Endogenous sulfur dioxide is a novel adipocyte-derived inflammatory inhibitor

The present study was designed to determine whether sulfur dioxide (SO2) could be endogenously produced in adipocyte and served as a novel adipocyte-derived inflammatory inhibitor. SO2 was detected in adipose tissue using high-performance liquid chromatography with fluorescence detection. SO2 synthase aspartate aminotransferase (AAT1 and AAT2) mRNA and protein expressions in adipose tissues were measured. For in vitro study, 3T3-L1 adipocytes were cultured, infected with adenovirus carrying AAT1 gene or lentivirus carrying shRNA to AAT1, and then treated with tumor necrosis factor-α (TNF-α). We found that endogenous SO2/AAT pathway existed in adipose tissues including perivascular, perirenal, epididymal, subcutaneous and brown adipose tissue. AAT1 overexpression significantly increased SO2 production and inhibited TNF-α-induced inflammatory factors, monocyte chemoattractant protein-1 (MCP-1) and interleukin-8 (IL-8) secretion from 3T3-L1 adipocytes. By contrast, AAT1 knockdown decreased SO2 production and exacerbated TNF-α-stimulated MCP-1 and IL-8 secretion. Mechanistically, AAT1 overexpression attenuated TNF-α-induced IκBα phosphorylation and degradation, and nuclear factor-κB (NF-κB) p65 phosphorylation, while AAT1 knockdown aggravated TNF-α-activated NF-κB pathway, which was blocked by SO2. NF-κB inhibitors, PDTC or Bay 11-7082, abolished excessive p65 phosphorylation and adipocyte inflammation induced by AAT1 knockdown. This is the first report to suggest that endogenous SO2 is a novel adipocyte-derived inflammatory inhibitor.


Results
Endogenous SO 2 /AAT pathway existed in adipose tissue of rats. We detected the concentration of SO 2 in respective rat adipose tissues, including perivascular, adipose tissue (1.53 ± 0.33 μ mol/g protein), perirenal adipose tissue (1.54 ± 0.17 μ mol/g protein), epididymal adipose tissue (0.65 ± 0.26 μ mol/g protein), subcutaneous adipose tissue (0.67 ± 0.32 μ mol/g protein), and brown adipose tissue (1.34 ± 0.37 μ mol/g protein) (Fig. 1a). The content was comparable to that in the spleen and kidney, but lower than that in the heart, lung, liver and aorta (Fig. 1a). SO 2 generation in mammals mainly depends on two enzymes AAT1 and AAT2. RT-PCR revealed that both AAT1 and AAT2 mRNA were expressed in perivascular, perirenal, epididymal, subcutaneous and brown adipose tissue, with heart, lung, liver, spleen, kidney and aorta used as a positive control (Fig. 1b). Moreover, AAT1 and AAT2 protein expressions were also detected in respective rat adipose tissues by western blot analysis (Fig. 1c). The evaluation by the production of SO 2 in the presence of L-cysteine in the adipose tissue homogenate showed that AAT activities were 747.37 ± 227.09 nmol/min/g in perivascular adipose tissue, 1745.18 ± 279.90 nmol/min/g in perirenal adipose tissue, 1516.22 ± 286.60 nmol/min/g in epididymal adipose tissue, 1253.75 ± 152.93 nmol/ min/g in subcutaneous adipose tissue, and 927.41 ± 179.24 nmol/min/g in brown adipose tissue (Fig. 1d). Protein locations of AAT1 and AAT2 in different adipose tissues stained by immunohistochemical analysis are shown in Fig. 1e-g.

Endogenous SO 2 inhibited TNF-α-induced NF-κB pathway activation in adipocytes.
The NF-κ B pathway plays an important role in regulating inflammation in adipocytes. We then studied the effect of endogenous SO 2 on the NF-κ B pathway during TNF-α -induced inflammatory factors secretion in adipocyte. Western blot analysis revealed that the phosphorylation of NF-κ B p65 was increased markedly during stimulation with TNF-α , whereas AAT1 overexpression significantly inhibited NF-κ B p65 phosphorylation induced by TNF-α in adipocytes (Fig. 4a).
The activation of NF-κ B is processed by the phosphorylation and degradation of Iκ Bα . Western blot analysis revealed that TNF-α treatment increased the phosphorylation of Iκ Bα and induced Iκ Bα degradation, while AAT1 overexpression blocked the effects of TNF-α on Iκ Bα degradation and phosphorylation (Fig. 4a).
On the contrary, a 1.5-fold increase in p65 phosphorylation was observed in response to TNF-α as early as 15 min after TNF-α addition in sh-AAT1-infected adipocytes. Increased levels persisted for at least 2 h. However, the maximal increase in p65 phosphorylation in response to TNF-α was observed at 30 min after TNF-α addition in sh-control-infected 3T3-L1 adipocytes, although it was not as high as that in adipocytes treated with AAT1 knockdown (Fig. 4b). Also, AAT1 knockdown aggravated basal and TNF-α -induced NF-κ B p65 phosphorylation, Iκ Bα phosphorylation and its degradation (Fig. 4c). TNF-α binds to TNF-α receptor 1 (TNFR1) or TNF-α receptor 2 (TNFR2), leading to the activation of NF-κ B. Herein, we investigated the effect of endogenous SO 2 (perivascular adipose tissue, perirenal adipose tissue, epididymal adipose tissue, subcutaneous adipose tissue, brown adipose tissue, heart, lung, liver, spleen, kidney and aorta). The bands of AAT1 and AAT2 were exposed twice. (d) Measurement of SO 2 production from different rat tissues by addition of L-cysteine plus pyridoxal 5′ -phosphate to tissue homogenate and incubation for 90 min. (e) Expression of AAT1 in different rat tissues using immunohistochemistry: i, perivascular adipose tissue; ii, perirenal adipose tissue; iii, epididymal adipose tissue; iv, subcutaneous adipose tissue; v, brown adipose tissue; vi, heart; vii, liver; viii, aorta; and ix IgG as a negative control. (f) Expression of AAT2 in different rat tissues using immunohistochemistry: i, perivascular adipose tissue; ii, perirenal adipose tissue; iii, epididymal adipose tissue; iv, subcutaneous adipose tissue; v, brown adipose tissue; vi, heart; vii, liver; viii, aorta; and ix IgG as a negative control. (g) Hematoxylin and eosin (HE) staining of different rat tissues: i, perivascular adipose tissue; ii, perirenal adipose tissue; iii, epididymal adipose tissue; iv, subcutaneous adipose tissue; v, brown adipose tissue; vi, heart; vii, liver; and viii aorta.
on the abundance of TNF-α receptors. The data showed that there was no statistically significant difference in protein expressions of TNFR1 and TNFR2 between adipocytes with AAT1 knockdown and control adipocytes ( Supplementary Fig. 1). These results indicated that endogenous SO 2 inhibited TNF-α -induced NF-κ B phosphorylation at least partly in association with suppression of Iκ Bα phosphorylation and degradation.

Discussion
SO 2 is a novel gasotransmitter in the cardiovascular system and plays vital roles in regulation of cardiovascular system homeostasis [5][6][7] . L-cysteine is the major precursor to endogenous SO 2 . It can be metabolized to L-cysteine sulfinate by cysteine dioxygenase (CDO). L-cysteine sulfinate is then metabolized to β -sulfinylpyruvate via AAT1 and AAT2, and finally spontaneously decomposes to pyruvate and SO 2 5,8,9 . Some of the endogenous SO 2 is hydrated to sulfite, whereas the other part stays in the gaseous form 5,10,11 . Our previous study demonstrated that AAT expression is accompanied by the generation of SO 2 in several tissues and organs such as the stomach, heart, cerebral gray matters, cerebral white matter, pancreas, lung, kidney, spleen, liver and aorta in mammals 12 . The present study showed that different adipose tissues could also generate endogenous SO 2 , such as epididymal, subcutaneous, perirenal, perivascular and brown adipose tissue.  Two types of isozymes of AAT exist, including AAT1 and AAT2. AAT1 is mainly localized in the cell cytoplasm and AAT2 is localized in the cell mitochondria 12,13 . Nowadays, both are found to be present in most cells but erythrocytes from animal tissues 12,14 . There are also some differences between AAT1 and AAT2 in terms of functional effects. For example, AAT1 and AAT2 perform different function in the malate-aspartate shuttle 12 . However, until now, we have not understood completely the distribution and function of the SO 2 generating enzymes in adipose tissues. Thus, the present study indicated that the SO 2 generating enzymes AAT1 and AAT2 were distributed in adipose tissues of rats detected by real-time PCR, Western blot and immunohistochemical analysis.
Many studies demonstrated that chronic inflammation in adipose tissue was mechanically linked to insulin resistance. Adipose tissue is considered to be an active endocrine organ. Inflammatory cytokines secreted from adipose tissue play vital roles in the development of insulin resistance, obesity and type 2 diabetes 1,15 . In particular, pro-inflammatory cytokines MCP-1 and IL-8 are the key regulators of immune cells such as macrophage recruitment, infiltration and activation in adipose tissue in obesity. This process leads to chronic inflammation in adipose tissue and the pathogenesis of obesity and insulin resistance. But the exact interaction between adipocytes and immune cells in the occurrence and development of chronic inflammation still requires further study. Inflammatory cytokines from adipocytes play an important role in the interaction between adipocytes and macrophages, resulting in chronic inflammation. We studied the possible impact of endogenous SO 2 on MCP-1 and IL-8 secretion in 3T3-L1 adipocytes. The results showed that AAT1 overexpression decreased MCP-1 and IL-8 secretion, whereas AAT1 knockdown increased MCP-1 and IL-8 secretion from TNF-α -stimulated 3T3-L1 adipocytes. These results suggested that endogenous SO 2 might inhibit secretion of MCP-1 and IL-8 in TNF-α -induced adipocytes, which might play an important role in the protection against adipose inflammation-related diseases such as insulin resistance and obesity.
We then studied how endogenous SO 2 inhibited MCP-1 and IL-8 secretion. It is widely accepted that nuclear factor-κ B (NF-κ B) is required for induction of MCP-1 and IL-8 expression and secretion in adipocytes. NF-κ B is a key transcription factor in the activation of inflammation in adipocytes. The p65 protein is the key transcriptionally active component of NF-κ B, which consists primarily of dimers of the two subunits p50 and p65. Iκ B, an inhibitory protein in the NF-κ B signaling pathway, exists in the cytoplasm in an inactive form associated with the dimer of p65 and p50. During activation of NF-κ B, Iκ B subunit is released from the complex and phosphorylation and translocation of the dimer to the nucleus take place 16,17 . In the nucleus, it regulates the transcription of inflammatory genes such as MCP-1 and IL-8. To test the hypothesis that NF-κ B was likely to be involved in the underlying mechanism by which endogenous SO 2 inhibited secretion of MCP-1 and IL-8 in TNF-α -stimulated adipocytes, we first examined the effect of endogenous SO 2 on the phosphorylation of NF-κ B p65 in adipocytes. The results demonstrated that AAT1 overexpression significantly inhibited phosphorylation of NF-κ B p65, whereas AAT1 knockdown aggravated it in 3T3-L1 adipocytes stimulated with TNF-α . Furthermore, we found that AAT1 overexpression could suppress the TNF-α -stimulated phosphorylation and degradation of Iκ Bα , which was prevented by AAT1 knockdown, suggesting that endogenous SO 2 inhibited NF-κ B activation in association with preventing Iκ Bα activation. NF-κ B inhibitors remarkably antagonized AAT1 deficiency-induced p65 phosphorylation and MCP-1 and IL-8 secretion in 3T3-L1 adipocytes. Thus, these findings implied that the role of SO 2 in the regulation of adipocyte inflammation was at least partly mediated by NF-κ B pathway.
To further investigate whether the effects of AAT manipulation on adipocyte NF-κ B pathway activation was caused by SO 2 , we added SO 2 derivatives to 3T3-L1 adipocytes with AAT1-knock down. The results showed that SO 2 derivatives at 100 μ mol/L significantly inhibited AAT1 silencing-aggravated NF-κ B p65 phosphorylation in 3T3-L1 adipocytes with or without TNF-α stimulation, demonstrating that SO 2 generated by AAT1 played a crucially protective role against adipocyte NF-κ B activation in adipocytes with AAT1 knock-down.
TNF-α is one of the primary mediator of the inflammatory response in obesity and insulin resistance. The present study showed that MCP-1 concentration in supernatant from sh-Control-infected and sh-AAT1-infected 3T3-L1 adipocytes increased by 60.62% and 58.98%, respectively, after stimulation with TNF-α for 2 h. And IL-8 concentration in supernatant from sh-Control-infected and sh-AAT1-infected 3T3-L1 adipocytes increased by 65.5% and 44.87%, respectively, after stimulation with TNF-α for 2 h. These findings suggested that it was not an over-sensitization of AAT1 silencing cells to TNF-α . Transducing TNF-α signals through TNF-α receptors resulted in the activation of NF-κ B pathway and downstream inflammatory gene transcription. Our data demonstrated that AAT1 deficiency did not affect the protein expressions of two TNF-α receptors, TNFR1 and TNFR2, compared to the control group in the presence or absence of TNF-α .
The limitation of the present study was that we could not exactly define if the effects of bisulfite or sulfite were involved in the gas SO 2 function. SO 2 is mainly derived from sulfite (SO 3 2− ) and bisulfite (HSO 3 − ) generated after SO 2 dissolves in cellular cytoplasm 18 . SO 2 can easily be hydrated to produce sulfurous acid, which subsequently dissociates to form its derivatives, sulfite and bisulfite (3:1 mole ratios in neutral fluids) [19][20][21] . Therefore, SO 2 , sulfite and bisulfite might exist in vivo as a mixture which could not be completely distinguished.
Scientific RepoRts | 6:27026 | DOI: 10.1038/srep27026 adipocyte-derived inflammatory inhibitor and might be a new therapeutic target for inflammation-related diseases, such as obesity and insulin resistance.

Methods
Animal preparation. Animal care and experimental protocols were in accordance with the Animal Management Rule of the Ministry of Health, China. All experimental protocols were approved by the Animal Research Ethics Committee of Peking University First Hospital, Beijing, China (Permit Number: J201335). Eightweek-old male Sprague-Dawley (SD) rats (180-200 g) were obtained from the Experimental Animal Center, Peking University Health Science Center, Beijing, China. The rats were housed under special pathogen-free conditions, at a temperature of 22 °C with 40% humidity and a 12 hour light/12 hour dark cycle.
Preparation of different tissue samples in rats. Ten male SD rats were anaesthetized using urethane (1 g/kg) by intraperitoneal injection, and then the perivascular adipose tissue, perirenal adipose tissue, epididymal adipose tissue, subcutaneous adipose tissue, brown adipose tissue, heart, lung, liver, spleen, kidney and aorta were harvested rapidly for the following analysis. For SO 2 determination via PCR and Western blot, samples were frozen and stored in liquid nitrogen. For immunohistochemical analysis, tissues were fixed in 4% polyoxymethylene.
Determination of SO 2 concentration of tissue and adipocyte supernatant. The tissues were homogenized in 0.1 mol/L phosphate-buffered saline (PBS, pH 7.4, 10 mL/g tissue), and then centrifuged at 12,000 g for 30 min at 4 °C. The supernatants obtained were prepared for SO 2 content and protein assay determination. Cell supernatants for SO 2 concentration determination were taken from TNF-α -stimulated adipocytes. SO 2 concentrations were measured using high-performance liquid chromatography with fluorescence detection (HPLC-FD, Agilent 1200 series, Agilent Technologies, Palo Alto, CA, USA) 22   The neutralized supernatant was used for HPLC-FD. Sulfite-bimane was measured by excitation at 392 nm and emission at 479 nm. Quantification was carried out by the standardization of sodium sulfite.

Determination of aspartate aminotransferase (AAT) 1 and AAT2 mRNA in tissues by real-time PCR.
Total RNA in tissues was extracted by Trizol reagent and reverse transcribed by oligo d(T) 18 primer and M-MuLV reverse transcriptase. Real-time PCR was performed on an ABI PRISM 7300 instrument (ABI USA Sales Corp, Los Angeles, CA, USA). Samples and standard DNA were determined in duplicate. The PCR condition was predenaturing at 95 °C for 5 min, then 95 °C for 15 s, and 60 °C for 1 min for 35 cycles. The amount of β -actin cDNA in the sample was used to calibrate the sample amount used for the determination. The PCR products were separated in agarose gel electrophoresis.
Sequences of the primers and TaqMan probes were provided as follows. For AAT1, forward, 5′-CCAGGGAGCTCGGATCGT-3′, reverse, 5′-GCCATTGTCTTCACGTTTCCTT-3′, TaqMam probe, 5′-CCACCACCCTCTCCAACCCTGA-3′, the product size is 79 bp. For AAT2, forward, 5′ -GAGGGTCGGAGCCAGCT T-3′ , reverse, 5′ -GT T TCCCCAGGATGGT T TGG-3′, TaqMan probe, 5′-TTTAAGTTCAGCCGAGATGTCTTTC-3′, the product size is 82 bp; for β-actin, forward, 5′-ACCCGCGAGTACAACCTTCTT-3′, reverse, 5′-TATCGTCATCCATGGCGAACT-3′, TaqMan probe, 5′-CCTCCGTCGCCGGTCCACAC-3′, the product size is 80 bp. Expression of AAT1 and AAT2 in different tissues using immunohistochemical analysis. We detected the expression of AAT1 and AAT2 in different tissues by using immunohistochemical analysis. Sections of tissues were dewaxed by dimethylbenzene, washed three times with PBS, and then treated with 3% H 2 O 2 for 10 min. The slides were blocked with 5% bovine serum albumin (BSA) working fluid at 37 °C for 30 min, and incubated overnight at 4 °C with AAT1 and AAT2 antibodies (dilutions of 1:25, respectively). Slides were then washed with PBS. Biotinylated anti-rabbit or anti-mouse IgG was incubated for 60 min at 37 °C. After the slides were rinsed in PBS three times, the sections were stained with 3,3′ -diaminobenzidine (DAB) to develop color. The sections were dehydrated and mounted. Positive signals were defined as brown granules in tissues under light microscopy. For negative controls, primary incubation was performed with non-immune goat serum instead of primary antibodies. 3T3-L1 cell culture. Murine 3T3-L1 cell line was obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The 3T3-L1 preadipocytes were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS, 2 mmol/L glutamine and 20 mmol/L HEPES (pH 7.4) in a humidified atmosphere of 5% CO 2 at 37 °C. After reaching 100% confluence, 3T3-L1 preadipocytes were stimulated to differentiate using a differentiation mixture containing 0.25 μ M dexamethasone, 0.5 mM IBMX and 1 μ M insulin in DMEM with 10% FBS. After 2 days, the medium was replaced with DMEM supplemented with 10% FBS and 1 μ M insulin. Cultures were incubated for 2 days; afterwards the culture medium was replaced again with DMEM supplemented with 10% FBS and replaced at 2 day intervals until the adenovirus or lentivirus infection was performed on days 6-8 23-25 . 3T3-L1 cell treatment. The adenovirus containing the cDNA encoding AAT1 was purchased from Vigene Biosciences (Shandong, China). The lentivirus carrying shRNA to AAT1 was purchased from Pregene (Beijing, China). 3T3-L1 adipocytes were transduced at a multiplicity of infection of 100 PFU/cell for 24 h 21 . Transduced cells were incubated for 48-72 h at 37 °C in 5% CO 2 , followed by a stimulation with TNF-α (10 ng/ml) for 30 min or 2 h. Determination of SO 2 production. The different rat tissue homogenate was prepared, then mixed with the reaction buffer including 100 mmol/L potassium phosphate buffer (pH 7.4), 10 mmol/L L-cysteine and 2 mmol/L pyridoxal 5′ -phosphate. The mixture was incubated at 37 °C for 90 min. For the production of SO 2 in adipocytes, at the end of treatment, cell medium was replaced with fresh medium containing 10 mmol/L L-cysteine and 2 mmol/L pyridoxal 5′ -phosphate, and then, the cells were incubated for further 4 h. Released SO 2 was detected by HPLC-FD. The rate of SO 2 production is expressed as nmol SO 2 formed from 1 g of protein per minute (nmol/ min/g protein) determined by the Bradford assay.

Determination of AAT, NF-κB and IκBα protein expression in tissues and adipocytes by
Statistical analysis. Results are expressed as means ± SEM. The analysis was done using GraphPad Prism 5. For multiple group comparisons, ANOVA followed by a post-hoc analysis (Newman-Keuls test) was used. P < 0.05 were considered statistically significant.