CPEB2-activated axonal translation of VGLUT2 mRNA promotes glutamatergic transmission and presynaptic plasticity.

BACKGROUND: Local translation at synapses is important for rapidly remodeling the synaptic proteome to sustain long-term plasticity and memory. While the regulatory mechanisms underlying memory-associated local translation have been widely elucidated in the postsynaptic/dendritic region, there is no direct evidence for which RNA-binding protein (RBP) in axons controls target-specific mRNA translation to promote long-term potentiation (LTP) and memory. We previously reported that translation controlled by cytoplasmic polyadenylation element binding protein 2 (CPEB2) is important for postsynaptic plasticity and memory. Here, we investigated whether CPEB2 regulates axonal translation to support presynaptic plasticity. METHODS: Behavioral and electrophysiological assessments were conducted in mice with pan neuron/glia- or glutamatergic neuron-specific knockout of CPEB2. Hippocampal Schaffer collateral (SC)-CA1 and temporoammonic (TA)-CA1 pathways were electro-recorded to monitor synaptic transmission and LTP evoked by 4 trains of high-frequency stimulation. RNA immunoprecipitation, coupled with bioinformatics analysis, were used to unveil CPEB2-binding axonal RNA candidates associated with learning, which were further validated by Western blotting and luciferase reporter assays. Adeno-associated viruses expressing Cre recombinase were stereotaxically delivered to the pre- or post-synaptic region of the TA circuit to ablate Cpeb2 for further electrophysiological investigation. Biochemically isolated synaptosomes and axotomized neurons cultured on a microfluidic platform were applied to measure axonal protein synthesis and FM4-64FX-loaded synaptic vesicles. RESULTS: Electrophysiological analysis of hippocampal CA1 neurons detected abnormal excitability and vesicle release probability in CPEB2-depleted SC and TA afferents, so we cross-compared the CPEB2-immunoprecipitated transcriptome with a learning-induced axonal translatome in the adult cortex to identify axonal targets possibly regulated by CPEB2. We validated that Slc17a6, encoding vesicular glutamate transporter 2 (VGLUT2), is translationally upregulated by CPEB2. Conditional knockout of CPEB2 in VGLUT2-expressing glutamatergic neurons impaired consolidation of hippocampus-dependent memory in mice. Presynaptic-specific ablation of Cpeb2 in VGLUT2-dominated TA afferents was sufficient to attenuate protein synthesis-dependent LTP. Moreover, blocking activity-induced axonal Slc17a6 translation by CPEB2 deficiency or cycloheximide diminished the releasable pool of VGLUT2-containing synaptic vesicles. CONCLUSIONS: We identified 272 CPEB2-binding transcripts with altered axonal translation post-learning and established a causal link between CPEB2-driven axonal synthesis of VGLUT2 and presynaptic translation-dependent LTP. These findings extend our understanding of memory-related translational control mechanisms in the presynaptic compartment.

Identification of a damage-associated molecular pattern (DAMP) receptor and its cognate peptide ligand in sweet potato (Ipomoea batatas).

Sweet potato (Ipomoea batatas) is an important tuber crop, but also target of numerous insect pests. Intriguingly, the abundant storage protein in tubers, sporamin, has intrinsic trypsin protease inhibitory activity. In leaves, sporamin is induced by wounding or a volatile homoterpene and enhances insect resistance. While the signalling pathway leading to sporamin synthesis is partially established, the initial event, perception of a stress-related signal is still unknown. Here, we identified an IbLRR-RK1 that is induced upon wounding and herbivory, and related to peptide-elicitor receptors (PEPRs) from tomato and Arabidopsis. We also identified a gene encoding a precursor protein comprising a peptide ligand (IbPep1) for IbLRR-RK1. IbPep1 represents a distinct signal in sweet potato, which might work in a complementary and/or parallel pathway to the previously described hydroxyproline-rich systemin (HypSys) peptides to strengthen insect resistance. Notably, an interfamily compatibility in the Pep/PEPR system from Convolvulaceae and Solanaceae was identified.

PRAP1 is a novel lipid-binding protein that promotes lipid absorption by facilitating MTTP-mediated lipid transport.

Microsomal triglyceride transfer protein (MTTP) is an endoplasmic reticulum resident protein that is essential for the assembly and secretion of triglyceride (TG)-rich, apoB-containing lipoproteins. Although the function and structure of mammalian MTTP have been extensively studied, how exactly MTTP transfers lipids to lipid acceptors and whether there are other biomolecules involved in MTTP-mediated lipid transport remain elusive. Here we identify a role in this process for the poorly characterized protein PRAP1. We report that PRAP1 and MTTP are partially colocalized in the endoplasmic reticulum. We observe that PRAP1 directly binds to TG and facilitates MTTP-mediated lipid transfer. A single amino acid mutation at position 85 (E85V) impairs PRAP1's ability to form a ternary complex with TG and MTTP, as well as impairs its ability to facilitate MTTP-mediated apoB-containing lipoprotein assembly and secretion, suggesting that the ternary complex formation is required for PRAP1 to facilitate MTTP-mediated lipid transport. PRAP1 is detectable in chylomicron/VLDL-rich plasma fractions, suggesting that MTTP recognizes PRAP1-bound TG as a cargo and transfers TG along with PRAP1 to lipid acceptors. Both PRAP1-deficient and E85V knock-in mutant mice fed a chow diet manifested an increase in the length of their small intestines, likely to compensate for challenges in absorbing lipid. Interestingly, both genetically modified mice gained significantly less body weight and fat mass when on high-fat diets compared with littermate controls and were prevented from hepatosteatosis. Together, this study provides evidence that PRAP1 plays an important role in MTTP-mediated lipid transport and lipid absorption.

JAK2-CHK2 signaling safeguards the integrity of the mitotic spindle assembly checkpoint and genome stability.

Checkpoint kinase 2 (CHK2) plays an important role in safeguarding the mitotic progression, specifically the spindle assembly, though the mechanism of regulation remains poorly understood. Here, we identified a novel mitotic phosphorylation site on CHK2 Tyr156, and its responsible kinase JAK2. Expression of a phospho-deficient mutant CHK2 Y156F or treatment with JAK2 inhibitor IV compromised mitotic spindle assembly, leading to genome instability. In contrast, a phospho-mimicking mutant CHK2 Y156E restored mitotic normalcy in JAK2-inhibited cells. Mechanistically, we show that this phosphorylation is required for CHK2 interaction with and phosphorylation of the spindle assembly checkpoint (SAC) kinase Mps1, and failure of which results in impaired Mps1 kinetochore localization and defective SAC. Concordantly, analysis of clinical cancer datasets revealed that deletion of JAK2 is associated with increased genome alteration; and alteration in CHEK2 and JAK2 is linked to preferential deletion or amplification of cancer-related genes. Thus, our findings not only reveal a novel JAK2-CHK2 signaling axis that maintains genome integrity through SAC but also highlight the potential impact on genomic stability with clinical JAK2 inhibition.

Cryo-EM structure of human hexameric MCM2-7 complex.

The central step in the initiation of eukaryotic DNA replication is the loading of the minichromosome maintenance 2-7 (MCM2-7) complex, the core of the replicative DNA helicase, onto chromatin at replication origin. Here, we reported the cryo-EM structure of endogenous human single hexameric MCM2-7 complex with a resolution at 4.4 Å, typically an open-ring hexamer with a gap between Mcm2 and Mcm5. Strikingly, further analysis revealed that human MCM2-7 can self-associate to form a loose double hexamer which potentially implies a novel mechanism underlying the MCM2-7 loading in eukaryote. The high-resolution cryo-EM structure of human MCM2-7 is critical for understanding the molecular mechanisms governing human DNA replication, especially the MCM2-7 chromatin loading and pre-replicative complex assembly.

CRM1-spike-mediated nuclear export of hepatitis B virus encapsidated viral RNA.

Hepatitis B virus (HBV) is a global pathogen. We report here that the cellular CRM1 machinery can mediate nuclear export of entire HBV core (HBc) particles containing encapsidated viral RNAs. Two CRM1-mediated nuclear export signals (NESCRM1) cluster at the conformationally flexible spike tips of HBc particles. Mutant NESCRM1 capsids exhibit strongly reduced associations with CRM1 and nucleoporin358 in vivo. CRM1 and NXF1 machineries mediate nuclear export of HBc particles independently. Inhibition of nuclear export has pleiotropic consequences, including nuclear accumulation of HBc particles, a significant reduction of encapsidated viral RNAs in the cytoplasm but not in the nucleus, and barely detectable viral DNA. We hypothesize an HBV life cycle where encapsidation of the RNA pregenome can initiate early in the nucleus, whereas DNA genome maturation occurs mainly in the cytoplasm. We identified a druggable target for HBV by blocking its intracellular trafficking.

Proline in transmembrane domain of type II protein DPP-IV governs its translocation behavior through endoplasmic reticulum.

A transmembrane domain (TMD) at the N-terminus of a membrane protein is a signal sequence that targets the protein to the endoplasmic reticulum (ER) membrane. Proline is found more frequently in TM helices compared to water-soluble helices. To investigate the effects of proline on protein translocation and integration in mammalian cells, we made proline substitutions throughout the TMD of dipeptidyl peptidase IV, a type II membrane protease with a single TMD at its N-terminus. The proteins were expressed and their capacities for targeting and integrating into the membrane were measured in both mammalian cells and in vitro translation systems. Three proline substitutions in the central region of the TMD resulted in various defects in membrane targeting and/or integration. The replacement of proline with other amino acids of similar hydrophobicity rescued both the translocation and anchoring defects of all three proline mutants, indicating that conformational change caused by proline is a determining factor. Increasing hydrophobicity of the TMD by replacing other residues with more hydrophobic residues also effectively reversed the translocation and integration defects. Intriguingly, increasing hydrophobicity at the C-terminal end of the TMD rescued much more effectively than it did at the N-terminal end. Thus, the effect of proline on translocation and integration of the TMD is not determined solely by its conformation and hydrophobicity, but also by the location of proline in the TMD, the location of highly hydrophobic residues, and the relative position of the proline to other proline residues in the TMD.

The extracellular K+ concentration dependence of outward currents through Kir2.1 channels is regulated by extracellular Na+ and Ca2+.

It has been known for more than three decades that outward Kir currents (I(K1)) increase with increasing extracellular K(+) concentration ([K(+)](o)). Although this increase in I(K1) can have significant impacts under pathophysiological cardiac conditions, where [K(+)](o) can be as high as 18 mm and thus predispose the heart to re-entrant ventricular arrhythmias, the underlying mechanism has remained unclear. Here, we show that the steep [K(+)](o) dependence of Kir2.1-mediated outward I(K1) was due to [K(+)](o)-dependent inhibition of outward I(K1) by extracellular Na(+) and Ca(2+). This could be accounted for by Na(+)/Ca(2+) inhibition of I(K1) through screening of local negative surface charges. Consistent with this, extracellular Na(+) and Ca(2+) reduced the outward single-channel current and did not increase open-state noise or decrease the mean open time. In addition, neutralizing negative surface charges with a carboxylate esterifying agent inhibited outward I(K1) in a similar [K(+)](o)-dependent manner as Na(+)/Ca(2+). Site-directed mutagenesis studies identified Asp(114) and Glu(153) as the source of surface charges. Reducing K(+) activation and surface electrostatic effects in an R148Y mutant mimicked the action of extracellular Na(+) and Ca(2+), suggesting that in addition to exerting a surface electrostatic effect, Na(+) and Ca(2+) might inhibit outward I(K1) by inhibiting K(+) activation. This study identified interactions of K(+) with Na(+) and Ca(2+) that are important for the [K(+)](o) dependence of Kir2.1-mediated outward I(K1).

Outer membrane protein I of Pseudomonas aeruginosa is a target of cationic antimicrobial peptide/protein.

Cationic antimicrobial peptides/proteins (AMPs) are important components of the host innate defense mechanisms against invading microorganisms. Here we demonstrate that OprI (outer membrane protein I) of Pseudomonas aeruginosa is responsible for its susceptibility to human ribonuclease 7 (hRNase 7) and alpha-helical cationic AMPs, instead of surface lipopolysaccharide, which is the initial binding site of cationic AMPs. The antimicrobial activities of hRNase 7 and alpha-helical cationic AMPs against P. aeruginosa were inhibited by the addition of exogenous OprI or anti-OprI antibody. On modification and internalization of OprI by hRNase 7 into cytosol, the bacterial membrane became permeable to metabolites. The lipoprotein was predicted to consist of an extended loop at the N terminus for hRNase 7/lipopolysaccharide binding, a trimeric alpha-helix, and a lysine residue at the C terminus for cell wall anchoring. Our findings highlight a novel mechanism of antimicrobial activity and document a previously unexplored target of alpha-helical cationic AMPs, which may be used for screening drugs to treat antibiotic-resistant bacterial infection.

Testing the balanced electrostatic interaction hypothesis of hepatitis B virus DNA synthesis by using an in vivo charge rebalance approach.

Previously, a charge balance hypothesis was proposed to explain hepatitis B virus (HBV) capsid stability, assembly, RNA encapsidation, and DNA replication. This hypothesis emphasized the importance of a balanced electrostatic interaction between the positive charge from the arginine-rich domain (ARD) of the core protein (HBc) and the negative charge from the encapsidated nucleic acid. It remains unclear if any of the negative charge involved in this electrostatic interaction could come from the HBc protein per se, in addition to the encapsidated nucleic acid. HBc ARD IV mutant 173GG and ARD II mutant 173RR/R157A/R158A are arginine deficient and replication defective. Not surprisingly, the replication defect of ARD IV mutant 173GG can be rescued by restoring positively charged amino acids at the adjacent positions 174 and 175. However, most interestingly, it can be at least partially rescued by reducing negatively charged residues in the assembly domain, such as by glutamic acid-to-alanine (E-to-A) substitutions at position 46 or 117 and to a much lesser extent at position 113. Similar results were obtained for ARD II mutant 173RR/R157A/R158A. These amino acids are located on the inner surfaces of HBc icosahedral particles, and their acidic side chains point toward the capsid interior. For HBV DNA synthesis, the relative amount of positive versus negative charge in the electrostatic interactions is more important than the absolute amount of positive or negative charge. These results support the concept that balanced electrostatic interaction is important during the viral life cycle.

Improving blood cells classification in peripheral blood smears using enhanced incremental training.

Peripheral Blood Smear (PBS) analysis is a vital routine test carried out by medical specialists to assess some health aspects of individuals. The automation of blood analysis has attracted the attention of researchers in recent years, as it will not only save time, money and reduce errors, but also protect and save lives of front-line workers, especially during pandemics. In this work, deep neural networks are trained on a synthetic blood smears dataset to classify fifteen different white blood cell and platelet subtypes and morphological abnormalities. For classifying platelets, a hybrid approach of deep learning and image processing techniques is proposed. This approach improved the platelet classification accuracy and macro-average precision from 82.6% to 98.6% and 76.6%-97.6% respectively. Moreover, for white blood cell classification, a novel scheme for training deep networks is proposed, namely, Enhanced Incremental Training, that automatically recognises and handles classes that confuse and negatively affect neural network predictions. To handle the confusable classes, we also propose a procedure called "training revert". Application of the proposed method has improved the classification accuracy and macro-average precision from 61.5% to 95% and 76.6%-94.27% respectively.

Oligomerization is crucial for the stability and function of heme oxygenase-1 in the endoplasmic reticulum.

Heme oxygenase-1 (HO-1), a stress-inducible enzyme anchored in the endoplasmic reticulum (ER) by a single transmembrane segment (TMS) located at the C terminus, interacts with NADPH cytochrome P450 reductase and biliverdin reductase to catalyze heme degradation to biliverdin and its metabolite, bilirubin. Previous studies suggested that HO-1 functions as a monomer. Using chemical cross-linking, co-immunoprecipitation, and fluorescence resonance energy transfer (FRET) experiments, here we showed that HO-1 forms dimers/oligomers in the ER. However, oligomerization was not observed with a truncated HO-1 lacking the C-terminal TMS (amino acids 266-285), which exhibited cytosolic and nuclear localization, indicating that the TMS is essential for the self-assembly of HO-1 in the ER. To identify the interface involved in the TMS-TMS interaction, residue Trp-270, predicted by molecular modeling as a potential interfacial residue of TMS alpha-helices, was mutated, and the effects on protein subcellular localization and activity assessed. The results showed that the W270A mutant was present exclusively in the ER and formed oligomers with similar activity to those of the wild type HO-1. Interestingly, the W270N mutant was localized not only in the ER, but also in the cytosol and nucleus, suggesting it is susceptible to proteolytic cleavage. Moreover, the microsomal HO activity of the W270N mutant was significantly lower than that of the wild type. The W270N mutation appears to interfere with the oligomeric state, as revealed by a lower FRET efficiency. Collectively, these data suggest that oligomerization, driven by TMS-TMS interactions, is crucial for the stabilization and function of HO-1 in the ER.

BGN/TLR4/NF-B Mediates Epigenetic Silencing of Immunosuppressive Siglec Ligands in Colon Cancer Cells.

Human Toll-like receptor (TLR) signaling plays a vital role in intestinal inflammation by activating the NF-B pathway. By querying GENT2 datasets, we identified the gene expression level of TLR2 and TLR4 as being substantially increased in colorectal cancer. Introduction of shRNAs for TLR4 but not TLR2 dramatically recovered disialyl Lewisa and sialyl 6-sulfo Lewisx glycans, which are preferentially expressed in non-malignant colonic epithelial cells and could serve as ligands for the immunosuppressive molecule Siglec-7. We screened several TLR4 ligands and found that among them BGN is highly expressed in cancers and is involved in the epigenetic silencing of Siglec-7 ligands. Suppression of BGN expression substantially downregulated NF-B activity and the marker H3K27me3 in the promoter regions of the SLC26A2 and ST6GalNAc6 genes, which are involved in the synthesis of those glycans, and restored expression of normal glycans as well as Siglec-7 binding activities. We show that in the presence of TLR4, inflammatory stimuli initiate a positive loop involving NF-B that activates BGN and further enhances TLR4 activity. Present findings indicate a putative mechanism for the promotion of carcinogenesis by loss of immunosuppressive ligands by the BGN/TLR4/ NF-B pathway.

Lumenal Galectin-9-Lamp2 interaction regulates lysosome and autophagy to prevent pathogenesis in the intestine and pancreas.

Intracellular galectins are carbohydrate-binding proteins capable of sensing and repairing damaged lysosomes. As in the physiological conditions glycosylated moieties are mostly in the lysosomal lumen but not cytosol, it is unclear whether galectins reside in lysosomes, bind to glycosylated proteins, and regulate lysosome functions. Here, we show in gut epithelial cells, galectin-9 is enriched in lysosomes and predominantly binds to lysosome-associated membrane protein 2 (Lamp2) in a Asn(N)-glycan dependent manner. At the steady state, galectin-9 binding to glycosylated Asn175 of Lamp2 is essential for functionality of lysosomes and autophagy. Loss of N-glycan-binding capability of galectin-9 causes its complete depletion from lysosomes and defective autophagy, leading to increased endoplasmic reticulum (ER) stress preferentially in autophagy-active Paneth cells and acinar cells. Unresolved ER stress consequently causes cell degeneration or apoptosis that associates with colitis and pancreatic disorders in mice. Therefore, lysosomal galectins maintain homeostatic function of lysosomes to prevent organ pathogenesis.

SLC38A2 Overexpression Induces a Cancer-like Metabolic Profile and Cooperates with SLC1A5 in Pan-cancer Prognosis.

Cancer cells have dramatically increased demands for energy as well as biosynthetic precursors to fuel their restless growth. Enhanced glutaminolysis is a hallmark of cancer metabolism which fulfills these needs. Two glutamine transporters, SLC1A5 and SLC38A2, have been previously reported to promote glutaminolysis in cancer with controversial perspectives. In this study, we harnessed the proximity labeling reaction to map the protein interactome using mass spectrometry-based proteomics and discovered a potential protein-protein interaction between SLC1A5 and SLC38A2. The SLC1A5/SLC38A2 interaction was further confirmed by bimolecular fluorescence complementation assay. We further investigated the metabolic influence of SLC1A5 and SLC38A2 overexpression in human cells, respectively, and found that only SLC38A2, but not SLC1A5, resulted in a cancer-like metabolic profile, where the intracellular concentrations of essential amino acids and lactate were significantly increased as quantified by nuclear magnetic resonance spectroscopy. Finally, we analyzed the 5-year survival rates in a large pan-cancer cohort and found that the SLC1A5hi /SLC38A2lo group did not relate to a poor survival rate, whereas the SLC1A5lo /SLC38A2hi group significantly aggravated the lethality. Intriguingly, the SLC1A5hi /SLC38A2hi group resulted in an even worse prognosis, suggesting a cooperative effect between SLC1A5 and SCL38A2. Our data suggest that SLC38A2 plays a dominant role in reprogramming the cancer-like metabolism and promoting the cancer progression, whereas SLC1A5 may augment this effect when co-overexpressed with SLC38A2. We propose a model to explain the relationship between SLC1A5, SLC38A2 and SCL7A5, and discuss their impact on glutaminolysis and mTOR signaling.

Novel Naturally Occurring Mutations of Enterovirus 71 Associated With Disease Severity.

Infection with the re-emerging enterovirus 71 (EV-A71) is associated with a wide range of disease severity, including herpangina, encephalitis, and cardiopulmonary failure. At present, there is no FDA-approved therapeutics for EV-A71. Early diagnosis for the high-risk children is the key to successful patient care. We examined viral genome sequences at the 5' untranslated region (UTR) and the capsid protein VP1 from 36 mild and 27 severe cases. We identified five EV-A71 mutations associated with severe diseases, including (1) the 5' UTR mutations C580U, A707G, C709U; (2) a VP1 alanine-to-threonine mutation at position 280 (280T), and (3) a VP1 glutamic acid-to-(non-glutamic acid) at position 145 [145(non-E)]. SCARB2 is a known entry receptor for EV-A71. Based on a recent cryoEM structure of the EV-A71-SCARB2 binding complex, VP1-280T is near the binding interface between the VP1-VP2 complex and its entry receptor SCARB2. A de novo created hydrogen bonding between the mutant VP1-280T and the VP2-139T, could help strengthen a web-like interaction structure of the VP1-VP2 complex. A stabilized loop turn of VP2, once in contact with SCARB2, can enhance interaction with the host SCARB2 receptor for viral entry. Our findings here could facilitate early detection of severe cases infected with EV-A71 in clinical medicine. In addition, it opens up the opportunity of functional studies via infectious cDNA cloning, site-directed mutagenesis, and animal models in the future.

Plant Cytosolic Ascorbate Peroxidase with Dual Catalytic Activity Modulates Abiotic Stress Tolerances.

Ascorbic acid-glutathione (AsA-GSH) cycle represents important antioxidant defense system in planta. Here we utilized Oncidium cytosolic ascorbate peroxidase (OgCytAPX) as a model to demonstrate that CytAPX of several plants possess dual catalytic activity of both AsA and GSH, compared with the monocatalytic activity of Arabidopsis APX (AtCytAPX). Structural modeling and site-directed mutagenesis identified that three amino acid residues, Pro63, Asp75, and Tyr97, are required for oxidization of GSH in dual substrate catalytic type. Enzyme kinetic study suggested that AsA and GSH active sites are distinctly located in cytosolic APX structure. Isothermal titration calorimetric and UV-visible analysis confirmed that cytosolic APX is a heme-containing protein, which catalyzes glutathione in addition to ascorbate. Biochemical and physiological evidences of transgenic Arabidopsis overexpressing OgCytAPX1 exhibits efficient reactive oxygen species-scavenging activity, salt and heat tolerances, and early flowering, compared with Arabidopsis overexpressing AtCytAPX. Thus results on dual activity CytAPX impose significant advantage on evolutionary adaptive mechanism in planta.

Epigenetic silencing of the synthesis of immunosuppressive Siglec ligand glycans by NF-κB/EZH2/YY1 axis in early-stage colon cancers.

Normal colonic epithelial cells express sialyl 6-sulfo Lewisx and disialyl Lewisa on their cell surface, which are ligands for the immunosuppressive molecule Siglec-7. Expression of these normal glycans is frequently lost upon malignant transformation by silencing DTDST and ST6GalNAc6 at the early stage of colorectal carcinogenesis, and leads to production of inflammatory mediators that facilitate carcinogenesis. Indeed, by querying The Cancer Genome Atlas datasets, we confirmed that the level of DTDST or ST6GalNAc6 mRNA is substantially decreased at the early stage of colorectal carcinogenesis. Cultured colon cancer cell lines were used in this study including DLD-1, HT-29, LS174T and SW620. Their promoter regions were strongly marked by repressive mark H3K27me3, catalyzed by EZH2 that was markedly upregulated in early stage of colorectal carcinogenesis. Suppression of EZH2 substantially downregulated H3K27me3 mark and upregulated DTDST and ST6GalNAc6 as well as expression of normal glycans and Siglec-binding activities. Transcription factor YY1 was vital for the recruitment of PRC2-containing EZH2 to both promoters. Inhibition of NF-κB substantially reduced EZH2 transcription and restored their mRNAs as well as the production of normal Siglec ligand glycans in the results obtained from in vitro studies on cultured colon cancer cell lines. These findings provide a putative mechanism for promotion of carcinogenesis by loss of immunosuppressive molecules by epigenetic silencing through NF-κB-mediated EZH2/YY1 axis.

Mass detection in digital breast tomosynthesis data using convolutional neural networks and multiple instance learning.

Digital breast tomosynthesis (DBT) was developed in the field of breast cancer screening as a new tomographic technique to minimize the limitations of conventional digital mammography breast screening methods. A computer-aided detection (CAD) framework for mass detection in DBT has been developed and is described in this paper. The proposed framework operates on a set of two-dimensional (2D) slices. With plane-to-plane analysis on corresponding 2D slices from each DBT, it automatically learns complex patterns of 2D slices through a deep convolutional neural network (DCNN). It then applies multiple instance learning (MIL) with a randomized trees approach to classify DBT images based on extracted information from 2D slices. This CAD framework was developed and evaluated using 5040 2D image slices derived from 87 DBT volumes. The empirical results demonstrate that this proposed CAD framework achieves much better performance than CAD systems that use hand-crafted features and deep cardinality-restricted Bolzmann machines to detect masses in DBTs.

Compressed Submanifold Multifactor Analysis.

Although widely used, Multilinear PCA (MPCA), one of the leading multilinear analysis methods, still suffers from four major drawbacks. First, it is very sensitive to outliers and noise. Second, it is unable to cope with missing values. Third, it is computationally expensive since MPCA deals with large multi-dimensional datasets. Finally, it is unable to maintain the local geometrical structures due to the averaging process. This paper proposes a novel approach named Compressed Submanifold Multifactor Analysis (CSMA) to solve the four problems mentioned above. Our approach can deal with the problem of missing values and outliers via SVD-L1. The Random Projection method is used to obtain the fast low-rank approximation of a given multifactor dataset. In addition, it is able to preserve the geometry of the original data. Our CSMA method can be used efficiently for multiple purposes, e.g. noise and outlier removal, estimation of missing values, biometric applications. We show that CSMA method can achieve good results and is very efficient in the inpainting problem as compared to [1], [2]. Our method also achieves higher face recognition rates compared to LRTC, SPMA, MPCA and some other methods, i.e. PCA, LDA and LPP, on three challenging face databases, i.e. CMU-MPIE, CMU-PIE and Extended YALE-B.