NeurostimML: a machine learning model for predicting neurostimulation-induced tissue damage.

Objective. The safe delivery of electrical current to neural tissue depends on many factors, yet previous methods for predicting tissue damage rely on only a few stimulation parameters. Here, we report the development of a machine learning approach that could lead to a more reliable method for predicting electrical stimulation-induced tissue damage by incorporating additional stimulation parameters.Approach. A literature search was conducted to build an initial database of tissue response information after electrical stimulation, categorized as either damaging or non-damaging. Subsequently, we used ordinal encoding and random forest for feature selection, and investigated four machine learning models for classification: Logistic Regression, K-nearest Neighbor, Random Forest, and Multilayer Perceptron. Finally, we compared the results of these models against the accuracy of the Shannon equation.Main Results. We compiled a database with 387 unique stimulation parameter combinations collected from 58 independent studies conducted over a period of 47 years, with 195 (51%) categorized as non-damaging and 190 (49%) categorized as damaging. The features selected for building our model with a Random Forest algorithm were: waveform shape, geometric surface area, pulse width, frequency, pulse amplitude, charge per phase, charge density, current density, duty cycle, daily stimulation duration, daily number of pulses delivered, and daily accumulated charge. The Shannon equation yielded an accuracy of 63.9% using akvalue of 1.79. In contrast, the Random Forest algorithm was able to robustly predict whether a set of stimulation parameters was classified as damaging or non-damaging with an accuracy of 88.3%.Significance. This novel Random Forest model can facilitate more informed decision making in the selection of neuromodulation parameters for both research studies and clinical practice. This study represents the first approach to use machine learning in the prediction of stimulation-induced neural tissue damage, and lays the groundwork for neurostimulation driven by machine learning models.

What can we learn when fitting a simple telegraph model to a complex gene expression model?

In experiments, the distributions of mRNA or protein numbers in single cells are often fitted to the random telegraph model which includes synthesis and decay of mRNA or protein, and switching of the gene between active and inactive states. While commonly used, this model does not describe how fluctuations are influenced by crucial biological mechanisms such as feedback regulation, non-exponential gene inactivation durations, and multiple gene activation pathways. Here we investigate the dynamical properties of four relatively complex gene expression models by fitting their steady-state mRNA or protein number distributions to the simple telegraph model. We show that despite the underlying complex biological mechanisms, the telegraph model with three effective parameters can accurately capture the steady-state gene product distributions, as well as the conditional distributions in the active gene state, of the complex models. Some effective parameters are reliable and can reflect realistic dynamic behaviors of the complex models, while others may deviate significantly from their real values in the complex models. The effective parameters can also be applied to characterize the capability for a complex model to exhibit multimodality. Using additional information such as single-cell data at multiple time points, we provide an effective method of distinguishing the complex models from the telegraph model. Furthermore, using measurements under varying experimental conditions, we show that fitting the mRNA or protein number distributions to the telegraph model may even reveal the underlying gene regulation mechanisms of the complex models. The effectiveness of these methods is confirmed by analysis of single-cell data for E. coli and mammalian cells. All these results are robust with respect to cooperative transcriptional regulation and extrinsic noise. In particular, we find that faster relaxation speed to the steady state results in more precise parameter inference under large extrinsic noise.

STAT6-targeting antisense oligonucleotides against solitary fibrous tumor.

Solitary fibrous tumor (SFT) is a rare, non-hereditary soft tissue sarcoma thought to originate from fibroblastic mesenchymal stem cells. The etiology of SFT is thought to be due to an environmental intrachromosomal gene fusion between NGFI-A-binding protein 2 (NAB2) and signal transducer and activator protein 6 (STAT6) genes on chromosome 12, wherein the activation domain of STAT6 is fused with the DNA-binding domain of NAB2 resulting in the oncogenesis of SFT. All NAB2-STAT6 fusion variations discovered in SFTs contain the C-terminal of STAT6 transcript, and thus can serve as target site for antisense oligonucleotides (ASOs)-based therapies. Indeed, our in vitro studies show the STAT6 3' untranslated region (UTR)-targeting ASO (ASO 993523) was able to reduce expression of NAB2-STAT6 fusion transcripts in multiple SFT cell models with high efficiency (half-maximal inhibitory concentration: 116-300 nM). Encouragingly, in vivo treatment of SFT patient-derived xenograft mouse models with ASO 993523 resulted in acceptable tolerability profiles, reduced expression of NAB2-STAT6 fusion transcripts in xenograft tissues (21.9%), and, importantly, reduced tumor growth (32.4% decrease in tumor volume compared with the untreated control). Taken together, our study established ASO 993523 as a potential agent for the treatment of SFTs.

NeurostimML: A machine learning model for predicting neurostimulation-induced tissue damage.

Objective: The safe delivery of electrical current to neural tissue depends on many factors, yet previous methods for predicting tissue damage rely on only a few stimulation parameters. Here, we report the development of a machine learning approach that could lead to a more reliable method for predicting electrical stimulation-induced tissue damage by incorporating additional stimulation parameters. Approach: A literature search was conducted to build an initial database of tissue response information after electrical stimulation, categorized as either damaging or non-damaging. Subsequently, we used ordinal encoding and random forest for feature selection, and investigated four machine learning models for classification: Logistic Regression, K-nearest Neighbor, Random Forest, and Multilayer Perceptron. Finally, we compared the results of these models against the accuracy of the Shannon equation. Main Results: We compiled a database with 387 unique stimulation parameter combinations collected from 58 independent studies conducted over a period of 47 years, with 195 (51%) categorized as non-damaging and 190 (49%) categorized as damaging. The features selected for building our model with a Random Forest algorithm were: waveform shape, geometric surface area, pulse width, frequency, pulse amplitude, charge per phase, charge density, current density, duty cycle, daily stimulation duration, daily number of pulses delivered, and daily accumulated charge. The Shannon equation yielded an accuracy of 63.9% using a k value of 1.79. In contrast, the Random Forest algorithm was able to robustly predict whether a set of stimulation parameters was classified as damaging or non-damaging with an accuracy of 88.3%. Significance: This novel Random Forest model can facilitate more informed decision making in the selection of neuromodulation parameters for both research studies and clinical practice. This study represents the first approach to use machine learning in the prediction of stimulation-induced neural tissue damage, and lays the groundwork for neurostimulation driven by machine learning models.

RNA "COURIERs": Enabling synthetic cell-to-cell communication in human cells.

The development of molecular couriers to selectively package, export, and recover RNA molecules within human cells is a significant challenge. In this issue of Cell, Horns et al.1 introduce cellular RNA exporters, termed COURIERs, that package, secrete, and protect RNA cargo and establish the foundation for sophisticated cell-to-cell RNA communication.

Impact of variability in cell cycle periodicity on cell population dynamics.

The cell cycle consists of a series of orchestrated events controlled by molecular sensing and feedback networks that ultimately drive the duplication of total DNA and the subsequent division of a single parent cell into two daughter cells. The ability to block the cell cycle and synchronize cells within the same phase has helped understand factors that control cell cycle progression and the properties of each individual phase. Intriguingly, when cells are released from a synchronized state, they do not maintain synchronized cell division and rapidly become asynchronous. The rate and factors that control cellular desynchronization remain largely unknown. In this study, using a combination of experiments and simulations, we investigate the desynchronization properties in cervical cancer cells (HeLa) starting from the G1/S boundary following double-thymidine block. Propidium iodide (PI) DNA staining was used to perform flow cytometry cell cycle analysis at regular 8 hour intervals, and a custom auto-similarity function to assess the desynchronization and quantify the convergence to an asynchronous state. In parallel, we developed a single-cell phenomenological model the returns the DNA amount across the cell cycle stages and fitted the parameters using experimental data. Simulations of population of cells reveal that the cell cycle desynchronization rate is primarily sensitive to the variability of cell cycle duration within a population. To validate the model prediction, we introduced lipopolysaccharide (LPS) to increase cell cycle noise. Indeed, we observed an increase in cell cycle variability under LPS stimulation in HeLa cells, accompanied with an enhanced rate of cell cycle desynchronization. Our results show that the desynchronization rate of artificially synchronized in-phase cell populations can be used a proxy of the degree of variance in cell cycle periodicity, an underexplored axis in cell cycle research.

Reduction of Tumor Growth with RNA-Targeting Treatment of the NAB2-STAT6 Fusion Transcript in Solitary Fibrous Tumor Models.

Solitary fibrous tumor (SFT) is a rare soft-tissue sarcoma. This nonhereditary cancer is the result of an environmental intrachromosomal gene fusion between NAB2 and STAT6 on chromosome 12, which fuses the activation domain of STAT6 with the repression domain of NAB2. Currently there is not an approved chemotherapy regimen for SFTs. The best response on available pharmaceuticals is a partial response or stable disease for several months. The purpose of this study is to investigate the potential of RNA-based therapies for the treatment of SFTs. Specifically, in vitro SFT cell models were engineered to harbor the characteristic NAB2-STAT6 fusion using the CRISPR/SpCas9 system. Cell migration as well as multiple cancer-related signaling pathways were increased in the engineered cells as compared to the fusion-absent parent cells. The SFT cell models were then used for evaluating the targeting efficacies of NAB2-STAT6 fusion-specific antisense oligonucleotides (ASOs) and CRISPR/CasRx systems. Our results showed that fusion specific ASO treatments caused a 58% reduction in expression of fusion transcripts and a 22% reduction in cell proliferation after 72 h in vitro. Similarly, the AAV2-mediated CRISPR/CasRx system led to a 59% reduction in fusion transcript expressions in vitro, and a 55% reduction in xenograft growth after 29 days ex vivo.

Non-viral nitric oxide-based gene therapy improves perfusion and liposomal doxorubicin sonopermeation in neuroblastoma models.

Neuroblastoma (NB) is a pediatric malignancy that accounts for 15% of cancer-related childhood mortality. High-risk NB requires an aggressive chemoradiotherapy regimen that causes significant off-target toxicity. Despite this invasive treatment, many patients either relapse or do not respond adequately. Recent studies suggest that improving tumor perfusion can enhance drug accumulation and distribution within the tumor tissue, potentially augmenting treatment effects without inflicting systemic toxicity. Accordingly, methods that transiently increase tumor perfusion prior to treatment may help combat this disease. Here, we show the use of gene therapy to confer inducible nitric oxide synthase (iNOS) expression solely in the tumor space, using focused ultrasound targeting. NOS catalyzes the reaction that generates nitric oxide (NO), a potent endogenous vasodilator. This study reports the development of a targeted non-viral image-guided platform to deliver iNOS-expressing plasmid DNA (pDNA) to vascular endothelial cells encasing tumor blood vessels. Following transfection, longitudinal quantitative contrast-enhanced ultrasound (qCEUS) imaging revealed an increase in tumor perfusion over 72 h, attributed to elevated intratumoral iNOS expression. Methods: To construct a gene delivery vector, cationic ultrasound-responsive agents (known as "microbubbles") were employed to carry pDNA in circulation and transfect tumor vascular endothelial cells in vivo using focused ultrasound (FUS) energy. This was followed by liposomal doxorubicin (L-DOX) treatment. The post-transfection tumor response was monitored longitudinally using qCEUS imaging to determine relative changes in blood volumes and perfusion rates. After therapy, ex vivo analysis of tumors was performed to examine the bioeffects associated with iNOS expression. Results: By combining FUS therapy with cationic ultrasound contrast agents (UCAs), we achieved selective intratumoral transfection of pDNA encoding the iNOS enzyme. While transitory, the degree of expression was sufficient to induce significant increases in tumoral perfusion, to appreciably enhance the chemotherapeutic payload and to extend survival time in an orthotopic xenograft model. Conclusion: We have demonstrated the ability of a novel targeted non-viral gene therapy strategy to enhance tumor perfusion and improve L-DOX delivery to NB xenografts. While our results demonstrate that transiently increasing tumor perfusion improves liposome-encapsulated chemotherapeutic uptake and distribution, we expect that our iNOS gene delivery paradigm can also significantly improve radio and immunotherapies by increasing the delivery of radiosensitizers and immunomodulators, potentially improving upon current NB treatment without concomitant adverse effects. Our findings further suggest that qCEUS imaging can effectively monitor changes in tumor perfusion in vivo, allowing the identification of an ideal time-point to administer therapy.

Clinical gene therapy development for the central nervous system: Candidates and challenges for AAVs.

Many diseases affecting the central nervous system (CNS) are deadly but less understood, leading to impaired mental and motor capabilities and poor patient prospects. Gene therapy is a promising therapeutic modality for correcting many genetic disorders, expanding in breadth and scope with further advances. This review summarizes the candidate CNS disorders for gene therapy, mechanisms of gene therapy, and recent clinical advances and limitations of gene therapy in CNS disorders. We highlight that improving delivery across CNS barriers, safety, monitoring techniques, and multiplexing therapies are predominant factors in advancing long-term outcomes from gene therapy.

Genetic physical unclonable functions in human cells.

A physical unclonable function (PUF) is a physical entity that provides a measurable output that can be used as a unique and irreproducible identifier for the artifact wherein it is embedded. Popularized by the electronics industry, silicon PUFs leverage the inherent physical variations of semiconductor manufacturing to establish intrinsic security primitives for attesting integrated circuits. Owing to the stochastic nature of these variations, photolithographically manufactured silicon PUFs are impossible to reproduce (thus unclonable). Inspired by the success of silicon PUFs, we sought to create the first generation of genetic PUFs in human cells. We demonstrate that these PUFs are robust (i.e., they repeatedly produce the same output), unique (i.e., they do not coincide with any other identically produced PUF), and unclonable (i.e., they are virtually impossible to replicate). Furthermore, we demonstrate that CRISPR-engineered PUFs (CRISPR-PUFs) can serve as a foundational principle for establishing provenance attestation protocols.

Plasmonic LAMP: Improving the Detection Specificity and Sensitivity for SARS-CoV-2 by Plasmonic Sensing of Isothermally Amplified Nucleic Acids.

The ability to detect pathogens specifically and sensitively is critical to combat infectious diseases outbreaks and pandemics. Colorimetric assays involving loop-mediated isothermal amplification (LAMP) provide simple readouts yet suffer from the intrinsic non-template amplification. Herein, a highly specific and sensitive assay relying on plasmonic sensing of LAMP amplicons via DNA hybridization, termed as plasmonic LAMP, is developed for the severe acute respiratory syndrome-related coronavirus 2 (SARS-CoV-2) RNA detection. This work has two important advances. First, gold and silver (Au-Ag) alloy nanoshells are developed as plasmonic sensors that have 4-times stronger extinction in the visible wavelengths and give a 20-times lower detection limit for oligonucleotides over Au counterparts. Second, the integrated method allows cutting the complex LAMP amplicons into short repeats that are amendable for hybridization with oligonucleotide-functionalized Au-Ag nanoshells. In the SARS-CoV-2 RNA detection, plasmonic LAMP takes ≈75 min assay time, achieves a detection limit of 10 copies per reaction, and eliminates the contamination from non-template amplification. It also shows better detection specificity and sensitivity over commercially available LAMP kits due to the additional sequence identification. This work opens a new route for LAMP amplicon detection and provides a method for virus testing at its early representation.

Machine learning-based approaches for identifying human blood cells harboring CRISPR-mediated fetal chromatin domain ablations.

Two common hemoglobinopathies, sickle cell disease (SCD) and ß-thalassemia, arise from genetic mutations within the ß-globin gene. In this work, we identified a 500-bp motif (Fetal Chromatin Domain, FCD) upstream of human ϒ-globin locus and showed that the removal of this motif using CRISPR technology reactivates the expression of ϒ-globin. Next, we present two different cell morphology-based machine learning approaches that can be used identify human blood cells (KU-812) that harbor CRISPR-mediated FCD genetic modifications. Three candidate models from the first approach, which uses multilayer perceptron algorithm (MLP 20-26, MLP26-18, and MLP 30-26) and flow cytometry-derived cellular data, yielded 0.83 precision, 0.80 recall, 0.82 accuracy, and 0.90 area under the ROC (receiver operating characteristic) curve when predicting the edited cells. In comparison, the candidate model from the second approach, which uses deep learning (T2D5) and DIC microscopy-derived imaging data, performed with less accuracy (0.80) and ROC AUC (0.87). We envision that equivalent machine learning-based models can complement currently available genotyping protocols for specific genetic modifications which result in morphological changes in human cells.

Transcriptomics and solid tumors: The next frontier in precision cancer medicine.

Transcriptomics, which encompasses assessments of alternative splicing and alternative polyadenylation, identification of fusion transcripts, explorations of noncoding RNAs, transcript annotation, and discovery of novel transcripts, is a valuable tool for understanding cancer mechanisms and identifying biomarkers. Recent advances in high-throughput technologies have enabled large-scale gene expression profiling. Importantly, RNA expression profiling of tumor tissue has been successfully used to determine clinically actionable molecular alterations. The WINTHER precision medicine clinical trial was the first prospective trial in diverse solid malignancies that assessed both genomics and transcriptomics to match treatments to specific molecular alterations. The use of transcriptome analysis in WINTHER and other trials increased the number of targetable -omic changes compared to genomic profiling alone. Other applications of transcriptomics involve the evaluation of tumor and circulating noncoding RNAs as predictive and prognostic biomarkers, the improvement of risk stratification by the use of prognostic and predictive multigene assays, the identification of fusion transcripts that drive tumors, and an improved understanding of the impact of DNA changes as some genomic alterations are silenced at the RNA level. Finally, RNA sequencing and gene expression analysis have been incorporated into clinical trials to identify markers predicting response to immunotherapy. Many issues regarding the complexity of the analysis, its reproducibility and variability, and the interpretation of the results still need to be addressed. The integration of transcriptomics with genomics, proteomics, epigenetics, and tumor immune profiling will improve biomarker discovery and our understanding of disease mechanisms and, thereby, accelerate the implementation of precision oncology.

Single-Molecule Detection of SARS-CoV-2 by Plasmonic Sensing of Isothermally Amplified Nucleic Acids.

Single-molecule detection of pathogens such as SARS-CoV-2 is key to combat infectious diseases outbreak and pandemic. Currently colorimetric sensing with loop-mediated isothermal amplification (LAMP) provides simple readouts but suffers from intrinsic non-template amplification. Herein, we report that plasmonic sensing of LAMP amplicons via DNA hybridization allows highly specific and single-molecule detection of SARS-CoV-2 RNA. Our work has two important advances. First, we develop gold and silver alloy (Au-Ag) nanoshells as plasmonic sensors that have 4-times stronger extinction in the visible wavelengths and give 20-times lower detection limit for oligonucleotides than Au nanoparticles. Second, we demonstrate that the diagnostic method allows cutting the complex LAMP amplicons into short repeats that are amendable for hybridization with oligonucleotide-functionalized nanoshells. This additional sequence identification eliminates the contamination from non-template amplification. The detection method is a simple and single-molecule diagnostic platform for virus testing at its early representation.

Cell morphology-based machine learning models for human cell state classification.

Herein, we implement and access machine learning architectures to ascertain models that differentiate healthy from apoptotic cells using exclusively forward (FSC) and side (SSC) scatter flow cytometry information. To generate training data, colorectal cancer HCT116 cells were subjected to miR-34a treatment and then classified using a conventional Annexin V/propidium iodide (PI)-staining assay. The apoptotic cells were defined as Annexin V-positive cells, which include early and late apoptotic cells, necrotic cells, as well as other dying or dead cells. In addition to fluorescent signal, we collected cell size and granularity information from the FSC and SSC parameters. Both parameters are subdivided into area, height, and width, thus providing a total of six numerical features that informed and trained our models. A collection of logistical regression, random forest, k-nearest neighbor, multilayer perceptron, and support vector machine was trained and tested for classification performance in predicting cell states using only the six aforementioned numerical features. Out of 1046 candidate models, a multilayer perceptron was chosen with 0.91 live precision, 0.93 live recall, 0.92 live f value and 0.97 live area under the ROC curve when applied on standardized data. We discuss and highlight differences in classifier performance and compare the results to the standard practice of forward and side scatter gating, typically performed to select cells based on size and/or complexity. We demonstrate that our model, a ready-to-use module for any flow cytometry-based analysis, can provide automated, reliable, and stain-free classification of healthy and apoptotic cells using exclusively size and granularity information.

Robust Filtering and Noise Suppression in Intragenic miRNA-Mediated Host Regulation.

MicroRNAs (miRNAs) are short non-coding RNA molecules that regulate gene expression post-transcriptionally by binding to target messenger RNAs (mRNAs). Many human miRNAs are intragenic, located within introns of protein-coding sequence (host). Intriguingly, a percentage of intragenic miRNAs downregulate the host transcript forming an incoherent feedforward motif topology. Here, we study intragenic miRNA-mediated host gene regulation using a synthetic gene circuit stably integrated within a safe-harbor locus of human cells. When the intragenic miRNA is directed to inhibit the host transcript, we observe a reduction in reporter expression accompanied by output filtering and noise reduction. Specifically, the system operates as a filter with respect to promoter strength, with the threshold being robust to promoter strength and measurement time. Additionally, the intragenic miRNA regulation reduces expression noise compared to splicing-alone architecture. Our results provide a new insight into miRNA-mediated gene expression, with direct implications to gene therapy and synthetic biology applications.

Uncoupling gene expression noise along the central dogma using genome engineered human cell lines.

Eukaryotic protein synthesis is an inherently stochastic process. This stochasticity stems not only from variations in cell content between cells but also from thermodynamic fluctuations in a single cell. Ultimately, these inherently stochastic processes manifest as noise in gene expression, where even genetically identical cells in the same environment exhibit variation in their protein abundances. In order to elucidate the underlying sources that contribute to gene expression noise, we quantify the contribution of each step within the process of protein synthesis along the central dogma. We uncouple gene expression at the transcriptional, translational, and post-translational level using custom engineered circuits stably integrated in human cells using CRISPR. We provide a generalized framework to approximate intrinsic and extrinsic noise in a population of cells expressing an unbalanced two-reporter system. Our decomposition shows that the majority of intrinsic fluctuations stem from transcription and that coupling the two genes along the central dogma forces the fluctuations to propagate and accumulate along the same path, resulting in increased observed global correlation between the products.

Mechanistic insights into host adaptation, virulence and epidemiology of the phytopathogen Xanthomonas.

Xanthomonas is a well-studied genus of bacterial plant pathogens whose members cause a variety of diseases in economically important crops worldwide. Genomic and functional studies of these phytopathogens have provided significant understanding of microbial-host interactions, bacterial virulence and host adaptation mechanisms including microbial ecology and epidemiology. In addition, several strains of Xanthomonas are important as producers of the extracellular polysaccharide, xanthan, used in the food and pharmaceutical industries. This polymer has also been implicated in several phases of the bacterial disease cycle. In this review, we summarise the current knowledge on the infection strategies and regulatory networks controlling virulence and adaptation mechanisms from Xanthomonas species and discuss the novel opportunities that this body of work has provided for disease control and plant health.