Fig 2: Endothelial cell differentiation of DS-iPSCs(A) Diagram showing the close developmental relationship between hematopoietic and endothelial cell developmental pathways. Defects in hematopoietic lineage cells were characterized by Chiang et al. (2018). Areas of interest (i.e., production of EPCs from the hemangioblast and EC function) highlighted in red boxes.(B) Experimental schematic of endothelial cell differentiation and downstream assays. Dox is treated at day 0 of differentiation and the EPC population is assessed before purification. After purification, EPCs are matured and later tested for angiogenic function.(C) Representative flow analysis of CD31+ and CD34+ cell populations. Relevant CD31+CD34+ endothelial progenitor cells are in quadrant 2. XIST− (red) and XIST+ cells (blue) are overlaid.(D) Quantification of the CD31+CD34+ population of each sample (−/+ dox; paired t test; means ± SDs).(E) Immunofluorescence staining of endothelial cell specific markers and RNA FISH of XIST (scale bar, 10 μm).

Fig 3: Characterization of on-chip endothelial barrier. a) mvECs seeded onto the membrane in the medium channel formed uniform, tight monolayers as visualized by CD31 staining (fixed on d7). Scale bars equal 2 mm (tile scan of entire chip) and 200 µm (one-chamber view). b) Endothelial barrier integrity determined by fluorescence macromolecule tracing. Time difference is measured in 4 kDa FITC-dextran signal equilibria in tissue chambers versus media channels for chips with endothelial barrier versus chips without endothelial barrier. Endothelial barriers are less permeable than plain, acellularized membranes. c) Exposure of the on-chip endothelial barrier to inflammatory stimuli leads to i) a significant increase in CD106 expression for TNF-a stimulation (unpaired Two Sample t-test with p < 0.05); scale bar equals 200 µm; larger representation in Figure S6d, Supporting Information), ii) altered inflammatory cytokine secretion (n = 2 for each stimulus), and iii) decreased uptake/intracellular retention of acetylated low-density lipoprotein (AcLDL) (unpaired Two Sample t-test with p < 0.05; scale bar equals 200 µm).

Fig 4: Concept of the human patient-specific WAT-on-chip model. a) Schematic of WAT in vivo anatomy and integration of all cellular components into the microfluidic platform: Mature adipocytes, progenitors, stem cells, and tissue-resident immune cells are encapsulated in a hydrogel and cultured in the chips’ tissue compartment. MvECs are seeded via the media channel onto the membrane shielding tissue chambers from the constant perfusion. To study immune cell recruitment, circulating immune cells are perfused through the media channels. b) Patient-specific cell sources for building the WAT-on-chip model: Mature human adipocytes, mvECs as well as cells from the SVF, including tissue-resident immune cells such as CD14+-cells, are isolated from skin biopsies with subcutaneous fat. Circulating immune cells, such as T-cells, are retrieved by isolating PBMCs from the patients’ blood. Scale bars equal 100 µm (adipocytes, CD31, CD45, and T cell visualization) or 50 µm (SVF and CD14+-cell visualization). *CD14+-cells are isolated using MACS and could therefore not be stained for CD14 but for CD86, another marker expressed on macrophages.

Fig 5: Modular mix-and-match toolbox to build fit-for-purpose autologous WAT-on-chip models. a) To create WAT-on-chip models specifically tailored to certain scientific questions, the authors propose a spectrum of on-chip culture conditions with varying degrees of complexity ranging from simple adipocyte-only systems (A) to highly complex full WAT-on-chip models (ASE). b) Structural characterization of the full WAT-on-chip model (ASE) on d12. Anti-CD31 immunofluorescence staining showed a tight endothelial barrier on the chip's membrane. To visualize the membrane area over the adipocyte-filled tissue chambers, the chip had to be inverted (i). Adipocytes are displayed by staining their lipid droplets. To uncover stromovascular cells, SVF is labeled with a cell tracker prior to injection into the chip. SVF is 3D distributed among the adipocytes in the tissue chamber (adipocytes not shown for better visibility of SVF) (ii; orthogonal view). Moreover, a tracking of SVF motion within the first 6 h after injection revealed dynamic migration for some of the labeled cells while others remained stationary (ii; Video S1, Supporting Information). Scale bars equal 200 µm (one-chamber view) and 100 µm (zoomed-in orthogonal view and video). c) Monitoring of FA trafficking properties for A, AS, and AE systems uncovered noticeable differences in FA uptake comparing A and AS to AE. Representative images of A and AE conditions at time points 0, 10.5, 38.5, and 52.5 min. d) Comparison of adipokine release by different co-/multi-culture WAT-on-chips from the same patient, measured from media effluents collected for 24 h from d10 to d11 of on-chip culture. Even though the analyzed adipokines are exclusively produced by adipocytes, there are considerable differences in their release regarding the models’ cellular composition. e) Identification of on-chip ATMs by visualizing CD11c and Trem2 (i), CD86 and CD45 (ii), and CD206 and CD45 (iii) expression. ATMs are positive for the investigated markers and clustered to, sometimes even enwrapping, mature adipocytes. Moreover, intracellular lipid droplets in CD86+-cells might be an indicator of lipid scavenging activity. Z-stack imaging data are also represented as supplemental videos (Videos S2–S4, Supporting Information) to ensure full elucidation of all events in one stack. Scale bars equal 50 µm.