We enrolled 28 C57BL/6 male mice aged 3 weeks weighting 14 (median)±2 g, purchased from experimental animal center of Shandong University, Ji’nan, China. All procedures were approved by Animal Care and Use Committee of Shandong University. Animals group-housed 3 per cage under standard laboratory conditions (25°C, 60%–70% relative humidity and 12 h of light and dark circulation) were tested for intraperitoneal glucose tolerance tests (IPGTT) after 1 week of adaptation.

After IPGTT, 28 mice were randomly assigned to the control group (Ctrl, n =14) and type 2 diabetes group (T2DM, n =14). The control group were fed a normal chow while the T2DM group received a high-fat diet (HFD, 34.5% fat, 17.5% protein, and 48% carbohydrate; Beijing HFK Bio-Technology, China). After 6 weeks, insulin tolerance was evaluated by intraperitoneal insulin tolerance tests (IPITT). After 8 weeks, IPITT and IPGTT were performed again to confirm the induction of IR. Mice exhibiting IR were given once low-dose of streptozotocin (STZ) injection intraperitoneally (Sigma, St. Louis, MO; 75–80 mg/kg i. p. in 0.1 mol/L citrate buffer, pH 4.5); the control group was administered with citrate buffer intraperitoneally only. IPGTT was measured again 2 weeks after injection of STZ. Most HFD/STZ-treated mice exhibited hyperglycemia (random blood glucose >11.1 mmol/L), IR and abnormal glucose tolerance.

At the age of 14 week, mice with similar degrees of hyperglycemia and body weight were randomly divided into the vehicle group (T2DM, n = 7) and PTPN2-overexpression group (T2DM + PTPN2, n = 7). Mice fed on a normal diet were classified as the non-diabetic control group, divided into the vehicle group (Ctrl, n = 7) and PTPN2-overexpression group (Ctrl + PTPN2, n = 7).

After food intervention, body weight was recorded every week and blood glucose level was examined every 2 weeks. The fasting glucose level was measured by OneTouch Glucometer (LifeScan, Milpitas, CA) following 6 h of fasting. IPGTT and IPITT were conducted according to the previously reported method (Li et al., 2019; Fukumura et al., 2021).

A recombinant pAdxsi adenovirus constitutively expressing PTPN2 systemically was constructed using the pAdxsi adenovirus system (SinoGenoMax, Beijing, China). Mouse-derived PTPN2 cDNAs were inserted into the pShuttle-CMV-EGFP vector and the pAdxsi vector adenovirus was used as a control vector. At 20 weeks mice were treated with 5 × 10⁹ plaque-forming units of virus by caudal intravenous injection. Adenovirus re-transferred at 22 weeks and the control group was injected with control virus (vehicle). All mice were euthanized at 24 weeks of age for further study as below.

Samples were taken from epididymal, subcutaneous and brown adipose tissue (SAT, VAT, and BAT) and each sample was cut into 2 parts on average, one of which was fixed with 4% paraformaldehyde, embedded with paraffin and sliced into 5um-thick sections. The morphology of adipocytes was analyzed from sections staining with hematoxylin and eosin (H&E). Adipocyte areas were assessed under X400 magnification within adipose tissue, and the number of adipocytes per area was obtained by quantitative morphometry with automated image analysis (Image-Pro Plus, Version 5.0; Media Cybernatics, Houston, TX) to assess the size of adipocytes.

Paraffin sections were treated with immunohistochemistry using microwave-based antigen retrieval method. These sections were incubated with primary rabbit polyclonal antibodies at 4°C overnight, and then with a matching biotinylated secondary antibody for 20 min at 37°C. The stained sections were developed with diaminobenzidin (DAB) and restained with hematoxylin. The staining results were observed under a confocal FV 1000 SPD laser scanning microscope (Olympus, Japan) and the integrated optical density values (IOD)/total IOD were evaluated by quantitative morphometry with automated image analysis (Image-Pro Plus, Version 5.0; Media Cybernatics, Houston, TX).

Total protein of adipose tissue and cells was extracted using Cell Lysis (C2978, Sigma-Aldrich, USA) added with a protease inhibitor cocktail (1:100, CW2200S, CWBIO in Beijing, China). After centrifuging at 13,000×g, 4°C for 15 min, supernatant was collected and protein concentration was measured using BCA Protein Assay Kit (23227; ThermoFisher Scientific, United States). Then loading buffer was added, followed by incubating at 99°C for 5 min and subject to Western blotting analysis.

Proteins extracted from tissues or cells were separated by 10% as well as 12% SDS-PAGE and transferred on PVDF membrane. After incubation with 5% BSA (Albumin from bovine serum), transferred blots were incubated with primary antibodies (listed in Table 1) overnight at 4°C. On the next day, the PVDF membrane was added with anti-IgG horseradish peroxidase-conjugated secondary antibody at room temperature for 2 h. Blot images were detected by chemiluminescent reagent (WBKLS0500, Millipore, United States) via a luminescent image instrument (Amersham Imager 680, GE, United States). Total protein levels were normalized to β-actin levels.

Total RNA was extracted from adipocytes using the Total RNA Extraction Kit (220011, Fastagen in Shanghai, China) based on protocols. The PrimeScript RT reagent kit was used for reversed-transcribtion including gDNA Eraser (RR047A, TaKaRa, Japan). The cDNA samples were subject to quantitative PCR process for indicators detection using the LightCycler^® 480 SYBR^® Green I Master (04887352001, Roche, Switzerland) by Roche LightCycler480 according to protocols. The level of β-actin was used for data normalization and the 2^−ΔΔCT method was used for calculation. Primer sequences are listed in Table 2.

The concentrations of IL6 (SM6000B, R&D) and IL1β (MHSLB00, R&D) of adipocyte supernatant was measured using Enzyme-linked immunosorbent assay (ELISA) kits according to instructions.

Primary preadipocytes were isolated from SAT according to a standard protocol (Macotela et al., 2012). Then cells differentiated into mature adipocytes (hereafter primary adipocytes) exposed to serum-containing media supplemented with insulin, 1-methyl-3-isobutylxanthine (IBMX), indomethacin, dexamethasone, triiodothyronine (T3) and rosiglitazone and then switched to insulin, T3 and rosiglitazone for maintenance (Armani et al., 2014). Mouse 3T3-L1 cell line were purchased from the cellbank of Chinese Academy of Sciences. The mouse 3T3-L1 cell line is immortalised and does not undergo senescence as primary adipocytes. Then, cells were cultured and induced to differentiate (hereafter 3T3-L1-derived adipocytes) as previously reported (Molinari et al., 2021). After differentiation, Adipocytes were treated with tumor necrosis factor-alpha (TNFα) stimulation.

Human HEK293T cells and Hela cells were obtained from KeyGene BioTech (China) and cultured in complete Dulbecco’s Modified Eagle’s Medium (DMEM; C11995500BT, Gibco, United States) medium.

Plasmids were transiently transfected into cells with the help of lipofectamine 3000 reagent (100022052, Invitrogen, USA) based on manufacturer’s protocols. The lipofectamine RNAiMAX reagent (56532, Invitrogen) was used for siRNA transfection. After plasmid or siRNA transfection for 24 h, cells could be treated for further processing, then lysed for protein collection. For adipocytes, siRNA and plasmids were transfected during initial plating and grown to confluence, and transfected once again after 48 h. After transfection, primary adipocytes and the 3T3-L1 cell line were differentiated into mature adipocytes. Plasmids are listed in Table 3.

For exogenous Co-IP analysis, plasmids expressing PTPN2 and other indicators were transfected into HEK293T cells. After transfection with 24 h, cells were lysed with 1 mL Cell Lysis for protein extraction. After centrifugation for 15 min at 13,000g, 4°C, 100ul supernatant was removed as “Input” and the rest was incubated with agarose immunoprecipitation beads for tag combination with 4 °C incubation overnight on a rotary mixer. Beads were then washed for 5 times using Radio Immunoprecipitation Assay (RIPA) buffer (P0013c, Beyotime Biotechnology, China). Precipitated proteins were separated from beads after incubation at 99°C for 5min with 2x loading buffer. Before the incubation at 99°C, c-Myc peptide (M2435, sigma) or 3X-Flag peptide (F4799, sigma) was added to compete out the binding of c-Myc or 3X-Flag for necessary.

The protocol of endogenous Co-IP in 3T3-L1-derived adipocytes was similar to exogenous analysis described above. After supernatant preparation, corresponding antibodies were added for endogenous protein combination at 4°C for 1 h. Then protein A/G Plus Agarose (sc-2003, Santa Cruz Biotechnology, United States) was used for antibody-combination at 4°C overnight with gentle mixture. Notably, normal rabbit IgG (2729S, Cell Signaling Technology, United States) was used as a negative control (Ida et al., 2021).

Hela cells were seeded on climbing slices for staining. Plasmids expressing PTPN2 and TAK1 were co-transfected into Hela cells for 24 h. After fixation, 0.1% Triton X-100 was used for cell membranes permeabilization for 5 min, and then 10% goat serum was used for blocking. Then cells were incubated using primary antibodies at 4°C overnight. On the next day, corresponding fluoresce secondary antibodies were used to multiple signaling cascades. DAPI was used for nuclei stain and mounting. Samples incubated with IgG and secondary antibody were used as negative control. Images were captured via laser confocal microscope or electric upright microscope.

SA-β-gal staining was performed with Senescence β-Galactosidase Staining Kit (Beyotime, China) according to the manufacturer’s instruction. Primary adipocytes were fixed with β-Galactosidase fixing solution for 15 min after washed with PBS, and then maintained with the staining solution. The blue areas were recognized to be SA-β-gal-positive.

All data were statistically analyzed using GraphPad Prism 8 (GraphPad, San Diego, CA) and expressed as mean ± standard deviation (SD). The number of replicates was indicated in figure legends (n-numbers) and we did independent biological repeats for each experiment. The Shapiro-wilk test was used to evaluate the normality assumption of the data distribution. For normally distribution, the statistical difference was determined by unpaired two-tailed Student’s t-tests, one-way ANOVA test or two-way ANOVA accordingly. If p-value was less than 0.05, the comparison was considered to be statistically significant. * P < 0.05 , * * P < 0.01 , * * * P < 0.001 ; P # < 0.05 , P # # < 0.01 , P # # # < 0.001 .

The expression level of PTPN2 in the serum is downregulated in T2DM patients (Zheng et al., 2018). As TNFα is a triggering factor in IR (Pedersen, 2017) and involved in several pathogenesis programs, we chose TNFα as the stimulation factor for adipocytes. 3T3-L1 cells were used to differentiate into mature adipocytes. Then, 3T3-L1-derived adipocytes were incubated with TNFα (40 ng/mL) for different time points (0, 4, 8, 12, 24, and 36 h) and we found that the mRNA and protein level of PTPN2 by TNFα stimulation was decreased in a time-dependent manner, especially significant after 24 h stimulation (Figures 1A, B). The mRNA and protein levels of PTPN2 also declined in a concentration-dependent manner in response to TNFα (Figures 1C, D) in 3T3-L1-derived adipocytes. Furthermore, the time-dependent downregulation was confirmed in mouse primary adipocytes (Figures 1E, F).

To further investigate the role of PTPN2 in adipocytes, a whole genome RNA-sequence analysis was performed using primary adipocytes with or without PTPN2 knockdown. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis in cellular processes revealed a strong association between PTPN2 and cellular senescence (Figure 1G). Therefore, we silenced the PTPN2 gene in primary adipocytes from mice using small interfering RNA (siRNA). PTPN2-deficient (siPTPN2) primary adipocytes exhibited an increase in senescence‐associated β‐galactosidase (SA‐β‐gal) activity compared with the control group (siCtrl) as demonstrated with SA‐β‐gal staining (Figure 1H). Interleukin-6 (IL6) and Interleukin-1β (IL1β) are two key components of SASP (Gorgoulis et al., 2019). Excessive IL6 and IL1β were secreted from 3T3-L1-derived adipocytes after TNFα stimulation and increased in siPTPN2 group (Figure 1I). Then, we tested the protein level of P53, P16, P27 and P21, which often gradually increase during cellular senescence. PTPN2 deficiency increased the level of senescence-related proteins (Figure 1J), and 3T3-L1-derived adipocytes with overexpressed PTPN2 confirmed these findings (Figure 1K). Thus, PTPN2 might alleviate senescent phenotype in adipocytes.

As the MAPK/NF-κB pathway implements key traits of cellular senescence (Anerillas et al., 2020), we investigated the role of PTPN2 in the regulation of the MAPK/NF-κB pathway. We treated 3T3-L1-derived adipocytes with TNFα as an inducer and the phosphorylation level of Erk (p-Erk), P38 (p-P38), JNK (p-JNK), IκB (p- IκB) and P65 (p-P65) was upregulated in siRNA-mediated PTPN2-silent 3T3-L1-derived adipocytes after TNFα treatment (Figure 2A). Consistently, when 3T3-L1-derived adipocytes were transfected with Flag-tagged PTPN2 plasmid (Flag-PTPN2) for PTPN2 overexpression, the phosphorylation level of Erk, P38, JNK, IκB, and P65 was decreased significantly in response to TNFα (Figure 2B). To further confirm the effect of PTPN2 on MAPK/NF-κB pathway, we first silenced PTPN2 expression via siRNA, where the phosphorylation level was upregulated, and then rescued PTPN2 protein level by overexpression so that the phosphorylation level was inhibited subsequently (Figure 2C).

To identify exact molecules regulated by PTPN2, we co-transfected HEK293T cells with Myc-tagged PTPN2 plasmid and other plasmids encoding upstream regulators of MAPK/NF-κB pathway for the co-immunoprecipitation (Co-IP) experiment. PTPN2 interacted with TAK1 directly, but not with TRAF6, TAB2 and TAB1 (Figures 3A, B). Then, we confirmed the interaction between PTPN2 and TAK1 using tag-altered plasmids in HEK293T cells (Figures 3C, D) and Hela cells (Figures 3E, F), respectively. Endogenous Co-IP experiments in 3T3-L1-derived adipocytes (Figures 3G, H) and primary adipocytes (Figures 3I, J) confirmed the interaction between endogenous PTPN2 and TAK1. Moreover, we co-transfected PTPN2 and TAK1 plasmids into HeLa cells and confocal microscopy revealed that PTPN2 was colocalized with TAK1 (Figure 3K). Therefore, PTPN2 could bind to TAK1.

To explore the effect of PTPN2 on TAK1, we silenced PTPN2 in 3T3-L1-derived adipocytes using siRNA to test the phosphorylation level of TAK1. After the stimulation of TNFα, the level of phosphorylated TAK1 was upregulated in the PTPN2-deficient group compared with the control group (Figure 4A). Consistently, the phosphorylation level of TAK1 was significantly downregulated when PTPN2 was overexpressed in 3T3-L1-derived adipocytes (Figure 4B). Furthermore, we transfected Flag-tagged PTPN2 plasmid in 3T3-L1-derived adipocytes after PTPN2 silencing for rescue, and the result revealed that PTPN2-silence promoted the phosphorylation level of TAK1 while its phosphorylation was inhibited after PTPN2 overexpression (Figure 4C). A similar trend was also observed for the level of IL6 and IL1β secreted into the supernatant from cultured 3T3-L1-derived adipocytes (Figure 4D).

Finally, to investigate whether TAK1 has a role in the regulation of PTPN2 for the MAPK/NF-κB pathway, a recovery experiment was performed where TAK1 protein level was rescued in PTPN2-deficient 3T3-L1-derived adipocytes. Obviously, when stimulated with TNFα, the diminished SASP protein level of the PTPN2-overexpressed group was rescued by TAK1 overexpression (Figure 4E) and similar results were observed from the phosphorylation level detection of proteins of the MAPK/NF-κB pathway (Erk, P38, JNK, IκB and P65) (Figure 4F). Therefore, TAK1 acts as an essential regulator mediating the dephosphorylation effect of PTPN2 on the MAPK/NF-κB pathway.

The results above suggested that PTPN2 ameliorated adipocyte senescence and as adipocytes are the major cell type in adipose tissue, we proposed that PTPN2 could improve senescence in adipose tissue.

A non-genetic rodent model closely resembling human T2DM was generated by giving male C57BL/6 mice HFD as well as the STZ treatment (referred to as T2DM hereafter) while normal-diet fed mice served as controls (Ctrl hereafter). Multiple metabolic parameters of mice were examined to confirm the model generation. The body weight gain in the T2DM group was significantly increased compared with that in the control group after week-10 as expected (Figure 5A). IR in mice was verified by IPITT which showed that the blood glucose levels of T2DM group were higher than that of the control group at week-12 (Figure 5B). IPGTT was also tested at the age of 10 and 12 weeks, showing that blood glucose levels in the T2DM group were significantly elevated compared to the control group at all of the 5 time points (Figures 5C, D). Therefore, the T2DM model induced by HFD and STZ showed typical features of IR, hyperglycaemia and obesity, resembling the state of human T2DM.

For the T2DM model, IPGTT and IPITT demonstrated that compared with the control group, blood glucose levels in T2DM group were higher at all of the points tested at the age of 24 weeks, while levels in the PTPN2-overexpression group (T2DM + PTPN2) were lower than those in T2DM mice, indicating an improvement role of PTPN2 in IR and hyperglycaemia (Figures 5E, F). The fasting blood glucose level also showed that PTPN2 overexpression improved hyperglycaemia in T2DM mice (Figure 5G). We also discovered that the body weight, total cholesterol level and triglycerides level were decreased in the PTPN2-overexpression group compared with that in the T2DM group (Figures 5H, I). As adipose tissue plays a pivotal role in the development of insulin resistance, we isolated BAT, SAT and VAT respectively for subsequent experiments. The H&E staining demonstrated that the number of adipocytes in SAT counted in the T2DM group was lower than that of the control group in a field of identical size, indicating the bigger size and more lipid content of adipocytes in T2DM group. However, after Ad-PTPN2 treatment, the adipocyte number of SAT in the T2DM + PTPN2 group was significantly higher than that in T2DM group and a similar trend was also observed in VAT (data not shown). For BAT, we found that there were unilocular adipocytes akin to white adipocytes in T2DM mice while after PTPN2 intervention, there was a switch from unilocular to multilocular morphology in the T2DM + PTPN2 group (Figure 5J).

We have demonstrated that PTPN2 ameliorated cellular senescence in adipocytes in vitro, then we assessed the effects of PTPN2 on adipose tissue in vivo and proteins related to senescence were analyzed in T2DM mice. Western blotting showed that the protein levels of P16, P53, P27 and P21 was downregulated in BAT of the PTPN2-overexpressing mice compared with the T2DM group without PTPN2-overexpression (Figure 5K). Furthermore, two main indicators, P16 and P53, were characterized by immunohistochemical staining in adipose tissue which showed that the protein level of p16 and P53 in the T2DM + PTPN2 group after PTPN2 intervention was significantly decreased compared with the T2DM group (Figures 5L, M). Similar results were also obtained from SAT of T2DM mice, indicating the role of PTPN2 in improving senescence in adipose tissue (Figures 5N–P).

As aging aggravates the loss of brown adipocytes (Rogers et al., 2012) and PTPN2 was demonstrated to regulate senescence in adipose tissue, we evaluated regulatory genes that promoted adipocyte browning. Firstly, PTPN2 was silenced in 3T3-L1-derived adipocytes and then stimulated with TNFα. The expression of specific markers including UCP1, PGC1α [proliferator-activated receptor gamma coactivator-1-alpha, which regulates mitochondrial biogenesis and fatty acid oxidation (Yan et al., 2016)], FABP4 [fatty acid-binding protein-4, which regulates adipocyte differentiation and β-oxidation of fatty acids (Walenna et al., 2018)] and INSR (insulin receptor, relating to adipocyte differentiation and function) were assessed at both of the protein (Figure 6A) and mRNA (Figure 6B) levels. The results revealed that PTPN2 knockdown inhibited the browning process of adipocytes. White adipocyte browning develops predominantly in response to cold exposure and adrenergic stress, and it has become a therapeutic target for weight gain and metabolic dysfunction (Sidossis et al., 2015; Sakers et al., 2022).

Then we assessed the influence of PTPN2 on adipose tissue browning activity in vivo. The protein levels of Acox1 (acyl-coenzyme A oxidase 1, which regulates β-oxidation of adipose tissue (De Carvalho et al., 2021)), INSR, PGC1α, PPARγ [peroxisome-proliferator-activated receptor-gamma, which activates the thermogenic brown fat gene program (Ohno et al., 2012)], SIRT3 [Sirtuin 3, a NAD-dependent deacetylase which regulates mitochondrial fatty-acid oxidation (Hirschey et al., 2010)], SLC27A1 [fatty acid transport protein 1, which promotes the WAT-browning process (Guo et al., 2022)] and UCP1 in adipose tissue were detected by Western blotting. Both mice and humans have thermogenic BAT. A major murine BAT depot is located in the interscapular region and is also found in cervical, axillary, perivascular, and perirenal regions. Human infants also possess BAT depot in the interscapular region, however, it later regresses and is absent in adults. Adult humans possess variable BAT and beige adipose tissue in the paravertebral junctions, cervical/axillary regions and in perirenal/adrenal locations (Sakers et al., 2022).

For BAT, our results showed that overexpressed PTPN2 promoted the protein levels of these browning-related regulators (Figure 6C). Moreover, the protein level of the best characterized thermogenic effector, UCP1, was re-confirmed by immunohistochemical staining, which indicated that overexpressed PTPN2 in T2DM + PTPN2 group facilitated the protein level of UCP1 compared with the T2DM group (Figure 6D). The same analysis was performed in SAT and the protein level of Acox1, INSR, PGC1α, PPARγ, SIRT3, SLC27A1 and UCP1 was upregulated by PTPN2 overexpression (Figures 6E, F), illustrating the role of PTPN2 in promoting the WAT-browning process.

Adipose tissue remodeling is a complex process and is regulated by various metabolic challenges. Adipocytes have the ability to act rapidly in response to excess caloric intake and hypertrophy and hyperplasia are the main mechanisms of adipocyte plasticity (Zhu et al., 2022). Adipose tissue remodeling can be pathologically accelerated in the obese state, featuring reduced angiogenic remodeling, a heightened state of immune cell infiltration and subsequent proinflammatory responses (Sun et al., 2011). Disturbance of lipid metabolism contributes to the progression of IR in T2DM, so we investigated the role of PTPN2 in controlling lipid metabolism. Experiments in vitro were performed using 3T3-L1-derived adipocytes transfected with PTPN2-siRNA, and lipid metabolic gene expression was analyzed at the protein and mRNA levels. The results of Western blotting demonstrated that the levels of ATGL (adipose triglyceride lipase, which hydrolyzes intracellular triglyceride), HSL (hormone-sensitive lipase, which is involved in lipid catabolism (Jocken et al., 2016)) and LPL (lipoprotein lipase, which catalyzes the hydrolysis of triglyceride-rich lipoproteins (Jocken et al., 2016)) was reduced in PTPN2-silent 3T3-L1-derived adipocytes compared with the control group (Figures 7A, B).

Then we assessed the effect of PTPN2 on lipid homeostasis in adipose tissue in vivo. Compared with the T2DM group, the protein levels of LPL, ATGL and HSL was significantly higher in BAT in T2DM + PTPN2 group (Figures 7C–E). Consistently, these regulators hydrolyzing triglycerides and promoting thermogenesis were upregulated with PTPN2 overexpression in SAT (Figures 7F–H), confirming the function of PTPN2 on preventing lipid retention in adipocytes.

The recombinant pAdxsi adenovirus system we used in this study is not specifically expressed on adipose tissue so the non-specific expression in other tissues may influence the regulation of glucose and lipid metabolism. Further studies need to be conducted in the future to determine the function of PTPN2 in adipocytes in vivo using gene conditional knockout mice or recombinant pAdxsi adenovirus specifically expressed.