Main

It was recently demonstrated that Sirt5, one of the four sirtuins with weak deacetylase activity, preferentially hydrolyses succinyl and malonyl lysine4,5. The discovery of the novel Sirt5 activity suggested that other sirtuins with weak deacetylase activity may use alternative substrates as well. We therefore set out to investigate whether SIRT6 has any new enzymatic activity. SIRT6 has been reported to be important for DNA repair, transcriptional regulation of genes important for metabolism and immune responses, and for lifespan6,7,8,9,10. In most cases, the biological functions have been linked to the sequence-specific deacetylase activity of SIRT6 on histone H3K9 and K56, but not other peptide sequences9,11,12. The deacetylase activity of SIRT6 on H3K9 and K56 can indeed be detected in our assay. However, the catalytic efficiency is low (Table 1), indicating that the deacetylase activity may only account for a part of its biological function.

Table 1 Catalytic efficiencies of SIRT6 on different acyl peptides
Full size table

To discover possible novel activity of SIRT6, we synthesized H3K9 peptides with different acyl groups (acetyl, malonyl, succinyl, butyryl, myristoyl and palmitoyl) and assayed these peptides with recombinant SIRT6. SIRT6 can hydrolyse long-chain fatty acyl groups efficiently (Fig. 1a). The kcat/Km for demyristoylation (1,400 s−1 M−1) is approximately 300-fold better than that for deacetylation (4.8 s−1 M−1). The increased catalytic efficiency comes mainly from the decrease in Km. For deacetylation, the Km is 810 μM, but for demyristoylation, the Km is 3.4 μM. The activity of SIRT6 (a class IV sirtuin) is similar to that of PfSir2A5, which was classified as a Class III sirtuin13 but lacks the conserved Arg and Tyr residues that recognize negatively charged succinyl or malonyl groups4. This emphasizes that it is important to examine the activity of each sirtuin experimentally as bioinformatics predictions may not be sufficient.

Figure 1: SIRT6 preferentially hydrolyses long-chain fatty acyl lysine in vitro.
figure 1

a, High-performance liquid chromatography traces showing SIRT6-catalysed hydrolysis of different acyl peptides based on the H3K9 sequence. b, H2BK12 myristoyl peptide can be hydrolysed by SIRT6, whereas the corresponding acetyl peptide cannot. Reactions were carried out with 50 μM peptide, 1 μM SIRT6, 20 mM Tris pH 8.0, 0.5 mM NAD and 1 mM dithiothreitol (DTT) at 37 °C for 30 min.

PowerPoint slide

Full size image

The conclusion that SIRT6 is better at hydrolysing long-chain fatty acyl groups was also supported by its ability to hydrolyse myristoyl group from other peptide sequences. The deacetylase activity of SIRT6 was reported to be sequence-specific. Only H3K9 and H3K56 acetyl peptides, but not other peptides, were reported to be deacetylated by SIRT69,11,12. To test whether the more efficient demyristoylation activity allows SIRT6 to use more peptide substrates, we synthesized acetyl and myristoyl peptides based on the H2B K12 sequence. Consistent with previous reports9,11,12, the hydrolysis of the acetyl peptide by SIRT6 was undetectable. In contrast, the hydrolysis of the corresponding myristoyl peptide could be readily detected (Fig. 1b). These results confirm the preference of SIRT6 for long-chain fatty acyl groups.

To determine the structural basis for enhanced SIRT6 activity with fatty acyl peptides, we obtained a crystal structure of SIRT6 in complex with a H3K9 myristoyl peptide and ADP-ribose (ADPR) at 2.2 Å resolution (PDB 3ZG6, Fig. 2a, Supplementary Fig. 1). The overall structure is similar to the published SIRT6 structure14 (PDB 3K35) (Supplementary Fig. 1). Residues 2–10 and 166–174 of SIRT6 are visible in the H3K9 myristoyl bound structure, whereas the corresponding regions are missing in the published SIRT6 structure without peptide bound14 (Supplementary Fig. 2). The way that H3K9 myristoyl and ADPR bind to SIRT6 is similar to that seen in other ternary complex structures of other sirtuins, such as Sir2Tm in complex with the p53 acetyl peptide and NAD (Supplementary Fig. 3)15. The peptide interacts with SIRT6 through many hydrogen-bonding interactions. Similar to what was observed with other sirtuins, including Sirt5, most of the hydrogen bonds come from main chain C = O and N–H of the myristoyl peptide, with the only side chain hydrogen-bonding coming from Trp 11 (Supplementary Fig. 4). Therefore, the selectivity for peptide sequences is not high, which is consistent with our enzymology data. The myristoyl group is located in a hydrophobic pocket (Fig. 2b) formed by hydrophobic residues from several flexible loops, including Ala 11, Pro 60, Phe 62, Trp 69, Pro 78, Phe 80, Phe 84, Val 113, Leu 130, Leu 184 and Ile 217. The substrate-binding sites in other available human sirtuin structures do not possess such a big hydrophobic pocket (Supplementary Fig. 5). The structural data are thus consistent with the biochemical data and provide a reasonable explanation for the preference of SIRT6 for long-chain fatty acyl groups.

Figure 2: Structure basis for SIRT6 activity with long-chain fatty acyl groups.
figure 2

a, Overall structure of SIRT6 with myristoyl H3K9 (Myr-H3K9, green) peptide and ADP-ribose (ADPR, yellow) bound. b, Hydrophobic residues in SIRT6 that accommodate the myristoyl (Myr) group.

PowerPoint slide

Full size image

The next important question was whether this activity is physiologically relevant. There are a few proteins known to be modified by fatty acyl groups on lysine, although the function of the modification is unclear16,17. One of the fatty acylated proteins is TNF-α16, a type II membrane protein with a single transmembrane domain linking the amino-terminal intracellular domain and the carboxy-terminal extracellular domain18. When cleaved by proteases, the extracellular domain is released. The released TNF-α can then bind its receptors and induce various signalling pathways18. It was reported that SIRT6 can regulate the synthesis of TNF-α, but the mechanism was unclear19,20. We propose that SIRT6 may regulate TNF-α secretion via defatty-acylation.

To test this, we measured the fatty acylation level on TNF-α in SIRT6 wild-type (WT) and knockout (KO) mouse embryonic fibroblast (MEF) cells. Flag-tagged TNF-α was transfected into the cells. The cells were cultured in the presence of an alkyne-tagged fatty acid analogue, Alk14, which can covalently label fatty acylated proteins21,22. TNF-α was immunoprecipitated and conjugated to rhodamine-azide (Rh-N3) using click chemistry (Fig. 3a). A protein will be fluorescently labelled if it is fatty acylated by Alk14. Because Alk14 can also label cysteine residues22,23, we treated TNF-α with hydroxylamine to remove cysteine fatty acylation (Fig. 3a). As shown in Fig. 3b, TNF-α from SIRT6 KO MEF cells had significantly increased Alk14 labelling compared to TNF-α from SIRT6 WT MEF cells, indicating that SIRT6 regulates the fatty acylation level on TNF-α. When human SIRT6 WT was overexpressed in SIRT6 KO MEF cells, TNF-α had lower fatty acylation level than TNF-α from cells without overexpression of human SIRT6, whereas overexpression of human SIRT6(H133Y) catalytic mutant did not have much effect on TNF-α fatty acylation (Fig. 3c), indicating that enzymatic activity of SIRT6 is required for controlling TNF-α fatty acylation.

Figure 3: SIRT6 regulates TNF-α fatty acylation and secretion.
figure 3

a, Method of using Alk14 to detect TNF-α fatty acylation. b, SIRT6 controls TNF-α fatty acylation on K19 and K20. KR, TNF-α(K19R, K20R) mutant. c, H133 of SIRT6 is required for TNF-α defatty-acylation. KI, knock-in. d, SIRT6 defatty-acylates TNF-α in vitro. e, SIRT6 regulates secretion of TNF-α in MEF cells. n = 6; ***P < 0.001. f–i, SIRT6 regulates fatty acylation level and secretion of endogenous TNF-α in THP-1 cells (f and g; n = 3; **P < 0.01) and bone-marrow-derived macrophages (h and i; n = 5; *P < 0.02). Secretion data were expressed as mean ± s.d. IP, immunoprecipitation; shRNA, short hairpin RNA; WB, western blot.

PowerPoint slide

Full size image

To further confirm that fatty acylation occurs on lysine residues of TNF-α, we mutated K19 and K20, the reported sites of myristoylation on TNF-α16, to Arg (labelled as ‘KR’ mutant in Fig. 3b). The labelling intensities of the mutant from SIRT6 KO MEF cells dropped to background level (Fig. 3b), confirming that K19 and K20 are the major fatty acylation sites. Furthermore, TNF-α isolated from SIRT6 KO MEF cells can be defatty-acylated in vitro in an NAD-dependent manner (Fig. 3d). Synthetic TNF-α K19 and K20 myristoyl peptides can be efficiently hydrolysed by SIRT6 (Table 1). To rule out that TNF-α could also be regulated by lysine acetylation, we examined the acetylation level using a pan-specific acetyl lysine antibody. No acetylation was detected on K19 and K20 of TNF-α (Supplementary Fig. 6). Thus, SIRT6 regulates the fatty acylation level, but not acetylation level on K19 and K20 of TNF-α.

We next investigated the function of TNF-α fatty acylation on K19 and K20. It was reported that the amount of TNF-α detected in the media of SIRT6 KO cells is less than that in the media of SIRT6 WT cells20. This was attributed to the regulation of TNF-α synthesis by SIRT620. Given that SIRT6 regulates lysine fatty acylation, we propose that SIRT6 may regulate the secretion of TNF-α. To test this, Flag-tagged TNF-α was transiently transfected into SIRT6 WT and KO MEF cells. The amount of secreted TNF-α in the medium and the amount of TNF-α in the cells were measured by ELISA (Supplementary Fig. 7). The percentage of secreted TNF-α was then calculated. TNF-α secretion efficiency was lower in SIRT6 KO MEF cells than in SIRT6 WT MEF cells (Fig. 3E). The secretion efficiency is due to lysine fatty acylation because the secretion efficiency of TNF-α KR mutant was not affected by SIRT6 knockout. These data indicate that lysine fatty acylation regulates TNF-α secretion and SIRT6 promotes TNF-α secretion by defatty-acylation.

We further investigated whether endogenous TNF-α are also regulated by SIRT6. For this purpose, we used both human THP-1 cells and bone-marrow-derived mouse macrophages. Two SIRT6 knockdown (KD) THP-1 cell lines and one control KD THP-1 cell line were generated (Fig. 3f). TNF-α from SIRT6 KD cells contained more fatty acylation than TNF-α from the control KD cells (Fig. 3f). The amount of secreted TNF-α in the medium and the amount of TNF-α in the cells were measured (Supplementary Fig. 8) and the percentage of secreted TNF-α was calculated (Fig. 3g). The secretion efficiency of TNF-α was lower in SIRT6 KD THP-1 cells than in control KD cells, especially when the cells were supplemented with palmitic acid (Fig. 3g). Similar results were obtained for TNF-α in mouse macrophages. TNF-α in SIRT6 WT macrophages had lower lysine fatty acylation than in SIRT6 KO macrophages (Fig. 3h), whereas TNF-α secretion efficiency in SIRT6 WT macrophages was higher than in SIRT6 KO macrophages (Fig. 3i).

SIRT6 was reported to be mainly localized in the nucleus9. The regulation of TNF-α secretion by SIRT6 indicated that SIRT6 might be present in secretory organelles, such as the endoplasmic reticulum. We indeed detected SIRT6 in the endoplasmic reticulum fraction of MEF and THP-1 cell lysates (Supplementary Fig. 9). In SIRT6 KD THP-1 cells or SIRT6 KO MEF cells, less or no SIRT6 was detected in the endoplasmic reticulum fraction compared with SIRT6 WT cells (Supplementary Fig. 9). The presence of SIRT6 in the endoplasmic reticulum provided further support for the conclusion that SIRT6 regulates TNF-α secretion through defatty-acylation of TNF-α.

In summary, our enzymological and structural studies show that human SIRT6, which has weak deacetylase activities in vitro, catalyses the hydrolysis of fatty acyl lysine modifications more efficiently than deacetylation. SIRT6 regulates the fatty acylation level on K19 and K20 of TNF-α and this modulates the secretion of TNF-α. It has been reported that SIRT6 regulates the acetylation level on histone H3K9 and K569,11,12. Our results are not in conflict with these reports because it is possible that the association with chromatin can regulate the activity of SIRT6. Alternatively, fatty acylation of lysine residues may occur to histones22,23, and SIRT6 can regulate both acetylation and fatty acylation of histones. The significance of our finding is several fold. First, it reveals a novel physiological activity for SIRT6. Second, it demonstrates, as far as we know for the first time, that lysine fatty acylation, a protein post-translational modification that has been little studied until now, has an important role in regulating protein secretion. Other proteins (Supplementary Fig. 10), such as insulin-like growth factor 1 (IGF1) may be regulated by similar mechanisms6,24. Third, the discovery of SIRT6 activity on protein lysine fatty acylation provides an avenue to further investigate this under-recognized protein modification16,17 and may reveal interesting connections to the well-known cysteine palmitoylation and N-terminal glycine myristoylation25.

Methods Summary

Expression, purification and crystallization

Human SIRT6 was overexpressed in Escherichia coli and purified using Ni-NTA affinity chromatography, ion exchange and gel filtration. Crystals were grown from 12% PEG6K, 0.1 M MES (2-(N-morpholino)ethanesulphonic acid), pH 6.5.

Data collection and structure determination

Data were collected at the Shanghai Synchrotron Radiation Facility (SSRF). The structure was determined by molecular replacement.

Activity assay of SIRT6 on different acyl peptides

Acyl peptides were synthesized as previously described5. The hydrolysis of different acyl peptides catalysed by SIRT6 was monitored by high-performance liquid chromatography. Kinetic parameters were determined by varying the concentrations of the acyl peptides.

Detection of long-chain fatty acylation on TNF-α

Flag-tagged TNF-α was transiently transfected into either SIRT6 WT or KO MEF cells7. Endogenous TNF-α in THP-1 cells and SIRT6 WT or KO mouse macrophages was stimulated by LPS. The fatty acyl lysine modification was detected using alkyne-tagged fatty acid analogue Alk14 using a procedure modified from a previously published method21.

Online Methods

Reagents

Mouse monoclonal anti-Flag M2 antibody conjugated with horseradish peroxidase, anti-Flag M2 affinity gel, and human/mouse SIRT6 antibody (S4322) were from Sigma. Human TNF-α antibody (D5G9) and mouse TNF-α antibody (D2D4) were from Cell Signaling Technology. TNF-α Affibody immobilized on agarose (ab31909) and human SIRT6 antibody (ab88494) were from Abcam. Lamin A/C antibody (636), β-actin antibody (C4) and GRP 78 antibody (H-129) were from Santa Cruz Biotechnology. The rabbit pan-specific anti-acetyl lysine antibody was from ImmuneChem Pharmaceuticals. The goat anti-rabbit and mouse IgG conjugated with horseradish peroxidase and protein A/G plus-agarose were from Santa Cruz Biotechnology. Human and mouse TNF-α ELISA kits were from eBioscience. Brefeldin A (BFA), phorbol 12-myristate 13-acetate (PMA), lipopolysaccharides from Escherichia coli 0111:B4 (LPS) and palmitic acid were purchased from Sigma. Alk14 was synthesized according to reported procedures21. SIRT6 WT or KO MEF cells were generated as previously reported7. THP-1 cells were purchased from ATCC.

Cloning, expression and purification of full-length SIRT6 for activity assay

The open reading frame of full-length human SIRT6 (1–355) was inserted into a pET28a vector between the BamHI and NotI sites. This plasmid was transformed into E. coli Arctic Express (DE3) cells. The cells were cultured at 37 °C in 2 × YT culture medium (5 g of NaCl, 16 g of bactotrypton and 10 g of yeast extract per litre). Isopropyl-β-d-1-thiogalactopyranoside (IPTG, 0.2 mM) was used to induce expression when attenuance D600 nm was 0.6, and the culture was grown for 20 h at 289 K. Cells were collected by centrifugation at 7,330g for 10 min and then re-suspended in lysis buffer (20 mM Tris-HCl, pH 7.2, 500 mM NaCl and 2% glycerol). Cells were lysed using a cell disrupter. After centrifugation at 29,300g for 25 min at 277 K, the supernatant was loaded onto a nickel column (QFF-Sepharose, Amersham Biosciences) pre-equilibrated with 20 mM Tris-HCl pH 7.2 with 500 mM NaCl. The protein was eluted with a linear gradient of imidazole (0–500 mM). The desired fractions were pooled, concentrated and buffer exchanged to cation exchange buffer (20 mM Tris pH 7.2, 80 mM NaCl, 5% glycerol). The protein was loaded onto a cation exchange column (Amersham Biosciences) and eluted with 1 M NaCl, 20 mM Tris-HCl, pH 7.2, 2% glycerol. The purified protein was stored at −80 °C.

Cloning, expression and purification of truncated SIRT6 for crystallization

Human SIRT6 (1–294) was inserted into a pET28a vector between the NdeI and NotI sites. The protein was overproduced at 37 °C in E. coli Rosetta (DE3) strain using 2 × YT medium. The expression and purification method was the same as that used for the full-length SIRT6 except that after Ni column purification the protein was further purified by gel filtration using a Superdex75 column (Amersham Biosciences). The protein was eluted with 20 mM Tris-HCl, pH 7.0, 100 mM NaCl. After concentration to 8 mg ml−1, the target protein was frozen at −80 °C.

Synthesis of acyl peptides

The peptides were synthesized and purified as described previously5 The identity of the peptides was confirmed using LCQ Fleet ThermoFisher Mass Spectrometer. The acetyl, butyryl peptides were dissolved in 25% (v/v) DMSO in water. The longer chain fatty acyl peptides were dissolved in DMSO. The concentrations of peptides were determined at 280 nm using extinctions coefficient of the two tryptophans attached at the C termini of the peptides.

Deacylation activity assay

The activity of SIRT6 was analysed using reverse phase high-performance liquid chromatography on a Kinetex XB-C18 column (100A, 75 mm × 4.60 mm, 2.6 μm, Phenomenex). SIRT6 full length (1 μM) was incubated in a reaction mixture (60 μl) containing 20 mM Tris pH 8.0, 1 mM DTT, 0.5 mM NAD and 50 μM H3K9 acyl peptides at 37 °C for 30 min. Total DMSO content in the reaction was <2.5% unless mentioned otherwise. The reactions were quenched with 60 μl of 0.5 N HCl in methanol. The reactions were then monitored by high-performance liquid chromatography as described later in the kinetics assay.

Kinetics assay for acetyl and butyryl peptides

Peptide concentrations were varied from 0 to 250 μM for H3K9 butyryl peptide and from 0 to 600 μM for the acetyl peptide. The reaction mixtures (60 μl with 2 mM NAD, 1 mM DTT, 20 mM Tris, pH 8.0, 4 μM SIRT6 full length, and acyl peptides at various concentrations) were incubated for 30 min at 37 °C.The reaction was stopped using 60 μl of 0.5 N HCl in methanol. The reaction mixtures were spun at 18,000g for 10 min and were analysed on a Kinetex XB-C18 column (100A, 100 mm × 4.60 mm, 2.6 μm, Phenomenex). The gradient of 20–40% B (acetonitrile with 0.1% TFA) in 17 min at 0.5 ml min−1 was used.

Kinetics assay for long-chain fatty acyl peptides

The peptide concentrations were varied from 1 to 20 μM. The reactions (60 μl with 2 mM NAD, 1 mM DTT, 20 mM Tris pH 8.0, 0.2 μM recombinant SIRT6 full length, and acyl peptides at different concentrations) was incubated for 15 min at 37 °C. The reaction was stopped using 60 μl of 0.5N HCl in methanol. The reaction mixtures were spun at 18,000g for 10 min and were analysed on a Kinetex XB-C18 column (100A, 75 mm × 4.60 mm, 2.6 μm, Phenomenex). The gradient of 0–55% B in 10 min at 0.5 ml min−1 was used. The product and the substrate peaks were quantified using absorbance at 280 nm and converted to initial rates, which were then plotted against the acyl peptide concentrations and fitted using Kaleidagraph.

Crystallization, X-ray data collection and structure determination

Crystals of complex SIRT6 with H3K9-Myr and ADPR were obtained by hanging drop vapour-diffusion method at 291 K using commercial screens from Hampton Research. Each drop consisting of 1 μl of 10 mg ml−1 protein complex solution (20 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM DTT) and 1 μl reservoir solution was equilibrated against 400 μl reservoir solution. The qualified crystals of SIRT6 grew with a cube profile within 1 week with a reservoir containing 12% PEG6K, 0.1 M MES, pH 6.5. The mixture of 30% glycerol with reservoir solution above was used as cryogenic liquor. The X-ray diffraction data were collected at 100 K in a liquid nitrogen gas stream at the Shanghai Synchrotron Radiation Facility BL17U. 180 frames were collected with a 1° oscillation and the data were indexed and integrated using the program HKL200026. The complex structure of SIRT6 with H3K9 myristoyl peptide and ADPR was solved by molecular replacement using the program Molrep from the CCP4 Suite27 with the published SIRT6 structure (PDB: 3K35)29 as the search model. Refinement and model building were performed with REFMAC5 and COOT from CCP4. The X-ray diffraction data collection and structure refinement statistics are shown in Supplementary Table 1.

Generation of SIRT6 KO MEF cells with human SIRT6 WT and H133Y mutant knock-in

Human SIRT6 WT or H133Y complementary DNA was inserted to lentiviral vector (engineered from pSIN-EF2-OCT4 vector with OCT4 deletion to convert into gateway destination vector, and provided by C. Zhang) using Gateway Cloning. After co-transfection of SIRT6 WT or H133Y lentiviral plasmid, pCMV-dR8.2, and pMD2.G into 293T cells, the medium was collected to infect SIRT6 KO MEF cells. The SIRT6 WT or H133Y knock-in cells were selected using 1.5 µg ml−1 puromycin in complete cell culture medium. Overexpressed SIRT6 WT and H133Y mutant have a C-terminal V5 tag.

Cloning of TNF-α

For construction of human TNF-α expression vector to express wild-type TNF-α protein (TNF-α WT) with N-terminal Flag tag and C-terminal haemagglutinin tag, human full-length TNF-α cDNA were generated by PCR and inserted into pCMV4A vector at EcoRV and XhoI sites. The plasmid of TNF-α double lysine mutant (K19R, K20R), TNF-α KR, was made by overlap extension PCR.

Transfection of TNF-α into MEF cells

SIRT6 WT and KO MEF cells were maintained in DMEM medium containing glucose and l-glutamine (Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (Invitrogen). The pCMV vectors containing target genes were transfected into cells using FuGene 6 (Promega) according to the manufacturer’s protocol. Empty pCMV vector was transfected into cells as negative control. This transfection method was also applied to SIRT6 KO MEF cells with human SIRT6 WT or H133Y mutant knock-in for TNF-α overexpression.

Labelling of TNF-α in MEF cells with Alk14

SIRT6 WT or KO MEF cells were cultured with fresh medium containing 20 μM Alk14 and 5 µg ml−1 brefeldin A for 12 h after transient transfection of TNF-α. Cells without Alk14 or without overexpression of TNF-α were used as negative controls. Cells were collected by cell scraper and resuspended in 1 × PBS buffer. After centrifugation at 500g for 5 min at 4 °C, the cell pellet was dissolved in lysis buffer (25 mM Tris, pH 7.4, 150 mM NaCl, 10% glycerol, 1% Nonidet P-40) with protease inhibitor cocktail (Sigma). The supernatant was collected after centrifugation at 16,000g for 10 min at 4 °C, and used for western blotting, immunoprecipitation or TNF-α ELISA experiment.

Generation of stable SIRT6 KD THP-1 cells

SIRT6 shRNA lentiviral plasmids in pLKO.1-puro vector were purchased from Sigma. SIRT6 shRNA 1 (TRCN0000378253) ccggcagtacgtccgagacacagtcctcgaggactgtgctcggacgtactgtttttg and shRNA 2 (TRCN0000232528) ccgggaagaatgtgccaagtgtaagctcgagcttacacttggcacattcttctttttg were used. After co-transfection of SIRT6 shRNA plasmid, pCMV-dR8.2 and pMD2.G into 293T cells, the medium was collected to infect THP-1 cells. The SIRT6 KD cells were selected using 1.5 µg ml−1 puromycin in complete cell culture medium. THP-1 cells infected with lentivirus containing control shRNA plasmid were carried out similarly.

Labelling of TNF-α in THP-1 cells with Alk14

THP-1 cells (with control shRNA, SIRT6 shRNA 1 orSIRT6 shRNA 2, at 0.6 × 106 cells ml−1) were treated with PMA (200 ng ml−1) for 24 h. Then cells were cultured with fresh medium containing Alk14 (50 μM) and LPS (1 µg ml−1). After 1 h, brefeldin A was added to culture medium (5 µg ml−1). Cells were collected after 10 h, and dissolved in lysis buffer (25 mM Tris, pH 7.4, 150 mM NaCl, 10% glycerol, 1% Nonidet P-40) with protease inhibitor cocktail (Sigma). The supernatant was collected after centrifugation at 16,000g for 10 min at 4 °C and used for western blotting or immunoprecipitation.

Isolation of SIRT6 WT and KO macrophages and TNF-α secretion

Bone-marrow-derived macrophages were isolated from SIRT6 WT and KO mice as described14 with some modifications. Briefly, femurs and tibiae were dissected and marrow tissue eluted by irrigation with DMEM. Cells were cultured in Petri dishes containing complete media (DMEM + 20% FCS + 10ng ml−1 recombinant murine M-CSF + penicillin/streptomycin) for 7–10 days at 37 °C in a humidified 5% CO2 atmosphere. For TNF-α mRNA expression analysis, 1.5 × 105 macrophages were plated in six-well plates containing complete media and incubated overnight. Next day, LPS (100 ng ml−1) was added and the cells were cultured for 12 h. For TNF-α protein expression and secretion, cells were plated as above and, after 12 h of LPS stimulation, the supernatants were collected and used for ELISA analysis. To measure TNF-α fatty acylation levels, 1.5 × 106 macrophages were plated in 6-cm plates in complete media and incubated overnight. Next day, cells were treated using the same procedure as described above for THP-1 cells.

Western blot for TNF-α

Protein samples were separated by 18% SDS–PAGE and transferred to PVDF membrane. The membrane was blocked with 5% BSA in TBST (25 mM Tris, pH 7.4, 150 mM NaCl, 0.1% Tween-20), incubated with antibodies in TBST, and developed in ECL Plus western blotting detection reagents (GE Healthcare). The chemiluminescence signal was recorded by Storm 860 Imager (Amersham Biosciences) and analysed with ImageQuant TL v2005.

Immunoprecipitation of Flag-TNF-α using anti-Flag affinity gel

Cell lysate (250–400 µg from SIRT6 WT or KO MEF cells) was incubated with 20 μl suspension of anti-Flag M2 affinity gel for 2 h at 4 °C. After centrifugation at 500g for 2 min at 4 °C, the affinity gel was washed three times with 500 μl washing buffer (25 mM Tris, pH 7.4, 150 mM NaCl, 0.2% Nonidet P-40) and used for later experiments.

Immunoprecipitation of endogenous TNF-α from THP-1 cells

Cell lysate (∼500 µg) was incubated with 20 μl suspension of TNF-α Affibody immobilized agarose for 4 h at 4 °C. After centrifugation at 500g for 2 min at 4 °C, the agarose was washed three times with 500 μl washing buffer (25 mM Tris, pH 7.4, 150 mM NaCl, 0.2% Nonidet P-40) and used for later experiments.

Immunoprecipitation of endogenous TNF-α from bone-marrow-derived mouse macrophage

Cell lysate (∼100 µg from SIRT6 WT or KO macrophages) was incubated with 2 µg mouse TNF-α antibody for 2 h at 4 °C. Then the mixture was incubated with 40 μl suspension of protein A/G plus-agarose for 8 h at 4 °C. After centrifugation at 500g for 2 min at 4 °C, the agarose was washed three times with 500 μl washing buffer (25 mM Tris, pH 7.4, 150 mM NaCl, 0.2% Nonidet P-40) and used for later experiments.

Detection of fatty acylation on TNF-α by fluorescence

After immunoprecipitation, the gel/agarose was re-suspended in 10 μl buffer (25 mM Tris, pH 7.4, 50 mM NaCl, 1% Nonidet P-40) for click chemistry. Rh-N3 (in DMF) was added to the above suspension to a final concentration of 200 μM, followed by the addition of Tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (in DMF, final concentration 600 μM), CuSO4 (in water, final concentration 2 mM), and TCEP (in water, final concentration 2 mM). The click chemistry was allowed to proceed at room temperature for 60 min. The reaction mixture was mixed with protein loading buffer (final 2 ×) and heated at 95 °C for 10 min. After centrifugation at 16,000g for 2 min at room temperature, the supernatant was collected and heated with hydroxylamine (pH 7.2, 60 mM) at 95 °C for 7 min. The samples were resolved by SDS–PAGE using 18% acrylamide gel. Rhodamine fluorescence signal was recorded by Typhoon 9400 Variable Mode Imager (GE Healthcare Life Sciences) with setting of Green (532 nm)/580BP30 PMT 500 V (normal sensitivity), and analysed by ImageQuant TL v2005.

Detection of acetyl-lysine on TNF-α by western blot using anti-acetyl lysine antibody

SIRT6 KO MEF cell lysate (∼30 mg), with/without overexpression of TNF-α WT or TNF-α KR, was incubated with 40 μl suspension of anti-Flag M2 affinity gel for 2 h at 4 °C. After centrifugation at 500g for 2 min at 4 °C, the affinity gel was washed three times with 500 μl washing buffer (25 mM Tris, pH 7.4, 150 mM NaCl, 0.2% Nonidet P-40) and then heated with 2 × protein loading buffer at 100 °C for 10 min. The samples were resolved by SDS–PAGE using 18% acrylamide gel and examined by western-blotting using anti-acetyl-lysine antibody. Acetylated BSA was used for positive control to demonstrate acetyl-lysine signal by western-blotting. After recording acetyl-lysine signal, PVDF membrane was washed with water and stained with Coomassie blue to detect TNF-α protein. Another western-blotting using anti-Flag antibody was also carried out to demonstrate the equal loading of TNF-α WT and TNF-α KR.

Defatty-acylation of TNF-α by SIRT6 in vitro

TNF-α WT was immunoprecipitated from lysate of SIRT6 KO MEF cells (∼1 mg) overexpressing Flag–TNF-α and cultured with Alk14 using 40 μl suspension of anti-Flag M2 affinity gel following procedure describe above. After washing, the gel was divided into four equal aliquots. Each aliquot was re-suspended in 10 μl assay buffer (50 mM Tris, pH 7.4, 100 mM NaCl, 2 mM MgCl2, 1% Nonidet P-40) with 20 μM of SIRT6 or BSA and with or without 2 mM NAD and incubated at 37 °C for 1 h. Then detection of fatty acylation by fluorescence after hydroxylamine treatment was carried out with each aliquot as described earlier.

Secretion of TNF-α and TNF-α KR from SIRT6 WT and KO MEF Cells

After transient transfection of TNF-α WT or TNF-α KR, SIRT6 WT and KO MEF cells were further incubated with fresh medium for 12 h. Then culture medium and cells were collected separately for detection of TNF-α by using human TNF-α ELISA kit. Secretion efficiency was calculated by TNF-α amount in culture medium versus total amount of TNF-α in culture medium and cells. Six independent experiments were carried out.

TNF-α secretion from THP-1 cells

Stable shRNA-infected THP-1 cells (at 0.6 × 106 cells ml−1) were treated with PMA (200 ng ml−1) for 20 h. Treatment without palmitic acid: cells were incubated with fresh medium containing LPS (100 ng ml−1). After 2 h, culture medium and cells were collected for human TNF-α ELISA assay. Treatment with palmitic acid: cells were incubated with palmitic acid (final 50 μM) in culture medium for 2 h, and then with fresh medium containing LPS (100 ng ml−1) and palmitic acid (50 μM). After 2 h, culture medium and cells were collected for human TNF-α ELISA assay. Three independent experiments were carried out.

Endoplasmic reticulum fractions from THP-1 cells and SIRT6 WT/KO MEF cells

Endoplasmic reticulum fraction was obtained from THP-1 cells, SIRT6 WT/KO MEF cells, or SIRT6 KO MEF cells with human SIRT6 WT or H133Y knock-in, using an isopycnic flotation method29. Endoplasmic reticulum fraction was solubilized in lysis buffer (25 mM Tris, pH 7.4, 150 mM NaCl, 10% glycerol, 1% Nonidet P-40) with protease inhibitor cocktail (Sigma). The supernatant was collected after centrifugation at 16,000g for 10 min at 4 °C, and used for western blotting.

Statistical analysis

Data were expressed as mean ± s.d. (standard deviation, shown as error bars). Differences were examined by two-tailed Student’s t-test between two groups; and *P < 0.02, **P < 0.01, ***P < 0.001.