Abstract

   Aging is associated with a decline in the browning capacity of white
   adipose tissue (WAT), contributing to metabolic dysfunction. Beige
   adipocytes, which dissipate excess energy as heat, are a key feature of
   this process. In this study, we investigate the role of adipose stem
   and progenitor cells (ASPCs), specifically the Aregs (CD142+)
   subpopulation, in regulating beige adipocyte formation in aged mice
   under cold stimulation. Our findings reveal that Aregs significantly
   increase in the subcutaneous WAT (sWAT) of aged mice following cold
   exposure. We further demonstrate that Aregs secrete insulin-like growth
   factor binding protein 3 (IGFBP3), which appears to play a pivotal role
   in the cross-talk between adipogenesis-regulatory cells (Aregs) and
   smooth muscle cell-like (SMC-like) cells, thereby leading to the
   inhibition of beige adipocytes formation. Functional enrichment
   analysis highlighted the activation of TGFβ, MAPK and p53 signaling
   pathways in SMC-like cells, all of which are known to induce cell
   apoptosis and fibrosis. Moreover, IGFBP3 was found to interact with
   receptors and signaling molecules, including Egfr, Irf1 and Cdkn1a, in
   SMC-like cells, enhancing their apoptosis. Co-culture experiments
   confirmed that IGFBP3 significantly suppressed the formation of beige
   adipocytes, further corroborating its role in impairing browning.
   Overall, our study provides novel insights into the molecular
   mechanisms by which Aregs and IGFBP3 contribute to the age-related
   decline in WAT browning. These findings suggest potential therapeutic
   targets for reversing impaired WAT browning in aging and related
   metabolic disorders.

   Keywords: Aregs, IGFBP3, SMC-Like cells, Browning, Aging

Graphical abstract

   [45]Image 1
   [46]Open in a new tab

Highlights

     * •
       Aregs accumulation in aged mice under cold exposure.
     * •
       IGFBP3 secreted by Aregs induces apoptosis in SMC-like cells.
     * •
       Aregs-IGFBP3 signaling pathway impairs beige adipocyte formation.

1. Introduction

   Adipose tissue plays a pivotal role not only in energy storage but also
   as an active endocrine organ regulating systemic metabolism. Among its
   subtypes, brown adipose tissue (BAT) and beige adipose tissue have
   garnered attention for their unique ability to dissipate energy via
   thermogenesis, mediated by mitochondrial uncoupling through uncoupling
   protein 1 (UCP1) [[47]1,[48]2]. Beige adipose tissue, in particular,
   has drawn significant interest due to its remarkable plasticity. It can
   transdifferentiate from widespread white adipose tissue (WAT) or de
   novo differentiate from beige progenitor cells scattered within WAT
   under specific stimuli, such as classical cold exposure. De novo
   differentiated beige progenitor cells exhibit smooth muscle-like
   (SMC-like) characteristics [[49][3], [50][4], [51][5]]. This entire
   process, referred to as “browning” [[52]6,[53]7], holds great potential
   for combating obesity, cardiovascular diseases (CVD), and other
   metabolic disorders [[54][8], [55][9], [56][10]].

   However, the ability of adipose tissue to undergo browning declines
   with aging. Aged individuals exhibit a reduced capacity for beige fat
   formation, which is often associated with impaired thermogenesis and
   metabolic dysfunction [[57]11,[58]12]. This decline is thought to
   contribute to the increased risk of obesity, insulin resistance, and
   CVD observed in the elderly [[59]12,[60]13]. The diminished browning
   capacity in aging fat depots has been linked to changes in the adipose
   tissue microenvironment, including altered stem cell function
   [[61]14,[62]15], the accumulation of inflammatory mediators [[63]8] and
   a reduced response to cold exposure [[64]16]. Understanding the
   cellular and molecular mechanisms behind this decline in browning
   ability could provide valuable insights into how aging exacerbates
   metabolic diseases and identify potential therapeutic targets to
   reverse these age-related changes.

   Adipogenesis-regulatory cells (Aregs), a CD142+ subpopulation of
   adipose stem and precursor cells (ASPCs), were first identified and
   molecularly characterized in 2018 by Schwalie et al. These cells are
   refractory to adipogenesis and exhibit a paracrine capacity to suppress
   adipocytes formation in vitro and in vivo, suggesting a critical role
   in regulating adipose tissue plasticity and metabolic functions
   [[65]17]. A recent study has suggested that Aregs may play a pivotal
   role in the remodeling of WAT under cold stimulation [[66]18].
   Additionally, it has been shown that the anti-adipogenic activity of
   Aregs increases with aging, which may contribute to the age-related
   decline in adipose tissue function and thermogenic capacity [[67]19].

   Insulin-like growth factor binding protein 3 (IGFBP3) has been
   identified as a critical mediator secreted by Aregs in this study. It
   is well known for regulating cell proliferation and differentiation by
   modulating the bioavailability of insulin-like growth factors (IGFs).
   Specifically, IGFBP3 binds to IGFs, inhibiting its interaction with the
   IGFs receptors, thereby reducing IGFs signaling activity [[68][20],
   [69][21], [70][22], [71][23]]. In fact, IGFBP3 is also a known
   inhibitor of adipogenesis and can induce adipocytes apoptosis [[72]24].

   These studies underscore the remarkable plasticity of Aregs in
   modulating the adipose tissue microenvironment across different ages
   and conditions. However, the roles of Aregs and IGFBP3 in the impaired
   browning capacity of aging adipose tissue remains unexplored.
   Understanding whether Aregs and IGFBP3 are involved in the diminished
   browning capacity of aged adipose tissue under cold exposure, and the
   underlying molecular mechanisms, is crucial for future research and
   warrants further investigation. In this study, we further analyze Aregs
   cells and beige adipocytes precursor SMC-like cells using publicly
   available single-cell RNA sequencing (scRNA-seq) data (GEO:
   [73]GSE227439). Through integrated analysis and experimental
   validation, we present the first detailed description of the increased
   presence of the Aregs subpopulation in aged mice under cold
   stimulation. We demonstrate that Aregs inhibit the number of SMC-like
   cells through the secretion and action of IGFBP3, leading to a novel
   mechanism of impaired browning ability in response to cold stimulation
   in aged mice.

2. Materials and methods

2.1. Animal experiments

   All animal procedures received approval from the Animal Care and Use
   Committee of Kunming Medical University (Approval No. KMMU-20230939)
   and conducted in compliance with institutional guidelines. Male C57BL/6
   mice, classified as young (4 weeks) and old (52 weeks), were obtained
   from the Experimental Animal Center of Kunming Medical University. Mice
   were housed under controlled conditions at 30 °C for 3 weeks to ensure
   acclimatization. Following this period, they were either maintained at
   thermoneutrality (30 °C) for an additional 14 days (TN group) or
   subjected to cold exposure at 6 °C for 14 days (cold group). Throughout
   the experiment, all mice were provided ad libitum access to standard
   chow and maintained under a 12 h-light/dark cycle.

2.2. Primary adipose stem and progenitor cells (ASPCs) isolation

   Murine sWAT was dissected from 9-week-old male and female mice
   separately into ice-cold PBS with 1% (v/v) penicillin-streptomycin
   solution (P/S) (Gibco, 15070063). The tissue was minced with scissors
   and digested in DMEM (Gibco, 11965092) containing 1 mg/mL collagenase
   type I at 37 °C with constant shaking at 1200 rpm for 90 min. After
   digestion, the tissue was filtered through a 70 μm nylon mesh (Absin,
   abs7232) and centrifuged at 1200 rpm for 10 min. The floating lipid
   layer was carefully removed, and the stromal vascular fraction (SVF)
   pellet was resuspended in red blood cell lysis buffer (Gibco,
   00-4333-57). The recovered cells were washed with pre-warmed PBS
   (Gibco, 10010023) containing 3% fetal bovine serum (FBS) (Gibco,
   10270-106). Finally, primary ASPCs were cultured in complete medium
   (DMEM supplemented with 10% FBS and 1% P/S) in a CO[2] incubator.

2.3. Lentiviral transduction

   We used a lentivirus-based CRISPR/Cas9 system to knockout (KO) Igfbp3
   in Aregs. The lentivirus employed two gRNAs (gRNA1:
   CTGACTTGCGCGCTGCGCGA, gRNA2: CCGCGGAGCAGTACCCGCTG) targeting the
   full-length sequence of Igfbp3. Additionally, we included a control
   group of Aregs treated with non-targeting CRISPR (KO ctrl) using a
   scramble gRNA
   (GGTGTAGTTCGACCATTCGTGGTTTTAGAGCTAGAAATAGCAAGTTAAAATAAGGCTAGTCCGTTATCAA
   CTTGAAAAAGTGGCACCGAGTCGGTGC) (Ubigene, Guangzhou). Briefly, the
   lentivirus, along with the lentiviral transduction enhancer Lentiblast
   reagent (OZ Biosciences, LBPX500), was used to infect ASPCs. 2 days
   post-infection, transduced cells were selected with puromycin
   (1 μg/mL). Surviving cells were then used for subsequent experiments.

2.4. Browning differentiation

   Fully confluent ASPCs were induced to differentiate by switching to
   differentiation medium I, which consisted of DMEM complete medium
   supplemented with 850 nM insulin (Solarbio, I8830), 0.5 mM
   isobutylmethylxanthine (Solarbio, II0010), 1 μM dexamethasone
   (Solarbio, ID0170), 125 nM indomethacin (Solarbio, SI9020), 1 μM
   rosiglitazone (Solarbio, IR0130), and 1 nM 3,3′,5-triiodo-l-thyronine
   (T3) (MCE, HY-A0070A) for 2 days. Following this, the cells were
   transferred to differentiation medium II, which contained DMEM complete
   medium supplemented with 850 nM insulin, 1 μM rosiglitazone, and 1 nM
   T3, and incubated for an additional 6 days.

2.5. RNA isolation and quantitative PCR (qPCR)

   Cells were harvested using pre-chilled TRIzol reagent (Invitrogen,
   15596018), and total RNA was extracted following the
   chloroform-isopropanol-ethanol method according to the manufacturer's
   protocol. For sWAT, samples were first minced using a TissueLyser II
   (Qiagen, 85300), and RNA extraction was performed similarly to the
   cellular RNA extraction procedure. One microgram of RNA was reverse
   transcribed into complementary DNA (cDNA) using the MightyScript Plus
   First Strand cDNA Synthesis Master Mix (Sangon Biotech, B639252). qPCR
   was conducted using SGExcel FastSYBR Mixture (Sangon Biotech, B532955)
   and the ABI PRISM™ 7900HT system (Applied Biosystems). Primer sequences
   are provided in ([74]Supplementary Table 1). Relative mRNA expression
   levels were calculated accordingly using the 2^−ΔCt method and
   normalized to GAPDH.

2.6. Immunohistochemistry fluorescence

   sWAT was carefully excised and fixed in 4% paraformaldehyde (Beyotime,
   P0099) for 18–24 h. The tissue was then immersed in phosphate-buffered
   sucrose solutions (10%, 20%, and 30%) until it sank. After removing
   excess moisture with absorbent paper, the samples were embedded in
   O.C.T. compound (Sakura Finetek, TISSUE-TEK 4583), fully covered with
   additional O.C.T., and flash-frozen in isopentane and liquid nitrogen
   for ∼30 s. The frozen samples were equilibrated for 5 min in a cryostat
   chamber before sectioning into 15 μm slices using a Leica CM1850
   cryostat. The sections were mounted on adhesive glass slides,
   air-dried, stored at −20 °C, and stained and analyzed promptly. For
   staining, sections were thawed, washed with PBST, and blocked for 2 h
   at room temperature (RT) with 5% bovine serum albumin (BSA) (sigma
   Aldrich, A7030) and 0.1% Triton X-100 in PBS. Primary antibodies (UCP1,
   1:100, CST, 72298; CD142, 1:100, R&D Systems, AF3178) were applied and
   incubated overnight at 4 °C. After washing with PBST, sections were
   incubated for 3 h at RT in the dark with anti-goat FITC secondary
   antibody (1:400, Proteintech, SA00003-3) for CD142, anti-rabbit
   CoraLite594 secondary antibody (1:400, Proteintech, SA00013-8) for
   UCP1, and DAPI (Invitrogen, D1306). Immunofluorescence images were
   captured using a Nikon TiE-A1 Plus confocal microscope (Nikon, Japan).

2.7. Immunofluorescence staining

   Primary adipocytes were stained with MitoTracker™ Deep Red FM
   (Invitrogen, A66440) to label mitochondria, diluted in pre-warmed
   serum-free DMEM, and incubated for 30 min at 37 °C. And then the
   adipocytes were fixed in 4% paraformaldehyde at RT for 15 min, followed
   by 0.3% Triton™ X-100 (Sigma, T8787) for 10 min, and then blocked in 1%
   BSA for 1 h at RT. Next, adipocytes were subjected to an overnight
   incubation at 4 °C with CD142 primary antibody (1:100, R&D Systems,
   AF3178). Subsequently, adipocytes were stained with anti-goat FITC
   secondary antibody (1:400, Proteintech, SA00003-3) for 1 h at RT. Then,
   the coverslips were mounted with mounting reagent including DAPI.
   Finally, the coverslips were imaged using confocal microscope (Nikon,
   Japan).

2.8. Immunoblotting

   Primary adipocytes were scraped, and adipose tissue was homogenized in
   cold RIPA buffer supplemented with a protease inhibitor cocktail
   (Roche, 11836170001) on ice for 1 h. The lysates were centrifuged at
   12,000 rpm for 15 min at 4 °C, and the supernatant protein
   concentration was measured using the BCA Protein Assay (Thermo Fisher
   Scientific, 23225). Protein samples (20 μg) were mixed with 4 × loading
   buffer, heated at 95 °C for 5 min, and separated on a 15% SDS-PAGE gel.
   Proteins were transferred to 0.45 μm nitrocellulose (NC) membranes (GE
   Healthcare, 10600002), blocked with 5% non-fat milk in TBS for 1 h, and
   incubated overnight at 4 °C with primary antibodies: anti-CD142
   (1:1000, R&D Systems, AF3178), anti-IGFBP3 (1:1000, Proteintech,
   10189-2-AP), anti-UCP1 (1:1000, CST, 72298S), anti-TOMM20 (1:1000, CST,
   D8T4N), anti-P-SMAD2/3 (1:1000, CST, 8828S), anti-SMAD2/3 (1:1000, CST,
   8685S), anti-P-p38 (1:1000, CST, 4511S), anti-p38 (1:1000, CST, 8690S),
   anti-p53 (1:1000, CST, 30313S), anti-p38 (1:1000, Proteintech,
   28248-1-AP) and anti-GAPDH (1:1000, CST, 97166). After washing with
   TBST, membranes were incubated with secondary antibodies for 1 h at RT.
   Protein bands were detected using Pierce ECL substrate (Thermo Fisher
   Scientific, 32106) and visualized with a BioLight Gelview 6000 Pro
   imaging system.

2.9. Flow cytometry

   To isolate the CD142+ cell population from mouse ASPCs, flow cytometry
   was performed. Briefly, ASPCs were collected and resuspended in FACS
   buffer (Biolegend 420201). After filtering through a 70 μm nylon mesh
   (Absin, abs7232), the cell density was adjusted to 1 × 10^ˆ6 cells/mL.
   Cells were incubated with anti-CD142 antibody (1:50, R&D Systems,
   AF3178) at 4 °C for 1 h. Following washing with FACS buffer, cells were
   incubated with anti-goat FITC secondary antibody (1:200, Proteintech,
   SA00003-3) at 4 °C for 30 min, followed by additional washing with FACS
   buffer. Cell sorting was conducted using a BD FACSAria™ flow cytometer
   (BD Biosciences), equipped with a 100 μm nozzle and lasers/filters for
   FITC detection (488 nm and 515/20 nm).

2.10. scRNA-seq data preprocessing and quality control

   Raw data for [75]GSE227439 were downloaded from the GEO database and
   processed using CellRanger (version 7.0.1) from 10× Genomics. Sequence
   alignment was performed against the mm10 genome using the CellRanger
   count module with default parameters. To ensure data quality,
   low-quality cells were excluded based on the following criteria: (1)
   total UMI count <200; (2) total UMI count >5000; and (3) mitochondrial
   transcript proportion >15%.

2.11. Transwell co-culture of Aregs and non-Aregs cells

   After flow cytometric sorting, CD142+ Aregs cells and CD142− non-Aregs
   cells from ASPCs (the ASPCs were not passaged and were defined as
   passage 0) were directly seeded into the upper chambers and the bottom
   layer of 24-well transwell plates (Corning, 3413), respectively, at
   approximately 60% confluence (about 1.2 × 10^5 cells, passage 1). The
   Aregs from Group IV and its corresponding control group (KO ctrl) in
   the upper chambers were then transferred to another 24-well plate for
   lentiviral infection. Two days post-infection, an additional 2-day
   selection with puromycin (1 μg/mL) was performed until 90% of the
   control uninfected cells had died. Subsequently, the selected Aregs
   together with the upper chamber were moved back to the original plate
   containing non-Aregs in bottom layer for continued co-culture (day 0).
   Meanwhile, on day 0, 6 μg/mL of anti-IGFBP3 antibody (Proteintech,
   10189-2-AP) was added to the co-culture medium of Group V. Two days
   later (day 2), when cell confluence reached 100%, either the apoptosis
   of non-Aregs in the bottom layer of plates was assessed, or they were
   subjected to beige differentiation for subsequent experiments.

2.12. Annexin V/propidium iodide (PI) apoptosis assay

   After 48 h of transwell co-culture, ASPCs from each group were digested
   with 0.25% trypsin (EDTA-free) and stained using the Annexin V/PI
   Apoptosis Detection Kit (BD Biosciences, AB_2869082). Briefly, ASPCs
   were washed with cold PBS and resuspended in 100 μL of 1 × binding
   buffer at a concentration of approximately 1 × 10^6 cells/mL. Annexin
   V-FITC (5 μL) and PI (5 μL) were added, followed by 15 min of
   incubation at RT in the dark. After incubation, 400 μL of 1 × binding
   buffer was added to each tube, and apoptosis was immediately analyzed
   using flow cytometry (Beckman Coulter).

2.13. RNA velocity analysis

   To investigate cellular dynamics and lineage evolution, RNA velocity
   analysis was performed on the scRNA-seq data. Unspliced and spliced
   mRNA transcripts were quantified using velocyto (version 0.17.17)
   [[76]25]. Subsequently, scVelo (version 0.2.5) was used to identify
   highly variable genes and conduct dynamic modeling. Loom files were
   generated from BAM files in the CellRanger output directory using the
   Velocyto Run10× function. Data preprocessing was conducted with
   Scv.pp.filter_and_Normalize, followed by RNA velocity estimation using
   Scv.pp.moments and Scv.tl.velocity in “stochastic mode”. RNA velocity
   results were integrated with UMAP dimensionality reduction to predict
   the direction and rate of differentiation from progenitor cells to
   terminally differentiated cells.

2.14. snRNA-seq batch correction, dimensionality reduction, and clustering

   The UMI count matrix was imported into Seurat (version 4.4.0) for
   further analysis [[77]26]. Batch correction across multiple samples was
   performed using Canonical Correlation Analysis (CCA). Key steps
   included: (1) log-normalization of the UMI count matrix using
   NormalizeData() (scale.factor = 10,000) to mitigate differences in
   total gene expression levels between cells; (2) identification of
   highly variable genes with FindVariableFeatures() (nFeatures = 2000);
   (3) identification of integration anchors using
   FindIntegrationAnchors(); and (4) data integration using
   IntegrateData() with 30 principal components (nPCs = 30). Cellular
   subpopulations were identified by applying FindNeighbors() and
   FindClusters() (resolution = 0.4). Finally, UMAP visualization was
   performed using RunUMAP() based on the top 15 adjusted CCA dimensions.

2.15. Differential expression analysis and cell type annotation

   After filtering, a total of 58,714 single cells were grouped into 13
   distinct clusters. Differentially expressed genes (DEGs) were
   identified using the FindAllMarkers() function with parameters
   “logfc.threshold = 0.25” and “min.pct = 0.1.” P-values were adjusted
   using the Bonferroni correction method, with a significance threshold
   of 0.05.

2.16. Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and
Genomes (KEGG) analysis

   Gene sets were annotated using GO terms and KEGG pathways. Enrichment
   analyses utilized the clusterProfiler R package (version 4.6.2) with
   the enrichGO and enrichKEGG functions. Annotation was condudted with
   the org.Mm.eg.db package, and multiple testing corrections were applied
   using the Benjamini-Hochberg (BH) method (OrgDb = org.Mm.eg.db,
   pAdjustMethod = “BH”, pvalueCutoff = 0.05). GO terms and KEGG pathways
   with adjusted p-values (p-adjust) < 0.05 were considered significantly
   enriched, reflecting their biological relevance to the dataset.

2.17. Transcription factor (TF) analysis

   TF enrichment and regulon activity were evaluated using the pySCENIC
   package (version 0.12.0) in Python (version 3.9.20). Gene regulatory
   networks were constructed based on co-expression and DNA motif
   analysis. The gene-motif ranking, spanning 10 kb on either side of the
   transcription start site, was used to delineate the search space for TF
   regulatory networks. The SCENIC workflow was implemented with default
   parameters, using the raw count matrix from all samples as input. The
   analysis consisted of three steps: First, co-expression modules were
   computed, and relationships between TFs and target genes were weighted
   with GRNBoost. Second, RcisTarget was applied to identify transcription
   factors with direct targets (regulons). Finally, AUCell assessed
   regulon activity in individual cells, and heatmaps visualized the
   activity of regulons using ggheatmap.

2.18. Gene set enrichment analysis (GSEA)

   GSEA was conducted on gene expression data from different cell types
   using the RunGSEA() function in the SCP package (version 0.5.6) to
   explore biological functions and regulatory mechanisms. The GO
   Biological Process (GO_BP) database was used for gene set annotation,
   with Mus musculus as the species reference. Cells were grouped by cell
   type using the parameter group_by = “CellType”, and only DEGs with
   adjusted p-values (p_val_adj <0.05) were included for enrichment
   analysis. Gene set sizes were restricted to a minimum of 10 and a
   maximum of 500 (minGSSize = 10, maxGSSize = 500), and normalized
   enrichment scores (scoreType = “std”) were computed. To enhance
   computational efficiency, parallel processing was enabled using the
   default configuration of the BiocParallel package.

2.19. Inference and analysis of intercellular communication networks

   To accurately explore cell–cell interactions within the dataset, three
   cell communication analysis methods were applied in this study. First,
   CellChat (version 2.1.2) was used to analyze the differences in
   intercellular communication between young and aged mice under TN
   conditions and after 3 or 14 days of cold stimulation. Cell–cell
   communication probabilities were calculated using the computeCommunProb
   function (parameter: type = “triMean”) and network centrality scores
   were derived using the netAnalysis_computeCentrality function
   (parameter: slot.name = “netP”) to quantify the roles and importance of
   different cell types within the signaling network. Next, CellPhoneDB
   was run by converting rat gene names to their human homologs using the
   homologene tool. The analysis was conducted in a Linux environment,
   executing CellPhoneDB scripts to generate key output files, including
   ligand-receptor pair communication significance (pvalues.txt) and
   average expression levels (means.txt). Visualization was performed
   using ktplots (version 2.4.0), where heatmaps were created to display
   the significance and expression levels of ligand-receptor interactions
   across different cell types. A communication network was then
   constructed to uncover the signaling patterns between cell types, with
   a focus on the key interactions between Aregs and SMC-like cells.
   Lastly, NicheNet (version 2.2.0) was employed to further infer
   ligand-receptor and ligand–target relationships between Aregs and
   SMC-like cells. Using the nichenet_seuratobj_aggregate function, Seurat
   objects were input, with Aregs designated as the signaling sender and
   SMC-like cells as the receiver. The average expression and standard
   deviation of ligands were calculated to identify Aregs-specific
   ligands, from which a ligand-target gene network from Aregs to SMC-like
   cells was built. A filtering threshold was set at the 40th percentile
   of the ligand-target gene pair weight distribution, selecting pairs
   with higher weights. Finally, a Circos plot was used to visualize the
   interactions between Aregs-specific ligands and their corresponding
   target genes and receptors in SMC-like cells.

2.20. Virtual KO of key genes

   In this study, the scTenifoldKnk method [[78]27] was used to perform
   virtual KO analysis on scRNA-seq data to explore changes in gene
   regulatory networks and their functional impacts. First, the gene
   expression matrix was extracted from the scRNA-seq data, and
   low-expression or noise genes (such as ribosomal genes and
   mitochondrial genes) were filtered out. Only genes expressed in at
   least 10 cells were retained to ensure data quality. A wild-type gene
   regulatory network (WT GRN) was constructed based on scTenifoldKnk, and
   a KO gene regulatory network (KO GRN) was generated by virtually
   knocking out the target genes. The KO GRN was then compared to the WT
   GRN using manifold alignment methods to identify differentially
   regulated genes (DRGs) associated with the KO of the target gene. DRGs
   with a significance level meeting the threshold (FDR-adjusted p-value
   <0.05) were selected for downstream functional analysis. Functional
   enrichment and pathway analysis of the DRGs were conducted using the
   STRING protein interaction database and the Enrichr tool to reveal the
   potential biological significance of the differentially regulated
   genes. Various visualization methods, including Rank plots, GSEA
   enrichment plots, and network analysis diagrams, were used to
   comprehensively display the impact of gene KO on the regulatory network
   and its potential biological functions.

2.21. Statistical analysis

   Data are expressed as mean ± SD. Statistical analysis was conducted
   using one-way ANOVA with GraphPad Prism 9. Statistical significance is
   indicated as: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

3. Results

3.1. Aging leads to an increase in Aregs cells and a decrease in SMC-like
cells in murine sWAT following cold exposure

   To assess how aging affects sWAT browning, we integrated scRNA-seq data
   (GEO: [79]GSE227439) of sWAT from young (9-week-old) and old
   (57-week-old) mice under different environmental temperatures
   [thermoneutral (30 °C, TN) and cold conditions (6 °C, Cold)]. After
   excluding doublets and residual immune cells, a total of 58,714 cells
   were obtained (29,390 cells from young mice and 29,324 cells from old
   mice) for downstream analysis. After unbiased clustering using UMAP, we
   classified all cells into 7 cell types based on the expression of
   specific marker genes in each cluster. These included five adipose
   progenitor cell types: Progenitor (Dpp4+), Aregs (Cd142+), Preadipocyte
   (Cxcl14+), SMC-like cells (Acta2+), and Spp1+ cells. Additionally, two
   vascular-related cell types were identified: Pericytes (Pdgfrb+) and
   Endothelial cells (ECs, Flt1+) ([80]Figure 1A,B).

Figure 1.

   [81]Figure 1
   [82]Open in a new tab

   Aging induces an elevation in Aregs and a reduction in SMC-like cells
   among ASPCs following cold exposure. (A) UMAP plot showing the
   clustering distribution of 58,714 cells from young and aged mice sWAT.
   (B) Dot plot displaying the expression levels and proportions of
   selected marker genes across different cell types. The size of the dot
   represents the percentage of cells expressing the gene, while the color
   intensity reflects the average gene expression level. (C) Stacked bar
   chart showing the proportion of cell types in sWAT at different time
   points (TN, 3 days, and 14 days post-cold exposure). (D) Ucp1 and Cd142
   mRNA expression levels analyzed by qPCR in sWAT (n = 11). (E)
   Immunoblots for CD142, UCP1, TOMM20 and GAPDH in sWAT (n = 3). (F)
   Quantification of immunoblots of CD142, UCP1 and TOMM20, protein levels
   are normalized to GAPDH. (G) Immunohistochemistry fluorescence for UCP1
   (red) and CD142 (green) (scale bar = 50 μm or 10 μm). Data represent
   mean ± SD, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

   Our analysis revealed changes in the cell composition of sWAT at
   different time points following cold exposure in mice of different
   ages. The results showed that in aged mice, the number of Aregs
   increased with prolonged cold exposure (14D) ([83]Figure 1C and
   [84]Supplementary Fig. S1), whereas the number of beige
   adipocyte-specific precursor cells [[85][3], [86][4], [87][5]],
   SMC-like cells, decreased (14D) ([88]Figure 1C). Notably, this
   phenomenon was almost absent in young mice. Next, to validate the
   results of bioinformatics analysis, we first used qPCR to assess the
   mRNA expression level of the beige adipocyte marker gene Ucp1,
   confirming the successful establishment of our cold stimulation mouse
   model. Subsequently, we evaluated the level of Aregs by analyzing the
   mRNA expression of Cd142. The results showed a significant increase in
   the Ucp1 mRNA in young mice following cold exposure (14D), whereas the
   increase in aged mice was less pronounced and remained at lower levels.
   The result is consistent with numerous previous well-established
   studies of aging and browning in mice. More importantly, qPCR results
   also validated a significant upregulation of Cd142 in the cold-exposed
   aged group (14D) ([89]Figure 1D). Western blot analysis further
   supported these findings, showing increased expression of UCP1 and the
   mitochondrial outer membrane protein TOMM20 (used to indicate
   mitochondrial content) in the cold-exposed sWAT, confirming the
   successful induction of browning. Additionally, CD142 expression at the
   protein level was also significantly higher in the cold-exposed aged
   group ([90]Figure 1E). Similarly, immunohistochemical fluorescence
   results further corroborated these findings ([91]Figure 1F).

   We performed GO and KEGG pathway analyses for each cell type, revealing
   that the functions of Aregs were predominantly enriched in “Positive
   regulation of fat cell differentiation,” “Insulin-like growth factor
   receptor signaling pathway,” and “Regulation of smooth muscle cell
   migration.” In contrast, SMC-like cells were enriched in “Positive
   regulation of ion transport,” “Multicellular organismal signaling,” and
   “Negative regulation of vascular-associated smooth muscle cell
   proliferation” ([92]Figure 2A). The pathways enriched in Aregs and
   SMC-like cells suggest a potential complementary or corresponding
   relationship between these two cell types in sWAT, highlighting a
   coordinated regulatory network among different cell subpopulations to
   enhance adaptability. Further investigation into their interactions may
   reveal new therapeutic targets for metabolic and vascular diseases.

Figure 2.

   [93]Figure 2
   [94]Open in a new tab

   Analysis of functional properties and molecular characteristics of
   ASPCs under aged-cold stimulation. (A) Functional enrichment analysis
   of different cell types. Left: Heatmap of DEGs for each cell type,
   showing some DEGs contributing to enrichment. Middle: GO biological
   processes. Right: KEGG pathways. (B) Volcano plot showing the top 10
   upregulated DEGs in Aregs and SMC-like cells. (C) Heatmap of
   transcription factor analysis for Aregs and SMC-like cells, with colors
   representing the relative Z-scores of the corresponding transcription
   factors (red indicating higher reliability and blue representing lower
   reliability). “+” indicates upregulation.

   Subsequently, we performed DEGs analysis on the two significantly
   altered cell populations, Aregs and SMC-like cells, and visualized the
   DEGs using volcano plots. Among the top 10 upregulated genes, we
   identified several noteworthy candidates that, aside from the marker
   genes, have been rarely emphasized in prior studies but may play roles
   in the aging-associated impairment of sWAT browning. For instance,
   Aregs exhibited upregulation of Igfbp3 and complement-related genes
   (C7, C2, C4b), while SMC-like cells showed increased expression of cell
   cycle regulatory genes such as Cdkn1a and cytoskeletal genes (Tpm1,
   Tpm2) ([95]Figure 2B). Studies have shown that complement activation
   induces inflammation in adipose tissue [[96]28], which negatively
   impacts the browning process of WAT. The cytoskeleton is essential for
   maintaining low inflammation levels, regulating proliferation and
   migration, and ensuring the normal functions of smooth muscle cells
   [[97][29], [98][30], [99][31]]. This undoubtedly plays a significant
   role in the browning of WAT. To further reveal the unique
   transcriptional features of each cell type, we analyzed transcription
   factors with high transcriptional activity in each cell type
   ([100]Supplementary Fig. S2), particularly in Aregs and SMC-like cells
   ([101]Figure 2C). In Aregs (14D), the transcription factor Tfdp1 was
   highly expressed, with the most significant differential expression
   observed (encoding PD1), playing an important role in promoting cell
   apoptosis [[102]32]. Additionally, we also observed high expression of
   Rarg (encoding retinoic acid receptor) in Aregs (14D) and SMC-like
   cells (14D), suggesting that Aregs in the sWAT of aged individuals
   under cold stimulation may further suppress the browning process by
   inhibiting adipogenesis, potentially leading to the reduced browning
   capacity of aged adipose tissue. This finding further extends previous
   research [[103]19]. Moreover, in SMC-like cells (14D), we also observed
   the expression of pro-apoptotic genes, such as Alx4 and Npas2, which
   are frequently reported in cancer research [[104]33,[105]34],
   suggesting they may have an adverse impact on adipose tissue browning.
   In summary, under aging cold exposure conditions, distinct cell types
   of ASPCs, especially Aregs and the beige adipocyte precursor SMC-like
   cells, exhibit specific changes in cell numbers, gene expression, and
   functional enrichment. These changes may collectively impact the
   plasticity of adipose progenitor cells and their browning capacity,
   providing new insights into the imbalance of energy metabolism in aging
   tissues.

3.2. Aregs regulate surrounding ASPCs by IGFBP3-mediated apoptotic signaling

   In our previous results, we identified Aregs as one of the two cell
   types significantly affected by aging in response to cold stimulation,
   with potential involvement in regulating aging-associated browning
   processes. To further elucidate the regulatory capacity of Aregs on
   surrounding cells, we conducted secondary clustering analysis on 48,003
   single cells, including Aregs and two closely related cell types,
   Progenitor and Preadipocyte. Using Partition-based graph abstraction
   (PAGA), we performed cell dynamics analysis, which revealed that Aregs,
   as intermediate-state cells, are likely to regulate the other two cell
   types. Velocity pseudotime analysis showed that Preadipocytes exhibit a
   significantly higher degree of differentiation compared to Aregs and
   Progenitors ([106]Figure 3A), aligning with previous understandings of
   these cell types [[107]18,[108]35]. Next, we visualized the major
   upregulated and specifically expressed genes, excluding marker genes,
   for the three cell types on a UMAP plot. These genes provide molecular
   evidence for understanding the functions and interconnections of these
   cells. Among them, Aregs specifically expressed high levels of Gas6 and
   Igf1, Progenitors showed high expression of Axl and Cd55, while
   Preadipocytes exhibited specific upregulation of Icam1 and Col15a1
   ([109]Figure 3B). Gas6 and Axl, as a ligand-receptor pair, play roles
   in promoting the activation of BAT, WAT browning [[110]36], and the
   reduction of inflammation through the promotion of macrophage M2
   polarization [[111]37]. The increased expression of Icam1 and Col15a1
   in Preadipocytes represents a rise in adhesion molecule and collagen
   fiber expression during the differentiation of adipocytes toward mature
   white adipocytes. This change significantly enhances the adhesive
   capacity of adipocytes, providing a more stable and favorable
   extracellular matrix environment for their growth and maturation
   [[112]35]. To further analyze the regulatory capacity of Aregs on
   surrounding cells, we focused on transition-state genes between
   different cell types. Pseudotime analysis, with Aregs as the starting
   cell state, revealed transition-state genes such as Igf1, Sfrp2, Gas1,
   and Epha3. The GO functional enrichment analysis of these genes showed
   significant enrichment in pathways such as “regulation of apoptotic
   signaling pathway” ([113]Figure 3C). Combining these findings with the
   earlier results and the role of IGFs-mediated apoptotic signaling
   [[114][21], [115][22], [116][23],[117]38], we hypothesize that Aregs
   play a critical role in modulating the microenvironment between
   different ASPCs subtypes through the IGFBP3-IGFs signaling-apoptotic
   pathway.

Figure 3.

   [118]Figure 3
   [119]Open in a new tab

   Aregs influences adjacent ASPCs via IGFBP3-mediated apoptosis
   signaling. (A) RNA velocity analysis. Left: PAGA cell dynamics analysis
   displays differentiation trajectories of three ASPCs subtypes on the
   UMAP plot. Right: Velocity pseudotime analysis indicates the
   differentiation extent of the three ASPCs subtypes. (B) UMAP showing
   the expression distribution of unique genes across the three ASPCs
   subtypes, with darker colors indicating higher expression levels. (C)
   Pseudotime analysis of ASPCs. Left: Dynamics of key transition genes
   along differentiation pathways. Middle: Gene Ontology (GO, biological
   process). Right: Key enriched genes in the matched pathways. (D)
   Differential expression analysis highlights the top 5 upregulated and
   downregulated genes in Aregs cells of aged mice after 14 days of cold
   stimulation. (E) Violin plots display the expression levels of Igf1 and
   Fabp4 in Aregs cells under various cold stimulation conditions. (F)
   Network diagram illustrating the functional enrichment analysis and
   pathway-related genes specific to upregulated genes in Aregs cells. (G)
   GSEA analysis based on the MSigDB database shows pathway enrichment for
   Aregs-specific gene sets. Data represent mean ± SD, ns represents
   p > 0.05, ∗∗∗∗p < 0.0001.

   Given the significant increase in the number of Aregs in aged mice
   after 14 days of cold stimulation, we specifically examined the DEGs in
   Aregs under these conditions. Notable upregulated genes included Fabp4,
   Sema3e, Igfbp5, Actb, and Gsn, while downregulated genes included Igf1,
   Egr1, Fbln1, Il11ra1, and Srpx ([120]Figure 3D). We observed that the
   expression of Igf1 was significantly reduced, and Fabp4 was
   significantly increased in aged mouse Aregs after 14 days of cold
   stimulation, a phenomenon that was exclusive to aged mice
   ([121]Figure 3E). Further exploration of the biological processes
   enriched for these DEGs revealed pathways such as “Negative regulation
   of protein phosphorylation”, “Negative regulation of cell migration”,
   “Negative regulation of smooth muscle cell proliferation”, “Positive
   regulation of muscle cell apoptotic process”, “Lipid transport” and
   “Positive regulation of insulin-like growth factor receptor signaling
   pathway”. These findings suggest that Aregs play a crucial role in
   negatively regulating cell proliferation and migration, with Igfbp3
   identified as a central factor linking these pathways ([122]Figure 3F).
   GSEA analysis using the MSigDB database showed that the gene expression
   profile of Aregs was significantly enriched in gene sets associated
   with adipose tissue aging. Enrichment curves highlighted a strong
   positive correlation for genes such as C4b, Apod, Cst3, Igfbp3 and
   Cxcl12, indicating that Aregs are closely linked to adipose tissue
   aging. Additionally, Aregs were highly enriched in Reactome gene sets
   related to IGFs signaling, with positively correlated cumulative
   effects for genes including Igf1, Cp, C4b, Igfbp3 and Gas6
   ([123]Figure 3G). These results indicate that Aregs, as one of the key
   cell types responding to aging under cold stimulation, exhibit critical
   functions in regulating the microenvironment of ASPCs subtypes through
   Igfbp3 and its modulation of the IGFs-apoptotic pathway.

3.3. Apoptotic signaling plays a critical role in SMC-like cells under
aged-cold conditions

   SMC-like cells, as another cell type with significantly reduced numbers
   in the aged-cold stimulation group, were analyzed alongside Pericytes,
   which share a similar expression profile, resulting in a total of 3,823
   cells for secondary clustering to refine the analysis of SMC-like
   cells. UMAP and Velocity trajectory analyses revealed that a subset of
   less-differentiated SMC-like cells transitions into Pericytes, while
   another subset progresses toward more differentiated SMC-like cells
   ([124]Figure 4A). Trajectory gene exploration showed that Pericytes
   specifically expressed Pdgfrb and Postn, which are consistent with
   their typical characteristics [[125]39,[126]40]. Meanwhile, SMC-like
   cells exhibited not only the conventional features but also high
   expression of Cdkn1a and Tgfb3. TGFβ signaling, which can upregulate
   Cdkn1a through Smad-dependent mechanisms and is also associated with
   fibrosis in adipose tissue aging [[127]41,[128]42]. Furthermore, the
   elevated expression of Cdkn1a suggests its association with the aging
   and functional exhaustion of SMC-like cells. Additionally, the
   transition-state genes Id1 and Mob2 are likely to play key roles in the
   process of SMC-like cell conversion into Pericytes ([129]Figure 4B)
   ([130]Supplementary Fig. S2).

Figure 4.

   [131]Figure 4
   [132]Open in a new tab

   Apoptotic signaling is pivotal in regulating SMC-like cells under
   aged-cold conditions. (A) UMAP integrated with PAGA trajectory analysis
   illustrating the differentiation trajectories of 3,823 SMC-like and
   Pericyte cells. (B) UMAP plot showing the expression distribution of
   specific genes in SMC-like and Pericyte cells, with darker colors
   indicating higher expression levels. (C) Network diagram depicting GO
   and KEGG functional enrichment pathways associated with upregulated
   genes specific to SMC-like cells. (D) GSEA analysis based on the MSigDB
   database revealing pathway enrichment for SMC-like-specific gene sets.
   (E) UMAP plot visualizing gene expression of the TGFβ, MAPK, and P53
   signaling pathways based on AUCell scoring. (F) Differential expression
   analysis showing upregulated and downregulated genes in SMC-like cells
   from aged mice after 14 days of cold stimulation. (G) Violin plot
   illustrating the expression levels of Dcn and Celf2 in SMC-like cells
   from aged mice under different cold stimulation conditions. Data
   represent mean ± SD, ns represents p > 0.05, ∗∗∗∗p < 0.0001.

   Subsequently, we performed gene function enrichment analysis on
   SMC-like cells. GO-BP analysis revealed that SMC-like cells were
   primarily enriched in pathways such as “regulation of smooth muscle
   cell proliferation”, “regulation of protein phosphorylation”,
   “regulation of apoptotic signaling pathway” and “regulation of
   serine/threonine kinase activity”, suggesting their potential role in
   maintaining adipose tissue homeostasis and mediating signal
   transduction. Additionally, KEGG pathway enrichment analysis
   highlighted key functional modules of SMC-like cells, including
   “adrenergic signaling in cardiomyocytes”, “cGMP-PKG signaling pathway”,
   “MAPK signaling pathway”, “TGFβ signaling pathway” and “regulation of
   lipolysis in adipocytes” ([133]Figure 4C). Based on these initial
   enrichment findings, we conducted GSEA to evaluate the activation
   states of signaling pathways ([134]Figure 4E). The analysis
   demonstrated significant enrichment of the TGFβ signaling pathway
   (NES = 1.678, p = 8.64e-03). Concurrently, the MAPK signaling pathway
   also showed significant enrichment (NES = 1.478, p = 9.85e-03),
   reflecting its critical role in cell proliferation and apoptosis
   [[135]43]. Although the p53 signaling pathway did not reach statistical
   significance (NES = 1.44, p = 7.04e-02), the upregulation of key genes
   such as Cdkn1a, Gadd45b, and Thbs1 suggests its collaborative role with
   TGFβ and MAPK signaling in inhibiting cell cycling or promoting
   apoptosis, contributing to the reduction of SMC-like cells. These
   findings collectively support the hypothesis that apoptotic signaling
   plays a central regulatory role in the transition and reduction of
   SMC-like cells ([136]Figure 4D). We further utilized AUCell scoring to
   project the gene expression patterns of TGFβ, MAPK, and p53 signaling
   pathways onto the UMAP plot. After 14 days of cold stimulation, a
   subset of highly differentiated SMC-like cells in aged mice exhibited
   significant activation of these three pathways ([137]Figure 4E).

   To better understand the fate changes of SMC-like cells in the
   aged-cold group, we performed DEGs analysis on these cells. The results
   showed that after 14 days of cold stimulation, aged SMC-like cells
   exhibited upregulation of genes such as Dcn, Col3a1, Igfbp6, Celf2, C3,
   Dpt, and Col1a2, while genes such as Ubb, Ppp1r15a, and Ncapd3 were
   downregulated, indicating that these genes may play critical roles in
   the reduction of SMC-like cells ([138]Figure 4F). The upregulation of
   fibrosis-related genes (e.g., Dcn, Col3a1, and Col1a2) and
   inflammation-associated genes (e.g., C3) suggests that cold exposure
   accelerates the aging and fibrotic processes in these cells.
   Conversely, the downregulation of genes like Ubb, Ppp1r15a, and Ncapd3,
   which are involved in protein degradation and cellular stress responses
   [[139][44], [140][45], [141][46]], further points to the functional
   exhaustion and senescence of SMC-like cells. Among them, Dcn expression
   continued to rise significantly as cold stimulation progressed, and
   Celf2 exhibited a marked increase at the late stage of cold stimulation
   ([142]Figure 4G). Both DCN and CELF2 have been reported to inhibit cell
   proliferation and promote cell apoptosis [[143][47], [144][48],
   [145][49]]. Together, these molecular changes indicate that
   cold-induced stress exacerbates the reduction of SMC-like cells and
   contributes to the impaired browning capacity of adipose tissue in aged
   individuals.

3.4. Aregs-SMC-like cells cross-talk regulates apoptosis and impairs beige
adipocytes formation

   Considering the opposing trends in cell numbers of Aregs (upregulated)
   and SMC-like cells (downregulated) under aged-cold stimulation, with
   Aregs acting as a pro-apoptotic signal donor and SMC-like cells as its
   recipient, we decided to explore the cross-talk between these two cell
   types. To systematically elucidate this interaction, we employed three
   cell communication analysis methods: CellChat v2, NicheNet v2, and
   CellPhoneDB v5. First, we used CellChat to calculate the number of
   communications between different cell types. The results revealed that
   overall intercellular communication was relatively lower in aged mice
   after 14 days of cold stimulation. Focusing on IGF signaling, which had
   been highlighted in our previous analyses, we observed that the number
   of IGF signal-mediated communications progressively decreased with
   prolonged cold stimulation in the aged group, further corroborating our
   earlier findings ([146]Figure 5A). Next, we used CellPhoneDB to
   identify specific ligand-receptor interactions between Aregs (as
   ligand-producing cells) and SMC-like cells (as recipient cells). This
   analysis uncovered several unique pro-apoptotic ligand-receptor
   interactions, such as THBS2-CD36 and IGFBP3-TMEM219 ([147]Figure 5B).
   Finally, we applied NicheNet to further investigate key ligand-receptor
   and ligand-target interactions between Aregs and SMC-like cells after
   14 days of cold stimulation in aged mice. Strikingly, comparisons
   between the aged-14 days cold stimulation and the young-14 days cold
   stimulation group, or between the 14-day and 3-day cold stimulation in
   aged mice, revealed that Aregs consistently secreted the ligand Igfbp3.
   This finding underscores the critical role of Igfbp3, which may
   regulate apoptosis in SMC-like cells by targeting genes such as Cdkn1a
   and Irf1, as well as the receptor Egfr ([148]Figure 5C). To further
   explore the functional and regulatory effects of Igfbp3 and Cdkn1a in
   Aregs and SMC-like cells, respectively, we utilized scTenifoldKnk to
   perform virtual KO of these genes and examine the resulting gene
   perturbations within each cell type. In Aregs, the virtual Igfbp3 KO
   identified Neat1, Fabp4, Nav3, Celf2, and Ebf1 as the top five
   perturbed genes. Functional enrichment analysis revealed that these
   perturbed genes were enriched in pathways such as “BDNF signaling
   pathway”, “white adipocyte differentiation”, “adipogenesis”, “FSH
   regulation of apoptosis”, “regulation of insulin-like growth factor
   receptor signaling pathway”, “negative regulation of smooth muscle cell
   proliferation”, “response to TGFβ” and “RhoA activity regulation”
   ([149]Figure 5D). Similarly, in SMC-like cells, the virtual KO of
   Cdkn1a identified Ccn1, Fos, Fth1, Jun, and Ptma as the top five
   perturbed genes. Functional enrichment analysis indicated that these
   perturbed genes were associated with pathways such as “BDNF signaling
   pathway”, “RANKL regulation of apoptosis”, “immune response”, “RhoA
   signaling pathway”, “regulation of apoptosis” and “TGFβ receptor
   signaling in EMT” ([150]Figure 5E). In summary, our results demonstrate
   that under aged-cold stimulation, Aregs secrete Igfbp3, which interacts
   with multiple targets in SMC-like cells, including Cdkn1a, to establish
   specific intercellular communication. This interaction enhances
   apoptosis in SMC-like cells through multiple signaling pathways,
   ultimately impairing beige adipocytes formation.

Figure 5.

   [151]Figure 5
   [152]Open in a new tab

   The cross-talk between aged Aregs and SMC-like cells following cold
   stimulation. (A) CellChat: Bar chart showing overall differences in the
   number of cell–cell interactions following cold stimulation. (B)
   CellPhoneDB: Chord diagram illustrating apoptosis-related
   ligand-receptor interactions specific to Aregs and SMC-like cells. (C)
   NicheNet: Chord diagram displaying ligand-receptor (target)
   interactions specifically upregulated in the aged 14D group compared to
   the young 14D group (top) and aged 14D group compared to the aged 3D
   group (bottom). (D) scTenifoldKnk analysis of Aregs. Left: Rank plot
   showing 30 perturbed genes influenced by Igfbp3 (p < 0.05); Right:
   Network diagram depicting the central network and functional enrichment
   pathways of perturbed genes. (E) scTenifoldKnk analysis of SMC-like
   cells. Left: Rank plot showing 27 perturbed genes influenced by Cdkn1a
   (p < 0.05); Right: Network diagram depicting the central network and
   functional enrichment pathways of perturbed genes.

3.5. Aregs and IGFBP3 inhibit beige adipocytes formation by regulating
SMC-like cells in murine ASPCs

   To clarify the impact of Aregs and its secreted IGFBP3 on the number of
   SMC-like cells and the subsequent impairment of beige adipocytes
   formation, we performed flow cytometric sorting of primary ASPCs
   isolated from murine sWAT to separate CD142+ (Aregs) and CD142−
   (non-Aregs), which were then co-cultured in different groups using
   transwell assays ([153]Figure 6B). Flow sorting results showed that,
   compared to the negative control group incubated only with
   FITC-conjugated secondary antibody ([154]Figure 6A left), the
   experimental group yielded approximately 7.82% CD142+ cells
   ([155]Figure 6A right), which closely matched the proportion of Aregs
   cells observed in our previous bioinformatic analysis, confirming
   successful separation of Aregs cells. To further validate the success
   of flow sorting, we collected groups I, II, and Aregs for qPCR and
   immunoblotting. Results showed that CD142 mRNA and protein levels in
   group II were significantly lower than in group I, while Aregs group
   showed significantly higher levels compared to group I
   ([156]Supplementary Figs. S4A, B, C). This further confirmed the
   successful separation of Aregs cells. Additionally, we verified IGFBP3
   KO in the Aregs of group IV using qPCR and immunoblotting. The results
   showed that, compared to group I and Aregs group, IGFBP3 mRNA and
   protein levels were significantly lower in IGFBP3 KO of Aregs,
   confirming successful KO of IGFBP3 ([157]Supplementary Figs. S4D, E,
   F). Next, we compared the expression levels of SMC-like cell markers
   (Myh11, Acta2 and Tagln) in the bottom layers of groups I, II, III, IV,
   and V. Strikingly, qPCR results showed that Acta2 and Tagln mRNA levels
   were significantly higher in the groups II, bottom layers of IV and V
   compared to groups I and bottom layers of III, although Myh11 did not
   show significant differences in all groups ([158]Figure 6C). To further
   clarify the pro-apoptotic effects of Aregs and their secreted factor
   IGFBP3 on SMC-like cells, we performed an Annexin V/PI apoptosis assay
   on ASPCs from each group. Flow cytometry results demonstrated that the
   apoptosis rates of ASPCs in Group II and the bottom layers of Groups IV
   and V were significantly lower than those in Group I and the bottom
   layer of Group III ([159]Figure 6C, D). Meanwhile, to further explore
   and validate the bioinformatics analysis results, we examined the
   activation of the TGFβ, MAPK, and p53 pathways in ASPCs from each
   group. Immunoblotting results revealed that the protein levels of
   P-SMAD2/3, P-p38, p53, and p21 were suppressed to varying degrees in
   Group II and the bottom layers of Groups IV and V compared to Group I
   and the bottom layer of Group III, indicating that these three
   signaling pathways were inhibited either independently or cooperatively
   when Aregs or their secretory factor IGFBP3 were removed
   ([160]Figure 6F–J).

Figure 6.

   [161]Figure 6
   [162]Open in a new tab

   Aregs and IGFBP3 regulate SMC-like cells apoptosis. (A–G) Primary ASPCs
   were isolated and cultured from sWAT of 9-week-old mice and used in
   subsequent experiments. (A) The flow cytometry sorting results, the
   left panel represents the negative control group (NC) and the right
   panel represents the experimental group (Exp). (B) Schematic diagram of
   transwell co-culture groups (Ab-IGFBP3 was added at a concentration of
   6 μg/mL ([163]Supplementary Figs. S6A–D) [[164]50]). (C) Myh11, Acta2
   and Tagln mRNA expression levels analyzed by qPCR. (D) Annexin V/PI
   staining followed by flow cytometry analysis. (E) Proportion of Annexin
   V+ apoptotic cells. (F) Immunoblots for P-SMAD2/3, SMAD2/3, P-p38, p38,
   p53, p21 and GAPDH. (G–J) Quantification of P-SMAD2/3 immunoblotting,
   normalized to SMAD2/3; P-p38 immunoblotting, normalized to p38; and p53
   and p21 immunoblotting, normalized to GAPDH. Data represent mean ± SD,
   ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

   Subsequently, all five groups underwent the same browning
   differentiation protocol. After differentiation, the results showed
   that the mRNA and protein levels of the beige adipocytes marker UCP1 in
   the groups II, bottom layers of IV and V were significantly higher than
   those in groups I and bottom layers of III ([165]Figure 7A, B, C).
   Another marker of beige adipocytes, high mitochondrial content, was
   also observed in the groups II, bottom layers of IV and V
   ([166]Figure 7B, D, E). In conclusion, our results demonstrate that
   Aregs and their secreted factor IGFBP3 in murine primary ASPCs play a
   significant role in inhibiting beige adipocytes formation, which may be
   a key factor contributing to the impaired sWAT browning capacity of
   aged mice.

Figure 7.

   [167]Figure 7
   [168]Open in a new tab

   Aregs and IGFBP3 inhibit beige adipocyte formation. (A–E) All five
   groups of ASPCs underwent the same browning differentiation protocol.
   (A) Ucp1 mRNA expression levels analyzed by qPCR. (B) Immunoblots for
   UCP1, TOMM20 and GAPDH. (C,D) Quantification of immunoblots of UCP1 and
   TOMM20, protein levels are normalized to GAPDH. (E) Immunofluorescence
   staining for CD142 and Mitotracker for mitochondria (scale
   bar = 20 μm). Data represent mean ± SD, ∗p < 0.05, ∗∗p < 0.01,
   ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

4. Discussion

   Our study demonstrates the crucial role of Aregs and their secreted
   factor IGFBP3 in the impaired browning capacity of aged sWAT under cold
   stimulation. We show that Aregs, which are upregulated in aged mice
   following cold exposure, secrete IGFBP3, which acts as a pro-apoptotic
   signal for SMC-like cells. This interaction between Aregs and SMC-like
   cells leads to apoptosis, impairs the function of SMC-like cells, and
   ultimately reduces the formation of beige adipocytes.

   Adipose tissue serves not only as an energy reservoir but also plays a
   crucial role in maintaining systemic health through the secretion of
   adipokines and extracellular vesicles [[169]51,[170]52]. Beige
   adipocytes, in addition to sharing these functions, promote lipid and
   glucose metabolism by dissipating excess energy as heat, a process
   mediated by UCP1. This enhanced metabolic activity helps mitigate or
   prevent metabolic diseases, including CVD, through mechanisms such as
   lowering blood triglycerides and non-HDL cholesterol levels, increasing
   HDL cholesterol levels, accelerating the clearance of triglyceride-rich
   lipoprotein (TRL) remnants, and secreting cardioprotective adipokines
   [[171]9,[172]10,[173]53]. These findings highlight the potential
   therapeutic value of targeting WAT browning in the management of
   metabolic diseases.

   Given the promising therapeutic potential of beige adipose tissue, an
   increasing number of researchers are exploring factors that can promote
   browning process from various perspectives, including pharmacological
   interventions, genetic modifications, and GPCRs [[174][54], [175][55],
   [176][56], [177][57], [178][58]]. However, the inevitable decline in
   browning capacity observed in aging individuals exacerbates metabolic
   dysregulation and disrupts homeostasis [[179]11,[180]12,[181]59]. This
   underscores the importance of understanding the molecular mechanisms
   behind adipose browning in aging individuals and how these processes
   can be leveraged for therapeutic interventions. WAT browning occurs
   through two primary mechanisms: one involves the trans-differentiation
   of white adipocytes into beige adipocytes, and the other involves the
   de novo differentiation of beige adipocytes from specialized progenitor
   cells, SMC-like cells [[182]60]. These SMC-like cells exhibit
   characteristics of both adipose progenitor cells and smooth muscle-like
   cells, playing a critical role in the direct formation of beige
   adipocytes [[183]3,[184]60,[185]61]. In practice, both mechanisms
   coexist, but the predominance of each varies depending on external
   conditions. Reports have shown that in mice maintained at ambient
   temperature (20–23 °C) before cold stimulation, trans-differentiation
   is predominant, whereas in mice preconditioned at thermoneutrality
   (30 °C) before cold exposure, de novo differentiation dominates the
   process [[186]62]. The publicly available scRNA-seq data and the mouse
   models used in our experiments were derived from TN-to-cold stimulation
   conditions, which enabled us to focus on the impact of SMC-like cells
   on browning capacity.

   Aregs, as a novel cell subtype in ASPCs, were first reported by
   Schwalie et al. and have since been primarily studied for their
   anti-adipogenic effects in white adipocytes [[187]17]. The mechanisms
   involved mainly focus on the paracrine actions of Aregs, with key genes
   such as Rtp3, Spink2, and Vitthe, as well as the retinoic acid pathway
   [[188]17,[189]19]. In our study, we provide the first evidence of an
   increase in Aregs cell numbers in aged individuals' sWAT under cold
   stimulation and, for the first time, report their inhibitory role in
   beige adipocytes formation. Our findings extend this understanding by
   showing that Aregs secrete IGFBP3 and act as an important mediator in
   the cross-talk between Aregs and SMC-like cells, emphasizing its
   potential as a therapeutic target for reversing age-related adipose
   dysfunction. However, we noted that Holman et al. did not emphasize
   changes in ASPC subpopulation proportions, including Aregs and SMC-like
   cells, despite our scRNA-seq raw data being sourced from their study
   (Holman et al., 2024, eLife) [[190]14]. To understand the origin of
   these discrepancies, we compared the analytical processes and found
   that differences in raw data processing, including the Cell Ranger
   version and reference genome selection, resulted in variations in cell
   clustering. Additionally, the strict filtering criteria employed by
   Holman et al. excluded moderately expressed cells, such as those from
   the aged-cold 14-day group, leading to a reduced number of analyzed
   cells. In contrast, we applied a two-stage quality control (QC)
   strategy, allowing us to obtain a larger and more comprehensive
   dataset. Regarding CD142+ ASPCs, Holman et al. reported normal
   adipogenesis, while other studies [[191]17,[192]19,[193]63,[194]64],
   including ours, demonstrate that Aregs inhibit adipogenesis. These
   discrepancies may stem from differences in experimental protocols. For
   instance, we exclusively used primary ASPCs isolated from young mice
   and differentiated them under higher insulin concentrations (850 nM).
   Moreover, during cell sorting, it is possible that a minor proportion
   of CD142+ cells with adipogenic potential (CD142+, ICAM1+, and VAP1+),
   as reported in a previous study [[195]63], were more or less
   inadvertently included.

   IGFBP3 influences the apoptosis of SMC-like cells through IGF-dependent
   or IGF-independent signaling pathways. Our analysis reveals that
   SMC-like cells in aged mice exhibit increased expression of apoptosis
   markers, such as Cdkn1a, and functional enrichment analysis highlights
   the activation of TGFβ and MAPK signaling pathways, both of which are
   known to induce apoptosis and fibrosis in adipose tissue
   [[196]42,[197]43,[198]65,[199]66]. Notably, our results suggest that
   IGFBP3 secreted by Aregs interacts with receptors and signaling
   molecules in SMC-like cells, including EGFR and Cdkn1a, thereby
   enhancing the apoptosis process [[200][67], [201][68], [202][69]].
   These findings provide a mechanistic link between Aregs and the reduced
   thermogenic capacity of aged adipose tissue. Additionally, our
   co-culture experiments demonstrate that IGFBP3 significantly inhibits
   the differentiation of beige adipocytes, which is consistent with the
   reduced expression of SMC-like cell markers (Acta2, Myh11, and Tagln),
   further confirming its role in the decline of sWAT browning capacity.

   It is worth mentioning, in addition to Aregs and SMC-like cells, there
   is another significantly altered cell subpopulation, ECs. ECs and
   SMC-like cells are reported to have a mutually regulatory role as the
   primary cell types in the intima and media of blood vessel walls,
   respectively, influencing the formation of vascularization [[203][70],
   [204][71], [205][72], [206][73], [207][74]], which is crucial for
   supporting the development of beige adipose tissue [[208]75,[209]76].
   Therefore, we conducted preliminary explorations, and our results
   indicated that despite a general reduction in ECs, a specific
   subpopulation of arterial endothelial cells (Mgp + ArtEC) was
   significantly upregulated in aged mice following cold exposure. Based
   on our initial findings, we hypothesize that changes in SMC-like cells
   may lead to a reduction in other endothelial cells through the
   upregulation of Mgp + ArtEC, ultimately impairing vascularization and
   affecting the formation of beige adipose tissue in aged mice
   ([210]Supplementary Fig. S7). This warrants further exploration in
   future studies.

   Although our study provides important insights into the roles of Aregs
   and SMC-like cells, several limitations remain to be addressed. First,
   while our research offers valuable understanding of the molecular
   mechanisms underlying impaired adipose browning during aging and has
   conducted preliminary explorations of the apoptotic signaling pathways
   involved, further studies are needed to pinpoint whether the TGFβ,
   MAPK, and p53 signaling pathways have synergistic, partially
   synergistic, or independent effects on the apoptosis of SMC-like cells.
   Second, the use of in vitro co-culture systems may not fully replicate
   the physiologically complex interactions occurring in vivo. Future
   studies utilizing lineage tracing models and conditional KO of CD142 or
   IGFBP3 in specific cell populations, as well as the development of
   peptides targeting CD142+ cells, will provide more definitive evidence
   of their role in adipose tissue aging and offer a more accurate
   understanding of the molecular mechanisms involved, including the
   signaling pathways. Moreover, the long-term effects of regulating Aregs
   or IGFBP3 on systemic metabolism and thermogenesis remain to be
   explored.

   In conclusion, our findings elucidate the key role of Aregs and IGFBP3
   in mediating the impaired browning capacity of aged sWAT under cold
   stimulation. By revealing the molecular mechanisms underlying this
   process, our study lays the foundation for developing therapeutic
   strategies to combat age-related metabolic dysfunction.

5. Conclusion

   In this study, we identified the critical role of Aregs and its
   secreted factor IGFBP3 in regulating the browning process of aged
   adipose tissue. Our results indicate that Aregs, through
   IGFBP3-mediated apoptosis of SMC-like cells, significantly impairs the
   formation of beige adipocytes in aged mice, leading to the loss of
   thermogenic capacity in aged WAT. These findings provide insights into
   the molecular mechanisms underlying the reduction in adipose tissue
   browning with aging, particularly from the perspective of de novo
   differentiation, and highlight Aregs and IGFBP3 as potential
   therapeutic targets for preventing or reversing age-related metabolic
   dysfunctions.

CRediT authorship contribution statement

   Shifeng Wang: Writing – original draft, Visualization, Validation,
   Resources, Methodology, Investigation, Formal analysis, Data curation.
   Yuanxu Cui: Writing – review & editing, Writing – original draft,
   Visualization, Investigation, Funding acquisition. Limei Wang: Writing
   – review & editing, Investigation, Funding acquisition. Chun Feng:
   Writing – review & editing, Investigation, Funding acquisition. Yifei
   Sun: Methodology, Formal analysis. Bangyun Huo: Investigation, Formal
   analysis. Honglu Jiang: Investigation, Formal analysis. Mingyu Zhao:
   Resources, Data curation. Yingying Tu: Resources. Qiyue Wang:
   Resources. Yutao Yang: Resources. Qiang Zhang: Writing – review &
   editing, Supervision, Resources, Methodology, Investigation, Funding
   acquisition, Data curation, Conceptualization.

Funding sources

   This research was funded by the Kunming Medical University (China) for
   High-Level Talent Research Supporting Fund (grant K132310515 to Q.Z.
   and grant K132310535 to Y.X.C.), Yunnan Provincial Department of
   Education Scientific Research Fund Project (grant 2025J0287 to Q.Z. and
   grant 2025J0255 to Y.X.C.), National Natural Science Foundation of
   China (grant 82460164 to L.M.W.) and The cultivation of the discipline
   team in Medical Experimental Animal Science at Kunming Medical
   University (grant 2024XKTDPY17 to Q.Z. and L.M.W.).

Declaration of competing interest

   The authors declare that they have no known competing financial
   interests or personal relationships that could have appeared to
   influence the work reported in this paper.

Acknowledgements