Insulin-stimulated glucose uptake is impaired in senescent human adipocytes

Anonymous samples of SAT were obtained as a surgical biopsy from the abdominal region of non-obese female individuals undergoing elective surgery at Sahlgrenska University Hospital in Gothenburg, Sweden. All study subjects received written and oral information before giving written informed consent for the use of their tissue. The study was approved by the Regional Ethical Review Board in Gothenburg, Sweden. A description of the patient cohort used in the study can be found in. 1

Human SAT was processed for adipose cell isolation as described previously (19, 20). After separating the infranatant containing the stromal vascular fraction (SVF) from the floating adipocytes, the SVF was pelleted by centrifugation (200xg, 7 min). To remove red blood cells, the pellet was incubated (RT, 5 min) in red blood cell lysing buffer (Sigma-Aldrich, Cat. R7757) followed by centrifugation (200xg, 5 min). The pellet was resuspended in subcutaneous basal medium (BM-1; ZenBio, BM-1), 10% FBS (Gibco, Cat. 10270-106), 1% penicillin and streptomycin (Pen/Strep), and 1 nM bFGF (Sigma-Aldrich, Cat. F0291). The cells were seeded at a density of 20,000 cells/cm² and expanded in flasks or cell factories. The cells were cultured at 37 °C with 5% CO₂ and 95% relative humidity.

Human preadipocytes from 10 female adults were isolated from human SAT surgical biopsies (n = 7) or purchased from Lonza (n = 3) as Human Adipose-derived stem cells from liposuction (Lonza, Cat. PT-5006). The preadipocytes were seeded for experimental use in subcutaneous preadipocyte medium (PM-1; ZenBio, PM-1) supplemented with 1 nM bFGF. Medium was changed the following day to PM-1 supplemented with 1 nM bFGF and 15 ng/ml BMP4 (R&D systems, Cat. 314-BP-010/CF). After becoming confluent (usually after 4–5 days), differentiation was initiated by changing the medium to subcutaneous basal medium (BM-1; ZenBio, BM-1), 3% FBS (Gibco, Cat. 10270-106), 1% Pen/Strep, 1 µM dexamethasone (Sigma-Aldrich, Cat. D2915), 0.5 mM IBMX (Sigma-Aldrich, Cat. I5879), 100 nM insulin (Actrapid Penfill; Novo Nordisk), and 1 µM pioglitazone (AstraZeneca compound management). Medium was changed on day 4. On day 7 of differentiation, the medium was changed but with the exclusion of IBMX and pioglitazone. On day ten of differentiation, a medium change was done and 5 µM nutlin-3a (Sigma-Aldrich, Cat. SML0580), 50 µM etoposide (Sigma-Aldrich, Cat. E1383) or 0.3 µM doxorubicin (AstraZeneca compound management) was added to the cells to induce senescence. The specified concentrations were chosen from pilot studies set up to define the concentrations to induce senescence without affecting cell viability in the human adipocytes (). Cytotoxicity was assessed by alamarBlue staining. Medium was changed again 4 days later. Further analyses were performed after 7 days of treatment, unless indicated differently. As a control, 0.1% dimethylsulfoxide (DMSO) was used. 1

For washout experiments, on day 7 of treatment with nutlin-3a, etoposide and doxorubicin, adipocytes were washed twice with PBS. The cells were thereafter cultured for 4 days in BM-1 supplemented with 3% FBS, 1% Pen/Strep, 1 µM dexamethasone, 100 nM insulin, until further processing.

For the acute insulin signaling experiments, adipocytes were washed twice with PBS and starved of insulin and serum for 4 hours in DMEM (Gibco, Cat. 31966021), 25 mM HEPES (Gibco, Cat. 15630056), and 0.1% BSA (Sigma-Aldrich, Cat. A6003). The cells were incubated with 1 or 10 nM of insulin for 10 mins at 37°C with 5% CO₂. Plates with adipocytes were put on ice, medium was removed followed by washing with cold PBS, and the cells were lysed in RIPA buffer for Western blot analysis.

Total RNA was isolated using the RNeasy Mini kits (Qiagen, Cat. 74106), following the manufacturer’s instructions. Complementary DNA (cDNA) was reverse transcribed, from RNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Cat. 4368814) and a heat cycler. Gene expression was measured using real-time PCR with SYBR Green PCR Master Mix (Applied Biosystems, Cat. 4309155) or TaqMan Fast Advanced Master mix (Applied Biosystems, Cat. 4444557) on a Quantstudio 7 Flex Real-Time PCR machine (Applied Biosystems). The following Real Time PCR protocols were used: For SYBR: polymerase activation (95°C, 10 min), followed by 40 cycles of denaturation (95°C, 15 s) and annealing/extension (60°C, 1 min), and TaqMan: polymerase activation (95°C, 20 s), followed by 40 cycles of denaturation (95°C, 1 s) and annealing/extension (60°C, 20 s). Tata-binding protein (TBP) was used as an internal normalization control. Primer sequences and probes are listed in. 1

To obtain protein extracts from differentiated adipocytes, cells were washed with PBS and lysed in RIPA buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EGTA, 1 mM EDTA, 1% Triton X-100, 0.1% SDS) containing protease inhibitors (Roche, Cat. 11836170001). For measurement of insulin signaling proteins, protease inhibitors, and PhosStop (Roche, Cat. 4906845001) were added to the lysis buffer. Lysates were snap-frozen and thawed rapidly at 90°C twice, and centrifuged (10,000xg, 5 min, 4°C). The clarified lysate was removed from lipids and pellet and was centrifuged (10,000xg, 5 min, 4°C) to remove all remaining carryover lipid. Protein concentration was measured using a BCA Protein Assay Kit (ThermoFisher Scientific, Cat. 23227). Proteins were separated on 4-12% Bis-Tris NuPAGE gels (Invitrogen) and transferred to PVDF membranes (Invitrogen, Cat. LC2005). Membranes were blocked for 60 min at RT in 1X TBST with 5% w/v nonfat dry milk and incubated overnight at 4°C with primary antibodies. Membranes were washed and incubated with secondary antibodies for 1 hour at RT. Bands were detected using ECL Western Blotting Substrate (Pierce, Cat. 32106) or SuperSignal West Femto (ThermoFisher Scientific, Cat. 34096) in a ChemiDoc MP Imaging system (Bio-Rad). Image Lab version 6.0.1 software (Bio-Rad) was used to quantify the bands. Quantified protein levels were normalized to β-actin, and phosphorylated protein levels are normalized to β-actin or respective total protein. Primary antibodies are listed in Supplementary Table 3.

Assessment of SA-βgal activity was performed using a cellular senescence activity kit (Enzo Life Sciences, Cat. ENZ-KIT129-0120) according to the manufacturer’s instructions. In brief, cell lysates were incubated with SA-βgal substrate for 2 hours at 37°C. Fluorescence was read at 360 nm (excitation)/465 nm (emission) with a CLARIOstar (BMG LABTECH) plate reader. Protein concentration was measured in the cell lysates using a BCA Protein Assay Kit. Recorded relative fluorescence units were normalized to protein concentration.

Cell culture media was collected on day 7 of treatment with nutlin-3a, doxorubicin, and etoposide, and was filtered through a 0.2 µm syringe filter. The Olink^® Target 96 inflammation panel (Olink Proteomics AB), which enables simultaneous analysis of 92 analytes in 1 µL of sample, was used following manufacturer’s instructions. Out of 92 analytes, 39 were detected in at least 75% of the samples. Using statistical analyses, 32 analytes were found to display a significant difference between the groups, P < 0.05 (Friedman test), and were selected for further analysis. For the hierarchical clustering of 24 samples (six donors; four treatment groups), the R-package pheatmap (version:1.0.12) was used. Averaged, centered, and scaled values were used as input, rows (proteins) were ordered based on euclidean distances using the clustering method “complete”. For a detailed list, see Supplementary Table 4. Adiponectin levels were determined using an ELISA kit (R&D systems, Cat. DRP300) according to the manufacturer’s instructions.

Lipolysis activity was measured by quantifying glycerol in cell culture medium after stimulation with forskolin or isoproterenol. Human preadipocytes were seeded, differentiated, and treated with nutlin-3a, doxorubicin and etoposide in 96-well plates (Corning, Cat. 3595). On day 7 of treatment with nutlin-3a, doxorubicin, and etoposide, human adipocytes were washed and incubated (37°C, 5% CO₂, 3 hours) in sterile-filtered serum- and insulin-free medium (DMEM [Gibco, Cat. 31966], 1% BSA [Sigma-Aldrich, Cat. A8806], 1% Pen/Strep). Medium was changed to KRHB buffer (120 mM NaCl, 4.7 mM KCl, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄·7H2O, and 2.5 mM CaCl₂·2H₂O, 25 mM HEPES, 2 mM glucose, 1% BSA [Sigma-Aldrich, Cat. A6003], pH 7.4). Lipolysis was stimulated by adding forskolin (Sigma-Aldrich, Cat. F6886, 5 concentrations, 1:10 dilutions, final conc. 0.01-100 µM) or isoproterenol (Sigma-Aldrich, Cat. 16504-100MG, 5 concentrations, 1:10 dilutions, final conc. 0.01–100 nM) using a Biomek FX autosampler workstation (Beckman Coulter). The adipocytes were incubated for 2 hours (37°C, 5% CO₂). The Biomek FX autosampler workstation was used to transfer 14 µL of medium in four replicates to a 384-well plate (Greiner Bio-one, Cat. 781101) containing a 6.25-200 µM glycerol standard curve (Sigma-Aldrich, Cat. G7793). Sixty microliters of Free Glycerol Reagent (Sigma-Aldrich, Cat. F6428) was added to each well followed by incubation (10 min, RT). The absorbance was analyzed at 540 nm using a Paradigm™ Detection Platform (Beckman Coulter). Quantified glycerol was normalized by total cell number (Hoechst staining).

Human preadipocytes were differentiated and treated with nutlin-3a, doxorubicin, and etoposide in CytoStar-T 96-well plates (Perkin Elmer, Cat. RPNQ0162). The adipocytes were washed twice and incubated (37°C, 5% CO₂, 3 hours) in serum- and insulin-free media (70% DMEM [Gibco, Cat. 31966], 27% DMEM [Gibco, Cat. 11966], 25 mM HEPES [Gibco, Cat. Cat. 15630056], 0.1% BSA [Sigma-Aldrich, Cat. A6003]). Medium was changed to assay medium (DMEM [Gibco, Cat. 11966], 25 mM HEPES [Gibco, Cat. 15630056], 0.1% BSA [Sigma-Aldrich, Cat. A6003], 2 nM Sodium pyruvate [Gibco, Cat. 11360070]). Glucose uptake was stimulated by adding insulin (Novo Nordisk, 9 concentrations in 1:4 dilutions, final conc. 0.015 nM-1 µM) in 8 replicates using a Biomek FX autosampler workstation (Beckman Coulter) followed by incubation (37°C, 5% CO₂, 45 min). Non-specific binding controls received cytochalasin B (Sigma-Aldrich, Cat. C2743, final conc. 20 µM). Twenty-five microliters of 2.5 μCi/ml 2-[1-¹⁴C]-Deoxy-D-Glucose (Perkin Elmer, Cat. NEC495A250UC) and the cells were incubated (37°C, 5% CO₂, 10 min). Glucose uptake was stopped by addition of cytochalasin B (final conc. 20 µM), and the plate was read using a Microbeta2 plate reader (Perkin Elmer). Glucose uptake was normalized by total cell number (Hoechst staining). Cytochalasin B control values were subtracted.

For fluorescent analysis of ɣH2AX, p21 and lipid accumulation, human adipocytes were cultured and treated as previously described in 96-well plates (PerkinElmer, Cat. 6055302). Adipocytes were washed twice with PBS and fixed with 4% paraformaldehyde for 20 min at RT followed by washing thrice with PBS. For analysis of ɣH2AX and p21, cells were permeabilized using 0.1% Triton X-100 in PBS for 10 min at RT. After washing thrice with PBS, adipocytes were blocked in 2% BSA in PBS for 30 min at RT. The cells were incubated with primary antibody in 1% BSA/PBS for 60 min. After washing thrice with PBS, adipocytes were incubated with secondary antibody in 1% BSA/PBS for 60 min at RT and washed thrice with PBS. Adipocytes were counterstained with Hoechst (1:5000 in 1% BSA/PBS, ThermoFisher Scientific, Cat. 33342) for 30 min at RT followed by washing thrice with PBS. For fluorescent analysis of lipid accumulation, fixated cells were left to stain for 20 min in PBS containing 0.1 µg/ml BODIPY 493/503 (Invitrogen, Cat. D3922) and Hoechst (1:5000) followed by washing thrice with PBS. Images were acquired with a robotic Yokogawa CV8000 (ɣH2AX and BODIPY) or Yokogawa CV7000 spinning disc confocal microscope (Wako Automation) at 20× magnification (NA.75, 2 × 2 binning) using ZYLA 5.5 sCMOS cameras Andor Technology). Images were acquired with 16 bits image depth and a resolution of 1000 × 1000 (ɣH2AX and BODIPY) or 1280 x 1080 (p21), with a pixel dwell of ~1.02 μs. ɣH2AX, p21, and BODIPY intensity was quantified using Columbus 2.9.1.532 software (PerkinElmer). Primary and secondary antibodies are listed in. 1

Results are presented as mean ± SEM. The number of replicates, shown as dots, represents individual donors and are stated in figure legends of each measurement. For comparisons between three or more groups, a Friedman test for non-parametric analysis with the assumption of non-normally distributed data was performed, followed by Dunn’s multiple comparisons post-hoc test between groups. When assuming normal distribution, parametric tests were used. Multiple paired t-test (one per row) with False Discovery Rate approach were used to compare glucose uptake and lipolysis data between two groups. Two-way ANOVA was used to test for a significant effect of two factors and the interaction, followed by Dunnett’s multiple comparisons post-hoc test between groups. P < 0.05 was considered statistically significant. EC50 was calculated using GraphPad Prism 8.4.3 software (GraphPad Software). GraphPad Prism 8.4.3 software was used for statistical analysis.

To induce cellular senescence, we treated human adipocytes with compounds that have previously been reported to induce senescence in other cell types (21). Human preadipocytes were differentiated for 10 days prior to treatment with a p53 activator (5 µM nutlin-3a, Nutlin) and two DNA-damaging agents (0.3 µM doxorubicin, Dox; 50 µM etoposide, Etop) for 7 days. Nutlin and Dox induced expression of senescence markers p21, p53 and MDM2 (Figures 1A, B). Nutlin induced a 7-fold increase of p21 protein levels, and a 40-fold and 380-fold increase of MDM2 and p53 protein levels, respectively. Although not statistically significant, the expression of senescence markers after Etop treatment exhibited a pattern similar to those observed with Nutlin and Dox treatment. In addition, treatment with Dox, Etop, and Nutlin increased the activity of Senescence-associated β-galactosidase (SA-βgal) in the adipocytes by 1.4- to 2-fold (Figure 1C). Furthermore, all three compounds at the given concentrations (especially Nutlin and Dox) caused morphological changes in the human adipocytes, which became more elongated and fibroblast-like (Supplementary Figure 1B).

DNA damage is a trigger of senescence which activates a DNA damage response (5). We investigated the expression of Ser139 phosphorylated H2AX (ɣH2AX) – a marker of double-stranded DNA breaks (22). Immunostaining revealed that ɣH2AX expression was 1.6- and 1.3-fold higher in cells treated with the DNA damaging agent Dox and Etop, respectively (Figures 1D, E), in addition to confirming the elevated expression of p21 in treated cells (Figures 1F, G). The number of p21-positive adipocytes increased from 20% in control cells to 60-80% in adipocytes treated with Dox, Etop and Nutlin (Figure 1G). To validate that the induced senescence phenotype was stable compounds were washed out at day 7 of treatment. Indeed, the number of p21-positive cells was still around 70% four days after compounds had been washed out (Supplementary Figure 1C). Taken together, these findings demonstrate that induction of senescence in human adipocytes can be achieved following treatment with Dox, Etop, and Nutlin.

To study the SASP of senescent human adipocytes, the composition of inflammatory proteins was assessed in cell media from senescent adipocytes using a targeted inflammatory proteome platform. Proteins that displayed a statistically significant difference (Friedman test) between treatment groups are shown in Figure 2 and are clustered based on similar responses to treatment. Although the strongest response was observed in Dox-treated adipocytes, increased secretion of inflammatory proteins could be detected in cell media from adipocytes treated with all three compounds (Figure 2). The senescence-inducing compounds increased the secretion of several known SASP factors, including IL6, CCL2, CCL3, CCL7, CCL8, CCL13, CXCL1, CXCL5, KITLG, CSF1, MMP1, and MMP10 (18, 23–26). Other factors that were increased, including CASP8, ADA and FLT3LG, have not, to the best of our knowledge, been appointed as SASP previously. Furthermore, secretion of other proteins, including FGF5, TGFB1, OPG and the angiogenic factor VEGFA were decreased in senescent adipocytes.

To determine if cellular senescence has an impact on lipolysis, human adipocytes treated with Dox, Etop, and Nutlin for 7 days were exposed to isoproterenol or forskolin. Isoproterenol stimulates lipolysis by binding and activating β-adrenergic receptors, while forskolin stimulates adenylyl cyclase leading to increased lipolytic activity (27, 28). Lipolysis was determined as the release of glycerol in cell culture media. Isoproterenol stimulation yielded a numerically lower EC50 in senescent adipocytes compared with control cells, whereas this was not consistently observed with forskolin stimulation (Figures 3A, B). Maximal stimulated lipolysis, both with forskolin and isoproterenol stimulation, trended to be slightly lower in senescent adipocytes compared to control cells. No significant difference in basal or stimulated lipolysis was found between controls and Nutlin- and Etop-treated cells using isoproterenol (Figure 3A) or forskolin (Figure 3B). However, Dox-treated adipocytes displayed a reduction in glycerol release with both forskolin and isoproterenol stimulation compared to control cells. The lower lipolysis by Dox-treated adipocytes is most likely explained by a reduction in basal lipolysis. These results suggest that the sensitivity to stimulation of lipolysis and lipolytic capacity of senescent adipocytes remain largely unaffected.

Next, we investigated if cellular senescence impacts basal and insulin-responsive glucose uptake in adipocytes by measuring uptake of a ¹⁴C-labeled glucose analogue, deoxy-D-glucose. To stimulate glucose uptake, we exposed Dox-, Etop-, and Nutlin-treated adipocytes to insulin for 45 min. As expected, insulin stimulation resulted in a dose-dependent increase in glucose uptake in control cells with an EC₅₀ of ~2 nM (Figure 4). Whereas the basal glucose uptake of control and senescent cells were similar, insulin-stimulated glucose uptake was markedly decreased in senescent adipocytes. Treatment with Dox-, Etop-, and Nutlin caused a reduction of the maximal insulin-stimulated glucose uptake by 66-82% compared to control treatment. Thus, these results indicate that insulin-stimulated glucose uptake in senescent human adipocytes is impeded.

Reduced insulin-stimulated glucose uptake could be a consequence of reduced insulin receptor signaling. Key insulin signaling proteins and their phosphorylation were measured in human adipocytes treated with the senescence-inducing compounds, followed by insulin stimulation for 10 min (Figure 5A). Expression of AKT, insulin receptor β (IRβ) and P70S6K could be detected in samples from both control and senescent cells, as well as phosphorylated levels of IRβ (Tyr1150/1151) and AKT (Ser473 and Thr308) (Figure 5A). The ratio of phosphorylated levels of AKT (Ser473 and Thr308) and IRβ (Tyr1150/1151) over total protein were similar between control and senescent cells at 1nM insulin stimulation, indicating preserved receptor sensitivity to insulin. However, incubating Dox-treated cells with 10 nM of insulin yielded a reduced pIRβ response (Figure 5B). Dox-treatment reduced the total protein levels of both IRβ and P70S6K (Figure 5C). Investigating the phosphorylated proteins with β-actin ratio showed that dox-treated adipocytes had a 76% reduction in phosphorylation of AKT(Ser473) and 83% in phosphorylation of IRβ (Tyr1150/1151) at 10 nM of insulin (Figure 5D). This finding indicates that the phosphorylation ratio in Dox-treated adipocytes visualized in Figure 5B, is overestimating the total amount of phosphorylated signaling proteins in dox-treated cells. Together, these results indicate that the insulin action capacity in senescent adipocytes induced by dox is reduced, while the receptor signaling sensitivity to insulin is generally preserved.

GLUT4 is the glucose transporter mediating insulin-responsive glucose uptake. We therefore investigated GLUT4 expression and found that treatment with Dox, Nutlin and Etop reduced the expression of GLUT4 mRNA (encoded by SLC2A4) by 70 to almost 100%, and GLUT4 protein level by 60 to almost 100% (Figures 6A, B). The marked decrease in GLUT4 expression likely contributes to the reduced insulin-stimulated glucose uptake in senescent adipocytes.

Given the reduced GLUT4 mRNA and protein expression, and the slightly lower total expression of insulin-responsive proteins, especially IRβ, we hypothesized that the adipocytes could be partly dedifferentiated as they become senescent. To investigate this further, the expression of adipocyte markers was measured. At the mRNA level, Dox treatment reduced the expression of adiponectin (ADIPOQ) and perilipin1 (PLIN1) (Figure 6C). Both Dox and Nutlin reduced the mRNA levels of PPARɣ2 - a key regulator of adipogenesis, while Nutlin and Etop decreased levels of leptin (LEP) (Figure 6C). Expression of ATGL (PNPLA2) and HSL (LIPE), two key lipolytic enzymes, were reduced by both Nutlin and Dox-treatment (Figure 6C). At the protein level, Dox but not Nutlin and Etop treatment lowered expression of PPARɣ2 (Figures 6D, E). Furthermore, secreted levels of adiponectin were found to be reduced by 65-87% in senescent human adipocytes treated with Nutlin and Dox compared to controls (Figure 6F). Last, we used imaging to evaluate whether senescence impacted adipocyte lipid content and lipid droplet morphology as measured by lipid droplet number and size (Figures 6G–I). No difference in lipid droplet number or size was observed between senescent adipocytes and control cells. These findings suggest that some, but not all, characteristics of the adipocyte phenotype are altered by senescence.

The original contributions presented in the study are included in the article/. Further inquiries can be directed to the corresponding author. 1