Human adipose-derived stem cells (ADSCs) (HUXMD-01001) and adipogenic differentiation medium (HUXMD-90031) were obtained from Cyagen (Santa Clara, California, USA). TNF-α (C008) was purchased from Novoprotein (Shanghai, China). Anti-DAB2IP (ab87811), TNF-α (ab6671), and β-tubulin (ab6046) antibodies, Glucose Uptake Assay Kit (ab136955) were purchased from Abcam (Cambridge, UK). DAB Substrate Kit (8059S), and IHC Detection Reagent (8125S, 8114S), anti-phospho-SAPK/JNK (9255S), SAPK/JNK (9252S), Phospho-p38 MAPK (Thr180/Tyr182) (9211S), p38 MAPK(8690S), phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (9106S), p44/42 MAPK (Erk1/2) (4695S), and rabbit IgG (2729S), HRP-linked (7074S) antibodies were purchased from Cell Signaling Technology (Danvers, Massachusetts, USA). Antifade mountant (P36934), Bicinchoninic acid (BCA) Protein Assay Kit (23225), BODIPY (D3922) and Hoechst (H21486) were purchased from Invitrogen (Waltham, Massachusetts, USA). TNF-α (EK0525) and C-reactive protein (CRP) (EK1316) ELISA kits were purchased from BOSTER (Wuhan, China). SB203580 (S1076), KO-947 (S8569), and SP600125 (S1460) were purchased from Selleck (Houston, Texas, USA). LentiCRISPRv2, VSVG, RA2, TAT, and HEPM2 plasmids were purchased from Addgene (Watertown, Massachusetts, USA). Glucose Assay Kit (Glucose oxidase and Peroxidase (GOD-POD) Method) (BC2505) were purchased from Solarbio (Beijing, China). Radio-immunoprecipitation assay buffer (RIPA) buffer (P0013B), Hematoxylin and Eosin (HE) Staining Kit (C0105S), Citrate Antigen Retrieval solution (P0083), Puromycin (ST551) were purchased from Beyotime (Shanghai, China). Polyvinylidene difluoride (PVDF) (IPVH00010) membranes and enhanced chemiluminescence (ECL) (WBKLS00100) substrates were purchased from Merck Millipore (Burlington, Massachusetts, USA). TMT-reagent (Art.No.90111) were purchased from ThermoFisher Scientific (Waltham, Massachusetts, USA).

46 obese patients and 46 obese patients with T2D who underwent laparoscopic sleeve gastrectomy or laparoscopic Roux-en-Y gastric bypass (RYGB) surgery were recruited from Jinshazhou Hospital of Guangzhou University of Traditional Chinese Medicine during July 2019 to October 2020. All subjects were of Asian. As recommended by the Working Group on Obesity in China (19), subjects were categorized by body mass index (BMI) ≥ 28 kg/m² as obese. As recommended by Chinese Diabetes Society, subjects were categorized by Diagnostic Criteria for Type 2 Diabetes launched by WHO (1999) (20) as T2D. The obese participants didn’t suffer from T2D. All participants provided written consent, and the study was performed in accordance with the principles outlined in the Declaration of Helsinki. Ethical approval was obtained from the Jinshazhou Hospital of Guangzhou University of Traditional Chinese Medicine Ethics Committee (Reference Number, 19060101). Key eligibility criteria were male or female participants aged between 18 and 70 years who required bariatric surgery. Key exclusion criteria were any comorbidities such as acute infection, cancer, and any other consuming disease that could compromise the validity and safety of the study.

Demographic and clinical features were collected at the time of recruitment as described previously (21). Briefly, Weight was measured without shoes, bulky clothing, and accessories. Height was measured without shoes. BMI (kg/m²) was calculated as weight (kg) divided height (m) squared. Waistline (cm) referred as the perimeter of the plane formed by the top of the hip bone and the navel. Hipline (cm) referred as perimeter of the plane formed by widest part of hip. Waist-to-Hip Ratio (WHR) was calculated as Waistline (cm) divided Hipline (cm). After fasting overnight, fasting blood glucose (FBG) (mmol/L), glycated hemoglobin (HbA1c) (%), insulin (μU/mL), and connecting peptide (C-peptide) (ng/mL) were measured in clinical laboratory of Jinshazhou Hospital of Guangzhou University of Traditional Chinese Medicine. Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) was calculated by FBG (mmol/L)×insulin (μU/mL)/22.5.

After fasting overnight, pre-surgery blood samples were collected into blood collection tube containing ethylene diamine tetraacetic acid (EDTA), then centrifuged at 1000 g for 15 minutes. The serum was collected and stored at −80°C. Total serum TNF-α and C-reactive protein (CRP) was measured using an ELISA kit according to the manufactures protocol.

All subjects were of Asian origin. Omental adipose tissue (OAT) was obtained from 46 obese and 46 obese T2D patients who underwent laparoscopic sleeve gastrectomy or laparoscopic RYGB surgery. In addition, 3 of the 46 obese T2D patients received other routine operations at 6 months or more post-RYGB surgery due to other physical discomfort. During the 1st RYGB surgery, and the 2nd routine surgery, we obtained their OAT specimens.

Briefly, proteins were extracted from OAT of obese patients or obese patients with T2D, and protein concentration was determined using BCA Protein Assay Kit. After reduction, cysteine alkylation, and digestion, the samples were labeled with TMT-reagent. After C18 solid-phase extraction desalting, peptides were used for nano-liquid chromatography-mass spectrometry/mass spectrometry (LC–MS/MS) analysis using Q Exactive Plus quadrupole orbitrap mass spectrometer (Thermo fisher). The MS proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (22) partner repository with the dataset identifier PXD033235. Next, the expressed proteins were identified. A fold change threshold (>1.2 or <0.83) was used to identify differentially expressed proteins. Gene Ontology (GO) (www.blast2go.com/b2ghome) was used to annotate all identified proteins. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD033235.

When ADSCs reached approximately 100% confluence for 24 h, adipocyte differentiation was induced with ADSC adipogenic differentiation medium A for the first 3 days and ADSC adipogenic differentiation medium B for the next 2 days. After 15 days of this cycle, ADSC differentiation was verified by staining using BODIPY493/503. Briefly, the differentiated adipocytes were fixed with 4% paraformaldehyde for 12 minutes at room temperature. Then incubated with PBS containing 2 μmol/L BODIPY493/503 and 5 μg/mL Hoechst for 10 min. Cells were mounted with antifade mountant. The fluorescence microscope (IX71, Olympus) was used to capture the specific signals of six randomly selected, non-overlapping fields. Image J (Fiji) with the plugin Adiposoft (23) was used to measure the diameters(μm) of lipid droplet size, and the lipid stain percent (%, characterized by the percentage of BODIPY positive cells). This evaluation has been performed blindly by two independent individuals. The differentiated adipocytes are defined as cells whose cytoplasm is filled with lipid droplets. Only cultures with >70% adipocytes were used in further experiments. The differentiated adipocytes were maintained with adipogenic differentiation medium B.

The endogenous AIP1 was knocked-down by clustered regularly interspaced short palindromic repeats/clustered regularly interspaced short palindromic repeat–associated 9 (CRISPR/Cas9) in human ADSCs. The sequences of sgRNA of AIP1 were as followed: 5’-3’: caccgAAGTACCTGCAGGACGCCCT, aaacAGGGCGTCCTGCAGGTACTTc, targeting AIP1 exons were inserted into lentiCRISPRv2 vector, respectively. Non-targeting sgRNA (sgRNA-Ctrl) sequences were inserted into lentiCRISPRv2 vector as negative control. Thereafter, we packaged LentiCRISPRv2 vector with VSVG, RA2, TAT, and HEPM2 plasmid to generate lentivirus (negative control, LV-sgRNA-Ctrl; to knockout AIP1, LV-sgRNA-AIP1) and infected ADSCs. Puromycin (1.5 ug/mL) is used to screen stable AIP1 knockout and negative control ADSCs.

The ADSCs were infected with LV-sgRNA-AIP1 or LV-sgRNA-Ctrl. Then, the ADSCs were differentiated to matured adipocytes. The matured adipocytes were treated with TNF-α (10 ng/mL). And the supernatant was collected at 4, 8, 12, and 24 h after treatment.

The supernatant glucose concentration was measured using Glucose Assay Kit (GOD-POD Method) following the manufacturer’s instructions.

For TNF-α measurement, in vitro cultured adipocytes were pre-treated with TNF-α (10 ng/mL) for 4, 8, 12, and 24 hours and supernatant was replaced with serum-free medium, and the medium was collected after 24 hours. ELISA assays for TNF-α was performed following the manufacturer’s instructions.

Glucose Uptake Assay Kit was used to evaluate the 2-DG uptake of matured adipocytes. Briefly, ADSCs infected with LV-sgRNA-AIP1 or LV-sgRNA-Ctrl were seeded into 96 well plate and differentiated to adipocytes. Matured adipocytes were pre-treated with or without TNF-α (10 ng/mL) for 12 hours and then starved with serum free medium overnight, followed by pre-incubating with 100 µL Krebs-Ringer-Phosphate-HEPES (KRPH) Buffer (20 mM HEPES, 5 mM KH₂PO₄, 1 mM MgSO₄, 1 mM CaCl₂, 136 mM NaCl, and 4.7 mM KCl, pH 7.4) containing 2% Bovine serum albumin (BSA) for 40 minutes. Then, 10 µl of 2-DG (10 mM) was added into the KRPH Buffer and incubated for 30 minutes at 37°C. And the 2-DG uptake was measured following the manufacturer’s instructions.

For the inhibitor experiments, the ADSCs were infected with LV-sgRNA-AIP1 or LV-sgRNA-Ctrl. Then, the ADSCs were differentiated to matured adipocytes and pre-treated with specific inhibitors, including 10 μM p38 MAPK inhibitor SB203580, 10 μM ERK1/2 inhibitor KO-947, and 10 μM JNK inhibitor SP600125 for 1 hour. And the adipocytes were further stimulated by 10 ng/mL TNF-α combined with the specific inhibitors for 12 hours. The supernatant glucose concentration was analyzed using Glucose Assay Kit. Further, supernatant was replaced with serum-free medium and incubated for another 24h, TNF-α concentration in the serum-free medium was analyzed using ELISA kit. For 2-DG uptake measurement, cells seeded in 96 well plate were starved overnight with serum free medium, then pre-incubated with 100 µL KRPH buffer containing 2% BSA for 40 minutes. And followed by adding 10 µL 2-DG (10 mM) into the KRPH Buffer and incubating for 30 min at 37°C. Glucose Uptake Assay Kit was used to measure the 2-DG uptake.

Proteins were extracted from adipocytes in radio-immunoprecipitation assay (RIPA) lysis buffer. Protein concentrations were evaluated using a BCA Protein Assay Kit. Samples with the same protein quantity were separated using SDS-PAGE and transferred onto polyvinylidene difluoride membranes. The membranes were incubated with primary antibodies against AIP1, JNK, p-JNK, Erk1/2, p-Erk1/2, P38, p-P38, and β-tubulin, and then with the appropriate HRP-conjugated secondary antibodies, the specific signals were detected by iBright CL1500 Imaging System (Invitrogen) using ECL substrates. And the quantification analyses were accessed by using the Image J software (Fiji).

OAT was isolated from the omentum majus of obese or obese T2D patients who underwent laparoscopic sleeve gastrectomy or laparoscopic RYGB surgery and fixed with 4% paraformaldehyde for 24 h, paraffin-embedded, and cut into 5-μm-tick sections.

For immunohistochemistry, six serial sections of each tissue were deparaffinized, rehydrated, retrieved at 95°C for 20 minutes using Citrate Antigen Retrieval solution and cooled down to room temperature. After blocking endogenous peroxidase activity with 3% H₂O₂ in methanol for 12 minutes, washing thrice with PBS, and blocking with 10% goat serum for 1 hour, the slides were incubated with primary antibodies (rabbit anti-AIP1, 0.5 μg/ml; rabbit anti-TNF-α, 2.5 μg/ml) at 4°C overnight, the same concentration of rabbit IgG antibody was used as negative control. Then washed thrice with PBS and incubated with secondary antibodies. Finally, the DAB and hematoxylin staining were used to detect the specific signals and the nucleus. After dehydration by using increasing concentrations of ethanol (70%, 80%, 90% and 100%) and xylene, the sections were mounted with neutral balsam and the specific signals were captured using AxioImagerZ2 Microscope (Zeiss).

For HE staining, six serial sections of each tissue were deparaffinized, rehydrated, and conducted using the Hematoxylin and Eosin Staining Kit according to the manufactures protocol.

Image J (Fiji) was used to measure the mean intensity of positive signals of six randomly selected, non-overlapping fields of each image. And the adipocyte size was measured by the Image J (Fiji) with the plugin Adiposoft (23). The staining intensity score was as follows: 0: negative, 1: weak; 2, moderate; 3, strong. The positive area was defined as follows: 0, <5%, 1: 5%–25%, 2: 26%–50%, 3: 51%–75%; and 4, >75%. The histochemical score (H-SCORE) was calculated by multiplying the score of the staining intensity and the score of the positive area. This evaluation has been performed blindly by two independent individuals.

Statistical analyses of clinical data were performed using SPSS software (Version 22.0). Other analyses were performed using GraphPad Prism 9.0 software (GraphPad Software, San Diego, California, USA). All Figures are representative of three experiments unless otherwise stated. Clinical data are presented as mean ± standard deviation (SD). Other data are presented as mean ± standard error of the mean (SEM). We firstly determined whether the data were normality using D’Agostino & Pearson omnibus normality test. For the normally distributed data, comparisons between two groups were performed using an unpaired Student’s t-test. Comparisons among more than two groups were performed using two-way ANOVA with Tukey’s multiple comparisons test. And Kruskal Wallis test was used for the abnormally distributed data. Pearson’s correlation test was used to analyze the correlation among multiple variables. P-values were two-tailed, and results with P < 0.05 were considered to indicate statistically significant (indicated with an asterisk *).

To determine the relationship between AIP1 and T2D, we collected the OAT from 46 consecutively admitted patients with obesity and 46 consecutively admitted BMI-paired obese patients with T2D, recruited from clinics. Their baseline characteristics are shown in Table 1 . Briefly, there were no differences in age, weight, and BMI between these groups. Furthermore, the characteristics related to T2D, such as waistline, WHR, FBG, HbA1c, and HOMA-IR, were significantly higher in obese patients T2D than in obese patient. We then assessed the expression of AIP1 in the OAT of obese patients with or without T2D. The results showed that the expression of AIP1 was significantly decreased in the OAT of obese patients with T2D ( Figures 1A, B ). Furthermore, Pearson’s correlation test was used to reveal the correlation between AIP1 expression and the baseline characteristics in these two groups. As shown in Figures 1C, D , AIP1 expression was negatively correlated with HOMA-IR (r = -0.4829) and WHR (r = -0.2614), which are major characteristics of T2D. These results indicated that the expression of AIP1 was negatively correlated with T2D.

To identify the regulating pathways of AIP1 in T2D, we performed an analysis using a proteomic array of OAT between obese patients with T2D and obese patients. The proteomic analysis indicated that obese patients with T2D presented a higher inflammatory response than obese patients ( Figure 2A ). TNF-α is one of the critical inflammatory factors related to T2D. Therefore, we further confirmed the expression of TNF-α in the OAT. The results revealed that there was no difference in the size of adipocytes between obese patients and obese patients with T2D ( Figures 2B, C ). However, TNF-α expression in the OAT of obese patient with T2D was significantly increased compared with that in the OAT of obese patients ( Figures 2D, E ). Then we accessed the concentration of TNF-α and CRP, two critical indexes of inflammation, in the serum of obese patients or obese patients with T2D using ELISA kit. Interestingly, the concentration of TNF-α and CRP was significantly increased in the serum of obese patients with T2D compared with that in obese patients ( Figures 2F, H ). Furthermore, Pearson’s correlation test revealed that the concentrations of TNF-α ( Figures 2G ) and CRP ( Figures 2I ) were positively correlated with HOMA-IR, respectively (TNF-α, r = 0.4728; CRP, r = 0.5522). These results suggest that the inflammatory status of OAT may be involved in the pathogenesis of T2D.

To investigate the role of AIP1 in TNF-α expression and T2D, AIP1 knockout (KO) human ADSCs were obtained using CRISPR-Cas9. ADSCs were differentiated into adipocytes, western blot analysis revealed that AIP1 was effectively knockout in adipocytes infected with LV-sgRNA-AIP1 ( Figures 3A, B ). And BODIPY was used to stain the differentiated adipocyte lipid droplets ( Figure 3C ). We found that there was no significant difference between AIP1-KO and vector-control adipocytes in the lipid droplet size ( Figure 3D ) and lipid staining percentage area ( Figure 3E ). Subsequently, we measured the concentration of TNF-α in the supernatant of differentiated adipocytes after TNF-α (10 ng/mL) stimulation at different time points. As expected, significantly higher TNF-α secretion ( Figure 3F ) was observed in the supernatant of AIP1 KO adipocytes. We then tested the glucose level in the cell supernatant at different time points after TNF-α (10 ng/mL) stimulation and found that compared with the control adipocytes, the AIP1 KO group had a higher sugar level in the supernatant ( Figure 3G ). The sugar uptake ability was decreased in AIP1 KO adipocytes compared with that in the control adipocytes after TNF-α (10 ng/mL) stimulation ( Figure 3H ). These results showed that the AIP1 deficiency in adipocytes resulted in less sugar consumption under TNF-α stimulation. The result also showed higher insulin resistance.

Previous research showed AIP1 modulated the inflammatory signaling including JNK and MAPK signaling. To further understand the molecular mechanism by which AIP1 modulates adipocyte TNF-α secretion and insulin resistance phenotype, the differentiated adipocytes were treated with TNF-α. As a result, AIP1 deletion caused enhanced TNF-α-induced activation of JNK, P38 and ERK1/2 in human adipocytes ( Figure 4 ). Furthermore, suppression of P38 MAPK, ERK1/2 and JNK using their inhibitor SB203580, KO-947, and SP600125, respectively, abolished the secretion of TNF-α induced by AIP1 deficiency in TNF-α treated adipocytes ( Figures 5A–C ), and restored the glucose level in the cell supernatant ( Figure 5D ) and 2-DG uptake ( Figure 5E ) in the TNF-α treated AIP1-KO adipocytes. The results indicated that AIP1 modulated TNF-α production and insulin resistance in adipocytes via JNK and MAPK/ERK signaling.

It was well known that metabolic surgery could considerably improve the diabetic state and metabolic syndrome of patients. Therefore, we collected the OAT from patients with T2D pre- and post-surgery ( Table 2 ). After surgery, all of the characteristics related to patients with T2D, including weight, BMI, waistline, WHR, FBG, HbA1c, and HOMA-IR, were significantly decreased in obese T2D patients. We then performed IHC staining and found that compared with that in pre-surgery, the expression of AIP1 in adipose tissue after metabolic surgery was increased ( Figures 6A, B ), accompanied by lower TNF-α expression ( Figures 6C, D ). Taken together, these findings suggested that AIP1 is an integral part of the feedback loop of low-grade inflammation in OAT of patients with T2D.