Nicholas Panchy’s research while affiliated with The University of Tennessee Medical Center at Knoxville and other places

Simple Summary Small cell lung cancer (SCLC) is an aggressive cancer that is difficult to treat. There are at least five subtypes of SCLC cells, defined by gene expression signatures. The transitions between these subtypes and cooperation between them contribute to the progression of SCLC. Particularly, transitions between neuroendocrine (NE) states, including A, A2, and N subtypes, and the non-NE states, including P and Y subtypes, are hallmarks of SCLC plasticity. In this study, the relationship between SCLC subtypes and epithelial to mesenchymal transition (EMT) was analyzed. EMT is a well-known form of cellular plasticity that contributes to cancer invasiveness and resistance. The results showed that the SCLC-A2 subtype is epithelial while SCLC-A and SCLC-N are mesenchymal but distinct from the non-NE mesenchymal states. This study provides a basis for understanding the gene regulatory mechanisms of SCLC tumor plasticity and its applicability to other cancer types. Abstract Small cell lung cancer (SCLC) is an aggressive cancer recalcitrant to treatment, arising predominantly from epithelial pulmonary neuroendocrine (NE) cells. Intratumor heterogeneity plays critical roles in SCLC disease progression, metastasis, and treatment resistance. At least five transcriptional SCLC NE and non-NE cell subtypes were recently defined by gene expression signatures. Transition from NE to non-NE cell states and cooperation between subtypes within a tumor likely contribute to SCLC progression by mechanisms of adaptation to perturbations. Therefore, gene regulatory programs distinguishing SCLC subtypes or promoting transitions are of great interest. Here, we systematically analyze the relationship between SCLC NE/non-NE transition and epithelial to mesenchymal transition (EMT)—a well-studied cellular process contributing to cancer invasiveness and resistance—using multiple transcriptome datasets from SCLC mouse tumor models, human cancer cell lines, and tumor samples. The NE SCLC-A2 subtype maps to the epithelial state. In contrast, SCLC-A and SCLC-N (NE) map to a partial mesenchymal state (M1) that is distinct from the non-NE, partial mesenchymal state (M2). The correspondence between SCLC subtypes and the EMT program paves the way for further work to understand gene regulatory mechanisms of SCLC tumor plasticity with applicability to other cancer types.

Colorectal cancer has proven to be difficult to treat as it is the second leading cause of cancer death for both men and women worldwide. Recent work has shown the importance of microRNA (miRNA) in the progression and metastasis of colorectal cancer. Here, we develop a metric based on miRNA-gene target interactions, previously validated to be associated with colorectal cancer. We use this metric with a regularized Cox model to produce a small set of top-performing genes related to colon cancer. We show that using the miRNA metric and a Cox model led to a meaningful improvement in colon cancer survival prediction and correct patient risk stratification. We show that our approach outperforms existing methods and that the top genes identified by our process are implicated in NOTCH3 signaling and general metabolism pathways, which are essential to colon cancer progression.

Epithelial–mesenchymal transition (EMT) is a cellular process involved in development and disease progression. Intermediate EMT states were observed in tumors and fibrotic tissues, but previous in vitro studies focused on time-dependent responses with single doses of signals; it was unclear whether single-cell transcriptomes support stable intermediates observed in diseases. Here, we performed single-cell RNA-sequencing with human mammary epithelial cells treated with multiple doses of TGF-β. We found that dose-dependent EMT harbors multiple intermediate states at nearly steady state. Comparisons of dose- and time-dependent EMT transcriptomes revealed that the dose-dependent data enable higher sensitivity to detect genes associated with EMT. We identified cell clusters unique to time-dependent EMT, reflecting cells en route to stable states. Combining dose- and time-dependent cell clusters gave rise to accurate prognosis for cancer patients. Our transcriptomic data and analyses uncover a stable EMT continuum at the single-cell resolution, and complementary information of two types of single-cell experiments.

Small cell lung cancer (SCLC) is an aggressive cancer recalcitrant to treatment. SCLC arises in lung epithelium, and the phenotypic heterogeneity of SCLC cells plays critical roles in disease progression and drug resistance. Particularly, the transition from neuroendocrine (NE) cells to non-neuroendocrine (non-NE) cells, as well as the cooperations between these cell types, are observed in SCLC progression. At least five subtypes of SCLC tumors have recently been defined by signature gene expression and these subtypes have been extended to individual SCLC tumor cells. However, the molecular and cellular programs that distinguish the cell types or allow transitions among them are unclear. Here, we systematically analyzed the relationship between SCLC and epithelial to mesenchymal transition (EMT), a cellular process well known to contribute to invasiveness and drug resistance of cancer cells, using multiple transcriptome datasets from a mouse tumor model, human cancer cell lines, and human tumor samples. Our analysis of these datasets consistently showed that an extreme epithelial state corresponds to the SCLC-A2 subtype. Furthermore, two NE subtypes have significant expression of a subset of mesenchymal genes, with surprisingly limited overlap with the mesenchymal genes highly expressed in non-NE cells. Our work reveals a previously unknown connection between an EMT program and a specific SCLC subtype and an underappreciated divergence of EMT trajectories contributing to the SCLC phenotypic heterogeneity.

Epithelial-mesenchymal transition (EMT) is a key cellular process involved in development and disease progression. Single-cell transcriptomes can characterize intermediate EMT states observed in tumors and fibrotic tissues, but previous in vitro models focused on time-dependent responses after stimulation with single dose of EMT signals. It was therefore unclear whether single-cell transcriptomes support stable intermediate EMT phenotypes crucial for disease progression. We performed single-cell RNA-sequencing with human mammary epithelial cells treated with various concentrations of TGF-β. We found that the dose-dependent EMT harbors multiple intermediate states at the single-cell level after two weeks of treatment, suggesting a stable continuum. After correcting batch effects from experiments, we performed comparative analyses of the dose- and time-dependent EMT. We found that the dose-dependent EMT shows a stronger anti-correlation between epithelial and mesenchymal transcriptional programs and a better resolution of transition stages compared to the time-dependent process. These properties enable higher sensitivity to detect genes whose expressions are associated with core EMT regulatory networks. Nonetheless, we found cell clusters unique to the time-dependent EMT, which correspond to en route cell populations that do not appear at steady states. Furthermore, combining dose- and time-dependent cell clusters gave rise to more accurate prognosis for cancer patients compared to individual EMT spectrum. Our new data and analyses reveal a stable EMT continuum at the single-cell resolution and the transcriptomic level. The dose-dependent experimental model can complement the widely used time-course experiments to reflect physiologically or pathologically relevant EMT phenotypes in a comprehensive manner.

Revealing the contributions of genes to plant phenotype is frequently challenging because loss‐of‐function effects may be subtle or masked by varying degrees of genetic redundancy. Such effects can potentially be detected by measuring plant fitness, which reflects the cumulative effects of genetic changes over the lifetime of a plant. However, fitness is challenging to measure accurately, particularly in species with high fecundity and relatively small propagule sizes such as Arabidopsis thaliana. An image segmentation‐based method using the software ImageJ and an object detection‐based method using the Faster Region‐based Convolutional Neural Network (R‐CNN) algorithm were used for measuring two Arabidopsis fitness traits: seed and fruit counts. The segmentation‐based method was error‐prone (correlation between true and predicted seed counts, r² = 0.849) because seeds touching each other were undercounted. By contrast, the object detection‐based algorithm yielded near perfect seed counts (r² = 0.9996) and highly accurate fruit counts (r² = 0.980). Comparing seed counts for wild‐type and 12 mutant lines revealed fitness effects for three genes; fruit counts revealed the same effects for two genes. Our study provides analysis pipelines and models to facilitate the investigation of Arabidopsis fitness traits and demonstrates the importance of examining fitness traits when studying gene functions.

Revealing the contributions of genes to plant phenotype is frequently challenging because the effects of loss of gene function may be subtle or be masked by genetic redundancy. Such effects can potentially be detected by measuring plant fitness, which reflects the cumulative effects of genetic changes over the lifetime of a plant. However, fitness is challenging to measure accurately, particularly in species with high fecundity and relatively small propagule sizes such as Arabidopsis thaliana . An image segmentation-based (ImageJ) and a Faster Region Based Convolutional Neural Network (R-CNN) approach were used for measuring two Arabidopsis fitness traits: seed and fruit counts. Although straightforward to use, ImageJ was error-prone (correlation between true and predicted seed counts, r ² =0.849) because seeds touching each other were undercounted. In contrast, Faster R-CNN yielded near perfect seed counts (r ² =0.9996) and highly accurate fruit counts (r ² =0.980). By examining seed counts, we were able to reveal fitness effects for genes that were previously reported to have no or condition-specific loss-of-function phenotypes. Our study provides models to facilitate the investigation of Arabidopsis fitness traits and demonstrates the importance of examining fitness traits in the study of gene functions.

Large-scale transcriptome data, such as single-cell RNA-sequencing data, have provided unprecedented resources for studying biological processes at the systems level. Numerous dimensionality reduction methods have been developed to visualize and analyze these transcriptome data. In addition, several existing methods allow inference of functional variations among samples using gene sets with known biological functions. However, it remains challenging to analyze transcriptomes with reduced dimensions that are interpretable in terms of dimensions’ directionalities, transferrable to new data, and directly expose the contribution or association of individual genes. In this study, we used gene set non-negative principal component analysis (gsPCA) and non-negative matrix factorization (gsNMF) to analyze large-scale transcriptome datasets. We found that these methods provide low-dimensional information about the progression of biological processes in a quantitative manner, and their performances are comparable to existing functional variation analysis methods in terms of distinguishing multiple cell states and samples from multiple conditions. Remarkably, upon training with a subset of data, these methods allow predictions of locations in the functional space using data from experimental conditions that are not exposed to the models. Specifically, our models predicted the extent of progression and reversion for cells in the epithelial-mesenchymal transition (EMT) continuum. These methods revealed conserved EMT program among multiple types of single cells and tumor samples. Finally, we demonstrate this approach is broadly applicable to data and gene sets beyond EMT and provide several recommendations on the choice between the two linear methods and the optimal algorithmic parameters. Our methods show that simple constrained matrix decomposition can produce to low-dimensional information in functionally interpretable and transferrable space, and can be widely useful for analyzing large-scale transcriptome data.

Large-scale transcriptome data, such as single-cell RNA-sequencing data, have provided unprecedented resources for studying biological processes at the systems level. Numerous dimensionality reduction methods have been developed to visualize and analyze these transcriptome data. In addition, several existing methods allow inference of functional variations among samples using gene sets with known biological functions. However, it remains challenging to analyze transcriptomes with reduced dimensions that are both interpretable in terms of dimensions, directionalities and transferrable to new data. In this study, we used gene set non-negative principal component analysis (gsPCA) and non-negative matrix factorization (gsNMF) to analyze large-scale transcriptome datasets. We found that these methods provide low-dimensional information about the progression of biological processes in a quantitative manner, and their performances are comparable to existing functional variation analysis methods in terms of distinguishing multiple cell states and samples from multiple conditions. Remarkably, upon training with a subset of data, these methods allow predictions of locations in the functional space using data from experimental conditions that are not exposed to the models. Specifically, our models predicted the extent of progression and reversion for cells in the epithelial-mesenchymal transition (EMT) continuum. These methods revealed conserved EMT program among multiple types of single cells and tumor samples. Finally, we provide several recommendations on the choice between the two linear methods and the optimal algorithmic parameters. Our methods show that simple constrained matrix decomposition can produce to low-dimensional information in functionally interpretable and transferrable space, and can be widely useful for analyzing large-scale transcriptome data.

Light-entrained circadian clocks confer rhythmic dynamics of cellular and molecular activities to animals and plants. These intrinsic clocks allow stable anticipations to light-dark (diel) cycles. Many genes in the model plant Arabidopsis thaliana are regulated by diel cycles via pathways independent of the clock, suggesting that the integration of circadian and light signals is important for the fitness of plants. Previous studies of light-clock signal integrations have focused on moderate phase adjustment of the two signals. However, dynamical features of integrations across a broad range of phases remain elusive. Phosphorylation of ribosomal protein of the small subunit 6 (eS6), a ubiquitous post-translational modification across kingdoms, is influenced by the circadian clock and the light-dark (diel) cycle in an opposite manner. To understand this striking phenomenon and its underlying information processing capabilities, we built a mathematical model for the eS6 phosphorylation (eS6-P) control circuit. We found that the dynamics of eS6-P can be explained by a feedforward circuit with inputs from both circadian and diel cycles. Furthermore, the early day response of this circuit with dual rhythmic inputs is sensitive to the changes in daylength, including both transient and gradual changes observed in realistic light intervals across a year, because of weather and seasons. By analyzing published gene expression data, we found that the dynamics produced by the eS6-P control circuit can be observed in the expression profiles of a large number of genes. Our work provides mechanistic insights into the complex dynamics of a ribosomal protein, and it proposes a previously underappreciated function of the circadian clock, which not only prepares organisms for normal diel cycles but also helps to detect both transient and seasonal changes with a predictive power.