Dipeptidyl peptidase‐4 disturbs adipocyte differentiation via the negative regulation of the glucagon‐like peptide‐1/adiponectin‐cathepsin K axis in mice under chronic stress conditions

Exposure to chronic psychosocial stress is a risk factor for metabolic disorders. Because dipeptidyl peptidase‐4 (DPP4) and cysteinyl cathepsin K (CTSK) play important roles in human pathobiology, we investigated the role(s) of DPP4 in stress‐related adipocyte differentiation, with a focus on the glucagon‐like peptide‐1 (GLP‐1)/adiponectin‐CTSK axis in vivo and in vitro. Plasma and inguinal adipose tissue from non‐stress wild‐type (DPP4+/+), DPP4‐knockout (DPP4−/−) and CTSK‐knockout (CTSK−/−) mice, and stressed DPP4+/+, DPP4−/−, CTSK−/−, and DPP4+/+ mice underwent stress exposure plus GLP‐1 receptor agonist exenatide loading for 2 weeks and then were analyzed for stress‐related biological and/or morphological alterations. On day 14 under chronic stress, stress decreased the weights of adipose tissue and resulted in harmful changes in the plasma levels of DPP4, GLP‐1, CTSK, adiponectin, and tumor necrosis factor‐α proteins and the adipose tissue levels of CTSK, preadipocyte factor‐1, fatty acid binding protein‐4, CCAAT/enhancer binding protein‐α, GLP‐1 receptor, peroxisome proliferator‐activated receptor‐γ, perilipin2, secreted frizzled‐related protein‐4, Wnt5α, Wnt11 and β‐catenin proteins and/or mRNAs as well as macrophage infiltration in adipose tissue; these changes were rectified by DPP4 deletion. GLP‐1 receptor activation and CTSK deletion mimic the adipose benefits of DPP4 deficiency. In vitro, CTSK silencing and overexpression respectively prevented and facilitated stress serum and oxidative stress‐induced adipocyte differentiation accompanied with changes in the levels of pref‐1, C/EBP‐α, and PPAR‐γ in 3T3‐L1 cells. Thus, these findings indicated that increased DPP4 plays an essential role in stress‐related adipocyte differentiation, possibly through a negative regulation of GLP‐1/adiponectin‐CTSK axis activation in mice under chronic stress conditions.


| INTRODUCTION
6][7] In addition, CPS causes changes in the size of adipocytes and an inflammatory response in subcutaneous and inguinal adipose tissue in mice, leading to insulin resistance and perturbation of glucose uptake. 2 It is thus very important to understand the exact mechanism by which CPS interferes with adipocyte differentiation and adipose inflammation.
Dipeptidyl peptidase-4 (DPP4) is a multifunctional enzyme that is commonly expressed in visceral, epididymal, and omental adipose tissue. 8DPP4, also known as T cell activation antigen CD26, is a serine protease that acts as a membrane-anchored extracellular peptidase. 9DPP4 cell surface expression marks an abundant adipose stem/progenitor cell population with high stemness in human white adipose tissue. 102][13] DPP4 degrades GLP-1 as well as several other peptides and cytokines 14 and is well known for its regulatory role in glucose metabolism. 15DPP4 is upregulated in proinflammatory states such as obesity, diabetes, and atherosclerotic coronary artery diseases. 16,17DPP4 is also highly expressed in human adipocytes, and its knockdown has been shown to retard preadipocytes proliferation and maturation by mimicking growth factor withdrawal. 18 Studies performed over the past five years have revealed that increased DPP4 negatively modulates bone marrow hematopoietic stem cell proliferation and output, adipose inflammation, vascular damage, thrombosis and ischemic angiogenesis in mice under chronic stress conditions. 7,19,20GLP-1 has been shown to stimulate adipogenic potential and the ability of human adipose tissue to expand, and both these effects involve activation of the canonical GLP-1 receptor. 21Although laboratory and clinical evidence indicate DPP4 and GLP-1 have a broad function in metabolic disorders, 8 the mechanism by which the DPP4/GLP-1 axis induces adipocyte differentiation and adipose remodeling in humans and animals under chronic stress conditions remains largely uncertain.
Cysteine protease cathepsins were originally found to be localized to lysosomes and endosomes in humans and animals. 22The main physiological function of cathepsins is to break down unwanted intracellular or endocytic proteins, but cathepsins also play an important role in pathological processes by regulating inflammation and immune responses through extracellular matrix remodeling, leukocyte recruitment, and cell adhesion. 23,246][27] There has also been a single report showing that, among cathepsin family members, only CTSL contributes to controlling adipogenesis via the production of fibronectin in mice. 28In addition, CTSK activity has been reported to slow the development of obesity-associated cardiac hypertrophy. 29Deficiency and inhibition of CTSK can reduce body weight gain and increase glucose metabolism in mice.Furthermore, we previously reported that the CTSK gene was more highly expressed in stressed than in adipose tissue of non-stress mice. 20Both DPP4 and GLP-1 have been shown to modulate adipocyte differentiation and insulin sensitivity. 18,21owever, it is unclear whether there is close interaction between DPP4/GLP-1 and CTSK in adipocyte differentiation and adipose remodeling in animals with and without chronic stress.
In the present study, we explore the role(s) of DPP4 in the pathogenesis of stress-related adipocyte differentiation and adipose remodeling, with a focus on the GLP-1/ adiponectin-CTSK axis.For this purpose, we subjected wild-type (DPP4 +/+ ), DPP4-knockout (DPP4 −/− ), wildtype (CTSK +/+ ), and CTSK-knockout (CTSK −/− ) mice to 2 weeks of chronic variable stress or a non-stress condition (control mice) followed by biological and morphological analysis.In a separate GLP-1 agonist experiment, DPP4 +/+ mice under stress conditions were assigned to receive a vehicle or an exenatide treatment for 2 weeks.To investigate the molecular mechanisms underlying the observed effects of DPP4, after the silencing or overexpression of CTSK, 3T3-L1 cells were treated with the stress serum (Sserum) or superoxide and subjected to biological analysis.

| Animal care and use
The animal study protocols were approved by the Institutional Animal Care and Use Committees of Yanbian University and Nagoya University.Male DPP4 +/+ (C57BL/6J) and DPP4-knockout (DPP4 −/−30 ) mice, and male CTSK +/+ (C57BL/6J) and CTSK-knockout (CTSK −/−5 ) mice were used in the animal experiments; all mice were aged 8 weeks old and weighed 22-26 g.Both genotype-matched pairs of mice were provided a growing diet and tap water ad libitum and housed three or four per cage under standard conditions (12 h light/dark cycle in a viral pathogen-free facility) at the Animal Center of Yanbian University (for the former pair of genotypematched mice) and the Animal Center of Nagoya University Graduate School of Medicine (for the latter pair of genotype-matched mice).

| Chronic variable stress model and mice grouping
The chronic variable stress procedure was reported previously. 31DPP4 +/+ and DPP4 −/− mice were randomly divided into a non-stress group and stressed group.The mice in the non-stress group were fed normally without any stimulation and were left undisturbed and allowed contact with each other.Stress procedures were performed between 7 a.m. and 6 p.m.The following stressors were applied. 31For cage tilt, the cage was tilted at a 45° angle and kept in this position for 4 h.For isolation, mice were individually housed in a fixed stress cage (catalog G-13302; Natsume Seisakusho, Tokyo).For horizontal cage and damping, we removed the sawdust from the floor of the stress cage and placed some water in the cage; the stress cage was then suspended horizontally, with the mouse's tail in the water for 4 h.For overnight illumination, mice were housed in a separate room with illumination from 9 p.m. to 9 a.m.All stressors were randomly shuffled over 2 consecutive weeks.In a separate experiment, DPP4 +/+ mice were randomly assigned to three groups and given (by subcutaneous injection) either a control NaCl (vehicle, C-NaCl group), NaCl vehicle with stress (S-NaCl group) or the GLP-1R agonist exenatide with stress (5 μg/ kg/d; S-Exe group) for 2 weeks with constant pressure fixation for 4 h daily.In a separate experiment, CTSK +/+ and CTSK −/− mice that had undergone the same stress protocol for two weeks were subjected to biological and morphological analyses.

| Sample collections
Before sampling, mice were anesthetized with isoflurane (3-5% in an induction chamber and between 0.5% and 2% for maintenance, depending on the procedure and its duration) at the Animal Center of Yanbian University.For the experiments with CTSK +/+ and CTSK −/− mice, mice were euthanized by intraperitoneal injection of a single dose of sodium pentobarbital (50 mg/kg; Sanofi, Belgium) at the Animal Center of Nagoya University Graduate School of Medicine.As shown in Figure 1A, on days 0 and 14 of chronic restraint stress, the mice were weighed, and then both the stressed and non-stress mice were anesthetized as above.The stressed mice were left undisturbed for 2 h before sacrifice.Blood samples were obtained directly from the left ventricles of the mice and the plasma was collected; then both sides of inguinal white adipose tissue (iWAT) were isolated and weighed (Figure 1B).Plasma was subjected to enzyme-linked immunosorbent assay (ELISA) for the evaluations of DPP4, adiponectin, GLP-1, CTSK and TNF-α, and was subjected to biochemical detection for the evaluations of plasma triglycerides (TG), total cholesterol (TC) and free fatty acids (FFA).For the pathological evaluation, adipose tissue was embedded in paraffin; in the biological analyses, the adipose tissue was kept in RNAlater solution for quantitative polymerase chain reaction (qPCR) and in liquid nitrogen for Western blotting.

| Plasma and tissue DPP4 activity analysis
We measured the DPP4 activity by using the DPP4 Glo Protease Assay with an aminoluciferin substrate. 19The mice adipose tissues (about 10 mg) were minced using a razor blade and extracted in the lysis buffer (100 mM Tris-HCl, 20 mM NaCl, 0.5% SDS, pH 8.0) without serine protease inhibitors, by means of a tissue homogenizer.The protein concentration for each result was determined using a protein assay system according to the manufacturer (Bio-Rad).For the blood DPP4 activity assays, the plasma was isolated using VENOJectII vacuum blood collection tubes containing anticoagulants without serine protease inhibitor (Terumo, Tokyo) and then diluted in 0.1 mM Tris-HCl buffer (pH 8.0) by 30-fold.Equal amounts of protein (200 μg/25 μL) and diluted plasma (25 μL) were subjected to a DPP4 Glo assay (Promega) in the presence or absence of the DPP4 inhibitor anagliptin (20 μmol/L).Human recombinant DPP4 (Sigma-Aldrich) was used to draw a standard curve.The luminescence intensity was calculated using a POWERSCAN4 (BioTek Instruments Inc., Winooski, VT).The anagliptin-sensitive value (i.e., the absence value minus the presence value as an absolute value of its DPP4 activity) in relative light units per 200 μg of protein (tissue extract) or per mL of plasma was calculated with the standard curve to represent the DPP4 level.

| Gene expression assay
Total RNA from adipocytes was extracted using TRIZOL Reagent and a Super Script III First Strand system following the manufacturer's protocol.Quantitative polymerase chain reaction (qPCR) analysis was performed using a RNeasy Micro Kit and SYBR™ Green Master Mix.Relative gene expressions of interest were determined using GAPDH as the housekeeping gene and an ABI 7300 PCR System with GraphPad Prism software (the formula equation: 2 −ΔΔCt ; Applied Biosystems, Foster City, CA).All qPCR experiments were performed in accordance with the MIQE guidelines.All experiments were performed in triplicate.The sequences of primers specific to CCAAT/ enhancer binding protein α (C/EBP-α), CCAAT/enhancer binding protein β (C/EBP-β), Wnt family member 5α (Wnt5α), Wnt11, beta catenin (β-catenin), peroxisome proliferator-activated receptor γ (PPAR-γ), lipid droplet coating proteins-2 (perilipin2), secreted frizzled-related protein 4 (SFRP4), and adiponectin used in this study are listed in Table 1.

| Immunoblotting assay
Samples of adipose tissue or cells were homogenized or lysed, respectively, and then the proteins were extracted and the total protein concentration was measured using RIPA lysis buffer and a Pierce BCA Protein Assay Kit.Equal amounts of proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes, and then blocked in 5% skim milk in Tris-HCl-buffered saline containing 0.2% Tween for 1 h.After washing three times, the membranes were incubated with primary antibodies directed against the targeted proteins-i.e., CTSK, Pref-1, GLP-1R, adiponectin, fatty acid binding protein 4 (FABP4), PPAR-γ, C/EBP-α, and βtubulin-at 4°C overnight, following the manufacturer's instructions.Then, the membranes were further incubated with horseradish peroxidaseconjugated secondary antibody (dilution, 1:5000) at room temperature for 2 h.After washing with TBST three times, the membranes were subjected to chemiluminescent protein detection.An Amersham ECL Prime Western Blotting Detection kit was used for the evaluation of targeted proteins.Proteins were visualized using an Azure 500 Bioanalytical Imaging System.The loading βtubulin levels were used as an internal control.

| Histological characterization of the adipose tissue
For the histological analysis, iWAT was fixed in ice-cold 4% paraformaldehyde solution for 24 h, embedded in paraffin, and processed for histology and immunofluorescence.Adipose tissue samples were cut into sections (4 μm thickness) and stained with hematoxylin-eosin (H&E).The size and distribution of adipocyte areas in different groups were analyzed in stained sections under 400-fold magnification.Adipose tissue samples were stained with CD4, CD8 and CD68 (dilution, 1:100) for immunofluorescence staining.After washing with PBS three times, the tissue sections were treated with the secondary antibody against anti-Mouse IgG for 1 h at room temperature and then the nuclei were counterstained with anti-fluorescence quenching solution including DAPI.CD4 + and CD8 + T cells and CD68 + macrophages that infiltrated the adipose tissue were counted in four random microscopic fields from three independent sections of each animal (n = 6-8), and the infiltration was expressed as the number of these cells per high-power field (×400).The area per adipocyte was measured in three randomly chosen microscopic fields from 5 to 7 different sections in each adipose tissue block and averaged as an index of the adipocyte size (500-4000 μm 2 ) of each mouse.

| ELISA and biochemical detection assays
We determined the contents of plasma GLP-1, adiponectin and tumor necrosis factor-α (TNF-α) in experimental groups by using a commercially available ELISA kit according to the manufacturer's instructions.The levels of plasma TG, TC, and FFA were also examined.

| Assay of NADPH oxidase-dependent superoxide production
Specific adipose tissue NADPH oxidase activity was measured in total homogenates of the fresh iWAT with the use of a lucigenin-based enhanced chemiluminescence assay as described previously. 23A low lucigenin concentration (5 μmol/L) was used to minimize artifactual O 2 − production attributable to redox cycling.In brief, 100 μg of homogenate protein diluted in 1 mL of lysis buffer (20 mmol/L Tris-HCl, 150 mmol/L NaCl, 1 mmol EDTA, 1 mmol/L EGTA, and 1% Triton X-100; pH 7.5) was transferred to an assay tube, and NADPH and dark-adapted lucigenin were added to final concentrations of 100 and 5 μmol/L, respectively, immediately before the measurement of chemiluminescence.The chemiluminescence signal was sampled every min for 12 min with a tube luminometer (20/20; Tuner Designs), and the respective background counts were subtracted from the experimental values.All of the assays were performed in triplicate.

| Cell culture
White preadipocytes were isolated from iWAT of mice.In brief, adipose tissue was collected from mice under sterile conditions.The tissue blocks were cut into about 1 mm 3 sections with scissors and digested in a shaking bath at 37°C for 90 min with 1 mg/mL type I collagenase.The digestion was stopped using high glucose Dulbecco's modified eagle medium (DMEM) supplemented with 20% new born calf serum (NCS) and 1% Pen/Strep.To remove undigested tissue fragments, the digested tissue suspension was filtered with a 200 μm nylon mesh, and cell sediments (preadipocytes) were collected by three 10-min rounds of centrifugation at 1500 rpm.Then, different groups of preadipocytes were induced to become mature adipocytes and stained with Oil Red O.The 3T3-L1 cells were cultured in DMEM with 15% NCS in a humidified atmosphere of 5% CO 2 and 95% air.The 3T3-L1 cells were then seeded into six-well plates (5 × 10 5 cell/well) and cultured in DMEM for 24 h.The cells were cultured in the presence of 5% non-stress-serum (NS-serum), 5% stress-serum (Sserum) or H 2 O 2 (0, 250, 500 μmol/L) for 24 h and then subjected to the cellular assays.Following pretreatment with various concentrations of exenatide (0, 10, 30 nmol/L) for 30 min, the cells were cultured for 24 h in DMEM with Sserum.In separate experiments, cells were treated with 30 nmol/L exenatide and further incubated for 24 h in S-serum DMEM with or without 10 μg/mL neutralizing adiponectin antibody (nAdipo); they were then subjected to the biological analyses.For all cell culture assays, at least three independent experiments were performed in triplicate.

| Oil Red O staining
The 3T3-L1 cells were seeded in six-well culture plates with DMEM containing 15% NCS at a density of 5 × 10 5 cells/well and cultured at 37°C in a humidified atmosphere with 5% CO 2 .Two days after the cells reached confluence, 3T3-L1 cell differentiation was initiated with a mixture of 10 μg/mL insulin, 0.5 mM 3-isobutyl-1-methylxanthine and 1 μM dexamethasone in DMEM with 15% NCS.After 72 h, the medium was maintained with 15% NCS DMEM supplemented with 20 μg/mL insulin.After 48 h, the medium was maintained with 15% NCS and cells were kept in maintenance media for another 3-7 days.The adipocyte differentiation pattern is shown in Figure S6B.The formation of lipid droplets in the differentiated adipocytes was analyzed using Oil Red O staining, and observed under a light microscope.To quantify lipid droplets, the average areas of red-stained droplets were measured by the image analysis system.The progress of adipocyte differentiation was assessed by lipid drop density (pixels per pixels).

| Gene silencing and overexpression
3T3-L1 cells were transfected with CTSK-418, pcDNA3.1(+)-CTSKor the corresponding nontargeting control, using Lipofectamine 3000 reagent following the manufacturer's instructions.We first designed and tested three short interfering RNA against CTSK (siCTSK) sequences (shown in Table 2) and chose siCTSK-418 for the following experiments because it had the strongest knockdown activity.After transfection for 6 h, cells were treated with or without 30 nmol/L exenatide, further incubated for 24 h in S-serum DMEM, and then subjected to the biological analyses.Levels of CTSK expression were assessed at 48 h or 72 h post-transfection by qPCR and Western blotting.In separate experiments, 3T3-L1 cells were transfected with siCTSK-418 or pcDNA3.1(+)-CTSKfor 6 h, and the fluorescence was observed to confirm the success of CTSK transfection.Following transfection with a non-targeted control, the cells were subjected to immunofluorescent staining as a negative control.

| Statistical analysis
Data are expressed as mean ± SEM.Differences between groups were assessed by Student's t-test.Comparisons of three or more groups were conducted using oneway ANOVA followed by Tukey's post hoc tests for the statistical analyses.The body weight data were subjected to a two-way ANOVA and Bonferroni's post hoc tests.
After the distribution status of the test data was determined, the data were subjected to the statistical analyses.
The adipose tissue immunofluorescence staining as well as the adipocytes area (H&E staining) were evaluated by two observers in a blind manner, and the values they obtained were averaged.Statistical analyses were performed using GraphPad Prism version 9.0.0.Probability (p)-values <.05 were considered significant.

| Stress increased plasma DPP4 and adipose CTSK levels and adipose wasting
Chronic stress models have frequently been used to investigate chronic psychological stress-related metabolic disorders.In the present study, to examine the impacts of stress on adipocyte differentiation and remodeling, wild-type (DPP4 +/+ ) mice were subjected to chronic variable stress protocols (Figure 1A).Consistent with previous studies, 19 the stressed mice maintained decreased body weight throughout the 2-week follow-up period (Figure S1A) and also exhibited a reduction in iWAT volumes (Figure 1B).Compared with those in non-stress mice, the protein and gene levels of CTSK were clearly increased in the stressed mice (Figure 1C,D).The biological detection and ELISA data showed that plasma DPP4 activity was higher in stressed mice, while the level of GLP-1 was lower in non-stress mice (Figure 1E).The quantitative H&E staining results revealed that the stressed mice had a higher frequency of smaller adipocytes (<1500 μm 2 ) in the iWAT compared with the control mice (Figure 1F).Thus, these observations indicate that the changes in CTSK and DPP4/GLP-1 might be associated with stress-related adipocyte dedifferentiation and remodeling.In addition, the biological detection data showed an increase in the levels of plasma TG, TC, and FFA in stressed mice (Figure 1G).
To further identify the molecular change of adipose remodeling in response to stress, we analyzed the expression levels of adipocyte differentiation-and adipogenesisrelated transcriptional factors.As compared with those in non-stress mice, the quantitative PCR data of iWAT showed that the levels of proteolysis (CTSK), inflammation (TNF-α) and adipose differentiation inhibitorrelated genes (Pref-1, Wnt5α, Wnt11, and βcatenin) were markedly elevated in stressed mice.Furthermore, a number of adipogenesis-related (C/EBP-α, C/EBP-β, and PPAR-γ) and adipocyte differentiation-related (per-ilipin2, SFRP4, adiponectin) genes showed a significant reduction (Figure S1B-D).Quantitative western blotting data similarly showed that the expressions of adiponectin, fatty acid binding protein 4 (FABP4), GLP-1R, PPAR-γ and C/EBP-α were significantly reduced by chronic stress, while the protein level of Pref-1 was increased in iWAT (Figure S2A,B), indicating that chronic stress may inhibit adipocyte differentiation and adipogenesis in vivo.On the other hand, stress enhanced adipose NADPH oxidase activity and lowered plasma adiponectin levels (Figures S2C  and 2D).

| DPP4 deletion prevented adipose wasting, oxidative stress, and inflammation in response to stress
As shown in Figure 2A,B, compared with those in DPP4 +/+ -stressed mice, the augments in body weight and iWAT weight were higher in DPP4 −/− -stressed mice.The levels of TNF-α protein and DPP4 activity in the plasma were higher in the DPP4 +/+ -stressed mice than in the DPP4 −/− -stressed mice, whereas the opposite results were observed for GLP-1 and adiponectin levels (Figure 2C,D).Quantitative H&E staining showed that DPP4 deficiency prevented the frequency of smaller adipocytes (Figure 2E).The inguinal adipose tissue is characterized by inflammatory cell accumulation.We observed that stress increased the numbers of CD68 + macrophages and CD4 + and CD8 + T cells in the iWAT of stressed mice, and these changes were reversed by DPP4 deletion (Figure 3A,B).Analysis of NADPH oxidase activity in the same set of samples confirmed a reduction in the oxidative stress production in DPP4 −/− iWAT (Figure S3A), indicating that the DPP4 deficiency-mediated adipose benefit might be due to the reduction of stress-related oxidative stress production and inflammatory action.However, DPP4 deficiency had no effect on the T-cell and macrophage infiltration and oxidative stress production in either genotype of non-stress mice (Figures 3A,B and S3A).

| DPP4 deficiency prevented stress-induced adipocyte dedifferentiation
As compared with those in DPP4 +/+ -stressed mice, qPCR data showed that the levels of the CTSK, TNF-α, Wnt5α, Wnt11, βcatenin, and Pref-1 genes were markedly decreased, whereas the levels of C/EBP-α, C/EBP-β, PPAR-γ, perilipin2, SFRP4, and adiponectin were dramatically elevated in DPP4 −/− -stressed mice (Figure S3B-D).Quantitative immunoblot analysis confirmed these results for the targeted proteins (i.e., PPAR-γ, C/EBPα, FABP4, GLP-1R, CTSK and Pref-1) (Figure 3C,D).there were no changes in these investigated proteins between non-stress DPP4 +/+ and DPP4 −/− mice.Immunofluorescent staining with CD29 and CD44 confirmed the presence of preadipocytes in the iWAT of the stressed DPP4 +/+ mice (Figure S4A).Representative microscopic images of the adipocyte differentiation patterns at various time points are shown in Figure S4B.Consistent results were observed in the Oil Red O staining analysis of the same set of samples, confirming that DPP4 deficiency dramatically reduced S-serum-induced lipid droplet loss in preadipocytes (Figure S4C,D).Taken together, these results indicate that DPP4 deficiency is resistant to adipocyte de-differentiation induced by stress.In addition, the DPP4 −/− genotype resulted in a marked reduction in the levels of TC, TG, and FFA in not only non-stress but also stressed mice (Figure S3E).

| A GLP-1 agonist mimicked DPP4 −/− - mediated adipose benefits in mice under stress conditions
We observed that the body weight and iWAT volumes were higher in stressed mice with exenatide loading (S-Exe) than in stressed mice that received NaCl vehicle alone (S-NaCl) (Figure 4A,B).Plasma TNF-α was significantly reduced in S-Exe mice, whereas plasma adiponectin levels were significantly increased in S-Exe mice (Figure S5A,B).Quantitative H&E staining revealed that exenatide treatment prevented reductions in adipocyte size (Figure 4C).Similar to the findings in DPP4 −/− mice, exenatide produced a benefit on the expressions of the targeted proteins (FABP4, adiponectin, C/EBP-α, PPAR-γ, Pref-1, and CTSK) in S-Exe mice (Figure 4D,E).Interestingly, 30-nM exenatide had a benefit on the investigated molecular and lipid droplet changes in 3T3-L1 cells induced by 5% S-serum (Figure S6A-C).As shown in Figure S6D, the results of an in vitro experiment showed that exenatide had beneficial effects on C/EBP-α and Pref-1 protein expression in a dose-dependent manner in 3T3 cells under a 5% S-serum condition, indicating that GLP-1 receptor activation can prevent adipocyte dedifferentiation in mice under stress conditions.Quantitative immunofluorescent staining demonstrated that exenatide markedly lowered the infiltration of CD68 + macrophages and CD4 + and CD8 + T cells in the iWAT of stressed mice (Figure S5C,D).Analysis of the plasma lipid profile showed that the TG, TC and FFA levels in the S-Exe group were significantly lower than those the S-NaCl group (Figure S5E).

| Stress serum disturbed 3T3-L1 cell differentiation
To investigate the effects of stress serum, we cultured 3T3-L1 cells in the presence of 5% S-serum (extracted from 2-week stressed mice) or 5% NS-serum (extracted from non-stress mice).S-serum augmented CTSK and Pref-1 protein expressions (Figure S7A,B).Conversely, it lowered the levels of adiponectin, FABP4, PPAR-γ and C/EBP-α proteins.Oil Red O staining revealed that S-serum markedly reduced the number of lipid droplets (Figure S7C).

| CTSK deficiency mitigated adipocyte stress-induced de-differentiation in vivo and in vitro
To further test our hypothesis, we also investigated the role of CTSK in mice that underwent variable stress for 2 weeks.Although CTSK deficiency had no effect on body weight, the iWAT volumes in stressed CTSK −/− mice were significantly better-maintained compared to those in stressed CTSK +/+ mice (Figure 5A,B).Stressed CTSK −/− mice exhibited a reduction in the levels of adipotissue DPP4 and plasma TNF-α, whereas they had increased levels of plasma GLP-1 and adiponectin (Figure 5C,D).Immunofluorescent staining revealed the lower numbers of CD4 + and CD68 + macrophages in the iWAT of CTSK −/− -stressed mice (Figure 5E,F).CTSK deletion lowered the frequency of small-size adipocytes (less 1500 μm 2 ) in the iWAT of stressed mice (Figure 5G), indicating that CTSK may be involved in stress-induced adipocyte dedifferentiation and remodeling.
Next, we investigated whether genetic modification of CTSK affected adipocyte differentiation in vitro.We first determined which of three siCTSKs would be most effective for use in silencing experiments; siCTSK-418 was found to be the most impactful (Figure S7D).Immunofluorescence results showed that the CTSK gene knockout (Figure 6A) and CTSK overexpression (Figure 7A) had succeeded.Consistent with these findings, the CTSK gene modifications respectively suppressed or enhanced gene and protein expression in 3T3-L1 cells (Figures 6B,C and  7B,C).CTSK silencing increased the levels of adipocyte differentiation-related factors (C/EBP-α and PPARlowered the level of a de-differentiation-related factor (Pref-1) in 3T3-L1 cells in response to S-serum, and these effects were facilitated by exenatide (Figure 6D,E).Quantitative Oil Red O staining revealed that siCTSK enhanced lipid droplet accumulation in 3T3-L1 cells treated with S-serum (Figure 6F).Conversely, CTSK overexpression had a harmful effect on S-serum-induced targeted molecular changes, and these effects were mitigated by exenatide (Figure 7D,E).As anticipated, CTSK overexpression also suppressed lipid droplet accumulation (Figure 7F).

| Adiponectin blocking disturbed the exenatide-mediated preventive effect on S-serum-induced adipocyte dedifferentiation and lipid droplet loss
To further examine the consequence of neutralizing adiponectin antibody (nAdipo) on exenatide-mediated adipose benefit, the 3T3-L1 cells were cultured in the presence of S-serum containing Exe alone or combined with nAdipo for 24 h, and then were applied to Western blotting assay.We observed that exenatide still reduced Pref-1 protein expression and enhanced C/EBP-α and PPAR-γ protein expressions in 3T3-L1 cells under S-serum conditions; these beneficial effects were diminished by nAdipo treatment (Figure 8A,B).Similarly, as shown in Figure 8C, exenatide significantly lowered S-serum-induced CTSK gene expression, and this benefit was diminished by nAdipo.As anticipated, quantitative data of Oil Red O staining showed that nAdipo suppressed lipid droplet accumulation in 3T3-L1 cells treated by exenatide (Figure 8D,E).Thus, these findings indicated that GLP-1 can reverse Sserum-induced adipocyte dedifferentiation via the modulation of adiponectin/CTSK axis expression.

3T3-L1 cell differentiation and reduced lipid droplets
We stimulated 3T3-L1 cells with H 2 O 2 at the indicated concentrations (0-, 250-, and 500-μM) for 24 h, and observed that the protein levels of FABP4, PPAR-γ, C/EBP-α, adiponectin were significantly lower in 3T3-cells treated with H 2 O 2 in a dose-dependent manner.In contrast, the proteins levels of CTSK and Pref-1 were significantly increased by the H 2 O 2 treatment (Figure S8A,B).Photomicrographs of quantitative Oil Red O staining also revealed that H 2 O 2 realized a dose-dependent reduction of lipid droplet accumulation (Figure S8C,D).

| DISCUSSION
This study focused on novel role(s) of the interaction between the DPP4-GLP-1/GLP-1R and adiponectin/CTSK axes in adipose differentiation and remodeling in mice under restraint stress conditions.The most significant finding of this investigation is that mice lacking the DPP4 gene were resistant to stress-related adipose loss and remodeling.At the molecular and cellular levels, DPP4 deficiency was shown to prevent harmful changes in the following ways: 1) increasing TNF-α and CTSK levels and decreasing GLP-1 and adiponectin levels in the plasma and/or adipose tissue; 2) enhancing the levels of adipose differentiation inhibition-related molecules (Pref-1, Wnt5α, Wnt11, and βcatenin); 3) reducing the levels of adipogenesis-related molecules (C/EBP-α, C/EBP-β, and PPAR-γ) and adipocyte differentiation-related molecules (perilipin-2, SFRP4, adiponectin, and FABP4); and 4) increasing the levels of macrophage and T-cell (CD4 + and CD8 + ) infiltration.All of these beneficial effects were reproduced by a pharmacological GLP-1 receptor agonist, exenatide, in vivo and in vitro.In addition, CTSK deficiency was also found to have a beneficial effect on adipose tissue in mice in response to stress.In vitro, the silencing of CTSK facilitated and the overexpression of CTSK disturbed lipid accumulation in association with changes in the levels of the targeted proteins (C/EBP-α, PPAR-γ, and Pref-1), and these results were prevented by treatment with a GLP-1 analogue in 3T3-L1 cells in the presence of S-serum, providing evidence and a mechanistic explanation for the participation of CTSK in DPP4/ GLP-1-adiponectin signaling in stressed adipose differentiation and remodeling.
It has been reported that plasma levels of DPP4 were increased in patients with metabolic cardiovascular disease with and without diabetes mellitus. 16,17Under our experimental conditions, the ability of stress to enhance DPP4 activity is likely to have contributed to the activation of the adipocyte lipid droplet loss and dedifferentiation.The quantitative qPCR and histological data demonstrated that chronic stress caused adipocyte dedifferentiation and accompanied by the harmful changes of the targeted adipocyte dedifferentiation/differentiationrelated (Pref-1, Wnt5α, Wnt11, βcatenin/perilipin2, SFRP4, and FABP4) and adipogenesis-related (C/EBP-α, C/EBP-β, PPAR-γ) two-side factors, and these changes were significantly reversed by DPP4 deletion in stressed mice.However, DPP4 knockout had no influence on the changes to the adipose tissue of mice under non-stress conditions.An in vitro study reported a metabolic role of DPP4 in primary human adipocyte proliferation. 18ecause DPP4 inhibition can have beneficial effects against metabolic disorders including obesity and diabetes mellitus, 32,33 we propose that DPP4 works as an important regulator of stress-induced adipose responses.
It is well-known that inflammation plays a fundamental role in obesity and metabolic disease. 34Accumulating evidence indicates that chronic stress can cause an inflammatory overaction in different tissues, including adipose and vascular tissues. 7,20Our observations here show that the stressed inguinal adipose tissue had increased plasma TNF-α levels and increased numbers of macrophages as well as CD4 + and CD8 + immune T cells, and these changes were rectified by DPP4 deletion.Hepatocyte-secreted DPP4 in obesity promotes adipose inflammation and insulin resistance. 35DPP4 has been shown to participate in Lats1/2-deficient adipocytes dedifferentiation into progenitor cells and conversion to myofibroblasts upon TGF-β stimulation. 36Moreover, a DPP4 inhibitor was reported to increase serum levels of anti-inflammatory sFRP5, which might be beneficial in the treatment of COVID-19, reflecting a state of sFRP5 deficiency. 37Thus, the ability of DPP4 inhibition to lessen over-activated proteolysis may have a salutary effect on adipose tissue by inhibiting inflammatory responses, thereby ameliorating adipocyte dedifferentiation and remodeling in mice under chronic stress conditions.
Accumulating evidence shows that oxidative stress also plays an important role in inflammatory and metabolic disease inception and progression. 38We previously reported that DPP4 inhibition lowered the components of NAD(P)H oxidase (gp91 phox , p22 phox ) mRNAs and/or proteins and oxidative stress production in the injured arterial tissues of mice under chronic stress conditions. 20,23Our observations here show that hydrogen peroxide lowered the levels of FABP4, adiponectin, C-EBP-α, and PPAR-γ and increased the levels of Pref-1 and CTSK, leading to lipid droplet loss L1 cells.In other experiments we found that the stressed adipose tissue had increased NADPH oxidase activity; this change was rectified by DPP4 deletion in mice under our experimental conditions.Oxidative stress can disturb adipocyte differentiation and adipogenesis via the modulation of C-X-C motif chemokine ligand 5 production. 39Collectively, these observations suggested that DPP4 inhibition-mediated amelioration of adipose dysfunction and remodeling might be in part due to a reduction of oxidative stress production in mice exposed to chronic stress.
Our preclinical mouse data revealed that the stressed mice had lower plasma GLP-1 and adiponectin levels and lower adipose adiponectin gene expression compared to the non-stress mice.Similar to the findings in DPP4deficient mice, our present experiments showed that exenatide therapy ameliorated the stressed adipose morphological changes accompanied with an improvement of plasma and adipose adiponectin levels as well as an amelioration of the biological changes of the investigated adipocyte differentiation-and adipogenesis-related molecules in mice.Exenatide also lowered plasma TNF-α levels and adipose inflammation responses, including both macrophage and T-cell infiltrations in response to stress.
A previous report showed that GLP-1-mediated antiinflammatory effects are associated with changes in the adipogenic potential and ability of human adipose tissue to expand, via activation of the canonical GLP-1R. 21In addition, adiponectin is a classical adipocytokine, mainly secreted by adipose tissue, which regulates adipocyte function and differentiation. 40,41Anti-inflammatory and anti-oxidative effects of GLP-1 analogues in the initiation and progression of metabolic diseases such as obesity and diabetes mellitus have been sufficiently demonstrated in previous studies. 42In addition, DPP4 inhibition has been shown to exert pleiotropic effects via an augmentation of GLP-1/GLP-1R signaling. 14Taken together, these results show that the up-regulation of adipose GLP-1/GLP-1Radiponectin signaling by DPP4 inhibition could represent a common mechanism in the treatment of adipose differentiation and remodeling in response to stress.This notion is further supported by our finding that GLP-1R stimulation with exenatide or adiponectin depletion with nAdipo, respectively, altered C/EBP-α, PPAR-γ, and Pref-1 protein levels, leading to an increase or decrease in lipid accumulation in 3T3-L1 cells under S-serum conditions.
The ability of chronic stress to enhance CTSK expression and activity probably contributed to adipose dysfunction mice under our conditions.A recent comprehensive review documented the role cysteinyl cathepsins in metabolic cardiovascular disease. 22Cathepsins function as modulators of cellular events (e.g., migration, proliferation, apoptosis) in various cell lines, and this modulation is dependent on the proteolytic and nonproteolytic functions of the respective cells. 43It was reported that CTSL activity leads to an enhancement of human and murine pre-adipocyte lipid accumulation and adipogenesis by modulating fibronectin degradation. 28CTSS has been shown to control angiogenesis and tumor growth via the production of type IV collagen-derived anti-angiogenic peptides and the generation of bioactive pro-angiogenic γ2 fragments from laminin-5. 44CTSK can also promote vascular endothelial cell and progenitor cell proliferation and neovascularization activity by Notch1 activation in vivo and in vitro under hypoxic conditions. 25In agreement with previous studies reporting that CTSK deletion conferred protection against vascular and kidney damages in an experimental animal model, 5,26 we observed that mice whose adipose tissue lacked CTSK were resistant to chronic stress.Our present findings demonstrated that the iWAT of stressed CTSK −/− mice exhibited lower CD4 + and CD68 + macrophage infiltration as well as lower frequency of small-size adipocytes compared with the iWAT of stressed CTSK +/+ mice.As anticipated, stressed CTSK −/− mice had a reduction in the levels of plasma DPP4 and TNF-α, whereas they had increased levels of plasma GLP-1 and adiponectin.In 3T3-L1 cells, CTSK silencing or overexpression, respectively, reduced or increased the Pref-1, C-EBP-α, and PPAR-γ protein levels and lipid accumulation, and these effects were positively reversed by exenatide.Pref-1 is a transcription factor known to inhibit adipocyte differentiation and is tightly regulated by GLP-1. 45PPAR-γ and C/EBP-α are considered to be important activators of adipogenesis, and these two factors have been shown to function synergistically in the adipogenesis process. 46It was reported that full-length adiponectin binding to both types of its receptors mediated PPAR-γ activation and, consequently, fattyacid oxidation. 47Taking these results together, although other factors may be involved in the stress-induced adipose response, we favor our hypothesis that a stressmediated increase in DPP4 had a harmful effect on the GLP-1/adiponectin axis, leading to enhancement of CTSK expression and activity and thereby promoting the adipose dedifferentiation and remodeling under conditions of chronic stress.
Another implication of this study is the potential use of elevated plasma DPP4, TNF-α, and CTSK and decreased GLP-1 and to the presence of stress in animals.Our findings indicate that plasma DPP4, TNF-α, GLP-1, adiponectin, and CTSK levels were sensitive to the chronic stress, and that the noninvasive measurement of the plasma levels of these five molecules would be helpful for assessing adipose-tissue injury in animals exposed to chronic stress.However, the roles of these inflammatory and metabolic factors in the initiation and progression of various diseases associated with modern stressors (including work-related stress, social anxiety, environmental stress, and natural disasters) should be further studied in prospective and retrospective cohort clinical trials, in which many of the factors that influence case control trials are not an issue.
This study had several limitations.First, the animal model of chronic variable stress used herein cannot completely mimic the status of human psychological stress.Second, it was too hard to separate and evaluate each adipose tissue mass around the organs (i.e., epicardial, periadventitial, and perirenal adipose tissues) and we did not evaluate the plasma glucocorticoid levels of the mice.Finally, although Joffin and colleagues demonstrated that PDGFRβ + DPP4 − subpopulations are major adipogenic precursors and PDGFRβ + DPP4 + subpopulations are key precursors of inflammation, 48 we were not able to isolate PDGFRβ + DPP4 − and PDGFRβ + /DPP4 + cell subpopulations to confirm their conclusions in vitro.Further research is necessary to investigate these issues.
Our current findings help explain how stress interferes with metabolic adipose disorder, and they also clarify the role of the interaction between the DPP4-GLP-1/ GLP-1R and adiponectin/CTSK axes in the stress-related adipocyte dedifferentiation and remodeling processes.In animals exposed to stress, the inactivation of GLP-1/ GLP-1R caused by elevated DPP4 signaled adipocytes to enhance CTSK expression and activity through adiponectin reduction, accelerated adipocyte dedifferentiation, and facilitated adipose wasting.These events resulted in extensive lipolysis and blood hyperlipidemia (TG, TC, and FFA) and promoted metabolic disorder.DPP4 inhibition or the activation of the GLP-1/GLP-1R signaling limited the imbalance between DPP4 and GLP-1 in the blood and adipose, supporting the notion that GLP-1/GLP-1R signaling and targeting of the adiponectin/CTSK axis should be explored as a potential avenue for clinical therapy.

F I G U R E 1
The 2-week chronic stress protocol elevated the CTSK expression and increased the plasma DPP4 level.(A) Diagram of the weighing and sampling process of iWAT under variable stress at specified time points.(B) Photograph of iWAT and quantitative data showing the weights of iWAT in both groups (n = 10/group).Scale bar: 5 mm.(C) Representative gel images and quantitative data showing the protein level of CTSK in two experimental groups (n = 3/group).(D) Relative mRNA levels of the CTSK in both groups (n = 7/group).(E) The plasma levels of DPP4 and GLP-1 in the two experimental groups (n = 9/group).(F) H&E staining images and the combined quantitative data showing the frequency of adipocyte area in the two experimental groups (n = 6/group).Scale bar: 75 μm.(G) The plasma levels of TC, TG and FFA in both groups (n = 9/group).Data are expressed as mean ± SEM.Statistical significance was assessed by Student's t-test for (B) through (E), and (G).**p < .01,***p < .001.

F I G U R E 2
DPP4 deletion decreased adipose wasting in the mice subjected to the 2-week stress protocol.(A,B) Quantitative data showing the body weights and iWAT weights in DPP4 +/+ and DPP4 −/− non-stress or stressed mice (n = 10/group).(C,D) The levels of plasma DPP4 activity, GLP-1, adiponectin and TNF-α in the four experimental groups (n = 7-8/group).(E) Representative H&E staining images and the combined quantitative data showing the frequency of adipocyte area in the four groups (n = 6/group).Scale bar: 75 μm.Results are expressed as mean ± SEM.Statistical significance was assessed by unpaired Student's t-test (B), one-way ANOVA (C,D), or two-way ANOVA (A).NS indicates not significant.*p < .05,**p < .01,***p < .001.

F I G U R E 4
The pharmacological GLP-1 receptor agonist exenatide mitigated adipose wasting and modulated adipocyte differentiation signaling in DPP4 +/+ mice that were subjected to 2 weeks of stress.(A,B) Body weights and the weights of iWAT in the three experimental groups (n = 10/group).(C) H&E staining images and the combined quantitative data showing the frequency of adipocyte area in the three experimental groups (n = 6/group).(D,E) Representative immunoblot images and combined quantitative data showing the levels of FABP4, adiponectin, CTSK, Pref-1, C/EBP-α and PPAR-γ proteins in the three experimental groups (n = 4/group).Scale bar: 75 μm.Results are expressed as mean ± SEM.Statistical significance was assessed by one-way ANOVA for (B,E), and two-way ANOVA for (A).NS indicates not significant.*p < .05,**p < .01,***p < .001.

F
I G U R E 5 CTSK deficiency ameliorated adipose wasting, DPP4 activity and adipose inflammation.(A,B) The body weights and weights of iWAT in both groups (n = 8/group).(C,D) The DPP4 activity of adipose tissue, and the levels of plasma TNF-α, GLP-1 and adiponectin after stress in the two experimental groups (n = 8/group).(E,F) Quantitative data showing CD4 + T cells and CD68 + macrophages expression in the two groups (n = 8/group).(G) H&E staining images and the combined quantitative data showing the frequency of adipocyte area in the two experimental groups (n = 6/group).Scale bar: 100 μm.Data are shown as mean ± SEM.Statistical significance was assessed by Student's t-test for (A) through (F).*p < .05,**p < .01,***p < .001.

F
I G U R E 6 CTSK silencing alleviated 3T3-L1 cell differentiation in response to S-serum.Following transfection with a nontargeting negative control (NC) and siCTSK-418 for 48 h, the 3T3-L1 cells were treated with 5% stress serum (S-serum) for 24 h with or without exenatide.(A) 3T3-L1 cells were transfected with siCTSK-418 for 6 h, and the fluorescence was observed under a microscope for the confirmation of successful siCTSK-418 transfection.Scale bar: 50 μm.(B) Western blot analysis of the investigated molecular protein levels (CTSK) in the 5%S-serum-NC group and 5%S-serum-siCTSK-418 group (n = 4/group).(C) Relative mRNA levels of CTSK in both groups (n = 5/group).(D,E) Representative images and quantitative data of the levels of Pref-1, C/EBP-α and PPAR-γ in the four experimental groups (n = 3/group).(F) Oil Red O staining images combined with the quantitative data showing the area of lipid droplets in the 5%Sserum-NC group and 5%S-serum-siCTSK group (n = 8/group).Scale bar: 100 μm.Data are shown as mean ± SEM.Statistical significance was assessed by unpaired Student's t-test (B, C and F) or one-way ANOVA (E).NS indicates not significant.*p < .05,**p < .01,***p < .001.

F
I G U R E 7 CTSK overexpression by pl-CTSK accelerated S-serum-induced adipocyte dedifferentiation in 3T3-L1 cells.Following transfection with empty vector (Cont) and pl-CTSK for 48 h, the 3T3-L1 cells were treated by 5% S-serum stimulations for 24 h.(A) 3T3-L1 cells were transfected with GFP-plCTSK for 6 h, and the fluorescence was observed under a microscope for the confirmation of successful pl-CTSK transfection.Scale bar: 50 μm.(B) Western blot analysis of the levels of CTSK in the 5%S-serum-Cont group and 5%S-serum-plCTSK group (n = 4/group).(C) Relative mRNA levels of CTSK in both groups (n = 5/group).(D,E) Western blot images and quantification of the investigated molecular protein levels of Pref-1, C/EBP-α and PPAR-γ in the four experimental groups (n = 3/group).(F) Oil Red O staining images combined with the quantitative data showing the area of lipid droplets in the 5%S-serum-Cont group and 5%S-serum-plCTSK group (n = 8/group).Scale bar: 100 μm.Data are expressed as mean ± SEM.Statistical significance was assessed by unpaired Student's t-test (B, C and F) or one-way ANOVA (E).NS indicates not significant.*p < .05,**p < .01,***p < .001.

F I G U R E 8
Mouse neutralizing antibody against adiponectin (nAdipo) prevented an increase in adipocyte differentiation-related protein levels in 3T3-L1 cells.3T3-L1 cells were treated with 5% stress serum (S-Serum) or 5% non-stress serum (NS-Serum) and then further incubated in 30 nmol exenatide with or without nAdipo.Four groups were established, i.e., the 5%NS-Serum (C), 5%S-Serum (S), 5%S-Serum-Exe (E) and 5%S-Serum-Exe-nAdipo (A) groups.(A,B) Representative Western blotting images and quantitative data of the investigated molecular protein levels of Pref-1, C/EBP-α and PPAR-γ in the four experimental groups (n = 3/group).(C) Relative mRNA levels of CTSK in the four experimental groups (n = 5/group).(D,E) Oil Red O images and the area of lipid droplets in the four groups (n = 8/ group).Scale bars: 100 μm.Data are expressed as mean ± SEM.Statistical significance was assessed by one-way ANOVA for the experiments in panels (B, C and E).NS indicates not significant.*p < .05,**p < .01,***p < .001.(F) Summary scheme: DPP4 disturbs adipocyte differentiation via the negative regulation of the GLP-1/adiponectin-CTSK axis under chronic stress conditions.
Mouse primer sequences used for the quantitative real-time PCR.