Nick Weir

Identification of road networks and optimal routes directly from remote sensing is of critical importance to a broad array of humanitarian and commercial applications. Yet while identification of road pixels has been attempted before, estimation of route travel times from overhead imagery remains a novel problem, particularly for off-nadir overhead imagery. To this end, we extract road networks with travel time estimates from the SpaceNet MVOI dataset. Utilizing the CRESIv2 framework, we… Show more

Identification of road networks and optimal routes directly from remote sensing is of critical importance to a broad array of humanitarian and commercial applications. Yet while identification of road pixels has been attempted before, estimation of route travel times from overhead imagery remains a novel problem, particularly for off-nadir overhead imagery. To this end, we extract road networks with travel time estimates from the SpaceNet MVOI dataset. Utilizing the CRESIv2 framework, we demonstrate the ability to extract road networks in various observation angles and quantify performance at 27 unique nadir angles with the graph-theoretic APLS_length and APLS_time metrics. A minimal gap of 0.03 between APLS_length and APLS_time scores indicates that our approach yields speed limits and travel times with very high fidelity. We also explore the utility of incorporating all available angles during model training, and find a peak score of APLS_time = 0.56. The combined model exhibits greatly improved robustness over angle-specific models, despite the very different appearance of road networks at extremely oblique off-nadir angles versus images captured from directly overhead. Show less

Detection and segmentation of objects in overheard imagery is a challenging task. The variable density, random orientation, small size, and instance-to-instance heterogeneity of objects in overhead imagery calls for approaches distinct from existing models designed for natural scene datasets. Though new overhead imagery datasets are being developed, they almost universally comprise a single view taken from directly overhead (“at nadir”), failing to address a critical variable: look angle. By… Show more

Detection and segmentation of objects in overheard imagery is a challenging task. The variable density, random orientation, small size, and instance-to-instance heterogeneity of objects in overhead imagery calls for approaches distinct from existing models designed for natural scene datasets. Though new overhead imagery datasets are being developed, they almost universally comprise a single view taken from directly overhead (“at nadir”), failing to address a critical variable: look angle. By contrast, views vary in real-world overhead imagery, particularly in dynamic scenarios such as natural disasters where first looks are often over 40◦ off-nadir. This represents an important challenge to computer vision methods, as changing view angle adds distortions, alters resolution, and changes lighting. At present, the impact of these perturbations for algorithmic detection and segmentation of objects is untested. To address this problem, we present an open source MultiView Overhead Imagery dataset, termed SpaceNet MVOI, with 27 unique looks from a broad range of viewing angles (−32.5◦ to 54.0◦). Each of these images cover the same 665 sq km geographic extent and are annotated with 126,747 building footprint labels, enabling direct assessment of the impact of viewpoint perturbation on model performance. We benchmark multiple leading segmentation and object detection models on: (1) building detection, (2) generalization to unseen viewing angles and resolutions, and (3) sensitivity of building footprint extraction to changes in resolution. We find that state of the art segmentation and object detection models struggle to identify buildings in off-nadir imagery and generalize poorly to unseen views, presenting an important benchmark to explore the broadly relevant challenge of detecting small, heterogeneous target objects in visually dynamic contexts. Show less

Selective autophagy comprises cytoplasm-to-lysosome trafficking routes that transport cargos using double-membrane vesicles (autophagosomes). Cargos are detected by receptor proteins, which typically also bind to lipid-conjugated LC3 proteins on autophagosome membranes. We dissected lysosomal delivery of four SQSTM1-like receptors by genome-wide CRISPR screening looking for novel autophagy-related (ATG) factors and trafficking routes. We uncovered new mammalian ATG factors including TMEM41B, an… Show more

Selective autophagy comprises cytoplasm-to-lysosome trafficking routes that transport cargos using double-membrane vesicles (autophagosomes). Cargos are detected by receptor proteins, which typically also bind to lipid-conjugated LC3 proteins on autophagosome membranes. We dissected lysosomal delivery of four SQSTM1-like receptors by genome-wide CRISPR screening looking for novel autophagy-related (ATG) factors and trafficking routes. We uncovered new mammalian ATG factors including TMEM41B, an endoplasmic reticulum membrane protein required for autophagosome membrane expansion and/or closure. Furthermore, we found that certain receptors remain robustly targeted to the lysosome even in the absence of ATG7 or other LC3 conjugation factors. Lastly, we identified a unique genetic fingerprint behind receptor flux in ATG7KO cells, which includes factors implicated in nucleating autophagosome formation and vesicle trafficking factors. Our work uncovers new ATG factors, reveals a malleable network of autophagy receptor genetic interactions, and provides a valuable resource for further mining of novel autophagy mechanisms. Show less

Msp1 is a conserved AAA ATPase in budding yeast localized to mitochondria where it prevents accumulation of mistargeted tail-anchored (TA) proteins, including the peroxisomal TA protein Pex15. Msp1 also resides on peroxisomes but it remains unknown how native TA proteins on mitochondria and peroxisomes evade Msp1 surveillance. We used live-cell quantitative cell microscopy tools and drug-inducible gene expression to dissect Msp1 function. We found that a small fraction of peroxisomal Pex15… Show more

Msp1 is a conserved AAA ATPase in budding yeast localized to mitochondria where it prevents accumulation of mistargeted tail-anchored (TA) proteins, including the peroxisomal TA protein Pex15. Msp1 also resides on peroxisomes but it remains unknown how native TA proteins on mitochondria and peroxisomes evade Msp1 surveillance. We used live-cell quantitative cell microscopy tools and drug-inducible gene expression to dissect Msp1 function. We found that a small fraction of peroxisomal Pex15, exaggerated by overexpression, is turned over by Msp1. Kinetic measurements guided by theoretical modeling revealed that Pex15 molecules at mitochondria display age-independent Msp1 sensitivity. By contrast, Pex15 molecules at peroxisomes are rapidly converted from an initial Msp1-sensitive to an Msp1-resistant state. Lastly, we show that Pex15 interacts with the peroxisomal membrane protein Pex3, which shields Pex15 from Msp1-dependent turnover. In sum, our work argues that Msp1 selects its substrates on the basis of their solitary membrane existence. Show less

Presented early findings in the Denic Lab's Msp1 project to globally renowned experts at the Keystone Mitochondrial Communication research conference. This work was later published in eLife.

An excerpt from the Abstract:

Hundreds of tail-anchored proteins, including soluble N-ethylmaleimide-sensitive factor attachment receptors (SNAREs) involved in vesicle fusion, are inserted post-translationally into the endoplasmic reticulum membrane by a dedicated protein-targeting pathway.

Here we use cell reporters and biochemical reconstitution to define mutations in the Get1/2 transmembrane domain that disrupt tail-anchored protein insertion without interfering with Get1/2… Show more

An excerpt from the Abstract:

Hundreds of tail-anchored proteins, including soluble N-ethylmaleimide-sensitive factor attachment receptors (SNAREs) involved in vesicle fusion, are inserted post-translationally into the endoplasmic reticulum membrane by a dedicated protein-targeting pathway.

Here we use cell reporters and biochemical reconstitution to define mutations in the Get1/2 transmembrane domain that disrupt tail-anchored protein insertion without interfering with Get1/2 cytosolic domain function. These mutations reveal a novel Get1/2 insertase function, in the absence of which substrates stay bound to Get3 despite their proximity to the lipid bilayer; as a consequence, the notion of spontaneous transmembrane domain insertion is a non sequitur. Instead, the Get1/2 transmembrane domain helps to release substrates from Get3 by capturing their transmembrane domains, and these transmembrane interactions define a bona fide pre-integrated intermediate along a facilitated route for tail-anchor entry into the lipid bilayer. Our work sheds light on the fundamental point of convergence between co-translational and post-translational endoplasmic-reticulum membrane protein targeting and insertion: a mechanism for reducing the ability of a targeting factor to shield its substrates enables substrate handover to a transmembrane-domain-docking site embedded in the endoplasmic-reticulum membrane. Show less

Two-component signaling elements play important roles in plants, including a central role in cytokinin signaling. We characterized two-component elements from the monocot rice (Oryza sativa) using several complementary approaches. Phylogenetic analysis reveals relatively simple orthologous relationships among the histidine kinases in rice and Arabidopsis. In contrast, the histidine-containing phosphotransfer proteins (OsHPs) and response regulators (OsRRs) display a higher degree of… Show more

Two-component signaling elements play important roles in plants, including a central role in cytokinin signaling. We characterized two-component elements from the monocot rice (Oryza sativa) using several complementary approaches. Phylogenetic analysis reveals relatively simple orthologous relationships among the histidine kinases in rice and Arabidopsis. In contrast, the histidine-containing phosphotransfer proteins (OsHPs) and response regulators (OsRRs) display a higher degree of lineage-specific expansion. The intracellular localization of several OsHPs and OsRRs was examined in rice and generally found to correspond to the localizations of their dicot counterparts. The functionality of rice type-B OsRRs was tested in Arabidopsis; one from a clade comprised of both monocot and dicot type-B OsRRs complemented an Arabidopsis type-B ARR mutant, but a type-B OsRR from a monocot-specific subfamily generally did not. The expression of genes encoding two-component elements and proteins involved in cytokinin biosynthesis and degradation was analyzed in rice roots and shoots and in response to phytohormones. Nearly all type-A OsRRs and OsHK4 were up-regulated in response to cytokinin, but other cytokinin signaling elements were not appreciably affected. Further, multiple OsCKX genes were up-regulated by cytokinin. ABA treatment decreased the expression of several genes involved in cytokinin biosynthesis and degradation. Auxin affected the expression of a few genes; brassinosteroid and GA had only modest effects. Our results support a shared role for two-component elements in mediating cytokinin signaling in monocots and dicots, and reveal how phytohormones can impact cytokinin function through modulating gene expression. Show less