Lab on a Chip

Hematogenous metastasis is initiated when tumor cells (TCs) intravasate into the vasculature, yet intravasation remains poorly understood because it is difficult to observe in vivo and intravasated TCs are challenging to isolate. To address these challenges, we developed IntravChip, a continuously perfused microfluidic platform containing a vascularized primary tumor microenvironment (TME) enabling the observation of TC intravasation, and a downstream chamber to collect intravasated TCs. The IntravChip can support a high TC concentration in the TME while maintaining complete vascular perfusion, which we found was necessary to collect intravasated cells. Using MDA-MB-231 breast TCs, we identified an optimal initial TC seeding density that, by day 9, yields a densely populated TME and 100-440 collected intravasated TCs. We validated the IntravChip across several TC types, showing that MDA-MB-231 and MV3 TCs have the highest intravasation rates while MCF-7 TCs have low intravasation efficiency. We also show that the IntravChip is compatible with super-resolution nano-imaging. Our devices enabled high-quality STORM imaging, which revealed that H3K9me3 nanodomains are significantly differentially distributed in intravasated MDA-MB-231 tumor cells compared to those residing in the TME. Finally, the IntravChip was validated as a platform to test the effects of anti-cancer drugs on tumor cells and on the vasculature. We showed that a 5 M concentration of sorafenib reduced intravasation events by 69% without impacting the morphology of the microvascular networks (MVNs), while a 10 M concentration led to a significant decrease in vessel diameter. This platform enables quantitative analysis of TC intravasation, collection of intravasated TCs for characterization, and screening of anti-metastatic therapies.

Microwell microfluidics has emerged as powerful platforms for high-precision biological and chemical investigations, bridging microscale fluid handling with compartmentalized reaction environments. Achieving robust and reproducible performance in such studies requires substantial effort to optimize microwell array design. This burden could be markedly alleviated by the availability of a curated database of microwell array parameters. Such a resource would enable the application of machine-learning models for performance prediction and automated design, leveraging knowledge accumulated from prior microfluidics research. However, constructing such a database entails a considerable investment of time and extensive manual curation, as microwell performance is governed by numerous critical design parameters that are reported inconsistently across a broad and largely unstructured body of literature. In this study, we introduce MicrowellMicrofluidicsMiner (M3), a framework that employs large language model (LLM) agents for autonomous knowledge extraction in microwell microfluidics. To evaluate its performance, we curate a ground-truth database and establish an LLM-driven assessment approach. Our results demonstrate that M3 achieves a peak accuracy of approximately 78%, representing more than a twofold improvement over the lowest observed accuracy (32%) obtained using a standalone LLM model (LLAMA 3.1). This study provides a foundational reference for researchers seeking to apply LLM agents to data-driven microfluidics research. The insights presented have the potential to substantially improve how scientists across microfluidics-related disciplines access, interpret, and leverage scientific information, thereby accelerating the development of innovative microfluidic devices and associated discoveries.

Embryo adhesion represents the first step of implantation, yet understanding this process has been hindered by the lack of human in vitro platforms that replicate endometrial physiology. Here, we present a dual-channel microfluidic platform containing organoid-derived endometrial epithelium and primary stromal cells. Our model recapitulates important endometrial hallmarks including epithelial polarization, stromal decidualization, extracellular vesicle release, and hormone-induced receptivity. We tested its function by introducing mouse and human blastocysts and showed that embryos displayed features of initial adhesion. These included establishment of embryo-epithelial contacts initiated via the polar trophectoderm, inner cell mass repositioning, and lineage segregation. Moreover, human embryos secreted {beta}hCG indicating a functional trophoblast. Thus, this work provides a platform to study key features of embryo adhesion and endometrial receptivity, and disorders affecting embryo-endometrium interactions. TeaserEndometrium-on-a-chip shows detailed human embryo adhesion dynamics.

Extracellular vesicles (EVs) carry molecular cargo that can reflect the real-time state of parental cells, yet most in vitro EV analyses rely on bulk approaches and therefore average over pronounced heterogeneity in both cell and EV populations. Here, we present a semi-open microfluidic platform that enables multi-marker profiling of EVs released from single-cell-derived clones, allowing EV signatures to be linked to clonal progeny originating from a single parental cell. The platform integrates aligned cell and EV arrays containing 17,305 wells, assembled with a 3D-printed housing to capture released EVs in one-to-one matched wells. Captured EVs are immunolabeled for canonical tetraspanin markers (CD9, CD63, CD81) and EpCAM, imaged by high-resolution fluorescence microscopy, and quantified using an automated image-analysis pipeline. Applying the platform to single-cell-derived PC3 clones revealed substantial heterogeneity in EV marker co-expression, with hierarchical clustering identifying four distinct tetraspanin co-expression profiles. The fraction of EpCAM-positive EVs increased with PC3 cell proliferation, as assessed by endpoint cell number, whereas free (non-EV-associated) EpCAM showed no correlation. This platform enables near single-EV-level, multi-marker profiling from single-cell lineages and provides a practical approach to simultaneously dissect both cellular and EV heterogeneity.

Currently in vitro models of microvascular biology rely on self-assembly of vascular cells in compatible gels. However, the stochastic nature of this process results in large variations in lumen sizes, perfusion continuity, and shear stresses making systematic and reproducible analysis challenging. Here, we report a new technology to generate artificial capillaries on a chip with custom control over lumen sizes and architectures using a combination of femtosecond laser cavitation and collagen casting within multi-chambered microfluidic chips. The design allows seeding of endothelial cells within capillary-sized microchannels and seeding of stromal cells within top-open silos, with independent control over seeding sequence and media compositions. Results show that endothelialized microchannels, coined as artificial capillaries, exhibit excellent barrier function with reproducible control over lumen sizes ({phi}=8-35{micro}m) and their architectures (straight, curvatures, tapered, branched). The physical flow parameters measured across the lumen (namely, flow shear) and at the channel outlets (flow velocities) have been validated against high-fidelity numerical assessments from the Large Eddy Simulation scheme within the digitized versions of the microchannels. The experiment-computation compatibility enabled us to predict changes in regional velocity and wall shear stresses within artificial capillaries, for various capillary architectures. We also show that in situ editing of artificial capillaries in the form of adding new branches or adding occlusions is possible. Lastly, we developed a co-culture model which enables the study of stromal cells with artificial capillaries using conventional imaging methods. We envision that acellular chips with two seeding ports can be readily shipped worldwide and could potentially be adopted as a new technology to study microvascular biology in a reproducible manner.

Microphysiological systems (MPS) are essential for modeling tissue barriers, yet integrating electrical readouts often requires permanently sealed microfluidic architectures that limit access to open-well (direct-access) workflows used in bioscience laboratories. To resolve this issue, we present a modular approach in which functional components are added and removed from a standard MPS core using a magnetic interface. This design preserves compatibility with established open-well protocols for seeding and downstream analysis, while microfluidic perfusion or electrical sensing capabilities are added only when needed. We demonstrate this approach with an impedance-sensing module that enables continuous impedance measurements to assess barrier function. By fitting spectra to an equivalent circuit model, we quantify junctional and non-junctional electrical contributions to barrier integrity over time, alongside conventional single-frequency TEER, and complementary permeability and imaging readouts. We apply this platform across three representative use cases, including LPS-induced disruption, shear stress-mediated strengthening, and compatibility with barrier models formed above a 3D hydrogel matrix.

Polymeric 3D printing of microfluidic devices for biosensing is an appealing fabrication alternative for rapid manufacturing of biosensing devices with complex geometry in a streamlined, repeatable and cost-effective manner without the need for expensive instrumentation such as those employed in photochemical etching and soft lithography. Hybrid 3D printed paper-based microfluidics is an emerging area which harnesses the unique properties of both, merging the construction of microfluidic structures and the inherent capillary-driven flow within paper substrates. In this work, we have fabricated hydrophobic barriers by 3D printing a single layer of machinable wax, thermoplastic polyurethane, polylactic acid and polypropylene directly on chromatography paper to create open microchannels and determine the most suitable material. Characterization of each open microchannel using the four materials revealed polypropylene as the most reliable material with high hydrophobic barrier integrity and resolution. Polypropylene achieved functional microchannels with a resolution of 621 {+/-} 33{micro}m, hydrophobic barrier integrity of (93.75 {+/-} 9.16%), wicking speed of 0.38mm/s and optimal hydrophilicity of channels (51.4 {+/-} 8.36 {degrees}) with minimal embedding during thermal curing. To demonstrate proof of principle, a fluorescence assay demonstrating the formation of a dimeric g-quadruplex structure from a g-rich sequence which significantly enhances fluorescence of thioflavin T was implemented.

Neurons in the brain are organized and connected into complex networks in which electrochemical signaling forms the basis for all brain function. Cortical neuronal net-works are arranged in distinct modular, layered, and hierarchical structures, underlying its diverse functions such as learning, memory, or vision. Modern biotechnology has enabled an array of techniques to culture human neural cells, ranging from discreet co-cultures to complex developmental organoids, but all of which almost exclusively form unstructured and hypersynchronous networks. Overcoming this and capturing the functional and anatomical properties of the brain in vitro has proven to be a great challenge. Current techniques for guiding neuronal connectivity in vitro is often limited to a small fraction of the total population of neural cells, leaving most of the culture effectively unguided. To provide large-scale guidance of neurons in culture, we developed a microtunnel device which allows large-scale cell entry through an array of perforations, and guides neuronal network formation through a series of tunnels. Human neural stem cells capable of forming extensive neuronal projections were used to investigate several different microtunnel designs. One particularity noteworthy design which produced predominantly unidirectional growth was used to successfully validate its effect on propagation of neural activity on microelectrode arrays. Serendipitously, we found that our microtunnels had an extraordinary effect on signal-to-noise ratio and the quality of electrophysiological recordings with regards to number of active channels and detected spikes. Since we often found the neuronal growth surprising, we developed a simple computer model which could reproduce neuronal growth in the various tunnels, allowing computer aided design (CAD) of future projects.

Extracellular vesicles (EVs) are lipid spheres released from cells. Research utilizing EVs has met several hurdles owing to the small size of the majority of EVs and other nanoparticles (<150 nm) and the lack of detection technologies capable of providing high-throughput single particle measurements at this scale. The use of high-throughput single particle measurements is critical for the assessment of EV heterogeneity and abundance which are features often used to assess the development of isolation protocols or particle characterization. The Coulter principle, known in the field as resistive pulse sensing (RPS), has been used for several decades to size and count cells. More recently, this technology has evolved to accommodate nanoparticle analysis. In the last decade a platform utilizing microfluidic resistive pulse sensing (MRPS) has been demonstrated for nanoparticles, offering ergonomic characterization of nanoparticles along with utilizing open format data. To date, assessment of MRPS accuracy and reporting standards have not been assessed. With the aim of increasing data accuracy, ergonomics, and reporting transparency, we developed a microfluidic resistive pulse sensing post-acquisition analysis software (RPSPASS) application for automated cohort calibration, population gating, statistical output, QC plot generation, alternative data file outputs, and standardized reporting templates.

The fallopian tube serves as a sperm reservoir, and it is the site where the oocytes become fertilized. Here, we describe development of an organ-on-a-chip microfluidic model of the fallopian tube (FT Chip) lined by primary human epithelial cells and stromal fibroblasts derived from the FT ampulla. Abundant tissue folds lined by hormone-responsive, epithelial cells resembling those seen in vivo formed on-chip, but not in epithelial organoids cultured in gel cultures. Comparative time-resolved analysis of human sperm versus oocyte-sized microparticles introduced into the epithelial channel in the presence of estradiol revealed that sperm movement was significantly reduced, while the oocyte-sized particles increased, relative to movements in acellular chips. When the non-hormonal contraceptive TDI-11861 was administered to the chip, dose-dependent inhibition of human sperm motility was detected. Thus, this FT Chip may offer a human preclinical tool to study FT physiology and assess the efficacy and mechanism of action of contraceptives.

1.Adipocyte size is an independent predictor of several metabolic disorders, including type 2 diabetes, liver and cardiovascular diseases. However, technical limitations due to the fragile nature of mature adipocytes have restricted the functional analyses of size-separated adipocytes using conventional methods. Therefore, we have developed a microfluidic device, based on deterministic lateral displacement, for sorting intact, mature adipocytes. Cell-size distribution was determined from time-lapse recordings inside the device, in separate outlets, and by Coulter counter analysis of the collected cell fractions. This approach allowed size-separation with minimal size-overlap with mean diameters of (small fraction) 47 {micro}m and (large fraction) 82 {micro}m based on Coulter counter measurements. Viability of the separated cells was verified by insulin stimulation and western blotting of key insulin signaling proteins. The sample recovery, comparing input versus output material, was relatively high, 42% for the large fraction with a purity of 93%. We demonstrate that microfluidics is a suitable approach to overcome the limitations of sorting mature adipocytes according to size. Together, the high recovery rate, high throughput capacity, accurate separation and the fact that the cells maintained hormonal response after sorting provides compelling evidence of the strength and usability of the microfluidic approach for exploring adipocyte function in relation to size.

Organ chips offer a disruptive innovation to study human diseases with tissue-specific resolution within a predictable and tunable in vitro environment. However, these platform technologies have for the most part failed to translate to broad use in the private sector due to a lack of high-throughput, user-friendly platforms. Here we present an automated high-throughput organ chip seeded with iPSC-derived cardiomyocytes transduced with GCaMP6f and interface with translational technologies to bridge the current academia-industry gap. Cardiomyocytes were seeded on-chip fully hands-free using an entry-level fluid handling robot to significantly reduce user handling requirements. Pipette interfaces were paramount to facilitating seeding and feeding through improved tolerances for establishing a functional connection to dispense and collect small fluidic volumes. Following successful seeding, GCaMP6f activity on-chip was monitored with our automated, non-invasive fiber-optic sensing platform. We show a significant decrease in cardiomyocyte beat rate in response to decreased ambient culture temperature using data collected with our optical sensing platform. This study provides a potential translational blueprint for academia-industry partnership toward broad adoption of organ chip technology in drug development and disease modeling.

Organ-on-chip (OoC) platforms are increasingly adopted for predictive in vitro testing. However, most remain limited by soft-lithography-derived 2.5D microfluidic architectures and non-physiological rigid materials, or bioprinting approaches that require complex and failure-prone post-fabrication assembly. Here, we present a versatile approach that integrates tomographic volumetric additive manufacturing (TVAM) directly within preassembled microfluidic chips, enabling rapid, contactless fabrication of freeform 3D OoCs. Leveraging our open-source optical simulation framework, Dr.TVAM, we perform TVAM in custom-designed chips, eliminating post-printing manual assembly steps that commonly lead to leakage, contamination, and poor reproducibility. This strategy, termed TVAM-in-a-chip, supports the generation of diverse 3D channel architectures in multiple biocompatible photoresins spanning a wide range of chemistries and mechanical properties, including cell-laden formulations. We demonstrate multi-channel designs, compatibility with confocal imaging, and dynamic culture of epithelial and endothelial models. Overall, TVAM-in-a-chip overcomes key limitations of current OoC technologies and paves the way for a new generation of scalable, biomimetic 3D platforms for advanced in vitro modeling.

A systematic optimization of throughput and operational stability in Sorting by Interfacial Tension (SIFT) is presented. Reducing droplet size and enabling a broader distribution of droplet trajectories increased the number of droplets processed per sorting element, resulting in about a four fold improvement in throughput from 30 to 125 droplets per second. Throughput was further enhanced through device parallelization, with devices incorporating two and four independent sorting regions demonstrated. These configurations distributed droplets evenly across sorting elements that exhibited comparable pH sorting thresholds, indicating similar flow conditions and drag forces within each region. Among the designs evaluated, the two-element configuration provided the optimal balance of throughput, robustness, and simplicity, achieving maximum throughputs of about 250 droplets per second. Throughput and pH sorting thresholds were preserved throughout two hours of continuous sorting. The improved platform was applied to examine the relationship between cellular glycolysis and iron homeostasis at the single-cell level for Jurkat cells, revealing a subpopulation of highly glycolytic cells with significantly elevated iron uptake, consistent with prior reports linking iron regulation and T cell metabolism. Collectively, these advances expand the scale, stability, and biological applicability of SIFT, enabling large-scale functional studies while facilitating the capture of rare and metabolically distinct cell populations.

Small extracellular vesicles (sEV) are membrane-enclosed nanoparticles found in body fluids that carry molecular cargo from their cells of origin. Their stability and disease-associated molecular signatures make them promising targets for the development of non-, or minimally invasive liquid biopsies, yet scalable approaches enabling single-vesicle quantification of sEV while resolving their heterogeneity remain limited. Here, we present PICO (Protein Interaction Coupling), a reference-free quantitative assay adapted for sensitive multiplex profiling of individual intact vesicles. PICO detects vesicle markers by requiring colocalization of two or more copies of the same protein or of distinct proteins (e.g., CD9 or CD9/CD63) on individual vesicles, using DNA-barcoded antibodies and digital PCR (dPCR) for quantitative readout. We demonstrate that this unique architecture of the assay provides high specificity by distinguishing EV-bound proteins from soluble counterparts, and can be adapted to target either surface-exposed or intravesicular biomarkers. PICO requires minimal sample input (1 {micro}l) and no specialized instrumentation beyond standard digital PCR. In a head-to-head comparison with nano-flow cytometry, PICO achieved a comparable limit of detection for sEV subpopulations. Profiling sEV isolates from blood for canonical markers (CD9, CD63 and CD81) and HER2 demonstrates precise, high-resolution quantification of sEV subpopulations in complex clinical samples and supports integration of scalable single-EV analysis into research and diagnostic workflows.

Elastic porous membranes are essential components of mechanically active organ-on-a-chip and microphysiological system (MPS) platforms, where cyclic strain is required to recapitulate physiologically relevant tissue mechanics. However, existing fabrication methods are often difficult to reproduce, low throughput, or dependent on specialized infrastructure, limiting their adoption across laboratories. Many protocols also lack quality control steps for ensuring device assembling and reproducibility. In this paper, we present a robust and accessible fabrication and quality control workflow for the consistent production of elastic porous PDMS membranes. The method uses commercially available heat presses, release liners, and pre-patterned membrane wafers to enable rapid membrane molding. We describe a quality control framework, including visual verification of porous regions and wettability testing for surface activation, to ensure irreversible PDMS bonding and reliable device assembly. Together, this workflow improves fabrication yield, reduces device failure, and supports reproducible implementation of elastic porous membrane in organ-on-a-chip applications.

Urinary tract infections (UTIs) remain a major health burden, yet mechanistic studies are limited by the lack of experimental models that enable high spatiotemporal resolution tracking of infection dynamics, while recapitulating the stratified architecture of the bladder epithelium, urine tolerance and fluid dynamics. Here, we present a modular microphysiological platform integrating a fully stratified, urine-tolerant human urothelium cultured on standard transwell inserts within a custom-designed perfusion device compatible with live imaging. Urine flow enables real-time, high-resolution imaging of uropathogenic Escherichia coli (UPEC) infections under physiologically relevant conditions, including clearance of planktonic bacteria and nutrients replenishment, while retaining tissue-associated populations. This system revealed UPEC attachment via the type 1 fimbrial adhesin FimH and its inhibition by D-mannose treatment. Moreover, the platform captured L-form formation upon treatment with the frontline antibiotic fosfomycin and regrowth of walled bacteria following drug withdrawal. The platform further uncovered strain-specific lysis through bacteriophages in contrast to the activity of broad-spectrum antibiotics. In summary, this system constitutes a scalable platform with high predictive power for studying UTI pathogenesis and preclinical therapeutic testing.

Precise diagnosis of high-risk conditions such as cardiovascular and cerebrovascular diseases still remains a challenge. We previously developed a microfluidic chip for sperm selection and observed that sperm motility is highly sensitive to environmental changes. Building on this finding, we hypothesized that motility traits of sperm could be differentially modulated by body fluids from healthy versus diseased individuals, thereby serving as potential biomarkers for disease diagnosis. To test this hypothesis, we designed a diagnostic system in which mouse sperm were co-incubated with serum samples from patients with myocardial infarction, cerebral infarction, and pancreatitis, along with matched healthy controls. Key kinematic parameters--including motility rate (MR), curvilinear velocity (VCL), straight-line velocity (VSL), linearity (LIN), and amplitude of lateral head displacement (ALH)--were analyzed using a multiparameter sperm quality analysis system. The results revealed that disease-specific serum induced distinct and reproducible changes in sperm motility patterns, enabling accurate discrimination between healthy and pathological conditions. Evaluation of these motility parameters demonstrated high diagnostic performance, with area under the receiver operating characteristic curve (AUC) values ranging from 0.719 to 0.888. This sperm-based bioassay offers a non-invasive, rapid, and cost-effective platform for disease detection and personalized health assessment, with the potential to complement existing diagnostic approaches.

Microfluidic devices with surface-bound biomolecular patterns enable localised detection arrays, enzymatic catalysis, and gene expression. Photolithography is a contactless patterning method with high spatial control. However, while patterning open surfaces by photolithography is well-established, patterning enclosed microfluidic channels remains technically challenging. Such capability would enable in situ surface modification and precise pattern alignment to channel geometries. Here, we present a photolithographic method using commercially available reagents to pattern sealed microfluidic devices. We first coat surfaces with (3-Aminopropyl)triethoxysilane (APTES) to bond microfluidic chips and provide surface amine groups onto which photocleavable polyethylene glycol (PC PEG) compounds are bound. UV exposure using standard photolithography equipment selectively deprotects the amine groups, which can subsequently bind amine-reactive cargos. We demonstrate this methods versatility by patterning both glass and poly(dimethylsiloxane) (PDMS) surfaces with diverse cargoes: DNA, proteins and gold nanoparticles. We also compare covalent versus noncovalent DNA patterning. Covalently bound DNA patterns were denser and could be used for sequence-specific target DNA capture. However, noncovalently bound DNA yielded higher cell-free gene expression from surface-bound GFP templates.

Metastasis is the leading cause of death in breast cancer patients, yet there are no drugs specifically designed to block cancer cell intravasation, an early step of the metastatic cascade that originates circulating tumour cells (CTCs). A major challenge in developing anti-intravasation drugs is the scarcity of relevant in vitro platforms suitable for predictable drug discovery. Intravasation is a fundamental step of metastasis and involves the crossing of cancer cells through an endothelial barrier to enter the blood circulation. Here we developed an intravasation-on-a-chip model with controlled extracellular matrix composition, fluid flow and shear stress, which mimics the dynamic tumour-endothelium interface. The systems allows real-time imaging of intravasation and the isolation and quantification of intravasated cancer cells. As a proof-of-concept for drug testing, we show that perfusion with the PI3K/mTOR inhibitor Dactolisib, significantly reduced intravasation without compromising endothelial cell viability. The system also provides the capability to evaluate inhibitor on-target activity via imaging analysis. This intravasation-on-a-chip model offers a powerful, scalable, and imaging-compatible platform for discovering and evaluating anti-intravasation compounds.