Compressive forces driven by lateral actin fibers are a key to the nuclear deformation under uniaxial cell-substrate stretching.

Cells sense the direction of mechanical stimuli including substrate stretching and show morphological and functional responses. The nuclear deformation with respect to the direction of mechanical stimuli is thought of as a vital factor in mechanosensitive intracellular signaling and gene transcription, but the detailed relationship between the direction of stimuli and nuclear deformation behavior is not fully solved yet. Here, we assessed the role of actin cytoskeletons in nuclear deformation caused by cell substrate stretching with different directions. Cells on a PDMS stretching chamber were subjected to a step-strain and changes of long- and short-axes of nucleus before and after stretching were evaluated in terms of nuclear orientation against the direction of stretching. Nuclei oriented parallel to the stretching direction showed elongation and shrinkage in the long and short axes, respectively, and vice versa. However, calculation of the aspect ratio (ratio of long- and short-axes) changes revealed orientation-depend nuclear deformation: The nucleus oriented parallel to the stretching direction showed a greater aspect ratio change than it aligned in the perpendicular direction of the stretching. A decrease in actin cytoskeletal tension significantly changed the nuclear deformation only in the short axis direction, thereby abolishing the orientation-depend deformation of the nucleus. These results suggest that lateral compressive forces exerted by the actin cytoskeleton is a key factor of orientation-depend deformation in short axis of the nucleus under the cell-substrate stretching condition, and may be crucial for mechano-sensing and responses to the cell-substrate stretching direction.

Cellular ageing of oral fibroblasts differentially modulates extracellular matrix organization.

BACKGROUND AND OBJECTIVES: Ageing is associated with an impaired cellular function that can affect tissue architecture and wound healing in gingival and periodontal tissues. However, the impact of oral fibroblast ageing on the structural organization of the extracellular matrix (ECM) proteins is poorly understood. Hence, in this study, we investigated the impact of cellular ageing of oral fibroblasts on the production and structural organization of collagen and other ECM proteins. METHODS: Oral fibroblasts were serially subcultured, and replicative cellular senescence was assessed using population doubling time, Ki67 counts and expression of P21WAFI . The production and structural organization of ECM proteins were assessed at early (young-oFB) and late (aged-oFB) passages. The thickness and pattern of collagen produced by live cultures of young- and aged-oFB were assessed using a label-free and non-invasive second harmonic generation (SHG)-based multiphoton imaging. Expression of other ECM proteins (fibronectin, fibrillin, collagen-IV and laminins) was evaluated using immunocytochemistry and confocal microscopy-based depth profile analysis. RESULTS: Aged-oFB displayed a higher population doubling time, lower Ki67+ cells and higher expression of P21WAFI indicative of slower proliferation rate and senescence phenotype. SHG imaging demonstrated that young-oFB produced a thick, interwoven network of collagen fibres, while the aged-oFB produced thin and linearly organized collagen fibres. Similarly, analysis of immunostained cultures showed that young-oFB produced a rich, interwoven mesh of fibronectin, fibrillin and collagen-IV fibres. In contrast, the aged-oFB produced linearly organized fibronectin, fibrillin and collagen-IV fibres. Lastly, there was no observable difference in production and organization of laminins among the young- and aged-oFB. CONCLUSION: Our results suggest that oral fibroblast ageing impairs ECM production and more importantly the organization of ECM fibres, which could potentially impair wound healing in the elderly.

Regulation of flagellar motor switching by c-di-GMP phosphodiesterases in Pseudomonas aeruginosa.

The second messenger cyclic diguanylate (c-di-GMP) plays a prominent role in regulating flagellum-dependent motility in the single-flagellated pathogenic bacterium Pseudomonas aeruginosa The c-di-GMP-mediated signaling pathways and mechanisms that control flagellar output remain to be fully unveiled. Studying surface-tethered and free-swimming P. aeruginosa PAO1 cells, we found that the overexpression of an exogenous diguanylate cyclase (DGC) raises the global cellular c-di-GMP concentration and thereby inhibits flagellar motor switching and decreases motor speed, reducing swimming speed and reversal frequency, respectively. We noted that the inhibiting effect of c-di-GMP on flagellar motor switching, but not motor speed, is exerted through the c-di-GMP-binding adaptor protein MapZ and associated chemotactic pathways. Among the 22 putative c-di-GMP phosphodiesterases, we found that three of them (DipA, NbdA, and RbdA) can significantly inhibit flagellar motor switching and swimming directional reversal in a MapZ-dependent manner. These results disclose a network of c-di-GMP-signaling proteins that regulate chemotactic responses and flagellar motor switching in P. aeruginosa and establish MapZ as a key signaling hub that integrates inputs from different c-di-GMP-signaling pathways to control flagellar output and bacterial motility. We rationalized these experimental findings by invoking a model that postulates the regulation of flagellar motor switching by subcellular c-di-GMP pools.

Structural analyses unravel the molecular mechanism of cyclic di-GMP regulation of bacterial chemotaxis via a PilZ adaptor protein.

The bacterial second messenger cyclic di-GMP (c-di-GMP) has emerged as a prominent mediator of bacterial physiology, motility, and pathogenicity. c-di-GMP often regulates the function of its protein targets through a unique mechanism that involves a discrete PilZ adaptor protein. However, the molecular mechanism for PilZ protein-mediated protein regulation is unclear. Here, we present the structure of the PilZ adaptor protein MapZ cocrystallized in complex with c-di-GMP and its protein target CheR1, a chemotaxis-regulating methyltransferase in Pseudomonas aeruginosa This cocrystal structure, together with the structure of free CheR1, revealed that the binding of c-di-GMP induces dramatic structural changes in MapZ that are crucial for CheR1 binding. Importantly, we found that restructuring and repositioning of two C-terminal helices enable MapZ to disrupt the CheR1 active site by dislodging a structural domain. The crystallographic observations are reinforced by protein-protein binding and single cell-based flagellar motor switching analyses. Our studies further suggest that the regulation of chemotaxis by c-di-GMP through MapZ orthologs/homologs is widespread in proteobacteria and that the use of allosterically regulated C-terminal motifs could be a common mechanism for PilZ adaptor proteins. Together, the findings provide detailed structural insights into how c-di-GMP controls the activity of an enzyme target indirectly through a PilZ adaptor protein.

A computational model for how cells choose temporal or spatial sensing during chemotaxis.

Cell size is thought to play an important role in choosing between temporal and spatial sensing in chemotaxis. Large cells are thought to use spatial sensing due to large chemical difference at its ends whereas small cells are incapable of spatial sensing due to rapid homogenization of proteins within the cell. However, small cells have been found to polarize and large cells like sperm cells undergo temporal sensing. Thus, it remains an open question what exactly governs spatial versus temporal sensing. Here, we identify the factors that determines sensing choices through mathematical modeling of chemotactic circuits. Comprehensive computational search of three-node signaling circuits has identified the negative integral feedback (NFB) and incoherent feedforward (IFF) circuits as capable of adaptation, an important property for chemotaxis. Cells are modeled as one-dimensional circular system consisting of diffusible activator, inactivator and output proteins, traveling across a chemical gradient. From our simulations, we find that sensing outcomes are similar for NFB or IFF circuits. Rather than cell size, the relevant parameters are the 1) ratio of cell speed to the product of cell diameter and rate of signaling, 2) diffusivity of the output protein and 3) ratio of the diffusivities of the activator to inactivator protein. Spatial sensing is favored when all three parameters are low. This corresponds to a cell moving slower than the time it takes for signaling to propagate across the cell diameter, has an output protein that is polarizable and has a local-excitation global-inhibition system to amplify the chemical gradient. Temporal sensing is favored otherwise. We also find that temporal sensing is more robust to noise. By performing extensive literature search, we find that our prediction agrees with observation in a wide range of species and cell types ranging from E. coli to human Fibroblast cells and propose that our result is universally applicable.

Cell-Cell Adhesion and Cortical Actin Bending Govern Cell Elongation on Negatively Curved Substrates.

Physiologically, cells experience and respond to a variety of mechanical stimuli such as rigidity and topography of the extracellular matrix. However, little is known about the effects of substrate curvature on cell behavior. We developed a novel, to our knowledge, method to fabricate cell culture substrates with semicylindrical grooves of negative curvatures (radius of curvature, Rc = 20-100 µm). We found that negative substrate curvatures induced elongation of mesenchymal and epithelial cells along the cylinder axis. As Rc decreases, mesenchymal National Institutes of Health 3T3 fibroblasts increasingly elongate along the long axis of the grooves, whereas elongation of epithelial Madin-Darby Canine Kidney (MDCK) cells is biphasic with maximal cell elongation when Rc = 40 µm. Addition of blebbistatin to MDCK cells to reduce cortical actin rigidity resulted in a decrease in cell elongation across all curvatures while preserving the biphasic trend. However, addition of calyculin A or ethylene glycol-bis(2-aminoethylether)-N,N,N',N'-tetraacetic acid, to increase cortical rigidity or reduce intercellular adhesion, respectively, resulted in a monotonic increase in MDCK cell elongation with decreasing Rc. Using an energy minimization model, we showed that cell elongation in epithelial cell sheet is governed by the competition between two energies as Rc decreases: curvature-dependent intercellular adhesion that prevents elongation; and intracellular cortical actin bending that enhances elongation. Therefore, our results of cellular elongation induced by negatively curved substrates offer insights into how tubule elongation or growth of tubular structures such as kidney tubules can be controlled by the substrate curvature in vivo.

MEKK1-dependent phosphorylation of calponin-3 tunes cell contractility.

MEKK1 (also known as MAP3K1), which plays a major role in MAPK signaling, has been implicated in mechanical processes in cells, such as migration. Here, we identify the actin-binding protein calponin-3 as a new MEKK1 substrate in the signaling that regulates actomyosin-based cellular contractility. MEKK1 colocalizes with calponin-3 at the actin cytoskeleton and phosphorylates it, leading to an increase in the cell-generated traction stress. MEKK1-mediated calponin-3 phosphorylation is attenuated by the inhibition of myosin II activity, the disruption of actin cytoskeletal integrity and adhesion to soft extracellular substrates, whereas it is enhanced upon cell stretching. Our results reveal the importance of the MEKK1-calponin-3 signaling pathway to cell contractility.

POPX2 phosphatase regulates the KIF3 kinesin motor complex.

The kinesin motors are important in the regulation of cellular functions such as protein trafficking, spindle organization and centrosome separation. In this study, we have identified POPX2, a serine-threonine phosphatase, as an interacting partner of the KAP3 subunit of the kinesin-2 motor. The kinesin-2 motor is a heterotrimeric complex composed of KIF3A, KIF3B motor subunits and KAP3, the non-motor subunit, which binds the cargo. Here we report that the phosphatase POPX2 is a negative regulator of the trafficking of N-cadherin and other cargoes; consequently, it markedly influences cell-cell adhesion. POPX2 affects trafficking by determining the phosphorylation status of KIF3A at serine 690. This is consistent with the observation that the KIF3A-S690A mutant is defective in cargo trafficking. Our studies also implicate CaMKII as the kinase that phosphorylates KIF3A at serine 690. These results strongly suggest that POPX2 and CaMKII are a phosphatase-kinase pair that regulates kinesin-mediated transport and cell-cell adhesion.

Loss of p53 enhances NF-κB-dependent lamellipodia formation.

Tumor suppressor p53 prevents tumorigenesis and tumor growth by suppressing the activation of several transcription factors, including nuclear factor-κB (NF-κB) and STAT3. On the other hand, p53 stimulates actin cytoskeleton remodeling and integrin-related signaling cascades. Here, we examined the p53-mediated link between regulation of the actin cytoskeleton and activation of NF-κB and STAT3 in MCF-7 cells and mouse embryonic fibroblasts (MEFs). In the absence of p53, STAT3 was constitutively activated. This activation was attenuated by depleting the expression of p65, a component of NF-κB. Integrin ß3 expression and lamellipodia formation were also downregulated by NF-κB depletion. Inhibition of integrin αvß3, Rac1 or Arp2/3, which diminished lamellipodia formation, suppressed STAT3 activation induced by p53 depletion. These results suggest that loss of p53 leads to STAT3 activation via NF-κB-dependent lamellipodia formation. Our study proposes a novel role for p53 in modulating the actin cytoskeleton through suppression of NF-κB, which restricts STAT3 activation.

Discovery of plasma biomarkers related to blood-brain barrier dysregulation in Alzheimer's disease.

Introduction: Blood-based biomarkers are quantitative, non-invasive diagnostic tools. This study aimed to identify candidate biomarkers for Alzheimer's disease (AD) using publicly available omics datasets, using the hypothesis that with blood-brain barrier dysfunction in AD, brain-synthesized proteins can leak into plasma for detection. Methods: Differential abundance results of plasma and brain proteomic datasets were integrated to obtain a list of potential biomarkers. Biological validity was investigated with intercellular communication and gene regulatory analyses on brain single-cell transcriptomics data. Results: Five proteins (APOD, B2M, CFH, CLU, and C3) fit biomarker criteria. 4 corresponding transcripts (APOD, B2M, CLU, and C3) were overexpressed in AD astrocytes, mediated by AD-related dysregulations in transcription factors regulating neuroinflammation. Additionally, CLU specifically induced downstream expression of neuronal death genes. Discussion: In conclusion, a 5-protein panel is shown to effectively identify AD patients, with evidence of disease specificity and biological validity. Future research should investigate the mechanism of protein leakage through the blood-brain barrier.

The role of cell-matrix adhesion and cell migration in breast tumor growth and progression.

During breast cancer progression, there is typically increased collagen deposition resulting in elevated extracellular matrix rigidity. This results in changes to cell-matrix adhesion and cell migration, impacting processes such as the epithelial-mesenchymal transition (EMT) and metastasis. We aim to investigate the roles of cell-matrix adhesion and cell migration on breast tumor growth and progression by studying the impacts of different types of extracellular matrices and their rigidities. We embedded MCF7 spheroids within three-dimensional (3D) collagen matrices and agarose matrices. MCF7 cells adhere to collagen but not agarose. Contrasting the results between these two matrices allows us to infer the role of cell-matrix adhesion. We found that MCF7 spheroids exhibited the fastest growth rate when embedded in a collagen matrix with a rigidity of 5.1 kPa (0.5 mg/mL collagen), whereas, for the agarose matrix, the rigidity for the fastest growth rate is 15 kPa (1.0% agarose) instead. This discrepancy is attributable to the presence of cell adhesion molecules in the collagen matrix, which initiates collagen matrix remodeling and facilitates cell migration from the tumor through the EMT. As breast tumors do not adhere to agarose matrices, it is suitable to simulate the cell-cell interactions during the early stage of breast tumor growth. We conducted further analysis to characterize the stresses exerted by the expanding spheroid on the agarose matrix. We identified two distinct MCF7 cell populations, namely, those that are non-dividing and those that are dividing, which exerted low and high expansion stresses on the agarose matrix, respectively. We confirmed this using Western blot which showed the upregulation of proliferating cell nuclear antigen, a proliferation marker, in spheroids grown in the 1.0% agarose (≈13 kPa). By treating the embedded MCF7 spheroids with an inhibitor or activator of myosin contractility, we showed that the optimum spheroids' growth can be increased or decreased, respectively. This finding suggests that tumor growth in the early stage, where cell-cell interaction is more prominent, is determined by actomyosin tension, which alters cell rounding pressure during cell division. However, when breast tumors begin generating collagen into the surrounding matrix, collagen remodeling triggers EMT to promote cell migration and invasion, ultimately leading to metastasis.

Cellular response to substrate rigidity is governed by either stress or strain.

Cells sense the rigidity of their substrate; however, little is known about the physical variables that determine their response to this rigidity. Here, we report traction stress measurements carried out using fibroblasts on polyacrylamide gels with Young's moduli ranging from 6 to 110 kPa. We prepared the substrates by employing a modified method that involves N-acryloyl-6-aminocaproic acid (ACA). ACA allows for covalent binding between proteins and elastomers and thus introduces a more stable immobilization of collagen onto the substrate when compared to the conventional method of using sulfo-succinimidyl-6-(4-azido-2-nitrophenyl-amino) hexanoate (sulfo-SANPAH). Cells remove extracellular matrix proteins off the surface of gels coated using sulfo-SANPAH, which corresponds to lower values of traction stress and substrate deformation compared to gels coated using ACA. On soft ACA gels (Young's modulus <20 kPa), cell-exerted substrate deformation remains constant, independent of the substrate Young's modulus. In contrast, on stiff substrates (Young's modulus >20 kPa), traction stress plateaus at a limiting value and the substrate deformation decreases with increasing substrate rigidity. Sustained substrate strain on soft substrates and sustained traction stress on stiff substrates suggest these may be factors governing cellular responses to substrate rigidity.

Bacterial tethering analysis reveals a "run-reverse-turn" mechanism for Pseudomonas species motility.

We have developed a program that can accurately analyze the dynamic properties of tethered bacterial cells. The program works especially well with cells that tend to give rise to unstable rotations, such as polar-flagellated bacteria. The program has two novel components. The first dynamically adjusts the center of the cell's rotational trajectories. The second applies piecewise linear approximation to the accumulated rotation curve to reduce noise and separate the motion of bacteria into phases. Thus, it can separate counterclockwise (CCW) and clockwise (CW) rotations distinctly and measure rotational speed accurately. Using this program, we analyzed the properties of tethered Pseudomonas aeruginosa and Pseudomonas putida cells for the first time. We found that the Pseudomonas flagellar motor spends equal time in both CCW and CW phases and that it rotates with the same speed in both phases. In addition, we discovered that the cell body can remain stationary for short periods of time, leading to the existence of a third phase of the flagellar motor which we call "pause." In addition, P. aeruginosa cells adopt longer run lengths, fewer pause frequencies, and shorter pause durations as part of their chemotactic response. We propose that one purpose of the pause phase is to allow the cells to turn at a large angle, where we show that pause durations in free-swimming cells positively correlate with turn angle sizes. Taken together, our results suggest a new "run-reverse-turn" paradigm for polar-flagellated Pseudomonas motility that is different from the "run-and-tumble" paradigm established for peritrichous Escherichia coli.

A review of algorithmic approaches for cell culture media optimization.

Cell culture media composition and culture conditions play a crucial role in product yield, quality and cost of production. Culture media optimization is the technique of improving media composition and culture conditions to achieve desired product outcomes. To achieve this, there have been many algorithmic methods proposed and used for culture media optimization in the literature. To help readers evaluate and decide on a method that best suits their specific application, we carried out a systematic review of the different methods from an algorithmic perspective that classifies, explains and compares the available methods. We also examine the trends and new developments in the area. This review provides recommendations to researchers regarding the suitable media optimization algorithm for their applications and we hope to also promote the development of new cell culture media optimization methods that are better suited to existing and upcoming challenges in this biotechnology field, which will be essential for more efficient production of various cell culture products.

Subcellular Force Imbalance in Actin Bundles Induces Nuclear Repositioning and Durotaxis.

Durotaxis is a phenomenon in which cells migrate toward substrates of increasing stiffness. However, how cells assimilate substrate stiffness as a directional cue remains poorly understood. In this study, we experimentally show that mouse embryonic fibroblasts can discriminate between different substrate stiffnesses and develop higher traction forces at regions of the cell adhering to the stiffer pillars. In this way, the cells generate a force imbalance between adhesion sites. It is this traction force imbalance that drives durotaxis by providing directionality for cell migration. Significantly, we found that traction forces are transmitted via LINC complexes to the cell nucleus, which serves to maintain the global force imbalance. In this way, LINC complexes play an essential role in anterograde nuclear movement and durotaxis. This conclusion is supported by the fact that LINC complex-deficient cells are incapable of durotaxis and instead migrate randomly on substrates featuring a stiffness gradient.

ContrastivePose: A contrastive learning approach for self-supervised feature engineering for pose estimation and behavorial classification of interacting animals.

In recent years, supervised machine learning models trained on videos of animals with pose estimation data and behavior labels have been used for automated behavior classification. Applications include, for example, automated detection of neurological diseases in animal models. However, we identify two potential problems of such supervised learning approach. First, such models require a large amount of labeled data but the labeling of behaviors frame by frame is a laborious manual process that is not easily scalable. Second, such methods rely on handcrafted features obtained from pose estimation data that are usually designed empirically. In this paper, we propose to overcome these two problems using contrastive learning for self-supervised feature engineering on pose estimation data. Our approach allows the use of unlabeled videos to learn feature representations and reduce the need for handcrafting of higher-level features from pose positions. We show that this approach to feature representation can achieve better classification performance compared to handcrafted features alone, and that the performance improvement is due to contrastive learning on unlabeled data rather than the neural network architecture. The method has the potential to reduce the bottleneck of scarce labeled videos for training and improve performance of supervised behavioral classification models for the study of interaction behaviors in animals.

Fungus-derived protein particles as cell-adhesive matrices for cell-cultivated food.

Cell-adhesive factors mediate adhesion of cells to substrates via peptide motifs such as the Arg-Gly-Asp (RGD) sequence. With the onset of sustainability issues, there is a pressing need to find alternatives to animal-derived cell-adhesive factors, especially for cell-cultivated food applications. In this paper, we show how data mining can be a powerful approach toward identifying fungal-derived cell-adhesive proteins and present a method to isolate and utilize these proteins as extracellular matrices (ECM) to support cell adhesion and culture in 3D. Screening of a protein database for fungal and plant proteins uncovered that ~5.5% of the unique reported proteins contain RGD sequences. A plot of fungi species vs RGD percentage revealed that 98% of the species exhibited an RGD percentage > = 1%. We observed the formation of protein particles in crude extracts isolated from basidiomycete fungi, which could be correlated to their stability towards particle aggregation at different temperatures. These protein particles were incorporated in 3D fiber matrices encapsulating mouse myoblast cells, showing a positive effect on cell alignment. We demonstrated a cell traction stress on the protein particles (from Flammulina velutipes) that was comparable to cells on fibronectin. A snapshot of the RGD-containing proteins in the fungal extracts was obtained by combining SDS-PAGE and mass spectrometry of the peptide fragments obtained by enzymatic cleavage. Therefore, a sustainable source of cell-adhesive proteins is widely available in the fungi kingdom. A method has been developed to identify candidate species and produce cell-adhesive matrices, applicable to the cell-cultivated food and healthcare industries.

Zyxin regulates embryonic stem cell fate by modulating mechanical and biochemical signaling interface.

Biochemical signaling and mechano-transduction are both critical in regulating stem cell fate. How crosstalk between mechanical and biochemical cues influences embryonic development, however, is not extensively investigated. Using a comparative study of focal adhesion constituents between mouse embryonic stem cell (mESC) and their differentiated counterparts, we find while zyxin is lowly expressed in mESCs, its levels increase dramatically during early differentiation. Interestingly, overexpression of zyxin in mESCs suppresses Oct4 and Nanog. Using an integrative biochemical and biophysical approach, we demonstrate involvement of zyxin in regulating pluripotency through actin stress fibres and focal adhesions which are known to modulate cellular traction stress and facilitate substrate rigidity-sensing. YAP signaling is identified as an important biochemical effector of zyxin-induced mechanotransduction. These results provide insights into the role of zyxin in the integration of mechanical and biochemical cues for the regulation of embryonic stem cell fate.

Transcription factor oscillations induce differential gene expressions.

Intracellular protein levels of diverse transcription factors (TFs) vary periodically with time. However, the effects of TF oscillations on gene expression, the primary role of TFs, are poorly understood. In this study, we determined these effects by comparing gene expression levels induced in the presence and in the absence of TF oscillations under same mean intracellular protein level of TF. For all the nonlinear TF transcription kinetics studied, an oscillatory TF is predicted to induce gene expression levels that are distinct from a nonoscillatory TF. The conditions dictating whether TF oscillations induce either higher or lower average gene expression levels were elucidated. Subsequently, the predicted effects from an oscillatory TF, which follows sigmoid transcription kinetics, were applied to demonstrate how oscillatory dynamics provide a mechanism for differential target gene transactivation. Generally, the mean TF concentration at which oscillations occur relative to the promoter binding affinity of a target gene determines whether the gene is up- or downregulated whereas the oscillation amplitude amplifies the magnitude of the differential regulation. Notably, the predicted trends of differential gene expressions induced by oscillatory NF-κB and glucocorticoid receptor match the reported experimental observations. Furthermore, the biological function of p53 oscillations is predicted to prime the cell for death upon DNA damage via differential upregulation of apoptotic genes. Lastly, given N target genes, an oscillatory TF can generate between (N-1) and (2N-1) distinct patterns of differential transactivation. This study provides insights into the mechanism for TF oscillations to induce differential gene expressions, and underscores the importance of TF oscillations in biological regulations.

Integration of a microfluidic multicellular coculture array with machine learning analysis to predict adverse cutaneous drug reactions.

Adverse cutaneous reactions are potentially life-threatening skin side effects caused by drugs administered into the human body. The availability of a human-specific in vitro platform that can prospectively screen drugs and predict this risk is therefore of great importance to drug safety. However, since adverse cutaneous drug reactions are mediated by at least 2 distinct mechanisms, both involving systemic interactions between liver, immune and dermal tissues, existing in vitro skin models have not been able to comprehensively recapitulate these complex, multi-cellular interactions to predict the skin-sensitization potential of drugs. Here, we report a novel in vitro drug screening platform, which comprises a microfluidic multicellular coculture array (MCA) to model different mechanisms-of-action using a collection of simplistic cellular assays. The resultant readouts are then integrated with a machine-learning algorithm to predict the skin sensitizing potential of systemic drugs. The MCA consists of 4 cell culture compartments connected by diffusion microchannels to enable crosstalk between hepatocytes that generate drug metabolites, antigen-presenting cells (APCs) that detect the immunogenicity of the drug metabolites, and keratinocytes and dermal fibroblasts, which collectively determine drug metabolite-induced FasL-mediated apoptosis. A single drug screen using the MCA can simultaneously generate 5 readouts, which are integrated using support vector machine (SVM) and principal component analysis (PCA) to classify and visualize the drugs as skin sensitizers or non-skin sensitizers. The predictive performance of the MCA and SVM classification algorithm is then validated through a pilot screen of 11 drugs labelled by the US Food and Drug Administration (FDA), including 7 skin-sensitizing and 4 non-skin sensitizing drugs, using stratified 4-fold cross-validation (CV) on SVM. The predictive performance of our in vitro model achieves an average of 87.5% accuracy (correct prediction rate), 75% specificity (prediction rate of true negative drugs), and 100% sensitivity (prediction rate of true positive drugs). We then employ the MCA and the SVM training algorithm to prospectively identify the skin-sensitizing likelihood and mechanism-of-action for obeticholic acid (OCA), a farnesoid X receptor (FXR) agonist which has undergone clinical trials for non-alcoholic steatohepatitis (NASH) with well-documented cutaneous side effects.