Integrative chemical biology approaches to examine ‘erasers’ for protein lysine fatty-acylation

Abstract

Protein fatty-acylation is important for the regulation of membrane association, trafficking, subcellular localization and activity of many cellular proteins. While significant progress has been made on our understanding of the two major forms of protein fatty-acylation, N-myristoylation and S-palmitoylation, studies of fatty-acylation of protein lysine residues have lagged behind. This is in large part due to a lack of knowledge about the enzymes that regulate the addition and removal of lysine fatty-acylation. To fill this knowledge gap, here we demonstrate the use of integrative chemical biology approaches to examine human sirtuins as a de-fatty-acylase in vitro and in cells. We used a photo-cross-linking-based chemical approach to investigate enzymes that recognize lysine fatty-acylation. This approach enables the identification of human Sirt2 as a robust lysine de-fatty-acylase in vitro. Using a bioorthogonal chemical reporter for protein fatty-acylation, we show that Sirt2 can regulate protein lysine fatty-acylation in cells. This finding not only opens a new avenue to investigate the biological significances of protein lysine fatty-acylation, but also may help to unravel unrecognized cellular functions of Sirt2 that has been considered solely as a deacetylase until now. Furthermore, the photo-cross-linking-based chemical approach used in this study should also be useful to examine the interactions between other protein posttranslational modifications and their ‘erasers’.

Fatty-acylation of proteins regulates many fundamental cellular processes in eukaryotes such as signal transduction and membrane trafficking.^[1] While diverse fatty acyl modifications can occur on different residues of proteins, our current understanding of this type of posttranslational modification (PTM) is mainly based on the studies of two well-known fatty-acylation, myristoylation of N-terminal glycine residues (N-myristoylation)^[2] and palmitoylation of cysteine residues (S-palmitoylation)^[3]. In contrast, fatty-acylation of lysine residues, although identified more than two decades ago,^[4] has been little studied. The regulatory mechanisms and cellular functions of lysine fatty-acylation remains poorly understood, in large part due to a lack of knowledge of cellular enzymes that catalyze the addition (i.e., ‘writers’) or removal (i.e., ‘erasers’) of this PTM.

Recently, Lin and coworkers found that sirtuin 6 (Sirt6), one of seven members of mammalian nicotinamide adenine dinucleotide (NAD)-dependent lysine deacetylases^[5], can catalyze the removal of fatty acyl modifications on the lysine residues of tumor necrosis factor-α (TNF-α), through which the secretion of TNF-α can be regulated.^[6] A later study by Denu and coworkers revealed that several other mammalian sirtuins also have ability to catalyze the hydrolysis of lysine fatty-acylation on a model peptide substrate.^[7] However, due to lack of further characterizations, our understanding of their de-fatty-acylase activity and substrate selectivity toward acetylated vs. fatty-acylated lysine is still incomplete. More importantly, since the de-fatty-acylase activities of these sirtuins were only tested using a peptide-based in vitro assay, it is essential to determine whether they can regulate lysine fatty-acylation in vivo, and if so, what their endogenous fatty-acylated protein substrates are.

To address these questions, we first sought to examine the ability of human sirtuins to recognize lysine fatty-acylation. We have previously developed a photo-cross-linking-based chemical approach to study protein-protein interactions mediated by PTMs, such as lysine methylation and threonine phosphorylation.^[8] We reasoned that this method would also be useful to investigate the recognition process of lysine fatty-acylation by sirtuins, which involves transient interactions between the PTM and the enzymes. To this end, we designed and synthesized a chemical probe (probe 1) based on a Lys9-myristoylated histone H3 peptide (H3K9Myr), in which residue Thr6 was replaced by a diazirine-containing photoreactive amino acid, photo-Leu, to capture enzymes that recognize the myristoyl modification by converting transient protein-protein interactions into irreversible covalent linkages. The probe also has a terminal alkyne-containing amino acid (propargylglycine) at the peptide C-terminus to enable bioorthorgonal conjugation of fluorescence tags for the detection of captured proteins (Figure 1a and 1b).

Figure 1.

(a) Schematic for photo-cross-linking-based chemical approach to capture and label ‘erasers’ of protein lysine myristoylation. (b) Chemical structures of probes 1 and C. (c) Sirt6, a known lysine de-fatty-acylase, was selectively labeled by probe 1 but not probe C. (d) Labeling of the indicated sirtuins (20 µg/mL) by probe 1 (2 µM). (e) Probe 1 (1 µM) selectively labeled Sirt2 from a mixture of the indicated sirtuins (1 µM) and BSA (10 µM). (f) Probe 1 (1 µM) selectively labeled Sirt2 from whole cell lysate (1.5 mg/ml) containing the indicated recombinant sirtuins (1 µM).The proteins were incubated with probe 1 or probe C. After UV irradiation (365 nm) for 20 min, the labeled proteins were conjugated to rhodamine-azide, resolved by SDS-PAGE, and detected by in-gel fluorescence scanning. The Coomassie blue staining was used as loading controls. Rho, rhodamine fluorescence; CB, Coomassie blue.

We first tested if probe 1 can covalently label Sirt6 that is known to catalyze the removal of myristoyl group from the H3K9Myr peptide in vitro.^[6] The recombinant Sirt6 was incubated with probes 1 and a K9-unmodified control probe (probe C), respectively. After the irradiation by UV light for 20 min, the probe-labeled Sirt6 was then conjugated with rhodamine azide (Rho-N₃) via Cu(I)-catalyzed azide-alkyne cycloaddition (“click chemistry”). The resulting reaction mixtures were resolved by SDS-PAGE and analyzed by in-gel fluorescence scanning. As expected, Sirt6 was selectively labeled by myristoylated probe 1 but not by unmodified probe C (Figure 1c), suggesting that probe 1 can be used to examine the interactions between N-myristoylated lysine and its ‘erasers’. We then used probe 1 to investigate the ability of other sirtuins to recognize lysine myristoylation. In addition to Sirt6, Sirt2 and Sirt3 were also labeled by probe 1 (Figure 1d), indicating they are likely to recognize lysine myristoylation.

Notably, Sirt2 showed a much higher labeling signal than Sirt6 and Sirt3 did. This result prompted us to further compare the relative ability of these sirtuins to recognize and bind myristoyl lysine. We then mixed Sirt1, Sirt2, Sirt3 and Sirt6 at the same concentration together with a 10-fold excess amount of BSA, or total cell lysates of human HeLa S3 cells, respectively. We found that Sirt2 was preferentially labeled by probe 1 from these complex protein mixtures (Figure 1e and 1f), suggesting that among the tested sirtuins, Sirt2 has the strongest interaction with H3K9Myr. Consistent with this result, the labeling of Sirt2 was competed by a native H3K9Myr peptide with an IC₅₀ = 1.0 µM, while a much higher value of IC₅₀ (31.5 µM) was determined for the competition of Sirt6 labeling by the same peptide (Figure 2a and 2b). In addition, we found that the labeling of Sirt2 by probe 1 can also be competed by the H3 peptides with a longer (i.e., palmitoyl) or a shorter (i.e., lauroyl) acyl groups at Lys9, as well as a histone H4 peptide with myristoyl at Lys16 (Figure S1), suggesting that Sirt2 may preferentially recognize a broad scope of long-chain fatty-acylated proteins. Finally, the direct measurement of binding affinity using isothermal titration calorimetry (ITC) revealed that Sirt2 (K_d = 2.1 µM) bound the H3K9Myr peptide much tighter than Sirt6 (K_d = 32.1 µM) did (Figure 2c, 2d and S2). Interestingly, Sirt2’s interaction with this H3K9Myr peptide was also much stronger than its interaction with a K9-acetylated histone H3 (H3K9Ac) peptide (K_d > 100 µM) (Figure 2c and 2d), a known in vitro substrate of Sirt2.^[9] Taken together, Sirt2 demonstrated a distinguished ability over other sirtuin family members to recognize and interact with the lysine fatty-acylated peptide substrate. Therefore, we focused on this enzyme in our further studies.

Figure 2.

The H3K9Myr peptide inhibited the labeling of (a) Sirt2 (20 µg/mL) and (b) Sirt6 (20 µg/mL) by probe 1 (2 µM). Data are averages ± s.e. (n=3). (c) Isothermal titration calorimetry (ITC) measurement for the binding affinity of Sirt2 toward the H3K9Myr (left) and H3K9Ac (right) peptide. (d) Summary of dissociation constants (K_d), enthalpy changes (ΔH) and entropy changes (ΔS) of Sirt2 and Sirt6 for the histone peptides.

We next examined the de-fatty-acylase activity of Sirt2. High-performance liquid chromatography (HPLC), in combination with mass spectrometry and ultraviolet spectroscopy, respectively, was used to monitor the hydrolysis of the H3K9Myr peptides. Like Sirt6, Sirt2 efficiently catalyzed the removal of the myristoyl group from the peptide (Figure 3a and 3b). To quantitatively compare the demyristoylase activity of Sirt2 and Sirt6, we carried out kinetic studies on these enzymes. The steady-state kinetics data (Figure 3c) suggested that the catalytic efficiency of Sirt2 (k_cat/K_m = 1700 s⁻¹ M⁻¹) for the hydrolysis of the H3K9Myr peptide was around 2.8-fold higher than that of Sirt6 (k_cat/K_m = 608 s⁻¹ M⁻¹). Interestingly, the demyristoylation activity of Sirt2 is close to its deacetylation activity toward an H3K9Ac peptide (k_cat/K_m = 3160 s⁻¹ M⁻¹, Figure S5). The K_m value of Sirt2 for the myristoyl peptide (8.6 µM) was much lower than that for the acetylated peptide (66.7 µM), agreeing well with the determined binding affinities of the enzyme toward these two peptides (Table 1). These enzymology data demonstrated that Sirt2 can function as a lysine de-fatty-acylase in vitro.

Figure 3.

(a) LC-MS traces and (b) HPLC-UV traces showing de-myristoylation of histone H3K9-myristoylated peptide ARTKQTARK(Myr)STGGKA and KQTARK(Myr)STGGWW, respectively, catalyzed by Sirt2 and Sirt6. The enzyme (1 µM) was incubated with the myristoylated peptide (200 µM) for 60 min in the presence of 1 mM NAD, and then analyzed by LC-MS (a) and HPLC b), respectively. In (a), black traces show total ion intensity for all ion species with m/z from 300 to 2000 (i.e., total ion counts, TIC); pink traces show ion intensity (5× magnified) for the masses of demyristoylated (unmodified) H3 peptides; and blue traces show ion intensity (5× magnified) for the masses of K9-myristoylated H3 peptide. (c) Michaelis-Menten curves show the catalytic efficiencies of Sirt2 and Sirt6 toward the H3K9Myr peptide. Data are average ± s.e., n = 3.

Table 1.

The kinetic parameters of Sirt2 and Sirt6 on the acetyl and myristoyl H3K9 peptides.

Enzyme	Substrate	kcat (s−1)	Km (µM)	kcat/Km (s−1 M−1)
Sirt2	H3K9Myr	1.47 ± 0.03 × 10−2	8.62 ± 1.08	1.70 × 103
	H3K9Ac	0.211 ± 0.008	66.7 ± 9.12	3.16 × 103
Sirt6	H3K9Myr	1.12 ± 0.04 × 10−2	18.4 ± 3.16	6.08 × 102

Finally, we sought to determine whether Sirt2 can regulate lysine fatty-acylation in cells. To test this, we perturbed the enzymatic activity of Sirt2 in HeLa cells by siRNA-induced Sirt2 knockdown and a specific small-molecule inhibitor of Sirt2 (AGK2)^[10], respectively, and monitored the changes in protein fatty-acylation levels in cells using a previously developed chemical reporter, alk-14, for fatty-acylation^[11]. The HeLa cells were cultured in the presence of alk-14 that potentially labels Cys-, Lys- and N-terminal Gly-fatty-acylated proteins. After harvesting the cells, the cytosolic and nuclear lysates were prepared and subjected to an azide-alkyne click chemistry to conjugate the alk-14-labeled proteins with a rhodamine dye. The labeled proteins were then treated with hydroxylamine (NH₂OH) to uncouple the Cys-fatty-acylated proteins. The remaining Lys- and N-terminal Gly-fatty-acylated proteins were finally resolved by SDS-PAGE and visualized by in-gel fluorescent scanning (Figure 4a). As shown in Figure 4b and S6, the fluorescence signals of many protein bands labeled by alk-14 from the cytosolic lysates were significantly increased by both Sirt2 knockdown and inhibition. Given that Sirt2 cannot catalyze the hydrolysis of N-terminal Gly-fatty-acylation (Figure S3), it is likely that the lysine fatty-acylation levels of these cytosolic proteins were regulated by Sirt2 in cells. For the nuclear lysates, we focused on examining whether lysine fatty-acylation occurs on histones, which are known to carry diverse PTMs. Indeed, we found that the labeling of core histones by alk-14 was enhanced in the cells treated with Sirt2 siRNA or its inhibitor (Figure 4b), suggesting that Sirt2 could regulate lysine fatty-acylation on histones.

Figure 4.

(a) Schematic for the detection of fatty-acylated proteins using chemical reporter alk-14. (b) Fatty-acylation level of cytosolic proteins (upper panel) and core histones (lower panel) extracted from HeLa cells was regulated by Sirt2. The arrows highlight the cytosolic proteins whose fatty-acylation levels were increased after the cells were treated with Sirt2 siRNA and AGK2, a selective Sirt2 inhibitor. Rho, rhodamine fluorescence; CB, Coomassie blue, WB, Western blot;

In summary, we have used integrative chemical biology approaches to demonstrate that human Sirt2 catalyzes the hydrolysis of lysine fatty-acyl modifications in vitro and in cells. This finding opens new opportunities to investigate the biological significances of protein lysine fatty-acylation that has been under-recognized for more than two decades. The photo-cross-linking strategy used in this study should also be applicable to examine interactions between other PTMs and their ‘erasers’. On the other hand, the identification of Sirt2 as a robust de-fatty-acylase may also help to unravel unknown cellular mechanisms controlled by this enzyme that has been considered solely as a deacetylase until now. Comprehensive profiling of cellular lysine-fatty-acylated substrates of Sirt2 using chemical reporter for fatty-acylation in combination with quantitative proteomics methods is an important next step and will be reported in due course.