Deletion of Trim28 in committed adipocytes promotes obesity but preserves glucose tolerance

Obesity leads to metabolic complications including insulin resistance, type 2 diabetes and NAFLD, driven in part by ectopic lipid deposition when adipocytes become hypertrophic and dysfunctional. Promoting healthy WAT expansion via adipocyte hyperplasia can preserve metabolic health. PPARγ is a master regulator of adipogenesis, but many pathways converge to regulate adipocyte function. TRIM28 (KAP1/TIF1β), a transcriptional corepressor with SUMO E3 ligase activity, has been genetically associated with increased fat mass in mice and humans. Global Trim28 haploinsufficiency in mice exhibits variable obesity with reports of both metabolic syndrome and metabolically healthy obesity, potentially via an imprinted gene network modulated by environment. Prior limited data suggested post-developmental adipocyte-specific Trim28 deletion might not recapitulate obesity, implying a developmental mechanism. The research question: does deletion of Trim28 in committed (post-developmental) adipocytes alter adiposity and metabolism, and are there sex-specific effects and molecular mechanisms (e.g., lipolysis, TG metabolism, Klf14) underlying these phenotypes?

Previous studies identified Trim28 as a corepressor interacting with KRAB-ZFPs and involved in autophagy regulation. Global Trim28 haploinsufficient mice showed increased fat mass and variable metabolic outcomes; some studies reported glucose intolerance and metabolic syndrome, others found metabolically healthy obesity linked to alterations in an imprinted gene network (e.g., Peg3, Nnat) and environmental triggers. Dalgaard et al. proposed that post-developmental adipocyte-specific deletion did not induce obesity, supporting a developmental origin for the phenotype. KLF14 has been implicated as a female-specific master regulator of adipocyte gene networks influencing adipocyte size and T2D risk. DJ-1 (Park7) has been linked to suppression of adipocyte β-oxidation. ELOVL family enzymes (e.g., Elovl3) regulate elongation of saturated/monounsaturated fatty acids and adipose lipid recruitment. These studies frame the investigation into whether adipocyte-specific Trim28 deletion affects adiposity via lipid metabolism pathways and sex-specific regulators like Klf14.

Animals: Conditional adipose-specific Trim28 deletion was achieved by crossing Trim28 floxed mice (C57BL/6J) with AdipoQ-Cre mice to generate Trim28fl/fl;AdipoQ-Cre+/- (adi-KO) and Trim28fl/fl;AdipoQ-Cre-/- (WT) littermates. Both sexes were studied. Mice were housed at 22°C, 12 h light/dark, chow ad libitum; acclimatized 2 weeks. Chow cohorts studied to 24 weeks of age. High-fat diet (HFD; 43% energy from fat, SF04-001) was administered from 6 weeks for up to 18 weeks. Groups were age- and sex-matched, randomly allocated. Ethics approval E/1618/2016/B. Genotyping and deletion validation: WAT immunoblotting confirmed significant reduction of TRIM28 protein in male and female adi-KO, with residual expression attributed to non-adipocyte cells. qPCR showed reduced Trim28 mRNA in WAT and BAT, not in quadriceps or liver. Phenotyping: Body mass and composition (lean mass, fat mass) measured longitudinally by EchoMRI 4in1. Glucose metabolism assessed via fasting blood glucose and oral glucose tolerance test (OGTT) at 6, 10, and 22 weeks (chow) or 0, 4, 16 weeks on HFD. OGTT dose 2 g/kg by body weight; an additional OGTT in HFD females at 24 weeks used 2 g/kg by lean mass (LM) to control for dosing. Insulin tolerance tests (ITT; 1 U/kg by LM) performed after 12 weeks HFD (18 weeks age) following 5 h fast. Plasma insulin measured by ultrasensitive ELISA (ALPCO). Blood glucose measured at baseline and 15–120 min time points post gavage/injection. Energy expenditure and respirometry: CLAMS metabolic cages used at 11 weeks (5 weeks post diet) with 72 h housing, last 24 h analyzed. VO2, VCO2, respiratory exchange ratio (RER), energy expenditure (EE), and activity (beam breaks) recorded. EE analyzed by ANCOVA with LM and total body mass as covariates. Lipidomics: ESI-MS/MS lipidomics on plasma, liver, and gonadal WAT from chow-fed female WT and adi-KO (male subset also analyzed). Modified Folch extraction; MassHunter quantification. Triglyceride subclasses analyzed by total acyl carbon length and unsaturation; free fatty acids measured by WAKO NEFA kit and ESI-MS/MS. In vitro adipocyte studies: 3T3-L1 preadipocytes differentiated (DMEM + 10% FBS; induction with insulin 20 nM, GW1929 50 nM, IBMX 0.5 mM, dexamethasone 1 μM for 48 h; then DMEM + 10% FBS + insulin 20 nM up to 10 days). Stable Trim28 knockdown (shTrim28; TRCN0000302256) and control (shLuc; SHC002V) via lentiviral transduction (MOI 20, polybrene 10 μg/mL), puromycin selection (4 μg/mL, 4 days). Lipolysis assay at day 8: serum-free media 6 h ± isoproterenol (0, 0.5, 1, 5 μM); glycerol in media normalized to protein. Microscopy performed on day 10 cells. Protein and gene analyses: SDS-PAGE/immunoblotting for TRIM28, HSL, pHSL-Ser563, ACC, DJ-1 (Park7), pAkt-Ser473, tAkt, pGSK3β-Ser9, tGSK3β, PPARγ, C/EBPα, 14-3-3, β-actin. Densitometry normalized to loading controls. qPCR using SYBR Green; normalization to Cyclophilin A (Ppia); primers span exon junctions; delta-CT method. RNA-seq: RNA from gonadal WAT (chow-fed male and female WT and adi-KO; n=6/group males, n=5 females after QC). Libraries prepared with Kapa Stranded RNA-seq; sequenced on Illumina HiSeq 2500 (single-end, ~35–50 bp). Reads <45 bp removed (0.3–8%). STAR aligner (mm10); featureCounts for gene counts. Low-expression filter (<30 CPM average) removed; 12,940 genes retained. Differential expression: DESeq2 and iDEP; FDR correction (Benjamini-Hochberg). Enrichment: DAVID, GSEA, MSigDB C5. Data deposited at ArrayExpress E-MTAB-9809. In silico PPI modeling: STRING-based human PPI datasets for KLF14 and TRIM28 queried; two shells of interaction. Overlap identified common interactors (CREBBP, EP300, SIN3A, HDAC1, PHF12, NCOR1, TP53; plus SUDS3, MXD1). SIN3A was a direct interactor with both. Statistics: Mean ± SEM. Repeated measures two-way ANOVA or mixed-effects for longitudinal data; ANCOVA for EE; Student's t-tests or ANOVA with Fisher’s LSD for group comparisons. Significance p<0.05. Predefined exclusion criteria applied (e.g., technical failures, implausible values, Tukey outliers).

• TRIM28 expression dynamics: In 3T3-L1 adipogenesis, TRIM28 protein and mRNA were high in precursors and progressively decreased, lowest by day 6 post-differentiation. In HMDP mouse panel, adipose Trim28 mRNA negatively correlated with fat-related traits (e.g., gonadal fat percent r≈-0.364, adjusted p≈2.9×10^-4).
• Validation of adipocyte-specific deletion: WAT TRIM28 protein significantly reduced in adi-KO males and females; Trim28 mRNA reduced in WAT and BAT but not quadriceps or liver.
• Chow diet (24 weeks): Body weight similar across genotypes. Fat mass increased in adi-KO: males showed a trend from ~8 weeks; females had significant FM increase (genotype p=0.0034; genotype×time p=0.0215) from ~8 weeks onward. Lean mass largely unchanged in absolute terms; LM%BW differences driven by increased FM. Fasting glucose tended to be lower in male adi-KO at all time points and was significantly lower at study end; females showed lower fasting glucose at study end. OGTTs showed no significant genotype differences at 6, 10, or 22 weeks; females trended toward improved tolerance despite higher FM. Fasting insulin unchanged; liver insulin signaling (pAkt-Ser473) suggested increased sensitivity (trend).
• High-fat diet (16 weeks): Body weight increased in both sexes. Males: no significant total body weight difference, but FM significantly higher in adi-KO across the study (genotype p=0.017; genotype×time p=0.025). Females: adi-KO heavier than WT across study (genotype p=0.024; genotype×time p<0.0001) with significantly higher FM (genotype p=0.0012; genotype×time p<0.0001). Fasting glucose unchanged by genotype. At 16 weeks HFD, apparent OGTT impairment in adi-KO was eliminated when glucose dosing was normalized to LM; ITT after 12 weeks HFD showed no genotype differences.
• Energy metabolism: Males showed no genotype differences in RER or EE. Females: chow-fed adi-KO had significantly lower RER during light period (p<0.05), indicating greater resting fatty acid utilization; HFD-fed females showed lower RER in light and higher at night, suggesting reduced metabolic flexibility. Female adi-KO displayed higher EE patterns independent of LM (ANCOVA p=0.018 chow; p=0.04 HFD) and body mass (p=0.007 chow; p=0.0008 HFD), particularly pronounced in smaller HFD-fed animals. Activity was similar overall with trends toward reduced dark-period activity in adi-KO.
• Lipidomics (chow-fed females): Total WAT TG unchanged; liver/plasma TG showed consistent but nonsignificant reductions. Specific WAT TG species altered with a shift toward more saturated species and fewer polyunsaturated species, notably within TG(54:x): increases in TG(54:0)–TG(54:2) and decreases in TG(54:7). Liver and plasma showed an overall reduction in short-chain TGs (TG48:x–TG54:x), consistent with improved hepatic lipid profile. Trends partially recapitulated in males.
• Lipolysis and lipid handling: In WAT of HFD-fed females, pHSL-Ser563 was significantly reduced in adi-KO, consistent with decreased HSL-mediated lipolysis; ACC protein abundance was significantly reduced. Insulin signaling in WAT (pAkt-Ser473; pGSK3β-Ser9) unchanged, indicating insulin-independent effects. Park7 (DJ-1) mRNA and protein were significantly increased. Plasma NEFA showed ~35% reduction trend (p=0.1) with selective decreases in lower abundance FFAs (e.g., 14:0, 16:1, 20:2, 20:3). In vitro, Trim28-depleted 3T3-L1 adipocytes exhibited significantly reduced basal and isoproterenol-stimulated glycerol release and reduced pHSL-Ser563.
• Gene expression networks: IGN1 genes Peg3 and Nnat were not altered in WAT of adi-KO, suggesting the phenotype is not driven by this imprinted network. Elovl3 mRNA was markedly upregulated (up to ~100-fold) across adipose depots and sexes/diets. Other browning markers were not broadly increased. RNA-seq revealed differential expression enriched for adipogenesis, differentiation, and fatty acid metabolism pathways in adi-KO. Elovl family members (Elovl6, Elovl7), Lpl, Pck1, and Park7 were altered. Olr1 was upregulated ~30–40-fold; Klf14 expression was almost completely ablated in both sexes’ WAT.
• Sex-specific effects: Female mice exhibited more pronounced increases in adiposity, altered RER and EE patterns, and near-complete loss of Klf14, consistent with known female-specific roles of KLF14 in adipocyte biology.

Deleting Trim28 specifically in committed adipocytes increases adiposity without impairing glucose tolerance, directly addressing whether Trim28 acts post-developmentally to regulate adipose function. The phenotype, more pronounced in females, mirrors aspects of global haploinsufficiency while avoiding developmental confounds, indicating a bona fide role for Trim28 in mature adipocytes. Mechanistically, Trim28 loss reduces adrenergic-stimulated lipolysis (lower pHSL-Ser563), alters triglyceride composition toward saturation in WAT, and reduces short-chain TGs in liver/plasma, suggesting a reprogramming of lipid handling that may protect against ectopic lipid accumulation. Elevated DJ-1 (Park7) may contribute to reduced fatty acid oxidation and favor storage. Upregulation of Elovl enzymes aligns with observed TG species shifts and may promote adipose expansion. Crucially, Klf14 expression was nearly abolished in adi-KO WAT, providing a plausible explanation for female-biased adiposity and altered adipocyte function, given KLF14’s established female-specific regulatory role in human and rodent adipocytes. Olr1 upregulation may further influence lipid metabolism pathways. Insulin signaling in WAT remained unchanged, and whole-body glucose homeostasis was preserved even with increased fat mass, supporting the concept of metabolically healthy adiposity when adipose tissue can adaptively store lipids. Changes in RER and EE patterns in females indicate reduced metabolic flexibility, potentially reflecting impaired substrate switching in adipocytes lacking Trim28. Collectively, these findings extend Trim28’s role beyond development, positioning it as a key modulator of adipocyte lipid metabolism and sex-specific adiposity.

Adipocyte-specific Trim28 deletion promotes increased adiposity, particularly in females, while preserving glucose tolerance. The phenotype is associated with insulin-independent reductions in lipolysis (reduced HSL phosphorylation), altered triglyceride composition with increased saturation in WAT, decreased short-chain TGs in liver/plasma, and transcriptomic changes implicating fatty acid elongation and lipid metabolism. Notably, Trim28 loss profoundly reduces Klf14 expression, suggesting Trim28 is necessary for Klf14 expression in adipocytes and providing a mechanistic link to female-specific adiposity. These results redefine Trim28 as a post-developmental regulator of adipocyte function and lipid handling. Future studies should: (1) delineate direct molecular mechanisms linking TRIM28 to KLF14 transcriptional control (e.g., TRIM28–SIN3A–HDAC1 complexes), (2) assess adipocyte-specific Klf14 loss-of-function to test causality for the sex-specific phenotype, (3) quantify food intake and energy balance components, and (4) explore therapeutic modulation of TRIM28–KLF14 axis and lipid metabolism enzymes (e.g., ELOVLs) to promote metabolically healthy adiposity.

• Food intake was not measured, limiting interpretation of energy balance contributions to increased adiposity.
• OGTT dosing by total body weight initially confounded glucose tolerance comparisons; LM-adjusted dosing corrected this but indicates sensitivity of results to dosing strategy.
• Residual Trim28 expression in WAT likely reflects non-adipocyte cells; AdipoQ-Cre targets committed adipocytes but not stromal/progenitor or immune/vascular cells.
• Mechanistic link between TRIM28 and Klf14 expression is correlative; direct binding or complex formation remains speculative and requires validation.
• Primary phenotyping was in mice; translation to humans, while supported by genetic data on KLF14, requires further study.
• Browning marker analysis suggests no broad beigeing; depot-specific effects or temporal dynamics may have been missed.
• Some lipid changes and NEFA differences were trends rather than statistically significant; study powered primarily for adiposity phenotypes.