Deep neural network based histological scoring of lung fibrosis and inflammation in the mouse model system.

Preclinical studies of novel compounds rely on quantitative readouts from animal models. Frequently employed readouts from histopathological tissue scoring are time consuming, require highly specialized staff and are subject to inherent variability. Recent advances in deep convolutional neural networks (CNN) now allow automating such scoring tasks. Here, we demonstrate this for the case of the Ashcroft fibrosis score and a newly developed inflammation score to characterize fibrotic and inflammatory lung diseases. Sections of lung tissue from mice exhibiting a wide range of fibrotic and inflammatory states were stained with Masson trichrome. Whole slide scans using a 20x objective were acquired and cut into smaller tiles of 512x512 pixels. The tiles were subsequently classified by specialized CNNs, either an "Ashcroft fibrosis CNN" or an "inflammation CNN". For the Ashcroft fibrosis score the CNN was fine-tuned by using 14000 labelled tiles. For the inflammation score the CNN was trained with 3500 labelled tiles. After training, the Ashcroft fibrosis CNN achieved an accuracy of 79.5% and the inflammation CNN an accuracy of 80.0%. An error analysis revealed that misclassifications are almost exclusively with neighboring scores, which reflects the inherent ambiguity of parts of the data. The variability between two experts was found to be larger than the variability between the CNN classifications and the ground truth. The CNN generated Ashcroft score was in very good agreement with the score of a pathologist (r2 = 0.92). Our results demonstrate that costly and time consuming scoring tasks can be automated and standardized with deep learning. New scores such as the inflammation score can be easily developed with the approach presented here.

Evaluation of Quantitative Imaging Biomarkers in the DSS Colitis Model.

PURPOSE: In humans, colonoscopy is the gold standard for the diagnosis of inflammatory changes of the colon wall. Aim of this study was the identification of less invasive imaging biomarkers in the dextran sodium sulfate (DSS) colitis model to provide additional information on transmural changes of the colon wall. PROCEDURES: Colitis was induced in C57BL/6 mice by administration of 2, 3, and 4 % DSS over a period of 5 days. Colon wall thickness was measured using magnetic resonance imaging (MRI), ultrasound (US), and x-ray computed tomography (CT), gut inflammation by positron emission tomography/CT, and mucosal changes of the colon wall by colonoscopy. Colon samples were examined histologically. RESULTS: MRI, CT, US, and histological data revealed increased colon wall thickness in DSS-treated mice compared to healthy controls. Elevated 2-deoxy-2[(18)F]fluoro-D-glucose uptake and colonoscopy confirmed high inflammatory load in the guts of colitis mice. CONCLUSIONS: The established quantitative imaging readouts offer promising perspectives to develop new compounds and to translate these methods into the clinical setting.

Pivotal role of serum- and glucocorticoid-inducible kinase 1 in vascular inflammation and atherogenesis.

OBJECTIVE: Atherosclerosis, an inflammatory disease of arterial vessel walls, requires migration and matrix metalloproteinase (MMP)-9-dependent invasion of monocytes/macrophages into the vascular wall. MMP-9 expression is stimulated by transcription factor nuclear factor-κB, which is regulated by inhibitor κB (IκB) and thus IκB kinase. Regulators of nuclear factor-κB include serum- and glucocorticoid-inducible kinase 1 (SGK1). The present study explored involvement of SGK1 in vascular inflammation and atherogenesis. APPROACH AND RESULTS: Gene-targeted apolipoprotein E (ApoE)-deficient mice without (apoe(-/-)sgk1(+/+)) or with (apoe(-/-)sgk1(-/-)) additional SGK1 knockout received 16-week cholesterol-rich diet. According to immunohistochemistry atherosclerotic lesions in aorta and carotid artery, vascular CD45(+) leukocyte infiltration, Mac-3(+) macrophage infiltration, vascular smooth muscle cell content, MMP-2, and MMP-9 positive areas in atherosclerotic tissue were significantly less in apoe(-/-)sgk1(-/-)mice than in apoe(-/-)sgk1(+/+)mice. As determined by Boyden chamber, thioglycollate-induced peritonitis and air pouch model, migration of SGK1-deficient CD11b(+)F4/80(+) macrophages was significantly diminished in vitro and in vivo. Zymographic MMP-2 and MMP-9 production, MMP-9 activity and invasion through matrigel in vitro were significantly less in sgk1(-/-) than in sgk1(+/+)macrophages and in control plasmid-transfected or inactive (K127N)SGK1-transfected than in constitutively active (S422D)SGK1-transfected THP-1 cells. Confocal microscopy revealed reduced macrophage number and macrophage MMP-9 content in plaques of apoe(-/-)sgk1(-/-) mice. In THP-1 cells, MMP-inhibitor GM6001 (25 µmol/L) abrogated (S422D)SGK1-induced MMP-9 production and invasion. According to reverse transcription polymerase chain reaction, MMP-9 transcript levels were significantly reduced in sgk1(-/-)macrophages and strongly upregulated in (S422D)SGK1-transfected THP-1 cells compared with control plasmid-transfected or (K127N)SGK1-transfected THP-1 cells. According to immunoblotting and confocal microscopy, phosphorylation of IκB kinase and inhibitor κB and nuclear translocation of p50 were significantly lower in sgk1(-/-)macrophages than in sgk1(+/+)macrophages and significantly higher in (S422D)SGK1-transfected THP-1 cells than in control plasmid-transfected or (K127N)SGK1-transfected THP-1 cells. Treatment of (S422D)SGK1-transfected THP-1 cells with IκB kinase-inhibitor BMS-345541 (10 µmol/L) abolished (S422D)SGK1-induced increase of MMP-9 transcription and gelatinase activity. CONCLUSIONS: SGK1 plays a pivotal role in vascular inflammation during atherogenesis. SGK1 participates in the regulation of monocyte/macrophage migration and MMP-9 transcription via regulation of nuclear factor-κB.

PI3 kinase-dependent stimulation of platelet migration by stromal cell-derived factor 1 (SDF-1).

Platelets have been regarded as static cells that do not move once they adhere to a matrix. The present study explored, whether platelets are able to migrate. In contrast to the current opinion, we found that platelets were mobile, able to migrate over a surface, and transmigrate through a transwell membrane and endothelium toward a source of stromal cell-derived factor 1 (SDF-1). Platelet migration was stimulated by SDF-1, which led to the downstream activation and phosphorylation of Wiskott-Aldrich syndrome protein. SDF-1 signaling and subsequent platelet migration could be inhibited by CXCR4-receptor blocker AMD3100, pertussis toxin, inhibition of phosphoinositol 3-kinase (PI3 kinase) with LY294002 or wortmannin, and disruption of actin polymerization with cytochalasin B. The potential of platelets to migrate in an SDF-1-mediated fashion may redefine the role of platelets in the pathophysiology of vascular inflammation, subsequent atherosclerotic degeneration, and vascular regeneration.

Capture of endothelial progenitor cells by a bispecific protein/monoclonal antibody molecule induces reendothelialization of vascular lesions.

Tissue injury is inevitably accompanied by disruption of the endothelium and exposure of the subendothelial matrix. To generate a guidance molecule directing progenitor cells to sites of vascular lesions, we designed a bifunctional protein. The protein consists of the soluble platelet collagen receptor glycoprotein VI and an antibody to CD133 (hereafter called GPVI-CD133). In vitro and in vivo, this construct substantially mediates endothelial progenitor cell (EPC) homing to vascular lesions. Exposure of EPCs to GPVI-CD133 did not impair their capability to differentiate toward mature endothelial cells as verified by the formation of colony-forming units, the upregulation of endothelial markers CD31 and CD146 analyzed by flow cytometry or von Willebrand factor and endoglin assessed by immunofluorescence microscopy, as well as the presence of Weibel-Palade bodies using transmission electron microscopy. In vivo, GPVI-CD133 augments reendothelialization of vascular lesions. Thus, this bifunctional protein could be a potential new therapeutic option for cardiovascular diseases.

Platelet-derived stromal cell-derived factor-1 regulates adhesion and promotes differentiation of human CD34+ cells to endothelial progenitor cells.

BACKGROUND: Peripheral homing of progenitor cells in areas of diseased organs is critical for tissue regeneration. The chemokine stromal cell-derived factor-1 (SDF-1) regulates homing of CD34+ stem cells. We evaluated the role of platelet-derived SDF-1 in adhesion and differentiation of human CD34+ cells into endothelial progenitor cells. METHODS AND RESULTS: Adherent platelets express substantial amounts of SDF-1 and recruit CD34+ cells in vitro and in vivo. A monoclonal antibody to SDF-1 or to its counterreceptor, CXCR4, inhibits stem cell adhesion on adherent platelets under high arterial shear in vitro and after carotid ligation in mice, as determined by intravital fluorescence microscopy. Platelets that adhere to human arterial endothelial cells enhance the adhesion of CD34+ cells on endothelium under flow conditions, a process that is inhibited by anti-SDF-1. During intestinal ischemia/reperfusion in mice, anti-SDF-1 and anti-CXCR4, but not isotype control antibodies, abolish the recruitment of CD34+ cells in microcirculation. Moreover, platelet-derived SDF-1 binding to CXCR4 receptor promotes platelet-induced differentiation of CD34+ cells into endothelial progenitor cells, as verified by colony-forming assays in vitro. CONCLUSIONS: These findings imply that platelet-derived SDF-1 regulates adhesion of stem cells in vitro and in vivo and promotes differentiation of CD34+ cells to endothelial progenitor cells. Because tissue regeneration depends on recruitment of progenitor cells to peripheral vasculature and their subsequent differentiation, platelet-derived SDF-1 may contribute to vascular and myocardial regeneration.