To study the impact of aging on beige adipose tissue development, we exposed young (9-week-old) and middle aged (57-week-old) C57BL/6 mice to 6 °C for either 3 or 14 days. All mouse groups were first acclimated to 30 °C (thermoneutrality [TN]) for 3 weeks to reduce beige adipose tissue to baseline levels. Following acclimation, TN-housed mice remained at 30 °C; acute cold mice (3D) were transitioned to 6 °C after 11 days for the final 3 days; and chronic cold mice (14D) were moved to 6 °C for 2 weeks (Figure 1A). As expected, the aged mice weighed more and had larger iWAT depots than the young mice (Figure 1—figure supplement 1A-B). Cold exposure progressively increased the expression levels of thermogenic genes Ucp1, Cidea, Dio2, and Ppargc1a in iWAT from young mice, and the activation of these genes was significantly blunted in aged mice, especially at the 3D time point (Figure 1B). Immunofluorescence (IF) staining showed a robust induction of UCP1 protein in multilocular adipocytes of young iWAT at 3D of cold exposure, which was further increased at 14D. The induction of UCP1⁺ beige adipocytes at 3D was severely reduced in aged animals, with few UCP1 +adipocytes detected. At 14D, the beige adipocytes were morphologically similar in young and aged mice, although there were many fewer in aged animals (Figure 1C). At both ages, beige adipocytes were more prominent in the inguinal versus dorsolumbar region of iWAT, consistent with other reports (Barreau et al., 2016; Chi et al., 2018; Dichamp et al., 2019), and beiging was largely absent in the dorsolumbar region of aged mice (Figure 1—figure supplement 1C–F). To determine if the beiging response was delayed in aged mice, we exposed young and aged mice at 6 °C for 6 weeks. At this time point, the iWAT of aged mice exhibited a larger deficit in thermogenic gene expression compared to young animals (Figure 1D). Thermogenic gene levels in interscapular BAT were similar between young and aged mice at TN and after cold exposure, indicating that the inhibitory effects of aging were selective to WAT (Figure 1—figure supplement 1E).

Next, we examined beige fat formation in young and aged animals upon treatment with the β3-selective adrenergic agonist CL-316,243 (CL). CL acts in an adipose tissue autonomous manner to stimulate beige fat biogenesis, bypassing the central nervous system pathways that mediate the cold response. Acute CL treatment for only 1 hr increased Ucp1 expression in in iWAT of young mice to a much greater extent than in aged mice (Figure 1E). Chronic CL exposure for 5 days also induced much higher expression levels of Ucp1 and Cidea in iWAT of young compared to aged mice (Figure 1F). Taken together, these results demonstrate that beige adipose tissue induction is severely impaired in middle aged mice.

To determine the contribution of fibroblastic ASPCs to beige adipocytes during cold exposure, we performed lineage tracing using Pdgfra-Cre^ERT2; R26R^tdTomato reporter mice. Pdgfra expression marks multiple ASPC populations, including preadipocytes (Merrick et al., 2019; Sakers et al., 2022). Young and aged reporter mice were treated with tamoxifen for 5 days at TN (30 °C; ‘pulse’) to activate Cre and induce tdTomato expression in Pdgfra⁺ cells. Following a 9-day washout period, mice were transferred to 6 °C (cold) for 2 weeks (‘chase’; Figure 2A). We observed near complete and specific labeling of ASPCs during the pulse period, with ~95% of PDGFRα⁺ cells in iWAT from young and aged mice displaying tdTomato expression (Figure 2B, Figure 2—figure supplement 1). The proportion of PDGFRα+ cells in iWAT was similar between young and aged mice (Figure 2B). No tdTomato-expressing adipocytes were observed after the pulse (Figure 2—figure supplement 1). After 14 days of cold exposure, we detected many newly developed beige adipocytes from ASPCs in young mice (visible as tdTomato⁺/UCP1⁺ multilocular adipocytes). By contrast, very few ASPC-derived (tdTomato⁺) adipocytes were detected in the beige fat areas of aged iWAT at day 14 (Figure 2C). Quantifying across the entire length of iWAT pads revealed that most beige adipogenesis occurred in the inguinal region and was ~12-fold lower in aged compared to young mice (Figure 2D and E). However, the overall contribution of Pdgfra⁺ ASPCs to beige adipocytes was relatively low, even in young animals, with <20% of beige adipocytes expressing tdTomato.

We previously identified three main fibroblastic ASPC populations in iWAT: DPP4⁺ cells, ICAM1⁺ preadipocytes, and CD142⁺ cells. All these cell types express Pdgfra and have the capacity to undergo adipogenic differentiation (Merrick et al., 2019). To test whether aging dysregulates one or more of these ASPC types, we performed scRNA-seq on stromal vascular cells from iWAT of young and aged animals, maintained at TN, or following transition to cold for 3 or 14 days (Figure 1A). ASPCs were enriched by removing immune (CD45⁺) cells using fluorescence activated cell sorting (FACS). We integrated the datasets from all conditions together and performed clustering analysis. The following cell populations were annotated based on their expression of cell-type-specific marker genes: four fibroblast populations (Dpp4⁺; Icam1⁺ preadipocytes; Cd142⁺, Spp1⁺), two populations of endothelial cells (Pecam1⁺); smooth muscle cells/pericytes (Myh11⁺, Pdgfrb⁺); Schwann cells (Mpz⁺); and residual immune cells (Ptprc⁺; Figure 3A–C). We did not identify any cell population specific to either aging or cold exposure. In this regard, we did not identify ‘aging-dependent regulatory cells (ARCs)’, which were previously defined as ASPCs expressing Lgals3 and other inflammatory genes (Figure 3—figure supplement 1; Nguyen et al., 2021). The expression levels of identity markers of the ASPC populations were not modulated during cold exposure or aging (Figure 3—figure supplement 1).

Differential gene expression analyses identified aging-modulated genes in ASPCs (Figure 3D). Notably, expression of Cd9, previously identified as a fibrogenic marker, was upregulated with age in Dpp4⁺ cells and preadipocytes (Marcelin et al., 2017). Pltp and Gpnmb were also elevated by aging across all ASPC populations and temperature conditions. Genes downregulated by aging in all ASPC populations included Meg3, Itm2a and Gpc3 and Postn. Of note, Postn encodes an extracellular matrix protein that was previously reported to regulate adipose tissue expansion and decrease in expression during aging (Graja et al., 2018).

We next evaluated if ASPCs from young and aged animals exhibit cell-autonomous differences in adipogenic differentiation capacity. We FACS-purified DPP4⁺, ICAM1⁺, and CD142⁺ cells from the iWAT of young and aged mice, plated them in culture and induced adipocyte differentiation. Using a minimal differentiation stimulus consisting of insulin only (Min), ICAM1⁺ and CD142⁺ cells underwent more efficient differentiation into lipid droplet-containing adipocytes, and expressed higher levels of adipocyte genes (Adipoq and Fabp4) than DPP4⁺ cells, consistent with prior work (Figure 4A and B; Merrick et al., 2019). DPP4⁺ and CD142⁺ cells from young and aged mice underwent adipocyte differentiation and induced adipocyte genes with equivalent efficiency. Unexpectedly, ICAM1⁺ cells from aged mice exhibited greater differentiation capacity than those from young mice, as evidenced by higher expression levels of Adipoq and Fabp4 (Figure 4A and B). Maximal stimulation with a full cocktail of adipogenic inducers (Max) produced similar and robust differentiation in all ASPC populations from young or aged mice (Figure 4C and D). To assess whether young and aged ASPCs behave differently when cultured as a mixed heterogeneous population, we isolated the stromal vascular fraction (SVF) for adipogenesis assays. Again, SVF cell cultures from young and aged mice displayed similar adipogenic differentiation efficiency following either Min or Max stimulation (Figure 4E and F). Finally, we treated differentiated adipocyte cultures with the pan-adrenergic agonist isoproterenol for 4 hours to evaluate thermogenic gene activation (i.e. beiging). Basal levels of Ucp1 expression were lower in DPP4⁺ cells compared to other ASPC types, but all ASPC populations activated Ucp1 expression to similarly high levels in response to isoproterenol treatment and did not differ by age (Figure 4G). We also did not observe an aging-related difference in the levels of Ucp1 induction in SVF-derived adipocyte cultures (Figure 4H). Together, these data suggest that the beige adipogenic capacity of ASPCs is not intrinsically compromised in aged mice, and therefore the in vivo deficit in beige adipogenesis could be due to non-ASPC-autonomous effects.

To determine the effects of aging and cold exposure on adipocyte gene profiles, we performed snRNA-seq analyses of iWAT samples using the same experimental paradigm described above (Figure 1A). We integrated all the conditions together for analyses from two separate runs (Figure 5A and C). Similar cell types were captured as with scRNA-seq (Figure 3A), but with the addition of mature adipocyte populations. This dataset also has increased representation from immune cells since there was no negative selection against CD45⁺ cells. As with the single-cell data set, we did not identify any aging-specific cell populations (Figure 5—figure supplement 1). However, we observed striking gene expression differences in the adipocyte cluster across age and temperature. Most obvious, and expectedly, was the emergence and expansion of a distinct beige adipocyte population, marked by expression of Ucp1 and other thermogenic genes, during cold exposure (Figure 5B).

To focus on adipocyte responses, we reintegrated the snRNA-seq data using only the adipocytes, which revealed four main clusters (Figure 5D–F). All adipocyte clusters displayed similarly high mRNA levels of canonical adipocyte markers Fabp4 and Plin1. Beige adipocytes, marked by high expression of thermogenic genes (i.e. Ppargc1a, Esrrg, Cidea, Gk, Prdm16, and Ucp1), were the most distinctive cluster and were largely absent at TN in young and aged mice. These cells began to appear in young mice after 3 days of cold exposure, and were further increased at 14 days. By contrast, in aged mice, beige cells were barely detectable at 3 days of cold exposure and were present at greatly reduced numbers than in young mice at 14 days (Figure 5E). This analysis also revealed three sub-populations of ‘white’ adipocytes. ‘Npr3-high’ adipocytes were enriched for expression of Npr3, Synpo2, Prr16, and Tshr, expressed higher levels of white fat marker genes Leptin (Lep) and Nnat, and exhibited the lowest expression levels of thermogenic (beige) genes (Gesta et al., 2007; Rosell et al., 2014). Two additional white adipocyte clusters were designated as ‘de novo lipogenesis (DNL)-low’ and ‘DNL-high’ cells, both of which expressed lower levels of Npr3 and shared selective expression of Fgf14. DNL-high cells uniquely expressed Ces1f and Gsta3 and activated high levels of DNL pathway genes (i.e. Fasn, Acss2 and Acly) upon cold exposure (Figure 5F). Interestingly, Adiponectin (Adipoq) was differentially expressed across adipocyte clusters, with higher levels in Npr3-high and DNL-high cells. Quantification of adipocyte nuclei from this data set suggested that the proportions of Npr3-high and DNL-high adipocytes remain stable across temperature, with aged mice having more Npr3-high adipocytes. The proportion of beige adipocytes increased during cold exposure, while DNL-low adipocytes decreased with cold exposure in both young and aged mice (Figure 5G).

To evaluate the global effects of cold exposure and aging on adipocytes, we performed differential gene expression analysis between young and aged adipocytes within each cluster. DNL-high and beige adipocytes exhibited the most dramatic expression changes between young and aged animals (Figure 6A and B, Figure 6—figure supplement 1A and B). At TN, DNL-high cells from aged animals expressed lower levels of several genes, including Fkbp5, Spon1, and Adam12. Interestingly, Npr3, in addition to marking Npr3-high cells, was increased by aging in DNL-high adipocytes and to a lesser extent in other adipocyte populations (Figure 6C and D). In young animals, Npr3 expression was downregulated by cold exposure in the three white adipocyte populations, and this downregulation was blunted in aged animals (Figure 6D). Gene expression analysis of whole iWAT pads confirmed that Npr3 mRNA levels were progressively decreased by cold exposure and elevated in aged versus young mice under all temperature conditions (Figure 6E). Npr3 expression levels were also increased in isolated primary adipocytes from aged relative to young mice (Figure 6F). Expression levels of the G-protein-coupled NP receptors Npr1 or Npr2 were not modulated by cold or aging in iWAT or iWAT adipocytes (Figure 6—figure supplement 1).

We also observed a striking activation of the DNL gene program (Acly, Fasn, Acaca, Scd1, etc.) in DNL-high and beige adipocytes during cold exposure (Figure 6G and H). The induction of these genes during cold exposure, exemplified by Acly expression, was a cluster-defining attribute of DNL-high cells, which did not express beige markers like Ucp1 even after 14 days of cold exposure. Of note, we found two types of beige (Ucp1⁺) adipocytes, distinguished by the presence vs. absence of high DNL gene levels (i.e. Ucp1⁺; DNL⁺ and Ucp1⁺; DNL(^-)), with the latter arising first during cold exposure (3D vs. 14D) (Figure 6G, Figure 6—figure supplement 1E, F). Importantly, the induction of DNL genes was nearly completely blocked in DNL-high cells and reduced in beige cells of aged animals (Figure 6G). Indeed, the top aging downregulated genes in adipocytes from cold exposed mice correspond to DNL and related pathways, especially in DNL-high cells (Figure 6—figure supplement 1G). Lastly, at the whole tissue level, we observed robust induction of Acly in iWAT of young relative to aged mice with increasing duration of cold exposure (Figure 6—figure supplement 1H). Taken together, these results implicate the suppression of natriuretic peptide signaling and DNL in the aging-related impairment of beige fat formation.

All animal procedures were approved and performed under the guidance of the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC) (protocol #805649). Young (4 weeks) and aged (52 weeks) C57BL/6 male mice were obtained from the National Institute of Aging (C57BL/6JN) or Jackson Laboratories (C57BL/6 J, stock number 000664). Mice were housed at 30 °C for 3 weeks, then were either: maintained at 30 °C for 2 weeks (TN); kept at 30 °C for 11 more days before moving to 6 °C for 3 days (3D cold) or moved to 6 °C for 14 days (14D cold). Mice were single housed during the final 2-week temperature treatment and provided with a nestlet and shepherd shack. For experiments with CL316,243 (CL, Sigma-C5976), mice were housed at 30 °C for 5 weeks, followed by intraperitoneal (IP) injection of 1 mg/kg/d CL either 1 hr prior to tissue harvest or for 5 days. Pdgfra^CreERT2 mice were obtained from Dr. Brigid Hogan (Duke University) (Chung et al., 2018) and crossed with Rosa26^tdTomato (strain: B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J, stock no. 007914). To induce Cre activity, tamoxifen (Sigma, T5648) dissolved in corn oil (Sigma, C8267) was injected intraperitonially (IP) into mice at a dose of 100 mg/kg/d for 5 days. For all iWAT processing other than histology, the inguinal lymph node was removed.

Tissues were fixed overnight in 4% paraformaldehyde, washed with PBS, dehydrated in ethanol, paraffin-embedded and sectioned. Following deparaffinization, slides were subjected to heat antigen retrieval in a pressure cooker with Bulls Eye Decloaking buffer (Biocare), unless otherwise noted. Slides were incubated in primary antibody overnight and secondary antibody conjugated to peroxidase and then developed using Tyramide Signal Amplification (TSA, Akoya Biosciences). Samples were stained with either hematoxylin and eosin or the following antibodies: anti-red fluorescent protein (RFP) (rabbit; 1:500; Rockland #600-401-379), anti-UCP1 (rabbit, 1:2000, AstraZeneca), and anti-PLIN1 (rabbit, 1:200 Cell Signaling #3470). Slides were imaged on an inverted fluorescence microscope (Keyence BZ-X710). For quantification of tdTomato-expressing adipocytes, full-length iWAT slices were tile imaged, stitched, exported as a BigTiff, and quantified in a blinded-manner using the Count Tool in Photoshop (Adobe).

DPP4⁺, ICAM1⁺, and CD142⁺ cells were isolated as previously described (Merrick et al., 2019). Briefly, SVCs from the subcutaneous adipose of mice (n=2–5) were pooled and resuspended in FACS buffer (HBSS containing 3% FBS; Fisher), then incubated for 1 hr at 4 °C with the following antibodies: CD26 (DPP4)-fluorescein isothiocyanate (FITC) (Biolegend, 137806; 1:200), anti-mouse ICAM1-phycoerythrin (PE)/Cy7 (Biolegend, 116122; 1:100), anti-mouse CD45-allophycocyanin (APC)/Cy7 (Biolegend, 103116; 1:1000), anti-mouse CD31-APC-Fire (Biolegend, 102528; 1:1000), and anti-mouse CD142 (Sino Biological, 50413-R001, 1:100; or R&D Systems, AF3178, 1:50). Anti-mouse CD142 antibodies were conjugated with Biotium Mix-n-Stain CF647 (Sigma, MX647S100). For lineage tracing pulse analysis, SVCs were isolated from individual mice without pooling. SVCs were stained with anti-mouse CD31, anti-mouse CD45, and anti-mouse CD140a (PDGFRΑ) (PE/Cy7) (Biolegend, 135912; 1:100). In all FACS experiments, cells were stained with 4′,6-diamidino-2-phenylindole (DAPI) (Roche, 10236276001; 1:10,000) for 5 min, then washed three times with FACS buffer to remove unbound antibodies. Cells were sorted with a BD FACS Aria cell sorter (BD Biosciences) equipped with a 100 μm nozzle and the following lasers and filters: DAPI, 405 and 450/50 nm; FITC, 488 and 515/20 nm; mTomato, 532 and 610/20 nm; PE/Cy7, 532 and 780/60 nm; CF647, 640 and 660/20 nm; and APC/Cy7 and APC-Fire, 640 and 780/60 nm. All compensation was performed at the time of acquisition in Diva software by using compensation beads (BioLegend, A10497) for single-color staining and SVCs for negative staining and fluorescence (DAPI and tdTomato).

Mouse studies were performed with >n = 5 per group for p=0.05 with 95% power given the expected variability of examined phenotypes. Each experiment was independently replicated at least twice. Sample sizes are reported in figure legends. All bar graphs represent the mean ± SEM. A Student’s t-test was used when two groups were compared. Where multiple conditions were compared, we applied two-way ANOVA with a Tukey correction for multiple comparisons. Only the Young vs. Aged comparisons were depicted on graphs for clarity, with additional multiple comparisons provided below. p Values are indicated by asterisks and defined as *p<0.05, **p<0.01 and ***p<0.001. All statistics were calculated with GraphPad Prism Version 10.0.3.