We perform a large scale benchmark of machine learning methods for the prediction of the thermodynamic stability of solids. We start by constructing a data set that comprises density functional theory calculations of around 250000 cubic perovskite systems. This includes all possible perovskite and antiperovskite crystals that can be generated with elements from hydrogen to bismuth, excluding rare gases and lanthanides. Incidentally, these calculations already reveal a large number of systems (around 500) that are thermodynamically stable but that are not present in crystal structure databases. Moreover, some of these phases have unconventional compositions and define completely new families of perovskites. This data set is then used to train and test a series of machine learning algorithms to predict the energy distance to the convex hull of stability. In particular, we study the performance of ridge regression, random forests, extremely randomized trees (including adaptive boosting), and neural networks. We find that extremely randomized trees give the smallest mean absolute error of the distance to the convex hull (121 meV/atom) in the test set of 230000 perovskites, after being trained in 20000 samples. Surprisingly, the machine already works if we give it as sole input features the group and row in the periodic table of the three elements composing the perovskite. Moreover, we find that the prediction accuracy is not uniform across the periodic table, being worse for first-row elements and elements forming magnetic compounds. Our results suggest that machine learning can be used to speed up considerably (by at least a factor of 5) high-throughput DFT calculations, by restricting the space of relevant chemical compositions without degradation of the accuracy.

Emerging evidence indicates that in addition to native pentameric C-reactive protein (CRP), monomeric CRP (mCRP) also plays an active role in inflammation associated with cardiovascular diseases. mCRP activates endothelial cells, one of the critical events in cardiovascular diseases; however, the underlying molecular mechanisms are incompletely understood. Here we report that association of mCRP with human aortic and coronary artery endothelial cells is predominantly due to membrane insertion rather than binding to the surface proteins Fc gammaRs and proteoglycans. We identify lipid rafts as the preferential membrane microdomains for mCRP anchorage. mCRP binding depends on membrane cholesterol content and is synergistically mediated by the putative cholesterol binding consensus sequence of CRP (aa 35-47) and the C-terminal octapeptide (aa 199-206). Conversely, disrupting lipid rafts with methyl-beta cyclodextrin or nystatin abrogated mCRP-induced cytokine release, reactive oxygen species generation, and adhesion molecule expression in endothelial cells. Furthermore, ex vivo treatment of rabbit thoracic aorta and carotid artery segments with nystatin prevented mCRP-induced IL-8 release. Our data identify mCRP-lipid raft interaction as an important mechanism in mediating cellular responses to mCRP and lend further support to the notion of mCRP regulation of endothelial cell function during inflammation.

No abstract

C-reactive protein (CRP) has been implicated in the regulation of inflammation underlying coronary artery disease; however, little is known about the molecular mechanisms responsible for the expression of its pro- or anti-inflammatory activities. Here, we have identified the intrasubunit disulfide bond as a conserved switch that controls the structure and functions of CRP. Conformational rearrangement in human pentameric CRP to monomeric CRP (mCRP) is the prerequisite for this switch to be activated by reducing agents, including thioredoxin. Immunohistochemical analysis revealed 36-79% colocalization of thioredoxin and mCRP in human advanced coronary atherosclerotic lesions. Nonreduced mCRP was largely inert in activating human coronary artery endothelial cells (HCAECs), whereas reduced or cysteine-mutated mCRP evoked marked release of IL-8 and monocyte chemoattractant protein-1 from HCAECs, with ~50% increase at a concentration of 1 μg/ml. Reduced mCRP was ~4 to 40-fold more potent than mCRP in up-regulating adhesion molecule expression, promoting U937 monocyte adhesion to HCAECs, and inducing cytokine release from rabbit arteries ex vivo and in mice. These actions were primarily due to unlocking the lipid raft interaction motif. Therefore, expression of proinflammatory properties of CRP on endothelial cells requires sequential conformational changes, i.e., loss of pentameric symmetry followed by reduction of the intrasubunit disulfide bond.

No abstract

Pt/Au alloy nanoparticles (NPs) in a wide composition range have been synthesized by room-temperature simultaneous sputter deposition from two independent magnetron sources onto liquid PEG (MW = 600). The prepared NPs were alloyed with the face-centered cubic (fcc) structure. In addition, the particle sizes, composition, and shape are strongly correlated but can be tailored by an appropriate variation of the sputtering parameters. No individual particle but large agglomerates with partial alloy structure formed at Pt content of less than 16 atom %. Highly dispersed NPs with no agglomeration were observed in PEG when the quantity of Pt is more than 26 atom %. On the other hand, a small amount of Pt could terminate the agglomeration of Au when sputtering on the grids for transmission electron microscope observation. Our experiment and computer simulation carried out by two different methods indicate that the composition-dependent particle size of Pt/Au can be explained by the atomic concentration, formation energy of the cluster, and interaction between different metal atoms and the PEG molecule.

Background: Diabetic retinopathy (DR) is currently the leading cause of blindness and visual disability in adults with diabetes mellitus (DM). Neovascularization has been identified as an important clinical property in DR, however, the exact mechanisms in DR neovascularization are still unclear and need further elucidation. Methods: Quantitative real-time PCR (qRT-PCR) was conducted to detect the expression level of long non-coding RNA (lncRNA)-metastasis associated lung adenocarcinoma transcript 1 (MALAT1), miR-125b and vascular endothelial-cadherin (VE-cadherin) in human retina microvascular endothelial cells (hRMECs) treated with high glucose (HG). Luciferase assay was used to detect interaction of MALAT1 with miR-125b and miR-125b with VE-cadherin. MTT assay, transwell assay, tube formation assay and vascular permeability assay were conducted to detect the cell viability, migration tube formation ability and permeability of hRMECs, respectively. ELISA was used to examine the release of VE-cadherin and vascular endothelial growth factor (VEGF). Western blotting was used to access the protein expression of VE-cadherin, VEGF, β-catenin, matrix metalloproteinase (MMP) 2 (MMP2) and MMP9. Results: MALAT1 and VE-cadherin were up-regulated while miR-125b was down-regulated in hRMECs treated with HG. MALAT1 could competitively bind to miR-125b against VE-cadherin at the site of 3′-untranslated region (3′-UTR), leading to the up-regulation of VE-cadherin. Knockdown of MALAT1 inhibited the proliferation, migration, tube formation and vascular permeability of hRMECs induced by HG through up-regulating miR-125b. Furthermore, we found the deletion of MALAT1 suppressed the VE-cadherin/β-catenin complex and neovascularization related proteins expression, which was up-regulated by HG. Conclusion: Knockdown of MALAT1 inhibited cell proliferation, migration and angiogenesis of hRMECs via suppressing the VE-cadherin/β-catenin complex through targeting miR-125b. Inhibition of MALAT1 may serve as a potential target for anti-angiogenic therapy for DR.

Cell-based therapies are increasingly recognized as a potential strategy to treat retinal neurodegenerative disease. Their administration, however, is normally indirect and complex, often with an inability to assess in real time their effects on cell death and their migration/integration into the host retina. In the present study, using a partial optic nerve transection (pONT) rat model, we describe a new method of Schwann cell (SC) delivery (direct application to injured optic nerve sheath, SC/DONS), which was compared with intravitreal SC delivery (SC/IVT). Both SC/DONS and SC/IVT were able to be assessed in vivo using imaging to visualize retinal ganglion cell (RGC) apoptosis and SC retinal integration. RGC death in the pONT model was best fitted to the one-phase exponential decay model. Although both SC/DONS and SC/IVT altered the temporal course of RGC degeneration in pONT, SC/DONS resulted in delayed but long-lasting effects on RGC protection, compared with SC/IVT treatment. In addition, their effects on primary and secondary degeneration, and axonal regeneration, were also investigated, by histology, whole retinal counting, and modelling of RGC loss. SC/DONS was found to significantly reduce RGC apoptosis in vivo and significantly increase RGC survival by targeting secondary rather than primary degeneration. Both SC/DONS and SC/IVT were found to promote RGC axonal regrowth after optic nerve injury, with evidence of GAP-43 expression in RGC somas and axons. SC/DONS may have the potential in the treatment of optic neuropathies, such as glaucoma. We show that SC transplantation can be monitored in real time and that the protective effects of SCs are associated with targeting secondary degeneration, with implications for translating cell-based therapies to the clinic.