Application of Model-Informed Drug Development in Dose Selection and Optimization for siRNA Therapies.

The application of model-informed drug development (MIDD) has revolutionized drug development and regulatory decision making, transforming the process into one that is more efficient, effective, and patient centered. A critical application of MIDD is to facilitate dose selection and optimization, which play a pivotal role in improving efficacy, safety, and tolerability profiles of a candidate drug. With the surge of interest in small interfering RNA (siRNA) drugs as a promising class of therapeutics, their applications in various disease areas have been extensively studied preclinically. However, dosing selection and optimization experience for siRNA in humans is limited. Unique challenges exist for the dose evaluation of siRNA due to the temporal discordance between pharmacokinetic and pharmacodynamic profiles, as well as limited available clinical experience and considerable interindividual variability. This review highlights the pivotal role of MIDD in facilitating dose selection and optimization for siRNA therapeutics. Based on past experiences with approved siRNA products, MIDD has demonstrated its ability to aid in dose selection for clinical trials and enabling optimal dosing for the general patient population. In addition, MIDD presents an opportunity for dose individualization based on patient characteristics, enhancing the precision and effectiveness of siRNA therapeutics. In conclusion, the integration of MIDD offers substantial advantages in navigating the complex challenges of dose selection and optimization in siRNA drug development, which in turn accelerates the development process, supports regulatory decision making, and ultimately improves the clinical outcomes of siRNA-based therapies, fostering advancements in precision medicine across a diverse range of diseases.

Trellis tree-based analysis reveals stromal regulation of patient-derived organoid drug responses.

Patient-derived organoids (PDOs) can model personalized therapy responses; however, current screening technologies cannot reveal drug response mechanisms or how tumor microenvironment cells alter therapeutic performance. To address this, we developed a highly multiplexed mass cytometry platform to measure post-translational modification (PTM) signaling, DNA damage, cell-cycle activity, and apoptosis in >2,500 colorectal cancer (CRC) PDOs and cancer-associated fibroblasts (CAFs) in response to clinical therapies at single-cell resolution. To compare patient- and microenvironment-specific drug responses in thousands of single-cell datasets, we developed "Trellis"-a highly scalable, tree-based treatment effect analysis method. Trellis single-cell screening revealed that on-target cell-cycle blockage and DNA-damage drug effects are common, even in chemorefractory PDOs. However, drug-induced apoptosis is rarer, patient-specific, and aligns with cancer cell PTM signaling. We find that CAFs can regulate PDO plasticity-shifting proliferative colonic stem cells (proCSCs) to slow-cycling revival colonic stem cells (revCSCs) to protect cancer cells from chemotherapy.

GEODESIC SINKHORN FOR FAST AND ACCURATE OPTIMAL TRANSPORT ON MANIFOLDS.

Efficient computation of optimal transport distance between distributions is of growing importance in data science. Sinkhorn-based methods are currently the state-of-the-art for such computations, but require On2 computations. In addition, Sinkhorn-based methods commonly use an Euclidean ground distance between datapoints. However, with the prevalence of manifold structured scientific data, it is often desirable to consider geodesic ground distance. Here, we tackle both issues by proposing Geodesic Sinkhorn-based on diffusing a heat kernel on a manifold graph. Notably, Geodesic Sinkhorn requires only O(nlog⁡n) computation, as we approximate the heat kernel with Chebyshev polynomials based on the sparse graph Laplacian. We apply our method to the computation of barycenters of several distributions of high dimensional single cell data from patient samples undergoing chemotherapy. In particular, we define the barycentric distance as the distance between two such barycenters. Using this definition, we identify an optimal transport distance and path associated with the effect of treatment on cellular data.

A Heat Diffusion Perspective on Geodesic Preserving Dimensionality Reduction.

Diffusion-based manifold learning methods have proven useful in representation learning and dimensionality reduction of modern high dimensional, high throughput, noisy datasets. Such datasets are especially present in fields like biology and physics. While it is thought that these methods preserve underlying manifold structure of data by learning a proxy for geodesic distances, no specific theoretical links have been established. Here, we establish such a link via results in Riemannian geometry explicitly connecting heat diffusion to manifold distances. In this process, we also formulate a more general heat kernel based manifold embedding method that we call heat geodesic embeddings. This novel perspective makes clearer the choices available in manifold learning and denoising. Results show that our method outperforms existing state of the art in preserving ground truth manifold distances, and preserving cluster structure in toy datasets. We also showcase our method on single cell RNA-sequencing datasets with both continuum and cluster structure, where our method enables interpolation of withheld timepoints of data. Finally, we show that parameters of our more general method can be configured to give results similar to PHATE (a state-of-the-art diffusion based manifold learning method) as well as SNE (an attraction/repulsion neighborhood based method that forms the basis of t-SNE).

Single-cell analysis reveals inflammatory interactions driving macular degeneration.

Due to commonalities in pathophysiology, age-related macular degeneration (AMD) represents a uniquely accessible model to investigate therapies for neurodegenerative diseases, leading us to examine whether pathways of disease progression are shared across neurodegenerative conditions. Here we use single-nucleus RNA sequencing to profile lesions from 11 postmortem human retinas with age-related macular degeneration and 6 control retinas with no history of retinal disease. We create a machine-learning pipeline based on recent advances in data geometry and topology and identify activated glial populations enriched in the early phase of disease. Examining single-cell data from Alzheimer's disease and progressive multiple sclerosis with our pipeline, we find a similar glial activation profile enriched in the early phase of these neurodegenerative diseases. In late-stage age-related macular degeneration, we identify a microglia-to-astrocyte signaling axis mediated by interleukin-1ß which drives angiogenesis characteristic of disease pathogenesis. We validated this mechanism using in vitro and in vivo assays in mouse, identifying a possible new therapeutic target for AMD and possibly other neurodegenerative conditions. Thus, due to shared glial states, the retina provides a potential system for investigating therapeutic approaches in neurodegenerative diseases.

A trailing ribosome speeds up RNA polymerase at the expense of transcript fidelity via force and allostery.

In prokaryotes, translation can occur on mRNA that is being transcribed in a process called coupling. How the ribosome affects the RNA polymerase (RNAP) during coupling is not well understood. Here, we reconstituted the E. coli coupling system and demonstrated that the ribosome can prevent pausing and termination of RNAP and double the overall transcription rate at the expense of fidelity. Moreover, we monitored single RNAPs coupled to ribosomes and show that coupling increases the pause-free velocity of the polymerase and that a mechanical assisting force is sufficient to explain the majority of the effects of coupling. Also, by cryo-EM, we observed that RNAPs with a terminal mismatch adopt a backtracked conformation, while a coupled ribosome allosterically induces these polymerases toward a catalytically active anti-swiveled state. Finally, we demonstrate that prolonged RNAP pausing is detrimental to cell viability, which could be prevented by polymerase reactivation through a coupled ribosome.

Monitoring the compaction of single DNA molecules in Xenopus egg extract in real time.

DNA compaction is required for the condensation and resolution of chromosomes during mitosis, but the relative contribution of individual chromatin factors to this process is poorly understood. We developed a physiological, cell-free system using high-speed Xenopus egg extracts and optical tweezers to investigate real-time mitotic chromatin fiber formation and force-induced disassembly on single DNA molecules. Compared to interphase extract, which compacted DNA by ~60%, metaphase extract reduced DNA length by over 90%, reflecting differences in whole-chromosome morphology under these two conditions. Depletion of the core histone chaperone ASF1, which inhibits nucleosome assembly, decreased the final degree of metaphase fiber compaction by 29%, while depletion of linker histone H1 had a greater effect, reducing total compaction by 40%. Compared to controls, both depletions reduced the rate of compaction, led to more short periods of decompaction, and increased the speed of force-induced fiber disassembly. In contrast, depletion of condensin from metaphase extract strongly inhibited fiber assembly, resulting in transient compaction events that were rapidly reversed under high force. Altogether, these findings support a speculative model in which condensin plays the predominant role in mitotic DNA compaction, while core and linker histones act to reduce slippage during loop extrusion and modulate the degree of DNA compaction.

Resistance of MMTV-NeuT/ATTAC mice to anti-PD-1 immune checkpoint therapy is associated with macrophage infiltration and Wnt pathway expression.

One of the central challenges for cancer therapy is the identification of factors in the tumor microenvironment that increase tumor progression and immune tolerance. In breast cancer, fibrosis is a histopathologic criterion for invasive cancer and poor survival that results from inflammatory factors and remodeling of the extracellular matrix to produce an immune tolerant microenvironment. To determine whether tolerance is associated with the immune checkpoint, Programmed Cell Death 1 (PD-1), NeuT/ATTAC mice, a conditional model of mammary fibrosis that we recently developed, were administered a murine-specific anti-PD-1 mAb related to pembrolizumab, and drug response was monitored by tumor development, imaging mass cytometry, immunohistochemistry and tumor gene expression by RNAseq. Tumor progression in NeuT/ATTAC mice was unaffected by weekly injection of anti-PD-1 over four months. Insensitivity to anti-PD-1 was associated with several processes, including increased tumor-associated macrophages (TAM), epithelial to mesenchymal transition (EMT), fibroblast proliferation, an enhanced extracellular matrix and the Wnt signaling pathway, including increased expression of Fzd5, Wnt5a, Vimentin, Mmp3, Col2a1 and Tgfß1. These results suggest potential therapeutic avenues that may enhance PD-1 immune checkpoint sensitivity, including the use of tumor microenvironment targeted agents and Wnt pathway inhibitors.

Immune cells and their inflammatory mediators modify ß cells and cause checkpoint inhibitor-induced diabetes.

Checkpoint inhibitors (CPIs) targeting programmed death 1 (PD-1)/programmed death ligand 1 (PD-L1) and cytotoxic T lymphocyte antigen 4 (CTLA-4) have revolutionized cancer treatment but can trigger autoimmune complications, including CPI-induced diabetes mellitus (CPI-DM), which occurs preferentially with PD-1 blockade. We found evidence of pancreatic inflammation in patients with CPI-DM with shrinkage of pancreases, increased pancreatic enzymes, and in a case from a patient who died with CPI-DM, peri-islet lymphocytic infiltration. In the NOD mouse model, anti-PD-L1 but not anti-CTLA-4 induced diabetes rapidly. RNA sequencing revealed that cytolytic IFN-γ+CD8+ T cells infiltrated islets with anti-PD-L1. Changes in ß cells were predominantly driven by IFN-γ and TNF-α and included induction of a potentially novel ß cell population with transcriptional changes suggesting dedifferentiation. IFN-γ increased checkpoint ligand expression and activated apoptosis pathways in human ß cells in vitro. Treatment with anti-IFN-γ and anti-TNF-α prevented CPI-DM in anti-PD-L1-treated NOD mice. CPIs targeting the PD-1/PD-L1 pathway resulted in transcriptional changes in ß cells and immune infiltrates that may lead to the development of diabetes. Inhibition of inflammatory cytokines can prevent CPI-DM, suggesting a strategy for clinical application to prevent this complication.

Friction-driven membrane scission by the human ESCRT-III proteins CHMP1B and IST1.

The endosomal sorting complexes required for transport (ESCRT) system is an ancient and ubiquitous membrane scission machinery that catalyzes the budding and scission of membranes. ESCRT-mediated scission events, exemplified by those involved in the budding of HIV-1, are usually directed away from the cytosol ("reverse topology"), but they can also be directed toward the cytosol ("normal topology"). The ESCRT-III subunits CHMP1B and IST1 can coat and constrict positively curved membrane tubes, suggesting that these subunits could catalyze normal topology membrane severing. CHMP1B and IST1 bind and recruit the microtubule-severing AAA+ ATPase spastin, a close relative of VPS4, suggesting that spastin could have a VPS4-like role in normal-topology membrane scission. Here, we reconstituted the process in vitro using membrane nanotubes pulled from giant unilamellar vesicles using an optical trap in order to determine whether CHMP1B and IST1 are capable of membrane severing on their own or in concert with VPS4 or spastin. CHMP1B and IST1 copolymerize on membrane nanotubes, forming stable scaffolds that constrict the tubes, but do not, on their own, lead to scission. However, CHMP1B-IST1 scaffolded tubes were severed when an additional extensional force was applied, consistent with a friction-driven scission mechanism. We found that spastin colocalized with CHMP1B-enriched sites but did not disassemble the CHMP1B-IST1 coat from the membrane. VPS4 resolubilized CHMP1B and IST1 without leading to scission. These observations show that the CHMP1B-IST1 ESCRT-III combination is capable of severing membranes by a friction-driven mechanism that is independent of VPS4 and spastin.

Structure of cytoplasmic ring of nuclear pore complex by integrative cryo-EM and AlphaFold.

INTRODUCTION The nuclear pore complex (NPC) is the molecular conduit in the nuclear membrane of eukaryotic cells that regulates import and export of biomolecules between the nucleus and the cytosol, with vertebrate NPCs ~110 to 125 MDa in molecular mass and ~120 nm in diameter. NPCs are organized into four main rings: the cytoplasmic ring (CR) at the cytosolic side, the inner ring and the luminal ring on the plane of the nuclear membrane, and the nuclear ring facing the nucleus. Each ring possesses an approximate eightfold symmetry and is composed of multiple copies of different nucleoporins. NPCs have been implicated in numerous biological processes, and their dysfunctions are associated with a growing number of serious human diseases. However, despite pioneering studies from many groups over the past two decades, we still lack a full understanding of NPCs' organization, dynamics, and complexity. RATIONALE We used the Xenopus laevis oocyte as a model system for the structural characterization because each oocyte possesses a large number of NPC particles that can be visualized on native nuclear membranes without the aid of detergent extraction. We used single-particle cryo-electron microscopy (cryo-EM) analysis on data collected at different stage tilt angles for three-dimensional reconstruction and structure prediction with AlphaFold for model building. RESULTS We reconstructed the CR map of X. laevis NPC at 6.9 and 6.7 Å resolutions for the full CR protomer and a core region, respectively, and predicted the structures of the individual nucleoporins using AlphaFold because no high-resolution models of X. laevis Nups were available. For any ambiguous subunit interactions, we also predicted complex structures, which further guided model fitting of the CR protomer. We placed the nucleoporin or complex structures into the CR density to obtain an almost full CR atomic model, composed of the inner and outer Y-complexes, two copies of Nup205, two copies of the Nup214-Nup88-Nup62 complex, one Nup155, and five copies of Nup358. In particular, we predicted the largest protein in the NPC, Nup358, as having an S-shaped globular domain, a coiled-coil domain, and a largely disordered C-terminal region containing phenylalanine-glycine (FG) repeats previously shown to form a gel-like condensate phase for selective cargo passage. Four of the Nup358 copies clamp around the inner and outer Y-complexes to stabilize the CR, and the fifth Nup358 situates in the center of the cluster of clamps. AlphaFold also predicted a homo-oligomeric, likely specifically pentameric, coiled-coil structure of Nup358 that may provide the avidity for Nup358 recruitment to the NPC and for lowering the threshold for Nup358 condensation in NPC biogenesis. CONCLUSION Our studies offer an example of integrative cryo-EM and structure prediction as a general approach for attaining more precise models of megadalton protein complexes from medium-resolution density maps. The more accurate and almost complete model of the CR presented here expands our understanding of the molecular interactions in the NPC and represents a substantial step forward toward the molecular architecture of a full NPC, with implications for NPC function, biogenesis, and regulation. [Figure: see text].

Multiscale PHATE identifies multimodal signatures of COVID-19.

As the biomedical community produces datasets that are increasingly complex and high dimensional, there is a need for more sophisticated computational tools to extract biological insights. We present Multiscale PHATE, a method that sweeps through all levels of data granularity to learn abstracted biological features directly predictive of disease outcome. Built on a coarse-graining process called diffusion condensation, Multiscale PHATE learns a data topology that can be analyzed at coarse resolutions for high-level summarizations of data and at fine resolutions for detailed representations of subsets. We apply Multiscale PHATE to a coronavirus disease 2019 (COVID-19) dataset with 54 million cells from 168 hospitalized patients and find that patients who die show CD16hiCD66blo neutrophil and IFN-γ+ granzyme B+ Th17 cell responses. We also show that population groupings from Multiscale PHATE directly fed into a classifier predict disease outcome more accurately than naive featurizations of the data. Multiscale PHATE is broadly generalizable to different data types, including flow cytometry, single-cell RNA sequencing (scRNA-seq), single-cell sequencing assay for transposase-accessible chromatin (scATAC-seq), and clinical variables.

Manifold Interpolating Optimal-Transport Flows for Trajectory Inference.

We present a method called Manifold Interpolating Optimal-Transport Flow (MIOFlow) that learns stochastic, continuous population dynamics from static snapshot samples taken at sporadic timepoints. MIOFlow combines dynamic models, manifold learning, and optimal transport by training neural ordinary differential equations (Neural ODE) to interpolate between static population snapshots as penalized by optimal transport with manifold ground distance. Further, we ensure that the flow follows the geometry by operating in the latent space of an autoencoder that we call a geodesic autoencoder (GAE). In GAE the latent space distance between points is regularized to match a novel multiscale geodesic distance on the data manifold that we define. We show that this method is superior to normalizing flows, Schrödinger bridges and other generative models that are designed to flow from noise to data in terms of interpolating between populations. Theoretically, we link these trajectories with dynamic optimal transport. We evaluate our method on simulated data with bifurcations and merges, as well as scRNA-seq data from embryoid body differentiation, and acute myeloid leukemia treatment.

EMBEDDING SIGNALS ON GRAPHS WITH UNBALANCED DIFFUSION EARTH MOVER'S DISTANCE.

In modern relational machine learning it is common to encounter large graphs that arise via interactions or similarities between observations in many domains. Further, in many cases the target entities for analysis are actually signals on such graphs. We propose to compare and organize such datasets of graph signals by using an earth mover's distance (EMD) with a geodesic cost over the underlying graph. Typically, EMD is computed by optimizing over the cost of transporting one probability distribution to another over an underlying metric space. However, this is inefficient when computing the EMD between many signals. Here, we propose an unbalanced graph EMD that efficiently embeds the unbalanced EMD on an underlying graph into an L 1 space, whose metric we call unbalanced diffusion earth mover's distance (UDEMD). Next, we show how this gives distances between graph signals that are robust to noise. Finally, we apply this to organizing patients based on clinical notes, embedding cells modeled as signals on a gene graph, and organizing genes modeled as signals over a large cell graph. In each case, we show that UDEMD-based embeddings find accurate distances that are highly efficient compared to other methods.

A DNA packaging motor inchworms along one strand allowing it to adapt to alternative double-helical structures.

Ring ATPases that translocate disordered polymers possess lock-washer architectures that they impose on their substrates during transport via a hand-over-hand mechanism. Here, we investigate the operation of ring motors that transport ordered, helical substrates, such as the bacteriophage ϕ29 dsDNA packaging motor. This pentameric motor alternates between an ATP loading dwell and a hydrolysis burst wherein it packages one turn of DNA in four steps. When challenged with DNA-RNA hybrids and dsRNA, the motor matches its burst to the shorter helical pitches, keeping three power strokes invariant while shortening the fourth. Intermittently, the motor loses grip on the RNA-containing substrates, indicating that it makes optimal load-bearing contacts with dsDNA. To rationalize these observations, we propose a helical inchworm translocation mechanism in which, during each cycle, the motor increasingly adopts a lock-washer structure during the ATP loading dwell and successively regains its planar form with each power stroke during the burst.

Diffusion Earth Mover's Distance and Distribution Embeddings.

We propose a new fast method of measuring distances between large numbers of related high dimensional datasets called the Diffusion Earth Mover's Distance (EMD). We model the datasets as distributions supported on common data graph that is derived from the affinity matrix computed on the combined data. In such cases where the graph is a discretization of an underlying Riemannian closed manifold, we prove that Diffusion EMD is topologically equivalent to the standard EMD with a geodesic ground distance. Diffusion EMD can be computed in $\tilde{O}(n)$ time and is more accurate than similarly fast algorithms such as tree-based EMDs. We also show Diffusion EMD is fully differentiable, making it amenable to future uses in gradient-descent frameworks such as deep neural networks. Finally, we demonstrate an application of Diffusion EMD to single cell data collected from 210 COVID-19 patient samples at Yale New Haven Hospital. Here, Diffusion EMD can derive distances between patients on the manifold of cells at least two orders of magnitude faster than equally accurate methods. This distance matrix between patients can be embedded into a higher level patient manifold which uncovers structure and heterogeneity in patients. More generally, Diffusion EMD is applicable to all datasets that are massively collected in parallel in many medical and biological systems.

Quantifying the effect of experimental perturbations at single-cell resolution.

Current methods for comparing single-cell RNA sequencing datasets collected in multiple conditions focus on discrete regions of the transcriptional state space, such as clusters of cells. Here we quantify the effects of perturbations at the single-cell level using a continuous measure of the effect of a perturbation across the transcriptomic space. We describe this space as a manifold and develop a relative likelihood estimate of observing each cell in each of the experimental conditions using graph signal processing. This likelihood estimate can be used to identify cell populations specifically affected by a perturbation. We also develop vertex frequency clustering to extract populations of affected cells at the level of granularity that matches the perturbation response. The accuracy of our algorithm at identifying clusters of cells that are enriched or depleted in each condition is, on average, 57% higher than the next-best-performing algorithm tested. Gene signatures derived from these clusters are more accurate than those of six alternative algorithms in ground truth comparisons.

Helical inchworming: a novel translocation mechanism for a ring ATPase.

Ring ATPases perform a variety of tasks in the cell. Their function involves complex communication and coordination among the often identical subunits. Translocases in this group are of particular interest as they involve both chemical and mechanical actions in their operation. We study the DNA packaging motor of bacteriophage φ29, and using single-molecule optical tweezers and single-particle cryo-electron microscopy, have discovered a novel translocation mechanism for a molecular motor.

MULTIMODAL DATA VISUALIZATION AND DENOISING WITH INTEGRATED DIFFUSION.

We propose a method called integrated diffusion for combining multimodal data, gathered via different sensors on the same system, to create a integrated data diffusion operator. As real world data suffers from both local and global noise, we introduce mechanisms to optimally calculate a diffusion operator that reflects the combined information in data by maintaining low frequency eigenvectors of each modality both globally and locally. We show the utility of this integrated operator in denoising and visualizing multimodal toy data as well as multi-omic data generated from blood cells, measuring both gene expression and chromatin accessibility. Our approach better visualizes the geometry of the integrated data and captures known cross-modality associations. More generally, integrated diffusion is broadly applicable to multimodal datasets generated by noisy sensors collected in a variety of fields.