The technology belongs to a family known as sorption-based atmospheric water harvesting, or SAWH. Unlike condensation systems, which chill air below its dew point and demand substantial electrical power, sorption-based systems use hygroscopic materials that soak up vapor at ambient temperature and then release it when gently heated, often by nothing more than sunlight. That makes them attractive for off-grid communities, remote islands, and agricultural settings. But the leading materials, salt-composite hydrogels, have long suffered from a kinetic bottleneck. Loading more hygroscopic salt raises the total water capacity, yet it also thickens the internal diffusion barrier, so the material takes longer to reach its capacity. A sorbent that needs hours to saturate simply cannot cycle often enough to be productive.

The new work traces that bottleneck to a transport problem with two coupled stages. Water vapor must first find and bind to hygroscopic sites, and the resulting salt solution must then redistribute through the polymer network. Previous analyses have suggested that, under representative conditions, liquid diffusion inside the hydrogel is often the dominant resistance, not the arrival of vapor from the surrounding air. Earlier attempts to speed things up focused on porous architecture: ice-templating, foaming, and directional freeze-drying all shorten diffusion distances. But those approaches are energy-intensive, add processing complexity, and never fully solve the deeper issue that much of the salt remains buried inside the polymer matrix, out of direct contact with the air it is supposed to be drinking from.

The inspiration for a different solution came from the grey mangrove, Avicennia marina, a plant that thrives while rooted in seawater. The mangrove transports salt ions from its internal tissues to specialized glands on its leaf surface, where the salt accumulates in microstructured regions. Under humid conditions, that surface salt deliquesces and pulls moisture straight out of the atmosphere. The research team realized that the most valuable feature was not the plant’s physiology but its geometry: a spatial separation between internal salt transport and storage on one hand, and surface-exposed, air-accessible hygroscopic sites on the other. A synthetic material built on that principle could let vapor meet salt immediately, without waiting for moisture to fight its way through a dense gel.

Translating that idea into hardware, the researchers fabricated what they call a multi-layer heterogeneous hygroscopic membrane, or MHH membrane. It is a sandwich structure made by alternating electrospinning and spraying. First, a porous mat of polyacrylonitrile nanofibers is electrospun onto a collector. Then, pre-formed brush-like hydrogel microspheres, roughly 100 micrometers across, are sprayed onto the fiber scaffold. A second round of electrospinning encapsulates the hydrogel, producing a nanofiber-hydrogel-nanofiber stack. The outer nanofiber layer is coated with polypyrrole, a black polymer that absorbs sunlight efficiently, and the whole membrane is loaded with lithium chloride, one of the most powerful hygroscopic salts known. The final membrane carries about 170 grams of salt per square meter.

The clever part is what happens during drying. When the saturated membrane is regenerated, salt solution is driven out of the hydrogel and crystallizes in the voids of the outer nanofiber layer, right at the air interface. On the next adsorption cycle, that surface-enriched salt meets humid air directly and deliquesces almost immediately, and the resulting solution is then drawn into the hydrogel reservoir below. Energy-dispersive X-ray spectroscopy confirmed the effect: the membrane surface showed a chlorine-to-carbon atomic ratio of 10.80, compared with just 1.27 on the surface of conventional hydrogel microspheres loaded with the same amount of salt. Cross-sectional imaging found substantial salt deposits occupying the inter-fiber voids of the outer layer, consistent with the redistribution mechanism.

The kinetic payoff is striking. At 25 degrees Celsius and 70 percent relative humidity, both the MHH membrane and the hydrogel microsphere control reach a similar equilibrium uptake of about 2.4 grams of water per gram of material, proving that the layered design sacrifices no storage capacity. But the membrane absorbs roughly 1.4 grams per gram within the first 30 minutes, nearly double the control, and hits 85 percent of equilibrium within 90 minutes. The microspheres need about 300 minutes to reach the same fraction. Fitting the curves with a unified adsorption model yields a characteristic time constant of 40.12 minutes for the membrane versus 110.72 minutes for the microspheres, and an initial sorption rate about 2.1 times higher. Independent measurements of the elementary steps showed why: the hydrogel takes up saline solution faster than the nanofiber layer can generate it by vapor-driven deliquescence, so the interface never floods and vapor capture can continue.

That rapid liquid uptake is itself engineered at the molecular scale. The hydrogel microspheres are not ordinary polyacrylamide; they contain polyethylene glycol side chains grafted onto the network, forming a brush-like architecture. Flexible PEG segments increase chain mobility and accessible free volume, easing the passage of water molecules. In cross-flow vapor permeation experiments, a PAM-co-PEGMA film 0.46 millimeters thick showed a permeation lag time of only about 2 minutes, whereas a plain polyacrylamide film of just 0.14 millimeters lagged roughly 13 minutes. Normalized to a common thickness, the time-lag-derived effective diffusion coefficient of the brush-like hydrogel came out approximately 70 times that of conventional PAM. Swelling tests reinforced the picture: in 10 percent lithium chloride solution, the brush-like hydrogel swelled to about 1000 percent after 240 minutes, versus roughly 600 percent for PAM, with kinetics consistent with Fickian transport rather than sluggish polymer relaxation.

The membrane also regenerates efficiently under sunlight. Coated with polypyrrole on its outer surface, it releases about 70 percent of its stored water within 90 minutes under one-sun illumination, compared with about 60 percent for the microspheres, and its surface temperature exceeds 70 degrees Celsius within 20 minutes, roughly 10 degrees hotter than the control. Because the photothermal layer sits at the evaporation interface rather than dispersed through a hydrated gel, heat is generated exactly where water leaves. Durability held up too: after 20 adsorption-desorption cycles, the membrane retained about 2.5 grams per gram of uptake with overlapping kinetic curves, and salt remained visible in the nanofiber interstices. One honest caveat emerged from a 48-hour exposure to 100 percent humidity, which caused slight saline leakage; the design does not fully confine salt solution under prolonged saturation, though performance recovered after drying.

Crucially, the material-level speed translated into device-level output. The team built a semi-continuous, three-bed harvester in which one bed desorbs under a solar lamp while the others adsorb, cycling every 2 hours of capture and 1 hour of release. In the laboratory at 25 degrees Celsius and 90 percent relative humidity, the device produced 3.47 liters per square meter over 8 hours, an average of 0.43 liters per square meter per hour. Outdoors in Xiamen, under 60 to 65 percent humidity and peak solar irradiance near 830 watts per square meter, the same prototype yielded about 1.62 liters per square meter in a day, and a second field test under cloudier, hotter conditions reached about 1.96 liters per square meter per day. Condensate analysis by ICP-OES found low inorganic contamination, though the authors caution that microbiological safety requires separate system-level measures. The broader lesson may outlast the specific membrane: by engineering interfacial accessibility and internal transport together, across scales from nanofibers to polymer side chains, sorbent designers can finally stop choosing between capacity and speed.

Subject of Research: Biomimetic multi-layer hygroscopic membranes for rapid sorption-based atmospheric water harvesting

Article Title: Interfacial Salt Redistribution Enables Rapid Atmospheric Water Harvesting in a Biomimetic Multi‐Layer Hygroscopic Membrane

Article References: Zhong, F.-Y., Guo, C.-Y., Shao, Z.-D., Zhong, L.-B., Chen, G. Z., Zheng, Y., Zhou, J. L., Zheng, Y.-M., & He, J. (2026). Interfacial Salt Redistribution Enables Rapid Atmospheric Water Harvesting in a Biomimetic Multi‐Layer Hygroscopic Membrane. Advanced Science, Article e78196. https://doi.org/10.1002/advs.78196

Image Credits: AI Generated

DOI: 10.1002/advs.78196

Keywords: atmospheric water harvesting, hygroscopic membrane, mangrove biomimicry, lithium chloride, hydrogel, electrospinning, solar desorption, water scarcity, sorption kinetics, polyethylene glycol, nanofibers, photothermal materials

The research team, working at the Korea Research Institute of Bioscience and Bioscience-affiliated laboratories, took an approach that sidesteps a long-standing obstacle in the field. Bone marrow from children with Legg-Calvé-Perthes disease is scarce and difficult to study, and even when mesenchymal stromal cells, the body’s principal skeletal progenitors, are isolated from osteonecrotic bone, they carry the imprint of a diseased, inflamed, poorly perfused tissue environment. Any abnormality observed in such cells could reflect that hostile microenvironment rather than an intrinsic flaw. To separate the two, the researchers reprogrammed skin fibroblasts from two patients with the disease, and from a healthy donor, into induced pluripotent stem cells using a non-integrating Sendai virus system that delivers the reprogramming factors OCT4, SOX2, KLF4 and c-MYC. Because skin cells can be reset to an embryonic-like state and then steered into mesenchymal stem cells under identical, standardized conditions, any difference that emerges between patient-derived and healthy cells must be written into the cells themselves rather than imposed by the tissue around them.

The quality control on this cellular reset was rigorous. The resulting induced pluripotent stem cell lines formed compact colonies with the characteristic morphology of embryonic stem cells, stained strongly for alkaline phosphatase, and expressed the canonical pluripotency markers OCT4, NANOG, TRA-1-60, TRA-1-81, SSEA-3 and SSEA-4 at levels comparable across all three lines. Quantitative PCR confirmed expression of OCT4 and REX1 in the undifferentiated state, and when the cells were allowed to spontaneously differentiate into embryoid bodies, they activated ectodermal markers such as SOX1 and PAX6, mesodermal markers including HAND1 and PDGFRA, and endodermal markers SOX7 and AFP. Immunostaining for TUJ1, NESTIN, DESMIN, alpha-SMA, FOXA2 and SOX17 confirmed protein-level differentiation into all three germ layers, and teratoma assays in immunodeficient mice produced tissues representing ectoderm, mesoderm and endoderm. G-band karyotyping showed normal diploid chromosomes in every line, ensuring that any phenotype found later was not an artefact of culture stress or genomic instability.

With validated stem cell lines in hand, the team differentiated them into induced mesenchymal stem cells using a defined, xeno-free protocol, first patterning the cells toward the mesodermal lineage for four days and then maturing them for a further seventeen days on fibronectin-coated plates. Flow cytometry confirmed that the resulting cells expressed the canonical mesenchymal markers CD73 and CD105 and lacked the haematopoietic markers CD34 and CD45, consistent with bona fide MSC identity. One subtlety stood out: the two patient-derived lines expressed CD90, a surface glycoprotein involved in adhesion and MSC immunophenotype, at only about sixty to seventy percent of the level seen in healthy donor cells, which reached roughly ninety-five percent. Proliferation rates, however, were indistinguishable between the groups, meaning the patient cells grew normally even as they showed this subtle phenotypic deviation, hinting that the real differences lay deeper, in what the cells could become rather than how fast they could multiply.

What the cells could become turned out to be the central finding. Under osteogenic induction, healthy donor cells produced abundant mineralized matrix, revealed by intense Alizarin Red S staining, and strongly upregulated the osteoblast genes RUNX2 and BGLAP. Both patient-derived lines formed dramatically fewer mineral deposits and mounted only weak induction of the same genes. Adipogenesis fared no better: healthy cells accumulated plentiful lipid droplets stained by Oil Red O and induced the adipogenic regulators PPARG and ADIPOQ, while patient cells showed markedly reduced lipid accumulation and gene induction. Chondrogenesis, assessed by Alcian Blue staining of sulfated glycosaminoglycans and by expression of SOX9 and COL2A1, was also reduced but comparatively better preserved. Crucially, this pattern of a simultaneous, broad suppression of all three mesenchymal lineages argues against the classic idea of a simple lineage shift, in which bone-forming cells are diverted into fat cells. Instead, it points to a deeper loss of overall differentiation competence within the patient-derived progenitors themselves.

The molecular culprit emerged when the researchers examined the Wnt/beta-catenin pathway, the master regulator of mesenchymal lineage commitment. Beta-catenin, the transcriptional co-activator at the heart of canonical Wnt signalling, promotes osteogenesis and chondrogenesis while restraining adipogenesis, and its intracellular abundance is normally kept in check by glycogen synthase kinase-3, which phosphorylates beta-catenin and tags it for proteasomal destruction. In the patient-derived cells, the team found a striking double abnormality. Transcripts for CTNNB1, the gene encoding beta-catenin, were significantly reduced, while GSK3B mRNA was markedly elevated. At the protein level, phosphorylation of GSK3alpha at Tyr279 and GSK3beta at Tyr216, the modifications associated with enhanced enzymatic activity, was substantially increased, reaching roughly 1.8- to 2.2-fold above healthy donor levels in the more severely affected patient line. Total beta-catenin protein and intracellular beta-catenin fluorescence were correspondingly diminished, painting a coherent picture of a kinase pushed into overdrive and a key transcriptional regulator destroyed faster than it should be.

The disease signature extended into inflammation, a domain long suspected to matter in this condition because children with active disease show elevated IL-6 and HMGB1 in synovial fluid. Even without any inflammatory stimulus, the patient-derived mesenchymal cells expressed higher baseline levels of IL6, PTGS2 and CCL2, indicating a primed inflammatory state. When the cells were challenged with IL-1beta, the patient cells responded with larger fold increases in these cytokines than healthy cells, and the Wnt target genes AXIN2 and LEF1, already slightly reduced at baseline, were driven down even further. Western blotting showed that IL-1beta pushed GSK3 phosphorylation higher in patient cells than in controls while beta-catenin, already low, declined still more. These exaggerated responses persisted after osteogenic differentiation, where patient-derived osteoblasts displayed stronger NF-kappaB phosphorylation and greater degradation of its inhibitor I-kappa-B-alpha upon inflammatory stimulation. Transcriptomic sequencing reinforced the theme, revealing enriched alterations in inflammatory signalling, extracellular matrix organization and Wnt-related regulatory pathways in both patient fibroblasts and patient-derived mesenchymal cells, with upregulation of multiple extracellular Wnt antagonists including DKK1, SFRP1, SFRP2, WIF1 and SOST.

The most striking result, and the one with the clearest translational implication, came next. When the researchers treated the patient-derived cells with lithium chloride, a long-established inhibitor of GSK3, beta-catenin fluorescence recovered to levels comparable to untreated healthy donor cells, with the rescue being strongest precisely where the deficit had been greatest. More importantly, the functional defect yielded to the same treatment. Lithium chloride substantially increased mineralized nodule formation and Oil Red O staining in both patient lines during osteogenic and adipogenic induction, and quantitative PCR confirmed that either lithium chloride or the more selective inhibitor CHIR99021 significantly boosted expression of BGLAP, RUNX2, ADIPOQ and PPARG. The differentiation programs had not been deleted; they had merely been silenced by an overactive kinase, and silencing the kinase switched them back on. This established that the mesenchymal failure in Legg-Calvé-Perthes disease reflects a reversible, kinase-dependent signalling imbalance rather than irreversible cellular damage.

The study’s authors are careful to place their findings in context. Observations in patients, such as elevated circulating leptin and increased adipose tissue within the affected femoral head, reflect systemic and tissue-level changes that arise from ischemia, inflammation and altered marrow mechanics, and the in vitro reduction in adipogenesis seen here does not contradict that clinical picture. Rather, the platform isolates the intrinsic cellular component of the disease, complementing animal models in which ischemia was surgically induced in piglets and rodents. Those models have shown that ischemia activates hypoxia-inducible factor 1-alpha and drives inflammatory and angiogenic responses, and the new human data now connect those tissue-level events to a specific, druggable intracellular node inside the progenitors that must rebuild the bone. The authors propose a reciprocal feedback loop in which inflammatory activation sustains GSK3 activity, further destabilizing beta-catenin and progressively compromising skeletal regeneration.

For a disease whose current treatments range from conservative containment to major reconstructive surgery, the prospect of a pharmacological strategy that restores the bone-building capacity of a child’s own progenitor cells is genuinely exciting. Lithium is an old drug with a well-characterized safety profile in other contexts, and the iPSC-based model developed here provides a patient-specific testing ground for optimizing GSK3-targeted regenerative therapies before they reach the clinic. Much work remains, including validating the findings in larger patient cohorts and determining how a GSK3 inhibitor could be delivered safely to the developing femoral head. But the central message stands: the cellular failure underlying a devastating childhood hip disease is not a life sentence written into the genome, but a reversible regulatory state, and with the right key, the machinery of skeletal repair can be switched back on.

Subject of Research: Reversible GSK3/beta-catenin dysregulation in patient-derived iPSC-mesenchymal stem cells from children with Legg-Calvé-Perthes disease

Article Title: Patient‐Derived iPSC‐MSC Modelling Reveals Reversible GSK3‐Mediated Suppression of Mesenchymal Lineage Differentiation in Legg‐Calvé‐Perthes Disease

Article References: Seol, B., Lim, H. J., Song, C. L., Lee, J., & Cho, Y. S. (2026). Patient‐Derived iPSC ‐ MSC Modelling Reveals Reversible GSK3 ‐Mediated Suppression of Mesenchymal Lineage Differentiation in Legg‐Calvé‐Perthes Disease. Journal of Cellular and Molecular Medicine, 30(17), Article e71350. https://doi.org/10.1111/jcmm.71350

Image Credits: AI Generated

DOI: 10.1111/jcmm.71350

Keywords: Legg-Calvé-Perthes disease, induced pluripotent stem cells, mesenchymal stem cells, GSK3, beta-catenin, Wnt signalling, osteonecrosis, osteogenesis, inflammation, lithium chloride, CHIR99021, paediatric orthopaedics