The research team, led by Min-Ji Kim and Kyung-Sun Kang of Seoul National University together with Seunghee Lee, created what they call vascularized skin assembloids, or VSAs. The strategy builds on two earlier stem-cell technologies. Skin organoids grown from human induced pluripotent stem cells can recapitulate epidermal and dermal differentiation and even generate appendages such as hair follicles, melanocytes and neurons. Blood vessel organoids, likewise derived from pluripotent stem cells, self-assemble into endothelial networks with lumina and mural support cells. Each system on its own, however, falls short of real skin. Skin organoids often develop inverted polarity, with the epidermis facing inward, and can undergo ectopic cartilage formation, while neither model on its own provides the vascular and immune components needed to model inflammation.

The assembloid approach sidesteps these problems by treating the two organoid types as building blocks rather than finished products. The researchers first optimized the extracellular matrix in which the blood vessel organoids were embedded, testing different ratios of Matrigel to collagen. Expression of the endothelial genes PECAM1 and CDH5, markers of vascular density and junctional organization, peaked at a one-to-three Matrigel-to-collagen ratio, while alpha-SMA, a marker of aberrant mural-cell activation, rose in collagen-rich matrices. With this optimized composition in hand, the team cut mature, cyst-like skin organoids into eight pieces each and placed them onto pre-vascularized matrix, then cultured the combined tissues at an air–liquid interface for several weeks.

Over 28 days, the fused structures underwent a striking transformation. Initially the skin organoid fragments and vascular matrix retained distinct boundaries, but within a week the skin tissue spread across the vascularized surface. By day 15 the construct had thickened and consolidated, and by day 28 branched CD31-positive vascular networks had arborized extensively through the dermal compartment. Gene-expression analysis tracked this maturation, with PECAM1 and CDH5 rising steadily while the barrier genes filaggrin and loricrin were significantly upregulated by the end of the culture period. Three-dimensional reconstructions showed that the vascular networks maintained relatively uniform vessel diameters, and hair-follicle-like structures were frequently found sitting adjacent to vessels, echoing the intimate spatial relationship between skin appendages and blood supply in native tissue. The vascularized constructs also secreted more vascular endothelial growth factor and nitric oxide than non-vascularized controls, indicating genuine endothelial activity rather than passive decoration.

Single-cell RNA sequencing confirmed that the assembloids had gained cellular complexity, not just structural features. Compared with conventional skin organoids, VSAs contained increased proportions of cells expressing endothelial markers such as PECAM1, CDH5, KDR, VWF and CLDN5, as well as perivascular markers including PDGFRB, RGS5 and ACTA2. Enrichment analysis of genes upregulated in the assembloids pointed to programs of endothelial proliferation and migration, vasculature development, angiogenesis, basement membrane organization and cytokine response. Histology reinforced the picture: the assembloids developed a defined KRT5-positive basal layer and KRT10-positive suprabasal layer, expressed filaggrin and loricrin more prominently than ordinary skin organoids, and restricted penetration of Lucifer Yellow dye largely to the outer epidermis, evidence of a more mature barrier. When transplanted onto wounds in immunodeficient mice, the grafts persisted for four weeks, and their human CD31-positive vessels connected with mouse host vasculature and became accessible to an intravenously injected lectin, suggesting the engineered vessels can be incorporated into a living circulatory system.

The most consequential addition, however, was immunological. The team incorporated mononuclear cells isolated from umbilical cord blood into the assembloids and challenged the system with lipopolysaccharide, a bacterial endotoxin. In vascularized constructs, the immune cells migrated into the upper dermis and clustered near CD31-positive vascular structures, whereas in matched avascular organoids the infiltrating cells rarely progressed beyond the lower dermis. Molecular readouts told the same story: after LPS stimulation, vascularized assembloids with immune cells showed markedly higher expression of the adhesion molecules ICAM1 and VCAM1 and the inflammatory cytokines IL-6 and TNF-alpha than any other condition, with ICAM1 upregulation localized to the endothelial structures themselves. In other words, the vasculature was not merely a passive conduit but an active participant in recruiting and activating immune cells, exactly as it does in inflamed human skin.

With an immune-responsive, vascularized skin model in hand, the researchers turned to atopic dermatitis. They applied house dust mite extract, a major environmental allergen, to the epidermal surface of the assembloids in three treatments spaced two days apart. The response was dramatic and reproducible. Transcriptomic profiling separated treated and control samples cleanly, revealing activation of inflammatory, cornification and type 2 immune-regulation programs, along with TNF and IL-17 signalling pathways. Crucially, when the team compared their treated assembloids with published gene-expression data from skin biopsies of atopic dermatitis patients, the patterns partially paralleled each other: both showed elevated type 2 and alarmin-associated genes such as IL33, IL4R, CCL17 and CCL26, keratinocyte stress markers including KRT16, S100A8 and S100A9, and adhesion and infiltration mediators such as CXCL8 and ICAM1. At the tissue level, allergen exposure thickened the epidermis, reduced filaggrin, raised KRT16 and drove immune cells along vascular structures into the upper dermis. Flow cytometry identified CD4-positive and CD8-positive T cells, CD14-positive myeloid cells and CD11c-positive HLA-DR-positive antigen-presenting-like populations, with a rightward shift in HLA-DR intensity suggesting myeloid activation. The treated assembloids also secreted elevated levels of TSLP, CCL17, IL-4, IL-13, VEGF and nitric oxide.

The final test was pharmacological, and here the model passed with a precision that will excite drug developers. The team applied upadacitinib, a JAK1 inhibitor, and dupilumab, an antibody that blocks the IL-4 receptor alpha subunit, two drugs with proven efficacy in moderate-to-severe atopic dermatitis. Both agents reduced allergen-induced epidermal hyperplasia, restored expression of the barrier proteins filaggrin, loricrin and claudin-1, suppressed TSLP and KRT16, and reduced the infiltration of CD4-positive T cells into the upper dermis. Mechanistically, house dust mite exposure strongly increased phosphorylation of STAT6, the central transcription factor downstream of IL-4 and IL-13 signalling, and both drugs significantly diminished this activation, along with the expression of STAT6 target genes including TARC, MDC, CCL26 and POSTN. Cytokine profiling showed broad suppression of the allergen-induced mediators IL-33, ST2, IL-4, IL-13, CCL17, RANTES, VCAM1 and VEGF. The responses matched the known mechanisms of action of both therapeutics, which is precisely what a preclinical screening platform should do.

The authors are candid about the limits of their system. The vascular networks, while organized and associated with basement membrane and perivascular cells, do not yet form a fully mature, perfusable artery-vein-capillary hierarchy, because the static air–liquid culture lacks blood flow and haemodynamic cues; integrating microfluidic perfusion could change that. Definitive lymphatic vessels were not observed, the immune compartment derives from cord blood rather than skin-resident cells and lacks bona fide Langerhans cells and tissue-resident memory T cells, and the allergen model captures an acute response rather than the chronic, relapsing reality of clinical atopic dermatitis. Even so, the achievement is substantial: a modular, human-cell-derived skin model that assembles epithelium, dermis, vessels and immune cells in a tunable way, reproduces key molecular features of a common allergic disease, and responds to real drugs in a mechanism-consistent manner. Beyond disease modelling, the preformed vascular networks suggest future roles in regenerative medicine, from wound healing to bioengineered skin grafts, where vascularization remains the central bottleneck. For a field long forced to choose between oversimplified dishes and imperfect animal models, vascularized skin assembloids offer a third option that looks increasingly like the real thing.

Subject of Research: Vascularized, immune-integrated human skin assembloids for modelling atopic dermatitis and drug testing in vitro

Article Title: Vascularized human skin assembloids model compartmental skin organization and immune responsiveness in vitro

Article References: Kim, M.-J., Lee, S., & Kang, K.-S. (2026). Vascularized human skin assembloids model compartmental skin organization and immune responsiveness in vitro. Experimental & Molecular Medicine. https://doi.org/10.1038/s12276-026-01840-x

Image Credits: AI Generated

DOI: 10.1038/s12276-026-01840-x

Keywords: skin assembloids, organoids, atopic dermatitis, vascularization, immune cells, house dust mite, upadacitinib, dupilumab, STAT6 signalling, induced pluripotent stem cells, skin barrier, inflammatory skin disease

The scale of the problem is well documented. Pre-clinical animal models have proven poor predictors of human toxicities in phase 1 oncology trials, and quantitative proteomics has revealed substantial differences between humans and laboratory species in the abundance of non-cytochrome P450 drug-metabolizing enzymes. Species divergence extends deep into immunology, where functional differences between rodent and human PD-1 have been linked to evolutionary divergence, and into catastrophic safety failures such as the TGN1412 monoclonal antibody trial, which was later explained by species differences in CD28 expression on human CD4-positive effector memory T-cells. Because absorption, distribution, metabolism, excretion and toxicity, collectively known as ADMET, arise from coordinated activities across multiple organs rather than from any single tissue, even the best single-organ cultures cannot capture the systemic picture. A drug that is harmless in an isolated liver culture may become toxic only after intestinal metabolism, hepatic processing and renal clearance interact in sequence.

The Review identifies two persistent challenges that have kept multi-organ chips in the proof-of-concept stage. The first is how to preserve the distinct biochemical and biomechanical niches of each organ while enabling controlled exchange between them. The liver, for example, depends on a specific extracellular matrix composition, oxygen gradient and shear stress regime, while the gut requires peristalsis-like motion, a mucus layer and a resident microbial community. The second challenge is manufacturing: most existing devices are hand-built, low-throughput and difficult to reproduce, which makes them unsuitable for the standardized, data-rich drug evaluation pipelines that regulators and pharmaceutical companies require. The authors organize their analysis around these two axes, first surveying niche-preserving engineering strategies developed within single-organ systems and then outlining how those components can be integrated into scalable, multimodal platforms.

Niche preservation begins with materials. Biological hydrogels act as selective diffusion barriers in living tissue, and the Review highlights how engineered hydrogels can replicate this function, controlling the release of growth factors through synthetic glycosaminoglycans in modular macroporous scaffolds. Growth factor presentation matters as much as its presence: engineered variants with enhanced syndecan binding generate tonic signalling that promotes tissue healing, and superior extracellular matrix binding motifs have been shown to enhance the regenerative activity and safety of therapeutic proteins. Spatial control is equally critical, with techniques now available for patterning multiple growth factors simultaneously in three-dimensional hydrogels and for photopatterning biomolecule immobilization to guide three-dimensional cell fate in protein-based hydrogels. The mechanical properties of the matrix, including its viscoelasticity, are increasingly recognized as central regulators of cellular behaviour, meaning that a chip’s scaffold must be tuned not only chemically but physically to match each organ’s native stiffness and stress-relaxation profile.

Membranes and barriers form a second pillar of niche engineering. Organ-on-chip devices rely on polymeric and biological membranes to separate tissue compartments while permitting controlled molecular traffic, and recent work has produced trumpet-shaped porous PDMS membranes, extracellular matrix-derived membranes synthesized in situ with tunable thickness, and hydrogel membranes formed directly in microchannels to generate stable chemical gradients. These advances matter because barrier function is organ-defining: the intestinal epithelium decides what enters the bloodstream, the blood-brain barrier decides what enters the central nervous system, and the renal tubular barrier decides what is excreted. A validated Caco-2 microfluidic chip model has already demonstrated the ability to predict intestinal absorption of BCS class I through IV drugs, and ultrathin membranes have been shown to mediate tissue-specific morphogenesis and barrier function in a human kidney chip. Microfluidic gradient devices have even been used to generate spatially patterned human neural tube-like structures, showing that developmental signalling landscapes can be reconstructed on chip with remarkable fidelity.

Vascularization represents perhaps the most demanding niche to engineer, and also the most important for ADMET prediction, because the vasculature is the highway along which every drug travels. The Review emphasizes that vascular endothelium is not a uniform pipe but a heterogeneous, organ-adapted tissue: tight junction claudin composition varies across vascular beds and guides organotropic metastasis, the glycocalyx serves as a central regulator of vascular function, and endothelial cells in different organs perform strikingly different tasks. Pericytes, which wrap around capillaries and modulate barrier properties, are increasingly recognized as essential players in disease models. Engineering efforts have produced functional perfusable three-dimensional microvascular networks on chip, microfluidic platforms integrating functional vascularized organoids, and open-top chips generating two distinct interconnected three-dimensional microvascular networks without barriers. Physiological perfusion itself has been shown to rescue regressed microvascular networks and extend their longevity, and recent preprint work links intraluminal flow to endothelial state transitions through a YAP/TAZ-Apelin switch, underscoring how fluid forces actively sculpt vascular identity rather than merely nourishing it.

With single-organ niches increasingly well engineered, the field has moved toward integration. Landmark demonstrations include fluidically coupled vascularized organ chips that quantitatively predicted human pharmacokinetic responses to drugs, a physiomimetic model integrating gut, liver and brain systems for studies of neurodegenerative disease, and a multi-organ chip with matured tissue niches linked by vascular flow. The ambition keeps growing: an eighteen-organ microphysiological system coupling a vascular network and excretion system has been reported for drug discovery, alongside pumpless designs that use gravity-driven flow to avoid external pumps, including a fourteen-compartment system and modular platforms for co-culturing gastrointestinal epithelium with three-dimensional primary liver tissue. Robotic fluidic coupling and interrogation of multiple vascularized organ chips has automated what was once delicate manual work, and closed-loop modular platforms now achieve self-sustaining, tightly controlled oxygenation. Four-organ chips connecting intestine, liver, skin and kidney equivalents have supported long-term co-culture, and multi-organoid body-on-a-chip systems have enabled compound screening in both single and integrated configurations.

Scalable manufacturing is where the Review sees the decisive bottleneck. Microfabrication techniques such as ultra-resolution scalable microprinting and continuous liquid interface production, which achieves single-digit-micrometer resolution and can operate in injection mode, promise to move chip production from artisanal to industrial. Adaptable manufacturing of milliscale organ chips with perfusable vascular beds, scalable microfluidic platforms for flexible configuration of microtissue multiorgan models, and three-dimensional printed multi-compartment platforms with tubing-free pumps all point toward platforms that could be produced, configured and run reproducibly across laboratories. Material choices matter here too: systematic studies of sorption and release of small molecules in PDMS and cyclic olefin copolymer are helping engineers select polymers that do not sequester the very drugs being tested, a subtle but consequential source of error in pharmacokinetic measurements.

The final ingredient is data. The Review highlights multimodal readouts as the basis for making mOoCs robust, reproducible and data-rich technologies for drug response evaluation. Optical methods now span label-free refractive index tomography, holotomography, hyperspectral CARS microscopy for chemical imaging, dynamic full-field optical coherence tomography for live imaging of retinal organoids, and photoacoustic techniques capable of volumetric imaging of neuromelanin in live brain organoids and all-optical three-dimensional scanning of vasculature. Electrical readouts include impedance mapping with high-density microelectrode arrays that reveal dynamic heterogeneity of epithelial barriers, and microelectrode array-integrated cantilever platforms that compare real-time electrophysiology and contractility of cardiomyocytes under drug exposure. Biosensors can now selectively detect dopamine and serotonin at nanomolar concentrations from complex in vitro systems, while nanoplasmonic microwell arrays monitor single-cell secretion with spatiotemporal resolution. These streams of data feed directly into artificial intelligence workflows: AI models trained on liver organoid brightfield images already predict drug-induced liver injury levels, machine learning strategies reduce variability in stem cell differentiation, and active learning frameworks leveraging transcriptomics identify modulators of disease phenotypes.

The regulatory and cultural context is shifting rapidly in favour of these technologies. The US Food and Drug Administration has formalized its support for new approach methodologies, the National Institutes of Health has announced a priority on human-based research technologies, and major funders and institutions are moving away from primate research. The authors argue that the human-relevant datasets generated by niche-preserving, vascularly linked, multimodally monitored multi-organ chips are exactly what next-generation AI-driven drug discovery needs: physiologically grounded training data that reflect how a compound behaves in a human body rather than in a mouse. If the engineering challenges of scalability and standardization can be met, the multi-organ chip may finally close the gap between a promising molecule in a dish and a safe, effective medicine in a patient, replacing a century of imperfect animal surrogates with a living, breathing model of ourselves, built one microfluidic channel at a time.

Subject of Research: Multi-organ-on-a-chip systems engineered to preserve organ-specific niches for predicting human drug absorption, distribution, metabolism, excretion and toxicity

Article Title: Engineering human multi-organ tissue chip niches for drug absorption, distribution, metabolism, excretion and toxicity prediction

Article References: Li, J., Madrigal Gamboa, S., Chao, J. T., Heilshorn, S. C., & Wu, J. C. (2026). Engineering human multi-organ tissue chip niches for drug absorption, distribution, metabolism, excretion and toxicity prediction. Nature Biomedical Engineering. https://doi.org/10.1038/s41551-026-01806-4

Image Credits: AI Generated

DOI: 10.1038/s41551-026-01806-4

Keywords: organ-on-a-chip, multi-organ chips, ADMET, drug toxicity prediction, microfluidics, tissue engineering, new approach methodologies, vascularization, organoids, pharmacokinetics, artificial intelligence, drug development

The focus of the investigation was the lipid droplet, a dynamic intracellular organelle consisting of a core of neutral lipids such as triglycerides and cholesteryl esters wrapped in a phospholipid monolayer. Lipid droplets bud off from the endoplasmic reticulum with the help of proteins like seipin and FIT2, and their size and turnover are managed by an array of peripheral and transmembrane proteins. Although lipid droplets are best known as energy reserves in fat cells, they also serve as protective buffers, sequestering potentially toxic lipids and shielding cells from oxidative stress. In the brain, their role has become a hot topic, particularly because lipid droplet accumulation in microglia, the resident immune cells of the central nervous system, has been linked to aging and to neurodegenerative diseases such as Alzheimer’s and Parkinson’s.

Previous work had left the field divided. One influential study introduced the term lipid droplet accumulating microglia, or LDAM, and showed that these fat-laden immune cells produce reactive oxygen species and perform phagocytosis poorly, suggesting they drive neurodegeneration. Other research pointed in the opposite direction, proposing that transferring lipid droplets from neurons to glial cells is a protective response that spares neurons from lipotoxicity. The new study adds crucial comparative evidence to this debate by examining a species that ages remarkably slowly. The researchers used coronal brain sections from naked mole rats aged about four and thirteen years, corresponding to young and middle-aged mice of twenty-six and fifty-two to fifty-four weeks, and stained the tissue with the neutral lipid probe BODIPY 493/503 together with markers for microglia and astrocytes. To confirm that the fluorescent signal truly came from lipids, they also performed Stimulated Raman Scattering imaging, a label-free technique that targets the methylene vibrational band of lipid acyl chains, and found near-complete overlap between the two signals.

The results in the pyramidal cell layers of the hippocampus were revealing. In mice, the lipid droplets inside neurons of the CA3 region grew significantly larger with age, expanding by sixty-eight percent from an average cross-sectional area of 0.19 to 0.32 square micrometers. In naked mole rats, by contrast, neuronal lipid droplet size was unchanged between the young and middle-aged groups, remaining at roughly 0.15 square micrometers in both. The density of droplets per unit area of the pyramidal layer was similar between the species and unaffected by aging, meaning the key difference lay in droplet size and in where the fat was being stored. In the CA1 region, the naked mole rat droplets were actually smaller than those of mice at both ages, and neither species showed an age-related change there.

The most dramatic contrast emerged when the researchers examined microglia. Naked mole rat microglia were packed with lipid droplets that clustered extensively, making individual droplets impossible to distinguish. Measured as the percentage of microglial area covered by lipid droplets, the naked mole rat values dwarfed those of mice: in the CA3 region, young naked mole rats showed 12.07 percent coverage versus 2.41 percent in young mice, and middle-aged naked mole rats reached 20.06 percent versus just 1.79 percent in mice. In the CA1 region, the coverage was generally fifty to sixty-three percent lower than in CA3, but the pattern held, with middle-aged naked mole rat microglia carrying nine times the droplet load of young mice and fourteen times that of middle-aged mice. Critically, the number of microglia was similar across groups, so the difference reflected genuine fat storage rather than a change in cell counts.

When the researchers calculated the ratio of lipid droplet content in microglia to that in neurons, the species difference became stark. Naked mole rats showed six- to eightfold higher droplet coverage in microglia than in neurons in CA3, and sixteen- to twentyfold higher in CA1, while mice distributed their droplets more or less equally between the two cell types. The authors propose that this pattern represents a shuttling mechanism: the naked mole rat moves lipids out of neurons and into microglia, preventing the accumulation of toxic fats inside the cells that cannot be replaced easily. In mice, this transfer appears far less efficient, leaving neuronal droplets to enlarge with age. Because apolipoproteins, including the Alzheimer’s risk factor APOE4, govern lipid transport between neurons and glia, the finding raises the possibility that the efficiency of this neuron-to-glia lipid shuttle is a determinant of how vulnerable a brain is to lipotoxicity and degeneration.

Importantly, the study challenges the assumption that fat-laden microglia are inherently harmful. The naked mole rat, a species with exceptional resistance to brain aging, carried the heaviest microglial lipid load, yet its microglia showed no age-related loss of ramification, no increase in cell body area, and no rise in cell density. In mice, by contrast, microglial area increased with age in CA3 and microglial density tended to rise, hinting at early neuroinflammation. Correlation analyses showed a weak inverse relationship between droplet load and branching in both species, suggesting mildly reactive microglia hold more fat, but the causality remains unresolved. The authors note that lipid droplets may buffer peroxidized lipids, sequester fatty acids released during autophagy, and even detoxify misfolded proteins, supporting the view that droplet accumulation is protective at physiological levels and becomes damaging only when excessive.

The second major discovery concerned blood supply. Using collagen IV staining to visualize capillaries, the team quantified vascular density in the dentate hilus, the CA1 stratum radiatum, and the sensory and motor cortex. In every region and at every age, naked mole rats had dramatically sparser capillary networks than mice, with densities as low as one quarter to one half of mouse levels. In the dentate hilus, for example, young mice showed about 1.00 percent capillary coverage against 0.35 percent in naked mole rats, and middle-aged mice rose to 1.31 percent while the naked mole rat stayed flat at 0.34 percent. Mouse capillary density actually increased with age in the hilus and cortex, possibly a compensatory response to declining blood oxygen that later fails in old age, whereas the naked mole rat vasculature was completely stable across the lifespan.

Far from being a deficit, the authors argue, this sparse vasculature is likely part of a coordinated adaptation. Naked mole rats inhabit chronically hypoxic, carbon dioxide-rich burrows, have a markedly reduced basal metabolic rate, possess hemoglobin with high oxygen affinity, and can survive eighteen minutes of complete anoxia without brain damage, partly by switching to fructose-fueled anaerobic metabolism. Their brains appear to run on a minimal but sufficient capillary network optimized for stable function under low oxygen rather than maximal perfusion. The low capillary density may even connect back to the lipid droplet story: reduced perfusion can promote reactive oxygen species and lipid peroxidation, and microglial lipid droplets may be the naked mole rat’s intrinsic defense against exactly that damage, soaking up peroxidized fats before they can injure neurons.

Together, the findings sketch a coherent picture of a brain built for endurance. Where the aging mouse accumulates enlarging fat droplets in its neurons and shows early signs of microglial activation, the naked mole rat keeps its neurons lean and parks the fat in microglia, all while maintaining a stable, sparse blood supply that never changes with age. The work suggests that lipid droplet accumulating microglia are not intrinsically deleterious and that the traffic of lipids between brain cells, mediated by apolipoproteins, may be a decisive factor in neural resilience. If future research can confirm and harness this shuttling mechanism, it could open new avenues for protecting the human brain against Alzheimer’s disease, stroke, and the slow attrition of aging itself.

Subject of Research: Age-related lipid droplet distribution and capillary density in the hippocampus of naked mole rats compared with mice

Article Title: Age-dependent changes in lipid droplet distribution and vascularization in naked mole rat vs. mouse hippocampus

Article References: Krogstad, L. S., Teppen, M. A., Mjønes, H. S., Brewer, J. R., Park, T., Geiseler, S., & Morland, C. (2026). Age-dependent changes in lipid droplet distribution and vascularization in naked mole rat vs. mouse hippocampus. GeroScience. https://doi.org/10.1007/s11357-026-02548-6

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02548-6

Keywords: naked mole rat, lipid droplets, microglia, hippocampus, aging, neurodegeneration, capillary density, vascularization, lipotoxicity, BODIPY, hypoxia tolerance, GeroScience

A new study published in Bioengineering & Translational Medicine offers an elegant solution inspired by anatomy itself. A research team led by Yi Chieh Chang, Cherng-Shyang Chang, and Chung-Kan Tsao, supported by Chang Gung Memorial Hospital and Taiwan’s National Science and Technology Council, developed a living, bioactive wrapper made from fibroblasts—the connective tissue cells that naturally sheath and support organs. Their insight came from the trachea, the cartilage-ringed windpipe. In the native airway, each cartilaginous ring is enclosed by intercartilaginous ligaments and surrounded by an adventitial layer, a fibrovascular compartment composed largely of fibroblasts that both protects the cartilage and spatially regulates the blood vessels around it. The team hypothesized that recreating this compartment around an implanted cartilage scaffold could tame the chaotic host response.

To build their living interface, the researchers turned to a layer-by-layer assembly technique. Rabbit tracheal fibroblasts, isolated from the fibrovascular compartment of the airway, were repeatedly coated in alternating layers of fibronectin and gelatin, then seeded onto culture dishes over four consecutive days. The coatings allowed the cells to accumulate bottom-up into a dense, cohesive multilayered sheet that, after three additional days of culture, could be peeled off the dish intact, like a delicate living fabric. Three-dimensional confocal microscopy confirmed that fibroblast surface protein and fibronectin were distributed throughout the thickness of the sheet, giving it both structural cohesion and a matrix-rich character reminiscent of natural connective tissue.

But protection alone was not the goal. The team knew that cartilage, being avascular, depends on diffusion from surrounding vascularized tissue for oxygen and nutrients, and that engineered grafts in non-joint sites such as the trachea or the ear require early vascularization to survive. So they incorporated a second cell type: endothelial progenitor cells, or EPCs, isolated from rabbit white adipose tissue. These cells expressed the endothelial marker CD31 and the progenitor marker CD34, and in laboratory assays they formed capillary-like tube structures, confirming their vascular potential. When mixed into the top layer of the fibroblast sheet, the EPCs localized predominantly to the outer surface, creating a spatially organized construct: a fibroblast bulk facing the scaffold and a progenitor-rich vascular layer facing the host.

The testing platform was a polycaprolactone, or PCL, scaffold, a widely used biodegradable polymer known to permit host cell infiltration after implantation—precisely the vulnerability the team wanted to address. The salt-leached, laser-perforated scaffolds were seeded with rabbit chondrocytes, which within days expressed the master chondrogenic transcription factor SOX9 and, by two weeks, began producing aggrecan and type II collagen, the hallmark molecules of cartilage matrix. Each scaffold was then wrapped in a cell sheet, fibroblast layer inward, EPC layer outward, and implanted under the skin of nude mice, whose lack of mature T cells allowed the researchers to isolate innate immune responses in a deliberately hostile, vascularized, connective-tissue-rich environment.

The protective results were striking. After fourteen days, unwrapped cell-free scaffolds were riddled with host cells, and their polymer structure showed partial degradation. Wrapped scaffolds, by contrast, showed dramatically reduced infiltration, with quantified hematoxylin-positive and DAPI-positive areas significantly lower than in the unwrapped controls, and their structural integrity was largely preserved. Immunostaining told a similar immunological story: CD45-positive leukocytes and F4/80-positive macrophages, abundant throughout unwrapped scaffolds, were markedly reduced in the wrapped ones. Type I collagen staining revealed a discontinuous boundary-like pattern at the periphery of wrapped scaffolds, suggesting the cell sheet persisted, at least partially, as a transient interface between implant and host.

Yet protection came at a price. When the researchers examined chondrocyte-laden scaffolds, they found that fibroblast-only wrapping sharply reduced cartilage matrix deposition. Safranin O and Alcian blue staining, which highlight the sulfated proteoglycans of cartilage, were markedly diminished in wrapped scaffolds compared with unwrapped controls. The most likely explanation, the authors suggest, is diffusion limitation: the dense cellular barrier that kept host cells out may also have throttled the transport of oxygen and nutrients into the scaffold interior, starving the chondrocytes of the metabolic support they need to build matrix. This mirrors a well-known problem in other encapsulation contexts, from islet transplantation, where semipermeable membranes can protect grafts while limiting their survival, to myocardial tissue engineering, where stacked cell sheets become hypoxic without adequate vascular integration.

This is where the EPCs earned their place. In scaffolds wrapped with the EPC-incorporated sheet, cartilage matrix deposition was restored to levels comparable with unwrapped controls, with robust Safranin O and Alcian blue staining throughout the construct. Critically, the vascular benefit was spatially controlled. CD31 immunostaining revealed that EPC-containing scaffolds developed significantly more CD31-positive vascular structures in the peri-scaffold region, the roughly 200-micrometer zone surrounding the implant, while the scaffold interior showed no significant increase in vascularization compared with the other groups. In other words, the engineered interface created a vascularized sleeve around the cartilage—much like the perichondrium and subchondral bone do in native tissue—without letting vessels invade and disrupt the cartilage itself, a balance that excessive vascular ingrowth is known to upset.

The team also tested simpler alternatives, and the comparison proved instructive. A porous polycarbonate membrane seeded with fibroblasts and EPCs kept host cells out entirely, but cartilage-like tissue formed only near the scaffold boundary, leaving the interior sparse and matrix-poor—a reminder that a rigid physical barrier can be too effective, sealing the scaffold off from the very interactions it needs. A GelMA hydrogel encapsulation, meanwhile, largely degraded within the two-week implantation window, allowing extensive heterogeneous cellular infiltration and producing no cartilage-like tissue at all. The living cell sheet, by contrast, achieved something neither synthetic membrane nor hydrogel could: a cohesive, biologically integrated boundary that both shields and communicates, contacting the scaffold and host tissue directly while modulating what passes between them.

The implications reach well beyond the laboratory bench. Cartilage scaffolds implanted in joint defects benefit from regular mechanical loading and an avascular, relatively homogeneous environment, but constructs destined for tracheal repair or craniofacial reconstruction, such as treating microtia, face heterogeneous surroundings teeming with connective tissue, immune activity, and dynamic mechanical forces—conditions that have doomed many previous attempts. By demonstrating that a fibroblast-based cell sheet can act as a bioactive regulator of scaffold-host interactions, the study points toward a design principle: rather than engineering the scaffold alone, engineers the boundary. The authors acknowledge that their subcutaneous model does not fully replicate the mechanical and tissue-specific demands of orthotopic sites, that only male mice were used, and that future work should combine the approach with 3D-printed or electrospun scaffolds and test it in tracheal and auricular implantation models. But the core message is clear and potentially transformative: a living wrapper, borrowed from the body’s own anatomy, can protect an engineered implant, choreograph its vascular supply, and let cartilage grow—turning the hostile host environment from an adversary into a partner in regeneration.

Subject of Research: Bioactive fibroblast and endothelial progenitor cell sheet encapsulation for regulating scaffold-host interactions in cartilage tissue engineering

Article Title: A bioactive cell sheet interface regulates scaffold–host interactions for cartilage regeneration

Article References: Chang, Y. C., Chang, C.-S., & Tsao, C.-K. (2026). A bioactive cell sheet interface regulates scaffold–host interactions for cartilage regeneration. Bioengineering & Translational Medicine, Article e70177. https://doi.org/10.1002/btm2.70177

Image Credits: AI Generated

DOI: 10.1002/btm2.70177

Keywords: cartilage regeneration, cell sheet engineering, tissue engineering, fibroblasts, endothelial progenitor cells, PCL scaffold, vascularization, host immune response, tracheal repair, chondrocytes, extracellular matrix, biomaterials

The authors argue that the term bone organoid has been applied far too loosely. Spheroids, which are useful modular building blocks that can enhance osteogenic differentiation, often lack tissue-level organization altogether. Engineered constructs can reproduce geometry or mechanical behavior yet remain dominated by an exogenous scaffold rather than by cells. Under the framework proposed in the review, a genuine bone organoid must satisfy four core criteria: it must be a viable cell-derived three-dimensional tissue that develops organization beyond simple aggregation; it must produce its own collagenous or osteoid-like matrix with spatially and temporally resolved mineralization; it must show that new mineral is biologically deposited rather than inherited from the material it was grown on; and it must display at least one dynamic skeletal function, such as regulated matrix formation, formation-resorption coupling or a controlled mechanobiological response. Engineered guidance is permitted, but only when it is transient or demonstrably permissive to endogenous organization.

The review goes further by proposing a tiered classification system with five categories: osteogenic spheroids, bone-like microtissues, engineered skeletal constructs, bone organoids and high-fidelity bone organoids. The top designation demands evidence across three interdependent domains. Structural mimicry requires hierarchical extracellular matrix organization and progressive mineralization that mirrors the sequence seen in native bone, where cells first deposit unmineralized osteoid and mineral then advances from discrete fronts. Functional fidelity requires coupled activity between osteoblast-lineage cells that build matrix and osteoclast-lineage cells that resorb it, linked through signaling pathways such as RANKL, RANK and osteoprotegerin. Biomechanical integrity requires that the construct maintain structural coherence and exhibit measurable mechanosensation, reflecting the fact that bone cells continuously convert mechanical loading, fluid flow and matrix stiffness into biological responses. Crucially, the authors stress that a sophisticated scaffold-dominated construct can be clinically useful while still falling outside the organoid class; placement is a statement about the source of organization, not about quality.

When the authors applied this classification to representative original studies, the results were sobering. Of the ten studies examined in their evidence map, several constructs labeled organoids by their original authors were reassigned as osteogenic spheroids because self-assembly was limited to the spheroid unit and tissue-level hierarchy was never demonstrated. Others were reclassified as engineered skeletal constructs because pre-existing material architecture dominated organization. Only a handful, including a 2021 woven bone organoid grown from human bone marrow stromal cells and a 2025 system using dynamic dual-network hydrogels in which cell migration generated spatiotemporal woven-bone architecture, met the core organoid criteria. Notably, no current model satisfies every high-fidelity domain, and the reviewers conclude that transparent reporting of what was tested, and what was not, is essential for the field to advance.

A recurring problem the review highlights is confounding between biological mineralization and material-derived mineral signals. Many popular bone engineering strategies incorporate nano-hydroxyapatite, bioactive glass or black-phosphorus nanosheets that release phosphate ions, and any of these can generate calcium deposits that look like bone formation on a stain but are actually chemical precipitation. The framework therefore demands that active, biologically regulated mineralization be distinguished from material contributions using acellular baselines, a requirement that several widely cited nanomaterial-based studies do not currently meet. The authors also emphasize that late osteogenic marker expression alone does not establish functional bone-like tissue; marker data must be interpreted alongside matrix organization and mineral distribution.

Beyond classification, the review provides a detailed account of the biology that a faithful model must capture. Osteoblasts deposit type I collagen-rich osteoid that later mineralizes as hydroxyapatite crystals grow; a subset of these cells becomes embedded as osteocytes, which sense strain through dendritic processes in canalicular networks and regulate both building and resorption through mediators such as sclerostin. Osteoclasts, formed by fusion of monocyte-macrophage precursors, acidify the resorption compartment and release enzymes that remove mineral and collagen while also releasing matrix-stored growth factors like TGF-beta and BMP-related signals. Meanwhile, endothelial cells provide angiocrine signaling that supports osteoprogenitor maintenance, and neural elements release calcitonin gene-related peptide and substance P, which influence osteoblast, osteoclast and vascular function. The review cautions that multilineage marker expression alone is insufficient; functional integration must be demonstrated with lineage-appropriate readouts such as lumen formation and perfusion for vessels, or innervation-dependent regulation of remodeling for nerves.

On the fabrication side, the authors take a clear-eyed view of bioprinting, distinguishing between printing preformed organoids and printing for organoid formation, with only the former deserving the label unless post-print self-organization is demonstrated. They survey the four major modalities: inkjet printing offers fine patterning but limited viscosity range, extrusion printing handles mineral-filled bioinks and is the most scalable but exposes cells to shear stress, laser-assisted printing achieves micrometer-scale nozzle-free patterning but is limited by throughput, and light-based stereolithography and digital light processing generate complex channels but suffer from optical attenuation in mineral-rich inks. Across all modalities, the review finds that evidence that printing improves organoid maturation remains substantially weaker than evidence that it improves initial geometry. Hybrid strategies, such as temporary scaffold-guided self-assembly, printed channels coupled with self-organized microvasculature, and modular assembly of developmentally primed cartilage microtissues, are highlighted as the most promising path forward, exemplified by recent bioprinted bone-organoid grafts with guided vascularization that matured after implantation.

The translational outlook is cautiously encouraging. Patient-derived chondrosarcoma organoids have faithfully recapitulated histological and genetic features of parental tumors and shown sensitivity to the SHH pathway inhibitor vismodegib, while a biobank of 44 sarcoma organoid lines has enabled high-throughput drug screening. A three-dimensional vascularized humanized bone organoid has revealed that estrogen withdrawal drives vessel-like structure formation and mineral deposition, offering mechanistic insight into postmenopausal osteoporosis that non-vascularized models could not capture. Induced pluripotent stem cell-derived jawbone organoids have reproduced phenotypic features of osteogenesis imperfecta, pointing toward precision modeling of genetic skeletal disease. For regenerative applications, engineered ossification center-like organoids and periosteum-derived organoids combined with printed scaffolds have achieved bone repair in rodent critical-sized defects, though the authors note that systematic comparison against autograft controls remains an essential next step.

The review closes with a series of practical proposals intended to move the field from anecdote to standard. These include a minimum reporting checklist covering cell source, passage number, matrix composition, induction schedules, regional rather than whole-construct viability, quantitative mineralization assessment and predefined exclusion criteria for poorly formed organoids, together with application-specific potency assays: coupled remodeling endpoints for osteoporosis models, reproducible dose-response data for drug screening platforms and vascularized bone formation for regenerative grafts. The authors also flag regulatory hurdles, noting that clinical-grade organoids will require good manufacturing practice compliance, validated potency assays and jurisdiction-specific pathways such as the European Advanced Therapy Medicinal Product framework. Their overarching message is that progress will come not from adding complexity indiscriminately but from controlling when and where complexity is introduced, with stimuli-responsive matrices that yield to endogenous tissue, machine-learning-optimized printing and microfluidic platforms that connect bone organoids to vascular, immune and metabolic modules. Bone organoids, the review concludes, occupy a genuinely valuable intermediate position between dish and animal, provided the label is reserved for constructs that can prove they deserve it.

Subject of Research: Development of an operational classification framework and biofabrication standards for engineering high-fidelity bone organoids

Article Title: Engineering high-fidelity bone organoids: Operational classification, multilineage crosstalk, biofabrication evidence, and translational validation

Article References: Huang, Y., Zhang, T., Chen, S., Zhou, H., Xu, H., Zhang, F., Li, L., & Lyu, F. (2026). Engineering high-fidelity bone organoids: Operational classification, multilineage crosstalk, biofabrication evidence, and translational validation. Materials Today Bio, 41, Article 103676. https://doi.org/10.1016/j.mtbio.2026.103676

Image Credits: AI Generated

DOI: 10.1016/j.mtbio.2026.103676

Keywords: bone organoids, organoid classification, bioprinting, osteoblasts, osteoclasts, mineralization, vascularization, hydrogels, disease modeling, drug screening, bone regeneration, bioinks

The case for better models is straightforward. Mouse and rat hearts, the traditional workhorses of cardiovascular research, differ from human hearts in fundamental ways. They beat far faster, burn energy at a higher basal rate, and rely predominantly on a different myosin heavy chain isoform than the adult human ventricle, which is enriched in the beta form. These species differences help explain why promising preclinical results so often collapse in human trials, producing either unexpected cardiotoxicity or disappointing therapeutic benefit. Two-dimensional cultures, meanwhile, are cheap and fast but strip away the three-dimensional architecture, multicellular crosstalk and maturation cues that define living heart tissue. Cardiac MPS emerged to fill this void, and the review traces their remarkable evolution from the first beating embryoid bodies generated from stem cells in 1993, through engineered heart tissue in 2004, to the landmark 2021 creation of scaffold-free, self-organizing cardiac organoids called cardioids that recapitulate key features of early human heart development.

The review organizes cardiac MPS into two construction paradigms. The first is developmentally guided self-organization, in which human pluripotent stem cells are coaxed through timed bursts and pauses of WNT signaling, together with factors such as BMP4, activin A and FGF2, to form chamber-like organoids containing cardiomyocytes, fibroblasts, endothelial cells, neural crest cells and more. Elaborate versions add vascular-promoting cues like VEGF, foregut patterning, or even blood-generating signals to produce organoids with vascular compartments and hematopoietic progenitors. The second paradigm is engineering-driven assembly, which imposes external control through biomaterial scaffolds, 3D bioprinting and microfluidic heart-on-a-chip platforms. Self-organizing organoids offer unmatched developmental fidelity, making them ideal for studying embryonic cardiogenesis and congenital defects, while engineered tissues deliver superior reproducibility, spatial control and compatibility with quantitative drug screening. The review stresses that neither approach is universally superior; the choice depends entirely on the scientific question.

Biomaterials sit at the heart of the engineering approach, and the review provides an unusually detailed tour of the options. Naturally derived materials such as collagen, fibrin, hyaluronic acid, gelatin and decellularized extracellular matrix offer rich biological signals but suffer from batch variability and weak mechanics. Synthetic polymers like PEG, PCL and PLGA provide precise control and manufacturability but must be functionalized with cell-adhesive motifs. Hybrid systems blend the two philosophies. Electroconductive hydrogels laced with carbon nanotubes or conductive polymers can boost electrical coupling and connexin 43 expression, improving contraction synchrony, though dosing must be tuned carefully to avoid cytotoxicity. Scaffold stiffness matters enormously: native myocardium sits near 10 to 30 kilopascals, while scarred, fibrotic tissue stiffens to 35 to 70 kilopascals or more, and abnormal stiffness can actively drive disease-like cell states through epigenetic remodeling rather than merely accompanying them. Topographic cues from electrospun and micropatterned scaffolds guide cardiomyocyte alignment and restore the anisotropic architecture of real heart muscle.

Bioprinting adds another layer of control, depositing cell-laden bioinks layer by layer to fabricate vascular-like cardiac constructs with defined geometry. The review distinguishes carefully between biomaterial scaffolds, which can be processed under cell-free conditions, and true bioinks, which must keep living cells viable during extrusion or photopolymerization. Extrusion printing demands shear-thinning materials that recover quickly after deposition, while light-based printing requires finely tuned cross-linking kinetics. The technology’s strengths include spatial patterning of endothelialized structures, reproducible geometry, multicellular integration and controllable anisotropy. But a persistent trade-off between printing resolution and cell survival remains unresolved, and constructing a full hierarchical vascular network, from large vessels down to capillaries, still eludes the field. Microfluidics complements these efforts by providing perfusion, controlled shear stress, real-time biosensing and, in multi-organ chip formats, the ability to couple heart tissue with liver or kidney modules to capture systemic drug responses.

Perhaps the review’s most important contribution is its rigorous framework for assessing maturation, the property that ultimately determines whether a heart model is physiologically meaningful. Structural benchmarks include sarcomere length approaching the adult range of roughly 2.2 micrometers, organized Z-discs and A-bands, polarized connexin 43 at cell junctions, chamber-like cavity formation, and organized populations of fibroblasts, endothelial cells, macrophages and sympathetic neurons. Molecularly, the developmental shift from the TNNI1 to the TNNI3 troponin isoform is tracked quantitatively, though even the most mature organoids reported so far reach only about 18 percent TNNI3, corresponding roughly to a twenty-week human fetal heart, a sobering reminder that in vitro maturation still falls short of adulthood.

Functional maturation is assessed through an equally demanding battery of readouts. Electrophysiological maturity shows up as a more hyperpolarized resting membrane potential, adult-like action potentials and conduction velocities around 25 centimeters per second, with properly responsive beta-adrenergic signaling. Calcium handling reveals whether cardiomyocytes have built the T-tubule and dyad architecture needed for efficient excitation-contraction coupling, a feature most current systems still lack. Contractile performance is quantified by force generation; one progressively stretched engineered heart tissue achieved a specific twitch force of 11.28 millinewtons per square millimeter, a 5.1-fold improvement over controls. Metabolic maturation is measured through oxygen consumption rates and the shift from fetal glycolysis toward fatty acid beta-oxidation, which can be encouraged by substrate switching with fatty acids and galactose, thyroid hormone, or chronic electrical stimulation. In one conditioned tissue, mitochondria occupied 30 percent of the analyzed area, matching adult human myocardium. No single metric suffices, the review insists; maturity is a convergent, multidimensional phenotype.

The applications already demonstrated are striking. Multi-chamber cardioids have disentangled the regulatory relationship between the transcription factors HAND1 and NKX2-5 during early chamber formation. Organoids with microRNA-187 overexpression have reproduced congenital heart defect phenotypes that could be rescued by NIPBL supplementation. Gene-edited organoids carrying C10orf71 frameshift variants replicated the contractile defects of dilated cardiomyopathy, and the cardiac myosin activator omecamtiv mecarbil later restored function in a matching mouse model. Zonally oxygenated engineered tissues have recreated the infarct border zone with region-specific metabolic, fibrotic and calcium abnormalities that uniform hypoxia models miss entirely. Vascularized organoids showed that the blood pressure drug captopril protects cryoinjured tissue, while doxorubicin cardiotoxicity and even nanoplastic-induced injury have been modeled in organoid-on-chip formats. On the regenerative frontier, engineered heart muscle allografts have achieved durable remuscularization in rhesus macaques, though post-transplant arrhythmias highlight that electrical integration, not just survival, must be solved.

The review closes with a clear-eyed assessment of the remaining barriers: incomplete vascularization beyond the diffusion limit, fetal-like maturation, line-to-line variability, Matrigel batch effects, missing systemic physiology and the absence of standardized benchmarking. Its prescription is pragmatic rather than maximalist. Automated bioreactors, biosensors, AI-assisted closed-loop cultivation and standardized reference datasets could make cardiac MPS reproducible enough for regulated drug testing, while regenerative applications await perfusable vasculature and safe graft-host coupling. The message to the field is that progress should be judged not by how elaborate these tissues look, but by whether they become reliable, interpretable and genuinely predictive instruments of human cardiac medicine.

Subject of Research: Cardiac microphysiological systems for modeling human cardiac development, disease and drug responses

Article Title: Cardiac microphysiological systems in cardiovascular research: Construction paradigms, maturation trajectories, and translational frontiers

Article References: Xie, R., Zhao, H., Lei, Y., & Wang, X. (2026). Cardiac microphysiological systems in cardiovascular research: Construction paradigms, maturation trajectories, and translational frontiers. Bioengineering & Translational Medicine, Article e70167. https://doi.org/10.1002/btm2.70167

Image Credits: AI Generated

DOI: 10.1002/btm2.70167

Keywords: cardiac organoids, microphysiological systems, engineered heart tissue, cardiomyocyte maturation, 3D bioprinting, heart-on-a-chip, cardiotoxicity, disease modeling, vascularization, regenerative medicine, biomaterials, drug screening

The review begins with the fundamentals. Brain organoids are generated from human pluripotent stem cells, either embryonic stem cells or induced pluripotent stem cells, which aggregate into embryoid bodies and self-organize into neural tissue. In unguided protocols, these aggregates rely on intrinsic developmental cues, producing heterogeneous tissues that can contain features of several brain regions, including cortical, telencephalic, and choroid plexus-like identities. Guided protocols, by contrast, apply defined morphogens that modulate key signaling pathways such as WNT, sonic hedgehog, FGF, BMP, and retinoic acid to steer cells toward cortical, ventral forebrain, midbrain, hypothalamic, or choroid plexus-like lineages. The resulting tissues can recapitulate striking features of early cortical development, including ventricular-like lumens lined by SOX2-, PAX6-, and NESTIN-positive neural progenitors, from which differentiating neurons migrate outward toward cortical plate-like zones that echo the layered architecture of the human cortex.

Yet morphology, the authors stress, is a deceptive benchmark. The construction variables that shape an organoid’s identity include the cell source and genetic background, extracellular matrix support, embryoid body formation method, patterning strategy, culture platform, vascularization approach, and the incorporation of additional cell types such as astrocytes, microglia, or endothelial cells. Each choice ripples through regional identity, cellular composition, maturation state, and viability. Matrigel, the most widely used matrix, supports neuroepithelial expansion and three-dimensional morphogenesis, but its undefined composition and batch-to-batch variability actively undermine reproducibility. The review calls for defined hydrogels with tunable stiffness, degradability, and ligand composition as more controllable alternatives for future organoid engineering, framing this substitution as a prerequisite for translational standardization rather than a cosmetic improvement.

Culture conditions compound these challenges. Because organoids lack vasculature, oxygen and nutrients must diffuse inward from the surrounding medium, and interior cells frequently succumb to hypoxia and necrosis during long-term culture. Suspension and rotating bioreactor systems, first popularized in the landmark 2013 protocols, improve medium mixing and mass transfer, but dish-based approaches still provide limited control over aggregate size and oxygen gradients, contributing to inter-organoid variability. Microfluidic organoid-on-a-chip platforms go further, using miniaturized channels, valves, and sensors to regulate flow, compartmentalization, temperature, pH, nutrient supply, and drug exposure with unprecedented precision. Pillar and perfusion plate systems offer a parallel, higher-throughput middle ground, with evidence that continuous flow can improve oxygen supply and reduce necrotic regions compared with static culture. The review cautions, however, that every engineering benefit must be demonstrated with quantitative readouts, including organoid size distribution, hypoxia markers, viability, lineage composition, and electrophysiological maturation, rather than assumed from device sophistication.

Nowhere is the gap between engineering ambition and functional evidence more pronounced than in vascularization. The living brain is an energy-hungry organ wrapped in a dense capillary network, and avascular organoids inevitably suffer restricted proliferation, premature differentiation, impaired neurogenesis, and aberrant cortical patterning. The review details two principal strategies for closing this gap. In vitro, researchers embed endothelial cells in the matrix surrounding organoids, induce the endothelial transcription factor ETV2 to generate vascular-like networks, or fuse brain organoids with vascular spheroids, with Wnt/β-catenin pathway activation further promoting vessel formation. In vivo, transplantation into rodent brains allows host vasculature to infiltrate the graft, with vascular ingrowth detectable within 7 to 10 days and robust blood flow confirmed by two-photon microscopy. In one striking study, transplanted human brain organoids survived for up to 233 days in mice, with 85.4 percent of grafts becoming vascularized, while nonvascularized organoids failed to survive; vascularized grafts showed reduced apoptosis, larger size, and greater numbers of mature neurons.

The authors draw a sharp line, however, between vascularization and functional blood-brain barrier formation. Endothelial-like cells inside organoids may express tight-junction proteins and transporter markers such as Claudin-5, GLUT1, and P-gp, but calling this a reconstructed BBB requires permeability assays, transendothelial electrical resistance measurements, and transporter activity studies. Barrier maturation in vivo depends on orchestrated signaling, including Wnt/β-catenin for endothelial specification, PDGF-B/PDGFRβ for pericyte recruitment, and sonic hedgehog for barrier integrity, and no current organoid platform reproduces the full cellular composition, perfusion dynamics, or regional heterogeneity of the human neurovascular unit. The review’s message is that strong claims about BBB-like function demand multi-level evidence, not marker expression alone.

Functional validation receives an equally systematic treatment. Morphological quality control should track size, volume, surface regularity, neuroepithelial bud formation, ventricular zone-like structures, and necrotic or cystic regions, but these endpoints cannot establish regional identity or synaptic function. For that, the review turns to electrophysiology. Patch-clamp recording resolves single-cell excitability, revealing, for example, functionally mature neurons with voltage-dependent sodium and potassium currents in microglia-containing organoids and pacemaking dopaminergic neurons in midbrain-like models relevant to Parkinson’s disease. Multielectrode arrays provide non-invasive, longitudinal recordings of spikes, local field potentials, network bursts, and synchrony, capturing the emergence of network-level activity during long-term culture, including epileptiform-like dynamics in organoids derived from patients with neurodevelopmental disorders. Imaging adds another layer: marker panels spanning SOX2 and Ki67 for progenitors, DCX and TUJ1 for immature neurons, MAP2 and NeuN for mature neurons, and GFAP and IBA1 for astrocytes and microglia, benchmarked increasingly against single-cell transcriptomic references of the developing human brain.

The translational payoff of this framework is illustrated across a remarkable range of applications. TP53 knockdown organoids revealed disorganized neural stem cell layers and cell-cycle accumulation in G1 phase, illuminating the tumor suppressor’s role in human brain development. CDK5RAP2-mutant organoids reproduced the microcephaly phenotype with reduced volume and thinner neuroepithelium. Patient-derived midbrain organoids carrying the LRRK2 G2019S Parkinson’s mutation showed impaired dopaminergic markers, phosphorylated α-synuclein accumulation, and elevated mitophagy, with an LRRK2 kinase inhibitor rescuing the pathology. Familial Alzheimer’s organoids developed amyloid plaques and neurofibrillary tangles, while glioblastoma-organoid co-cultures preserved invasive tumor-brain interactions for drug testing. Perhaps most provocatively, the Brainoware platform coupled cortical organoids to high-density multielectrode arrays, using organoid network dynamics as a biological reservoir for speech recognition and nonlinear equation prediction tasks, an early but tangible step toward biohybrid computing.

The review’s closing argument is that none of these advances will translate reliably without confronting variability head-on. Donor genetics, reprogramming procedures, embryoid body size, Matrigel batches, morphogen timing, and culture parameters all accumulate into substantial inter-organoid and inter-batch heterogeneity that can obscure genotype-phenotype relationships, produce screening artifacts, and distort barrier measurements. The authors call for comprehensive reporting standards covering every stage from pluripotency validation to statistical strategy, automated platforms for cell maintenance and medium exchange, AI-assisted high-content quality control, and multi-omics molecular benchmarking. Crucially, they reject the idea of a single universal organoid: developmental studies, disease models, drug screens, BBB platforms, and biohybrid interfaces each require application-specific acceptance criteria. Brain organoids, the review concludes, should be understood not as miniature brains but as engineered biological platforms, designed, validated, and interpreted with the same discipline that governs any other precision technology destined for clinical and computational frontiers.

Subject of Research: Engineering strategies, vascularization, and functional standardization of human brain organoids for biomedical applications

Article Title: Engineering brain organoids for functional validation and translational applications: Construction strategies, vascularization, and standardization

Article References: Huang, G., Lu, C., Jin, Z., Huang, Y., Du, X., Hu, X., Peng, H., Wang, S., Wen, L., Qiu, J., Wang, G., & Zhao, C. (2026). Engineering brain organoids for functional validation and translational applications: Construction strategies, vascularization, and standardization. Bioengineering & Translational Medicine, Article e70164. https://doi.org/10.1002/btm2.70164

Image Credits: AI Generated

DOI: 10.1002/btm2.70164

Keywords: brain organoids, pluripotent stem cells, vascularization, blood-brain barrier, electrophysiology, microfluidics, drug screening, disease modeling, standardization, reproducibility, biohybrid computing, organoid-on-a-chip