Search

The learning and inference efficiencies of an artificial neural network represented by a cross-point synaptic memristor array can be achieved using a selector, with high selectivity (I on /I off ) and sufficient death region, stacked vertically on a synaptic memristor. This can prevent a sneak current in the memristor array. A selector with multiple jar-shaped conductive Cu filaments in the resistive switching layer is precisely fabricated by designing the Cu ion concentration depth profile of the CuGeSe layer as a filament source, TiN diffusion barrier layer, and Ge 3 Se 7 switching layer. The selector performs super-linear-threshold-switching with a selectivity of > 10 7 , death region of -0.70-0.65 V, holding time of 300 ns, switching speed of 25 ns, and endurance cycle of > 10 6 . In addition, the mechanism of switching is proven by the formation of conductive Cu filaments between the CuGeSe and Ge 3 Se 7 layers under a positive bias on the top Pt electrode and an automatic rupture of the filaments after the holding time. Particularly, a spiking deep neural network using the designed one-selector-one-memory cross-point array improves the Modified National Institute of Standards and Technology classification accuracy by ≈3.8% by eliminating the sneak current in the cross-point array during the inference process., (© 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH.)

The response criteria for complete remission (CR) in acute myeloid leukemia (AML) are currently based on morphology and blood cell counts. However, these criteria are insufficient to establish a diagnosis in cases with poor quality bone marrow (BM) samples demonstrating a loss of cellular morphology. We investigated whether the sera of patients contained biomarkers that indicate disease response status. First, we performed multidimensional liquid chromatography-differential gel electrophoresis (MDLC-DIGE) to generate protein profiles of two pooled, paired serum samples from patients who had achieved CR; one collected at diagnosis (PreCR) and the other collected after chemotherapy (CR). Then, with the biomarker candidates found, ELISA was carried out for individual PreCR and CR samples, and for other verification sets including nonremission (NR) patients and normal samples. We selected two proteins, complement factor H (CFH) and apolipoprotein H (ApoH), with dye (Cy) ratios showing greater than 2.0-fold differences between the pooled samples. ELISA showed that CFH and ApoH are useful for distinguishing between the recovered (CR and normal) and nonrecovered (PreCR, PreNR, and NR) states in AML (p <0.001). We successfully applied a protein profiling technology of MDLC-DIGE and LC-MS/MS to discover two biomarkers for CR which needs further validation for a clinical setting., (© 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.)

Mesenchymal stem cells (MSCs) are multipotent cells, which have the capability to differentiate into various mesenchymal tissues such as bone, cartilage, fat, tendon, muscle, and marrow stroma. However, they lose the capability of multi-lineage differentiation after several passages. It is known that basic fibroblast growth factor (bFGF) increases growth rate, differentiation potential, and morphological changes of MSCs in vitro. In this report, we have used 2-DE coupled to MS to identify differentially expressed proteins at the cell membrane level in MSCs growing in bFGF containing medium. The cell surface proteins isolated by the biotin-avidin affinity column were separated by 2-DE in triplicate experiments. A total of 15 differentially expressed proteins were identified by quadrupole-time of flight tandem MS. Nine of the proteins were upregulated and six proteins were downregulated in the MSCs cultured with bFGF containing medium. The expression level of three actin-related proteins, F-actin-capping protein subunit alpha-1, actin-related protein 2/3 complex subunit 2, and myosin regulatory light chain 2, was confirmed by Western blot analysis. The results indicate that the expression levels of F-actin-capping protein subunit alpha-1, actin-related protein 2/3 complex subunit 2, and myosin regulatory light chain 2 are important in bFGF-induced morphological change of MSCs.

A protein identified in multiple separate bands of a 1-D gel reflects variation in the molecular weight caused by alternative splicing, endoproteolytic cleavage, or PTMs, such as glycosylation or ubiquitination. To characterize such a protein distribution over the bands, we defined an entity called an 'island' as the band region including the bands of the same protein identified sequentially. We quantified the island distribution using a new variable called an Iscore. Previously, as described in Park et al.. (Proteomics 2006, 6, 4978-4986.), we analyzed human brain tissue using a multidimensional MS/MS separation method. Here, the new method of island analysis was applied to the previous proteome data. The soluble and membrane protein fractions of human brain tissue were reanalyzed using the island distribution. The proteome of the soluble fraction exhibited more variation in island positions than that of the membrane fraction. Through the island analysis, we identified protein modifications and protein complexes over the 1-D gel bands.

Gram-negative bacteria constitutively secrete native outer membrane vesicles (OMVs) into the extracellular milieu. Although recent progress in this area has revealed that OMVs are essential for bacterial survival and pathogenesis, the mechanism of vesicle formation and the biological roles of OMVs have not been clearly defined. Using a proteomics approach, we identified 141 protein components of Escherichia coli-derived native OMVs with high confidence; two separate analyses yielded identifications of 104 and 117 proteins, respectively, with 80 proteins overlapping between the two trials. In the group of identified proteins, the outer membrane proteins were highly enriched, whereas inner membrane proteins were lacking, suggesting that a specific sorting mechanism for vesicular proteins exists. We also identified proteins involved in vesicle formation, the removal of toxic compounds and attacking phage, and the elimination of competing organisms, as well as those involved in facilitating the transfer of genetic material and protein to other bacteria, targeting host cells, and modulating host immune responses. This study provides a global view of native bacterial OMVs. This information will help us not only to elucidate the biogenesis and functions of OMV from nonpathogenic and pathogenic bacteria but also to develop vaccines and antibiotics effective against pathogenic strains.

In our initial attempt to analyze the human brain proteome, we applied multi-dimensional protein separation and identification techniques using a combination of sample fractionation, 1-D SDS-PAGE, and MS analysis. The complexity of human brain proteome requires multiple fractionation strategies to extend the range and total number of proteins identified. According to the method of Klose (Methods Mol. Biol. 1999, 112, 67), proteins of the temporal lobe of human brain were fractionated into (i) cytoplasmic and nucleoplasmic, (ii) membrane and other structural, and (iii) DNA-binding proteins. Each fraction was then separated by SDS-PAGE, and the resulting gel line was cut into approximately 50 bands. After trypsin digestion, the resulting peptides from each band were analyzed by RP-LC/ESI-MS/MS using an LTQ spectrometer. The SEQUEST search program, which searched against the IPI database, was used for peptide sequence identification, and peptide sequences were validated by reversed sequence database search and filtered by the Protein Hit Score. Ultimately, 1533 proteins could be detected from the human brain. We classified the identified proteins according to their distribution on cellular components. Among these proteins, 24% were membrane proteins. Our results show that the multiple separation strategy is effective for high-throughput characterization of proteins from complex proteomic mixtures.

In shotgun proteomics, proteins can be fractionated by 1-D gel electrophoresis and digested into peptides, followed by liquid chromatography to separate the peptide mixture. Mass spectrometry generates hundreds of thousands of tandem mass spectra from these fractions, and proteins are identified by database searching. However, the search scores are usually not sufficient to distinguish the correct peptides. In this study, we propose a confident protein identification method for high-throughput analysis of human proteome. To build a filtering protocol in database search, we chose Pseudomonas putida KT2440 as a reference because this bacterial proteome contains fewer modifications and is simpler than the human proteome. First, the P. putida KT2440 proteome was filtered by reversed sequence database search and correlated by the molecular weight in 1-D-gel band positions. The characterization protocol was then applied to determine the criteria for clustering of the human plasma proteome into three different groups. This protein filtering method, based on bacterial proteome data analysis, represents a rapid way to generate higher confidence protein list of the human proteome, which includes some of heavily modified and cleaved proteins.

Proteomic analysis of Pseudomonas putida KT2440 cultured in monocyclic aromatic compounds was performed using 2-DE/MS and cleavable isotope-coded affinity tag (ICAT) to determine whether proteins involved in aromatic compound degradation pathways were altered as predicted by genomic analysis (Jiménez et al., Environ Microbiol. 2002, 4, 824-841). Eighty unique proteins were identified by 2-DE/MS or MS/MS analysis from P. putida KT2440 cultured in the presence of six different organic compounds. Benzoate dioxygenase (BenA, BenD) and catechol 1,2-dioxygenase (CatA) were induced by benzoate. Protocatechuate 3,4-dixoygenase (PcaGH) was induced by p-hydroxybenzoate and vanilline. beta-Ketoadipyl CoA thiolase (PcaF) and 3-oxoadipate enol-lactone hydrolase (PcaD) were induced by benzoate, p-hydroxybenzoate and vanilline, suggesting that benzoate, p-hydroxybenzoate and vanilline were degraded by different dioxygenases and then converged in the same beta-ketoadipate degradation pathway. An additional 110 proteins, including 19 proteins from 2-DE analysis, were identified by cleavable ICAT analysis for benzoate-induced proteomes, which complemented the 2-DE results. Phenylethylamine exposure induced beta-ketoacyl CoA thiolase (PhaD) and ring-opening enzyme (PhaL), both enzymes of the phenylacetate (pha) biodegradation pathway. Phenylalanine induced 4-hydroxyphenyl-pyruvate dioxygenase (Hpd) and homogentisate 1,2-dioxygenase (HmgA), key enzymes in the homogentisate degradation pathway. Alkyl hydroperoxide reductase (AphC) was induced under all aromatic compounds conditions. These results suggest that proteome analysis complements and supports predictive information obtained by genomic sequence analysis.

The proteome of a HUPO human serum reference sample was analyzed using multidimensional separation techniques at both the protein and the peptide levels. To eliminate false-positive identifications from the search results, we employed a data filtering method using molecular weight (MW) correlations derived from denaturing 1-DE. First, the six most abundant serum proteins were removed from the sample using immunoaffinity chromatography. 1-DE was then used to fractionate the remaining serum proteins according to the MW. Gel bands were isolated and in-gel digested with trypsin, and the resulting peptides were analyzed by 2-D LC/ESI-MS/MS. A SEQUEST search using the MS/MS results identified 494 proteins. Of these, 202 were excluded formally using protein data filtering as they were single-assignment proteins and their theoretical and electrophoretically-derived MWs did not correlate at high confidence. To evaluate this method, the results were compared with those of 1-D LC/MALDI-TOF/TOF and HUPO Plasma Proteome Project analyses. Our data filtering approach proved valuable in analysis of complex, large-scale proteomes such as human serum.

We compared peptide identification by database (DB) search methods with de novo sequencing results for proteomics study in an organism without genome sequence information. When the former was done by searching the Expressed Sequence Tag (EST) DB of the sample organism or the NCBI nonredundant (nr) protein DB of green plants using either the MASCOT or SEQUEST software program, it was confirmed that the former is as accurate as the latter. Peptides identified from EST DB were twice as many as those from the nr protein DB, in spite of the fact that the EST DB has less data (26 222 EST) than the NCBI nr protein DB (224 238). This study demonstrates that EST DB with tandem mass spectra can be used reliably for high-throughput proteomics studies in an organism without genome information.

We performed comparative proteomic analyses in order to understand the physiological responses of ginseng (Panax ginseng C. A. Meyer) to high light (HL). As a first step, we analyzed the proteins expressed in ginseng leaves. Proteins extracted from leaves were separated by two-dimensional polyacrylamide gel electrophoresis. Protein spots were identified by tandem mass spectra analysis using electrospray ionization quadrupole-time of flight mass spectrometry (ESI Q-TOF MS). We used a ginseng expressed sequence tag (EST) database as well as a nonredundant protein database from NCBI to identify proteins. Eighty-one proteins were identified using the nr protein database, 51 of which were also verified from the ginseng EST database. An additional 66 proteins were identified only from the ginseng EST database. Proteins that function in energy metabolism, protein stabilization, and protection against oxidative stress were abundant. To understand the light responses of ginseng leaves, we studied time dependent changes in expressed proteins produced by 0-4 h of HL exposure. Six HL-responsive proteins were identified: three proteins were up-regulated (cytosolic small heat-shock protein, cytosolic ascorbate peroxidase, and putative major latex-like protein) and three proteins were down-regulated (Rieske Fe/S protein, putative 3-beta hydroxysteroid dehydrogenase/isomerase-like protein, and oxygen-evolving enhancer-like protein). Our results show that the ginseng EST database combined with ESI Q-TOF MS analysis can be used to identify ginseng proteins and to elucidate the protective mechanism of ginseng against HL induced damage.

As an initial step to the comprehensive proteomic analysis of Panax ginseng C. A. Meyer, protein mixtures extracted from the cultured hairy root of Panax ginseng were separated by two-dimensional polyacrylamide gel electrophoresis (2-DE). The protein spots were analyzed and identified by peptide finger printing and internal amino acid sequencing by matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) and electrospray ionization quadrupole-time of flight mass spectrometry (ESI Q-TOF MS), respectively. More than 300 protein spots were detected on silver stained two-dimensional (2-D) gels using pH 3-10, 4-7, and 4.5-5.5 gradients. Major protein spots (159) were analyzed by peptide fingerprinting or de novo sequencing and the functions of 91 of these proteins were identified. Protein identification was achieved using the expressed sequence tag (EST) database from Panax ginseng and the protein database of plants like Arabidopsis thaliana and Oryza sativa. However, peptide mass fingerprinting by MALDI-TOF MS alone was insufficient for protein identification because of the lack of a genome database for Panax ginseng. Only 17 of the 159 protein spots were verified by peptide mass fingerprinting using MALDI-TOF MS whereas 87 out of 102 protein spots, which included 13 of the 17 proteins identified by MALDI-TOF MS, were identified by internal amino acid sequencing using tandem mass spectrometry analysis by ESI Q-TOF MS. When the internal amino acid sequences were used as identification markers, the identification rate exceeded 85.3%, suggesting that a combination of internal sequencing and EST data analysis was an efficient identification method for proteome analysis of plants having incomplete genome data like ginseng. The 2-D patterns of the main root and leaves of Panax ginseng differed from that of the cultured hairy root, suggesting that some proteins are exclusively expressed by different tissues for specific cellular functions. Proteome analysis will undoubtedly be helpful for understanding the physiology of Panax ginseng.