Biology subjects

Schulz, V.

Publications and source records attributed to Schulz, V..

8 recordsLinked to original sources

bioRxiv2024

A novel isoform of Tensin1 promotes actin filament assembly for efficient erythroblast enucleation

Mammalian red blood cells are generated via a terminal erythroid differentiation pathway culminating in cell polarization and enucleation. Actin filament polymerization is critical for enucleation, but the molecular regulatory mechanisms remain poorly understood. We utilized publicly available RNA-seq and proteomics datasets to mine for actin-binding proteins and actin- nucleation factors differentially expressed during human erythroid differentiation and discovered that a focal adhesion protein--Tensin-1--dramatically increases in expression late in differentiation. Remarkably, we found that differentiating human CD34+ cells express a novel truncated form of Tensin-1 (eTNS1; Mr [~]125 kDa) missing the N-terminal half of the protein, due to an internal mRNA translation start site resulting in a unique exon 1. eTNS1 localized to the cytoplasm during terminal erythroid differentiation, with no apparent membrane association or focal adhesion formation. Knocking out eTNS1 had no effect on assembly of the spectrin membrane skeleton but led to impaired enucleation and absent or mis-localized actin filament foci in enucleating erythroblasts. We conclude that eTNS1 is a novel regulator of actin filament assembly during human erythroid terminal differentiation required for efficient enucleation.

cell biology↗

bioRxiv2022

Adversarial attacks and adversarial robustness in computational pathology

Artificial Intelligence (AI) can support diagnostic workflows in oncology by aiding diagnosis and providing biomarkers. AI applications are therefore expected to evolve from academic prototypes to commercial products in the coming years. However, AI applications are vulnerable to adversarial attacks, such as malicious interference with test data aiming to cause misclassifications. Therefore, it is essential for the use of AI-based diagnostic devices to secure them against such attacks before widespread use. Unfortunately, no resistant systems exist in computational pathology so far. To address this problem, we investigate the susceptibility of convolutional neural networks (CNNs) to multiple types of white- and black-box attacks. We demonstrate that both attacks can easily confuse CNNs in clinically relevant pathology tasks and impair classification performance. Classical adversarially robust training and dual batch normalization (DBN) are possible mitigation strategies but require precise knowledge of the type of attack used in the inference. We demonstrate that vision transformers (ViTs) perform equally well compared to CNNs at baseline and are orders of magnitude more robust to different types of white-box and black-box attacks. At a mechanistic level, we show that this is associated with a more robust latent representation of clinically relevant categories in ViTs compared to CNNs. Our results are in line with previous theoretical studies. We show that ViTs are robust learners in computational pathology. This implies that large-scale rollout of AI models in computational pathology should rely on ViTs rather than CNN-based classifiers to provide inherent protection against adversaries.

bioinformatics↗

bioRxiv2021

Benchmarking artificial intelligence methods for end-to-end computational pathology

Artificial intelligence (AI) can extract subtle visual information from digitized histopathology slides and yield scientific insight on genotype-phenotype interactions as well as clinically actionable recommendations. Classical weakly supervised pipelines use an end-to-end approach with residual neural networks (ResNets), modern convolutional neural networks such as EfficientNet, or non-convolutional architectures such as vision transformers (ViT). In addition, multiple-instance learning (MIL) and clustering-constrained attention MIL (CLAM) are being used for pathology image analysis. However, it is unclear how these different approaches perform relative to each other. Here, we implement and systematically compare all five methods in six clinically relevant end-to-end prediction tasks using data from N=4848 patients with rigorous external validation. We show that histological tumor subtyping of renal cell carcinoma is an easy task which approaches successfully solved with an area under the receiver operating curve (AUROC) of above 0.9 without any significant differences between approaches. In contrast, we report significant performance differences for mutation prediction in colorectal, gastric and bladder cancer. Weakly supervised ResNet-and ViT-based workflows significantly outperformed other methods, in particular MIL and CLAM for mutation prediction. As a reason for this higher performance we identify the ability of ResNet and ViT to assign high prediction scores to highly informative image regions with plausible histopathological image features. We make all source codes publicly available at https://github.com/KatherLab/HIA, allowing easy application of all methods on any end-to-end problem in computational pathology.

bioinformatics↗

bioRxiv2020

Study of flow effects on temperature-controlled radio-frequency ablation using phantom experiments and forward simulations

PurposeBlood perfusion is known to add variability to hepatic radiofrequency ablation (RFA) treatment outcomes. Simulation-assisted treatment planning taking into account blood perfusion may solve this problem in the future. Hence, this study aims to study perfusion effects on RFA in a controlled environment and to compare the outcome to a prediction made using finite volume simulations. MethodsAblation zones were induced in tissue-mimicking, thermochromic ablation phantoms with a single flow channel, using a RF generator with needle temperature controlled power delivery and a monopolar needle electrode. Channel radius and saline flow rate were varied and the impact of saline flow on the ablated cross-sectional area, on a potential occurrence of directional effects as well as on the delivered generator power input was studied. Finite-volume simulations reproducing the experimental geometry, flow conditions and generator power input were conducted in a second step and compared to the experimental ablation outcomes. ResultsVessels of different radii affected the ablation result in different manners. For the channel radius of 0.275 mm both the ablated area and energy input reduced with increasing flow rate. For radius 0.9 mm the ablated area reduced with increasing flow rate but the energy input increased. An increasing area and energy input were observed towards larger flow rates for the channel radius of 2.3 mm. Directional effects, i.e., shrinking of the lesion upstream of the needle and an extension thereof downstream, were observed only for the smallest channel radius. The simulations qualitatively confirmed these observations. When using the simulations to make a prediction of ablation outcomes with flow, the mean absolute error between experimental and predicted ablation outcomes was reduced from 23% to 12% as compared to neglecting flow effects. ConclusionSimulations can improve the prediction of RFA ablation regions in the presence of various blood flow effects. Our findings therefore underline the potential of simulation-assisted, patient-individual RFA treatment planning and guidance for the prediction of RFA outcomes in the presence of blood flow. Additional comments-Teresa Nolte and Nikhil Vaidya contributed equally to this work. -Volkmar Schulz and Karen Veroy contributed equally to this work. -A single reference experiment, i.e., not using a flow channel, and the image in the upper left corner of Figure 4 were included into a publication submitted to Int. J. Hyperthermia for model validation purposes. O_FIG O_LINKSMALLFIG WIDTH=178 HEIGHT=200 SRC="FIGDIR/small/172957v1_fig4.gif" ALT="Figure 4"> View larger version (122K): org.highwire.dtl.DTLVardef@935182org.highwire.dtl.DTLVardef@1cc4eedorg.highwire.dtl.DTLVardef@51ad9aorg.highwire.dtl.DTLVardef@5c9625_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOFig. 4C_FLOATNO Experimentally obtained ablation zones for all RFA experiments and the corresponding reference experiments. The experimental and simulated 70{degrees}C temperature contours are marked by a blue and a black line, respectively. The ablation needle was inserted from the left side. Saline flow was established from right to left through the flow channel located below the ablation area. Radius and flow rate for each experiment were as specified within the figure below the ablation zones. In four experiments, the experimental ablation zone shows an unexpected bump structure, which we marked by white arrows. The numbering of the experiments refers to Table 1. C_FIG

biophysics↗

bioRxiv2020

Mitochondrial nucleoid organization and biogenesis of complex I require mTERF18/SHOT1 and ATAD3 in Arabidopsis thaliana.

Mitochondria play critical roles in eukaryotes in ATP generation through oxidative phosphorylation (OXPHOS) and also produce both damaging and signaling reactive oxygen species (ROS). Originating from endosymbiosis, mitochondria have their own reduced genomes that encode essential subunits of the OXPHOS machinery. MTERF (Mitochondrial Transcription tERmination Factor-related) proteins have been shown to be involved in organelle gene expression by interacting with organellar DNA or RNA in multicellular eukaryotes. We previously identified mutations in Arabidopsis thaliana MTERF18/SHOT1 that enable plants to better tolerate heat and oxidative stresses, presumably due to low ROS and reduced oxidative damage. To understand molecular mechanisms leading to shot1 phenotypes, we investigated mitochondrial defects of shot1 mutants and targets of the SHOT1 protein. shot1 mutants have problems accumulating OXPHOS complexes that contain mitochondria-encoded subunits, with complex I and complex IV most affected. SHOT1 binds specific mitochondrial DNA sequences and localizes to mitochondrial nucleoids, which are diffuse in shot1 mutants. Furthermore, three homologues of mammalian ATAD3A proteins, which are suggested to be involved in mitochondrial nucleoid organization, were identified as SHOT1-interacting proteins (designated SHOT1 BINDING ATPASES (SBA)1, 2 and 3). Importantly, disrupting SBA function also disrupts nucleoids, compromises accumulation of complex I and enhances heat tolerance. We conclude that proper nucleoid organization is critical for correct expression and accumulation of complex I, and propose that nucleoid disruption results in unique changes in mitochondrial metabolism and signaling that lead to heat tolerance. SignificanceIn all eukaryotes, mitochondria are critical organelles that supply chemical energy for life, which is produced by the oxidative phosphorylation (OXPHOS) machinery on the inner mitochondrial membrane. The OXPHOS machinery comprises multiple protein complexes with subunits encoded by both nuclear and mitochondrial genes. Nuclear-encoded mTERF proteins are important for expression of mitochondrial genes, interacting with mitochondrial DNA or RNA. Our study reveals that the Arabidopsis mTERF18/SHOT1 protein interacts with mtDNA and homologs of human ATAD3A proteins, and that both proteins are critical for mitochondrial nucleoid organization and accumulation of OXPHOS Complex I. Further, the data indicate nucleoid disruption leads to unique mitochondrial and cellular responses such that mutant plants have enhanced heat tolerance.

plant biology↗

bioRxiv2019

Breaking Medical Data Sharing Boundaries by EmployingArtificial Radiographs

Artificial intelligence (AI) has the potential to change medicine fundamentally. Here, expert knowledge provided by AI can enhance diagnosis by comprehensive and user independent integration of multiple image features. Unfortunately, existing algorithms often stay behind expectations, as databases used for training are usually too small, incomplete, and heterogeneous in quality. Additionally, data protection constitutes a serious obstacle to data sharing. We propose to use generative models (GM) to produce high-resolution artificial radiographs, which are free of personal identifying information. Blinded analyses by computer vision and radiology experts proved the high similarity of artificial and real radiographs. The combination of multiple GM improves the performance of computer vision algorithms and the integration of artificial data into patient data repositories can compensate for underrepresented disease entities. Furthermore, the low computational effort of our method complies with existing IT infrastructure in hospitals and thus facilitates its dissemination. We envision that our approach could lead to scalable databases of anonymous medical images enabling standardized radiomic analyses at multiple sites.

bioinformatics↗

bioRxiv2019

Hypoxia-Induced PIM Kinase and Laminin-Activated Integrin α6 Mediate Resistance to PI3K Inhibitors in Bone-Metastatic CRPC

Bone-metastatic castration-resistant prostate cancer (CRPC) is lethal due to inherent resistance to androgen deprivation therapy, chemotherapy, and targeted therapies. Despite the fact that a majority of CRPC patients (approximately 70%) harbor a constitutively active PI3K survival pathway, targeting the PI3K/mTOR pathway has failed to increase overall survival in clinical trials. Here, we identified two separate and independent survival pathways induced by the bone tumor microenvironment that are hyperactivated in CRPC and confer resistance to PI3K inhibitors. The first pathway involves integrin 6{beta}1-mediated adhesion to laminin and the second involves hypoxia-induced expression of PIM kinases. In vitro and in vivo models demonstrate that these pathways transduce parallel but independent signals that promote survival by reducing oxidative stress and preventing cell death. We further demonstrate that both pathways drive resistance to PI3K inhibitors in PTEN-negative tumors. These results provide preclinical evidence that combined inhibition of integrin 6{beta}1 and PIM kinase in CRPC is required to overcome tumor microenvironment-mediated resistance to PI3K inhibitors in PTEN-negative tumors within the hypoxic and laminin-rich bone microenvironment.

cancer biology↗

bioRxiv2018

Androgen Receptor-Induced Integrin α6β1 and Adhesion to Laminin Promotes Survival andDrug Resistance in Castration-Resistant Prostate Cancer through BNIP3

Although castration-resistant prostate cancers no longer respond to anti-androgen therapies, the androgen receptor (AR) is still required to promote tumor survival. However, the signaling pathways downstream of AR that promote this survival are not well known. We recently identified an AR-dependent survival pathway whereby AR induction of integrin 6{beta}1 and adhesion to laminin activates NF-kB/RelA signaling and Bcl-xL. This pathway acts in parallel with the PI3K/Akt pathway in Pten-null tumor cells such that combined inhibition of both PI3K and integrin 6{beta}1 is required to kill tumor cells adherent to laminin. However, PTEN-null castration-resistant tumors were not effectively inhibited by this combination. We discovered that BNIP3, a hypoxia-induced BH3-only, pro-mitophagic Bcl2 family member, is induced by androgen in castration-resistant cells through integrin 6{beta}1 signaling to HIF1. Furthermore, castration-resistant cells adherent to laminin were much more efficient at inducing autophagy in response to androgen. Androgen blocked the ability of the PI3K inhibitor PX-866 to kill castration-resistant tumors, but this was reversed by loss of BNIP3. Although BNIP3 was dispensable for androgen-induced autophagy, its mitophagy function was required for BNIP3 to promote resistance to PI3K inhibition. Thus, adhesion to laminin triggers signaling through AR/6{beta}1/HIF1 in castration-resistant cells to drive the expression of BNIP3 and cooperates with AR/6{beta}1-mediated autophagy, both of which contribute to PI3K resistance through induction of mitophagy.

cancer biology↗