Implantation of engineered adipocytes suppresses tumor progression in cancer models

Main

Tumors are complex tissues composed of cancerous and non-cancerous cells in a hypoxic and nutrient-deprived microenvironment. The tumor microenvironment contains heterogeneous cell populations, including immune cells, mesenchymal support cells and matrix components that contribute to tumor growth and progression¹. To survive this environment, tumors are capable of reprogramming metabolic pathways to better use available substrates in the surrounding tumor microenvironment, ultimately becoming dependent on these pathways for continued growth and survival². In contrast to normal cells, the main pathway of glucose metabolism in cancer cells is aerobic glycolysis, termed the Warburg effect³. Glucose uptake and lactate production are increased in these cells, even in the presence of oxygen and functional mitochondria³. The increase in glycolytic flux allows glycolytic intermediates to supply subsidiary pathways to fulfil the metabolic demands of proliferating cells. During hypoxia, cancer cells also undergo metabolic reprogramming to increase lipid use, as fatty acids produce twice the energy of glucose^4,5.

There have been many efforts to target cancer glucose and fatty acid metabolism for therapeutic purposes. For glycolysis, these include drugs that target hexokinase 2 (HK2), which is involved in the initial steps in glycolysis, using ATP from the mitochondria to phosphorylate glucose, or drugs that target glucose transporters (GLUT1 and GLUT4)^2,6,7,8. Several drugs are also used to target lipid metabolism in cancer^9,10,11,12. These include drugs that target lipid uptake (targeting proteins such as LXR, CD36 and FABP4/5), lipogenic enzymes (such as ACC, ACLY, FASN and SCD1) and proteins involved in intracellular lipid homeostasis (such as CPT1A, PPARG and others)^13,14. In addition, recent work has shown that cold activation of brown adipose tissue (BAT), which dissipates energy by non-shivering thermogenesis, increases adipocyte glucose uptake and lipid metabolism and significantly inhibits tumor growth¹⁵. However, situating cancer patients in cold conditions for extended periods is challenging.

Here, we set out to develop a therapeutic approach, termed adipose manipulation transplantation (AMT), that uses two unique abilities of white adipose tissue (WAT). First, it can be readily extracted in the clinic by liposuction and implanted through reconstructive surgery; second, it can change into a BAT-like tissue, called browning or beiging^16,17, by upregulating essential transcriptional regulators or enzymes, such as the uncoupling protein 1 (UCP1), PPARG coactivator 1 alpha (PPARGC1A) or PR/SET domain 16 (PRDM16) genes^{18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33}. Similar to BAT, beige adipocytes have the capacity to convert energy to heat and contribute to whole-body energy expenditure³⁴. We show that CRISPR activation (CRISPRa) of either UCP1, PRDM16 or PPARGC1A induces browning, subsequently increasing glucose and fat metabolism in human white adipocytes and adipose organoids. Co-culturing of these CRISPRa-modulated adipocytes with various cancer cell lines (breast, colon, pancreatic or prostate cancer) significantly suppresses cancer cell proliferation as well as decreases glucose uptake, glycolysis and fatty acid oxidation (FAO) capacity in the cancer cells. Subcutaneously co-transplanting CRISPRa-modulated human adipose organoids and cancer cell xenografts (two different breast cancer lines, pancreatic or prostate) into immune-compromised mice leads to significantly reduced tumor size with decreased hypoxia and angiogenesis. Implantation of engineered adipose organoids into pancreatic or breast cancer genetic mouse models significantly suppresses cancer progression. Furthermore, in the breast cancer model, we show that their implantation both near and distal to the tumor leads to similar results. To further demonstrate the therapeutic potential of this approach, we show that adipocytes isolated from resected human breast tissues can be similarly manipulated with CRISPRa and can inhibit the growth of patient-derived breast cancer organoids as well as the proliferation of high-risk non-cancerous breast tissues such as those from patients with BRCA1/2 mutations. In addition, we also show how induction of CRISPRa using tetracycline or the implantation of modulated adipose organoids in an integrated cell-scaffold delivery platform can control this therapeutic approach in a reversible manner. Finally, to show that this approach can be customizable for other tumor-associated metabolic programs, we show how CRISPRa upregulation of uridine phosphorylase 1 (UPP1) in adipose organoids can suppress xenograft growth of a uridine-dependent pancreatic ductal adenocarcinoma (PDA)³⁵. Combined, our results introduce a cancer therapeutic approach that has the potential to treat numerous cancer types.

Results

CRISPRa browning of human white adipocytes

To induce browning in human adipocytes, we used CRISPRa to upregulate UCP1, PPARGC1A or PRDM16, all known genes involved in BAT development and function. Using CRISPick³⁶, we designed five guide RNAs (gRNAs) targeting each gene’s promoter and cloned them into an adeno-associated virus (AAV)-based expression vector. Differentiated adipocytes derived from human white preadipocytes were co-transfected with the gRNAs along with a Staphylococcus aureus endonuclease-deficient Cas9 (dCas9) fused to the VP64 transcriptional activator. We used an S. aureus dCas9 due to its smaller size and the VP64 transcriptional activator (which carries four copies of VP16, a herpes simplex virus type 1 transcriptional activator³⁷) because it provides moderate gene upregulation and is small enough to fit into AAV, which has a 4.7 kb optimal packaging capacity³⁸. To generate mature human adipocytes, preadipocytes were subjected to adipocyte differentiation using a cocktail of 3-isobutyl-1-methylxanthine (IBMX), dexamethasone and insulin before being subjected to CRISPRa. After 4 days, we measured gene expression by quantitative PCR with reverse transcription (RT–qPCR), finding several gRNAs to significantly increase the expression levels of UCP1, PPARGC1A and PRDM16 in the transfected cells compared to the control cells (treated with dCas9–VP64 only) (Extended Data Fig. 1a). We selected the top two gRNAs for each gene for AAV serotype 9 (AAV9) packaging. We used AAV9 because it was shown to effectively infect various adipose depots³⁹. We infected human differentiated adipocytes with these viruses along with AAV9 dCas9–VP64, finding for each gene at least one gRNA that significantly increased expression levels compared to the dCas9–VP64-only control (Extended Data Fig. 1b). We used the top-activating gRNA AAV for all subsequent experiments.

We next examined whether our CRISPRa treatment increases browning in these human white adipocytes. Human adipocytes transduced with the top gRNA for UCP1, PRDM16 or PPARGC1A showed significantly increased expression of their target genes compared to dCas9–VP64-only infected cells (Extended Data Fig. 1c). In addition, we also observed increased mRNA levels for brown fat marker genes, including TFAM, DIO2, CPT1b and NRF1 upon upregulation of either of the three genes (Extended Data Fig. 1d); PRDM16 CRISPRa did not show upregulation of CPT1b and NRF1. We next examined the oxygen consumption rate (OCR) in these cells using Seahorse (Methods), by initially blocking oxygen consumption and then adding 1 μM oligomycin, followed by the introduction of 1 μM carbonyl cyanide-p-trifluoromethoxy-phenylhydrazone (FCCP) to measure the maximal respiratory capacity. We found that CRISPRa-AAV treatment targeting all these genes increased overall OCR levels in human white adipocytes, with UCP1-gRNA-AAV-treated cells having the largest increase (Extended Data Fig. 1e). In addition, these cells showed increased uncoupled respiration (under oligomycin, an ATP synthase inhibitor), indicating a brown fat-like phenotype (Extended Data Fig. 1e). CRISPRa-treated cells also had elevated maximal respiration following FCCP treatment (Extended Data Fig. 1e). Furthermore, these CRISPRa-AAV-engineering adipocytes showed increased glucose uptake in both basal and insulin-stimulated conditions (Extended Data Fig. 1f). We also tested the FAO capacity of these cells by performing a similar Seahorse OCR assay. In both BSA and BSA-conjugated-palmitate (saturated fatty acid complex) media, we found that our CRISPRa-treated adipocytes had an overall OCR increase in BSA-conjugated-palmitate (Extended Data Fig. 1g). Under FCCP treatment, in which the increase of OCR in palmitate-containing media compared to BSA-containing media is thought to be caused by exogenous FAO, we found that upregulation of either gene increased exogenous FAO capacity in human adipocytes (Extended Data Fig. 1h). Similarly, UCP1-upregulated human adipocytes had the largest increase in FAO. We also found that UCP1-upregulated human adipocytes had higher fatty acid uptake than the dCas9–VP64 control cells (Extended Data Fig. 1i). Taken together, we show that AAV-based CRISPRa upregulation of UCP1, PPARGC1A or PRDM16 induces browning in human adipocytes, leading them to have increased glucose uptake and FAO.

CRISPRa-modulated adipocytes suppress tumor growth in vitro

We next evaluated whether our CRISPRa ‘browned’ adipocytes could inhibit cancer growth in vitro using a co-culturing system (Fig. 1a). We initially treated differentiated adipocytes with CRISPRa-AAV for either of the three genes (UCP1, PPARGC1A or PRDM16) and replated these adipocytes on the top chamber of a 12-well or 24-well Transwell plate, which has inserts of 0.4 µm membranes, so that the adipocytes did not contact the cells on the lower chamber (Fig. 1a). In the lower chamber, we grew five different cancer cell lines: breast cancer cells MCF-7 (estrogen receptor-positive (ER⁺), progesterone receptor-positive (PR⁺), glucocorticoid receptor-positive (GR⁺)) and MDA-MB-436 (triple negative), colon cancer (SW-1417), pancreatic cancer (Panc 10.05) and prostate cancer (DU-145). As a negative control, we used adipocytes infected with dCas9–VP64 only. After 3 days, we observed that all five cancer cell lines that were co-cultured with UCP1, PPARGC1A or PRDM16 CRISPRa-treated human adipocytes showed significantly lower cell numbers than cancer cells co-cultured with dCas9–VP64-treated adipocytes (Fig. 1b). We also found that the number of cancer cells co-cultured with CRISPRa-treated adipocytes was threefold to fivefold lower than cancer cells co-cultured with the control adipocytes (Fig. 1c). By RT–qPCR, we observed that all cancer cells co-cultured with CRISPRa-treated adipocytes had significantly reduced levels of the proliferation marker MKI67 (other than Panc 10.05 and DU-145 treated with PRDM16 CRISPRa) compared to control cells, with UCP1-CRISPRa having the greatest effect (Fig. 1d). We also performed a BrdU incorporation assay, finding that after 24 h, CRISPRa co-cultured cancer cells showed significantly reduced proliferation compared to cells co-cultured with control adipocytes (Extended Data Fig. 1j).

**Fig. 1: CRISPRa-modulated adipocytes inhibit cancer cell growth in vitro.**

We next analyzed the glucose and fatty acid metabolism of the co-cultured cancer cells. To measure glycolysis, we used the extracellular acidification rate (ECAR) assay in which the basal glycolytic rate is measured by adding glucose and the maximal glycolytic rate is measured by oligomycin A addition. We found that most cancer cell lines co-cultured with CRISPRa-AAV-treated adipocytes showed a significant reduction in both basal and maximal glycolytic rate (Fig. 1e,f) and lower glucose uptake in both basal and insulin conditions (Fig. 1g,h). Using RT–qPCR, we also found that the expression of key glycolysis genes, such as GCK and GLUT4, a major glucose transporter, was significantly lower in most cancer cells co-cultured with CRISPRa-AAV adipocytes than in the negative control (Fig. 1i). We also examined FAO using Seahorse. In BSA-only media, we found that all cancer cell lines that were co-cultured with UCP1-CRISPRa-modulated adipocytes had lower OCR. For PPARGC1A and PRDM16 CRISPRa-modulated adipocytes, this was only observed in MDA-MB-436 and Panc 10.05 cells (Extended Data Fig. 1k). In palmitate-containing media, we found that all five cancer cell lines that were co-cultured with CRISPRa-AAV adipocytes had reduced FAO compared to the negative control (Fig. 1j). Moreover, using RT–qPCR, we found in several CRISPRa conditions that the cancer cells had lower expression of both CD36, a fatty acid transporter on the cell membrane, and CPT1b, a key regulator of FAO in the mitochondria, further confirming decreased FAO (Fig. 1k). To show that this approach works with other adipocyte lines, we also upregulated UCP1 in human adipocytes differentiated from primary preadipocytes (Extended Data Fig. 1l) and found that they inhibit tumor growth of all five cancer cells (Extended Data Fig. 1m). Combined, our data shows that our CRISPRa-modulated adipocytes reduce glycolysis and FAO in five different cancer cell lines and can significantly suppress cancer growth.

We next compared the tumor suppression capability of CRISPRa-AAV-modulated adipocytes to known metabolic cancer drugs. We grew MCF-7 cells and treated them with the following: (1) CRISPRa-UCP1-AAV adipocytes at the top layer of a Transwell model; (2) 6-aminonicotinamide, which is known to target glycolysis and reduce cell growth in a variety of tumors^40,41,42, including MCF-7 breast cancer cells⁴¹, using similar or higher drug concentrations (50 µM, 100 µM and dimethylsulfoxide (DMSO) as a control)⁴¹; and (3) etomoxir, an inhibitor of FAO⁴³ that suppresses tumor cell growth^44,45, including breast cancer^46,47, using concentrations previously used for MCF-7 cells (100 µM, 200 µM and DMSO as a control)⁴⁷. Using a luminescent cell viability assay (Methods), we observed an increased reduction in the number of cancer cells cultured with CRISPRa-UCP1-AAV adipocytes compared to 6-aminonicotinamide and a slightly increased reduction compared to etomoxir (two-tailed t-test P = 0.0163 versus P = 0.0319 for CRISPRa-UCP1-AAV or etomoxir, respectively; Extended Data Fig. 1n).

Modulated human adipose organoids suppress xenograft growth

To examine whether CRISPRa-modulated adipocytes could inhibit cancer growth in tumor xenograft models, we co-transplanted four cancer cell lines (MCF-7 and MDA-MB-436 (breast), Panc 10.05 (pancreas) and DU-145 (prostate)) with CRISPRa-modulated human adipose organoids. Although adipocytes could be used for co-transplantation, adipose organoids offer added advantages, such as providing a three-dimensional (3D) culture that can better recapitulate the heterogeneity of adipose tissue, enhanced response to endogenous stimuli and the ability to form tissue microenvironments that could better integrate with cancer cells following transplantation. We initially established culturing conditions for human adipose organoids. We used immortalized human preadipocytes grown using methods that combine features from three different 3D culturing protocols^48,49. In brief, we cultured human preadipocytes in basal DMEM media supplemented with 10% FBS in Nunc 96-well plates treated with Nunclon Delta. Organoids formed after 48 h and were then differentiated into adipose organoids using a differentiation cocktail containing IBMX, dexamethasone, insulin, T3 and rosiglitazone. Adipocytes formed 21 days post differentiation (Extended Data Fig. 2a). Using RT–qPCR, we analyzed these organoids for various adipogenic markers, including FABP4, PLIN1 and ADIPOQ, finding all to be expressed (Extended Data Fig. 2b). We further tested our ability to upregulate UCP1, PPARGC1A and PRDM16 in these organoids using similar methods and AAV gRNAs as in the adipocytes, finding all three genes to show significant upregulation of the target genes (Extended Data Fig. 2c). With the establishment of these adipose organoid-culturing and CRISPRa conditions, we next set out to test whether they can suppress xenograft cancer growth.

To generate xenografts, cancer cells were subcutaneously implanted into immuno-compromised SCID mice. After 6–8 weeks, UCP1-CRISPRa-treated human adipose organoids were mixed with Matrigel and co-transplanted adjacent to palpable tumors. For all subsequent assays, we only used UCP1-CRISPRa, as it showed the most optimal results in our cell culture experiments. Some reports have suggested that brown fat could be linked to cancer-associated cachexia (loss of skeletal muscle and fat)⁵⁰; therefore, each week we measured the body weight of mice co-transplanted with cancer cells and UCP1-modulated adipose organoids. We found no significant differences in body weight between CRISPRa-treated and control mice (Extended Data Fig. 2d). Tumors and human adipose organoids were collected after 3 weeks (Fig. 2a). Adipose organoids were stained with LipidTox, showing adipocytes were still present in them 3 weeks after implantation (Extended Data Fig. 2e). These organoids also had increased gene expression levels of UCP1, some brown fat genes including PPARGC1A and CPT1b and the GLUT4 glucose transporter compared to control adipose organoids (Extended Data Fig. 2f). All tumor types co-transplanted with CRISPRa-modulated human adipose organoids were significantly smaller than dCas9–VP64-only transplanted human adipose organoids, having over 50% reduction in volume (Fig. 2b). Gene analysis showed that tumors co-transplanted with CRISPRa-treated adipose organoids had decreased expression of the proliferation marker gene MIK67 (Extended Data Fig. 2g) as well as reduced marker gene expression for glycolysis (GLUT4 (except for DU-145), GCK) and FAO (CD36 (except for DU-145), CPT1B) (Extended Data Fig. 2h). Using immunofluorescence, we examined additional tumor marker genes and found that all tumors co-transplanted with UCP1-CRISPRa-modulated human adipose organoids had markedly reduced Ki67⁺ cells (Fig. 2c). In addition, cancer cells showed decreased levels of hypoxia, identified by having a lower carbonic anhydrase (CA9⁺) area per image view (Fig. 2d). Furthermore, these tumors exhibited decreased levels of CD31⁺ area per image view, indicating reduced microvessel density and suggesting a corresponding lower metastatic potential (Fig. 2e). We also found higher levels of caspase-3⁺ in both MCF-7 and MDA-MB-436 xenografts co-implanted with UCP1-CRISPRa adipose organoids compared to controls, indicating higher apoptotic rates (Extended Data Fig. 2i). To show that this approach can work with another adipocyte cell line, we repeated these experiments for MCF-7 xenografts with UCP1-CRISPRa primary adipocytes and found that they also significantly suppressed tumor growth (Extended Data Fig. 2j). Taken together, these results show that UCP1-CRISPRa-modulated human adipose organoids significantly reduce glycolysis and FAO, reduce hypoxia and inhibit tumor growth for four different cancer types in vivo.

**Fig. 2: Co-transplantation of xenografts with *UCP1*-CRISPRa-modulated human adipose organoids suppresses tumor growth.**

Modulated adipose organoids outcompete tumors for nutrients

We next set out to characterize the effect of UCP1-CRISPRa adipose organoids on various metabolic parameters and test whether resource competition is involved in the observed cancer suppression. We implanted UCP1-CRISPRa or dCas9–VP64 only (negative control) adipose organoids in SCID mice and, after 6 weeks, used the comprehensive lab animal monitoring system (CLAMS) to measure their whole-body oxygen consumption. Mice implanted with UCP1-CRISPRa adipose organoids exhibited increased whole-body oxygen consumption at all temperatures (Extended Data Fig. 3a). We also carried out both a glucose and insulin tolerance test on these mice, which revealed that all UCP1-CRISPRa-treated mice had increased glucose tolerance and insulin sensitivity (Extended Data Fig. 3b). Following up on these results, we next examined the insulin plasma levels in all four of our xenograft lines, which showed that mice co-transplanted with UCP1-CRISPRa adipose organoids had significantly lower insulin levels than those co-transplanted with dCas9–VP64 control adipose organoids and were comparable to levels of wild-type SCID mice (Extended Data Fig. 3c). Combined, these data suggest that UCP1-CRISPRa-modulated adipose organoids lead to robust energy consumption, enhanced glucose tolerance and insulin sensitivity and reduced insulin levels.

To examine whether UCP1-modulated adipose organoids can prevent glucose uptake in the co-transplanted xenograft tumor, we measure glucose levels in both the adipose organoids and co-implanted MCF-7 tumor. We found that UCP1-modulated adipose organoids had higher glucose levels than controls and that xenograft tumors co-implanted with UCP1-modulated adipose organoids had lower levels of glucose than control tumors (Fig. 3a and Supplementary Table 1). Metabolomics analysis of these tumors found that glucose levels and glycolytic intermediates, including glucose-6-phosphate, fructose-6-phosphate, 3-phosphoglycerate and phosphoenolpyruvate, were lower in the tumors co-implanted with UCP1-modulated adipose organoids than in the control tumors (Fig. 3b and Supplementary Table 1). In addition, the tumors co-implanted with UCP1-modulated adipose organoids had lower fatty acid levels, including oleic acid and palmitoleic acid, than the control tumors (Fig. 3c and Supplementary Table 1).

**Fig. 3: Increasing nutrients reduces *UCP1*-CRISPRa human adipose organoid cancer suppression.**

To further test whether nutrient competition is contributory to the ability of UCP1-CRISPRa-treated adipose organoids to suppress tumor growth, we performed similar MCF-7 xenograft experiments with mice fed with either standard chow, a high-fat diet (HFD) or 15% glucose containing water (Fig. 3d). We found that tumors co-implanted with UCP1-CRISPRa adipose organoids from mice fed with standard chow had significantly lower volume than mice co-implanted with control (dCas9–VP64 only) (Fig. 3e and Extended Data Fig. 3c), which showed significant reduction in volume starting 2 weeks post implantation (Extended Data Fig. 3d). By contrast, HFD-treated or 15% glucose-treated mice showed no apparent difference in tumor growth compared to the negative control (Fig. 3e). Analysis of tumors from mice on a HFD or 15% glucose found that they had expression patterns of the MKI67 proliferation marker, glycolysis (GLUT4, GCK) and FAO (CD36, CPT1B) that are similar to the negative control, whereas mice on a regular chow diet had significantly lower expression for all these markers (Fig. 3f). Similarly, tumors from mice fed with standard chow displayed reduced levels of Ki67⁺ cells, CA9⁺ area, and CD31⁺ area compared to controls, whereas there was no apparent difference in tumors from HFD-treated or 15% glucose-treated mice (Fig. 3g–i and Extended Data Fig. 3e).

We next performed RNA-seq on tumors from all conditions (standard chow, HFD, 15% glucose and negative controls). Principal component analysis revealed that tumors co-implanted with UCP1-CRISPRa-modulated adipose organoids in mice fed with a standard chow diet were significantly different than controls, whereas there was no consistent difference in tumors co-implanted with UCP1-CRISPRa-modulated adipose organoids and control tumors in mice fed with HFD or 15% glucose water (Extended Data Fig. 3f). Tumors co-implanted with UCP1-CRISPRa-modulated adipose organoids in mice fed with a standard chow diet had significantly more global gene expression changes (7,102 differentially expressed genes with 6,623 downregulated and 479 upregulated genes) (Fig. 3j) compared to zero genes for mice treated with a HFD or 15% glucose water (Extended Data Fig. 3g). Among the downregulated genes in the standard chow tumors co-implanted with UCP1-CRISPRa-modulated adipose organoids, we found many metabolic genes, including CPT1b, SCD (stearoyl-CoA desaturase) and FASN (fatty acid synthase), and cancer growth and progression regulated genes, such as MKI67, ITGB4 (ref. ⁵¹), LLGL2 (ref. ⁵²), GATA3 (ref. ⁵³), EEF1A2 (refs. ^54,55) and TBX2 (ref. ⁵⁶) (Fig. 3j). Upregulated genes included tumor suppressor genes, such as HOXD10 (ref. ⁵⁷), GAS7 (ref. ⁵⁸) and MAP2K4 (ref. ⁵⁹), and cancer progression genes, including WNT2 (ref. ⁶⁰) and RAC2 (ref. ⁶¹) (Fig. 3j). We next performed gene ontology enrichment analysis using Geneontology.org (https://geneontology.org/)^62,63 on differentially expressed genes in tumors of mice fed with standard chow and found downregulated differentially expressed genes to be enriched in pathways involved in metabolic regulation, including FAO, oxidative phosphorylation, cell growth, DNA replication and cell division (Fig. 3k, top). Upregulated enriched pathways included those involved in cellular response to stress response and cell apoptosis (Fig. 3k, bottom). We found no significant pathway changes in tumors in mice fed with HFD or 15% glucose water. In summary, these experiments suggest that UCP1-CRISPRa-modulated adipose organoids outcompete tumors for glucose and fatty acids, as increasing fatty acid or glucose levels abolished cancer suppression.

AMT suppresses cancer development in genetic mouse models

To examine whether our AMT approach can prevent cancer development, we used pancreatic and breast cancer genetic mouse models. For pancreatic cancer, we used the KPC mouse model that, upon tamoxifen treatment, develops PDA caused by conditional mutations in Kras and Trp53 (ref. ⁶⁴). For breast cancer, we used MMTV-PyMT mice on an FVB background, containing a mouse mammary tumor virus (MMTV) long terminal repeat upstream of the polyomavirus middle T antigen (PyVmT); these mice develop mammary tumors with a mean latency of 53 days⁶⁵. We first designed five gRNAs targeting the mouse Ucp1 gene and transfected them into mouse 3T3-L1 adipocytes, finding all five to upregulate Ucp1 (Extended Data Fig. 4a). We next selected two gRNA for AAV9 packaging and infected 3T3-L1 differentiated adipocytes with these viruses along with AAV9 dCas9–VP64, finding that all of them significantly increased Ucp1 gene expression levels compared to dCas9–VP64-only control (Extended Data Fig. 4b). We used the top-activating gRNA AAV9 for all subsequent experiments. As organoids have several of the aforementioned advantages, we generated adipose organoids using mouse preadipocytes, using techniques similar to those described for human adipose organoids (Methods). We then infected them with AAV9 dCas9–VP64 and Ucp1-gRNA and observed both mCherry expression from our gRNA virus and significant Ucp1 upregulation (Extended Data Fig. 4c).

We next set out to test whether we can suppress pancreatic cancer development using KPC mice. KPC mice were treated with tamoxifen on postnatal day 0–4. At 4 weeks of age, Ucp1-CRISPRa or dCas9–VP64-only (negative control) adipose organoids were orthotopically implanted next to the pancreas (Fig. 4a). After 6 weeks, the pancreases were removed and analyzed. Over those 6 weeks, we observed no significant difference in body weight between mice implanted with Ucp1-CRISPRa organoids or control organoids (dCas9–VP64 only) (Extended Data Fig. 4d). We found that KPC mice implanted with Ucp1-CRISPRa-upregulated adipose organoids had significantly smaller tumors than those in control dCas9–VP64-implanted mice (Fig. 4b), reduced pancreatic mass and Ck19⁺ staining (Fig. 4c and Extended Data Fig. 4e). They also showed reduced expression of the proliferation marker Mki67 as well as genes involved in FAO (Cd36, Cpt1b) (Fig. 4d). In addition, we found Ucp1-CRISPRa mice to have lower expression levels of the pancreas’ main glucose transporter, Glut2, but no significant change for the Gck glycolysis marker (Fig. 4d). Ucp1-CRISPRa mice also had a lower number of Ki67⁺ cells than control mice as determined by immunofluorescence (Fig. 4e and Extended Data Fig. 4e). In addition, we observed lower CA9⁺ and CD31⁺ area per image view in mouse tumors implanted with Ucp1-CRISPR adipose organoids, suggesting reduced hypoxia and angiogenesis (Fig. 4e and Extended Data Fig. 4e). These tumors also had slightly increased levels of caspase-3⁺ cells compared to controls (Extended Data Fig. 4f), suggesting increased apoptosis. Furthermore, we found Ucp1-CRISPRa adipose organoids of KPC mice to have significantly reduced insulin levels compared to control mice (Extended Data Fig. 4g).

**Fig. 4: Implantation of *Ucp1*-CRISPRa adipose organoids in pancreatic and breast cancer genetic mouse models suppresses cancer development.**

To examine whether our treatment might provide systematic therapeutic effect in suppressing breast cancer growth, we next used MMTV-PyMT female mice, implanting Ucp1-CRISPRa or dCas9–VP64 adipose organoids near the third nipples of 4-week-old mice (Fig. 4f). Given that our experiments suggest a nutrient competition model and cold treatment was previously shown to cause widespread BAT activation and subsequent cancer suppression¹⁵, we also examined whether distal implantation of adipose organoids could suppress cancer development by implanting organoids in the back of 4-week-old mice. At 6 weeks post implantation, tumors were dissected and analyzed. We found no difference in body weight of mice implanted with Ucp1-CRISPRa organoids compared to those implanted with control organoids (Extended Data Fig. 4h). Remarkably, we found that both strategies of organoid implantation resulted in significantly reduced tumor size (Fig. 4g) and volume (Fig. 4h) regardless of the site of implantation. Tumors also had decreased expression of the Mki67 proliferation marker and metabolic genes Glut4, Gck, Cd36 and Cpt1b (Fig. 4i). The tumors of mice implanted with Ucp1-CRISPRa adipose organoids had fewer Ki67⁺ cells than control mice (Fig. 4j). These tumors also had lower CA9⁺ and CD31⁺ area per image view and increased caspase-3⁺ (Fig. 4k–m), suggesting that they have reduced hypoxia and angiogenesis and increased apoptosis. Additionally, mice implanted with Ucp1-CRISPRa adipose organoids had significantly lower plasma insulin levels than control mice (Extended Data Fig. 4i). Taken together, our results indicate that our CRISPRa-modulated adipose organoids could have systematic therapeutic effects in suppressing cancer growth.

Breast dissected modulated adipocytes suppress tumor growth

To further demonstrate the therapeutic potential of AMT, we treated adipocytes obtained from dissected human breast tissues with UCP1-CRISPRa AAV9 and tested their ability to suppress tumor progression by co-culturing them with breast cancer organoids generated from dissected breast tumors or grown from metastatic pleural effusions (Fig. 5a). Samples were obtained from patients who underwent breast surgery or thoracentesis. For the co-culture experiments, we used five different breast cancer organoids generated from patients with early stage or metastatic triple-negative breast cancer (ER⁻, PR⁻ and human epidermal growth factor receptor 2-negative (HER2⁻)) or hormone receptor-positive (ER⁺ and/or PR⁺), HER2⁻ breast cancer (Supplementary Table 2). Organoids from breast tumor tissue were generated by digesting cells for 1 h in collagenase and organoids from metastatic pleural effusions were generated by isolating tumor spheroids by centrifugation before, in both cases, embedding cancer cells in organoid culture using established protocols that enable long-term propagation of tumor organoids⁶⁶. In parallel, primary human adipocytes were isolated from human breast tissue using an established protocol⁶⁷. For the triple-negative breast cancer organoids derived from primary tumors, we used two different cases, including one in which we generated organoids and isolated mammary gland adipose tissue from the same individual (TOR41). Adipocytes were infected with dCas9–VP64 only (negative control) or UCP1-CRISPRa AAV9. After 5 days, they showed strong mCherry expression, a fluorescent marker that is part of the gRNA AAV9 (Extended Data Fig. 5a), suggesting that they can be readily infected by our AAVs. In addition, gene analysis showed that both mCherry and UCP1 expression levels were significantly higher in the UCP1-CRISPRa-treated adipose organoids than in the control organoids (Extended Data Fig. 5b). Owing to the limited amount of obtainable dissected tissues, we measured glucose and fatty acid uptake of adipocytes isolated from one sample for which we had more abundant material (TOR40). Upon insulin treatment, we found that UCP1-CRISPRa-modulated mammary gland adipocytes exhibited increased glucose and fatty acid uptake compared to dCas9–VP64-treated adipocytes (Extended Data Fig. 5c).

**Fig. 5: Cancer organoids co-cultured with *UCP1*-CRISPRa adipocytes, both from dissected breast tissue, lead to tumor suppression and prevent cancer development.**

To test whether UCP1-CRISPRa AAV9 adipocytes can suppress cancer growth, we added them to a hydrogel dome on top of the breast cancer organoids with a fully defined organoid medium⁶⁶ and co-cultured for 5 days (Fig. 5a). In all five cases, we observed a significant reduction in cancer organoid size and, in most organoids, also a reduction in organoid number (TOR124 and TOR41 were not significant) upon incubation with adipocytes infected with UCP1-CRISPRa AAV9 compared to the dCas9–VP64-only treated cells (Fig. 5b and Extended Data Fig. 5d). All five cancer organoid cases co-cultured with UCP1-CRISPRa had significantly lower proliferation marker MKI67 expression compared to the negative control (Fig. 5c). We also observed significantly decreased levels of GLUT4, GCK, CD36 and CPT1b for most of the UCP1-CRISPRa co-cultured cancer organoids (Fig. 5d), suggesting that they have lower glycolysis and fatty acid metabolism.

We next tested whether UCP1-CRISPRa primary breast adipocytes can suppress breast cancer growth in vivo. We implanted triple-negative breast cancer organoids derived from resected breast tumor into SCID mice for 3 weeks. We used the floating fraction of mature adipocytes in breast tissue. The cells were cultured, treated with UCP1-CRISPRa and mixed with Matrigel before their co-implantation for 3 weeks into mice with tumors (Fig. 5e). After 3 weeks, the tumors were dissected and examined. We found that tumors co-implanted with UCP1-CRISPRa adipocytes were significantly smaller than those co-implanted with dCas9–VP64-treated adipocytes (Fig. 5f). In addition, tumors co-implanted with UCP1-CRISPRa adipocytes had lower expression of MKI67, glycolytic genes (GLUT4 and GCK) and FAO genes (CD36 and CPT1b) (Fig. 5g). Overall, our results demonstrate that human adipocytes from dissected breast tissue can upregulate UCP1 via AAV9-CRISPRa and are able to reduce glycolysis and fatty acid metabolism and suppress breast cancer organoid growth, both in cell culture and xenografts. Furthermore, this work demonstrates the potential clinical utility of an ex vivo autologous transplantation of CRISPRa-modulated adipocytes to treat cancer.

AMT suppresses high breast cancer risk cell proliferation

To examine whether UCP1-CRISPRa mammary adipocytes might prevent cancer development in individuals at high lifetime risk of cancer, we co-cultured them alongside breast organoids from dissected patient-matched breast tissue of BRCA1/BRCA2/RAD51D mutation carriers (Fig. 5h). We obtained three samples dissected from individual donors carrying a BRCA1 mutation (ORG158), a BRCA2 mutation (ORG150) and a RAD51D mutation (ORG164) (details of specific mutations and patient history in Supplementary Table 2). In all three cases, we observed that breast organoids co-cultured with donor-matched UCP1-CRISPRa adipocytes were significantly smaller and had lower organoid numbers (other than ORG150) than in the negative control (Fig. 5i). In addition, we found that BRCA1/BRCA2 breast organoids co-cultured with UCP1-CRISPRa adipose organoids had lower expression levels of known proliferation markers MKI67 and MTOR than that of controls (Fig. 5j). Furthermore, these breast organoids exhibited reduced expression levels of CK5 and CK17 (other than ORG158) suggesting reduced basal–luminal phenotypes in these BRCA1/BRCA2/RAD51D-heterozygous luminal progenitor-predominant breast cells^68,69 (Fig. 5j). In summary, our results suggest that UCP1-CRISPRa breast adipocytes may inhibit premalignant phenotypes in breast organoids from donors with inherited cancer predisposition syndromes.

Inducible or cell-scaffold AMT suppresses cancer progression

To further enhance the translation capabilities of AMT, we developed inducible AAV vectors designed to turn on dCas9–VP64 upon tetracycline treatment, thereby enabling tight regulation of UCP1 gene expression in adipocytes or adipose organoids. We cloned the reverse tetracycline-control transactivator (rtTA) downstream of the CMV promoter into the gRNA-AAV vector and the tetracycline response element (TRE) and a minimal promoter upstream of dCas9–VP64 in the dCas9–VP64 AAV vector. Upon tetracycline treatment, rtTA binds to TRE and induces expression of dCas9–VP64 (Fig. 6a). To first test this system, we infected TRE-dCas9-VP64-AAV with or without the rtTA-UCP1-gRNA-AAV, followed by the addition of DMSO or doxycycline to human adipocytes. We found that only cells in which both TRE-dCas9-VP64-AAV and rtTA-UCP1-gRNA-AAV were transduced and treated with doxycycline had increased expression levels of Cas9 and UCP1 (Extended Data Fig. 6a). We next generated MCF-7 xenografts in SCID mice and, after 4 weeks, implanted inducible UCP1-CRISPRa human adipose organoids. Following adipose organoid implantation, mice were fed a doxycycline diet for 3 weeks and tumors were subsequently dissected and examined. We found that tumors co-implanted with the inducible UCP1-CRISPRa human adipose organoids were significantly smaller than ones co-implanted with dCas9–VP64-treated adipose organoids (Fig. 6b). These tumors also had lower expression of the proliferation marker MKI67, glycolytic genes (GLUT4 and GCK) and FAO genes (CD36 and CPT1b) (Fig. 6c).

**Fig. 6: Development of inducible AMT systems and the use of AMT to upregulate *UPP1* to suppress PDA.**

To further showcase the clinical applicability of AMT, we inserted UCP1-CRISPRa human adipose organoids into a polymer microwell scaffold generated from polycaprolactone (PCL) and implanted it alongside tumors (Fig. 6d). The microwell scaffold organoid delivery platform is made from biodegradable polyester PCL and manufactured using a microfabrication process involving photolithography and micromolding^70,71,72. We made PCL scaffolds patterned with arrays of cylindrical microwells^{73,74,75,76,77}. The microwells present a controlled 3D microenvironment that provides a supportive niche for the survival, function and enhanced integration of transplanted organoids^72,78,79. Here, we manufactured microwells of 500 μm diameter, specifically designed to accommodate and spatially organize adipose organoids for transplantation. Using scanning electron microscopy (SEM), we found that adipose organoids attached to the microwell scaffold with filopodia and lamellipodia, suggesting that the implant is stable and can be extracted after implantation without leaving residual organoids (Extended Data Fig. 6b). We next generated MCF-7 xenografts in SCID mice for 4 weeks, loaded UCP1-CRISPRa or dCas9–VP64-only (negative control) human adipose organoids onto the device and co-implanted them next to the xenograft (Fig. 6d,e). After 3 weeks, tumors were dissected and examined. We found that tumors co-implanted with the scaffold containing UCP1-CRISPRa human adipose organoids were significantly smaller than tumors co-implanted with the control scaffold adipose organoids containing dCas9–VP64 only (Fig. 6f). Furthermore, these tumors had lower expression levels of the proliferation marker MKI67, glycolytic genes (GLUT4 and GCK) and FAO genes (CD36 and CPT1b) (Fig. 6g). Combined, these data show that AMT can be used in an inducible manner or delivered by an integrated cell-scaffold delivery platform that can be readily removed or replaced to fit the changing tumor metabolic landscape.

Uridine-based AMT suppresses PDA

To examine whether AMT can be used to treat other cancer-associated metabolic pathways, we upregulated the uridine phosphorylase 1 (UPP1) gene in human adipocytes and adipose organoids and tested whether they can suppress PDA, which is known to use uridine-derived ribose in glucose-restricted conditions³⁵. Using CRISPick³⁶, we designed five gRNAs to upregulate UPP1 gene expression in human adipocytes and found that all gRNAs were able to induce UPP1 expression in these cells (Extended Data Fig. 6c). We next generated an AAV9 virus for two gRNAs and transduced human adipocytes with AAV9, finding that both gRNAs significantly induced UPP1 gene expression (Extended Data Fig. 6c). We used the top-activating gRNA AAV for all subsequent experiments. We next transduced human adipocytes with UPP1-CRISPRa AAV9 and incubated these cells with ³H-labelled uridine. We found these adipocytes to significantly uptake uridine during a 12-h incubation (Extended Data Fig. 6d) and have increased lactate levels, a byproduct of uridine catabolism (Extended Data Fig. 6e). We next co-cultured a PDA cell line, PANC-1, with UPP1-CRISPRa or dCas9–VP64 only (negative control) human adipocytes on top of Transwell plates. Cells were cultured in low-glucose-containing media with or without an excess amount of uridine. We found that PANC-1 cells that were co-cultured with the UPP1-CRISPRa adipocytes had a significantly lower number of cells than those co-cultured with dCas9–VP64 adipocytes (Fig. 6h). This growth suppression was abolished with the addition of uridine (Fig. 6h). PANC-1 cells that were co-cultured with UPP1-CRISPRa adipocytes had lower ATP levels than control PANC-1 cells. By contrast, with an excess amount of uridine, there was no apparent difference in ATP levels between PANC-1 cells with UPP1-CRISPRa adipocytes and control cells (Fig. 6i). Furthermore, PANC-1 cells that were co-cultured with UPP1-CRISPRa-modulated adipocytes had lower NADH and lactate levels than control cells (Fig. 6j). We next carried out a similar study but grew the cells in a high-glucose-containing media and found that the growth suppression of PANC-1 by UPP1-CRISPRa adipocytes was reduced (Extended Data Fig. 6f). Combined, these assays suggest that UPP1-CRISPRa adipocytes have increased uridine use and can outcompete PDA for uridine use in low-glucose conditions.

We next set out to test whether UPP1-CRISPRa can suppress PDA xenograft growth. We implanted PANC-1 cells in SCID mice for 4 weeks, followed by the implantation of UPP1-CRISPRa and dCas9–VP64 only (negative control) adipose organoids (Fig. 6k). After 3 weeks, tumors were dissected and examined. We found that tumors co-implanted with UPP1-CRISPRa human adipose organoids were significantly smaller than those co-implanted with dCas9–VP64 adipose organoids (Fig. 6l). In addition, these tumors had lower expression of the proliferation marker MKI67 (Fig. 6m). All the byproducts of uridine catabolism, including NADH and lactate, were found to be significantly reduced in tumors co-implanted with UPP1-CRISPRa human adipose organoids (Fig. 6n). In summary, these results suggest that CRISPRa upregulation of UPP1 in adipocytes and adipose organoids reduces uridine in the tumor microenvironment of PDA and leads to cancer growth suppression. These results also demonstrate the clinical versatility and capacity of AMT to target various cancer metabolic pathways.

Discussion

Cancer cells are fast-proliferating cells that require large amounts of nutrients, including glucose and fatty acids. They can reprogram metabolic pathways to use available substrates in the surrounding environment. Targeting their metabolism can be a potent cancer treatment. Here, we developed a cell-based cancer therapeutic approach that uses modified adipocytes to target various cancer metabolic pathways and has the potential to treat a wide variety of cancers. Adipocytes offer a unique ex vivo therapeutic system, with many of the needed procedures already established in the clinic. Liposuction and fat transplantation are commonly used in many surgical procedures, such as aesthetic and reconstructive surgery. Owing to successful engraftment, adipose tissue transplantation has progressively evolved, not only in plastic and reconstructive surgery but also for therapeutic treatments⁸⁰. Several reports using rodent models have shown that BAT transplantation has beneficial metabolic outcomes^81,82,83,84. These also include the use of UCP1-CRISPRa modulation in human white preadipocytes to induce browning, followed by their transplantation in mice on a HFD, leading to improved body weight, glucose tolerance and insulin sensitivity³⁴. Our work showcases how these ‘brown’ adipocytes and adipose organoids can be used for cancer treatment. In addition, given that adipocytes are ‘metabolic engines’, they can also be engineered to outcompete tumors that use various metabolic pathways, as we show for PDA. Adipocytes are also extensive secretors⁸⁵, and AMT could be used to secrete tumor-targeting factors. The use of cell scaffolds, as done here, could be leveraged by removing and replacing cell scaffolds that contain adipose organoids that target different cancer metabolic pathways, further ‘personalizing’ this therapeutic approach.

There is a growing interest in implementing human adipose tissue grafting by using adipose stem cells or progenitors because of their resistance to trauma and long-term survival following transplantation. One such example is the use of CRISPRa to upregulate the relaxin family peptide receptor 1 (RXFP1) gene in adipose-derived stem cells and their transplantation in a diabetes mellitus-induced erectile dysfunction rat model, showing amelioration of the erectile dysfunction phenotype⁸⁶. Organoids could be advantageous for these cells, as they provide a 3D culture that can better recapitulate the heterogeneity of adipose tissue, respond better to endogenous stimuli and form tissue microenvironments that could more efficiently integrate with cancer cells following transplantation. Several groups have successfully grown adipose organoids from mouse adipose stem cells^33,48,49,87. Here, we were able to culture human adipose organoids from preadipocytes using various conditions from previous studies^33,48,49,87. These human adipose organoids exhibited mature adipocyte markers, including FABP4 and PLIN1. Finally, in our study, we show that the CRISPRa modulation can significantly reduce tumor size, glycolysis, fatty acid metabolism and uridine use and improve hypoxia and angiogenesis in cancer mouse models. Further development of these organoids and their modulation as well as determination of the number of cells needed to achieve a therapeutic benefit could improve these attributes and their therapeutic use for a wide range of diseases.

Our work builds on the recent observation that activating BAT through cold exposure increases adipocyte glucose uptake and lipid metabolism and significantly inhibits tumor progression¹⁵. Here, instead of placing tumor models in cold conditions, we took advantage of CRISPRa to increase the gene expression of key BAT regulators, including UCP1, PPARGC1A and PRDM16, to engineer adipocytes to have increased glucose and fatty acid uptake and metabolism and then used them to deplete resources from cancer cells. Similar to the BAT cold activation studies¹⁵, we observed that high glucose feeding mitigates cancer suppression. In addition, we show that MCF-7 xenograft mice on a HFD have reduced cancer suppression, suggesting that fatty acid competition is also involved in this mechanism. Implanting CRISPRa-modulated adipose organoids distal to the tumors also led to suppression of cancer growth, suggesting that resource competition, similar to BAT cold activation studies¹⁵, can be carried out distal to the tumor and that complex surgical implantation procedures for tumors with limited access might not be needed for AMT. The observed tumor suppression could also be a result of additional mechanisms. The CRISPRa-modulated adipose organoids could modulate whole-body metabolism. Previous studies have shown that hyperinsulinemia can lead to cancer growth due to insulin being a powerful mitogen and survival factor^88,89,90. For example, the administration of dapagliflozin, an SGLT2 inhibitor that lowers blood glucose, and a controlled-release mitochondrial protonophore (CRMP) suppresses cancer growth in mice by reversing hyperinsulinemia⁹¹. Given that BAT is widely known to reduce whole-body blood glucose and insulin levels in humans^15,83,92, we reason that the CRISPRa-modulated adipose organoids could also reduce cancer progression by lowering plasma insulin levels. Indeed, our data, using both xenograft and genetic mouse models, show that mice implanted with CRISPRa-modulated adipose organoids exhibit reduced plasma insulin levels compared to the dCas9–VP64 control mice (Extended Data Figs. 3c and 4g,i).

Among BAT-activating genes, UCP1 showed the most robust effect in terms of cancer suppression. It would be interesting to further develop this AMT approach to upregulate additional genes that could aid in cancer therapy. These could include, for example, upregulation of GLUT1 and GLUT4, which are the main glucose transporters in adipose cells, with GLUT4 being the most abundant and insulin-responsive⁹³; glucose-metabolism-associated genes, such as the transcription factor FOXO1 (ref. ⁹⁴) and the G-protein coupled receptors GPR40 and GPR120, which have been implicated in improved glucose uptake and insulin resistance^95,96; AIFM2, which promotes glycolysis in BAT⁹⁷; and FAO-associated genes, including, for example, the fatty acid transporter CD36, a key transporter for FAO, CPT1b and the fatty acid breakdown enzyme, ACC1. Finally, as shown in our study with UPP1-CRISPRa and PDA, AMT can be customized to fit different cancer metabolic programs. Additional modifications could also be engineered in these adipocytes and adipose organoids, including the use of their endocrine capabilities⁹⁸ to secrete chemotherapeutic drugs or other cancer therapeutic-associated compounds or take advantage of their extracellular vesicles, which are known to have an important role in metabolic regulation⁹⁹. AMT could also be readily implemented with various cancer treatments (surgery, drugs, chemotherapy, radiation and others) that, combined, will increase therapeutic impact.

In this study, we used AAV-based CRISPRa to upregulate genes. CRISPRa has several advantages, including tight regulation owing to the use of the endogenous regulatory machinery¹⁰⁰ and the ability to simultaneously upregulate multiple genes. However, it is worth noting that both upregulation and delivery could also be carried out using other modalities. For example, gene upregulation could be carried out using zinc fingers, TALENs, generation of specific mutations using regular CRISPR editing or base or prime editing in promoters or enhancers, or standard overexpression using a cDNA mammalian expression construct of the gene of interest. Adipocytes could also be ‘browned’ using various differentiation cocktails¹⁰¹, cold activation¹⁵ or drugs, such as β3 agonists^102,103, which could also be used to achieve higher competition for nutrients. In addition, various drugs could be used to induce different metabolic pathways. Delivery could be carried out with other viruses, such as lentiviruses that are widely used for chimeric antigen receptor (CAR)-T cell therapy but have a major caveat of genomic integration, or various non-viral nucleic acid delivery vehicles such as nanoparticles¹⁰⁴ or virus-like particles¹⁰⁵. Various drugs could also be used to upregulate specific genes in adipocytes in a global manner in cancer patients. In addition, the downregulation of certain genes in adipocytes or adipose organoids using CRISPR inactivation (CRISPRi), short interfering RNA, CRISPR editing or other techniques could also be used for AMT. For example, a recent study used CRISPR to deplete the nuclear receptor interacting protein 1 (NRIP1) gene to make ‘brown’ adipocytes that, upon implantation, decreased the adiposity of mice on a HFD¹⁰⁶.

A link between obesity, excess amount of WAT and cancer development and progression has been established, with nearly 40% of all cancer deaths in the United States being attributed to obesity¹⁰⁷. There have been numerous mechanisms proposed to explain how WAT is linked to cancer development and progression, including chronic inflammation, hyperinsulinemia, steroid hormones and adipokines^{108,109,110,111,112,113,114}. In addition, in glucose-rich conditions, cancer cells synthesize de novo fatty acids from intermediates of the glycolysis–TCA cycle (lipogenesis)^115,116. The synthesized fatty acids are then used to synthesize triglycerides and are stored as lipid droplets in cancer cells. When energy is needed, these lipid droplets undergo lipolysis to release fatty acids, which are subjected to β-oxidation^116,117,118. Given that BAT is highly associated with improved glucose tolerance and insulin sensitivity¹¹⁹, one could envision a personalized treatment developed for patients with cancer and obesity, whereby AMT is used not only to target cancer and its unique metabolism but also to treat the patients’ metabolic disease. One major hurdle in our approach that needs to be taken into account is cancer-associated cachexia⁵⁰. Although we did not observe weight loss in our mouse models, a longer treatment time could potentially lead to a reduction in body weight, as was shown for UCP1-CRISPRa mice on a HFD³⁴. Having the ability to control transgene expression in the modified adipocytes and adipose organoids using drugs (such as tetracycline used in our assays) or the ability to remove these cells with the use of an integrated cell-scaffold delivery platform (as also done in our study) could allow the use of this therapeutic approach in a tightly regulated and reversible manner^{120,121,122,123,124,125}. This potential to turn therapeutic intervention on and off holds the key to effectively addressing concerns regarding unforeseen clinical complications, adverse outcomes, changing patient conditions or emerging developments in treatment regimens or tumor metabolic pathways.

In summary, our results provide proof-of-principle results for a cancer therapeutic approach, termed AMT, that can be further developed and personalized for specific cancers and patients. Similar to CAR-T cell therapy, AMT can be readily used in the clinic because cells can be obtained from cancer patients through liposuction or other procedures, engineered and transplanted back into the same individual for therapeutic benefit. The use of adipocytes from dissected breast tissue, as performed in our study, further showcases the clinical utility of such an ex vivo approach. In particular, this could be particularly straightforward for breast cancer, as many mastectomies are followed up by reconstructive surgery with autologous tissue¹²⁶, which could be manipulated before this procedure. Unlike T cells, adipocytes have a lower immune response^34,127, which could allow more straightforward development of ‘off-the-shelf’ adipocytes or adipose organoids for cancer and other treatments. Their ease of growth in culture, longevity, robustness, lower multiplicity¹²⁸ and endocrine capabilities, along with existing clinical procedures for removal and transplantation, make them a beneficial cell type for cancer and other cellular-based disease therapies.

Methods

Human and mouse adipocytes

For adipocyte differentiation, human preadipocytes or mouse 3T3-L1 preadipocytes were used (both kind gifts from H. Sook Sul, UC Berkeley). For human cells, the cell line was immortalized and a single-cell clone was established from primary preadipocytes (Cell Applications, 802S-05A). Cells were cultured to 100% confluency in DMEM, supplemented with 10% FBS, and fresh media was replaced. After 48 h, cells were subjected to adipocyte differentiation by adding a cocktail of IBMX (0.5 M) (Sigma-Aldrich, 410957), dexamethasone (1 µM) (Sigma-Aldrich, D1756) and insulin (10 µg ml⁻¹) (Sigma-Aldrich, I9278). The media was replaced every 2 days with insulin-containing DMEM complete media (Fisher Scientific, 11-965-118) during differentiation.

Cancer cell lines

All cancer cells were acquired from American Type Culture Collection (ATCC). MCF-7 cells (ATCC, HTB-22) were cultured in Eagle’s minimum essential medium (ATCC, 30-2003), supplemented with 10% FBS and 10 µg ml⁻¹ human recombinant insulin (Sigma-Aldrich, I9278). MDA-MB-436 cells (ATCC, HTB-130) were cultured in Leibovitz’s L-15 medium (ATCC, 30-2008) with 10 µg ml⁻¹ insulin,16 µg ml⁻¹ glutathione (Sigma-Aldrich, G6013) and 10% FBS. SW-1417 cells (ATCC, CCL-238) were grown in Leibovitz’s L-15 medium supplemented with 10% FBS. PANC 10.05 cells (ATCC, CRL-2547) were cultured in RPMI-1640 medium (ATCC, 30-2001) with 10 µg ml⁻¹ human insulin and 15% FBS. DU-145 cells (ATCC, HTB-81) were cultured in Eagle’s minimum essential medium with 10% FBS. PANC-1 cells (ATCC, CRL-1469) were cultured in DMEM medium (ATCC, 30-2002) supplemented with 10% FBS.

CRISPRa-AAV in vitro optimization

Five gRNAs targeting the promoter of human UCP1, PPARGC1A, PRDM16 and UPP1 or mouse Ucp1 were designed using CRISPick¹²⁹ (Supplementary Table 3). These guides were individually cloned into pAAV-U6-sasgRNA-CMV-mCherry-WPREpA¹⁰⁰ (Addgene, 217015) at the BstXI and XhoI restriction enzyme sites using the In-Fusion (Takara Bio, 638910) cloning method¹⁰⁰. A total of 5 × 10⁵ human preadipocytes or mouse 3T3-L1 cells were plated onto 12-well plates and then subjected to the adipocyte differentiation protocol after 2 days of confluency. At day 4 of differentiation, cells were transfected with 0.5 µg of dCas9–VP64 and 0.5 µg of gRNA plasmid using X-tremeGene (Sigma-Aldrich, 6366236001) and CombiMag reagent (OZ Biosciences, CM21000) following the manufacturer’s protocol. At day 8 of differentiation, cells were lysed with Trizol, RNA was collected and cDNA and RT–qPCR were performed as described in the ‘RNA isolation and RT–qPCR’ section below. The two gRNAs with the highest upregulation for each gene were packaged into rAAV9 serotype virions. rAAV9 serotype virions were produced by transfecting AAVpro 293T cells (Takara, 632273) with pCMV-sadCas9-VP64 (Addgene, 115790) or pAAV-U6-sasgRNA-CMV-mCherry-WPREpA¹⁰⁰ along with packaging vectors, including PAAV2/9n (Addgene, 112865) and pHelper vectors using TransIT293 reagent (Mirus, 2700). After 72 h, AAV particles were collected and purified using the AAVpro Cell & Sup. Purification Kit Maxi (Takara, 6676) and quantified by the AAVpro Titration Kit (Takara, 6233). gRNA AAV (1 × 10⁶ multiplicity of infection (MOI)) and dCas9–VP64 AAV (1 × 10⁶ MOI) were used to infect human and mouse differentiated adipocytes. After 5 days, RNA was collected, cDNA was prepared and RT–qPCR was carried out as described below.

RNA isolation and RT–qPCR

Total RNA from sorted or cultured cells was extracted using Trizol reagent (Thermo Fisher, 15596026). Reverse transcription was performed with 1 μg of total RNA using the qScript cDNA Synthesis Kit (Quantabio, 95047) following the manufacturer’s protocol. RT–qPCR was performed on the QuantStudio 6 Real-Time PCR System (Thermo Fisher) using Sso Fast (Bio-Rad, 1725205). Statistical analysis was performed using ΔΔCt method with Gapdh primers as a control (see primer sequences in Supplementary Table 3).

Seahorse assay

Human preadipocytes (5 × 10⁵ cells per well) were plated in 12-well plates and subjected to the adipocyte differentiation protocol. Adipocytes and cancer cell lines were trypsinized and reseeded in XF96 plates (Agilent, 102905-100) at 2 × 10⁴ cells per well and assayed the next day. On the day of the experiments, the cells were washed two times and maintained in XF base medium (Agilent, 103334) supplemented with either 1 mM sodium pyruvate (Thermo Fisher, 11360070) and 17.5 mM glucose (Sigma-Aldrich, G7021) for the mitochondria stress test or 2 mM glutamine (StemCell technologies, 07100) for the glycolysis test. For FAO tests, cells were washed and incubated with substrate-limiting medium DMEM (Corning, 17-207-CV) supplemented with 0.5 mM glucose, 1 mM GlutaMAX (Thermo Fisher, 35050061), 0.5 mM carnitine (Sigma-Aldrich, C0283) and 1% FBS. Cells were incubated at 37 °C in a non-CO₂ incubator for 1 h. The assay was performed using a Seahorse XFe96 analyzer (Agilent, XFe96). Oxygen consumption was measured under 1.5 μM oligomycin (Sigma-Aldrich, 75351), 2 μM FCCP (Sigma-Aldrich, C2920) and 0.5 μM Rotenone (Sigma-Aldrich, R8875) and antimycin (Sigma-Aldrich, A8674). The ECAR was measured under 10 mM glucose, 1 µM oligomycin and 50 mM 2-deoxyglucose (Sigma-Aldrich, D8375).

Glucose uptake assay

Human preadipocytes (3 × 10⁵ cells per well) were plated in 96-well tissue culture plates and subjected to the adipocyte differentiation protocol. The day before the assay, the media was replaced with low serum-free media. On the day of the assay, using glucose uptake-Glo assay (Promega, J1342), cells were incubated with 1 nM insulin (Sigma-Aldrich, I9278) for 1 h before the assay following the manufacturer’s protocol.

Fatty acid uptake assay

Human preadipocytes (3 × 10⁵ cells per well) were plated in 96-well tissue culture plates and subjected to the adipocyte differentiation protocol or 1 × 10⁵ human mammary gland adipocytes were plated in 96-well plates. On the day of the assay, the media was replaced with serum-free media. Using the fatty acid uptake kit (Sigma-Aldrich, MAK156), cells were incubated with 1 nM insulin (Sigma-Aldrich, I9278) for 30 min before the assay following the manufacturer’s protocol.

Western blot

To generate a UCP1-positive control, 2 µg of a FLAG-tagged human UCP1 plasmid (Origene, RC218901) was mixed with 200 µl of Opti-MEM (Fisher Scientific, 31985062) and 4 µl of X-tremeGENE HP DNA Transfection Reagent (Roche, XTGHP-RO) and incubated for 15 min. The mixture was added onto AAVpro 293T cells (Takara, 632273) at 6 × 10⁵ per well in a six-well plate and cells were collected 48 h after the transfection. Cells for all conditions were washed twice with DPBS (Sigma-Aldrich, D8537) and dissolved with RIPA buffer (Fisher Scientific, PI89900) containing proteinase inhibitor cocktails and EDTA (Fisher Scientific, PI78440). After shaking for 5 min at 4 °C, cell lysates were centrifuged at 12,000g at 4 °C for 10 min to collect the supernatant, and protein concentrations were determined with a BCA protein assay kit (Fisher Scientific, 23227). The cell lysate was mixed with Laemmli SDS sample buffer, reducing (6X) (Fisher Scientific, J61337.AD) to prepare a 1 µg µl⁻¹ concentration and placed at 95 °C for 5 min in a programmed heat block. SDS–PAGE was performed on the denatured lysate (10 µg per lane for GAPDH and 40 µg ml⁻¹ for other conditions) using Bolt Bis-Tris Mini Protein Gels, 4–12%, 1.0 mm (Fisher Scientific, NW04120BOX) and Bolt MES SDS Running Buffer (Fisher Scientific, B0002). PageRuler Plus Prestained Protein Ladder (Fisher Scientific, 26619) was used as a molecular weight marker. The gels were transferred onto membranes using iBlot2 (Fisher Scientific, IB23002). The membranes were blocked with PVDF Blocking Reagent for Can Get Signal (Toyobo, NYPBR01) for 1 h and treated with primary antibody solutions for hGAPDH and hUCP1(antibody information in Supplementary Table 4) diluted 1,000-fold with Can Get Signal solution 1 (Toyobo, NKB-101) overnight at 4 °C. The membranes were washed twice with DPBS containing 0.02% Tween 20 (Bio-Rad, 1706531) and a secondary antibody solution diluted with Can Get Signal solution 2 was added (Toyobo, NKB-101). After incubation for 1 h at 20 °C room temperature, the membranes were washed twice with DPBS containing 0.02% Tween 20. Antigens were detected with the ECL 2 Western Blotting Substrate detection reagent (Fisher Scientific, PI80196) and imaged on an ImageQuant 800 (Cytiva).

Co-culturing CRISPRa-modulated adipocytes with cancer cells

A total of 5 × 10⁵ human preadipocytes were plated in 12-well tissue culture plates and subjected to the adipocyte differentiation protocol. On day 2 of differentiation, cells were transduced with gRNA AAV (1 × 10⁶ MOI) and dCas9–VP64 AAV (1 × 10⁶ MOI). After 6 days, cells were collected and replated into the upper well of a 12-well Transwell plate (Corning, 07-200-150) in which 3 × 10⁵ cancer cells had been plated in the lower well 1 day earlier. The cells were cultured in the adipocyte differentiation media (as described previously) and designated media for each cancer cell line (1:1 ratio). For PANC-1, cells were incubated with or without the addition of 1 mM uridine. After 3 days, cancer cells were collected for imaging, cell viability assay or seahorse assay. RNA was collected, cDNA was prepared and RT–qPCR was carried out as previously described. Differential expression was determined using the ΔΔCt method with GAPDH primers as control (primer sequences in Supplementary Table 3).

Luminescent cell viability

MCF-7 cells were grown as mentioned above and treated for 3 days with either CRISPRa-UCP1-AAV adipocytes or dCas9–VP64 as a negative control; 6-aminonicotinamide (Sigma-Aldrich, A68203) at 50 µM or 100 µM with DMSO as negative control; and etomoxir (Sigma-Aldrich, E1905) at 100 µM or 200 μM with DMSO as a negative control. The luminescent cell viability assay was performed using the CellTiter-Glo Luminescent Cell Viability Assay (Promega, G7570) following the manufacturer’s protocol on four biological replicates per condition.

Human and mouse adipose organoid culture

Human or mouse 3T3-L1 preadipocytes (0.5 × 10⁶ cells) were plated in 96-well Nunclon Sphera ULA U-bottom plates (Thermo Fisher, 174929). Organoids formed after 48 h and were then differentiated into adipose organoids using a differentiation cocktail containing IBMX (0.5 M), dexamethasone (1 µM) and insulin (10 µg ml⁻¹). Adipocytes formed 21 days post differentiation. Human adipose organoids were then transduced with gRNA AAV (1 × 10⁶ MOI) and dCas9–VP64 AAV (1 × 10⁶ MOI). After 5 days, organoids were collected and mixed with Matrigel (Corning, 354234) and subcutaneously injected into mice (ten organoids per mouse).

Co-transplantation of CRISPRa-modulated human adipose organoids and cancer cells

All animal studies were carried out in accordance with the University of California San Francisco (UCSF) Institutional Animal Care and Use Committee, protocol number AN197608. Mice were housed in a 12 h:12 h light-to-dark cycle; chow diet (Envigo, 2018S) and water were provided ad libitum. Immune-deficient SCID mice (5 weeks old; JAX, 001303) were anesthetized using isoflurane and subcutaneously injected with 2 × 10⁶–6 × 10⁶ cancer cells, or ~2 × 10⁶ cells after trypsinization of tumor organoids and mixed 1:1 with Matrigel, in PBS. After 6–12 weeks, depending on the cancer cell lines, the mice were subcutaneously injected with CRISPRa-AAV human adipocytes or adipose organoids to a site adjacent to the tumor. After 3–6 weeks, mice were killed and tumors and adipose implants were collected. Tumor size was measured with calipers and tumor volume was determined according to the standard formula (length × width² × 0.52).

Immunofluorescence

A portion of the isolated tumors was fixed with 1% paraformaldehyde for 2 h and prepared for cryostat sectioning. Each sample was cut into 5 µm-thick sections and blocked with 3% BSA (Miltenyi Biotec, 130-091-376) blocking solution for 30 min. For Ki67 staining, slides were permeabilized with 1% Triton X-100 PBS solution for 15 min and washed with 0.1% Tween 20 PBS solution before blocking. Slides were then incubated with primary antibodies (information and concentrations are listed in Supplementary Table 4) overnight at 4 °C and washed three times with 0.1% Tween20 PBS. The slides were then incubated with secondary antibodies (information and concentrations are listed in Supplementary Table 4) for 60 min at room temperature and then washed three times with 0.1% Tween 20 PBS. Images were obtained using a confocal microscope (Zeiss, LSM880) and analyzed with Fiji¹³⁰.

Whole-mount staining

Adipose organoids collected after implantation were fixed with 2% paraformaldehyde overnight and then stained with LipidTox Red (Thermo Fisher, H34476) (1:1,000) for 2 h. Images were acquired using a confocal microscope (Zeiss, LSM880).

Metabolic and thermogenic measurements

Oxymax–CLAMS (Columbus Instruments) was used to measure oxygen consumption and carbon dioxide production of individual mice. Mice were housed individually at 30 °C, 16 °C and 4 °C under a 12 h:12 h light-to-dark cycle. Food and water were available ad libitum. We calculated the respiratory exchange ratio by dividing the volume of CO₂ produced by the volume of O₂ consumed.

RNA-seq analysis

Total RNA was isolated from tumors using the RNeasy Plus kit (Qiagen, 74136). Sequencing libraries were prepared using TruSeq Stranded Total RNA Library Prep Kit (Illumina) and sequenced by Novogene using Illumina NovaSeq 6000. Using Partek Flow (Illumina), reads were trimmed and aligned to the human genome (hg38) using STAR¹³¹. Differentially expressed genes were annotated using Limma-voom¹³² and were defined as those that were induced at least ±fourfold, and their expression was significantly different from the basal (5% false discovery rate-adjusted P < 0.01). We used Geneontology.org (https://geneontology.org/)^62,63 to determine gene ontology enrichment using biological processes annotations with Fisher’s Exact test and a 5% false discovery rate-adjusted P value of <0.0001. RNA-seq data are available under GEO accession number GSE246231 (ref. ¹³³).

CRISPRa-modulated mouse adipose organoid implantation in genetic mouse models

KPC, a tamoxifen-inducible Kras^LSL-G12D; p53^LoxP; Pdx1-creER triple-mutant model of tamoxifen-inducible PDA on a C57/BL6 background, was acquired from the Jackson Laboratory (strain no. 032429). To induce Cre recombination, tamoxifen was administered to pups through lactation following oral gavage of the mother with 6 mg tamoxifen (suspended in corn oil) on postnatal days 0, 1, 2 and 4. We orthotopically implanted mouse adipose organoids (ten organoids per mouse) mixed with Matrigel (Corning, 354234) near the pancreas in 4-week-old mice, using a method similar to a previously published protocol¹³⁴, and dissected the pancreas 6 weeks after the implantation. The MMTV-PyMT breast cancer mice were acquired from the Jackson Laboratory (strain no. 002374). Female mice were implanted with mouse adipose organoids at 4 weeks of age (ten organoids per mouse) mixed with Matrigel (Corning, 354234) and subcutaneously injected into mice either at the third nipple or on the back. The tumors were collected 6 weeks post implantation.

Cancer organoids

Organoids were generated from breast tumor tissue or tumors from malignant effusions. Patients provided their consent for specimen collection using IRB-approved tissue collection protocols at the UCSF and the Brigham and Women’s Hospital. Surgical specimens were obtained from the UCSF Medical Center or Brigham and Women’s Hospital on the day of their procedure, viably frozen as tissue pieces or used to generate formalin-fixed, paraffin-embedded sections or organoid cultures. Each tissue was minced using razor blades and digested in a solution containing DMEM/F12 (Gibco, 11330), 2 mM GlutaMAX (Gibco, 35050), 10 mM HEPES (Gibco, 15630), 50 U ml⁻¹ penicillin–streptomycin (Gibco, 15070) and 1 mg ml⁻¹ collagenase XI (Sigma-Aldrich, C9407). Tissue digestion was performed at 37 °C with constant shaking at 150 rpm for 1–2 h. Cells were then pelleted by centrifugation, further dissociated by sequentially pipetting with 10 ml, 5 ml and 1 ml pipette tips and then recentrifuged. The resulting cell pellet was used directly to establish organoid cultures by embedding in basement membrane extract, allowing this to harden at 37 °C for 20 min to form a hydrogel dome and then overlaying this dome with Type 1 Organoid Medium as previously described^69,135. All metastatic cancer organoids were derived from malignant pleural effusions that were collected from patients with metastatic breast cancer who were undergoing thoracentesis at the UCSF Medical Center. The fluid samples were placed on ice within 4 h after collection. Organoids were generated by washing malignant effusions with PBS, collecting tumor spheroids by centrifugation and incubating with 3–5 ml RBC lysis buffer (BioLegend, 420301) for 10–15 min when there were visible RBCs, followed by embedding the cell pellet in organoid culture as described above.

Metabolomics

The tumors were collected and subjected to metabolomics analysis to measure primary metabolites by the University of California Davis West Coast Metabolomics Center. The data were normalized to tumor mass (Supplementary Table 1).

Mammary gland adipocyte isolation

For adipocyte extraction, mammary gland adipose tissues from excess tissue removed during breast surgeries were washed with PBS three times and mechanically minced into smaller pieces. The tissue mixture was incubated with PBS with 3% BSA and collagenase I (1 mg ml⁻¹) for 45 min at 37 °C and then centrifuged; the top lipid layer was then discarded. The remaining mixture was filtered twice through a 200 µm strainer and centrifuged. Mature adipocytes were isolated and washed twice with PBS and then plated in suspension in six-well plates in high-glucose DMEM media supplemented with 10% FBS. For transduction of CRISPRa, UCP1 AAV9 (1 × 10⁶ MOI) and dCas9–VP64 AAV9 (1 × 10⁶ MOI) were mixed with 20 µl of AdeoMag (OZ Biosciences, AM71000) and adipocytes and incubated for 15 min. The virus and cell mixture was distributed onto a six-well plate placed upon the magnetic plate and incubated for 15 more minutes.

Co-culturing of cancer organoids and adipocytes

At 5 days post infection of CRISPRa AAV9, adipocytes were placed in a tissue Transwell culture plate (Corning, 07-200-150) of tumor organoids at a 1:1 ratio of adipocyte media and breast cancer organoid media. Cells were incubated for up to 7 days and then examined for adipocyte and tumor phenotypes.

Co-implantation of cancer organoids and adipocytes

Immune-deficient SCID mice (5 weeks old; JAX, 001303) were anesthetized using isoflurane and subcutaneously injected with 3 × 10⁵ cancer organoids mixed 1:1 with Matrigel in PBS. After 3 weeks, we subcutaneously injected CRISPRa-AAV mammary adipocytes mixed with Matrigel to a site adjacent to the tumor. After another 3 weeks, tumors and adipose implants were dissected. We measured tumor size with calipers and determined tumor volume according to the standard formula (length × width² × 0.52).

Tet-On CRISPRa

The rtTA was amplified from pSBtet-GP (Addgene, 60495) along with a sequence encoding for T2A and cloned downstream of the CMV promoter in the gRNA-AAV vector digested with NheI and AgeI. The TRE and a minimal promoter were amplified and cloned upstream of dCas9–VP64 in the dCas9–VP64 AAV vector digested with XhoI and AgeI. After adipose organoids were infected with Tet-On UCP1-CRISPRa and implanted, mice were fed with a doxycycline diet (Bio-Serv, S3888).

PCL microwell scaffold fabrication

Microwell scaffolds were fabricated using a combination of microfabrication techniques, including photolithography and micromolding. We used PCL (Sigma-Aldrich, 440744), a biocompatible polymer to fabricate microwell scaffolds. We used a polymer blend of 150 mg ml⁻¹ PCL (molecular weight, 80K; Sigma-Aldrich, 9016-00-6) and polyethylene glycol (molecular weight, 2.5K; Sigma-Aldrich, 438197) in trifluoroethanol (Sigma-Aldrich, 8.08259). The polymer solution was stirred on the roller bank overnight at room temperature. We then used photolithography to fabricate the SU8 (Microchem/Kayaku, Y111075) microwell array master mold. The microwell mold was then silanized and polydimethylsiloxane (PDMS) (Thermo Fisher, 178442500) and was used to prepare the microwell template. The PDMS template had a micropatterned spatially organized array of dome-like structures called microwells, each 700 μm in height and 500 μm diameter. Micromolding was used to prepare the PCL microwell scaffold using a PDMS microwell template. Two-step spin casting was used to cast the PCL solution on the PDMS microwell template at 100 rpm and 1,000 rpm for 10 s to regulate the height of the microwell; the microwell diameter was defined by the microwells on the template. The PCL-coated template was allowed to dry at room temperature overnight to evaporate the organic solvent, followed by immersion in deionized water. The polyethylene glycol in the PCL polymer blend was allowed to leach out in a deionized water bath for 5 days to introduce the porosity in the microwell PCL scaffold. Subsequently, the microwell scaffolds were washed and stored in deionized water till further use. The size of the microwells on the PCL microwell scaffold was designed to be suitable to host adipose organoids.

Adipose organoids microwell scaffold transplantation

The microwell scaffold organoid delivery platform was primed for the transplantation of adipose organoids. The microwell scaffolds were thoroughly washed and dried followed by treatment with air plasma for 3 min in aseptic conditions. We then soaked the scaffolds in a solution of adipocyte differentiation media and Matrigel (Corning, 354234) at 4 °C for 24 h. The adipose organoids were then spin-loaded individually in each microwell at 1,000 rpm and 4 °C. The microwell scaffolds loaded with adipose organoids were then layered with a 1:1 ratio of adipocyte differentiation media and Matrigel. The scaffold-organoids system was then cultured at 37 °C at 5% CO₂ for 24 h. The immunodeficient SCID mice (JAX, 001303) were antecedently prepped with orthotopic MCF-7 tumor. The adipose organoids-loaded microwell organoid delivery scaffolds were then transplanted into the mice adjacent to the orthotopic tumor. After 3 weeks, tumors and microwell scaffold adipose organoid implants were collected. The tumor size was measured with calipers and tumor volume was determined according to the standard formula (length × width² × 0.52).

Scanning electron microscopy

Adipose organoids on the microwell scaffolds were meticulously prepared for analysis and visualization of their ultrastructural characteristics using SEM. The organoids on the scaffolds were fixed by immersing them in a 2.5% glutaraldehyde (Electron Microscopic Sciences, 16210) solution in a 0.1 M sodium cacodylate (Sigma-Aldrich, C0250) buffer with a pH of 7.2 at 4 °C over 24 h. Subsequently, samples were thoroughly rinsed for 30 min with 0.1 M sodium cacodylate buffer with a pH of 7.4 to remove residual fixative. To enhance structural contrast, post-fixation was performed using a 2% phosphotungstic acid (Sigma-Aldrich, P4006) in 0.1 M sodium cacodylate buffer at 4 °C for 1 h followed by thoroughly rinsing for 30 min with 0.1 M sodium cacodylate buffer with a pH of 7.4. The samples were then dehydrated by immersing for 20 min each in a graded series of ethanol (35%, 50%, 70%, 95%, 100%) solutions, enabling the gradual removal of water content. Critical-point drying over 24 h using hexamethyldisilazane (Sigma-Aldrich, 440191) ensured the complete removal of moisture from the specimens without causing any damage. Finally, a thin iridium coating of ~2 nm thickness was sputter-coated on the samples to enhance conductivity during SEM imaging. This systematic sample preparation method facilitated high-resolution SEM imaging at 1 kV on Phenom Pharos G2 Desktop FEG-SEM (Thermo Fisher Scientific), enabling a comprehensive exploration of the ultrastructural morphology.

Uridine and uridine triphosphate (UTP) uptake studies

Adipocytes differentiated and treated with or without UPP1-CRISPRa were maintained in a 24-well plate using cell culture medium (see ‘Cell culture media’ section). Before the uptake experiments, the culture medium was discarded and the cells were incubated in 1.0 ml of Hanks’ balanced salt solution (HBSS; Gibco, 14025092) for 10–20 min at 37 °C. After this pre-incubation, trace amounts of either ³H-uridine (128 nM) or 3H-UTP (240 nM) were introduced to the cells and incubation continued for 1 h or 12 h. Subsequently, the cells were washed twice with ice-cold HBSS. After washing, the cells were lysed using a lysis buffer composed of 0.1 N NaOH and 0.1% SDS. The radioactivity in the lysate was then quantified through liquid scintillation counting.

Statistical analysis

Statistical analyses were performed using one-way ANOVA or two-tailed t-tests and are described in each figure legend. A P value of <0.05 was considered statistically significant. The number of mice or replicates used in each experiment are indicated in the figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All data are available upon request. Requests for materials should be directed to the corresponding author. RNA-seq data are available under GEO accession number GSE246231 (ref. ¹³³). Source data are provided with this paper.

References

Lin, X., Xiao, Z., Chen, T., Liang, S. H. & Guo, H. Glucose metabolism on tumor plasticity, diagnosis, and treatment. Front. Oncol. 10, 317 (2020).
Article PubMed PubMed Central Google Scholar
Hay, N. Reprogramming glucose metabolism in cancer: Can it be exploited for cancer therapy? Nat. Rev. Cancer 16, 635–649 (2016).
Article CAS PubMed PubMed Central Google Scholar
Liberti, M. V. & Locasale, J. W. The Warburg effect: How does it benefit cancer cells? Trends Biochem. Sci. 41, 211–218 (2016).
Article CAS PubMed PubMed Central Google Scholar
Nagarajan, S. R., Butler, L. M. & Hoy, A. J. The diversity and breadth of cancer cell fatty acid metabolism. Cancer Metab. 9, 2 (2021).
Article PubMed PubMed Central Google Scholar
Palm, W. & Thompson, C. B. Nutrient acquisition strategies of mammalian cells. Nature 546, 234–242 (2017).
Article CAS PubMed PubMed Central Google Scholar
DeWaal, D. et al. Hexokinase-2 depletion inhibits glycolysis and induces oxidative phosphorylation in hepatocellular carcinoma and sensitizes to metformin. Nat. Commun. 9, 446 (2018).
Article PubMed PubMed Central Google Scholar
Chan, D. A. et al. Targeting GLUT1 and the Warburg effect in renal cell carcinoma by chemical synthetic lethality. Sci. Transl. Med. 3, 94ra70 (2011).
Article CAS PubMed PubMed Central Google Scholar
Liu, Y. et al. A small-molecule inhibitor of glucose transporter 1 downregulates glycolysis, induces cell-cycle arrest, and inhibits cancer cell growth in vitro and in vivo. Mol. Cancer Ther. 11, 1672–1682 (2012).
Article CAS PubMed Google Scholar
Fernandez, L. P., Gomez de Cedron, M. & Ramirez de Molina, A. Alterations of lipid metabolism in cancer: implications in prognosis and treatment. Front. Oncol. 10, 577420 (2020).
Article PubMed PubMed Central Google Scholar
Khan, W. et al. Lipid metabolism in cancer: a systematic review. J. Carcinog. 20, 4 (2021).
Article CAS PubMed PubMed Central Google Scholar
Wang, W., Bai, L., Li, W. & Cui, J. The lipid metabolic landscape of cancers and new therapeutic perspectives. Front. Oncol. 10, 605154 (2020).
Article PubMed PubMed Central Google Scholar
Snaebjornsson, M. T., Janaki-Raman, S. & Schulze, A. Greasing the wheels of the cancer machine: the role of lipid metabolism in cancer. Cell Metab. 31, 62–76 (2020).
Article CAS PubMed Google Scholar
Mason, P. et al. SCD1 inhibition causes cancer cell death by depleting mono-unsaturated fatty acids. PLoS ONE 7, e33823 (2012).
Article CAS PubMed PubMed Central Google Scholar
Guseva, N. V., Rokhlin, O. W., Glover, R. A. & Cohen, M. B. TOFA (5-tetradecyl-oxy-2-furoic acid) reduces fatty acid synthesis, inhibits expression of AR, neuropilin-1 and Mcl-1 and kills prostate cancer cells independent of p53 status. Cancer Biol. Ther. 12, 80–85 (2011).
Article CAS PubMed Google Scholar
Seki, T. et al. Brown-fat-mediated tumour suppression by cold-altered global metabolism. Nature 608, 421–428 (2022).
Article CAS PubMed PubMed Central Google Scholar
Symonds, M. E., Aldiss, P., Pope, M. & Budge, H. Recent advances in our understanding of brown and beige adipose tissue: the good fat that keeps you healthy. F1000Res 7, F1000 (2018).
Article PubMed PubMed Central Google Scholar
Virtanen, K. A. et al. Functional brown adipose tissue in healthy adults. N. Engl. J. Med. 360, 1518–1525 (2009).
Article CAS PubMed Google Scholar
Cannon, B. & Nedergaard, J. Brown adipose tissue: function and physiological significance. Physiol. Rev. 84, 277–359 (2004).
Article CAS PubMed Google Scholar
Klingenberg, M. Uncoupling protein—a useful energy dissipator. J. Bioenerg. Biomembr. 31, 419–430 (1999).
Article CAS PubMed Google Scholar
Suzuki, D., Murata, Y. & Oda, S. Changes in Ucp1, D2 (Dio2) and Glut4 (Slc2a4) mRNA expression in response to short-term cold exposure in the house musk shrew (Suncus murinus). Exp. Anim. 56, 279–288 (2007).
Article CAS PubMed Google Scholar
Vimaleswaran, K. S., Radha, V., Deepa, R. & Mohan, V. Absence of association of metabolic syndrome with PPARGC1A, PPARG and UCP1 gene polymorphisms in Asian Indians. Metab. Syndr. Relat. Disord. 5, 153–162 (2007).
Article CAS PubMed Google Scholar
Feldmann, H. M., Golozoubova, V., Cannon, B. & Nedergaard, J. UCP1 ablation induces obesity and abolishes diet-induced thermogenesis in mice exempt from thermal stress by living at thermoneutrality. Cell Metab. 9, 203–209 (2009).
Article CAS PubMed Google Scholar
Tabuchi, C. & Sul, H. S. Corrigendum: signaling pathways regulating thermogenesis. Front. Endocrinol. (Lausanne) 12, 698619 (2021).
Article PubMed Google Scholar
Yi, D. et al. Zc3h10 acts as a transcription factor and is phosphorylated to activate the thermogenic program. Cell Rep. 29, 2621–2633.e4 (2019).
Article CAS PubMed PubMed Central Google Scholar
Puigserver, P. & Spiegelman, B. M. Peroxisome proliferator–activated receptor-γ coactivator 1α (PGC-1α): transcriptional coactivator and metabolic regulator. Endocr. Rev. 24, 78–90 (2003).
Article CAS PubMed Google Scholar
Lin, J. et al. Defects in adaptive energy metabolism with CNS-linked hyperactivity in PGC-1α null mice. Cell 119, 121–135 (2004).
Article CAS PubMed Google Scholar
Kajimura, S. Promoting brown and beige adipocyte biogenesis through the PRDM16 pathway. Int. J. Obes. Suppl. 5, S11–S14 (2015).
Article CAS PubMed PubMed Central Google Scholar
Harms, M. J. et al. PRDM16 binds MED1 and controls chromatin architecture to determine a brown fat transcriptional program. Genes Dev. 29, 298–307 (2015).
Article CAS PubMed PubMed Central Google Scholar
Ohno, H., Shinoda, K., Spiegelman, B. M. & Kajimura, S. PPARgamma agonists induce a white-to-brown fat conversion through stabilization of PRDM16 protein. Cell Metab. 15, 395–404 (2012).
Article CAS PubMed PubMed Central Google Scholar
Kajimura, S. et al. Initiation of myoblast to brown fat switch by a PRDM16-C/EBP-β transcriptional complex. Nature 460, 1154–1158 (2009).
Article CAS PubMed PubMed Central Google Scholar
Seale, P. et al. PRDM16 controls a brown fat/skeletal muscle switch. Nature 454, 961–967 (2008).
Article CAS PubMed PubMed Central Google Scholar
Kajimura, S. et al. Regulation of the brown and white fat gene programs through a PRDM16/CtBP transcriptional complex. Genes Dev. 22, 1397–1409 (2008).
Article CAS PubMed PubMed Central Google Scholar
Seale, P. et al. Transcriptional control of brown fat determination by PRDM16. Cell Metab. 6, 38–54 (2007).
Article CAS PubMed PubMed Central Google Scholar
Wang, C. H. et al. CRISPR-engineered human brown-like adipocytes prevent diet-induced obesity and ameliorate metabolic syndrome in mice. Sci. Transl. Med. 12, eaaz8664 (2020).
Article CAS PubMed PubMed Central Google Scholar
Nwosu, Z. C. et al. Uridine-derived ribose fuels glucose-restricted pancreatic cancer. Nature 618, 151–158 (2023).
Article CAS PubMed PubMed Central Google Scholar
Kim, H. K. et al. Deep learning improves prediction of CRISPR-Cpf1 guide RNA activity. Nat. Biotechnol. 36, 239–241 (2018).
Article CAS PubMed Google Scholar
Flint, J. & Shenk, T. Viral transactivating proteins. Annu. Rev. Genet. 31, 177–212 (1997).
Article CAS PubMed Google Scholar
Wu, Z., Yang, H. & Colosi, P. Effect of genome size on AAV vector packaging. Mol. Ther. 18, 80–86 (2010).
Article CAS PubMed Google Scholar
Bates, R., Huang, W. & Cao, L. Adipose tissue: an emerging target for adeno-associated viral vectors. Mol. Ther. Methods Clin. Dev. 19, 236–249 (2020).
Article CAS PubMed PubMed Central Google Scholar
Kaushik, N., Kaushik, N. K., Choi, E. H. & Kim, J. H. Blockade of cellular energy metabolism through 6-aminonicotinamide reduces proliferation of non-small lung cancer cells by inducing endoplasmic reticulum stress. Biology (Basel) 10, 1088 (2021).
CAS PubMed Google Scholar
Li, Y. et al. Targeting glucose-6-phosphate dehydrogenase by 6-AN induces ROS-mediated autophagic cell death in breast cancer. FEBS J. 290, 763–779 (2023).
Article CAS PubMed Google Scholar
Varshney, R., Dwarakanath, B. & Jain, V. Radiosensitization by 6-aminonicotinamide and 2-deoxy-d-glucose in human cancer cells. Int. J. Radiat. Biol. 81, 397–408 (2005).
Article CAS PubMed Google Scholar
O’Connor, R. S. et al. The CPT1a inhibitor, etomoxir induces severe oxidative stress at commonly used concentrations. Sci. Rep. 8, 6289 (2018).
Article PubMed PubMed Central Google Scholar
Shim, J.-K. et al. Etomoxir, a carnitine palmitoyltransferase 1 inhibitor, combined with temozolomide reduces stemness and invasiveness in patient-derived glioblastoma tumorspheres. Cancer Cell Int. 22, 309 (2022).
Article CAS PubMed PubMed Central Google Scholar
Lee, C.-K. et al. Tumor metastasis to lymph nodes requires YAP-dependent metabolic adaptation. Science 363, 644–649 (2019).
Article CAS PubMed Google Scholar
Loo, S. Y. et al. Fatty acid oxidation is a druggable gateway regulating cellular plasticity for driving metastasis in breast cancer. Sci. Adv. 7, eabh2443 (2021).
Article CAS PubMed PubMed Central Google Scholar
Manerba, M. et al. Metabolic activation triggered by cAMP in MCF-7 cells generates lethal vulnerability to combined oxamate/etomoxir. Biochim. Biophys. Acta Gen. Subj. 1863, 1177–1186 (2019).
Article CAS PubMed Google Scholar
Strobel, H. A., Gerton, G. & Hoying, J. B. Vascularized adipocyte organoid model using isolated human microvessel fragments. Biofabrication 13, 035022 (2021).
Article CAS Google Scholar
Taylor, J. et al. Generation of immune cell containing adipose organoids for in vitro analysis of immune metabolism. Sci. Rep. 10, 21104 (2020).
Article CAS PubMed PubMed Central Google Scholar
Kir, S. & Spiegelman, B. M. Cachexia and brown fat: a burning issue in cancer. Trends Cancer 2, 461–463 (2016).
Article PubMed PubMed Central Google Scholar
Bierie, B. et al. Integrin-β4 identifies cancer stem cell-enriched populations of partially mesenchymal carcinoma cells. Proc. Natl Acad. Sci. USA 114, E2337–E2346 (2017).
Article CAS PubMed PubMed Central Google Scholar
Saito, Y. et al. LLGL2 rescues nutrient stress by promoting leucine uptake in ER⁺ breast cancer. Nature 569, 275–279 (2019).
Article CAS PubMed Google Scholar
Takaku, M., Grimm, S. A. & Wade, P. A. GATA3 in breast cancer: Tumor suppressor or oncogene? Gene Expr. 16, 163–168 (2015).
Article CAS PubMed PubMed Central Google Scholar
Jia, L. et al. EEF1A2 interacts with HSP90AB1 to promote lung adenocarcinoma metastasis via enhancing TGF-β/SMAD signalling. Br. J. Cancer 124, 1301–1311 (2021).
Article CAS PubMed PubMed Central Google Scholar
Giudici, F. et al. Elevated levels of eEF1A2 protein expression in triple negative breast cancer relate with poor prognosis. PLoS ONE 14, e0218030 (2019).
Article CAS PubMed PubMed Central Google Scholar
Abrahams, A., Parker, M. I. & Prince, S. The T-box transcription factor Tbx2: its role in development and possible implication in cancer. IUBMB Life 62, 92–102 (2010).
Article CAS PubMed Google Scholar
Yang, H. et al. HOXD10 acts as a tumor-suppressive factor via inhibition of the RHOC/AKT/MAPK pathway in human cholangiocellular carcinoma. Oncol. Rep. 34, 1681–1691 (2015).
Article CAS PubMed PubMed Central Google Scholar
Chang, J. W. et al. Wild-type p53 upregulates an early onset breast cancer-associated gene GAS7 to suppress metastasis via GAS7–CYFIP1-mediated signaling pathway. Oncogene 37, 4137–4150 (2018).
Article CAS PubMed PubMed Central Google Scholar
Liu, S. et al. MAP2K4 interacts with Vimentin to activate the PI3K/AKT pathway and promotes breast cancer pathogenesis. Aging (Albany NY) 11, 10697–10710 (2019).
Article CAS PubMed Google Scholar
Aizawa, T. et al. Cancer-associated fibroblasts secrete Wnt2 to promote cancer progression in colorectal cancer. Cancer Med. 8, 6370–6382 (2019).
Article CAS PubMed PubMed Central Google Scholar
Joshi, S. et al. Rac2 controls tumor growth, metastasis and M1–M2 macrophage differentiation in vivo. PLoS ONE 9, e95893 (2014).
Article PubMed PubMed Central Google Scholar
Ashburner, M. et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet. 25, 25–29 (2000).
Article CAS PubMed PubMed Central Google Scholar
Gene Ontology Consortium et al. The gene ontology knowledgebase in 2023. Genetics 224, iyad031 (2023).
Article Google Scholar
Maddipati, R. & Stanger, B. Z. Pancreatic cancer metastases harbor evidence of polyclonality. Cancer Discov. 5, 1086–1097 (2015).
Article CAS PubMed PubMed Central Google Scholar
Guy, C. T., Cardiff, R. D. & Muller, W. J. Induction of mammary tumors by expression of polyomavirus middle T oncogene: a transgenic mouse model for metastatic disease. Mol. Cell. Biol. 12, 954–961 (1992).
CAS PubMed PubMed Central Google Scholar
Dekkers, J. F. et al. Long-term culture, genetic manipulation and xenotransplantation of human normal and breast cancer organoids. Nat. Protoc. 16, 1936–1965 (2021).
Article CAS PubMed PubMed Central Google Scholar
Picon-Ruiz, M., Marchal, J. A. & Slingerland, J. M. Obtaining human breast adipose cells for breast cancer cell co-culture studies. STAR Protoc. 1, 100197 (2020).
Article PubMed PubMed Central Google Scholar
Shalabi, S. F. et al. Evidence for accelerated aging in mammary epithelia of women carrying germline BRCA1 or BRCA2 mutations. Nat. Aging 1, 838–849 (2021).
Article PubMed PubMed Central Google Scholar
Gray, G. K. et al. A human breast atlas integrating single-cell proteomics and transcriptomics. Dev. Cell 57, 1400–1420.e7 (2022).
Article CAS PubMed PubMed Central Google Scholar
Nyitray, C. E., Chavez, M. G. & Desai, T. A. Compliant 3D microenvironment improves β-cell cluster insulin expression through mechanosensing and β-catenin signaling. Tissue Eng. Part A 20, 1888–1895 (2014).
Article CAS PubMed PubMed Central Google Scholar
Jeong, G. S. et al. Viscoelastic lithography for fabricating self-organizing soft micro-honeycomb structures with ultra-high aspect ratios. Nat. Commun. 7, 11269 (2016).
Article CAS PubMed PubMed Central Google Scholar
Girgin, M. U. et al. Bioengineered embryoids mimic post-implantation development in vitro. Nat. Commun. 12, 5140 (2021).
Article CAS PubMed PubMed Central Google Scholar
Wise, K. D. & Najafi, K. Microfabrication techniques for integrated sensors and microsystems. Science 254, 1335–1342 (1991).
Article CAS PubMed Google Scholar
Leong, T. G., Zarafshar, A. M. & Gracias, D. H. Three-dimensional fabrication at small size scales. Small 6, 792–806 (2010).
Article CAS PubMed PubMed Central Google Scholar
Steedman, M. R., Tao, S. L., Klassen, H. & Desai, T. A. Enhanced differentiation of retinal progenitor cells using microfabricated topographical cues. Biomed. Microdevices 12, 363–369 (2010).
Article PubMed PubMed Central Google Scholar
Kharbikar, B. N., Kumar, S. H., Kr, S. & Srivastava, R. Hollow silicon microneedle array based trans-epidermal antiemetic patch for efficient management of chemotherapy induced nausea and vomiting. In Micro + Nano Materials, Devices, and Systems Vol. 9668 (eds Eggleton, B. J. & Palomba, S.) 256–272 (SPIE, 2015).
Bernards, D. A. et al. Injectable devices for delivery of liquid or solid protein formulations. ACS Materials Au 3, 255–264 (2023).
Article CAS PubMed PubMed Central Google Scholar
Kharbikar, B. N., Chendke, G. S. & Desai, T. A. Modulating the foreign body response of implants for diabetes treatment. Adv. Drug Deliv. Rev. 174, 87–113 (2021).
Article CAS PubMed PubMed Central Google Scholar
Kharbikar, B. N., Mohindra, P. & Desai, T. A. Biomaterials to enhance stem cell transplantation. Cell Stem Cell 29, 692–721 (2022).
Article CAS PubMed PubMed Central Google Scholar
Shukla, L., Yuan, Y., Shayan, R., Greening, D. W. & Karnezis, T. Fat therapeutics: the clinical capacity of adipose-derived stem cells and exosomes for human disease and tissue regeneration. Front. Pharmacol. 11, 158 (2020).
Article CAS PubMed PubMed Central Google Scholar
White, J. D., Dewal, R. S. & Stanford, K. I. The beneficial effects of brown adipose tissue transplantation. Mol. Aspects Med. 68, 74–81 (2019).
Article CAS PubMed PubMed Central Google Scholar
Liu, X. et al. Brown adipose tissue transplantation improves whole-body energy metabolism. Cell Res. 23, 851–854 (2013).
Article CAS PubMed PubMed Central Google Scholar
Stanford, K. I. et al. Brown adipose tissue regulates glucose homeostasis and insulin sensitivity. J. Clin. Invest. 123, 215–223 (2013).
Article CAS PubMed Google Scholar
Liu, X. et al. Brown adipose tissue transplantation reverses obesity in Ob/Ob mice. Endocrinology 156, 2461–2469 (2015).
Article CAS PubMed Google Scholar
Pogodzinski, D., Ostrowska, L., Smarkusz-Zarzecka, J. & Zysk, B. Secretome of adipose tissue as the key to understanding the endocrine function of adipose tissue. Int. J. Mol. Sci. 23, 2309 (2022).
Article CAS PubMed PubMed Central Google Scholar
Sun, T. et al. Engineered adipose-derived stem cells overexpressing RXFP1 via CRISPR activation ameliorate erectile dysfunction in diabetic rats. Antioxidants 12, 171 (2023).
Article CAS PubMed PubMed Central Google Scholar
Rudolph, M. C., Wellberg, E. A. & Anderson, S. M. Adipose-depleted mammary epithelial cells and organoids. J. Mammary Gland Biol. Neoplasia 14, 381–386 (2009).
Article PubMed PubMed Central Google Scholar
Currie, C. J., Poole, C. D. & Gale, E. A. The influence of glucose-lowering therapies on cancer risk in type 2 diabetes. Diabetologia 52, 1766–1777 (2009).
Article CAS PubMed Google Scholar
Hemkens, L. G. et al. Risk of malignancies in patients with diabetes treated with human insulin or insulin analogues: a cohort study. Diabetologia 52, 1732–1744 (2009).
Article CAS PubMed PubMed Central Google Scholar
Godsland, I. F. Insulin resistance and hyperinsulinaemia in the development and progression of cancer. Clin. Sci. (Lond) 118, 315–332 (2009).
Article PubMed Google Scholar
Nasiri, A. R., Rodrigues, M. R., Li, Z., Leitner, B. P. & Perry, R. J. SGLT2 inhibition slows tumor growth in mice by reversing hyperinsulinemia. Cancer Metab. 7, 10 (2019).
Article PubMed PubMed Central Google Scholar
Chondronikola, M. et al. Brown adipose tissue improves whole-body glucose homeostasis and insulin sensitivity in humans. Diabetes 63, 4089–4099 (2014).
Article CAS PubMed PubMed Central Google Scholar
Chadt, A. & Al-Hasani, H. Glucose transporters in adipose tissue, liver, and skeletal muscle in metabolic health and disease. Pflugers Arch. 472, 1273–1298 (2020).
Article CAS PubMed PubMed Central Google Scholar
Kousteni, S. FoxO1, the transcriptional chief of staff of energy metabolism. Bone 50, 437–443 (2012).
Article CAS PubMed Google Scholar
Schilperoort, M. et al. The GPR120 agonist TUG-891 promotes metabolic health by stimulating mitochondrial respiration in brown fat. EMBO Mol. Med. 10, e8047 (2018).
Article PubMed PubMed Central Google Scholar
Satapati, S. et al. GPR120 suppresses adipose tissue lipolysis and synergizes with GPR40 in antidiabetic efficacy. J. Lipid Res. 58, 1561–1578 (2017).
Article CAS PubMed PubMed Central Google Scholar
Nguyen, H. P. et al. Aifm2, a NADH oxidase, supports robust glycolysis and is required for cold- and diet-induced thermogenesis. Mol. Cell 77, 600–617.e4 (2020).
Article CAS PubMed PubMed Central Google Scholar
Wang, P., Mariman, E., Renes, J. & Keijer, J. The secretory function of adipocytes in the physiology of white adipose tissue. J. Cell. Physiol. 216, 3–13 (2008).
Article CAS PubMed Google Scholar
Bond, S. T., Calkin, A. C. & Drew, B. G. Adipose-derived extracellular vesicles: systemic messengers and metabolic regulators in health and disease. Front. Physiol. 13, 837001 (2022).
Article PubMed PubMed Central Google Scholar
Matharu, N. et al. CRISPR-mediated activation of a promoter or enhancer rescues obesity caused by haploinsufficiency. Science 363, eaau0629 (2019).
Article CAS PubMed Google Scholar
Wang, X. et al. Evaluation and optimization of differentiation conditions for human primary brown adipocytes. Sci. Rep. 8, 5304 (2018).
Article PubMed PubMed Central Google Scholar
Lowell, B. B. & Flier, J. S. Brown adipose tissue, β3-adrenergic receptors, and obesity. Annu. Rev. Med. 48, 307–316 (1997).
Article CAS PubMed Google Scholar
Sun, X. et al. Mirabegron displays anticancer effects by globally browning adipose tissues. Nat. Commun. 14, 7610 (2023).
Article CAS PubMed PubMed Central Google Scholar
Chen, J., Guo, Z., Tian, H. & Chen, X. Production and clinical development of nanoparticles for gene delivery. Mol. Ther. Methods Clin. Dev. 3, 16023 (2016).
Article PubMed PubMed Central Google Scholar
Banskota, S. et al. Engineered virus-like particles for efficient in vivo delivery of therapeutic proteins. Cell 185, 250–265.e16 (2022).
Article CAS PubMed PubMed Central Google Scholar
Tsagkaraki, E. et al. CRISPR-enhanced human adipocyte browning as cell therapy for metabolic disease. Nat. Commun. 12, 6931 (2021).
Article CAS PubMed PubMed Central Google Scholar
Steele, C. B. et al. Vital signs: trends in incidence of cancers associated with overweight and obesity—United States, 2005–2014. MMWR Morb. Mortal Wkly Rep. 66, 1052–1058 (2017).
Article PubMed PubMed Central Google Scholar
Renehan, A. G., Zwahlen, M. & Egger, M. Adiposity and cancer risk: new mechanistic insights from epidemiology. Nat. Rev. Cancer 15, 484–498 (2015).
Article CAS PubMed Google Scholar
Paz-Filho, G., Lim, E. L., Wong, M. L. & Licinio, J. Associations between adipokines and obesity-related cancer. Front. Biosci. (Landmark Ed.) 16, 1634–1650 (2011).
Article CAS PubMed Google Scholar
Park, J., Morley, T. S., Kim, M., Clegg, D. J. & Scherer, P. E. Obesity and cancer—mechanisms underlying tumour progression and recurrence. Nat. Rev. Endocrinol. 10, 455–465 (2014).
Article CAS PubMed PubMed Central Google Scholar
Sirin, O. & Kolonin, M. G. Treatment of obesity as a potential complementary approach to cancer therapy. Drug Discov. Today 18, 567–573 (2013).
Article PubMed Google Scholar
Khandekar, M. J., Cohen, P. & Spiegelman, B. M. Molecular mechanisms of cancer development in obesity. Nat. Rev. Cancer 11, 886–895 (2011).
Article CAS PubMed Google Scholar
Huffman, D. M. et al. Cancer progression in the transgenic adenocarcinoma of mouse prostate mouse is related to energy balance, body mass, and body composition, but not food intake. Cancer Res. 67, 417–424 (2007).
Article CAS PubMed Google Scholar
Zhang, Y. et al. Stromal progenitor cells from endogenous adipose tissue contribute to pericytes and adipocytes that populate the tumor microenvironment. Cancer Res. 72, 5198–5208 (2012).
Article CAS PubMed Google Scholar
Wang, Y. Y. et al. Mammary adipocytes stimulate breast cancer invasion through metabolic remodeling of tumor cells. JCI Insight 2, e87489 (2017).
Article PubMed PubMed Central Google Scholar
Rohrig, F. & Schulze, A. The multifaceted roles of fatty acid synthesis in cancer. Nat. Rev. Cancer 16, 732–749 (2016).
Article PubMed Google Scholar
Beloribi-Djefaflia, S., Vasseur, S. & Guillaumond, F. Lipid metabolic reprogramming in cancer cells. Oncogenesis 5, e189 (2016).
Article CAS PubMed PubMed Central Google Scholar
Ye, H. et al. Leukemic stem cells evade chemotherapy by metabolic adaptation to an adipose tissue niche. Cell Stem Cell 19, 23–37 (2016).
Article CAS PubMed PubMed Central Google Scholar
Cypess, A. M. Reassessing human adipose tissue. N. Engl. J. Med. 386, 768–779 (2022).
Article CAS PubMed Google Scholar
Bashor, C. J., Hilton, I. B., Bandukwala, H., Smith, D. M. & Veiseh, O. Engineering the next generation of cell-based therapeutics. Nat. Rev. Drug Discov. 21, 655–675 (2022).
Article CAS PubMed PubMed Central Google Scholar
Chendke, G. S. et al. Replenishable prevascularized cell encapsulation devices increase graft survival and function in the subcutaneous space. Bioeng. Transl. Med. 8, e10520 (2023).
Article CAS PubMed PubMed Central Google Scholar
Kharbikar, B. N., Zhong, J. X., Cuylear, D. L., Perez, C. A. & Desai, T. A. Theranostic biomaterials for tissue engineering. Curr. Opin. Biomed. Eng. 19, 100299 (2021).
Article CAS PubMed PubMed Central Google Scholar
Herberts, C. A., Kwa, M. S. & Hermsen, H. P. Risk factors in the development of stem cell therapy. J. Transl. Med. 9, 29 (2011).
Article PubMed PubMed Central Google Scholar
Buitinga, M. et al. Micro-fabricated scaffolds lead to efficient remission of diabetes in mice. Biomaterials 135, 10–22 (2017).
Article CAS PubMed Google Scholar
Carpenter, R. et al. Scaffold-assisted ectopic transplantation of internal organs and patient-derived tumors. ACS Biomater. Sci. Eng. 5, 6667–6678 (2019).
Article CAS PubMed PubMed Central Google Scholar
Cordeiro, P. G. Breast reconstruction after surgery for breast cancer. N. Engl. J. Med. 359, 1590–1601 (2008).
Article CAS PubMed Google Scholar
Juntunen, M. et al. Evaluation of the effect of donor weight on adipose stromal/stem cell characteristics by using weight-discordant monozygotic twin pairs. Stem Cell Res. Ther. 12, 516 (2021).
Article CAS PubMed PubMed Central Google Scholar
Chu, D. T. et al. Adipose tissue stem cells for therapy: an update on the progress of isolation, culture, storage, and clinical application. J. Clin. Med. 8, 917 (2019).
Article CAS PubMed PubMed Central Google Scholar
Doench, J. G. et al. Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR–Cas9. Nat. Biotechnol. 34, 184–191 (2016).
Article CAS PubMed PubMed Central Google Scholar
Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012).
Article CAS PubMed Google Scholar
Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).
Article CAS PubMed Google Scholar
Law, C. W., Chen, Y., Shi, W. & Smyth, G. K. voom: precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biol. 15, R29 (2014).
Article PubMed PubMed Central Google Scholar
H. P. Nguyen, et al. Project ID GSE246231. Gene Expression Omnibus https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE246231 (2024).
Qiu, W. & Su, G. H. Development of orthotopic pancreatic tumor mouse models. Methods Mol. Biol. 980, 215–223 (2013).
Article CAS PubMed PubMed Central Google Scholar
Rosenbluth, J. M. et al. Organoid cultures from normal and cancer-prone human breast tissues preserve complex epithelial lineages. Nat. Commun. 11, 1711 (2020).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

This paper is dedicated to Dr. Hai P. Nguyen, an outstanding scientist, colleague and friend who left this world prematurely. We thank the Shkedy sisters (Nurit Ahituv, Liora Parzelina), Felix Feng and Rachel Chan, Ethan Freed, Cameron Pitzak and the Las Lomas High School cancer club for inspiring this study. We thank Hei Sook Sul for her support and kindly providing the human preadipocyte cell line, the UC Davis West Coast Metabolomics Center for metabolomic assays and the patients who provided clinical material. We would also like to thank Coline Arnould, Felix Feng and Alan Ashworth from UCSF for their support and advice. This work was funded in part by the UCSF Sandler Program for Breakthrough Biomedical Research, which is partially funded by the Sandler Foundation (to J.M.R. and N.A.), Susan G. Komen Foundation (J.M.R.), METAvivor (J.M.R.), DF/HCC Breast SPORE: Specialized Program of Research Excellence (SPORE), a National Cancer Institute (NCI) funded program, Grant 1P50CA168504, UCSF Benioff Initiative for Prostate Cancer Research (N.A.), National Institute of Diabetes and Digestive and Kidney Disease 1R01DK124769 (N.A.), the UCSF Living Therapeutics Initiative (N.A.), NCI grant number 1R01CA283826 (J.M.R., N.A.), University of California Office of the President (UCOP) California Breast Cancer Research Program (CBCRP) IDEA grant (M.J.M.M.) and NCI grant number R01CA292019 (M.J.M.M., J.M.R.), and the California Institute for Regenerative Medicine (CIRM) postdoctoral fellowship (H.P.N., S.L.F.), CIRM Bridges Science Master’s Fellowship (SFSU EDUC2-12693) (B.P.) and the National Human Genome Research Institute 1K99HG012576 (A.U.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Author information

Author notes

Deceased: Hai P. Nguyen

Authors and Affiliations

Department of Bioengineering and Therapeutic Sciences, University of California San Francisco, San Francisco, CA, USA
Hai P. Nguyen, Kelly An, Yusuke Ito, Bhushan N. Kharbikar, Rory Sheng, Breanna Paredes, Elizabeth Murray, Kimberly Pham, Xujia Zhou, Cassandra Biellak, Aki Ushiki, Mai Nobuhara, Sarah L. Fong, Daniel A. Bernards, Tejal A. Desai, Sook Wah Yee & Nadav Ahituv
Institute for Human Genetics, University of California San Francisco, San Francisco, CA, USA
Hai P. Nguyen, Kelly An, Yusuke Ito, Rory Sheng, Breanna Paredes, Elizabeth Murray, Kimberly Pham, Xujia Zhou, Cassandra Biellak, Aki Ushiki, Mai Nobuhara, Sarah L. Fong & Nadav Ahituv
Department of Nutritional Sciences, University of Texas at Austin, Austin, TX, USA
Hai P. Nguyen & Jace Anton Klein
Division of Hematology/Oncology, Department of Medicine, University of California San Francisco, San Francisco, CA, USA
Michael Bruck, Laura A. Huppert & Jennifer M. Rosenbluth
Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, USA
Filipa Lynce
Harvard Medical School, Boston, MA, USA
Filipa Lynce
Department of Pathology, Brigham and Women’s Hospital, Boston, MA, USA
Deborah A. Dillon
Department of Laboratory Medicine, University of California San Francisco, San Francisco, CA, USA
Mark Jesus M. Magbanua
Department of Pharmaceutical Chemistry, University of California San Francisco, San Francisco, CA, USA
Heinz Hammerlindl
University of California Davis West Coast Metabolomics Center, Davis, CA, USA
Luis Valdiviez & Oliver Fiehn
Department of Surgery, University of California San Francisco, San Francisco, CA, USA
Laura Esserman
School of Engineering, Brown University, Providence, RI, USA
Tejal A. Desai
Chan Zuckerberg Biohub, San Francisco, CA, USA
Jennifer M. Rosenbluth

Authors

Hai P. Nguyen
Kelly An
Yusuke Ito
Bhushan N. Kharbikar
Rory Sheng
Breanna Paredes
Elizabeth Murray
Kimberly Pham
Michael Bruck
Xujia Zhou
Cassandra Biellak
Aki Ushiki
Mai Nobuhara
Sarah L. Fong
Daniel A. Bernards
Filipa Lynce
Deborah A. Dillon
Mark Jesus M. Magbanua
Laura A. Huppert
Heinz Hammerlindl
Jace Anton Klein
Luis Valdiviez
Oliver Fiehn
Laura Esserman
Tejal A. Desai
Sook Wah Yee
Jennifer M. Rosenbluth
Nadav Ahituv

Contributions

H.P.N. and N.A. conceived the study. H.P.N., K.A., Y.I., R.S., M.B., B.P., K.P., H.H., E.M., X.Z., C.B., A.U., J.A.K. and S.W.Y. devised the methodology. J.M.R., F.L., D.A.D., M.J.M.M., L.A.H. and L.E. collected and worked with the human samples. B.N.K., D.A.B. and T.A.D. generated and worked on the integrated cell-scaffold delivery platform device. L.V. and O.F. performed the metabolomics analysis. H.P.N., M.N., S.L.F. and N.A. visualized the study. N.A. supervised the project. N.A. and J.M.R. acquired funding. H.P.N. and N.A. wrote, reviewed and edited the paper with contributions from all authors.

Corresponding author

Correspondence to Nadav Ahituv.

Ethics declarations

Competing interests

N.A. is a cofounder and on the scientific advisory board of Regel Therapeutics. N.A. receives funding from BioMarin Pharmaceutical Incorporate. H.P.N. and N.A. have filed a patent application covering embodiments and concepts disclosed in the paper. The other authors declare no competing interests.

Peer review

Peer review information

Nature Biotechnology thanks the anonymous reviewers for their contribution to the peer review of this work

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 CRISPRa upregulation of UCP1, PPARGC1A, and PRDM16 in human adipocytes suppresses cancer progression in various cancer types.

a, qRT-PCR of UCP1, PPARGC1A, and PRDM16 in white adipocytes transfected with CRISPRa using five different gRNAs per gene (n = 2 biological replicates). b, qRT-PCR of UCP1, PPARGC1A, and PRDM16 in human adipocytes transduced by AAV9-CRISPRa with the top two gRNAs per gene (n = 4 biological replicates). c, qRT-PCR of UCP1, PPARGC1A, and PRDM16 in human white adipocytes transduced with CRISPRa targeting UCP1, PPARGC1A, and PRDM16 (n = 8–10 biological replicates). d, qRT-PCR of TFAM, DIO2, CPT1b, and NRF1 in CRISPRa-modulated adipocytes (n = 8–10 biological replicates). e, Oxygen consumption rate (OCR) of CRISPRa-modulated adipocytes measured by the seahorse assay (n = 6 biological replicates). Uncoupled OCR was measured under oligomycin treatment, while maximal OCR was measured under FCCP. f, Glucose uptake of CRISPRa-modulated adipocytes with or without insulin (n = 3 biological replicates). g, OCR of CRISPRa-modulated cells measured by the seahorse assay in BSA- or BSA-Palmitate- medium (n = 5 biological replicates). h, Exogenous fatty acid oxidation of CRISPRa-modulated adipocytes calculated by the difference of area under the curve of OCR between BSA- and BSA-Palmitate media upon FCCP treatment (n = 4-5 biological replicates). i, Fatty acid (FA) uptake of control and UCP1-CRISPRa-modulated adipocytes (n = 4 biological replicates). j, BrdU incorporation of five cancers cell lines co-cultured with CRISPRa-modulated adipocytes (n = 4-5 biological replicates). k, OCR under FCCP of cancer cells co-cultured with CRISPRa-modulated cells measured by the seahorse assay in BSA- medium (n = 4-5 biological replicates). l, qRT-PCR (n= 3 biological replicates) and immunoblotting (n = 2 biological replicates; uncropped images in Source Data Extended Data Fig. 1) for UCP1 in primary adipocytes transduced with UCP1-CRISPRa. m, Cell numbers of various cancer cells co-cultured with CRISPRa-modulated primary adipocytes (n = 5-6 biological replicates). n, Cell viability assay on MCF-7 cells treated with CRISPRa-UCP1-AAV adipocytes (AMT), 6-aminonicotinamide (6-AN; purple bars) at 50 μM, 100 μM and DMSO as negative control and Etomoxir (ETO; yellow bars) at100 μM, 200 μM and DMSO as negative control (n = 4 biological replicates). All statistical tests in d,e,f,j,k were carried out using a one-way ANOVA and the rest using a two tailed t-test. Data are represented as mean ± standard deviation (SD) *≤0.05, **≤0.01, ***≤0.001.

Source data

Extended Data Fig. 2 CRISPRa upregulation of UCP1, PPARGC1A, and PRDM16 in human adipose organoids and proliferation and metabolic gene expression in tumors co-transplanted with CRISPRa-modulated adipose organoids.

a, Representative images of human adipose organoids transduced with UCP1, PPARGC1A, and PRDM16 AAV9 CRISPRa showing mCherry expression. White scale bar on lower left depicts 176.6μm. b, qRT-PCR of FABP4, PLIN1, and ADIPOQ in adipose organoids (n = 4 biological replicates). c, qRT-PCR of UCP1, PPARGC1A, and PRDM16 in human adipose organoids transduced by AAV9 CRISPRa (n = 5-7 biological replicates). d, Body weight of mice co-transplanted with UCP1-modulated adipose organoids and cancer cells (n = 8 mice). e, Immunofluorescence of LipidTox-stained-adipose organoids after implantation near MCF-7 xenografts. White scale bar on lower right bar depicts 100μm. f, qRT-PCR of UCP1, GLUT4, PPARGC1A, TFAM, DIO2, and CPT1B of adipose organoids after implantation near MCF-7 xenografts (n = 8 biological replicates). g, qRT-PCR of MKI67 in xenograft tumors derived from MCF-7, MDA-MB-436, Panc 10.05, and DU-145 cancer cells co-transplanted with UCP1-CRISPRa modulated adipose organoids (n = 6–8 biological replicates). h, qRT-PCR of GLUT4, GCK, CD36, and CPT1B in xenograft tumors (n = 6–8 biological replicates). i, Representative images and quantification of immunofluorescence staining for Caspase 3 in tumors co-implanted with UCP1-CRISPRa adipose organoids (n = 6 cryosections). White scale bar on lower right bar depicts 10μm. j, Representative images, volume, qRT-PCR of proliferation maker, MKI67 and metabolic genes, including GLUT4, GCK, CD36, and CPT1b of MCF tumors co-implanted with primary adipose organoids (N = 3 mice). All statistical tests were carried out using a two tailed t-test and data are represented as mean ± standard deviation (SD) *≤0.05, **≤0.01, ***≤0.001.

Extended Data Fig. 3 UCP1-CRISPRa human adipose organoids increased whole-body energy expenditure, glucose tolerance, and insulin sensitivity.

a, Whole-body oxygen consumption at 30 °C, 16 °C, and 4 °C during day and night time of mice implanted with UCP1-CRISPRa modulated adipose organoids (n = 4 mice). b, Glucose tolerance test and insulin tolerance test (n = 4 mice). c, Plasma insulin levels of wild type SCID mice (WT) and mice that were co-transplanted with dCas9-VP64 (control) and UCP1-CRISPRa adipose organoids and xenografts (n = 6–8 mice). d, tumor size of MCF7 tumor xenograft and UCP1-CRISPRa treated human adipose organoids in immune-deficient SCID mice fed with standard chow, high-fat diet (HFD), and 15% glucose containing water during 9 weeks (n = 4-5 mice). e, Immunofluorescence images from cryosections of xenograft tumors (n = 5 sections per treatment) of Ki67, CA9, and CD31. White scale bar on lower right bar depicts 20μm. f, Principle component analysis (PCA) of MCF7 tumors that were co-implanted with UCP1-CRISPRa treated human adipose organoids in mice fed with standard chow (left) high-fat diet (middle) or 15% glucose (right); (N = 3). g,Volcano plots showing p-value versus fold change of MCF-7 tumors that were co-implanted with UCP1-CRISPRa treated human adipose organoids in mice fed with high-fat diet or 15% glucose. No differentially expressed genes were identified, defined as those exhibiting at least a +/− 4.0 fold change, with their expression being significantly different from the basal level (false discovery rate (FDR) adjusted, p < 0.01). All statistical tests in a-d were carried out using a two tailed t-test and data are represented as mean ± standard deviation (SD) *≤0.05, **≤0.01, ***≤0.001.

Extended Data Fig. 4 CRISPRa upregulation of Ucp1 in mouse white adipocytes and phenotypic analysis of mouse genetic cancer models.

a, qRT-PCR of Ucp1, in white adipocytes transfected with CRISPRa using five different gRNAs (n = 4–6 biological replicates). b, qRT-PCR of Ucp1 in mouse adipocytes transduced by AAV9-CRISPRa with two gRNAs (n = 5-6 biological replicates). c, qRT-PCR of mCherry and Ucp1 in mouse adipose organoids (n = 5 biological replicates). d, Body weight of pancreatic cancer KPC mice implanted with control (dCas9-VP64) or Ucp1-upregulated mouse adipose organoids (n = 5-6 mice). e, Immunofluorescence staining of CK19, Ki67, CA9, and CD31 in cryosections of pancreatic tumors (N = 6 sections). White scale bar on lower right bar depicts 20μm. f, Representative images and quantification of Caspase3 in cryosections of pancreatic tumors (N = 6 sections). White scale bar on lower right bar depicts 10μm. g, Plasma insulin levels of pancreatic cancer genetic mice implanted with either Ucp1-CRISPRa adipose organoids or control organoids (n = 4 mice). h, Body weight of breast cancer (MMTV-PyMT) mice implanted with control (dCas9-VP64) or Ucp1-CRISPRa mouse adipose organoids nearby mammary gland (MG) or distal (DOR)(n = 4 mice). i, Plasma insulin levels of breast cancer genetic mice implanted with either Ucp1-CRISPRa adipose organoids or control organoids nearby mammary gland (MG) or distal (DOR)(n = 4 mice). All statistical tests were carried out using a two tailed t-test and data are represented as mean ± standard deviation (SD) *≤0.05, **≤0.01, ***≤0.001.

Extended Data Fig. 5 CRISPRa upregulation of UCP1 in human mammary gland adipocytes.

a, Representative images of adipocytes isolated from mammary glands that were transduced with UCP1-CRISPRa AAV9. Scale bar on lower left depicts 265.2μm. b, qRT-PCR of mcherry and UCP1 in CRISPRa-modulated mammary gland adipocytes (n = 5 biological replicates). c, Glucose uptake and fatty acid uptake of UCP1-CRISPRa adipocytes of sample TOR40 (n = 3 biological replicates). d, Bright-field images of adipose organoids that were co-cultured with UCP1-CRISPRa modulated mammary gland adipocytes. Scale bars at bottom left depicts 265.2μm (TOR40, TOR120, TOR127) and 530.4μm (TOR41, TOR124). All statistical tests were carried out using a two tailed t-test and data are represented as mean ± standard deviation (SD) **≤0.01, ***≤0.001.

Extended Data Fig. 6 Inducible CRISPRa-modulated adipose organoids and CRISPRa-upregulation of UPP1 in adipocytes to suppress cancer growth of pancreatic ductal adenocarcinoma.

a, qRT-PCR of Cas9 and UCP1 in adipocytes transduced with inducible CRISPRa AAV and treated with either DMSO or doxycycline (Dox) (n = 3 biological replicates). b, Electron microscopy image of adipose organoids implanted in a microwell scaffold showing the organoid attached to the scaffold with filopodia and lamellipodia (1.0 kV, 1500x, and 7.861 mm lens). c, qRT-PCR of UPP1 in white adipocytes transfected with CRISPRa using five different gRNAs (left) and AAV transduced (right) with two gRNAs targeting the UPP1 promoter (n = 3 biological replicates). d, Uridine uptake after 1 or 12 hours measured with incubating UPP1-CRISPRa modulated adipocytes with 3H-uridine (n = 3 biological replicates). e, Lactate levels of UPP1-CRISPRa modulated adipocytes (n = 4 biological replicates). f, Representative images, cell numbers/view, and ATP levels of PANC-1 cells that were co-cultured with UPP1-modulated adipocytes or control (dCas9-VP64 only) adipocytes in high-glucose media (n = 5 biological replicates). White scale bar on lower right bar depicts 100μm. All statistical tests were carried out using a two tailed t-test and data are represented as mean ± standard deviation (SD) *≤0.05, **≤0.01, ***≤0.001.

Supplementary information

Source data

Source Data Extended Data Fig. 1

Source Data Fig. 1. Raw data for western blots. a,b, Uncropped images of the western blot displayed in Extended Data Fig. 1l. The same blot was cut and stained on the left for GAPDH (a) and UCP1 on the right (b) because both proteins were close in size (36 kDa for GAPDH and 33 kDa for UCP1). GAPDH was used as a loading marker because of its use also in RT–qPCR. A total of 10 µg per lane GAPDH, 40 µg per lane dCas9–VP64 only (VP64 (neg.)), 40 µg per lane dCas9–VP64 + UCP1-sgRNA (UCP1-CRISPRa) and 10 µg per lane UCP1 overexpression were loaded in their respective lanes. c, Additional western blot using the same samples in a and b showing 40 µg per lane dCas9–VP64 only (VP64 (neg.)), 40 µg per lane dCas9–VP64 + UCP1-sgRNA (UCP1-CRISPRa) and 10 µg per lane UCP1 overexpression. The PageRuler Plus Protein Ladder (Thermo Fisher) was used as a marker in all western blots.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Nguyen, H.P., An, K., Ito, Y. et al. Implantation of engineered adipocytes suppresses tumor progression in cancer models. Nat Biotechnol 43, 1979–1995 (2025). https://doi.org/10.1038/s41587-024-02551-2

Download citation

Received: 19 November 2024
Accepted: 19 December 2024
Published: 04 February 2025
Version of record: 04 February 2025
Issue date: December 2025
DOI: https://doi.org/10.1038/s41587-024-02551-2