG I differing in thermal stability was included in the sample after purification. Therefore, we investigated the disulfide bonds of recombinant EndoPG I by LC-MS analysis. As a result, peptides without a second or third disulfide bond were detected. This result is the first indicating that there are incomplete enzymes in terms of disulfide bonds in the Pichia expression system.

Transient cardiomyopathy in a patient with congenital contractural arachnodactyly (Beals syndrome).

We report on an infant with Beals syndrome (congenital contractural arachnodactyly [CCA], MIM 121050) with transient cardiomyopathy showing ballon-like dilatation of the left ventricle that was similar to noncompaction. The patients father and two of his brothers were also found to have CCA without cardiovascular complications. CCA, which is caused by a mutation of the gene for fibrillin 2 protein is similar to Marfan syndrome (MIM 154700), which is caused by a mutation of fibrillin 1 but produces a life-threatening cardiovascular complications. This is the first report of CCA with transient cardiomyopathy. We discuss the mechanism of the spontaneous improvement of cardiomyopathy in this case on the basis of expression of the responsible gene.