Cancer-associated adipomes promote invadopodia formation and enhance metastatic potential in triple-negative breast cancer Authors: Hariprasad Thangavel, Dhanya Dhanyalayam, Kezia Lizardo, Shivani Bansal, Junfeng Ma, Robert Glazer, Jyothi Nagajyothi Published: 2026 ABSTRACT AbstractTriple-Negative Breast Cancer (TNBC), which accounts for 15–20% of all breast cancer cases, remains one of the most aggressive and therapeutically challenging subtypes due to its lack of hormone receptors and HER2 expression, leaving chemotherapy as the primary treatment option. TNBC is characterized by early relapses and visceral metastasis, underscoring the urgent need to elucidate the mechanisms driving metastasis. The metastatic cascade begins with cancer cell invasion of the surrounding stroma, a process mediated by actin-rich protrusions called invadopodia that degrade the extracellular matrix (ECM). While the molecular machinery of invadopodia is well characterized, the upstream tumor microenvironment (TME) signals that regulate their formation remain poorly understood. In breast cancer, the TME is predominantly composed of mammary adipose tissue, where adipocytes undergo reprogramming into cancer-associated adipocytes (CAAs) that release extracellular vesicles, termed 'adipomes'. These adipomes serve as intercellular messengers, transferring bioactive molecules that can alter tumor cell behavior. Using a novel purification method, large (L)-adipomes were isolated from the plasma of TNBC subjects as well as from tumor-associated mammary fat (TAMF) using a murine TNBC model (EO771-C57BL/6) to investigate their role in invasion. Circulating L-adipomes from metastatic TNBC subjects induced epithelial-to-mesenchymal transition (EMT) and stemness signaling and promoted invadopodia formation in human non-tumorigenic breast epithelial cells, while TAMF-derived L-adipomes from metastatic-stage mice were enriched in phospholipids and promoted transcriptional and functional reprogramming in murine TNBC cells. Integrated transcriptomic and proteomic analyses revealed that adipomes activate stress-responsive MAPK signaling and upregulate proteins involved in actin cytoskeletal remodeling, mitochondrial metabolism, and translational machinery, pathways known to support invadopodia formation and extracellular matrix (ECM) degradation. Together, these findings establish adipocyte-derived adipomes as potent regulators of TNBC invasion and metastasis and reveal a previously unrecognized tumor–adipocyte signaling axis that may present new opportunities for therapeutic targeting.Subject terms: Cancer, Cell biology, Oncology Introduction Triple-Negative Breast Cancer (TNBC) constitutes a significant clinical challenge, accounting for 15–20% of breast cancer diagnoses1,2. Lacking the expression of estrogen, progesterone, and HER2 receptors, TNBC is refractory to targeted therapies, leaving systemic chemotherapy as the primary treatment3,4. Unfortunately, TNBC is characterized by a high rate of early relapse and a predilection for visceral metastasis, leading to a significantly poorer prognosis compared to other subtypes5–7. As metastatic dissemination is the principal cause of mortality8–10, there is an urgent unmet need to elucidate the fundamental mechanisms that drive this lethal process. The metastatic cascade is a multifaceted process in which the initial invasion of cancer cells into the surrounding stroma represents a critical rate-limiting step11–13. This requires cancer cells to breach the basement membrane and navigate the dense extracellular matrix (ECM). A central mechanism driving this invasion is ECM degradation mediated by specialized, actin-rich membrane protrusions known as invadopodia14–17. These structures localize and secrete matrix-degrading proteases, generating focal zones of ECM breakdown that enable cellular penetration18–20. The ability to form functional invadopodia is tightly linked to metastatic potential21–23. Although the core molecular machinery underlying invadopodia formation is well characterized14,24–26, the upstream signals from the tumor microenvironment (TME) that initiate and regulate their formation and maturation remain poorly understood, constituting a critical knowledge gap in metastasis research. It is now well established that invasion is not a purely cell-autonomous process but is orchestrated through complex interactions within the TME11,27,28, a dynamic ecosystem of stromal cells, signaling molecules, and ECM components that actively shape every stage of cancer progression29–32. In breast cancer, TME is largely composed of mammary adipose tissue33,34 where tumors develop in close contact with adipocytes that undergo reprogramming into cancer-associated adipocytes (CAAs)35–38. CAAs are phenotypically and functionally distinct from normal adipocytes in that they undergo dedifferentiation, exhibit delipidation, secrete altered profiles of cytokines and growth factors, and adopt an altered metabolic state that fuels tumor progression through the continuous release of free fatty acids and pro-inflammatory mediators35–38. Although this intimate tumor–adipocyte crosstalk positions the adipose TME as a key regulator of breast cancer biology, our current understanding remains limited, as most studies have utilized extracellular vesicles (EVs) from in-vitro cultured adipocytes and ex-vivo cultured adipose tissue39–42, which may not recapitulate the biology of TME-associated adipomes. Furthermore, significant technical challenges in isolating pure adipomes directly from tissue have hindered a deep mechanistic understanding of their specific composition and the precise pathways they regulate in tumor cells. To address this deficiency, a novel purification method was developed for isolating adipomes directly from intact tissue and plasma43–45. Leveraging this approach to investigate their role in metastatic progression, we hypothesized that adipomes from the TNBC microenvironment function not merely as metabolic fuel but as sophisticated paracrine signaling vehicles carrying specific biomolecules that directly activate the invasive machinery in both non-cancerous breast epithelial cells and cancer cells. The present study utilized both clinical samples and murine TNBC models to examine whether circulating adipomes from metastatic TNBC subjects enhance the invasive potential of normal epithelial cells and to elucidate the mechanisms through which TME-derived adipomes regulate metastatic behavior in breast cancer cells. Our results indicate that circulating L-adipomes from metastatic-stage TNBC patients, but not from healthy controls, enhance invasive behavior in nonmalignant mammary epithelial cells. This effect is associated with the induction of EMT- and stemness-related markers and increased invadopodia-associated matrix degradation compared to L-adipomes from early-stage TNBC patients. Similarly, L-adipomes isolated from tumor-associated mammary fat (TAMF) in a murine TNBC model increase the invasive potential of EO771 triple-negative breast cancer cells, accompanied by enhanced matrix degradation. Integrated transcriptomic and proteomic analyses indicate that cancer-associated L-adipomes are enriched in phospholipids and activate molecular programs linked to actin cytoskeletal remodeling, localized protein synthesis, mitochondrial bioenergetics, and MAPK signaling, which are known to support invadopodia formation and ECM degradation. Collectively, our findings reveal a previously unrecognized mode of tumor–adipocyte communication and provide a mechanistic framework for how cancer-associated adipomes contribute to the invasive phenotype in TNBC. Results Circulating L-adipomes from metastatic TNBC subjects reprogram non-tumorigenic mammary epithelial cells toward an invasive, stem-like phenotype We previously demonstrated that circulating L-adipomes can modulate recipient cell functions in a murine infection model43. To explore their clinical relevance in breast cancer, we investigated whether circulating L-adipomes from TNBC subjects differentially regulate cell invasion compared to those from healthy subjects. Purified L-adipomes (Fig. 1a) were used to treat the non-tumorigenic mammary epithelial cell line MCF-10A to evaluate their intrinsic capacity to induce invasion. As shown in Matrigel-coated transwell invasion assays, adipomes from TNBC subjects significantly enhanced MCF-10A cell invasion relative to both vehicle-treated cells and those treated with adipomes from healthy individuals (Fig. 1b). To further determine whether this effect is stage-specific, L-adipomes were isolated from early-stage (E-Adi; stages I–II) and metastatic (M-Adi; stages III–IV) TNBC subjects and used to treat MCF-10A cells to assess their ability to reprogram epithelial cells toward a malignant, invasive phenotype. Circulating adipomes from metastatic TNBC subjects reprogram non-tumorigenic mammary epithelial cells toward an invasive, stem-like phenotype. a Workflow demonstrating L-adipome isolation from the plasma of TNBC subjects. A representative transmission electron micrograph (TEM) shows intact circulating L-adipomes upon immuno-magnetic purification. Scale bar, 100 nm. b Quantification of transwell Matrigel invasion of MCF-10A, non-tumorigenic human mammary epithelial cells treated with either PBS vehicle (Ctrl) or adipomes isolated from healthy donors (HA) or subjects with TNBC (TA). Invaded cells were quantified by Calcein AM fluorescence. c Relative mRNA expression of canonical EMT-associated transcription factors (TWIST, SNAI1, SNAI2), mesenchymal markers (CDH2, VIM, FN1), matrix metalloproteinases (MMP3), and hypoxia-inducible factor-1 (HIF1) in MCF-10A cells following 48 h treatment with adipomes from either early-stage I-II (E-Adi) or late-stage III-IV TNBC (M-Adi) subjects. Expression was determined by quantitative real-time PCR (qPCR) and normalized to the housekeeping gene GAPDH. d Relative mRNA expression of CSC-associated markers (CD44, ALDH1A3, EPCAM, MYC) in MCF-10A cells treated as in (c). e Immunoblot analysis and corresponding densitometric quantification of EpCAM, a key marker for cancer stemness assessed in MCF-10A cells treated with E-Adi or M-Adi. GDI served as the loading control. f Representative multi-channel fluorescence micrographs of an invadopodia-mediated matrix degradation assay. MCF-10A cells were pre-treated with E-Adi or M-Adi, and cultured for 24 h on fluorescein-conjugated gelatin (grayscale). Sites of proteolytic degradation appear as dark, non-fluorescent spots, which co-localize with F-actin-rich invadopodia (magenta staining with Phalloidin-iFluor 647). Nuclei are counterstained with Hoechst 33342 (blue). Scale bar, 20 µm. Box numbered 1 (E-Adi+) and 2 (M-Adi+) in the fluoro-gelatin channel show magnified cell structures with black arrows pointing to dark degradation puncta mediated by invadopodia activity. g Quantification of the number of invadopodia puncta per cell, showing a significant increase in discrete degradation events following M-Adi treatment. h Quantification of the percentage of gelatin degraded area per field of view (n ≥ 4 independent images per condition). Data in (b–e) are presented as the mean ± SEM from a minimum of three independent biological replicates. Statistical significance was determined using a two-tailed, unpaired Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Quantitative PCR analysis revealed that treatment with M-Adi for 48 hours resulted in a robust upregulation of a panel of canonical EMT-associated genes (Fig. 1c). The expression of key EMT-driving transcription factors (TWIST, SNAI1, SNAI2) and mesenchymal markers (CDH2, VIM, FN1) was markedly increased in M-Adi-treated cells compared to those treated with E-Adi. Notably, expression of hypoxia-inducible factor-1 (HIF1) was also significantly increased, suggesting that M-Adi treatment may induce a hypoxic signaling state that facilitates the transition toward a mesenchymal and stem-like identity46–50. Furthermore, the expression of matrix metalloproteinase-3 (MMP3), an enzyme critical for ECM degradation, was significantly elevated. Concurrently, M-Adi treatment promoted a shift toward a cancer stem cell (CSC)–like phenotype, as evidenced by significant upregulation of established stemness-associated markers, including CD44, ALDH1A3, and EPCAM (Fig. 1d). In contrast, expression of the proto-oncogene MYC did not differ significantly between E-Adi– and M-Adi–treated MCF-10A cells. Consistent with the transcriptomic findings, EpCAM protein levels were elevated in M-Adi–treated cells (Fig. 1e), further supporting the induction of stemness-associated features in nonmalignant epithelial cells. The acquisition of both EMT and stem-like properties is a hallmark of highly aggressive, metastasis-initiating cells that were expressed in M-Adi-treated MCF-10A cells as an increase in invadopodia-mediated matrix degradation using a gelatin degradation assay (Fig. 1f, Supplementary Fig. 1a). M-Adi-treated MCF-10A cells displayed a marked increase in invadopodia activity, forming numerous F-actin-rich puncta (Fig. 1g, Supplementary Fig. 1b) that colocalized with distinct, dark areas of degraded fluorescent gelatin, indicating strong proteolytic activity (mean degradation area: 39.8% ± 1.2%; 95% CI [35.98, 43.62]; Fig. 1h, Supplementary Fig. 1). In contrast, cells treated with E-Adi showed minimal to no matrix degradation. Together, these results demonstrate that circulating L-adipomes from subjects with metastatic TNBC are capable of reprogramming non-tumorigenic mammary epithelial cells into a pro-invasive state. TNBC TME is characterized by progressive adipocyte loss and remodeling To understand the origin and context of pro-metastatic adipome release in vivo, the dynamics of the adipose TME were examined using the syngeneic EO771-C57BL/6 TNBC mouse model. This model allows for the temporal analysis of the interaction between mammary tumor cells and mammary adipocytes in an immunocompetent host. Histological analysis of the tumor-associated mammary fat (TAMF) was performed at 0-, 3-, and 5-weeks post-tumor implantation (wpi), time points corresponding to pre-tumor, early invasion, and established metastasis, respectively51. Histological analysis of TAMF at different time points revealed a progressive and dramatic remodeling of the adipose tissue architecture (Fig. 2a). At 0-wpi, the pre-tumoral mammary fat (MF) pad consisted of healthy, mature adipocytes with large, unilocular lipid droplets (Fig. 2a). By week 3, a clear interface between the tumor and adipose tissue was visible, characterized by adipocytes with smaller lipid droplets and increased stromal infiltration. By week 5, there was a profound loss of mature adipocytes in the TAMF, which were largely replaced by dense, fibrotic tissue (Fig. 2a). Quantification of adipocyte area fraction by morphometric analysis revealed a significant decline in adipocyte coverage from 95.21% ± 4.88% (95% CI [79.68, 110.74]) at baseline (0-wpi) to 25.53% ± 7.97% (95% CI [5.05, 46.01]) at 5-wpi (Fig. 2b). Concurrently, a progressive expansion of the tumor and fibrotic stroma occurred that was characterized by dense tumor nests and collagenous stroma peaking at 72.44% ± 7.98% (95% CI [51.92, 92.96]) area coverage by 5-wpi (Fig. 2c). TNBC tumor microenvironment drives progressive adipocyte loss associated with apoptosis/necroptosis in peritumoral fat tissue and the release of large-adipomes. a Representative H&E-stained sections of the mammary fat pad from EO771-C57BL/6 mice at 0-, 3-, and 5-weeks post-implantation (wpi). Magnified images demonstrate progressive remodeling of the TAMF, characterized by a profound loss of mature adipocytes and replacement by dense, fibrotic stroma. Scale bar, 100 µm. b Quantification of adipocyte area by morphometric analysis reveals a significant loss in adipocytes at 3- and 5-wpi. c Morphometric quantification of tumor progression and fibrotic infiltration shows a significant increase in non-adipose pathological tissue by 5-wpi. d Immunoblot analysis (left) and corresponding densitometric quantification (right) of the apoptotic markers cleaved caspase-3, annexin V, and Bnip3 in protein lysates from MF and TAMF at 5-wpi. β-Actin served as the loading control. e Workflow of the isolation of large adipomes (L-adipomes) and small adipomes (S-adipomes) from murine pre-tumoral and tumor-associated mammary fat. Representative TEM micrographs show particle size difference. Scale bar, 100 nm. f Concentration Spectral Density (CSD) plot shows the particle size distribution and concentration profile of MF- and TAMF-derived L-adipomes across 250–500 nm size range, measured by a Spectradyne nCS1 particle analyzer using a C-2000 cartridge. g Quantification of L-adipome abundance (particles/mL) in TAMF compared to MF that correlated with tumor progression. h Representative immunofluorescence micrographs of adipocyte apoptosis in TAMF. MF and TAMF sections were immunostained for perilipin (green; lipid droplets marker) and cleaved-caspase 3 (red; apoptosis marker). Nuclei were counterstained with Hoechst 33342 (blue). In the TAMF merged image, the white arrow highlights the nuclear colocalization of caspase-3, indicating apoptosis within the peritumoral adipose tissue. Scale bar, 50 µm. Data in (d, f, g) are presented as mean ± SEM (n = 4–7 mice per group). Statistical analysis was performed using a two-tailed, unpaired Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001. To determine the mechanism underlying this progressive loss of adipocytes, the expression of key apoptotic and necroptotic markers were assessed by Western blotting in protein lysates from TAMF and control MF. As shown in Fig. 2d, TAMF displayed significantly elevated levels of the apototic markers cleaved caspase-3 and Annexin V, with increased expression of the Bcl-2 family member BNIP3, indicating enhanced programmed cell death within the TAMF. Collectively, the data suggest that adipocyte loss in the TME is associated with increased apoptosis and potentially other cell death pathways. Adipocyte loss within the TME appears not to be merely a passive consequence of tumor expansion but is associated with the active production and release of functionally relevant adipomes, particularly large-adipomes (L-adipomes; > 200 nm in diameter) (Fig. 2e–g). This fraction includes apoptotic bodies and other macro–extracellular vesicles that are likely generated from adipocytes undergoing apoptosis or other regulated cell death processes in proximity to the growing tumor (Fig. 2h). Quantification of adipomes isolated from the TAMF revealed that the progressive apoptosis of adipocytes was temporally correlated with a 3-fold increase (95% CI [2.14, 4.20]) in the abundance of L-adipomes compared to pre-tumoral MF (Fig. 2f, g, Supplementary Fig. 2). The size of these vesicles is consistent with that of apoptotic bodies, a known class of extracellular vesicles43,52. Previously, we reported increased apoptotic markers in L-adipome fractions derived from pathological adipose tissue43,45. This establishes a clear biological link between tumor-induced programmed cell death (apoptosis/necroptosis) of adipocytes as a primary mechanism for the generation and release of L-adipomes to mediate tumor-stroma crosstalk, and reframes adipocyte death from a bystander effect to an active, pro-tumorigenic process.. TAMF-Adipomes promote a pro-metastatic phenotype in TNBC cells Having established that the TNBC TME contains elevated levels of L-adipomes, we next investigated their functional effects on malignant TNBC cells. L-adipomes were isolated from TAMF of mice bearing late-stage tumors (3–5 wpi) and from pre-tumoral MF (Fig. 2e) and used to treat EO771 murine TNBC cells to assess their impact on tumor cell behavior. Initial experiments demonstrated that treatment of EO771 cells with a 1:25 ratio of cell-to-TAMF adipomes resulted in increased proliferation (data not shown) and were used for subsequent experiments. L-Adipomes isolated from TAMF were found to be functionally distinct and significantly more potent than adipomes from control MF in regulating TNBC cells (Fig. 3). Treatment of EO771 cells for 48 h with TAMF-adipomes significantly enhanced their proliferation (Fig. 3a), migration (Fig. 3b), and invasive capacity (Fig. 3c) compared to treatment with MF-adipomes. Conversely, TAMF-adipome treatment led to a significant reduction in cell adhesion, a hallmark of cells undergoing EMT, and acquired increased motility (Fig. 3d), suggesting a pro-metastatic phenotype. TAMF-derived adipomes promote a comprehensive pro-metastatic phenotype in murine TNBC cells. a Proliferation of murine EO771 TNBC cells assessed by CyQUANT™ assay after 48 h treatment with adipomes from MF or TAMF. b Transwell migration of pre-treated EO771 cells (48 h) toward a 10% FBS gradient over 18–20 h and cell migration was quantified by Calcein AM fluorescence. c Transwell invasion of pre-treated EO771 cells through Matrigel coated inserts over 24 h. Invaded cells were quantified by Calcein AM fluorescence. d Adhesion of pre-treated EO771 cells to fibronectin-coated plates. Adherent cells were quantified by Calcein AM fluorescence after a 60-minute incubation. e Invadopodia-mediated matrix degradation assay in EO771 cells treated with MF or TAMF adipomes. Representative fluorescence micrographs show fluorescein-conjugated gelatin (grayscale), F-actin (magenta), and merge. Dark spots indicated by black arrows represent areas of focal degradation mediated by invadopodia activity. Scale bar, 10 µm. f Quantitative analysis of the percentage area of degraded gelatin per field of view (n ≥ 4 independent images per condition). g Relative mRNA expression of cancer stemness-associated genes (Cd44, Cd133, Aldh1a1, Epcam, Myc) in EO771 cells treated with MF or TAMF adipomes for 48 hours. Data were normalized to the housekeeping gene Pum1. All treatments (a–g) were performed at a cell-to-adipome ratio of 1:25. Data are presented as mean ± SEM from n = 3 independent biological replicates. Statistical significance was determined by a two-tailed, unpaired Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001. EO771 cells possess an intrinsic invasive capacity. Treatment with TAMF-derived adipomes significantly enhanced invadopodia-associated matrix degradation compared to MF-derived adipomes (mean degradation area: 40.1% ± 3.4%; 95% CI [29.28, 50.92]; Fig. 3e, f, Supplementary Fig. 3). Notably, the effects observed in the murine model closely paralleled the findings in human nonmalignant epithelial cells treated with adipomes from TNBC patients, which also demonstrated a marked increase in invadopodia-associated degradation (Fig. 1f–h). This functional change was accompanied by a corresponding upregulation of stemness-associated genes, including Cd44, Cd133, Aldh1a1, Epcam, and Myc (Fig. 3g). Our data indicate that L-adipomes generated in association with adipocyte loss within the tumor-associated stroma function as active modulators of tumor cell behavior. These TAMF-derived adipomes promote molecular and phenotypic changes in TNBC cells consistent with a more aggressive and invasive state. Taken together, these findings identify TAMF-derived L-adipomes as a critical and previously under-appreciated mechanism by which the TME actively supports and fuels TNBC progression, motility, and the acquisition of stem-like, aggressive traits. Lipidomic profiling identifies a pro-invasive lipid signature in TAMF-adipomes Since, adipomes are rich in lipids, we performed a targeted lipidomic analysis of TAMF- and MF-adipomes using liquid chromatography–tandem mass spectrometry (LC-MS/MS) to identify the lipids associated with a pro-tumorigenic response. This analysis profiled 350 lipid species across 19 lipid classes, and revealed a distinct lipidomic signature in TAMF-adipomes (Supplementary Table 1). This resulted in characterization of several lipids that were significantly enriched ≥2-fold (p ≤ 0.05) and known to regulate cell signaling and mitochondrial function (Fig. 4a). Palmitic acid–containing phospholipids were among the most highly enriched species, including an 8-fold increase in phosphatidic acid, PA(16:0/16:1) and a 5-fold increase in phosphatidylserine, PS(16:0/16:0). Other classes showing significant and specific enrichment included plasmalogens, ether lipids, sphingomyelins, and cardiolipins, which are known regulators of pro-tumorigenic pathways (Fig. 4b). For instance, PS(16:0/16:0) influences MAPK/ERK signaling through interactions with serine/threonine kinases53,54. We also observed enrichment of several plasmalogen-containing phosphatidylethanolamines, including PE(P-18:1/20:4), PE(P-18:0/22:4), PE(P-18:0/20:1), and PE(P-16:0/20:4), which are implicated in ERK and JNK pathway regulation during inflammation55–57. Additionally, TAMF-adipomes were rich in PE(O-18:0/22:4), PE(15:0/14:1), LPE(18:0), and cardiolipins (-2)CL76:4-FA18:2 and (-2)CL74:5-FA16:0, which are essential for protein post-translational modifications, mitochondrial bioenergetics, membrane integrity, and mitophagy58. Lipidomic profiling identifies a distinct pro-invasive lipid signature in TAMF-adipomes. a Targeted lipidomic profile (volcano plot) of 350 lipid species analyzed by liquid LC-MS/MS, comparing adipomes from MF versus TAMF (n = 3/group). Red and blue dots indicate lipids that are significantly enriched or depleted, respectively, in TAMF-adipomes (fold change ≥ 2, p ≤ 0.05). b Relative abundance (normalized response) of significantly enriched key lipid species in TAMF-adipomes known to regulate pro-tumorigenic pathways. Together, the specific enrichment of these lipids suggests a “lipid code” transmitted from the CAA to the tumor cell to engage specific signaling pathways and metabolic reprogramming required for tumor cell invasion. TAMF-adipomes activate the MAPK signaling axis Based on the lipidomic data suggesting enrichment of signaling lipids, we first investigated their downstream transcriptional effects by performing RNA sequencing on EO771 cells treated with MF- and TAMF-derived L-adipomes. This analysis identified 4432 differentially expressed genes (adjusted p ≤ 0.05), including 2295 upregulated and 2195 downregulated genes, with a stringent subset (≥ 2-fold change, transcript length ≥ 300) used for downstream analysis (Fig. 5a, Supplementary Fig. 4, Table 1). TAMF-adipomes activate the MAPK signaling axis and integrated biosynthetic pathways. a Volcano plot of differentially expressed genes (DEGs) identified by RNA sequencing in EO771 cells treated with MF and TAMF adipomes for 48 h. Red (n = 2295) and blue (n = 2195) dots represent significantly altered and unaltered genes, respectively (adjusted p (padj) ≤ 0.05). b Gene Set Enrichment Analysis (GSEA) plot for upregulated pathways in TAMF adipomes treated EO771 cells, showing significant enrichment for Gene Ontology (GO) terms related to ribosome biogenesis and RNA metabolic processes. c GSEA plot for downregulated pathways, showing significant suppression of MAP kinase tyrosine/serine/threonine phosphatase activity corresponding to the transcriptional downregulation of MAPK inhibitors (DUSPs). d Gene Ontology based pathway analysis of significantly enriched proteins in TAMF adipomes treated EO771 cells, showing significant enrichment of translation, intermediate filament organization, and mitochondrial respiration. e Immunoblot analysis (left) and corresponding densitometric quantification (right) of the ribosomal protein RPL19, a key upregulated protein involved in translation. f Immunoblot analysis (left) and corresponding densitometric quantification (right) validating the activation of the MAPK signaling cascade (p-ERK, p-JNK, p-p38, p-MEK1) in EO771 cells treated with MF or TAMF adipomes. GDI1 served as a loading control. For (e, f), data are presented as mean ± SEM from n = 3 independent replicates. Statistical significance was determined by a two-tailed, unpaired Student’s t-test. *p < 0.05, **p < 0.01, ****p < 0.0001. Table 1Genes significantly upregulated (selection criteria: transcript length ≥ 300, FDR-adjusted p ≤ 0.05, and fold change > 2.0) in TAMF-adipome treated EO771 cells compared to MF-adipome treatmentGene_IDbaseMeanFold changep-valueGeneCardsSTRING (Biological process)Hspa1b12963.795.53E−15Heat Shock Protein Family A (Hsp70) Member 1BATP-dependent protein folding chaperoneSnord554183.452.19E−13Sno RNA, C/D Box 55RNA processingRsrp193653.306.58E−16Arginine and Serine Rich Protein 1Protein serine/ threonine/tyrosine kinase activityGadd45b29322.978.48E−12Growth Arrest and DNA Damage Inducible BetaStem cell proliferation: MAPK regulationPpp1r15a63022.732.65E−17Protein Phosphatase 1 Regulatory Subunit 15 AProtein autophosphorylationmt-Tp9112.608.42E−12mitochondrial t-RNA for prolinemitochondrial t-RNA for prolineCiart10192.543.99E−20Circadian Associated Repressor of TranscriptionE-box bindingRhov3352.411.79E−09Signaling by Rho-GTPases to regulate the actin cytoskeletonRegulation of cytoskeleton organizationGadd45g45542.402.86E−08Regulation of growth and apoptosis via ERK1/2 signalingStem cell proliferation: MAPK regulationArhgap3312152.284.03E−15Rho GTPase-Activating Protein 33Small GTPase-mediated signal transductionThe names of the genes, description, and roles as indicated by GeneCards and STRING database analysis. Genes significantly upregulated (selection criteria: transcript length ≥ 300, FDR-adjusted p ≤ 0.05, and fold change > 2.0) in TAMF-adipome treated EO771 cells compared to MF-adipome treatment The names of the genes, description, and roles as indicated by GeneCards and STRING database analysis. Gene set enrichment analysis (GSEA) revealed two major trends: a strong upregulation of pathways related to protein biosynthesis and a clear activation of the MAPK signaling axis. Pathways that were upregulated included RNA metabolism, RNA processing, RNA splicing and gene expression (Fig. 5b). Notably, small nucleolar RNAs (snoRNAs) like Snord55, which are known to orchestrate rRNA modification and ribosome maturation, were significantly enriched (Table 1). Concurrently, transcriptomic data pointed to activation of the MAPK pathway linked to migration and invasion that was associated with the downregulation (> 2-fold; p < 0.001) of several key tumor suppressor genes that negatively regulate this cascade (Supplementary data). Among these genes were the dual-specificity phosphatases DUSP2, DUSP5, and DUSP6, immediate early response 3 (IER3), and Tribbles-1 (TRIB1) that specifically inactivate MAPK/ERK signaling (Supplementary data). GSEA analysis of the downregulated genes confirmed a significant reduction in MAP kinase tyrosine/serine/threonine phosphatase activity (Fig. 5c). Overall, the transcriptomic data suggest that TAMF-adipomes promote a pro-tumorigenic state, in part, by activating kinase cascades while simultaneously suppressing phosphatase activity. To determine if these transcriptional changes translated to the protein level, proteomic profiling was conducted, which identified 379 differentially expressed proteins between MF- and TAMF-adipome treated EO771 cells. This analysis confirmed the transcriptomic findings, revealing a significant overexpression of proteins involved in macromolecule biosynthesis and translation, including ribosomal-binding proteins, intermediate filament organization and mitochondrial respiration (Fig. 5d). Key proteins identified included keratins and other proteins involved in cytoskeletal architecture and cellular motility and ribosomal protein RPL19, a 60S subunit component indicative of ribosome biogenesis (Table 2, Fig. 5e). These highlight the role of TAMF-adipomes in cytoskeletal remodeling and enhanced protein synthesis capacity. Table 2Proteins significantly upregulated (selection criteria: FDR-adjusted q ≤ 0.05) in TAMF-adipome treated EO771 cells compared to MF-adipome treatment, as demonstrated by proteomic analysisGenesProtein DescriptionsFold ChangeCytoskeletal reorganization, protrusion, motility, migrationMyh11Myosin-117.61Dnai7Dynein axonemal intermediate chain 76.63Krt76Keratin, type II cytoskeletal 2 oral3.88Per3Period circadian protein homolog 33.61Dynll2Dynein light chain 2, cytoplasmic2.57Krt2Keratin, type II cytoskeletal 2 epidermal2.45DesDesmin2.32Krt13Keratin, type I cytoskeletal 132.22Wasf2Actin-binding protein WASF22.07RhocRho-related GTP-binding protein RhoC2.05Tuba3bTubulin alpha-3 chain2.03Krt1Keratin, type II cytoskeletal 12.00DeptorDEP domain-containing mTOR-interacting protein1.90Tubb2aTubulin beta-2A chain1.90Krt80Keratin, type II cytoskeletal 801.80Krt79Keratin, type II cytoskeletal 791.63Rab1bRas-related protein Rab-1B1.32Tube1Tubulin epsilon chain1.30Cyto- and Myto- ribosomal biogenesis/RNA processing/Cell cycleLsm6U6 snRNA-associated Sm-like protein LSm63.74Lsm3U6 snRNA-associated Sm-like protein LSm33.69Rrp15RRP15-like protein3.58Osbpl6Oxysterol-binding protein-related protein 63.36EtfbElectron transfer flavoprotein subunit beta3.18Rpl1960S ribosomal protein L193.14NfyaNuclear transcription factor Y subunit alpha3.09Ankrd40Ankyrin repeat domain-containing protein 402.95Arglu1Arginine and glutamate-rich protein 12.70Pcgf2Polycomb group RING finger protein 22.65H2az1; H2az2Histone H2A.Z; Histone H2A.V2.63Rps1240S ribosomal protein S122.56Mrps1728S ribosomal protein S17, mitochondrial2.51H3c11Histone H3.12.44Mrpl3439S ribosomal protein L34, mitochondrial2.44Mrpl1839S ribosomal protein L18, mitochondrial2.42Mitochondrial functionsCyp4a12b; Cyp4a12aCytochrome P450 4A12B; Cytochrome P450 4A12A4.21Hdhd3Haloacid dehalogenase-like hydrolase domain-containing protein 33.45Higd1aHIG1 domain family member 1 A, mitochondrial2.68Tmem109Transmembrane protein 1092.55Tmem186Transmembrane protein 1862.47Cox16Cytochrome c oxidase assembly protein COX16 homolog, mitochondrial2.46Gramd1bProtein Aster-B2.37Edem2ER degradation-enhancing alpha-mannosidase-like protein 22.37Rpl3660S ribosomal protein L362.36The gene name of the protein and fold change is indicated. The proteins are grouped based on their molecular functions. Proteins significantly upregulated (selection criteria: FDR-adjusted q ≤ 0.05) in TAMF-adipome treated EO771 cells compared to MF-adipome treatment, as demonstrated by proteomic analysis The gene name of the protein and fold change is indicated. The proteins are grouped based on their molecular functions. To confirm that the observed changes in gene and protein expression resulted in definitive pathway activation, MAPK signaling was examined by Western blotting analysis of phosphorylated MAPK kinases in adipome treated EO771 cells (Fig. 5f). This analysis provided clear validation of the significant increase in ERK, JNK, and p38 activation in cells treated with TAMF-adipomes. In addition, phosphoMEK1, an upstream regulator of ERK59–61 was significantly activated in TAMF-adipome treated cells compared to MF-adipome treated cells. Taken together, these data suggest that TAMF-adipomes induce transcriptional suppression of key phosphatases and MAPK inhibitors such as DUSP, IER3, and TRIB1, collectively with increased phosphorylation of MAPKs, likely promote three key signaling cascades: translation, mitochondrial oxidative phosphorylation (OXPHOS), and cytoskeletal rearrangement that drive the metastatic phenotype. TAMF-adipomes induce a coordinated bioenergetic and proteomic shift to support invasion The process of invasion involves extensive cytoskeletal remodeling and protein secretion, which are energetically demanding. To investigate whether TAMF-adipomes are energy demanding, mitochondrial bioenergetics of adipome treated EO771 cells were profiled using a Seahorse XF analyzer. The Seahorse XF Mitochondrial Stress Test demonstrated that cells treated with TAMF-adipomes underwent significant metabolic reprogramming toward a more energetic state (Fig. 6a). These cells exhibited a marked increase in their oxygen consumption rate (OCR) compared to MF-adipome treated cells. This included basal respiration, ATP-linked respiration, and maximal respiratory capacity, which were all significantly elevated and indicative of enhanced mitochondrial OXPHOS and a greater capacity for ATP generation. qPCR analysis revealed a significant upregulation of genes involved in mitochondrial function, including mt-Nd1, mt-Nd2, Sdhc, Cytb, Cox1a, Cox5a and Nadhd in TAMF-adipome treated cells compared to MF-adipome treated controls (Fig. 6b). Western blot analysis revealed modest but statistically significant increases in cytochrome c and HSP60 levels in TAMF-adipome treated cells compared to MF-adipome treatment (Fig. 6c), suggesting alterations in mitochondrial homeostasis and bioenergetic activity associated with L-adipome exposure. TAMF-adipomes induce a coordinated bioenergetic shift to enhance mitochondrial oxidative phosphorylation. a Mitochondrial bioenergetics profile of EO771 cells pre-treated (48 h) with MF or TAMF adipomes, assessed using a Seahorse XFe96 Mito Stress Test. The kinetic plot (left) shows the normalized real-time oxygen consumption rate (OCR) following sequential injections of 1.5 µM Oligomycin (ATP synthase inhibitor), 0.5 µM FCCP (uncoupler), and 0.5 µM Rotenone/Antimycin A (Complex I/III inhibitors). Bar graphs (right) quantify basal respiration, ATP-linked respiration, and maximal respiratory capacity. Data were normalized to cell number (Hoechst 33342). b Relative mRNA expression of key mitochondrial (mt) and nuclear-encoded genes involved in mitochondrial function and oxidative phosphorylation in EO771 cells treated as in (a). Data were normalized to the housekeeping gene Pum1. c Immunoblot analysis (left) and corresponding densitometric quantification (right) of the mitochondrial proteins HSP60 and Cytochrome c in EO771 cells treated with MF or TAMF adipomes. GDI1 served as the loading control. Data in (a–c) are presented as mean ± SEM from n = 3 independent biological replicates. Statistical significance was determined using a two-tailed, unpaired Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Discussion The findings of this study establish adipomes as potent and previously unrecognized regulators of the metastatic cascade in TNBC. This work challenges the traditional view of tumor-adjacent adipocytes as passive lipid reservoirs and reveals their active and dynamic role as key orchestrators of the mammary TME. Adipocytes transmit specific pro-invasive cues through a signaling system composed of lipids packaged within adipomes. While earlier studies implicated adipocyte-derived EVs in tumor proliferation and metabolic adaptation39,41,62–67, our study provides the first direct evidence that circulating L-adipomes from metastatic TNBC patients promote an invasive phenotype in non-tumorigenic mammary epithelial cells. In parallel, L-adipomes derived from tumor-associated mammary fat enhance invadopodia-associated matrix degradation and invasion potential in TNBC cells. These conclusions are supported by integrated transcriptomic, proteomic, and functional analyses across clinical samples and murine TNBC models, collectively highlighting a previously underappreciated role for cancer-associated adipomes in regulating tumor cell invasion. We demonstrate that both the composition and abundance of L-adipomes evolve during tumor progression, with metastatic stage L-adipomes specifically promoting invadopodia formation and invasion. Importantly, L-adipomes derived from both the local TME and systemic circulation could induce a fully invasive phenotype, encompassing EMT, acquisition of stem-like traits, and ECM degradation, underscoring the broad influence of this signaling axis on TNBC progression. Correlative multi-omics analyses revealed that TAMF–derived adipomes activate MAPK signaling, remodel the actin cytoskeleton, enhance mitochondrial oxidative phosphorylation, and stimulate protein biosynthesis pathways, collectively driving the metabolic and structural reprogramming necessary for metastasis. Previous studies investigating adipocyte-derived EVs have typically relied on in vitro cultured adipocytes or ex vivo adipose explants39–42, which do not accurately replicate the complex microenvironment of mammary adipose tissue. By leveraging our novel adipome isolation method, we overcame this limitation and demonstrated that CAA-derived L-adipomes differentially reprogram both normal mammary epithelial cells and tumor cells toward an invasive phenotype likely via a lipid-driven molecular mechanism. Notably, increased adipocyte dedifferentiation, programmed cell death, and fibrosis in TAMF are associated with elevated release of L-adipomes (> 200 nm). Targeted lipidomic analysis identified specific phospholipid species enriched in TAMF-adipomes, which, upon uptake by recipient cells, are known to activate stress-responsive MAPK pathways that modulate cell growth, post-translational modifications, and mitochondrial function53–58. The MAPK pathway plays a multifaceted role in metastatic breast cancer, influencing tumor progression, metastasis, and therapeutic resistance68–70. Our data show that TAMF-adipomes robustly upregulate ERK, JNK, and p38 activation in EO771 cells. Upstream MEK1 was also elevated, consistent with prior reports indicating JNK-mediated activation of the MEK1–ERK axis71. These results establish a clear link between CAA-derived adipomes and activation of the JNK/MEK1/ERK signaling cascade that correlated with increased invadopodia formation in tumor cells. Transcriptomic analysis revealed increased MAPK signaling with diminished phosphatase levels, such as DUSPs. Integrated transcriptomic and proteomic analyses further revealed three highly interconnected biological processes altered by TAMF-adipome treatment: (1) ribosome biogenesis and translation, (2) mitochondrial organization and function, and (3) actin cytoskeleton regulation. Proteomic analysis confirmed the upregulation of proteins involved in intermediate filament organization, ribosomal biosynthesis, and the mitochondrial respiratory chain. The simultaneous activation of protein synthesis, energy production, and cytoskeletal machinery suggests a coordinated cellular strategy, whereby adipomes initiate a global “invasion module” that endows cells with the structural, biosynthetic, and energetic capacity for metastasis. MAPK activation promotes cytoskeletal reprogramming, a process essential for cancer cell stemness and motility72–75. Our findings show that metastatic adipomes upregulate both EMT and stemness signaling in non-transformed mammary epithelial cells. Because cancer stem cells rely on invadopodia for ECM degradation and metastatic dissemination14, these findings establish a direct molecular link between adipome-induced signaling and the functional invasion machinery. Although this study establishes a conceptual and mechanistic framework for adipome-mediated tumor–adipocyte communication in TNBC, several important limitations should be acknowledged and carefully considered. First, although our lipidomic and functional analyses implicate specific lipid classes in promoting invasive phenotypes, these associations remain correlative. Definitive causal validation will require gain- and loss-of-function approaches, synthetic lipid supplementation, and targeted perturbation studies in TNBC models. Similarly, validation of the lipidomic findings in independent adipome cohorts would further strengthen the robustness of the dataset, but such efforts would necessitate large-scale orthogonal lipid profiling, repeated adipome isolations, and extensive functional testing beyond the scope of the current work. Second, while our ex vivo and murine model systems establish a functional association between cancer-associated adipomes and invasive behavior, in vivo lineage-tracing and metastasis assays are required to determine whether adipome-primed TNBC cells exhibit enhanced metastatic capacity in physiological settings. Additional mechanistic interrogation, including MAPK pathway inhibition, cytoskeletal regulatory analyses, and validation in multiple human TNBC cell lines, would further clarify pathway dependency. While the role of the MAPK pathway in regulating MMP expression and tumor invasion is well-documented76–81, further investigation is required to define the specific upstream adipome-driven regulatory mechanisms that trigger this signaling, which represents a substantial extension of the present study. We also did not perform functional stemness assays or evaluate the effects of adipomes on murine nonmalignant epithelial cells, which should be addressed in future investigations. Finally, patient-specific metabolic variables such as BMI, diabetes status, and systemic inflammation may influence adipome composition. Larger, well-annotated, and racially diverse clinical cohorts will be necessary to validate adipome-associated lipid signatures as clinically meaningful biomarkers. We also did not explore additional lipid-regulated mechanisms, such as post-translational lipidation of signaling proteins (e.g., palmitoylation of Rho GTPases) that may influence cytoskeletal dynamics and invadopodia assembly. Despite these limitations, the study provides a strong foundation for future research aimed at defining the causal lipid effectors, validating in vivo relevance, and translating these findings into biomarker and therapeutic applications in TNBC. In conclusion, these results demonstrate that adipomes orchestrate a comprehensive reprogramming of cellular physiology to sustain invasion. The adipome-induced MAPK axis enhances translation of cytoskeletal proteins required for invadopodia formation, while mitochondrial activation supplies the ATP necessary for this energy-intensive process. This establishes a feed-forward loop in which metabolic and structural networks reinforce one another to drive invasion. It is plausible that this pathway intersects with other lipid-sensitive networks, such as PKC or PI3K/Akt signaling, to further amplify the invasive phenotype. The finding that the lipidomic profile of circulating adipomes changes dynamically with disease stage offers potential for their use as liquid biopsy biomarkers for TNBC risk stratification and disease monitoring. Methods Clinical samples This study utilized de-identified human plasma samples procured from the Hackensack Meridian Health Biorepository (Bio-R). All research was performed in accordance with the Declaration of Helsinki. The Bio-R operates under an Institutional Review Board (IRB)-approved protocol (IRB study# Pro2018-1022) wherein all donors provided broad written informed consent for the future, unspecified research use of their de-identified biospecimens and associated clinical data. Because this study utilized de-identified samples from a biorepository, no human subjects were directly enrolled for this specific research. Plasma samples from three distinct cohorts were acquired: (1) female subjects diagnosed with metastatic triple-negative breast cancer (M-TNBC; stages III and IV), (2) female subjects with early-stage, non-metastatic TNBC (E-TNBC; stages I and II), and (3) healthy female subjects. All samples were collected and processed by the Bio-R according to standardized procedures to separate plasma, which was then aliquoted and stored at −80 °C. The samples were shipped on dry ice from the Bio-R upon request and subsequently used for adipome isolation. Cell culture and maintenance The human non-tumorigenic mammary epithelial cells, MCF-10A (acquired from Dr. Robert Glazer’s lab), were maintained in DMEM/F12 medium (#11330-032, Gibco) supplemented with 5% horse serum (#16050-130, Gibco), 20 ng/mL human epidermal growth factor (EGF, AF-100-15-100UG, Gibco), 0.5 µg/mL hydrocortisone (#07925, STEMCELL Technologies), 100 ng/mL cholera toxin (#C8052-0.5MG, Sigma-Aldrich), 10 µg/mL insulin (#12585-014, Gibco), and 1% penicillin-streptomycin solution (#15140-122, Gibco). The murine triple-negative breast cancer cells, EO771 (originally from American Type Culture Collection), were maintained in DMEM medium (#10-013-CV, Corning) supplemented with 10% heat-inactivated fetal bovine serum (#16140-071, Gibco), 1% penicillin-streptomycin solution (#15140-122, Gibco), and 20 mM HEPES buffer solution (#15630-080, Gibco)51. All cell lines were cultured in a humidified incubator at 37 °C with an atmosphere of 5% CO2. Cell line authenticity was periodically confirmed by short tandem repeat (STR) profiling, and cultures were routinely tested for mycoplasma contamination using a PCR-based assay to ensure experimental validity. In vivo animal model and tumor implantation All animal experiments were performed in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee (AUP #283.03) at the Hackensack Meridian Health–Center for Discovery and Innovation (HMH–CDI). Female C57BL/6J mice (n = 24), aged 7–8 weeks, were procured from The Jackson Laboratory (Bar Harbor, ME) and housed at CDI Research Animal Facility (RAF) in sterilized filter top cages. Animals were maintained under a controlled 12-hour dark-light cycle with ad libitum access to standard rodent chow and water. Upon arrival at the CDI RAF, all mice were allowed a minimum of 4 days for acclimatization to the environment prior to the commencement of any experimental procedures. Food and water were not withheld prior to anesthesia induction. The study utilized the syngeneic EO771-C57BL/6 murine model of TNBC. The selection of this model was critical, as it provides a fully immunocompetent host environment. This is essential for accurately recapitulating the complex and dynamic interactions within the mammary TME. This approach circumvents the limitations of immunodeficient models, where the absence of an intact immune system would preclude the study of inflammation-driven adipocyte reprogramming and other immune-stromal crosstalk central to metastasis. Prior to any procedures, all animals underwent visual assessment and baseline weighing to verify their healthy physical condition. For orthotopic tumor implantation, EO771 cells were harvested from sub-confluent cultures (~80% confluency). Cells were washed once with sterile, ice-cold Hank’s balanced salt solution (HBSS, #14175-095, Gibco), and resuspended at a final concentration of 2.5 × 105 cells/100 µL/mouse. Mice were anesthetized using inhaled isoflurane delivered via a precision vaporizer. Anesthesia was induced at 2–4% in a clean anesthetic chamber and maintained at 2% via a tight-fitting nose cone. The prepared cell suspension was injected subcutaneously on the right fat pad of the fourth mammary gland using a sterile 25-gauge needle. During the procedure, thermal support was provided by using a heated pad with the temperature set at 37 °C to prevent hypothermia. Animals were observed continuously for respiratory rate, movement, and response to toe pinch to ensure adequate anesthetic depth. Post-implantation, animals were monitored on a heat source until they were fully ambulatory and recovered from anesthesia. Tumor growth was monitored bi-weekly by measuring the length and width of the tumors with digital calipers as previously reported51. At specified time points corresponding to distinct stages of disease progression: 0 weeks (n = 4, pre-implantation), 3 weeks post-implantation (wpi; n = 8, early invasion), and 5-wpi (n = 6, established metastasis), mice were transported to the procedure room for terminal blood collection. The sample size for each group was determined based on previous experience with EO771 syngeneic model51 to ensure sufficient statistical power to detect significant differences in tumor growth and adipocyte remodeling while minimizing animal usage. To minimize bias, mice were randomly assigned to experimental groups at the time of implantation. For euthanasia, animals were placed in an induction chamber and anesthetized with 3–4% isoflurane, increasing to 5% as needed to achieve a deep surgical plane, confirmed by the absence of a pedal reflux. While maintained under deep terminal anesthesia, a terminal blood sample was collected via cardiac puncture using a sterile 25-gauge needle inserted through the chest wall into the heart ventricle. Immediately following blood collection, isoflurane flow was maintained for at least one minute after the cessation of respiration. Death was confirmed by verifying the absence of a heartbeat, respiration, and pedal reflux. As a secondary physical method to ensure death, cervical dislocation was performed by applying pressure at the base of the skull while simultaneously pulling the base of the tail. Immediately post-euthanasia, tissue samples such as the tumor, mammary gland, lung, and liver were harvested for histological and biochemical analyses. Tumor-associated mammary fat (TAMF) from EO771 tumor-bearing mice and healthy mammary fat (MF) from age-matched control mice were reserved for adipome isolation. Adipome isolation from plasma and mammary adipose tissue A rigorous, multi-step purification protocol was employed to isolate intact and functional adipomes, addressing the critical need for methods that effectively separate adipomes from other extracellular vesicles (EVs) released by diverse cell populations in the TME and from abundant contaminants like lipoproteins and soluble proteins, a known challenge in the field43–45. This methodological advancement was central to the study, as it ensured that the observed biological effects were attributable to the adipomes themselves, thereby validating the subsequent functional and multi-omics analyses. For isolation from plasma, frozen samples were thawed on ice and were first treated with thrombin (5 U/mL) for 5 minutes at room temperature to enzymatically convert fibrinogen to fibrin, followed by centrifugation at 10,000 rpm to pellet and remove the fibrin clot. The resulting defibrinated plasma was then re-centrifuged at 21,000 × g for one hour at 4 °C to fractionate large EVs (L-EVs) in the pellet and small EVs (S-EVs) in the supernatant. The L-EVs deposited on the tube were resuspended in PBS and processed using the ExoQuick ULTRA EV Isolation Kit for Serum and Plasma (#EQULTRA-20A-1, System Biosciences). Briefly, L-EVs were precipitated with the ExoQuick reagent and subsequently purified using the size exclusion chromatography (SEC) columns to remove abundant plasma proteins. This purified L-EV fraction was then subjected to immunoaffinity capture using the Exo-Flow Capture Kit (#CSFLOWBASICA-1, System Biosciences). Streptavidin-coated magnetic beads were functionalized with a cocktail of biotinylated antibodies against human adiponectin (#130-10216B-50, RayBiotech) and FABP4 (#C49511-Biotin, Signalway Antibody) and incubated with the L-EV fraction overnight at 4 °C. After extensive washing, bead-bound large adipomes (L-adipomes) were either lysed for molecular analyses or eluted as intact vesicles for functional studies and characterization43–45. For isolation from mammary adipose tissue, freshly harvested MF or TAMF was mechanically minced in ice-cold PBS to liberate interstitial extracellular vesicles43–45. The resulting biofluid was cleared of large debris by filtration through a 70 μm cell strainer and centrifugation at 500 × g. The supernatant was then transferred and re-centrifuged at 21,000 × g for one hour at 10 °C to fractionate L-EVs and S-EVs as described earlier. The S-EV fraction was stored at −80 °C for future studies. The L-EV pellet was resuspended and enriched using the ExoQuick-TC Ultra EV Isolation Kit (#EQULTRA-20TC-1, System Biosciences). From this L-EV fraction, L-adipomes were immuno-purified using the Exo-Flow Capture Kit with magnetic streptavidin beads functionalized with biotinylated antibodies against mouse adiponectin (#BAF1119, R&D Systems) and FABP4 (#C49511-Biotin, Signalway Antibody). Following capture, the purified L-adipomes were either directly processed on beads for lipidomic analysis or eluted and reprecipitated for characterization and functional studies in murine EO771 cells. Adipome quantification The size distribution and concentration of purified adipomes were determined by Spectradyne nCS1 particle analyzer using C-2000 cartridges with a size detection range of 250–2000 nm. Adipome samples were diluted 1:1 (v/v) in 0.22 µm-filtered 1X PBST to achieve a particle concentration of approximately 1 × 10^8 particles/mL (20–100 particles per frame), which is within the optimal range for accurate measurement. Particle acquisition was performed in continuous mode until the count reached a set stop point of 5000 particles with <1.4% statistical error. The acquired stats files were analyzed for particle size distribution and concentration (per mL) using Spectradyne Data Viewer software to generate peak-filtered, background-subtracted CSD plots. Based on the observation that tumor progression was associated with an increase in the release of larger vesicles/apoptotic bodies, adipomes were categorized by size, with “large-adipomes” defined as those with a diameter greater than 200 nm. For all adipome treatment experiments, the L-Adipome concentration was standardized based on the total particle count within a defined size range of 250–500 nm. This 250 nm span was utilized to ensure consistent L-adipome dosing across all experimental groups, with concentrations calculated as particles per mL (mL−1). Transmission electron microscopy Transmission electron microscopy (TEM) images of adipomes were obtained at the Albert Einstein College of Medicine Analytical Imaging Facility43. Samples were fixed in 2% Glutaraldehyde in 0.1 M Phosphate Buffer, adsorbed onto plasma-cleaned 400-mesh carbon grids, and negatively stained with 1% aqueous uranyl acetate. Images were acquired on a Tecnai 20 TEM (ThermoFisher Scientific) at 120 kV. Histological analysis Freshly harvested mammary fat tissues (0-, 3-, and 5-wpi) were fixed in 10% neutral buffered formalin for a minimum of 48 hours at room temperature. Following fixation, tissues were dehydrated through a graded series of ethanol (70% to 100%), cleared with xylene, and embedded in paraffin wax. 5 µm-thick sections were cut using a rotary microtome, mounted on charged glass slides, and dried overnight. Hematoxylin and Eosin (H&E) staining was performed on mammary fat tissue sections obtained from mice at distinct stages of tumor progression: 0 weeks (pre-implantation), 3-wpi (early invasion), and 5-wpi (established metastases). Stained sections were imaged using an Olympus BX43 brightfield microscope (Olympus, Tokyo, Japan)51. Quantitative morphometric analysis of the mammary tissue composition (3 images/section; n = 4/group) was performed using ImageJ software. All images were processed using a standardized macro-enabled workflow to ensure unbiased quantification. Immunofluorescence analysis Formalin-fixed mouse mammary fat tissue sections (0- and 3-wpi) were processed for multicolor immunofluorescence analysis (IFA). Adipocytes were stained using anti-Perilipin 1 antibody (1:250 dilution, #4854, Vala Sciences), followed by incubation with Alexa Fluor 488-conjugated anti-mouse secondary antibody (green) to visualize the adipocyte lipid droplet membrane. To assess programmed cell death within the adipose niche, sections were co-stained with anti-Caspase-3 antibody (1:100, #9662, Cell Signaling Technology), followed by an Alexa Fluor 555-conjugated anti-rabbit secondary antibody (red), serving as a marker for apoptotic activity. Cell nuclei were counterstained using 2 µg/mL Hoechst 33342 (blue) (#62249, Thermo Fisher Scientific) to provide structural context within the mammary architecture. High-resolution images were acquired using a Nikon A1R confocal microscope (Nikon Instruments, Tokyo, Japan). Proliferation assay Cell proliferation was assessed using the CyQUANT™ Direct Cell Proliferation Assay (#C35011, Thermo Fisher Scientific), which measures nucleic acid content. EO771 cells were seeded into 96-well plates at a density of 5000 cells per well and allowed to adhere overnight. The following day, the growth medium was replaced with fresh medium containing either MF or TAMF adipomes at a cell-to-adipome ratio of 1:25, based on initial dose-response experiments. After a 48-hour incubation period, the assay was performed according to the manufacturer’s protocol. Fluorescence (ex/em maxima 508/527 nm), proportional to the cell number, was measured using an Infinite M Plex multimode plate reader (Tecan, Switzerland). Transwell migration and invasion assay Cell migration and invasion were evaluated using 24-well Corning FluoroBlok Transwell inserts with 8.0 µm pore size Polyethylene terephthalate (PET) membranes (#351152, Corning). This suite of assays was chosen to dissect distinct aspects of cell motility. The migration assay measures the intrinsic ability of cells to move through a porous membrane, while the invasion assay adds a layer of complexity by requiring cells to actively degrade an extracellular matrix (ECM) barrier, a process central to metastasis. To prepare the cells, EO771 (or MCF-10A cells treated with human plasma-derived adipomes from healthy subjects or TNBC subjects) were first pre-treated with MF or TAMF adipomes at a cell-to-adipome ratio of 1:25 for 48 hours. Following treatment, the cells were harvested and resuspended in serum-free medium at a density of 2.5 × 10^5 cells/mL. For the migration assay, the lower chamber of each well was filled with 600 µL of complete DMEM medium containing 10% FBS as a chemoattractant. A 100 µL volume of this cell suspension (2.5 × 10^4 cells) was added to the upper chamber of the FluoroBlok insert. The plates were incubated for 18–20 hours at 37 °C. For the invasion assay, the procedure was identical, with the critical exception that the upper surface of the inserts was first coated with 70 µL of Matrigel (#356234, Corning), diluted 1:3 in serum-free medium, and allowed to solidify for 1 hour at 37 °C before cell seeding. The incubation time for the invasion assay was extended to 24 hours to allow for matrix degradation. For quantification of both assays, cells that had traversed the membrane and adhered to the lower surface of the inserts were washed, and labeled with Calcein AM (#354216, Corning) in serum-free medium for 60 minutes at 37 °C. After incubation, the fluorescence of the Calcein AM-labeled cells was quantified by reading the plate from the bottom using an Infinite M Plex multimode plate reader (Tecan, Switzerland) at an excitation of 494 nm and emission of 517 nm. Adhesion assay Cell adhesion was assessed using the Vybrant™ Cell Adhesion Assay Kit (#V13181, Thermo Fisher Scientific). 96-well plates were coated overnight at 4 °C with 10 µg/mL human fibronectin (#354008, Corning) in PBS and subsequently blocked with 1% bovine serum albumin (BSA) for 1 hour at 37 °C. EO771 cells, pre-treated with adipomes (MF or TAMF) for 24 hours, were harvested and labeled with the fluorescent dye, Calcein AM (#354216, Corning) according to the manufacturer’s protocol. Labeled cells were seeded at 2.5 × 10^4 cells per well in serum-free medium onto the prepared plates and were incubated for 60 minutes at 37 °C to permit cell adhesion. After incubation, non-adherent cells were removed by three gentle washes with PBS. The fluorescence of the remaining adherent cells was measured using an Infinite M Plex multimode plate reader (Tecan, Switzerland) with ex/em set at 494/517 nm. Invadopodia formation and matrix degradation assay The ability of cells to form functional, matrix-degrading invadopodia was assessed using fluorescently labeled gelatin-coated coverslips, providing a direct visual and quantitative readout of the cellular machinery responsible for invasion82,83. Glass coverslips were coated with 50 µg/mL poly-D-lysine (#A3890401, Gibco), cross-linked with 0.5% glutaraldehyde (#16536-05, Electron Microscopy Sciences), and then coated with 0.2 mg/mL fluorescein-conjugated gelatin (#G13187, Thermo Fisher Scientific). After polymerization, the matrix was quenched with 100 mM glycine, washed extensively with PBS, and sterilized with 70% ethanol. For the assay, MCF-10A cells were pre-treated for 48 hours with either vehicle (PBS) or adipomes isolated from the plasma of early-stage or metastatic TNBC subjects. Similarly, EO771 cells were pre-treated for 48 hours with vehicle (PBS) or adipomes isolated from MF or TAMF. Following pre-treatment, cells were harvested and seeded at a density of 2 × 10^4 cells/well onto the prepared fluorescent gelatin-coated coverslips placed in a 24-well plate with complete medium. The cells were incubated for 24 hours at 37 °C to allow for invadopodia formation and matrix degradation. At the end of the incubation, cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 in PBS, and stained for F-actin with Phalloidin-iFluor 647 reagent (#ab176759, Abcam) to visualize the actin-rich cores of invadopodia. Nuclei were counterstained with Hoechst 33342 (#62249, Thermo Fisher Scientific). The coverslips were then mounted onto glass slides with Vectashield Vibrance anti-fade mounting medium (#H-1700, Vector Laboratories). Imaging was performed on a Nikon ECLIPSE Ti2-E inverted microscope (Nikon Instruments, Tokyo, Japan). Invadopodia were identified as distinct, F-actin-rich puncta that co-localized with dark, non-fluorescent spots in the green gelatin matrix, which represent sites of proteolytic degradation. For quantification, matrix degradation was analyzed using ImageJ software by thresholding the gelatin channel to isolate and measure the non-fluorescent regions of matrix clearing. The percentage of degraded area was quantified by normalizing the total area of degradation spots to the total cell footprint, while the number of invadopodia was determined using the ‘Analyze Particles’ tool to count discrete degradation puncta per cell. All measurements were averaged across 4–6 representative fields per treatment condition to ensure robust quantitative comparison. Quantitative real-time PCR Total RNA was extracted from adipome treated cells (MCF-10A and EO771) using the RNeasy Mini Kit (#74106, Qiagen) according to the manufacturer’s instructions, including an on-column DNase I digestion step to eliminate genomic DNA contamination. RNA concentration and purity were assessed using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA). For cDNA synthesis, 0.5 to 1 µg of total RNA was reverse-transcribed using the iScript cDNA Synthesis Kit (#1708890, Bio-Rad). The cDNA samples were diluted at 1:10, and qPCR was performed using the iTaq Universal SYBR Green Supermix (#1725124, Bio-Rad). All assays were performed on Applied Biosystems QuantStudio™ 3 Real-Time PCR System (ThermoFisher Scientific, Waltham, MA) according to the following cycling conditions: 30 seconds at 95 °C (1 cycle, Polymerase activation and initial denaturation), followed by 15 seconds at 95 °C and 1 minute at 60 °C (40 cycles, PCR amplification and data collection). A melt curve analysis was performed at the end of each run to verify the specificity of the amplified product. Relative gene expression was calculated using the ΔΔCT method, with expression levels normalized to the housekeeping gene GAPDH (for human samples) or Pum1 (for murine samples). Primer sequences used for this study are listed in Supplementary Table 2. Western blotting Mouse mammary adipose tissues and adipomes treated cells were lysed in ice-cold cell lysis buffer (#9803, Cell Signaling Technology) supplemented with Pierce protease inhibitor cocktail (#A32963, ThermoFisher Scientific). Lysates were cleared by centrifugation at 14,000 × g for 15 minutes at 4 °C. Protein concentration in the supernatant was determined using the Pierce BCA Protein Assay Kit (#23225, ThermoFisher Scientific). For electrophoresis, 20–30 µg of total protein per sample was mixed with Laemmli sample buffer, heated at 95 °C for 5 minutes, and resolved on polyacrylamide gels (n = 3–6/group). Separated proteins were transferred to a polyvinylidene difluoride (PVDF) membrane for immunoblot analysis. For the detection of phosphorylated proteins, membranes were blocked for 1 hour at room temperature in 5% BSA in Tris-buffered saline containing 0.1% Tween-20 (TBST) to minimize background from phosphoproteins in milk. For all other targets, membranes were blocked in 5% non-fat dry milk in TBST. Membranes were incubated overnight at 4 °C with primary antibodies diluted in the appropriate blocking buffer. The primary antibodies used were against: EpCAM (#2626, Cell Signaling Technology), cleaved caspase-3 (#9664, Cell Signaling Technology), annexin V (#8555, Cell Signaling Technology), Bnip3 (#3769, Cell Signaling Technology), RPL19 (#PA5-79935, ThermoFisher Scientific), phospho-ERK1/2 (#28733-1-AP, Proteintech), phospho-JNK (#80024-1-RR, Proteintech), phospho-p38 MAPK (#4511, Cell Signaling Technology), phosphor-MEK1 (#67873-1-Ig, Proteintech), HSP60 (##12165, Cell Signaling Technology), and Cytochrome C (#4280, Cell Signaling Technology). After incubation, membranes were washed three times with TBST and then incubated for 1 hour at room temperature with the appropriate HRP-conjugated secondary antibody (Cell Signaling Technology). Following final washes in TBST, immunoreactive bands were visualized using the SuperSignal™ West Pico PLUS Chemiluminescent Substrate (#34580, ThermoFisher Scientific) and imaged on an iBright CL1500 Imaging Systems (ThermoFisher Scientific). Densitometric analysis of the bands was performed using Image Studio software (LI-COR Biosciences), with protein levels normalized to a loading control such as β-Actin (#4967, Cell Signaling Technology) or GDI1 (#71-0300, ThermoFisher Scientific). RNA sequencing and bioinformatics analysis Total RNA samples were extracted from EO771 cells treated with MF- or TAMF-adipomes as described for qPCR, and sent to Novogene Corporation (Sacramento, CA) for library preparation, sequencing, and bioinformatic analysis. Upon receipt, samples underwent quality control. Strand-specific mRNA-seq libraries were prepared from total RNA using poly-T oligo-attached magnetic beads to purify mRNA. The mRNA was then fragmented, and first-strand cDNA was synthesized using random hexamer primers. Second-strand cDNA synthesis was performed using dUTP, a key step for preserving strand information. The resulting cDNA fragments were end-repaired, A-tailed, and ligated to sequencing adapters. After size selection, the libraries were treated with USER enzyme (to degrade the dUTP-containing second strand), amplified by PCR, and purified. Library quality and concentration were assessed using a Qubit and a Bioanalyzer. The qualified, pooled libraries were sequenced on an Illumina platform to generate paired-end reads. Raw sequencing data (FASTQ files) were first processed to obtain high-quality “clean reads” by removing reads containing sequencing adapters, reads with poly-N sequences, and low-quality reads. All downstream analyses were based on these clean reads. The clean, paired-end reads were aligned to the mouse GRCm38/mm10 reference genome using the splice-aware aligner Hisat2 (v2.0.5). Gene expression levels were quantified using featureCounts (v1.5.0-p3), which counts the number of reads mapped to each gene. The resulting values were calculated as FPKM (Fragments Per Kilobase of transcript sequence per Millions base pairs sequenced). Differential expression analysis between experimental groups (e.g., TAMF- vs. MF-adipome treated cells) was performed using the DESeq2 R package (v1.20.0). This method models read counts using a negative binomial distribution, and p-values were adjusted for multiple testing using the Benjamini-Hochberg method. Genes with an adjusted p-value (padj) ≤0.05 were considered differentially expressed. To interpret the biological functions of these differentially expressed genes (DEGs), Gene Ontology (GO) and KEGG pathway enrichment analyses were performed using the clusterProfiler R package. GO terms and KEGG pathways with a corrected p-value < 0.05 were considered significantly enriched. Furthermore, Gene Set Enrichment Analysis (GSEA) was conducted using the GSEA software (Broad Institute) to identify statistically significant, concordant changes in pre-defined gene sets (e.g., GO, KEGG, Reactome) between the experimental conditions. Targeted Lipidomics by LC-MS/MS A targeted lipidomic analysis was performed to quantify 21 classes of lipids—including diacylglycerols (DAG), cholesterol esters (CE), sphingomyelins (SM), phosphatidylcholine (PC), triacylglycerols (TAG), and ceramides (CER)—using a QTRAP® 5500 LC-MS/MS System (Sciex) as previously reported43. For lipid extraction, adipomes (TAMF or MF) bound to magnetic beads were resuspended in 40 µL PBS and subjected to three rapid freeze-thaw cycles by alternating between dry ice (30 sec) and a 37 °C water bath (90 sec), followed by sonication for 30 seconds. Lipids were then extracted by adding 100 µL of chilled isopropanol containing a comprehensive suite of internal standards (including EquiSPLASH® LIPIDOMIX® (Avanti) and standards from the Sciex Lipidyzer platform). The mixture was vortexed, incubated on ice for (20 min), and then at −20 °C (20 min) to precipitate proteins. After centrifugation at 13,000 rpm for 20 minutes at 4 °C, the lipid-containing supernatant was transferred to an MS vial for analysis. LC-MS/MS analysis was conducted using a Shimadzu LC system (SIL-30AC autosampler, LC-30AD solvent delivery) coupled to the QTRAP 5500. A 5 µL aliquot of the extract was injected onto a Xbridge amide column (3.5 µm, 4.6 ×100 mm, Waters) maintained at 35 °C. Lipids were resolved at 0.7 mL/min using a binary solvent system: Solvent A (95:5 acetonitrile/water, 10 mM ammonium acetate) and Solvent B (50:50 acetonitrile/water, 10 mM ammonium acetate). The gradient ran from 100% A to 99.9% A (3 min), to 94% A (6 min), and to 25% A (10 min), followed by a wash with 100% B (6 min) and re-equilibration with 100% A (6 min). The QTRAP 5500 was operated in both positive and negative ion modes using Multiple Reaction Monitoring (MRM). Key source settings were: curtain gas = 30, CAD gas = medium, ion spray voltage = +5.5 kV (positive) and −4.5 kV (negative), temperature = 550 °C, nebulizing gas = 50, and heater gas = 60. Data was acquired and processed using MultiQuant 3.0.3 (Sciex). Lipid concentrations were quantified as a normalized response (area under the curve for each metabolite divided by its respective internal standard area). The raw data was preprocessed by applying a signal/noise (>20:1) and a retention time tolerance (5 seconds). Data quality and reproducibility were ensured by the periodic injection of pooled QC samples, NIST plasma standards, and blank solvent runs. Lipids displaying a coefficient of variation (CV) > 20% in the pooled QC samples were filtered from the dataset. Statistical analysis was performed to identify lipids that were significantly different between TAMF- and MF-adipomes, with a fold change of ≥ 2 and a p-value of ≤ 0.05 considered significant. Final figures were generated using MetaboAnalyst 5.0. Label-free quantitative proteomics by diaPASEF Adipomes treated EO771 cells were lysed, and total protein concentration in each sample was determined using a BCA assay. Equal amounts of protein from each sample were processed for proteomic analysis as previously described84. In brief, samples were resuspended in 50 μL of 5% SDS buffer (50 mM TEABC, 20 mM DTT) and heated at 95 °C for 10 min. After cooling to room temperature, 4 μL of 500 mM iodoacetamide (IAA) was added to a final concentration of 40 mM, and samples were incubated in the dark for 30 min. Undissolved matter was pelleted (13,000 × g, 8 min), and the supernatant was processed using S-Trap columns (ProtiFi, LLC). Proteins were digested overnight at 37 °C with sequencing-grade Lys-C/trypsin (Promega). The resulting peptides were eluted and dried using a SpeedVac (ThermoFisher Scientific). Dried peptides were resuspended in 0.1% formic acid (FA) and analyzed using a reversed-phase nanoElute 2 (Bruker Daltonics GmbH) coupled to a timsTOF Ultra mass spectrometer (Bruker Daltonics GmbH). Samples were loaded and separated on an analytical column (PepSep C18 packed with ReproSil AQ as a stationary phase, pore size 120 Å, 10 cm × 75 μm, 1.9 μm) maintained at 50 °C. Peptides were eluted at a flow rate of 300 nL/min using a 40-minute gradient of buffer A (0.1% FA) and buffer B (0.1% FA in ACN) as follows: 2% B at 0 min, 23% B at 28 min, 30% B at 30 min, and 90% B from 36 to 40 min. A CaptiveSpray source was used with an ion spray voltage of 1600 V. All data were acquired in diaPASEF mode. Full MS data were acquired in a range of 100–1700 m/z and 1.45 1/K0 [V-s/cm2] to 0.64 1/K0 [V-s/cm2] in diaPASEF. DIA windows ranged from m/z 400–1000 and were acquired with ramp times of 100 ms. High sensitivity detection was enabled without diaPASEF data denoising. Analysis of diaPASEF raw files was performed using Spectronaut (v19, Biognosys) in DirectDIA mode with default settings. Dynamic retention time prediction with local regression calibration was selected, and interference correction was enabled at both MS and MS2 levels. The Q-value cutoff was set to 1% at both peptide precursor and protein levels, using scrambled decoy generation and a dynamic size of 0.1 fraction of the library size. Quantification was performed using MS2-based area, with local cross-run normalization enabled. Mitochondrial bioenergetics analysis Real-time mitochondrial respiration was measured using a Seahorse XFe96 Extracellular Flux Analyzer (Agilent Technologies) and the Seahorse XF Cell Mito Stress Test kit (#103015-100, Agilent Technologies). EO771 cells were first treated in culture with MF- or TAMF-adipomes for 48 hours. A day prior to the assay, these pre-treated cells were harvested, counted, and seeded onto a XFe96 cell culture microplate at a density of 20,000 cells/well in complete medium. The plate was then incubated overnight at 37 °C in a 5% CO2 incubator to allow for adherence. The XFe96 sensor cartridge was hydrated overnight with sterile water in a humidified non-CO2 37 °C incubator and submerged in pre-warmed XF-calibrant for 45–60 min prior to the assay. On the day of the assay, the culture medium was replaced with Seahorse XF DMEM assay medium (pH 7.4) supplemented with 10 mM XF glucose, 1 mM XF pyruvate, and 2 mM XF glutamine. The cell plate was incubated at 37 °C in a non-CO2 incubator for 1 hour to allow for temperature and pH equilibration. Mitochondrial function was assessed by sequential injections of 1.5 μM Oligomycin (Port A), 0.5 μM FCCP (Port B), and 0.5 μM Rotenone/Antimycin A (Port C). The oxygen consumption rate (OCR) was measured in real-time following the default measurement protocol. After the assay, data were normalized to nuclear content by staining the live cells with Hoechst 33342 and performing fluorescence cell counting using a BioTek Cytation C10 (Agilent Technologies). Statistical analysis All quantitative data from functional and biochemical assays are presented as the mean ± standard error of the mean (SEM) derived from a minimum of three independent biological replicates, unless stated otherwise in the figure legends. The statistical framework for this study was designed to ensure robust and appropriate analysis of the data generated. Statistical analyses were performed using GraphPad Prism Ver 10.0 (GraphPad Software, LLC). For comparisons between two experimental groups (e.g., MF-adipome vs. TAMF-adipome treatment and E-Adipome vs. M-Adipome treatment), a two-tailed, unpaired Student’s t-test was used. The assumptions of each test (e.g., normal distribution) were evaluated prior to analysis. In accordance with ARRIVE Essential 10 guidelines, 95% confidence intervals (CI) were calculated for primary morphological and functional outcomes to indicate the precision of the estimated effects. For the analysis of high-dimensional multi-omics data (RNA-seq, proteomics, and lipidomics), statistical significance was determined using methods that account for multiple hypothesis testing. Specifically, an adjusted p-value, calculated using the Benjamini-Hochberg method to control the false discovery rate (FDR), was used. A threshold of FDR ≤ 0.05 was considered significant. For all analyses, a p-value < 0.05 was considered statistically significant. While researchers were not blinded to group allocation during the tumor implantation or bi-weekly tumor monitoring, all subsequent histological and biochemical analyses, including adipocyte quantification and multi-omics data processing, were performed by investigators blinded to the sample identity until final statistical analysis was completed to ensure unbiased interpretation. Significance levels are denoted in figures as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. The specific statistical tests and the number of replicates (n) for each experiment are detailed in the corresponding figure legends. Supplementary information Supplemental material Supplemental data_Proteomics_ MF vs TAMF. Supplemental data_Transcriptomics_ TAMF vs MF. Acknowledgements We are grateful to Steven Park and Apolinar Quezada at the Center for Discovery and Innovation (CDI) for their dedicated support in laboratory operations. We thank Yael Kramer and Dr. David Chow at the HMH Network Biorepository for the provision of clinical specimens. We also extend our gratitude to Dr. Amrita Cheema at the Georgetown Metabolomics Core Facility for her expert guidance on lipidomics analysis. Additionally, we thank Drs. Leslie Cummins and Frank Macaluso of the Analytical Imaging Facility at the Albert Einstein College of Medicine for their technical expertise and assistance with electron microscopy studies. This work has been supported by Hackensack Meridian Health–Center for Discovery and Innovation (HMH–CDI) startup funds granted to Jyothi Nagajyothi. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Author contributions Conceptualization and Study Design: H.T. and J.N.; Investigation: H.T. and J.N.; Methodology: H.T., D.D., and K.L.; Formal Analysis: H.T., D.D., S.B., J.M., R.G., and J.N.; Validation: H.T. and J.N.; Writing–original draft: H.T.; Writing–review and editing: R.G. and J.N.; Funding Acquisition: J.N.; Resources: J.N.; Project Administration: H.T.; Supervision: J.N. All authors reviewed and approved the final manuscript. Data availability The authors declare that the data supporting the findings of this study are available within the paper and its Supplementary Information files. Any other minimal datasets necessary to interpret or replicate the findings are available from the corresponding author upon reasonable request. Competing interests The authors J.N. and H.T. are inventors on a pending United States patent application (Application Number: 19/233,485, 2025, Status: Pending) filed by Hackensack Meridian Health, which covers the methods for adipome isolation and their use in disease models described in this manuscript. The other authors do not have a competing interest. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41523-026-00985-2. References Associated Data Supplementary Materials Supplemental material Supplemental data_Proteomics_ MF vs TAMF. Supplemental data_Transcriptomics_ TAMF vs MF. Data Availability Statement The authors declare that the data supporting the findings of this study are available within the paper and its Supplementary Information files. Any other minimal datasets necessary to interpret or replicate the findings are available from the corresponding author upon reasonable request.