Biological neurons act as generalization filters in reservoir computing.

Reservoir computing is a machine learning paradigm that transforms the transient dynamics of high-dimensional nonlinear systems for processing time-series data. Although the paradigm was initially proposed to model information processing in the mammalian cortex, it remains unclear how the nonrandom network architecture, such as the modular architecture, in the cortex integrates with the biophysics of living neurons to characterize the function of biological neuronal networks (BNNs). Here, we used optogenetics and calcium imaging to record the multicellular responses of cultured BNNs and employed the reservoir computing framework to decode their computational capabilities. Micropatterned substrates were used to embed the modular architecture in the BNNs. We first show that the dynamics of modular BNNs in response to static inputs can be classified with a linear decoder and that the modularity of the BNNs positively correlates with the classification accuracy. We then used a timer task to verify that BNNs possess a short-term memory of several 100 ms and finally show that this property can be exploited for spoken digit classification. Interestingly, BNN-based reservoirs allow categorical learning, wherein a network trained on one dataset can be used to classify separate datasets of the same category. Such classification was not possible when the inputs were directly decoded by a linear decoder, suggesting that BNNs act as a generalization filter to improve reservoir computing performance. Our findings pave the way toward a mechanistic understanding of information representation within BNNs and build future expectations toward the realization of physical reservoir computing systems based on BNNs.

Enhanced responses to inflammatory cytokine interleukin-6 in micropatterned networks of cultured cortical neurons.

Cortical neurons in dissociated cultures are an indispensable model system for pharmacological research that provides insights into chemical responses in well-defined environments. However, cortical neurons plated on homogeneous substrates develop an unstructured network that exhibits excessively synchronized activity, which occasionally masks the consequences induced by external substances. Here, we show that hyperactivity and excessive synchrony in cultured cortical networks can be effectively suppressed by growing neurons in microfluidic devices. These devices feature a hierarchically modular design that resembles the in vivo network. We focused on interleukin-6, a pro-inflammatory cytokine, and assessed its acute and chronic effects. Fluorescence calcium imaging of spontaneous neural activity for up to 20 days of culture showed detectable modulation of collective activity events and neural correlation in micropatterned neurons, which was not apparent in neurons cultured on homogeneous substrates. Our results indicate that engineered neuronal networks provide a unique platform for detecting and understanding the fundamental effects of biochemical compounds on neuronal networks.

Model-free idealization: Adaptive integrated approach for idealization of ion-channel currents.

Single-channel electrophysiological recordings provide insights into transmembrane ion permeation and channel gating mechanisms. The first step in the analysis of the recorded currents involves an "idealization" process, in which noisy raw data are classified into two discrete levels corresponding to the open and closed states of channels. This provides valuable information on the gating kinetics of ion channels. However, the idealization step is often challenging in cases of currents with poor signal-to-noise ratios and baseline drifts, especially when the gating model of the target channel is not identified. We report herein on a highly robust model-free idealization method for achieving this goal. The algorithm, called adaptive integrated approach for idealization of ion-channel currents (AI2), is composed of Kalman filter and Gaussian mixture model clustering and functions without user input. AI2 automatically determines the noise reduction setting based on the degree of separation between the open and closed levels. We validated the method on pseudo-channel-current datasets that contain either computed or experimentally recorded noise. We also investigated the relationship between the noise reduction parameter of the Kalman filter and the cutoff frequency of the low-pass filter. The AI2 algorithm was then tested on actual experimental data for biological channels including gramicidin A, a voltage-gated sodium channel, and other unidentified channels. We compared the idealization results with those obtained by the conventional methods, including the 50%-threshold-crossing method.

Advances in Artificial Cell Membrane Systems as a Platform for Reconstituting Ion Channels.

An artificial cell membrane that is composed of bilayer lipid membranes (BLMs) with transmembrane proteins incorporated within them represents a well-defined system for the analysis of membrane proteins, especially ion channel proteins that are major targets for drug design. Because the BLM system has a high compatibility with recently developed cell-free expression systems, it has attracted attention as a next-generation drug screening system. However, three issues associated with BLM systems, i. e., their instability, the need for non-volatile organic solvents and a low efficiency of ion channel incorporation, have limited their use as a drug screening platform. In this personal account, we discuss our recent approaches to address these issues based on microfabrication. We also discuss the potential for using the BLM system combined with cell-free expression systems as a drug screening system for future personalized medicine.

Suppression of hypersynchronous network activity in cultured cortical neurons using an ultrasoft silicone scaffold.

The spontaneous activity pattern of cortical neurons in dissociated culture is characterized by burst firing that is highly synchronized among a wide population of cells. The degree of synchrony, however, is excessively higher than that in cortical tissues. Here, we employed polydimethylsiloxane (PDMS) elastomers to establish a novel system for culturing neurons on a scaffold with an elastic modulus resembling brain tissue, and investigated the effect of the scaffold's elasticity on network activity patterns in cultured rat cortical neurons. Using whole-cell patch clamp to assess the scaffold effect on the development of synaptic connections, we found that the amplitude of excitatory postsynaptic current, as well as the frequency of spontaneous transmissions, was reduced in neuronal networks grown on an ultrasoft PDMS with an elastic modulus of 0.5 kPa. Furthermore, the ultrasoft scaffold was found to suppress neural correlations in the spontaneous activity of the cultured neuronal network. The dose of GsMTx-4, an antagonist of stretch-activated cation channels (SACs), required to reduce the generation of the events below 1.0 event per min on PDMS substrates was lower than that for neurons on a glass substrate. This suggests that the difference in the baseline level of SAC activation is a molecular mechanism underlying the alteration in neuronal network activity depending on scaffold stiffness. Our results demonstrate the potential application of PDMS with biomimetic elasticity as cell-culture scaffold for bridging the in vivo-in vitro gap in neuronal systems.

Hepatic Expression of Niemann-Pick C1-Like 1, a Cholesterol Reabsorber from Bile, Exacerbates Western Diet-Induced Atherosclerosis in LDL Receptor Mutant Mice.

Westernization of dietary habits increases lipid intake and is responsible for increased numbers of patients with atherosclerotic diseases. Niemann-Pick C1-Like 1 (NPC1L1)-a cholesterol importer-plays a crucial role in dietary cholesterol absorption in the intestine and is closely associated with several lipid-related diseases, including atherosclerosis. NPC1L1 is highly expressed in the liver and intestine in humans, whereas NPC1L1 expression is low in the rodent liver. Due to species differences in the tissue distribution of NPC1L1, there are limited studies on the pathophysiological role of hepatic NPC1L1, a cholesterol reabsorber from bile. In the present study, to explore whether hepatic NPC1L1 is involved in the development/progression of atherosclerosis, we compared four kinds of atherosclerosis mouse models with different expression levels of NPC1L1 in the intestinal and liver tissues in a genetic background of dysfunctional low-density lipoprotein receptor mutation. Western diet (WD)-induced hyperlipidemia and atherosclerotic plaque formation were more severe in mice expressing NPC1L1 in both the liver and intestine (plasma cholesterol, 839.5 mg/dl; plaque area, 29.5% of total aorta), compared with mice expressing NPC1L1 only in the intestine (plasma cholesterol, 573.1 mg/dl; plaque area, 13.3% of total aorta). Such hepatic NPC1L1-mediated promotion of hyperlipidemia and atherosclerosis was not observed in mice not expressing intestinal NPC1L1 and mice treated with ezetimibe, an NPC1L1 inhibitor used clinically for dyslipidemia. These results suggested that hepatic NPC1L1 promotes WD-induced dyslipidemia and atherosclerosis in concert with intestinal NPC1L1. Our findings provide novel insights into the pathophysiological importance of hepatic NPC1L1 in development/progression of atherosclerosis. SIGNIFICANCE STATEMENT: Niemann-Pick C1-Like 1 (NPC1L1) protein, a cholesterol importer and a molecular target of ezetimibe clinically used for dyslipidemia, is highly expressed not only in the intestine, but also in the liver in humans, although the pathophysiological importance of hepatic NPC1L1 in atherosclerotic diseases remained unclear. By using novel mouse models to separately analyze the effects of hepatic and intestinal NPC1L1 on the development/progression of atherosclerosis, we first demonstrated that hepatic NPC1L1 accelerates Western diet-induced atherosclerotic plaque formation in an intestinal NPC1L1-dependent and an ezetimibe-sensitive manner.

Mean-field analysis of directed modular networks.

We considered a modular network with a binomial degree distribution and related the analytical relationships of the network properties (modularity, average clustering coefficient, and small-worldness) with structural parameters that define the network, i.e., number of nodes, number of modules, average node degree, and ratio of intra-modular to total connections. Even though modular networks are universally found in real-world systems and are consequently of broad interest in complex network science, the relationship between network properties and structural parameters has not yet been formulated. Here, we show that a series of equations for predicting the network properties can be related using a mean-field connectivity matrix that is defined on the basis of the structural parameters in the network generation algorithm. The theoretical results are then compared with values calculated numerically using the original connectivity matrix and are found to agree well, except when the connections between modules are sparse. Representation of the structure of the network using simple parameters is expected to be conducive for elucidating the structure-dynamics relationship.

Amphiphobic Septa Enhance the Mechanical Stability of Free-Standing Bilayer Lipid Membranes.

Artificial bilayer lipid membranes (BLMs) provide well-defined systems for investigating the fundamental properties of membrane proteins, including ion channels, and for screening the effect of drugs that act on them. However, the application of this technique is limited due to the low stability and low reconstitution efficiency of the process. We previously reported on improving the stability of BLM based on the fabrication of microapertures having a tapered edge in SiO2/Si3N4 septa and efficient ion channel incorporation based on vesicle fusion accelerated by a centrifugal force. Although the BLM stability and incorporation probability were dramatically improved when these approaches were used, some BLMs were ruptured when subjected to a centrifugal force. To further improve the BLM stability, we investigated the effect of modifying the surface of the SiO2/Si3N4 septa on the stability of BLM suspended in the septa. The modified surfaces were characterized in terms of hydrophobicity, lipophobicity, and surface roughness. Diffusion coefficients of the lipid monolayers formed on the modified surfaces were also determined. Highly fluidic lipid monolayers were formed on the amphiphobic substrates that had been modified with long-chain perfluorocarbons. Free-standing BLMs formed in amphiphobic septa showed a much higher mechanical stability, including tolerance to water movement and applied centrifugal forces with and without proteoliposomes, than those formed in the septa that had been modified with a short alkyl chain. These results demonstrate that highly stable BLMs are formed when the surface of the septa has amphiphobic properties. Because highly fluidic lipid monolayers that are formed on the septa seamlessly connect with BLMs in a free-standing region, the high fluidity of the lipids contributes to decreasing potential damage to BLMs when mechanical stresses are applied. This approach to improve the BLM stability increases the experimental efficiency of the BLM systems and will contribute to the development of high-throughput platforms for functional assays of ion channel proteins.

Roles of afadin in the formation of the cellular architecture of the mouse hippocampus and dentate gyrus.

The hippocampal formation with tightly packed neurons, mainly at the dentate gyrus, CA3, CA2, and CA1 regions, constitutes a one-way neural circuit, which is associated with learning and memory. We previously showed that the cell adhesion molecules nectins and its binding protein afadin play roles in the formation of the mossy fiber synapses which are formed between the mossy fibers of the dentate gyrus granule cells and the dendrites of the CA3 pyramidal cells. We showed here that in the afadin-deficient hippocampal formation, the dentate gyrus granules cells and the CA3, CA2, and CA1 pyramidal cells were abnormally located; the mossy fiber trajectory was abnormally elongated; the CA3 pyramidal cells were abnormally differentiated; and the densities of the presynaptic boutons on the mossy fibers and the apical dendrites of the CA3 pyramidal cells were decreased. These results indicate that afadin plays roles not only in the formation of the mossy fiber synapses but also in the formation of the cellular architecture of the hippocampus and the dentate gyrus.

Astrocytes with previous chronic exposure to amyloid ß-peptide fragment 1-40 suppress excitatory synaptic transmission.

Synaptic dysfunction and neuronal death are responsible for cognitive and behavioral deficits in Alzheimer's disease (AD). It is well known that such neurological abnormalities are preceded by long-term exposure of amyloid ß-peptide (Aß) and/or hyperphosphorylated tau prior. In addition to the neurological deficit, astrocytes as a major glial cell type in the brain, significantly participate in the neuropathogenic mechanisms underlying synaptic modulation. Although astrocytes play a significant key role in modulating synaptic transmission, little is known on whether astrocyte dysfunction caused by such long-term Aß exposure affects synapse formation and function. Here, we show that synapse formation and synaptic transmission are attenuated in hippocampal-naïve neurons co-cultured with astrocytes that have previously experienced chronic Aß1-40 exposure. In this abnormal astrocytic condition, hippocampal neurons exhibit decrements of evoked excitatory post-synaptic currents (EPSCs) and miniature EPSC frequency. Furthermore, size of readily releasable synaptic pools and number of excitatory synapses were also significantly decreased. Contrary to these negative effects, release probability at individual synapses was significantly increased in the same astrocytic condition. Taken together, our data indicate that lower synaptic transmission caused by astrocytes previously, and chronically, exposed to Aß1-40 is attributable to a small number of synapses with higher release probability.

Reconstitution of Human Ion Channels into Solvent-free Lipid Bilayers Enhanced by Centrifugal Forces.

Artificially formed bilayer lipid membranes (BLMs) provide well-defined systems for functional analyses of various membrane proteins, including ion channels. However, difficulties associated with the integration of membrane proteins into BLMs limit the experimental efficiency and usefulness of such BLM reconstitution systems. Here, we report on the use of centrifugation to more efficiently reconstitute human ion channels in solvent-free BLMs. The method improves the probability of membrane fusion. Membrane vesicles containing the human ether-a-go-go-related gene (hERG) channel, the human cardiac sodium channel (Nav1.5), and the human GABAA receptor (GABAAR) channel were formed, and the functional reconstitution of the channels into BLMs via vesicle fusion was investigated. Ion channel currents were recorded in 67% of the BLMs that were centrifuged with membrane vesicles under appropriate centrifugal conditions (14-55 × g). The characteristic channel properties were retained for hERG, Nav1.5, and GABAAR channels after centrifugal incorporation into the BLMs. A comparison of the centrifugal force with reported values for the fusion force revealed that a centrifugal enhancement in vesicle fusion was attained, not by accelerating the fusion process but by accelerating the delivery of membrane vesicles to the surface of the BLMs, which led to an increase in the number of membrane vesicles that were available for fusion. Our method for enhancing the probability of vesicle fusion promises to dramatically increase the experimental efficiency of BLM reconstitution systems, leading to the realization of a BLM-based, high-throughput platform for functional assays of various membrane proteins.

Ticlopidine, a cholestatic liver injury-inducible drug, causes dysfunction of bile formation via diminished biliary secretion of phospholipids: involvement of biliary-excreted glutathione-conjugated ticlopidine metabolites.

The antiplatelet drug, ticlopidine (TIC), reportedly causes cholestatic liver injuries. The present study analyzed the effect of TIC on bile formation, revealing that the biliary secretion of phospholipids was significantly decreased in TIC-administered Sprague Dawley (SD) rats. However, the effect of TIC on biliary phospholipids was not observed in SD rats pretreated with diethylaminoethyl diphenylpropylacetate that inhibits cytochrome P450s (P450), or in Eisai hyperbilirubinemic rats (EHBR) lacking functional multidrug resistance-associated protein 2 (MRP2/ABCC2). These results suggest that glutathione-conjugated TIC metabolites (TIC-SGs), which were formed in the liver after P450s-mediated metabolism and were excreted extensively into bile by MRP2, mediated the observed alterations of the bile composition. Administration of TIC caused significant liver injuries in SD rats, with decreased biliary phospholipids, but not in EHBR, consistent with the in vitro observation that phospholipid-bile acid-mixed micelles moderated the cytotoxic effects of bile acids. Further analyses revealed that TIC-SGs did not directly inhibit multidrug resistance 3 P-glycoprotein (MDR3/ABCB4)-mediated phosphatidylcholine efflux in vitro. Because the diminished biliary secretion of phospholipids with TIC administration was restored by taurocholate infusion in SD rats, the decreased biliary concentration of bile acids, due to the stimulation of bile acid-independent bile flow driven by TIC-SGs, might have indirectly attenuated phospholipid secretion. In conclusion, extensive biliary excretion of TIC-SGs decreased the biliary secretion of phospholipids, which might have increased the risk of TIC-induced cholestatic liver injury.

Human mesenchymal stem-derived extracellular vesicles improve body growth and motor function following severe spinal cord injury in rat.

BACKGROUND: Spinal cord injury (SCI) in young adults leads to severe sensorimotor disabilities as well as slowing of growth. Systemic pro-inflammatory cytokines are associated with growth failure and muscle wasting. Here we investigated whether intravenous (IV) delivery of small extracellular vesicles (sEVs) derived from human mesenchymal stem/stromal cells (MSC) has therapeutic effects on body growth and motor recovery and can modulate inflammatory cytokines following severe SCI in young adult rats. METHODS: Contusional SCI rats were randomized into three different treatment groups (human and rat MSC-sEVs and a PBS group) on day 7 post-SCI. Functional motor recovery and body growth were assessed weekly until day 70 post-SCI. Trafficking of sEVs after IV infusions in vivo, the uptake of sEVs in vitro, macrophage phenotype at the lesion and cytokine levels at the lesion, liver and systemic circulation were also evaluated. RESULTS: An IV delivery of both human and rat MSC-sEVs improved functional motor recovery after SCI and restored normal body growth in young adult SCI rats, indicating a broad therapeutic benefit of MSC-sEVs and a lack of species specificity for these effects. Human MSC-sEVs were selectively taken up by M2 macrophages in vivo and in vitro, consistent with our previous observations of rat MSC-sEV uptake. Furthermore, the infusion of human or rat MSC-sEVs resulted in an increase in the proportion of M2 macrophages and a decrease in the production of the pro-inflammatory cytokines tumour necrosis factor-alpha (TNF-α) and interleukin (IL)-6 at the injury site, as well as a reduction in systemic serum levels of TNF-α and IL-6 and an increase in growth hormone receptors and IGF-1 levels in the liver. CONCLUSIONS: Both human and rat MSC-sEVs promote the recovery of body growth and motor function after SCI in young adult rats possibly via the cytokine modulation of growth-related hormonal pathways. Thus, MSC-sEVs affect both metabolic and neurological deficits in SCI.

Two-dimensional water-molecule-cluster layers at nanobubble interfaces.

HYPOTHESIS: Bulk nanobubbles (NBs) have high surface charge densities and long lifetimes. Despite several attempts to understand the lifetime of NBs, their interfacial layer structure remains unknown. It is hypothesized that a specific interfacial layer exists with a hydrogen bond network that stabilizes NBs. EXPERIMENTS: In situ infrared reflectance-absorption spectroscopy and density functional theory were used to determine the interfacial layer structure of NBs. Furthermore, nuclear magnetic resonance spectroscopy was used to examine the interfacial layer hardness of bubbles filled with N2, O2, and CO2, which was expected to depend on the encapsulated gas species. FINDINGS: The interfacial layer was composed of three-, four-, and five-membered ring clusters of water molecules. An interface model was proposed in which a two-dimensional layer of clusters with large electric dipole moments is oriented toward the endohedral gas, and the hydrophobic surface is adjacent to the free water. The interfacial layer hardness was dependent on the interaction with the gas (N2 > O2 > CO2), which supports the proposed interface model. These findings can be generalized to the structure of water at gas-water interfaces.

Modular architecture facilitates noise-driven control of synchrony in neuronal networks.

High-level information processing in the mammalian cortex requires both segregated processing in specialized circuits and integration across multiple circuits. One possible way to implement these seemingly opposing demands is by flexibly switching between states with different levels of synchrony. However, the mechanisms behind the control of complex synchronization patterns in neuronal networks remain elusive. Here, we use precision neuroengineering to manipulate and stimulate networks of cortical neurons in vitro, in combination with an in silico model of spiking neurons and a mesoscopic model of stochastically coupled modules to show that (i) a modular architecture enhances the sensitivity of the network to noise delivered as external asynchronous stimulation and that (ii) the persistent depletion of synaptic resources in stimulated neurons is the underlying mechanism for this effect. Together, our results demonstrate that the inherent dynamical state in structured networks of excitable units is determined by both its modular architecture and the properties of the external inputs.

Differential neurite outgrowth is required for axon specification by cultured hippocampal neurons.

Formation of an axon is the first morphological evidence of neuronal polarization, visible as a profound outgrowth of the axon compared with sibling neurites. One unsolved question on the mechanism of axon formation is the role of axon outgrowth in axon specification. This question was difficult to assess, because neurons freely extend their neurites in a conventional culture. Here, we leveraged surface nano/micro-modification techniques to fabricate a template substrate for constraining neurite lengths of cultured neurons. Using the template, we asked (i) Do neurons polarize even if all neurites cannot grow sufficiently long? (ii) Would the neurite be fated to become an axon if only one was allowed to grow long? A pattern with symmetrical short paths (20 µm) was used to address the former question, and an asymmetrical pattern with one path extended to 100 µm for the latter. Axon formation was evaluated by tau-1/MAP2 immunostaining and live-cell imaging of constitutively-active kinesin-1. We found that (1) neurons cannot polarize when extension of all neurites is restricted and that (2) when only a single neurite is permitted to grow long, neurons polarize and the longest neurite becomes the axon. These results provide clear evidence that axon outgrowth is required for its specification.

Microfluidic cell engineering on high-density microelectrode arrays for assessing structure-function relationships in living neuronal networks.

Neuronal networks in dissociated culture combined with cell engineering technology offer a pivotal platform to constructively explore the relationship between structure and function in living neuronal networks. Here, we fabricated defined neuronal networks possessing a modular architecture on high-density microelectrode arrays (HD-MEAs), a state-of-the-art electrophysiological tool for recording neural activity with high spatial and temporal resolutions. We first established a surface coating protocol using a cell-permissive hydrogel to stably attach a polydimethylsiloxane microfluidic film on the HD-MEA. We then recorded the spontaneous neural activity of the engineered neuronal network, which revealed an important portrait of the engineered neuronal network-modular architecture enhances functional complexity by reducing the excessive neural correlation between spatially segregated modules. The results of this study highlight the impact of HD-MEA recordings combined with cell engineering technologies as a novel tool in neuroscience to constructively assess the structure-function relationships in neuronal networks.

Metabolic engineering for the production of prenylated polyphenols in transgenic legume plants using bacterial and plant prenyltransferases.

Prenylated polyphenols are secondary metabolites beneficial for human health because of their various biological activities. Metabolic engineering was performed using Streptomyces and Sophora flavescens prenyltransferase genes to produce prenylated polyphenols in transgenic legume plants. Three Streptomyces genes, NphB, SCO7190, and NovQ, whose gene products have broad substrate specificity, were overexpressed in a model legume, Lotus japonicus, in the cytosol, plastids or mitochondria with modification to induce the protein localization. Two plant genes, N8DT and G6DT, from Sophora flavescens whose gene products show narrow substrate specificity were also overexpressed in Lotus japonicus. Prenylated polyphenols were undetectable in these plants; however, supplementation of a flavonoid substrate resulted in the production of prenylated polyphenols such as 7-O-geranylgenistein, 6-dimethylallylnaringenin, 6-dimethylallylgenistein, 8-dimethylallynaringenin, and 6-dimethylallylgenistein in transgenic plants. Although transformants with the native NovQ did not produce prenylated polyphenols, modification of its codon usage led to the production of 6-dimethylallylnaringenin and 6-dimethylallylgenistein in transformants following naringenin supplementation. Prenylated polyphenols were not produced in mitochondrial-targeted transformants even under substrate feeding. SCO7190 was also expressed in soybean, and dimethylallylapigenin and dimethylallyldaidzein were produced by supplementing naringenin. This study demonstrated the potential for the production of novel prenylated polyphenols in transgenic plants. In particular, the enzymatic properties of prenyltransferases seemed to be altered in transgenic plants in a host species-dependent manner.

Successive deaths of a captive snow leopard (Uncia uncia) and a serval (Leptailurus serval) by infection with feline panleukopenia virus at Sapporo Maruyama Zoo.

Feline parvoviruses were isolated from frozen samples of intestines taken from a snow leopard (Uncia uncia) and a serval (Leptailurus serval) that died successively at Sapporo Maruyama Zoo in Hokkaido, Japan. Isolates possessed an antigenic epitope for both the feline panleukopenia virus (FPLV) and mink enteritis virus, identified with a hemagglutination inhibition test. Sequencing analyses of the VP2 region of the isolates revealed that the two isolates were identical and of the FPLV-type. These results suggested that FPLV was introduced from a feral cat which entered the zoo and transmitted the virus inside the zoo.

Computational Efficiency of a Modular Reservoir Network for Image Recognition.

Liquid state machine (LSM) is a type of recurrent spiking network with a strong relationship to neurophysiology and has achieved great success in time series processing. However, the computational cost of simulations and complex dynamics with time dependency limit the size and functionality of LSMs. This paper presents a large-scale bioinspired LSM with modular topology. We integrate the findings on the visual cortex that specifically designed input synapses can fit the activation of the real cortex and perform the Hough transform, a feature extraction algorithm used in digital image processing, without additional cost. We experimentally verify that such a combination can significantly improve the network functionality. The network performance is evaluated using the MNIST dataset where the image data are encoded into spiking series by Poisson coding. We show that the proposed structure can not only significantly reduce the computational complexity but also achieve higher performance compared to the structure of previous reported networks of a similar size. We also show that the proposed structure has better robustness against system damage than the small-world and random structures. We believe that the proposed computationally efficient method can greatly contribute to future applications of reservoir computing.