Using partial least squares regression to analyze cellular response data

Quantitative multiplexing of uric acid and creatinine using polydisperse plasmonic nanoparticles enabled by electrochemical-SERS and machine learning

Surface-enhanced Raman spectroscopy (SERS) is a promising technique for the detection of biomarkers, but it can struggle to quantify multiple analytes in complex fluids. This study combines electrochemical SERS (E-SERS) and machine learning for the quantitative multiplexed detection of uric acid (UA) and creatinine (CRN). Using classical polydisperse Ag nanoparticles (NPs) made by scalable synthesis, we achieved quantitative multiplexing with low limits of detection (LoDs) and high prediction accuracy, comparable to those made by sophisticated approaches. The E-SERS LoDs at the optimal applied potentials were 0.127 μM and 0.354 μM for UA and CRN respectively, compared to 0.504 μM and 1.02 μM for conventional SERS (recorded at 0 V). By collecting a multi-dimensional E-SERS dataset and applying a two-step partial least squares regression - multilayer perceptron (PLSR-MLP) machine learning algorithm, we were able to identify the analyte concentrations in unseen spectra with a prediction accuracy of 0.94. This research demonstrates the potential of E-SERS and machine learning for multiplexed detection in clinical settings.

The impact of breastfeeding on facial appearance in adolescent children

Evidence that breastfeeding impacts the facial features of children is conflicting. Most studies to date have focused on dental and skeletal malocclusion. It currently remains unclear whether such effects are of sufficient magnitude to be detectable on outward facial appearance. Here, we evaluate the extent to which maternally reported breastfeeding is associated with 3D facial shape in a large adolescent cohort. After extracting 3D facial surfaces from MR scans in 2275 9- and 10-year-old children and aligning the surfaces in dense correspondence, we analyzed the effect of breastfeeding on shape as a dichotomous (no/yes) and semi-quantitative (to assess duration in months) variable using partial least squares regression. Our results showed no effect (p = 0.532) when breastfeeding was dichotomized. However, when treated as a semi-quantitative variable, breastfeeding duration was associated with statistically significant changes in shape (p = 3.61x 10-4). The most prominent facial changes included relative retrusion of the central midface, zygomatic arches, and orbital regions along with relative protrusion of forehead, cheek, and mandible. The net effect was that as breastfeeding duration increased, the facial profile in children became flatter (less convex). The observed effects on the face, however, were subtle and likely not conspicuous enough to be noticed by most observers. This was true even when comparing the faces of children breastfed for 19-24 months to children with no reported breastfeeding. Thus, breastfeeding does appear to have detectable effect on outward facial appearance in adolescent children, but its practical impact appears to be minimal.

The structure is the message: Preserving experimental context through tensor decomposition

Recent biological studies have been revolutionized in scale and granularity by multiplex and high-throughput assays. Profiling cell responses across several experimental parameters, such as perturbations, time, and genetic contexts, leads to richer and more generalizable findings. However, these multidimensional datasets necessitate a reevaluation of the conventional methods for their representation and analysis. Traditionally, experimental parameters are merged to flatten the data into a two-dimensional matrix, sacrificing crucial experiment context reflected by the structure. As Marshall McLuhan famously stated, "the medium is the message." In this work, we propose that the experiment structure is the medium in which subsequent analysis is performed, and the optimal choice of data representation must reflect the experiment structure. We review how tensor-structured analyses and decompositions can preserve this information. We contend that tensor methods are poised to become integral to the biomedical data sciences toolkit.

Rational design of HER2-targeted combination therapies to reverse drug resistance in fibroblast-protected HER2+ breast cancer cells

Introduction

Fibroblasts, an abundant cell type in the breast tumor microenvironment, interact with cancer cells and orchestrate tumor progression and drug resistance. However, the mechanisms by which fibroblast-derived factors impact drug sensitivity remain poorly understood. Here, we develop rational combination therapies that are informed by proteomic profiling to overcome fibroblast-mediated therapeutic resistance in HER2+ breast cancer cells.

Methods

Drug sensitivity to the HER2 kinase inhibitor lapatinib was characterized under conditions of monoculture and exposure to breast fibroblast-conditioned medium. Protein expression was measured using reverse phase protein arrays. Candidate targets for combination therapy were identified using differential expression and multivariate regression modeling. Follow-up experiments were performed to evaluate the effects of HER2 kinase combination therapies in fibroblast-protected cancer cell lines and fibroblasts.

Results

Compared to monoculture, fibroblast-conditioned medium increased the expression of plasminogen activator inhibitor-1 (PAI1) and cell cycle regulator polo like kinase 1 (PLK1) in lapatinib-treated breast cancer cells. Combination therapy of lapatinib with inhibitors targeting either PAI1 or PLK1, eliminated fibroblast-protected cancer cells, under both conditions of direct coculture with fibroblasts and protection by fibroblast-conditioned medium. Analysis of publicly available, clinical transcriptomic datasets revealed that HER2-targeted therapy fails to suppress PLK1 expression in stroma-rich HER2+ breast tumors and that high PAI1 gene expression associates with high stroma density. Furthermore, we showed that an epigenetics-directed approach using a bromodomain and extraterminal inhibitor to globally target fibroblast-induced proteomic adaptions in cancer cells, also restored lapatinib sensitivity.

Conclusions

Our data-driven framework of proteomic profiling in breast cancer cells identified the proteolytic degradation regulator PAI1 and the cell cycle regulator PLK1 as predictors of fibroblast-mediated treatment resistance. Combination therapies targeting HER2 kinase and these fibroblast-induced signaling adaptations eliminates fibroblast-protected HER2+ breast cancer cells.

Determination of key events in mouse hepatocyte maturation at the single-cell level

Hepatocytes, the liver's predominant cells, perform numerous essential biological functions. However, crucial events and regulators during hepatocyte maturation require in-depth investigation. In this study, we performed single-cell RNA sequencing (scRNA-seq) and single-nucleus RNA sequencing (snRNA-seq) to explore the precise hepatocyte development process in mice. We defined three maturation stages of postnatal hepatocytes, each of which establishes specific metabolic functions and exhibits distinct proliferation rates. Hepatic zonation is gradually formed during hepatocyte maturation. Hepatocytes or their nuclei with distinct ploidies exhibit zonation preferences in distribution and asynchrony in maturation. Moreover, by combining gene regulatory network analysis with in vivo genetic manipulation, we identified critical maturation- and zonation-related transcription factors. This study not only delineates the comprehensive transcriptomic profiles of hepatocyte maturation but also presents a paradigm to identify genes that function in the development of hepatocyte maturation and zonation by combining genetic manipulation and measurement of coordinates in a single-cell developmental trajectory.

MOBILE pipeline enables identification of context-specific networks and regulatory mechanisms

Robust identification of context-specific network features that control cellular phenotypes remains a challenge. We here introduce MOBILE (Multi-Omics Binary Integration via Lasso Ensembles) to nominate molecular features associated with cellular phenotypes and pathways. First, we use MOBILE to nominate mechanisms of interferon-γ (IFNγ) regulated PD-L1 expression. Our analyses suggest that IFNγ-controlled PD-L1 expression involves BST2, CLIC2, FAM83D, ACSL5, and HIST2H2AA3 genes, which were supported by prior literature. We also compare networks activated by related family members transforming growth factor-beta 1 (TGFβ1) and bone morphogenetic protein 2 (BMP2) and find that differences in ligand-induced changes in cell size and clustering properties are related to differences in laminin/collagen pathway activity. Finally, we demonstrate the broad applicability and adaptability of MOBILE by analyzing publicly available molecular datasets to investigate breast cancer subtype specific networks. Given the ever-growing availability of multi-omics datasets, we envision that MOBILE will be broadly useful for identification of context-specific molecular features and pathways.

Defects in protective cytokine profiles in spontaneous miscarriage in the first trimester

OBJECTIVE

To study differences in cytokine expression profiles between women with ongoing pregnancy and those experiencing spontaneous miscarriage, among women who presented with threatened miscarriage before week 16 of gestation.

DESIGN

Prospective cohort study.

SETTING

Academic hospital.

PATIENT(S)

In this prospective cohort study, 155 pregnant women, comprising normal pregnant women recruited from antenatal clinics (n = 97) and women with threatened miscarriage recruited from an emergency walk-in clinic (n = 58).

INTERVENTION(S)

None.

MAIN OUTCOME MEASURE(S)

Sixty-five serum cytokines quantified using multiplex immunoassay correlated with miscarriage outcomes.

RESULT(S)

Among women presenting with threatened miscarriage, those who eventually miscarried had significantly lower levels of interleukin (IL)-2, IL-12p70, IL-17A, B-cell-activating factor, B lymphocyte chemoattractant, basic nerve growth factor, interferon-γ, tumor necrosis factor-related apoptosis-inducing ligand, thymic stromal lymphopoietin, and tumor necrosis factor-α and higher levels of vascular endothelial growth factor A, IL-21, and stromal cell-derived factor 1α than those with ongoing pregnancy. Comparisons between normal pregnancies and women with threatened miscarriage who eventually miscarried revealed significant differences across 7 cytokines: B-cell-activating factor; B lymphocyte chemoattractant; basic nerve growth factor; IL-17A; fractalkine/CX3CL1; vascular endothelial growth factor A; and CCL22. Vascular endothelial growth factor A exhibited a negative correlation with the progesterone level (r = -0.270). The cluster of significant cytokines alludes to T cell proliferation, B-cell proliferation, natural killer cell-mediated cytotoxicity, and apoptosis as important pathways that determine pregnancy outcomes. Bioinformatic analysis further revealed alteration of the suppressor of cytokine signaling proteins family of Janus kinase-signal transducer and activator of transcription signaling axis by cytokines as a plausible key molecular mechanism in spontaneous miscarriage.

CONCLUSION(S)

This study demonstrates that the regulated balance between the proinflammatory and anti-inflammatory pathways is crucial to maintaining pregnancy. A better understanding of the cytokines associated with immunomodulatory effects may lead to novel targets for the prediction and treatment of spontaneous miscarriage.

Predicting miRNA-Disease Association Based on Improved Graph Regression

Recently, as a growing number of associations between microRNAs (miRNAs) and diseases are discovered, researchers gradually realize that miRNAs are closely related to several complicated biological processes and human diseases. Hence, it is especially important to construct availably models to infer associations between miRNAs and diseases. In this study, we presented Improved Graph Regression for miRNA-Disease Association Prediction (IGRMDA) to observe potential relationship between miRNAs and diseases. In order to reduce the inherent noise existing in the acquired biological datasets, we utilized matrix decomposition algorithm to process miRNA functional similarity and disease semantic similarity and then combining them with existing similarity information to obtain final miRNA similarity data and disease similarity data. Then, we applied miRNA-disease association data, miRNA similarity data and disease similarity data to form corresponding latent spaces. Furthermore, we performed improved graph regression algorithm in latent spaces, which included miRNA-disease association space, miRNA similarity space and disease similarity space. Non-negative matrix factorization and partial least squares were used in the graph regression process to obtain important related attributes. The cross validation experiments and case studies were also implemented to prove the effectiveness of IGRMDA, which showed that IGRMDA could predict potential associations between miRNAs and diseases.

Transcription factor SP1 regulates haptoglobin fucosylation via induction of GDP-fucose transporter 1 in the hepatoma cell line HepG2

Fucosylation is involved in cancer and inflammation, and several fucosylated proteins, such as AFP-L3 for hepatocellular carcinoma, are used as cancer biomarkers. We previously reported an increase in serum fucosylated haptoglobin (Fuc-Hp) as a biomarker for several cancers, including pancreatic and colon cancer and hepatocellular carcinoma. The regulation of fucosylated protein production is a complex cellular process involving various fucosylation regulatory genes. In this report, we investigated the molecular mechanisms regulating Fuc-Hp production in cytokine-treated hepatoma cells using a partial least squares (PLS) regression model. We found that SLC35C1, which encodes GDP-fucose transporter 1 (GFT1), is the most responsible factor for Fuc-Hp production among various fucosylation regulatory genes. Furthermore, the transcription factor SP1 was essential in regulating SLC35C1 expression. We also found that an SP1 inhibitor was able to suppress Fuc-Hp production without affecting total Hp levels. Taken together, Fuc-Hp production was regulated by SP1 via induction of GFT1 in the hepatoma cell line HepG2.

•

PLS analysis identified SLC35C1 as a critical gene to promote Hp fucosylation.

•

SP1 regulates Fuc-Hp production via inducing SLC35C1.

•

SP1 inhibition decreases Fuc-Hp production in HepG2 cells.

Immune cell mediated cabozantinib resistance for patients with renal cell carcinoma

Renal cell carcinoma (RCC) is the third most common genitourinary cancer in the USA. Despite recent advances in the treatment for advanced and metastatic clear cell RCC (ccRCC), the 5-year relative survival rate for the distant disease remains at 12%. Cabozantinib, a tyrosine kinase inhibitor (TKI), which is one of the first-line therapies approved to treat advanced ccRCC as a single agent, is now being investigated as a combination therapy with newer immunotherapeutic agents. However, not much is known about how cabozantinib modulates the immune system. Here, we present a high throughput tri-culture model that incorporates cancer cells, endothelial cells, and patient-derived immune cells to study the effect of immune cells from patients with ccRCC on angiogenesis and cabozantinib resistance. We show that circulating immune cells from patients with ccRCC induce cabozantinib resistance via increased secretion of a set of pro-angiogenic factors. Using multivariate partial least square regression modeling, we identified CD4+ T cell subsets that are correlated with cabozantinib resistance and report the changes in the frequency of these populations in ccRCC patients who are undergoing cabozantinib therapy. These findings provide a potential set of biomarkers that should be further investigated in the current TKI-immunotherapy combination clinical trials to improve personalized treatments for patients with ccRCC.

Defining Proteomic Signatures to Predict Multidrug Persistence in Pseudomonas aeruginosa

Bacterial persisters are difficult to eradicate because of their ability to survive prolonged exposure to a range of different antibiotics. Because they often represent small subpopulations of otherwise drug-sensitive bacterial populations, studying their physiological state and antibiotic stress response remains challenging. Sorting and enrichment procedures of persister fractions introduce experimental biases limiting the significance of follow-up molecular analyses. In contrast, proteome analysis of entire bacterial populations is highly sensitive and reproducible and can be employed to explore the persistence potential of a given strain or isolate. Here, we summarize methodology to generate proteomic signatures of persistent Pseudomonas aeruginosa isolates with variable fractions of persisters. This includes proteome sample preparation, mass spectrometry analysis, and an adaptable machine learning regression pipeline. We show that this generic method can determine a common proteomic signature of persistence among different P. aeruginosa hyper-persister mutants. We propose that this approach can be used as diagnostic tool to gauge antimicrobial persistence of clinical isolates.

Early identification of Aspergillus spp. contamination in milled rice by E-nose combined with chemometrics

BACKGROUND

Rice grains can be contaminated easily by certain fungi during storage and in the market chain, thus generating a risk for humans. Most classical methods for identifying and rectifying this problem are complex and time-consuming for manufacturers and consumers. However, E-nose technology provides analytical information in a non-destructive and environmentally friendly manner. Two-feature fusion data combined with chemometrics were employed for the determination of Aspergillus spp. contamination in milled rice.

RESULTS

Linear discriminant analysis (LDA) indicated that the efficiency of fusion signals ('80^th s values' and 'area values') outperformed that of independent E-nose signals. Linear discriminant analysis showed clear discrimination of fungal species in stored milled rice for four groups on day 2, and the discrimination accuracy reached 92.86% by using an extreme learning machine (ELM). Gas chromatography-mass spectrometry (GC-MS) analysis showed that the volatile compounds had close relationships with fungal species in rice. The quantification results of colony counts in milled rice showed that the monitoring models based on ELM and the genetic algorithm optimized support vector machine (GA-SVM) (R²  = 0.924-0.983) achieved better performances than those based on partial least squares regression (PLSR) (R²  = 0.877-0.913). The ability of the E-nose to monitor fungal infection at an early stage would help to prevent contaminated rice grains from entering the food chains.

CONCLUSIONS

The results indicated that an E-nose coupled with ELM or GA-SVM algorithm could be a useful tool for the rapid detection of fungal infection in milled rice, to prevent contaminated rice from entering the food chain. © 2021 Society of Chemical Industry.

Perivascular Secretome Influences Hematopoietic Stem Cell Maintenance in a Gelatin Hydrogel

Adult hematopoietic stem cells (HSCs) produce the body's full complement of blood and immune cells. They reside in specialized microenvironments, or niches, within the bone marrow. The perivascular niche near blood vessels is believed to help maintain primitive HSCs in an undifferentiated state but demonstration of this effect is difficult. In vivo studies make it challenging to determine the direct effect of the endosteal and perivascular niches as they can be in close proximity, and two-dimensional in vitro cultures often lack an instructive extracellular matrix environment. We describe a tissue engineering approach to develop and characterize a three-dimensional perivascular tissue model to investigate the influence of the perivascular secretome on HSC behavior. We generate 3D endothelial networks in methacrylamide-functionalized gelatin hydrogels using human umbilical vein endothelial cells (HUVECs) and mesenchymal stromal cells (MSCs). We identify a subset of secreted factors important for HSC function, and examine the response of primary murine HSCs in hydrogels to the perivascular secretome. Within 4 days of culture, perivascular conditioned media promoted maintenance of a greater fraction of hematopoietic stem and progenitor cells. This work represents an important first-generation perivascular model to investigate the role of niche secreted factors on the maintenance of primary HSCs.

Ovarian cancer cells direct monocyte differentiation through a non-canonical pathway

BACKGROUND

Alternatively-activated macrophages (AAMs), an anti-inflammatory macrophage subpopulation, have been implicated in the progression of high grade serous ovarian carcinoma (HGSOC). Increased levels of AAMs are correlated with poor HGSOC survival rates, and AAMs increase the attachment and spread of HGSOC cells in vitro. However, the mechanism by which monocytes in the HGSOC tumor microenvironment are differentiated and polarized to AAMs remains unknown.

METHODS

Using an in vitro co-culture device, we cultured naïve, primary human monocytes with a panel of five HGSOC cell lines over the course of 7 days. An empirical Bayesian statistical method, EBSeq, was used to couple RNA-seq with observed monocyte-derived cell phenotype to explore which HGSOC-derived soluble factors supported differentiation to CD68+ macrophages and subsequent polarization towards CD163+ AAMs. Pathways of interest were interrogated using small molecule inhibitors, neutralizing antibodies, and CRISPR knockout cell lines.

RESULTS

HGSOC cell lines displayed a wide range of abilities to generate AAMs from naïve monocytes. Much of this variation appeared to result from differential ability to generate CD68+ macrophages, as most CD68+ cells were also CD163+. Differences in tumor cell potential to generate macrophages was not due to a MCSF-dependent mechanism, nor variance in established pro-AAM factors. TGFα was implicated as a potential signaling molecule produced by tumor cells that could induce macrophage differentiation, which was validated using a CRISPR knockout of TGFA in the OVCAR5 cell line.

CONCLUSIONS

HGSOC production of TGFα drives monocytes to differentiate into macrophages, representing a central arm of the mechanism by which AAMs are generated in the tumor microenvironment.

Connecting secretome to hematopoietic stem cell phenotype shifts in an engineered bone marrow niche

Hematopoietic stem cells (HSCs) primarily reside in the bone marrow, where they receive external cues from their local microenvironment. The complex milieu of biophysical cues, cellular components and cell-secreted factors regulates the process by which HSC produce the blood and immune system. We previously showed direct coculture of primary murine hematopoietic stem and progenitor cells with a population of marrow-derived mesenchymal stromal and progenitor cells (MSPCs) in a methacrylamide-functionalized gelatin (GelMA) hydrogel improves hematopoietic progenitor maintenance. However, the mechanism by which MSPCs influenced HSC fate decisions remained unknown. Herein, we report the use of proteomic analysis to correlate HSC phenotype to a broad candidate pool of 200 soluble factors produced by combined mesenchymal and hematopoietic progeny. Partial least squares regression (PLSR), along with an iterative filter method, identified TGFβ-1, MMP-3, c-RP and TROY as positively correlated with HSC maintenance. Experimentally, we then observe exogenous stimulation of HSC monocultures in GelMA hydrogels with these combined cytokines increases the ratio of hematopoietic progenitors to committed progeny after a 7-day culture 7.52 ± 3.65-fold compared to non-stimulated monocultures. Findings suggest a cocktail of the downselected cytokines amplifies hematopoietic maintenance potential of HSCs beyond that of MSPC-secreted factors alone. This work integrates empirical and computation methods to identify cytokine combinations to improve HSC maintenance within an engineered HSC niche, suggesting a route toward identifying feeder-free culture platforms for HSC expansion. Insight Hematopoietic stem cells within an artificial niche receive maintenance cues in the form of soluble factors from hematopoietic and mesenchymal progeny. Applying a proteomic regression analysis, we identify a reduced set of soluble factors correlated to maintenance of a hematopoietic phenotype during culture in a biomaterial model of the bone marrow niche. We identify a minimum factor cocktail that promotes hematopoietic maintenance potential in a gelatin-based culture, regardless of the presence of mesenchymal feeder cells. By combining empirical and computational methods, we report an experimentally feasible number of factors from a large dataset, enabling exogenous integration of soluble factors into an engineered hematopoietic stem cell for enhanced maintenance potential of a quiescent stem cell population.

The Many Microenvironments of Ovarian Cancer

High-grade serous ovarian cancer (HGSOC) is the most common and deadly subtype of ovarian cancer as it is commonly diagnosed after substantial metastasis has already occurred. The past two decades have been an active era in HGSOC research, with new information on the origin and genomic signature of the tumor cell. Additionally, studies have begun to characterize changes in the HGSOC microenvironment and examine the impact of these changes on tumor progression and response to therapies. While this knowledge may provide valuable insight into better prognosis and treatments for HGSOCs, its collection, synthesis, and application are complicated by the number of unique microenvironments in the disease-the initiating site (fallopian tube), first metastasis (ovary), distal metastases (peritoneum), and recurrent/platinum-resistant setting. Here, we review the state of our understanding of these diverse sites and highlight remaining questions.

Data-driven analysis of a mechanistic model of CAR T cell signaling predicts effects of cell-to-cell heterogeneity

Due to the variability of protein expression, cells of the same population can exhibit different responses to stimuli. It is important to understand this heterogeneity at the individual level, as population averages mask these underlying differences. Using computational modeling, we can interrogate a system much more precisely than by using experiments alone, in order to learn how the expression of each protein affects a biological system. Here, we examine a mechanistic model of CAR T cell signaling, which connects receptor-antigen binding to MAPK activation, to determine intracellular modulations that can increase cellular response. CAR T cell cancer therapy involves removing a patient's T cells, modifying them to express engineered receptors that can bind to tumor-associated antigens to promote tumor cell killing, and then injecting the cells back into the patient. This population of cells, like all cell populations, would have heterogeneous protein expression, which could affect the efficacy of treatment. Thus, it is important to examine the effects of cell-to-cell heterogeneity. We first generated a dataset of simulated cell responses via Monte Carlo simulations of the mechanistic model, where the initial protein concentrations were randomly sampled. We analyzed the dataset using partial least-squares modeling to determine the relationships between protein expression and ERK phosphorylation, the output of the mechanistic model. Using this data-driven analysis, we found that only the expressions of proteins relating directly to the receptor and the MAPK cascade, the beginning and end of the network, respectively, are relevant to the cells' response. We also found, surprisingly, that increasing the amount of receptor present can actually inhibit the cell's ability to respond due to increasing the strength of negative feedback from phosphatases. Overall, we have combined data-driven and mechanistic modeling to generate detailed insight into CAR T cell signaling.

The impact of tumor receptor heterogeneity on the response to anti-angiogenic cancer treatment

Multiple promoters and inhibitors mediate angiogenesis, the formation of new blood vessels, and these factors represent potential targets for impeding vessel growth in tumors. Vascular endothelial growth factor (VEGF) is a potent angiogenic factor targeted in anti-angiogenic cancer therapies. In addition, thrombospondin-1 (TSP1) is a major endogenous inhibitor of angiogenesis, and TSP1 mimetics are being developed as an alternative type of anti-angiogenic agent. The combination of bevacizumab, an anti-VEGF agent, and ABT-510, a TSP1 mimetic, has been tested in clinical trials to treat advanced solid tumors. However, the patients' responses are highly variable and show disappointing outcomes. To obtain mechanistic insight into the effects of this combination anti-angiogenic therapy, we have constructed a novel whole-body systems biology model including the VEGF and TSP1 reaction networks. Using this molecular-detailed model, we investigated how the combination anti-angiogenic therapy changes the amounts of pro-angiogenic and anti-angiogenic complexes in cancer patients. We particularly focus on answering the question of how the effect of the combination therapy is influenced by tumor receptor expression, one aspect of patient-to-patient variability. Overall, this model complements the clinical administration of combination anti-angiogenic therapy, highlights the role of tumor receptor variability in the heterogeneous responses to anti-angiogenic therapy, and identifies the tumor receptor profiles that correlate with a high likelihood of a positive response to the combination therapy. Our model provides novel understanding of the VEGF-TSP1 balance in cancer patients at the systems-level and could be further used to optimize combination anti-angiogenic therapy.

Alternatively-Activated Macrophages Upregulate Mesothelial Expression of P-Selectin to Enhance Adhesion of Ovarian Cancer Cells

Peritoneal metastasis of high-grade serous ovarian cancer (HGSOC) occurs when tumor cells suspended in ascites adhere to mesothelial cells. Despite the strong relationship between metastatic burden and prognosis in HGSOC, there are currently no therapies specifically targeting the metastatic process. We utilized a coculture model and multivariate analysis to examine how interactions between tumor cells, mesothelial cells, and alternatively-activated macrophages (AAM) influence the adhesion of tumor cells to mesothelial cells. We found that AAM-secreted MIP-1β activates CCR5/PI3K signaling in mesothelial cells, resulting in expression of P-selectin on the mesothelial cell surface. Tumor cells attached to this de novo P-selectin through CD24, resulting in increased tumor cell adhesion in static conditions and rolling underflow. C57/BL6 mice treated with MIP-1β exhibited increased P-selectin expression on mesothelial cells lining peritoneal tissues, which enhanced CaOV3 adhesion ex vivo and ID8 adhesion in vivo Analysis of samples from patients with HGSOC confirmed increased MIP-1β and P-selectin, suggesting that this novel multicellular mechanism could be targeted to slow or stop metastasis in HGSOC by repurposing anti-CCR5 and P-selectin therapies developed for other indications.Significance: This study reports novel insights on the peritoneal dissemination occurring during progression of ovarian cancer and has potential for therapeutic intervention.Graphical Abstract: http://cancerres.aacrjournals.org/content/canres/78/13/3560/F1.large.jpg Cancer Res; 78(13); 3560-73. ©2018 AACR.

Population-based mechanistic modeling allows for quantitative predictions of drug responses across cell types

Quantitative mismatches between human physiology and experimental models can be problematic for the development of effective therapeutics. When the effects of drugs on human adult cardiac electrophysiology are of interest, phenotypic differences with animal cells, and more recently stem cell-derived models, can present serious limitations. We addressed this issue through a combination of mechanistic mathematical modeling and statistical analyses. Physiological metrics were simulated in heterogeneous populations of models describing cardiac myocytes from adult ventricles and those derived from induced pluripotent stem cells (iPSC-CMs). These simulated measures were used to construct a cross-cell type regression model that predicts adult myocyte drug responses from iPSC-CM behaviors. We found that (1) quantitatively accurate predictions of responses to selective or non-selective ion channel blocking drugs could be generated based on iPSC-CM responses under multiple experimental conditions; (2) altering extracellular ion concentrations is an effective experimental perturbation for improving the model’s predictive strength; (3) the method can be extended to predict and contrast drug responses in diseased as well as healthy cells, indicating a broader application of the concept. This cross-cell type model can be of great value in drug development, and the approach, which can be applied to other fields, represents an important strategy for overcoming experimental model limitations.

The quantitative limitations of experimental models, which can impair the development of effective therapeutics, can be overcome through a combination of mechanistic simulations and statistical analysis. A team from Icahn School of Medicine at Mount Sinai led by Eric Sobie devised a computational method to quantitatively translate drug responses across cell types. The method involves mechanism-based simulations of heterogeneous populations combined with a multivariable regression model that translates between cell types. Simulation results presented in the study show that the response to a drug in one cell type can be predicted with quantitative accuracy from physiological recordings made in another cell type, as can differential drug responses observed in diseased compared with healthy cells. This methodology can be used in drug development to better predict clinical responses based on experiments performed in preclinical models.

GRMDA: Graph Regression for MiRNA-Disease Association Prediction

Nowadays, as more and more associations between microRNAs (miRNAs) and diseases have been discovered, miRNA has gradually become a hot topic in the biological field. Because of the high consumption of time and money on carrying out biological experiments, computational method which can help scientists choose the most likely associations between miRNAs and diseases for further experimental studies is desperately needed. In this study, we proposed a method of Graph Regression for MiRNA-Disease Association prediction (GRMDA) which combines known miRNA-disease associations, miRNA functional similarity, disease semantic similarity, and Gaussian interaction profile kernel similarity. We used Gaussian interaction profile kernel similarity to supplement the shortage of miRNA functional similarity and disease semantic similarity. Furthermore, the graph regression was synchronously performed in three latent spaces, including association space, miRNA similarity space, and disease similarity space, by using two matrix factorization approaches called Singular Value Decomposition and Partial Least-Squares to extract important related attributes and filter the noise. In the leave-one-out cross validation and five-fold cross validation, GRMDA obtained the AUCs of 0.8272 and 0.8080 ± 0.0024, respectively. Thus, its performance is better than some previous models. In the case study of Lymphoma using the recorded miRNA-disease associations in HMDD V2.0 database, 88% of top 50 predicted miRNAs were verified by experimental literatures. In order to test the performance of GRMDA on new diseases with no known related miRNAs, we took Breast Neoplasms as an example by regarding all the known related miRNAs as unknown ones. We found that 100% of top 50 predicted miRNAs were verified. Moreover, 84% of top 50 predicted miRNAs in case study for Esophageal Neoplasms based on HMDD V1.0 were verified to have known associations. In conclusion, GRMDA is an effective and practical method for miRNA-disease association prediction.

Mechanistic modeling quantifies the influence of tumor growth kinetics on the response to anti-angiogenic treatment

Tumors exploit angiogenesis, the formation of new blood vessels from pre-existing vasculature, in order to obtain nutrients required for continued growth and proliferation. Targeting factors that regulate angiogenesis, including the potent promoter vascular endothelial growth factor (VEGF), is therefore an attractive strategy for inhibiting tumor growth. Computational modeling can be used to identify tumor-specific properties that influence the response to anti-angiogenic strategies. Here, we build on our previous systems biology model of VEGF transport and kinetics in tumor-bearing mice to include a tumor compartment whose volume depends on the “angiogenic signal” produced when VEGF binds to its receptors on tumor endothelial cells. We trained and validated the model using published in vivo measurements of xenograft tumor volume, producing a model that accurately predicts the tumor’s response to anti-angiogenic treatment. We applied the model to investigate how tumor growth kinetics influence the response to anti-angiogenic treatment targeting VEGF. Based on multivariate regression analysis, we found that certain intrinsic kinetic parameters that characterize the growth of tumors could successfully predict response to anti-VEGF treatment, the reduction in tumor volume. Lastly, we use the trained model to predict the response to anti-VEGF therapy for tumors expressing different levels of VEGF receptors. The model predicts that certain tumors are more sensitive to treatment than others, and the response to treatment shows a nonlinear dependence on the VEGF receptor expression. Overall, this model is a useful tool for predicting how tumors will respond to anti-VEGF treatment, and it complements pre-clinical in vivo mouse studies.

One hallmark of cancer is angiogenesis, the formation of new blood capillaries from pre-existing vessels. Angiogenesis promotes tumor growth by enabling the tumor to obtain oxygen and nutrients from the surrounding microenvironment. Cancer drugs that inhibit angiogenesis ("anti-angiogenic therapies") have focused on inhibiting proteins that promote the growth of new blood vessels. The response to anti-angiogenic therapy is highly variable, and some tumors do not respond at all. Therefore, identifying a biomarker that predicts how specific tumors will respond would be extremely valuable. This work uses a computational model of tumor-bearing mice to investigate the response to anti-angiogenic treatment that targets the potent promoter of angiogenesis, vascular endothelial growth factor (VEGF), and how the response is influenced by tumor growth kinetics. We show that certain properties of tumor growth can be used to predict how much the tumor volume will be reduced upon administration of an anti-VEGF drug. This work identifies tumor growth parameters that may be reliable biomarkers for predicting how tumors will respond to anti-VEGF therapy. Our computational model generates novel, testable hypotheses and nicely complements pre-clinical studies of anti-angiogenic therapeutics.

An improved partial least-squares regression method for Raman spectroscopy

It is known that the performance of partial least-squares (PLS) regression analysis can be improved using the backward variable selection method (BVSPLS). In this paper, we further improve the BVSPLS based on a novel selection mechanism. The proposed method is based on sorting the weighted regression coefficients, and then the importance of each variable of the sorted list is evaluated using root mean square errors of prediction (RMSEP) criterion in each iteration step. Our Improved BVSPLS (IBVSPLS) method has been applied to leukemia and heparin data sets and led to an improvement in limit of detection of Raman biosensing ranged from 10% to 43% compared to PLS. Our IBVSPLS was also compared to the jack-knifing (simpler) and Genetic Algorithm (more complex) methods. Our method was consistently better than the jack-knifing method and showed either a similar or a better performance compared to the genetic algorithm.

Data-Modeling Identifies Conflicting Signaling Axes Governing Myoblast Proliferation and Differentiation Responses to Diverse Ligand Stimuli

Introduction

Skeletal muscle tissue development and regeneration relies on the proliferation, maturation and fusion of muscle progenitor cells (myoblasts), which arise transiently from muscle stem cells (satellite cells). Following muscle damage, myoblasts proliferate and differentiate in response to temporally-varying inflammatory cytokines, growth factors, and extracellular matrix cues, which stimulate a shared network of intracellular signaling pathways. Here we present an integrated data-modeling approach to elucidate synergies and antagonisms among proliferation and differentiation signaling axes in myoblasts stimulated by regeneration-associated ligands.

Methods

We treated mouse primary myoblasts in culture with combinations of eight regeneration-associated growth factors and cytokines in mixtures that induced additive, synergistic, and antagonistic effects on myoblast proliferation and differentiation responses. For these combinatorial stimuli, we measured the activation dynamics of seven signal transduction pathways using multiplexed phosphoprotein assays and scored proliferation and differentiation responses based on expression of myogenic commitment factors to assemble a cue-signaling-response data compendium. We interrogated the relationship between these signals and responses by partial least-squares (PLS) regression modeling.

Results

Partial least-squares data-modeling accurately predicted response outcomes in cross-validation on the training compendium (cumulative R ² = 0.96). The PLS model highlighted signaling axes that distinctly govern myoblast proliferation (MEK-ERK, Stat3) and differentiation (JNK) in response to these combinatorial cues, and we confirmed these signal-response associations with small molecule perturbations. Unexpectedly, we observed that a negative feedback circuit involving the phosphatase DUSP6/MKP-3 auto-regulates MEK-ERK signaling in myoblasts.

Conclusion

This data-modeling approach identified conflicting signaling axes that underlie muscle progenitor cell proliferation and differentiation.

Partial Least Squares Regression Models for the Analysis of Kinase Signaling

Partial least squares regression (PLSR) is a data-driven modeling approach that can be used to analyze multivariate relationships between kinase networks and cellular decisions or patient outcomes. In PLSR, a linear model relating an X matrix of dependent variables and a Y matrix of independent variables is generated by extracting the factors with the strongest covariation. While the identified relationship is correlative, PLSR models can be used to generate quantitative predictions for new conditions or perturbations to the network, allowing for mechanisms to be identified. This chapter will provide a brief explanation of PLSR and provide an instructive example to demonstrate the use of PLSR to analyze kinase signaling.

TNF-insulin crosstalk at the transcription factor GATA6 is revealed by a model that links signaling and transcriptomic data tensors

Signal transduction networks coordinate transcriptional programs activated by diverse extracellular stimuli, such as growth factors and cytokines. Cells receive multiple stimuli simultaneously, and mapping how activation of the integrated signaling network affects gene expression is a challenge. We stimulated colon adenocarcinoma cells with various combinations of the cytokine tumor necrosis factor (TNF) and the growth factors insulin and epidermal growth factor (EGF) to investigate signal integration and transcriptional crosstalk. We quantitatively linked the proteomic and transcriptomic data sets by implementing a structured computational approach called tensor partial least squares regression. This statistical model accurately predicted transcriptional signatures from signaling arising from single and combined stimuli and also predicted time-dependent contributions of signaling events. Specifically, the model predicted that an early-phase, AKT-associated signal downstream of insulin repressed a set of transcripts induced by TNF. Through bioinformatics and cell-based experiments, we identified the AKT-repressed signal as glycogen synthase kinase 3 (GSK3)-catalyzed phosphorylation of Ser(37) on the long form of the transcription factor GATA6. Phosphorylation of GATA6 on Ser(37) promoted its degradation, thereby preventing GATA6 from repressing transcripts that are induced by TNF and attenuated by insulin. Our analysis showed that predictive tensor modeling of proteomic and transcriptomic data sets can uncover pathway crosstalk that produces specific patterns of gene expression in cells receiving multiple stimuli.

Hierarchy of cellular decisions in collective behavior: Implications for wound healing

Collective processes such as wound re-epithelialization result from the integration of individual cellular decisions. To determine which individual cell behaviors represent the most promising targets to engineer re-epithelialization, we examined collective and individual responses of HaCaT keratinocytes seeded upon polyacrylamide gels of three stiffnesses (1, 30, and 100 kPa) and treated with a range of epidermal growth factor (EGF) doses. Wound closure was found to increase with substrate stiffness, but was responsive to EGF treatment only above a stiffness threshold. Individual cell behaviors were used to create a partial least squares regression model to predict the hierarchy of factors driving wound closure. Unexpectedly, cell area and persistence were found to have the strongest correlation to the observed differences in wound closure. Meanwhile, the model predicted a relatively weak correlation between wound closure with proliferation, and the unexpectedly minor input from proliferation was successfully tested with inhibition by aphidicolin. Combined, these results suggest that the poor clinical results for growth factor-based therapies for chronic wounds may result from a disconnect between the individual cellular behaviors targeted in these approaches and the resulting collective response. Additionally, the stiffness-dependency of EGF sensitivity suggests that therapies matched to microenvironmental characteristics will be more efficacious.

Cell-to-cell variability in cell death: can systems biology help us make sense of it all?

One of the most common observations in cell death assays is that not all cells die at the same time, or at the same treatment dose. Here, using the perspective of the systems biology of apoptosis and the context of cancer treatment, we discuss possible sources of this cell-to-cell variability as well as its implications for quantitative measurements and computational models of cell death. Many different factors, both within and outside of the apoptosis signaling networks, have been correlated with the variable responses to various death-inducing treatments. Systems biology models offer us the opportunity to take a more synoptic view of the cell death process to identify multifactorial determinants of the cell death decision. Finally, with an eye toward 'systems pharmacology', we discuss how leveraging this new understanding should help us develop combination treatment strategies to compel cancer cells toward apoptosis by manipulating either the biochemical state of cancer cells or the dynamics of signal transduction.