Formononetin inhibits lipopolysaccharide-induced release of high mobility group box 1 by upregulating SIRT1 in a PPARδ-dependent manner

Materials

Formononetin, actinomycin D (Act D), cycloheximide (CHX), Ponceau S solution, resveratrol, sirtinol, MTT, LPS (Escherichia coli 0111:B4), curcumin, genistein, and an anti-β-actin polyclonal antibody were obtained from Sigma-Aldrich Co. (St. Louis, MO, USA). GSK0660 and the luciferase assay system were purchased from Tocris Bioscience (Bristol, UK) and Promega (Madison, WI, USA), respectively. Monoclonal antibodies specific for HMGB1 and PPARδ were supplied by Epitomics (Burlingame, CA, USA). Monoclonal antibodies specific for acetyl-lysine, lamin B, and α-tubulin as well as a polyclonal antibody specific for SIRT1 were supplied by Santa Cruz Biotechnology (Dallas, TX, USA).

Cell culture

Human primary peripheral blood macrophages and RAW264.7 murine macrophage-like cells were purchased from STEMCELL Technologies (Vancouver, BC, Canada) and the Korean Cell Line Bank (Seoul, Korea), respectively. Primary human macrophages and RAW264.7 cells were maintained in RPMI (Roswell Park Memorial Institute) 1640 and DMEM (Dulbecco’s modified Eagle medium) containing antibiotics and 10% FCS at 37 °C in a 5% humidified CO₂ incubator, respectively.

Cell viability assays

RAW264.7 cells were stimulated with 30 μM formononetin for the indicated duration or the indicated dose of formononetin for 24 h in 24-well plates. Thereafter, MTT assays and trypan blue exclusion were performed to determine the cell viability. For trypan blue exclusion, the collected cells were mixed with trypan blue solution (0.4%), and then viable cells were determined by a hemocytometer. For MTT assay, the cells were incubated for final 2h in medium containing MTT solution (0.1 mg/ml). Following removing the medium, the absorbance at 570 nm was measured using formazan crystals solution dissolved in acidified isopropanol.

Western blot analysis

Protein levels were analyzed by immunoblot as described previously (Hwang et al., 2015). Briefly, RAW264.7 cells washed with ice-cold PBS were lysed and aliquots of the resulting whole-cell lysates or conditioned media were analyzed by immunoblot with indicated antibodies. Immuno-reactive bands were detected using WesternBright ECL (Advansta Co., Menlo Park, CA, USA).

Measurement of extracellular HMGB1

Levels of HMGB1 released into culture media were determined using a previously described method (Hwang et al., 2012). Briefly, the relative amounts of HMGB1 were determined in the conditioned media of RAW264.7 cells treated with the indicated reagents for the indicated durations. The 80% ice-cold acetone was used to precipitate the proteins in the conditioned media. After centrifugation, the pellets were obtained and washed with 80% ice-cold acetone. Following resuspension in SDS-PAGE sample buffer, the levels of HMGB1 released into culture media were analyzed by immunoblot.

Fractionation of nuclear and cytoplasmic proteins

Cellular fractions were prepared using a previously described method (Hwang et al., 2015). Briefly, RAW264.7 cells were washed in PBS, suspended in lysis solution for 15 min at 4 °C to swell. Nonidet P-40 (final 0.1% concentration) was added to the lysates and then vortexed vigorously for 20 s. Following centrifugation (13,000g) for 20 s, the supernatant containing cytosolic fraction was obtained and the resulting pellet was lysed by a PRO-PREP Protein Extraction Solution. Following standing for 20 min on ice, the nuclear fraction (supernatant) was obtained by centrifugation.

Reporter gene assay

The luciferase construct containing mouse SIRT1 promoter was a gift from Dr. Toren Finkel (NIH, MD, USA). The promoter activity of SIRT1 was measured as described previously (Hwang et al., 2014). Briefly, 1 μg of the SIRT1 luciferase reporter plasmid and 0.5 μg of pSV β-Gal (SV40 β-galactosidase expression vector) were introduced into RAW264.7 cells by SuperFect reagent (Qiagen, Valencia, CA, USA). After 38 h, the cells were treated with GSK0660 for 30 min prior to stimulation with formononetin for 24 h. Then, the cells were lysed by adding the luciferase reporter lysis buffer (Promega, Madison, WI, USA) and then aliquots of the lysates were used to determine luciferase activity.

Small interfering RNA-mediated gene silencing

The indicated small interfering RNA (siRNA) was introduced into RAW264.7 cells in serum-containing medium using SuperFect (Qiagen, Valencia, CA, USA) as described previously (Hwang et al., 2014). Briefly, siRNA targeting scrambled non-specific sequences (Ambion, Austin, TX, USA), PPARδ (Ambion, Austin, TX, USA), or SIRT1 designed against nucleotides (5′-TAATATCTGAGGCACTTCA-3′ and 5′-TGAAGTGCCTCAGATATTA-3′) of mouse (Bioneer, Daejeon, Korea) was introduced into the cells for 6 h. The cells were then cultured for further 38 h in fresh medium. At which time, the indicated reagents were added into the cells for the indicated durations. Gene silencing was analyzed by immunoblot.

Real-time polymerase chain reaction

Levels of SIRT1 and HMGB1 mRNA were analyzed by real-time polymerase chain reaction (PCR) as described previously (Hwang et al., 2014). Briefly, total RNA was converted into cDNA by a reverse transcription kit (TOPscript RT DryMIX; Enzynomics, Seoul, Korea). Real-time PCR was carried out using equal amount of cDNA in a 20 μl reaction solution containing primers and 1 × SYBR PCR mix (Takara Bio Inc., Otsu, Japan). The PCR condition: initial denaturation at 94 °C for 20 min, followed by 42 cycles of 25 s at 95 °C, 44 s at 58.2 °C, and 40 s at 72 °C. The primers were as follows: SIRT1, 5′-AGAACCACCAAAGCGGAAA-3′ and 5′-TCCCACAGGAGACAGAAACC-3′; HMGB1, 5′-TACCGCCCCAAAATCAAAGG-3′ and 5′-TCTCATAGGGCTGCTTGTCA-3′; and GAPDH, 5′-CATGGCCTTCCGTGTTCCTA-3′ and 5′-CCTGCTTCACCACCTTCTTGAT-3′.

Co-immunoprecipitation

Immunoprecipitation was performed using a previously described method (Hwang et al., 2015). Briefly, the protein G Sepharose was added to whole-cell lysates to pre-clear and then the pre-cleared lysates were mixed with 1 μg of an anti-HMGB1 antibody. After incubation overnight at 4 °C, the mixture was reacted with protein G Sepharose for 4 h. Mixtures were extensively washed with PBS and then boiled in gel-loading buffer. The immunoblot analysis was performed using an anti-acetyl-lysine antibody (Santa Cruz Biotechnology, Dallas, TX, USA).

Statistical analysis

Data are expressed as mean ± standard error (SE). The significance in statistical analysis was evaluated by a one-way ANOVA, followed by Tukey–Kramer test. A value of p < 0.05 was considered statistically significant.

Formononetin inhibits LPS-induced release of HMGB1 in both murine and human macrophages

To determine the optimal concentration of formononetin, we determined the viability of RAW264.7 cells treated with various concentrations of formononetin for 24 h or with 30 μM formononetin for various durations. Treatment with concentrations of formononetin up to 30 μM did not elicit cytotoxic effects on RAW264.7 cells, and cell viability remained high following treatment with 30 μM formononetin for up to five days (Figs. 1A and 1B). Thus, we selected 30 μM formononetin as the optimal concentration for subsequent experiments using RAW264.7 cells.

Because many herbal compounds including curcumin and genistein reported to exhibit anti-inflammatory activity (Biswas & Rahman, 2008; Dharmappa et al., 2010), we compared the effect of formononetin on the HMGB1 release with curcumin and genistein in RAW264.7 cells treated with LPS. The level of HMGB1 released into culture media was increased in RAW264.7 cells exposed to LPS, and this increase was markedly reduced in the presence of herbal compounds. In particular, the effects of formononetin and curcumin were superior to those of genistein (Fig. 1C). By contrast, neither LPS nor formononetin affected the expression level of endogenous HMGB1 (Fig. 1D). Similar results were obtained from the human primary macrophages, indicating that formononetin affects LPS-induced HMGB1 release, but not HMGB1 expression, in both murine and human macrophages (Fig. 1E).

HMGB1 is reported to translocate from the nucleus into the cytoplasm in response to inflammatory signals such as LPS (Bonaldi et al., 2003; Youn & Shin, 2006). Therefore, we examined whether formononetin affects this translocation of HMGB1 in LPS-stimulated RAW264.7 cells. While translocation of HMGB1 into the cytoplasm was increased in cells exposed to LPS, this was significantly suppressed by formononetin (Fig. 1F). However, the expression of HMGB1 mRNA was not affected by formononetin in cells treated with or without LPS (Fig. 1G). These results suggest that formononetin decreases the release of HMGB1 by inhibiting its translocation rather than expression in LPS-primed RAW264.7 cells.

Formononetin upregulates SIRT1 expression in RAW264.7 cells

Formononetin increased protein expression of SIRT1 in RAW264.7 cells in a concentration- and time-dependent manner. SIRT1 protein expression was significantly increased in cells treated with 20–30 μM formononetin for 24 h (Fig. 2A) and peaked at 12–24 h in cells treated with 30 μM formononetin (Fig. 2B). Similarly, the mRNA level of SIRT1 was time-dependently upregulated by formononetin (Fig. 2C). In addition, the inhibitory effect of formononetin on the LPS-stimulated release of HMGB1 was significant at 6 h pre-treatment and the maximal inhibitory effect of formononetin was observed with a pre-treatment of 24 h which corresponds to the time of maximal induction of SIRT1 expression upon formononetin treatment (Fig. 2D).

To elucidate the mechanisms by which formononetin induces SIRT1 expression, we determined the effects of Act D (a RNA synthesis inhibitor) and CHX (a protein synthesis inhibitor). While formononetin significantly increased mRNA expression of SIRT1, this was significantly reduced in the presence of Act D or CHX (Fig. 2E). These results indicate that de novo synthesis of mRNA as well as of proteins that act on the SIRT1 gene promoter is indispensable for the induction of SIRT1 mRNA by formononetin in RAW264.7 cells.

Formononetin induces SIRT1 expression via PPARδ in RAW264.7 cells

To further examine the mechanisms by which formononetin upregulates SIRT1 expression, we evaluated the role of PPARδ, a nuclear receptor that regulates the transcription of a variety of target genes (Kidani & Bensinger, 2012; Mangelsdorf et al., 1995), by transfecting RAW264.7 cells with siRNA against PPARδ. The protein level of PPARδ was reduced in cells transfected with PPARδ-targeting siRNA, but not in cells transfected with control siRNA composed of a pool of nonspecific sequences (Fig. 2F). Transfection of PPARδ-targeting siRNA attenuated the induction of SIRT1 expression by formononetin, whereas transfection of control siRNA did not (Fig. 2G). In line with these findings, GSK0660, a specific inhibitor of PPARδ, significantly attenuated the formononetin-induced increase in SIRT1 promoter activity (Fig. 2H). These results suggest that formononetin upregulates SIRT1 expression via PPARδ at the transcriptional level.

SIRT1 is essential for inhibition of LPS-induced HMGB1 release by formononetin

To investigate the direct effect of SIRT1 on LPS-induced HMGB1 release, we examined the levels of SIRT1 protein and released HMGB1 in RAW264.7 cells exposed to LPS in the presence or absence of formononetin. A high level of HMGB1 was released upon LPS treatment, whereas this was reduced in the presence of formononetin. On the other hand, the level of SIRT1 protein was significantly suppressed in LPS-treated RAW264.7 cells. However, this LPS-mediated repression of SIRT1 was recovered in the presence of formononetin, indicating that SIRT1 is critical for modulation of LPS-induced HMGB1 release by formononetin (Fig. 3A).

To further clarify the functional significance of formononetin-mediated upregulation of SIRT1 in RAW264.7 cells, we manipulated the expression and activity of SIRT1 using siRNA or chemicals. The levels of SIRT1 protein were diminished in cells transfected with SIRT1 siRNA, however control siRNAs had no effect on the levels of either protein (Fig. 3B). Transfection of SIRT1-targeting siRNA significantly attenuated the inhibitory effect of formononetin on LPS-induced HMGB1 release (Fig. 3C). Consistently, inhibition of SIRT1 activity by sirtinol also prevented inhibition of HMGB1 release by formononetin (Fig. 3D). By contrast, activation of SIRT1 by resveratrol inhibited LPS-induced HMGB1 release. Furthermore, resveratrol treatment potentiated the inhibitory effects of formononetin, suggesting that SIRT1 plays a role in the suppression of HMGB1 release by formononetin (Fig. 3E). These results indicate that formononetin inhibits LPS-induced HMGB1 release by regulating SIRT1 expression.

SIRT1-mediated deacetylation of HMGB1 underlies the inhibition of its release by formononetin

Inflammatory signal-mediated acetylation of HMGB1 is critical for its release into the extracellular compartment and acetylated HMGB1 is a substrate of SIRT1 (Bonaldi et al., 2003; Hwang et al., 2014; Rickenbacher et al., 2014); therefore, we evaluated whether formononetin affects LPS-induced acetylation of HMGB1. When RAW264.7 cells were stimulated with LPS for 6 h, the level of acetylated HMGB1 in an immunoprecipitate obtained using an anti-HMGB1 antibody was significantly enhanced. However, formononetin reduced this increase in acetylated HMGB1 in a concentration-dependent manner, indicating that formononetin is involved in the deacetylation of HMGB1 primed by LPS (Fig. 4A).

To evaluate whether this inhibition of LPS-induced HMGB1 acetylation by formononetin correlates with the level of SIRT1 in RAW264.7 cells, we knocked down SIRT1. Transfection of SIRT1-targeting siRNA significantly prevented the decrease in acetylated HMGB1 by formononetin in LPS-exposed RAW264.7 cells (Fig. 4B). Transfection of PPARδ-targeting siRNA elicited the same effect (Fig. 4C). These results indicate that formononetin reduces HMGB1 acetylation via PPARδ-mediated upregulation of SIRT1, thereby inhibiting the release of HMGB1 into the extracellular milieu.