Application of convolutional neural network for analyzing hepatic fibrosis in mice.

Recently, with the development of computer vision using artificial intelligence (AI), clinical research on diagnosis and prediction using medical image data has increased. In this study, we applied AI methods to analyze hepatic fibrosis in mice to determine whether an AI algorithm can be used to analyze lesions. Whole slide image (WSI) Sirius Red staining was used to examine hepatic fibrosis. The Xception network, an AI algorithm, was used to train normal and fibrotic lesion identification. We compared the results from two analyses, that is, pathologists' grades and researchers' annotations, to observe whether the automated algorithm can support toxicological pathologists efficiently as a new apparatus. The accuracies of the trained model computed from the training and validation datasets were greater than 99%, and that obtained by testing the model was 100%. In the comparison between analyses, all analyses showed significant differences in the results for each group. Furthermore, both normalized fibrosis grades inferred from the trained model annotated the fibrosis area, and the grades assigned by the pathologists showed significant correlations. Notably, the deep learning algorithm derived the highest correlation with the pathologists' average grade. Owing to the correlation outcomes, we conclude that the trained model might produce results comparable to those of the pathologists' grading of the Sirius Red-stained WSI fibrosis. This study illustrates that the deep learning algorithm can potentially be used for analyzing fibrotic lesions in combination with Sirius Red-stained WSIs as a second opinion tool in non-clinical research.

A comparative study on the implementation of deep learning algorithms for detection of hepatic necrosis in toxicity studies.

Deep learning has recently become one of the most popular methods of image analysis. In non-clinical studies, several tissue slides are generated to investigate the toxicity of a test compound. These are converted into digital image data using a slide scanner, which is then studied by researchers to investigate abnormalities, and the deep learning method has been started to adopt in this study. However, comparative studies evaluating different deep learning algorithms for analyzing abnormal lesions are scarce. In this study, we applied three algorithms, SSD, Mask R-CNN, and DeepLabV3+, to detect hepatic necrosis in slide images and determine the best deep learning algorithm for analyzing abnormal lesions. We trained each algorithm on 5750 images and 5835 annotations of hepatic necrosis including validation and test, augmented with 500 image tiles of 448 × 448 pixels. Precision, recall, and accuracy were calculated for each algorithm based on the prediction results of 60 test images of 2688 × 2688 pixels. The two segmentation algorithms, DeepLabV3+ and Mask R-CNN, showed over 90% of accuracy (0.94 and 0.92, respectively), whereas SSD, an object detection algorithm, showed lower accuracy. The trained DeepLabV3+ outperformed all others in recall while also successfully separating hepatic necrosis from other features in the test images. It is important to localize and separate the abnormal lesion of interest from other features to investigate it on a slide level. Therefore, we suggest that segmentation algorithms are more appropriate than object detection algorithms for use in the pathological analysis of images in non-clinical studies. Supplementary Information: The online version contains supplementary material available at 10.1007/s43188-023-00173-5.

Segmentation algorithm can be used for detecting hepatic fibrosis in SD rat.

BACKGROUND: Liver fibrosis is an early stage of liver cirrhosis. As a reversible lesion before cirrhosis, liver failure, and liver cancer, it has been a target for drug discovery. Many antifibrotic candidates have shown promising results in experimental animal models; however, due to adverse clinical reactions, most antifibrotic agents are still preclinical. Therefore, rodent models have been used to examine the histopathological differences between the control and treatment groups to evaluate the efficacy of anti-fibrotic agents in non-clinical research. In addition, with improvements in digital image analysis incorporating artificial intelligence (AI), a few researchers have developed an automated quantification of fibrosis. However, the performance of multiple deep learning algorithms for the optimal quantification of hepatic fibrosis has not been evaluated. Here, we investigated three different localization algorithms, mask R-CNN, DeepLabV3+, and SSD, to detect hepatic fibrosis. RESULTS: 5750 images with 7503 annotations were trained using the three algorithms, and the model performance was evaluated in large-scale images and compared to the training images. The results showed that the precision values were comparable among the algorithms. However, there was a gap in the recall, leading to a difference in model accuracy. The mask R-CNN outperformed the recall value (0.93) and showed the closest prediction results to the annotation for detecting hepatic fibrosis among the algorithms. DeepLabV3+ also showed good performance; however, it had limitations in the misprediction of hepatic fibrosis as inflammatory cells and connective tissue. The trained SSD showed the lowest performance and was limited in predicting hepatic fibrosis compared to the other algorithms because of its low recall value (0.75). CONCLUSIONS: We suggest it would be a more useful tool to apply segmentation algorithms in implementing AI algorithms to predict hepatic fibrosis in non-clinical studies.

Preparing pathological data to develop an artificial intelligence model in the nonclinical study.

Artificial intelligence (AI)-based analysis has recently been adopted in the examination of histological slides via the digitization of glass slides using a digital scanner. In this study, we examined the effect of varying the staining color tone and magnification level of a dataset on the result of AI model prediction in hematoxylin and eosin stained whole slide images (WSIs). The WSIs of liver tissues with fibrosis were used as an example, and three different datasets (N20, B20, and B10) were prepared with different color tones and magnifications. Using these datasets, we built five models trained Mask R-CNN algorithm by a single or mixed dataset of N20, B20, and B10. We evaluated their model performance using the test dataset of three datasets. It was found that the models that were trained with mixed datasets (models B20/N20 and B10/B20), which consist of different color tones or magnifications, performed better than the single dataset trained models. Consequently, superior performance of the mixed models was obtained from the actual prediction results of the test images. We suggest that training the algorithm with various staining color tones and multi-scaled image datasets would be more optimized for consistent remarkable performance in predicting pathological lesions of interest.

Preparation of bottom-up graphene oxide using citric acid and tannic acid, and its application as a filler for polypropylene nanocomposites.

The production of graphene oxide (GO) in large amounts for commercialization in the chemical industry has been limited because harsh and tedious process conditions are required. In this study, a novel carbon nanomaterial called 'bottom-up graphene oxide (BGO)' could be easily prepared for the first time by heat treatment of the mixture of citric acid (CA) and tannic acid (TA) with different weight ratios for the first time. BGO3 prepared using a 50/50 weight ratio of CA/TA was found to have an average lateral size of 250.0 nm and an average thickness of 7.2 nm, and it was further functionalized with cardanol to prepare cardanol functionalized BGO3 (CBGO3) to be used as a filler for the polypropylene (PP) nanocomposite, where cardanol was used to increase the compatibility between BGO3 and PP. The improved mechanical properties and thermal stability of PP nanocomposites containing CBGO3 could be ascribed to the intrinsic mechanical properties of the carbon nanomaterial and the increased compatibility by the attached cardanol on BGO3.

Runx2 regulates survivin expression in prostate cancer cells.

Previously we described that bone morphogenetic protein-7 (BMP7) could protect prostate cancer C4-2B cells from serum starvation-induced apoptosis via survivin induction. Here, for the first time, we identify Runx2 as a key regulator of survivin transcription. In C4-2B cells grown normally, suppression of Runx2 reduced survivin expression. Using ChIP assays, two regions of the survivin promoter, -1953 to -1812 (I) and -1485 to -1119 (II) encompassing consensus Runx-binding sites were examined. Runx2 was found to be associated with both regions, with a stronger affinity to region-I. In serum-starved cells neither region was occupied, but BMP7 restored association to region-II and not region-I. In reporter assays, transcription activity by BMP7 was significantly reduced when sequences including binding sites of region-II were deleted. Additionally, Runx2 expression was enhanced by BMP7 in these cells. Along with a strong survivin expression, a trend in increased Runx2 expression in human prostate cancer cells and tissues was noted. In the conditional Pten-knockout mouse, Runx2 level increased with growth of prostate tumor. The data define a novel role of Runx2 in regulating survivin expression in malignant epithelial cells and identify it as a critical factor in BMP signaling that protects cancer cells against apoptosis.

Hygroscopy-induced nanoparticle reshuffling in ionic-gold-residue-stabilized gold suprananoparticles.

Polyethyleneimine (PEI)-stabilized gold nanoparticles were used as a model to understand the roles of ionic precursors in the formation of nanoparticles and the impact of their presence on the nanoparticle properties. The low availability of elemental gold and the stabilization of the just-generated gold nanoparticles by the excess gold ions contributed to the production of ultra-small nearly neutral gold nanoparticles, resulting in properties significantly different from those prepared by conventional methods. The cross-linking between gold ions/PEI/nanoparticles further led to the assembly of these small gold nanoparticles into suprananoparticles that were stable in water. The hygroscopic Au(iii) residues in the suprananoparticles absorbed moisture to form a micro-water pool and the nanoparticles in the new aqueous solution reshuffled to generate larger nanoparticles, leading to significant changes in their optical properties. Such a phenomenon was formulated into a fast, sensitive and straightforward method for the detection of water content in organic solvents.

Mechanistic Investigation of the Androgen Receptor DNA-Binding Domain Inhibitor Pyrvinium.

Pyrvinium was identified as the first small molecule inhibitor of the androgen receptor (AR) DNA-binding domain (DBD). It was also among the first small molecules shown to directly inhibit the activity of AR splice variants (ARVs), which has important clinical implications in the treatment of castration-resistant prostate cancer. Important questions about pyrvinium's mechanism of action remain. Here, we demonstrate through mutational analysis that amino acids 609 and 612 are important for pyrvinium action. Nuclear magnetic resonance demonstrates a specific interaction between a soluble pyrvinium derivative and the AR DBD homodimer-DNA complex. Chromatin immunoprecipitation and electrophoretic mobility shift assay experiments demonstrate that, despite an interaction with this complex, pyrvinium does not alter the DNA-binding kinetics in either assay. AR immunoprecipitation followed by mass spectrometry was used to identify proteins whose interaction with AR is altered by pyrvinium. Several splicing factors, including DDX17, had reduced interactions with AR in the presence of pyrvinium. RNA sequencing of prostate cancer cells treated with pyrvinium demonstrated changes in splicing, as well as in several other pathways. However, pyrvinium did not alter the levels of ARVs in several prostate cancer cell lines. Taken together, our new data pinpoint the direct interaction between pyrvinium and AR DBD and shed light on the mechanism by which it inhibits AR transcriptional activity.

Bone morphogenetic protein 7 protects prostate cancer cells from stress-induced apoptosis via both Smad and c-Jun NH2-terminal kinase pathways.

We reported earlier that exposure to exogenous bone morphogenetic protein 7 (BMP7) could strongly inhibit serum starvation-induced apoptosis to C4-2B cell line, a variant of the LNCaP human prostate cancer cell line with propensity for bone metastasis. Whereas serum starvation suppressed the expression of survivin, a member of the inhibitor of apoptosis protein family, its expression was sustained in the presence of BMP7. In this study, we present evidence that BMP7 exposure up-regulated survivin promoter activity, an effect that was associated with activation of Smad, and could be repressed by dominant-negative Smad5. Additionally, serum starvation-induced suppression of c-jun NH2-terminal kinase (JNK) activity in C4-2B cells could be mostly restored by BMP7, and a JNK inhibitor could counteract the antiapoptotic effect of BMP7, without a significant effect on the level of survivin expression. Thus, we identified JNK pathway as another signaling mode for the antiapoptotic function of BMP7. To test the effect of endogenous up-regulation of BMP7, we genetically modulated the C4-2B cell line to overexpress BMP7 protein. Not only was this altered cell line resistant to serum starvation-induced apoptosis but it also exhibited patterns of Smad activation, survivin up-regulation, and JNK activation similar to those of the parental C4-2B cells exposed to exogenous BMP7. Consistent with these in vitro findings of BMP7 action, we acquired correlative results of Smad activation, survivin expression, and JNK activation in the progression of prostate cancer in the conditional Pten deletion mouse model, in which we first obtained the evidence of BMP7 overexpression.

Sulfonated Poly(Arylene Ether Sulfone) and Perfluorosulfonic Acid Composite Membranes Containing Perfluoropolyether Grafted Graphene Oxide for Polymer Electrolyte Membrane Fuel Cell Applications.

Sulfonated poly(arylene ether sulfone) (SPAES) and perfluorosulfonic acid (PFSA) composite membranes were prepared using perfluoropolyether grafted graphene oxide (PFPE-GO) as a reinforcing filler for polymer electrolyte membrane fuel cell (PEMFC) applications. PFPE-GO was obtained by grafting poly(hexafluoropropylene oxide) having a carboxylic acid end group onto the surface of GO via ring opening reaction between the carboxylic acid group in poly(hexafluoropropylene oxide) and the epoxide groups in GO, using 4-dimethylaminopyridine as a base catalyst. Both SPAES and PFSA composite membranes containing PFPE-GO showed much improved mechanical strength and dimensional stability, compared to each linear SPAES and PFSA membrane, respectively. The enhanced mechanical strength and dimensional stability of composite membranes can be ascribed to the homogeneous dispersion of rigid conjugated carbon units in GO through the increased interfacial interactions between PFPE-GO and SPAES/PFSA matrices.

Dual Roles of Graphene Oxide To Attenuate Inflammation and Elicit Timely Polarization of Macrophage Phenotypes for Cardiac Repair.

Development of localized inflammatory environments by M1 macrophages in the cardiac infarction region exacerbates heart failure after myocardial infarction (MI). Therefore, the regulation of inflammation by M1 macrophages and their timely polarization toward regenerative M2 macrophages suggest an immunotherapy. Particularly, controlling cellular generation of reactive oxygen species (ROS), which cause M1 differentiation, and developing M2 macrophage phenotypes in macrophages propose a therapeutic approach. Previously, stem or dendritic cells were used in MI for their anti-inflammatory and cardioprotective potentials and showed inflammation modulation and M2 macrophage progression for cardiac repair. However, cell-based therapeutics are limited due to invasive cell isolation, time-consuming cell expansion, labor-intensive and costly ex vivo cell manipulation, and low grafting efficiency. Here, we report that graphene oxide (GO) can serve as an antioxidant and attenuate inflammation and inflammatory polarization of macrophages via reduction in intracellular ROS. In addition, GO functions as a carrier for interleukin-4 plasmid DNA (IL-4 pDNA) that propagates M2 macrophages. We synthesized a macrophage-targeting/polarizing GO complex (MGC) and demonstrated that MGC decreased ROS in immune-stimulated macrophages. Furthermore, DNA-functionalized MGC (MGC/IL-4 pDNA) polarized M1 to M2 macrophages and enhanced the secretion of cardiac repair-favorable cytokines. Accordingly, injection of MGC/IL-4 pDNA into mouse MI models attenuated inflammation, elicited early polarization toward M2 macrophages, mitigated fibrosis, and improved heart function. Taken together, the present study highlights a biological application of GO in timely modulation of the immune environment in MI for cardiac repair. Current therapy using off-the-shelf material GO may overcome the shortcomings of cell therapies for MI.

Registered report: Interactions between cancer stem cells and their niche govern metastatic colonization.

The Reproducibility Project: Cancer Biology seeks to address growing concerns about reproducibility in scientific research by replicating selected results from a substantial number of high-profile papers in the field of cancer biology published between 2010 and 2012. This Registered report describes the proposed replication plan of key experiments from 'Interactions between cancer stem cells and their niche govern metastatic colonization' by Malanchi and colleagues, published in Nature in 2012 (Malanchi et al., 2012). The key experiments that will be replicated are those reported in Figures 2H, 3A, 3B, and S13. In these experiments, Malanchi and colleagues analyze messenger RNA levels of periostin (POSTN) in pulmonary fibroblasts, endothelial cells, and immune cells isolated from mice with micrometastases to determine which cell type is producing POSTN in the metastatic niche (Figure 2H; Malanchi et al., 2012). Additionally, they examine MMTV-PyMT control or POSTN null mice to test the effect of POSTN on primary tumor growth and metastasis (Figures 3A, 3B, and S13; Malanchi et al., 2012). The Reproducibility Project: Cancer Biology is a collaboration between the Center for Open Science and Science Exchange, and the results of the replications will be published in eLife.

Ligand-independent and tissue-selective androgen receptor inhibition by pyrvinium.

Pyrvinium pamoate (PP) is a potent noncompetitive inhibitor of the androgen receptor (AR). Using a novel method of target identification, we demonstrate that AR is a direct target of PP in prostate cancer cells. We demonstrate that PP inhibits AR activity via the highly conserved DNA binding domain (DBD), the only AR inhibitor that functions via this domain. Furthermore, computational modeling predicts that pyrvinium binds at the interface of the DBD dimer and the minor groove of the AR response element. Because PP acts through the DBD, PP is able to inhibit the constitutive activity of AR splice variants, which are thought to contribute to the growth of castration resistant prostate cancer (CRPC). PP also inhibits androgen-independent AR activation by HER2 kinase. The antiandrogen activity of pyrvinium manifests in the ability to inhibit the in vivo growth of CRPC xenografts that express AR splice variants. Interestingly, PP was most potent in cells with endogenous AR expression derived from prostate or bone. PP was able to inhibit several other hormone nuclear receptors (NRs) but not structurally unrelated transcription factors. PP inhibition of other NRs was similarly cell-type selective. Using dual-energy X-ray absorptiometry, we demonstrate that the cell-type specificity of PP manifests in tissue-selective inhibition of AR activity in mice, as PP decreases prostate weight and bone mineral density but does not affect lean body mass. Our results suggest that the noncompetitive AR inhibitor pyrvinium has significant potential to treat CRPC, including cancers driven by ligand-independent AR signaling.

PI3K, Erk signaling in BMP7-induced epithelial-mesenchymal transition (EMT) of PC-3 prostate cancer cells in 2- and 3-dimensional cultures.

We reported previously that bone morphogenetic protein 7 (BMP7) could induce epithelial-mesenchymal transition (EMT) in PC-3 prostate cancer cells grown in tissue culture plates. In this study, we examined BMP7-induced morphological and molecular expression changes that are characteristic of EMT using these cells under both two- (2D) and three-dimensional (3D) culture conditions. Filamentous outgrowths from spheroid structures that were formed from PC-3 cells in 3D cultures were strikingly evident when the spheroids were exposed to extracellular BMP7. This morphological change in 3D was accompanied by down-regulation of E-cadherin, which is an essential adhesion molecule for the integrity of epithelial phenotype. Invasiveness of the cancer cells was significantly enhanced with BMP7 treatment along with activation and up-regulation of proteases such as MMP1, MMP13, and urokinase plasminogen activator. Signal transduction of EMT conversion was examined by the use of certain pathway-specific inhibitors. Of the chemical inhibitors tested, inhibitors of PI3 kinase and Erk were found to suppress BMP-induced morphological changes both in 2D and 3D conditions. These results suggest that, besides the Smad signaling pathways, BMP-induced activation of PI3K and Erk contribute to EMT morphologic conversion of the PC-3 prostate cancer cells. Together, the results support the notion that the complexity of EMT may be better evaluated in terms of both spatial and temporal processes in 3D cell culture models that are physiologically more relevant than the cell growth in tissue culture plates.

Temperature stability of the interfacial structure between a sulfonated crystalline alkyl side-chain polymer and a soft adhesive.

The fracture toughness of interfaces between a sulfonated alkyl side-chain graft copolymer and a soft acrylic random copolymer containing acrylic acid monomers was investigated with a probe test method. Interfaces between a thin (100 nm) layer of the block copolymer and a thick (100 microm) layer of the acrylic copolymer were prepared at room temperature and subsequently annealed for 7 h at different temperatures. After the annealing step, the interface was quenched to room temperature and tested, a strategy that provides the advantage of keeping constant the mechanical properties of the materials on both sides of the interface so that any major difference in adhesive behavior can only be attributed to a change in the interfacial structure. For annealing temperatures below the crystalline to liquid crystalline transition temperature (86 degrees C), the adhesion energy remained very low and failure occurred by interfacial crack propagation. However when the interface was annealed above that temperature, a much higher adhesion energy was observed at room temperature because of the formation of a fibrillar structure upon debonding. The results indicate that the crystalline order at low temperature is very stable presumably because of the strong interactions between the sulfone groups in the side chains. However, when these interactions weaken and the side chains become liquid crystalline, the surface reconstruction mechanism cannot be prevented and strong interactions formed between the polar parts of the copolymer and the acrylic acid. These strong interactions remain during the cooling step, and a mechanism of surface reconstruction is proposed.