Physiological aging and rejuvenation of vital human organ systems in vitro

Abstract

Implementations of a microphysiological system may include at least one first inlet for receiving a fluid medium; a fat module including adipocytes, the fat module configured to receive the fluid medium; a liver module including hepatocytes; a channel between the fat module and the liver module, the channel configured to promote communication of the fluid medium from the fat module to the liver module; and the channel configured for exchange of one or more of nutrients, signaling molecules, and drugs between the fat module and the liver module.

Description

PCT / US25 / 5230902 December 2025 (02.12.2025)Attorney Reference: BERK-524WOPHYSIOLOGICAL AGING AND REJUVENATION OF VITAL HUMAN ORGAN SYSTEMS IN VITROGOVERNMENT SUPPORT

[0001] This invention was made with government support under Grant Number AG071787 awarded by the National Institutes of Health. The government has certain rights in the invention.CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This Pursuant to 35 U.S.C. § 119 (e), this application claims priority to the filing date of United States Provisional Patent Application Serial No. 63 / 719,531 filed November 12, 2024, the disclosure of which application is herein incorporated by reference.BACKGROUND OF THE INVENTION

[0003] The steadily increasing incidence of obesity, affecting approximately 40% of adults in the US, urges for a better mechanistic understanding of human obesity-associated metabolic diseases and the development of effective and safe treatments. Increased visceral adipose tissue mass is linked to multiple disease conditions, including type 2 diabetes mellitus (T2DM) and metabolic dysfunction-associated steatotic liver disease (MASLD). Although obesity is strongly correlated with the risk of MASLD development, not all obese individuals develop metabolic dysfunctions, highlighting the need for comprehensive studies of disease triggers.

[0004] Expansion of white adipose tissue (WAT) is frequently associated with inflammation caused by accumulation of tissue-resident activated (Ml) macrophages. WAT inflammation leads to dysregulated lipolysis, insulin resistance, increased secretion of proinflammatory cytokines and altered secretion of adipokines, all of which affect other organs, including the liver. Whether WAT expansion or its inflammation is the main trigger in WAT-driven MASLD and hepatic insulin resistance (HIR) has been difficult to address in animal models because lipid accumulation and lipotoxic events often go hand in hand with inflammation of WAT and liver. The clinical relevance of data from conventional in vitro models in whichPCT / US25 / 5230902 December 2025 (02.12.2025)cellular monocultures are challenged with fatty acids or proinflammatory cytokines is also limited. Therefore, there is a need for advanced human-specific in vitro models that address the complexity of obesity-induced HIR and define the role of WAT expansion vs inflammation.

[0005] Microphy siological systems (MPS) are superior to conventional in vitro models in mimicking human liver tissue and WAT by allowing for increased cellular complexity, 3D tissue structure and dynamic culture conditions. These characteristics proved to be instrumental for modeling of MASLD using hepatic cell lines, primary cells and iPSC-derived cells.

[0006] Multi-tissue MPS are particularly clinically relevant because they address the importance of tissue interaction in metabolism. A pioneering study identified adipocyte-derived factors that cause metabolic alterations in hepatocytes in a primary human cell -based WAT-liver MPS, including under culture conditions mimicking T2DM and obesity. Similarly, media transfer experiments between murine adipocytes and hepatocytes showed that secreted factors act as mediators of communication between WAT and liver to regulate lipid metabolism. Although primary cells are considered the gold-standard for disease modeling and drug testing, their limited availability, variable quality and genetic heterogeneity represent substantial hurdles for the development of multi -tissue MPS.

[0007] Recent protocols can produce human induced pluripotent stem cell (iPSC)-derived hepatocytes (iHEPs), adipocytes (iADIPOs) and macrophages (iMACs) that closely resemble primary cells in differentiation and function. Here, we use iADIPOs and iHEPs to develop an interconnected WAT-liver MPS, including modeling of WAT inflammation by exposure to proinflammatory Ml-iMACs. We show that this isogenic iPSC-based MPS allows for dissecting the effects of WAT expansion and inflammation on metabolic hepatocyte function and rational screening for MASLD drugs.SUMMARY OF THE INVENTION

[0008] In a first aspect, the invention provides a microphysiological system comprising: at least one first inlet for receiving a fluid medium; a fat module comprising adipocytes, the fat module configured to receive the fluid medium; a liver module comprising hepatocytes; a channel between the fat module and the liver module, the channel configured to promote communication of the fluid medium fromPCT / US25 / 5230902 December 2025 (02.12.2025)the fat module to the liver module, the channel configured for exchange of one or more of nutrients, signaling molecules, and drugs between the fat module and the liver module.

[0009] In a second aspect, the invention provides a method of testing a drug or a combination of drugs for treatment of an adipocyte-associated disease or a hepatocyte-associated disease, comprising: contacting the microphysiological system (MPS) of a preceding claim with the drug or the combination of drugs; and assessing an attribute of the adipocytes, the hepatocytes, and / or the fluid medium; thereby testing the drug or the combination of drugs for treatment of the adipocyte-associated disease or the hepatocyte-associated disease.

[0010] In a third aspect, the invention provides a method for determining the biological age of a biological sample, comprising: a) obtaining the biological sample; b) assessing a property of the biological sample; c) comparing the property of the biological sample with a database of the property in a human population; and d) determining the biological age of the biological sample based on step c); thereby determining the biological age of the biological sample.

[0011] In a fourth aspect the i n vention provides a method for determining the biological age of a human subject or a biological sample obtained from the human subject, comprising: a) obtaining the biological sample from the human subject; b) contacting the biological sample with a microphysiological system (MPS) of the disclosure; c) assessing a property of the adipocytes, the hepatocytes, and / or the fluid medium of the MPS; d) comparing the property of the adipocytes, the hepatocytes, and / or the fluid medium of the MPS with a database of the property in a human population; and e) determining the biological age of the biological sample based on step d); thereby determining the biological age of the subject or the biological sample.

[0012] In a fifth aspect, the invention provides a method of testing a drug or a combination of drugs for the attenuation or reversal of the biological aging of adipose and / or liver tissue, comprising: a) contacting the microphysiological system (MPS) of the disclosure with a fluid medium comprising human serum from a male or female donor with a chronological age of at least 62 years; b) contacting the product of step a) with the drug or the combination of drugs; and c) assessing an attribute of the adipocytes, the hepatocytes, and / or the fluid medium; thereby testing the drug or the combination of drugs for the attenuation or reversal of the biological aging of adiposePCT / US25 / 5230902 December 2025 (02.12.2025)and / or liver tissue.

[0013] In a sixth aspect, the invention provides a method for controlling cell and tissue aging, comprising: al) contacting the microphysiological system (MPS) of the disclosure with fluid medium comprising human serum from a male or female donor with a chronological age of at least 62 years, thus creating an old circulatory milieu; bl) assessing the effects of the old circulatory milieu; cl) contacting the product of step al) with fluid medium comprising human serum from a male or female donor with a chronological age of from about 21 years to about 34 years, thus creating a young circulatory milieu; dl) assessing the effects of the young circulatory milieu; el) establishing a hallmark and / or a process of aging, which is controlled significantly differently by the young circulatory milieus versus the old circulatory milieus; fl) establishing the timing and the dose of the human serum that control a hallmark and / or a process of aging upward by the old circulatory milieu as compared to the young circulatory milieu. Alternatively, the invention provides a method for controlling cell and tissue aging, comprising: a2) contacting the microphysiological system (MPS) of any one of claims 1-37 with fluid medium comprising human serum from a male or female donor with a chronological age of from about 21 years to about 34 years, thus creating a young circulatory milieu; b2) assessing the effects of the young circulatory milieu; c2) contacting the product of step al) with fluid medium comprising human serum from a male or female donor with a biological age of at least 62 years, thus creating an old circulatory milieu; d2) assessing the effects of the old circulatory milieu; e2) establishing a hallmark and / or a process of aging, which is controlled significantly differently by the old circulatory milieu versus the young circulator}' milieu; f2) establishing the timing and the dose of the human serum that control a hallmark and / or a process of aging upward by the young circulatory milieu as compared to the old circulatory milieu.BRIEF DESCRIPTION OF THE DRAWINGS

[0014] FIG. 1A-1J shows interconnection and scaling of iHEP-MPS and iADIPO-MPS. (FIGS. 1A, IB) Albumin (FIG. 1A) and urea (FIG. IB) in media of iHEP-MPS under perfused (dashed line) and static (solid line) conditions for 15 days. n= 14. 12, 12, 14, 5, 10, 5 in the perfused conditions and n= 13, 14, 14, 11, 5, 11, 7 in the static conditions for D3-D15 in (FIG. 1 A). n=3 in all conditions in (FIG. IB), n referring to media collected from both conditions, p-values were calculated by unpaired two-PCT / US25 / 5230902 December 2025 (02.12.2025)tailed t-tests for the comparison of perfused vs static each day. (FIGS. 1C, ID) Glucose in media (FIG. 1C) (n=4 for Basal, Insulin; n=3 for Glucagon) and metabolic gene expression in iHEPs (FIG. ID) (n=4 except n=3 for Glucagon PC KI, G6PC, GCK) of iHEP-MPS after 1-hour hormonal stimulation, n referring to media in (FIG.1C) and MPS in (FIG. ID), p-values were calculated by one-way ANOVA followed by Tukey’s test in (FIG. 1C) and by unpaired two-tailed t-tests in (FIG. ID). (FIG. IE) Schematic and photo of interconnected iADIPO-MPS and iHEP-MPS. (FIGS. 1F-1H) Representative fluorescent images of iADIPO-MPS (left image in FIG. IF, FIG. 1G) and iHEP-MPS (right image in FIG. IF, FIG. 1H). Blue, nuclei; red, F-actin; green, fatty acids. (FIG. II) Total non-esterified fatty acid and fluorescent fatty acid concentrations in media at different iADIPO-iHEP ratios. Insets show representative fluorescent images of iHEP-MPS at 1:1 and 30:1 ratios. n=3 except n=4 for Total fatty acids 1,5. n referring to media. Two-tailed correlation test shows Pearson r= 0.9996 and p=0.173 between total and fluorescent fatty acids. (FIG. 1J) Glucose levels in media of 1-hour insulin-stimulated iHEP-MPS 48 hours after interconnection with iADIPO-MPS at different cell ratios. n=4 except n=7 for 5: 1. n referring to media, p-values were calculated by one-way ANOVA then Dunnett’s test. All data are mean ± SD. *p<0.05; **p<0.005; ns, p>0.05.

[0015] FIG. 2A-2J shows modeling the link between WAT inflammation and MASLD in iADIPO-iHEP-MPS. (FIGS.2A-2D) Schematic of iADIPO-MPS and iHEP-MPS interconnection (FIG. 2A) and representative images and quantification of fatty acids in iADIPO-MPS (FIGS. 2B, 2D) and iHEP-MPS (FIGS. 2C, 2D). Ml-iADIPO, iADIPO-MPS containing Ml-iMACs; iADIPO, iADIPO-MPS without iMACs. n=4 in iADIPO and n=3 in iHEP. n referring to media. (FIG. 2E) LDH and lipolysis products in circulating media. n=4 except n=3 for LDH. n referring to media.(FIGS. 2F, 2G) Gene expression in iADIPO-MPS (FIG. 2F) (n=4) and iHEP-MPS (FIG.2G) (n=3 except n=4 for iADIPO NFKB2). n referring to MPS. (FIG. 2H) Cytokine levels in circulating media. n=3 except n=4 for TNFa. n referring to media. (**) indicates values that were set to detection limit of the assay. (FIG. 21) Glucose uptake in iADIPO-MPS measured by the clearance of glucose in the assay media. n=4 referring to media. (FIG. 2J) Glucose levels in media of 1-hour insulin-stimulated iHEP-MPS. n=5 referring to media. All data are mean ± SD. p-values were determined by paired two-tailed t-test in (FIG. 21) and unpaired two-tailed t-test elsewhere. * p<0.05; ** p<0.005; ns, p>0.05.PCT / US25 / 5230902 December 2025 (02.12.2025)

[0016] FIG. 3A-3H shows drugs normalizing inflammation and glucose and lipid metabolism in Ml-iADIPO-iHEP-MPS. (FIG.3A) Schematic of MPS interconnection and drug administration. (FIGS.3B, 3C) Representative images (FIG. 3B) and quantification (FIG. 3C) of fluorescent fatty acids in Ml-iADIPO-MPS and iHEP-MPS. Met, metformin; Rosi, rosiglitazone; Dex, dexamethasone. n=4 for Ml-iADIPO-MPS except n=3 for Rosi. n=4 for iHEP-MPS except n=3 for Dex. n=3 for circulating media, n referring to MPS. (FIG. 3D) Glucose uptake in Ml-iADIPO-MPS measured by the clearance of glucose in the assay media. n=3 referring to media.(FIG. 3E) Gene expression in iMAC-iADIPO-MPS. n=3 referring to MPS. (FIG. 3F) TNFa and adiponectin levels in circulating media. n=3 except n=4 for TNFa-Dex and Adiponectin-Control. n referring to media. (FIG. 3G) Gene expression in iHEP-MPS. n=4 except n=3 for Met, Rosi, Dex TNF and Met, Rosi NFKB2. n referring to MPS.(FIG. 3H) Glucose levels in media of iHEP-MPS without and with 1-hour insulinstimulation. n=3 referring to media. All data are mean ± SD. p-values were determined by paired two-tailed t-tests in Basal vs Insulin comparison for each drag in (FIG. 3D) and (FIG. 3H) and by one-way ANOVA followed by Tukey’s test in multiple comparisons for each group in (FIGS. 3C, 3E, 3F, 3G) and two-way ANOVA followed by Tukey’s test for cross-group comparison in (FIG. 3H). *p<0.05; **p<0.005.

[0017] FIG. 4A-4K shows identification of the effects of semaglutide on WAT inflammation-induced MASLD in Ml-iADIPO-iHEP-MPS. (FIGS. 4A, 4B) Representative images (FIG. 4A) and quantification (FIG. 4B) of fluorescent fatty acids in Ml-iADIPO-MPS and iHEP-MPS. n=6 except n=5 for circulating media, n referring to MPS. (FIG. 4C) Gene expression in iMAC-iADIPO-MPS. n=5 referring to MPS. (FIG. 4D) and adiponectin levels in circulating media. n=3 referring to media. (FIG. 4E) Gene expression in iHEP-MPS. n=3 except n=4 for ACACA to CYP2E1 and n=5 for Semaglutide PPARA. n referring to MPS. (FIG. 4F) Glucose uptake in Ml-iADIPO-MPS in response to insulin. n=3 referring to media. (FIG. 4G) Glucose levels in media of iHEP-MPS without and with 1-hour insulin-stimulation. n=3 for Control and n=4 for Semaglutide. n referring to MPS. (FIG. 4H) Schematic of selective semaglutide treatment of iHEP-MPS (top) or Ml-iADIPO-MPS, the latter including semaglutide antagonism with exendin 9-39 and compound 5D in iHEP-MPS (bottom). (FIGS. 41, 4J) Quantification of fatty acids (FIG. 41) (n=8 except n=6 for No semaglutide and Semaglutide on Ml-iADIPO-iHEP-MPS. n referring to MPS)PCT / US25 / 5230902 December 2025 (02.12.2025)and glucose without and with 1-hour insulin-stimulation in media of iHEP-MPS (FIG.4J) (n=3 referring to media). (FIG.4K) Gene expression in iHEP-MPS. n=3 referring to MPS. All data are mean ± SD. p-values were determined by paired two-tailed t-test in Basal vs Insulin comparison in (FIGS. 4F, 4G, 4J), one-way ANOVA followed by Tukey’s test in (FIG. 41), two-way ANOVA followed by Tukey’s test in cross-group comparison in (FIG. 4G, 4J) and by unpaired two-tailed t-test elsewhere. *p<0.05; **p<0.005; ns, p>0.05.

[0018] FIG. 5A-5E shows iADIPO-iHEP transwell co-culture. (FIG.5A) Schematic of transwell co-culture of iHEPs and iADIPOs. (FIG. 5B) Isoproterenol (Isop)-induced release of fatty acids into media by iADIPOs with and without insulin stimulation. n=8 referring to media. (FIG. 5C) Time course of fatty acid uptake into iHEPs co-cultured with iADIPOs. n=3 referring to iHEPs. (FIG. 5D) Time course of glucose release into media of insulin-stimulated iHEPs after co-culture with iADIPOs. Insulin was added at the beginning (0 minutes, min). n=4 referring to media. (FIG. 5E) Normalized glucose release into media by iHEPs after 3 hours (h) of insulin stimulation after co-culture with iADIPOs. n=3 referring to media. All data are mean ± SD. p-values were calculated by one-way ANOVA followed by Tukey’s test in (FIGS. 5B, 5E) and two-way ANOVA then Tukey’s test in (FIGS. 5C, 5D). *p<0.05; **p<0.005.

[0019] FIG. 6A, 6B shows PDMS absorption in cell-free MPS. (FIG. 6A) Fluorescent fatty acid concentrations in media circulating for up to 6 days. n=3 referring to media, p-values were determined by one-way ANOVA followed by Dunnett’s test by comparing circulating days to day 0. (FIG. 6B) Relative drug concentrations in media before and after circulation for 48 hours. n=3 referring to media, p-values were determined by unpaired two-tailed t-tests. *p<0.05; **p<0.005; ns, p>0.05. All data are mean ± SD.

[0020] FIG. 7A-7H shows specific effects of iMACs and iADIPOs on glucose and lipid metabolisms in iHEP-MPS. (FIG. 7A) Cytokine levels in media circulating between iHEP-MPS and iADIPO-MPS without iMACs, with Ml-iMACs or with M0-iMACs 48 hours after interconnection. n=3 except n=4 for i ADIPO and Ml-i ADIPO TNFa. n referring to media. (**) indicate values that were set to detection limit of the assay. (FIG. 7B) Glucose uptake in iADIPO-MPS cultured without iMACs, with Ml-iMACs or with MO-iMACs for 48 hours. n=4 referring to media. (FIG. 7C) GlucosePCT / US25 / 5230902 December 2025 (02.12.2025)levels in media of insulin-stimulated iHEP-MPS after interconnection with iADIPO-MPS without iMACs, with Ml-iMACs or with MO-iMACs for 48 hours. n=5 except n=6 for MO-iADIPO. n referring to media. (FIG. 7D) Schematic of interconnected MPS conditions with (top) or without (bottom) iADIPOs. (FIGS. 7E, 7F) Glucose levels in circulating media 24 hours (FIG. 7E) and 48 hours (FIG. 7F) after interconnection. n=3 referring to media. (FIGS. 7G, 7H) Basal lipid uptake (FIG. 7G) (n=8 referring to MPS) and gene expression (FIG. 7H) (n=3 referring to MPS) in iHEP-MPS after 48 hours of interconnection, p-values were determined by paired two-tailed t-test in (FIGS. 7B, 7E, 7F) in Basal vs Insulin comparison, one-way ANOVA followed by Tukey’s test in comparison of three conditions in (FIGS. 7A, 7C) and two-way ANOVA then Tukey’s test in cross-group comparison in (FIG. 7E), and unpaired two-tailed t-test in (FIGS. 7G, 7H). The significance of i AD IPO effectiveness in (FIG. 7F) was determined by ANOVA without post-hoc test. All data are mean ± SD. *p<0.05; **p<0.005; ns, p>0.05.

[0021] FIG. 8A-8F shows effects of semaglutide treatment on individual cell types.(FIGS. 8A, 8B) Gene expression in iADIPO-MPS after 48 hours (FIG. 8A) (n=3 referring to MPS) and 7 days (FIG. 8B) (n=4 except n=3 for EBF2 referring to MPS).(FIG. 8C) Adiponectin levels in circulating media. n=3 referring to media. (FIGS.8D, 8E) Gene expression in iHEP-MPS (FIG. 8D) (n=3 referring to MPS) and Ml-iMACs in monoculture (MC; FIG. 8E) (n=3 referring to iMACs). (FIG.8F) TNFa levels in circulating media. n=4 in Ml-MC referring to wells and n=3 in iADIPO-MPS referring to MPS. p-values were determined by unpaired two-tailed t-test.*p<0.05; **p<0.005; ns, p>0.05. All data are mean ± SD.

[0022] FIG. 9A, 9B shows GLP1R expression in iPSC-derived and primary human cells. (FIG. 9A) GLP1R gene expression in iPSC-derived cells. n=9 referring to MPS for iADIPOs and iHEPs, and to wells for iMACs. *p<0.05; **p<0.005. p-values were determined by one-way ANOVA followed by Tukey’s test. (FIG. 9B) GLP1R gene expression in primary human pancreatic islets (plslets), primary human WAT (pWAT) and primary human hepatocytes (pHeps). pWAT contains adipocytes, tissueresident macrophages and other cell types. n=l referring to cells / tissue. All data are mean ± SD.

[0023] FIG. 10A-10N, 10P, 10Q shows human serum induced aging in WAT- and liver-MPS. After 4-days of perfusion of the WAT-MPS (FIGS. 10A-10H) and liver-PCT / US25 / 5230902 December 2025 (02.12.2025)MPS (FIGS. 10I-10N, 10P, 10Q) with OCM, as compared to conventional MPS medium and compared to YCM, several clinically significant aging-associated tissue dysfunctions were established, including accumulation of senescent cells (FIGS 10A-10D, 10I-10L) in measures of aging-associated gene expression, SA-beta Gal senescence, increase in p 16 marker of senescence, inflammatory proteins and senescence associated secretory phenotype, SASP (FIGS. 10E, 10M), increase in oxidative DNA damage (FIGS. 10F, ION), perturbation of adipogenesis and fat depot (FIG. 10G), perturbation of lipid and insulin regulated glucose metabolisms (FIG. 10H, 10P, 10Q). all gene markers were normalized to negative control, which was cultured in the same batch with the conventional MPS medium.

[0024] FIG. 11A-11E shows aged Fat MPS propagates aging to Liver MPS: fat-liver organome influences. Transfer of aging from OCM-pretreated fat to liver was successfully established in the WAT-liver organome chip (FIG. 11 A), with respect to the altered gene expression (FIG. 11B), induction of SA-beta-Gal accumulation (FIG. 11C), and perturbation of glucose and lipid metabolisms (FIG. 11D), showing a different genotype pattern comparing to OCM treatment in liver (FIG. HE).

[0025] FIG. 12A-12E shows transcriptomics of the serum-treated heterochronic WAT-MPS and human data trained ML model predicts the age of WAT-MPS (FIG.12A) PCA grouped by age conditions. n=3 for young-male, young-female and old-male conditions, n=5 for old-female condition. (FIG. 12B) Genes of aging-associated alterations selected from top 25% of differential gene list. (FIG. 12C) Top 10 GO terms agreed between public database (GTEx subcutaneous adipose tissue, SAT) and WAT-MPSs with OCM and YCM treatments. (FIG. 12D) Human SAT age prediction model training workflow based on machine learning from GTEx database. The model was trained based on 70% random datapoints of GTEx, then validated using the rest 30% datapoints. (FIG. 12E) Age prediction of male OCM- and YCM-treated WAT-MPS.

[0026] FIG. 13A-13D shows serum induced WAT-MPS aging recreates human in vivo aging, suggesting novel biomarkers and pathways. (FIGS. 13A, 13C) Workflow and visualization of aging biomarkers on expression levels (FIG. 13 A) and biological noises (FIG. 13C). (FIGS. 13B, 13D) Protein-protein interaction networks enriched from the aging biomarkers of expression levels (FIG. 13B) and biological noises (FIG. 13D). (FIG. 13B) shows the network in degree sorted layout. (FIG. 13D) onlyPCT / US25 / 5230902 December 2025 (02.12.2025)shows genes with interactions. STRING analysis predicted 10 associated partner genes from original 37 genes of noise with confidence score >0.99.

[0027] FIG. 14A-14G shows drug screening and approaches for improving tissue health in the aging-on-chip MPS hiPSC-based MPS (WAT and liver) were perfused with human serum from old donors for 4 days (FIG. 14A), after which several approaches, such as senolytics, Alk5 inhibitor of TGF-beta (Alk5i), oxytocin (OT), and heterochronic microfluidics (as indicated in the X axes) were examined for their capacity to attenuate the experimentally induced aging dysfunctions. Significant experimental attenuation of aging-on-chip was observed for both, fat and liver MPS, with respect to the normalized gene expression (FIGS. 14B, 14D, 14F) and the recovery of the insulin-regulated glucose and lipid metabolisms (FIGS. 14C, 14E, 14G).

[0028] FIG. 15A-15H shows sexual dimorphism in serum-treated MPSs (FIGS. 15A, 15B) gene expression characterized by qPCR in WAT- (FIG. 15A) and liver-MPS (FIG. 15B), replotting from the same data set of FIG. 10 but in the measure of sex. (FIG. 15C) Principal components analysis and sample cluster tree regarding on both age and sex. (FIGS. 15D-15G) Top 10 KEGG pathways of upregulated genes in male (FIG. 15D) and female (FIG. 15F), and GO terms overlapped to GTEx subcutaneous adipose tissue database in male (FIG. 15E) and female (FIG. 15G). (FIG. 15H) Aging biomarkers of expression level and biological noises in male and female conditions of WAT-MPS.

[0029] FIG. 16A-16F shows additional analyses of 4-day-serum-treated WAT- and liver-MPS. (FIGS. 16A, 16B) Full expression pattern of serum-treated WAT-MPS (FIG. 16 A) and liver-MPS (FIG. 16B) including those gene without significant changes. (FIGS. 16C-16F) GO (FIGS. 16C, 16E) and KEGG (FIGS. 16D, 16F) enriched from WAT-MPS under OCM (FIGS. 16C, 16D) and YCM conditions (FIGS. 16E, 16F).

[0030] FIG. 17A-17F shows differential expression and enrichment analyses of age and sex variables, using public database GTEx. (FIG. 17A) Volcano plot of differentiation genes in old vs young. (FIGS. 17B, 17C) Top 10 GO terms enriched from upregulated genes in young (FIG. 17B) and old (FIG. 17C). (FIG. 17D) Volcano plot of differentiation genes in male vs female. (FIGS. 17E, 17F) Top 10 GO terms enriched from upregulated genes in male (FIG. 17E) and female (FIG.PCT / US25 / 5230902 December 2025 (02.12.2025)17F). 6 of top 10 GO terms enriched from female SAT dataset related to immune responses (FIG. 17F), confirming the markers of elevated inflammation in the MPS set-up with serum from old females (shown in FIG. 16A. 16F).

[0031] FIG. 18A-18D shows human female SAT and VAT age prediction models.(FIG. 18A) model validation of male and female VAT. (FIG. 18B) female SAT age prediction of WAT -MPS in female condition. (FIGS. 18C, 18D) VAT age prediction of WAT-MPS in male (FIG. 18C) and female condition (FIG. 18D).

[0032] FIG. 19A-19D shows GO and KEGG enriched from aging markers. (FIGS.19A, 19B) GO (FIG. 19A) and KEGG (FIG. 19B) from the biomarkers of transcriptomic levels. (FIGS. 19C, 19D) GO (FIG. 19C) and KEGG (FIG. 19D) from biological noises.

[0033] FIG. 20A-20D shows transcriptomic comparison of old serum-induced and pro-inflammatory Ml macrophage induced alterations in WAT. (FIG. 20A) Venn plot of DEGs upregulated in old serum treated iADIPO-MPS and DEGs upregulated in Ml macrophage co-cultured i ADIPO. (FIG. 20B) Heatmaps of genes uniquely induced by old serum. (FIGS. 20C, 20D) Selected GO terms enriched from upregulated genes in Ml -macrophage-induced (FIG. 20C) and old-serum-induced conditions (FIG. 20D).

[0034] FIG. 21A-21H shows transcriptional alterations after anti-aging treatment. FIG.21 A-C shows whole gene patterns in response to anti-aging drugs in three administration settings measured by qPCR. (FIG. 21D) Transcriptomics of heterochronic serum exchange in WAT-MPS. Genes of aging-associated alterations selected from differential gene list in OY vs OO comparison. (FIGS. 21E, 21F) GO terms enriched from upregulated genes in OY (FIG. 21E) and OO (FIG. 21F). (FIGS.21G, 21H) GO terms enriched from upregulated genes in YO (FIG. 21G) and YY (FIG. 21H) after differential comparison of YO vs YY.

[0035] FIG. 22A-22I shows additional transcriptomic analysis of heterochronic serum exchange in iADIPO-MPS (FIG. 22A) Principal components analysis regarding on all heterochronic conditions. (FIGS. 22B, 22C) KEGG pathways of upregulated genes in OY (FIG. 22B) and OO (FIG. 22C). (FIG. 22D) Since enriched KEGG pathways are very limited, supervised enrichment PLSDA was performed. (FIG.22E) PLSDA scatter plots of OY vs OO and YO vs YY, showing improved separation of two conditions in each comparison. (FIG. 22F, 22G) PLSDA KEGGPCT / US25 / 5230902 December 2025 (02.12.2025)pathways of upregulated genes in OY (FIG. 22F) and 00 (FIG. 22G) in OY vs 00 comparison. (FIGS. 22H, 221) PLSDA KEGG pathways of upregulated genes in YO (FIG. 22H) and YY (FIG. 221) in YO vs YY comparison.

[0036] FIG. 23A-23C shows (A) Workflow schematic of the detection of de novo proteins in WAT-MPSs derived from MetRSL274GhiPSCs. (B,C) Profiles of de novo proteins under OCM (B) and YCM (C) conditions. Proteins are shown as their encoding gene names in the enrichment analysis. Data represent merged results from both male and female samples. Singleton without enriched functions are not shown.

[0037] FIG. 24 shows transcriptomic analysis and identification of novel biomarkers and pathways in liver-MPS aging induced by CMs alone and CMs-pretreated-WAT. (A) Cross-comparison of GO terms to GTEx male liver aging. (B, C) Identification of aging biomarkers on expression levels (B) and biological noises (C) from GTEx liver aging of both sexes, aligning to FIG 13. Both only show genes with protein-protein interactions, excepting shared singleton CDKN2B. STRING analysis predicted 5 associated partner genes for noise genes with confidence score >0.99 in two aging conditions, for CDKN2B with confidence score >0.95. Blue hallmark genes are shared by GTEx and CMs alone, red by GTEx and CMs-pretreated-WAT, purple by all three.

[0038] FIG. 25 shows rejuvenation approaches in the aging-on-chip MPS (a-c) Drug screening in WAT-MPS by perfusing with OCM plus drugs for 4 days (a), OCM for 4 days then OCM plus drugs for additional 4 days (b), or OCM for 4 days then medium without old serum and plus drugs (c). (d,e) Knockdown P16 by siRNA together with 4 days OCM (d) or after OCM for 2 days (e). (f,g) Heterochronic microfluidics in WAT-MPS (f) and liver-MPS (g) by perfusion with OCM for 4 days then YCM for 4 days, (a-c) only show significant changes comparing to respective old controls.

[0039] FIG. 26 shows principal components analysis regarding on both age and sex in WAT-MPS and liver-MPS.

[0040] FIG. 27A-27D shows signature KEGG pathways of upregulated genes in male (A, B) and female (C, D), and GO terms overlapped to GTEx liver database, all in their young conditions due to diminished sex dimorphism in old.

[0041] FIG. 28A-D shows additional analyses of serum heterochronicity in WAT- / liver-MPS and fat-liver organome influences. (A,B) KEGG enriched liver-MPSPCT / US25 / 5230902 December 2025 (02.12.2025)under OCM (A) and YCM (B) conditions. (C, D) KEGG in liver-MPS induced by OCM-pretreated (C) and YCM-pretreated WAT-MPS (D).

[0042] FIG. 29A-D shows top 10 GO terms enriched from upregulated genes in liver young (30-39 year-old) (A) vs old (50-69 year-old) (B) and male (C) vs female (D). Young vs old comparison of two tissues merged both male and female data. Male vs female comparison of SAT merged both young and old data, but of liver only used young data due to diminished sexual dimorphism in old liver. The age range in liver was off the proposed range because limited sample number and disease-biased 20-29-year-old cohort. The trend also match the GO results enriched from GTEx alone.

[0043] FIG. 30 shows model validation of female VAT and liver.

[0044] FIG. 31 shows age prediction of both sex conditions of liver-MPS in human liver model. Noting that only 65 female liver samples, 46 for machine -learning and 19 for validation, highly compromised the reliability of the model, even with a nominal 100% accuracy.

[0045] FIG. 32A-32F shows GO and KEGG enriched from aging markers. GO and KEGG from the WAT biomarkers of transcriptomic noises with log2 (CVoid / CVyoUng) >0.5 (A, B). (C, D, E, F) GO from transcriptomic levels and noises in liver aging induced by CMs alone (C, D) and CMs-pretreated-WAT (E, F).

[0046] FIG. 33A-33C shows secretomic analysis based on SASP genes and de novo proteogenesis. (A) Summary of major SASP expression in all measured aging conditions. P as “none” because the raw counts with numerous zero-expression samples could not be processed through the transcriptomic analysis pipeline. While there are still two meaningful cases: expression detected only in young samples (extreme blue) or only in old samples (extreme red). (B, C) De novo proteogenesis in WAT-MPS derived from MetRSL274GhiPSCs for comparisons of female (B) and male (C) YCM conditions. Singleton genes without enriched functions were hidden.

[0047] FIG. 34 shows the time-course pilot test of OCM induced aging. The same batch of WAT-MPS s was treated with OCM (HPLM supplemented with 5% pooled male serum) for 0-4 days. Aging status was assessed by senescence-associated p-galactosidase staining. One-way ANOVA followed by Dunnett’s test was performed to determine the critical time point for senescence induction.

[0048] FIG. 35 shows a full plasmid construct.PCT / US25 / 5230902 December 2025 (02.12.2025)DETAILED DESCRIPTION OF THE INVENTIONI. Introduction

[0049] The invention provides microphysiological systems, as well as methods of making and using them.II. Microph siological System

[0050] In one aspect, the invention provides a microphysiological system comprising: at least one first inlet for receiving a fluid medium; a fat module comprising adipocytes, the fat module configured to receive the fluid medium; a liver module comprising hepatocytes; a channel between the fat module and the liver module, the channel configured to promote communication of the fluid medium from the fat module to the liver module, the channel configured for exchange of one or more of nutrients, signaling molecules, and drugs between the fat module and the liver module. In an exemplary embodiment, the fluid medium is a hepatocyte culture media, wherein the fluid medium comprises hepatocyte growth factor and oncostatin M, and wherein the fluid medium essentially does not comprise epidermal growth factor and dexamethasone. In some embodiments, the hepatocyte culture media may comprise hepatocyte growth factor in the range of from about 1-100 ng / ml, e.g., from about 15-25 ng / ml. In some embodiments, the hepatocyte culture media may comprise oncostatin M in the range of from about 1-100 ng / ml, e.g., from about 15-25 ng / ml. In an exemplary embodiment, the fluid medium is a human plasma like media. In an exemplary embodiment, the fluid medium comprises human serum in a concentration of from about 4% to about 12%, e.g., the fluid medium may comprise human serum in a concentration of from about 5% to about 11 %, from about 6% to about 10%, or from about 7% to about 9%. In an exemplary embodiment, the adipocyte to the hepatocyte cell number ratio is from about 1:1 to about 1:30. In an exemplary embodiment, the adipocyte to the hepatocyte cell number ratio is from about 1:5 to about 1:30. In an exemplary embodiment, the adipocyte to the hepatocyte cell number ratio is from about 1 :4 to about 1 :6. In an exemplar}' embodiment, the adipocyte to the hepatocyte cell number ratio is from about 1:1 to about 30:1, e.g., the adipocyte to the hepatocyte cell number ratio may be from about about 3:1 to about 28:1, from about 5:1 to about 26:1, from about 7:1 to about 24:1, from about 9:1 to about 22:1, from about 10:1 to about 20:1, from about 12:1 to about 18:1, or from about 14:1 to about 16:1. In some cases, the adipocyte to the hepatocyte cell numberPCT / US25 / 5230902 December 2025 (02.12.2025)ratio may be from about 5:1 to about 30:1, e.g., from about 7:1 to about 28:1, from about 9:1 to about 26:1, from about 11:1 to about 24:1, from about 12:1 to about 22:1, from about 14:1 to about 20:1, ofror m about 16:1 to about 18:1. In some cases, the adipocyte to the hepatocyte cell number ratio may be from about 4:1 to about 6:1. In an exemplary embodiment, the hepatocytes are derived from human induced pluripotent stem cells. Methods for producing hepatocytes from induced pluripotent stem cells are known in the art and described, e.g., in WO2020227711A1, WO2010149597A2, and Groeger, M. et al. Nature Communications 14, 3902 (2023), the disclosures of which are incorporated herein by reference. In an exemplary embodiment, the human induced pluripotent stem cells are cultured on a substrate comprising a member selected from Matrigel, RGD-BME, Vitronectin, and Laminin 521. In an exemplary embodiment, the human induced pluripotent stem cells are cultured on Matrigel. In an exemplary embodiment, the human induced pluripotent stem cells are contacted with endoderm-induction media which comprises a member selected from CHIR99021, PI103, sodium butyrate, monothioglycerol, doxycycline, Y27432, and a combination thereof. In some cases, the endoderm-induction media may comprise CHIR99021 in the range of from about 0.1-10 pM, e.g., from about 2-4 pM. In some cases, the endoderm-induction media may comprise PI103 in the range of from about l-100nM, e.g., from about 40-60nM. In some cases, the endoderminduction media may comprise sodium butyrate in the range from about 0.05-5 mM, e.g., from about 0.4-0.6mM. In some cases, the endoderm-induction media may comprise Y27432 in the range of from about 0.1-10 pM, e.g., from about 9-11 pM. In an exemplary embodiment, the liver module comprises a plurality of liver module cell culture chambers and the hepatocytes are introduced into the liver module cell culture chambers as mature hepatocytes. In an exemplary embodiment, the adipocytes are derived from induced pluripotent stem cells (e.g., human induced pluripotent stem cells. Methods for producing adipocytes from induced pluripotent stem cells are known in the art and described, e.g., in WO2024066156A1 and Qi, L. et al. Small, 2103157 (2021), the disclosures of which are incorporated herein by reference. In an exemplary embodiment, the adipocytes are derived from human mesenchymal stem cells (hMSCs) (e.g., hMSCs derived from induced pluripotent stem cells). In an exemplary embodiment, the adipocytes are white adipocytes derived from human mesenchymal stem cells (hMSCs). In an exemplary embodiment, the hMSCs are transduced with a nucleic acid that codes for peroxisome proliferator-actrivated receptPCT / US25 / 5230902 December 2025 (02.12.2025)gamma (PPARy) (e.g., human PPARy). In some embodiments, the nucleic acid may comprise an expression vector, e.g., a lentivirus compatible expression vector. In some embodiments, the expression of the encoded PPARy may be under the control of an inducible promoter (e.g., a chemically inducible promoter such as tetO, inducible by doxycycline). In an exemplary embodiment, the hMSCs are transduced by a lentivirus. Lentivirus, lentivirus compatible expression expression vectors, methods of transducing cells with lentivirus are known in the art and described, e.g., in Ahfeldt, Tim, et al. " Nature cell biology 14.2 (2012): 209-219, the disclosure of which is incorporated herein by reference. In an exemplary embodiment, the transduction occurs in a transduction medium comprising SB431542 and / or doxycycline. In an exemplary embodiment, 80% or more of the adipocytes have lipid droplets, e.g., in some cases 82% or more, 84% or more, 86% or more, 88% or more, 90% or more, 92% or more, 94% or more, 96% or more, 98% or more, or 99% or more of the adipocytes have lipid droplets (e.g., as evaluated by known methods in the art such as microscopy). In an exemplary embodiment, the fat module comprises a plurality of fat module cell culture chambers, and the adipocytes are present in the fat module cell culture chambers, and the fat module cell culture chambers have a shape which ranges from cylindrical to spherical. In an exemplary embodiment, the fat module cell culture chambers have a diameter of from about 450 micrometers to about 2200 micrometers, e.g., from about 700 pm to about 2200 pm, from about 900 pm to about 2100 pm, from about 1100 pm to about 1900 pm, or from about 1300 pm to about 1700 pm . In an exemplary embodiment, the fat module cell culture chamber diameter if from about 1400 micrometers to about 1600 micrometers. In an exemplary embodiment, there are from 6 to 12 cell culture chambers in the fat module. In an exemplary embodiment, there are 8 cell culture chambers in the fat module. In exemplary embodiments, the MPS of the present disclosure may comprise one or more ports, configured to allow loading of cells into the fat module and / or liver module. In some embodiments, the MPS may comprise at least one outlet configured to allow exit of the fluid media from the MPS. In some cases, the at least one outlet is configured to allow exit of fluid media from the liver chamber. In exemplary embodiments, the MPS of the present disclosure may be functionally connected to a microfluidic pump (e.g., a syringe pump or a peristaltic pump), the microfluidic pump configured to perfuse fluid medium through the MPS. In some embodiments, the microfluidic pump may be configured to perfuse fluid medium through the MPS in aPCT / US25 / 5230902 December 2025 (02.12.2025)continuous closed circuit (i.e., fluid medium which exits the MPS, e.g., via the at least one outlet, is then re-introduced to the MPS via, e.g., the at least one inlet). In some embodiments, the microfluidic pump may be configured to perfuse fluid medium through the MPS in a single-pass flow (i.e., such that any fluid medium exiting the MPS, e.g., via the at least one outlet, is not reintroduced into MPS). In an exemplary embodiment, the microphysiological system further comprises macrophages. In an exemplary embodiment, the macrophages are derived from induced pluripotent stem cells (e.g., human induced pluripotent stem cells). Methods for producing macrophages from induced pluripotent stem cells are known in the art and described, e.g., in W02020051453A1 and Matsuo et al. Bone 153, 116129 (2021), the disclosures of which are incorporated herein by reference. In an exemplary embodiment, the human induced pluripotent stem cells are cultured on Matrigel. In an exemplary embodiment, the human induced pluripotent stem cells are contacted with hematopoietic progenitor-induction media, macrophage induction media comprising M-CSF, and a combination thereof. In some embodiments, the macrophage induction media may comprise M-CSF in the range of from about 1-1000 ng / mL, e.g., from about 50-150 ng / ml. In an exemplary embodiment, the macrophages are polarized macrophages (e.g., Ml -macrophages). In an exemplary embodiment, the microphysiological system further comprises Ml -macrophages. In some embodiments, the macrophages may be polarized to Ml -macrophages by contacting the macrophages with LPS (lipopolysaccharide), IFNy, or a combination thereof. In some cases, the macrophages are contacted with LPS in the range of from about 1-100 ng / ml LPS, e.g., from about 5-15 ng / ml LPS. In some cases, the macrophages are contacted with IFNY inthe range of from about 1-100 ng / ml, e.g., from about 10-40 ng / ml. In an exemplary embodiment, the macrophages are present in the fat module (e.g., in the fat module cell culture chambers). In an exemplary embodiment, the macrophage to the adipocyte cell number ratio is from about 1:1 to 1:30. For example, in some cases, the macrophage to the adipocyte cell number ratio is from about 1:2 to 1:28, e.g., from about 1:4 to 1:26, from about 1:6 to 1:24, from about 1:8 to 1:22, from about 1:10 to 1:20, from about 1:12 to 1:18, or from about 1:14 to 1:16. In an exemplary embodiment, the macrophage to the adipocyte cell number ratio is from about 1:5 to about 1:30.

[0051] In an exemplary embodiment, the macrophage to the adipocyte cell number ratio is from about 1 :9 to about 1 :11. In an exemplary embodiment, the adipocytesPCT / US25 / 5230902 December 2025 (02.12.2025)and / or hepatocytes comprise a nucleic acid that codes for methionyl-tRNA synthetase 1 (MARS1), e.g., MARS1 (NM_004990.4):MRLFVSDGVPGCLPVLAAAGRARGRAEVLISTVGPEDCVVPFLTRPKVPVLQ LDSGNYLFSTSAICRYFFLLSGWEQDDLTNQWLEWEATELQPALSAALYYLV VQGKKGEDVLGSVRRALTHIDHSLSRQNCPFLAGETESLADIVLWGALYPLL QDPAYEPEEESAEHSWFQTESTQEPCQRAAETVEKQQGVEAERPYEQKQPQP SPAEGRAVTNEPEEEEEATESEEEIAMAVTAWEKGEESEPPLRPQQNPVEPVA GERNVLITSALPYVNNVPHLGNIIGCVLSADVFARYSRLRQWNTLYLCGTDE YGTATETKALEEGLTPQEICDKYHIIHADIYRWFNISFDIFGRTTTPQQTKITQD IFQQEEKRGFVEQDTVEQERCEHCARFEADRFVEGVCPFCGYEEARGDQCDK CGKLINAVELKKPQCKVCRSCPVVQSSQHLFLDLPKLEKRLEEWLGRTLPGS DWTPNAQFITRSWLRDGLKPRCITRDLKWGTPVPLEGFEDKVFYVWFDATIG YLSITANYTDQWERWWKNPEQVDLYQFMAKDNVPFHSLVFPCSALGAEDN YTLVSHLIATEYLNYEDGKFSKSRGVGVFGDMAQDTGIPADIWRFYLLYIRPE GQDSAFSWTDLLLKNNSELLNNLGNHNRAGMFVSKFFGGYVPEMVLTPDDQ RLLAHVTLELQHYHQLLEKVRIRDALRSILTISRHGNQYIQVNEPWKRIKGSE ADRQRAGTVTGLAVNIAALLSVMLQPYMPTVSATIQAQLQLPPPACSILLTNF LCTLPAGHQIGTVSPLFQKLENDQIESLRQRFGGGQAKTSPKPAVVETVTTAK PQQIQALMDEVTKQGNIVRELKAQKADKNEVAAEVAKLLDLKKQLAVAEG KPPEAPKGKKKK (SEQ ID NO:1). In exemplary embodiments, the encoded MARS1 comprises a E274G substitution according to the amino acid numbering of SEQ ID NO:1. In some embodiments, the nucleic acid that codes for MARS1 may be stably integrated into the genome of the adipocytes and / or hepatocytes (e.g., via lentiviral transduction) and may be under the control of any suitable constitutive or inducible promoter known in the art. The methionyl-tRNA synthetase encoded by the subject nucleic acid may find use in, e.g., measuring de novo protein synthesis in the adipocytes and / or hepatocytes by facilitating the incorporation of non-canonical amino acids into newly synthesized proteins in the adipocytes and / or hepatocytes.

[0052] The microphysiological system (MPS) of the present disclosure may be fabricated from any suitable material(s). In some cases, the MPS of the present disclosure is comprised of a polymer. In some cases, the polymer is a thermoplastic polymer. Suitable thermoplastic polymers include, without limitation, polystyrene, polyether ketone, polyethylene terephthalate, polyvinyl chloride, polymethylmethacrylate, cyclic olefin copolymer, polycarbonate, polyetherimide, andPCT / US25 / 5230902 December 2025 (02.12.2025)polydimethylsiloxane (PDMS). In some embodiments, the polymer is PDMS. In some cases, the MPS may be fabricated using soft lithography techniques. Methods and materials for fabricating microfluidic devices, suitable for fabricating an MPS of the present disclosure, are known in the art and described, e.g., in US Pat. App. No. 2004 / 0115838, Fiorini, Gina S., and Daniel T. Chiu. BioTechniques 38.3 (2005): 429-446, and Qi, L. et al. Small, 2103157 (2021), the disclosures of which are incorporated herein by reference.III. Methods of testing a drug or a combination of drugs for disease treatment

[0053] In another aspect, the invention provides a method of testing a drug or a combination of drugs for treatment of an adipocyte-associated disease or a hepatocyte-associated disease, comprising: contacting the microphysiological system (MPS) of a preceding claim with the drug or the combination of drugs; and assessing an attribute of the adipocytes, the hepatocytes, and / or the fluid medium; thereby testing the drug or the combination of drugs for treatment of the adipocyte-associated disease or the hepatocyte-associated disease. In some cases, the the drag or the combination of drags may, e.g., be contacted to the MPS via the at least one inlet of the MPS. In an exemplary embodiment, the disease (e.g., the adipocyte-associated disease and / or a hepatocyte-associated disease ) is metabolic dysfunction-associated steatotic liver disease (MASLD), metabolic dysfunction-associated steatotohepatitis (MASH), insulin resistance (pre-diabetes), or type 2 diabetes (T2DM). In an exemplary embodiment, the assessing an attribute comprises measuring the intracellular uptake of fatty acids in the adipocytes and / or the hepatocytes as compared to a suitable control. The intracellular uptake of fatty acids may be measured according to any suitable method known in the art. In an exemplary embodiment, the assessing an attribute comprises measuring the levels of fatty acids and / or glycerol in the fluid medium as compared to a suitable control. The level of fatty acids and / or glycerol in the fluid medium may be indicative of lipolytic activity and may be measured according to any suitable method known in the art. In an exemplary embodiment, the assessing an attribute comprises measuring the expression level of one or more adipocyte-associated disease or hepatocyte-associated disease markers in the adipocytes and / or the hepatocytes as compared to a suitable control. The expression level of the one or more markers may be measured according to any suitable method known in the art (e.g., qPCR, next-generation sequencing, etc.). The one or more adipocyte-associated disease or hepatocyte-associated diseasePCT / US25 / 5230902 December 2025 (02.12.2025)markers may, e.g., include markers associated with adipocyte and / or hepatocyte metabolic functions (e.g., gluconeogenesis and / or lipolysis), and inflammation-associated markers. In an exemplary embodiment, the one or more adipocyte-associated disease or hepatocyte-associated disease markers comprise GLUT4, FATP1, HSL, ADIPOQ, LEP, TNF, IL6, NFKB1, NFKB2, PCK1, GCK, SREBPlc, and any combination thereof. In an exemplary embodiment, the assessing an attribute comprises measuring the level of one or more cytokines in the fluid medium as compared to a suitable control. The level of the one or more cytokines may be measured according to any suitable method known in the art. The cytokines may be, e.g., inflammatory' cytokines and / or adipokines. In an exemplary embodiment, the one or more cytokines comprise TNF-a, ILip, IL6, IL8, IL10, MCP1, Adiponectin, and any combination thereof. In an exemplary embodiment, the assessing an attribute comprises measuring insulin mediated glucose uptake in the adipocytes as compared to a suitable control. The insulin mediated glucose uptake may be measured according to any suitable method known in the art. In an exemplary embodiment, the assessing an attribute comprises measuring insulin mediated glucose production in the hepatocytes as compared to a suitable control. The insulin mediated glucose production may be measured according to any suitable method known in the art. In some embodiments, the suitable control may include a comparable MPS not contacted with the drug or the combination of drugs. In some embodiments, the suitable control may include the MPS prior to contacting the MPS with the drug or the combination of drugs. The drug or combination of drugs to be tested may be any agent or compound of interest. A suitable drug may include, e.g., a small molecule (e.g., any organic or inorganic compound having a molecular weight of 1000 atomic mass units (amu) or less), a peptide or protein (e.g., an antibody or antibody-derivative), a nucleic acid / oligonucleotide (e.g., an anti-sense oligonucleotide, siRNA), a nucleic acid protein complex (e.g., CRISPR / Cas complex), etc.IV. Methods of determining the biological age or a biological sample or a human subject

[0054] In another aspect, the invention provides a method for determining the biological age of a biological sample, comprising: a) obtaining the biological sample; b) assessing a property of the biological sample; c) comparing the property of the biological sample with a database of the property in a human population; and d) determining the biological age of the biological sample based on step c); therebyPCT / US25 / 5230902 December 2025 (02.12.2025)determining the biological age of the biological sample. In an exemplary embodiment, the biological sample is obtained from a human subject (e.g., via a biopsy or blood draw). In an exemplary embodiment, the biological sample is an adipose tissue sample (e.g., subcutaneous or visceral) or a liver tissue sample. The biological sample may be obtained by any suitable method (e.g., biopsy method) known in the art. In an exemplary embodiment, the biological sample comprises in vitro cells. In an exemplary embodiment, the in vitro cells are induced pluripotent stem cells. In an exemplary embodiment, the one or more in vitro cells are adipocytes and / or hepatocytes. In an exemplary embodiment, the adipocytes and / or hepatocytes are derived from induced pluripotent stem cells. Methods for producing adipocytes and / or hepatocytes from induced pluripotent stem cells are known in the art and described elsewhere herein. In an exemplary embodiment, the induced pluripotent stem cells are derived from somatic cells obtained from a human subject. Methods for producing pluripotent stem cells from somatic cells are known in the art and described, e.g., in W02013082509A1 and WO2018026723A1, the disclosures of which are incorporated herein by reference. In another aspect, the invention provides a method for determining the biological age of a human subject or a biological sample obtained from the human subject, comprising: a) obtaining the biological sample from the human subject; b) contacting the biological sample with a microphysiological system (MPS) of the disclosure; c) assessing a property of the adipocytes, the hepatocytes, and / or the fluid medium of the MPS; d) comparing the property of the adipocytes, the hepatocytes, and / or the fluid medium of the MPS with a database of the property in a human population; and e) determining the biological age of the biological sample based on step d); thereby determining the biological age of the subject or the biological sample. In an exemplary embodiment, the biological sample is a serum sample. In an exemplary embodiment of the methods, the property is the expression level, and / or the variance of the expression level, of one or more genes. The expression level, and or the variance (e.g., the coefficient of variation) of the expression level, of the one or more genes may be measured according to any suitable method known in the art (e.g., qPCR (e.g., RT-qPCR), next generation sequencing (e.g., RNA-seq), single-cell sequencing (e.g., scRNA-seq, etc.). In an exemplary embodiment of the methods, the assessing a property comprises measuring the expression level of the one or more genes. The one or more genes may include genes associated with cellular senescence, metabolic function, inflammation, programmedPCT / US25 / 5230902 December 2025 (02.12.2025)cell death, mitochondrial translation, translation, signaling (e.g., cytokine / chemokine signaling), etc. In an exemplary embodiment, the assessing a property comprises measuring the expression level of IL6, CTGF, PTGS2, POU3F1, CXCL8, EGR1, NOG, FOS, CDKN2A, HMGA2, and / or JUN. In an exemplary embodiment, the assessing a property comprises measuring the expression level of GDF15, TNFAIP3, TRPS1, MMP10, CCL2, BDNF, CDKN2B, NRG1, SFN, DDB2, ALOX15B, EGR1, GDF15, CCL20, CXCL2, IL6, PTGS2, and / or PAPP A. In an exemplary embodiment, the assessing a property comprises measuring the variance of the expression level of the one or more genes. In an exemplary embodiment, the assessing a property comprises measuring the variance of the expression level of CCDC107, HCAR2, SPON1, NDRG4, CNBP, RHEB, PELI2, MRPL33, TNIK, PTCH2, RBM12B, TWSG1, ALKBH8, DHRS3, GPX7, FZD7, KCNN3, BCAN, TFPI, ZFP62, ZBED3 AS1, KLHL23, CNTFR, USP3 AS1, PKN0X2, IRS2, AZIN2, MYL12B, VGLL3, CHRNA5, PENK, TMEM269, SRRM3, ZNF589, LINC01560, ELOVL7, and / or PDZD2. In an exemplary embodiment, the assessing a property comprises measuring the variance of the expression level of SLC9A8, AEN, DNAH11, EID2, DMRTA1, EX01, PCMTD2, UBE2E2, KIF2C, MAPKAP1, DUSP10, RGPD3, MLST8, MTOR, MAPK14, PRR5, RICTOR, AURKB, BLM, MSH2, MLH1, and / or DNA2. In some embodiments, the one or more genes may include genes with sexually dimorphic expression. In an exemplar}' embodiment, the assessing a property comprises measuring the expression level, and / or the variance of the expression level, of CYP3A4, CYP2A6, SLC3A1, CYP7A1, ACSL4, BCL6, UTG2B17, ADH1C, CYP2C16, CYP2D6, LEP, SALL1, ESRI, PNPLA4, PGR, AR, PNPLA3, MMP3M GREB1L, FABP4, LPL, GLUT4, FATP1, ADIPOQ, TNF, SREBPlc, CDKN2A, CDKN1A, and a combination thereof.

[0055] In an exemplary embodiment of the methods, the database comprises expression level data, and / or variance of the expression level data, of the one or more genes in a human population. In an exemplary embodiment, the database comprises expression level data, and / or variance of the expression level data, of the one or more genes in adipose tissue samples of a human population. In an exemplary embodiment, the database comprises expression level data, and / or variance of the expression level data, of the one or more genes in liver tissue samples of a human population. In an exemplary embodiment, the human population comprises male and female individuals. In an exemplary embodiment, the human population comprisesPCT / US25 / 5230902 December 2025 (02.12.2025)individuals of different ages. In an exemplary embodiment, the human population comprises individuals from 18 to 80 years of age. In an exemplary embodiment, the methods for determining the biological age of a biological sample or for determining the biological age of a human subject or a biological sample obtained from the human subject, further comprise determining the similarity of the biological sample to different age groups in the human population of the database based on the expression level, and / or variance of the expression level, of one or more genes. In certain embodiments, the different age groups may include ‘young’ and ‘old’ groups. For example, a ‘young’ age group may include individuals from 18-40 years of age, e.g., from 18-30 years of age, or from 20-29 years of age. An ‘old’ age group may include individuals >50 years of age, e.g., >55 years of age, or >60 years of age. In an exemplary embodiment, the methods for determining the biological age of a biological sample or for determining the biological age of a human subject or a biological sample obtained from the human subject, further comprise determining the activity of a functional gene group in the biological sample based on the expression level, and / or variance of the expression level, of one or more genes. In exemplary embodiments, the functional gene group is associated with a biological function (e.g., as annotated by KEGG or GO annotations). Suitable databases include, without limitation publically available GTEx datasets.

[0056] In an exemplary embodiment of the subject methods, the comparing the property comprises applying a computer implemented model to: i) the expression level of the one or more genes in the biological sample, or ii) the expression level of the one or more genes in the adipocytes, hepatocytes, and / or fluid medium of the MPS; wherein the model is configured to: a) determine the biological age of the biological sample; b) determine the similarity of the biological sample to different age groups in the human population of the database; and / or; c) determine the activity of a functional gene group in the biological sample, based on the expression level, and / or variance of the expression level, of the one or more genes, and wherein the model has been trained on the expression level data, and / or variance of the expression level data, of the database of the human population. The computer implemented model described in connection with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or combinations of both. In some embodiments, the computer implemented model may be trained using an unsupervised learning technique, a semi -supervised learning technique, a supervisedPCT / US25 / 5230902 December 2025 (02.12.2025)learning technique, a parallel training technique, a round robin training technique, an attention-based training technique and combinations thereof, as each such technique is known in the art. In some embodiments, the model is trained using one or more of: supervised training, unsupervised training or semi-supervised training. In some embodiments, the model is trained using unsupervised training. Unsupervised learning is a machine learning technique known in the art for training a model to, for example, identify or recognize patterns. Unsupervised learning comprises training a model where pre-assigned labels are not provided to the model with respect to data used to train the model. As a result, applying unsupervised learning to train a model entails the model itself discovering patterns among the training data. In other embodiments, the model is trained using semi-supervised training. Semi-supervised learning is a machine learning technique known in the art for training a model to, for example, identify or recognize patterns. Semi-supervised learning comprises training a model using both labeled and unlabeled training data. In other embodiments, the model is trained using supervised training. Supervised learning is a machine learning technique known in the art for training a model to, for example, identify or recognize patterns. Supervised learning involves training a model here labels are provided to the model. In some embodiments, the model may comprise a classification function such as, without limitation, Random Forest, k-nearest neighbors (KNN), support vector machine (SVM), multilayer perceptron (MLP), and XGBoost classification methods. The dimensionality of the training data may be reduced according to known methods. In some embodiments, the training data (e.g., expression level data) may be reduced by partial least squares discriminant analysis (PLS-DA).

[0057] In exemplary embodiments of the methods, the assessing an attribute comprises detecting de novo synthesized proteins in the adipocytes, hepatocytes, and / or the fluid medium of the MPS. In exemplary embodiments, the de novo synthesized proteins are intracellular and / or extracellular secreted proteins. In exemplary embodiments, a noncanonical amino acid is introduced to the fluid medium of the MPS. In exemplary embodiments, the de novo synthesized proteins incorporate the noncanonical amino acid. In exemplary embodiments, the noncanonical amino acid comprises an azide moiety. In exemplary embodiments, the noncanonical amino acid is azidonorleucine. In exemplary embodiments, the de novo synthesized proteins are labeled with fluorescent or radioactive probes. In exemplary embodiments, the labeling comprises conjugating the fluorescent or radioactivePCT / US25 / 5230902 December 2025 (02.12.2025)probes to the azide moiety via click chemistry. In exemplary embodiments, the de novo synthesized proteins are detected by immunological array and / or mass spectrometry. Methods for labeling de novo synthesized proteins are known in the art and described, e.g., in Dieterich, Daniela C., et al. Proceedings of the National Academy of Sciences 103.25 (2006): 9482-9487, the disclosure of which is incorporated herein by reference.V. Methods of testing a drug or a combination of drugs for attenuation or reversal of biological aging

[0058] In another aspect, the invention provides a method of testing a drug or a combination of drugs for the attenuation or reversal of the biological aging of adipose and / or liver tissue, comprising: a) contacting the microphysiological system (MPS) of the disclosure with a fluid medium comprising human serum from a male or female donor with a chronological age of at least 62 years; b) contacting the product of step a) with the drug or the combination of drugs; and c) assessing an attribute of the adipocytes, the hepatocytes, and / or the fluid medium; thereby testing the drug or the combination of drugs for the attenuation or reversal of the biological aging of adipose and / or liver tissue. In some cases, the the drug or the combination of drugs may, e.g., be introduced to the MPS via the at least one inlet. In an exemplary embodiment, the fluid medium further comprises human plasma-like medium. In an exemplary embodiment, the fluid medium comprises from about 4% to about 25% of human serum, e.g., the fluid medium may comprise human serum in a concentration of from about 5% to about 11%, from about 6% to about 10%, or from about 7% to about 9%. In an exemplary embodiment, the contacting of step b) is immediately following the contacting of step a). In an exemplar}' embodiment, prior to the contacting of step b), the product of step a) is incubated for one or more days (e.g., one or more, two or more, three or more, or four or more days). In an exemplary embodiment, prior to the assessing of step c), the product of step b) is incubated for one or more days (e.g., one or more, two or more, three or more, or four or more days). In a particular embodiment, the product of step b) is incubated for 3-5 days. In a particular embodiment, the assessing an attribute comprises measuring the intracellular uptake of fatty acids in the adipocytes and / or the hepatocytes as compared to a suitable control. The intracellular uptake of fatty acids may be measured according to any suitable method known in the art. In a particular embodiment, the assessing an attribute comprises measuring the levels of fatty acids and / or glycerol in the fluidPCT / US25 / 5230902 December 2025 (02.12.2025)medium as compared to a suitable control. The level of fatty acids and / or glycerol in the fluid medium may be indicative of lipolytic activity and may be measured according to any suitable method known in the art. In a particular embodiment, the assessing an attribute comprises measuring insulin mediated glucose uptake in the adipocytes as compared to a suitable control. The insulin mediated glucose uptake may be measured according to any suitable method known in the art. In an exemplary embodiment, the assessing an attribute comprises measuring insulin mediated glucose production in the hepatocytes as compared to a suitable control. The insulin mediated glucose production may be measured according to any suitable method known in the art. In an exemplary embodiment, the assessing an attribute comprises measuring oxidative DNA damage in the adipocytes and / or the hepatocytes as compared to a suitable control. The oxidative DNA damage may be measured according to any suitable method known in the art. In an exemplary embodiment, the assessing an attribute comprises measuring the expression level of one or more biological age-associated and / or metabolic markers in the adipocytes and / or the hepatocytes as compared to a suitable control. In an exemplary embodiment, the one or more biological age- associated and / or metabolic markers comprise CDKN1A, CDKN1B, LIFR, TNF, ILG, PPARG, FABP4, LPL, HSL, LEP, EBF2, NFKB1, SREBP1C, PCK1, and any combination thereof. In an exemplary embodiment, the assessing an attribute comprises measuring the expression level of one or more sexually dimorphic markers as compared to a suitable control. In an exemplary embodiment, the one or more sexually dimorphic markers comprise CYP3A4, CYP2A6, SLC3A1, CYP7A1, ACSL4, BCL6, UTG2B17, ADH1C, CYP2C16, CYP2D6, LEP, SALL1, ESRI, PNPLA4, PGR, AR, PNPLA3, MMP3M GREB1L, FABP4, LPL, GLUT4, FATP1, ADIPOQ, TNF, SREBPlc, CDKN2A, CDKN1 A, and a combination thereof. In exemplary embodiments of the subject methods, the assessing an attribute comprises measuring the variance of the expression level of the markers discussed herein. The expression level, and / or variance of the expression level, of the markers may be measured according to any suitable technique known in the art (e.g., qPCR (e.g., RT-qPCR), next generation sequencing (e.g., RNA-seq), single-cell sequencing (e.g., scRNA-seq, etc.). In an exemplary embodiment, the assessing an attribute comprises measuring the level of one or more cytokines in the fluid medium as compared to a suitable control. In an exemplary embodiment, the one or more cytokines comprise TNF-a, IL1 P, IL6, IL8, IL10, MCP1, Adiponectin, and any combination thereof. In anPCT / US25 / 5230902 December 2025 (02.12.2025)exemplary embodiment, the assessing an attribute comprises measuring SA-p-Gal activity in the adipocytes and / or the hepatocytes as compared to a suitable control. SA-0-Gal activity may be measured according to any suitable method known in the art. In an exemplary embodiment, the assessing an attribute comprises measuring the nuclear accumulation of p 16 in the adipocytes and / or the hepatocytes as compared to a suitable control. The nuclear accumulation of pl 6 may be measured according to any suitable method known in the art. In some exemplary embodiments, the assessing an attribute comprises applying the subject methods for determining the biological age of a biological sample, as disclosed herein, to the adipocytes and / or the hepatocytes of the MPS. In some embodiments, the suitable control may include a comparable MPS not contacted with the drug or the combination of drugs. In some embodiments, the suitable control may include the MPS prior to contacting the MPS with the drug or the combination of drugs. The drug or combination of drugs to be tested may be any agent or compound of interest. The drug or combination of drugs to be tested may be any agent or compound of interest. A suitable drug may include, e.g., a small molecule (e.g., any organic or inorganic compound having a molecular weight of 1000 atomic mass units (amu) or less), a peptide or protein (e.g., an antibody or antibody-derivative), a nucleic acid / oligonucleotide (e.g., an anti-sense oligonucleotide, siRNA), a nucleic acid protein complex (e.g., CRISPR / Cas complex), etc.

[0059] In exemplary embodiments of the methods, the assessing an attribute comprises detecting de novo synthesized proteins in the adipocytes, hepatocytes, and / or the fluid medium of the MPS (e.g., as compared to a suitable control). In exemplary embodiments, the de novo synthesized proteins are intracellular and / or extracellular secreted proteins. In exemplary embodiments, a noncanonical amino acid is introduced to the fluid medium of the MPS. In exemplary embodiments, the de novo synthesized proteins incorporate the noncanonical amino acid. In exemplary embodiments, the noncanonical amino acid comprises an azide moiety. In exemplary embodiments, the noncanonical amino acid is azidonorleucine. In exemplary embodiments, the de novo synthesized proteins are labeled with fluorescent or radioactive probes. In exemplary embodiments, the labeling comprises conjugating the fluorescent or radioactive probes to the azide moiety via click chemistry. In exemplary embodiments, the de novo synthesized proteins are detected by immunological array and / or mass spectrometry. Methods for labeling de novoPCT / US25 / 5230902 December 2025 (02.12.2025)synthesized proteins are known in the art and described, e.g., in Dieterich, Daniela C., et al. Proceedings of the National Academy of Sciences 103.25 (2006): 9482-9487, the disclosure of which is incorporated herein by reference.

[0060] In another aspect, the invention provides a method for controlling cell and tissue aging, comprising: al) contacting the microphysiological system (MPS) of the disclosure with fluid medium comprising human serum from a male or female donor with a chronological age of at least 62 years, thus creating an old circulatory milieu; bl) assessing the effects of the old circulatory milieu; cl) contacting the product of step al) with fluid medium comprising human serum from a male or female donor with a chronological age of from about 21 years to about 34 years, thus creating a young circulatory milieu; dl) assessing the effects of the young circulatory milieu; el) establishing a hallmark and / or a process of aging, which is controlled significantly differently by the young circulatory milieus versus the old circulatory milieus; fl) establishing the timing and the dose of the human serum that control a hallmark and / or a process of aging upward by the old circulatory milieu as compared to the young circulatory milieu. Alternatively, the invention provides a method for controlling cell and tissue aging, comprising: a2) contacting the microphysiological system (MPS) of any one of claims 1-37 with fluid medium comprising human serum from a male or female donor with a chronological age of from about 21 years to about 34 years, thus creating a young circulatory milieu; b2) assessing the effects of the young circulatory milieu; c2) contacting the product of step al) with fluid medium comprising human serum from a male or female donor with a biological age of at least 62 years, thus creating an old circulatory milieu; d2) assessing the effects of the old circulatory milieu; e2) establishing a hallmark and / or a process of aging, which is controlled significantly differently by the old circulatory milieu versus the young circulatory milieu; f2) establishing the timing and the dose of the human serum that control a hallmark and / or a process of aging upward by the young circulatory milieu as compared to the old circulatory milieu.

[0061] In an exemplary embodiment of the methods, the fluid medium further comprises human plasma-like medium. In an exemplary embodiment, wherein the fluid medium comprises from about 4% to about 25% of human serum (e.g., the fluid medium may comprise human serum in a concentration of from about 5% to about 11%, from about 6% to about 10%, or from about 7% to about 9%). In an exemplaryPCT / US25 / 5230902 December 2025 (02.12.2025)embodiment, prior to the assessing of step dl) or d2), the product of step al) or a2) is incubated for one or more days (e.g., two or more, three or more, or four or more days). In an exemplary embodiment, prior to the assessing of step bl) or b2), the product of step cl) or c2) is incubated for one or more days (e.g., two or more, three or more, or four or more days). In an exemplary embodiment, the assessing the effects comprises measuring the intracellular uptake of fatty acids in the adipocytes and / or the hepatocytes as compared to a suitable control. In an exemplary embodiment, the assessing the effects comprises measuring the levels of fatty acids and / or glycerol in the fluid medium as compared to a suitable control. In an exemplary embodiment, the assessing the effects comprises measuring insulin mediated glucose uptake in the adipocytes as compared to a suitable control. In an exemplary embodiment, the assessing the effects comprises measuring insulin mediated glucose production in the hepatocytes as compared to a suitable control. In an exemplary embodiment, the assessing the effects comprises measuring oxidative DNA damage in the adipocytes and / or the hepatocytes as compared to a suitable control. In exemplary embodiments, the assessing the effects comprises measuring the expression level of one or more biological age-associated and / or metabolic markers in the adipocytes and / or the hepatocytes as compared to a suitable control. In exemplary embodiments, the one or more biological age-associated and / or metabolic markers comprise CDKN1A, CDKN1B, LIFR, TNF, ILG, PPARG, FABP4, LPL, HSL, LEP, EBF2, NFKB1, SREBP1C, PCK1, and any combination thereof. In exemplary embodiments, the assessing the effects of comprises measuring the expression level of one or more sexually dimorphic markers. In exemplary embodiments, the one or more sexually dimorphic markers comprise CYP3A4, CYP2A6, SLC3A1, CYP7A1, ACSL4, BCL6, UTG2B17, ADH1C, CYP2C16, CYP2D6, LEP, SALL1, ESRI, PNPLA4, PGR, AR, PNPLA3, MMP3M GREB1L, FABP4, LPL, GLUT4, FATP1, ADIPOQ, TNF, SREBPlc, CDKN2A, CDKN1A, and a combination thereof. In an exemplary embodiment, the assessing the effects comprises measuring the level of one or more cytokines in the fluid medium as compared to a suitable control. In an exemplary embodiment, the one or more cytokines comprise TNF-a, IL10, IL6, IL8, IL10, MCP1, Adiponectin, and any combination thereof. In an exemplary embodiment, the assessing the effects comprises measuring SA-0-Gal activity in the adipocytes and / or the hepatocytes as compared to a suitable control. In an exemplary embodiment, thePCT / US25 / 5230902 December 2025 (02.12.2025)assessing the effects comprises measuring the nuclear accumulation of pl 6 in the adipocytes and / or the hepatocytes as compared to a suitable control.

[0062] In exemplary embodiments of the methods, the assessing an attribute comprises detecting de novo synthesized proteins in the adipocytes, hepatocytes, and / or the fluid medium of the MPS. In exemplary embodiments, the de novo synthesized proteins are intracellular and / or extracellular secreted proteins. In exemplary embodiments, a noncanonical amino acid is introduced to the fluid medium of the MPS. In exemplary embodiments, the de novo synthesized proteins incorporate the noncanonical amino acid. In exemplary embodiments, the noncanonical amino acid comprises an azide moiety. In exemplary embodiments, the noncanonical amino acid is azidonorleucine. In exemplary embodiments, the de novo synthesized proteins are labeled with fluorescent or radioactive probes. In exemplary embodiments, the labeling comprises conjugating the fluorescent or radioactive probes to the azide moiety via click chemistry. In exemplary embodiments, the de novo synthesized proteins are detected by immunological array and / or mass spectrometry. Methods for labeling de novo synthesized proteins are known in the art and described, e.g., in Dieterich, Daniela C., et al. Proceedings of the National Academy of Sciences 103.25 (2006): 9482-9487, the disclosure of which is incorporated herein by reference.

[0063] In exemplary embodiments of the methods, the methods further comprise transfecting the adipocytes and / or hepatocytes of the MPS with a nucleic acid. In exemplary embodiments, the nucleic acid is a DNA or RNA. In exemplary embodiments, the RNA is a small interfering RNA (siRNA). In exemplary embodiments, the transfecting occurs prior to the contacting of step al) or step c2). In exemplary embodiments the transfecting occurs following the contacting of step al) or step c2).

[0064] The invention is further illustrated by the Examples that follow. The Examples are not intended to define or limit the scope of the invention.EXAMPLES

[0065] The following Examples illustrate the synthesis of representative compounds used in the invention and the following Reference Examples illustrate the synthesis of intermediates in their preparation. These examples are not intended, nor are they toPCT / US25 / 5230902 December 2025 (02.12.2025)be construed, as limiting the scope of the invention. It will be clear that the invention may be practiced otherwise than as particularly described herein. Numerous modifications and variations of the invention are possible in view of the teachings herein and, therefore, are within the scope of the invention.EXAMPLE 1Functional characterization and scaling of iADIPO-MPS and iHEP-MPS

[0066] The effects of MPS culture on iHEPs generated with a protocol designed to optimize metabolic function were determined (Groeger, et al. Nature communications 14, 3902 (2023)). For this, the cells were loaded into an MPS designed as previously reported (Lee-Montiel, et al. Frontiers in Pharmacology 12, 667010 (2021)). It was found that iHEPs benefit from MPS culture as evidenced by higher albumin and urea secretion over 15 days compared to conventional static culture (FIG. 1A, IB). In addition, the iHEP-MPS showed physiological responses to insulin and glucagon of hepatic glucose production (HGP) and glucose and lipid metabolism-associated gene expression (FIG. 1C, ID).

[0067] Next, it was investigated whether iADIPO-iHEP co-culture using a transwell system allows for analysis of lipid uptake and release dynamics (FIG. 5A). For this, iADIPOs were cultured in 24-well plates and preloaded them with fluorescent fatty acids for 24 hours, after which they were co-cultured with iHEPs in transwell inserts for 24 hours. Fatty acid uptake by iHEPs increased after induction of lipolysis with isoproterenol in iADIPOs (FIG. 5B, 5C). Moreover, insulin-mediated reduction of glucose output by iHEPs was impaired under lipolysis-inducing conditions when the iADIPO-iHEP ratio was increased to 5:1; HGP continued to be insulin responsive under basal conditions, i.e., without induction of lipolysis (FIG. 5D, 5E).

[0068] Next, an interconnected iADIPO-iHEP MPS allowing for scalable cell ratios was developed (FIG. IE, IF). iADIPO-iHEP ratios ranging from 1:1 to 30:1 were investigated under basal conditions. To reliably quantify lipid accumulation in iHEPs, absorption of fluorescent fatty acids by the PDMS used to build the iHEP-MPS and iADIPO-MPS was analyzed, which showed no bias for up to 4 days (FIG. 6A). It was found that preloaded fluorescent fatty acids were spontaneously released from the iADIPO-MPS into the flow-through media and transferred to the iHEP-MPS (FIG.1G, 1H) Both total and fluorescent fatty acids in iHEPs increased as a linear function of iADIPO number (Pearson R>0.99 and p-values<0.05), which confirmed the utilityPCT / US25 / 5230902 December 2025 (02.12.2025)of fluorescent fatty acid uptake as a proxy for total fatty acid accumulation (FIG. II).Analysis of insulin regulation of HGP after 48 hours of iADIPO-iHEP-MPS interconnection showed insulin resistance only at the highest iADIPO-iHEP ratio of 30:1 (FIG. 1J). Because this ratio is unphysiologically high (Hatton, I. A. et al. Proceedings of the National Academy of Sciences 120, e2303077120 (2023)), these findings suggest that accumulation of metabolically healthy WAT mass alone is not sufficient to drive HIR and its hepatic complications.Inflamed iADIPO-MPS causes steatosis and insulin resistance in iHEP-MPS

[0069] Recently, it was shown that activated proinflammatory Ml-iMACs cause insulin resistance independent of steatosis in iHEPs in transwell culture (Groeger et al. Nature communications 14, 3902 (2023)). To determine the effect of WAT inflammation on glucose and lipid metabolism in hepatocytes, Ml-iMACs were introduced into iADIPO-MPS for 24 hours at an Ml-iMAC-i ADIPO ratio of 1:10 before interconnection to iHEP-MPS at an iADIPO-iHEP ratio of 5:1. After 48 hours of interconnection, the inflamed Ml -iADIPO-MPS showed decreased intracellular uptake of fluorescent fatty acids compared to the iADIPO-MPS cultured without iMACs (FIG.2A-2D), reflecting higher lipolytic activity (FIG. 2E). Consequently, the iHEP-MPS showed increased fatty acid uptake and lipid accumulation under inflamed iADIPO-MPS conditions (FIG. 2C, 2D). As expected, inflammation markers were increased in the Ml -iADIPO-MPS, including gene expression of the proinflammatory cytokine TNF. Key adipokines were also altered, with increased gene expression of LEP, which encodes leptin and inhibits lipogenesis and increases lipolysis, and decreased gene expression of ADIPOQ, which encodes insulinsensitizing adiponectin (FIG. 2F). Similarly, the iHEP-MPS showed increased expression of the inflammatory genes TNF and NFKB1 / 2 (FIG.2G). In accordance, higher levels of proinflammatory cytokines were found in the media circulating between the iADIPO-MPS and iHEP-MPS (FIG. 2H). Analysis of metabolic effects revealed decreased expression of the insulin-responsive glucose and fatty-acid transporter genes GLUT4 and FATP1 in the iADIPO-MPS under inflammatory conditions (FIG. 2F), whereas PCK1 gene expression was increased in the iHEP-MPS, suggesting activation of gluconeogenesis and development of insulin resistance (FIG. 2F, 2G). This finding was confirmed by analysis of insulin-mediated glucose uptake in the iADIPO-MPS (FIG. 21) and HGP in the iHEP-MPS (FIG. 2J), both of which were pathologically altered by inflammation compared to the control condition.PCT / US25 / 5230902 December 2025 (02.12.2025)In contrast, the addition of non-activated (MO) iMACs to the iADIPO-MPS neither increased TNFa secretion nor disrupted insulin-mediated regulation of glucose uptake in the iADIPO-MPS or HGP in the iHEP-MPS, similar to control conditions without iMACs (FIG. 7A-7C).

[0070] To determine the contribution of adipocytes to the alterations in glucose and lipid metabolism in hepatocytes caused by WAT inflammation, iHEP-MPS was interconnected with MPS only containing Ml -iMACs and the results were compared to iHEP-MPS interconnected with Ml-iADIPO-MPS (FIG. 7D). It was found that inclusion of iADIPOs worsened the metabolic alterations in iHEP-MPS, including higher HGP, earlier onset of insulin resistance, higher lipid levels and higher expression of genes reflecting inflammation, gluconeogenesis, lipogenesis and lipid transport (FIG. 7E-7H). These findings show that alterations of adipocytes caused by proinflammatory macrophages disrupt glucose and lipid metabolism in hepatocytes.Rational drug screening in iADIPO-iHEP-MPS reveals distinct effects on inflammation and lipid and glucose metabolism

[0071] To demonstrate the utility of MPS representing the WAT-liver axis for drug discovery, two well-recognized insulin sensitizers (metformin at 10 mM and 1 pM rosiglitazone) and one anti-inflammatory drug (dexamethasone at 0.5 pM) were selected to assess their efficacy in treating obesity-induced MASLD / T2DM in the Ml -iADIPO-iHEP-MPS (FIG. 3A). Drug effects were analyzed 48 hours after MPS connection and excluded drug absorption by PDMS at this time point (FIG.6B). It was found that all of these drugs reduced lipid accumulation in iHEPs (FIG. 3B, 3C).Metformin and rosiglitazone were equally effective in increasing intracellular fatty acids in iADIPOs, resulting in lower amounts of fatty acids circulating in the media, and normalization of insulin-stimulated glucose uptake in the iADIPO-MPS (FIG. 3C, 3D). Dexamethasone had no effect on fatty acid or glucose uptake in the iADIPO-MPS (FIG. 3C, 3D). These findings were consistent with drug-induced changes in GLUT4 and FATP1 gene expression observed in the iADIPO-MPS (FIG. 3E). All of the drugs decreased TNF gene expression in the iADIPO-MPS; they all also decreased secretion of TNFa and increased secretion of adiponectin into the media (FIG. 3E, 3F). However, only metformin and rosiglitazone decreased TNF and PCK1 gene expression in the iHEP-MPS and improved insulin response (FIG. 3G, 3H). These findings show that metformin and rosiglitazone prevent inflammation-induced dysfunction of lipid and glucose metabolism in WAT and hepatocytes.PCT / US25 / 5230902 December 2025 (02.12.2025)Adipose tissue-specific incretin prevents Metabolic dysfunction-Associated Steatotic Liver Disease (MASLD)

[0072] Treatment of obese patients with glucagon-like peptide- 1 receptor (GLP1R) agonists has shown hepatoprotective effects (Newsome, P. N. et al. The New England journal of medicine 384, 1113-1124 (2021)). However, the specific mechanisms of action in the liver are incompletely understood. Moreover, in contrast to other metabolically active cells such as adipocytes (Ejarque, M. et al. Scientific reports 9, 6274 (2019); Jiang, Y. et al. Frontiers in pharmacology 9, 1168 (2018)), whether GLP1R is expressed at relevant levels and metabolically active in hepatocytes is controversial (Jin, et al. American Journal of Physiology-Endocrinology and Metabolism 311, E620-E627 (2016)). Therefore, it was investigated whether the GLP1R agonist semaglutide (at 1 pM) acts differently on iADIPOs, iHEPs and iMACs in separate MPS or conventional monocultures before testing whether it can improve metabolic function in the Ml-iADIPO-iHEP-MPS (FIG.8). It was found that only the iADIPO-MPS responded to semaglutide. Gene expression changes in the iADIPO-MPS included increased expression of GLUT4 and HSL, suggesting higher glucose clearance and lipolysis (FIG. 8A, 8B). Increased gene expression of the general WAT marker PPARG and the subcutaneous WAT marker PNPLA3 as well as ADIPOQ was found, which translated into increased secretion of adiponectin into the media (FIG. 8A-8C). In accordance, GLP1R gene expression was significantly higher in iADIPOs than iHEPs and iMACs, matching its expression levels in primary human WAT and primary' human hepatocytes (FIG. 9A, 9B). None of the iPSC-derived cell types showed a difference in INF gene expression and TNFa secretion was not altered by semaglutide treatment in iADIPO-MPS or Ml-iMAC monocultures (FIG.8A, 8B, 8D-8F).

[0073] To determine whether the observed WAT-specific effects of semaglutide improve metabolic function of hepatocytes, Ml-iADIPO-iHEP-MPS were treated with semaglutide (at 1 pM) in the circulating media for 48 hours. Increased lipolysis was observed in the iADIPO-MPS, as evidenced by reduced fluorescence intensity and increased fatty acids in the circulating media (FIG. 4A, 4B), which was accompanied by increased HSL gene expression (FIG. 4C). Surprisingly, lipid accumulation in the iHEP-MPS also decreased after semaglutide treatment compared to the untreated control (FIG. 4A, 4B). These changes were accompanied by significantly higher GLUT4 and ADIPOQ gene expression in the iADIPO-MPS,PCT / US25 / 5230902 December 2025 (02.12.2025)increased secretion of adiponectin into the media, and overall reduced inflammatory activity (FIG.4C, 4D). Foremost, TNFa secretion from the iADIPO-MPS into the circulating media decreased 30-fold (FIG. 4D). In accordance, after semaglutide treatment, the iHEP-MPS showed decreased expression of the inflammatory genes TNF and NF KB 1 / 2, which was accompanied by reduced PC KI gene expression (FIG.4E). Expression of the lipogenesis-associated genes ACACA, ACACB and FASN was also decreased, whereas expression of PPARA was increased in the iHEP-MPS, suggesting increased 0-oxidation, together explaining the finding of reduced lipid accumulation in the iHEP-MPS after semaglutide treatment (FIG.4A, 4B, 4E). In addition, semaglutide improved insulin sensitivity, resulting in restoration of insulin-mediated glucose uptake in the iADIPO-MPS and insulin-mediated suppression of HGP in the iHEP-MPS (FIG. 4F, 4G), probably mediated by its anti-inflammatory effect reducing TNFa levels (Qi, L. et al. Small, 2203725 (2023); Groeger, M. et al. Nature communications 14, 3902 (2023)).

[0074] To confirm the tissue-specific effects of semaglutide in the Ml-iADIPO-iHEP-MPS, a non-circulating flow-through interconnection was employed to selectively stimulate the Ml -iADIPO-MPS or the iHEP-MPS component (FIG.4H).For this, semaglutide (at 1 pM) was added for 48 hours either after or before the Ml -iADIPO-MPS component. The latter configuration was used in combination with the semaglutide antagonists exendin 9-39 (at 1 pM) and compound 5D (at 1 pM) (FIG.4H). While selective semaglutide treatment of the Ml -iADIPO-MPS component was as effective as treatment of the entire Ml-iADIPO-iHEP-MPS, selective semaglutide treatment of the iHEP-MPS component was largely ineffective as evidenced by higher lipid accumulation and impaired insulin-mediated suppression of HGP (FIG. 41, 4J).Furthermore, gene expression markers of inflammation, gluconeogenesis and lipogenesis were higher in the iHEP-MPS -targeting condition compared to the condition targeting the Ml -iADIPO-MPS (FIG. 4K). These findings show that semaglutide prevents lipid accumulation and insulin resistance in hepatocytes by reducing WAT inflammation.Discussion

[0075] The WAT-liver axis maintains metabolic homeostasis and plays an important role in driving metabolic dysfunction in obese patients (Duwaerts et al. Cellular and molecular gastroenterology and hepatology 7, 749-761 (2019). To model thePCT / US25 / 5230902 December 2025 (02.12.2025)interaction of WAT and hepatocytes in the development of MASLD, an iPSC-based multi-tissue MPS that not only allows for upscaling of the WAT component but also considers inflammation as a disease driver was developed.

[0076] Consistent with previous results (Qi, L. et al. Small, 2203725 (2023); Groeger, M. et al. Nature communications 14, 3902 (2023)) and human studies (Du Plessis, J. et al. PloS one 11, e0166048 (2016)), the model herein shows that Ml-iMAC-mediated inflammation readily induces metabolic dysfunction, including increased WAT lipolysis and induction of systemic insulin resistance. In contrast, it was found that the cellular ratio of iADIPOs to iHEPs needs to be escalated to 30:1 to induce HIR in the absence of Ml macrophage-mediated inflammation. Given that ratios of adipocytes to hepatocytes range from 1:10 in lean to 1:1 in obese humans (Hatton, I. A. et al. Proceedings of the National Academy of Sciences 120, e2303077120 (2023); Salans, et al. The Journal of clinical investigation 52, 929-941 (1973)), these findings show that healthy WAT causes metabolic dysfunction in hepatocytes only if the mass is expanded to clinically irrelevant levels. Therefore, these findings suggest that WAT inflammation, rather than mere WAT expansion, is required for the development of MASLD in obese individuals (Fabbrini, et al. Hepatology 51, 679-689 (2010)).

[0077] To further understand the relationship of inflammation and steatosis, it was tested tested how the model responds to drugs clinically used for T2DM and MASLD therapy. Metformin, a widely used insulin sensitizer reported to promote AMPK activation and inhibit mitochondrial respiration (Barbe et al. British journal of pharmacology 117, 907-913 (1996)), and rosiglitazone, a PPARy agonist that increases adipogenesis (Benvenuti, S. et al. Journal of endocrinological investigation 30, RC26-RC30 (2007)), was used. In addition to these insulin-sensitizing drugs, dexamethasone was tested because of its anti-inflammatory properties (Tsurufuji, S. et al. Nature 280, 408-410 (1979)). Although dexamethasone was effective in reducing inflammatory markers in the Ml-iADIPO-iHEP-MPS, it failed to improve insulin sensitivity, probably because of its known insulin-desensitizing effects on adipocytes and hepatocytes (Qi, L. et al. Probing Insulin Sensitivity with Metabolically Competent Human Stem Cell-Derived White Adipose Tissue Microphysiological Systems. Small, 2103157 (2021); Buren, et al. European Journal of Endocrinology 146, 419-429 (2002); Luan, G. et al. Molecules 24, 1982 (2019). Metformin and rosiglitazone showed comparable beneficial effects on lipolysis, inflammation and insulin resistance in the iADIPO-MPS, but rosiglitazone was more effective thanPCT / US25 / 5230902 December 2025 (02.12.2025)metformin in reversing inflammation and insulin resistance in the iHEP-MPS. These findings suggest superior efficacy of rosiglitazone against MASLD, particularly when considering the high metformin concentration used here as compared to serum levels found in patients treated with this drug (Slaughter, et al. Scientific Reports 11, 13159 (2021); Howell, J. J. et al. Cell metabolism 25, 463-471 (2017); Cao, J. et al. J Biol Chem 289, 20435-20446 (2014); Tiikkainen, M. et al. Diabetes 53, 2169-2176 (2004)).

[0078] Previous studies showed that GLP1R agonists are effective in improving obesity-related hepatic dysfunction (Petit, et al. Diabetes & Metabolism 43, 2S28-22S33 (2017); Bifari, F. et al. Pharmacological research 137, 219-229 (2018); Niu, S. et al. Frontiers in Endocrinology 13, 1046130 (2022)) but presumably through indirect effects (Newsome, et al. Incretins (GLP1 r agonists and dual, triple agonists) and the liver. Journal of Hepatology (2023). Studying the specific role of the WAT-liver axis in the semaglutide response using mouse models has been difficult due to the strong effects on feeding behavior and body weight (Garvey, W. T. et al. Nat Med 28, 2083-2091 (2022)) and because GLP1R expression has been found in various cell types, including macrophages (Lee, Y.-S. et al. Diabetologia 55, 2456-2468 (2012); Wan, S. & Sun, H. Experimental and Therapeutic Medicine 17, 3573-3579 (2019)) and adipocytes (Ejarque, M. et al. Scientific reports 9, 6274 (2019); Jiang, Y. et al. Frontiers in pharmacology 9, 1168 (2018); Seino, et al. Journal of diabetes investigation 1, 8-23 (2010)). The human iPSC-based system iADIPOs, but not iHEPs or iMACs, responded to semaglutide treatment in monoculture (FIG. 8). In accordance, little GLP1R gene expression in iHEPs and primary human hepatocytes was found, which confirms previous findings made in whole liver (Pyke, C. & Knudsen, L. Endocrinology 154, 4-8 (2013); Pyke, C. et al. Endocrinology 155, 1280-1290 (2014)). Thus, the multi-tissue model enabled the investigation of the adipocytespecific effects of semaglutide on inflammation-driven MASLD.

[0079] Although semaglutide appears to only target adipocytes directly, it had profound effects on WAT inflammation and HIR. These effects could be due to altered adipocyte-macrophage interactions dampening secretion of proinflammatory cytokines, such as TNFa, from either macrophages and / or adipocytes, similar to what was reported in a mouse model of T2DM (Lee, et al. Diabetologia 55, 2456-2468 (2012)). While semaglutide induced lipolysis in WAT, resulting in increasedPCT / US25 / 5230902 December 2025 (02.12.2025)circulating fatty acids, the interconnected iHEP-MPS showed reduced steatosis and improved insulin sensitivity.

[0080] By generating different cell types from the same human iPSC line, this approach reduces genetic biases such as MASLD risk factors, many of which remain to be discovered (Chen, et al. Nature Genetics 55, 1640-1650 (2023)). However, care needs to be taken to identify and eliminate mechanistically relevant deficiencies in cell maturity and subsequent function of human iPSC-derived cells, including the use of optimized protocols for directed differentiation and in-depth characterization by comparison to primary cells, as was carried out herein for iHEPs, iADIPOs and iMACs (Qi, L. et al. Human iPSC-Derived Proinflammatory Macrophages cause Insulin Resistance in an Isogenic White Adipose Tissue Microphysiological System. Small, 2203725 (2023); Groeger, M. et al. Nature communications 14, 3902 (2023); Qi, L. et al. Probing Insulin Sensitivity with Metabolically Competent Human Stem Cell-Derived White Adipose Tissue Microphysiological Systems. Small, 2103157 (2021); Matsuo, K. et al. Bone 153, 116129 (2021))

[0081] In summary, human iPSC-based multi-tissue MPS faithfully recapitulates the WAT-liver axis allowing for modeling of obesity-induced disorders and analysis of tissue-specific effects of obesity-targeting drugs. This model provides an authentic and scalable in vitro approach to advancing the understanding and therapy of obesity-induced hepatic disease and sets the stage for the next generation of isogenic, multiorgan disease models.Methods

[0082] MPS device fabrication. MPS devices were fabricated in accordance with a previous protocol (Qi, L. et al. Probing Insulin Sensitivity with Metabolically Competent Human Stem Cell-Derived White Adipose Tissue Microphysiological Systems. Small, 2103157 (2021)). Briefly, patterned master templates were fabricated with a thickness of 60 pm by standard photolithography using SU-8 (#3100, MicroChem Corp). The cell chambers were circular with diameter of 1500 pm to provide better adipogenesis support. The microfluidic patterns were replica-molded from the master templates to polydimethylsiloxane (PDMS, Sylgard 184, #NC9285739, FisherScientific) slabs by soft lithography. The inlet / outlets were holed on the media-channel slab using a 0.75 mm biopsy punch (#504529, World Precision Instruments LLC). A polyethylene terephthalate (PET) isoporous membranePCT / US25 / 5230902 December 2025 (02.12.2025)(#030060, TRAKETCH, SABEU GmbH & Co. KG) was activated by oxygen plasma (Plasma Equipment Technical Services) at 60 W under -0.6 Torr for 60 s and then chemically decorated in 2% bis(3-(trimethoxysilyl)propyl)anime solution (#413356, Sigma- Aldrich) in 97% isopropyl alcohol (#A451-4, FisherScientific) and 1% deionized water (Arium Mini, Sartorius) for 30 min at 80°C. After decoration, the membrane was rinsed in pure isopropyl alcohol and stored in anhydrous ethanol solution (#AC611050040, FisherScientific) until further use. For MPS assembly, the PDMS slabs of the cell chamber and media channel were activated by oxygen plasma at 60 W under ~0.6 Torr for 30 s and immediately sandwiched with a size-trimmed decorated PET membrane. To collect cell pellets out of the device, the cell chamber slab was activated for 60 s, resulting -50% higher bonding strength to the membrane than that of 30 s-treated media channel slab. The device was then baked at 110°C for 30 min for ethanol removal, bonding stabilization, and device sterilization.

[0083] iPSC cell culture and differentiation. All experiments used the normal human male GM25256 iPSC line (WTC, hPSCreg: UCSFiOOl-A, Gladstone Institutes, distributed by Coriell Institutes) (Kreitzer et al. American journal of stem cells 2, 119 (2013)). All iPSCs were used at passage numbers ranging from 40 to 80. When iPSC colonies reached 80% confluence, they were dissociated using ReLeSR (#100-0483, Stemcell Technologies) and subcultured at a 1 :10 ratio in mTeSR Plus medium (#100-0276, Stemcell Technologies) on Matrigel-coated substrates (#356231, Corning). Authentication was confirmed prior to experimentation by SNP analysis. Mycoplasma testing was conducted before all experiments and annually. Sterility was checked daily. The undifferentiated state, characterized by colony morphology, was verified before each differentiation. Pluripotency was assessed before all experiments and regularly, based on signature gene expression comparisons to endodermal and mesodermal differentiated states, and statistically analyzed using t-tests.

[0084] iMACs were generated as previously described (Matsuo et al. Bone 153, 116129 (2021)). Briefly, undifferentiated iPSCs were directed to hematopoietic stem cells using STEMdiff Hematopoietic Kit (#05310, Stemcell Technologies). Floating HSCs were collected, then magnetically sorted using anti-CD45 coated beads (#130-045-801, Miltenyi Biotec), and further differentiated to macrophages in ImmunoCult-SF Macrophage Media (#10961, Stemcell Technologies) and 100 ng / mL macrophagecolony stimulating factors (M-CSF; #300-25, Peprotech). Unpolarized MO-iMACsPCT / US25 / 5230902 December 2025 (02.12.2025)were polarized to Ml-iMACs using culture media supplemented with 10 ng / mL LPS (E. coli 0111 :B4; #LPS25, Millipore) and 20 ng / mL IFNy (#300-02, Peprotech) or 24 hours before use in subsequent experiments. Cells were washed and medium was changed to remove the LPS prior to use in experiments.

[0085] Differentiation of iADIPOs were described in a previous study (Qi, L. et al. Probing Insulin Sensitivity with Metabolically Competent Human Stem Cell-Derived White Adipose Tissue Microphysiological Systems. Small, 2103157 (2021)). Briefly, all iPSCs were firstly differentiated into mesenchymal progenitors (iPSC-MSCs) (STEMdiff Mesenchymal Progenitor Kit, #05240, Stemcell Technologies) and then transduced for chemically inducible PPARv. After 48 hours of post-confluent culture, differentiation of iPSC-MSCs was induced in complete media (DMEM / F12 (#11320033, Gibco) containing 1% HEPES (#15630080, Gibco), 1% penicillin / streptomycin (#15140122, Gibco) and 10% fetal bovine serum (#EF-0500-A, Equafetal)) with supplements of 0.25 pM dexamethasone (#D1756, Sigma-Aldrich), 0.25 mM 3-isobutyl-l-methylxanthine (#15879, Sigma- Aldrich), 100 nM rosiglitazone (#R2408, Sigma-Aldrich), and 500 nM insulin (#0002-8315-01, Humulin R, Eli Lilly) for 4 days (Day 0 to 3). Subsequent differentiation was completed in media supplemented with insulin and rosiglitazone at the same concentration (Day 4 to 14). Exogeneous PPARv was induced by 1 pg / mL doxycycline (#D5207, Sigma- Aldrich) from Day 0 until the end. On Day 14, the iADIPOs expressed hallmark genes comparable to primary adipocytes collected from human biopsies (Qi, L. et al. Probing Insulin Sensitivity with Metabolically Competent Human Stem Cell-Derived White Adipose Tissue Microphysiological Systems. Small, 2103157 (2021)).

[0086] iHEPs were generated as previously described (Groeger et al. Nature communications 14, 3902 (2023)). Differentiation was performed at 37°C in 5% CO2 and 5% O2 unless stated otherwise. Endoderm was induced using endoderm-induction media (EIM), consisting of RPMI 1640 (#11875093, Gibco) containing 2% Gem21 without insulin (#400-962-010, GeminiBio), 1% Glutamax (#35050061, Gibco), 1% non-essential amino acids solution (#11140050, NEAA; Gibco), 0.5 mM sodium butyrate (#B5887, Sigma-Aldrich) and 100 ng / ml activin A (#78001, StemCell Technologies), for 7 days in 20% O2. The following compounds were added to EIM during the first 3 days: 3 pM CHIR99201 (#100-1042, StemCell Technologies) on day 1, 20 ng / ml basic fibroblast growth factor (FGFb; #100-18B, Peprotech) and 10PCT / US25 / 5230902 December 2025 (02.12.2025)ng / ml bone morphogenetic protein 4 (BMP4; # 120-05ET, Peprotech) on days 1 and 2, 50 nM PI103 (#501932056, Thermo Fisher Scientific) on days 1 to 3, knockout serum replacement (KSR; # 10828010, Gibco) at 2% on day 1, 1% on day 2 and 0.2% on day 3. On days 8 to 17, cells were cultured in hepatic induction media (HIM), consisting of IMDM (#12440053, Gibco) containing 2% Gem21 without insulin, 1% Glutamax, 1% NEAA, 100 nM dexamethasone (#D4902, Sigma- Aldrich), 100 nM insulin (#800-112-005, GeminiBio) and 0.5 mM 1 -thioglycerol (#M1753, Sigma-Aldrich). The following compounds were added to HIM: 10 ng / ml FGFb and 20 ng / ml BMP4 between days 8 and 17, 20 ng / ml hepatocyte growth factor (HGF; #100-39, Peprotech) between days 12 and 17. On day 10, cells were detached using 0.25% trypsin-EDTA (#25200056, Gibco) and split 1:2 into matrigel-coated 12- or 24-well plates. On days 18 to 22, cells were cultured in Hepatocyte Culture Media BulletKit (HCM; #CC-3198, Lonza) without epidermal growth factor, including 20 ng / ml oncostatin M (#300-10, Peprotech) and 20 ng / ml HGF in 20% O2. Media was changed daily during differentiation and maintenance.

[0087] Primary cells and tissues. Cryopreserved, plateable primary human hepatocytes (Lot: BMO) were purchased from BioIVT; the cells were isolated from a 45-year-old Caucasian male with BMI of 22.6 and no history of excessive alcohol consumption or smoking. Cryopreserved, human pancreatic islets (Lot: 2201233) were purchased from Celprogen; the cells were isolated from a 55-year-old Caucasian male with no history of excessive alcohol consumption or smoking. Primary white adipose tissue, which was used as the standard in a previous study (Qi et al. Probing Insulin Sensitivity with Metabolically Competent Human Stem Cell-Derived White Adipose Tissue Microphysiological Systems. Small, 2103157 (2021)), was biopsied from the subcutaneous fat of a 47 -year-old Caucasian female with BMI of 25.6; fasting blood glucose of 94 mg / dL, A1C of 4.8%; exclusion factors include smoking, unstable weight within the last 3 months (>3% weight gain or loss), a diagnosed inflammatory or infectious disease, liver failure, renal dysfunction, cancer, and reported alcohol consumption of >20 grams per day.

[0088] RNA isolation from MPS and gene expression. Cells were collected by first cutting the sandwiched area off the device. Due to unequal plasma activation on the surfaces of media channel and cell chamber slabs, preferable debonding on the interface between media channel slab and the sandwiched membrane was easily achievable by exfoliation. The membrane was sequentially exfoliated off the cellPCT / US25 / 5230902 December 2025 (02.12.2025)chamber slab to allow direct exposure of the cell pellet. The entire cell chamber slab was then immersed in Trizol (#15596026, Invitrogen) lysis solution with active pipetting to allow efficient lysis of the cell pellet. RNA isolation (#T2010S, NEB), synthesis of cDNA via qRT-PCR (Maxima First Strand cDNA Synthesis Kit for RT- 5 qPCR, # K1641, ThermoFisher), and qPCR (TaqMan™ Universal PCR Master Mix, # 4304437, ThermoFisher) in QuantStudio5 (Applied Biosystems, ThermoFisher) were done according to the manufacturer’s instructions. Oligonucleotide primers for each target gene were pre-designed and synthesized by Integrated DNA Technologies.Relative mRNA expression was determined by the delta-delta-Ct method normalized 10 to PPIA or RPLPO. Primers for qRT-PCR are listed in Table 1. Table 2 shows examples of extracted RNA concentrations and efficiency. All genomic analyses were done within two weeks of collection.Table 1. Primers for qRT-PCR.PCT / US25 / 5230902 December 2025 (02.12.2025)Table 2. RNA yield from 1 well of a 6-well tissue culture plate and 1 MPS. Cells from 1 well of a 6-well plate can be loaded into approximately 4 MPS (ratio 1:4).Measurements are based on iHEPs.PCT / US25 / 5230902 December 2025 (02.12.2025)

[0089] Circulating media analysis. Cytokines in circulating media were analyzed by TNFa ELISA (KHC3011), adiponectin ELISA (KHP0041, both from ThermoFisher) and sent to Eve Technologies (Canada) for analysis with Human5 Cytokine / Chemokines 65-Plex panel (HD65). Additionally, cytokine concentrations were measured in co-culture supernatants using LEGENDplex Human Inflammation Panel 1 (#740809, Biolegend) according to the manufacturer’s instructions, analyzed using a LSRFortessa flow cytometer and quantified using LEGENDplex software (Biolegend). LDH in media was measured using LDH Cytotoxicity Assay (#8078, ScienCell) according to the manufacturer’s instructions. Non-esterified fatty acids (NEFA) were measured using enzymatic colorimetric assay (NEFA-HR(2) kit including reagents of # 434-91795 and # 436-91995, and standard #270-77000, WAKO) according to the manufacturer’s instructions. Fluorescent fatty acids were quantified by measuring fluorescent intensity (EX / EM 480 / 520 nm) of the circulating media. Albumin in media was measured using the Human Albumin ELISA Kit (#E88-129, Fortis Life Sciences) according to the manufacturer’s instructions. Urea in media was measured using the QuantiChrom Urea Assay Kit (#DIUR-100, Bio Assay Systems) according to the manufacturer’s instructions.

[0090] MPS culture. To ensure the adipogenesis of loaded adipocyte progenitor cells, the initial stage of differentiation was pre-induced in a tissue culture flask.Differentiating iPSC-MSCs on day 4, which contain tiny lipid droplets in >80% cells, were dissociated (TrypLE Express, Gibco) and centrifuged as cell pellets. Prior to cell loading, 10 wt% MMP cleavable peptides (CQPQGLAKC, GenScript) were dissolved in triethanolamine-buffer (TEOA; 0.3 M, pH 8, #T0449, Sigma-Aldrich) and directly 25 added to the cell pellet with dissolved 3% adhesion side-chain conjugated collagen I short peptide sequence Cl (Knight et al. J Biol Chem 275, 35-40 (2000)), (CGGGF(HYP)GER, GenScript) under a peptide-to-precursor volume ratio of 1:10.PCT / US25 / 5230902 December 2025 (02.12.2025)The hydrogel-cell slurry with density of 8xl07cells / mL was injected into the cell chamber and maintained for 30 min at room temperature to remove uneven loading stress. The cell number in MPS hereafter is estimated based on cell density in every loading then multiply loading volume. The MPS was incubated at 37°C for 30 min for crosslinking and then connected with catheter couplers (#SC20 / 15 and #SP20 / l 2, Instech Laboratories) and tubes (#06422-00, Cole-Parmer). Culture media was perfused at a flow rate of 10 pL / hour using a syringe pump (#703007, Harvard Apparatus). After Day 14, freely floating cells were flushed out by a continuous media flow at rate of 30 pL / min. Subsequently, macrophages were loaded at density of 5xl06cells / mL into the cell chamber at flow rate of 5 pL / min to yield the ratio of iMACs:iADIPOs at 1 : 10, matching the physiological ratios in obese WAT (Rosen et al. Cell 156, 20-44 (2014); Weisberg et al. J Clin Invest 112, 1796-1808 (2003)). At Day 22, fully differentiated iHEPs were loaded into the MPS and cultured for 1 day to allow tissue formation. iMACs and iADIPOs were co-cultured in iADIPO media, including 50 ng / ml M-CSF on Day 14. After 1-day coculture with iMACs, the iMAC-iADIPO-MPS was interconnected with iHEP-MPS at cell ratio of 5:1 (Fig. 2-4). The interconnected iMAC-iADIPO-iHEP-MPS was cultured in 0.2 ml HCM (without insulin and dexamethasone) supplemented with 50 ng / mL M-CSF and 10 ng / mL HGF and cultured for 48 hours at 20 pL / hour.

[0091] Drug administration. For drug testing, 10 mM metformin (#PHR1084, Sigma-Aldrich), 1 pM rosiglitazone, 0.5 pM dexamethasone, or 1 pM semaglutide (#S9697, Selleckchem) was added to the 0.2 mL circulating media at the beginning of MPS interconnection. To test the tissue-specific effect of GLP1R activation in the Ml-iADIPO-iHEP-MPS, 1 pM semaglutide was added to the medium after leaving the Ml-iADIPO-MPS or before entering the Ml-iADIPO-MPS. The GLP1R antagonists exendin 9-39 (E7269, Sigma) and compound 5D (HY-101116, MedChemExpress) were added to the medium after leaving the Ml-iADIPO-MPS at 1 pM each.

[0092] Absorption of fluorescent fatty acids and drug by PDMS. BSA-complexed fluorescent fatty acids, metformin, rosiglitazone, dexamethasone and semaglutide were diluted in 200 pL deionized water and circulated for 6 or 2 days through a set of interconnected iADIPO-MPS and iHEP-MPS at a flow rate of 20 pL / hour. Fatty acid concentration was measured by fluorescence at excitation / emission 480 / 520 nm and drug concentrations were measured by UV / Vis spectrometry at 240 nm using a monochromator-based microplate reader (SpectraMax i3x, Molecular Devices).PCT / US25 / 5230902 December 2025 (02.12.2025)

[0093] Functional assays for hormone- stimulated iADIPOs and iHEPs. For dynamic faty acid assay, adipocytes were preloaded with 2 pM fluorescent fatty acids (Bodipy 500 / 510 Cl, C12, #D3823, ThermoFisher) in iADIPO media for 48 hours and rinsed by iADIPO media without fluorescent fatty acid for 1 hour before interconnection to avoid fluorescent carry-over. Fatty acids were quantified based on the relative fluorescence unit (RFU) of the respective MPS. Lipolysis in iADIPOs was induced by I pM isoproterenol (#16504, Sigma-Aldrich) or 1 pM isoproterenol plus 1 pM insulin (#12643, Sigma- Aldrich). For adipogenic glucose uptake, iADIPO-MPS were infused with low glucose DMEM (1 g / L, #11885084, Gibco) at 37°C for 3 hours for basal uptake, then with 1 pM insulin stimulation for additional 3 hours for insulin-stimulated uptake. For hepatic glucose production, iHEP-MPS were infused by low glucose DMEM with 2 mM sodium pyruvate (#11360070, Gibco), 10 mM sodium lactate (#L7022, Sigma-Aldrich) including 100 nm glucagon (#05-23-2700, Sigma-Aldrich), or 100 nm insulin for 1 hour. Glucose was measured using the Amplex Red Glucose / Glucose Oxidase Assay kit (#A22189, ThermoFisher) according to the manufacturer’s instructions.

[0094] Immunofluorescence, lipid staining and image acquisition. MPS were fixed using 4% paraformaldehyde (#15710, Electron Microscopy Sciences) for 2 hours. Nuclei and F-actin were stained overnight in PBS including 1% BSA with 300 nM DAPI (#D1306, ThermoFisher) or 1 unit of phalloidin-iFluor 647 reagent (#abl76759, Abeam), respectively. Samples were mounted using Diamond Antifade (#P36965, Invitrogen) after staining and rinsing. Preloaded fluorescent fatty acids were used to track lipid droplet formation in iADIPO-MPS and iHEP-MPS.Brightfield and fluorescent images were captured by a wide-field fluorescence microscope (AxioObserver Zl, Zeiss) and confocal fluorescence microscopes (LSM710, LSM880, Zeiss). To enable fluorescent comparison, all staining and imaging procedures were kept consistent. Zen (Zeiss) and ImageJ (no plug-ins) were used to analyze fluorescent intensity with consistent settings. Absorbance and fluorescence of assay solutions and circulating media were analyzed using a microplate reader (SpectraMax i3x, Molecular Devices).

[0095] Statistics and reproducibility. All results were statistically analyzed using IMP I I (SAS Institute) or Prism (GraphPad). Paired two-tailed t-test with assumption of equal variance was used to compare the difference between two measures from the same MPS. Unpaired two-tailed t-test with the same assumption was used for resultsPCT / US25 / 5230902 December 2025 (02.12.2025)from different MPS. One-way analysis of variance (ANOVA) for 3 or more samples in one group and two-way ANOVA for multiple groups with 3 or more samples each group, and then post-hoc test for multiple comparisons with corrections were performed to determine significant pair(s): Tukey’s test was used in most analyses except in FIG 1J where Dunnett’s test was used to determine the significance between iADIPO:iHEP ratios and iADIPO control. p<0.05 was considered significant (*), p<0.005 highly significant (**). Results are shown as mean values with error bars representing standard deviation (SD). All experiments were repeated independently (biological replicates, n) with similar results at least 3 times (except FIG. 9B).EXAMPLE 2Functional characterization and scaling of iADIPO-MPS and iHEP-MPS Aging of WAT-MPS and Liver-MPS by human sera and pro-gerontic knock-on effects of aged fat on liver.

[0096] hiPSC-derived MPSs (iADIPO- and iHEP-MPS, also simplified as WAT- and liver-MPS) were perfused for 4 days with Human Plasma-Like medium (HPLM) containing 5% serum from old male or female human donors (>62 years), which we term old circulatory milieu (OCM). Alternatively, MPSs were exposed to serum from sex matched young donors (21-34 years) to establish a young circulatory milieu (YCM. See Methods for details). We assessed multiple hallmarks of aging, including cellular senescence, oxidative DNA damage, i nflammatory gene expression, as well as changes in the expression of the aging-associated adipocyte progenitor marker LIFR ( Qiong, W. Innov Aging 6(Suppl 1), doi:10.1093 / geroni / igac059.1263. (2022)), and several key molecular and functional features of WAT and liver metabolisms. Results demonstrate that OCM quickly and robustly established multiple gerontic phenotypes in WAT- and liver-MPS, both general and tissue-specific (FIG. 10).

[0097] With respect to the general hallmarks of aging, human tissue senescence was increased by OCM based on senescence associated beta-galactosidase (SA-beta-gal) staining, expression of CDKN1A, and CDKN2A as well as the increased nucleus-localized and total protein levels of p 16 as determined by immunofluorescent intensity and quantification by ELISA (FIG. 10A-10D, 10L10L). We detected higher levels of senescence-associated secretory protein (SASP, TNF-a, IL-6) expression and secretion (FIG. 10E, 10M). Oxidative DNA damage, measured by 8-OHdG assay, was also robustly induced by OCM in the WAT-MPS with a similar, albeit weaker, trend observed in the liver MPS (FIG. 10F, 10N).PCT / US25 / 5230902 December 2025 (02.12.2025)

[0098] Next, we examined adipose and hepatic metabolic markers and functions. As compared to YCM, OCM induced markers of adipogenesis and adipose lipid metabolism (PPARG, FABP4, LPL, HSL), as well as obesity / aging-related adipokines (LEP), and increased visceral fat depot signatures, i.e., higher EBF2 (FIG. 10G). Furthermore, OCM not only increased lipid synthesis but also elevated the uptake of exogenous fatty acids by the WAT- and liver-MPSs (FIG. 10H, 10P, 10Q). In liver-MPS OCM induced a trend for elevated expression of the lipogenic master regulator SREBPlc (0.05<p<0.1 FIG. 10P, 10Q).

[0099] In addition to lipid metabolism, glucose metabolism was also perturbed by OCM, indicative of an insulin resistant state: insulin-regulated hepatic gluconeogenesis marker PCK1 was increased, insulin-induced adipose glucose uptake and regulation of hepatic glucose production were both impaired (FIG. 10H, 10P, 10Q). Of note, when compared to the serum free conventional MPS medium, YCM and OCM drive WAT-MPS genes (CDKN2A, CDKN1A, LIFR, TNF, IL6, FABP4, LPL, HSL, EBF2) and TNF-a secretion, as well as liver-MPS genes (Cl) KN I A. SREBPlc, PCK1) in opposite directions (FIG.10).

[0100] Importantly, both YCM and OCM also established dimorphic patterns based on the sex of serum donors (FIG. 15A, 15B). In the WAT-MPS this included differential gene expression for adipogenesis markers LPL, GLUT4, FATP1, AD1POQ, LEP), adipokines (ADIPOQ, LEP), insulin-sensitive glucose and lipid metabolisms (GLUT4 and FATP1), as well as trends for sex-based differences in inflammation and subcutaneous adipogenesis. Similarly, liver MPS that were exposed to male YCM exhibited higher inflammation, lipogenesis, and senescence, as compared to the MPS that were exposed to female YCM.

[0101] Aging-induced dysfunction in visceral adipose tissue adipokine and metabolite secretion directly impacts the liver via the portal vein. Here, we recreated this process by first establishing a cellular memory of exposure to YCM or OCM in WAT-MPS and then interconnected the fat chips with untreated liver MPS via a unidirectional perfusion from the WAT to the liver with the basal HPLM medium (FIG. 11A). The data demonstrated that the OCM, but not YCM, pretreated WAT-MPS quickly induced aging hallmarks in the liver MPS, including higher expression of IL6, CDKN2A, PCK1 and SREBPlc (FIG. 11B), senescence (FIG. 11C), and insulin resistance indicating elevated hepatic glucose production and higher lipidPCT / US25 / 5230902 December 2025 (02.12.2025)accumulation (FIG. 11D). Interestingly, this knock-on hepatocyte dysfunction differed from the OCM driven aging phenotype of liver-MPS based on key signature genes of senescence, inflammation, and glucose metabolism (FIG. HE).

[0102] To further dissect the proteomic drivers of these on-chip aging patterns, we employed WAT-MPSs derived from the hiPSC line stably expressing the methionyl-tRNA synthetase mutant (MetRSL274G) for bio-orthogonal non-canonical amino acid tagging (BONCAT) (FIG. 23A). This system enables incorporation of azido norleucine (ANL) into newly synthesized proteins, which can be then detected via click chemistry on the azide group (Liu, Y. et al. Nature communications 8, 643 (2017); Sviercovich et al. Ageing Research Reviews, 102641 (2024): Liu, C. et al. Rejuvenation Research 25, 95-109 (2022); Burgess, J. D. et al. Stem Cell Research & Therapy 14, 289 (2023)). Using this approach, we found that OCM rapidly promoted de-novo protein synthesis along activation of inflammatory signaling pathways (FIG.23B). Conversely, YCM induced the groups of newly synthesized growth factors that promote cell proliferation, developmental pathways, and core homeostatic biological processes (FIG.23C). Notably, the changes in proteomic output under OCM also mirrors the broadly observed age-associated decline in gene expression reported in natural aging and parabiosis studies (Palovics, R. et al. Nature 603, 309-314 (2022)).

[0103] Taken together, these results establish a rapid and sex-specific in vitro system of aging using human iPSC derived adipocytes and hepatocytes that, within 4 days of OCM treatment, shows key features of aging including senescence with secretion of inflammatory / SASP markers, oxidative DNA damage, blunted insulin sensitivity with dysregulated glucose and lipid metabolism. Further, OCM exposure established a cellular memory that propagated aging related dysfunctions from one tissue to another even in the absence of continued exposure to OCM.On-chip aging closely recreates natural human aging based on genome-wide transcriptomics

[0104] Bulk RNAseq analysis was performed on the 4-day serum-treated WAT-MPS. Principal component analysis (FIG 12A) (PCA) distinguished two signature clusters grouped by YCM and OCM. Differential analysis demonstrated that the on-chip heterochronicity of human serum influenced biomedically relevant and known to be age-specific gene expression patterns: immune-tissue cross-talks and regulators of inflammation (TNF, interleukins, chemokines, interferons, their receptors andPCT / US25 / 5230902 December 2025 (02.12.2025)associated proteins), metabolism (adipokine signaling pathways, SLCs, PCK1, CYP2E1, ACCs), the aging hallmarks of ADP-ribosylase activity, cytosolic DNA-sensing, DNA damage, lifespan-relevant T0RC2, regulators of cell fate and cell proliferation (SOXs, POU5Fs, FATs), and tissue remodeling factors (COLs, LAMAs, MMP28, FBN2) (FIG 12B, FIG 16).

[0105] To determine whether on-chip aging faithfully recreates patterns of human aging, we compared our RNAseq data with the human subcutaneous adipose tissue (SAT) dataset of GTEx (gtex_v8) (Lonsdale, J. et al. Nature genetics 45, 580-585 (2013)). We found that the top 10 overlapping GO terms agreed between the human studies and the heterochronic WAT MPS and, importantly, suggested increased inflammation in OCM and improved metabolic processes in YCM (FIG 12C).

[0106] In an even more stringent test of the biomedical relevance of the on-chip aging, we compared the absolute age of fat MPS in YCM vs. OCM with the chronological age of human fat donors. To this end, we developed a transcriptomebased age-predictor machine learning model for human WAT. This approach avoids biases in transcript feature selection, minimizing arbitrary errors and is comprehensive: it includes hundreds to thousands of transcript features per input, which is more than sufficient for the number of studied aging biomarkers. This machine learning pipeline was designed to be universally applicable to the studies of WAT aging. The models were trained on 70% of randomly selected samples from the GTEx dataset, with separate training based on sex and fat depots; in testing on the remaining 30%, the models demonstrated 90 % accuracy for male SAT and 92% for female SAT; 97% for male VAT, and 94% for female VAT (FIG 12D, 18, 30, 31).Next, we used trained models on the SAT human datasets and tested them for predicting the biological age of the WAT MPS that were perfused with YCM vs OCM. We found a remarkably quick equilibration of hiPSC-derived tissue age toward the age of the serum donors (FIG 12E).

[0107] Fat tissue ages early in life (Ou et al. Cell death & disease 13, 300 (2022); Ding, Y. et al. Comprehensive human proteome profiles across a 50-year lifespan reveal aging trajectories and signatures. Cell (2025)), and is instrumental to the aging of other organ systems, thus, we focused our transcriptomics on WAT-MPS; yet interestingly, we found that several pathways in aging, e.g. cytokine interaction, were similarly changed by the serum heterochronicity in the liver-MPS (FIG 28C).PCT / US25 / 5230902 December 2025 (02.12.2025)Similarly, transcriptomic analyses further confirms that the aged fat-liver interactions share numerous characteristics with serum heterochronicity but also have distinctive features. (FIG 28G, 28H).

[0108] To further investigate these phenomena, we compared liver aging phenotypes induced directly by serum (YCM- vs. OCM-treated liver, or CMs-alone) and indirectly via aged fat (OCM-pretreated-WAT interconnected liver, vs. YCM or CMs-pretreated-WAT interconnected to hepatocyte MPS). In a rigorous validation of biomedical significance, aging associated genes from these data sets were further compared with the in-vivo human liver aging signatures from the GTEx database (FIG 24A) All three aging contexts showed some overlapping GO terms, e.g., tissue development and immune responses, but the two liver-MPS models exhibited distinct aging signatures. Liver aging induced by OCM shared metabolic and oxidative DNA damage signatures with the GTEx human aging in vivo, as well as uniquely featured TGF superfamily signaling. In contrast, OCM-treated-WAT induced liver aging with a more pronounced inflammaging profile than the direct OCM treatment; importantly, the inflammatory terms were also reflected in the in-vivo GTEx liver data. A biological age prediction model for human liver was limited in its performance, due to the small sample size (161 male and 65 female in GTEx), yet it slightly differentiated the predicted ages between male YCM- and OCM-treated liver-MPSs, (FIG 31).

[0109] Summarily, the global gene expression that is rapidly established by heterochronic serum on-chip reflects the patterns of aging in human populations, and the age of human MPS is influenced by the age of serum donors.Novel biomarkers and mechanisms of plasticity in human WAT aging

[0110] To confirm and extrapolate the relevance of our findings to the fundamental process of human aging, we performed more comprehensive comparisons between the RNAseq on the heterochronic MPS and the public datasets, CellGen, GenAge, Aging Atlas.

[0111] First, we focused on identifying the biomarkers of human aging that show the same differences on-chip (YCM vs OCM) and in vivo (young vs old humans), the conserved determinants behind systemically -influenced plasticity of human WAT aging. We gated the union set of the Aging datasets by the GTEx SAT dataset (FIG 13A), which yielded 24 overlapped genes, 11 of which also overlapped with our WAT-MPS RNAseq, the 11 novel biomarkers of aging that exist in human populationPCT / US25 / 5230902 December 2025 (02.12.2025)and are established in 4 days on-chip by human serum. These 11 genes had an average 2.57-fold increase with aging in the GTEx and had on-average 1.91-fold increase in the OCM comparing to YCM, in strong agreement with the human tissue aging in the MPS. Confirming the biomedical significance of our findings and expanding the understanding of aging, these 11 biomarkers include central hub genes CXCL8 (IL8), JUN, FOS, CDKN2A, IL6, and the networks of aging-associated pathways, such as TNF, cellular senescence, NOD-like, TGF-beta / SMAD, (FIG 13B, FIG 19A-19B).

[0112] Not only did the global mRNA levels of on-chip aging correlated with the natural human aging but the same was also true for another hallmark of aging, transcriptional noise. Comparison between the GTEx and the MPS RNAseq datasets revealed 37 novel biomarkers of transcriptional noise, which have increased standard deviation among older people and in the WAT -MPS exposed to the OCM, as compared to YCM, (FIG 13C). These biomarkers of transcriptional noise, plus 10 STRING-predicted partner genes (all with combined confidence scores>0.99), underlay the systemically regulated plasticity of human tissue aging, interestingly providing novel links to mTOR and insulin signaling and outlining several networks that were less studied in aging, (FIG 13D, FIG 19C-19D). With advanced threshold (log2(fold change)>0.5), 5 most fluctuated genes together with 5 predicted partners identified mTOR signaling pathway as the core network with aging-related increase in transcriptional noise.Novel biomarkers and mechanisms in human liver aging

[0113] Based on a similar workflow, we identified and compared aging in liver-MPS under two conditions. At the expression level (FIG 24B), 8 aging-associated genes induced by OCMs-alone were primarily growth factors (BDNF, GDF15, NRG1), whereas 15 genes induced by OCM-pretreated-WAT included not only growth factors (BDNF, GDF15) but also inflammatory mediators, most prominently IL6. Both conditions shared CDKN2B, which, although isolated from the two primary networks, could independently form a PPI network with five predicted partners (confidential score>0.95), enriched in TGFp-associated senescence pathways. Focusing on biological noise (FIG 24C), the OCMs-alone condition showed strong association with mTOR signaling, consistent with the core PPI network identified in WAT-MPS and highlighting mTOR dysregulation as a central feature of heterochronic CMs. In contrast, the heterochronic CM-pretreated-WAT showed dysregulated maintenancePCT / US25 / 5230902 December 2025 (02.12.2025)and repair of chromosomal components without involvement of mTOR, a reasonable outcome given that the heterochronic serum factors were no longer present in the culture environment.

[0114] Summarily, we describe novel biomarkers and mechanisms behind the systemically influenced plasticity of human tissue aging, both, on the levels of gene expression and transcriptional noise.Testing rejuvenation strategies for vital human tissues

[0115] MPS are well-known for their ability to support drug screens but had not been applied to testing rejuvenation strategies. Here, we used the human on-chip aging for drug screens in two key directions: attenuation of aging and reversal of aging. In the first approach, WAT -MPS was aged by OCM for 4-days, in the presence vs. absence of a senolytics drug combination of dasatinib (200 nM) plus quercetin (20 pM) called DQ, or an ALK5 inhibitor of aging-elevated TGF-beta signaling (ALK5i, 342 nM), or the aging-diminished hormone oxytocin (OT, 15 pM), or mTOR inhibitor rapamycin (Rapa, 20 nM), (FIG 25A). In the second approach on pharmacological reversal of aging, after 4-days in OCM alone, i.e. the aging, WAT-MPS were continued in the OCM for 4 more days in the presence or absence of DQ, Alk5i, OT, or Rapa (FIG 25B) As a variant of the approach on reversal of aging, we also examined if dilution of old serum would synergize with the DQ, Alk5i, OT, Rapa, i.e., after the 4-days in OCM, MPS were washed with the base medium (5% KSR, 0% old serum) and then kept for additional 4-days, in the presence vs. absence of DQ, Alk5i, OT, or Rapa, (FIG 14A-C, 25C)

[0116] All four drugs, at all regimens as well as dilution of old serum attenuated expression of the senescence marker CDKN2A in the WAT-MPS, (FIG. 14A-C, 21A-C, 25A-C). However, not all parameters of aging were attenuated in the continues presence of OCM by the tested drugs with ALK5i and OT showing weak effects on adipogenesis (FIG 25A,B). The most robust rejuvenation that included diminished SASP and improved metabolic parameters was observed when old serum dilution was combined with the tested longevity drug candidates (FIG 14C, 25C). Moreover, in this setting, DQ and oxytocin robustly increased expression of the insulin-regulated glucose transporter, GLUT4, leading to functional recovery of insulin sensitivity in the glucose uptake assay. Oxytocin was the most effective in overriding old serum, decreasing senescence, reducing inflammation, promoting the highest expression ofPCT / US25 / 5230902 December 2025 (02.12.2025)multiple important adipogenesis and function markers, FABP4, LPL, HSL, ADIPOQ, PNPLA3, and improving both glucose and lipid metabolisms (FIG 14C, 25C).

[0117] To test the functional relevance of regulatory nodes identified by transcriptomic analysis, we established in-device siRNA based knockdown strategies for the adipocyte MPS. Based on our aging hallmark analyses, a key biomarker of senescence, CDKN2A (pl 6) emerged as a central regulatory node, showing strong associations with aging and functionally linking the other 10 markers through upstream and downstream signaling pathways. For example, knockdown of CDKN2A suppresses senescence-associated cell cycle arrest, attenuates the autocrine pro-aging effects of SASPs, such as IL6 and IL8 (Buj, et al. Aging (Albany NY) 13, 3290 (2021), and enhances the long-lasting metabolic health in mice (Wu, R. et al. Nature communications 14, 2731 (2023)). To test if these pl6-related rejuvenation features would manifest in our human aging-on-chip, we performed siRNA-mediated knockdown of CDKN2A both concurrently with and following OCM treatment (FIG 25D,E). In both scenarios, CDKN2A knockdown markedly reduced aging phenotypes and enhanced metabolic function, with greater efficacy observed when knockdown was applied after OCM removal (i.e., in combination with serum dilution).

[0118] Inspired by the systems of heterochronic parabiosis (Conboy et al. Nature 433, 760-764 (2005)) and heterochronic blood exchange (Rebo, J. et al. Nature communications 1, 13363 (2016); Jeon, O. H. et al. Nature metabolism 4, 995-1006 (2022)), we next examined whether switching from OCM to YCM can reverse the experimental on-chip aging for WAT-MPS and liver-MPS. The MPS were sequentially perfused with medium containing 5% serum from old donors for 4 days and then were switched to medium containing 5% serum from young donors for another 4 days (OY). Control isochronic MPSs were treated for 8 days with medium containing 5% serum from old donors (OO).

[0119] As compared to OO, the OY serum switch in WAT-MPS resulted in lower levels of CDKN2a, and higher levels of the metabolism markers: GLUT4, FATP1, HSL, accompanied by recovery of insulin sensitivity in glucose uptake assay and reduced fatty acid uptake (FIG 14D,E). Based on the bulk RNAseq of WAT-MPS, the OY switch promoted transcriptional rejuvenation of immune responses, metabolism, cell fates, tissue remodeling, rejuvenated CEBPB, CEBPD, SLC2A4 (GLUT4) network, and normalized TGF-beta / Smad signaling, which suggestsPCT / US25 / 5230902 December 2025 (02.12.2025)improvements in adipogenesis, insulin sensitivity, glucose metabolism, and signal transduction (FIG. 21D,E). The exposure of WAT-MPS to OCM after YCM pretreatment (YO) induced pro-gerontic features, as compared to (YY), including DNA damage responses via p53 (FIG. 21G), and the patterns of higher adiposity and insulin resistance (FIG. 22H)

[0120] For the liver-MPS, the switch from old to young serum (OY) reduced several inflammation markers, TNF, NFKB1, NFKB2 and decreased senescence, based on the levels of CDKN1A, all as compared to OO (FIG 14F,G). A key marker of glucose metabolism, PCK1, and a determinant of lipid metabolism, SREBPlc, were attenuated, accompanied by the healthier physiologic responses: less glucose production, better insulin sensitivity and lower lipid accumulation.

[0121] Overall, above results demonstrate that not only is human tissue aging rapidly induced on-chip, but that human tissue rejuvenation is equally rapid, and moreover, by only some approaches, enabling selective screens for longevity therapeutics.Discussion

[0122] This study provides insights on the paradigm of plasticity in aging, its inducibility and attenuation, and, for the first time, applies these concepts to vital human tissues ex vivo. While we acknowledge that WAT and liver consist of more cell types than adipocytes and hepatocytes, respectively, and thus that the WAT- and liver-MPS are a simplification of complex tissues, OCM driven on-chip aging of these human iPSC-derived cells accurately reflects physiological in vivo process, and is in agreement with clinical observations while significantly improving our understanding beyond animal models and cell cultures (Harms, M. J. et al. Cell reports 27, 213-225. e215 (2019); Jeon, O. H. et al. Nature metabolism 4, 995-1006 (2022); Huffman, D. M., Csiszar, A. & Ungvari, Z. Vol. 43 111-113 (Springer, 2021); Qi, L. et al. Nature communications 15, 7991 (2024)). It informs on the key age-related influences of human systemic milieu on health, function, and cross-talks of human tissues, in realtime. The discovered attributes of human aging, e.g. novel signaling networks, knock-on effects of fat aging on liver, sexual polymorphism and tissue memory of age would have been hard or impossible to uncover using previous approaches (Park, S. et al. Aging Cell 23, el4070 (2024)), including previously published reports with animal alternative aging models, provided in the following Table 3.PCT / US25 / 5230902 December 2025 (02.12.2025)Table 3. Comparison to animal alternative aging models&PCT / US25 / 5230902 December 2025 (02.12.2025)&

[0123] Although sex chromosomes are considered the primary drivers of sexual dimorphism, this study demonstrated that sexually dimorphic responses can also be elicited by the biological sex of the serum donor, even when using a single XY hiPSC 5 line. To our knowledge, such a strategy for inducing sex-biased gene expression in iPSC-derived adipocytes and hepatocytes has not been previously reported. Moreover, serum-induced transcriptomic signatures of sexual dimorphisms in MPSs were, intriguingly, found to be tissue-specific and, when benchmarked against published human datasets (van den Munckhof, I. C. et al. International Journal of Obesity 48, 10 523-532 (2024); Li, Y. et al. Lipids in Health and Disease 24, 50 (2025); Kuipery, A.et al. Frontiers in immunology 13, 818612 (2022)), (FIG 15E-G, 27A-27D), confirmed that sex-specific patterns found in human WAT and liver are faithfully recapitulated on-chip. Specifically, inflammatory and immunoregulatory pathways were higher, while metabolic processes were lower in female serum WAT-MPS (van 15 den Munckhof, I. C. et al. International Journal of Obesity 48, 523-532 (2024); Li, Y.et al. Lipids in Health and Disease 24, 50 (2025)), but such pattern was observed in male serum liver-MPS (Kuipery, A. et al. Frontiers in immunology 13, 818612 (2022)) suggesting that on-chip sexual dimorphisms are driven by complex serum- derived factors with tissue specific effects that go beyond inflammatory cytokines.PCT / US25 / 5230902 December 2025 (02.12.2025)

[0124] Notably, the extended list of SASPs also revealed sex-biased aging phenotypes (FIG 33A). Under OCM versus YCM conditions, male MPSs exhibited stronger differential expression of inflammatory markers, indicating a more pronounced aging response than females, particularly in WAT-MPS. This aligns with the well-documented observation that males generally age faster and have shorter lifespans than females (Blagosklonny, M. V. Aging (Albany NY) 2, 265 (2010)). While at the same YCM condition, female WAT-MPS demonstrated more pronounced proinflammatory responses compared to male WAT-MPS (FIG 33B & 33C).

[0125] Another sex-specific influence was a decreased resolution of the phenotypes that were induced by OCM vs. YCM in the female serum treated MPS. This included a broader distribution in the PCA (FIG 15, 26), higher transcriptional noise (FIG 15H) and lower accuracy or reliability of age-predictions for MPSs (FIG 18, 30). Such on-chip sex-differences agree with the clinical observations that female aging is more complicated, varied (FIG. 15C), and less predictable (FIG. 18B) perhaps due to hormonal fluctuations, pregnancy, and menopause (Shirazi, et al. Scientific reports 10, 20522 (2020); Ziomkiewicz, A. et al. PloS one 11, e0145753 (2016); Ryan, C. P. et al. Scientific reports 8, 11100 (2018)). Additionally, the models we used for female aging phenotypes are expected to be less powerful in predictive power due to fewer available samples for machine learning (445 male and 218 female SAT samples; 371 male and 170 female VAT samples; 161 male and 65 female liver samples in GTEx). The limited sample size highlights that subcutaneous WAT is a far more practical tissue for model training and aging prediction than liver, as liver sampling via intraperitoneal procedures is considerably more invasive and carries greater risks compared to the relative ease of subcutaneous adipose sampling.

[0126] The comprehensive comparison of the MPS RNAseq with human transcriptomics, including through our novel ML age-predictor, demonstrate that the on-chip-controlled aging closely recreates in vivo human aging. This transforms the paradigm of biological aging, reducing its reliance on progression of time and highlighting the pivotal role of circulatory environment in time-independent induction and reduction of human tissue aging.

[0127] As a technical limitation we should note that our results are based on bulk RNAseq data rather than single-cell RNAseq approaches (Bartz, et al. InternationalPCT / US25 / 5230902 December 2025 (02.12.2025)journal of molecular sciences 24, 3701 (2023) and thus studied noise in gene expression using population noise rather than cell-to-cell noise. However, given the limited availability of larger cohorts of single-cell RNAseq data for mature adipocytes, our analysis strategy was the currently best option.

[0128] Our system showed that inflammatory cytokines and growth factors, the two major components of SASPs (Yue, Z. et al. Frontiers in Immunology 13, 1019313 (2022); Wang, et al. Nature Reviews Molecular Cell Biology 25, 958-978 (2024)), were primary drivers of aging, but their roles differed between WAT and liver: OCM-induced liver aging showed a stronger dependence on growth factors (BDNF, GDF15, NRG) rather than inflammatory cytokine (CCL2). This was accompanied by activation of the growth-factor-responsive cell cycle regulator CDKN2B (pl 5) via the TGF-p / SMAD pathway, and induction of an inflammation suppressor, TNFAIP3 (A20) to inhibit the NF-KB pathway. In contrast, WAT aging showed a clear inflammatory propensity, characterized by an aging loop involving immediate -early transcription factors JUN, FOS, EGR1), the senescence center CDKN2A (pl 6), and key inflammatory triggers, modulators, and SASPs (PTGS2, IL6, IL8) that reinforce autocrine senescence and propagate paracrine aging signals.

[0129] TGF-p, a growth factor and SASP component that is elevated during aging (Wang et al. Nature Reviews Molecular Cell Biology 25, 958-978 (2024)), may serve as a central regulator distinguishing WAT from liver aging, as it suppresses adipogenesis (Lee, M.-J. Biochimica et Biophy sica Acta (BBA)-Molecular Basis of Disease 1864, 1160-1171 (2018)) while promoting hepatogenesis (Sanchez, A. etal. Experimental cell research 242, 27-37 (1998); Sanchez, A. & Fabregat, I. World journal of gastroenterology: WJG 16, 5148 (2010). This role is strongly supported by our findings: in WAT-MPS, YCM and OCM induced divergent genotypic and phenotypic outcomes, whereas in liver-MPS, both conditions consistently converged toward enhanced developmental and hepatogenic programs.

[0130] Interestingly, OCM-pretreated WAT caused liver aging with similar features to that of OCM-induced aging and both processes closely resembled in-vivo aging (overlapped GO terms and aging biomarkers of expression and biological noise). Growth factors (BDNF, GDF15) and aging-related regulatory pathways (CDKN2B) as well as inflammation suppressor (TNFAIP3) were upregulated in the liver exposed to OCM-aged WAT. However, inflammaging signatures (EGR1, PTGS2, IL6),PCT / US25 / 5230902 December 2025 (02.12.2025)proinflammatory cytokines / SASPs (CCL20, CXCL2) were also elevated. Based on the secretome analyses of OCM vs. YCM-treated-WAT (FIG 23B, 23C, 33) and TNFa ELISA (FIG 10E), the heterochronic interorgan signatures were due to SASPs secreted from the OCM-aged WAT and hepatogenic growth factors secreted from the YCM-exposed youthfill WAT (BMP4 (Fan, J. et al. J Cell Physiol 220, 72-81 (2009)), FGFs (Padrissa-Altes, S. etal. Gut 64, 1444-1453 (2015)).

[0131] WAT inflammation is tightly linked with aging, but can also be induced by other pathological events, such as obesity resulting in the recruitment of proinflammatory (Ml) macrophages (Qi, L. et al. Small, 2203725 (2023)). Although obesity and aging share many similar hallmarks and feed forward each other, termed by some as obesageing or adipaging (Perez, L. M. et al. The Journal of physiology 594, 3187-3207 (2016); Ghosh, et al. Alzheimer's & Dementia 20, e085889 (2024)), the understanding of their differences is still limited. Here we were able to expand on this important topic by comparing the on-chip aging to on-chip inflammation driven by isogenic iPSC derived Ml -like activated macrophages, as in our previous study (Qi, L. et al. Small, 2203725 (2023)) (FIG 20). Both conditions manifested inflammatory features, however, WAT remodeling with a strong interleukin production occurred after the Ml -infiltration, whereas OCM-treated WAT had severe alterations in transcriptional and translational processes that are common in aging and are observed in heterochronic parabiosis (Palovics, R. et al. Nature 603, 309-314 (2022)). This approach highlights the capacity of our system to distinguish the clinically relevant details of inflammation, aging and obesity (Ou, et al. Cell death & disease 13, 300 (2022)).

[0132] While not exhaustive, we tested multiple anti-gerontic strategies to demonstrate the value of our aging-on-a-chip system for longevity drug discovery applications. Serum dilution, though a simple approach, proved highly effective in removing aging-increased systemic factors, thereby attenuating gerontic phenotypes on chip and enhancing the efficacy of additional interventions. This finding aligns with the growing clinical interest in therapeutic plasma exchange (TPE) as an emerging rejuvenation modality (Fuentealba, M. et al. Aging Cell, e70103 (2025)). Oxytocin that had efficacy in reversing the on-chip aging, demonstrated anti-gerontic effects in vivo, in mice: simultaneously promoting lipolysis, reducing inflammation, and correcting visceral fat redistribution (Li, E. et al. Nature 625, 175-180 (2024); Szeto, et al. Lipids in health and disease 19, 1-11 (2020); Yuan, et al. Molecular andPCT / US25 / 5230902 December 2025 (02.12.2025)cellular endocrinology 514, 110903 (2020); Cho, S. Y. et al. British Journal of Dermatology 181, 1216-1225 (2019); Buemann, B. Current Aging Science 15, 218-228 (2022)). Inhibition of TGF-J3 signaling through ALK5 also produced robust rejuvenation effects on-chip that are in agreement with the in vivo alleviated TGF-p / SMAD3-driven senescence, and simultaneously rejuvenation of muscle, liver, and brain in mice (Mehdipour, M. et al. Aging (Albany NY) 11, 5628 (2019); Yousef, H. et al. Oncotarget 6, 11959 (2015)). Furthermore, in an excellent evolutionary and in-vivo to MPS conservation, treatment of already old and frail mice with oxytocin plus Alk5i, extended their lifespan and healthspan by over 72%; intriguingly, only in males (Kato, C. et al. Sex-specific longitudinal reversal of aging in old frail mice. Aging 17) and allowed both male and female old mice to exercise without heart fibrosis and inflammation (Cruz, J. M. C. et al. J Cell Physiol 240, e70054 (2025)).

[0133] Our WAT-MPS exhibited a clear tissue memory of aging, including complex processes ranging from epigenetic regulation to persistent senescence. Although certain phenotypes were partially alleviated, aging signatures persisted in OCM-treated WAT for at least four days after serum withdrawal, as shown in the OO control versus dilution condition (FIG 14A-C, 25C). This retained aging memory was propagated to interconnected liver-MPSs through the circulating medium in a para- / endocrine-like manner (FIG 11A-E), resulting in phenotypes uniquely observed in OCM-treated WAT-MPS (e.g., aging markers IL6, PTGS2, EGR1) but absent in OCM-treated liver-MPS (FIG 24A-C). Such persistence of memory may also explain the blunted rejuvenating effect of YCM in heterochronic microfluidic conditions (OY vs. OO in (FIG 14D-G), where TNF and IL6 unchanged; FIG 22B, where TNF signaling remained upregulated), as well as the reduced pro-aging impact of OCM that still kept good metabolisms and upregulated PPAR signaling pathway (YO vs. YY in FIG 22H). These phenomena bear similarities to the obesogenic memory observed in human WAT that leaves the tissue prone to long-term inflammation and insulin resistance even after obesity is reduced (Schmitz, J. et al. Molecular metabolism 5, 328-339 (2016); Blaszczak, A. M. et al. Immunometabolism 2, e200023 (2020)). Similarly, a history of obesity, particularly in infancy and childhood, makes it easier to regain and harder to lose weight (Rundle, A. G. et al. Childhood Obesity 16, 226-233 (2020); Morales et al. Nutrients 13, 2132 (2021).

[0134] Given our success with rapidly inducing and reducing aging phenotypes in hiPSC derived adipocytes and hepatocytes in the MPS, this approach can be likelyPCT / US25 / 5230902 December 2025 (02.12.2025)expanded to other iPSC derived cell types and organ-on-a-chip models. There is an expectation of a new, powerful, widely applicable tool for advancing the understanding of human aging, lifestyle-related negative and positive factors: diet (Li, V. L. et al. Nature 606, 785-790 (2022)), exercise (Wei, W. et al. Cell metabolism 35, 5 1261-1279. el211 (2023)), alcohol (Oh, H. S.-H. et al. Nature Medicine, 1-9 (2025)), etc., as well as systematic evaluation of diverse anti-gerontic strategies.MethodsEthical Statement

[0135] This research complies with applicable ethical regulations. The study protocols receive annual approval from the Committee on Laboratory and Environmental Biosafety (CLEB), the Stem Cell Research Oversight (SCRO) Office, and the Office of Environment, Health & Safety (EH&S) at the University of California, Berkeley.

[0136] All human serum samples used in this research were obtained from a commercial vendor (Metaphor Laboratory LLC), ensuring compliance with U.S. Department of Health and Human Services regulations. As these samples were fully de-identified by the vendor, no identifying information is accessible, and they do not qualify as "human subjects" under research regulations. Essentially, the samples are anonymous and not linked to any living individual.

[0137] This study utilized de-identified blood serum samples from healthy donors of sub-Saharan African descent, including young donors (ages 21-34) and older donors (ages 62 and above), sourced from Metaphor Laboratory LLC. Table 4 provides detailed catalog and biometric information, including donor sex (male or female) and exact ages. BOLD font labels the sera that were individually tested for the25 universality of serum-induced effects in WAT-MPS. After confirming, pooled serum was made by mixing equal volume of the rest of five sera in each condition.Table 4PCT / US25 / 5230902 December 2025 (02.12.2025)

[0138] MPS fabrication. MPS devices were fabricated in accordance with our previous protocol. Briefly, patterned master templates were fabricated by standard photolithography using SU-8 (#3100, MicroChem Corp) then replica-molded to polydimethylsiloxane (PDMS, Sylgard 184, #NC9285739, FisherScientific) slabs by soft lithography. The inlet / outlets were holed using a 0.75 mm biopsy punch (#504529, World Precision Instruments LLC). The cell chambers were circular with diameter of 1500 pm and thickness of 60 pm. A polyethylene terephthalate (PET) isoporous membrane (#030060, TRAKETCH, SABEU GmbH & Co. KG) was activated by oxygen plasma (Plasma Equipment Technical Services) at 60 W under ~0.6 Torr for 60 s and then chemically decorated in 2% bis(3-(trimethoxysilyl)propyl)anime solution (#413356, Sigma- Aldrich) in 97% isopropyl alcohol (#A451-4, FisherScientific) and 1% deionized water (Arium Mini, Sartorius) for 30 min at 80°C. After decoration, the membrane was rinsed in pure isopropyl alcohol and sandwiched by oxygen plasma activated PDMS slabs of the cell chamber and media channel. The device was then baked at 110°C for 30 min for solvent removal, bonding stabilization, and device sterilization.

[0139] iPSC cell culture and differentiation. All experiments used the healthy human male G15.AO iPSC line (RRID:CVCL_V192), excepting for the de novo proteomic analysis using health human male WTC11 iPSC line (GM25256 hPSCreg:UCSFiOOl-A) with stable expression of a mutant methionyl-tRNA synthetase (MetRSL274G, detailed cell line establishment protocol is described herein). All iPSCs were used at passage numbers ranging from 40 to 80. When iPSC colonies reached 80% confluence, they were dissociated using ReLeSR (#100-0483, Stemcell Technologies) and subcultured at a 1:10 ratio in mTeSR Plus medium (#100-0276, Stemcell Technologies) on Matrigel-coated substrates (#356231, Corning).Authentication was confirmed prior to experimentation by SNP analysis. Mycoplasma testing was conducted before all experiments and annually. Sterility was checked daily. The undifferentiated state, characterized by colony morphology, was verified before each differentiation. Pluripotency was assessed before all experiments andPCT / US25 / 5230902 December 2025 (02.12.2025)regularly, based on signature gene expression comparisons to endodermal and mesodermal differentiated states, and statistically analyzed using t-tests.

[0140] MetRSL274GhiPSCs line establishment. Human iPSCs WTC11 (GM25256 hPSCreg: UCSFiOOl-A) were maintained on vitronectin-coated plates in Essential 8 medium (Thermo Fisher) and transduced at -60-70% confluence with custom lentiviral cDNA expression construct encoding hMARS_L274G (MetRSL274G, RefSeq#: NM_004990.4, Gene ID: 4141) together with TagRFP and a puromycin resistance cassette (pRCDCMURP-CMV-hMars_L274G-UbiC-TagRFP-2A-Puro, packaged lentivirus, Lot#231127004, Cellecta, CA, USA). Viral supernatant (2xl06TU / mE, 500 pl) was mixed 1:1 with culture medium containing polybrene (5 pg / ml) and applied for 6 h, followed by dilution with fresh medium and incubation for a total of 24 h. Puromycin selection was initiated 48-72 h post-transduction, and TagRFP fluorescence was used to monitor transduction efficiency and enrich positive colonies. Stable integration and expression of MetRS were confirmed by fluorescence microscopy, with clonal colonies hand-picked for expansion. All lentiviral manipulations were performed under BSE-2 conditions, with waste decontaminated in 10% bleach. The full plasmid construct is described in FIG. 35.

[0141] Differentiation of iADIPOs were described in our previous study (Qi, L. et al. Probing Insulin Sensitivity with Metabolically Competent Human Stem Cell-Derived White Adipose Tissue Microphysiological Systems. Small, 2103157 (2021)). Briefly, all iPSCs were firstly differentiated into mesenchymal progenitors (iPSC-MSCs) (STEMdiff Mesenchymal Progenitor Kit, #05240, Stemcell Technologies) and then transduced for chemically inducible PPARy. After 48 hours of post-confluent culture, differentiation of iPSC-MSCs was induced in complete media (DMEM / F12 (#11320033, Gibco) containing 1% HEPES (#15630080, Gibco), 1% penicillin / streptomycin (#15140122, Gibco) and 10% fetal bovine serum (#EF-0500-A, Equafetal)) with supplements of 0.25 pM dexamethasone (#D1756, Sigma-Aldrich), 0.25 mM 3-isobutyl-l-methylxanthine (#15879, Sigma- Aldrich), 100 nM rosiglitazone (#R2408, Sigma-Aldrich), and 500 nM insulin (#0002-8315-01, Humulin R, Eli Lilly) for 4 days (Day 0 to 3). Subsequent differentiation was completed in media supplemented with insulin and rosiglitazone at the same concentration (Day 4 to 14). Exogeneous PPARy was induced by 1 pg / mL doxycycline (#D5207, Sigma- Aldrich) from Day 0 until the end. On Day 14, thePCT / US25 / 5230902 December 2025 (02.12.2025)iADIPOs expressed hallmark genes comparable to primary adipocytes collected from human biopsies.

[0142] iHEPs were generated as previously described (Groeger, M. et al. Modeling and therapeutic targeting of inflammation-induced hepatic insulin resistance using human iPSC-derived hepatocytes and macrophages. Nature Communications 14, 3902 (2023)). Differentiation was performed at 37°C in 5% CO2 and 5% O2 unless stated otherwise. Endoderm was induced using endoderm-induction media (EIM), consisting of RPMI 1640 (#11875093, Gibco) containing 2% B27 without insulin (#A1895601, ThermoFisher), 1% Glutamax (#35050061, Gibco), 1% non-essential amino acids solution (#11140050, NEAA; Gibco), 0.5 mM sodium butyrate (#B5887, Sigma-Aldrich) and 100 ng / ml activin A (#78001, StemCell Technologies), for 7 days. The following compounds were added to EIM during the first 3 days: 3 pM CHIR99201 (#100-1042, StemCell Technologies) on day 1, 20 ng / ml basic fibroblast growth factor (FGFb; #100-18B, Peprotech) and 10 ng / ml bone morphogenetic protein 4 (BMP4; # 120-05ET, Peprotech) on days 1 and 2, 50 nM PH 03 (#501932056, Thermo Fisher Scientific) on days 1 to 3, knockout serum replacement (KSR; # 10828010, Gibco) at 2% on day 1, 1% on day 2 and 0.2% on day 3. On days 8 to 17, cells were cultured in hepatic induction media (HIM), consisting of IMDM (#12440053, Gibco) containing 2% B27 without insulin, 1% Glutamax, 1% NEAA, 100 nM dexamethasone, 100 nM insulin, and 0.5 mM 1 -thioglycerol (#M1753, Sigma- Aldrich). The following compounds were added to HIM: 10 ng / ml FGFb and 20 ng / ml BMP4 between days 8 and 17, 20 ng / ml hepatocyte growth factor (HGF; #100-39, Peprotech) between days 12 and 17. On days 18 to 22, cells were cultured in Hepatocyte Culture Media BulletKit (HCM; #CC-3198, Lonza) without epidermal growth factor, including 20 ng / ml oncostatin M (#300-10, Peprotech) and 20 ng / ml HGF. Media was changed daily during differentiation and maintenance.

[0143] MPS culture and coculture. For iADIPO-MPS, differentiating iPSC-MSCs on day 4, which contain tiny lipid droplets in >80% cells, were dissociated (TrypLE Express, Gibco) and centrifuged as cell pellets. The cell slurry with density of 8xl07cells / mL was injected into the cell chamber and maintained for 30 min at room temperature to remove uneven loading stress. The cell number in MPS hereafter is estimated based on cell density in every loading then multiply loading volume. The MPS was connected to catheter couplers (#SC20 / 15 and#SP20 / 12, InstechPCT / US25 / 5230902 December 2025 (02.12.2025)Laboratories) and tubes (#06422-00, Cole-Parmer). Culture media was perfused at a flow rate of 10 pL / hour using a syringe pump (#703007, Harvard Apparatus).

[0144] For iHEP-MPS, fully differentiated iHEPs on Day 22 were loaded into the MPS following same procedures as iADIPO-MPS and cultured in HCM. In knock-on test, the OCM-pretreated iADIPO-MPS was interconnected with iHEP-MPS at cell ratio of 5:1 and infused by HPLM supplemented with 5% KSR at 20 pL / hour.Old and young circulatory milieu treatment

[0145] The effectiveness of young and old circulatory environments was firstly assessed and validated in the WAT-MPSs, which were perfused for 4 days using HPLM (Human Plasma-Like Medium, #A4899101, Gibco) containing 5% human sera from 8 donors: 4 males (2 young and 2 old) and 4 females (2 young and 2 old), with n=3 or 4 per donor. A time-course pilot study with OCM revealed that a 4-day exposure was sufficient to induce robust aging-associated alterations (FIG 34), which was therefore adopted as the standard condition in all subsequent experiments. We confirmed that the induction of aging was consistent across donors despite individual variations, and the response was more pronounced in male sera (FIG 10, 12, 13). Although sexual dimorphism was less prominent in the liver-MPS compared to the WAT-MPS, it was still evident when comparing young male and young female CMs (FIG 15A). To focus on aging and to minimize donor and sex-related variability, HPLM containing 5% pooled male sera from five donors per age group were used as OCM and YCM for liver-MPS, knock-on treatments, and anti-aging interventions (FIG 10, 11, 14).Interconnection of WAT and liver MPS

[0146] Based on our previously established design (Qi, L. et al. Adipocyte inflammation is the primary driver of hepatic insulin resistance in a human iPSC-based microphysiological system. Nature communications 15, 7991 (2024), we interconnected CM-pretreated WAT-MPS and liver-MPS via a microliter-scale tubing system that enabled a single-pass flow from WAT to liver. The cell ratio between i AD IPOs and iHEPs was maintained at 5: 1, as optimized in our prior study, which ensured that the liver-MPS was not directly burdened while remaining sensitive to Epidemic flux generated by lipolysis-activated WAT-MPS. Prior to connection, the WAT-MPS was flushed with HPLM supplemented with 5% KSR under continuous flow at 30 pL / min for 1 minute to remove residual serum-containing medium andPCT / US25 / 5230902 December 2025 (02.12.2025)eliminate potential air bubbles. Afterward, the interconnected WAT-liver-MPS system was perfused with HPLM supplemented with 5% KSR at a controlled rate of 20 pL / hour from WAT to liver using a syringe pump.Drug administration

[0147] For drug testing, senolytic drug combination of 200 nM dasatinib (#SML2589) plus 20 pM quercetin (#Q4951 , both from Sigma-Aldrich); 342 nM ALK5 inhibitor (TGF-a type I receptor kinase inhibitor, #ALX-270-445, Enzo Life Sciences); 15 pM oxytocin acetate (#4016373 or old #H-2510, Bachem), 20 nM rapamycin (#R8781, Sigma Aldrich) was separately added to the testing medium for 4 days. Drugs were administrated in three scenarios: together with OCM for 4 days in non-treated WAT-MPS (OCM+drugs); together with OCM for 4 days in 4day-OCM-pretreated WAT-MPS (OO+drugs); in plain HPLM without old serum for 4 days in 4day-OCM-pretreated WAT-MPS (Dilution+dmgs). Vehicle conditions were set as OCM for 4 or 8 days without drug. All treatments started on Day 14.Functional assays for iADIPOs and iHEPs

[0148] Lor basic lipid accumulation assay, low glucose DMEM (1 g / L glucose, #11885084, Gibco) with 2 pM fluorescent fatty acids (Bodipy 500 / 510 Cl, C12, #D3823, ThermoEisher) was infused in MPSs at 37°C for 2 hours. After rinse, fatty acids within MPS chamber were quantified based on the relative fluorescence unit (RFU) of the respective MPS. For adipogenic glucose uptake, iADIPO-MPS were infused with low glucose DMEM at 37°C for 2 hours for basal uptake, then with 1 pM insulin stimulation (#12643, Sigma- Aldrich) for additional 2 hours for insulin-stimulated uptake. For hepatic glucose production, iHEP-MPS were infused by low glucose DMEM with 2 mM sodium pyruvate (#11360070, Gibco), 10 mM sodium lactate (#L7022, Sigma- Aldrich) at 37°C for 2 hours for basal production then including 100 nm insulin for 2 hours for insulin-inhibited production. Glucose content in each infusion medium was measured using the Amplex Red Glucose / Glucose Oxidase Assay kit (#A22189, ThermoFisher) according to the manufacturer’s instructions. For cytokine TNFa secretion, circulating media were analyzed by ELISA (#KHC3011, ThermoFisher), and assessed relative to a serum-free control condition, with negative values indicating comparatively reduced production. For DNA damage and Pl 6 expression, cell lysate in RIPA buffer (#89901, ThermoFisher) were measured by 8-OHdG ELISA (#IT7974, G-Biosciences) and P16 ELISA (#ab227903,PCT / US25 / 5230902 December 2025 (02.12.2025)Abeam), then normalized to protein concentration (BCA Protein Assay Kit, # 23225, ThermoFisher). All kits were used according to the manufacturers’ instructions. Absorbance and fluorescence of assay solutions and circulating media were analyzed using a microplate reader (SpectraMax i3x, Molecular Devices).Immunofluorescence, lipid staining and image acquisition

[0149] MPS were fixed using 4% paraformaldehyde (#15710, Electron Microscopy Sciences) for 2 hours. Non-specific binding was blocked in PBS containing 1% BSA (#A8806, Sigma) overnight. Permeabilization was activated by 1% Saponin (#558255, Sigma) for 1 hour then maintained until final rinse. Human Pl 6 was sequentially stained by 1% human anti -pl 6 (rabbit recombinant anti-CDKN2A / pl6INK4a, #abl08349, Abeam) overnight then 0.2% anti-rabbit IgG conjugated to Alexa Fluor 647 (#B40926, ThermoFisher) for 2 hours. Nuclei and lipid droplets were stained overnight by 300 nM DAPI (#D1306, ThermoFisher) and 1 pM fatty acid dye (Bodipy 493 / 503, #D3923, ThermoFisher). Samples were mounted using Diamond Antifade (#P36965, Invitrogen) after staining and rinsing. Brightfield and fluorescent images were captured by a wide-field fluorescence microscope (EVOS M5000 Imaging System, ThermoFisher) and confocal fluorescence microscopes (LSM710, LSM900, Zeiss). To enable fluorescent comparison, all staining and imaging procedures were kept consistent. Zen (Zeiss) and ImageJ (no plug-ins) were used to analyze colorimetric (SA-beta-gal) or fluorescent intensity (pl 6) with consistent settings. Imaging analyses were based on whole cell chamber (under lOx objective lens, one chamber as one data point, 8 chambers per MPS).RNA isolation from MPS and gene expression.

[0150] Cells were exposed by cutting and exfoliating the cell chamber slab off the device then lysed in Trizol (#15596026, Invitrogen). RNA isolation (#T2010S, NEB), synthesis of cDNA via qRT-PCR (Maxima First Strand cDNA Synthesis Kit for RT-qPCR, # K1641, ThermoFisher), and qPCR (TaqMan™ Universal PCR Master Mix, # 4304437, ThermoFisher) in QuantStudio5 (Applied Biosystems, ThermoFisher) were done according to the manufacturer’s instructions. Oligonucleotide primers for each target gene were pre-designed and synthesized by Integrated DNA Technologies. Relative mRNA expression was determined by the delta-delta-Ct method normalized to PPIA. All genomic analyses were done within two weeks of collection. Primers for qRT-PCR are listed in Table 5.PCT / US25 / 5230902 December 2025 (02.12.2025)Table 5. Primers for qRT-PCR.PCT / US25 / 5230902 December 2025 (02.12.2025)De novo proteomes

[0151] WAT-MPSs derived from the hiPSC WTC11 line stably expressing the methionyl-tRNA synthetase transgene (MetRSL274G) were differentiated and treated following the same protocol. 2mM azidonorleucine (ANL) was added to the medium at the onset of YCM and OCM treatments to label newly synthesized proteins. After cell lysis in RIPA buffer, ANL-labeled proteins were conjugated to alkyne-Cy3 using the Click-iT Protein Reaction Buffer Kit (#C 10276, Thermo Fisher Scientific) and subsequently probed on a human antibody array targeting 507 serum proteins (Human L507 Array, RayBiotech Life, GA, USA) according to the manufacturer’s instructions. Arrays were scanned at 532 nm using a GenePix 4000B scanner (Molecular Devices) and analyzed with GenePix Pro 6.1 software.

[0152] Data processing included background correction using positive and negative control probes with an offset to stabilize low-intensity values, normalization to a common reference array as specified in the RayBio® L-Series Human Antibody Array user manual, and appropriate averaging of technical and biological duplicates. Further data handling was performed with the limma package in R, with supplementary organization in Python using pandas and numpy.Statistics and reproducibility

[0153] All results were statistically analyzed by Prism 10 (GraphPad). All experiments contain 3 or more biological replicates (n) each group. Paired two-tailed t-test with assumption of equal variance was used to compare the difference between two groups from the same MPS. Unpaired two-tailed t-test with the same assumption was used for results from different MPS. One-way analysis of variance (ANOVA) for 3 or more groups comparison with one testing variable and two-way ANOVA for 3 or more groups with two testing variables, and then post-hoc test for multiple comparisons with corrections were performed to determine significant pair(s):Dunnett’s test was used to determine the significance between drug treatments and old control (FIG 14, 21, 25); Sidak's test to determine cross-group difference each condition (FIG 11); Tukey’s test to determine all significant pairs (lipid metabolism in FIG 14, 21, 25). p<0.1 was labeled in comparison. p<0.05 was consideredPCT / US25 / 5230902 December 2025 (02.12.2025)significant (*), p<0.005 highly significant (**). Results are shown as mean values with error bars representing standard deviation (SD). All experiments were repeated independently (biological replicates, n) with similar results at least 3 times.Bioinformatic analyses methods:Bulk RNA-seq

[0154] RNA samples harvested from MPS were frozen and shipped in dry-ice to Novogene for RNA sequencing. Libraries were sequenced on Illumina after quality control via Qubit and real-time PCR. Alignments were performed with HISAT2 (Kim et al. Nature biotechnology 37, 907-915 (2019)) software by Novogene. Subsequent analyses of gene expression were carried out using Python (3.10.13) with methods described below.Differential Expression Analysis

[0155] RNA-seq gene counts were first normalized using median of ratios, which is the default normalization method of popular differential expression analysis package DESeq2 (Cho et al. British Journal of Dermatology 181, 1216-1225 (2019)). For 4-day serum treatment samples, we also performed an additional ‘within batch normalization’ step, in which we normalized gene counts of each sample to the control sample of the corresponding batch, to reduce batch effects of WAT-MPS before proceeding to differential expression analysis.

[0156] Differential expression analysis (DEA) was conducted with pyDESeq2 (Muzellec et al. Bioinformatics 39, btad547 (2023)), a python implementation of DESeq2. Since normalization had already been performed using the method described above, the built-in normalization step in pyDESeq2 pipeline was omitted.Differentially expressed genes (DEGs) were identified with Wald test using pyDESeq2 (with threshold false discovery rate < 0.05; absolute log2 fold change > 1) after multiple testing correction by Benjamini-Hochberg method. Visualization of DEGs is performed with Matplotlib (3.8.3), Seaborn (0.13.0), and Scipy (1.11.3). Identification of aging associated genes was based on the public databases, HAGR (Human Ageing Genomic Resources).Differential Variability Analysis of Gene Expression

[0157] Based on the same normalized datasets utilized in Differential Expression Analysis (DEA), we identified gene markers that exhibit a significant increase in variability, or biological noise, with age, also known as Increasingly VariablyPCT / US25 / 5230902 December 2025 (02.12.2025)Expressed Genes (IVEGs). IVEGs were identified in the GTEx Adipose Subcutaneous and Liver datasets as well as our WAT-MPS and liver-MPS RNAseq data through Levene’s test with a significance threshold of p < 0.05, i.e., the variance is unequal, and a log2 fold change of the coefficient of variation (CV) > 0, calculated as log2(CVoid CVyoung), i.e., old CV is larger than young. CV was used to account for relative variation regarding mean expression values, acknowledging that higher means may exhibit greater variation, while lower means have less capacity for variation. Similarly, aging associated genes was selected based on HAGR. Since the highest log2(CVoid / CVyoung) is 0.76 in GTEx, an advanced threshold with log2(CVoid / CVyOung) >0.5 were tested to further identify the core and mostly dysregulated genes. Data analysis and statistical tests were conducted using Python with packages Pandas, Scikit-leam (1.3.2) (Li, etal. Nature 606, 785-790 (2022)), and SciPy (1.11.3) (Virtanen, P. et al. Nature methods 17, 261-272 (2020).

[0158] For comprehensive analysis of all WAT-MPS RNA-seq data, only the threshold of log2 fold change of CVold / CVyoung > 0 was applied to select, compare, and visualize IVEGs from the GTEx dataset and the 4-day and 8-day serum treatment sample data to identify signature genes that exhibit increasing variability with age. Data analysis and statistical tests were conducted using Python with packages Pandas, Scikit-learn (1.3.2) (Li, etal. Nature 606, 785-790 (2022)), and SciPy (1.11.3) (Virtanen et al. Nature methods 17, 261-272 (2020)).Gating threshold for liver samples in GTEx

[0159] For cross-comparison with physiological aging using the GTEx database, we initially aimed to match the same age ranges for young (20-29) and old (60-69) cohorts to serum donors as in the WAT analysis. However, the liver dataset presented limitations, including small sample sizes and uneven distributions (161 male and 65 female in age range from 20-79), and importantly, disease-biased phenotypes (e.g., only 7 samples with high inflammation in the 20-29 cohort). To balance biases and statistical power, differential genes (DEGs) and variables (IVEGs) for liver aging was adjusted to compare 30-39 vs. 50-69 age groups, with a lowered significance threshold (p < 0.05 instead of FDR < 0.05). To minimize sex-related confounding, only male GTEx liver samples were used for subsequent GO enrichment and comparisons with the two MPS aging conditions induced by male CMs or male CM-pretreated WAT. For consistency with the WAT analysis, aging biomarkers ofPCT / US25 / 5230902 December 2025 (02.12.2025)expression and noise were still identified in both sexes of GTEx liver samples, but to compensate for sex-related variances, the selection threshold was lowered to log2 fold change >0.5.Principle Component Analysis

[0160] Principle component analysis (PCA) was conducted on log-transformed normalized gene counts using Numpy (1.26.1) and Scikit-learn (1.3.2) (Virtanen et al. Nature methods 17, 261-272 (2020)), and then visualized using Matplotlib (3.8.3). Ellipses in the PCA plots indicate 95% confidence intervals.Euclidean Distance Measurement

[0161] Euclidean distances of each WAT-MPS sample to its corresponding batch control were measured using Numpy (1.26.1) on log-transformed normalized counts space. Visualization is conducted using Matplotlib (3.8.3) and Seaborn (0.13.0). Hierarchical Clustering Analysis

[0162] To assess similarities between serum-treated WAT-MPS samples, we computed a Euclidean distance matrix with log-transformed normalized counts of each sample. Hierarchical clustering was then implemented and plotted with Scipy (1.11.3).Enrichment Analysis

[0163] To gain thorough understanding of biological functions that have been altered, we performed enrichment analysis using both DAVID (Sherman et al. Nucleic acids research 50, W216-W221 (2022); Huang et al. Nature protocols 4, 44-57 (2009)) and SRplot (Tang et al. PloS one 18, e0294236 (2023)) based on the DEGs extracted from differential expression analysis and sorted by log2 fold change. The two web-based tools yielded similar enriched terms, and we extracted the terms enriched from categories including Gene Ontology Biological Process (GO_BP), Gene Ontology Molecular Function (G0_MF), Gene Ontology Cellular Component (GO_CC), and KEGG pathways for further analysis and visualization. After Benjamini-Hochberg correction, terms and pathways that were significantly enriched were identified (with threshold adjusted p-value <0.05). Significantly enriched terms were selected based on biological functions and plotted in Python using Matplotlib (3.8.3) and Seaborn (0.13.0).PCT / US25 / 5230902 December 2025 (02.12.2025)STRING Network Analysis

[0164] To gain a deeper understanding of the biological and molecular pathways that have been affected, we used the STRING Data Resource (version 12.0) (Szklarczyk et al. Nucleic acids research 51, D638-D646 (2023); Szklarczyk et al. Nucleic acids5 research 47, D607-D613 (2019)) to present and analyze the Protein-Protein Interaction(PPI) networks. A medium confidence score of 0.4 was applied asthreshold to identify significant interactions. The Log2 fold changes from gene expression analysis and significant pathways enriched in KEGG category - consistent with the results from DAVID enrichment analysis - were incorporated into the PPI 10 visualization. Only genes involved in significant pathways were identified and highlighted to ensure clarity and focus. Identification of aging associated genes was based on the public databases, HAGR (Human Ageing Genomic Resources).

[0165] In our PPI prediction of noise genes by STRING, we observed that the latest version of STRING (vl2, updated in July 2023) is inconsistent with the previous ten years of versions, ranging from v9.1 (released in December 2013) to vll.5 (archived in July 2023), as detailed below.Differences in the prediction of 10 partner genes by different versions of STRING.BOLD font labels shared genes in v9.1. Italicized font labels confidential score >or = 20 0.999; underlined for score < 0.99; * for score < 0.90.PCT / US25 / 5230902 December 2025 (02.12.2025)STRING database version history<< <5

[0166] The 10 genes predicted by vl2 have significantly lower confidence scores,with 9 of them scoring below 0.99 and 4 below 0.9, and none overlapping withpredicted genes from the earlier versions. In contrast, versions from v9.1 to vl 1.5 are consistent, with all predicted genes having confidence scores above 0.99, andpathways enriched in mTOR signaling / longevity and insulin signaling, none of which 10 were enriched by vl2.

[0167] We believe this inconsistency is primarily due to vl2.0 being the first or alpha subversion of vl2, making it less reliable. Additionally, vl2 has a reduced protein database (59,309,604 proteins from 12,535 organisms) compared to vl 1.5(67,592,464 proteins from 14,094 organisms). To ensure reliability, greater predictability, and consistency with studies from the past decade that used STRING analysis, we decided to use vl 1.5, the third subversion of vl 1, to predict 10biomarkers from the 37 noise markers. While the PPI network was plotted using vl2 (27,541,372,833 interactions) due to its inclusion of more interactions than vl 1.5 (20,052,394,042 interactions).20 Transcriptome-based Age Group Prediction Model

[0168] To assess our aging induction outcomes, we obtained bulk RNA-seq data of subcutaneous (SAT) and visceral adipose tissue (VAT) from the GTEx dataset, which is by far the largest human bulk RNA-seq project (Lonsdale et al. Nature genetics 45,PCT / US25 / 5230902 December 2025 (02.12.2025)580-585 (2013)), and normalized the acquired data of each depot with median of ratios method from DESeq2. We then split the normalized data by sex, which resulted in 445 male SAT samples and 218 female SAT samples, 371 male VAT samples and 170 female VAT samples. For each sex and fat depot, genes with fewer than 100 total counts were first filtered out. Mitochondrial DNA-encoded genes were also excluded to reduce model complexity. After filtering, a log transformation was applied to the remaining transcripts. Next, highly variable genes were selected based on the variance-to-mean ratio (VMR) and the mean expression level of each transcript, yielding 530 genes in male SAT, 267 genes in female SAT, 697 genes in male VAT, 309 genes in female VAT. Dimensionality were further reduced using linear discriminant analysis (LDA) from Scikit-leam. To classify each sample to the age of its donor, k-nearest neighbors (kNN)-based classifiers imported from Scikit-learn were trained on the LDA-transformed data, using 70% randomly selected samples of each sex and depot as training set, with 6 age groups (20-29, 30-39, 40-49, 50-59, 60-69, 70-79) as prediction outcomes. The trained age group prediction models were then validated using the remaining 30% of the samples (FIG 18, FIG 30).

[0169] In following iteration, the classification function was updated to include three algorithm options: Random Forest, KNN, and Support Vector Machine (SVM). The model achieving the best performance during cross-validation on the GTEx dataset with the Random Forest that was selected for downstream prediction of MPS ages and shared to GitHub. Other classification methods, multilayer perceptron (MLP) and XGBoost, were also tested during the prototyping stage, but they exhibited signs of overfitting and performed poorly in cross-validation.

[0170] When applying the trained age group prediction model to our WAT-MPS RNA-seq data, we first preprocessed the WAT-MPS counts matrix using the same steps applied to the GTEx data. This included median of ratios normalization and subsequent logarithmic transformation. Next, we extracted the highly variable genes identified during the model training process and applied the same LDA model used during model training to reduce dimensionality. The LDA-transformed data was used as input for our pre-trained age group prediction models.

[0171] To visualize the prediction outcome, we created a scatter plot using LD1 and LD2 from the LDA output. The age group predictions of WAT-MPS samples were represented using large, opaque dots, colored by their predicted age groups. ForPCT / US25 / 5230902 December 2025 (02.12.2025)comparison, the prediction results of the GTEx samples used for model training were plotted as smaller, semi-transparent dots, colored by their predicted age groups as well, to highlight the similarity between each WAT-MPS samples and GTEx samples across different age groups. Scatter plots are created using Matplotlib (3.8.3).Partial Least Squares Discriminant Analysis (PLS-DA)

[0172] Partial least squares discriminant analysis (PLS-DA) (Barker et al. Journal of Chemometrics: A Journal of the Chemometrics Society 17, 166-173 (2003)) is a multivariable statistical technique that has been widely used in RNA-seq analysis to classify samples based on their gene expression profiles and to reveal the biological mechanisms that differentiate these groups (Rohart et al. PLoS computational biology 13, el005752 (2017); Ruiz-Perez et al. BMC bioinformatics 21, 1-10 (2020)). In our pairwise PLS-DA analysis, we began by normalizing the samples with median of ratios method, and then applied a standard scaler to minimize bias toward genes with higher expression levels. We identified the major contributors to the latent variable that best distinguished the two conditions, selecting those with an absolute weight threshold of > 0.01, which typically yielded the top 1,000 genes. Selected major contributor genes were then ranked based on their log2 fold changes across the two conditions in the comparison. Up and down-altered major contributor genes were subsequently used for enrichment analysis, including GO and KEGG, as well as STRING network analysis.Data Availability

[0173] Transcriptomic raw FASTQ data are available on public repository, Gene Expression Omnibus, GSE280361. Age prediction model code is shared on Github via link: https: / / github.com / yuchen-he2000 / WAT-MPS-aging.

[0174] It is understood that the examples and embodiments described herein are for illustrative purposes only and that various modifications or changes in light thereof will be suggested to persons skilled in the art and are to be included within the spirit and purview of this application and scope of the appended claims. All publications, patents, and patent applications cited herein are hereby incorporated by reference in their entirety for all purposes.

Claims

PCT / US25 / 5230902 December 2025 (02.12.2025)WHAT IS CLAIMED IS:

1. A microphy siological system comprising:at least one first inlet for receiving a fluid medium;a fat module comprising adipocytes, the fat module configured to receive the fluid medium;a liver module comprising hepatocytes;a channel between the fat module and the liver module, the channel configured to promote communication of the fluid medium from the fat module to the liver module,the channel configured for exchange of one or more of nutrients, signaling molecules, and drugs between the fat module and the liver module.

2. The microphysiological system of claim 1, wherein the fluid medium is a hepatocyte culture media, wherein the fluid medium comprises hepatocyte growth factor and oncostatin M, and wherein the fluid medium essentially does not comprise epidermal growth factor and dexamethasone.

3. The microphysiological system of claim 1, wherein the fluid medium is a human plasma like media.

4. The microphysiological system of claim 3, wherein the fluid medium comprises human serum in a concentration of from about 4% to about 12%.

5. The microphysiological system of claim 1, wherein the adipocyte to the hepatocyte cell number ratio is from about 1:1 to about 30:1.

6. The microphysiological system of claim 1, wherein the adipocyte to the hepatocyte cell number ratio is from about 5:1 to about 30:1.

7. The microphysiological system of claim 1, wherein the adipocyte to the hepatocyte cell number ratio is from about 4:1 to about 6:1.

8. The microphysiological system of claim 1, wherein the hepatocytes are derived from induced pluripotent stem cells.

9. The microphysiological system of claim 1, wherein the hepatocytes are derived from human induced pluripotent stem cells.PCT / US25 / 5230902 December 2025 (02.12.2025)10. The microphysiological system of claim 8 or 9, wherein the induced pluripotent stem cells are cultured on a substrate comprising a member selected from Matrigel, RGD-BME, Vitronectin, and Laminin 521.

11. The microphysiological system of claims 8-10, wherein the induced pluripotent stem cells are cultured on Matrigel.

12. The microphysiological system of claims 8-11, wherein the induced pluripotent stem cells are contacted with endoderm-induction media which comprises a member selected from CHIR99021, PI103, sodium butyrate, monothioglycerol, doxycycline, Y27432, and a combination thereof.

13. The microphysiological system of claims 8-12, wherein the hepatocytes are mature hepatocytes.

14. The microphysiological system of a preceding claim, wherein the liver module comprises a plurality of liver module cell culture chambers, and the hepatocytes are present in the liver module cell culture chambers.

15. The microphysiological system of claim 1, wherein the adipocytes are derived from induced pluripotent stem cells.

16. The microphysiological system of claim 1, wherein the adipocytes are derived from human induced pluripotent stem cells.

17. The microphysiological system of claim 15 or 16, wherein the induced pluripotent stem cells are transduced with a nucleic acid that codes for peroxisome proliferator-actrivated recept gamma (PPARy).

18. The microphysiological system of claims 15-17, wherein the induced pluripotent stem cells are transduced by a lenti virus.

19. The microphysiological system of claims 15-18, wherein the transduction occurs in a transduction medium comprising SB431542 and / or doxycycline.

20. The microphysiological system of claims 15-19, wherein 80% or more of the adipocytes have lipid droplets.PCT / US25 / 5230902 December 2025 (02.12.2025)21. The microphysiological system of a preceding claim, wherein the fat module comprises a plurality of fat module cell culture chambers, and the adipocytes are present in the fat module cell culture chambers.

22. The microphysiological system of claim 21, wherein the fat module cell culture chambers have a shape which ranges from cylindrical to spherical.

23. The microphysiological system of claim 21, wherein the fat module cell culture chambers have a diameter of from about 450 micrometers to about 2200 micrometers.

24. The microphysiological system of claim 21, wherein the fat module cell culture chamber diameter if from about 1400 micrometers to about 1600 micrometers.

25. The microphysiological system of claim 21, wherein there are from 6 to 12 cell culture chambers in the fat module.

26. The microphysiological system of claim 21, wherein there are 8 cell culture chambers in the fat module.

27. The microphysiological system of claim 1, further comprising macrophages.

28. The microphysiological system of claim 27, wherein the macrophages are derived from induced pluripotent stem cells.

29. The microphysiological system of claim 27, wherein the macrophages are derived from human induced pluripotent stem cells.

30. The microphysiological system of claim 28 or 29, wherein the induced pluripotent stem cells are cultured on Matrigel.

31. The microphysiological system of any of claims 28-30, wherein the induced pluripotent stem cells are contacted with a hematopoietic progenitor-induction media and / or a macrophage induction media comprising M-CSF.

32. The microphysiological system of any of claims 27-31, wherein the macrophages are polarized macrophages.PCT / US25 / 5230902 December 2025 (02.12.2025)33. The microphysiological system of claim 32, wherein the polarized macrophages are Ml -macrophages.

34. The microphysiological system of any of claims 27-33, wherein the macrophages are present in the fat module.

35. The microphysiological system of any of claims 27-34, wherein the macrophage to the adipocyte cell number ratio is from about 1:1 to 1:30.

36. The microphysiological system of claim 35, wherein the macrophage to the adipocyte cell number ratio is from about 1 :5 to about 1 :30.

37. The microphysiological system of claim 36, wherein the macrophage to the adipocyte cell number ratio is from about 1 :9 to about 1:11.

38. The microphysiological system of any of claims 1-37, wherein the adipocytes and / or hepatocytes comprise a nucleic acid that codes for methionyl-tRNA synthetase 1 (MARS1), and wherein the encoded MARS1 comprises a L274G substitution according to the amino acid numbering of SEQ ID NO:(1).

39. A method of testing a drug or a combination of drugs for treatment of an adipocyte-associated disease or a hepatocyte-associated disease, comprising:contacting the microphysiological system (MPS) of a preceding claim with the drug or the combination of drugs; andassessing an attribute of the adipocytes, the hepatocytes, and / or the fluid medium; thereby testing the drug or the combination of drugs for treatment of the adipocyte- associated disease or the hepatocyte-associated disease.

40. The method of claim 39, wherein the disease is metabolic dysfunction-associated steatotic liver disease (MASLD), metabolic dysfunction-associated steatotohepatitis (MASH), insulin resistance (pre-diabetes), or type 2 diabetes (T2DM).

41. The method of claim 39 or 40, wherein the assessing an attribute comprises measuring the intracellular uptake of fatty acids in the adipocytes and / or the hepatocytes as compared to a suitable control.PCT / US25 / 5230902 December 2025 (02.12.2025)42. The method of claim 39 or 40, wherein the assessing an attribute comprises measuring the levels of fatty acids and / or glycerol in the fluid medium as compared to a suitable control.

43. The method of claim 39 or 40, wherein the assessing an attribute comprises measuring the expression level of one or more adipocyte-associated disease or hepatocyte-associated disease markers in the adipocytes and / or the hepatocytes as compared to a suitable control.

44. The method of claim 43, wherein the one or more adipocyte-associated disease or hepatocyte-associated disease markers comprise GLUT4, FATP1, HSL, ADIPOQ, LEP, TNF, IL6, NFKB1, NFKB2, PCK1, GCK, SREBPlc, and any combination thereof.

45. The method of claim 39 or 40, wherein the assessing an attribute comprises measuring the level of one or more cytokines in the fluid medium as compared to a suitable control.

46. The method claim 45, wherein the one or more cytokines comprise TNF-a, ILip, IL6, IL8, IL10, MCP1, Adiponectin, and any combination thereof.

47. The method of claim 39 or 40, wherein the assessing an attribute comprises measuring insulin mediated glucose uptake in the adipocytes as compared to a suitable control.

48. The method of claim 39 or 40, wherein the assessing an attribute comprises measuring insulin mediated glucose production in the hepatocytes as compared to a suitable control.PCT / US25 / 5230902 December 2025 (02.12.2025)49. A method for determining the biological age of a biological sample, comprising:a) obtaining the biological sample;b) assessing a property of the biological sample,c) comparing the property of the biological sample with a database of the property in a human population;d) determining the biological age of the biological sample based on step c); thereby determining the biological age of the biological sample.

50. The method of claim 49, wherein the biological sample is obtained from a human subject.

51. The method of claim 49 or 50, wherein the biological sample is an adipose tissue sample or a liver tissue sample.

52. The method of claim 49, wherein the biological sample comprises in vitro cells.

53. The method of claim 52, wherein the in vitro cells are induced pluripotent stem cells.

54. The method of claim 52, wherein the one or more in vitro cells are adipocytes and / or hepatocytes.

55. The method of claim 54, wherein the adipocytes and / or hepatocytes are derived from induced pluripotent stem cells.

56. The method of claim 53 or 55, wherein the induced pluripotent stem cells are derived from somatic cells obtained from a human subject.

57. A method for determining the biological age of a human subject or a biological sample obtained from the human subject, comprising:a) obtaining the biological sample from the human subject;b) contacting the biological sample with a microphysiological system (MPS) of any one of claims 1-38;c) assessing a property of the adipocytes, the hepatocytes, and / or the fluid medium of the MPS;d) comparing the property of the adipocytes, the hepatocytes, and / or the fluid medium of the MPS with a database of the property in a human population;PCT / US25 / 5230902 December 2025 (02.12.2025)e) determining the biological age of the biological sample based on step d); thereby determining the biological age of the subject or the biological sample.

58. The method of claim 57, wherein the biological sample is a serum sample.

59. The method of any of claims 49-58, wherein the property is the expression level, and / or the variance of the expression level, of one or more genes.

60. The method of any of claims 49-59, wherein the assessing a property comprises measuring the expression level of the one or more genes.

61. The method of claim 60, wherein the assessing a property comprises measuring the expression level of IL6, CTGF, PTGS2, POU3F1, CXCL8, EGR1, NOG, FOS, CDKN2A, HMGA2, and / or JUN.

62. The method of claim 60, wherein the assessing a property comprises measuring the expression level of GDF15, TNFAIP3, TRPS1, MMP10, CCL2, BDNF, CDKN2B, NRG1, SFN, DDB2, ALOX15B, EGR1, GDF15, CCL20, CXCL2, IL6, PTGS2, and / or PAPP A.

63. The method of any of claims 49-59, wherein the assessing a property comprises measuring the variance of the expression level of the one or more genes.

64. The method of claim 63, wherein the assessing a property comprises measuring the variance of the expression level of CCDC107, HCAR2, SPON1, NDRG4, CNBP, RHEB, PELI2, MRPL33, TNIK, PTCH2, RBM12B, TWSG1, ALKBH8, DHRS3, GPX7, FZD7, KCNN3, BCAN, TFPI, ZFP62, ZBED3-AS1, KLHL23, CNTFR, USP3-AS1, PKNOX2, IRS2, AZIN2, MYL12B, VGLL3, CHRNA5, PENK, TMEM269, SRRM3, ZNF589, LINC01560, ELOVL7, and / or PDZD2.

65. The method of claim 63, wherein the assessing a property comprises measuring the variance of the expression level of SLC9A8, AEN, DNAH11, EID2, DMRTA1, EXO1, PCMTD2, UBE2E2, KIF2C, MAPKAP1, DUSP10, RGPD3, MLST8, MTOR, MAPK14, PRR5, RICTOR, AURKB, BLM, MSH2, MLH1, and / or DNA2.

66. The method of any of claims 49-65, wherein the database comprises expression level data, and / or variance of the expression level data, of the one or more genes in a human population.PCT / US25 / 5230902 December 2025 (02.12.2025)67. The method of claim 66, wherein the database comprises expression level data, and / or variance of the expression level data, of the one or more genes in adipose tissue samples of a human population.

68. The method of claim 66, wherein the database comprises expression level data, and / or variance of the expression level data, of the one or more genes in liver tissue samples of a human population.

69. The method of any of claims 66-68, wherein the human population comprises male and female individuals.

70. The method of any of claims 66-69, wherein the human population comprises individuals of different ages.

71. The method of claim 70, wherein the human population comprises individuals from 18 to 80 years of age.

72. The method of claim 70 or 71, wherein the method further comprises determining the similarity of the biological sample to different age groups in the human population of the database based on the expression level, and / or variance of the expression level, of one or more genes.

73. The method of any of claims 66-72, wherein the method further comprises determining the activity of a functional gene group in the biological sample based on the expression level, and / or variance of the expression level, of one or more genes.

74. The method of claim 73, wherein the functional gene group is associated with a biological function.

75. The method of any of claims 59-74, wherein the comparing the property comprises applying a computer implemented model to:i) the expression level of the one or more genes in the biological sample, or ii) the expression level of the one or more genes in the adipocytes, hepatocytes, and / or fluid medium of the MPS;wherein the model is configured to1. a) determine the biological age of the biological sample;2. b) determine the similarity of the biological sample to different age groups in the human population of the database; and / orPCT / US25 / 5230902 December 2025 (02.12.2025)3. c) determine the activity of a functional gene group in the biological sample,4. based on the expression level, and / or variance of the expression level, of the one or more genes, and wherein the model has been trained on the expression level data, and / or variance of the expression level data, of the database of the human population.

76. A method of testing a drug or a combination of drugs for the attenuation or reversal of the biological aging of adipose and / or liver tissue, comprising:a) contacting the microphysiological system (MPS) of any one of claims 1-38 with fluid medium comprising human serum from a male or female donor with a chronological age of at least 62 years;b) contacting the product of step a) with the drug or the combination of drugs; and c) assessing an attribute of the adipocytes, the hepatocytes, and / or the fluid medium;thereby testing the drug or the combination of drugs for the attenuation or reversal of the biological aging of adipose and / or liver tissue.

77. The method of claim 76, wherein the fluid medium further comprises human plasma-like medium.

78. The method of claim 76, wherein the fluid medium comprises from about 4% to about 25% of human serum.

79. The method of claim 76, wherein the contacting of step b) is immediately following the contacting of step a).

80. The method of claim 76, wherein, prior to the contacting of step b), the product of step a) is incubated for one or more days.

81. The method of any of claims 76-80, wherein prior to the assessing of step c), the product of step b) is incubated for one or more days.

82. The method of claim 81, wherein the product of step b) is incubated for 3-5 days.PCT / US25 / 5230902 December 2025 (02.12.2025)83. The method of any of claims 76-82, wherein the assessing an attribute comprises measuring the intracellular uptake of fatty acids in the adipocytes and / or the hepatocytes as compared to a suitable control.

84. The method of any of claims 76-82, wherein the assessing an attribute comprises measuring the levels of fatty acids and / or glycerol in the fluid medium as compared to a suitable control.

85. The method of any of claims 76-82, wherein the assessing an attribute comprises measuring insulin mediated glucose uptake in the adipocytes as compared to a suitable control.

86. The method of any of claims 76-82, wherein the assessing an attribute comprises measuring insulin mediated glucose production in the hepatocytes as compared to a suitable control.

87. The method of any of claims 76-82, wherein the assessing an attribute comprises measuring oxidative DNA damage in the adipocytes and / or the hepatocytes as compared to a suitable control.

88. The method of any of claims 76-82, wherein the assessing an attribute comprises measuring the expression level of one or more biological age-associated and / or metabolic markers in the adipocytes and / or the hepatocytes as compared to a suitable control.

89. The method of claim 88, wherein the one or more biological age-associated and / or metabolic markers comprise CDKN1A, CDKN1B, LIFR, TNF, ILG, PPARG, FABP4, LPL, HSL, LEP, EBF2, NFKB1, SREBP1C, PCK1, and any combination thereof.

90. The method of any of claims 76-82, wherein the assessing an attribute comprises measuring the level of one or more cytokines in the fluid medium as compared to a suitable control.

91. The method of claim 90, wherein the one or more cytokines comprise TNF-a, ILip, IL6, IL8, IL10, MCP1, Adiponectin, and any combination thereof.PCT / US25 / 5230902 December 2025 (02.12.2025)92. The method of any of claims 76-82, wherein the assessing an attribute comprises measuring SA-P-Gal activity in the adipocytes and / or the hepatocytes as compared to a suitable control.

93. The method of any of claims 76-82, wherein the assessing an attribute comprises measuring the nuclear accumulation of p 16 in the adipocytes and / or the hepatocytes as compared to a suitable control.

94. The method of claim 76, wherein the assessing an attribute comprises applying the method of claim 48 to the adipocytes and / or the hepatocytes of the MPS.

95. A method for controlling cell and tissue aging, comprising:al) contacting the microphysiological system (MPS) of any one of claims 1-38 with fluid medium comprising human serum from a male or female donor with a chronological age of at least 62 years, thus creating an old circulatory milieu; bl) assessing the effects of the old circulatory milieu;cl) contacting the product of step al) with fluid medium comprising human serum from a male or female donor with a chronological age of from about 21 years to about 34 years, thus creating a young circulatory milieu;dl) assessing the effects of the young circulatory milieu;el) establishing a hallmark and / or a process of aging, which is controlled significantly differently by the young circulatory milieus versus the old circulatory milieus;fl) establishing the timing and the dose of the human serum that control a hallmark and / or a process of aging upward by the old circulatory milieu as compared to the young circulatory milieu.ora2) contacting the microphysiological system (MPS) of any one of claims 1-37 with fluid medium comprising human serum from a male or female donor with a chronological age of from about 21 years to about 34 years, thus creating a young circulatory milieu;b2) assessing the effects of the young circulatory milieu;c2) contacting the product of step al) with fluid medium comprising human serum from a male or female donor with a biological age of at least 62 years, thus creating an old circulatory milieu;PCT / US25 / 5230902 December 2025 (02.12.2025)d2) assessing the effects of the old circulatory milieu;e2) establishing a hallmark and / or a process of aging, which is controlled significantly differently by the old circulatory milieu versus the young circulatory milieu;f2) establishing the timing and the dose of the human serum that control a hallmark and / or a process of aging upward by the young circulatory milieu as compared to the old circulatory milieu.

96. The method of claim 95, wherein the fluid medium further comprises human plasma-like medium.

97. The method of claim 95, wherein the fluid medium comprises from about 4% to about 25% of human serum.

98. The method of claim 95, wherein, prior to the assessing of step dl) or d2), the product of step al) or a2) is incubated for one or more days.

99. The method of claim 95, wherein, prior to the assessing of step bl) or b2), the product of step cl) or c2) is incubated for one or more days.

100. The method of any of claims 95-99, wherein the assessing the effects comprises measuring the intracellular uptake of fatty acids in the adipocytes and / or the hepatocytes as compared to a suitable control.

101. The method of any of claims 95-99, wherein the assessing the effects comprises measuring the levels of fatty acids and / or glycerol in the fluid medium as compared to a suitable control.

102. The method of any of claims 95-99, wherein the assessing the effects comprises measuring insulin mediated glucose uptake in the adipocytes as compared to a suitable control.

103. The method of any of claims 95-99, wherein the assessing the effects comprises measuring insulin mediated glucose production in the hepatocytes as compared to a suitable control.PCT / US25 / 5230902 December 2025 (02.12.2025)104. The method of any of claims 95-99, wherein the assessing the effects comprises measuring oxidative DNA damage in the adipocytes and / or the hepatocytes as compared to a suitable control.

105. The method of any of claims 95-99, wherein the assessing the effects comprises measuring the expression level of one or more biological age-associated and / or metabolic markers in the adipocytes and / or the hepatocytes as compared to a suitable control.

106. The method of claim 105, wherein the one or more biological age-associated and / or metabolic markers comprise CDKN1A, CDKN1B, LIFR, TNF, ILG, PPARG, FABP4, LPL, HSL, LEP, EBF2, NFKB1, SREBP1C, PCK1, and any combination thereof.

107. The method of any of claims 95-99, wherein the assessing the effects comprises measuring the level of one or more cytokines in the fluid medium as compared to a suitable control.

108. The method of claim 107, wherein the one or more cytokines comprise TNF-a, ILip, IL6, IL8, IL10, MCP1, Adiponectin, and any combination thereof.

109. The method of any of claims 95-99, wherein the assessing the effects comprises measuring S A-P-Gal activity in the adipocytes and / or the hepatocytes as compared to a suitable control.

110. The method of any of claims 95-99, wherein the assessing the effects comprises measuring the nuclear accumulation of p 16 in the adipocytes and / or the hepatocytes as compared to a suitable control.

111. The method of any of claims 49, 57, 75, or 95, wherein the assessing an attribute or the assessing the effects comprises detecting de novo synthesized proteins in the adipocytes, hepatocytes, and / or the fluid medium of the MPS.

112. The method of claim 111, wherein the de novo synthesized proteins are intracellular and / or extracellular secreted proteins.PCT / US25 / 5230902 December 2025 (02.12.2025)113. The method of claim 111 or 112, wherein a noncanonical amino acid is introduced to the fluid medium of the MPS.

114. The method of claim 113, wherein the de novo synthesized proteins incorporate the noncanonical amino acid.

115. The method of claim 113 or 114, wherein the noncanonical amino acid comprises an azide moiety.

116. The method of claim 115, wherein the noncanonical amino acid is azidonorleucine.

117. The method of claim 115 or 116, wherein the de novo synthesized proteins are labeled with fluorescent or radioactive probes.

118. The method of claim 117, wherein the labeling comprises conjugating the fluorescent or radioactive probes to the azide moiety via click chemistry.

119. The method of any of claims 111-118, wherein the de novo synthesized proteins are detected by immunological array and / or mass spectrometry.

120. The method of claim 95, wherein the method further comprises transfecting the adipocytes and / or hepatocytes of the MPS with a nucleic acid.

121. The method of claim 120, wherein the nucleic acid is a DNA or RNA.

122. The method of claim 121, wherein the RNA is a small interfering RNA (siRNA).

123. The method of any of claims 120-122, wherein the transfecting occurs prior to the contacting of step al) or step c2).

124. The method of any of claims 120-122, wherein the transfecting occurs following the contacting of step al) or step c2).

Images