Mapping the landscape of histomorphological cancer phenotypes using self-supervised learning on unannotated pathology slides.

Cancer diagnosis and management depend upon the extraction of complex information from microscopy images by pathologists, which requires time-consuming expert interpretation prone to human bias. Supervised deep learning approaches have proven powerful, but are inherently limited by the cost and quality of annotations used for training. Therefore, we present Histomorphological Phenotype Learning, a self-supervised methodology requiring no labels and operating via the automatic discovery of discriminatory features in image tiles. Tiles are grouped into morphologically similar clusters which constitute an atlas of histomorphological phenotypes (HP-Atlas), revealing trajectories from benign to malignant tissue via inflammatory and reactive phenotypes. These clusters have distinct features which can be identified using orthogonal methods, linking histologic, molecular and clinical phenotypes. Applied to lung cancer, we show that they align closely with patient survival, with histopathologically recognised tumor types and growth patterns, and with transcriptomic measures of immunophenotype. These properties are maintained in a multi-cancer study.

Blueprinting extendable nanomaterials with standardized protein blocks.

A wooden house frame consists of many different lumber pieces, but because of the regularity of these building blocks, the structure can be designed using straightforward geometrical principles. The design of multicomponent protein assemblies, in comparison, has been much more complex, largely owing to the irregular shapes of protein structures1. Here we describe extendable linear, curved and angled protein building blocks, as well as inter-block interactions, that conform to specified geometric standards; assemblies designed using these blocks inherit their extendability and regular interaction surfaces, enabling them to be expanded or contracted by varying the number of modules, and reinforced with secondary struts. Using X-ray crystallography and electron microscopy, we validate nanomaterial designs ranging from simple polygonal and circular oligomers that can be concentrically nested, up to large polyhedral nanocages and unbounded straight 'train track' assemblies with reconfigurable sizes and geometries that can be readily blueprinted. Because of the complexity of protein structures and sequence-structure relationships, it has not previously been possible to build up large protein assemblies by deliberate placement of protein backbones onto a blank three-dimensional canvas; the simplicity and geometric regularity of our design platform now enables construction of protein nanomaterials according to 'back of an envelope' architectural blueprints.

Self-Supervised Learning Reveals Clinically Relevant Histomorphological Patterns for Therapeutic Strategies in Colon Cancer.

Self-supervised learning (SSL) automates the extraction and interpretation of histopathology features on unannotated hematoxylin-and-eosin-stained whole-slide images (WSIs). We trained an SSL Barlow Twins-encoder on 435 TCGA colon adenocarcinoma WSIs to extract features from small image patches. Leiden community detection then grouped tiles into histomorphological phenotype clusters (HPCs). HPC reproducibility and predictive ability for overall survival was confirmed in an independent clinical trial cohort (N=1213 WSIs). This unbiased atlas resulted in 47 HPCs displaying unique and sharing clinically significant histomorphological traits, highlighting tissue type, quantity, and architecture, especially in the context of tumor stroma. Through in-depth analysis of these HPCs, including immune landscape and gene set enrichment analysis, and association to clinical outcomes, we shed light on the factors influencing survival and responses to treatments like standard adjuvant chemotherapy and experimental therapies. Further exploration of HPCs may unveil new insights and aid decision-making and personalized treatments for colon cancer patients.

Inflammation in the tumor-adjacent lung as a predictor of clinical outcome in lung adenocarcinoma.

Approximately 30% of early-stage lung adenocarcinoma patients present with disease progression after successful surgical resection. Despite efforts of mapping the genetic landscape, there has been limited success in discovering predictive biomarkers of disease outcomes. Here we performed a systematic multi-omic assessment of 143 tumors and matched tumor-adjacent, histologically-normal lung tissue with long-term patient follow-up. Through histologic, mutational, and transcriptomic profiling of tumor and adjacent-normal tissue, we identified an inflammatory gene signature in tumor-adjacent tissue as the strongest clinical predictor of disease progression. Single-cell transcriptomic analysis demonstrated the progression-associated inflammatory signature was expressed in both immune and non-immune cells, and cell type-specific profiling in monocytes further improved outcome predictions. Additional analyses of tumor-adjacent transcriptomic data from The Cancer Genome Atlas validated the association of the inflammatory signature with worse outcomes across cancers. Collectively, our study suggests that molecular profiling of tumor-adjacent tissue can identify patients at high risk for disease progression.

Unlatching of the stem domains in the Staphylococcus aureus pore-forming leukocidin LukAB influences toxin oligomerization.

Staphylococcus aureus (S. aureus) is a serious global pathogen that causes a diverse range of invasive diseases. S. aureus utilizes a family of pore-forming toxins, known as bi-component leukocidins, to evade the host immune response and promote infection. Among these is LukAB (leukocidin A/leukocidin B), a toxin that assembles into an octameric ß-barrel pore in the target cell membrane, resulting in host cell death. The established cellular receptor for LukAB is CD11b of the Mac-1 complex. Here, we show that hydrogen voltage-gated channel 1 is also required for the cytotoxicity of all major LukAB variants. We demonstrate that while each receptor is sufficient to recruit LukAB to the plasma membrane, both receptors are required for maximal lytic activity. Why LukAB requires two receptors, and how each of these receptors contributes to pore-formation remains unknown. To begin to resolve this, we performed an alanine scanning mutagenesis screen to identify mutations that allow LukAB to maintain cytotoxicity without CD11b. We discovered 30 mutations primarily localized in the stem domains of LukA and LukB that enable LukAB to exhibit full cytotoxicity in the absence of CD11b. Using crosslinking, electron microscopy, and hydroxyl radical protein footprinting, we show these mutations increase the solvent accessibility of the stem domain, priming LukAB for oligomerization. Together, our data support a model in which CD11b binding unlatches the membrane penetrating stem domains of LukAB, and this change in flexibility promotes toxin oligomerization.

Blueprinting expandable nanomaterials with standardized protein building blocks.

A wooden house frame consists of many different lumber pieces, but because of the regularity of these building blocks, the structure can be designed using straightforward geometrical principles. The design of multicomponent protein assemblies in comparison has been much more complex, largely due to the irregular shapes of protein structures 1 . Here we describe extendable linear, curved, and angled protein building blocks, as well as inter-block interactions that conform to specified geometric standards; assemblies designed using these blocks inherit their extendability and regular interaction surfaces, enabling them to be expanded or contracted by varying the number of modules, and reinforced with secondary struts. Using X-ray crystallography and electron microscopy, we validate nanomaterial designs ranging from simple polygonal and circular oligomers that can be concentrically nested, up to large polyhedral nanocages and unbounded straight "train track" assemblies with reconfigurable sizes and geometries that can be readily blueprinted. Because of the complexity of protein structures and sequence-structure relationships, it has not been previously possible to build up large protein assemblies by deliberate placement of protein backbones onto a blank 3D canvas; the simplicity and geometric regularity of our design platform now enables construction of protein nanomaterials according to "back of an envelope" architectural blueprints.

Protein-protein interactions in the Mla lipid transport system probed by computational structure prediction and deep mutational scanning.

The outer membrane (OM) of Gram-negative bacteria is an asymmetric bilayer that protects the cell from external stressors, such as antibiotics. The Mla transport system is implicated in the Maintenance of OM Lipid Asymmetry by mediating retrograde phospholipid transport across the cell envelope. Mla uses a shuttle-like mechanism to move lipids between the MlaFEDB inner membrane complex and the MlaA-OmpF/C OM complex, via a periplasmic lipid-binding protein, MlaC. MlaC binds to MlaD and MlaA, but the underlying protein-protein interactions that facilitate lipid transfer are not well understood. Here, we take an unbiased deep mutational scanning approach to map the fitness landscape of MlaC from Escherichia coli, which provides insights into important functional sites. Combining this analysis with AlphaFold2 structure predictions and binding experiments, we map the MlaC-MlaA and MlaC-MlaD protein-protein interfaces. Our results suggest that the MlaD and MlaA binding surfaces on MlaC overlap to a large extent, leading to a model in which MlaC can only bind one of these proteins at a time. Low-resolution cryo-electron microscopy (cryo-EM) maps of MlaC bound to MlaFEDB suggest that at least two MlaC molecules can bind to MlaD at once, in a conformation consistent with AlphaFold2 predictions. These data lead us to a model for MlaC interaction with its binding partners and insights into lipid transfer steps that underlie phospholipid transport between the bacterial inner and OMs.

Self-supervised artificial intelligence predicts recurrence, metastasis and disease specific death from primary cutaneous squamous cell carcinoma at diagnosis.

Primary cutaneous squamous cell carcinoma (cSCC) is responsible for ~10,000 deaths annually in the United States. Stratification of risk of poor outcome (PO) including recurrence, metastasis and disease specific death (DSD) at initial biopsy would significantly impact clinical decision-making during the initial post operative period where intervention has been shown to be most effective. In this multi-institutional study, we developed a state-of-the-art self-supervised deep-learning approach with interpretability power and demonstrated its ability to predict poor outcomes of cSCCs at the time of initial biopsy. By highlighting histomorphological phenotypes, our approach demonstrates that poor differentiation and deep invasion correlate with poor prognosis. Our approach is particularly efficient at defining poor outcome risk in Brigham and Women's Hospital (BWH) T2a and American Joint Committee on Cancer (AJCC) T2 cSCCs. This bridges a significant gap in our ability to assess risk among T2a/T2 cSCCs and may be useful in defining patients at highest risk of poor outcome at the time of diagnosis. Early identification of highest-risk patients could signal implementation of more stringent surveillance, rigorous diagnostic work up and identify patients who might best respond to early postoperative adjunctive treatment.

Conformational states of the microtubule nucleator, the γ-tubulin ring complex.

Microtubules (MTs) perform essential functions in the cell, and it is critical that they are made at the correct cellular location and cell cycle stage. This nucleation process is catalyzed by the γ-tubulin ring complex (γ-TuRC), a cone-shaped protein complex composed of over 30 subunits. Despite recent insight into the structure of vertebrate γ-TuRC, which shows that its diameter is wider than that of a MT, and that it exhibits little of the symmetry expected for an ideal MT template, the question of how γ-TuRC achieves MT nucleation remains open. Here, we utilized single particle cryo-EM to identify two conformations of γ-TuRC. The helix composed of 14 γ-tubulins at the top of the γ-TuRC cone undergoes substantial deformation, which is predominantly driven by bending of the hinge between the GRIP1 and GRIP2 domains of the γ-tubulin complex proteins. However, surprisingly, this deformation does not remove the inherent asymmetry of γ-TuRC. To further investigate the role of γ-TuRC conformational change, we used cryo electron-tomography (cryo-ET) to obtain a 3D reconstruction of γ-TuRC bound to a nucleated MT, providing insight into the post-nucleation state. Rigid-body fitting of our cryo-EM structures into this reconstruction suggests that the MT lattice is nucleated by spokes 2 through 14 of the γ-tubulin helix, which entails spokes 13 and 14 becoming more structured than what is observed in apo γ-TuRC. Together, our results allow us to propose a model for conformational changes in γ-TuRC and how these may facilitate MT formation in a cell.

Structure and mechanism of the bacterial lipid ABC transporter, MlaFEDB.

The cell envelope of Gram-negative bacteria is composed of an inner membrane, outer membane, and an intervening periplasmic space. How the outer membrane lipids are trafficked and assembled there, and how the asymmetry of the outer membrane is maintained is an area of intense research. The Mla system has been implicated in the maintenance of lipid asymmetry in the outer membrane, and is generally thought to drive the removal of mislocalized phospholipids from the outer membrane and their retrograde transport to the inner membrane. At the heart of the Mla pathway is a structurally unique ABC transporter complex in the inner membrane, called MlaFEDB. Recently, an explosion of cryo-EM studies has begun to shed light on the structure and lipid translocation mechanism of MlaFEDB, with many parallels to other ABC transporter families, including human ABCA and ABCG, as well as bacterial lipopolysaccharide and O-antigen transporters. Here we synthesize information from all available structures, and propose a model for lipid trafficking across the cell envelope by MlaFEDB.

Role of Ring6 in the Function of the E. coli MCE Protein LetB.

LetB is a tunnel-forming protein found in the cell envelope of some double-membraned bacteria, and is thought to be important for the transport of lipids between the inner and outer membranes. In Escherichia coli the LetB tunnel is formed from a stack of seven rings (Ring1 - Ring7), in which each ring is composed of a homo-hexameric assembly of MCE domains. The primary sequence of each MCE domain of the LetB protein is substantially divergent from the others, making each MCE ring unique in nature. The role of each MCE domain and how it contributes to the function of LetB is not well understood. Here we probed the importance of each MCE ring for the function of LetB, using a combination of bacterial growth assays and cryo-EM. Surprisingly, we find that ΔRing3 and ΔRing6 mutants, in which Ring3 and Ring6 have been deleted, confer increased resistance to membrane perturbing agents. Specific mutations in the pore-lining loops of Ring6 similarly confer increased resistance. A cryo-EM structure of the ΔRing6 mutant shows that despite the absence of Ring6, which leads to a shorter assembly, the overall architecture is maintained, highlighting the modular nature of MCE proteins. Previous work has shown that Ring6 is dynamic and in its closed state, may restrict the passage of substrate through the tunnel. Our work suggests that removal of Ring6 may relieve this restriction. The deletion of Ring6 combined with mutations in the pore-lining loops leads to a model for the tunnel gating mechanism of LetB. Together, these results provide insight into the functional roles of individual MCE domains and pore-lining loops in the LetB protein.

Deep Learning and Pathomics Analyses Reveal Cell Nuclei as Important Features for Mutation Prediction of BRAF-Mutated Melanomas.

Image-based analysis as a method for mutation detection can be advantageous in settings when tumor tissue is limited or unavailable for direct testing. In this study, we utilize two distinct and complementary machine-learning methods of analyzing whole-slide images for predicting mutated BRAF. In the first method, whole-slide images of melanomas from 256 patients were used to train a deep convolutional neural network to develop a fully automated model that first selects for tumor-rich areas (area under the curve = 0.96) and then predicts for mutated BRAF (area under the curve = 0.71). Saliency mapping was performed and revealed that pixels corresponding to nuclei were the most relevant to network learning. In the second method, whole-slide images were analyzed using a pathomics pipeline that first annotates nuclei and then quantifies nuclear features, showing that mutated BRAF nuclei were significantly larger and rounder than BRAF‒wild-type nuclei. Finally, we developed a model that combines clinical information, deep learning, and pathomics that improves the predictive performance for mutated BRAF to an area under the curve of 0.89. Not only does this provide additional insights on how BRAF mutations affect tumor structural characteristics, but machine learning‒based analysis of whole-slide images also has the potential to be integrated into higher-order models for understanding tumor biology.

Targeting allostery in the Dynein motor domain with small molecule inhibitors.

Cytoplasmic dyneins are AAA (ATPase associated with diverse cellular activities) motor proteins responsible for microtubule minus-end-directed intracellular transport. Dynein's unusually large size, four distinct nucleotide-binding sites, and conformational dynamics pose challenges for the design of potent and selective chemical inhibitors. Here we use structural approaches to develop a model for the inhibition of a well-characterized S. cerevisiae dynein construct by pyrazolo-pyrimidinone-based compounds. These data, along with functional assays of dynein motility and mutagenesis studies, suggest that the compounds inhibit dynein by engaging the regulatory ATPase sites in the AAA3 and AAA4 domains, and not by interacting with dynein's main catalytic site in the AAA1 domain. A double Walker B mutation of the AAA3 and AAA4 sites substantially reduces enzyme activity, suggesting that targeting these regulatory domains is sufficient to inhibit dynein. Our findings reveal how chemical inhibitors can be designed to disrupt allosteric communication across dynein's AAA domains.

HEAL: an automated deep learning framework for cancer histopathology image analysis.

MOTIVATION: Digital pathology supports analysis of histopathological images using deep learning methods at a large-scale. However, applications of deep learning in this area have been limited by the complexities of configuration of the computational environment and of hyperparameter optimization, which hinder deployment and reduce reproducibility. RESULTS: Here, we propose HEAL, a deep learning-based automated framework for easy, flexible and multi-faceted histopathological image analysis. We demonstrate its utility and functionality by performing two case studies on lung cancer and one on colon cancer. Leveraging the capability of Docker, HEAL represents an ideal end-to-end tool to conduct complex histopathological analysis and enables deep learning in a broad range of applications for cancer image analysis. AVAILABILITY AND IMPLEMENTATION: The docker image of HEAL is available at https://hub.docker.com/r/docurdt/heal and related documentation and datasets are available at http://heal.erc.monash.edu.au. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Structure of the radial spoke head and insights into its role in mechanoregulation of ciliary beating.

Motile cilia power cell locomotion and drive extracellular fluid flow by propagating bending waves from their base to tip. The coordinated bending of cilia requires mechanoregulation by the radial spoke (RS) protein complexes and the microtubule central pair (CP). Despite their importance for ciliary motility across eukaryotes, the molecular function of the RSs is unknown. Here, we reconstituted the Chlamydomonas reinhardtii RS head that abuts the CP and determined its structure using single-particle cryo-EM to 3.1-Å resolution, revealing a flat, negatively charged surface supported by a rigid core of tightly intertwined proteins. Mutations in this core, corresponding to those involved in human ciliopathies, compromised the stability of the recombinant complex, providing a molecular basis for disease. Partially reversing the negative charge on the RS surface impaired motility in C. reinhardtii. We propose that the RS-head architecture is well-suited for mechanoregulation of ciliary beating through physical collisions with the CP.

Using Machine Learning Algorithms to Predict Immunotherapy Response in Patients with Advanced Melanoma.

PURPOSE: Several biomarkers of response to immune checkpoint inhibitors (ICI) show potential but are not yet scalable to the clinic. We developed a pipeline that integrates deep learning on histology specimens with clinical data to predict ICI response in advanced melanoma. EXPERIMENTAL DESIGN: We used a training cohort from New York University (New York, NY) and a validation cohort from Vanderbilt University (Nashville, TN). We built a multivariable classifier that integrates neural network predictions with clinical data. A ROC curve was generated and the optimal threshold was used to stratify patients as high versus low risk for progression. Kaplan-Meier curves compared progression-free survival (PFS) between the groups. The classifier was validated on two slide scanners (Aperio AT2 and Leica SCN400). RESULTS: The multivariable classifier predicted response with AUC 0.800 on images from the Aperio AT2 and AUC 0.805 on images from the Leica SCN400. The classifier accurately stratified patients into high versus low risk for disease progression. Vanderbilt patients classified as high risk for progression had significantly worse PFS than those classified as low risk (P = 0.02 for the Aperio AT2; P = 0.03 for the Leica SCN400). CONCLUSIONS: Histology slides and patients' clinicodemographic characteristics are readily available through standard of care and have the potential to predict ICI treatment outcomes. With prospective validation, we believe our approach has potential for integration into clinical practice.

Structure of bacterial phospholipid transporter MlaFEDB with substrate bound.

In double-membraned bacteria, phospholipid transport across the cell envelope is critical to maintain the outer membrane barrier, which plays a key role in virulence and antibiotic resistance. An MCE transport system called Mla has been implicated in phospholipid trafficking and outer membrane integrity, and includes an ABC transporter, MlaFEDB. The transmembrane subunit, MlaE, has minimal sequence similarity to other transporters, and the structure of the entire inner-membrane MlaFEDB complex remains unknown. Here, we report the cryo-EM structure of MlaFEDB at 3.05 Å resolution, revealing distant relationships to the LPS and MacAB transporters, as well as the eukaryotic ABCA/ABCG families. A continuous transport pathway extends from the MlaE substrate-binding site, through the channel of MlaD, and into the periplasm. Unexpectedly, two phospholipids are bound to MlaFEDB, suggesting that multiple lipid substrates may be transported each cycle. Our structure provides mechanistic insight into substrate recognition and transport by MlaFEDB.

Structure of MlaFB uncovers novel mechanisms of ABC transporter regulation.

ABC transporters facilitate the movement of diverse molecules across cellular membranes, but how their activity is regulated post-translationally is not well understood. Here we report the crystal structure of MlaFB from E. coli, the cytoplasmic portion of the larger MlaFEDB ABC transporter complex, which drives phospholipid trafficking across the bacterial envelope to maintain outer membrane integrity. MlaB, a STAS domain protein, binds the ABC nucleotide binding domain, MlaF, and is required for its stability. Our structure also implicates a unique C-terminal tail of MlaF in self-dimerization. Both the C-terminal tail of MlaF and the interaction with MlaB are required for the proper assembly of the MlaFEDB complex and its function in cells. This work leads to a new model for how an important bacterial lipid transporter may be regulated by small proteins, and raises the possibility that similar regulatory mechanisms may exist more broadly across the ABC transporter family.

The transition state and regulation of γ-TuRC-mediated microtubule nucleation revealed by single molecule microscopy.

Determining how microtubules (MTs) are nucleated is essential for understanding how the cytoskeleton assembles. While the MT nucleator, γ-tubulin ring complex (γ-TuRC) has been identified, precisely how γ-TuRC nucleates a MT remains poorly understood. Here, we developed a single molecule assay to directly visualize nucleation of a MT from purified Xenopus laevis γ-TuRC. We reveal a high γ-/αß-tubulin affinity, which facilitates assembly of a MT from γ-TuRC. Whereas spontaneous nucleation requires assembly of 8 αß-tubulins, nucleation from γ-TuRC occurs efficiently with a cooperativity of 4 αß-tubulin dimers. This is distinct from pre-assembled MT seeds, where a single dimer is sufficient to initiate growth. A computational model predicts our kinetic measurements and reveals the rate-limiting transition where laterally associated αß-tubulins drive γ-TuRC into a closed conformation. NME7, TPX2, and the putative activation domain of CDK5RAP2 h γ-TuRC-mediated nucleation, while XMAP215 drastically increases the nucleation efficiency by strengthening the longitudinal γ-/αß-tubulin interaction.

LetB Structure Reveals a Tunnel for Lipid Transport across the Bacterial Envelope.

Gram-negative bacteria are surrounded by an outer membrane composed of phospholipids and lipopolysaccharide, which acts as a barrier and contributes to antibiotic resistance. The systems that mediate phospholipid trafficking across the periplasm, such as MCE (Mammalian Cell Entry) transporters, have not been well characterized. Our ~3.5 Å cryo-EM structure of the E. coli MCE protein LetB reveals an ~0.6 megadalton complex that consists of seven stacked rings, with a central hydrophobic tunnel sufficiently long to span the periplasm. Lipids bind inside the tunnel, suggesting that it functions as a pathway for lipid transport. Cryo-EM structures in the open and closed states reveal a dynamic tunnel lining, with implications for gating or substrate translocation. Our results support a model in which LetB establishes a physical link between the two membranes and creates a hydrophobic pathway for the translocation of lipids across the periplasm.